diff --git a/lib/libprocstat/common_kvm.c b/lib/libprocstat/common_kvm.c
index 2889f802775d..0ce2a2900f66 100644
--- a/lib/libprocstat/common_kvm.c
+++ b/lib/libprocstat/common_kvm.c
@@ -1,216 +1,215 @@
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (c) 2009 Stanislav Sedov <stas@FreeBSD.org>
* Copyright (c) 1988, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/user.h>
#include <sys/stat.h>
#include <sys/vnode.h>
#include <sys/conf.h>
-#define _KERNEL
#include <sys/pipe.h>
+#define _WANT_MOUNT
#include <sys/mount.h>
#include <ufs/ufs/quota.h>
#include <ufs/ufs/inode.h>
#include <ufs/ufs/extattr.h>
#include <ufs/ufs/ufsmount.h>
#include <fs/devfs/devfs.h>
#include <fs/devfs/devfs_int.h>
-#undef _KERNEL
#include <nfs/nfsproto.h>
#include <nfsclient/nfs.h>
#include <nfsclient/nfsnode.h>
#include <assert.h>
#include <err.h>
#include <kvm.h>
#include <stddef.h>
#include <string.h>
#include <libprocstat.h>
#include "common_kvm.h"
int
kvm_read_all(kvm_t *kd, unsigned long addr, void *buf, size_t nbytes)
{
ssize_t error;
if (nbytes >= SSIZE_MAX)
return (0);
error = kvm_read(kd, addr, buf, nbytes);
return (error == (ssize_t)(nbytes));
}
int
kdevtoname(kvm_t *kd, struct cdev *dev, char *buf)
{
struct cdev si;
assert(buf);
if (!kvm_read_all(kd, (unsigned long)dev, &si, sizeof(si)))
return (1);
strlcpy(buf, si.si_name, SPECNAMELEN + 1);
return (0);
}
int
ufs_filestat(kvm_t *kd, struct vnode *vp, struct vnstat *vn)
{
struct inode inode;
struct ufsmount um;
if (!kvm_read_all(kd, (unsigned long)VTOI(vp), &inode, sizeof(inode))) {
warnx("can't read inode at %p", (void *)VTOI(vp));
return (1);
}
if (!kvm_read_all(kd, (unsigned long)inode.i_ump, &um, sizeof(um))) {
warnx("can't read ufsmount at %p", (void *)inode.i_ump);
return (1);
}
/*
* The st_dev from stat(2) is a dev_t. These kernel structures
* contain cdev pointers. We need to convert to dev_t to make
* comparisons
*/
vn->vn_fsid = dev2udev(kd, um.um_dev);
vn->vn_fileid = inode.i_number;
vn->vn_mode = (mode_t)inode.i_mode;
vn->vn_size = inode.i_size;
return (0);
}
int
devfs_filestat(kvm_t *kd, struct vnode *vp, struct vnstat *vn)
{
struct devfs_dirent devfs_dirent;
struct mount mount;
if (!kvm_read_all(kd, (unsigned long)getvnodedata(vp), &devfs_dirent,
sizeof(devfs_dirent))) {
warnx("can't read devfs_dirent at %p",
(void *)vp->v_data);
return (1);
}
if (!kvm_read_all(kd, (unsigned long)getvnodemount(vp), &mount,
sizeof(mount))) {
warnx("can't read mount at %p",
(void *)getvnodemount(vp));
return (1);
}
vn->vn_fsid = mount.mnt_stat.f_fsid.val[0];
vn->vn_fileid = devfs_dirent.de_inode;
vn->vn_mode = (devfs_dirent.de_mode & ~S_IFMT) | S_IFCHR;
vn->vn_size = 0;
return (0);
}
int
nfs_filestat(kvm_t *kd, struct vnode *vp, struct vnstat *vn)
{
struct nfsnode nfsnode;
mode_t mode;
if (!kvm_read_all(kd, (unsigned long)VTONFS(vp), &nfsnode,
sizeof(nfsnode))) {
warnx("can't read nfsnode at %p",
(void *)VTONFS(vp));
return (1);
}
vn->vn_fsid = nfsnode.n_vattr.va_fsid;
vn->vn_fileid = nfsnode.n_vattr.va_fileid;
vn->vn_size = nfsnode.n_size;
mode = (mode_t)nfsnode.n_vattr.va_mode;
switch (vp->v_type) {
case VREG:
mode |= S_IFREG;
break;
case VDIR:
mode |= S_IFDIR;
break;
case VBLK:
mode |= S_IFBLK;
break;
case VCHR:
mode |= S_IFCHR;
break;
case VLNK:
mode |= S_IFLNK;
break;
case VSOCK:
mode |= S_IFSOCK;
break;
case VFIFO:
mode |= S_IFIFO;
break;
default:
break;
};
vn->vn_mode = mode;
return (0);
}
/*
* Read the cdev structure in the kernel in order to work out the
* associated dev_t
*/
dev_t
dev2udev(kvm_t *kd, struct cdev *dev)
{
struct cdev_priv priv;
assert(kd);
if (kvm_read_all(kd, (unsigned long)cdev2priv(dev), &priv,
sizeof(priv))) {
return ((dev_t)priv.cdp_inode);
} else {
warnx("can't convert cdev *%p to a dev_t\n", dev);
return (-1);
}
}
void *
getvnodedata(struct vnode *vp)
{
return (vp->v_data);
}
struct mount *
getvnodemount(struct vnode *vp)
{
return (vp->v_mount);
}
diff --git a/lib/libprocstat/libprocstat.c b/lib/libprocstat/libprocstat.c
index a2de61855815..7ccf6c343705 100644
--- a/lib/libprocstat/libprocstat.c
+++ b/lib/libprocstat/libprocstat.c
@@ -1,2650 +1,2647 @@
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (c) 2017 Dell EMC
* Copyright (c) 2009 Stanislav Sedov <stas@FreeBSD.org>
* Copyright (c) 1988, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/elf.h>
#include <sys/time.h>
#include <sys/resourcevar.h>
#define _WANT_UCRED
#include <sys/ucred.h>
#undef _WANT_UCRED
#include <sys/proc.h>
#include <sys/user.h>
#include <sys/stat.h>
#include <sys/vnode.h>
#include <sys/socket.h>
#define _WANT_SOCKET
#include <sys/socketvar.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <sys/un.h>
#define _WANT_UNPCB
#include <sys/unpcb.h>
#include <sys/sysctl.h>
#include <sys/tty.h>
#include <sys/filedesc.h>
#include <sys/queue.h>
#define _WANT_FILE
#include <sys/file.h>
#include <sys/conf.h>
#include <sys/ksem.h>
#include <sys/mman.h>
#include <sys/capsicum.h>
#include <sys/ptrace.h>
-#define _KERNEL
+#define _WANT_MOUNT
#include <sys/mount.h>
#include <sys/filedesc.h>
#include <sys/pipe.h>
-#include <ufs/ufs/quota.h>
-#include <ufs/ufs/inode.h>
#include <fs/devfs/devfs.h>
#include <fs/devfs/devfs_int.h>
-#undef _KERNEL
#include <nfs/nfsproto.h>
#include <nfsclient/nfs.h>
#include <nfsclient/nfsnode.h>
#include <vm/vm.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <net/route.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#define _WANT_INPCB
#include <netinet/in_pcb.h>
#include <assert.h>
#include <ctype.h>
#include <err.h>
#include <fcntl.h>
#include <kvm.h>
#include <libutil.h>
#include <limits.h>
#include <paths.h>
#include <pwd.h>
#include <stdio.h>
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
#include <unistd.h>
#include <netdb.h>
#include <libprocstat.h>
#include "libprocstat_internal.h"
#include "common_kvm.h"
#include "core.h"
int statfs(const char *, struct statfs *); /* XXX */
#define PROCSTAT_KVM 1
#define PROCSTAT_SYSCTL 2
#define PROCSTAT_CORE 3
static char **getargv(struct procstat *procstat, struct kinfo_proc *kp,
size_t nchr, int env);
static char *getmnton(kvm_t *kd, struct mount *m);
static struct kinfo_vmentry * kinfo_getvmmap_core(struct procstat_core *core,
int *cntp);
static Elf_Auxinfo *procstat_getauxv_core(struct procstat_core *core,
unsigned int *cntp);
static Elf_Auxinfo *procstat_getauxv_sysctl(pid_t pid, unsigned int *cntp);
static struct filestat_list *procstat_getfiles_kvm(
struct procstat *procstat, struct kinfo_proc *kp, int mmapped);
static struct filestat_list *procstat_getfiles_sysctl(
struct procstat *procstat, struct kinfo_proc *kp, int mmapped);
static int procstat_get_pipe_info_sysctl(struct filestat *fst,
struct pipestat *pipe, char *errbuf);
static int procstat_get_pipe_info_kvm(kvm_t *kd, struct filestat *fst,
struct pipestat *pipe, char *errbuf);
static int procstat_get_pts_info_sysctl(struct filestat *fst,
struct ptsstat *pts, char *errbuf);
static int procstat_get_pts_info_kvm(kvm_t *kd, struct filestat *fst,
struct ptsstat *pts, char *errbuf);
static int procstat_get_sem_info_sysctl(struct filestat *fst,
struct semstat *sem, char *errbuf);
static int procstat_get_sem_info_kvm(kvm_t *kd, struct filestat *fst,
struct semstat *sem, char *errbuf);
static int procstat_get_shm_info_sysctl(struct filestat *fst,
struct shmstat *shm, char *errbuf);
static int procstat_get_shm_info_kvm(kvm_t *kd, struct filestat *fst,
struct shmstat *shm, char *errbuf);
static int procstat_get_socket_info_sysctl(struct filestat *fst,
struct sockstat *sock, char *errbuf);
static int procstat_get_socket_info_kvm(kvm_t *kd, struct filestat *fst,
struct sockstat *sock, char *errbuf);
static int to_filestat_flags(int flags);
static int procstat_get_vnode_info_kvm(kvm_t *kd, struct filestat *fst,
struct vnstat *vn, char *errbuf);
static int procstat_get_vnode_info_sysctl(struct filestat *fst,
struct vnstat *vn, char *errbuf);
static gid_t *procstat_getgroups_core(struct procstat_core *core,
unsigned int *count);
static gid_t * procstat_getgroups_kvm(kvm_t *kd, struct kinfo_proc *kp,
unsigned int *count);
static gid_t *procstat_getgroups_sysctl(pid_t pid, unsigned int *count);
static struct kinfo_kstack *procstat_getkstack_sysctl(pid_t pid,
int *cntp);
static int procstat_getosrel_core(struct procstat_core *core,
int *osrelp);
static int procstat_getosrel_kvm(kvm_t *kd, struct kinfo_proc *kp,
int *osrelp);
static int procstat_getosrel_sysctl(pid_t pid, int *osrelp);
static int procstat_getpathname_core(struct procstat_core *core,
char *pathname, size_t maxlen);
static int procstat_getpathname_sysctl(pid_t pid, char *pathname,
size_t maxlen);
static int procstat_getrlimit_core(struct procstat_core *core, int which,
struct rlimit* rlimit);
static int procstat_getrlimit_kvm(kvm_t *kd, struct kinfo_proc *kp,
int which, struct rlimit* rlimit);
static int procstat_getrlimit_sysctl(pid_t pid, int which,
struct rlimit* rlimit);
static int procstat_getumask_core(struct procstat_core *core,
unsigned short *maskp);
static int procstat_getumask_kvm(kvm_t *kd, struct kinfo_proc *kp,
unsigned short *maskp);
static int procstat_getumask_sysctl(pid_t pid, unsigned short *maskp);
static int vntype2psfsttype(int type);
void
procstat_close(struct procstat *procstat)
{
assert(procstat);
if (procstat->type == PROCSTAT_KVM)
kvm_close(procstat->kd);
else if (procstat->type == PROCSTAT_CORE)
procstat_core_close(procstat->core);
procstat_freeargv(procstat);
procstat_freeenvv(procstat);
free(procstat);
}
struct procstat *
procstat_open_sysctl(void)
{
struct procstat *procstat;
procstat = calloc(1, sizeof(*procstat));
if (procstat == NULL) {
warn("malloc()");
return (NULL);
}
procstat->type = PROCSTAT_SYSCTL;
return (procstat);
}
struct procstat *
procstat_open_kvm(const char *nlistf, const char *memf)
{
struct procstat *procstat;
kvm_t *kd;
char buf[_POSIX2_LINE_MAX];
procstat = calloc(1, sizeof(*procstat));
if (procstat == NULL) {
warn("malloc()");
return (NULL);
}
kd = kvm_openfiles(nlistf, memf, NULL, O_RDONLY, buf);
if (kd == NULL) {
warnx("kvm_openfiles(): %s", buf);
free(procstat);
return (NULL);
}
procstat->type = PROCSTAT_KVM;
procstat->kd = kd;
return (procstat);
}
struct procstat *
procstat_open_core(const char *filename)
{
struct procstat *procstat;
struct procstat_core *core;
procstat = calloc(1, sizeof(*procstat));
if (procstat == NULL) {
warn("malloc()");
return (NULL);
}
core = procstat_core_open(filename);
if (core == NULL) {
free(procstat);
return (NULL);
}
procstat->type = PROCSTAT_CORE;
procstat->core = core;
return (procstat);
}
struct kinfo_proc *
procstat_getprocs(struct procstat *procstat, int what, int arg,
unsigned int *count)
{
struct kinfo_proc *p0, *p;
size_t len, olen;
int name[4];
int cnt;
int error;
assert(procstat);
assert(count);
p = NULL;
if (procstat->type == PROCSTAT_KVM) {
*count = 0;
p0 = kvm_getprocs(procstat->kd, what, arg, &cnt);
if (p0 == NULL || cnt <= 0)
return (NULL);
*count = cnt;
len = *count * sizeof(*p);
p = malloc(len);
if (p == NULL) {
warnx("malloc(%zu)", len);
goto fail;
}
bcopy(p0, p, len);
return (p);
} else if (procstat->type == PROCSTAT_SYSCTL) {
len = 0;
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = what;
name[3] = arg;
error = sysctl(name, nitems(name), NULL, &len, NULL, 0);
if (error < 0 && errno != EPERM) {
warn("sysctl(kern.proc)");
goto fail;
}
if (len == 0) {
warnx("no processes?");
goto fail;
}
do {
len += len / 10;
p = reallocf(p, len);
if (p == NULL) {
warnx("reallocf(%zu)", len);
goto fail;
}
olen = len;
error = sysctl(name, nitems(name), p, &len, NULL, 0);
} while (error < 0 && errno == ENOMEM && olen == len);
if (error < 0 && errno != EPERM) {
warn("sysctl(kern.proc)");
goto fail;
}
/* Perform simple consistency checks. */
if ((len % sizeof(*p)) != 0 || p->ki_structsize != sizeof(*p)) {
warnx("kinfo_proc structure size mismatch (len = %zu)", len);
goto fail;
}
*count = len / sizeof(*p);
return (p);
} else if (procstat->type == PROCSTAT_CORE) {
p = procstat_core_get(procstat->core, PSC_TYPE_PROC, NULL,
&len);
if ((len % sizeof(*p)) != 0 || p->ki_structsize != sizeof(*p)) {
warnx("kinfo_proc structure size mismatch");
goto fail;
}
*count = len / sizeof(*p);
return (p);
} else {
warnx("unknown access method: %d", procstat->type);
return (NULL);
}
fail:
if (p)
free(p);
return (NULL);
}
void
procstat_freeprocs(struct procstat *procstat __unused, struct kinfo_proc *p)
{
if (p != NULL)
free(p);
p = NULL;
}
struct filestat_list *
procstat_getfiles(struct procstat *procstat, struct kinfo_proc *kp, int mmapped)
{
switch(procstat->type) {
case PROCSTAT_KVM:
return (procstat_getfiles_kvm(procstat, kp, mmapped));
case PROCSTAT_SYSCTL:
case PROCSTAT_CORE:
return (procstat_getfiles_sysctl(procstat, kp, mmapped));
default:
warnx("unknown access method: %d", procstat->type);
return (NULL);
}
}
void
procstat_freefiles(struct procstat *procstat, struct filestat_list *head)
{
struct filestat *fst, *tmp;
STAILQ_FOREACH_SAFE(fst, head, next, tmp) {
if (fst->fs_path != NULL)
free(fst->fs_path);
free(fst);
}
free(head);
if (procstat->vmentries != NULL) {
free(procstat->vmentries);
procstat->vmentries = NULL;
}
if (procstat->files != NULL) {
free(procstat->files);
procstat->files = NULL;
}
}
static struct filestat *
filestat_new_entry(void *typedep, int type, int fd, int fflags, int uflags,
int refcount, off_t offset, char *path, cap_rights_t *cap_rightsp)
{
struct filestat *entry;
entry = calloc(1, sizeof(*entry));
if (entry == NULL) {
warn("malloc()");
return (NULL);
}
entry->fs_typedep = typedep;
entry->fs_fflags = fflags;
entry->fs_uflags = uflags;
entry->fs_fd = fd;
entry->fs_type = type;
entry->fs_ref_count = refcount;
entry->fs_offset = offset;
entry->fs_path = path;
if (cap_rightsp != NULL)
entry->fs_cap_rights = *cap_rightsp;
else
cap_rights_init(&entry->fs_cap_rights);
return (entry);
}
static struct vnode *
getctty(kvm_t *kd, struct kinfo_proc *kp)
{
struct pgrp pgrp;
struct proc proc;
struct session sess;
int error;
assert(kp);
error = kvm_read_all(kd, (unsigned long)kp->ki_paddr, &proc,
sizeof(proc));
if (error == 0) {
warnx("can't read proc struct at %p for pid %d",
kp->ki_paddr, kp->ki_pid);
return (NULL);
}
if (proc.p_pgrp == NULL)
return (NULL);
error = kvm_read_all(kd, (unsigned long)proc.p_pgrp, &pgrp,
sizeof(pgrp));
if (error == 0) {
warnx("can't read pgrp struct at %p for pid %d",
proc.p_pgrp, kp->ki_pid);
return (NULL);
}
error = kvm_read_all(kd, (unsigned long)pgrp.pg_session, &sess,
sizeof(sess));
if (error == 0) {
warnx("can't read session struct at %p for pid %d",
pgrp.pg_session, kp->ki_pid);
return (NULL);
}
return (sess.s_ttyvp);
}
static int
procstat_vm_map_reader(void *token, vm_map_entry_t addr, vm_map_entry_t dest)
{
kvm_t *kd;
kd = (kvm_t *)token;
return (kvm_read_all(kd, (unsigned long)addr, dest, sizeof(*dest)));
}
static struct filestat_list *
procstat_getfiles_kvm(struct procstat *procstat, struct kinfo_proc *kp, int mmapped)
{
struct file file;
struct filedesc filed;
struct pwddesc pathsd;
struct fdescenttbl *fdt;
struct pwd pwd;
unsigned long pwd_addr;
struct vm_map_entry vmentry;
struct vm_object object;
struct vmspace vmspace;
vm_map_entry_t entryp;
vm_object_t objp;
struct vnode *vp;
struct filestat *entry;
struct filestat_list *head;
kvm_t *kd;
void *data;
int fflags;
unsigned int i;
int prot, type;
size_t fdt_size;
unsigned int nfiles;
bool haspwd;
assert(procstat);
kd = procstat->kd;
if (kd == NULL)
return (NULL);
if (kp->ki_fd == NULL || kp->ki_pd == NULL)
return (NULL);
if (!kvm_read_all(kd, (unsigned long)kp->ki_fd, &filed,
sizeof(filed))) {
warnx("can't read filedesc at %p", (void *)kp->ki_fd);
return (NULL);
}
if (!kvm_read_all(kd, (unsigned long)kp->ki_pd, &pathsd,
sizeof(pathsd))) {
warnx("can't read pwddesc at %p", (void *)kp->ki_pd);
return (NULL);
}
haspwd = false;
pwd_addr = (unsigned long)(PWDDESC_KVM_LOAD_PWD(&pathsd));
if (pwd_addr != 0) {
if (!kvm_read_all(kd, pwd_addr, &pwd, sizeof(pwd))) {
warnx("can't read fd_pwd at %p", (void *)pwd_addr);
return (NULL);
}
haspwd = true;
}
/*
* Allocate list head.
*/
head = malloc(sizeof(*head));
if (head == NULL)
return (NULL);
STAILQ_INIT(head);
/* root directory vnode, if one. */
if (haspwd) {
if (pwd.pwd_rdir) {
entry = filestat_new_entry(pwd.pwd_rdir, PS_FST_TYPE_VNODE, -1,
PS_FST_FFLAG_READ, PS_FST_UFLAG_RDIR, 0, 0, NULL, NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
/* current working directory vnode. */
if (pwd.pwd_cdir) {
entry = filestat_new_entry(pwd.pwd_cdir, PS_FST_TYPE_VNODE, -1,
PS_FST_FFLAG_READ, PS_FST_UFLAG_CDIR, 0, 0, NULL, NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
/* jail root, if any. */
if (pwd.pwd_jdir) {
entry = filestat_new_entry(pwd.pwd_jdir, PS_FST_TYPE_VNODE, -1,
PS_FST_FFLAG_READ, PS_FST_UFLAG_JAIL, 0, 0, NULL, NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
}
/* ktrace vnode, if one */
if (kp->ki_tracep) {
entry = filestat_new_entry(kp->ki_tracep, PS_FST_TYPE_VNODE, -1,
PS_FST_FFLAG_READ | PS_FST_FFLAG_WRITE,
PS_FST_UFLAG_TRACE, 0, 0, NULL, NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
/* text vnode, if one */
if (kp->ki_textvp) {
entry = filestat_new_entry(kp->ki_textvp, PS_FST_TYPE_VNODE, -1,
PS_FST_FFLAG_READ, PS_FST_UFLAG_TEXT, 0, 0, NULL, NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
/* Controlling terminal. */
if ((vp = getctty(kd, kp)) != NULL) {
entry = filestat_new_entry(vp, PS_FST_TYPE_VNODE, -1,
PS_FST_FFLAG_READ | PS_FST_FFLAG_WRITE,
PS_FST_UFLAG_CTTY, 0, 0, NULL, NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
if (!kvm_read_all(kd, (unsigned long)filed.fd_files, &nfiles,
sizeof(nfiles))) {
warnx("can't read fd_files at %p", (void *)filed.fd_files);
return (NULL);
}
fdt_size = sizeof(*fdt) + nfiles * sizeof(struct filedescent);
fdt = malloc(fdt_size);
if (fdt == NULL) {
warn("malloc(%zu)", fdt_size);
goto do_mmapped;
}
if (!kvm_read_all(kd, (unsigned long)filed.fd_files, fdt, fdt_size)) {
warnx("cannot read file structures at %p", (void *)filed.fd_files);
free(fdt);
goto do_mmapped;
}
for (i = 0; i < nfiles; i++) {
if (fdt->fdt_ofiles[i].fde_file == NULL) {
continue;
}
if (!kvm_read_all(kd, (unsigned long)fdt->fdt_ofiles[i].fde_file, &file,
sizeof(struct file))) {
warnx("can't read file %d at %p", i,
(void *)fdt->fdt_ofiles[i].fde_file);
continue;
}
switch (file.f_type) {
case DTYPE_VNODE:
type = PS_FST_TYPE_VNODE;
data = file.f_vnode;
break;
case DTYPE_SOCKET:
type = PS_FST_TYPE_SOCKET;
data = file.f_data;
break;
case DTYPE_PIPE:
type = PS_FST_TYPE_PIPE;
data = file.f_data;
break;
case DTYPE_FIFO:
type = PS_FST_TYPE_FIFO;
data = file.f_vnode;
break;
#ifdef DTYPE_PTS
case DTYPE_PTS:
type = PS_FST_TYPE_PTS;
data = file.f_data;
break;
#endif
case DTYPE_SEM:
type = PS_FST_TYPE_SEM;
data = file.f_data;
break;
case DTYPE_SHM:
type = PS_FST_TYPE_SHM;
data = file.f_data;
break;
case DTYPE_PROCDESC:
type = PS_FST_TYPE_PROCDESC;
data = file.f_data;
break;
case DTYPE_DEV:
type = PS_FST_TYPE_DEV;
data = file.f_data;
break;
case DTYPE_EVENTFD:
type = PS_FST_TYPE_EVENTFD;
data = file.f_data;
break;
default:
continue;
}
/* XXXRW: No capability rights support for kvm yet. */
entry = filestat_new_entry(data, type, i,
to_filestat_flags(file.f_flag), 0, 0, 0, NULL, NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
free(fdt);
do_mmapped:
/*
* Process mmapped files if requested.
*/
if (mmapped) {
if (!kvm_read_all(kd, (unsigned long)kp->ki_vmspace, &vmspace,
sizeof(vmspace))) {
warnx("can't read vmspace at %p",
(void *)kp->ki_vmspace);
goto exit;
}
vmentry = vmspace.vm_map.header;
for (entryp = vm_map_entry_read_succ(kd, &vmentry, procstat_vm_map_reader);
entryp != NULL && entryp != &kp->ki_vmspace->vm_map.header;
entryp = vm_map_entry_read_succ(kd, &vmentry, procstat_vm_map_reader)) {
if (vmentry.eflags & MAP_ENTRY_IS_SUB_MAP)
continue;
if ((objp = vmentry.object.vm_object) == NULL)
continue;
for (; objp; objp = object.backing_object) {
if (!kvm_read_all(kd, (unsigned long)objp,
&object, sizeof(object))) {
warnx("can't read vm_object at %p",
(void *)objp);
break;
}
}
/* We want only vnode objects. */
if (object.type != OBJT_VNODE)
continue;
prot = vmentry.protection;
fflags = 0;
if (prot & VM_PROT_READ)
fflags = PS_FST_FFLAG_READ;
if ((vmentry.eflags & MAP_ENTRY_COW) == 0 &&
prot & VM_PROT_WRITE)
fflags |= PS_FST_FFLAG_WRITE;
/*
* Create filestat entry.
*/
entry = filestat_new_entry(object.handle,
PS_FST_TYPE_VNODE, -1, fflags,
PS_FST_UFLAG_MMAP, 0, 0, NULL, NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
if (entryp == NULL)
warnx("can't read vm_map_entry");
}
exit:
return (head);
}
/*
* kinfo types to filestat translation.
*/
static int
kinfo_type2fst(int kftype)
{
static struct {
int kf_type;
int fst_type;
} kftypes2fst[] = {
{ KF_TYPE_PROCDESC, PS_FST_TYPE_PROCDESC },
{ KF_TYPE_DEV, PS_FST_TYPE_DEV },
{ KF_TYPE_FIFO, PS_FST_TYPE_FIFO },
{ KF_TYPE_KQUEUE, PS_FST_TYPE_KQUEUE },
{ KF_TYPE_MQUEUE, PS_FST_TYPE_MQUEUE },
{ KF_TYPE_NONE, PS_FST_TYPE_NONE },
{ KF_TYPE_PIPE, PS_FST_TYPE_PIPE },
{ KF_TYPE_PTS, PS_FST_TYPE_PTS },
{ KF_TYPE_SEM, PS_FST_TYPE_SEM },
{ KF_TYPE_SHM, PS_FST_TYPE_SHM },
{ KF_TYPE_SOCKET, PS_FST_TYPE_SOCKET },
{ KF_TYPE_VNODE, PS_FST_TYPE_VNODE },
{ KF_TYPE_EVENTFD, PS_FST_TYPE_EVENTFD },
{ KF_TYPE_UNKNOWN, PS_FST_TYPE_UNKNOWN }
};
#define NKFTYPES (sizeof(kftypes2fst) / sizeof(*kftypes2fst))
unsigned int i;
for (i = 0; i < NKFTYPES; i++)
if (kftypes2fst[i].kf_type == kftype)
break;
if (i == NKFTYPES)
return (PS_FST_TYPE_UNKNOWN);
return (kftypes2fst[i].fst_type);
}
/*
* kinfo flags to filestat translation.
*/
static int
kinfo_fflags2fst(int kfflags)
{
static struct {
int kf_flag;
int fst_flag;
} kfflags2fst[] = {
{ KF_FLAG_APPEND, PS_FST_FFLAG_APPEND },
{ KF_FLAG_ASYNC, PS_FST_FFLAG_ASYNC },
{ KF_FLAG_CREAT, PS_FST_FFLAG_CREAT },
{ KF_FLAG_DIRECT, PS_FST_FFLAG_DIRECT },
{ KF_FLAG_EXCL, PS_FST_FFLAG_EXCL },
{ KF_FLAG_EXEC, PS_FST_FFLAG_EXEC },
{ KF_FLAG_EXLOCK, PS_FST_FFLAG_EXLOCK },
{ KF_FLAG_FSYNC, PS_FST_FFLAG_SYNC },
{ KF_FLAG_HASLOCK, PS_FST_FFLAG_HASLOCK },
{ KF_FLAG_NOFOLLOW, PS_FST_FFLAG_NOFOLLOW },
{ KF_FLAG_NONBLOCK, PS_FST_FFLAG_NONBLOCK },
{ KF_FLAG_READ, PS_FST_FFLAG_READ },
{ KF_FLAG_SHLOCK, PS_FST_FFLAG_SHLOCK },
{ KF_FLAG_TRUNC, PS_FST_FFLAG_TRUNC },
{ KF_FLAG_WRITE, PS_FST_FFLAG_WRITE }
};
#define NKFFLAGS (sizeof(kfflags2fst) / sizeof(*kfflags2fst))
unsigned int i;
int flags;
flags = 0;
for (i = 0; i < NKFFLAGS; i++)
if ((kfflags & kfflags2fst[i].kf_flag) != 0)
flags |= kfflags2fst[i].fst_flag;
return (flags);
}
static int
kinfo_uflags2fst(int fd)
{
switch (fd) {
case KF_FD_TYPE_CTTY:
return (PS_FST_UFLAG_CTTY);
case KF_FD_TYPE_CWD:
return (PS_FST_UFLAG_CDIR);
case KF_FD_TYPE_JAIL:
return (PS_FST_UFLAG_JAIL);
case KF_FD_TYPE_TEXT:
return (PS_FST_UFLAG_TEXT);
case KF_FD_TYPE_TRACE:
return (PS_FST_UFLAG_TRACE);
case KF_FD_TYPE_ROOT:
return (PS_FST_UFLAG_RDIR);
}
return (0);
}
static struct kinfo_file *
kinfo_getfile_core(struct procstat_core *core, int *cntp)
{
int cnt;
size_t len;
char *buf, *bp, *eb;
struct kinfo_file *kif, *kp, *kf;
buf = procstat_core_get(core, PSC_TYPE_FILES, NULL, &len);
if (buf == NULL)
return (NULL);
/*
* XXXMG: The code below is just copy&past from libutil.
* The code duplication can be avoided if libutil
* is extended to provide something like:
* struct kinfo_file *kinfo_getfile_from_buf(const char *buf,
* size_t len, int *cntp);
*/
/* Pass 1: count items */
cnt = 0;
bp = buf;
eb = buf + len;
while (bp < eb) {
kf = (struct kinfo_file *)(uintptr_t)bp;
if (kf->kf_structsize == 0)
break;
bp += kf->kf_structsize;
cnt++;
}
kif = calloc(cnt, sizeof(*kif));
if (kif == NULL) {
free(buf);
return (NULL);
}
bp = buf;
eb = buf + len;
kp = kif;
/* Pass 2: unpack */
while (bp < eb) {
kf = (struct kinfo_file *)(uintptr_t)bp;
if (kf->kf_structsize == 0)
break;
/* Copy/expand into pre-zeroed buffer */
memcpy(kp, kf, kf->kf_structsize);
/* Advance to next packed record */
bp += kf->kf_structsize;
/* Set field size to fixed length, advance */
kp->kf_structsize = sizeof(*kp);
kp++;
}
free(buf);
*cntp = cnt;
return (kif); /* Caller must free() return value */
}
static struct filestat_list *
procstat_getfiles_sysctl(struct procstat *procstat, struct kinfo_proc *kp,
int mmapped)
{
struct kinfo_file *kif, *files;
struct kinfo_vmentry *kve, *vmentries;
struct filestat_list *head;
struct filestat *entry;
char *path;
off_t offset;
int cnt, fd, fflags;
int i, type, uflags;
int refcount;
cap_rights_t cap_rights;
assert(kp);
if (kp->ki_fd == NULL)
return (NULL);
switch(procstat->type) {
case PROCSTAT_SYSCTL:
files = kinfo_getfile(kp->ki_pid, &cnt);
break;
case PROCSTAT_CORE:
files = kinfo_getfile_core(procstat->core, &cnt);
break;
default:
assert(!"invalid type");
}
if (files == NULL && errno != EPERM) {
warn("kinfo_getfile()");
return (NULL);
}
procstat->files = files;
/*
* Allocate list head.
*/
head = malloc(sizeof(*head));
if (head == NULL)
return (NULL);
STAILQ_INIT(head);
for (i = 0; i < cnt; i++) {
kif = &files[i];
type = kinfo_type2fst(kif->kf_type);
fd = kif->kf_fd >= 0 ? kif->kf_fd : -1;
fflags = kinfo_fflags2fst(kif->kf_flags);
uflags = kinfo_uflags2fst(kif->kf_fd);
refcount = kif->kf_ref_count;
offset = kif->kf_offset;
if (*kif->kf_path != '\0')
path = strdup(kif->kf_path);
else
path = NULL;
cap_rights = kif->kf_cap_rights;
/*
* Create filestat entry.
*/
entry = filestat_new_entry(kif, type, fd, fflags, uflags,
refcount, offset, path, &cap_rights);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
if (mmapped != 0) {
vmentries = procstat_getvmmap(procstat, kp, &cnt);
procstat->vmentries = vmentries;
if (vmentries == NULL || cnt == 0)
goto fail;
for (i = 0; i < cnt; i++) {
kve = &vmentries[i];
if (kve->kve_type != KVME_TYPE_VNODE)
continue;
fflags = 0;
if (kve->kve_protection & KVME_PROT_READ)
fflags = PS_FST_FFLAG_READ;
if ((kve->kve_flags & KVME_FLAG_COW) == 0 &&
kve->kve_protection & KVME_PROT_WRITE)
fflags |= PS_FST_FFLAG_WRITE;
offset = kve->kve_offset;
refcount = kve->kve_ref_count;
if (*kve->kve_path != '\0')
path = strdup(kve->kve_path);
else
path = NULL;
entry = filestat_new_entry(kve, PS_FST_TYPE_VNODE, -1,
fflags, PS_FST_UFLAG_MMAP, refcount, offset, path,
NULL);
if (entry != NULL)
STAILQ_INSERT_TAIL(head, entry, next);
}
}
fail:
return (head);
}
int
procstat_get_pipe_info(struct procstat *procstat, struct filestat *fst,
struct pipestat *ps, char *errbuf)
{
assert(ps);
if (procstat->type == PROCSTAT_KVM) {
return (procstat_get_pipe_info_kvm(procstat->kd, fst, ps,
errbuf));
} else if (procstat->type == PROCSTAT_SYSCTL ||
procstat->type == PROCSTAT_CORE) {
return (procstat_get_pipe_info_sysctl(fst, ps, errbuf));
} else {
warnx("unknown access method: %d", procstat->type);
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
}
static int
procstat_get_pipe_info_kvm(kvm_t *kd, struct filestat *fst,
struct pipestat *ps, char *errbuf)
{
struct pipe pi;
void *pipep;
assert(kd);
assert(ps);
assert(fst);
bzero(ps, sizeof(*ps));
pipep = fst->fs_typedep;
if (pipep == NULL)
goto fail;
if (!kvm_read_all(kd, (unsigned long)pipep, &pi, sizeof(struct pipe))) {
warnx("can't read pipe at %p", (void *)pipep);
goto fail;
}
ps->addr = (uintptr_t)pipep;
ps->peer = (uintptr_t)pi.pipe_peer;
ps->buffer_cnt = pi.pipe_buffer.cnt;
return (0);
fail:
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
static int
procstat_get_pipe_info_sysctl(struct filestat *fst, struct pipestat *ps,
char *errbuf __unused)
{
struct kinfo_file *kif;
assert(ps);
assert(fst);
bzero(ps, sizeof(*ps));
kif = fst->fs_typedep;
if (kif == NULL)
return (1);
ps->addr = kif->kf_un.kf_pipe.kf_pipe_addr;
ps->peer = kif->kf_un.kf_pipe.kf_pipe_peer;
ps->buffer_cnt = kif->kf_un.kf_pipe.kf_pipe_buffer_cnt;
return (0);
}
int
procstat_get_pts_info(struct procstat *procstat, struct filestat *fst,
struct ptsstat *pts, char *errbuf)
{
assert(pts);
if (procstat->type == PROCSTAT_KVM) {
return (procstat_get_pts_info_kvm(procstat->kd, fst, pts,
errbuf));
} else if (procstat->type == PROCSTAT_SYSCTL ||
procstat->type == PROCSTAT_CORE) {
return (procstat_get_pts_info_sysctl(fst, pts, errbuf));
} else {
warnx("unknown access method: %d", procstat->type);
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
}
static int
procstat_get_pts_info_kvm(kvm_t *kd, struct filestat *fst,
struct ptsstat *pts, char *errbuf)
{
struct tty tty;
void *ttyp;
assert(kd);
assert(pts);
assert(fst);
bzero(pts, sizeof(*pts));
ttyp = fst->fs_typedep;
if (ttyp == NULL)
goto fail;
if (!kvm_read_all(kd, (unsigned long)ttyp, &tty, sizeof(struct tty))) {
warnx("can't read tty at %p", (void *)ttyp);
goto fail;
}
pts->dev = dev2udev(kd, tty.t_dev);
(void)kdevtoname(kd, tty.t_dev, pts->devname);
return (0);
fail:
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
static int
procstat_get_pts_info_sysctl(struct filestat *fst, struct ptsstat *pts,
char *errbuf __unused)
{
struct kinfo_file *kif;
assert(pts);
assert(fst);
bzero(pts, sizeof(*pts));
kif = fst->fs_typedep;
if (kif == NULL)
return (0);
pts->dev = kif->kf_un.kf_pts.kf_pts_dev;
strlcpy(pts->devname, kif->kf_path, sizeof(pts->devname));
return (0);
}
int
procstat_get_sem_info(struct procstat *procstat, struct filestat *fst,
struct semstat *sem, char *errbuf)
{
assert(sem);
if (procstat->type == PROCSTAT_KVM) {
return (procstat_get_sem_info_kvm(procstat->kd, fst, sem,
errbuf));
} else if (procstat->type == PROCSTAT_SYSCTL ||
procstat->type == PROCSTAT_CORE) {
return (procstat_get_sem_info_sysctl(fst, sem, errbuf));
} else {
warnx("unknown access method: %d", procstat->type);
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
}
static int
procstat_get_sem_info_kvm(kvm_t *kd, struct filestat *fst,
struct semstat *sem, char *errbuf)
{
struct ksem ksem;
void *ksemp;
char *path;
int i;
assert(kd);
assert(sem);
assert(fst);
bzero(sem, sizeof(*sem));
ksemp = fst->fs_typedep;
if (ksemp == NULL)
goto fail;
if (!kvm_read_all(kd, (unsigned long)ksemp, &ksem,
sizeof(struct ksem))) {
warnx("can't read ksem at %p", (void *)ksemp);
goto fail;
}
sem->mode = S_IFREG | ksem.ks_mode;
sem->value = ksem.ks_value;
if (fst->fs_path == NULL && ksem.ks_path != NULL) {
path = malloc(MAXPATHLEN);
for (i = 0; i < MAXPATHLEN - 1; i++) {
if (!kvm_read_all(kd, (unsigned long)ksem.ks_path + i,
path + i, 1))
break;
if (path[i] == '\0')
break;
}
path[i] = '\0';
if (i == 0)
free(path);
else
fst->fs_path = path;
}
return (0);
fail:
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
static int
procstat_get_sem_info_sysctl(struct filestat *fst, struct semstat *sem,
char *errbuf __unused)
{
struct kinfo_file *kif;
assert(sem);
assert(fst);
bzero(sem, sizeof(*sem));
kif = fst->fs_typedep;
if (kif == NULL)
return (0);
sem->value = kif->kf_un.kf_sem.kf_sem_value;
sem->mode = kif->kf_un.kf_sem.kf_sem_mode;
return (0);
}
int
procstat_get_shm_info(struct procstat *procstat, struct filestat *fst,
struct shmstat *shm, char *errbuf)
{
assert(shm);
if (procstat->type == PROCSTAT_KVM) {
return (procstat_get_shm_info_kvm(procstat->kd, fst, shm,
errbuf));
} else if (procstat->type == PROCSTAT_SYSCTL ||
procstat->type == PROCSTAT_CORE) {
return (procstat_get_shm_info_sysctl(fst, shm, errbuf));
} else {
warnx("unknown access method: %d", procstat->type);
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
}
static int
procstat_get_shm_info_kvm(kvm_t *kd, struct filestat *fst,
struct shmstat *shm, char *errbuf)
{
struct shmfd shmfd;
void *shmfdp;
char *path;
int i;
assert(kd);
assert(shm);
assert(fst);
bzero(shm, sizeof(*shm));
shmfdp = fst->fs_typedep;
if (shmfdp == NULL)
goto fail;
if (!kvm_read_all(kd, (unsigned long)shmfdp, &shmfd,
sizeof(struct shmfd))) {
warnx("can't read shmfd at %p", (void *)shmfdp);
goto fail;
}
shm->mode = S_IFREG | shmfd.shm_mode;
shm->size = shmfd.shm_size;
if (fst->fs_path == NULL && shmfd.shm_path != NULL) {
path = malloc(MAXPATHLEN);
for (i = 0; i < MAXPATHLEN - 1; i++) {
if (!kvm_read_all(kd, (unsigned long)shmfd.shm_path + i,
path + i, 1))
break;
if (path[i] == '\0')
break;
}
path[i] = '\0';
if (i == 0)
free(path);
else
fst->fs_path = path;
}
return (0);
fail:
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
static int
procstat_get_shm_info_sysctl(struct filestat *fst, struct shmstat *shm,
char *errbuf __unused)
{
struct kinfo_file *kif;
assert(shm);
assert(fst);
bzero(shm, sizeof(*shm));
kif = fst->fs_typedep;
if (kif == NULL)
return (0);
shm->size = kif->kf_un.kf_file.kf_file_size;
shm->mode = kif->kf_un.kf_file.kf_file_mode;
return (0);
}
int
procstat_get_vnode_info(struct procstat *procstat, struct filestat *fst,
struct vnstat *vn, char *errbuf)
{
assert(vn);
if (procstat->type == PROCSTAT_KVM) {
return (procstat_get_vnode_info_kvm(procstat->kd, fst, vn,
errbuf));
} else if (procstat->type == PROCSTAT_SYSCTL ||
procstat->type == PROCSTAT_CORE) {
return (procstat_get_vnode_info_sysctl(fst, vn, errbuf));
} else {
warnx("unknown access method: %d", procstat->type);
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
}
static int
procstat_get_vnode_info_kvm(kvm_t *kd, struct filestat *fst,
struct vnstat *vn, char *errbuf)
{
/* Filesystem specific handlers. */
#define FSTYPE(fst) {#fst, fst##_filestat}
struct {
const char *tag;
int (*handler)(kvm_t *kd, struct vnode *vp,
struct vnstat *vn);
} fstypes[] = {
FSTYPE(devfs),
FSTYPE(isofs),
FSTYPE(msdosfs),
FSTYPE(nfs),
FSTYPE(smbfs),
FSTYPE(udf),
FSTYPE(ufs),
#ifdef LIBPROCSTAT_ZFS
FSTYPE(zfs),
#endif
};
#define NTYPES (sizeof(fstypes) / sizeof(*fstypes))
struct vnode vnode;
char tagstr[12];
void *vp;
int error;
unsigned int i;
assert(kd);
assert(vn);
assert(fst);
vp = fst->fs_typedep;
if (vp == NULL)
goto fail;
error = kvm_read_all(kd, (unsigned long)vp, &vnode, sizeof(vnode));
if (error == 0) {
warnx("can't read vnode at %p", (void *)vp);
goto fail;
}
bzero(vn, sizeof(*vn));
vn->vn_type = vntype2psfsttype(vnode.v_type);
if (vnode.v_type == VNON || vnode.v_type == VBAD)
return (0);
error = kvm_read_all(kd, (unsigned long)vnode.v_lock.lock_object.lo_name,
tagstr, sizeof(tagstr));
if (error == 0) {
warnx("can't read lo_name at %p", (void *)vp);
goto fail;
}
tagstr[sizeof(tagstr) - 1] = '\0';
/*
* Find appropriate handler.
*/
for (i = 0; i < NTYPES; i++)
if (!strcmp(fstypes[i].tag, tagstr)) {
if (fstypes[i].handler(kd, &vnode, vn) != 0) {
goto fail;
}
break;
}
if (i == NTYPES) {
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "?(%s)", tagstr);
return (1);
}
vn->vn_mntdir = getmnton(kd, vnode.v_mount);
if ((vnode.v_type == VBLK || vnode.v_type == VCHR) &&
vnode.v_rdev != NULL){
vn->vn_dev = dev2udev(kd, vnode.v_rdev);
(void)kdevtoname(kd, vnode.v_rdev, vn->vn_devname);
} else {
vn->vn_dev = -1;
}
return (0);
fail:
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
/*
* kinfo vnode type to filestat translation.
*/
static int
kinfo_vtype2fst(int kfvtype)
{
static struct {
int kf_vtype;
int fst_vtype;
} kfvtypes2fst[] = {
{ KF_VTYPE_VBAD, PS_FST_VTYPE_VBAD },
{ KF_VTYPE_VBLK, PS_FST_VTYPE_VBLK },
{ KF_VTYPE_VCHR, PS_FST_VTYPE_VCHR },
{ KF_VTYPE_VDIR, PS_FST_VTYPE_VDIR },
{ KF_VTYPE_VFIFO, PS_FST_VTYPE_VFIFO },
{ KF_VTYPE_VLNK, PS_FST_VTYPE_VLNK },
{ KF_VTYPE_VNON, PS_FST_VTYPE_VNON },
{ KF_VTYPE_VREG, PS_FST_VTYPE_VREG },
{ KF_VTYPE_VSOCK, PS_FST_VTYPE_VSOCK }
};
#define NKFVTYPES (sizeof(kfvtypes2fst) / sizeof(*kfvtypes2fst))
unsigned int i;
for (i = 0; i < NKFVTYPES; i++)
if (kfvtypes2fst[i].kf_vtype == kfvtype)
break;
if (i == NKFVTYPES)
return (PS_FST_VTYPE_UNKNOWN);
return (kfvtypes2fst[i].fst_vtype);
}
static int
procstat_get_vnode_info_sysctl(struct filestat *fst, struct vnstat *vn,
char *errbuf)
{
struct statfs stbuf;
struct kinfo_file *kif;
struct kinfo_vmentry *kve;
char *name, *path;
uint64_t fileid;
uint64_t size;
uint64_t fsid;
uint64_t rdev;
uint16_t mode;
int vntype;
int status;
assert(fst);
assert(vn);
bzero(vn, sizeof(*vn));
if (fst->fs_typedep == NULL)
return (1);
if (fst->fs_uflags & PS_FST_UFLAG_MMAP) {
kve = fst->fs_typedep;
fileid = kve->kve_vn_fileid;
fsid = kve->kve_vn_fsid;
mode = kve->kve_vn_mode;
path = kve->kve_path;
rdev = kve->kve_vn_rdev;
size = kve->kve_vn_size;
vntype = kinfo_vtype2fst(kve->kve_vn_type);
status = kve->kve_status;
} else {
kif = fst->fs_typedep;
fileid = kif->kf_un.kf_file.kf_file_fileid;
fsid = kif->kf_un.kf_file.kf_file_fsid;
mode = kif->kf_un.kf_file.kf_file_mode;
path = kif->kf_path;
rdev = kif->kf_un.kf_file.kf_file_rdev;
size = kif->kf_un.kf_file.kf_file_size;
vntype = kinfo_vtype2fst(kif->kf_vnode_type);
status = kif->kf_status;
}
vn->vn_type = vntype;
if (vntype == PS_FST_VTYPE_VNON || vntype == PS_FST_VTYPE_VBAD)
return (0);
if ((status & KF_ATTR_VALID) == 0) {
if (errbuf != NULL) {
snprintf(errbuf, _POSIX2_LINE_MAX,
"? (no info available)");
}
return (1);
}
if (path && *path) {
statfs(path, &stbuf);
vn->vn_mntdir = strdup(stbuf.f_mntonname);
} else
vn->vn_mntdir = strdup("-");
vn->vn_dev = rdev;
if (vntype == PS_FST_VTYPE_VBLK) {
name = devname(rdev, S_IFBLK);
if (name != NULL)
strlcpy(vn->vn_devname, name,
sizeof(vn->vn_devname));
} else if (vntype == PS_FST_VTYPE_VCHR) {
name = devname(vn->vn_dev, S_IFCHR);
if (name != NULL)
strlcpy(vn->vn_devname, name,
sizeof(vn->vn_devname));
}
vn->vn_fsid = fsid;
vn->vn_fileid = fileid;
vn->vn_size = size;
vn->vn_mode = mode;
return (0);
}
int
procstat_get_socket_info(struct procstat *procstat, struct filestat *fst,
struct sockstat *sock, char *errbuf)
{
assert(sock);
if (procstat->type == PROCSTAT_KVM) {
return (procstat_get_socket_info_kvm(procstat->kd, fst, sock,
errbuf));
} else if (procstat->type == PROCSTAT_SYSCTL ||
procstat->type == PROCSTAT_CORE) {
return (procstat_get_socket_info_sysctl(fst, sock, errbuf));
} else {
warnx("unknown access method: %d", procstat->type);
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
}
static int
procstat_get_socket_info_kvm(kvm_t *kd, struct filestat *fst,
struct sockstat *sock, char *errbuf)
{
struct domain dom;
struct inpcb inpcb;
struct protosw proto;
struct socket s;
struct unpcb unpcb;
ssize_t len;
void *so;
assert(kd);
assert(sock);
assert(fst);
bzero(sock, sizeof(*sock));
so = fst->fs_typedep;
if (so == NULL)
goto fail;
sock->so_addr = (uintptr_t)so;
/* fill in socket */
if (!kvm_read_all(kd, (unsigned long)so, &s,
sizeof(struct socket))) {
warnx("can't read sock at %p", (void *)so);
goto fail;
}
/* fill in protosw entry */
if (!kvm_read_all(kd, (unsigned long)s.so_proto, &proto,
sizeof(struct protosw))) {
warnx("can't read protosw at %p", (void *)s.so_proto);
goto fail;
}
/* fill in domain */
if (!kvm_read_all(kd, (unsigned long)proto.pr_domain, &dom,
sizeof(struct domain))) {
warnx("can't read domain at %p",
(void *)proto.pr_domain);
goto fail;
}
if ((len = kvm_read(kd, (unsigned long)dom.dom_name, sock->dname,
sizeof(sock->dname) - 1)) < 0) {
warnx("can't read domain name at %p", (void *)dom.dom_name);
sock->dname[0] = '\0';
}
else
sock->dname[len] = '\0';
/*
* Fill in known data.
*/
sock->type = s.so_type;
sock->proto = proto.pr_protocol;
sock->dom_family = dom.dom_family;
sock->so_pcb = (uintptr_t)s.so_pcb;
/*
* Protocol specific data.
*/
switch(dom.dom_family) {
case AF_INET:
case AF_INET6:
if (proto.pr_protocol == IPPROTO_TCP) {
if (s.so_pcb) {
if (kvm_read(kd, (u_long)s.so_pcb,
(char *)&inpcb, sizeof(struct inpcb))
!= sizeof(struct inpcb)) {
warnx("can't read inpcb at %p",
(void *)s.so_pcb);
} else
sock->inp_ppcb =
(uintptr_t)inpcb.inp_ppcb;
sock->sendq = s.so_snd.sb_ccc;
sock->recvq = s.so_rcv.sb_ccc;
}
}
break;
case AF_UNIX:
if (s.so_pcb) {
if (kvm_read(kd, (u_long)s.so_pcb, (char *)&unpcb,
sizeof(struct unpcb)) != sizeof(struct unpcb)){
warnx("can't read unpcb at %p",
(void *)s.so_pcb);
} else if (unpcb.unp_conn) {
sock->so_rcv_sb_state = s.so_rcv.sb_state;
sock->so_snd_sb_state = s.so_snd.sb_state;
sock->unp_conn = (uintptr_t)unpcb.unp_conn;
sock->sendq = s.so_snd.sb_ccc;
sock->recvq = s.so_rcv.sb_ccc;
}
}
break;
default:
break;
}
return (0);
fail:
if (errbuf != NULL)
snprintf(errbuf, _POSIX2_LINE_MAX, "error");
return (1);
}
static int
procstat_get_socket_info_sysctl(struct filestat *fst, struct sockstat *sock,
char *errbuf __unused)
{
struct kinfo_file *kif;
assert(sock);
assert(fst);
bzero(sock, sizeof(*sock));
kif = fst->fs_typedep;
if (kif == NULL)
return (0);
/*
* Fill in known data.
*/
sock->type = kif->kf_sock_type;
sock->proto = kif->kf_sock_protocol;
sock->dom_family = kif->kf_sock_domain;
sock->so_pcb = kif->kf_un.kf_sock.kf_sock_pcb;
strlcpy(sock->dname, kif->kf_path, sizeof(sock->dname));
bcopy(&kif->kf_un.kf_sock.kf_sa_local, &sock->sa_local,
kif->kf_un.kf_sock.kf_sa_local.ss_len);
bcopy(&kif->kf_un.kf_sock.kf_sa_peer, &sock->sa_peer,
kif->kf_un.kf_sock.kf_sa_peer.ss_len);
/*
* Protocol specific data.
*/
switch(sock->dom_family) {
case AF_INET:
case AF_INET6:
if (sock->proto == IPPROTO_TCP) {
sock->inp_ppcb = kif->kf_un.kf_sock.kf_sock_inpcb;
sock->sendq = kif->kf_un.kf_sock.kf_sock_sendq;
sock->recvq = kif->kf_un.kf_sock.kf_sock_recvq;
}
break;
case AF_UNIX:
if (kif->kf_un.kf_sock.kf_sock_unpconn != 0) {
sock->so_rcv_sb_state =
kif->kf_un.kf_sock.kf_sock_rcv_sb_state;
sock->so_snd_sb_state =
kif->kf_un.kf_sock.kf_sock_snd_sb_state;
sock->unp_conn =
kif->kf_un.kf_sock.kf_sock_unpconn;
sock->sendq = kif->kf_un.kf_sock.kf_sock_sendq;
sock->recvq = kif->kf_un.kf_sock.kf_sock_recvq;
}
break;
default:
break;
}
return (0);
}
/*
* Descriptor flags to filestat translation.
*/
static int
to_filestat_flags(int flags)
{
static struct {
int flag;
int fst_flag;
} fstflags[] = {
{ FREAD, PS_FST_FFLAG_READ },
{ FWRITE, PS_FST_FFLAG_WRITE },
{ O_APPEND, PS_FST_FFLAG_APPEND },
{ O_ASYNC, PS_FST_FFLAG_ASYNC },
{ O_CREAT, PS_FST_FFLAG_CREAT },
{ O_DIRECT, PS_FST_FFLAG_DIRECT },
{ O_EXCL, PS_FST_FFLAG_EXCL },
{ O_EXEC, PS_FST_FFLAG_EXEC },
{ O_EXLOCK, PS_FST_FFLAG_EXLOCK },
{ O_NOFOLLOW, PS_FST_FFLAG_NOFOLLOW },
{ O_NONBLOCK, PS_FST_FFLAG_NONBLOCK },
{ O_SHLOCK, PS_FST_FFLAG_SHLOCK },
{ O_SYNC, PS_FST_FFLAG_SYNC },
{ O_TRUNC, PS_FST_FFLAG_TRUNC }
};
#define NFSTFLAGS (sizeof(fstflags) / sizeof(*fstflags))
int fst_flags;
unsigned int i;
fst_flags = 0;
for (i = 0; i < NFSTFLAGS; i++)
if (flags & fstflags[i].flag)
fst_flags |= fstflags[i].fst_flag;
return (fst_flags);
}
/*
* Vnode type to filestate translation.
*/
static int
vntype2psfsttype(int type)
{
static struct {
int vtype;
int fst_vtype;
} vt2fst[] = {
{ VBAD, PS_FST_VTYPE_VBAD },
{ VBLK, PS_FST_VTYPE_VBLK },
{ VCHR, PS_FST_VTYPE_VCHR },
{ VDIR, PS_FST_VTYPE_VDIR },
{ VFIFO, PS_FST_VTYPE_VFIFO },
{ VLNK, PS_FST_VTYPE_VLNK },
{ VNON, PS_FST_VTYPE_VNON },
{ VREG, PS_FST_VTYPE_VREG },
{ VSOCK, PS_FST_VTYPE_VSOCK }
};
#define NVFTYPES (sizeof(vt2fst) / sizeof(*vt2fst))
unsigned int i, fst_type;
fst_type = PS_FST_VTYPE_UNKNOWN;
for (i = 0; i < NVFTYPES; i++) {
if (type == vt2fst[i].vtype) {
fst_type = vt2fst[i].fst_vtype;
break;
}
}
return (fst_type);
}
static char *
getmnton(kvm_t *kd, struct mount *m)
{
struct mount mnt;
static struct mtab {
struct mtab *next;
struct mount *m;
char mntonname[MNAMELEN + 1];
} *mhead = NULL;
struct mtab *mt;
for (mt = mhead; mt != NULL; mt = mt->next)
if (m == mt->m)
return (mt->mntonname);
if (!kvm_read_all(kd, (unsigned long)m, &mnt, sizeof(struct mount))) {
warnx("can't read mount table at %p", (void *)m);
return (NULL);
}
if ((mt = malloc(sizeof (struct mtab))) == NULL)
err(1, NULL);
mt->m = m;
bcopy(&mnt.mnt_stat.f_mntonname[0], &mt->mntonname[0], MNAMELEN);
mt->mntonname[MNAMELEN] = '\0';
mt->next = mhead;
mhead = mt;
return (mt->mntonname);
}
/*
* Auxiliary structures and functions to get process environment or
* command line arguments.
*/
struct argvec {
char *buf;
size_t bufsize;
char **argv;
size_t argc;
};
static struct argvec *
argvec_alloc(size_t bufsize)
{
struct argvec *av;
av = malloc(sizeof(*av));
if (av == NULL)
return (NULL);
av->bufsize = bufsize;
av->buf = malloc(av->bufsize);
if (av->buf == NULL) {
free(av);
return (NULL);
}
av->argc = 32;
av->argv = malloc(sizeof(char *) * av->argc);
if (av->argv == NULL) {
free(av->buf);
free(av);
return (NULL);
}
return av;
}
static void
argvec_free(struct argvec * av)
{
free(av->argv);
free(av->buf);
free(av);
}
static char **
getargv(struct procstat *procstat, struct kinfo_proc *kp, size_t nchr, int env)
{
int error, name[4], argc, i;
struct argvec *av, **avp;
enum psc_type type;
size_t len;
char *p, **argv;
assert(procstat);
assert(kp);
if (procstat->type == PROCSTAT_KVM) {
warnx("can't use kvm access method");
return (NULL);
}
if (procstat->type != PROCSTAT_SYSCTL &&
procstat->type != PROCSTAT_CORE) {
warnx("unknown access method: %d", procstat->type);
return (NULL);
}
if (nchr == 0 || nchr > ARG_MAX)
nchr = ARG_MAX;
avp = (struct argvec **)(env ? &procstat->argv : &procstat->envv);
av = *avp;
if (av == NULL)
{
av = argvec_alloc(nchr);
if (av == NULL)
{
warn("malloc(%zu)", nchr);
return (NULL);
}
*avp = av;
} else if (av->bufsize < nchr) {
av->buf = reallocf(av->buf, nchr);
if (av->buf == NULL) {
warn("malloc(%zu)", nchr);
return (NULL);
}
}
if (procstat->type == PROCSTAT_SYSCTL) {
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = env ? KERN_PROC_ENV : KERN_PROC_ARGS;
name[3] = kp->ki_pid;
len = nchr;
error = sysctl(name, nitems(name), av->buf, &len, NULL, 0);
if (error != 0 && errno != ESRCH && errno != EPERM)
warn("sysctl(kern.proc.%s)", env ? "env" : "args");
if (error != 0 || len == 0)
return (NULL);
} else /* procstat->type == PROCSTAT_CORE */ {
type = env ? PSC_TYPE_ENVV : PSC_TYPE_ARGV;
len = nchr;
if (procstat_core_get(procstat->core, type, av->buf, &len)
== NULL) {
return (NULL);
}
}
argv = av->argv;
argc = av->argc;
i = 0;
for (p = av->buf; p < av->buf + len; p += strlen(p) + 1) {
argv[i++] = p;
if (i < argc)
continue;
/* Grow argv. */
argc += argc;
argv = realloc(argv, sizeof(char *) * argc);
if (argv == NULL) {
warn("malloc(%zu)", sizeof(char *) * argc);
return (NULL);
}
av->argv = argv;
av->argc = argc;
}
argv[i] = NULL;
return (argv);
}
/*
* Return process command line arguments.
*/
char **
procstat_getargv(struct procstat *procstat, struct kinfo_proc *p, size_t nchr)
{
return (getargv(procstat, p, nchr, 0));
}
/*
* Free the buffer allocated by procstat_getargv().
*/
void
procstat_freeargv(struct procstat *procstat)
{
if (procstat->argv != NULL) {
argvec_free(procstat->argv);
procstat->argv = NULL;
}
}
/*
* Return process environment.
*/
char **
procstat_getenvv(struct procstat *procstat, struct kinfo_proc *p, size_t nchr)
{
return (getargv(procstat, p, nchr, 1));
}
/*
* Free the buffer allocated by procstat_getenvv().
*/
void
procstat_freeenvv(struct procstat *procstat)
{
if (procstat->envv != NULL) {
argvec_free(procstat->envv);
procstat->envv = NULL;
}
}
static struct kinfo_vmentry *
kinfo_getvmmap_core(struct procstat_core *core, int *cntp)
{
int cnt;
size_t len;
char *buf, *bp, *eb;
struct kinfo_vmentry *kiv, *kp, *kv;
buf = procstat_core_get(core, PSC_TYPE_VMMAP, NULL, &len);
if (buf == NULL)
return (NULL);
/*
* XXXMG: The code below is just copy&past from libutil.
* The code duplication can be avoided if libutil
* is extended to provide something like:
* struct kinfo_vmentry *kinfo_getvmmap_from_buf(const char *buf,
* size_t len, int *cntp);
*/
/* Pass 1: count items */
cnt = 0;
bp = buf;
eb = buf + len;
while (bp < eb) {
kv = (struct kinfo_vmentry *)(uintptr_t)bp;
if (kv->kve_structsize == 0)
break;
bp += kv->kve_structsize;
cnt++;
}
kiv = calloc(cnt, sizeof(*kiv));
if (kiv == NULL) {
free(buf);
return (NULL);
}
bp = buf;
eb = buf + len;
kp = kiv;
/* Pass 2: unpack */
while (bp < eb) {
kv = (struct kinfo_vmentry *)(uintptr_t)bp;
if (kv->kve_structsize == 0)
break;
/* Copy/expand into pre-zeroed buffer */
memcpy(kp, kv, kv->kve_structsize);
/* Advance to next packed record */
bp += kv->kve_structsize;
/* Set field size to fixed length, advance */
kp->kve_structsize = sizeof(*kp);
kp++;
}
free(buf);
*cntp = cnt;
return (kiv); /* Caller must free() return value */
}
struct kinfo_vmentry *
procstat_getvmmap(struct procstat *procstat, struct kinfo_proc *kp,
unsigned int *cntp)
{
switch(procstat->type) {
case PROCSTAT_KVM:
warnx("kvm method is not supported");
return (NULL);
case PROCSTAT_SYSCTL:
return (kinfo_getvmmap(kp->ki_pid, cntp));
case PROCSTAT_CORE:
return (kinfo_getvmmap_core(procstat->core, cntp));
default:
warnx("unknown access method: %d", procstat->type);
return (NULL);
}
}
void
procstat_freevmmap(struct procstat *procstat __unused,
struct kinfo_vmentry *vmmap)
{
free(vmmap);
}
static gid_t *
procstat_getgroups_kvm(kvm_t *kd, struct kinfo_proc *kp, unsigned int *cntp)
{
struct proc proc;
struct ucred ucred;
gid_t *groups;
size_t len;
assert(kd != NULL);
assert(kp != NULL);
if (!kvm_read_all(kd, (unsigned long)kp->ki_paddr, &proc,
sizeof(proc))) {
warnx("can't read proc struct at %p for pid %d",
kp->ki_paddr, kp->ki_pid);
return (NULL);
}
if (proc.p_ucred == NOCRED)
return (NULL);
if (!kvm_read_all(kd, (unsigned long)proc.p_ucred, &ucred,
sizeof(ucred))) {
warnx("can't read ucred struct at %p for pid %d",
proc.p_ucred, kp->ki_pid);
return (NULL);
}
len = ucred.cr_ngroups * sizeof(gid_t);
groups = malloc(len);
if (groups == NULL) {
warn("malloc(%zu)", len);
return (NULL);
}
if (!kvm_read_all(kd, (unsigned long)ucred.cr_groups, groups, len)) {
warnx("can't read groups at %p for pid %d",
ucred.cr_groups, kp->ki_pid);
free(groups);
return (NULL);
}
*cntp = ucred.cr_ngroups;
return (groups);
}
static gid_t *
procstat_getgroups_sysctl(pid_t pid, unsigned int *cntp)
{
int mib[4];
size_t len;
gid_t *groups;
mib[0] = CTL_KERN;
mib[1] = KERN_PROC;
mib[2] = KERN_PROC_GROUPS;
mib[3] = pid;
len = (sysconf(_SC_NGROUPS_MAX) + 1) * sizeof(gid_t);
groups = malloc(len);
if (groups == NULL) {
warn("malloc(%zu)", len);
return (NULL);
}
if (sysctl(mib, nitems(mib), groups, &len, NULL, 0) == -1) {
warn("sysctl: kern.proc.groups: %d", pid);
free(groups);
return (NULL);
}
*cntp = len / sizeof(gid_t);
return (groups);
}
static gid_t *
procstat_getgroups_core(struct procstat_core *core, unsigned int *cntp)
{
size_t len;
gid_t *groups;
groups = procstat_core_get(core, PSC_TYPE_GROUPS, NULL, &len);
if (groups == NULL)
return (NULL);
*cntp = len / sizeof(gid_t);
return (groups);
}
gid_t *
procstat_getgroups(struct procstat *procstat, struct kinfo_proc *kp,
unsigned int *cntp)
{
switch(procstat->type) {
case PROCSTAT_KVM:
return (procstat_getgroups_kvm(procstat->kd, kp, cntp));
case PROCSTAT_SYSCTL:
return (procstat_getgroups_sysctl(kp->ki_pid, cntp));
case PROCSTAT_CORE:
return (procstat_getgroups_core(procstat->core, cntp));
default:
warnx("unknown access method: %d", procstat->type);
return (NULL);
}
}
void
procstat_freegroups(struct procstat *procstat __unused, gid_t *groups)
{
free(groups);
}
static int
procstat_getumask_kvm(kvm_t *kd, struct kinfo_proc *kp, unsigned short *maskp)
{
struct pwddesc pd;
assert(kd != NULL);
assert(kp != NULL);
if (kp->ki_pd == NULL)
return (-1);
if (!kvm_read_all(kd, (unsigned long)kp->ki_pd, &pd, sizeof(pd))) {
warnx("can't read pwddesc at %p for pid %d", kp->ki_pd,
kp->ki_pid);
return (-1);
}
*maskp = pd.pd_cmask;
return (0);
}
static int
procstat_getumask_sysctl(pid_t pid, unsigned short *maskp)
{
int error;
int mib[4];
size_t len;
mib[0] = CTL_KERN;
mib[1] = KERN_PROC;
mib[2] = KERN_PROC_UMASK;
mib[3] = pid;
len = sizeof(*maskp);
error = sysctl(mib, nitems(mib), maskp, &len, NULL, 0);
if (error != 0 && errno != ESRCH && errno != EPERM)
warn("sysctl: kern.proc.umask: %d", pid);
return (error);
}
static int
procstat_getumask_core(struct procstat_core *core, unsigned short *maskp)
{
size_t len;
unsigned short *buf;
buf = procstat_core_get(core, PSC_TYPE_UMASK, NULL, &len);
if (buf == NULL)
return (-1);
if (len < sizeof(*maskp)) {
free(buf);
return (-1);
}
*maskp = *buf;
free(buf);
return (0);
}
int
procstat_getumask(struct procstat *procstat, struct kinfo_proc *kp,
unsigned short *maskp)
{
switch(procstat->type) {
case PROCSTAT_KVM:
return (procstat_getumask_kvm(procstat->kd, kp, maskp));
case PROCSTAT_SYSCTL:
return (procstat_getumask_sysctl(kp->ki_pid, maskp));
case PROCSTAT_CORE:
return (procstat_getumask_core(procstat->core, maskp));
default:
warnx("unknown access method: %d", procstat->type);
return (-1);
}
}
static int
procstat_getrlimit_kvm(kvm_t *kd, struct kinfo_proc *kp, int which,
struct rlimit* rlimit)
{
struct proc proc;
unsigned long offset;
assert(kd != NULL);
assert(kp != NULL);
assert(which >= 0 && which < RLIM_NLIMITS);
if (!kvm_read_all(kd, (unsigned long)kp->ki_paddr, &proc,
sizeof(proc))) {
warnx("can't read proc struct at %p for pid %d",
kp->ki_paddr, kp->ki_pid);
return (-1);
}
if (proc.p_limit == NULL)
return (-1);
offset = (unsigned long)proc.p_limit + sizeof(struct rlimit) * which;
if (!kvm_read_all(kd, offset, rlimit, sizeof(*rlimit))) {
warnx("can't read rlimit struct at %p for pid %d",
(void *)offset, kp->ki_pid);
return (-1);
}
return (0);
}
static int
procstat_getrlimit_sysctl(pid_t pid, int which, struct rlimit* rlimit)
{
int error, name[5];
size_t len;
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = KERN_PROC_RLIMIT;
name[3] = pid;
name[4] = which;
len = sizeof(struct rlimit);
error = sysctl(name, nitems(name), rlimit, &len, NULL, 0);
if (error < 0 && errno != ESRCH) {
warn("sysctl: kern.proc.rlimit: %d", pid);
return (-1);
}
if (error < 0 || len != sizeof(struct rlimit))
return (-1);
return (0);
}
static int
procstat_getrlimit_core(struct procstat_core *core, int which,
struct rlimit* rlimit)
{
size_t len;
struct rlimit* rlimits;
if (which < 0 || which >= RLIM_NLIMITS) {
errno = EINVAL;
warn("getrlimit: which");
return (-1);
}
rlimits = procstat_core_get(core, PSC_TYPE_RLIMIT, NULL, &len);
if (rlimits == NULL)
return (-1);
if (len < sizeof(struct rlimit) * RLIM_NLIMITS) {
free(rlimits);
return (-1);
}
*rlimit = rlimits[which];
free(rlimits);
return (0);
}
int
procstat_getrlimit(struct procstat *procstat, struct kinfo_proc *kp, int which,
struct rlimit* rlimit)
{
switch(procstat->type) {
case PROCSTAT_KVM:
return (procstat_getrlimit_kvm(procstat->kd, kp, which,
rlimit));
case PROCSTAT_SYSCTL:
return (procstat_getrlimit_sysctl(kp->ki_pid, which, rlimit));
case PROCSTAT_CORE:
return (procstat_getrlimit_core(procstat->core, which, rlimit));
default:
warnx("unknown access method: %d", procstat->type);
return (-1);
}
}
static int
procstat_getpathname_sysctl(pid_t pid, char *pathname, size_t maxlen)
{
int error, name[4];
size_t len;
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = KERN_PROC_PATHNAME;
name[3] = pid;
len = maxlen;
error = sysctl(name, nitems(name), pathname, &len, NULL, 0);
if (error != 0 && errno != ESRCH)
warn("sysctl: kern.proc.pathname: %d", pid);
if (len == 0)
pathname[0] = '\0';
return (error);
}
static int
procstat_getpathname_core(struct procstat_core *core, char *pathname,
size_t maxlen)
{
struct kinfo_file *files;
int cnt, i, result;
files = kinfo_getfile_core(core, &cnt);
if (files == NULL)
return (-1);
result = -1;
for (i = 0; i < cnt; i++) {
if (files[i].kf_fd != KF_FD_TYPE_TEXT)
continue;
strncpy(pathname, files[i].kf_path, maxlen);
result = 0;
break;
}
free(files);
return (result);
}
int
procstat_getpathname(struct procstat *procstat, struct kinfo_proc *kp,
char *pathname, size_t maxlen)
{
switch(procstat->type) {
case PROCSTAT_KVM:
/* XXX: Return empty string. */
if (maxlen > 0)
pathname[0] = '\0';
return (0);
case PROCSTAT_SYSCTL:
return (procstat_getpathname_sysctl(kp->ki_pid, pathname,
maxlen));
case PROCSTAT_CORE:
return (procstat_getpathname_core(procstat->core, pathname,
maxlen));
default:
warnx("unknown access method: %d", procstat->type);
return (-1);
}
}
static int
procstat_getosrel_kvm(kvm_t *kd, struct kinfo_proc *kp, int *osrelp)
{
struct proc proc;
assert(kd != NULL);
assert(kp != NULL);
if (!kvm_read_all(kd, (unsigned long)kp->ki_paddr, &proc,
sizeof(proc))) {
warnx("can't read proc struct at %p for pid %d",
kp->ki_paddr, kp->ki_pid);
return (-1);
}
*osrelp = proc.p_osrel;
return (0);
}
static int
procstat_getosrel_sysctl(pid_t pid, int *osrelp)
{
int error, name[4];
size_t len;
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = KERN_PROC_OSREL;
name[3] = pid;
len = sizeof(*osrelp);
error = sysctl(name, nitems(name), osrelp, &len, NULL, 0);
if (error != 0 && errno != ESRCH)
warn("sysctl: kern.proc.osrel: %d", pid);
return (error);
}
static int
procstat_getosrel_core(struct procstat_core *core, int *osrelp)
{
size_t len;
int *buf;
buf = procstat_core_get(core, PSC_TYPE_OSREL, NULL, &len);
if (buf == NULL)
return (-1);
if (len < sizeof(*osrelp)) {
free(buf);
return (-1);
}
*osrelp = *buf;
free(buf);
return (0);
}
int
procstat_getosrel(struct procstat *procstat, struct kinfo_proc *kp, int *osrelp)
{
switch(procstat->type) {
case PROCSTAT_KVM:
return (procstat_getosrel_kvm(procstat->kd, kp, osrelp));
case PROCSTAT_SYSCTL:
return (procstat_getosrel_sysctl(kp->ki_pid, osrelp));
case PROCSTAT_CORE:
return (procstat_getosrel_core(procstat->core, osrelp));
default:
warnx("unknown access method: %d", procstat->type);
return (-1);
}
}
#define PROC_AUXV_MAX 256
#if __ELF_WORD_SIZE == 64
static const char *elf32_sv_names[] = {
"Linux ELF32",
"FreeBSD ELF32",
};
static int
is_elf32_sysctl(pid_t pid)
{
int error, name[4];
size_t len, i;
static char sv_name[256];
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = KERN_PROC_SV_NAME;
name[3] = pid;
len = sizeof(sv_name);
error = sysctl(name, nitems(name), sv_name, &len, NULL, 0);
if (error != 0 || len == 0)
return (0);
for (i = 0; i < sizeof(elf32_sv_names) / sizeof(*elf32_sv_names); i++) {
if (strncmp(sv_name, elf32_sv_names[i], sizeof(sv_name)) == 0)
return (1);
}
return (0);
}
static Elf_Auxinfo *
procstat_getauxv32_sysctl(pid_t pid, unsigned int *cntp)
{
Elf_Auxinfo *auxv;
Elf32_Auxinfo *auxv32;
void *ptr;
size_t len;
unsigned int i, count;
int name[4];
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = KERN_PROC_AUXV;
name[3] = pid;
len = PROC_AUXV_MAX * sizeof(Elf32_Auxinfo);
auxv = NULL;
auxv32 = malloc(len);
if (auxv32 == NULL) {
warn("malloc(%zu)", len);
goto out;
}
if (sysctl(name, nitems(name), auxv32, &len, NULL, 0) == -1) {
if (errno != ESRCH && errno != EPERM)
warn("sysctl: kern.proc.auxv: %d: %d", pid, errno);
goto out;
}
count = len / sizeof(Elf_Auxinfo);
auxv = malloc(count * sizeof(Elf_Auxinfo));
if (auxv == NULL) {
warn("malloc(%zu)", count * sizeof(Elf_Auxinfo));
goto out;
}
for (i = 0; i < count; i++) {
/*
* XXX: We expect that values for a_type on a 32-bit platform
* are directly mapped to values on 64-bit one, which is not
* necessarily true.
*/
auxv[i].a_type = auxv32[i].a_type;
ptr = &auxv32[i].a_un;
auxv[i].a_un.a_val = *((uint32_t *)ptr);
}
*cntp = count;
out:
free(auxv32);
return (auxv);
}
#endif /* __ELF_WORD_SIZE == 64 */
static Elf_Auxinfo *
procstat_getauxv_sysctl(pid_t pid, unsigned int *cntp)
{
Elf_Auxinfo *auxv;
int name[4];
size_t len;
#if __ELF_WORD_SIZE == 64
if (is_elf32_sysctl(pid))
return (procstat_getauxv32_sysctl(pid, cntp));
#endif
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = KERN_PROC_AUXV;
name[3] = pid;
len = PROC_AUXV_MAX * sizeof(Elf_Auxinfo);
auxv = malloc(len);
if (auxv == NULL) {
warn("malloc(%zu)", len);
return (NULL);
}
if (sysctl(name, nitems(name), auxv, &len, NULL, 0) == -1) {
if (errno != ESRCH && errno != EPERM)
warn("sysctl: kern.proc.auxv: %d: %d", pid, errno);
free(auxv);
return (NULL);
}
*cntp = len / sizeof(Elf_Auxinfo);
return (auxv);
}
static Elf_Auxinfo *
procstat_getauxv_core(struct procstat_core *core, unsigned int *cntp)
{
Elf_Auxinfo *auxv;
size_t len;
auxv = procstat_core_get(core, PSC_TYPE_AUXV, NULL, &len);
if (auxv == NULL)
return (NULL);
*cntp = len / sizeof(Elf_Auxinfo);
return (auxv);
}
Elf_Auxinfo *
procstat_getauxv(struct procstat *procstat, struct kinfo_proc *kp,
unsigned int *cntp)
{
switch(procstat->type) {
case PROCSTAT_KVM:
warnx("kvm method is not supported");
return (NULL);
case PROCSTAT_SYSCTL:
return (procstat_getauxv_sysctl(kp->ki_pid, cntp));
case PROCSTAT_CORE:
return (procstat_getauxv_core(procstat->core, cntp));
default:
warnx("unknown access method: %d", procstat->type);
return (NULL);
}
}
void
procstat_freeauxv(struct procstat *procstat __unused, Elf_Auxinfo *auxv)
{
free(auxv);
}
static struct ptrace_lwpinfo *
procstat_getptlwpinfo_core(struct procstat_core *core, unsigned int *cntp)
{
void *buf;
struct ptrace_lwpinfo *pl;
unsigned int cnt;
size_t len;
cnt = procstat_core_note_count(core, PSC_TYPE_PTLWPINFO);
if (cnt == 0)
return (NULL);
len = cnt * sizeof(*pl);
buf = calloc(1, len);
pl = procstat_core_get(core, PSC_TYPE_PTLWPINFO, buf, &len);
if (pl == NULL) {
free(buf);
return (NULL);
}
*cntp = len / sizeof(*pl);
return (pl);
}
struct ptrace_lwpinfo *
procstat_getptlwpinfo(struct procstat *procstat, unsigned int *cntp)
{
switch (procstat->type) {
case PROCSTAT_KVM:
warnx("kvm method is not supported");
return (NULL);
case PROCSTAT_SYSCTL:
warnx("sysctl method is not supported");
return (NULL);
case PROCSTAT_CORE:
return (procstat_getptlwpinfo_core(procstat->core, cntp));
default:
warnx("unknown access method: %d", procstat->type);
return (NULL);
}
}
void
procstat_freeptlwpinfo(struct procstat *procstat __unused,
struct ptrace_lwpinfo *pl)
{
free(pl);
}
static struct kinfo_kstack *
procstat_getkstack_sysctl(pid_t pid, int *cntp)
{
struct kinfo_kstack *kkstp;
int error, name[4];
size_t len;
name[0] = CTL_KERN;
name[1] = KERN_PROC;
name[2] = KERN_PROC_KSTACK;
name[3] = pid;
len = 0;
error = sysctl(name, nitems(name), NULL, &len, NULL, 0);
if (error < 0 && errno != ESRCH && errno != EPERM && errno != ENOENT) {
warn("sysctl: kern.proc.kstack: %d", pid);
return (NULL);
}
if (error == -1 && errno == ENOENT) {
warnx("sysctl: kern.proc.kstack unavailable"
" (options DDB or options STACK required in kernel)");
return (NULL);
}
if (error == -1)
return (NULL);
kkstp = malloc(len);
if (kkstp == NULL) {
warn("malloc(%zu)", len);
return (NULL);
}
if (sysctl(name, nitems(name), kkstp, &len, NULL, 0) == -1) {
warn("sysctl: kern.proc.pid: %d", pid);
free(kkstp);
return (NULL);
}
*cntp = len / sizeof(*kkstp);
return (kkstp);
}
struct kinfo_kstack *
procstat_getkstack(struct procstat *procstat, struct kinfo_proc *kp,
unsigned int *cntp)
{
switch(procstat->type) {
case PROCSTAT_KVM:
warnx("kvm method is not supported");
return (NULL);
case PROCSTAT_SYSCTL:
return (procstat_getkstack_sysctl(kp->ki_pid, cntp));
case PROCSTAT_CORE:
warnx("core method is not supported");
return (NULL);
default:
warnx("unknown access method: %d", procstat->type);
return (NULL);
}
}
void
procstat_freekstack(struct procstat *procstat __unused,
struct kinfo_kstack *kkstp)
{
free(kkstp);
}
diff --git a/lib/libprocstat/zfs_defs.c b/lib/libprocstat/zfs_defs.c
index c41054f05136..66acef743dc4 100644
--- a/lib/libprocstat/zfs_defs.c
+++ b/lib/libprocstat/zfs_defs.c
@@ -1,85 +1,85 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2020 Andriy Gapon <avg@FreeBSD.org>
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
#include <sys/types.h>
__FBSDID("$FreeBSD$");
/*
* Prevent some headers from getting included and fake some types
* in order to allow this file to compile without bringing in
* too many kernel build dependencies.
*/
#define _OPENSOLARIS_SYS_PATHNAME_H_
#define _OPENSOLARIS_SYS_POLICY_H_
#define _VNODE_PAGER_
enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD,
VMARKER };
/*
* Vnode attributes. A field value of VNOVAL represents a field whose value
* is unavailable (getattr) or which is not to be changed (setattr).
*/
struct vattr {
enum vtype va_type; /* vnode type (for create) */
u_short va_mode; /* files access mode and type */
u_short va_padding0;
uid_t va_uid; /* owner user id */
gid_t va_gid; /* owner group id */
nlink_t va_nlink; /* number of references to file */
dev_t va_fsid; /* filesystem id */
ino_t va_fileid; /* file id */
u_quad_t va_size; /* file size in bytes */
long va_blocksize; /* blocksize preferred for i/o */
struct timespec va_atime; /* time of last access */
struct timespec va_mtime; /* time of last modification */
struct timespec va_ctime; /* time file changed */
struct timespec va_birthtime; /* time file created */
u_long va_gen; /* generation number of file */
u_long va_flags; /* flags defined for file */
dev_t va_rdev; /* device the special file represents */
u_quad_t va_bytes; /* bytes of disk space held by file */
u_quad_t va_filerev; /* file modification number */
u_int va_vaflags; /* operations flags, see below */
long va_spare; /* remain quad aligned */
};
-
+#define _WANT_MOUNT
#include <sys/zfs_context.h>
#include <sys/zfs_znode.h>
size_t sizeof_znode_t = sizeof(znode_t);
size_t offsetof_z_id = offsetof(znode_t, z_id);
size_t offsetof_z_size = offsetof(znode_t, z_size);
size_t offsetof_z_mode = offsetof(znode_t, z_mode);
/* Keep pcpu.h satisfied. */
uintptr_t *__start_set_pcpu;
uintptr_t *__stop_set_pcpu;
diff --git a/lib/libufs/block.c b/lib/libufs/block.c
index 7accc8863693..3c99d288a402 100644
--- a/lib/libufs/block.c
+++ b/lib/libufs/block.c
@@ -1,203 +1,205 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2002 Juli Mallett. All rights reserved.
*
* This software was written by Juli Mallett <jmallett@FreeBSD.org> for the
* FreeBSD project. Redistribution and use in source and binary forms, with
* or without modification, are permitted provided that the following
* conditions are met:
*
* 1. Redistribution of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* 2. Redistribution in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
* INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
* IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/mount.h>
#include <sys/disk.h>
#include <sys/disklabel.h>
#include <sys/stat.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <libufs.h>
ssize_t
bread(struct uufsd *disk, ufs2_daddr_t blockno, void *data, size_t size)
{
void *p2;
ssize_t cnt;
ERROR(disk, NULL);
p2 = data;
/*
* XXX: various disk controllers require alignment of our buffer
* XXX: which is stricter than struct alignment.
* XXX: Bounce the buffer if not 64 byte aligned.
* XXX: this can be removed if/when the kernel is fixed
*/
if (((intptr_t)data) & 0x3f) {
p2 = malloc(size);
if (p2 == NULL) {
ERROR(disk, "allocate bounce buffer");
goto fail;
}
}
cnt = pread(disk->d_fd, p2, size, (off_t)(blockno * disk->d_bsize));
if (cnt == -1) {
ERROR(disk, "read error from block device");
goto fail;
}
if (cnt == 0) {
ERROR(disk, "end of file from block device");
goto fail;
}
if ((size_t)cnt != size) {
ERROR(disk, "short read or read error from block device");
goto fail;
}
if (p2 != data) {
memcpy(data, p2, size);
free(p2);
}
return (cnt);
fail: memset(data, 0, size);
if (p2 != data) {
free(p2);
}
return (-1);
}
ssize_t
bwrite(struct uufsd *disk, ufs2_daddr_t blockno, const void *data, size_t size)
{
ssize_t cnt;
int rv;
void *p2 = NULL;
ERROR(disk, NULL);
rv = ufs_disk_write(disk);
if (rv == -1) {
ERROR(disk, "failed to open disk for writing");
return (-1);
}
/*
* XXX: various disk controllers require alignment of our buffer
* XXX: which is stricter than struct alignment.
* XXX: Bounce the buffer if not 64 byte aligned.
* XXX: this can be removed if/when the kernel is fixed
*/
if (((intptr_t)data) & 0x3f) {
p2 = malloc(size);
if (p2 == NULL) {
ERROR(disk, "allocate bounce buffer");
return (-1);
}
memcpy(p2, data, size);
data = p2;
}
cnt = pwrite(disk->d_fd, data, size, (off_t)(blockno * disk->d_bsize));
if (p2 != NULL)
free(p2);
if (cnt == -1) {
ERROR(disk, "write error to block device");
return (-1);
}
if ((size_t)cnt != size) {
ERROR(disk, "short write to block device");
return (-1);
}
return (cnt);
}
#ifdef __FreeBSD_kernel__
static int
berase_helper(struct uufsd *disk, ufs2_daddr_t blockno, ufs2_daddr_t size)
{
off_t ioarg[2];
ioarg[0] = blockno * disk->d_bsize;
ioarg[1] = size;
return (ioctl(disk->d_fd, DIOCGDELETE, ioarg));
}
#else
static int
berase_helper(struct uufsd *disk, ufs2_daddr_t blockno, ufs2_daddr_t size)
{
char *zero_chunk;
off_t offset, zero_chunk_size, pwrite_size;
int rv;
offset = blockno * disk->d_bsize;
zero_chunk_size = 65536 * disk->d_bsize;
zero_chunk = calloc(1, zero_chunk_size);
if (zero_chunk == NULL) {
ERROR(disk, "failed to allocate memory");
return (-1);
}
while (size > 0) {
pwrite_size = size;
if (pwrite_size > zero_chunk_size)
pwrite_size = zero_chunk_size;
rv = pwrite(disk->d_fd, zero_chunk, pwrite_size, offset);
if (rv == -1) {
ERROR(disk, "failed writing to disk");
break;
}
size -= rv;
offset += rv;
rv = 0;
}
free(zero_chunk);
return (rv);
}
#endif
int
berase(struct uufsd *disk, ufs2_daddr_t blockno, ufs2_daddr_t size)
{
int rv;
ERROR(disk, NULL);
rv = ufs_disk_write(disk);
if (rv == -1) {
ERROR(disk, "failed to open disk for writing");
return(rv);
}
return (berase_helper(disk, blockno, size));
}
diff --git a/lib/libufs/cgroup.c b/lib/libufs/cgroup.c
index 1c3e271c4295..90b28eadad2c 100644
--- a/lib/libufs/cgroup.c
+++ b/lib/libufs/cgroup.c
@@ -1,314 +1,316 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2003 Juli Mallett. All rights reserved.
*
* This software was written by Juli Mallett <jmallett@FreeBSD.org> for the
* FreeBSD project. Redistribution and use in source and binary forms, with
* or without modification, are permitted provided that the following
* conditions are met:
*
* 1. Redistribution of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* 2. Redistribution in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
* INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
* IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/mount.h>
#include <sys/disklabel.h>
#include <sys/stat.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <libufs.h>
ufs2_daddr_t
cgballoc(struct uufsd *disk)
{
u_int8_t *blksfree;
struct cg *cgp;
struct fs *fs;
long bno;
fs = &disk->d_fs;
cgp = &disk->d_cg;
blksfree = cg_blksfree(cgp);
for (bno = 0; bno < fs->fs_fpg / fs->fs_frag; bno++)
if (ffs_isblock(fs, blksfree, bno))
goto gotit;
return (0);
gotit:
fs->fs_cs(fs, cgp->cg_cgx).cs_nbfree--;
ffs_clrblock(fs, blksfree, (long)bno);
ffs_clusteracct(fs, cgp, bno, -1);
cgp->cg_cs.cs_nbfree--;
fs->fs_cstotal.cs_nbfree--;
fs->fs_fmod = 1;
return (cgbase(fs, cgp->cg_cgx) + blkstofrags(fs, bno));
}
int
cgbfree(struct uufsd *disk, ufs2_daddr_t bno, long size)
{
u_int8_t *blksfree;
struct fs *fs;
struct cg *cgp;
ufs1_daddr_t fragno, cgbno;
int i, cg, blk, frags, bbase;
fs = &disk->d_fs;
cg = dtog(fs, bno);
if (cgread1(disk, cg) != 1)
return (-1);
cgp = &disk->d_cg;
cgbno = dtogd(fs, bno);
blksfree = cg_blksfree(cgp);
if (size == fs->fs_bsize) {
fragno = fragstoblks(fs, cgbno);
ffs_setblock(fs, blksfree, fragno);
ffs_clusteracct(fs, cgp, fragno, 1);
cgp->cg_cs.cs_nbfree++;
fs->fs_cstotal.cs_nbfree++;
fs->fs_cs(fs, cg).cs_nbfree++;
} else {
bbase = cgbno - fragnum(fs, cgbno);
/*
* decrement the counts associated with the old frags
*/
blk = blkmap(fs, blksfree, bbase);
ffs_fragacct(fs, blk, cgp->cg_frsum, -1);
/*
* deallocate the fragment
*/
frags = numfrags(fs, size);
for (i = 0; i < frags; i++)
setbit(blksfree, cgbno + i);
cgp->cg_cs.cs_nffree += i;
fs->fs_cstotal.cs_nffree += i;
fs->fs_cs(fs, cg).cs_nffree += i;
/*
* add back in counts associated with the new frags
*/
blk = blkmap(fs, blksfree, bbase);
ffs_fragacct(fs, blk, cgp->cg_frsum, 1);
/*
* if a complete block has been reassembled, account for it
*/
fragno = fragstoblks(fs, bbase);
if (ffs_isblock(fs, blksfree, fragno)) {
cgp->cg_cs.cs_nffree -= fs->fs_frag;
fs->fs_cstotal.cs_nffree -= fs->fs_frag;
fs->fs_cs(fs, cg).cs_nffree -= fs->fs_frag;
ffs_clusteracct(fs, cgp, fragno, 1);
cgp->cg_cs.cs_nbfree++;
fs->fs_cstotal.cs_nbfree++;
fs->fs_cs(fs, cg).cs_nbfree++;
}
}
return cgwrite(disk);
}
ino_t
cgialloc(struct uufsd *disk)
{
struct ufs2_dinode *dp2;
u_int8_t *inosused;
struct cg *cgp;
struct fs *fs;
ino_t ino;
int i;
fs = &disk->d_fs;
cgp = &disk->d_cg;
inosused = cg_inosused(cgp);
for (ino = 0; ino < fs->fs_ipg; ino++)
if (isclr(inosused, ino))
goto gotit;
return (0);
gotit:
if (fs->fs_magic == FS_UFS2_MAGIC &&
ino + INOPB(fs) > cgp->cg_initediblk &&
cgp->cg_initediblk < cgp->cg_niblk) {
char block[MAXBSIZE];
bzero(block, (int)fs->fs_bsize);
dp2 = (struct ufs2_dinode *)█
for (i = 0; i < INOPB(fs); i++) {
dp2->di_gen = arc4random();
dp2++;
}
if (bwrite(disk, ino_to_fsba(fs,
cgp->cg_cgx * fs->fs_ipg + cgp->cg_initediblk),
block, fs->fs_bsize))
return (0);
cgp->cg_initediblk += INOPB(fs);
}
setbit(inosused, ino);
cgp->cg_irotor = ino;
cgp->cg_cs.cs_nifree--;
fs->fs_cstotal.cs_nifree--;
fs->fs_cs(fs, cgp->cg_cgx).cs_nifree--;
fs->fs_fmod = 1;
return (ino + (cgp->cg_cgx * fs->fs_ipg));
}
int
cgread(struct uufsd *disk)
{
if (disk->d_ccg >= disk->d_fs.fs_ncg)
return (0);
return (cgread1(disk, disk->d_ccg++));
}
/* Short read/write error messages from cgget()/cgput() */
static const char *failmsg;
int
cgread1(struct uufsd *disk, int c)
{
if (cgget(disk->d_fd, &disk->d_fs, c, &disk->d_cg) == 0) {
disk->d_lcg = c;
return (1);
}
ERROR(disk, NULL);
if (failmsg != NULL) {
ERROR(disk, failmsg);
return (-1);
}
switch (errno) {
case EINTEGRITY:
ERROR(disk, "cylinder group checks failed");
break;
case EIO:
ERROR(disk, "read error from block device");
break;
default:
ERROR(disk, strerror(errno));
break;
}
return (-1);
}
int
cgget(int devfd, struct fs *fs, int cg, struct cg *cgp)
{
uint32_t cghash, calchash;
size_t cnt;
failmsg = NULL;
if ((cnt = pread(devfd, cgp, fs->fs_cgsize,
fsbtodb(fs, cgtod(fs, cg)) * (fs->fs_fsize / fsbtodb(fs,1)))) < 0)
return (-1);
if (cnt == 0) {
failmsg = "end of file from block device";
errno = EIO;
return (-1);
}
if (cnt != fs->fs_cgsize) {
failmsg = "short read from block device";
errno = EIO;
return (-1);
}
calchash = cgp->cg_ckhash;
if ((fs->fs_metackhash & CK_CYLGRP) != 0) {
cghash = cgp->cg_ckhash;
cgp->cg_ckhash = 0;
calchash = calculate_crc32c(~0L, (void *)cgp, fs->fs_cgsize);
cgp->cg_ckhash = cghash;
}
if (cgp->cg_ckhash != calchash || !cg_chkmagic(cgp) ||
cgp->cg_cgx != cg) {
errno = EINTEGRITY;
return (-1);
}
return (0);
}
int
cgwrite(struct uufsd *disk)
{
return (cgwrite1(disk, disk->d_cg.cg_cgx));
}
int
cgwrite1(struct uufsd *disk, int cg)
{
static char errmsg[BUFSIZ];
if (cg == disk->d_cg.cg_cgx) {
if (cgput(disk->d_fd, &disk->d_fs, &disk->d_cg) == 0)
return (0);
ERROR(disk, NULL);
if (failmsg != NULL) {
ERROR(disk, failmsg);
return (-1);
}
switch (errno) {
case EIO:
ERROR(disk, "unable to write cylinder group");
break;
default:
ERROR(disk, strerror(errno));
break;
}
return (-1);
}
snprintf(errmsg, BUFSIZ, "Cylinder group %d in buffer does not match "
"the cylinder group %d that cgwrite1 requested",
disk->d_cg.cg_cgx, cg);
ERROR(disk, errmsg);
errno = EDOOFUS;
return (-1);
}
int
cgput(int devfd, struct fs *fs, struct cg *cgp)
{
size_t cnt;
if ((fs->fs_metackhash & CK_CYLGRP) != 0) {
cgp->cg_ckhash = 0;
cgp->cg_ckhash =
calculate_crc32c(~0L, (void *)cgp, fs->fs_cgsize);
}
failmsg = NULL;
if ((cnt = pwrite(devfd, cgp, fs->fs_cgsize,
fsbtodb(fs, cgtod(fs, cgp->cg_cgx)) *
(fs->fs_fsize / fsbtodb(fs,1)))) < 0)
return (-1);
if (cnt != fs->fs_cgsize) {
failmsg = "short write to block device";
return (-1);
}
return (0);
}
diff --git a/lib/libufs/inode.c b/lib/libufs/inode.c
index 497ff4c854f6..c4a0cab9e95a 100644
--- a/lib/libufs/inode.c
+++ b/lib/libufs/inode.c
@@ -1,120 +1,122 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2002 Juli Mallett. All rights reserved.
*
* This software was written by Juli Mallett <jmallett@FreeBSD.org> for the
* FreeBSD project. Redistribution and use in source and binary forms, with
* or without modification, are permitted provided that the following
* conditions are met:
*
* 1. Redistribution of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* 2. Redistribution in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
* INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
* IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/mount.h>
#include <sys/disklabel.h>
#include <sys/stat.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <libufs.h>
int
getinode(struct uufsd *disk, union dinodep *dp, ino_t inum)
{
ino_t min, max;
caddr_t inoblock;
struct fs *fs;
ERROR(disk, NULL);
fs = &disk->d_fs;
if (inum >= (ino_t)fs->fs_ipg * fs->fs_ncg) {
ERROR(disk, "inode number out of range");
return (-1);
}
inoblock = disk->d_inoblock;
min = disk->d_inomin;
max = disk->d_inomax;
if (inoblock == NULL) {
inoblock = malloc(fs->fs_bsize);
if (inoblock == NULL) {
ERROR(disk, "unable to allocate inode block");
return (-1);
}
disk->d_inoblock = inoblock;
}
if (inum >= min && inum < max)
goto gotit;
bread(disk, fsbtodb(fs, ino_to_fsba(fs, inum)), inoblock,
fs->fs_bsize);
disk->d_inomin = min = inum - (inum % INOPB(fs));
disk->d_inomax = max = min + INOPB(fs);
gotit: switch (disk->d_ufs) {
case 1:
disk->d_dp.dp1 = &((struct ufs1_dinode *)inoblock)[inum - min];
if (dp != NULL)
*dp = disk->d_dp;
return (0);
case 2:
disk->d_dp.dp2 = &((struct ufs2_dinode *)inoblock)[inum - min];
if (dp != NULL)
*dp = disk->d_dp;
if (ffs_verify_dinode_ckhash(fs, disk->d_dp.dp2) == 0)
return (0);
ERROR(disk, "check-hash failed for inode read from disk");
return (-1);
default:
break;
}
ERROR(disk, "unknown UFS filesystem type");
return (-1);
}
int
putinode(struct uufsd *disk)
{
struct fs *fs;
fs = &disk->d_fs;
if (disk->d_inoblock == NULL) {
ERROR(disk, "No inode block allocated");
return (-1);
}
if (disk->d_ufs == 2)
ffs_update_dinode_ckhash(fs, disk->d_dp.dp2);
if (bwrite(disk, fsbtodb(fs, ino_to_fsba(&disk->d_fs, disk->d_inomin)),
disk->d_inoblock, disk->d_fs.fs_bsize) <= 0)
return (-1);
return (0);
}
diff --git a/lib/libufs/sblock.c b/lib/libufs/sblock.c
index 1d687d6d41aa..3b65e79b02b5 100644
--- a/lib/libufs/sblock.c
+++ b/lib/libufs/sblock.c
@@ -1,209 +1,211 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2002 Juli Mallett. All rights reserved.
*
* This software was written by Juli Mallett <jmallett@FreeBSD.org> for the
* FreeBSD project. Redistribution and use in source and binary forms, with
* or without modification, are permitted provided that the following
* conditions are met:
*
* 1. Redistribution of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* 2. Redistribution in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
* INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
* IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/mount.h>
#include <sys/disklabel.h>
#include <sys/stat.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include <errno.h>
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>
#include <libufs.h>
int
sbread(struct uufsd *disk)
{
struct fs *fs;
ERROR(disk, NULL);
if ((errno = sbget(disk->d_fd, &fs, STDSB)) != 0) {
switch (errno) {
case EIO:
ERROR(disk, "non-existent or truncated superblock");
break;
case ENOENT:
ERROR(disk, "no usable known superblock found");
break;
case ENOSPC:
ERROR(disk, "failed to allocate space for superblock "
"information");
break;
case EINVAL:
ERROR(disk, "The previous newfs operation on this "
"volume did not complete.\nYou must complete "
"newfs before using this volume.");
break;
default:
ERROR(disk, "unknown superblock read error");
errno = EIO;
break;
}
disk->d_ufs = 0;
return (-1);
}
memcpy(&disk->d_fs, fs, fs->fs_sbsize);
free(fs);
fs = &disk->d_fs;
if (fs->fs_magic == FS_UFS1_MAGIC)
disk->d_ufs = 1;
if (fs->fs_magic == FS_UFS2_MAGIC)
disk->d_ufs = 2;
disk->d_bsize = fs->fs_fsize / fsbtodb(fs, 1);
disk->d_sblock = fs->fs_sblockloc / disk->d_bsize;
disk->d_si = fs->fs_si;
return (0);
}
int
sbwrite(struct uufsd *disk, int all)
{
struct fs *fs;
int rv;
ERROR(disk, NULL);
rv = ufs_disk_write(disk);
if (rv == -1) {
ERROR(disk, "failed to open disk for writing");
return (-1);
}
fs = &disk->d_fs;
if ((errno = sbput(disk->d_fd, fs, all ? fs->fs_ncg : 0)) != 0) {
switch (errno) {
case EIO:
ERROR(disk, "failed to write superblock");
break;
default:
ERROR(disk, "unknown superblock write error");
errno = EIO;
break;
}
return (-1);
}
return (0);
}
/*
* These are the low-level functions that actually read and write
* the superblock and its associated data. The actual work is done by
* the functions ffs_sbget and ffs_sbput in /sys/ufs/ffs/ffs_subr.c.
*/
static int use_pread(void *devfd, off_t loc, void **bufp, int size);
static int use_pwrite(void *devfd, off_t loc, void *buf, int size);
/*
* Read a superblock from the devfd device allocating memory returned
* in fsp. Also read the superblock summary information.
*/
int
sbget(int devfd, struct fs **fsp, off_t sblockloc)
{
return (ffs_sbget(&devfd, fsp, sblockloc, "user", use_pread));
}
/*
* A read function for use by user-level programs using libufs.
*/
static int
use_pread(void *devfd, off_t loc, void **bufp, int size)
{
int fd;
fd = *(int *)devfd;
if ((*bufp = malloc(size)) == NULL)
return (ENOSPC);
if (pread(fd, *bufp, size, loc) != size)
return (EIO);
return (0);
}
/*
* Write a superblock to the devfd device from the memory pointed to by fs.
* Also write out the superblock summary information but do not free the
* summary information memory.
*
* Additionally write out numaltwrite of the alternate superblocks. Use
* fs->fs_ncg to write out all of the alternate superblocks.
*/
int
sbput(int devfd, struct fs *fs, int numaltwrite)
{
struct csum *savedcsp;
off_t savedactualloc;
int i, error;
if ((error = ffs_sbput(&devfd, fs, fs->fs_sblockactualloc,
use_pwrite)) != 0)
return (error);
if (numaltwrite == 0)
return (0);
savedactualloc = fs->fs_sblockactualloc;
savedcsp = fs->fs_csp;
fs->fs_csp = NULL;
for (i = 0; i < numaltwrite; i++) {
fs->fs_sblockactualloc = dbtob(fsbtodb(fs, cgsblock(fs, i)));
if ((error = ffs_sbput(&devfd, fs, fs->fs_sblockactualloc,
use_pwrite)) != 0) {
fs->fs_sblockactualloc = savedactualloc;
fs->fs_csp = savedcsp;
return (error);
}
}
fs->fs_sblockactualloc = savedactualloc;
fs->fs_csp = savedcsp;
return (0);
}
/*
* A write function for use by user-level programs using sbput in libufs.
*/
static int
use_pwrite(void *devfd, off_t loc, void *buf, int size)
{
int fd;
fd = *(int *)devfd;
if (pwrite(fd, buf, size, loc) != size)
return (EIO);
return (0);
}
diff --git a/lib/libufs/type.c b/lib/libufs/type.c
index 653ef8ceeea1..a060f8ec5abf 100644
--- a/lib/libufs/type.c
+++ b/lib/libufs/type.c
@@ -1,204 +1,206 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2002 Juli Mallett. All rights reserved.
*
* This software was written by Juli Mallett <jmallett@FreeBSD.org> for the
* FreeBSD project. Redistribution and use in source and binary forms, with
* or without modification, are permitted provided that the following
* conditions are met:
*
* 1. Redistribution of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* 2. Redistribution in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
* INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
* IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/mount.h>
#include <sys/disklabel.h>
#include <sys/stat.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include <errno.h>
#include <fcntl.h>
#include <fstab.h>
#include <paths.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <libufs.h>
/* Internally, track the 'name' value, it's ours. */
#define MINE_NAME 0x01
/* Track if its fd points to a writable device. */
#define MINE_WRITE 0x02
int
ufs_disk_close(struct uufsd *disk)
{
ERROR(disk, NULL);
close(disk->d_fd);
disk->d_fd = -1;
if (disk->d_inoblock != NULL) {
free(disk->d_inoblock);
disk->d_inoblock = NULL;
}
if (disk->d_mine & MINE_NAME) {
free((char *)(uintptr_t)disk->d_name);
disk->d_name = NULL;
}
if (disk->d_si != NULL) {
free(disk->d_si->si_csp);
free(disk->d_si);
disk->d_si = NULL;
}
return (0);
}
int
ufs_disk_fillout(struct uufsd *disk, const char *name)
{
if (ufs_disk_fillout_blank(disk, name) == -1) {
return (-1);
}
if (sbread(disk) == -1) {
ERROR(disk, "could not read superblock to fill out disk");
ufs_disk_close(disk);
return (-1);
}
return (0);
}
int
ufs_disk_fillout_blank(struct uufsd *disk, const char *name)
{
struct stat st;
struct fstab *fs;
struct statfs sfs;
const char *oname;
char dev[MAXPATHLEN];
int fd, ret;
ERROR(disk, NULL);
oname = name;
again: if ((ret = stat(name, &st)) < 0) {
if (*name != '/') {
snprintf(dev, sizeof(dev), "%s%s", _PATH_DEV, name);
name = dev;
goto again;
}
/*
* The given object doesn't exist, but don't panic just yet -
* it may be still mount point listed in /etc/fstab, but without
* existing corresponding directory.
*/
name = oname;
}
if (ret >= 0 && S_ISREG(st.st_mode)) {
/* Possibly a disk image, give it a try. */
;
} else if (ret >= 0 && S_ISCHR(st.st_mode)) {
/* This is what we need, do nothing. */
;
} else if ((fs = getfsfile(name)) != NULL) {
/*
* The given mount point is listed in /etc/fstab.
* It is possible that someone unmounted file system by hand
* and different file system is mounted on this mount point,
* but we still prefer /etc/fstab entry, because on the other
* hand, there could be /etc/fstab entry for this mount
* point, but file system is not mounted yet (eg. noauto) and
* statfs(2) will point us at different file system.
*/
name = fs->fs_spec;
} else if (ret >= 0 && S_ISDIR(st.st_mode)) {
/*
* The mount point is not listed in /etc/fstab, so it may be
* file system mounted by hand.
*/
if (statfs(name, &sfs) < 0) {
ERROR(disk, "could not find special device");
return (-1);
}
strlcpy(dev, sfs.f_mntfromname, sizeof(dev));
name = dev;
} else {
ERROR(disk, "could not find special device");
return (-1);
}
fd = open(name, O_RDONLY);
if (fd == -1) {
ERROR(disk, "could not open special device");
return (-1);
}
disk->d_bsize = 1;
disk->d_ccg = 0;
disk->d_fd = fd;
disk->d_inoblock = NULL;
disk->d_inomin = 0;
disk->d_inomax = 0;
disk->d_lcg = 0;
disk->d_mine = 0;
disk->d_ufs = 0;
disk->d_error = NULL;
disk->d_si = NULL;
if (oname != name) {
name = strdup(name);
if (name == NULL) {
ERROR(disk, "could not allocate memory for disk name");
return (-1);
}
disk->d_mine |= MINE_NAME;
}
disk->d_name = name;
return (0);
}
int
ufs_disk_write(struct uufsd *disk)
{
int fd;
ERROR(disk, NULL);
if (disk->d_mine & MINE_WRITE)
return (0);
fd = open(disk->d_name, O_RDWR);
if (fd < 0) {
ERROR(disk, "failed to open disk for writing");
return (-1);
}
close(disk->d_fd);
disk->d_fd = fd;
disk->d_mine |= MINE_WRITE;
return (0);
}
diff --git a/sbin/dump/main.c b/sbin/dump/main.c
index bbc54d7ab5d7..8752f2c1bea5 100644
--- a/sbin/dump/main.c
+++ b/sbin/dump/main.c
@@ -1,795 +1,797 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1980, 1991, 1993, 1994
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#ifndef lint
static const char copyright[] =
"@(#) Copyright (c) 1980, 1991, 1993, 1994\n\
The Regents of the University of California. All rights reserved.\n";
#endif /* not lint */
#ifndef lint
#if 0
static char sccsid[] = "@(#)main.c 8.6 (Berkeley) 5/1/95";
#endif
static const char rcsid[] =
"$FreeBSD$";
#endif /* not lint */
#include <sys/param.h>
#include <sys/stat.h>
#include <sys/mount.h>
#include <sys/disklabel.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ffs/fs.h>
#include <protocols/dumprestore.h>
#include <ctype.h>
#include <err.h>
#include <errno.h>
#include <fcntl.h>
#include <fstab.h>
#include <libufs.h>
#include <limits.h>
#include <signal.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <timeconv.h>
#include <unistd.h>
#include "dump.h"
#include "pathnames.h"
int mapsize; /* size of the state maps */
char *usedinomap; /* map of allocated inodes */
char *dumpdirmap; /* map of directories to be dumped */
char *dumpinomap; /* map of files to be dumped */
char *disk; /* name of the disk file */
char *tape; /* name of the tape file */
char *popenout; /* popen(3) per-"tape" command */
int level; /* dump level of this dump */
int uflag; /* update flag */
int diskfd; /* disk file descriptor */
int pipeout; /* true => output to standard output */
int density = 0; /* density in bytes/0.1" " <- this is for hilit19 */
long tapesize; /* estimated tape size, blocks */
long tsize; /* tape size in 0.1" units */
int etapes; /* estimated number of tapes */
int nonodump; /* if set, do not honor UF_NODUMP user flags */
int unlimited; /* if set, write to end of medium */
int cachesize = 0; /* block cache size (in bytes), defaults to 0 */
int rsync_friendly; /* be friendly with rsync */
int notify = 0; /* notify operator flag */
int blockswritten = 0; /* number of blocks written on current tape */
int tapeno = 0; /* current tape number */
int ntrec = NTREC; /* # tape blocks in each tape record */
long blocksperfile; /* number of blocks per output file */
int cartridge = 0; /* Assume non-cartridge tape */
char *host = NULL; /* remote host (if any) */
time_t tstart_writing; /* when started writing the first tape block */
time_t tend_writing; /* after writing the last tape block */
int passno; /* current dump pass number */
struct fs *sblock; /* the file system super block */
long dev_bsize = 1; /* recalculated below */
int dev_bshift; /* log2(dev_bsize) */
int tp_bshift; /* log2(TP_BSIZE) */
int snapdump = 0; /* dumping live filesystem, so use snapshot */
static char *getmntpt(char *, int *);
static long numarg(const char *, long, long);
static void obsolete(int *, char **[]);
static void usage(void) __dead2;
int
main(int argc, char *argv[])
{
struct stat sb;
ino_t ino;
int dirty;
union dinode *dp;
struct fstab *dt;
char *map, *mntpt;
int ch, mode, mntflags;
int i, ret, anydirskipped, bflag = 0, Tflag = 0, honorlevel = 1;
int just_estimate = 0;
ino_t maxino;
char *tmsg;
spcl.c_date = _time_to_time64(time(NULL));
tsize = 0; /* Default later, based on 'c' option for cart tapes */
dumpdates = _PATH_DUMPDATES;
popenout = NULL;
tape = NULL;
if (TP_BSIZE / DEV_BSIZE == 0 || TP_BSIZE % DEV_BSIZE != 0)
quit("TP_BSIZE must be a multiple of DEV_BSIZE\n");
level = 0;
rsync_friendly = 0;
if (argc < 2)
usage();
obsolete(&argc, &argv);
while ((ch = getopt(argc, argv,
"0123456789aB:b:C:cD:d:f:h:LnP:RrSs:T:uWw")) != -1)
switch (ch) {
/* dump level */
case '0': case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
level = 10 * level + ch - '0';
break;
case 'a': /* `auto-size', Write to EOM. */
unlimited = 1;
break;
case 'B': /* blocks per output file */
blocksperfile = numarg("number of blocks per file",
1L, 0L);
break;
case 'b': /* blocks per tape write */
ntrec = numarg("number of blocks per write",
1L, 1000L);
break;
case 'C':
cachesize = numarg("cachesize", 0, 0) * 1024 * 1024;
break;
case 'c': /* Tape is cart. not 9-track */
cartridge = 1;
break;
case 'D':
dumpdates = optarg;
break;
case 'd': /* density, in bits per inch */
density = numarg("density", 10L, 327670L) / 10;
if (density >= 625 && !bflag)
ntrec = HIGHDENSITYTREC;
break;
case 'f': /* output file */
if (popenout != NULL)
errx(X_STARTUP, "You cannot use the P and f "
"flags together.\n");
tape = optarg;
break;
case 'h':
honorlevel = numarg("honor level", 0L, 10L);
break;
case 'L':
snapdump = 1;
break;
case 'n': /* notify operators */
notify = 1;
break;
case 'P':
if (tape != NULL)
errx(X_STARTUP, "You cannot use the P and f "
"flags together.\n");
popenout = optarg;
break;
case 'r': /* store slightly less data to be friendly to rsync */
if (rsync_friendly < 1)
rsync_friendly = 1;
break;
case 'R': /* store even less data to be friendlier to rsync */
if (rsync_friendly < 2)
rsync_friendly = 2;
break;
case 'S': /* exit after estimating # of tapes */
just_estimate = 1;
break;
case 's': /* tape size, feet */
tsize = numarg("tape size", 1L, 0L) * 12 * 10;
break;
case 'T': /* time of last dump */
spcl.c_ddate = unctime(optarg);
if (spcl.c_ddate < 0) {
(void)fprintf(stderr, "bad time \"%s\"\n",
optarg);
exit(X_STARTUP);
}
Tflag = 1;
lastlevel = -1;
break;
case 'u': /* update /etc/dumpdates */
uflag = 1;
break;
case 'W': /* what to do */
case 'w':
lastdump(ch);
exit(X_FINOK); /* do nothing else */
default:
usage();
}
argc -= optind;
argv += optind;
if (argc < 1) {
(void)fprintf(stderr, "Must specify disk or file system\n");
exit(X_STARTUP);
}
disk = *argv++;
argc--;
if (argc >= 1) {
(void)fprintf(stderr, "Unknown arguments to dump:");
while (argc--)
(void)fprintf(stderr, " %s", *argv++);
(void)fprintf(stderr, "\n");
exit(X_STARTUP);
}
if (rsync_friendly && (level > 0)) {
(void)fprintf(stderr, "%s %s\n", "rsync friendly options",
"can be used only with level 0 dumps.");
exit(X_STARTUP);
}
if (Tflag && uflag) {
(void)fprintf(stderr,
"You cannot use the T and u flags together.\n");
exit(X_STARTUP);
}
if (popenout) {
tape = "child pipeline process";
} else if (tape == NULL && (tape = getenv("TAPE")) == NULL)
tape = _PATH_DEFTAPE;
if (strcmp(tape, "-") == 0) {
pipeout++;
tape = "standard output";
}
if (blocksperfile)
blocksperfile = rounddown(blocksperfile, ntrec);
else if (!unlimited) {
/*
* Determine how to default tape size and density
*
* density tape size
* 9-track 1600 bpi (160 bytes/.1") 2300 ft.
* 9-track 6250 bpi (625 bytes/.1") 2300 ft.
* cartridge 8000 bpi (100 bytes/.1") 1700 ft.
* (450*4 - slop)
* hilit19 hits again: "
*/
if (density == 0)
density = cartridge ? 100 : 160;
if (tsize == 0)
tsize = cartridge ? 1700L*120L : 2300L*120L;
}
if (strchr(tape, ':')) {
host = tape;
tape = strchr(host, ':');
*tape++ = '\0';
#ifdef RDUMP
if (strchr(tape, '\n')) {
(void)fprintf(stderr, "invalid characters in tape\n");
exit(X_STARTUP);
}
if (rmthost(host) == 0)
exit(X_STARTUP);
#else
(void)fprintf(stderr, "remote dump not enabled\n");
exit(X_STARTUP);
#endif
}
(void)setuid(getuid()); /* rmthost() is the only reason to be setuid */
if (signal(SIGHUP, SIG_IGN) != SIG_IGN)
signal(SIGHUP, sig);
if (signal(SIGTRAP, SIG_IGN) != SIG_IGN)
signal(SIGTRAP, sig);
if (signal(SIGFPE, SIG_IGN) != SIG_IGN)
signal(SIGFPE, sig);
if (signal(SIGBUS, SIG_IGN) != SIG_IGN)
signal(SIGBUS, sig);
if (signal(SIGSEGV, SIG_IGN) != SIG_IGN)
signal(SIGSEGV, sig);
if (signal(SIGTERM, SIG_IGN) != SIG_IGN)
signal(SIGTERM, sig);
if (signal(SIGINT, interrupt) == SIG_IGN)
signal(SIGINT, SIG_IGN);
dump_getfstab(); /* /etc/fstab snarfed */
/*
* disk can be either the full special file name,
* the suffix of the special file name,
* the special name missing the leading '/',
* the file system name with or without the leading '/'.
*/
dt = fstabsearch(disk);
if (dt != NULL) {
disk = rawname(dt->fs_spec);
if (disk == NULL)
errx(X_STARTUP, "%s: unknown file system", dt->fs_spec);
(void)strncpy(spcl.c_dev, dt->fs_spec, NAMELEN);
(void)strncpy(spcl.c_filesys, dt->fs_file, NAMELEN);
} else {
(void)strncpy(spcl.c_dev, disk, NAMELEN);
(void)strncpy(spcl.c_filesys, "an unlisted file system",
NAMELEN);
}
spcl.c_dev[NAMELEN-1]='\0';
spcl.c_filesys[NAMELEN-1]='\0';
if ((mntpt = getmntpt(disk, &mntflags)) != NULL) {
if (mntflags & MNT_RDONLY) {
if (snapdump != 0) {
msg("WARNING: %s\n",
"-L ignored for read-only filesystem.");
snapdump = 0;
}
} else if (snapdump == 0) {
msg("WARNING: %s\n",
"should use -L when dumping live read-write "
"filesystems!");
} else {
char snapname[BUFSIZ], snapcmd[BUFSIZ];
snprintf(snapname, sizeof snapname, "%s/.snap", mntpt);
if ((stat(snapname, &sb) < 0) || !S_ISDIR(sb.st_mode)) {
msg("WARNING: %s %s\n",
"-L requested but snapshot location",
snapname);
msg(" %s: %s\n",
"is not a directory",
"dump downgraded, -L ignored");
snapdump = 0;
} else {
snprintf(snapname, sizeof snapname,
"%s/.snap/dump_snapshot", mntpt);
snprintf(snapcmd, sizeof snapcmd, "%s %s %s",
_PATH_MKSNAP_FFS, mntpt, snapname);
unlink(snapname);
if (system(snapcmd) != 0)
errx(X_STARTUP, "Cannot create %s: %s\n",
snapname, strerror(errno));
if ((diskfd = open(snapname, O_RDONLY)) < 0) {
unlink(snapname);
errx(X_STARTUP, "Cannot open %s: %s\n",
snapname, strerror(errno));
}
unlink(snapname);
if (fstat(diskfd, &sb) != 0)
err(X_STARTUP, "%s: stat", snapname);
spcl.c_date = _time_to_time64(sb.st_mtime);
}
}
} else if (snapdump != 0) {
msg("WARNING: Cannot use -L on an unmounted filesystem.\n");
snapdump = 0;
}
if (snapdump == 0) {
if ((diskfd = open(disk, O_RDONLY)) < 0)
err(X_STARTUP, "Cannot open %s", disk);
if (fstat(diskfd, &sb) != 0)
err(X_STARTUP, "%s: stat", disk);
if (S_ISDIR(sb.st_mode))
errx(X_STARTUP, "%s: unknown file system", disk);
}
(void)strcpy(spcl.c_label, "none");
(void)gethostname(spcl.c_host, NAMELEN);
spcl.c_level = level;
spcl.c_type = TS_TAPE;
if (rsync_friendly) {
/* don't store real dump times */
spcl.c_date = 0;
spcl.c_ddate = 0;
}
if (spcl.c_date == 0) {
tmsg = "the epoch\n";
} else {
time_t t = _time64_to_time(spcl.c_date);
tmsg = ctime(&t);
}
msg("Date of this level %d dump: %s", level, tmsg);
if (!Tflag && (!rsync_friendly))
getdumptime(); /* /etc/dumpdates snarfed */
if (spcl.c_ddate == 0) {
tmsg = "the epoch\n";
} else {
time_t t = _time64_to_time(spcl.c_ddate);
tmsg = ctime(&t);
}
if (lastlevel < 0)
msg("Date of last (level unknown) dump: %s", tmsg);
else
msg("Date of last level %d dump: %s", lastlevel, tmsg);
msg("Dumping %s%s ", snapdump ? "snapshot of ": "", disk);
if (dt != NULL)
msgtail("(%s) ", dt->fs_file);
if (host)
msgtail("to %s on host %s\n", tape, host);
else
msgtail("to %s\n", tape);
sync();
if ((ret = sbget(diskfd, &sblock, STDSB)) != 0) {
switch (ret) {
case ENOENT:
warn("Cannot find file system superblock");
return (1);
default:
warn("Unable to read file system superblock");
return (1);
}
}
dev_bsize = sblock->fs_fsize / fsbtodb(sblock, 1);
dev_bshift = ffs(dev_bsize) - 1;
if (dev_bsize != (1 << dev_bshift))
quit("dev_bsize (%ld) is not a power of 2", dev_bsize);
tp_bshift = ffs(TP_BSIZE) - 1;
if (TP_BSIZE != (1 << tp_bshift))
quit("TP_BSIZE (%d) is not a power of 2", TP_BSIZE);
maxino = sblock->fs_ipg * sblock->fs_ncg;
mapsize = roundup(howmany(maxino, CHAR_BIT), TP_BSIZE);
usedinomap = (char *)calloc((unsigned) mapsize, sizeof(char));
dumpdirmap = (char *)calloc((unsigned) mapsize, sizeof(char));
dumpinomap = (char *)calloc((unsigned) mapsize, sizeof(char));
tapesize = 3 * (howmany(mapsize * sizeof(char), TP_BSIZE) + 1);
nonodump = spcl.c_level < honorlevel;
passno = 1;
setproctitle("%s: pass 1: regular files", disk);
msg("mapping (Pass I) [regular files]\n");
anydirskipped = mapfiles(maxino, &tapesize);
passno = 2;
setproctitle("%s: pass 2: directories", disk);
msg("mapping (Pass II) [directories]\n");
while (anydirskipped) {
anydirskipped = mapdirs(maxino, &tapesize);
}
if (pipeout || unlimited) {
tapesize += 10; /* 10 trailer blocks */
msg("estimated %ld tape blocks.\n", tapesize);
} else {
double fetapes;
if (blocksperfile)
fetapes = (double) tapesize / blocksperfile;
else if (cartridge) {
/* Estimate number of tapes, assuming streaming stops at
the end of each block written, and not in mid-block.
Assume no erroneous blocks; this can be compensated
for with an artificially low tape size. */
fetapes =
( (double) tapesize /* blocks */
* TP_BSIZE /* bytes/block */
* (1.0/density) /* 0.1" / byte " */
+
(double) tapesize /* blocks */
* (1.0/ntrec) /* streaming-stops per block */
* 15.48 /* 0.1" / streaming-stop " */
) * (1.0 / tsize ); /* tape / 0.1" " */
} else {
/* Estimate number of tapes, for old fashioned 9-track
tape */
int tenthsperirg = (density == 625) ? 3 : 7;
fetapes =
( (double) tapesize /* blocks */
* TP_BSIZE /* bytes / block */
* (1.0/density) /* 0.1" / byte " */
+
(double) tapesize /* blocks */
* (1.0/ntrec) /* IRG's / block */
* tenthsperirg /* 0.1" / IRG " */
) * (1.0 / tsize ); /* tape / 0.1" " */
}
etapes = fetapes; /* truncating assignment */
etapes++;
/* count the dumped inodes map on each additional tape */
tapesize += (etapes - 1) *
(howmany(mapsize * sizeof(char), TP_BSIZE) + 1);
tapesize += etapes + 10; /* headers + 10 trailer blks */
msg("estimated %ld tape blocks on %3.2f tape(s).\n",
tapesize, fetapes);
}
/*
* If the user only wants an estimate of the number of
* tapes, exit now.
*/
if (just_estimate)
exit(0);
/*
* Allocate tape buffer.
*/
if (!alloctape())
quit(
"can't allocate tape buffers - try a smaller blocking factor.\n");
startnewtape(1);
(void)time((time_t *)&(tstart_writing));
dumpmap(usedinomap, TS_CLRI, maxino - 1);
passno = 3;
setproctitle("%s: pass 3: directories", disk);
msg("dumping (Pass III) [directories]\n");
dirty = 0; /* XXX just to get gcc to shut up */
for (map = dumpdirmap, ino = 1; ino < maxino; ino++) {
if (((ino - 1) % CHAR_BIT) == 0) /* map is offset by 1 */
dirty = *map++;
else
dirty >>= 1;
if ((dirty & 1) == 0)
continue;
/*
* Skip directory inodes deleted and maybe reallocated
*/
dp = getino(ino, &mode);
if (mode != IFDIR)
continue;
(void)dumpino(dp, ino);
}
passno = 4;
setproctitle("%s: pass 4: regular files", disk);
msg("dumping (Pass IV) [regular files]\n");
for (map = dumpinomap, ino = 1; ino < maxino; ino++) {
if (((ino - 1) % CHAR_BIT) == 0) /* map is offset by 1 */
dirty = *map++;
else
dirty >>= 1;
if ((dirty & 1) == 0)
continue;
/*
* Skip inodes deleted and reallocated as directories.
*/
dp = getino(ino, &mode);
if (mode == IFDIR)
continue;
(void)dumpino(dp, ino);
}
(void)time((time_t *)&(tend_writing));
spcl.c_type = TS_END;
for (i = 0; i < ntrec; i++)
writeheader(maxino - 1);
if (pipeout)
msg("DUMP: %jd tape blocks\n", (intmax_t)spcl.c_tapea);
else
msg("DUMP: %jd tape blocks on %d volume%s\n",
(intmax_t)spcl.c_tapea, spcl.c_volume,
(spcl.c_volume == 1) ? "" : "s");
/* report dump performance, avoid division through zero */
if (tend_writing - tstart_writing == 0)
msg("finished in less than a second\n");
else
msg("finished in %jd seconds, throughput %jd KBytes/sec\n",
(intmax_t)tend_writing - tstart_writing,
(intmax_t)(spcl.c_tapea /
(tend_writing - tstart_writing)));
putdumptime();
trewind();
broadcast("DUMP IS DONE!\a\a\n");
msg("DUMP IS DONE\n");
Exit(X_FINOK);
/* NOTREACHED */
}
static void
usage(void)
{
fprintf(stderr,
"usage: dump [-0123456789acLnSu] [-B records] [-b blocksize] [-C cachesize]\n"
" [-D dumpdates] [-d density] [-f file | -P pipecommand] [-h level]\n"
" [-s feet] [-T date] filesystem\n"
" dump -W | -w\n");
exit(X_STARTUP);
}
/*
* Check to see if a disk is currently mounted.
*/
static char *
getmntpt(char *name, int *mntflagsp)
{
long mntsize, i;
struct statfs *mntbuf;
mntsize = getmntinfo(&mntbuf, MNT_NOWAIT);
for (i = 0; i < mntsize; i++) {
if (!strcmp(mntbuf[i].f_mntfromname, name)) {
*mntflagsp = mntbuf[i].f_flags;
return (mntbuf[i].f_mntonname);
}
}
return (0);
}
/*
* Pick up a numeric argument. It must be nonnegative and in the given
* range (except that a vmax of 0 means unlimited).
*/
static long
numarg(const char *meaning, long vmin, long vmax)
{
char *p;
long val;
val = strtol(optarg, &p, 10);
if (*p)
errx(1, "illegal %s -- %s", meaning, optarg);
if (val < vmin || (vmax && val > vmax))
errx(1, "%s must be between %ld and %ld", meaning, vmin, vmax);
return (val);
}
void
sig(int signo)
{
switch(signo) {
case SIGALRM:
case SIGBUS:
case SIGFPE:
case SIGHUP:
case SIGTERM:
case SIGTRAP:
if (pipeout)
quit("Signal on pipe: cannot recover\n");
msg("Rewriting attempted as response to unknown signal.\n");
(void)fflush(stderr);
(void)fflush(stdout);
close_rewind();
exit(X_REWRITE);
/* NOTREACHED */
case SIGSEGV:
msg("SIGSEGV: ABORTING!\n");
(void)signal(SIGSEGV, SIG_DFL);
(void)kill(0, SIGSEGV);
/* NOTREACHED */
}
}
char *
rawname(char *cp)
{
struct stat sb;
/*
* Ensure that the device passed in is a raw device.
*/
if (stat(cp, &sb) == 0 && (sb.st_mode & S_IFMT) == S_IFCHR)
return (cp);
/*
* Since there's only one device type now, we can't construct any
* better name, so we have to return NULL.
*/
return (NULL);
}
/*
* obsolete --
* Change set of key letters and ordered arguments into something
* getopt(3) will like.
*/
static void
obsolete(int *argcp, char **argvp[])
{
int argc, flags;
char *ap, **argv, *flagsp, **nargv, *p;
/* Setup. */
argv = *argvp;
argc = *argcp;
/*
* Return if no arguments or first argument has leading
* dash or slash.
*/
ap = argv[1];
if (argc == 1 || *ap == '-' || *ap == '/')
return;
/* Allocate space for new arguments. */
if ((*argvp = nargv = malloc((argc + 1) * sizeof(char *))) == NULL ||
(p = flagsp = malloc(strlen(ap) + 2)) == NULL)
err(1, NULL);
*nargv++ = *argv;
argv += 2;
for (flags = 0; *ap; ++ap) {
switch (*ap) {
case 'B':
case 'b':
case 'd':
case 'f':
case 'D':
case 'C':
case 'h':
case 's':
case 'T':
if (*argv == NULL) {
warnx("option requires an argument -- %c", *ap);
usage();
}
if ((nargv[0] = malloc(strlen(*argv) + 2 + 1)) == NULL)
err(1, NULL);
nargv[0][0] = '-';
nargv[0][1] = *ap;
(void)strcpy(&nargv[0][2], *argv);
++argv;
++nargv;
break;
default:
if (!flags) {
*p++ = '-';
flags = 1;
}
*p++ = *ap;
break;
}
}
/* Terminate flags. */
if (flags) {
*p = '\0';
*nargv++ = flagsp;
} else
free(flagsp);
/* Copy remaining arguments. */
while ((*nargv++ = *argv++));
/* Update argument count. */
*argcp = nargv - *argvp - 1;
}
diff --git a/sbin/ffsinfo/ffsinfo.c b/sbin/ffsinfo/ffsinfo.c
index 9bd3210986a9..33ec5f175cbd 100644
--- a/sbin/ffsinfo/ffsinfo.c
+++ b/sbin/ffsinfo/ffsinfo.c
@@ -1,649 +1,651 @@
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (c) 2000 Christoph Herrmann, Thomas-Henning von Kamptz
* Copyright (c) 1980, 1989, 1993 The Regents of the University of California.
* All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* Christoph Herrmann and Thomas-Henning von Kamptz, Munich and Frankfurt.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgment:
* This product includes software developed by the University of
* California, Berkeley and its contributors, as well as Christoph
* Herrmann and Thomas-Henning von Kamptz.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $TSHeader: src/sbin/ffsinfo/ffsinfo.c,v 1.4 2000/12/12 19:30:55 tomsoft Exp $
*
*/
#ifndef lint
static const char copyright[] =
"@(#) Copyright (c) 2000 Christoph Herrmann, Thomas-Henning von Kamptz\n\
Copyright (c) 1980, 1989, 1993 The Regents of the University of California.\n\
All rights reserved.\n";
#endif /* not lint */
#ifndef lint
static const char rcsid[] =
"$FreeBSD$";
#endif /* not lint */
/* ********************************************************** INCLUDES ***** */
#include <sys/param.h>
#include <sys/disklabel.h>
#include <sys/mount.h>
#include <sys/stat.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include <ctype.h>
#include <err.h>
#include <errno.h>
#include <fcntl.h>
#include <libufs.h>
#include <paths.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "debug.h"
/* *********************************************************** GLOBALS ***** */
#ifdef FS_DEBUG
int _dbg_lvl_ = (DL_INFO); /* DL_TRC */
#endif /* FS_DEBUG */
static struct uufsd disk;
#define sblock disk.d_fs
#define acg disk.d_cg
static union {
struct fs fs;
char pad[SBLOCKSIZE];
} fsun;
#define osblock fsun.fs
static char i1blk[MAXBSIZE];
static char i2blk[MAXBSIZE];
static char i3blk[MAXBSIZE];
static struct csum *fscs;
/* ******************************************************** PROTOTYPES ***** */
static void usage(void);
static void dump_whole_ufs1_inode(ino_t, int);
static void dump_whole_ufs2_inode(ino_t, int);
#define DUMP_WHOLE_INODE(A,B) \
( disk.d_ufs == 1 \
? dump_whole_ufs1_inode((A),(B)) : dump_whole_ufs2_inode((A),(B)) )
/* ************************************************************** main ***** */
/*
* ffsinfo(8) is a tool to dump all metadata of a file system. It helps to find
* errors is the file system much easier. You can run ffsinfo before and after
* an fsck(8), and compare the two ascii dumps easy with diff, and you see
* directly where the problem is. You can control how much detail you want to
* see with some command line arguments. You can also easy check the status
* of a file system, like is there is enough space for growing a file system,
* or how many active snapshots do we have. It provides much more detailed
* information then dumpfs. Snapshots, as they are very new, are not really
* supported. They are just mentioned currently, but it is planned to run
* also over active snapshots, to even get that output.
*/
int
main(int argc, char **argv)
{
DBG_FUNC("main")
char *device, *special;
int ch;
size_t len;
struct stat st;
struct csum *dbg_csp;
int dbg_csc;
char dbg_line[80];
int cylno,i;
int cfg_cg, cfg_in, cfg_lv;
int cg_start, cg_stop;
ino_t in;
char *out_file;
DBG_ENTER;
cfg_lv = 0xff;
cfg_in = -2;
cfg_cg = -2;
out_file = strdup("-");
while ((ch = getopt(argc, argv, "g:i:l:o:")) != -1) {
switch (ch) {
case 'g':
cfg_cg = strtol(optarg, NULL, 0);
if (errno == EINVAL || errno == ERANGE)
err(1, "%s", optarg);
if (cfg_cg < -1)
usage();
break;
case 'i':
cfg_in = strtol(optarg, NULL, 0);
if (errno == EINVAL || errno == ERANGE)
err(1, "%s", optarg);
if (cfg_in < 0)
usage();
break;
case 'l':
cfg_lv = strtol(optarg, NULL, 0);
if (errno == EINVAL||errno == ERANGE)
err(1, "%s", optarg);
if (cfg_lv < 0x1 || cfg_lv > 0x3ff)
usage();
break;
case 'o':
free(out_file);
out_file = strdup(optarg);
if (out_file == NULL)
errx(1, "strdup failed");
break;
case '?':
/* FALLTHROUGH */
default:
usage();
}
}
argc -= optind;
argv += optind;
if (argc != 1)
usage();
device = *argv;
/*
* Now we try to guess the (raw)device name.
*/
if (0 == strrchr(device, '/') && stat(device, &st) == -1) {
/*-
* No path prefix was given, so try in this order:
* /dev/r%s
* /dev/%s
* /dev/vinum/r%s
* /dev/vinum/%s.
*
* FreeBSD now doesn't distinguish between raw and block
* devices any longer, but it should still work this way.
*/
len = strlen(device) + strlen(_PATH_DEV) + 2 + strlen("vinum/");
special = (char *)malloc(len);
if (special == NULL)
errx(1, "malloc failed");
snprintf(special, len, "%sr%s", _PATH_DEV, device);
if (stat(special, &st) == -1) {
snprintf(special, len, "%s%s", _PATH_DEV, device);
if (stat(special, &st) == -1) {
snprintf(special, len, "%svinum/r%s",
_PATH_DEV, device);
if (stat(special, &st) == -1)
/* For now this is the 'last resort' */
snprintf(special, len, "%svinum/%s",
_PATH_DEV, device);
}
}
device = special;
}
if (ufs_disk_fillout(&disk, device) == -1)
err(1, "ufs_disk_fillout(%s) failed: %s", device, disk.d_error);
DBG_OPEN(out_file); /* already here we need a superblock */
if (cfg_lv & 0x001)
DBG_DUMP_FS(&sblock, "primary sblock");
/* Determine here what cylinder groups to dump */
if (cfg_cg==-2) {
cg_start = 0;
cg_stop = sblock.fs_ncg;
} else if (cfg_cg == -1) {
cg_start = sblock.fs_ncg - 1;
cg_stop = sblock.fs_ncg;
} else if (cfg_cg < sblock.fs_ncg) {
cg_start = cfg_cg;
cg_stop = cfg_cg + 1;
} else {
cg_start = sblock.fs_ncg;
cg_stop = sblock.fs_ncg;
}
if (cfg_lv & 0x004) {
fscs = (struct csum *)calloc((size_t)1,
(size_t)sblock.fs_cssize);
if (fscs == NULL)
errx(1, "calloc failed");
/* get the cylinder summary into the memory ... */
for (i = 0; i < sblock.fs_cssize; i += sblock.fs_bsize) {
if (bread(&disk, fsbtodb(&sblock,
sblock.fs_csaddr + numfrags(&sblock, i)),
(void *)(((char *)fscs)+i),
(size_t)(sblock.fs_cssize-i < sblock.fs_bsize ?
sblock.fs_cssize - i : sblock.fs_bsize)) == -1)
err(1, "bread: %s", disk.d_error);
}
dbg_csp = fscs;
/* ... and dump it */
for (dbg_csc = 0; dbg_csc < sblock.fs_ncg; dbg_csc++) {
snprintf(dbg_line, sizeof(dbg_line),
"%d. csum in fscs", dbg_csc);
DBG_DUMP_CSUM(&sblock,
dbg_line,
dbg_csp++);
}
}
if (cfg_lv & 0xf8) {
/* for each requested cylinder group ... */
for (cylno = cg_start; cylno < cg_stop; cylno++) {
snprintf(dbg_line, sizeof(dbg_line), "cgr %d", cylno);
if (cfg_lv & 0x002) {
/* dump the superblock copies */
if (bread(&disk, fsbtodb(&sblock,
cgsblock(&sblock, cylno)),
(void *)&osblock, SBLOCKSIZE) == -1)
err(1, "bread: %s", disk.d_error);
DBG_DUMP_FS(&osblock, dbg_line);
}
/*
* Read the cylinder group and dump whatever was
* requested.
*/
if (bread(&disk, fsbtodb(&sblock,
cgtod(&sblock, cylno)), (void *)&acg,
(size_t)sblock.fs_cgsize) == -1)
err(1, "bread: %s", disk.d_error);
if (cfg_lv & 0x008)
DBG_DUMP_CG(&sblock, dbg_line, &acg);
if (cfg_lv & 0x010)
DBG_DUMP_INMAP(&sblock, dbg_line, &acg);
if (cfg_lv & 0x020)
DBG_DUMP_FRMAP(&sblock, dbg_line, &acg);
if (cfg_lv & 0x040) {
DBG_DUMP_CLMAP(&sblock, dbg_line, &acg);
DBG_DUMP_CLSUM(&sblock, dbg_line, &acg);
}
#ifdef NOT_CURRENTLY
/*
* See the comment in sbin/growfs/debug.c for why this
* is currently disabled, and what needs to be done to
* re-enable it.
*/
if (disk.d_ufs == 1 && cfg_lv & 0x080)
DBG_DUMP_SPTBL(&sblock, dbg_line, &acg);
#endif
}
}
if (cfg_lv & 0x300) {
/* Dump the requested inode(s) */
if (cfg_in != -2)
DUMP_WHOLE_INODE((ino_t)cfg_in, cfg_lv);
else {
for (in = cg_start * sblock.fs_ipg;
in < (ino_t)cg_stop * sblock.fs_ipg;
in++)
DUMP_WHOLE_INODE(in, cfg_lv);
}
}
DBG_CLOSE;
DBG_LEAVE;
return 0;
}
/* ********************************************** dump_whole_ufs1_inode ***** */
/*
* Here we dump a list of all blocks allocated by this inode. We follow
* all indirect blocks.
*/
void
dump_whole_ufs1_inode(ino_t inode, int level)
{
DBG_FUNC("dump_whole_ufs1_inode")
union dinodep dp;
int rb;
unsigned int ind2ctr, ind3ctr;
ufs1_daddr_t *ind2ptr, *ind3ptr;
char comment[80];
DBG_ENTER;
/*
* Read the inode from disk/cache.
*/
if (getinode(&disk, &dp, inode) == -1)
err(1, "getinode: %s", disk.d_error);
if (dp.dp1->di_nlink == 0) {
DBG_LEAVE;
return; /* inode not in use */
}
/*
* Dump the main inode structure.
*/
snprintf(comment, sizeof(comment), "Inode 0x%08jx", (uintmax_t)inode);
if (level & 0x100) {
DBG_DUMP_INO(&sblock,
comment,
dp.dp1);
}
if (!(level & 0x200)) {
DBG_LEAVE;
return;
}
/*
* Ok, now prepare for dumping all direct and indirect pointers.
*/
rb = howmany(dp.dp1->di_size, sblock.fs_bsize) - UFS_NDADDR;
if (rb > 0) {
/*
* Dump single indirect block.
*/
if (bread(&disk, fsbtodb(&sblock, dp.dp1->di_ib[0]),
(void *)&i1blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment), "Inode 0x%08jx: indirect 0",
(uintmax_t)inode);
DBG_DUMP_IBLK(&sblock,
comment,
i1blk,
(size_t)rb);
rb -= howmany(sblock.fs_bsize, sizeof(ufs1_daddr_t));
}
if (rb > 0) {
/*
* Dump double indirect blocks.
*/
if (bread(&disk, fsbtodb(&sblock, dp.dp1->di_ib[1]),
(void *)&i2blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment), "Inode 0x%08jx: indirect 1",
(uintmax_t)inode);
DBG_DUMP_IBLK(&sblock,
comment,
i2blk,
howmany(rb, howmany(sblock.fs_bsize, sizeof(ufs1_daddr_t))));
for (ind2ctr = 0; ((ind2ctr < howmany(sblock.fs_bsize,
sizeof(ufs1_daddr_t))) && (rb > 0)); ind2ctr++) {
ind2ptr = &((ufs1_daddr_t *)(void *)&i2blk)[ind2ctr];
if (bread(&disk, fsbtodb(&sblock, *ind2ptr),
(void *)&i1blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment),
"Inode 0x%08jx: indirect 1->%d", (uintmax_t)inode,
ind2ctr);
DBG_DUMP_IBLK(&sblock,
comment,
i1blk,
(size_t)rb);
rb -= howmany(sblock.fs_bsize, sizeof(ufs1_daddr_t));
}
}
if (rb > 0) {
/*
* Dump triple indirect blocks.
*/
if (bread(&disk, fsbtodb(&sblock, dp.dp1->di_ib[2]),
(void *)&i3blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment), "Inode 0x%08jx: indirect 2",
(uintmax_t)inode);
#define SQUARE(a) ((a)*(a))
DBG_DUMP_IBLK(&sblock,
comment,
i3blk,
howmany(rb,
SQUARE(howmany(sblock.fs_bsize, sizeof(ufs1_daddr_t)))));
#undef SQUARE
for (ind3ctr = 0; ((ind3ctr < howmany(sblock.fs_bsize,
sizeof(ufs1_daddr_t))) && (rb > 0)); ind3ctr++) {
ind3ptr = &((ufs1_daddr_t *)(void *)&i3blk)[ind3ctr];
if (bread(&disk, fsbtodb(&sblock, *ind3ptr),
(void *)&i2blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment),
"Inode 0x%08jx: indirect 2->%d", (uintmax_t)inode,
ind3ctr);
DBG_DUMP_IBLK(&sblock,
comment,
i2blk,
howmany(rb,
howmany(sblock.fs_bsize, sizeof(ufs1_daddr_t))));
for (ind2ctr = 0; ((ind2ctr < howmany(sblock.fs_bsize,
sizeof(ufs1_daddr_t))) && (rb > 0)); ind2ctr++) {
ind2ptr=&((ufs1_daddr_t *)(void *)&i2blk)
[ind2ctr];
if (bread(&disk, fsbtodb(&sblock, *ind2ptr),
(void *)&i1blk, (size_t)sblock.fs_bsize)
== -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment),
"Inode 0x%08jx: indirect 2->%d->%d",
(uintmax_t)inode, ind3ctr, ind3ctr);
DBG_DUMP_IBLK(&sblock,
comment,
i1blk,
(size_t)rb);
rb -= howmany(sblock.fs_bsize,
sizeof(ufs1_daddr_t));
}
}
}
DBG_LEAVE;
return;
}
/* ********************************************** dump_whole_ufs2_inode ***** */
/*
* Here we dump a list of all blocks allocated by this inode. We follow
* all indirect blocks.
*/
void
dump_whole_ufs2_inode(ino_t inode, int level)
{
DBG_FUNC("dump_whole_ufs2_inode")
union dinodep dp;
int rb;
unsigned int ind2ctr, ind3ctr;
ufs2_daddr_t *ind2ptr, *ind3ptr;
char comment[80];
DBG_ENTER;
/*
* Read the inode from disk/cache.
*/
if (getinode(&disk, &dp, inode) == -1)
err(1, "getinode: %s", disk.d_error);
if (dp.dp2->di_nlink == 0) {
DBG_LEAVE;
return; /* inode not in use */
}
/*
* Dump the main inode structure.
*/
snprintf(comment, sizeof(comment), "Inode 0x%08jx", (uintmax_t)inode);
if (level & 0x100) {
DBG_DUMP_INO(&sblock, comment, dp.dp2);
}
if (!(level & 0x200)) {
DBG_LEAVE;
return;
}
/*
* Ok, now prepare for dumping all direct and indirect pointers.
*/
rb = howmany(dp.dp2->di_size, sblock.fs_bsize) - UFS_NDADDR;
if (rb > 0) {
/*
* Dump single indirect block.
*/
if (bread(&disk, fsbtodb(&sblock, dp.dp2->di_ib[0]),
(void *)&i1blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment), "Inode 0x%08jx: indirect 0",
(uintmax_t)inode);
DBG_DUMP_IBLK(&sblock, comment, i1blk, (size_t)rb);
rb -= howmany(sblock.fs_bsize, sizeof(ufs2_daddr_t));
}
if (rb > 0) {
/*
* Dump double indirect blocks.
*/
if (bread(&disk, fsbtodb(&sblock, dp.dp2->di_ib[1]),
(void *)&i2blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment), "Inode 0x%08jx: indirect 1",
(uintmax_t)inode);
DBG_DUMP_IBLK(&sblock,
comment,
i2blk,
howmany(rb, howmany(sblock.fs_bsize, sizeof(ufs2_daddr_t))));
for (ind2ctr = 0; ((ind2ctr < howmany(sblock.fs_bsize,
sizeof(ufs2_daddr_t))) && (rb>0)); ind2ctr++) {
ind2ptr = &((ufs2_daddr_t *)(void *)&i2blk)[ind2ctr];
if (bread(&disk, fsbtodb(&sblock, *ind2ptr),
(void *)&i1blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment),
"Inode 0x%08jx: indirect 1->%d",
(uintmax_t)inode, ind2ctr);
DBG_DUMP_IBLK(&sblock, comment, i1blk, (size_t)rb);
rb -= howmany(sblock.fs_bsize, sizeof(ufs2_daddr_t));
}
}
if (rb > 0) {
/*
* Dump triple indirect blocks.
*/
if (bread(&disk, fsbtodb(&sblock, dp.dp2->di_ib[2]),
(void *)&i3blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment), "Inode 0x%08jx: indirect 2",
(uintmax_t)inode);
#define SQUARE(a) ((a)*(a))
DBG_DUMP_IBLK(&sblock,
comment,
i3blk,
howmany(rb,
SQUARE(howmany(sblock.fs_bsize, sizeof(ufs2_daddr_t)))));
#undef SQUARE
for (ind3ctr = 0; ((ind3ctr < howmany(sblock.fs_bsize,
sizeof(ufs2_daddr_t))) && (rb > 0)); ind3ctr++) {
ind3ptr = &((ufs2_daddr_t *)(void *)&i3blk)[ind3ctr];
if (bread(&disk, fsbtodb(&sblock, *ind3ptr),
(void *)&i2blk, (size_t)sblock.fs_bsize) == -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment),
"Inode 0x%08jx: indirect 2->%d",
(uintmax_t)inode, ind3ctr);
DBG_DUMP_IBLK(&sblock,
comment,
i2blk,
howmany(rb,
howmany(sblock.fs_bsize, sizeof(ufs2_daddr_t))));
for (ind2ctr = 0; ((ind2ctr < howmany(sblock.fs_bsize,
sizeof(ufs2_daddr_t))) && (rb > 0)); ind2ctr++) {
ind2ptr = &((ufs2_daddr_t *)(void *)&i2blk) [ind2ctr];
if (bread(&disk, fsbtodb(&sblock, *ind2ptr),
(void *)&i1blk, (size_t)sblock.fs_bsize)
== -1) {
err(1, "bread: %s", disk.d_error);
}
snprintf(comment, sizeof(comment),
"Inode 0x%08jx: indirect 2->%d->%d",
(uintmax_t)inode, ind3ctr, ind3ctr);
DBG_DUMP_IBLK(&sblock, comment, i1blk, (size_t)rb);
rb -= howmany(sblock.fs_bsize, sizeof(ufs2_daddr_t));
}
}
}
DBG_LEAVE;
return;
}
/* ************************************************************* usage ***** */
/*
* Dump a line of usage.
*/
void
usage(void)
{
DBG_FUNC("usage")
DBG_ENTER;
fprintf(stderr,
"usage: ffsinfo [-g cylinder_group] [-i inode] [-l level] "
"[-o outfile]\n"
" special | file\n");
DBG_LEAVE;
exit(1);
}
diff --git a/sbin/fsck_ffs/suj.c b/sbin/fsck_ffs/suj.c
index aa2085e2fa77..4ad86b0dc14d 100644
--- a/sbin/fsck_ffs/suj.c
+++ b/sbin/fsck_ffs/suj.c
@@ -1,2454 +1,2456 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright 2009, 2010 Jeffrey W. Roberson <jeff@FreeBSD.org>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/disk.h>
#include <sys/disklabel.h>
#include <sys/mount.h>
#include <sys/stat.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ufs/dir.h>
#include <ufs/ffs/fs.h>
#include <assert.h>
#include <err.h>
#include <setjmp.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <libufs.h>
#include <string.h>
#include <strings.h>
#include <sysexits.h>
#include <time.h>
#include "fsck.h"
#define DOTDOT_OFFSET DIRECTSIZ(1)
struct suj_seg {
TAILQ_ENTRY(suj_seg) ss_next;
struct jsegrec ss_rec;
uint8_t *ss_blk;
};
struct suj_rec {
TAILQ_ENTRY(suj_rec) sr_next;
union jrec *sr_rec;
};
TAILQ_HEAD(srechd, suj_rec);
struct suj_ino {
LIST_ENTRY(suj_ino) si_next;
struct srechd si_recs;
struct srechd si_newrecs;
struct srechd si_movs;
struct jtrncrec *si_trunc;
ino_t si_ino;
char si_skipparent;
char si_hasrecs;
char si_blkadj;
char si_linkadj;
int si_mode;
nlink_t si_nlinkadj;
nlink_t si_nlink;
nlink_t si_dotlinks;
};
LIST_HEAD(inohd, suj_ino);
struct suj_blk {
LIST_ENTRY(suj_blk) sb_next;
struct srechd sb_recs;
ufs2_daddr_t sb_blk;
};
LIST_HEAD(blkhd, suj_blk);
struct suj_cg {
LIST_ENTRY(suj_cg) sc_next;
struct blkhd sc_blkhash[HASHSIZE];
struct inohd sc_inohash[HASHSIZE];
struct ino_blk *sc_lastiblk;
struct suj_ino *sc_lastino;
struct suj_blk *sc_lastblk;
struct bufarea *sc_cgbp;
struct cg *sc_cgp;
int sc_cgx;
};
static LIST_HEAD(cghd, suj_cg) cghash[HASHSIZE];
static struct suj_cg *lastcg;
static TAILQ_HEAD(seghd, suj_seg) allsegs;
static uint64_t oldseq;
static struct fs *fs = NULL;
static ino_t sujino;
/*
* Summary statistics.
*/
static uint64_t freefrags;
static uint64_t freeblocks;
static uint64_t freeinos;
static uint64_t freedir;
static uint64_t jbytes;
static uint64_t jrecs;
static jmp_buf jmpbuf;
typedef void (*ino_visitor)(ino_t, ufs_lbn_t, ufs2_daddr_t, int);
static void err_suj(const char *, ...) __dead2;
static void ino_trunc(ino_t, off_t);
static void ino_decr(ino_t);
static void ino_adjust(struct suj_ino *);
static void ino_build(struct suj_ino *);
static int blk_isfree(ufs2_daddr_t);
static void initsuj(void);
static void *
errmalloc(size_t n)
{
void *a;
a = Malloc(n);
if (a == NULL)
err(EX_OSERR, "malloc(%zu)", n);
return (a);
}
/*
* When hit a fatal error in journalling check, print out
* the error and then offer to fallback to normal fsck.
*/
static void
err_suj(const char * restrict fmt, ...)
{
va_list ap;
if (preen)
(void)fprintf(stdout, "%s: ", cdevname);
va_start(ap, fmt);
(void)vfprintf(stdout, fmt, ap);
va_end(ap);
longjmp(jmpbuf, -1);
}
/*
* Lookup a cg by number in the hash so we can keep track of which cgs
* need stats rebuilt.
*/
static struct suj_cg *
cg_lookup(int cgx)
{
struct cghd *hd;
struct suj_cg *sc;
struct bufarea *cgbp;
if (cgx < 0 || cgx >= fs->fs_ncg)
err_suj("Bad cg number %d\n", cgx);
if (lastcg && lastcg->sc_cgx == cgx)
return (lastcg);
cgbp = cglookup(cgx);
if (!check_cgmagic(cgx, cgbp, 0))
err_suj("UNABLE TO REBUILD CYLINDER GROUP %d", cgx);
hd = &cghash[HASH(cgx)];
LIST_FOREACH(sc, hd, sc_next)
if (sc->sc_cgx == cgx) {
sc->sc_cgbp = cgbp;
sc->sc_cgp = sc->sc_cgbp->b_un.b_cg;
lastcg = sc;
return (sc);
}
sc = errmalloc(sizeof(*sc));
bzero(sc, sizeof(*sc));
sc->sc_cgbp = cgbp;
sc->sc_cgp = sc->sc_cgbp->b_un.b_cg;
sc->sc_cgx = cgx;
LIST_INSERT_HEAD(hd, sc, sc_next);
return (sc);
}
/*
* Lookup an inode number in the hash and allocate a suj_ino if it does
* not exist.
*/
static struct suj_ino *
ino_lookup(ino_t ino, int creat)
{
struct suj_ino *sino;
struct inohd *hd;
struct suj_cg *sc;
sc = cg_lookup(ino_to_cg(fs, ino));
if (sc->sc_lastino && sc->sc_lastino->si_ino == ino)
return (sc->sc_lastino);
hd = &sc->sc_inohash[HASH(ino)];
LIST_FOREACH(sino, hd, si_next)
if (sino->si_ino == ino)
return (sino);
if (creat == 0)
return (NULL);
sino = errmalloc(sizeof(*sino));
bzero(sino, sizeof(*sino));
sino->si_ino = ino;
TAILQ_INIT(&sino->si_recs);
TAILQ_INIT(&sino->si_newrecs);
TAILQ_INIT(&sino->si_movs);
LIST_INSERT_HEAD(hd, sino, si_next);
return (sino);
}
/*
* Lookup a block number in the hash and allocate a suj_blk if it does
* not exist.
*/
static struct suj_blk *
blk_lookup(ufs2_daddr_t blk, int creat)
{
struct suj_blk *sblk;
struct suj_cg *sc;
struct blkhd *hd;
sc = cg_lookup(dtog(fs, blk));
if (sc->sc_lastblk && sc->sc_lastblk->sb_blk == blk)
return (sc->sc_lastblk);
hd = &sc->sc_blkhash[HASH(fragstoblks(fs, blk))];
LIST_FOREACH(sblk, hd, sb_next)
if (sblk->sb_blk == blk)
return (sblk);
if (creat == 0)
return (NULL);
sblk = errmalloc(sizeof(*sblk));
bzero(sblk, sizeof(*sblk));
sblk->sb_blk = blk;
TAILQ_INIT(&sblk->sb_recs);
LIST_INSERT_HEAD(hd, sblk, sb_next);
return (sblk);
}
static int
blk_overlaps(struct jblkrec *brec, ufs2_daddr_t start, int frags)
{
ufs2_daddr_t bstart;
ufs2_daddr_t bend;
ufs2_daddr_t end;
end = start + frags;
bstart = brec->jb_blkno + brec->jb_oldfrags;
bend = bstart + brec->jb_frags;
if (start < bend && end > bstart)
return (1);
return (0);
}
static int
blk_equals(struct jblkrec *brec, ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t start,
int frags)
{
if (brec->jb_ino != ino || brec->jb_lbn != lbn)
return (0);
if (brec->jb_blkno + brec->jb_oldfrags != start)
return (0);
if (brec->jb_frags < frags)
return (0);
return (1);
}
static void
blk_setmask(struct jblkrec *brec, int *mask)
{
int i;
for (i = brec->jb_oldfrags; i < brec->jb_oldfrags + brec->jb_frags; i++)
*mask |= 1 << i;
}
/*
* Determine whether a given block has been reallocated to a new location.
* Returns a mask of overlapping bits if any frags have been reused or
* zero if the block has not been re-used and the contents can be trusted.
*
* This is used to ensure that an orphaned pointer due to truncate is safe
* to be freed. The mask value can be used to free partial blocks.
*/
static int
blk_freemask(ufs2_daddr_t blk, ino_t ino, ufs_lbn_t lbn, int frags)
{
struct suj_blk *sblk;
struct suj_rec *srec;
struct jblkrec *brec;
int mask;
int off;
/*
* To be certain we're not freeing a reallocated block we lookup
* this block in the blk hash and see if there is an allocation
* journal record that overlaps with any fragments in the block
* we're concerned with. If any fragments have ben reallocated
* the block has already been freed and re-used for another purpose.
*/
mask = 0;
sblk = blk_lookup(blknum(fs, blk), 0);
if (sblk == NULL)
return (0);
off = blk - sblk->sb_blk;
TAILQ_FOREACH(srec, &sblk->sb_recs, sr_next) {
brec = (struct jblkrec *)srec->sr_rec;
/*
* If the block overlaps but does not match
* exactly this record refers to the current
* location.
*/
if (blk_overlaps(brec, blk, frags) == 0)
continue;
if (blk_equals(brec, ino, lbn, blk, frags) == 1)
mask = 0;
else
blk_setmask(brec, &mask);
}
if (debug)
printf("blk_freemask: blk %jd sblk %jd off %d mask 0x%X\n",
blk, sblk->sb_blk, off, mask);
return (mask >> off);
}
/*
* Determine whether it is safe to follow an indirect. It is not safe
* if any part of the indirect has been reallocated or the last journal
* entry was an allocation. Just allocated indirects may not have valid
* pointers yet and all of their children will have their own records.
* It is also not safe to follow an indirect if the cg bitmap has been
* cleared as a new allocation may write to the block prior to the journal
* being written.
*
* Returns 1 if it's safe to follow the indirect and 0 otherwise.
*/
static int
blk_isindir(ufs2_daddr_t blk, ino_t ino, ufs_lbn_t lbn)
{
struct suj_blk *sblk;
struct jblkrec *brec;
sblk = blk_lookup(blk, 0);
if (sblk == NULL)
return (1);
if (TAILQ_EMPTY(&sblk->sb_recs))
return (1);
brec = (struct jblkrec *)TAILQ_LAST(&sblk->sb_recs, srechd)->sr_rec;
if (blk_equals(brec, ino, lbn, blk, fs->fs_frag))
if (brec->jb_op == JOP_FREEBLK)
return (!blk_isfree(blk));
return (0);
}
/*
* Clear an inode from the cg bitmap. If the inode was already clear return
* 0 so the caller knows it does not have to check the inode contents.
*/
static int
ino_free(ino_t ino, int mode)
{
struct suj_cg *sc;
uint8_t *inosused;
struct cg *cgp;
int cg;
cg = ino_to_cg(fs, ino);
ino = ino % fs->fs_ipg;
sc = cg_lookup(cg);
cgp = sc->sc_cgp;
inosused = cg_inosused(cgp);
/*
* The bitmap may never have made it to the disk so we have to
* conditionally clear. We can avoid writing the cg in this case.
*/
if (isclr(inosused, ino))
return (0);
freeinos++;
clrbit(inosused, ino);
if (ino < cgp->cg_irotor)
cgp->cg_irotor = ino;
cgp->cg_cs.cs_nifree++;
if ((mode & IFMT) == IFDIR) {
freedir++;
cgp->cg_cs.cs_ndir--;
}
cgdirty(sc->sc_cgbp);
return (1);
}
/*
* Free 'frags' frags starting at filesystem block 'bno' skipping any frags
* set in the mask.
*/
static void
blk_free(ufs2_daddr_t bno, int mask, int frags)
{
ufs1_daddr_t fragno, cgbno;
struct suj_cg *sc;
struct cg *cgp;
int i, cg;
uint8_t *blksfree;
if (debug)
printf("Freeing %d frags at blk %jd mask 0x%x\n",
frags, bno, mask);
cg = dtog(fs, bno);
sc = cg_lookup(cg);
cgp = sc->sc_cgp;
cgbno = dtogd(fs, bno);
blksfree = cg_blksfree(cgp);
/*
* If it's not allocated we only wrote the journal entry
* and never the bitmaps. Here we unconditionally clear and
* resolve the cg summary later.
*/
if (frags == fs->fs_frag && mask == 0) {
fragno = fragstoblks(fs, cgbno);
ffs_setblock(fs, blksfree, fragno);
freeblocks++;
} else {
/*
* deallocate the fragment
*/
for (i = 0; i < frags; i++)
if ((mask & (1 << i)) == 0 && isclr(blksfree, cgbno +i)) {
freefrags++;
setbit(blksfree, cgbno + i);
}
}
cgdirty(sc->sc_cgbp);
}
/*
* Returns 1 if the whole block starting at 'bno' is marked free and 0
* otherwise.
*/
static int
blk_isfree(ufs2_daddr_t bno)
{
struct suj_cg *sc;
sc = cg_lookup(dtog(fs, bno));
return ffs_isblock(fs, cg_blksfree(sc->sc_cgp), dtogd(fs, bno));
}
/*
* Determine whether a block exists at a particular lbn in an inode.
* Returns 1 if found, 0 if not. lbn may be negative for indirects
* or ext blocks.
*/
static int
blk_isat(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int *frags)
{
struct inode ip;
union dinode *dp;
ufs2_daddr_t nblk;
ginode(ino, &ip);
dp = ip.i_dp;
if (DIP(dp, di_nlink) == 0 || DIP(dp, di_mode) == 0) {
irelse(&ip);
return (0);
}
nblk = ino_blkatoff(dp, ino, lbn, frags, NULL);
irelse(&ip);
return (nblk == blk);
}
/*
* Clear the directory entry at diroff that should point to child. Minimal
* checking is done and it is assumed that this path was verified with isat.
*/
static void
ino_clrat(ino_t parent, off_t diroff, ino_t child)
{
union dinode *dip;
struct direct *dp;
struct inode ip;
ufs2_daddr_t blk;
struct bufarea *bp;
ufs_lbn_t lbn;
int blksize;
int frags;
int doff;
if (debug)
printf("Clearing inode %ju from parent %ju at offset %jd\n",
(uintmax_t)child, (uintmax_t)parent, diroff);
lbn = lblkno(fs, diroff);
doff = blkoff(fs, diroff);
ginode(parent, &ip);
dip = ip.i_dp;
blk = ino_blkatoff(dip, parent, lbn, &frags, NULL);
blksize = sblksize(fs, DIP(dip, di_size), lbn);
irelse(&ip);
bp = getdatablk(blk, blksize, BT_DIRDATA);
if (bp->b_errs != 0)
err_suj("ino_clrat: UNRECOVERABLE I/O ERROR");
dp = (struct direct *)&bp->b_un.b_buf[doff];
if (dp->d_ino != child)
errx(1, "Inode %ju does not exist in %ju at %jd",
(uintmax_t)child, (uintmax_t)parent, diroff);
dp->d_ino = 0;
dirty(bp);
brelse(bp);
/*
* The actual .. reference count will already have been removed
* from the parent by the .. remref record.
*/
}
/*
* Determines whether a pointer to an inode exists within a directory
* at a specified offset. Returns the mode of the found entry.
*/
static int
ino_isat(ino_t parent, off_t diroff, ino_t child, int *mode, int *isdot)
{
struct inode ip;
union dinode *dip;
struct bufarea *bp;
struct direct *dp;
ufs2_daddr_t blk;
ufs_lbn_t lbn;
int blksize;
int frags;
int dpoff;
int doff;
*isdot = 0;
ginode(parent, &ip);
dip = ip.i_dp;
*mode = DIP(dip, di_mode);
if ((*mode & IFMT) != IFDIR) {
if (debug) {
/*
* This can happen if the parent inode
* was reallocated.
*/
if (*mode != 0)
printf("Directory %ju has bad mode %o\n",
(uintmax_t)parent, *mode);
else
printf("Directory %ju has zero mode\n",
(uintmax_t)parent);
}
irelse(&ip);
return (0);
}
lbn = lblkno(fs, diroff);
doff = blkoff(fs, diroff);
blksize = sblksize(fs, DIP(dip, di_size), lbn);
if (diroff + DIRECTSIZ(1) > DIP(dip, di_size) || doff >= blksize) {
if (debug)
printf("ino %ju absent from %ju due to offset %jd"
" exceeding size %jd\n",
(uintmax_t)child, (uintmax_t)parent, diroff,
DIP(dip, di_size));
irelse(&ip);
return (0);
}
blk = ino_blkatoff(dip, parent, lbn, &frags, NULL);
irelse(&ip);
if (blk <= 0) {
if (debug)
printf("Sparse directory %ju", (uintmax_t)parent);
return (0);
}
bp = getdatablk(blk, blksize, BT_DIRDATA);
if (bp->b_errs != 0)
err_suj("ino_isat: UNRECOVERABLE I/O ERROR");
/*
* Walk through the records from the start of the block to be
* certain we hit a valid record and not some junk in the middle
* of a file name. Stop when we reach or pass the expected offset.
*/
dpoff = rounddown(doff, DIRBLKSIZ);
do {
dp = (struct direct *)&bp->b_un.b_buf[dpoff];
if (dpoff == doff)
break;
if (dp->d_reclen == 0)
break;
dpoff += dp->d_reclen;
} while (dpoff <= doff);
if (dpoff > fs->fs_bsize)
err_suj("Corrupt directory block in dir ino %ju\n",
(uintmax_t)parent);
/* Not found. */
if (dpoff != doff) {
if (debug)
printf("ino %ju not found in %ju, lbn %jd, dpoff %d\n",
(uintmax_t)child, (uintmax_t)parent, lbn, dpoff);
brelse(bp);
return (0);
}
/*
* We found the item in question. Record the mode and whether it's
* a . or .. link for the caller.
*/
if (dp->d_ino == child) {
if (child == parent)
*isdot = 1;
else if (dp->d_namlen == 2 &&
dp->d_name[0] == '.' && dp->d_name[1] == '.')
*isdot = 1;
*mode = DTTOIF(dp->d_type);
brelse(bp);
return (1);
}
if (debug)
printf("ino %ju doesn't match dirent ino %ju in parent %ju\n",
(uintmax_t)child, (uintmax_t)dp->d_ino, (uintmax_t)parent);
brelse(bp);
return (0);
}
#define VISIT_INDIR 0x0001
#define VISIT_EXT 0x0002
#define VISIT_ROOT 0x0004 /* Operation came via root & valid pointers. */
/*
* Read an indirect level which may or may not be linked into an inode.
*/
static void
indir_visit(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, uint64_t *frags,
ino_visitor visitor, int flags)
{
struct bufarea *bp;
ufs_lbn_t lbnadd;
ufs2_daddr_t nblk;
ufs_lbn_t nlbn;
int level;
int i;
/*
* Don't visit indirect blocks with contents we can't trust. This
* should only happen when indir_visit() is called to complete a
* truncate that never finished and not when a pointer is found via
* an inode.
*/
if (blk == 0)
return;
level = lbn_level(lbn);
if (level == -1)
err_suj("Invalid level for lbn %jd\n", lbn);
if ((flags & VISIT_ROOT) == 0 && blk_isindir(blk, ino, lbn) == 0) {
if (debug)
printf("blk %jd ino %ju lbn %jd(%d) is not indir.\n",
blk, (uintmax_t)ino, lbn, level);
goto out;
}
lbnadd = 1;
for (i = level; i > 0; i--)
lbnadd *= NINDIR(fs);
bp = getdatablk(blk, fs->fs_bsize, BT_LEVEL1 + level);
if (bp->b_errs != 0)
err_suj("indir_visit: UNRECOVERABLE I/O ERROR");
for (i = 0; i < NINDIR(fs); i++) {
if ((nblk = IBLK(bp, i)) == 0)
continue;
if (level == 0) {
nlbn = -lbn + i * lbnadd;
(*frags) += fs->fs_frag;
visitor(ino, nlbn, nblk, fs->fs_frag);
} else {
nlbn = (lbn + 1) - (i * lbnadd);
indir_visit(ino, nlbn, nblk, frags, visitor, flags);
}
}
brelse(bp);
out:
if (flags & VISIT_INDIR) {
(*frags) += fs->fs_frag;
visitor(ino, lbn, blk, fs->fs_frag);
}
}
/*
* Visit each block in an inode as specified by 'flags' and call a
* callback function. The callback may inspect or free blocks. The
* count of frags found according to the size in the file is returned.
* This is not valid for sparse files but may be used to determine
* the correct di_blocks for a file.
*/
static uint64_t
ino_visit(union dinode *dp, ino_t ino, ino_visitor visitor, int flags)
{
ufs_lbn_t nextlbn;
ufs_lbn_t tmpval;
ufs_lbn_t lbn;
uint64_t size;
uint64_t fragcnt;
int mode;
int frags;
int i;
size = DIP(dp, di_size);
mode = DIP(dp, di_mode) & IFMT;
fragcnt = 0;
if ((flags & VISIT_EXT) &&
fs->fs_magic == FS_UFS2_MAGIC && dp->dp2.di_extsize) {
for (i = 0; i < UFS_NXADDR; i++) {
if (dp->dp2.di_extb[i] == 0)
continue;
frags = sblksize(fs, dp->dp2.di_extsize, i);
frags = numfrags(fs, frags);
fragcnt += frags;
visitor(ino, -1 - i, dp->dp2.di_extb[i], frags);
}
}
/* Skip datablocks for short links and devices. */
if (mode == IFBLK || mode == IFCHR ||
(mode == IFLNK && size < fs->fs_maxsymlinklen))
return (fragcnt);
for (i = 0; i < UFS_NDADDR; i++) {
if (DIP(dp, di_db[i]) == 0)
continue;
frags = sblksize(fs, size, i);
frags = numfrags(fs, frags);
fragcnt += frags;
visitor(ino, i, DIP(dp, di_db[i]), frags);
}
/*
* We know the following indirects are real as we're following
* real pointers to them.
*/
flags |= VISIT_ROOT;
for (i = 0, tmpval = NINDIR(fs), lbn = UFS_NDADDR; i < UFS_NIADDR; i++,
lbn = nextlbn) {
nextlbn = lbn + tmpval;
tmpval *= NINDIR(fs);
if (DIP(dp, di_ib[i]) == 0)
continue;
indir_visit(ino, -lbn - i, DIP(dp, di_ib[i]), &fragcnt, visitor,
flags);
}
return (fragcnt);
}
/*
* Null visitor function used when we just want to count blocks and
* record the lbn.
*/
ufs_lbn_t visitlbn;
static void
null_visit(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{
if (lbn > 0)
visitlbn = lbn;
}
/*
* Recalculate di_blocks when we discover that a block allocation or
* free was not successfully completed. The kernel does not roll this back
* because it would be too expensive to compute which indirects were
* reachable at the time the inode was written.
*/
static void
ino_adjblks(struct suj_ino *sino)
{
struct inode ip;
union dinode *dp;
uint64_t blocks;
uint64_t frags;
off_t isize;
off_t size;
ino_t ino;
ino = sino->si_ino;
ginode(ino, &ip);
dp = ip.i_dp;
/* No need to adjust zero'd inodes. */
if (DIP(dp, di_mode) == 0) {
irelse(&ip);
return;
}
/*
* Visit all blocks and count them as well as recording the last
* valid lbn in the file. If the file size doesn't agree with the
* last lbn we need to truncate to fix it. Otherwise just adjust
* the blocks count.
*/
visitlbn = 0;
frags = ino_visit(dp, ino, null_visit, VISIT_INDIR | VISIT_EXT);
blocks = fsbtodb(fs, frags);
/*
* We assume the size and direct block list is kept coherent by
* softdep. For files that have extended into indirects we truncate
* to the size in the inode or the maximum size permitted by
* populated indirects.
*/
if (visitlbn >= UFS_NDADDR) {
isize = DIP(dp, di_size);
size = lblktosize(fs, visitlbn + 1);
if (isize > size)
isize = size;
/* Always truncate to free any unpopulated indirects. */
ino_trunc(ino, isize);
irelse(&ip);
return;
}
if (blocks == DIP(dp, di_blocks)) {
irelse(&ip);
return;
}
if (debug)
printf("ino %ju adjusting block count from %jd to %jd\n",
(uintmax_t)ino, DIP(dp, di_blocks), blocks);
DIP_SET(dp, di_blocks, blocks);
inodirty(&ip);
irelse(&ip);
}
static void
blk_free_visit(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{
blk_free(blk, blk_freemask(blk, ino, lbn, frags), frags);
}
/*
* Free a block or tree of blocks that was previously rooted in ino at
* the given lbn. If the lbn is an indirect all children are freed
* recursively.
*/
static void
blk_free_lbn(ufs2_daddr_t blk, ino_t ino, ufs_lbn_t lbn, int frags, int follow)
{
uint64_t resid;
int mask;
mask = blk_freemask(blk, ino, lbn, frags);
resid = 0;
if (lbn <= -UFS_NDADDR && follow && mask == 0)
indir_visit(ino, lbn, blk, &resid, blk_free_visit, VISIT_INDIR);
else
blk_free(blk, mask, frags);
}
static void
ino_setskip(struct suj_ino *sino, ino_t parent)
{
int isdot;
int mode;
if (ino_isat(sino->si_ino, DOTDOT_OFFSET, parent, &mode, &isdot))
sino->si_skipparent = 1;
}
static void
ino_remref(ino_t parent, ino_t child, uint64_t diroff, int isdotdot)
{
struct suj_ino *sino;
struct suj_rec *srec;
struct jrefrec *rrec;
/*
* Lookup this inode to see if we have a record for it.
*/
sino = ino_lookup(child, 0);
/*
* Tell any child directories we've already removed their
* parent link cnt. Don't try to adjust our link down again.
*/
if (sino != NULL && isdotdot == 0)
ino_setskip(sino, parent);
/*
* No valid record for this inode. Just drop the on-disk
* link by one.
*/
if (sino == NULL || sino->si_hasrecs == 0) {
ino_decr(child);
return;
}
/*
* Use ino_adjust() if ino_check() has already processed this
* child. If we lose the last non-dot reference to a
* directory it will be discarded.
*/
if (sino->si_linkadj) {
if (sino->si_nlink == 0)
err_suj("ino_remref: ino %ld mode 0%o about to go "
"negative\n", sino->si_ino, sino->si_mode);
sino->si_nlink--;
if (isdotdot)
sino->si_dotlinks--;
ino_adjust(sino);
return;
}
/*
* If we haven't yet processed this inode we need to make
* sure we will successfully discover the lost path. If not
* use nlinkadj to remember.
*/
TAILQ_FOREACH(srec, &sino->si_recs, sr_next) {
rrec = (struct jrefrec *)srec->sr_rec;
if (rrec->jr_parent == parent &&
rrec->jr_diroff == diroff)
return;
}
sino->si_nlinkadj++;
}
/*
* Free the children of a directory when the directory is discarded.
*/
static void
ino_free_children(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{
struct suj_ino *sino;
struct bufarea *bp;
struct direct *dp;
off_t diroff;
int skipparent;
int isdotdot;
int dpoff;
int size;
sino = ino_lookup(ino, 0);
if (sino)
skipparent = sino->si_skipparent;
else
skipparent = 0;
size = lfragtosize(fs, frags);
bp = getdatablk(blk, size, BT_DIRDATA);
if (bp->b_errs != 0)
err_suj("ino_free_children: UNRECOVERABLE I/O ERROR");
dp = (struct direct *)&bp->b_un.b_buf[0];
for (dpoff = 0; dpoff < size && dp->d_reclen; dpoff += dp->d_reclen) {
dp = (struct direct *)&bp->b_un.b_buf[dpoff];
if (dp->d_ino == 0 || dp->d_ino == UFS_WINO)
continue;
if (dp->d_namlen == 1 && dp->d_name[0] == '.')
continue;
isdotdot = dp->d_namlen == 2 && dp->d_name[0] == '.' &&
dp->d_name[1] == '.';
if (isdotdot && skipparent == 1)
continue;
if (debug)
printf("Directory %ju removing ino %ju name %s\n",
(uintmax_t)ino, (uintmax_t)dp->d_ino, dp->d_name);
diroff = lblktosize(fs, lbn) + dpoff;
ino_remref(ino, dp->d_ino, diroff, isdotdot);
}
brelse(bp);
}
/*
* Reclaim an inode, freeing all blocks and decrementing all children's
* link counts. Free the inode back to the cg.
*/
static void
ino_reclaim(struct inode *ip, ino_t ino, int mode)
{
union dinode *dp;
uint32_t gen;
dp = ip->i_dp;
if (ino == UFS_ROOTINO)
err_suj("Attempting to free UFS_ROOTINO\n");
if (debug)
printf("Truncating and freeing ino %ju, nlink %d, mode %o\n",
(uintmax_t)ino, DIP(dp, di_nlink), DIP(dp, di_mode));
/* We are freeing an inode or directory. */
if ((DIP(dp, di_mode) & IFMT) == IFDIR)
ino_visit(dp, ino, ino_free_children, 0);
DIP_SET(dp, di_nlink, 0);
ino_visit(dp, ino, blk_free_visit, VISIT_EXT | VISIT_INDIR);
/* Here we have to clear the inode and release any blocks it holds. */
gen = DIP(dp, di_gen);
if (fs->fs_magic == FS_UFS1_MAGIC)
bzero(dp, sizeof(struct ufs1_dinode));
else
bzero(dp, sizeof(struct ufs2_dinode));
DIP_SET(dp, di_gen, gen);
inodirty(ip);
ino_free(ino, mode);
return;
}
/*
* Adjust an inode's link count down by one when a directory goes away.
*/
static void
ino_decr(ino_t ino)
{
struct inode ip;
union dinode *dp;
int reqlink;
int nlink;
int mode;
ginode(ino, &ip);
dp = ip.i_dp;
nlink = DIP(dp, di_nlink);
mode = DIP(dp, di_mode);
if (nlink < 1)
err_suj("Inode %d link count %d invalid\n", ino, nlink);
if (mode == 0)
err_suj("Inode %d has a link of %d with 0 mode\n", ino, nlink);
nlink--;
if ((mode & IFMT) == IFDIR)
reqlink = 2;
else
reqlink = 1;
if (nlink < reqlink) {
if (debug)
printf("ino %ju not enough links to live %d < %d\n",
(uintmax_t)ino, nlink, reqlink);
ino_reclaim(&ip, ino, mode);
irelse(&ip);
return;
}
DIP_SET(dp, di_nlink, nlink);
inodirty(&ip);
irelse(&ip);
}
/*
* Adjust the inode link count to 'nlink'. If the count reaches zero
* free it.
*/
static void
ino_adjust(struct suj_ino *sino)
{
struct jrefrec *rrec;
struct suj_rec *srec;
struct suj_ino *stmp;
union dinode *dp;
struct inode ip;
nlink_t nlink;
nlink_t reqlink;
int recmode;
int isdot;
int mode;
ino_t ino;
nlink = sino->si_nlink;
ino = sino->si_ino;
mode = sino->si_mode & IFMT;
/*
* If it's a directory with no dot links, it was truncated before
* the name was cleared. We need to clear the dirent that
* points at it.
*/
if (mode == IFDIR && nlink == 1 && sino->si_dotlinks == 0) {
sino->si_nlink = nlink = 0;
TAILQ_FOREACH(srec, &sino->si_recs, sr_next) {
rrec = (struct jrefrec *)srec->sr_rec;
if (ino_isat(rrec->jr_parent, rrec->jr_diroff, ino,
&recmode, &isdot) == 0)
continue;
ino_clrat(rrec->jr_parent, rrec->jr_diroff, ino);
break;
}
if (srec == NULL)
errx(1, "Directory %ju name not found", (uintmax_t)ino);
}
/*
* If it's a directory with no real names pointing to it go ahead
* and truncate it. This will free any children.
*/
if (mode == IFDIR && nlink - sino->si_dotlinks == 0) {
sino->si_nlink = nlink = 0;
/*
* Mark any .. links so they know not to free this inode
* when they are removed.
*/
TAILQ_FOREACH(srec, &sino->si_recs, sr_next) {
rrec = (struct jrefrec *)srec->sr_rec;
if (rrec->jr_diroff == DOTDOT_OFFSET) {
stmp = ino_lookup(rrec->jr_parent, 0);
if (stmp)
ino_setskip(stmp, ino);
}
}
}
ginode(ino, &ip);
dp = ip.i_dp;
mode = DIP(dp, di_mode) & IFMT;
if (nlink > UFS_LINK_MAX)
err_suj("ino %ju nlink manipulation error, new %ju, old %d\n",
(uintmax_t)ino, (uintmax_t)nlink, DIP(dp, di_nlink));
if (debug)
printf("Adjusting ino %ju, nlink %ju, old link %d lastmode %o\n",
(uintmax_t)ino, (uintmax_t)nlink, DIP(dp, di_nlink),
sino->si_mode);
if (mode == 0) {
if (debug)
printf("ino %ju, zero inode freeing bitmap\n",
(uintmax_t)ino);
ino_free(ino, sino->si_mode);
irelse(&ip);
return;
}
/* XXX Should be an assert? */
if (mode != sino->si_mode && debug)
printf("ino %ju, mode %o != %o\n",
(uintmax_t)ino, mode, sino->si_mode);
if ((mode & IFMT) == IFDIR)
reqlink = 2;
else
reqlink = 1;
/* If the inode doesn't have enough links to live, free it. */
if (nlink < reqlink) {
if (debug)
printf("ino %ju not enough links to live %ju < %ju\n",
(uintmax_t)ino, (uintmax_t)nlink,
(uintmax_t)reqlink);
ino_reclaim(&ip, ino, mode);
irelse(&ip);
return;
}
/* If required write the updated link count. */
if (DIP(dp, di_nlink) == nlink) {
if (debug)
printf("ino %ju, link matches, skipping.\n",
(uintmax_t)ino);
irelse(&ip);
return;
}
DIP_SET(dp, di_nlink, nlink);
inodirty(&ip);
irelse(&ip);
}
/*
* Truncate some or all blocks in an indirect, freeing any that are required
* and zeroing the indirect.
*/
static void
indir_trunc(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, ufs_lbn_t lastlbn,
union dinode *dp)
{
struct bufarea *bp;
ufs_lbn_t lbnadd;
ufs2_daddr_t nblk;
ufs_lbn_t next;
ufs_lbn_t nlbn;
int isdirty;
int level;
int i;
if (blk == 0)
return;
isdirty = 0;
level = lbn_level(lbn);
if (level == -1)
err_suj("Invalid level for lbn %jd\n", lbn);
lbnadd = 1;
for (i = level; i > 0; i--)
lbnadd *= NINDIR(fs);
bp = getdatablk(blk, fs->fs_bsize, BT_LEVEL1 + level);
if (bp->b_errs != 0)
err_suj("indir_trunc: UNRECOVERABLE I/O ERROR");
for (i = 0; i < NINDIR(fs); i++) {
if ((nblk = IBLK(bp, i)) == 0)
continue;
if (level != 0) {
nlbn = (lbn + 1) - (i * lbnadd);
/*
* Calculate the lbn of the next indirect to
* determine if any of this indirect must be
* reclaimed.
*/
next = -(lbn + level) + ((i+1) * lbnadd);
if (next <= lastlbn)
continue;
indir_trunc(ino, nlbn, nblk, lastlbn, dp);
/* If all of this indirect was reclaimed, free it. */
nlbn = next - lbnadd;
if (nlbn < lastlbn)
continue;
} else {
nlbn = -lbn + i * lbnadd;
if (nlbn < lastlbn)
continue;
}
isdirty = 1;
blk_free(nblk, 0, fs->fs_frag);
IBLK_SET(bp, i, 0);
}
if (isdirty)
dirty(bp);
brelse(bp);
}
/*
* Truncate an inode to the minimum of the given size or the last populated
* block after any over size have been discarded. The kernel would allocate
* the last block in the file but fsck does not and neither do we. This
* code never extends files, only shrinks them.
*/
static void
ino_trunc(ino_t ino, off_t size)
{
struct inode ip;
union dinode *dp;
struct bufarea *bp;
ufs2_daddr_t bn;
uint64_t totalfrags;
ufs_lbn_t nextlbn;
ufs_lbn_t lastlbn;
ufs_lbn_t tmpval;
ufs_lbn_t lbn;
ufs_lbn_t i;
int blksize, frags;
off_t cursize;
off_t off;
int mode;
ginode(ino, &ip);
dp = ip.i_dp;
mode = DIP(dp, di_mode) & IFMT;
cursize = DIP(dp, di_size);
if (debug)
printf("Truncating ino %ju, mode %o to size %jd from size %jd\n",
(uintmax_t)ino, mode, size, cursize);
/* Skip datablocks for short links and devices. */
if (mode == 0 || mode == IFBLK || mode == IFCHR ||
(mode == IFLNK && cursize < fs->fs_maxsymlinklen)) {
irelse(&ip);
return;
}
/* Don't extend. */
if (size > cursize) {
irelse(&ip);
return;
}
if ((DIP(dp, di_flags) & SF_SNAPSHOT) != 0) {
if (size > 0)
err_suj("Partial truncation of ino %ju snapshot file\n",
(uintmax_t)ino);
}
lastlbn = lblkno(fs, blkroundup(fs, size));
for (i = lastlbn; i < UFS_NDADDR; i++) {
if ((bn = DIP(dp, di_db[i])) == 0)
continue;
blksize = sblksize(fs, cursize, i);
blk_free(bn, 0, numfrags(fs, blksize));
DIP_SET(dp, di_db[i], 0);
}
/*
* Follow indirect blocks, freeing anything required.
*/
for (i = 0, tmpval = NINDIR(fs), lbn = UFS_NDADDR; i < UFS_NIADDR; i++,
lbn = nextlbn) {
nextlbn = lbn + tmpval;
tmpval *= NINDIR(fs);
/* If we're not freeing any in this indirect range skip it. */
if (lastlbn >= nextlbn)
continue;
if (DIP(dp, di_ib[i]) == 0)
continue;
indir_trunc(ino, -lbn - i, DIP(dp, di_ib[i]), lastlbn, dp);
/* If we freed everything in this indirect free the indir. */
if (lastlbn > lbn)
continue;
blk_free(DIP(dp, di_ib[i]), 0, fs->fs_frag);
DIP_SET(dp, di_ib[i], 0);
}
/*
* Now that we've freed any whole blocks that exceed the desired
* truncation size, figure out how many blocks remain and what the
* last populated lbn is. We will set the size to this last lbn
* rather than worrying about allocating the final lbn as the kernel
* would've done. This is consistent with normal fsck behavior.
*/
visitlbn = 0;
totalfrags = ino_visit(dp, ino, null_visit, VISIT_INDIR | VISIT_EXT);
if (size > lblktosize(fs, visitlbn + 1))
size = lblktosize(fs, visitlbn + 1);
/*
* If we're truncating direct blocks we have to adjust frags
* accordingly.
*/
if (visitlbn < UFS_NDADDR && totalfrags) {
long oldspace, newspace;
bn = DIP(dp, di_db[visitlbn]);
if (bn == 0)
err_suj("Bad blk at ino %ju lbn %jd\n",
(uintmax_t)ino, visitlbn);
oldspace = sblksize(fs, cursize, visitlbn);
newspace = sblksize(fs, size, visitlbn);
if (oldspace != newspace) {
bn += numfrags(fs, newspace);
frags = numfrags(fs, oldspace - newspace);
blk_free(bn, 0, frags);
totalfrags -= frags;
}
}
DIP_SET(dp, di_blocks, fsbtodb(fs, totalfrags));
DIP_SET(dp, di_size, size);
inodirty(&ip);
/*
* If we've truncated into the middle of a block or frag we have
* to zero it here. Otherwise the file could extend into
* uninitialized space later.
*/
off = blkoff(fs, size);
if (off && DIP(dp, di_mode) != IFDIR) {
long clrsize;
bn = ino_blkatoff(dp, ino, visitlbn, &frags, NULL);
if (bn == 0)
err_suj("Block missing from ino %ju at lbn %jd\n",
(uintmax_t)ino, visitlbn);
clrsize = frags * fs->fs_fsize;
bp = getdatablk(bn, clrsize, BT_DATA);
if (bp->b_errs != 0)
err_suj("ino_trunc: UNRECOVERABLE I/O ERROR");
clrsize -= off;
bzero(&bp->b_un.b_buf[off], clrsize);
dirty(bp);
brelse(bp);
}
irelse(&ip);
return;
}
/*
* Process records available for one inode and determine whether the
* link count is correct or needs adjusting.
*/
static void
ino_check(struct suj_ino *sino)
{
struct suj_rec *srec;
struct jrefrec *rrec;
nlink_t dotlinks;
nlink_t newlinks;
nlink_t removes;
nlink_t nlink;
ino_t ino;
int isdot;
int isat;
int mode;
if (sino->si_hasrecs == 0)
return;
ino = sino->si_ino;
rrec = (struct jrefrec *)TAILQ_FIRST(&sino->si_recs)->sr_rec;
nlink = rrec->jr_nlink;
newlinks = 0;
dotlinks = 0;
removes = sino->si_nlinkadj;
TAILQ_FOREACH(srec, &sino->si_recs, sr_next) {
rrec = (struct jrefrec *)srec->sr_rec;
isat = ino_isat(rrec->jr_parent, rrec->jr_diroff,
rrec->jr_ino, &mode, &isdot);
if (isat && (mode & IFMT) != (rrec->jr_mode & IFMT))
err_suj("Inode mode/directory type mismatch %o != %o\n",
mode, rrec->jr_mode);
if (debug)
printf("jrefrec: op %d ino %ju, nlink %ju, parent %ju, "
"diroff %jd, mode %o, isat %d, isdot %d\n",
rrec->jr_op, (uintmax_t)rrec->jr_ino,
(uintmax_t)rrec->jr_nlink,
(uintmax_t)rrec->jr_parent,
(uintmax_t)rrec->jr_diroff,
rrec->jr_mode, isat, isdot);
mode = rrec->jr_mode & IFMT;
if (rrec->jr_op == JOP_REMREF)
removes++;
newlinks += isat;
if (isdot)
dotlinks += isat;
}
/*
* The number of links that remain are the starting link count
* subtracted by the total number of removes with the total
* links discovered back in. An incomplete remove thus
* makes no change to the link count but an add increases
* by one.
*/
if (debug)
printf(
"ino %ju nlink %ju newlinks %ju removes %ju dotlinks %ju\n",
(uintmax_t)ino, (uintmax_t)nlink, (uintmax_t)newlinks,
(uintmax_t)removes, (uintmax_t)dotlinks);
nlink += newlinks;
nlink -= removes;
sino->si_linkadj = 1;
sino->si_nlink = nlink;
sino->si_dotlinks = dotlinks;
sino->si_mode = mode;
ino_adjust(sino);
}
/*
* Process records available for one block and determine whether it is
* still allocated and whether the owning inode needs to be updated or
* a free completed.
*/
static void
blk_check(struct suj_blk *sblk)
{
struct suj_rec *srec;
struct jblkrec *brec;
struct suj_ino *sino;
ufs2_daddr_t blk;
int mask;
int frags;
int isat;
/*
* Each suj_blk actually contains records for any fragments in that
* block. As a result we must evaluate each record individually.
*/
sino = NULL;
TAILQ_FOREACH(srec, &sblk->sb_recs, sr_next) {
brec = (struct jblkrec *)srec->sr_rec;
frags = brec->jb_frags;
blk = brec->jb_blkno + brec->jb_oldfrags;
isat = blk_isat(brec->jb_ino, brec->jb_lbn, blk, &frags);
if (sino == NULL || sino->si_ino != brec->jb_ino) {
sino = ino_lookup(brec->jb_ino, 1);
sino->si_blkadj = 1;
}
if (debug)
printf("op %d blk %jd ino %ju lbn %jd frags %d isat %d (%d)\n",
brec->jb_op, blk, (uintmax_t)brec->jb_ino,
brec->jb_lbn, brec->jb_frags, isat, frags);
/*
* If we found the block at this address we still have to
* determine if we need to free the tail end that was
* added by adding contiguous fragments from the same block.
*/
if (isat == 1) {
if (frags == brec->jb_frags)
continue;
mask = blk_freemask(blk, brec->jb_ino, brec->jb_lbn,
brec->jb_frags);
mask >>= frags;
blk += frags;
frags = brec->jb_frags - frags;
blk_free(blk, mask, frags);
continue;
}
/*
* The block wasn't found, attempt to free it. It won't be
* freed if it was actually reallocated. If this was an
* allocation we don't want to follow indirects as they
* may not be written yet. Any children of the indirect will
* have their own records. If it's a free we need to
* recursively free children.
*/
blk_free_lbn(blk, brec->jb_ino, brec->jb_lbn, brec->jb_frags,
brec->jb_op == JOP_FREEBLK);
}
}
/*
* Walk the list of inode records for this cg and resolve moved and duplicate
* inode references now that we have a complete picture.
*/
static void
cg_build(struct suj_cg *sc)
{
struct suj_ino *sino;
int i;
for (i = 0; i < HASHSIZE; i++)
LIST_FOREACH(sino, &sc->sc_inohash[i], si_next)
ino_build(sino);
}
/*
* Handle inodes requiring truncation. This must be done prior to
* looking up any inodes in directories.
*/
static void
cg_trunc(struct suj_cg *sc)
{
struct suj_ino *sino;
int i;
for (i = 0; i < HASHSIZE; i++) {
LIST_FOREACH(sino, &sc->sc_inohash[i], si_next) {
if (sino->si_trunc) {
ino_trunc(sino->si_ino,
sino->si_trunc->jt_size);
sino->si_blkadj = 0;
sino->si_trunc = NULL;
}
if (sino->si_blkadj)
ino_adjblks(sino);
}
}
}
static void
cg_adj_blk(struct suj_cg *sc)
{
struct suj_ino *sino;
int i;
for (i = 0; i < HASHSIZE; i++) {
LIST_FOREACH(sino, &sc->sc_inohash[i], si_next) {
if (sino->si_blkadj)
ino_adjblks(sino);
}
}
}
/*
* Free any partially allocated blocks and then resolve inode block
* counts.
*/
static void
cg_check_blk(struct suj_cg *sc)
{
struct suj_blk *sblk;
int i;
for (i = 0; i < HASHSIZE; i++)
LIST_FOREACH(sblk, &sc->sc_blkhash[i], sb_next)
blk_check(sblk);
}
/*
* Walk the list of inode records for this cg, recovering any
* changes which were not complete at the time of crash.
*/
static void
cg_check_ino(struct suj_cg *sc)
{
struct suj_ino *sino;
int i;
for (i = 0; i < HASHSIZE; i++)
LIST_FOREACH(sino, &sc->sc_inohash[i], si_next)
ino_check(sino);
}
static void
cg_apply(void (*apply)(struct suj_cg *))
{
struct suj_cg *scg;
int i;
for (i = 0; i < HASHSIZE; i++)
LIST_FOREACH(scg, &cghash[i], sc_next)
apply(scg);
}
/*
* Process the unlinked but referenced file list. Freeing all inodes.
*/
static void
ino_unlinked(void)
{
struct inode ip;
union dinode *dp;
uint16_t mode;
ino_t inon;
ino_t ino;
ino = fs->fs_sujfree;
fs->fs_sujfree = 0;
while (ino != 0) {
ginode(ino, &ip);
dp = ip.i_dp;
mode = DIP(dp, di_mode) & IFMT;
inon = DIP(dp, di_freelink);
DIP_SET(dp, di_freelink, 0);
inodirty(&ip);
/*
* XXX Should this be an errx?
*/
if (DIP(dp, di_nlink) == 0) {
if (debug)
printf("Freeing unlinked ino %ju mode %o\n",
(uintmax_t)ino, mode);
ino_reclaim(&ip, ino, mode);
} else if (debug)
printf("Skipping ino %ju mode %o with link %d\n",
(uintmax_t)ino, mode, DIP(dp, di_nlink));
ino = inon;
irelse(&ip);
}
}
/*
* Append a new record to the list of records requiring processing.
*/
static void
ino_append(union jrec *rec)
{
struct jrefrec *refrec;
struct jmvrec *mvrec;
struct suj_ino *sino;
struct suj_rec *srec;
mvrec = &rec->rec_jmvrec;
refrec = &rec->rec_jrefrec;
if (debug && mvrec->jm_op == JOP_MVREF)
printf("ino move: ino %ju, parent %ju, "
"diroff %jd, oldoff %jd\n",
(uintmax_t)mvrec->jm_ino, (uintmax_t)mvrec->jm_parent,
(uintmax_t)mvrec->jm_newoff, (uintmax_t)mvrec->jm_oldoff);
else if (debug &&
(refrec->jr_op == JOP_ADDREF || refrec->jr_op == JOP_REMREF))
printf("ino ref: op %d, ino %ju, nlink %ju, "
"parent %ju, diroff %jd\n",
refrec->jr_op, (uintmax_t)refrec->jr_ino,
(uintmax_t)refrec->jr_nlink,
(uintmax_t)refrec->jr_parent, (uintmax_t)refrec->jr_diroff);
sino = ino_lookup(((struct jrefrec *)rec)->jr_ino, 1);
sino->si_hasrecs = 1;
srec = errmalloc(sizeof(*srec));
srec->sr_rec = rec;
TAILQ_INSERT_TAIL(&sino->si_newrecs, srec, sr_next);
}
/*
* Add a reference adjustment to the sino list and eliminate dups. The
* primary loop in ino_build_ref() checks for dups but new ones may be
* created as a result of offset adjustments.
*/
static void
ino_add_ref(struct suj_ino *sino, struct suj_rec *srec)
{
struct jrefrec *refrec;
struct suj_rec *srn;
struct jrefrec *rrn;
refrec = (struct jrefrec *)srec->sr_rec;
/*
* We walk backwards so that the oldest link count is preserved. If
* an add record conflicts with a remove keep the remove. Redundant
* removes are eliminated in ino_build_ref. Otherwise we keep the
* oldest record at a given location.
*/
for (srn = TAILQ_LAST(&sino->si_recs, srechd); srn;
srn = TAILQ_PREV(srn, srechd, sr_next)) {
rrn = (struct jrefrec *)srn->sr_rec;
if (rrn->jr_parent != refrec->jr_parent ||
rrn->jr_diroff != refrec->jr_diroff)
continue;
if (rrn->jr_op == JOP_REMREF || refrec->jr_op == JOP_ADDREF) {
rrn->jr_mode = refrec->jr_mode;
return;
}
/*
* Adding a remove.
*
* Replace the record in place with the old nlink in case
* we replace the head of the list. Abandon srec as a dup.
*/
refrec->jr_nlink = rrn->jr_nlink;
srn->sr_rec = srec->sr_rec;
return;
}
TAILQ_INSERT_TAIL(&sino->si_recs, srec, sr_next);
}
/*
* Create a duplicate of a reference at a previous location.
*/
static void
ino_dup_ref(struct suj_ino *sino, struct jrefrec *refrec, off_t diroff)
{
struct jrefrec *rrn;
struct suj_rec *srn;
rrn = errmalloc(sizeof(*refrec));
*rrn = *refrec;
rrn->jr_op = JOP_ADDREF;
rrn->jr_diroff = diroff;
srn = errmalloc(sizeof(*srn));
srn->sr_rec = (union jrec *)rrn;
ino_add_ref(sino, srn);
}
/*
* Add a reference to the list at all known locations. We follow the offset
* changes for a single instance and create duplicate add refs at each so
* that we can tolerate any version of the directory block. Eliminate
* removes which collide with adds that are seen in the journal. They should
* not adjust the link count down.
*/
static void
ino_build_ref(struct suj_ino *sino, struct suj_rec *srec)
{
struct jrefrec *refrec;
struct jmvrec *mvrec;
struct suj_rec *srp;
struct suj_rec *srn;
struct jrefrec *rrn;
off_t diroff;
refrec = (struct jrefrec *)srec->sr_rec;
/*
* Search for a mvrec that matches this offset. Whether it's an add
* or a remove we can delete the mvref after creating a dup record in
* the old location.
*/
if (!TAILQ_EMPTY(&sino->si_movs)) {
diroff = refrec->jr_diroff;
for (srn = TAILQ_LAST(&sino->si_movs, srechd); srn; srn = srp) {
srp = TAILQ_PREV(srn, srechd, sr_next);
mvrec = (struct jmvrec *)srn->sr_rec;
if (mvrec->jm_parent != refrec->jr_parent ||
mvrec->jm_newoff != diroff)
continue;
diroff = mvrec->jm_oldoff;
TAILQ_REMOVE(&sino->si_movs, srn, sr_next);
free(srn);
ino_dup_ref(sino, refrec, diroff);
}
}
/*
* If a remove wasn't eliminated by an earlier add just append it to
* the list.
*/
if (refrec->jr_op == JOP_REMREF) {
ino_add_ref(sino, srec);
return;
}
/*
* Walk the list of records waiting to be added to the list. We
* must check for moves that apply to our current offset and remove
* them from the list. Remove any duplicates to eliminate removes
* with corresponding adds.
*/
TAILQ_FOREACH_SAFE(srn, &sino->si_newrecs, sr_next, srp) {
switch (srn->sr_rec->rec_jrefrec.jr_op) {
case JOP_ADDREF:
/*
* This should actually be an error we should
* have a remove for every add journaled.
*/
rrn = (struct jrefrec *)srn->sr_rec;
if (rrn->jr_parent != refrec->jr_parent ||
rrn->jr_diroff != refrec->jr_diroff)
break;
TAILQ_REMOVE(&sino->si_newrecs, srn, sr_next);
break;
case JOP_REMREF:
/*
* Once we remove the current iteration of the
* record at this address we're done.
*/
rrn = (struct jrefrec *)srn->sr_rec;
if (rrn->jr_parent != refrec->jr_parent ||
rrn->jr_diroff != refrec->jr_diroff)
break;
TAILQ_REMOVE(&sino->si_newrecs, srn, sr_next);
ino_add_ref(sino, srec);
return;
case JOP_MVREF:
/*
* Update our diroff based on any moves that match
* and remove the move.
*/
mvrec = (struct jmvrec *)srn->sr_rec;
if (mvrec->jm_parent != refrec->jr_parent ||
mvrec->jm_oldoff != refrec->jr_diroff)
break;
ino_dup_ref(sino, refrec, mvrec->jm_oldoff);
refrec->jr_diroff = mvrec->jm_newoff;
TAILQ_REMOVE(&sino->si_newrecs, srn, sr_next);
break;
default:
err_suj("ino_build_ref: Unknown op %d\n",
srn->sr_rec->rec_jrefrec.jr_op);
}
}
ino_add_ref(sino, srec);
}
/*
* Walk the list of new records and add them in-order resolving any
* dups and adjusted offsets.
*/
static void
ino_build(struct suj_ino *sino)
{
struct suj_rec *srec;
while ((srec = TAILQ_FIRST(&sino->si_newrecs)) != NULL) {
TAILQ_REMOVE(&sino->si_newrecs, srec, sr_next);
switch (srec->sr_rec->rec_jrefrec.jr_op) {
case JOP_ADDREF:
case JOP_REMREF:
ino_build_ref(sino, srec);
break;
case JOP_MVREF:
/*
* Add this mvrec to the queue of pending mvs.
*/
TAILQ_INSERT_TAIL(&sino->si_movs, srec, sr_next);
break;
default:
err_suj("ino_build: Unknown op %d\n",
srec->sr_rec->rec_jrefrec.jr_op);
}
}
if (TAILQ_EMPTY(&sino->si_recs))
sino->si_hasrecs = 0;
}
/*
* Modify journal records so they refer to the base block number
* and a start and end frag range. This is to facilitate the discovery
* of overlapping fragment allocations.
*/
static void
blk_build(struct jblkrec *blkrec)
{
struct suj_rec *srec;
struct suj_blk *sblk;
struct jblkrec *blkrn;
ufs2_daddr_t blk;
int frag;
if (debug)
printf("blk_build: op %d blkno %jd frags %d oldfrags %d "
"ino %ju lbn %jd\n",
blkrec->jb_op, (uintmax_t)blkrec->jb_blkno,
blkrec->jb_frags, blkrec->jb_oldfrags,
(uintmax_t)blkrec->jb_ino, (uintmax_t)blkrec->jb_lbn);
blk = blknum(fs, blkrec->jb_blkno);
frag = fragnum(fs, blkrec->jb_blkno);
sblk = blk_lookup(blk, 1);
/*
* Rewrite the record using oldfrags to indicate the offset into
* the block. Leave jb_frags as the actual allocated count.
*/
blkrec->jb_blkno -= frag;
blkrec->jb_oldfrags = frag;
if (blkrec->jb_oldfrags + blkrec->jb_frags > fs->fs_frag)
err_suj("Invalid fragment count %d oldfrags %d\n",
blkrec->jb_frags, frag);
/*
* Detect dups. If we detect a dup we always discard the oldest
* record as it is superseded by the new record. This speeds up
* later stages but also eliminates free records which are used
* to indicate that the contents of indirects can be trusted.
*/
TAILQ_FOREACH(srec, &sblk->sb_recs, sr_next) {
blkrn = (struct jblkrec *)srec->sr_rec;
if (blkrn->jb_ino != blkrec->jb_ino ||
blkrn->jb_lbn != blkrec->jb_lbn ||
blkrn->jb_blkno != blkrec->jb_blkno ||
blkrn->jb_frags != blkrec->jb_frags ||
blkrn->jb_oldfrags != blkrec->jb_oldfrags)
continue;
if (debug)
printf("Removed dup.\n");
/* Discard the free which is a dup with an alloc. */
if (blkrec->jb_op == JOP_FREEBLK)
return;
TAILQ_REMOVE(&sblk->sb_recs, srec, sr_next);
free(srec);
break;
}
srec = errmalloc(sizeof(*srec));
srec->sr_rec = (union jrec *)blkrec;
TAILQ_INSERT_TAIL(&sblk->sb_recs, srec, sr_next);
}
static void
ino_build_trunc(struct jtrncrec *rec)
{
struct suj_ino *sino;
if (debug)
printf("ino_build_trunc: op %d ino %ju, size %jd\n",
rec->jt_op, (uintmax_t)rec->jt_ino,
(uintmax_t)rec->jt_size);
sino = ino_lookup(rec->jt_ino, 1);
if (rec->jt_op == JOP_SYNC) {
sino->si_trunc = NULL;
return;
}
if (sino->si_trunc == NULL || sino->si_trunc->jt_size > rec->jt_size)
sino->si_trunc = rec;
}
/*
* Build up tables of the operations we need to recover.
*/
static void
suj_build(void)
{
struct suj_seg *seg;
union jrec *rec;
int off;
int i;
TAILQ_FOREACH(seg, &allsegs, ss_next) {
if (debug)
printf("seg %jd has %d records, oldseq %jd.\n",
seg->ss_rec.jsr_seq, seg->ss_rec.jsr_cnt,
seg->ss_rec.jsr_oldest);
off = 0;
rec = (union jrec *)seg->ss_blk;
for (i = 0; i < seg->ss_rec.jsr_cnt; off += JREC_SIZE, rec++) {
/* skip the segrec. */
if ((off % real_dev_bsize) == 0)
continue;
switch (rec->rec_jrefrec.jr_op) {
case JOP_ADDREF:
case JOP_REMREF:
case JOP_MVREF:
ino_append(rec);
break;
case JOP_NEWBLK:
case JOP_FREEBLK:
blk_build((struct jblkrec *)rec);
break;
case JOP_TRUNC:
case JOP_SYNC:
ino_build_trunc((struct jtrncrec *)rec);
break;
default:
err_suj("Unknown journal operation %d (%d)\n",
rec->rec_jrefrec.jr_op, off);
}
i++;
}
}
}
/*
* Prune the journal segments to those we care about based on the
* oldest sequence in the newest segment. Order the segment list
* based on sequence number.
*/
static void
suj_prune(void)
{
struct suj_seg *seg;
struct suj_seg *segn;
uint64_t newseq;
int discard;
if (debug)
printf("Pruning up to %jd\n", oldseq);
/* First free the expired segments. */
TAILQ_FOREACH_SAFE(seg, &allsegs, ss_next, segn) {
if (seg->ss_rec.jsr_seq >= oldseq)
continue;
TAILQ_REMOVE(&allsegs, seg, ss_next);
free(seg->ss_blk);
free(seg);
}
/* Next ensure that segments are ordered properly. */
seg = TAILQ_FIRST(&allsegs);
if (seg == NULL) {
if (debug)
printf("Empty journal\n");
return;
}
newseq = seg->ss_rec.jsr_seq;
for (;;) {
seg = TAILQ_LAST(&allsegs, seghd);
if (seg->ss_rec.jsr_seq >= newseq)
break;
TAILQ_REMOVE(&allsegs, seg, ss_next);
TAILQ_INSERT_HEAD(&allsegs, seg, ss_next);
newseq = seg->ss_rec.jsr_seq;
}
if (newseq != oldseq) {
TAILQ_FOREACH(seg, &allsegs, ss_next) {
printf("%jd, ", seg->ss_rec.jsr_seq);
}
printf("\n");
err_suj("Journal file sequence mismatch %jd != %jd\n",
newseq, oldseq);
}
/*
* The kernel may asynchronously write segments which can create
* gaps in the sequence space. Throw away any segments after the
* gap as the kernel guarantees only those that are contiguously
* reachable are marked as completed.
*/
discard = 0;
TAILQ_FOREACH_SAFE(seg, &allsegs, ss_next, segn) {
if (!discard && newseq++ == seg->ss_rec.jsr_seq) {
jrecs += seg->ss_rec.jsr_cnt;
jbytes += seg->ss_rec.jsr_blocks * real_dev_bsize;
continue;
}
discard = 1;
if (debug)
printf("Journal order mismatch %jd != %jd pruning\n",
newseq-1, seg->ss_rec.jsr_seq);
TAILQ_REMOVE(&allsegs, seg, ss_next);
free(seg->ss_blk);
free(seg);
}
if (debug)
printf("Processing journal segments from %jd to %jd\n",
oldseq, newseq-1);
}
/*
* Verify the journal inode before attempting to read records.
*/
static int
suj_verifyino(union dinode *dp)
{
if (DIP(dp, di_nlink) != 1) {
printf("Invalid link count %d for journal inode %ju\n",
DIP(dp, di_nlink), (uintmax_t)sujino);
return (-1);
}
if ((DIP(dp, di_flags) & (SF_IMMUTABLE | SF_NOUNLINK)) !=
(SF_IMMUTABLE | SF_NOUNLINK)) {
printf("Invalid flags 0x%X for journal inode %ju\n",
DIP(dp, di_flags), (uintmax_t)sujino);
return (-1);
}
if (DIP(dp, di_mode) != (IFREG | IREAD)) {
printf("Invalid mode %o for journal inode %ju\n",
DIP(dp, di_mode), (uintmax_t)sujino);
return (-1);
}
if (DIP(dp, di_size) < SUJ_MIN) {
printf("Invalid size %jd for journal inode %ju\n",
DIP(dp, di_size), (uintmax_t)sujino);
return (-1);
}
if (DIP(dp, di_modrev) != fs->fs_mtime) {
printf("Journal timestamp does not match fs mount time\n");
return (-1);
}
return (0);
}
struct jblocks {
struct jextent *jb_extent; /* Extent array. */
int jb_avail; /* Available extents. */
int jb_used; /* Last used extent. */
int jb_head; /* Allocator head. */
int jb_off; /* Allocator extent offset. */
};
struct jextent {
ufs2_daddr_t je_daddr; /* Disk block address. */
int je_blocks; /* Disk block count. */
};
static struct jblocks *suj_jblocks;
static struct jblocks *
jblocks_create(void)
{
struct jblocks *jblocks;
int size;
jblocks = errmalloc(sizeof(*jblocks));
jblocks->jb_avail = 10;
jblocks->jb_used = 0;
jblocks->jb_head = 0;
jblocks->jb_off = 0;
size = sizeof(struct jextent) * jblocks->jb_avail;
jblocks->jb_extent = errmalloc(size);
bzero(jblocks->jb_extent, size);
return (jblocks);
}
/*
* Return the next available disk block and the amount of contiguous
* free space it contains.
*/
static ufs2_daddr_t
jblocks_next(struct jblocks *jblocks, int bytes, int *actual)
{
struct jextent *jext;
ufs2_daddr_t daddr;
int freecnt;
int blocks;
blocks = btodb(bytes);
jext = &jblocks->jb_extent[jblocks->jb_head];
freecnt = jext->je_blocks - jblocks->jb_off;
if (freecnt == 0) {
jblocks->jb_off = 0;
if (++jblocks->jb_head > jblocks->jb_used)
return (0);
jext = &jblocks->jb_extent[jblocks->jb_head];
freecnt = jext->je_blocks;
}
if (freecnt > blocks)
freecnt = blocks;
*actual = dbtob(freecnt);
daddr = jext->je_daddr + jblocks->jb_off;
return (daddr);
}
/*
* Advance the allocation head by a specified number of bytes, consuming
* one journal segment.
*/
static void
jblocks_advance(struct jblocks *jblocks, int bytes)
{
jblocks->jb_off += btodb(bytes);
}
static void
jblocks_destroy(struct jblocks *jblocks)
{
free(jblocks->jb_extent);
free(jblocks);
}
static void
jblocks_add(struct jblocks *jblocks, ufs2_daddr_t daddr, int blocks)
{
struct jextent *jext;
int size;
jext = &jblocks->jb_extent[jblocks->jb_used];
/* Adding the first block. */
if (jext->je_daddr == 0) {
jext->je_daddr = daddr;
jext->je_blocks = blocks;
return;
}
/* Extending the last extent. */
if (jext->je_daddr + jext->je_blocks == daddr) {
jext->je_blocks += blocks;
return;
}
/* Adding a new extent. */
if (++jblocks->jb_used == jblocks->jb_avail) {
jblocks->jb_avail *= 2;
size = sizeof(struct jextent) * jblocks->jb_avail;
jext = errmalloc(size);
bzero(jext, size);
bcopy(jblocks->jb_extent, jext,
sizeof(struct jextent) * jblocks->jb_used);
free(jblocks->jb_extent);
jblocks->jb_extent = jext;
}
jext = &jblocks->jb_extent[jblocks->jb_used];
jext->je_daddr = daddr;
jext->je_blocks = blocks;
return;
}
/*
* Add a file block from the journal to the extent map. We can't read
* each file block individually because the kernel treats it as a circular
* buffer and segments may span mutliple contiguous blocks.
*/
static void
suj_add_block(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{
jblocks_add(suj_jblocks, fsbtodb(fs, blk), fsbtodb(fs, frags));
}
static void
suj_read(void)
{
uint8_t block[1 * 1024 * 1024];
struct suj_seg *seg;
struct jsegrec *recn;
struct jsegrec *rec;
ufs2_daddr_t blk;
int readsize;
int blocks;
int recsize;
int size;
int i;
/*
* Read records until we exhaust the journal space. If we find
* an invalid record we start searching for a valid segment header
* at the next block. This is because we don't have a head/tail
* pointer and must recover the information indirectly. At the gap
* between the head and tail we won't necessarily have a valid
* segment.
*/
restart:
for (;;) {
size = sizeof(block);
blk = jblocks_next(suj_jblocks, size, &readsize);
if (blk == 0)
return;
size = readsize;
/*
* Read 1MB at a time and scan for records within this block.
*/
if (pread(fsreadfd, &block, size, dbtob(blk)) != size) {
err_suj("Error reading journal block %jd\n",
(intmax_t)blk);
}
for (rec = (void *)block; size; size -= recsize,
rec = (struct jsegrec *)((uintptr_t)rec + recsize)) {
recsize = real_dev_bsize;
if (rec->jsr_time != fs->fs_mtime) {
#ifdef notdef
if (debug)
printf("Rec time %jd != fs mtime %jd\n",
rec->jsr_time, fs->fs_mtime);
#endif
jblocks_advance(suj_jblocks, recsize);
continue;
}
if (rec->jsr_cnt == 0) {
if (debug)
printf("Found illegal count %d\n",
rec->jsr_cnt);
jblocks_advance(suj_jblocks, recsize);
continue;
}
blocks = rec->jsr_blocks;
recsize = blocks * real_dev_bsize;
if (recsize > size) {
/*
* We may just have run out of buffer, restart
* the loop to re-read from this spot.
*/
if (size < fs->fs_bsize &&
size != readsize &&
recsize <= fs->fs_bsize)
goto restart;
if (debug)
printf("Found invalid segsize %d > %d\n",
recsize, size);
recsize = real_dev_bsize;
jblocks_advance(suj_jblocks, recsize);
continue;
}
/*
* Verify that all blocks in the segment are present.
*/
for (i = 1; i < blocks; i++) {
recn = (void *)((uintptr_t)rec) + i *
real_dev_bsize;
if (recn->jsr_seq == rec->jsr_seq &&
recn->jsr_time == rec->jsr_time)
continue;
if (debug)
printf("Incomplete record %jd (%d)\n",
rec->jsr_seq, i);
recsize = i * real_dev_bsize;
jblocks_advance(suj_jblocks, recsize);
goto restart;
}
seg = errmalloc(sizeof(*seg));
seg->ss_blk = errmalloc(recsize);
seg->ss_rec = *rec;
bcopy((void *)rec, seg->ss_blk, recsize);
if (rec->jsr_oldest > oldseq)
oldseq = rec->jsr_oldest;
TAILQ_INSERT_TAIL(&allsegs, seg, ss_next);
jblocks_advance(suj_jblocks, recsize);
}
}
}
/*
* Orchestrate the verification of a filesystem via the softupdates journal.
*/
int
suj_check(const char *filesys)
{
struct inodesc idesc;
struct csum *cgsum;
union dinode *jip;
struct inode ip;
uint64_t blocks;
int i, retval;
struct suj_seg *seg;
struct suj_seg *segn;
initsuj();
fs = &sblock;
if (real_dev_bsize == 0 && ioctl(fsreadfd, DIOCGSECTORSIZE,
&real_dev_bsize) == -1)
real_dev_bsize = secsize;
if (debug)
printf("dev_bsize %u\n", real_dev_bsize);
/*
* Set an exit point when SUJ check failed
*/
retval = setjmp(jmpbuf);
if (retval != 0) {
pwarn("UNEXPECTED SU+J INCONSISTENCY\n");
TAILQ_FOREACH_SAFE(seg, &allsegs, ss_next, segn) {
TAILQ_REMOVE(&allsegs, seg, ss_next);
free(seg->ss_blk);
free(seg);
}
if (reply("FALLBACK TO FULL FSCK") == 0) {
ckfini(0);
exit(EEXIT);
} else
return (-1);
}
/*
* Search the root directory for the SUJ_FILE.
*/
idesc.id_type = DATA;
idesc.id_fix = IGNORE;
idesc.id_number = UFS_ROOTINO;
idesc.id_func = findino;
idesc.id_name = SUJ_FILE;
ginode(UFS_ROOTINO, &ip);
if ((ckinode(ip.i_dp, &idesc) & FOUND) == FOUND) {
sujino = idesc.id_parent;
irelse(&ip);
} else {
printf("Journal inode removed. Use tunefs to re-create.\n");
sblock.fs_flags &= ~FS_SUJ;
sblock.fs_sujfree = 0;
irelse(&ip);
return (-1);
}
/*
* Fetch the journal inode and verify it.
*/
ginode(sujino, &ip);
jip = ip.i_dp;
printf("** SU+J Recovering %s\n", filesys);
if (suj_verifyino(jip) != 0 || (!preen && !reply("USE JOURNAL"))) {
irelse(&ip);
return (-1);
}
/*
* Build a list of journal blocks in jblocks before parsing the
* available journal blocks in with suj_read().
*/
printf("** Reading %jd byte journal from inode %ju.\n",
DIP(jip, di_size), (uintmax_t)sujino);
suj_jblocks = jblocks_create();
blocks = ino_visit(jip, sujino, suj_add_block, 0);
if (blocks != numfrags(fs, DIP(jip, di_size))) {
printf("Sparse journal inode %ju.\n", (uintmax_t)sujino);
irelse(&ip);
return (-1);
}
irelse(&ip);
suj_read();
jblocks_destroy(suj_jblocks);
suj_jblocks = NULL;
if (preen || reply("RECOVER")) {
printf("** Building recovery table.\n");
suj_prune();
suj_build();
cg_apply(cg_build);
printf("** Resolving unreferenced inode list.\n");
ino_unlinked();
printf("** Processing journal entries.\n");
cg_apply(cg_trunc);
cg_apply(cg_check_blk);
cg_apply(cg_adj_blk);
cg_apply(cg_check_ino);
}
if (preen == 0 && (jrecs > 0 || jbytes > 0) && reply("WRITE CHANGES") == 0)
return (0);
/*
* Recompute the fs summary info from correct cs summaries.
*/
bzero(&fs->fs_cstotal, sizeof(struct csum_total));
for (i = 0; i < fs->fs_ncg; i++) {
cgsum = &fs->fs_cs(fs, i);
fs->fs_cstotal.cs_nffree += cgsum->cs_nffree;
fs->fs_cstotal.cs_nbfree += cgsum->cs_nbfree;
fs->fs_cstotal.cs_nifree += cgsum->cs_nifree;
fs->fs_cstotal.cs_ndir += cgsum->cs_ndir;
}
fs->fs_pendinginodes = 0;
fs->fs_pendingblocks = 0;
fs->fs_clean = 1;
fs->fs_time = time(NULL);
fs->fs_mtime = time(NULL);
sbdirty();
ckfini(1);
if (jrecs > 0 || jbytes > 0) {
printf("** %jd journal records in %jd bytes for %.2f%% utilization\n",
jrecs, jbytes, ((float)jrecs / (float)(jbytes / JREC_SIZE)) * 100);
printf("** Freed %jd inodes (%jd dirs) %jd blocks, and %jd frags.\n",
freeinos, freedir, freeblocks, freefrags);
}
return (0);
}
static void
initsuj(void)
{
int i;
for (i = 0; i < HASHSIZE; i++)
LIST_INIT(&cghash[i]);
lastcg = NULL;
TAILQ_INIT(&allsegs);
oldseq = 0;
fs = NULL;
sujino = 0;
freefrags = 0;
freeblocks = 0;
freeinos = 0;
freedir = 0;
jbytes = 0;
jrecs = 0;
suj_jblocks = NULL;
}
diff --git a/sbin/mksnap_ffs/mksnap_ffs.c b/sbin/mksnap_ffs/mksnap_ffs.c
index 5d230701de85..0e3586fcb2e2 100644
--- a/sbin/mksnap_ffs/mksnap_ffs.c
+++ b/sbin/mksnap_ffs/mksnap_ffs.c
@@ -1,187 +1,191 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 2003 Networks Associates Technology, Inc.
* All rights reserved.
*
* This software was developed for the FreeBSD Project by Marshall
* Kirk McKusick and Network Associates Laboratories, the Security
* Research Division of Network Associates, Inc. under DARPA/SPAWAR
* contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS
* research program.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. The names of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD$
*/
#include <sys/param.h>
#include <sys/mount.h>
#include <sys/stat.h>
+
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
+
#include <err.h>
#include <errno.h>
#include <fcntl.h>
#include <grp.h>
#include <limits.h>
#include <mntopts.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sysexits.h>
#include <unistd.h>
static void
usage(void)
{
errx(EX_USAGE, "usage: mksnap_ffs snapshot_name");
}
static int
isdir(const char *path, struct stat *stbufp)
{
if (stat(path, stbufp) < 0)
return (-1);
if (!S_ISDIR(stbufp->st_mode))
return (0);
return (1);
}
static int
issamefs(const char *path, struct statfs *stfsp)
{
struct statfs stfsbuf;
struct stat stbuf;
if (isdir(path, &stbuf) != 1)
return (-1);
if (statfs(path, &stfsbuf) < 0)
return (-1);
if (fsidcmp(&stfsbuf.f_fsid, &stfsp->f_fsid) != 0)
return (0);
return (1);
}
int
main(int argc, char **argv)
{
char errmsg[255], path[PATH_MAX];
char *cp, *snapname;
struct statfs stfsbuf;
struct group *grp;
struct stat stbuf;
struct iovec *iov;
int fd, iovlen;
if (argc == 2)
snapname = argv[1];
else if (argc == 3)
snapname = argv[2]; /* Old usage. */
else
usage();
/*
* Check that the user running this program has permission
* to create and remove a snapshot file from the directory
* in which they have requested to have it made. If the
* directory is sticky and not owned by the user, then they
* will not be able to remove the snapshot when they are
* done with it.
*/
if (strlen(snapname) >= PATH_MAX)
errx(1, "pathname too long %s", snapname);
cp = strrchr(snapname, '/');
if (cp == NULL) {
strlcpy(path, ".", PATH_MAX);
} else if (cp == snapname) {
strlcpy(path, "/", PATH_MAX);
} else {
strlcpy(path, snapname, cp - snapname + 1);
}
if (statfs(path, &stfsbuf) < 0)
err(1, "%s", path);
switch (isdir(path, &stbuf)) {
case -1:
err(1, "%s", path);
case 0:
errx(1, "%s: Not a directory", path);
default:
break;
}
if (access(path, W_OK) < 0)
err(1, "Lack write permission in %s", path);
if ((stbuf.st_mode & S_ISTXT) && stbuf.st_uid != getuid())
errx(1, "Lack write permission in %s: Sticky bit set", path);
/*
* Work around an issue when mksnap_ffs is started in chroot'ed
* environment and f_mntonname contains absolute path within
* real root.
*/
for (cp = stfsbuf.f_mntonname; issamefs(cp, &stfsbuf) != 1;
cp = strchrnul(cp + 1, '/')) {
if (cp[0] == '\0')
errx(1, "%s: Not a mount point", stfsbuf.f_mntonname);
}
if (cp != stfsbuf.f_mntonname)
strlcpy(stfsbuf.f_mntonname, cp, sizeof(stfsbuf.f_mntonname));
/*
* Having verified access to the directory in which the
* snapshot is to be built, proceed with creating it.
*/
if ((grp = getgrnam("operator")) == NULL)
errx(1, "Cannot retrieve operator gid");
iov = NULL;
iovlen = 0;
build_iovec(&iov, &iovlen, "fstype", "ffs", 4);
build_iovec(&iov, &iovlen, "from", snapname, (size_t)-1);
build_iovec(&iov, &iovlen, "fspath", stfsbuf.f_mntonname, (size_t)-1);
build_iovec(&iov, &iovlen, "errmsg", errmsg, sizeof(errmsg));
build_iovec(&iov, &iovlen, "update", NULL, 0);
build_iovec(&iov, &iovlen, "snapshot", NULL, 0);
*errmsg = '\0';
if (nmount(iov, iovlen, stfsbuf.f_flags) < 0) {
errmsg[sizeof(errmsg) - 1] = '\0';
err(1, "Cannot create snapshot %s%s%s", snapname,
*errmsg != '\0' ? ": " : "", errmsg);
}
if ((fd = open(snapname, O_RDONLY)) < 0)
err(1, "Cannot open %s", snapname);
if (fstat(fd, &stbuf) != 0)
err(1, "Cannot stat %s", snapname);
if ((stbuf.st_flags & SF_SNAPSHOT) == 0)
errx(1, "File %s is not a snapshot", snapname);
if (fchown(fd, -1, grp->gr_gid) != 0)
err(1, "Cannot chown %s", snapname);
if (fchmod(fd, S_IRUSR | S_IRGRP) != 0)
err(1, "Cannot chmod %s", snapname);
exit(EXIT_SUCCESS);
}
diff --git a/sbin/newfs/newfs.c b/sbin/newfs/newfs.c
index 232436c0aa7f..a2d97131d3a7 100644
--- a/sbin/newfs/newfs.c
+++ b/sbin/newfs/newfs.c
@@ -1,512 +1,514 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 2002 Networks Associates Technology, Inc.
* All rights reserved.
*
* This software was developed for the FreeBSD Project by Marshall
* Kirk McKusick and Network Associates Laboratories, the Security
* Research Division of Network Associates, Inc. under DARPA/SPAWAR
* contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS
* research program.
*
* Copyright (c) 1983, 1989, 1993, 1994
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#if 0
#ifndef lint
static const char copyright[] =
"@(#) Copyright (c) 1983, 1989, 1993, 1994\n\
The Regents of the University of California. All rights reserved.\n";
#endif /* not lint */
#ifndef lint
static char sccsid[] = "@(#)newfs.c 8.13 (Berkeley) 5/1/95";
#endif /* not lint */
#endif
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/*
* newfs: friendly front end to mkfs
*/
#include <sys/param.h>
#include <sys/stat.h>
#include <sys/disk.h>
#include <sys/disklabel.h>
#include <sys/file.h>
#include <sys/mount.h>
#include <ufs/ufs/dir.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ctype.h>
#include <err.h>
#include <errno.h>
#include <inttypes.h>
#include <paths.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <syslog.h>
#include <unistd.h>
#include <libutil.h>
#include "newfs.h"
int Eflag; /* Erase previous disk contents */
int Lflag; /* add a volume label */
int Nflag; /* run without writing file system */
int Oflag = 2; /* file system format (1 => UFS1, 2 => UFS2) */
int Rflag; /* regression test */
int Uflag; /* enable soft updates for file system */
int jflag; /* enable soft updates journaling for filesys */
int Xflag = 0; /* exit in middle of newfs for testing */
int Jflag; /* enable gjournal for file system */
int lflag; /* enable multilabel for file system */
int nflag; /* do not create .snap directory */
int tflag; /* enable TRIM */
intmax_t fssize; /* file system size */
off_t mediasize; /* device size */
int sectorsize; /* bytes/sector */
int realsectorsize; /* bytes/sector in hardware */
int fsize = 0; /* fragment size */
int bsize = 0; /* block size */
int maxbsize = 0; /* maximum clustering */
int maxblkspercg = MAXBLKSPERCG; /* maximum blocks per cylinder group */
int minfree = MINFREE; /* free space threshold */
int metaspace; /* space held for metadata blocks */
int opt = DEFAULTOPT; /* optimization preference (space or time) */
int density; /* number of bytes per inode */
int maxcontig = 0; /* max contiguous blocks to allocate */
int maxbpg; /* maximum blocks per file in a cyl group */
int avgfilesize = AVFILESIZ;/* expected average file size */
int avgfilesperdir = AFPDIR;/* expected number of files per directory */
u_char *volumelabel = NULL; /* volume label for filesystem */
struct uufsd disk; /* libufs disk structure */
static char device[MAXPATHLEN];
static u_char bootarea[BBSIZE];
static int is_file; /* work on a file, not a device */
static char *dkname;
static char *disktype;
static void getfssize(intmax_t *, const char *p, intmax_t, intmax_t);
static struct disklabel *getdisklabel(void);
static void usage(void);
static int expand_number_int(const char *buf, int *num);
ufs2_daddr_t part_ofs; /* partition offset in blocks, used with files */
int
main(int argc, char *argv[])
{
struct partition *pp;
struct disklabel *lp;
struct stat st;
char *cp, *special;
intmax_t reserved;
int ch, i, rval;
char part_name; /* partition name, default to full disk */
part_name = 'c';
reserved = 0;
while ((ch = getopt(argc, argv,
"EJL:NO:RS:T:UXa:b:c:d:e:f:g:h:i:jk:lm:no:p:r:s:t")) != -1)
switch (ch) {
case 'E':
Eflag = 1;
break;
case 'J':
Jflag = 1;
break;
case 'L':
volumelabel = optarg;
i = -1;
while (isalnum(volumelabel[++i]) ||
volumelabel[i] == '_' || volumelabel[i] == '-');
if (volumelabel[i] != '\0') {
errx(1, "bad volume label. Valid characters "
"are alphanumerics, dashes, and underscores.");
}
if (strlen(volumelabel) >= MAXVOLLEN) {
errx(1, "bad volume label. Length is longer than %d.",
MAXVOLLEN);
}
Lflag = 1;
break;
case 'N':
Nflag = 1;
break;
case 'O':
if ((Oflag = atoi(optarg)) < 1 || Oflag > 2)
errx(1, "%s: bad file system format value",
optarg);
break;
case 'R':
Rflag = 1;
break;
case 'S':
rval = expand_number_int(optarg, §orsize);
if (rval < 0 || sectorsize <= 0)
errx(1, "%s: bad sector size", optarg);
break;
case 'T':
disktype = optarg;
break;
case 'j':
jflag = 1;
/* fall through to enable soft updates */
/* FALLTHROUGH */
case 'U':
Uflag = 1;
break;
case 'X':
Xflag++;
break;
case 'a':
rval = expand_number_int(optarg, &maxcontig);
if (rval < 0 || maxcontig <= 0)
errx(1, "%s: bad maximum contiguous blocks",
optarg);
break;
case 'b':
rval = expand_number_int(optarg, &bsize);
if (rval < 0)
errx(1, "%s: bad block size",
optarg);
if (bsize < MINBSIZE)
errx(1, "%s: block size too small, min is %d",
optarg, MINBSIZE);
if (bsize > MAXBSIZE)
errx(1, "%s: block size too large, max is %d",
optarg, MAXBSIZE);
break;
case 'c':
rval = expand_number_int(optarg, &maxblkspercg);
if (rval < 0 || maxblkspercg <= 0)
errx(1, "%s: bad blocks per cylinder group",
optarg);
break;
case 'd':
rval = expand_number_int(optarg, &maxbsize);
if (rval < 0 || maxbsize < MINBSIZE)
errx(1, "%s: bad extent block size", optarg);
break;
case 'e':
rval = expand_number_int(optarg, &maxbpg);
if (rval < 0 || maxbpg <= 0)
errx(1, "%s: bad blocks per file in a cylinder group",
optarg);
break;
case 'f':
rval = expand_number_int(optarg, &fsize);
if (rval < 0 || fsize <= 0)
errx(1, "%s: bad fragment size", optarg);
break;
case 'g':
rval = expand_number_int(optarg, &avgfilesize);
if (rval < 0 || avgfilesize <= 0)
errx(1, "%s: bad average file size", optarg);
break;
case 'h':
rval = expand_number_int(optarg, &avgfilesperdir);
if (rval < 0 || avgfilesperdir <= 0)
errx(1, "%s: bad average files per dir", optarg);
break;
case 'i':
rval = expand_number_int(optarg, &density);
if (rval < 0 || density <= 0)
errx(1, "%s: bad bytes per inode", optarg);
break;
case 'l':
lflag = 1;
break;
case 'k':
if ((metaspace = atoi(optarg)) < 0)
errx(1, "%s: bad metadata space %%", optarg);
if (metaspace == 0)
/* force to stay zero in mkfs */
metaspace = -1;
break;
case 'm':
if ((minfree = atoi(optarg)) < 0 || minfree > 99)
errx(1, "%s: bad free space %%", optarg);
break;
case 'n':
nflag = 1;
break;
case 'o':
if (strcmp(optarg, "space") == 0)
opt = FS_OPTSPACE;
else if (strcmp(optarg, "time") == 0)
opt = FS_OPTTIME;
else
errx(1,
"%s: unknown optimization preference: use `space' or `time'",
optarg);
break;
case 'r':
errno = 0;
reserved = strtoimax(optarg, &cp, 0);
if (errno != 0 || cp == optarg ||
*cp != '\0' || reserved < 0)
errx(1, "%s: bad reserved size", optarg);
break;
case 'p':
is_file = 1;
part_name = optarg[0];
break;
case 's':
errno = 0;
fssize = strtoimax(optarg, &cp, 0);
if (errno != 0 || cp == optarg ||
*cp != '\0' || fssize < 0)
errx(1, "%s: bad file system size", optarg);
break;
case 't':
tflag = 1;
break;
case '?':
default:
usage();
}
argc -= optind;
argv += optind;
if (argc != 1)
usage();
special = argv[0];
if (!special[0])
err(1, "empty file/special name");
cp = strrchr(special, '/');
if (cp == NULL) {
/*
* No path prefix; try prefixing _PATH_DEV.
*/
snprintf(device, sizeof(device), "%s%s", _PATH_DEV, special);
special = device;
}
if (is_file) {
/* bypass ufs_disk_fillout_blank */
bzero( &disk, sizeof(disk));
disk.d_bsize = 1;
disk.d_name = special;
disk.d_fd = open(special, O_RDONLY);
if (disk.d_fd < 0 ||
(!Nflag && ufs_disk_write(&disk) == -1))
errx(1, "%s: ", special);
} else if (ufs_disk_fillout_blank(&disk, special) == -1 ||
(!Nflag && ufs_disk_write(&disk) == -1)) {
if (disk.d_error != NULL)
errx(1, "%s: %s", special, disk.d_error);
else
err(1, "%s", special);
}
if (fstat(disk.d_fd, &st) < 0)
err(1, "%s", special);
if ((st.st_mode & S_IFMT) != S_IFCHR) {
warn("%s: not a character-special device", special);
is_file = 1; /* assume it is a file */
dkname = special;
if (sectorsize == 0)
sectorsize = 512;
mediasize = st.st_size;
/* set fssize from the partition */
} else {
if (sectorsize == 0)
if (ioctl(disk.d_fd, DIOCGSECTORSIZE, §orsize) == -1)
sectorsize = 0; /* back out on error for safety */
if (sectorsize && ioctl(disk.d_fd, DIOCGMEDIASIZE, &mediasize) != -1)
getfssize(&fssize, special, mediasize / sectorsize, reserved);
}
pp = NULL;
lp = getdisklabel();
if (lp != NULL) {
if (!is_file) /* already set for files */
part_name = special[strlen(special) - 1];
if ((part_name < 'a' || part_name - 'a' >= MAXPARTITIONS) &&
!isdigit(part_name))
errx(1, "%s: can't figure out file system partition",
special);
cp = &part_name;
if (isdigit(*cp))
pp = &lp->d_partitions[RAW_PART];
else
pp = &lp->d_partitions[*cp - 'a'];
if (pp->p_size == 0)
errx(1, "%s: `%c' partition is unavailable",
special, *cp);
if (pp->p_fstype == FS_BOOT)
errx(1, "%s: `%c' partition overlaps boot program",
special, *cp);
getfssize(&fssize, special, pp->p_size, reserved);
if (sectorsize == 0)
sectorsize = lp->d_secsize;
if (fsize == 0)
fsize = pp->p_fsize;
if (bsize == 0)
bsize = pp->p_frag * pp->p_fsize;
if (is_file)
part_ofs = pp->p_offset;
}
if (sectorsize <= 0)
errx(1, "%s: no default sector size", special);
if (fsize <= 0)
fsize = MAX(DFL_FRAGSIZE, sectorsize);
if (bsize <= 0)
bsize = MIN(DFL_BLKSIZE, 8 * fsize);
if (minfree < MINFREE && opt != FS_OPTSPACE) {
fprintf(stderr, "Warning: changing optimization to space ");
fprintf(stderr, "because minfree is less than %d%%\n", MINFREE);
opt = FS_OPTSPACE;
}
realsectorsize = sectorsize;
if (sectorsize != DEV_BSIZE) { /* XXX */
int secperblk = sectorsize / DEV_BSIZE;
sectorsize = DEV_BSIZE;
fssize *= secperblk;
if (pp != NULL)
pp->p_size *= secperblk;
}
mkfs(pp, special);
ufs_disk_close(&disk);
if (!jflag)
exit(0);
if (execlp("tunefs", "newfs", "-j", "enable", special, NULL) < 0)
err(1, "Cannot enable soft updates journaling, tunefs");
/* NOT REACHED */
}
void
getfssize(intmax_t *fsz, const char *s, intmax_t disksize, intmax_t reserved)
{
intmax_t available;
available = disksize - reserved;
if (available <= 0)
errx(1, "%s: reserved not less than device size %jd",
s, disksize);
if (*fsz == 0)
*fsz = available;
else if (*fsz > available)
errx(1, "%s: maximum file system size is %jd",
s, available);
}
struct disklabel *
getdisklabel(void)
{
static struct disklabel lab;
struct disklabel *lp;
if (is_file) {
if (read(disk.d_fd, bootarea, BBSIZE) != BBSIZE)
err(4, "cannot read bootarea");
if (bsd_disklabel_le_dec(
bootarea + (0 /* labeloffset */ +
1 /* labelsoffset */ * sectorsize),
&lab, MAXPARTITIONS))
errx(1, "no valid label found");
lp = &lab;
return &lab;
}
if (disktype) {
lp = getdiskbyname(disktype);
if (lp != NULL)
return (lp);
}
return (NULL);
}
static void
usage(void)
{
fprintf(stderr,
"usage: %s [ -fsoptions ] special-device%s\n",
getprogname(),
" [device-type]");
fprintf(stderr, "where fsoptions are:\n");
fprintf(stderr, "\t-E Erase previous disk content\n");
fprintf(stderr, "\t-J Enable journaling via gjournal\n");
fprintf(stderr, "\t-L volume label to add to superblock\n");
fprintf(stderr,
"\t-N do not create file system, just print out parameters\n");
fprintf(stderr, "\t-O file system format: 1 => UFS1, 2 => UFS2\n");
fprintf(stderr, "\t-R regression test, suppress random factors\n");
fprintf(stderr, "\t-S sector size\n");
fprintf(stderr, "\t-T disktype\n");
fprintf(stderr, "\t-U enable soft updates\n");
fprintf(stderr, "\t-a maximum contiguous blocks\n");
fprintf(stderr, "\t-b block size\n");
fprintf(stderr, "\t-c blocks per cylinders group\n");
fprintf(stderr, "\t-d maximum extent size\n");
fprintf(stderr, "\t-e maximum blocks per file in a cylinder group\n");
fprintf(stderr, "\t-f frag size\n");
fprintf(stderr, "\t-g average file size\n");
fprintf(stderr, "\t-h average files per directory\n");
fprintf(stderr, "\t-i number of bytes per inode\n");
fprintf(stderr, "\t-j enable soft updates journaling\n");
fprintf(stderr, "\t-k space to hold for metadata blocks\n");
fprintf(stderr, "\t-l enable multilabel MAC\n");
fprintf(stderr, "\t-n do not create .snap directory\n");
fprintf(stderr, "\t-m minimum free space %%\n");
fprintf(stderr, "\t-o optimization preference (`space' or `time')\n");
fprintf(stderr, "\t-p partition name (a..h)\n");
fprintf(stderr, "\t-r reserved sectors at the end of device\n");
fprintf(stderr, "\t-s file system size (sectors)\n");
fprintf(stderr, "\t-t enable TRIM\n");
exit(1);
}
static int
expand_number_int(const char *buf, int *num)
{
int64_t num64;
int rval;
rval = expand_number(buf, &num64);
if (rval < 0)
return (rval);
if (num64 > INT_MAX || num64 < INT_MIN) {
errno = ERANGE;
return (-1);
}
*num = (int)num64;
return (0);
}
diff --git a/sbin/tunefs/tunefs.c b/sbin/tunefs/tunefs.c
index 2eca8e0a3e36..e59b583ea58b 100644
--- a/sbin/tunefs/tunefs.c
+++ b/sbin/tunefs/tunefs.c
@@ -1,1140 +1,1142 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1983, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#if 0
#ifndef lint
static const char copyright[] =
"@(#) Copyright (c) 1983, 1993\n\
The Regents of the University of California. All rights reserved.\n";
#endif /* not lint */
#ifndef lint
static char sccsid[] = "@(#)tunefs.c 8.2 (Berkeley) 4/19/94";
#endif /* not lint */
#endif
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/*
* tunefs: change layout parameters to an existing file system.
*/
#include <sys/param.h>
#include <sys/mount.h>
#include <sys/disklabel.h>
#include <sys/stat.h>
+#include <ufs/ufs/extattr.h>
+#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include <ufs/ufs/dir.h>
#include <ctype.h>
#include <err.h>
#include <fcntl.h>
#include <fstab.h>
#include <libufs.h>
#include <mntopts.h>
#include <paths.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
/* the optimization warning string template */
#define OPTWARN "should optimize for %s with minfree %s %d%%"
static int blocks;
static char clrbuf[MAXBSIZE];
static struct uufsd disk;
#define sblock disk.d_fs
static void usage(void);
static void printfs(void);
static int journal_alloc(int64_t size);
static void journal_clear(void);
static void sbdirty(void);
int
main(int argc, char *argv[])
{
const char *avalue, *jvalue, *Jvalue, *Lvalue, *lvalue, *Nvalue, *nvalue;
const char *tvalue;
const char *special, *on;
const char *name;
int active;
int Aflag, aflag, eflag, evalue, fflag, fvalue, jflag, Jflag, kflag;
int kvalue, Lflag, lflag, mflag, mvalue, Nflag, nflag, oflag, ovalue;
int pflag, sflag, svalue, Svalue, tflag;
int ch, found_arg, i;
int iovlen = 0;
const char *chg[2];
struct statfs stfs;
struct iovec *iov = NULL;
char errmsg[255] = {0};
if (argc < 3)
usage();
Aflag = aflag = eflag = fflag = jflag = Jflag = kflag = Lflag = 0;
lflag = mflag = Nflag = nflag = oflag = pflag = sflag = tflag = 0;
avalue = jvalue = Jvalue = Lvalue = lvalue = Nvalue = nvalue = NULL;
evalue = fvalue = mvalue = ovalue = svalue = Svalue = 0;
active = 0;
found_arg = 0; /* At least one arg is required. */
while ((ch = getopt(argc, argv, "Aa:e:f:j:J:k:L:l:m:N:n:o:ps:S:t:"))
!= -1)
switch (ch) {
case 'A':
found_arg++;
Aflag++;
break;
case 'a':
found_arg++;
name = "POSIX.1e ACLs";
avalue = optarg;
if (strcmp(avalue, "enable") &&
strcmp(avalue, "disable")) {
errx(10, "bad %s (options are %s)",
name, "`enable' or `disable'");
}
aflag = 1;
break;
case 'e':
found_arg++;
name = "maximum blocks per file in a cylinder group";
evalue = atoi(optarg);
if (evalue < 1)
errx(10, "%s must be >= 1 (was %s)",
name, optarg);
eflag = 1;
break;
case 'f':
found_arg++;
name = "average file size";
fvalue = atoi(optarg);
if (fvalue < 1)
errx(10, "%s must be >= 1 (was %s)",
name, optarg);
fflag = 1;
break;
case 'j':
found_arg++;
name = "softdep journaled file system";
jvalue = optarg;
if (strcmp(jvalue, "enable") &&
strcmp(jvalue, "disable")) {
errx(10, "bad %s (options are %s)",
name, "`enable' or `disable'");
}
jflag = 1;
break;
case 'J':
found_arg++;
name = "gjournaled file system";
Jvalue = optarg;
if (strcmp(Jvalue, "enable") &&
strcmp(Jvalue, "disable")) {
errx(10, "bad %s (options are %s)",
name, "`enable' or `disable'");
}
Jflag = 1;
break;
case 'k':
found_arg++;
name = "space to hold for metadata blocks";
kvalue = atoi(optarg);
if (kvalue < 0)
errx(10, "bad %s (%s)", name, optarg);
kflag = 1;
break;
case 'L':
found_arg++;
name = "volume label";
Lvalue = optarg;
i = -1;
while (isalnum(Lvalue[++i]) || Lvalue[i] == '_' ||
Lvalue[i] == '-')
;
if (Lvalue[i] != '\0') {
errx(10, "bad %s. Valid characters are "
"alphanumerics, dashes, and underscores.",
name);
}
if (strlen(Lvalue) >= MAXVOLLEN) {
errx(10, "bad %s. Length is longer than %d.",
name, MAXVOLLEN - 1);
}
Lflag = 1;
break;
case 'l':
found_arg++;
name = "multilabel MAC file system";
lvalue = optarg;
if (strcmp(lvalue, "enable") &&
strcmp(lvalue, "disable")) {
errx(10, "bad %s (options are %s)",
name, "`enable' or `disable'");
}
lflag = 1;
break;
case 'm':
found_arg++;
name = "minimum percentage of free space";
mvalue = atoi(optarg);
if (mvalue < 0 || mvalue > 99)
errx(10, "bad %s (%s)", name, optarg);
mflag = 1;
break;
case 'N':
found_arg++;
name = "NFSv4 ACLs";
Nvalue = optarg;
if (strcmp(Nvalue, "enable") &&
strcmp(Nvalue, "disable")) {
errx(10, "bad %s (options are %s)",
name, "`enable' or `disable'");
}
Nflag = 1;
break;
case 'n':
found_arg++;
name = "soft updates";
nvalue = optarg;
if (strcmp(nvalue, "enable") != 0 &&
strcmp(nvalue, "disable") != 0) {
errx(10, "bad %s (options are %s)",
name, "`enable' or `disable'");
}
nflag = 1;
break;
case 'o':
found_arg++;
name = "optimization preference";
if (strcmp(optarg, "space") == 0)
ovalue = FS_OPTSPACE;
else if (strcmp(optarg, "time") == 0)
ovalue = FS_OPTTIME;
else
errx(10,
"bad %s (options are `space' or `time')",
name);
oflag = 1;
break;
case 'p':
found_arg++;
pflag = 1;
break;
case 's':
found_arg++;
name = "expected number of files per directory";
svalue = atoi(optarg);
if (svalue < 1)
errx(10, "%s must be >= 1 (was %s)",
name, optarg);
sflag = 1;
break;
case 'S':
found_arg++;
name = "Softdep Journal Size";
Svalue = atoi(optarg);
if (Svalue < SUJ_MIN)
errx(10, "%s must be >= %d (was %s)",
name, SUJ_MIN, optarg);
break;
case 't':
found_arg++;
name = "trim";
tvalue = optarg;
if (strcmp(tvalue, "enable") != 0 &&
strcmp(tvalue, "disable") != 0) {
errx(10, "bad %s (options are %s)",
name, "`enable' or `disable'");
}
tflag = 1;
break;
default:
usage();
}
argc -= optind;
argv += optind;
if (found_arg == 0 || argc != 1)
usage();
on = special = argv[0];
if (ufs_disk_fillout(&disk, special) == -1)
goto err;
/*
* Check for unclean filesystem.
*/
if ((sblock.fs_clean == 0 ||
(sblock.fs_flags & (FS_UNCLEAN | FS_NEEDSFSCK)) != 0) &&
(found_arg > 1 || !pflag))
errx(1, "%s is not clean - run fsck.\n", special);
if (disk.d_name != special) {
if (statfs(special, &stfs) != 0)
warn("Can't stat %s", special);
if (strcmp(special, stfs.f_mntonname) == 0)
active = 1;
}
if (pflag) {
printfs();
exit(0);
}
if (Lflag) {
name = "volume label";
strncpy(sblock.fs_volname, Lvalue, MAXVOLLEN);
}
if (aflag) {
name = "POSIX.1e ACLs";
if (strcmp(avalue, "enable") == 0) {
if (sblock.fs_flags & FS_ACLS) {
warnx("%s remains unchanged as enabled", name);
} else if (sblock.fs_flags & FS_NFS4ACLS) {
warnx("%s and NFSv4 ACLs are mutually "
"exclusive", name);
} else {
sblock.fs_flags |= FS_ACLS;
warnx("%s set", name);
}
} else if (strcmp(avalue, "disable") == 0) {
if ((~sblock.fs_flags & FS_ACLS) ==
FS_ACLS) {
warnx("%s remains unchanged as disabled",
name);
} else {
sblock.fs_flags &= ~FS_ACLS;
warnx("%s cleared", name);
}
}
}
if (eflag) {
name = "maximum blocks per file in a cylinder group";
if (sblock.fs_maxbpg == evalue)
warnx("%s remains unchanged as %d", name, evalue);
else {
warnx("%s changes from %d to %d",
name, sblock.fs_maxbpg, evalue);
sblock.fs_maxbpg = evalue;
}
}
if (fflag) {
name = "average file size";
if (sblock.fs_avgfilesize == (unsigned)fvalue) {
warnx("%s remains unchanged as %d", name, fvalue);
}
else {
warnx("%s changes from %d to %d",
name, sblock.fs_avgfilesize, fvalue);
sblock.fs_avgfilesize = fvalue;
}
}
if (jflag) {
name = "soft updates journaling";
if (strcmp(jvalue, "enable") == 0) {
if ((sblock.fs_flags & (FS_DOSOFTDEP | FS_SUJ)) ==
(FS_DOSOFTDEP | FS_SUJ)) {
warnx("%s remains unchanged as enabled", name);
} else if (sblock.fs_clean == 0) {
warnx("%s cannot be enabled until fsck is run",
name);
} else if (journal_alloc(Svalue) != 0) {
warnx("%s cannot be enabled", name);
} else {
sblock.fs_flags |= FS_DOSOFTDEP | FS_SUJ;
warnx("%s set", name);
}
} else if (strcmp(jvalue, "disable") == 0) {
if ((~sblock.fs_flags & FS_SUJ) == FS_SUJ) {
warnx("%s remains unchanged as disabled", name);
} else {
journal_clear();
sblock.fs_flags &= ~FS_SUJ;
sblock.fs_sujfree = 0;
warnx("%s cleared but soft updates still set.",
name);
warnx("remove .sujournal to reclaim space");
}
}
}
if (Jflag) {
name = "gjournal";
if (strcmp(Jvalue, "enable") == 0) {
if (sblock.fs_flags & FS_GJOURNAL) {
warnx("%s remains unchanged as enabled", name);
} else {
sblock.fs_flags |= FS_GJOURNAL;
warnx("%s set", name);
}
} else if (strcmp(Jvalue, "disable") == 0) {
if ((~sblock.fs_flags & FS_GJOURNAL) ==
FS_GJOURNAL) {
warnx("%s remains unchanged as disabled",
name);
} else {
sblock.fs_flags &= ~FS_GJOURNAL;
warnx("%s cleared", name);
}
}
}
if (kflag) {
name = "space to hold for metadata blocks";
if (sblock.fs_metaspace == kvalue)
warnx("%s remains unchanged as %d", name, kvalue);
else {
kvalue = blknum(&sblock, kvalue);
if (kvalue > sblock.fs_fpg / 2) {
kvalue = blknum(&sblock, sblock.fs_fpg / 2);
warnx("%s cannot exceed half the file system "
"space", name);
}
warnx("%s changes from %jd to %d",
name, sblock.fs_metaspace, kvalue);
sblock.fs_metaspace = kvalue;
}
}
if (lflag) {
name = "multilabel";
if (strcmp(lvalue, "enable") == 0) {
if (sblock.fs_flags & FS_MULTILABEL) {
warnx("%s remains unchanged as enabled", name);
} else {
sblock.fs_flags |= FS_MULTILABEL;
warnx("%s set", name);
}
} else if (strcmp(lvalue, "disable") == 0) {
if ((~sblock.fs_flags & FS_MULTILABEL) ==
FS_MULTILABEL) {
warnx("%s remains unchanged as disabled",
name);
} else {
sblock.fs_flags &= ~FS_MULTILABEL;
warnx("%s cleared", name);
}
}
}
if (mflag) {
name = "minimum percentage of free space";
if (sblock.fs_minfree == mvalue)
warnx("%s remains unchanged as %d%%", name, mvalue);
else {
warnx("%s changes from %d%% to %d%%",
name, sblock.fs_minfree, mvalue);
sblock.fs_minfree = mvalue;
if (mvalue >= MINFREE && sblock.fs_optim == FS_OPTSPACE)
warnx(OPTWARN, "time", ">=", MINFREE);
if (mvalue < MINFREE && sblock.fs_optim == FS_OPTTIME)
warnx(OPTWARN, "space", "<", MINFREE);
}
}
if (Nflag) {
name = "NFSv4 ACLs";
if (strcmp(Nvalue, "enable") == 0) {
if (sblock.fs_flags & FS_NFS4ACLS) {
warnx("%s remains unchanged as enabled", name);
} else if (sblock.fs_flags & FS_ACLS) {
warnx("%s and POSIX.1e ACLs are mutually "
"exclusive", name);
} else {
sblock.fs_flags |= FS_NFS4ACLS;
warnx("%s set", name);
}
} else if (strcmp(Nvalue, "disable") == 0) {
if ((~sblock.fs_flags & FS_NFS4ACLS) ==
FS_NFS4ACLS) {
warnx("%s remains unchanged as disabled",
name);
} else {
sblock.fs_flags &= ~FS_NFS4ACLS;
warnx("%s cleared", name);
}
}
}
if (nflag) {
name = "soft updates";
if (strcmp(nvalue, "enable") == 0) {
if (sblock.fs_flags & FS_DOSOFTDEP)
warnx("%s remains unchanged as enabled", name);
else if (sblock.fs_clean == 0) {
warnx("%s cannot be enabled until fsck is run",
name);
} else {
sblock.fs_flags |= FS_DOSOFTDEP;
warnx("%s set", name);
}
} else if (strcmp(nvalue, "disable") == 0) {
if ((~sblock.fs_flags & FS_DOSOFTDEP) == FS_DOSOFTDEP)
warnx("%s remains unchanged as disabled", name);
else {
sblock.fs_flags &= ~FS_DOSOFTDEP;
warnx("%s cleared", name);
}
}
}
if (oflag) {
name = "optimization preference";
chg[FS_OPTSPACE] = "space";
chg[FS_OPTTIME] = "time";
if (sblock.fs_optim == ovalue)
warnx("%s remains unchanged as %s", name, chg[ovalue]);
else {
warnx("%s changes from %s to %s",
name, chg[sblock.fs_optim], chg[ovalue]);
sblock.fs_optim = ovalue;
if (sblock.fs_minfree >= MINFREE &&
ovalue == FS_OPTSPACE)
warnx(OPTWARN, "time", ">=", MINFREE);
if (sblock.fs_minfree < MINFREE && ovalue == FS_OPTTIME)
warnx(OPTWARN, "space", "<", MINFREE);
}
}
if (sflag) {
name = "expected number of files per directory";
if (sblock.fs_avgfpdir == (unsigned)svalue) {
warnx("%s remains unchanged as %d", name, svalue);
}
else {
warnx("%s changes from %d to %d",
name, sblock.fs_avgfpdir, svalue);
sblock.fs_avgfpdir = svalue;
}
}
if (tflag) {
name = "issue TRIM to the disk";
if (strcmp(tvalue, "enable") == 0) {
if (sblock.fs_flags & FS_TRIM)
warnx("%s remains unchanged as enabled", name);
else {
sblock.fs_flags |= FS_TRIM;
warnx("%s set", name);
}
} else if (strcmp(tvalue, "disable") == 0) {
if ((~sblock.fs_flags & FS_TRIM) == FS_TRIM)
warnx("%s remains unchanged as disabled", name);
else {
sblock.fs_flags &= ~FS_TRIM;
warnx("%s cleared", name);
}
}
}
if (sbwrite(&disk, Aflag) == -1)
goto err;
ufs_disk_close(&disk);
if (active) {
build_iovec_argf(&iov, &iovlen, "fstype", "ufs");
build_iovec_argf(&iov, &iovlen, "fspath", "%s", on);
build_iovec(&iov, &iovlen, "errmsg", errmsg, sizeof(errmsg));
if (nmount(iov, iovlen,
stfs.f_flags | MNT_UPDATE | MNT_RELOAD) < 0) {
if (errmsg[0])
err(9, "%s: reload: %s", special, errmsg);
else
err(9, "%s: reload", special);
}
warnx("file system reloaded");
}
exit(0);
err:
if (disk.d_error != NULL)
errx(11, "%s: %s", special, disk.d_error);
else
err(12, "%s", special);
}
static void
sbdirty(void)
{
disk.d_fs.fs_flags |= FS_UNCLEAN | FS_NEEDSFSCK;
disk.d_fs.fs_clean = 0;
}
static ufs2_daddr_t
journal_balloc(void)
{
ufs2_daddr_t blk;
struct cg *cgp;
int valid;
static int contig = 1;
cgp = &disk.d_cg;
for (;;) {
blk = cgballoc(&disk);
if (blk > 0)
break;
/*
* If we failed to allocate a block from this cg, move to
* the next.
*/
if (cgwrite(&disk) < 0) {
warn("Failed to write updated cg");
return (-1);
}
while ((valid = cgread(&disk)) == 1) {
/*
* Try to minimize fragmentation by requiring a minimum
* number of blocks present.
*/
if (cgp->cg_cs.cs_nbfree > 256 * 1024)
break;
if (contig == 0 && cgp->cg_cs.cs_nbfree)
break;
}
if (valid)
continue;
/*
* Try once through looking only for large contiguous regions
* and again taking any space we can find.
*/
if (contig) {
contig = 0;
disk.d_ccg = 0;
warnx("Journal file fragmented.");
continue;
}
warnx("Failed to find sufficient free blocks for the journal");
return -1;
}
if (bwrite(&disk, fsbtodb(&sblock, blk), clrbuf,
sblock.fs_bsize) <= 0) {
warn("Failed to initialize new block");
return -1;
}
return (blk);
}
/*
* Search a directory block for the SUJ_FILE.
*/
static ino_t
dir_search(ufs2_daddr_t blk, int bytes)
{
char block[MAXBSIZE];
struct direct *dp;
int off;
if (bread(&disk, fsbtodb(&sblock, blk), block, bytes) <= 0) {
warn("Failed to read dir block");
return (-1);
}
for (off = 0; off < bytes; off += dp->d_reclen) {
dp = (struct direct *)&block[off];
if (dp->d_reclen == 0)
break;
if (dp->d_ino == 0)
continue;
if (dp->d_namlen != strlen(SUJ_FILE))
continue;
if (bcmp(dp->d_name, SUJ_FILE, dp->d_namlen) != 0)
continue;
return (dp->d_ino);
}
return (0);
}
/*
* Search in the UFS_ROOTINO for the SUJ_FILE. If it exists we can not enable
* journaling.
*/
static ino_t
journal_findfile(void)
{
union dinodep dp;
ino_t ino;
int i;
if (getinode(&disk, &dp, UFS_ROOTINO) != 0) {
warn("Failed to get root inode: %s", disk.d_error);
return (-1);
}
if (sblock.fs_magic == FS_UFS1_MAGIC) {
if ((off_t)dp.dp1->di_size >= lblktosize(&sblock, UFS_NDADDR)) {
warnx("UFS_ROOTINO extends beyond direct blocks.");
return (-1);
}
for (i = 0; i < UFS_NDADDR; i++) {
if (dp.dp1->di_db[i] == 0)
break;
if ((ino = dir_search(dp.dp1->di_db[i],
sblksize(&sblock, (off_t)dp.dp1->di_size, i))) != 0)
return (ino);
}
} else {
if ((off_t)dp.dp2->di_size >= lblktosize(&sblock, UFS_NDADDR)) {
warnx("UFS_ROOTINO extends beyond direct blocks.");
return (-1);
}
for (i = 0; i < UFS_NDADDR; i++) {
if (dp.dp2->di_db[i] == 0)
break;
if ((ino = dir_search(dp.dp2->di_db[i],
sblksize(&sblock, (off_t)dp.dp2->di_size, i))) != 0)
return (ino);
}
}
return (0);
}
static void
dir_clear_block(const char *block, off_t off)
{
struct direct *dp;
for (; off < sblock.fs_bsize; off += DIRBLKSIZ) {
dp = (struct direct *)&block[off];
dp->d_ino = 0;
dp->d_reclen = DIRBLKSIZ;
dp->d_type = DT_UNKNOWN;
}
}
/*
* Insert the journal at inode 'ino' into directory blk 'blk' at the first
* free offset of 'off'. DIRBLKSIZ blocks after off are initialized as
* empty.
*/
static int
dir_insert(ufs2_daddr_t blk, off_t off, ino_t ino)
{
struct direct *dp;
char block[MAXBSIZE];
if (bread(&disk, fsbtodb(&sblock, blk), block, sblock.fs_bsize) <= 0) {
warn("Failed to read dir block");
return (-1);
}
bzero(&block[off], sblock.fs_bsize - off);
dp = (struct direct *)&block[off];
dp->d_ino = ino;
dp->d_reclen = DIRBLKSIZ;
dp->d_type = DT_REG;
dp->d_namlen = strlen(SUJ_FILE);
bcopy(SUJ_FILE, &dp->d_name, strlen(SUJ_FILE));
dir_clear_block(block, off + DIRBLKSIZ);
if (bwrite(&disk, fsbtodb(&sblock, blk), block, sblock.fs_bsize) <= 0) {
warn("Failed to write dir block");
return (-1);
}
return (0);
}
/*
* Extend a directory block in 'blk' by copying it to a full size block
* and inserting the new journal inode into .sujournal.
*/
static int
dir_extend(ufs2_daddr_t blk, ufs2_daddr_t nblk, off_t size, ino_t ino)
{
char block[MAXBSIZE];
if (bread(&disk, fsbtodb(&sblock, blk), block,
roundup(size, sblock.fs_fsize)) <= 0) {
warn("Failed to read dir block");
return (-1);
}
dir_clear_block(block, size);
if (bwrite(&disk, fsbtodb(&sblock, nblk), block, sblock.fs_bsize)
<= 0) {
warn("Failed to write dir block");
return (-1);
}
return (dir_insert(nblk, size, ino));
}
/*
* Insert the journal file into the UFS_ROOTINO directory. We always extend the
* last frag
*/
static int
journal_insertfile(ino_t ino)
{
union dinodep dp;
ufs2_daddr_t nblk;
ufs2_daddr_t blk;
ufs_lbn_t lbn;
int size;
int off;
if (getinode(&disk, &dp, UFS_ROOTINO) != 0) {
warn("Failed to get root inode: %s", disk.d_error);
sbdirty();
return (-1);
}
blk = 0;
size = 0;
nblk = journal_balloc();
if (nblk <= 0)
return (-1);
/*
* For simplicity sake we aways extend the UFS_ROOTINO into a new
* directory block rather than searching for space and inserting
* into an existing block. However, if the rootino has frags
* have to free them and extend the block.
*/
if (sblock.fs_magic == FS_UFS1_MAGIC) {
lbn = lblkno(&sblock, dp.dp1->di_size);
off = blkoff(&sblock, dp.dp1->di_size);
blk = dp.dp1->di_db[lbn];
size = sblksize(&sblock, (off_t)dp.dp1->di_size, lbn);
} else {
lbn = lblkno(&sblock, dp.dp2->di_size);
off = blkoff(&sblock, dp.dp2->di_size);
blk = dp.dp2->di_db[lbn];
size = sblksize(&sblock, (off_t)dp.dp2->di_size, lbn);
}
if (off != 0) {
if (dir_extend(blk, nblk, off, ino) == -1)
return (-1);
} else {
blk = 0;
if (dir_insert(nblk, 0, ino) == -1)
return (-1);
}
if (sblock.fs_magic == FS_UFS1_MAGIC) {
dp.dp1->di_blocks += (sblock.fs_bsize - size) / DEV_BSIZE;
dp.dp1->di_db[lbn] = nblk;
dp.dp1->di_size = lblktosize(&sblock, lbn+1);
} else {
dp.dp2->di_blocks += (sblock.fs_bsize - size) / DEV_BSIZE;
dp.dp2->di_db[lbn] = nblk;
dp.dp2->di_size = lblktosize(&sblock, lbn+1);
}
if (putinode(&disk) < 0) {
warn("Failed to write root inode: %s", disk.d_error);
return (-1);
}
if (cgwrite(&disk) < 0) {
warn("Failed to write updated cg");
sbdirty();
return (-1);
}
if (blk) {
if (cgbfree(&disk, blk, size) < 0) {
warn("Failed to write cg");
return (-1);
}
}
return (0);
}
static int
indir_fill(ufs2_daddr_t blk, int level, int *resid)
{
char indirbuf[MAXBSIZE];
ufs1_daddr_t *bap1;
ufs2_daddr_t *bap2;
ufs2_daddr_t nblk;
int ncnt;
int cnt;
int i;
bzero(indirbuf, sizeof(indirbuf));
bap1 = (ufs1_daddr_t *)indirbuf;
bap2 = (void *)bap1;
cnt = 0;
for (i = 0; i < NINDIR(&sblock) && *resid != 0; i++) {
nblk = journal_balloc();
if (nblk <= 0)
return (-1);
cnt++;
if (sblock.fs_magic == FS_UFS1_MAGIC)
*bap1++ = nblk;
else
*bap2++ = nblk;
if (level != 0) {
ncnt = indir_fill(nblk, level - 1, resid);
if (ncnt <= 0)
return (-1);
cnt += ncnt;
} else
(*resid)--;
}
if (bwrite(&disk, fsbtodb(&sblock, blk), indirbuf,
sblock.fs_bsize) <= 0) {
warn("Failed to write indirect");
return (-1);
}
return (cnt);
}
/*
* Clear the flag bits so the journal can be removed.
*/
static void
journal_clear(void)
{
union dinodep dp;
ino_t ino;
ino = journal_findfile();
if (ino == (ino_t)-1 || ino == 0) {
warnx("Journal file does not exist");
return;
}
printf("Clearing journal flags from inode %ju\n", (uintmax_t)ino);
if (getinode(&disk, &dp, ino) != 0) {
warn("Failed to get journal inode: %s", disk.d_error);
return;
}
if (sblock.fs_magic == FS_UFS1_MAGIC)
dp.dp1->di_flags = 0;
else
dp.dp2->di_flags = 0;
if (putinode(&disk) < 0) {
warn("Failed to write journal inode: %s", disk.d_error);
return;
}
}
static int
journal_alloc(int64_t size)
{
union dinodep dp;
ufs2_daddr_t blk;
struct cg *cgp;
int resid;
ino_t ino;
int blks;
time_t utime;
int i;
cgp = &disk.d_cg;
ino = 0;
/*
* If the journal file exists we can't allocate it.
*/
ino = journal_findfile();
if (ino == (ino_t)-1) {
warnx("journal_findfile() failed.");
return (-1);
}
if (ino > 0) {
warnx("Journal file %s already exists, please remove.",
SUJ_FILE);
return (-1);
}
/*
* If the user didn't supply a size pick one based on the filesystem
* size constrained with hardcoded MIN and MAX values. We opt for
* 1/1024th of the filesystem up to MAX but not exceeding one CG and
* not less than the MIN.
*/
if (size == 0) {
size = (sblock.fs_size * sblock.fs_bsize) / 1024;
size = MIN(SUJ_MAX, size);
if (size / sblock.fs_fsize > sblock.fs_fpg)
size = sblock.fs_fpg * sblock.fs_fsize;
size = MAX(SUJ_MIN, size);
}
/* fsck does not support fragments in journal files. */
size = roundup(size, sblock.fs_bsize);
resid = blocks = size / sblock.fs_bsize;
if (sblock.fs_cstotal.cs_nbfree < blocks) {
warn("Insufficient free space for %jd byte journal", size);
return (-1);
}
/*
* Find a cg with enough blocks to satisfy the journal
* size. Presently the journal does not span cgs.
*/
while (cgread(&disk) == 1) {
if (cgp->cg_cs.cs_nifree == 0)
continue;
ino = cgialloc(&disk);
if (ino <= 0)
break;
printf("Using inode %ju in cg %d for %jd byte journal\n",
(uintmax_t)ino, cgp->cg_cgx, size);
if (getinode(&disk, &dp, ino) != 0) {
warn("Failed to get allocated inode: %s", disk.d_error);
sbdirty();
goto out;
}
/*
* We leave fields unrelated to the number of allocated
* blocks and size uninitialized. This causes legacy
* fsck implementations to clear the inode.
*/
time(&utime);
if (sblock.fs_magic == FS_UFS1_MAGIC) {
bzero(dp.dp1, sizeof(*dp.dp1));
dp.dp1->di_size = size;
dp.dp1->di_mode = IFREG | IREAD;
dp.dp1->di_nlink = 1;
dp.dp1->di_flags =
SF_IMMUTABLE | SF_NOUNLINK | UF_NODUMP;
dp.dp1->di_atime = utime;
dp.dp1->di_mtime = utime;
dp.dp1->di_ctime = utime;
} else {
bzero(dp.dp2, sizeof(*dp.dp2));
dp.dp2->di_size = size;
dp.dp2->di_mode = IFREG | IREAD;
dp.dp2->di_nlink = 1;
dp.dp2->di_flags =
SF_IMMUTABLE | SF_NOUNLINK | UF_NODUMP;
dp.dp2->di_atime = utime;
dp.dp2->di_mtime = utime;
dp.dp2->di_ctime = utime;
dp.dp2->di_birthtime = utime;
}
for (i = 0; i < UFS_NDADDR && resid; i++, resid--) {
blk = journal_balloc();
if (blk <= 0)
goto out;
if (sblock.fs_magic == FS_UFS1_MAGIC) {
dp.dp1->di_db[i] = blk;
dp.dp1->di_blocks++;
} else {
dp.dp2->di_db[i] = blk;
dp.dp2->di_blocks++;
}
}
for (i = 0; i < UFS_NIADDR && resid; i++) {
blk = journal_balloc();
if (blk <= 0)
goto out;
blks = indir_fill(blk, i, &resid) + 1;
if (blks <= 0) {
sbdirty();
goto out;
}
if (sblock.fs_magic == FS_UFS1_MAGIC) {
dp.dp1->di_ib[i] = blk;
dp.dp1->di_blocks += blks;
} else {
dp.dp2->di_ib[i] = blk;
dp.dp2->di_blocks += blks;
}
}
if (sblock.fs_magic == FS_UFS1_MAGIC)
dp.dp1->di_blocks *= sblock.fs_bsize / disk.d_bsize;
else
dp.dp2->di_blocks *= sblock.fs_bsize / disk.d_bsize;
if (putinode(&disk) < 0) {
warn("Failed to write allocated inode: %s",
disk.d_error);
sbdirty();
return (-1);
}
if (cgwrite(&disk) < 0) {
warn("Failed to write updated cg");
sbdirty();
return (-1);
}
if (journal_insertfile(ino) < 0) {
sbdirty();
return (-1);
}
sblock.fs_sujfree = 0;
return (0);
}
warnx("Insufficient free space for the journal.");
out:
return (-1);
}
static void
usage(void)
{
fprintf(stderr, "%s\n%s\n%s\n%s\n%s\n%s\n",
"usage: tunefs [-A] [-a enable | disable] [-e maxbpg] [-f avgfilesize]",
" [-J enable | disable] [-j enable | disable] [-k metaspace]",
" [-L volname] [-l enable | disable] [-m minfree]",
" [-N enable | disable] [-n enable | disable]",
" [-o space | time] [-p] [-s avgfpdir] [-t enable | disable]",
" special | filesystem");
exit(2);
}
static void
printfs(void)
{
warnx("POSIX.1e ACLs: (-a) %s",
(sblock.fs_flags & FS_ACLS)? "enabled" : "disabled");
warnx("NFSv4 ACLs: (-N) %s",
(sblock.fs_flags & FS_NFS4ACLS)? "enabled" : "disabled");
warnx("MAC multilabel: (-l) %s",
(sblock.fs_flags & FS_MULTILABEL)? "enabled" : "disabled");
warnx("soft updates: (-n) %s",
(sblock.fs_flags & FS_DOSOFTDEP)? "enabled" : "disabled");
warnx("soft update journaling: (-j) %s",
(sblock.fs_flags & FS_SUJ)? "enabled" : "disabled");
warnx("gjournal: (-J) %s",
(sblock.fs_flags & FS_GJOURNAL)? "enabled" : "disabled");
warnx("trim: (-t) %s",
(sblock.fs_flags & FS_TRIM)? "enabled" : "disabled");
warnx("maximum blocks per file in a cylinder group: (-e) %d",
sblock.fs_maxbpg);
warnx("average file size: (-f) %d",
sblock.fs_avgfilesize);
warnx("average number of files in a directory: (-s) %d",
sblock.fs_avgfpdir);
warnx("minimum percentage of free space: (-m) %d%%",
sblock.fs_minfree);
warnx("space to hold for metadata blocks: (-k) %jd",
sblock.fs_metaspace);
warnx("optimization preference: (-o) %s",
sblock.fs_optim == FS_OPTSPACE ? "space" : "time");
if (sblock.fs_minfree >= MINFREE &&
sblock.fs_optim == FS_OPTSPACE)
warnx(OPTWARN, "time", ">=", MINFREE);
if (sblock.fs_minfree < MINFREE &&
sblock.fs_optim == FS_OPTTIME)
warnx(OPTWARN, "space", "<", MINFREE);
warnx("volume label: (-L) %s",
sblock.fs_volname);
}
diff --git a/sys/fs/devfs/devfs.h b/sys/fs/devfs/devfs.h
index aef291601289..b62438ff6730 100644
--- a/sys/fs/devfs/devfs.h
+++ b/sys/fs/devfs/devfs.h
@@ -1,213 +1,220 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 1992, 1993
* The Regents of the University of California. All rights reserved.
* Copyright (c) 2000
* Poul-Henning Kamp. All rights reserved.
* Copyright (c) 2002
* Dima Dorfman. All rights reserved.
*
* This code is derived from software donated to Berkeley by
* Jan-Simon Pendry.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)kernfs.h 8.6 (Berkeley) 3/29/95
* From: FreeBSD: src/sys/miscfs/kernfs/kernfs.h 1.14
*
* $FreeBSD$
*/
#ifndef _FS_DEVFS_DEVFS_H_
#define _FS_DEVFS_DEVFS_H_
#define DEVFS_MAGIC 0xdb0a087a
/*
* Identifiers. The ruleset and rule numbers are 16-bit values. The
* "rule ID" is a combination of the ruleset and rule number; it
* should be able to univocally describe a rule in the system. In
* this implementation, the upper 16 bits of the rule ID is the
* ruleset number; the lower 16 bits, the rule number within the
* aforementioned ruleset.
*/
typedef uint16_t devfs_rnum;
typedef uint16_t devfs_rsnum;
typedef uint32_t devfs_rid;
/*
* Identifier manipulators.
*/
#define rid2rsn(rid) ((rid) >> 16)
#define rid2rn(rid) ((rid) & 0xffff)
#define mkrid(rsn, rn) ((rn) | ((rsn) << 16))
/*
* Plain DEVFS rule. This gets shared between kernel and userland
* verbatim, so it shouldn't contain any pointers or other kernel- or
* userland-specific values.
*/
struct devfs_rule {
uint32_t dr_magic; /* Magic number. */
devfs_rid dr_id; /* Identifier. */
/*
* Conditions under which this rule should be applied. These
* are ANDed together since OR can be simulated by using
* multiple rules. dr_icond determines which of the other
* variables we should process.
*/
int dr_icond;
#define DRC_DSWFLAGS 0x001
#define DRC_PATHPTRN 0x002
int dr_dswflags; /* cdevsw flags to match. */
#define DEVFS_MAXPTRNLEN 200
char dr_pathptrn[DEVFS_MAXPTRNLEN]; /* Pattern to match path. */
/*
* Things to change. dr_iacts determines which of the other
* variables we should process.
*/
int dr_iacts;
#define DRA_BACTS 0x001
#define DRA_UID 0x002
#define DRA_GID 0x004
#define DRA_MODE 0x008
#define DRA_INCSET 0x010
int dr_bacts; /* Boolean (on/off) action. */
#define DRB_HIDE 0x001 /* Hide entry (DE_WHITEOUT). */
#define DRB_UNHIDE 0x002 /* Unhide entry. */
uid_t dr_uid;
gid_t dr_gid;
mode_t dr_mode;
devfs_rsnum dr_incset; /* Included ruleset. */
};
/*
* Rule-related ioctls.
*/
#define DEVFSIO_RADD _IOWR('D', 0, struct devfs_rule)
#define DEVFSIO_RDEL _IOW('D', 1, devfs_rid)
#define DEVFSIO_RAPPLY _IOW('D', 2, struct devfs_rule)
#define DEVFSIO_RAPPLYID _IOW('D', 3, devfs_rid)
#define DEVFSIO_RGETNEXT _IOWR('D', 4, struct devfs_rule)
#define DEVFSIO_SUSE _IOW('D', 10, devfs_rsnum)
#define DEVFSIO_SAPPLY _IOW('D', 11, devfs_rsnum)
#define DEVFSIO_SGETNEXT _IOWR('D', 12, devfs_rsnum)
/* XXX: DEVFSIO_RS_GET_INFO for refcount, active if any, etc. */
#ifdef _KERNEL
#ifdef MALLOC_DECLARE
MALLOC_DECLARE(M_DEVFS);
#endif
+#endif /* _KERNEL */
+
struct componentname;
TAILQ_HEAD(devfs_dlist_head, devfs_dirent);
struct devfs_dirent {
struct cdev_priv *de_cdp;
int de_inode;
int de_flags;
#define DE_WHITEOUT 0x01
#define DE_DOT 0x02
#define DE_DOTDOT 0x04
#define DE_DOOMED 0x08
#define DE_COVERED 0x10
#define DE_USER 0x20
int de_holdcnt;
struct dirent *de_dirent;
TAILQ_ENTRY(devfs_dirent) de_list;
struct devfs_dlist_head de_dlist;
struct devfs_dirent *de_dir;
int de_links;
mode_t de_mode;
uid_t de_uid;
gid_t de_gid;
struct label *de_label;
struct timespec de_atime;
struct timespec de_mtime;
struct timespec de_ctime;
struct vnode *de_vnode;
char *de_symlink;
int de_usecount;
};
+#include <sys/_lock.h>
+#include <sys/_sx.h>
+
struct devfs_mount {
u_int dm_idx;
struct mount *dm_mount;
struct devfs_dirent *dm_rootdir;
unsigned dm_generation;
int dm_holdcnt;
struct sx dm_lock;
devfs_rsnum dm_ruleset;
};
#define DEVFS_ROOTINO 2
+#ifdef _KERNEL
+
extern unsigned devfs_rule_depth;
#define VFSTODEVFS(mp) ((struct devfs_mount *)((mp)->mnt_data))
#define DEVFS_DE_HOLD(de) ((de)->de_holdcnt++)
#define DEVFS_DE_DROP(de) (--(de)->de_holdcnt == 0)
#define DEVFS_DMP_HOLD(dmp) ((dmp)->dm_holdcnt++)
#define DEVFS_DMP_DROP(dmp) (--(dmp)->dm_holdcnt == 0)
#define DEVFS_DEL_VNLOCKED 0x01
#define DEVFS_DEL_NORECURSE 0x02
void devfs_rules_apply(struct devfs_mount *, struct devfs_dirent *);
void devfs_rules_cleanup(struct devfs_mount *);
int devfs_rules_ioctl(struct devfs_mount *, u_long, caddr_t,
struct thread *);
void devfs_ruleset_set(devfs_rsnum rsnum, struct devfs_mount *dm);
void devfs_ruleset_apply(struct devfs_mount *dm);
int devfs_allocv(struct devfs_dirent *, struct mount *, int,
struct vnode **);
char *devfs_fqpn(char *, struct devfs_mount *, struct devfs_dirent *,
struct componentname *);
void devfs_delete(struct devfs_mount *, struct devfs_dirent *, int);
void devfs_dirent_free(struct devfs_dirent *);
int devfs_populate_needed(struct devfs_mount *dm);
void devfs_populate(struct devfs_mount *);
void devfs_cleanup(struct devfs_mount *);
void devfs_unmount_final(struct devfs_mount *);
struct devfs_dirent *devfs_newdirent(char *, int);
struct devfs_dirent *devfs_parent_dirent(struct devfs_dirent *);
struct devfs_dirent *devfs_vmkdir(struct devfs_mount *, char *, int,
struct devfs_dirent *, u_int);
struct devfs_dirent *devfs_find(struct devfs_dirent *, const char *, int,
int);
void devfs_ctty_ref(struct vnode *);
void devfs_ctty_unref(struct vnode *);
int devfs_usecount(struct vnode *);
#endif /* _KERNEL */
#endif /* !_FS_DEVFS_DEVFS_H_ */
diff --git a/sys/fs/devfs/devfs_int.h b/sys/fs/devfs/devfs_int.h
index 5c3cb17eca61..e5afa311cbfd 100644
--- a/sys/fs/devfs/devfs_int.h
+++ b/sys/fs/devfs/devfs_int.h
@@ -1,103 +1,103 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2005 Poul-Henning Kamp. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD$
*/
/*
* This file documents a private interface and it SHALL only be used
* by kern/kern_conf.c and fs/devfs/...
*/
#ifndef _FS_DEVFS_DEVFS_INT_H_
#define _FS_DEVFS_DEVFS_INT_H_
#include <sys/queue.h>
-#ifdef _KERNEL
-
struct devfs_dirent;
struct devfs_mount;
struct cdev_privdata {
struct file *cdpd_fp;
void *cdpd_data;
void (*cdpd_dtr)(void *);
LIST_ENTRY(cdev_privdata) cdpd_list;
};
struct cdev_priv {
struct cdev cdp_c;
TAILQ_ENTRY(cdev_priv) cdp_list;
u_int cdp_inode;
u_int cdp_flags;
#define CDP_ACTIVE (1 << 0)
#define CDP_SCHED_DTR (1 << 1)
#define CDP_UNREF_DTR (1 << 2)
u_int cdp_inuse;
u_int cdp_maxdirent;
struct devfs_dirent **cdp_dirents;
struct devfs_dirent *cdp_dirent0;
TAILQ_ENTRY(cdev_priv) cdp_dtr_list;
void (*cdp_dtr_cb)(void *);
void *cdp_dtr_cb_arg;
LIST_HEAD(, cdev_privdata) cdp_fdpriv;
struct mtx cdp_threadlock;
};
#define cdev2priv(c) __containerof(c, struct cdev_priv, cdp_c)
+#ifdef _KERNEL
+
struct cdev *devfs_alloc(int);
int devfs_dev_exists(const char *);
void devfs_free(struct cdev *);
void devfs_create(struct cdev *);
void devfs_destroy(struct cdev *);
void devfs_destroy_cdevpriv(struct cdev_privdata *);
int devfs_dir_find(const char *);
void devfs_dir_ref_de(struct devfs_mount *, struct devfs_dirent *);
void devfs_dir_unref_de(struct devfs_mount *, struct devfs_dirent *);
int devfs_pathpath(const char *, const char *);
extern struct unrhdr *devfs_inos;
extern struct mtx devmtx;
extern struct mtx devfs_de_interlock;
extern struct sx clone_drain_lock;
extern struct mtx cdevpriv_mtx;
extern TAILQ_HEAD(cdev_priv_list, cdev_priv) cdevp_list;
#define dev_lock_assert_locked() mtx_assert(&devmtx, MA_OWNED)
#define dev_lock_assert_unlocked() mtx_assert(&devmtx, MA_NOTOWNED)
#endif /* _KERNEL */
#endif /* !_FS_DEVFS_DEVFS_INT_H_ */
diff --git a/sys/fs/msdosfs/denode.h b/sys/fs/msdosfs/denode.h
index b198a6a39d81..9a4de4acbe2e 100644
--- a/sys/fs/msdosfs/denode.h
+++ b/sys/fs/msdosfs/denode.h
@@ -1,289 +1,290 @@
/* $FreeBSD$ */
/* $NetBSD: denode.h,v 1.25 1997/11/17 15:36:28 ws Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1994, 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*-
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
* This software is provided "as is".
*
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
* October 1992
*/
#ifndef _FS_MSDOSFS_DENODE_H_
#define _FS_MSDOSFS_DENODE_H_
/*
* This is the pc filesystem specific portion of the vnode structure.
*
* To describe a file uniquely the de_dirclust, de_diroffset, and
* de_StartCluster fields are used.
*
* de_dirclust contains the cluster number of the directory cluster
* containing the entry for a file or directory.
* de_diroffset is the index into the cluster for the entry describing
* a file or directory.
* de_StartCluster is the number of the first cluster of the file or directory.
*
* Now to describe the quirks of the pc filesystem.
* - Clusters 0 and 1 are reserved.
* - The first allocatable cluster is 2.
* - The root directory is of fixed size and all blocks that make it up
* are contiguous.
* - Cluster 0 refers to the root directory when it is found in the
* startcluster field of a directory entry that points to another directory.
* - Cluster 0 implies a 0 length file when found in the start cluster field
* of a directory entry that points to a file.
* - You can't use the cluster number 0 to derive the address of the root
* directory.
* - Multiple directory entries can point to a directory. The entry in the
* parent directory points to a child directory. Any directories in the
* child directory contain a ".." entry that points back to the parent.
* The child directory itself contains a "." entry that points to itself.
* - The root directory does not contain a "." or ".." entry.
* - Directory entries for directories are never changed once they are created
* (except when removed). The size stays 0, and the last modification time
* is never changed. This is because so many directory entries can point to
* the physical clusters that make up a directory. It would lead to an
* update nightmare.
* - The length field in a directory entry pointing to a directory contains 0
* (always). The only way to find the end of a directory is to follow the
* cluster chain until the "last cluster" marker is found.
*
* My extensions to make this house of cards work. These apply only to the in
* memory copy of the directory entry.
* - A reference count for each denode will be kept since dos doesn't keep such
* things.
*/
/*
* Internal pseudo-offset for (nonexistent) directory entry for the root
* dir in the root dir
*/
#define MSDOSFSROOT_OFS 0x1fffffff
/*
* The FAT cache structure. fc_fsrcn is the filesystem relative cluster
* number that corresponds to the file relative cluster number in this
* structure (fc_frcn).
*/
struct fatcache {
u_long fc_frcn; /* file relative cluster number */
u_long fc_fsrcn; /* filesystem relative cluster number */
};
/*
* The FAT entry cache as it stands helps make extending files a "quick"
* operation by avoiding having to scan the FAT to discover the last
* cluster of the file. The cache also helps sequential reads by
* remembering the last cluster read from the file. This also prevents us
* from having to rescan the FAT to find the next cluster to read. This
* cache is probably pretty worthless if a file is opened by multiple
* processes.
*/
#define FC_SIZE 3 /* number of entries in the cache */
#define FC_LASTMAP 0 /* entry the last call to pcbmap() resolved
* to */
#define FC_LASTFC 1 /* entry for the last cluster in the file */
#define FC_NEXTTOLASTFC 2 /* entry for a close to the last cluster in
* the file */
#define FCE_EMPTY 0xffffffff /* doesn't represent an actual cluster # */
/*
* Set a slot in the FAT cache.
*/
#define fc_setcache(dep, slot, frcn, fsrcn) \
(dep)->de_fc[(slot)].fc_frcn = (frcn); \
(dep)->de_fc[(slot)].fc_fsrcn = (fsrcn);
/*
* This is the in memory variant of a dos directory entry. It is usually
* contained within a vnode.
*/
struct denode {
struct vnode *de_vnode; /* addr of vnode we are part of */
u_long de_flag; /* flag bits */
u_long de_dirclust; /* cluster of the directory file containing this entry */
u_long de_diroffset; /* offset of this entry in the directory cluster */
u_long de_fndoffset; /* offset of found dir entry */
int de_fndcnt; /* number of slots before de_fndoffset */
long de_refcnt; /* reference count */
struct msdosfsmount *de_pmp; /* addr of our mount struct */
u_char de_Name[12]; /* name, from DOS directory entry */
u_char de_Attributes; /* attributes, from directory entry */
u_char de_LowerCase; /* NT VFAT lower case flags */
u_char de_CHun; /* Hundredth of second of CTime*/
u_short de_CTime; /* creation time */
u_short de_CDate; /* creation date */
u_short de_ADate; /* access date */
u_short de_MTime; /* modification time */
u_short de_MDate; /* modification date */
u_long de_StartCluster; /* starting cluster of file */
u_long de_FileSize; /* size of file in bytes */
struct fatcache de_fc[FC_SIZE]; /* FAT cache */
u_quad_t de_modrev; /* Revision level for lease. */
uint64_t de_inode; /* Inode number (really byte offset of direntry) */
};
/*
* Values for the de_flag field of the denode.
*/
#define DE_UPDATE 0x0004 /* Modification time update request */
#define DE_CREATE 0x0008 /* Creation time update */
#define DE_ACCESS 0x0010 /* Access time update */
#define DE_MODIFIED 0x0020 /* Denode has been modified */
#define DE_RENAME 0x0040 /* Denode is in the process of being renamed */
/* Maximum size of a file on a FAT filesystem */
#define MSDOSFS_FILESIZE_MAX 0xFFFFFFFFLL
/*
* Transfer directory entries between internal and external form.
* dep is a struct denode * (internal form),
* dp is a struct direntry * (external form).
*/
#define DE_INTERNALIZE32(dep, dp) \
((dep)->de_StartCluster |= getushort((dp)->deHighClust) << 16)
#define DE_INTERNALIZE(dep, dp) \
(memcpy((dep)->de_Name, (dp)->deName, 11), \
(dep)->de_Attributes = (dp)->deAttributes, \
(dep)->de_LowerCase = (dp)->deLowerCase, \
(dep)->de_CHun = (dp)->deCHundredth, \
(dep)->de_CTime = getushort((dp)->deCTime), \
(dep)->de_CDate = getushort((dp)->deCDate), \
(dep)->de_ADate = getushort((dp)->deADate), \
(dep)->de_MTime = getushort((dp)->deMTime), \
(dep)->de_MDate = getushort((dp)->deMDate), \
(dep)->de_StartCluster = getushort((dp)->deStartCluster), \
(dep)->de_FileSize = getulong((dp)->deFileSize), \
(FAT32((dep)->de_pmp) ? DE_INTERNALIZE32((dep), (dp)) : 0))
#define DE_EXTERNALIZE(dp, dep) \
(memcpy((dp)->deName, (dep)->de_Name, 11), \
(dp)->deAttributes = (dep)->de_Attributes, \
(dp)->deLowerCase = (dep)->de_LowerCase, \
(dp)->deCHundredth = (dep)->de_CHun, \
putushort((dp)->deCTime, (dep)->de_CTime), \
putushort((dp)->deCDate, (dep)->de_CDate), \
putushort((dp)->deADate, (dep)->de_ADate), \
putushort((dp)->deMTime, (dep)->de_MTime), \
putushort((dp)->deMDate, (dep)->de_MDate), \
putushort((dp)->deStartCluster, (dep)->de_StartCluster), \
putulong((dp)->deFileSize, \
((dep)->de_Attributes & ATTR_DIRECTORY) ? 0 : (dep)->de_FileSize), \
putushort((dp)->deHighClust, (dep)->de_StartCluster >> 16))
#if defined(_KERNEL) || defined(MAKEFS)
#define VTODE(vp) ((struct denode *)(vp)->v_data)
#define DETOV(de) ((de)->de_vnode)
#define DETIMES(dep, acc, mod, cre) do { \
if ((dep)->de_flag & DE_UPDATE) { \
(dep)->de_flag |= DE_MODIFIED; \
timespec2fattime((mod), 0, &(dep)->de_MDate, \
&(dep)->de_MTime, NULL); \
(dep)->de_Attributes |= ATTR_ARCHIVE; \
} \
if ((dep)->de_pmp->pm_flags & MSDOSFSMNT_NOWIN95) { \
(dep)->de_flag &= ~(DE_UPDATE | DE_CREATE | DE_ACCESS); \
break; \
} \
if ((dep)->de_flag & DE_ACCESS) { \
uint16_t adate; \
\
timespec2fattime((acc), 0, &adate, NULL, NULL); \
if (adate != (dep)->de_ADate) { \
(dep)->de_flag |= DE_MODIFIED; \
(dep)->de_ADate = adate; \
} \
} \
if ((dep)->de_flag & DE_CREATE) { \
timespec2fattime((cre), 0, &(dep)->de_CDate, \
&(dep)->de_CTime, &(dep)->de_CHun); \
(dep)->de_flag |= DE_MODIFIED; \
} \
(dep)->de_flag &= ~(DE_UPDATE | DE_CREATE | DE_ACCESS); \
} while (0)
/*
* This overlays the fid structure (see mount.h)
*/
struct defid {
u_short defid_len; /* length of structure */
u_short defid_pad; /* force long alignment */
uint32_t defid_dirclust; /* cluster this dir entry came from */
uint32_t defid_dirofs; /* offset of entry within the cluster */
#if 0
uint32_t defid_gen; /* generation number */
#endif
};
extern struct vop_vector msdosfs_vnodeops;
#ifdef _KERNEL
int msdosfs_lookup(struct vop_cachedlookup_args *);
int msdosfs_inactive(struct vop_inactive_args *);
int msdosfs_reclaim(struct vop_reclaim_args *);
#endif
/*
* Internal service routine prototypes.
*/
+struct componentname;
int deget(struct msdosfsmount *, u_long, u_long, struct denode **);
int uniqdosname(struct denode *, struct componentname *, u_char *);
int readep(struct msdosfsmount *pmp, u_long dirclu, u_long dirofs, struct buf **bpp, struct direntry **epp);
int readde(struct denode *dep, struct buf **bpp, struct direntry **epp);
int deextend(struct denode *dep, u_long length, struct ucred *cred);
int fillinusemap(struct msdosfsmount *pmp);
void reinsert(struct denode *dep);
int dosdirempty(struct denode *dep);
int createde(struct denode *dep, struct denode *ddep, struct denode **depp, struct componentname *cnp);
int deupdat(struct denode *dep, int waitfor);
int removede(struct denode *pdep, struct denode *dep);
int detrunc(struct denode *dep, u_long length, int flags, struct ucred *cred);
int doscheckpath( struct denode *source, struct denode *target);
#endif /* _KERNEL || MAKEFS */
#endif /* !_FS_MSDOSFS_DENODE_H_ */
diff --git a/sys/sys/buf.h b/sys/sys/buf.h
index 2997560b9ab3..aacad3a057d2 100644
--- a/sys/sys/buf.h
+++ b/sys/sys/buf.h
@@ -1,605 +1,606 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)buf.h 8.9 (Berkeley) 3/30/95
* $FreeBSD$
*/
#ifndef _SYS_BUF_H_
#define _SYS_BUF_H_
#include <sys/bufobj.h>
#include <sys/queue.h>
#include <sys/lock.h>
#include <sys/lockmgr.h>
#include <vm/uma.h>
struct bio;
struct buf;
struct bufobj;
struct mount;
struct vnode;
struct uio;
/*
* To avoid including <ufs/ffs/softdep.h>
*/
LIST_HEAD(workhead, worklist);
/*
* These are currently used only by the soft dependency code, hence
* are stored once in a global variable. If other subsystems wanted
* to use these hooks, a pointer to a set of bio_ops could be added
* to each buffer.
*/
extern struct bio_ops {
void (*io_start)(struct buf *);
void (*io_complete)(struct buf *);
void (*io_deallocate)(struct buf *);
int (*io_countdeps)(struct buf *, int);
} bioops;
struct vm_object;
struct vm_page;
typedef uint32_t b_xflags_t;
/*
* The buffer header describes an I/O operation in the kernel.
*
* NOTES:
* b_bufsize, b_bcount. b_bufsize is the allocation size of the
* buffer, either DEV_BSIZE or PAGE_SIZE aligned. b_bcount is the
* originally requested buffer size and can serve as a bounds check
* against EOF. For most, but not all uses, b_bcount == b_bufsize.
*
* b_dirtyoff, b_dirtyend. Buffers support piecemeal, unaligned
* ranges of dirty data that need to be written to backing store.
* The range is typically clipped at b_bcount ( not b_bufsize ).
*
* b_resid. Number of bytes remaining in I/O. After an I/O operation
* completes, b_resid is usually 0 indicating 100% success.
*
* All fields are protected by the buffer lock except those marked:
* V - Protected by owning bufobj lock
* Q - Protected by the buf queue lock
* D - Protected by an dependency implementation specific lock
*/
struct buf {
struct bufobj *b_bufobj;
long b_bcount;
void *b_caller1;
caddr_t b_data;
int b_error;
uint16_t b_iocmd; /* BIO_* bio_cmd from bio.h */
uint16_t b_ioflags; /* BIO_* bio_flags from bio.h */
off_t b_iooffset;
long b_resid;
void (*b_iodone)(struct buf *);
void (*b_ckhashcalc)(struct buf *);
uint64_t b_ckhash; /* B_CKHASH requested check-hash */
daddr_t b_blkno; /* Underlying physical block number. */
off_t b_offset; /* Offset into file. */
TAILQ_ENTRY(buf) b_bobufs; /* (V) Buffer's associated vnode. */
uint32_t b_vflags; /* (V) BV_* flags */
uint8_t b_qindex; /* (Q) buffer queue index */
uint8_t b_domain; /* (Q) buf domain this resides in */
uint16_t b_subqueue; /* (Q) per-cpu q if any */
uint32_t b_flags; /* B_* flags. */
b_xflags_t b_xflags; /* extra flags */
struct lock b_lock; /* Buffer lock */
long b_bufsize; /* Allocated buffer size. */
int b_runningbufspace; /* when I/O is running, pipelining */
int b_kvasize; /* size of kva for buffer */
int b_dirtyoff; /* Offset in buffer of dirty region. */
int b_dirtyend; /* Offset of end of dirty region. */
caddr_t b_kvabase; /* base kva for buffer */
daddr_t b_lblkno; /* Logical block number. */
struct vnode *b_vp; /* Device vnode. */
struct ucred *b_rcred; /* Read credentials reference. */
struct ucred *b_wcred; /* Write credentials reference. */
union {
TAILQ_ENTRY(buf) b_freelist; /* (Q) */
struct {
- void (*b_pgiodone)(void *, vm_page_t *, int, int);
+ void (*b_pgiodone)(void *, struct vm_page **,
+ int, int);
int b_pgbefore;
int b_pgafter;
};
};
union cluster_info {
TAILQ_HEAD(cluster_list_head, buf) cluster_head;
TAILQ_ENTRY(buf) cluster_entry;
} b_cluster;
int b_npages;
struct workhead b_dep; /* (D) List of filesystem dependencies. */
void *b_fsprivate1;
void *b_fsprivate2;
void *b_fsprivate3;
#if defined(FULL_BUF_TRACKING)
#define BUF_TRACKING_SIZE 32
#define BUF_TRACKING_ENTRY(x) ((x) & (BUF_TRACKING_SIZE - 1))
const char *b_io_tracking[BUF_TRACKING_SIZE];
uint32_t b_io_tcnt;
#elif defined(BUF_TRACKING)
const char *b_io_tracking;
#endif
struct vm_page *b_pages[];
};
#define b_object b_bufobj->bo_object
/*
* These flags are kept in b_flags.
*
* Notes:
*
* B_ASYNC VOP calls on bp's are usually async whether or not
* B_ASYNC is set, but some subsystems, such as NFS, like
* to know what is best for the caller so they can
* optimize the I/O.
*
* B_PAGING Indicates that bp is being used by the paging system or
* some paging system and that the bp is not linked into
* the b_vp's clean/dirty linked lists or ref counts.
* Buffer vp reassignments are illegal in this case.
*
* B_CACHE This may only be set if the buffer is entirely valid.
* The situation where B_DELWRI is set and B_CACHE is
* clear MUST be committed to disk by getblk() so
* B_DELWRI can also be cleared. See the comments for
* getblk() in kern/vfs_bio.c. If B_CACHE is clear,
* the caller is expected to clear BIO_ERROR and B_INVAL,
* set BIO_READ, and initiate an I/O.
*
* The 'entire buffer' is defined to be the range from
* 0 through b_bcount.
*
* B_MALLOC Request that the buffer be allocated from the malloc
* pool, DEV_BSIZE aligned instead of PAGE_SIZE aligned.
*
* B_CLUSTEROK This flag is typically set for B_DELWRI buffers
* by filesystems that allow clustering when the buffer
* is fully dirty and indicates that it may be clustered
* with other adjacent dirty buffers. Note the clustering
* may not be used with the stage 1 data write under NFS
* but may be used for the commit rpc portion.
*
* B_INVALONERR This flag is set on dirty buffers. It specifies that a
* write error should forcibly invalidate the buffer
* contents. This flag should be used with caution, as it
* discards data. It is incompatible with B_ASYNC.
*
* B_VMIO Indicates that the buffer is tied into an VM object.
* The buffer's data is always PAGE_SIZE aligned even
* if b_bufsize and b_bcount are not. ( b_bufsize is
* always at least DEV_BSIZE aligned, though ).
*
* B_DIRECT Hint that we should attempt to completely free
* the pages underlying the buffer. B_DIRECT is
* sticky until the buffer is released and typically
* only has an effect when B_RELBUF is also set.
*
*/
#define B_AGE 0x00000001 /* Move to age queue when I/O done. */
#define B_NEEDCOMMIT 0x00000002 /* Append-write in progress. */
#define B_ASYNC 0x00000004 /* Start I/O, do not wait. */
#define B_DIRECT 0x00000008 /* direct I/O flag (pls free vmio) */
#define B_DEFERRED 0x00000010 /* Skipped over for cleaning */
#define B_CACHE 0x00000020 /* Bread found us in the cache. */
#define B_VALIDSUSPWRT 0x00000040 /* Valid write during suspension. */
#define B_DELWRI 0x00000080 /* Delay I/O until buffer reused. */
#define B_CKHASH 0x00000100 /* checksum hash calculated on read */
#define B_DONE 0x00000200 /* I/O completed. */
#define B_EINTR 0x00000400 /* I/O was interrupted */
#define B_NOREUSE 0x00000800 /* Contents not reused once released. */
#define B_REUSE 0x00001000 /* Contents reused, second chance. */
#define B_INVAL 0x00002000 /* Does not contain valid info. */
#define B_BARRIER 0x00004000 /* Write this and all preceding first. */
#define B_NOCACHE 0x00008000 /* Do not cache block after use. */
#define B_MALLOC 0x00010000 /* malloced b_data */
#define B_CLUSTEROK 0x00020000 /* Pagein op, so swap() can count it. */
#define B_INVALONERR 0x00040000 /* Invalidate on write error. */
#define B_IOSTARTED 0x00080000 /* buf_start() called */
#define B_00100000 0x00100000 /* Available flag. */
#define B_MAXPHYS 0x00200000 /* nitems(b_pages[]) = atop(MAXPHYS). */
#define B_RELBUF 0x00400000 /* Release VMIO buffer. */
#define B_FS_FLAG1 0x00800000 /* Available flag for FS use. */
#define B_NOCOPY 0x01000000 /* Don't copy-on-write this buf. */
#define B_INFREECNT 0x02000000 /* buf is counted in numfreebufs */
#define B_PAGING 0x04000000 /* volatile paging I/O -- bypass VMIO */
#define B_MANAGED 0x08000000 /* Managed by FS. */
#define B_RAM 0x10000000 /* Read ahead mark (flag) */
#define B_VMIO 0x20000000 /* VMIO flag */
#define B_CLUSTER 0x40000000 /* pagein op, so swap() can count it */
#define B_REMFREE 0x80000000 /* Delayed bremfree */
#define PRINT_BUF_FLAGS "\20\40remfree\37cluster\36vmio\35ram\34managed" \
"\33paging\32infreecnt\31nocopy\30b23\27relbuf\26maxphys\25b20" \
"\24iostarted\23invalonerr\22clusterok\21malloc\20nocache\17b14" \
"\16inval\15reuse\14noreuse\13eintr\12done\11b8\10delwri" \
"\7validsuspwrt\6cache\5deferred\4direct\3async\2needcommit\1age"
/*
* These flags are kept in b_xflags.
*
* BX_FSPRIV reserves a set of eight flags that may be used by individual
* filesystems for their own purpose. Their specific definitions are
* found in the header files for each filesystem that uses them.
*/
#define BX_VNDIRTY 0x00000001 /* On vnode dirty list */
#define BX_VNCLEAN 0x00000002 /* On vnode clean list */
#define BX_CVTENXIO 0x00000004 /* Convert errors to ENXIO */
#define BX_BKGRDWRITE 0x00000010 /* Do writes in background */
#define BX_BKGRDMARKER 0x00000020 /* Mark buffer for splay tree */
#define BX_ALTDATA 0x00000040 /* Holds extended data */
#define BX_FSPRIV 0x00FF0000 /* Filesystem-specific flags mask */
#define PRINT_BUF_XFLAGS "\20\7altdata\6bkgrdmarker\5bkgrdwrite\3cvtenxio" \
"\2clean\1dirty"
#define NOOFFSET (-1LL) /* No buffer offset calculated yet */
/*
* These flags are kept in b_vflags.
*/
#define BV_SCANNED 0x00000001 /* VOP_FSYNC funcs mark written bufs */
#define BV_BKGRDINPROG 0x00000002 /* Background write in progress */
#define BV_BKGRDWAIT 0x00000004 /* Background write waiting */
#define BV_BKGRDERR 0x00000008 /* Error from background write */
#define PRINT_BUF_VFLAGS "\20\4bkgrderr\3bkgrdwait\2bkgrdinprog\1scanned"
#ifdef _KERNEL
#ifndef NSWBUF_MIN
#define NSWBUF_MIN 16
#endif
/*
* Buffer locking
*/
extern const char *buf_wmesg; /* Default buffer lock message */
#define BUF_WMESG "bufwait"
#include <sys/proc.h> /* XXX for curthread */
#include <sys/mutex.h>
/*
* Initialize a lock.
*/
#define BUF_LOCKINIT(bp) \
lockinit(&(bp)->b_lock, PRIBIO + 4, buf_wmesg, 0, LK_NEW)
/*
*
* Get a lock sleeping non-interruptably until it becomes available.
*/
#define BUF_LOCK(bp, locktype, interlock) \
_lockmgr_args_rw(&(bp)->b_lock, (locktype), (interlock), \
LK_WMESG_DEFAULT, LK_PRIO_DEFAULT, LK_TIMO_DEFAULT, \
LOCK_FILE, LOCK_LINE)
/*
* Get a lock sleeping with specified interruptably and timeout.
*/
#define BUF_TIMELOCK(bp, locktype, interlock, wmesg, catch, timo) \
_lockmgr_args_rw(&(bp)->b_lock, (locktype) | LK_TIMELOCK, \
(interlock), (wmesg), (PRIBIO + 4) | (catch), (timo), \
LOCK_FILE, LOCK_LINE)
/*
* Release a lock. Only the acquiring process may free the lock unless
* it has been handed off to biodone.
*/
#define BUF_UNLOCK(bp) do { \
KASSERT(((bp)->b_flags & B_REMFREE) == 0, \
("BUF_UNLOCK %p while B_REMFREE is still set.", (bp))); \
\
BUF_UNLOCK_RAW((bp)); \
} while (0)
#define BUF_UNLOCK_RAW(bp) do { \
(void)_lockmgr_args(&(bp)->b_lock, LK_RELEASE, NULL, \
LK_WMESG_DEFAULT, LK_PRIO_DEFAULT, LK_TIMO_DEFAULT, \
LOCK_FILE, LOCK_LINE); \
} while (0)
/*
* Check if a buffer lock is recursed.
*/
#define BUF_LOCKRECURSED(bp) \
lockmgr_recursed(&(bp)->b_lock)
/*
* Check if a buffer lock is currently held.
*/
#define BUF_ISLOCKED(bp) \
lockstatus(&(bp)->b_lock)
/*
* Free a buffer lock.
*/
#define BUF_LOCKFREE(bp) \
lockdestroy(&(bp)->b_lock)
/*
* Print informations on a buffer lock.
*/
#define BUF_LOCKPRINTINFO(bp) \
lockmgr_printinfo(&(bp)->b_lock)
/*
* Buffer lock assertions.
*/
#if defined(INVARIANTS) && defined(INVARIANT_SUPPORT)
#define BUF_ASSERT_LOCKED(bp) \
_lockmgr_assert(&(bp)->b_lock, KA_LOCKED, LOCK_FILE, LOCK_LINE)
#define BUF_ASSERT_SLOCKED(bp) \
_lockmgr_assert(&(bp)->b_lock, KA_SLOCKED, LOCK_FILE, LOCK_LINE)
#define BUF_ASSERT_XLOCKED(bp) \
_lockmgr_assert(&(bp)->b_lock, KA_XLOCKED, LOCK_FILE, LOCK_LINE)
#define BUF_ASSERT_UNLOCKED(bp) \
_lockmgr_assert(&(bp)->b_lock, KA_UNLOCKED, LOCK_FILE, LOCK_LINE)
#else
#define BUF_ASSERT_LOCKED(bp)
#define BUF_ASSERT_SLOCKED(bp)
#define BUF_ASSERT_XLOCKED(bp)
#define BUF_ASSERT_UNLOCKED(bp)
#endif
#ifdef _SYS_PROC_H_ /* Avoid #include <sys/proc.h> pollution */
/*
* When initiating asynchronous I/O, change ownership of the lock to the
* kernel. Once done, the lock may legally released by biodone. The
* original owning process can no longer acquire it recursively, but must
* wait until the I/O is completed and the lock has been freed by biodone.
*/
#define BUF_KERNPROC(bp) \
_lockmgr_disown(&(bp)->b_lock, LOCK_FILE, LOCK_LINE)
#endif
#endif /* _KERNEL */
struct buf_queue_head {
TAILQ_HEAD(buf_queue, buf) queue;
daddr_t last_pblkno;
struct buf *insert_point;
struct buf *switch_point;
};
/*
* This structure describes a clustered I/O.
*/
struct cluster_save {
long bs_bcount; /* Saved b_bcount. */
long bs_bufsize; /* Saved b_bufsize. */
int bs_nchildren; /* Number of associated buffers. */
struct buf **bs_children; /* List of associated buffers. */
};
#ifdef _KERNEL
static __inline int
bwrite(struct buf *bp)
{
KASSERT(bp->b_bufobj != NULL, ("bwrite: no bufobj bp=%p", bp));
KASSERT(bp->b_bufobj->bo_ops != NULL, ("bwrite: no bo_ops bp=%p", bp));
KASSERT(bp->b_bufobj->bo_ops->bop_write != NULL,
("bwrite: no bop_write bp=%p", bp));
return (BO_WRITE(bp->b_bufobj, bp));
}
static __inline void
bstrategy(struct buf *bp)
{
KASSERT(bp->b_bufobj != NULL, ("bstrategy: no bufobj bp=%p", bp));
KASSERT(bp->b_bufobj->bo_ops != NULL,
("bstrategy: no bo_ops bp=%p", bp));
KASSERT(bp->b_bufobj->bo_ops->bop_strategy != NULL,
("bstrategy: no bop_strategy bp=%p", bp));
BO_STRATEGY(bp->b_bufobj, bp);
}
static __inline void
buf_start(struct buf *bp)
{
KASSERT((bp->b_flags & B_IOSTARTED) == 0,
("recursed buf_start %p", bp));
bp->b_flags |= B_IOSTARTED;
if (bioops.io_start)
(*bioops.io_start)(bp);
}
static __inline void
buf_complete(struct buf *bp)
{
if ((bp->b_flags & B_IOSTARTED) != 0) {
bp->b_flags &= ~B_IOSTARTED;
if (bioops.io_complete)
(*bioops.io_complete)(bp);
}
}
static __inline void
buf_deallocate(struct buf *bp)
{
if (bioops.io_deallocate)
(*bioops.io_deallocate)(bp);
}
static __inline int
buf_countdeps(struct buf *bp, int i)
{
if (bioops.io_countdeps)
return ((*bioops.io_countdeps)(bp, i));
else
return (0);
}
static __inline void
buf_track(struct buf *bp __unused, const char *location __unused)
{
#if defined(FULL_BUF_TRACKING)
bp->b_io_tracking[BUF_TRACKING_ENTRY(bp->b_io_tcnt++)] = location;
#elif defined(BUF_TRACKING)
bp->b_io_tracking = location;
#endif
}
#endif /* _KERNEL */
/*
* Zero out the buffer's data area.
*/
#define clrbuf(bp) { \
bzero((bp)->b_data, (u_int)(bp)->b_bcount); \
(bp)->b_resid = 0; \
}
/*
* Flags for getblk's last parameter.
*/
#define GB_LOCK_NOWAIT 0x0001 /* Fail if we block on a buf lock. */
#define GB_NOCREAT 0x0002 /* Don't create a buf if not found. */
#define GB_NOWAIT_BD 0x0004 /* Do not wait for bufdaemon. */
#define GB_UNMAPPED 0x0008 /* Do not mmap buffer pages. */
#define GB_KVAALLOC 0x0010 /* But allocate KVA. */
#define GB_CKHASH 0x0020 /* If reading, calc checksum hash */
#define GB_NOSPARSE 0x0040 /* Do not instantiate holes */
#define GB_CVTENXIO 0x0080 /* Convert errors to ENXIO */
#ifdef _KERNEL
extern int nbuf; /* The number of buffer headers */
extern u_long maxswzone; /* Max KVA for swap structures */
extern u_long maxbcache; /* Max KVA for buffer cache */
extern int maxbcachebuf; /* Max buffer cache block size */
extern long runningbufspace;
extern long hibufspace;
extern int dirtybufthresh;
extern int bdwriteskip;
extern int dirtybufferflushes;
extern int altbufferflushes;
extern int nswbuf; /* Number of swap I/O buffer headers. */
extern caddr_t unmapped_buf; /* Data address for unmapped buffers. */
static inline int
buf_mapped(struct buf *bp)
{
return (bp->b_data != unmapped_buf);
}
void runningbufwakeup(struct buf *);
void waitrunningbufspace(void);
caddr_t kern_vfs_bio_buffer_alloc(caddr_t v, long physmem_est);
void bufinit(void);
void bufshutdown(int);
void bdata2bio(struct buf *bp, struct bio *bip);
void bwillwrite(void);
int buf_dirty_count_severe(void);
void bremfree(struct buf *);
void bremfreef(struct buf *); /* XXX Force bremfree, only for nfs. */
#define bread(vp, blkno, size, cred, bpp) \
breadn_flags(vp, blkno, blkno, size, NULL, NULL, 0, cred, 0, \
NULL, bpp)
#define bread_gb(vp, blkno, size, cred, gbflags, bpp) \
breadn_flags(vp, blkno, blkno, size, NULL, NULL, 0, cred, \
gbflags, NULL, bpp)
#define breadn(vp, blkno, size, rablkno, rabsize, cnt, cred, bpp) \
breadn_flags(vp, blkno, blkno, size, rablkno, rabsize, cnt, cred, \
0, NULL, bpp)
int breadn_flags(struct vnode *, daddr_t, daddr_t, int, daddr_t *, int *,
int, struct ucred *, int, void (*)(struct buf *), struct buf **);
void bdwrite(struct buf *);
void bawrite(struct buf *);
void babarrierwrite(struct buf *);
int bbarrierwrite(struct buf *);
void bdirty(struct buf *);
void bundirty(struct buf *);
void bufstrategy(struct bufobj *, struct buf *);
void brelse(struct buf *);
void bqrelse(struct buf *);
int vfs_bio_awrite(struct buf *);
void vfs_busy_pages_acquire(struct buf *bp);
void vfs_busy_pages_release(struct buf *bp);
struct buf *incore(struct bufobj *, daddr_t);
bool inmem(struct vnode *, daddr_t);
struct buf *gbincore(struct bufobj *, daddr_t);
struct buf *gbincore_unlocked(struct bufobj *, daddr_t);
struct buf *getblk(struct vnode *, daddr_t, int, int, int, int);
int getblkx(struct vnode *vp, daddr_t blkno, daddr_t dblkno, int size,
int slpflag, int slptimeo, int flags, struct buf **bpp);
struct buf *geteblk(int, int);
int bufwait(struct buf *);
int bufwrite(struct buf *);
void bufdone(struct buf *);
void bd_speedup(void);
extern uma_zone_t pbuf_zone;
uma_zone_t pbuf_zsecond_create(const char *name, int max);
int cluster_read(struct vnode *, u_quad_t, daddr_t, long,
struct ucred *, long, int, int, struct buf **);
int cluster_wbuild(struct vnode *, long, daddr_t, int, int);
void cluster_write(struct vnode *, struct buf *, u_quad_t, int, int);
void vfs_bio_brelse(struct buf *bp, int ioflags);
void vfs_bio_bzero_buf(struct buf *bp, int base, int size);
void vfs_bio_clrbuf(struct buf *);
void vfs_bio_set_flags(struct buf *bp, int ioflags);
void vfs_bio_set_valid(struct buf *, int base, int size);
void vfs_busy_pages(struct buf *, int clear_modify);
void vfs_unbusy_pages(struct buf *);
int vmapbuf(struct buf *, void *, size_t, int);
void vunmapbuf(struct buf *);
void brelvp(struct buf *);
void bgetvp(struct vnode *, struct buf *);
void pbgetbo(struct bufobj *bo, struct buf *bp);
void pbgetvp(struct vnode *, struct buf *);
void pbrelbo(struct buf *);
void pbrelvp(struct buf *);
int allocbuf(struct buf *bp, int size);
void reassignbuf(struct buf *);
void bwait(struct buf *, u_char, const char *);
void bdone(struct buf *);
typedef daddr_t (vbg_get_lblkno_t)(struct vnode *, vm_ooffset_t);
typedef int (vbg_get_blksize_t)(struct vnode *, daddr_t);
int vfs_bio_getpages(struct vnode *vp, struct vm_page **ma, int count,
int *rbehind, int *rahead, vbg_get_lblkno_t get_lblkno,
vbg_get_blksize_t get_blksize);
#endif /* _KERNEL */
#endif /* !_SYS_BUF_H_ */
diff --git a/sys/sys/mount.h b/sys/sys/mount.h
index 9ae5f58e5218..ca3261a22e62 100644
--- a/sys/sys/mount.h
+++ b/sys/sys/mount.h
@@ -1,1159 +1,1163 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1989, 1991, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)mount.h 8.21 (Berkeley) 5/20/95
* $FreeBSD$
*/
#ifndef _SYS_MOUNT_H_
#define _SYS_MOUNT_H_
#include <sys/ucred.h>
#include <sys/queue.h>
#ifdef _KERNEL
#include <sys/lock.h>
#include <sys/lockmgr.h>
#include <sys/tslog.h>
#include <sys/_mutex.h>
#include <sys/_sx.h>
#endif
/*
* NOTE: When changing statfs structure, mount structure, MNT_* flags or
* MNTK_* flags also update DDB show mount command in vfs_subr.c.
*/
typedef struct fsid { int32_t val[2]; } fsid_t; /* filesystem id type */
#define fsidcmp(a, b) memcmp((a), (b), sizeof(fsid_t))
/*
* File identifier.
* These are unique per filesystem on a single machine.
*
* Note that the offset of fid_data is 4 bytes, so care must be taken to avoid
* undefined behavior accessing unaligned fields within an embedded struct.
*/
#define MAXFIDSZ 16
struct fid {
u_short fid_len; /* length of data in bytes */
u_short fid_data0; /* force longword alignment */
char fid_data[MAXFIDSZ]; /* data (variable length) */
};
/*
* filesystem statistics
*/
#define MFSNAMELEN 16 /* length of type name including null */
#define MNAMELEN 1024 /* size of on/from name bufs */
#define STATFS_VERSION 0x20140518 /* current version number */
struct statfs {
uint32_t f_version; /* structure version number */
uint32_t f_type; /* type of filesystem */
uint64_t f_flags; /* copy of mount exported flags */
uint64_t f_bsize; /* filesystem fragment size */
uint64_t f_iosize; /* optimal transfer block size */
uint64_t f_blocks; /* total data blocks in filesystem */
uint64_t f_bfree; /* free blocks in filesystem */
int64_t f_bavail; /* free blocks avail to non-superuser */
uint64_t f_files; /* total file nodes in filesystem */
int64_t f_ffree; /* free nodes avail to non-superuser */
uint64_t f_syncwrites; /* count of sync writes since mount */
uint64_t f_asyncwrites; /* count of async writes since mount */
uint64_t f_syncreads; /* count of sync reads since mount */
uint64_t f_asyncreads; /* count of async reads since mount */
uint64_t f_spare[10]; /* unused spare */
uint32_t f_namemax; /* maximum filename length */
uid_t f_owner; /* user that mounted the filesystem */
fsid_t f_fsid; /* filesystem id */
char f_charspare[80]; /* spare string space */
char f_fstypename[MFSNAMELEN]; /* filesystem type name */
char f_mntfromname[MNAMELEN]; /* mounted filesystem */
char f_mntonname[MNAMELEN]; /* directory on which mounted */
};
#if defined(_WANT_FREEBSD11_STATFS) || defined(_KERNEL)
#define FREEBSD11_STATFS_VERSION 0x20030518 /* current version number */
struct freebsd11_statfs {
uint32_t f_version; /* structure version number */
uint32_t f_type; /* type of filesystem */
uint64_t f_flags; /* copy of mount exported flags */
uint64_t f_bsize; /* filesystem fragment size */
uint64_t f_iosize; /* optimal transfer block size */
uint64_t f_blocks; /* total data blocks in filesystem */
uint64_t f_bfree; /* free blocks in filesystem */
int64_t f_bavail; /* free blocks avail to non-superuser */
uint64_t f_files; /* total file nodes in filesystem */
int64_t f_ffree; /* free nodes avail to non-superuser */
uint64_t f_syncwrites; /* count of sync writes since mount */
uint64_t f_asyncwrites; /* count of async writes since mount */
uint64_t f_syncreads; /* count of sync reads since mount */
uint64_t f_asyncreads; /* count of async reads since mount */
uint64_t f_spare[10]; /* unused spare */
uint32_t f_namemax; /* maximum filename length */
uid_t f_owner; /* user that mounted the filesystem */
fsid_t f_fsid; /* filesystem id */
char f_charspare[80]; /* spare string space */
char f_fstypename[16]; /* filesystem type name */
char f_mntfromname[88]; /* mounted filesystem */
char f_mntonname[88]; /* directory on which mounted */
};
#endif /* _WANT_FREEBSD11_STATFS || _KERNEL */
#ifdef _KERNEL
#define OMFSNAMELEN 16 /* length of fs type name, including null */
#define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */
/* XXX getfsstat.2 is out of date with write and read counter changes here. */
/* XXX statfs.2 is out of date with read counter changes here. */
struct ostatfs {
long f_spare2; /* placeholder */
long f_bsize; /* fundamental filesystem block size */
long f_iosize; /* optimal transfer block size */
long f_blocks; /* total data blocks in filesystem */
long f_bfree; /* free blocks in fs */
long f_bavail; /* free blocks avail to non-superuser */
long f_files; /* total file nodes in filesystem */
long f_ffree; /* free file nodes in fs */
fsid_t f_fsid; /* filesystem id */
uid_t f_owner; /* user that mounted the filesystem */
int f_type; /* type of filesystem */
int f_flags; /* copy of mount exported flags */
long f_syncwrites; /* count of sync writes since mount */
long f_asyncwrites; /* count of async writes since mount */
char f_fstypename[OMFSNAMELEN]; /* fs type name */
char f_mntonname[OMNAMELEN]; /* directory on which mounted */
long f_syncreads; /* count of sync reads since mount */
long f_asyncreads; /* count of async reads since mount */
short f_spares1; /* unused spare */
char f_mntfromname[OMNAMELEN];/* mounted filesystem */
short f_spares2; /* unused spare */
/*
* XXX on machines where longs are aligned to 8-byte boundaries, there
* is an unnamed int32_t here. This spare was after the apparent end
* of the struct until we bit off the read counters from f_mntonname.
*/
long f_spare[2]; /* unused spare */
};
+#endif /* _KERNEL */
+#if defined(_WANT_MOUNT) || defined(_KERNEL)
TAILQ_HEAD(vnodelst, vnode);
/* Mount options list */
TAILQ_HEAD(vfsoptlist, vfsopt);
struct vfsopt {
TAILQ_ENTRY(vfsopt) link;
char *name;
void *value;
int len;
int pos;
int seen;
};
struct mount_pcpu {
int mntp_thread_in_ops;
int mntp_ref;
int mntp_lockref;
int mntp_writeopcount;
};
_Static_assert(sizeof(struct mount_pcpu) == 16,
"the struct is allocated from pcpu 16 zone");
/*
* Structure per mounted filesystem. Each mounted filesystem has an
* array of operations and an instance record. The filesystems are
* put on a doubly linked list.
*
* Lock reference:
* l - mnt_listmtx
* m - mountlist_mtx
* i - interlock
* v - vnode freelist mutex
*
* Unmarked fields are considered stable as long as a ref is held.
*
*/
struct mount {
int mnt_vfs_ops; /* (i) pending vfs ops */
int mnt_kern_flag; /* (i) kernel only flags */
uint64_t mnt_flag; /* (i) flags shared with user */
struct mount_pcpu *mnt_pcpu; /* per-CPU data */
struct vnode *mnt_rootvnode;
struct vnode *mnt_vnodecovered; /* vnode we mounted on */
struct vfsops *mnt_op; /* operations on fs */
struct vfsconf *mnt_vfc; /* configuration info */
struct mtx __aligned(CACHE_LINE_SIZE) mnt_mtx; /* mount structure interlock */
int mnt_gen; /* struct mount generation */
#define mnt_startzero mnt_list
TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */
struct vnode *mnt_syncer; /* syncer vnode */
int mnt_ref; /* (i) Reference count */
struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */
int mnt_nvnodelistsize; /* (i) # of vnodes */
int mnt_writeopcount; /* (i) write syscalls pending */
struct vfsoptlist *mnt_opt; /* current mount options */
struct vfsoptlist *mnt_optnew; /* new options passed to fs */
int mnt_maxsymlinklen; /* max size of short symlink */
struct statfs mnt_stat; /* cache of filesystem stats */
struct ucred *mnt_cred; /* credentials of mounter */
void * mnt_data; /* private data */
time_t mnt_time; /* last time written*/
int mnt_iosize_max; /* max size for clusters, etc */
struct netexport *mnt_export; /* export list */
struct label *mnt_label; /* MAC label for the fs */
u_int mnt_hashseed; /* Random seed for vfs_hash */
int mnt_lockref; /* (i) Lock reference count */
int mnt_secondary_writes; /* (i) # of secondary writes */
int mnt_secondary_accwrites;/* (i) secondary wr. starts */
struct thread *mnt_susp_owner; /* (i) thread owning suspension */
#define mnt_endzero mnt_gjprovider
char *mnt_gjprovider; /* gjournal provider name */
struct mtx mnt_listmtx;
struct vnodelst mnt_lazyvnodelist; /* (l) list of lazy vnodes */
int mnt_lazyvnodelistsize; /* (l) # of lazy vnodes */
struct lock mnt_explock; /* vfs_export walkers lock */
TAILQ_ENTRY(mount) mnt_upper_link; /* (m) we in the all uppers */
TAILQ_HEAD(, mount) mnt_uppers; /* (m) upper mounts over us*/
};
+#endif /* _WANT_MOUNT || _KERNEL */
+#ifdef _KERNEL
/*
* Definitions for MNT_VNODE_FOREACH_ALL.
*/
struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp);
struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp);
void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp);
#define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \
for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \
(vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp)))
#define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \
do { \
MNT_ILOCK(mp); \
__mnt_vnode_markerfree_all(&(mvp), (mp)); \
/* MNT_IUNLOCK(mp); -- done in above function */ \
mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \
} while (0)
/*
* Definitions for MNT_VNODE_FOREACH_LAZY.
*/
typedef int mnt_lazy_cb_t(struct vnode *, void *);
struct vnode *__mnt_vnode_next_lazy(struct vnode **mvp, struct mount *mp,
mnt_lazy_cb_t *cb, void *cbarg);
struct vnode *__mnt_vnode_first_lazy(struct vnode **mvp, struct mount *mp,
mnt_lazy_cb_t *cb, void *cbarg);
void __mnt_vnode_markerfree_lazy(struct vnode **mvp, struct mount *mp);
#define MNT_VNODE_FOREACH_LAZY(vp, mp, mvp, cb, cbarg) \
for (vp = __mnt_vnode_first_lazy(&(mvp), (mp), (cb), (cbarg)); \
(vp) != NULL; \
vp = __mnt_vnode_next_lazy(&(mvp), (mp), (cb), (cbarg)))
#define MNT_VNODE_FOREACH_LAZY_ABORT(mp, mvp) \
__mnt_vnode_markerfree_lazy(&(mvp), (mp))
#define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx)
#define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx)
#define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx)
#define MNT_MTX(mp) (&(mp)->mnt_mtx)
#define MNT_REF(mp) do { \
mtx_assert(MNT_MTX(mp), MA_OWNED); \
mp->mnt_ref++; \
} while (0)
#define MNT_REL(mp) do { \
mtx_assert(MNT_MTX(mp), MA_OWNED); \
(mp)->mnt_ref--; \
if ((mp)->mnt_vfs_ops && (mp)->mnt_ref < 0) \
vfs_dump_mount_counters(mp); \
if ((mp)->mnt_ref == 0 && (mp)->mnt_vfs_ops) \
wakeup((mp)); \
} while (0)
#endif /* _KERNEL */
#if defined(_WANT_MNTOPTNAMES) || defined(_KERNEL)
struct mntoptnames {
uint64_t o_opt;
const char *o_name;
};
#define MNTOPT_NAMES \
{ MNT_ASYNC, "asynchronous" }, \
{ MNT_EXPORTED, "NFS exported" }, \
{ MNT_LOCAL, "local" }, \
{ MNT_NOATIME, "noatime" }, \
{ MNT_NOEXEC, "noexec" }, \
{ MNT_NOSUID, "nosuid" }, \
{ MNT_NOSYMFOLLOW, "nosymfollow" }, \
{ MNT_QUOTA, "with quotas" }, \
{ MNT_RDONLY, "read-only" }, \
{ MNT_SYNCHRONOUS, "synchronous" }, \
{ MNT_UNION, "union" }, \
{ MNT_NOCLUSTERR, "noclusterr" }, \
{ MNT_NOCLUSTERW, "noclusterw" }, \
{ MNT_SUIDDIR, "suiddir" }, \
{ MNT_SOFTDEP, "soft-updates" }, \
{ MNT_SUJ, "journaled soft-updates" }, \
{ MNT_MULTILABEL, "multilabel" }, \
{ MNT_ACLS, "acls" }, \
{ MNT_NFS4ACLS, "nfsv4acls" }, \
{ MNT_GJOURNAL, "gjournal" }, \
{ MNT_AUTOMOUNTED, "automounted" }, \
{ MNT_VERIFIED, "verified" }, \
{ MNT_UNTRUSTED, "untrusted" }, \
{ MNT_NOCOVER, "nocover" }, \
{ MNT_EMPTYDIR, "emptydir" }, \
{ MNT_UPDATE, "update" }, \
{ MNT_DELEXPORT, "delexport" }, \
{ MNT_RELOAD, "reload" }, \
{ MNT_FORCE, "force" }, \
{ MNT_SNAPSHOT, "snapshot" }, \
{ 0, NULL }
#endif
/*
* User specifiable flags, stored in mnt_flag.
*/
#define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */
#define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */
#define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */
#define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */
#define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */
#define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */
#define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */
#define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */
#define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */
#define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */
#define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */
#define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */
#define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */
#define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */
#define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */
#define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */
#define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */
#define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */
#define MNT_UNTRUSTED 0x0000000800000000ULL /* filesys metadata untrusted */
/*
* NFS export related mount flags.
*/
#define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */
#define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */
#define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */
#define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */
#define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */
#define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */
#define MNT_EXTLS 0x0000004000000000ULL /* require TLS */
#define MNT_EXTLSCERT 0x0000008000000000ULL /* require TLS with client cert */
#define MNT_EXTLSCERTUSER 0x0000010000000000ULL /* require TLS with user cert */
/*
* Flags set by internal operations,
* but visible to the user.
* XXX some of these are not quite right.. (I've never seen the root flag set)
*/
#define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */
#define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */
#define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */
#define MNT_USER 0x0000000000008000ULL /* mounted by a user */
#define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */
#define MNT_VERIFIED 0x0000000400000000ULL /* filesystem is verified */
/*
* Mask of flags that are visible to statfs().
* XXX I think that this could now become (~(MNT_CMDFLAGS))
* but the 'mount' program may need changing to handle this.
*/
#define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \
MNT_NOSUID | MNT_UNION | MNT_SUJ | \
MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \
MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \
MNT_LOCAL | MNT_USER | MNT_QUOTA | \
MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \
MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \
MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \
MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \
MNT_NFS4ACLS | MNT_AUTOMOUNTED | MNT_VERIFIED | \
MNT_UNTRUSTED)
/* Mask of flags that can be updated. */
#define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \
MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \
MNT_NOATIME | \
MNT_NOSYMFOLLOW | MNT_IGNORE | \
MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \
MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \
MNT_AUTOMOUNTED | MNT_UNTRUSTED)
/*
* External filesystem command modifier flags.
* Unmount can use the MNT_FORCE flag.
* XXX: These are not STATES and really should be somewhere else.
* XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL,
* but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2),
* and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2),
* it's harmless.
*/
#define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */
#define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */
#define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */
#define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */
#define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */
#define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */
#define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */
#define MNT_NOCOVER 0x0000001000000000ULL /* Do not cover a mount point */
#define MNT_EMPTYDIR 0x0000002000000000ULL /* Only mount on empty dir */
#define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \
MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \
MNT_BYFSID | MNT_NOCOVER | MNT_EMPTYDIR)
/*
* Internal filesystem control flags stored in mnt_kern_flag.
*
* MNTK_UNMOUNT locks the mount entry so that name lookup cannot
* proceed past the mount point. This keeps the subtree stable during
* mounts and unmounts. When non-forced unmount flushes all vnodes
* from the mp queue, the MNTK_UNMOUNT flag prevents insmntque() from
* queueing new vnodes.
*
* MNTK_UNMOUNTF permits filesystems to detect a forced unmount while
* dounmount() is still waiting to lock the mountpoint. This allows
* the filesystem to cancel operations that might otherwise deadlock
* with the unmount attempt (used by NFS).
*/
#define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */
#define MNTK_ASYNC 0x00000002 /* filtered async flag */
#define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */
#define MNTK_NOMSYNC 0x00000008 /* don't do msync */
#define MNTK_DRAINING 0x00000010 /* lock draining is happening */
#define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */
#define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */
#define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */
#define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads
and writes. Filesystem shall properly
handle i/o state on EFAULT. */
#define MNTK_VGONE_UPPER 0x00000200
#define MNTK_VGONE_WAITER 0x00000400
#define MNTK_LOOKUP_EXCL_DOTDOT 0x00000800
#define MNTK_MARKER 0x00001000
#define MNTK_UNMAPPED_BUFS 0x00002000
#define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */
#define MNTK_TEXT_REFS 0x00008000 /* Keep use ref for text */
#define MNTK_VMSETSIZE_BUG 0x00010000
#define MNTK_UNIONFS 0x00020000 /* A hack for F_ISUNIONSTACK */
#define MNTK_FPLOOKUP 0x00040000 /* fast path lookup is supported */
#define MNTK_SUSPEND_ALL 0x00080000 /* Suspended by all-fs suspension */
#define MNTK_NOASYNC 0x00800000 /* disable async */
#define MNTK_UNMOUNT 0x01000000 /* unmount in progress */
#define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */
#define MNTK_SUSPEND 0x08000000 /* request write suspension */
#define MNTK_SUSPEND2 0x04000000 /* block secondary writes */
#define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */
#define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs
mounts over this fs */
#define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */
#define MNTK_NOKNOTE 0x80000000 /* Don't send KNOTEs from VOP hooks */
#ifdef _KERNEL
static inline int
MNT_SHARED_WRITES(struct mount *mp)
{
return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0);
}
static inline int
MNT_EXTENDED_SHARED(struct mount *mp)
{
return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0);
}
#endif
/*
* Sysctl CTL_VFS definitions.
*
* Second level identifier specifies which filesystem. Second level
* identifier VFS_VFSCONF returns information about all filesystems.
* Second level identifier VFS_GENERIC is non-terminal.
*/
#define VFS_VFSCONF 0 /* get configured filesystems */
#define VFS_GENERIC 0 /* generic filesystem information */
/*
* Third level identifiers for VFS_GENERIC are given below; third
* level identifiers for specific filesystems are given in their
* mount specific header files.
*/
#define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */
#define VFS_CONF 2 /* struct: vfsconf for filesystem given
as next argument */
/*
* Flags for various system call interfaces.
*
* waitfor flags to vfs_sync() and getfsstat()
*/
#define MNT_WAIT 1 /* synchronously wait for I/O to complete */
#define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */
#define MNT_LAZY 3 /* push data not written by filesystem syncer */
#define MNT_SUSPEND 4 /* Suspend file system after sync */
/*
* Generic file handle
*/
struct fhandle {
fsid_t fh_fsid; /* Filesystem id of mount point */
struct fid fh_fid; /* Filesys specific id */
};
typedef struct fhandle fhandle_t;
/*
* Old export arguments without security flavor list
*/
struct oexport_args {
int ex_flags; /* export related flags */
uid_t ex_root; /* mapping for root uid */
struct xucred ex_anon; /* mapping for anonymous user */
struct sockaddr *ex_addr; /* net address to which exported */
u_char ex_addrlen; /* and the net address length */
struct sockaddr *ex_mask; /* mask of valid bits in saddr */
u_char ex_masklen; /* and the smask length */
char *ex_indexfile; /* index file for WebNFS URLs */
};
/*
* Not quite so old export arguments with 32bit ex_flags and xucred ex_anon.
*/
#define MAXSECFLAVORS 5
struct o2export_args {
int ex_flags; /* export related flags */
uid_t ex_root; /* mapping for root uid */
struct xucred ex_anon; /* mapping for anonymous user */
struct sockaddr *ex_addr; /* net address to which exported */
u_char ex_addrlen; /* and the net address length */
struct sockaddr *ex_mask; /* mask of valid bits in saddr */
u_char ex_masklen; /* and the smask length */
char *ex_indexfile; /* index file for WebNFS URLs */
int ex_numsecflavors; /* security flavor count */
int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
};
/*
* Export arguments for local filesystem mount calls.
*/
struct export_args {
uint64_t ex_flags; /* export related flags */
uid_t ex_root; /* mapping for root uid */
uid_t ex_uid; /* mapping for anonymous user */
int ex_ngroups;
gid_t *ex_groups;
struct sockaddr *ex_addr; /* net address to which exported */
u_char ex_addrlen; /* and the net address length */
struct sockaddr *ex_mask; /* mask of valid bits in saddr */
u_char ex_masklen; /* and the smask length */
char *ex_indexfile; /* index file for WebNFS URLs */
int ex_numsecflavors; /* security flavor count */
int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */
};
/*
* Structure holding information for a publicly exported filesystem
* (WebNFS). Currently the specs allow just for one such filesystem.
*/
struct nfs_public {
int np_valid; /* Do we hold valid information */
fhandle_t np_handle; /* Filehandle for pub fs (internal) */
struct mount *np_mount; /* Mountpoint of exported fs */
char *np_index; /* Index file */
};
/*
* Filesystem configuration information. One of these exists for each
* type of filesystem supported by the kernel. These are searched at
* mount time to identify the requested filesystem.
*
* XXX: Never change the first two arguments!
*/
struct vfsconf {
u_int vfc_version; /* ABI version number */
char vfc_name[MFSNAMELEN]; /* filesystem type name */
struct vfsops *vfc_vfsops; /* filesystem operations vector */
struct vfsops *vfc_vfsops_sd; /* ... signal-deferred */
int vfc_typenum; /* historic filesystem type number */
int vfc_refcount; /* number mounted of this type */
int vfc_flags; /* permanent flags */
int vfc_prison_flag; /* prison allow.mount.* flag */
struct vfsoptdecl *vfc_opts; /* mount options */
TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */
};
/* Userland version of the struct vfsconf. */
struct xvfsconf {
struct vfsops *vfc_vfsops; /* filesystem operations vector */
char vfc_name[MFSNAMELEN]; /* filesystem type name */
int vfc_typenum; /* historic filesystem type number */
int vfc_refcount; /* number mounted of this type */
int vfc_flags; /* permanent flags */
struct vfsconf *vfc_next; /* next in list */
};
#ifndef BURN_BRIDGES
struct ovfsconf {
void *vfc_vfsops;
char vfc_name[32];
int vfc_index;
int vfc_refcount;
int vfc_flags;
};
#endif
/*
* NB: these flags refer to IMPLEMENTATION properties, not properties of
* any actual mounts; i.e., it does not make sense to change the flags.
*/
#define VFCF_STATIC 0x00010000 /* statically compiled into kernel */
#define VFCF_NETWORK 0x00020000 /* may get data over the network */
#define VFCF_READONLY 0x00040000 /* writes are not implemented */
#define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */
#define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */
#define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */
#define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */
#define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */
#define VFCF_SBDRY 0x01000000 /* Stop at Boundary: defer stop requests
to kernel->user (AST) transition */
typedef uint32_t fsctlop_t;
struct vfsidctl {
int vc_vers; /* should be VFSIDCTL_VERS1 (below) */
fsid_t vc_fsid; /* fsid to operate on */
char vc_fstypename[MFSNAMELEN];
/* type of fs 'nfs' or '*' */
fsctlop_t vc_op; /* operation VFS_CTL_* (below) */
void *vc_ptr; /* pointer to data structure */
size_t vc_len; /* sizeof said structure */
u_int32_t vc_spare[12]; /* spare (must be zero) */
};
/* vfsidctl API version. */
#define VFS_CTL_VERS1 0x01
/*
* New style VFS sysctls, do not reuse/conflict with the namespace for
* private sysctls.
* All "global" sysctl ops have the 33rd bit set:
* 0x...1....
* Private sysctl ops should have the 33rd bit unset.
*/
#define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */
#define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */
#define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */
struct vfsquery {
u_int32_t vq_flags;
u_int32_t vq_spare[31];
};
/* vfsquery flags */
#define VQ_NOTRESP 0x0001 /* server down */
#define VQ_NEEDAUTH 0x0002 /* server bad auth */
#define VQ_LOWDISK 0x0004 /* we're low on space */
#define VQ_MOUNT 0x0008 /* new filesystem arrived */
#define VQ_UNMOUNT 0x0010 /* filesystem has left */
#define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */
#define VQ_ASSIST 0x0040 /* filesystem needs assistance from external
program */
#define VQ_NOTRESPLOCK 0x0080 /* server lockd down */
#define VQ_FLAG0100 0x0100 /* placeholder */
#define VQ_FLAG0200 0x0200 /* placeholder */
#define VQ_FLAG0400 0x0400 /* placeholder */
#define VQ_FLAG0800 0x0800 /* placeholder */
#define VQ_FLAG1000 0x1000 /* placeholder */
#define VQ_FLAG2000 0x2000 /* placeholder */
#define VQ_FLAG4000 0x4000 /* placeholder */
#define VQ_FLAG8000 0x8000 /* placeholder */
#ifdef _KERNEL
/* Point a sysctl request at a vfsidctl's data. */
#define VCTLTOREQ(vc, req) \
do { \
(req)->newptr = (vc)->vc_ptr; \
(req)->newlen = (vc)->vc_len; \
(req)->newidx = 0; \
} while (0)
#endif
struct iovec;
struct uio;
#ifdef _KERNEL
/*
* vfs_busy specific flags and mask.
*/
#define MBF_NOWAIT 0x01
#define MBF_MNTLSTLOCK 0x02
#define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK)
#ifdef MALLOC_DECLARE
MALLOC_DECLARE(M_MOUNT);
MALLOC_DECLARE(M_STATFS);
#endif
extern int maxvfsconf; /* highest defined filesystem type */
TAILQ_HEAD(vfsconfhead, vfsconf);
extern struct vfsconfhead vfsconf;
/*
* Operations supported on mounted filesystem.
*/
struct mount_args;
struct nameidata;
struct sysctl_req;
struct mntarg;
/*
* N.B., vfs_cmount is the ancient vfsop invoked by the old mount(2) syscall.
* The new way is vfs_mount.
*
* vfs_cmount implementations typically translate arguments from their
* respective old per-FS structures into the key-value list supported by
* nmount(2), then use kernel_mount(9) to mimic nmount(2) from kernelspace.
*
* Filesystems with mounters that use nmount(2) do not need to and should not
* implement vfs_cmount. Hopefully a future cleanup can remove vfs_cmount and
* mount(2) entirely.
*/
typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags);
typedef int vfs_unmount_t(struct mount *mp, int mntflags);
typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp);
typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg);
typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp);
typedef int vfs_sync_t(struct mount *mp, int waitfor);
typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags,
struct vnode **vpp);
typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp,
int flags, struct vnode **vpp);
typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam,
uint64_t *extflagsp, struct ucred **credanonp,
int *numsecflavors, int *secflavors);
typedef int vfs_init_t(struct vfsconf *);
typedef int vfs_uninit_t(struct vfsconf *);
typedef int vfs_extattrctl_t(struct mount *mp, int cmd,
struct vnode *filename_vp, int attrnamespace,
const char *attrname);
typedef int vfs_mount_t(struct mount *mp);
typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op,
struct sysctl_req *req);
typedef void vfs_susp_clean_t(struct mount *mp);
typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp);
typedef void vfs_purge_t(struct mount *mp);
struct vfsops {
vfs_mount_t *vfs_mount;
vfs_cmount_t *vfs_cmount;
vfs_unmount_t *vfs_unmount;
vfs_root_t *vfs_root;
vfs_root_t *vfs_cachedroot;
vfs_quotactl_t *vfs_quotactl;
vfs_statfs_t *vfs_statfs;
vfs_sync_t *vfs_sync;
vfs_vget_t *vfs_vget;
vfs_fhtovp_t *vfs_fhtovp;
vfs_checkexp_t *vfs_checkexp;
vfs_init_t *vfs_init;
vfs_uninit_t *vfs_uninit;
vfs_extattrctl_t *vfs_extattrctl;
vfs_sysctl_t *vfs_sysctl;
vfs_susp_clean_t *vfs_susp_clean;
vfs_notify_lowervp_t *vfs_reclaim_lowervp;
vfs_notify_lowervp_t *vfs_unlink_lowervp;
vfs_purge_t *vfs_purge;
vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */
};
vfs_statfs_t __vfs_statfs;
#define VFS_MOUNT(MP) ({ \
int _rc; \
\
TSRAW(curthread, TS_ENTER, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
_rc = (*(MP)->mnt_op->vfs_mount)(MP); \
TSRAW(curthread, TS_EXIT, "VFS_MOUNT", (MP)->mnt_vfc->vfc_name);\
_rc; })
#define VFS_UNMOUNT(MP, FORCE) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE); \
_rc; })
#define VFS_ROOT(MP, FLAGS, VPP) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP); \
_rc; })
#define VFS_CACHEDROOT(MP, FLAGS, VPP) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_cachedroot)(MP, FLAGS, VPP); \
_rc; })
#define VFS_QUOTACTL(MP, C, U, A) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A); \
_rc; })
#define VFS_STATFS(MP, SBP) ({ \
int _rc; \
\
_rc = __vfs_statfs((MP), (SBP)); \
_rc; })
#define VFS_SYNC(MP, WAIT) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT); \
_rc; })
#define VFS_VGET(MP, INO, FLAGS, VPP) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP); \
_rc; })
#define VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP); \
_rc; })
#define VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\
SEC); \
_rc; })
#define VFS_EXTATTRCTL(MP, C, FN, NS, N) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N); \
_rc; })
#define VFS_SYSCTL(MP, OP, REQ) ({ \
int _rc; \
\
_rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ); \
_rc; })
#define VFS_SUSP_CLEAN(MP) do { \
if (*(MP)->mnt_op->vfs_susp_clean != NULL) { \
(*(MP)->mnt_op->vfs_susp_clean)(MP); \
} \
} while (0)
#define VFS_RECLAIM_LOWERVP(MP, VP) do { \
if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) { \
(*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP)); \
} \
} while (0)
#define VFS_UNLINK_LOWERVP(MP, VP) do { \
if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) { \
(*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP)); \
} \
} while (0)
#define VFS_PURGE(MP) do { \
if (*(MP)->mnt_op->vfs_purge != NULL) { \
(*(MP)->mnt_op->vfs_purge)(MP); \
} \
} while (0)
#define VFS_KNOTE_LOCKED(vp, hint) do \
{ \
if (((vp)->v_vflag & VV_NOKNOTE) == 0) \
VN_KNOTE((vp), (hint), KNF_LISTLOCKED); \
} while (0)
#define VFS_KNOTE_UNLOCKED(vp, hint) do \
{ \
if (((vp)->v_vflag & VV_NOKNOTE) == 0) \
VN_KNOTE((vp), (hint), 0); \
} while (0)
#define VFS_NOTIFY_UPPER_RECLAIM 1
#define VFS_NOTIFY_UPPER_UNLINK 2
#include <sys/module.h>
/*
* Version numbers.
*/
#define VFS_VERSION_00 0x19660120
#define VFS_VERSION_01 0x20121030
#define VFS_VERSION_02 0x20180504
#define VFS_VERSION VFS_VERSION_02
#define VFS_SET(vfsops, fsname, flags) \
static struct vfsconf fsname ## _vfsconf = { \
.vfc_version = VFS_VERSION, \
.vfc_name = #fsname, \
.vfc_vfsops = &vfsops, \
.vfc_typenum = -1, \
.vfc_flags = flags, \
}; \
static moduledata_t fsname ## _mod = { \
#fsname, \
vfs_modevent, \
& fsname ## _vfsconf \
}; \
DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE)
/*
* exported vnode operations
*/
int dounmount(struct mount *, int, struct thread *);
int kernel_mount(struct mntarg *ma, uint64_t flags);
int kernel_vmount(int flags, ...);
struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len);
struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name);
struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...);
struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len);
void statfs_scale_blocks(struct statfs *sf, long max_size);
struct vfsconf *vfs_byname(const char *);
struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *);
void vfs_mount_destroy(struct mount *);
void vfs_event_signal(fsid_t *, u_int32_t, intptr_t);
void vfs_freeopts(struct vfsoptlist *opts);
void vfs_deleteopt(struct vfsoptlist *opts, const char *name);
int vfs_buildopts(struct uio *auio, struct vfsoptlist **options);
int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w,
uint64_t val);
int vfs_getopt(struct vfsoptlist *, const char *, void **, int *);
int vfs_getopt_pos(struct vfsoptlist *opts, const char *name);
int vfs_getopt_size(struct vfsoptlist *opts, const char *name,
off_t *value);
char *vfs_getopts(struct vfsoptlist *, const char *, int *error);
int vfs_copyopt(struct vfsoptlist *, const char *, void *, int);
int vfs_filteropt(struct vfsoptlist *, const char **legal);
void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...);
int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...);
int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value,
int len);
int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value,
int len);
int vfs_setopts(struct vfsoptlist *opts, const char *name,
const char *value);
int vfs_setpublicfs /* set publicly exported fs */
(struct mount *, struct netexport *, struct export_args *);
void vfs_periodic(struct mount *, int);
int vfs_busy(struct mount *, int);
int vfs_export /* process mount export info */
(struct mount *, struct export_args *);
void vfs_allocate_syncvnode(struct mount *);
void vfs_deallocate_syncvnode(struct mount *);
int vfs_donmount(struct thread *td, uint64_t fsflags,
struct uio *fsoptions);
void vfs_getnewfsid(struct mount *);
struct cdev *vfs_getrootfsid(struct mount *);
struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */
struct mount *vfs_busyfs(fsid_t *);
int vfs_modevent(module_t, int, void *);
void vfs_mount_error(struct mount *, const char *, ...);
void vfs_mountroot(void); /* mount our root filesystem */
void vfs_mountedfrom(struct mount *, const char *from);
void vfs_notify_upper(struct vnode *, int);
struct mount *vfs_ref_from_vp(struct vnode *);
void vfs_ref(struct mount *);
void vfs_rel(struct mount *);
struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *,
struct ucred *);
int vfs_suser(struct mount *, struct thread *);
void vfs_unbusy(struct mount *);
void vfs_unmountall(void);
extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */
extern struct mtx_padalign mountlist_mtx;
extern struct nfs_public nfs_pub;
extern struct sx vfsconf_sx;
#define vfsconf_lock() sx_xlock(&vfsconf_sx)
#define vfsconf_unlock() sx_xunlock(&vfsconf_sx)
#define vfsconf_slock() sx_slock(&vfsconf_sx)
#define vfsconf_sunlock() sx_sunlock(&vfsconf_sx)
struct vnode *mntfs_allocvp(struct mount *, struct vnode *);
void mntfs_freevp(struct vnode *);
/*
* Declarations for these vfs default operations are located in
* kern/vfs_default.c. They will be automatically used to replace
* null entries in VFS ops tables when registering a new filesystem
* type in the global table.
*/
vfs_root_t vfs_stdroot;
vfs_quotactl_t vfs_stdquotactl;
vfs_statfs_t vfs_stdstatfs;
vfs_sync_t vfs_stdsync;
vfs_sync_t vfs_stdnosync;
vfs_vget_t vfs_stdvget;
vfs_fhtovp_t vfs_stdfhtovp;
vfs_checkexp_t vfs_stdcheckexp;
vfs_init_t vfs_stdinit;
vfs_uninit_t vfs_stduninit;
vfs_extattrctl_t vfs_stdextattrctl;
vfs_sysctl_t vfs_stdsysctl;
void syncer_suspend(void);
void syncer_resume(void);
struct vnode *vfs_cache_root_clear(struct mount *);
void vfs_cache_root_set(struct mount *, struct vnode *);
void vfs_op_barrier_wait(struct mount *);
void vfs_op_enter(struct mount *);
void vfs_op_exit_locked(struct mount *);
void vfs_op_exit(struct mount *);
#ifdef DIAGNOSTIC
void vfs_assert_mount_counters(struct mount *);
void vfs_dump_mount_counters(struct mount *);
#else
#define vfs_assert_mount_counters(mp) do { } while (0)
#define vfs_dump_mount_counters(mp) do { } while (0)
#endif
enum mount_counter { MNT_COUNT_REF, MNT_COUNT_LOCKREF, MNT_COUNT_WRITEOPCOUNT };
int vfs_mount_fetch_counter(struct mount *, enum mount_counter);
void suspend_all_fs(void);
void resume_all_fs(void);
/*
* Code transitioning mnt_vfs_ops to > 0 issues IPIs until it observes
* all CPUs not executing code enclosed by thread_in_ops_pcpu variable.
*
* This provides an invariant that by the time the last CPU is observed not
* executing, everyone else entering will see the counter > 0 and exit.
*
* Note there is no barrier between vfs_ops and the rest of the code in the
* section. It is not necessary as the writer has to wait for everyone to drain
* before making any changes or only make changes safe while the section is
* executed.
*/
#define vfs_mount_pcpu(mp) zpcpu_get(mp->mnt_pcpu)
#define vfs_mount_pcpu_remote(mp, cpu) zpcpu_get_cpu(mp->mnt_pcpu, cpu)
#define vfs_op_thread_entered(mp) ({ \
MPASS(curthread->td_critnest > 0); \
struct mount_pcpu *_mpcpu = vfs_mount_pcpu(mp); \
_mpcpu->mntp_thread_in_ops == 1; \
})
#define vfs_op_thread_enter_crit(mp, _mpcpu) ({ \
bool _retval_crit = true; \
MPASS(curthread->td_critnest > 0); \
_mpcpu = vfs_mount_pcpu(mp); \
MPASS(mpcpu->mntp_thread_in_ops == 0); \
_mpcpu->mntp_thread_in_ops = 1; \
__compiler_membar(); \
if (__predict_false(mp->mnt_vfs_ops > 0)) { \
vfs_op_thread_exit_crit(mp, _mpcpu); \
_retval_crit = false; \
} \
_retval_crit; \
})
#define vfs_op_thread_enter(mp, _mpcpu) ({ \
bool _retval; \
critical_enter(); \
_retval = vfs_op_thread_enter_crit(mp, _mpcpu); \
if (__predict_false(!_retval)) \
critical_exit(); \
_retval; \
})
#define vfs_op_thread_exit_crit(mp, _mpcpu) do { \
MPASS(_mpcpu == vfs_mount_pcpu(mp)); \
MPASS(_mpcpu->mntp_thread_in_ops == 1); \
__compiler_membar(); \
_mpcpu->mntp_thread_in_ops = 0; \
} while (0)
#define vfs_op_thread_exit(mp, _mpcpu) do { \
vfs_op_thread_exit_crit(mp, _mpcpu); \
critical_exit(); \
} while (0)
#define vfs_mp_count_add_pcpu(_mpcpu, count, val) do { \
MPASS(_mpcpu->mntp_thread_in_ops == 1); \
_mpcpu->mntp_##count += val; \
} while (0)
#define vfs_mp_count_sub_pcpu(_mpcpu, count, val) do { \
MPASS(_mpcpu->mntp_thread_in_ops == 1); \
_mpcpu->mntp_##count -= val; \
} while (0)
#else /* !_KERNEL */
#include <sys/cdefs.h>
struct stat;
__BEGIN_DECLS
int fhlink(struct fhandle *, const char *);
int fhlinkat(struct fhandle *, int, const char *);
int fhopen(const struct fhandle *, int);
int fhreadlink(struct fhandle *, char *, size_t);
int fhstat(const struct fhandle *, struct stat *);
int fhstatfs(const struct fhandle *, struct statfs *);
int fstatfs(int, struct statfs *);
int getfh(const char *, fhandle_t *);
int getfhat(int, char *, struct fhandle *, int);
int getfsstat(struct statfs *, long, int);
int getmntinfo(struct statfs **, int);
int lgetfh(const char *, fhandle_t *);
int mount(const char *, const char *, int, void *);
int nmount(struct iovec *, unsigned int, int);
int statfs(const char *, struct statfs *);
int unmount(const char *, int);
/* C library stuff */
int getvfsbyname(const char *, struct xvfsconf *);
__END_DECLS
#endif /* _KERNEL */
#endif /* !_SYS_MOUNT_H_ */
diff --git a/sys/sys/pipe.h b/sys/sys/pipe.h
index f8b008151c6a..c9f222ffb01a 100644
--- a/sys/sys/pipe.h
+++ b/sys/sys/pipe.h
@@ -1,153 +1,153 @@
/*-
* Copyright (c) 1996 John S. Dyson
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice immediately at the beginning of the file, without modification,
* this list of conditions, and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Absolutely no warranty of function or purpose is made by the author
* John S. Dyson.
* 4. This work was done expressly for inclusion into FreeBSD. Other use
* is allowed if this notation is included.
* 5. Modifications may be freely made to this file if the above conditions
* are met.
*
* $FreeBSD$
*/
#ifndef _SYS_PIPE_H_
#define _SYS_PIPE_H_
-#ifndef _KERNEL
-#error "no user-serviceable parts inside"
-#endif
-
/*
* Pipe buffer size, keep moderate in value, pipes take kva space.
*/
#ifndef PIPE_SIZE
#define PIPE_SIZE 16384
#endif
#ifndef BIG_PIPE_SIZE
#define BIG_PIPE_SIZE (64*1024)
#endif
#ifndef SMALL_PIPE_SIZE
#define SMALL_PIPE_SIZE PAGE_SIZE
#endif
/*
* PIPE_MINDIRECT MUST be smaller than PIPE_SIZE and MUST be bigger
* than PIPE_BUF.
*/
#ifndef PIPE_MINDIRECT
#define PIPE_MINDIRECT 8192
#endif
#define PIPENPAGES (BIG_PIPE_SIZE / PAGE_SIZE + 1)
+#ifdef _KERNEL
/*
* See sys_pipe.c for info on what these limits mean.
*/
extern long maxpipekva;
extern struct fileops pipeops;
+#endif
/*
* Pipe buffer information.
* Separate in, out, cnt are used to simplify calculations.
* Buffered write is active when the buffer.cnt field is set.
*/
struct pipebuf {
u_int cnt; /* number of chars currently in buffer */
u_int in; /* in pointer */
u_int out; /* out pointer */
u_int size; /* size of buffer */
caddr_t buffer; /* kva of buffer */
};
/*
* Information to support direct transfers between processes for pipes.
*/
struct pipemapping {
vm_size_t cnt; /* number of chars in buffer */
vm_size_t pos; /* current position of transfer */
int npages; /* number of pages */
vm_page_t ms[PIPENPAGES]; /* pages in source process */
};
/*
* Bits in pipe_state.
*/
#define PIPE_ASYNC 0x004 /* Async? I/O. */
#define PIPE_WANTR 0x008 /* Reader wants some characters. */
#define PIPE_WANTW 0x010 /* Writer wants space to put characters. */
#define PIPE_WANT 0x020 /* Pipe is wanted to be run-down. */
#define PIPE_SEL 0x040 /* Pipe has a select active. */
#define PIPE_EOF 0x080 /* Pipe is in EOF condition. */
#define PIPE_LOCKFL 0x100 /* Process has exclusive access to pointers/data. */
#define PIPE_LWANT 0x200 /* Process wants exclusive access to pointers/data. */
#define PIPE_DIRECTW 0x400 /* Pipe direct write active. */
#define PIPE_DIRECTOK 0x800 /* Direct mode ok. */
/*
* Bits in pipe_type.
*/
#define PIPE_TYPE_NAMED 0x001 /* Is a named pipe. */
/*
* Per-pipe data structure.
* Two of these are linked together to produce bi-directional pipes.
*/
struct pipe {
struct pipebuf pipe_buffer; /* data storage */
struct pipemapping pipe_pages; /* wired pages for direct I/O */
struct selinfo pipe_sel; /* for compat with select */
struct timespec pipe_atime; /* time of last access */
struct timespec pipe_mtime; /* time of last modify */
struct timespec pipe_ctime; /* time of status change */
struct sigio *pipe_sigio; /* information for async I/O */
struct pipe *pipe_peer; /* link with other direction */
struct pipepair *pipe_pair; /* container structure pointer */
u_short pipe_state; /* pipe status info */
u_char pipe_type; /* pipe type info */
u_char pipe_present; /* still present? */
int pipe_waiters; /* pipelock waiters */
int pipe_busy; /* busy flag, mostly to handle rundown sanely */
int pipe_wgen; /* writer generation for named pipe */
ino_t pipe_ino; /* fake inode for stat(2) */
};
/*
* Values for the pipe_present.
*/
#define PIPE_ACTIVE 1
#define PIPE_CLOSING 2
#define PIPE_FINALIZED 3
/*
* Container structure to hold the two pipe endpoints, mutex, and label
* pointer.
*/
struct pipepair {
struct pipe pp_rpipe;
struct pipe pp_wpipe;
struct mtx pp_mtx;
struct label *pp_label;
};
#define PIPE_MTX(pipe) (&(pipe)->pipe_pair->pp_mtx)
#define PIPE_LOCK(pipe) mtx_lock(PIPE_MTX(pipe))
#define PIPE_UNLOCK(pipe) mtx_unlock(PIPE_MTX(pipe))
#define PIPE_LOCK_ASSERT(pipe, type) mtx_assert(PIPE_MTX(pipe), (type))
+#ifdef _KERNEL
void pipe_dtor(struct pipe *dpipe);
int pipe_named_ctor(struct pipe **ppipe, struct thread *td);
void pipeselwakeup(struct pipe *cpipe);
+#endif
#endif /* !_SYS_PIPE_H_ */
diff --git a/sys/ufs/ufs/extattr.h b/sys/ufs/ufs/extattr.h
index 781cc782f415..fa20f8c0b936 100644
--- a/sys/ufs/ufs/extattr.h
+++ b/sys/ufs/ufs/extattr.h
@@ -1,147 +1,148 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 1999-2001 Robert N. M. Watson
* All rights reserved.
*
* This software was developed by Robert Watson for the TrustedBSD Project.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD$
*/
/*
* Developed by the TrustedBSD Project.
* Support for extended filesystem attributes.
*/
#ifndef _UFS_UFS_EXTATTR_H_
#define _UFS_UFS_EXTATTR_H_
#define UFS_EXTATTR_MAGIC 0x00b5d5ec
#define UFS_EXTATTR_VERSION 0x00000003
#define UFS_EXTATTR_FSROOTSUBDIR ".attribute"
#define UFS_EXTATTR_SUBDIR_SYSTEM "system"
#define UFS_EXTATTR_SUBDIR_USER "user"
#define UFS_EXTATTR_MAXEXTATTRNAME 65 /* including null */
#define UFS_EXTATTR_ATTR_FLAG_INUSE 0x00000001 /* attr has been set */
#define UFS_EXTATTR_PERM_KERNEL 0x00000000
#define UFS_EXTATTR_PERM_ROOT 0x00000001
#define UFS_EXTATTR_PERM_OWNER 0x00000002
#define UFS_EXTATTR_PERM_ANYONE 0x00000003
#define UFS_EXTATTR_UEPM_INITIALIZED 0x00000001
#define UFS_EXTATTR_UEPM_STARTED 0x00000002
#define UFS_EXTATTR_CMD_START 0x00000001
#define UFS_EXTATTR_CMD_STOP 0x00000002
#define UFS_EXTATTR_CMD_ENABLE 0x00000003
#define UFS_EXTATTR_CMD_DISABLE 0x00000004
struct ufs_extattr_fileheader {
u_int uef_magic; /* magic number for sanity checking */
u_int uef_version; /* version of attribute file */
u_int uef_size; /* size of attributes, w/o header */
};
struct ufs_extattr_header {
u_int ueh_flags; /* flags for attribute */
u_int ueh_len; /* local defined length; <= uef_size */
u_int32_t ueh_i_gen; /* generation number for sanity */
/* data follows the header */
};
/*
* This structure defines the required fields of an extended-attribute header.
*/
struct extattr {
uint32_t ea_length; /* length of this attribute */
uint8_t ea_namespace; /* name space of this attribute */
uint8_t ea_contentpadlen; /* bytes of padding at end of attribute */
uint8_t ea_namelength; /* length of attribute name */
char ea_name[1]; /* attribute name (NOT nul-terminated) */
/* padding, if any, to align attribute content to 8 byte boundary */
/* extended attribute content follows */
};
/*
* These macros are used to access and manipulate an extended attribute:
*
* EXTATTR_NEXT(eap) returns a pointer to the next extended attribute
* following eap.
* EXTATTR_CONTENT(eap) returns a pointer to the extended attribute
* content referenced by eap.
* EXTATTR_CONTENT_SIZE(eap) returns the size of the extended attribute
* content referenced by eap.
*/
#define EXTATTR_NEXT(eap) \
((struct extattr *)(__DECONST(char *, (eap)) + (eap)->ea_length))
#define EXTATTR_CONTENT(eap) \
(void *)(((u_char *)(eap)) + EXTATTR_BASE_LENGTH(eap))
#define EXTATTR_CONTENT_SIZE(eap) \
((eap)->ea_length - EXTATTR_BASE_LENGTH(eap) - (eap)->ea_contentpadlen)
/* -1 below compensates for ea_name[1] */
#define EXTATTR_BASE_LENGTH(eap) \
roundup2((sizeof(struct extattr) - 1 + (eap)->ea_namelength), 8)
-#ifdef _KERNEL
-
-#include <sys/_sx.h>
-
struct vnode;
LIST_HEAD(ufs_extattr_list_head, ufs_extattr_list_entry);
struct ufs_extattr_list_entry {
LIST_ENTRY(ufs_extattr_list_entry) uele_entries;
struct ufs_extattr_fileheader uele_fileheader;
int uele_attrnamespace;
char uele_attrname[UFS_EXTATTR_MAXEXTATTRNAME];
struct vnode *uele_backing_vnode;
};
+#include <sys/_lock.h>
+#include <sys/_sx.h>
+
struct ucred;
struct ufs_extattr_per_mount {
struct sx uepm_lock;
struct ufs_extattr_list_head uepm_list;
struct ucred *uepm_ucred;
int uepm_flags;
};
+#ifdef _KERNEL
+
struct vop_getextattr_args;
struct vop_deleteextattr_args;
struct vop_setextattr_args;
void ufs_extattr_uepm_init(struct ufs_extattr_per_mount *uepm);
void ufs_extattr_uepm_destroy(struct ufs_extattr_per_mount *uepm);
int ufs_extattr_start(struct mount *mp, struct thread *td);
int ufs_extattr_autostart(struct mount *mp, struct thread *td);
int ufs_extattr_stop(struct mount *mp, struct thread *td);
int ufs_extattrctl(struct mount *mp, int cmd, struct vnode *filename,
int attrnamespace, const char *attrname);
int ufs_getextattr(struct vop_getextattr_args *ap);
int ufs_deleteextattr(struct vop_deleteextattr_args *ap);
int ufs_setextattr(struct vop_setextattr_args *ap);
void ufs_extattr_vnode_inactive(struct vnode *vp);
#endif /* !_KERNEL */
#endif /* !_UFS_UFS_EXTATTR_H_ */
diff --git a/sys/ufs/ufs/inode.h b/sys/ufs/ufs/inode.h
index 4515dcbed401..e00a89b012d0 100644
--- a/sys/ufs/ufs/inode.h
+++ b/sys/ufs/ufs/inode.h
@@ -1,308 +1,310 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1982, 1989, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)inode.h 8.9 (Berkeley) 5/14/95
* $FreeBSD$
*/
#ifndef _UFS_UFS_INODE_H_
#define _UFS_UFS_INODE_H_
#include <sys/lock.h>
#include <sys/queue.h>
#include <ufs/ufs/dinode.h>
#include <sys/seqc.h>
#ifdef DIAGNOSTIC
#include <sys/stack.h>
#endif
/*
* This must agree with the definition in <ufs/ufs/dir.h>.
*/
#define doff_t int32_t
#ifdef DIAGNOSTIC
struct iown_tracker {
struct thread *tr_owner;
struct stack tr_st;
struct stack tr_unlock;
int tr_gen;
};
#endif
/*
* The inode is used to describe each active (or recently active) file in the
* UFS filesystem. It is composed of two types of information. The first part
* is the information that is needed only while the file is active (such as
* the identity of the file and linkage to speed its lookup). The second part
* is the permanent meta-data associated with the file which is read in
* from the permanent dinode from long term storage when the file becomes
* active, and is put back when the file is no longer being used.
*
* An inode may only be changed while holding either the exclusive
* vnode lock or the shared vnode lock and the vnode interlock. We use
* the latter only for "read" and "get" operations that require
* changing i_flag, or a timestamp. This locking protocol allows executing
* those operations without having to upgrade the vnode lock from shared to
* exclusive.
*/
struct inode {
TAILQ_ENTRY(inode) i_nextsnap; /* snapshot file list. */
struct vnode *i_vnode;/* Vnode associated with this inode. */
struct ufsmount *i_ump;/* Ufsmount point associated with this inode. */
struct dquot *i_dquot[MAXQUOTAS]; /* Dquot structures. */
union {
struct dirhash *dirhash; /* Hashing for large directories. */
daddr_t *snapblklist; /* Collect expunged snapshot blocks. */
} i_un;
/*
* The real copy of the on-disk inode.
*/
union {
struct ufs1_dinode *din1; /* UFS1 on-disk dinode. */
struct ufs2_dinode *din2; /* UFS2 on-disk dinode. */
} dinode_u;
ino_t i_number; /* The identity of the inode. */
u_int32_t i_flag; /* flags, see below */
int i_effnlink; /* i_nlink when I/O completes */
/*
* Side effects; used during directory lookup.
*/
int32_t i_count; /* Size of free slot in directory. */
doff_t i_endoff; /* End of useful stuff in directory. */
doff_t i_diroff; /* Offset in dir, where we found last entry. */
doff_t i_offset; /* Offset of free space in directory. */
#ifdef DIAGNOSTIC
int i_lock_gen;
struct iown_tracker i_count_tracker;
struct iown_tracker i_endoff_tracker;
struct iown_tracker i_offset_tracker;
#endif
int i_nextclustercg; /* last cg searched for cluster */
/*
* Data for extended attribute modification.
*/
u_char *i_ea_area; /* Pointer to malloced copy of EA area */
unsigned i_ea_len; /* Length of i_ea_area */
int i_ea_error; /* First errno in transaction */
int i_ea_refs; /* Number of users of EA area */
/*
* Copies from the on-disk dinode itself.
*/
u_int64_t i_size; /* File byte count. */
u_int64_t i_gen; /* Generation number. */
u_int32_t i_flags; /* Status flags (chflags). */
u_int32_t i_uid; /* File owner. */
u_int32_t i_gid; /* File group. */
u_int16_t i_mode; /* IFMT, permissions; see below. */
int16_t i_nlink; /* File link count. */
};
/*
* These flags are kept in i_flag.
*/
#define IN_ACCESS 0x0001 /* Access time update request. */
#define IN_CHANGE 0x0002 /* Inode change time update request. */
#define IN_UPDATE 0x0004 /* Modification time update request. */
#define IN_MODIFIED 0x0008 /* Inode has been modified. */
#define IN_NEEDSYNC 0x0010 /* Inode requires fsync. */
#define IN_LAZYMOD 0x0020 /* Modified, but don't write yet. */
#define IN_LAZYACCESS 0x0040 /* Process IN_ACCESS after the
suspension finished */
#define IN_EA_LOCKED 0x0080 /* Extended attributes locked */
#define IN_EA_LOCKWAIT 0x0100 /* Want extended attributes lock */
#define IN_TRUNCATED 0x0200 /* Journaled truncation pending. */
#define IN_UFS2 0x0400 /* UFS2 vs UFS1 */
#define IN_IBLKDATA 0x0800 /* datasync requires inode block
update */
#define IN_SIZEMOD 0x1000 /* Inode size has been modified */
#define IN_ENDOFF 0x2000 /* Free space at the end of directory,
try to truncate when possible */
#define PRINT_INODE_FLAGS "\20\20b16\17b15\16b14\15sizemod" \
"\14iblkdata\13is_ufs2\12truncated\11ea_lockwait\10ea_locked" \
"\7lazyaccess\6lazymod\5needsync\4modified\3update\2change\1access"
#define UFS_INODE_FLAG_LAZY_MASK \
(IN_ACCESS | IN_CHANGE | IN_MODIFIED | IN_UPDATE | IN_LAZYMOD | \
IN_LAZYACCESS)
/*
* Some flags can persist a vnode transitioning to 0 hold count and being tkaen
* off the list.
*/
#define UFS_INODE_FLAG_LAZY_MASK_ASSERTABLE \
(UFS_INODE_FLAG_LAZY_MASK & ~(IN_LAZYMOD | IN_LAZYACCESS))
#define UFS_INODE_SET_MODE(ip, mode) do { \
struct inode *_ip = (ip); \
int _mode = (mode); \
\
ASSERT_VOP_IN_SEQC(ITOV(_ip)); \
atomic_store_short(&(_ip)->i_mode, _mode); \
} while (0)
#define UFS_INODE_SET_FLAG(ip, flags) do { \
struct inode *_ip = (ip); \
struct vnode *_vp = ITOV(_ip); \
int _flags = (flags); \
\
_ip->i_flag |= _flags; \
if (_flags & UFS_INODE_FLAG_LAZY_MASK) \
vlazy(_vp); \
} while (0)
#define UFS_INODE_SET_FLAG_SHARED(ip, flags) do { \
struct inode *_ip = (ip); \
struct vnode *_vp = ITOV(_ip); \
int _flags = (flags); \
\
ASSERT_VI_UNLOCKED(_vp, __func__); \
if ((_ip->i_flag & (_flags)) != _flags) { \
VI_LOCK(_vp); \
_ip->i_flag |= _flags; \
if (_flags & UFS_INODE_FLAG_LAZY_MASK) \
vlazy(_vp); \
VI_UNLOCK(_vp); \
} \
} while (0)
#define i_dirhash i_un.dirhash
#define i_snapblklist i_un.snapblklist
#define i_din1 dinode_u.din1
#define i_din2 dinode_u.din2
-#ifdef _KERNEL
-
#define ITOUMP(ip) ((ip)->i_ump)
#define ITODEV(ip) (ITOUMP(ip)->um_dev)
#define ITODEVVP(ip) (ITOUMP(ip)->um_devvp)
#define ITOFS(ip) (ITOUMP(ip)->um_fs)
#define ITOVFS(ip) ((ip)->i_vnode->v_mount)
+#ifdef _KERNEL
+
static inline _Bool
I_IS_UFS1(const struct inode *ip)
{
return ((ip->i_flag & IN_UFS2) == 0);
}
static inline _Bool
I_IS_UFS2(const struct inode *ip)
{
return ((ip->i_flag & IN_UFS2) != 0);
}
+#endif /* _KERNEL */
/*
* The DIP macro is used to access fields in the dinode that are
* not cached in the inode itself.
*/
#define DIP(ip, field) (I_IS_UFS1(ip) ? (ip)->i_din1->d##field : \
(ip)->i_din2->d##field)
#define DIP_SET(ip, field, val) do { \
if (I_IS_UFS1(ip)) \
(ip)->i_din1->d##field = (val); \
else \
(ip)->i_din2->d##field = (val); \
} while (0)
#define SHORTLINK(ip) (I_IS_UFS1(ip) ? \
(caddr_t)(ip)->i_din1->di_db : (caddr_t)(ip)->i_din2->di_db)
#define IS_SNAPSHOT(ip) ((ip)->i_flags & SF_SNAPSHOT)
/*
* Structure used to pass around logical block paths generated by
* ufs_getlbns and used by truncate and bmap code.
*/
struct indir {
ufs2_daddr_t in_lbn; /* Logical block number. */
int in_off; /* Offset in buffer. */
};
/* Convert between inode pointers and vnode pointers. */
#define VTOI(vp) ((struct inode *)(vp)->v_data)
#define VTOI_SMR(vp) ((struct inode *)vn_load_v_data_smr(vp))
#define ITOV(ip) ((ip)->i_vnode)
/* Determine if soft dependencies are being done */
#define DOINGSOFTDEP(vp) \
(((vp)->v_mount->mnt_flag & (MNT_SOFTDEP | MNT_SUJ)) != 0)
#define MOUNTEDSOFTDEP(mp) (((mp)->mnt_flag & (MNT_SOFTDEP | MNT_SUJ)) != 0)
#define DOINGSUJ(vp) (((vp)->v_mount->mnt_flag & MNT_SUJ) != 0)
#define MOUNTEDSUJ(mp) (((mp)->mnt_flag & MNT_SUJ) != 0)
/* This overlays the fid structure (see mount.h). */
struct ufid {
u_int16_t ufid_len; /* Length of structure. */
u_int16_t ufid_pad; /* Force 32-bit alignment. */
uint32_t ufid_ino; /* File number (ino). */
uint32_t ufid_gen; /* Generation number. */
};
+#ifdef _KERNEL
#ifdef DIAGNOSTIC
void ufs_init_trackers(struct inode *ip);
void ufs_unlock_tracker(struct inode *ip);
doff_t ufs_get_i_offset(struct inode *ip, const char *file, int line);
void ufs_set_i_offset(struct inode *ip, doff_t off, const char *file, int line);
#define I_OFFSET(ip) ufs_get_i_offset(ip, __FILE__, __LINE__)
#define SET_I_OFFSET(ip, off) ufs_set_i_offset(ip, off, __FILE__, __LINE__)
int32_t ufs_get_i_count(struct inode *ip, const char *file, int line);
void ufs_set_i_count(struct inode *ip, int32_t cnt, const char *file, int line);
#define I_COUNT(ip) ufs_get_i_count(ip, __FILE__, __LINE__)
#define SET_I_COUNT(ip, cnt) ufs_set_i_count(ip, cnt, __FILE__, __LINE__)
doff_t ufs_get_i_endoff(struct inode *ip, const char *file, int line);
void ufs_set_i_endoff(struct inode *ip, doff_t off, const char *file, int line);
#define I_ENDOFF(ip) ufs_get_i_endoff(ip, __FILE__, __LINE__)
#define SET_I_ENDOFF(ip, off) ufs_set_i_endoff(ip, off, __FILE__, __LINE__)
#else
#define I_OFFSET(ip) ((ip)->i_offset)
#define SET_I_OFFSET(ip, off) ((ip)->i_offset = (off))
#define I_COUNT(ip) ((ip)->i_count)
#define SET_I_COUNT(ip, cnt) ((ip)->i_count = cnt)
#define I_ENDOFF(ip) ((ip)->i_endoff)
#define SET_I_ENDOFF(ip, off) ((ip)->i_endoff = off)
#endif
#endif /* _KERNEL */
#endif /* !_UFS_UFS_INODE_H_ */
diff --git a/sys/ufs/ufs/ufsmount.h b/sys/ufs/ufs/ufsmount.h
index 57e163c11d77..0dfcecb178af 100644
--- a/sys/ufs/ufs/ufsmount.h
+++ b/sys/ufs/ufs/ufsmount.h
@@ -1,195 +1,197 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)ufsmount.h 8.6 (Berkeley) 3/30/95
* $FreeBSD$
*/
#ifndef _UFS_UFS_UFSMOUNT_H_
#define _UFS_UFS_UFSMOUNT_H_
/*
* Arguments to mount UFS-based filesystems
*/
struct ufs_args {
char *fspec; /* block special device to mount */
struct oexport_args export; /* network export information */
};
-#ifdef _KERNEL
-
#include <sys/_task.h>
+#ifdef _KERNEL
#ifdef MALLOC_DECLARE
MALLOC_DECLARE(M_UFSMNT);
MALLOC_DECLARE(M_TRIM);
#endif
+#endif /* _KERNEL */
struct buf;
struct inode;
struct nameidata;
struct taskqueue;
struct timeval;
struct ucred;
struct uio;
struct vnode;
struct ufs_extattr_per_mount;
struct jblocks;
struct inodedep;
TAILQ_HEAD(inodedeplst, inodedep);
LIST_HEAD(bmsafemaphd, bmsafemap);
LIST_HEAD(trimlist_hashhead, ffs_blkfree_trim_params);
struct fsfail_task {
struct task task;
fsid_t fsid;
};
+#include <sys/_lock.h>
+#include <sys/_mutex.h>
+
/*
* This structure describes the UFS specific mount structure data.
* The function operators are used to support different versions of
* UFS (UFS1, UFS2, etc).
*
* Lock reference:
* c - set at allocation then constant until freed
* i - ufsmount interlock (UFS_LOCK / UFS_UNLOCK)
* q - associated quota file is locked
* r - ref to parent mount structure is held (vfs_busy / vfs_unbusy)
* u - managed by user process fsck_ufs
*/
struct ufsmount {
struct mount *um_mountp; /* (r) filesystem vfs struct */
struct cdev *um_dev; /* (r) device mounted */
struct g_consumer *um_cp; /* (r) GEOM access point */
struct bufobj *um_bo; /* (r) Buffer cache object */
struct vnode *um_odevvp; /* (r) devfs dev vnode */
struct vnode *um_devvp; /* (r) mntfs private vnode */
u_long um_fstype; /* (c) type of filesystem */
struct fs *um_fs; /* (r) pointer to superblock */
struct ufs_extattr_per_mount um_extattr; /* (c) extended attrs */
u_long um_nindir; /* (c) indirect ptrs per blk */
u_long um_bptrtodb; /* (c) indir disk block ptr */
u_long um_seqinc; /* (c) inc between seq blocks */
struct mtx um_lock; /* (c) Protects ufsmount & fs */
pid_t um_fsckpid; /* (u) PID can do fsck sysctl */
struct mount_softdeps *um_softdep; /* (c) softdep mgmt structure */
struct vnode *um_quotas[MAXQUOTAS]; /* (q) pointer to quota files */
struct ucred *um_cred[MAXQUOTAS]; /* (q) quota file access cred */
time_t um_btime[MAXQUOTAS]; /* (q) block quota time limit */
time_t um_itime[MAXQUOTAS]; /* (q) inode quota time limit */
char um_qflags[MAXQUOTAS]; /* (i) quota specific flags */
int64_t um_savedmaxfilesize; /* (c) track maxfilesize */
u_int um_flags; /* (i) filesystem flags */
struct timeval um_last_fullmsg; /* (i) last full msg time */
int um_secs_fullmsg; /* (i) seconds since full msg */
struct timeval um_last_integritymsg; /* (i) last integrity msg */
int um_secs_integritymsg; /* (i) secs since integ msg */
u_int um_trim_inflight; /* (i) outstanding trim count */
u_int um_trim_inflight_blks; /* (i) outstanding trim blks */
u_long um_trim_total; /* (i) total trim count */
u_long um_trim_total_blks; /* (i) total trim block count */
struct taskqueue *um_trim_tq; /* (c) trim request queue */
struct trimlist_hashhead *um_trimhash; /* (i) trimlist hash table */
u_long um_trimlisthashsize; /* (i) trim hash table size-1 */
struct fsfail_task *um_fsfail_task; /* (i) task for fsfail cleanup*/
/* (c) - below function ptrs */
int (*um_balloc)(struct vnode *, off_t, int, struct ucred *,
int, struct buf **);
int (*um_blkatoff)(struct vnode *, off_t, char **, struct buf **);
int (*um_truncate)(struct vnode *, off_t, int, struct ucred *);
int (*um_update)(struct vnode *, int);
int (*um_valloc)(struct vnode *, int, struct ucred *,
struct vnode **);
int (*um_vfree)(struct vnode *, ino_t, int);
void (*um_ifree)(struct ufsmount *, struct inode *);
int (*um_rdonly)(struct inode *);
void (*um_snapgone)(struct inode *);
int (*um_check_blkno)(struct mount *, ino_t, daddr_t, int);
};
/*
* filesystem flags
*/
#define UM_CANDELETE 0x00000001 /* devvp supports TRIM */
#define UM_WRITESUSPENDED 0x00000002 /* suspension in progress */
#define UM_CANSPEEDUP 0x00000004 /* devvp supports SPEEDUP */
#define UM_FSFAIL_CLEANUP 0x00000008 /* need cleanup after
unrecoverable error */
/*
* function prototypes
*/
#define UFS_BALLOC(aa, bb, cc, dd, ee, ff) \
VFSTOUFS((aa)->v_mount)->um_balloc(aa, bb, cc, dd, ee, ff)
#define UFS_BLKATOFF(aa, bb, cc, dd) \
VFSTOUFS((aa)->v_mount)->um_blkatoff(aa, bb, cc, dd)
#define UFS_TRUNCATE(aa, bb, cc, dd) \
VFSTOUFS((aa)->v_mount)->um_truncate(aa, bb, cc, dd)
#define UFS_UPDATE(aa, bb) VFSTOUFS((aa)->v_mount)->um_update(aa, bb)
#define UFS_VALLOC(aa, bb, cc, dd) \
VFSTOUFS((aa)->v_mount)->um_valloc(aa, bb, cc, dd)
#define UFS_VFREE(aa, bb, cc) VFSTOUFS((aa)->v_mount)->um_vfree(aa, bb, cc)
#define UFS_IFREE(aa, bb) ((aa)->um_ifree(aa, bb))
#define UFS_RDONLY(aa) (ITOUMP(aa)->um_rdonly(aa))
#define UFS_SNAPGONE(aa) (ITOUMP(aa)->um_snapgone(aa))
#define UFS_CHECK_BLKNO(aa, bb, cc, dd) \
(VFSTOUFS(aa)->um_check_blkno == NULL ? 0 : \
VFSTOUFS(aa)->um_check_blkno(aa, bb, cc, dd))
#define UFS_LOCK(aa) mtx_lock(&(aa)->um_lock)
#define UFS_UNLOCK(aa) mtx_unlock(&(aa)->um_lock)
#define UFS_MTX(aa) (&(aa)->um_lock)
/*
* Filesystem types
*/
#define UFS1 1
#define UFS2 2
/*
* Flags describing the state of quotas.
*/
#define QTF_OPENING 0x01 /* Q_QUOTAON in progress */
#define QTF_CLOSING 0x02 /* Q_QUOTAOFF in progress */
#define QTF_64BIT 0x04 /* 64-bit quota file */
/* Convert mount ptr to ufsmount ptr. */
#define VFSTOUFS(mp) ((struct ufsmount *)((mp)->mnt_data))
#define UFSTOVFS(ump) (ump)->um_mountp
/*
* Macros to access filesystem parameters in the ufsmount structure.
* Used by ufs_bmap.
*/
#define MNINDIR(ump) ((ump)->um_nindir)
#define blkptrtodb(ump, b) ((b) << (ump)->um_bptrtodb)
#define is_sequential(ump, a, b) ((b) == (a) + ump->um_seqinc)
-#endif /* _KERNEL */
#endif
diff --git a/usr.sbin/makefs/ffs.c b/usr.sbin/makefs/ffs.c
index 4dfb37889b53..dcc6eb5cbed7 100644
--- a/usr.sbin/makefs/ffs.c
+++ b/usr.sbin/makefs/ffs.c
@@ -1,1189 +1,1189 @@
/* $NetBSD: ffs.c,v 1.45 2011/10/09 22:49:26 christos Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (c) 2001 Wasabi Systems, Inc.
* All rights reserved.
*
* Written by Luke Mewburn for Wasabi Systems, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed for the NetBSD Project by
* Wasabi Systems, Inc.
* 4. The name of Wasabi Systems, Inc. may not be used to endorse
* or promote products derived from this software without specific prior
* written permission.
*
* THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
/*
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)ffs_alloc.c 8.19 (Berkeley) 7/13/95
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#if HAVE_NBTOOL_CONFIG_H
#include "nbtool_config.h"
#endif
#include <sys/param.h>
#include <sys/mount.h>
#include <assert.h>
#include <errno.h>
#include <fcntl.h>
#include <stdarg.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
#include <util.h>
-#include "makefs.h"
-#include "ffs.h"
-
#if HAVE_STRUCT_STATVFS_F_IOSIZE && HAVE_FSTATVFS
#include <sys/statvfs.h>
#endif
#include <ufs/ufs/dinode.h>
#include <ufs/ufs/dir.h>
#include <ufs/ffs/fs.h>
-
#include "ffs/ufs_bswap.h"
#include "ffs/ufs_inode.h"
#include "ffs/newfs_extern.h"
#include "ffs/ffs_extern.h"
+#undef clrbuf
+#include "makefs.h"
+#include "ffs.h"
+
#undef DIP
#define DIP(dp, field) \
((ffs_opts->version == 1) ? \
(dp)->ffs1_din.di_##field : (dp)->ffs2_din.di_##field)
/*
* Various file system defaults (cribbed from newfs(8)).
*/
#define DFL_FRAGSIZE 4096 /* fragment size */
#define DFL_BLKSIZE 32768 /* block size */
#define DFL_SECSIZE 512 /* sector size */
#define DFL_CYLSPERGROUP 65536 /* cylinders per group */
#define DFL_FRAGSPERINODE 4 /* fragments per inode */
#define DFL_ROTDELAY 0 /* rotational delay */
#define DFL_NRPOS 1 /* rotational positions */
#define DFL_RPM 3600 /* rpm of disk */
#define DFL_NSECTORS 64 /* # of sectors */
#define DFL_NTRACKS 16 /* # of tracks */
typedef struct {
u_char *buf; /* buf for directory */
doff_t size; /* full size of buf */
doff_t cur; /* offset of current entry */
} dirbuf_t;
static int ffs_create_image(const char *, fsinfo_t *);
static void ffs_dump_fsinfo(fsinfo_t *);
static void ffs_dump_dirbuf(dirbuf_t *, const char *, int);
static void ffs_make_dirbuf(dirbuf_t *, const char *, fsnode *, int);
static int ffs_populate_dir(const char *, fsnode *, fsinfo_t *);
static void ffs_size_dir(fsnode *, fsinfo_t *);
static void ffs_validate(const char *, fsnode *, fsinfo_t *);
static void ffs_write_file(union dinode *, uint32_t, void *, fsinfo_t *);
static void ffs_write_inode(union dinode *, uint32_t, const fsinfo_t *);
static void *ffs_build_dinode1(struct ufs1_dinode *, dirbuf_t *, fsnode *,
fsnode *, fsinfo_t *);
static void *ffs_build_dinode2(struct ufs2_dinode *, dirbuf_t *, fsnode *,
fsnode *, fsinfo_t *);
/* publicly visible functions */
void
ffs_prep_opts(fsinfo_t *fsopts)
{
ffs_opt_t *ffs_opts = ecalloc(1, sizeof(*ffs_opts));
const option_t ffs_options[] = {
{ 'b', "bsize", &ffs_opts->bsize, OPT_INT32,
1, INT_MAX, "block size" },
{ 'f', "fsize", &ffs_opts->fsize, OPT_INT32,
1, INT_MAX, "fragment size" },
{ 'd', "density", &ffs_opts->density, OPT_INT32,
1, INT_MAX, "bytes per inode" },
{ 'm', "minfree", &ffs_opts->minfree, OPT_INT32,
0, 99, "minfree" },
{ 'M', "maxbpg", &ffs_opts->maxbpg, OPT_INT32,
1, INT_MAX, "max blocks per file in a cg" },
{ 'a', "avgfilesize", &ffs_opts->avgfilesize, OPT_INT32,
1, INT_MAX, "expected average file size" },
{ 'n', "avgfpdir", &ffs_opts->avgfpdir, OPT_INT32,
1, INT_MAX, "expected # of files per directory" },
{ 'x', "extent", &ffs_opts->maxbsize, OPT_INT32,
1, INT_MAX, "maximum # extent size" },
{ 'g', "maxbpcg", &ffs_opts->maxblkspercg, OPT_INT32,
1, INT_MAX, "max # of blocks per group" },
{ 'v', "version", &ffs_opts->version, OPT_INT32,
1, 2, "UFS version" },
{ 'o', "optimization", NULL, OPT_STRBUF,
0, 0, "Optimization (time|space)" },
{ 'l', "label", ffs_opts->label, OPT_STRARRAY,
1, sizeof(ffs_opts->label), "UFS label" },
{ 's', "softupdates", &ffs_opts->softupdates, OPT_INT32,
0, 1, "enable softupdates" },
{ .name = NULL }
};
ffs_opts->bsize= -1;
ffs_opts->fsize= -1;
ffs_opts->cpg= -1;
ffs_opts->density= -1;
ffs_opts->minfree= -1;
ffs_opts->optimization= -1;
ffs_opts->maxcontig= -1;
ffs_opts->maxbpg= -1;
ffs_opts->avgfilesize= -1;
ffs_opts->avgfpdir= -1;
ffs_opts->version = 1;
ffs_opts->softupdates = 0;
fsopts->fs_specific = ffs_opts;
fsopts->fs_options = copy_opts(ffs_options);
}
void
ffs_cleanup_opts(fsinfo_t *fsopts)
{
free(fsopts->fs_specific);
free(fsopts->fs_options);
}
int
ffs_parse_opts(const char *option, fsinfo_t *fsopts)
{
ffs_opt_t *ffs_opts = fsopts->fs_specific;
option_t *ffs_options = fsopts->fs_options;
char buf[1024];
int rv;
assert(option != NULL);
assert(fsopts != NULL);
assert(ffs_opts != NULL);
if (debug & DEBUG_FS_PARSE_OPTS)
printf("ffs_parse_opts: got `%s'\n", option);
rv = set_option(ffs_options, option, buf, sizeof(buf));
if (rv == -1)
return 0;
if (ffs_options[rv].name == NULL)
abort();
switch (ffs_options[rv].letter) {
case 'o':
if (strcmp(buf, "time") == 0) {
ffs_opts->optimization = FS_OPTTIME;
} else if (strcmp(buf, "space") == 0) {
ffs_opts->optimization = FS_OPTSPACE;
} else {
warnx("Invalid optimization `%s'", buf);
return 0;
}
break;
default:
break;
}
return 1;
}
void
ffs_makefs(const char *image, const char *dir, fsnode *root, fsinfo_t *fsopts)
{
struct fs *superblock;
struct timeval start;
assert(image != NULL);
assert(dir != NULL);
assert(root != NULL);
assert(fsopts != NULL);
if (debug & DEBUG_FS_MAKEFS)
printf("ffs_makefs: image %s directory %s root %p\n",
image, dir, root);
/* validate tree and options */
TIMER_START(start);
ffs_validate(dir, root, fsopts);
TIMER_RESULTS(start, "ffs_validate");
printf("Calculated size of `%s': %lld bytes, %lld inodes\n",
image, (long long)fsopts->size, (long long)fsopts->inodes);
/* create image */
TIMER_START(start);
if (ffs_create_image(image, fsopts) == -1)
errx(1, "Image file `%s' not created.", image);
TIMER_RESULTS(start, "ffs_create_image");
fsopts->curinode = UFS_ROOTINO;
if (debug & DEBUG_FS_MAKEFS)
putchar('\n');
/* populate image */
printf("Populating `%s'\n", image);
TIMER_START(start);
if (! ffs_populate_dir(dir, root, fsopts))
errx(1, "Image file `%s' not populated.", image);
TIMER_RESULTS(start, "ffs_populate_dir");
/* ensure no outstanding buffers remain */
if (debug & DEBUG_FS_MAKEFS)
bcleanup();
/* update various superblock parameters */
superblock = fsopts->superblock;
superblock->fs_fmod = 0;
superblock->fs_old_cstotal.cs_ndir = superblock->fs_cstotal.cs_ndir;
superblock->fs_old_cstotal.cs_nbfree = superblock->fs_cstotal.cs_nbfree;
superblock->fs_old_cstotal.cs_nifree = superblock->fs_cstotal.cs_nifree;
superblock->fs_old_cstotal.cs_nffree = superblock->fs_cstotal.cs_nffree;
/* write out superblock; image is now complete */
ffs_write_superblock(fsopts->superblock, fsopts);
if (close(fsopts->fd) == -1)
err(1, "Closing `%s'", image);
fsopts->fd = -1;
printf("Image `%s' complete\n", image);
}
/* end of public functions */
static void
ffs_validate(const char *dir, fsnode *root, fsinfo_t *fsopts)
{
int32_t ncg = 1;
#ifdef notyet
int32_t spc, nspf, ncyl, fssize;
#endif
ffs_opt_t *ffs_opts = fsopts->fs_specific;
assert(dir != NULL);
assert(root != NULL);
assert(fsopts != NULL);
assert(ffs_opts != NULL);
if (debug & DEBUG_FS_VALIDATE) {
printf("ffs_validate: before defaults set:\n");
ffs_dump_fsinfo(fsopts);
}
/* set FFS defaults */
if (fsopts->sectorsize == -1)
fsopts->sectorsize = DFL_SECSIZE;
if (ffs_opts->fsize == -1)
ffs_opts->fsize = MAX(DFL_FRAGSIZE, fsopts->sectorsize);
if (ffs_opts->bsize == -1)
ffs_opts->bsize = MIN(DFL_BLKSIZE, 8 * ffs_opts->fsize);
if (ffs_opts->cpg == -1)
ffs_opts->cpg = DFL_CYLSPERGROUP;
else
ffs_opts->cpgflg = 1;
/* fsopts->density is set below */
if (ffs_opts->nsectors == -1)
ffs_opts->nsectors = DFL_NSECTORS;
if (ffs_opts->minfree == -1)
ffs_opts->minfree = MINFREE;
if (ffs_opts->optimization == -1)
ffs_opts->optimization = DEFAULTOPT;
if (ffs_opts->maxcontig == -1)
ffs_opts->maxcontig =
MAX(1, MIN(MAXPHYS, FFS_MAXBSIZE) / ffs_opts->bsize);
/* XXX ondisk32 */
if (ffs_opts->maxbpg == -1)
ffs_opts->maxbpg = ffs_opts->bsize / sizeof(int32_t);
if (ffs_opts->avgfilesize == -1)
ffs_opts->avgfilesize = AVFILESIZ;
if (ffs_opts->avgfpdir == -1)
ffs_opts->avgfpdir = AFPDIR;
if (fsopts->maxsize > 0 &&
roundup(fsopts->minsize, ffs_opts->bsize) > fsopts->maxsize)
errx(1, "`%s' minsize of %lld rounded up to ffs bsize of %d "
"exceeds maxsize %lld. Lower bsize, or round the minimum "
"and maximum sizes to bsize.", dir,
(long long)fsopts->minsize, ffs_opts->bsize,
(long long)fsopts->maxsize);
/* calculate size of tree */
ffs_size_dir(root, fsopts);
fsopts->inodes += UFS_ROOTINO; /* include first two inodes */
if (debug & DEBUG_FS_VALIDATE)
printf("ffs_validate: size of tree: %lld bytes, %lld inodes\n",
(long long)fsopts->size, (long long)fsopts->inodes);
/* add requested slop */
fsopts->size += fsopts->freeblocks;
fsopts->inodes += fsopts->freefiles;
if (fsopts->freefilepc > 0)
fsopts->inodes =
fsopts->inodes * (100 + fsopts->freefilepc) / 100;
if (fsopts->freeblockpc > 0)
fsopts->size =
fsopts->size * (100 + fsopts->freeblockpc) / 100;
/* add space needed for superblocks */
/*
* The old SBOFF (SBLOCK_UFS1) is used here because makefs is
* typically used for small filesystems where space matters.
* XXX make this an option.
*/
fsopts->size += (SBLOCK_UFS1 + SBLOCKSIZE) * ncg;
/* add space needed to store inodes, x3 for blockmaps, etc */
if (ffs_opts->version == 1)
fsopts->size += ncg * DINODE1_SIZE *
roundup(fsopts->inodes / ncg,
ffs_opts->bsize / DINODE1_SIZE);
else
fsopts->size += ncg * DINODE2_SIZE *
roundup(fsopts->inodes / ncg,
ffs_opts->bsize / DINODE2_SIZE);
/* add minfree */
if (ffs_opts->minfree > 0)
fsopts->size =
fsopts->size * (100 + ffs_opts->minfree) / 100;
/*
* XXX any other fs slop to add, such as csum's, bitmaps, etc ??
*/
if (fsopts->size < fsopts->minsize) /* ensure meets minimum size */
fsopts->size = fsopts->minsize;
/* round up to the next block */
fsopts->size = roundup(fsopts->size, ffs_opts->bsize);
/* round up to requested block size, if any */
if (fsopts->roundup > 0)
fsopts->size = roundup(fsopts->size, fsopts->roundup);
/* calculate density if necessary */
if (ffs_opts->density == -1)
ffs_opts->density = fsopts->size / fsopts->inodes + 1;
if (debug & DEBUG_FS_VALIDATE) {
printf("ffs_validate: after defaults set:\n");
ffs_dump_fsinfo(fsopts);
printf("ffs_validate: dir %s; %lld bytes, %lld inodes\n",
dir, (long long)fsopts->size, (long long)fsopts->inodes);
}
/* now check calculated sizes vs requested sizes */
if (fsopts->maxsize > 0 && fsopts->size > fsopts->maxsize) {
errx(1, "`%s' size of %lld is larger than the maxsize of %lld.",
dir, (long long)fsopts->size, (long long)fsopts->maxsize);
}
}
static void
ffs_dump_fsinfo(fsinfo_t *f)
{
ffs_opt_t *fs = f->fs_specific;
printf("fsopts at %p\n", f);
printf("\tsize %lld, inodes %lld, curinode %u\n",
(long long)f->size, (long long)f->inodes, f->curinode);
printf("\tminsize %lld, maxsize %lld\n",
(long long)f->minsize, (long long)f->maxsize);
printf("\tfree files %lld, freefile %% %d\n",
(long long)f->freefiles, f->freefilepc);
printf("\tfree blocks %lld, freeblock %% %d\n",
(long long)f->freeblocks, f->freeblockpc);
printf("\tneedswap %d, sectorsize %d\n", f->needswap, f->sectorsize);
printf("\tbsize %d, fsize %d, cpg %d, density %d\n",
fs->bsize, fs->fsize, fs->cpg, fs->density);
printf("\tnsectors %d, rpm %d, minfree %d\n",
fs->nsectors, fs->rpm, fs->minfree);
printf("\tmaxcontig %d, maxbpg %d\n",
fs->maxcontig, fs->maxbpg);
printf("\toptimization %s\n",
fs->optimization == FS_OPTSPACE ? "space" : "time");
}
static int
ffs_create_image(const char *image, fsinfo_t *fsopts)
{
#if HAVE_STRUCT_STATVFS_F_IOSIZE && HAVE_FSTATVFS
struct statvfs sfs;
#endif
struct fs *fs;
char *buf;
int i, bufsize;
off_t bufrem;
int oflags = O_RDWR | O_CREAT;
time_t tstamp;
assert (image != NULL);
assert (fsopts != NULL);
/* create image */
if (fsopts->offset == 0)
oflags |= O_TRUNC;
if ((fsopts->fd = open(image, oflags, 0666)) == -1) {
warn("Can't open `%s' for writing", image);
return (-1);
}
/* zero image */
#if HAVE_STRUCT_STATVFS_F_IOSIZE && HAVE_FSTATVFS
if (fstatvfs(fsopts->fd, &sfs) == -1) {
#endif
bufsize = 8192;
#if HAVE_STRUCT_STATVFS_F_IOSIZE && HAVE_FSTATVFS
warn("can't fstatvfs `%s', using default %d byte chunk",
image, bufsize);
} else
bufsize = sfs.f_iosize;
#endif
bufrem = fsopts->size;
if (fsopts->sparse) {
if (ftruncate(fsopts->fd, bufrem) == -1) {
warn("sparse option disabled.");
fsopts->sparse = 0;
}
}
if (fsopts->sparse) {
/* File truncated at bufrem. Remaining is 0 */
bufrem = 0;
buf = NULL;
} else {
if (debug & DEBUG_FS_CREATE_IMAGE)
printf("zero-ing image `%s', %lld sectors, "
"using %d byte chunks\n", image, (long long)bufrem,
bufsize);
buf = ecalloc(1, bufsize);
}
if (fsopts->offset != 0)
if (lseek(fsopts->fd, fsopts->offset, SEEK_SET) == -1) {
warn("can't seek");
free(buf);
return -1;
}
while (bufrem > 0) {
i = write(fsopts->fd, buf, MIN(bufsize, bufrem));
if (i == -1) {
warn("zeroing image, %lld bytes to go",
(long long)bufrem);
free(buf);
return (-1);
}
bufrem -= i;
}
if (buf)
free(buf);
/* make the file system */
if (debug & DEBUG_FS_CREATE_IMAGE)
printf("calling mkfs(\"%s\", ...)\n", image);
if (stampst.st_ino != 0)
tstamp = stampst.st_ctime;
else
tstamp = start_time.tv_sec;
srandom(tstamp);
fs = ffs_mkfs(image, fsopts, tstamp);
fsopts->superblock = (void *)fs;
if (debug & DEBUG_FS_CREATE_IMAGE) {
time_t t;
t = (time_t)((struct fs *)fsopts->superblock)->fs_time;
printf("mkfs returned %p; fs_time %s",
fsopts->superblock, ctime(&t));
printf("fs totals: nbfree %lld, nffree %lld, nifree %lld, ndir %lld\n",
(long long)fs->fs_cstotal.cs_nbfree,
(long long)fs->fs_cstotal.cs_nffree,
(long long)fs->fs_cstotal.cs_nifree,
(long long)fs->fs_cstotal.cs_ndir);
}
if (fs->fs_cstotal.cs_nifree + UFS_ROOTINO < fsopts->inodes) {
warnx(
"Image file `%s' has %lld free inodes; %lld are required.",
image,
(long long)(fs->fs_cstotal.cs_nifree + UFS_ROOTINO),
(long long)fsopts->inodes);
return (-1);
}
return (fsopts->fd);
}
static void
ffs_size_dir(fsnode *root, fsinfo_t *fsopts)
{
struct direct tmpdir;
fsnode * node;
int curdirsize, this;
ffs_opt_t *ffs_opts = fsopts->fs_specific;
/* node may be NULL (empty directory) */
assert(fsopts != NULL);
assert(ffs_opts != NULL);
if (debug & DEBUG_FS_SIZE_DIR)
printf("ffs_size_dir: entry: bytes %lld inodes %lld\n",
(long long)fsopts->size, (long long)fsopts->inodes);
#define ADDDIRENT(e) do { \
tmpdir.d_namlen = strlen((e)); \
this = DIRSIZ_SWAP(0, &tmpdir, 0); \
if (debug & DEBUG_FS_SIZE_DIR_ADD_DIRENT) \
printf("ADDDIRENT: was: %s (%d) this %d cur %d\n", \
e, tmpdir.d_namlen, this, curdirsize); \
if (this + curdirsize > roundup(curdirsize, DIRBLKSIZ)) \
curdirsize = roundup(curdirsize, DIRBLKSIZ); \
curdirsize += this; \
if (debug & DEBUG_FS_SIZE_DIR_ADD_DIRENT) \
printf("ADDDIRENT: now: %s (%d) this %d cur %d\n", \
e, tmpdir.d_namlen, this, curdirsize); \
} while (0);
/*
* XXX this needs to take into account extra space consumed
* by indirect blocks, etc.
*/
#define ADDSIZE(x) do { \
fsopts->size += roundup((x), ffs_opts->fsize); \
} while (0);
curdirsize = 0;
for (node = root; node != NULL; node = node->next) {
ADDDIRENT(node->name);
if (node == root) { /* we're at "." */
assert(strcmp(node->name, ".") == 0);
ADDDIRENT("..");
} else if ((node->inode->flags & FI_SIZED) == 0) {
/* don't count duplicate names */
node->inode->flags |= FI_SIZED;
if (debug & DEBUG_FS_SIZE_DIR_NODE)
printf("ffs_size_dir: `%s' size %lld\n",
node->name,
(long long)node->inode->st.st_size);
fsopts->inodes++;
if (node->type == S_IFREG)
ADDSIZE(node->inode->st.st_size);
if (node->type == S_IFLNK) {
size_t slen;
slen = strlen(node->symlink) + 1;
if (slen >= (ffs_opts->version == 1 ?
UFS1_MAXSYMLINKLEN :
UFS2_MAXSYMLINKLEN))
ADDSIZE(slen);
}
}
if (node->type == S_IFDIR)
ffs_size_dir(node->child, fsopts);
}
ADDSIZE(curdirsize);
if (debug & DEBUG_FS_SIZE_DIR)
printf("ffs_size_dir: exit: size %lld inodes %lld\n",
(long long)fsopts->size, (long long)fsopts->inodes);
}
static void *
ffs_build_dinode1(struct ufs1_dinode *dinp, dirbuf_t *dbufp, fsnode *cur,
fsnode *root, fsinfo_t *fsopts)
{
size_t slen;
void *membuf;
struct stat *st = stampst.st_ino != 0 ? &stampst : &cur->inode->st;
memset(dinp, 0, sizeof(*dinp));
dinp->di_mode = cur->inode->st.st_mode;
dinp->di_nlink = cur->inode->nlink;
dinp->di_size = cur->inode->st.st_size;
#if HAVE_STRUCT_STAT_ST_FLAGS
dinp->di_flags = cur->inode->st.st_flags;
#endif
dinp->di_gen = random();
dinp->di_uid = cur->inode->st.st_uid;
dinp->di_gid = cur->inode->st.st_gid;
dinp->di_atime = st->st_atime;
dinp->di_mtime = st->st_mtime;
dinp->di_ctime = st->st_ctime;
#if HAVE_STRUCT_STAT_ST_MTIMENSEC
dinp->di_atimensec = st->st_atimensec;
dinp->di_mtimensec = st->st_mtimensec;
dinp->di_ctimensec = st->st_ctimensec;
#endif
/* not set: di_db, di_ib, di_blocks, di_spare */
membuf = NULL;
if (cur == root) { /* "."; write dirbuf */
membuf = dbufp->buf;
dinp->di_size = dbufp->size;
} else if (S_ISBLK(cur->type) || S_ISCHR(cur->type)) {
dinp->di_size = 0; /* a device */
dinp->di_rdev =
ufs_rw32(cur->inode->st.st_rdev, fsopts->needswap);
} else if (S_ISLNK(cur->type)) { /* symlink */
slen = strlen(cur->symlink);
if (slen < UFS1_MAXSYMLINKLEN) { /* short link */
memcpy(dinp->di_db, cur->symlink, slen);
} else
membuf = cur->symlink;
dinp->di_size = slen;
}
return membuf;
}
static void *
ffs_build_dinode2(struct ufs2_dinode *dinp, dirbuf_t *dbufp, fsnode *cur,
fsnode *root, fsinfo_t *fsopts)
{
size_t slen;
void *membuf;
struct stat *st = stampst.st_ino != 0 ? &stampst : &cur->inode->st;
memset(dinp, 0, sizeof(*dinp));
dinp->di_mode = cur->inode->st.st_mode;
dinp->di_nlink = cur->inode->nlink;
dinp->di_size = cur->inode->st.st_size;
#if HAVE_STRUCT_STAT_ST_FLAGS
dinp->di_flags = cur->inode->st.st_flags;
#endif
dinp->di_gen = random();
dinp->di_uid = cur->inode->st.st_uid;
dinp->di_gid = cur->inode->st.st_gid;
dinp->di_atime = st->st_atime;
dinp->di_mtime = st->st_mtime;
dinp->di_ctime = st->st_ctime;
#if HAVE_STRUCT_STAT_BIRTHTIME
dinp->di_birthtime = st->st_birthtime;
#else
dinp->di_birthtime = st->st_ctime;
#endif
#if HAVE_STRUCT_STAT_ST_MTIMENSEC
dinp->di_atimensec = st->st_atimensec;
dinp->di_mtimensec = st->st_mtimensec;
dinp->di_ctimensec = st->st_ctimensec;
#if HAVE_STRUCT_STAT_BIRTHTIME
dinp->di_birthnsec = st->st_birthtimensec;
#else
dinp->di_birthnsec = st->st_ctimensec;
#endif
#endif
/* not set: di_db, di_ib, di_blocks, di_spare */
membuf = NULL;
if (cur == root) { /* "."; write dirbuf */
membuf = dbufp->buf;
dinp->di_size = dbufp->size;
} else if (S_ISBLK(cur->type) || S_ISCHR(cur->type)) {
dinp->di_size = 0; /* a device */
dinp->di_rdev =
ufs_rw64(cur->inode->st.st_rdev, fsopts->needswap);
} else if (S_ISLNK(cur->type)) { /* symlink */
slen = strlen(cur->symlink);
if (slen < UFS2_MAXSYMLINKLEN) { /* short link */
memcpy(dinp->di_db, cur->symlink, slen);
} else
membuf = cur->symlink;
dinp->di_size = slen;
}
return membuf;
}
static int
ffs_populate_dir(const char *dir, fsnode *root, fsinfo_t *fsopts)
{
fsnode *cur;
dirbuf_t dirbuf;
union dinode din;
void *membuf;
char path[MAXPATHLEN + 1];
ffs_opt_t *ffs_opts = fsopts->fs_specific;
assert(dir != NULL);
assert(root != NULL);
assert(fsopts != NULL);
assert(ffs_opts != NULL);
(void)memset(&dirbuf, 0, sizeof(dirbuf));
if (debug & DEBUG_FS_POPULATE)
printf("ffs_populate_dir: PASS 1 dir %s node %p\n", dir, root);
/*
* pass 1: allocate inode numbers, build directory `file'
*/
for (cur = root; cur != NULL; cur = cur->next) {
if ((cur->inode->flags & FI_ALLOCATED) == 0) {
cur->inode->flags |= FI_ALLOCATED;
if (cur == root && cur->parent != NULL)
cur->inode->ino = cur->parent->inode->ino;
else {
cur->inode->ino = fsopts->curinode;
fsopts->curinode++;
}
}
ffs_make_dirbuf(&dirbuf, cur->name, cur, fsopts->needswap);
if (cur == root) { /* we're at "."; add ".." */
ffs_make_dirbuf(&dirbuf, "..",
cur->parent == NULL ? cur : cur->parent->first,
fsopts->needswap);
root->inode->nlink++; /* count my parent's link */
} else if (cur->child != NULL)
root->inode->nlink++; /* count my child's link */
/*
* XXX possibly write file and long symlinks here,
* ensuring that blocks get written before inodes?
* otoh, this isn't a real filesystem, so who
* cares about ordering? :-)
*/
}
if (debug & DEBUG_FS_POPULATE_DIRBUF)
ffs_dump_dirbuf(&dirbuf, dir, fsopts->needswap);
/*
* pass 2: write out dirbuf, then non-directories at this level
*/
if (debug & DEBUG_FS_POPULATE)
printf("ffs_populate_dir: PASS 2 dir %s\n", dir);
for (cur = root; cur != NULL; cur = cur->next) {
if (cur->inode->flags & FI_WRITTEN)
continue; /* skip hard-linked entries */
cur->inode->flags |= FI_WRITTEN;
if (cur->contents == NULL) {
if (snprintf(path, sizeof(path), "%s/%s/%s", cur->root,
cur->path, cur->name) >= (int)sizeof(path))
errx(1, "Pathname too long.");
}
if (cur->child != NULL)
continue; /* child creates own inode */
/* build on-disk inode */
if (ffs_opts->version == 1)
membuf = ffs_build_dinode1(&din.ffs1_din, &dirbuf, cur,
root, fsopts);
else
membuf = ffs_build_dinode2(&din.ffs2_din, &dirbuf, cur,
root, fsopts);
if (debug & DEBUG_FS_POPULATE_NODE) {
printf("ffs_populate_dir: writing ino %d, %s",
cur->inode->ino, inode_type(cur->type));
if (cur->inode->nlink > 1)
printf(", nlink %d", cur->inode->nlink);
putchar('\n');
}
if (membuf != NULL) {
ffs_write_file(&din, cur->inode->ino, membuf, fsopts);
} else if (S_ISREG(cur->type)) {
ffs_write_file(&din, cur->inode->ino,
(cur->contents) ? cur->contents : path, fsopts);
} else {
assert (! S_ISDIR(cur->type));
ffs_write_inode(&din, cur->inode->ino, fsopts);
}
}
/*
* pass 3: write out sub-directories
*/
if (debug & DEBUG_FS_POPULATE)
printf("ffs_populate_dir: PASS 3 dir %s\n", dir);
for (cur = root; cur != NULL; cur = cur->next) {
if (cur->child == NULL)
continue;
if ((size_t)snprintf(path, sizeof(path), "%s/%s", dir,
cur->name) >= sizeof(path))
errx(1, "Pathname too long.");
if (! ffs_populate_dir(path, cur->child, fsopts))
return (0);
}
if (debug & DEBUG_FS_POPULATE)
printf("ffs_populate_dir: DONE dir %s\n", dir);
/* cleanup */
if (dirbuf.buf != NULL)
free(dirbuf.buf);
return (1);
}
static void
ffs_write_file(union dinode *din, uint32_t ino, void *buf, fsinfo_t *fsopts)
{
int isfile, ffd;
char *fbuf, *p;
off_t bufleft, chunk, offset;
ssize_t nread;
struct inode in;
- struct buf * bp;
+ struct m_buf * bp;
ffs_opt_t *ffs_opts = fsopts->fs_specific;
- struct vnode vp = { fsopts, NULL };
+ struct m_vnode vp = { fsopts, NULL };
assert (din != NULL);
assert (buf != NULL);
assert (fsopts != NULL);
assert (ffs_opts != NULL);
isfile = S_ISREG(DIP(din, mode));
fbuf = NULL;
ffd = -1;
p = NULL;
in.i_fs = (struct fs *)fsopts->superblock;
- in.i_devvp = &vp;
+ in.i_devvp = (void *)&vp;
if (debug & DEBUG_FS_WRITE_FILE) {
printf(
"ffs_write_file: ino %u, din %p, isfile %d, %s, size %lld",
ino, din, isfile, inode_type(DIP(din, mode) & S_IFMT),
(long long)DIP(din, size));
if (isfile)
printf(", file '%s'\n", (char *)buf);
else
printf(", buffer %p\n", buf);
}
in.i_number = ino;
in.i_size = DIP(din, size);
if (ffs_opts->version == 1)
memcpy(&in.i_din.ffs1_din, &din->ffs1_din,
sizeof(in.i_din.ffs1_din));
else
memcpy(&in.i_din.ffs2_din, &din->ffs2_din,
sizeof(in.i_din.ffs2_din));
if (DIP(din, size) == 0)
goto write_inode_and_leave; /* mmm, cheating */
if (isfile) {
fbuf = emalloc(ffs_opts->bsize);
if ((ffd = open((char *)buf, O_RDONLY, 0444)) == -1) {
err(EXIT_FAILURE, "Can't open `%s' for reading", (char *)buf);
}
} else {
p = buf;
}
chunk = 0;
for (bufleft = DIP(din, size); bufleft > 0; bufleft -= chunk) {
chunk = MIN(bufleft, ffs_opts->bsize);
if (!isfile)
;
else if ((nread = read(ffd, fbuf, chunk)) == -1)
err(EXIT_FAILURE, "Reading `%s', %lld bytes to go",
(char *)buf, (long long)bufleft);
else if (nread != chunk)
errx(EXIT_FAILURE, "Reading `%s', %lld bytes to go, "
"read %zd bytes, expected %ju bytes, does "
"metalog size= attribute mismatch source size?",
(char *)buf, (long long)bufleft, nread,
(uintmax_t)chunk);
else
p = fbuf;
offset = DIP(din, size) - bufleft;
if (debug & DEBUG_FS_WRITE_FILE_BLOCK)
printf(
"ffs_write_file: write %p offset %lld size %lld left %lld\n",
p, (long long)offset,
(long long)chunk, (long long)bufleft);
/*
* XXX if holey support is desired, do the check here
*
* XXX might need to write out last bit in fragroundup
* sized chunk. however, ffs_balloc() handles this for us
*/
errno = ffs_balloc(&in, offset, chunk, &bp);
bad_ffs_write_file:
if (errno != 0)
err(1,
"Writing inode %d (%s), bytes %lld + %lld",
ino,
isfile ? (char *)buf :
inode_type(DIP(din, mode) & S_IFMT),
(long long)offset, (long long)chunk);
memcpy(bp->b_data, p, chunk);
errno = bwrite(bp);
if (errno != 0)
goto bad_ffs_write_file;
brelse(bp);
if (!isfile)
p += chunk;
}
write_inode_and_leave:
ffs_write_inode(&in.i_din, in.i_number, fsopts);
if (fbuf)
free(fbuf);
if (ffd != -1)
close(ffd);
}
static void
ffs_dump_dirbuf(dirbuf_t *dbuf, const char *dir, int needswap)
{
doff_t i;
struct direct *de;
uint16_t reclen;
assert (dbuf != NULL);
assert (dir != NULL);
printf("ffs_dump_dirbuf: dir %s size %d cur %d\n",
dir, dbuf->size, dbuf->cur);
for (i = 0; i < dbuf->size; ) {
de = (struct direct *)(dbuf->buf + i);
reclen = ufs_rw16(de->d_reclen, needswap);
printf(
" inode %4d %7s offset %4d reclen %3d namlen %3d name %s\n",
ufs_rw32(de->d_ino, needswap),
inode_type(DTTOIF(de->d_type)), i, reclen,
de->d_namlen, de->d_name);
i += reclen;
assert(reclen > 0);
}
}
static void
ffs_make_dirbuf(dirbuf_t *dbuf, const char *name, fsnode *node, int needswap)
{
struct direct de, *dp;
uint16_t llen, reclen;
u_char *newbuf;
assert (dbuf != NULL);
assert (name != NULL);
assert (node != NULL);
/* create direct entry */
(void)memset(&de, 0, sizeof(de));
de.d_ino = ufs_rw32(node->inode->ino, needswap);
de.d_type = IFTODT(node->type);
de.d_namlen = (uint8_t)strlen(name);
strcpy(de.d_name, name);
reclen = DIRSIZ_SWAP(0, &de, needswap);
de.d_reclen = ufs_rw16(reclen, needswap);
dp = (struct direct *)(dbuf->buf + dbuf->cur);
llen = 0;
if (dp != NULL)
llen = DIRSIZ_SWAP(0, dp, needswap);
if (debug & DEBUG_FS_MAKE_DIRBUF)
printf(
"ffs_make_dirbuf: dbuf siz %d cur %d lastlen %d\n"
" ino %d type %d reclen %d namlen %d name %.30s\n",
dbuf->size, dbuf->cur, llen,
ufs_rw32(de.d_ino, needswap), de.d_type, reclen,
de.d_namlen, de.d_name);
if (reclen + dbuf->cur + llen > roundup(dbuf->size, DIRBLKSIZ)) {
if (debug & DEBUG_FS_MAKE_DIRBUF)
printf("ffs_make_dirbuf: growing buf to %d\n",
dbuf->size + DIRBLKSIZ);
newbuf = erealloc(dbuf->buf, dbuf->size + DIRBLKSIZ);
dbuf->buf = newbuf;
dbuf->size += DIRBLKSIZ;
memset(dbuf->buf + dbuf->size - DIRBLKSIZ, 0, DIRBLKSIZ);
dbuf->cur = dbuf->size - DIRBLKSIZ;
} else if (dp) { /* shrink end of previous */
dp->d_reclen = ufs_rw16(llen,needswap);
dbuf->cur += llen;
}
dp = (struct direct *)(dbuf->buf + dbuf->cur);
memcpy(dp, &de, reclen);
dp->d_reclen = ufs_rw16(dbuf->size - dbuf->cur, needswap);
}
/*
* cribbed from sys/ufs/ffs/ffs_alloc.c
*/
static void
ffs_write_inode(union dinode *dp, uint32_t ino, const fsinfo_t *fsopts)
{
char *buf;
struct ufs1_dinode *dp1;
struct ufs2_dinode *dp2, *dip;
struct cg *cgp;
struct fs *fs;
int cg, cgino;
uint32_t i;
daddr_t d;
char sbbuf[FFS_MAXBSIZE];
uint32_t initediblk;
ffs_opt_t *ffs_opts = fsopts->fs_specific;
assert (dp != NULL);
assert (ino > 0);
assert (fsopts != NULL);
assert (ffs_opts != NULL);
fs = (struct fs *)fsopts->superblock;
cg = ino_to_cg(fs, ino);
cgino = ino % fs->fs_ipg;
if (debug & DEBUG_FS_WRITE_INODE)
printf("ffs_write_inode: din %p ino %u cg %d cgino %d\n",
dp, ino, cg, cgino);
ffs_rdfs(fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize, &sbbuf,
fsopts);
cgp = (struct cg *)sbbuf;
if (!cg_chkmagic_swap(cgp, fsopts->needswap))
errx(1, "ffs_write_inode: cg %d: bad magic number", cg);
assert (isclr(cg_inosused_swap(cgp, fsopts->needswap), cgino));
buf = emalloc(fs->fs_bsize);
dp1 = (struct ufs1_dinode *)buf;
dp2 = (struct ufs2_dinode *)buf;
if (fs->fs_cstotal.cs_nifree == 0)
errx(1, "ffs_write_inode: fs out of inodes for ino %u",
ino);
if (fs->fs_cs(fs, cg).cs_nifree == 0)
errx(1,
"ffs_write_inode: cg %d out of inodes for ino %u",
cg, ino);
setbit(cg_inosused_swap(cgp, fsopts->needswap), cgino);
ufs_add32(cgp->cg_cs.cs_nifree, -1, fsopts->needswap);
fs->fs_cstotal.cs_nifree--;
fs->fs_cs(fs, cg).cs_nifree--;
if (S_ISDIR(DIP(dp, mode))) {
ufs_add32(cgp->cg_cs.cs_ndir, 1, fsopts->needswap);
fs->fs_cstotal.cs_ndir++;
fs->fs_cs(fs, cg).cs_ndir++;
}
/*
* Initialize inode blocks on the fly for UFS2.
*/
initediblk = ufs_rw32(cgp->cg_initediblk, fsopts->needswap);
while (ffs_opts->version == 2 && cgino + INOPB(fs) > initediblk &&
initediblk < ufs_rw32(cgp->cg_niblk, fsopts->needswap)) {
memset(buf, 0, fs->fs_bsize);
dip = (struct ufs2_dinode *)buf;
for (i = 0; i < INOPB(fs); i++) {
dip->di_gen = random();
dip++;
}
ffs_wtfs(fsbtodb(fs, ino_to_fsba(fs,
cg * fs->fs_ipg + initediblk)),
fs->fs_bsize, buf, fsopts);
initediblk += INOPB(fs);
cgp->cg_initediblk = ufs_rw32(initediblk, fsopts->needswap);
}
ffs_wtfs(fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize, &sbbuf,
fsopts);
/* now write inode */
d = fsbtodb(fs, ino_to_fsba(fs, ino));
ffs_rdfs(d, fs->fs_bsize, buf, fsopts);
if (fsopts->needswap) {
if (ffs_opts->version == 1)
ffs_dinode1_swap(&dp->ffs1_din,
&dp1[ino_to_fsbo(fs, ino)]);
else
ffs_dinode2_swap(&dp->ffs2_din,
&dp2[ino_to_fsbo(fs, ino)]);
} else {
if (ffs_opts->version == 1)
dp1[ino_to_fsbo(fs, ino)] = dp->ffs1_din;
else
dp2[ino_to_fsbo(fs, ino)] = dp->ffs2_din;
}
ffs_wtfs(d, fs->fs_bsize, buf, fsopts);
free(buf);
}
void
panic(const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
vwarnx(fmt, ap);
va_end(ap);
exit(1);
}
diff --git a/usr.sbin/makefs/ffs/buf.c b/usr.sbin/makefs/ffs/buf.c
index 959734cacddf..ccbfd8ae1e23 100644
--- a/usr.sbin/makefs/ffs/buf.c
+++ b/usr.sbin/makefs/ffs/buf.c
@@ -1,216 +1,216 @@
/* $NetBSD: buf.c,v 1.13 2004/06/20 22:20:18 jmc Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (c) 2001 Wasabi Systems, Inc.
* All rights reserved.
*
* Written by Luke Mewburn for Wasabi Systems, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed for the NetBSD Project by
* Wasabi Systems, Inc.
* 4. The name of Wasabi Systems, Inc. may not be used to endorse
* or promote products derived from this software without specific prior
* written permission.
*
* THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/time.h>
#include <assert.h>
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <util.h>
#include "makefs.h"
#include "buf.h"
-static TAILQ_HEAD(buftailhead,buf) buftail;
+static TAILQ_HEAD(buftailhead, m_buf) buftail;
int
-bread(struct vnode *vp, daddr_t blkno, int size, struct ucred *u1 __unused,
- struct buf **bpp)
+bread(struct m_vnode *vp, daddr_t blkno, int size, struct ucred *u1 __unused,
+ struct m_buf **bpp)
{
off_t offset;
ssize_t rv;
fsinfo_t *fs = vp->fs;
assert (bpp != NULL);
if (debug & DEBUG_BUF_BREAD)
printf("%s: blkno %lld size %d\n", __func__, (long long)blkno,
size);
*bpp = getblk(vp, blkno, size, 0, 0, 0);
offset = (*bpp)->b_blkno * fs->sectorsize + fs->offset;
if (debug & DEBUG_BUF_BREAD)
printf("%s: blkno %lld offset %lld bcount %ld\n", __func__,
(long long)(*bpp)->b_blkno, (long long) offset,
(*bpp)->b_bcount);
if (lseek((*bpp)->b_fs->fd, offset, SEEK_SET) == -1)
err(1, "%s: lseek %lld (%lld)", __func__,
(long long)(*bpp)->b_blkno, (long long)offset);
rv = read((*bpp)->b_fs->fd, (*bpp)->b_data, (*bpp)->b_bcount);
if (debug & DEBUG_BUF_BREAD)
printf("%s: read %ld (%lld) returned %d\n", __func__,
(*bpp)->b_bcount, (long long)offset, (int)rv);
if (rv == -1) /* read error */
err(1, "%s: read %ld (%lld) returned %d", __func__,
(*bpp)->b_bcount, (long long)offset, (int)rv);
else if (rv != (*bpp)->b_bcount) /* short read */
err(1, "%s: read %ld (%lld) returned %d", __func__,
(*bpp)->b_bcount, (long long)offset, (int)rv);
else
return (0);
}
void
-brelse(struct buf *bp)
+brelse(struct m_buf *bp)
{
assert (bp != NULL);
assert (bp->b_data != NULL);
if (bp->b_lblkno < 0) {
/*
* XXX don't remove any buffers with negative logical block
* numbers (lblkno), so that we retain the mapping
* of negative lblkno -> real blkno that ffs_balloc()
* sets up.
*
* if we instead released these buffers, and implemented
* ufs_strategy() (and ufs_bmaparray()) and called those
* from bread() and bwrite() to convert the lblkno to
* a real blkno, we'd add a lot more code & complexity
* and reading off disk, for little gain, because this
* simple hack works for our purpose.
*/
bp->b_bcount = 0;
return;
}
TAILQ_REMOVE(&buftail, bp, b_tailq);
free(bp->b_data);
free(bp);
}
int
-bwrite(struct buf *bp)
+bwrite(struct m_buf *bp)
{
off_t offset;
ssize_t rv;
fsinfo_t *fs = bp->b_fs;
assert (bp != NULL);
offset = bp->b_blkno * fs->sectorsize + fs->offset;
if (debug & DEBUG_BUF_BWRITE)
printf("bwrite: blkno %lld offset %lld bcount %ld\n",
(long long)bp->b_blkno, (long long) offset,
bp->b_bcount);
if (lseek(bp->b_fs->fd, offset, SEEK_SET) == -1)
return (errno);
rv = write(bp->b_fs->fd, bp->b_data, bp->b_bcount);
if (debug & DEBUG_BUF_BWRITE)
printf("bwrite: write %ld (offset %lld) returned %lld\n",
bp->b_bcount, (long long)offset, (long long)rv);
if (rv == bp->b_bcount)
return (0);
else if (rv == -1) /* write error */
return (errno);
else /* short write ? */
return (EAGAIN);
}
void
bcleanup(void)
{
- struct buf *bp;
+ struct m_buf *bp;
/*
* XXX this really shouldn't be necessary, but i'm curious to
* know why there's still some buffers lying around that
* aren't brelse()d
*/
if (TAILQ_EMPTY(&buftail))
return;
printf("bcleanup: unflushed buffers:\n");
TAILQ_FOREACH(bp, &buftail, b_tailq) {
printf("\tlblkno %10lld blkno %10lld count %6ld bufsize %6ld\n",
(long long)bp->b_lblkno, (long long)bp->b_blkno,
bp->b_bcount, bp->b_bufsize);
}
printf("bcleanup: done\n");
}
-struct buf *
-getblk(struct vnode *vp, daddr_t blkno, int size, int u1 __unused,
+struct m_buf *
+getblk(struct m_vnode *vp, daddr_t blkno, int size, int u1 __unused,
int u2 __unused, int u3 __unused)
{
static int buftailinitted;
- struct buf *bp;
+ struct m_buf *bp;
void *n;
if (debug & DEBUG_BUF_GETBLK)
printf("getblk: blkno %lld size %d\n", (long long)blkno, size);
bp = NULL;
if (!buftailinitted) {
if (debug & DEBUG_BUF_GETBLK)
printf("getblk: initialising tailq\n");
TAILQ_INIT(&buftail);
buftailinitted = 1;
} else {
TAILQ_FOREACH(bp, &buftail, b_tailq) {
if (bp->b_lblkno != blkno)
continue;
break;
}
}
if (bp == NULL) {
bp = ecalloc(1, sizeof(*bp));
bp->b_bufsize = 0;
bp->b_blkno = bp->b_lblkno = blkno;
bp->b_fs = vp->fs;
bp->b_data = NULL;
TAILQ_INSERT_HEAD(&buftail, bp, b_tailq);
}
bp->b_bcount = size;
if (bp->b_data == NULL || bp->b_bcount > bp->b_bufsize) {
n = erealloc(bp->b_data, size);
memset(n, 0, size);
bp->b_data = n;
bp->b_bufsize = size;
}
return (bp);
}
diff --git a/usr.sbin/makefs/ffs/buf.h b/usr.sbin/makefs/ffs/buf.h
index 3bdd7ca2a9f1..31196b8b2fbe 100644
--- a/usr.sbin/makefs/ffs/buf.h
+++ b/usr.sbin/makefs/ffs/buf.h
@@ -1,78 +1,78 @@
/* $NetBSD: buf.h,v 1.3 2001/11/02 03:12:49 lukem Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (c) 2001 Wasabi Systems, Inc.
* All rights reserved.
*
* Written by Luke Mewburn for Wasabi Systems, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed for the NetBSD Project by
* Wasabi Systems, Inc.
* 4. The name of Wasabi Systems, Inc. may not be used to endorse
* or promote products derived from this software without specific prior
* written permission.
*
* THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*
* $FreeBSD$
*/
#ifndef _FFS_BUF_H
#define _FFS_BUF_H
#include <sys/param.h>
#include <sys/queue.h>
struct componentname;
struct makefs_fsinfo;
struct ucred;
-struct vnode {
+struct m_vnode {
struct makefs_fsinfo *fs;
void *v_data;
};
-struct buf {
+struct m_buf {
char * b_data;
long b_bufsize;
long b_bcount;
daddr_t b_blkno;
daddr_t b_lblkno;
struct makefs_fsinfo *b_fs;
- TAILQ_ENTRY(buf) b_tailq;
+ TAILQ_ENTRY(m_buf) b_tailq;
};
void bcleanup(void);
-int bread(struct vnode *, daddr_t, int, struct ucred *,
- struct buf **);
-void brelse(struct buf *);
-int bwrite(struct buf *);
-struct buf * getblk(struct vnode *, daddr_t, int, int, int, int);
+int bread(struct m_vnode *, daddr_t, int, struct ucred *,
+ struct m_buf **);
+void brelse(struct m_buf *);
+int bwrite(struct m_buf *);
+struct m_buf * getblk(struct m_vnode *, daddr_t, int, int, int, int);
#define bdwrite(bp) bwrite(bp)
#define clrbuf(bp) memset((bp)->b_data, 0, (u_int)(bp)->b_bcount)
#endif /* _FFS_BUF_H */
diff --git a/usr.sbin/makefs/ffs/ffs_alloc.c b/usr.sbin/makefs/ffs/ffs_alloc.c
index 88d95d6e5dda..c31106772b45 100644
--- a/usr.sbin/makefs/ffs/ffs_alloc.c
+++ b/usr.sbin/makefs/ffs/ffs_alloc.c
@@ -1,683 +1,683 @@
/* $NetBSD: ffs_alloc.c,v 1.14 2004/06/20 22:20:18 jmc Exp $ */
/* From: NetBSD: ffs_alloc.c,v 1.50 2001/09/06 02:16:01 lukem Exp */
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 2002 Networks Associates Technology, Inc.
* All rights reserved.
*
* This software was developed for the FreeBSD Project by Marshall
* Kirk McKusick and Network Associates Laboratories, the Security
* Research Division of Network Associates, Inc. under DARPA/SPAWAR
* contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS
* research program
*
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)ffs_alloc.c 8.19 (Berkeley) 7/13/95
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/time.h>
#include <errno.h>
#include <stdint.h>
#include "makefs.h"
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include "ffs/ufs_bswap.h"
#include "ffs/buf.h"
#include "ffs/ufs_inode.h"
#include "ffs/ffs_extern.h"
static int scanc(u_int, const u_char *, const u_char *, int);
static daddr_t ffs_alloccg(struct inode *, int, daddr_t, int);
-static daddr_t ffs_alloccgblk(struct inode *, struct buf *, daddr_t);
+static daddr_t ffs_alloccgblk(struct inode *, struct m_buf *, daddr_t);
static daddr_t ffs_hashalloc(struct inode *, u_int, daddr_t, int,
daddr_t (*)(struct inode *, int, daddr_t, int));
static int32_t ffs_mapsearch(struct fs *, struct cg *, daddr_t, int);
/*
* Allocate a block in the file system.
*
* The size of the requested block is given, which must be some
* multiple of fs_fsize and <= fs_bsize.
* A preference may be optionally specified. If a preference is given
* the following hierarchy is used to allocate a block:
* 1) allocate the requested block.
* 2) allocate a rotationally optimal block in the same cylinder.
* 3) allocate a block in the same cylinder group.
* 4) quadradically rehash into other cylinder groups, until an
* available block is located.
* If no block preference is given the following hierarchy is used
* to allocate a block:
* 1) allocate a block in the cylinder group that contains the
* inode for the file.
* 2) quadradically rehash into other cylinder groups, until an
* available block is located.
*/
int
ffs_alloc(struct inode *ip, daddr_t lbn __unused, daddr_t bpref, int size,
daddr_t *bnp)
{
struct fs *fs = ip->i_fs;
daddr_t bno;
int cg;
*bnp = 0;
if (size > fs->fs_bsize || fragoff(fs, size) != 0) {
errx(1, "ffs_alloc: bad size: bsize %d size %d",
fs->fs_bsize, size);
}
if (size == fs->fs_bsize && fs->fs_cstotal.cs_nbfree == 0)
goto nospace;
if (bpref >= fs->fs_size)
bpref = 0;
if (bpref == 0)
cg = ino_to_cg(fs, ip->i_number);
else
cg = dtog(fs, bpref);
bno = ffs_hashalloc(ip, cg, bpref, size, ffs_alloccg);
if (bno > 0) {
if (ip->i_fs->fs_magic == FS_UFS1_MAGIC)
ip->i_ffs1_blocks += size / DEV_BSIZE;
else
ip->i_ffs2_blocks += size / DEV_BSIZE;
*bnp = bno;
return (0);
}
nospace:
return (ENOSPC);
}
/*
* Select the desired position for the next block in a file. The file is
* logically divided into sections. The first section is composed of the
* direct blocks. Each additional section contains fs_maxbpg blocks.
*
* If no blocks have been allocated in the first section, the policy is to
* request a block in the same cylinder group as the inode that describes
* the file. If no blocks have been allocated in any other section, the
* policy is to place the section in a cylinder group with a greater than
* average number of free blocks. An appropriate cylinder group is found
* by using a rotor that sweeps the cylinder groups. When a new group of
* blocks is needed, the sweep begins in the cylinder group following the
* cylinder group from which the previous allocation was made. The sweep
* continues until a cylinder group with greater than the average number
* of free blocks is found. If the allocation is for the first block in an
* indirect block, the information on the previous allocation is unavailable;
* here a best guess is made based upon the logical block number being
* allocated.
*
* If a section is already partially allocated, the policy is to
* contiguously allocate fs_maxcontig blocks. The end of one of these
* contiguous blocks and the beginning of the next is physically separated
* so that the disk head will be in transit between them for at least
* fs_rotdelay milliseconds. This is to allow time for the processor to
* schedule another I/O transfer.
*/
/* XXX ondisk32 */
daddr_t
ffs_blkpref_ufs1(struct inode *ip, daddr_t lbn, int indx, int32_t *bap)
{
struct fs *fs;
u_int cg, startcg;
int avgbfree;
fs = ip->i_fs;
if (indx % fs->fs_maxbpg == 0 || bap[indx - 1] == 0) {
if (lbn < UFS_NDADDR + NINDIR(fs)) {
cg = ino_to_cg(fs, ip->i_number);
return (fs->fs_fpg * cg + fs->fs_frag);
}
/*
* Find a cylinder with greater than average number of
* unused data blocks.
*/
if (indx == 0 || bap[indx - 1] == 0)
startcg =
ino_to_cg(fs, ip->i_number) + lbn / fs->fs_maxbpg;
else
startcg = dtog(fs,
ufs_rw32(bap[indx - 1], UFS_FSNEEDSWAP(fs)) + 1);
startcg %= fs->fs_ncg;
avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg;
for (cg = startcg; cg < fs->fs_ncg; cg++)
if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree)
return (fs->fs_fpg * cg + fs->fs_frag);
for (cg = 0; cg <= startcg; cg++)
if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree)
return (fs->fs_fpg * cg + fs->fs_frag);
return (0);
}
/*
* We just always try to lay things out contiguously.
*/
return ufs_rw32(bap[indx - 1], UFS_FSNEEDSWAP(fs)) + fs->fs_frag;
}
daddr_t
ffs_blkpref_ufs2(struct inode *ip, daddr_t lbn, int indx, int64_t *bap)
{
struct fs *fs;
u_int cg, startcg;
int avgbfree;
fs = ip->i_fs;
if (indx % fs->fs_maxbpg == 0 || bap[indx - 1] == 0) {
if (lbn < UFS_NDADDR + NINDIR(fs)) {
cg = ino_to_cg(fs, ip->i_number);
return (fs->fs_fpg * cg + fs->fs_frag);
}
/*
* Find a cylinder with greater than average number of
* unused data blocks.
*/
if (indx == 0 || bap[indx - 1] == 0)
startcg =
ino_to_cg(fs, ip->i_number) + lbn / fs->fs_maxbpg;
else
startcg = dtog(fs,
ufs_rw64(bap[indx - 1], UFS_FSNEEDSWAP(fs)) + 1);
startcg %= fs->fs_ncg;
avgbfree = fs->fs_cstotal.cs_nbfree / fs->fs_ncg;
for (cg = startcg; cg < fs->fs_ncg; cg++)
if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) {
return (fs->fs_fpg * cg + fs->fs_frag);
}
for (cg = 0; cg < startcg; cg++)
if (fs->fs_cs(fs, cg).cs_nbfree >= avgbfree) {
return (fs->fs_fpg * cg + fs->fs_frag);
}
return (0);
}
/*
* We just always try to lay things out contiguously.
*/
return ufs_rw64(bap[indx - 1], UFS_FSNEEDSWAP(fs)) + fs->fs_frag;
}
/*
* Implement the cylinder overflow algorithm.
*
* The policy implemented by this algorithm is:
* 1) allocate the block in its requested cylinder group.
* 2) quadradically rehash on the cylinder group number.
* 3) brute force search for a free block.
*
* `size': size for data blocks, mode for inodes
*/
/*VARARGS5*/
static daddr_t
ffs_hashalloc(struct inode *ip, u_int cg, daddr_t pref, int size,
daddr_t (*allocator)(struct inode *, int, daddr_t, int))
{
struct fs *fs;
daddr_t result;
u_int i, icg = cg;
fs = ip->i_fs;
/*
* 1: preferred cylinder group
*/
result = (*allocator)(ip, cg, pref, size);
if (result)
return (result);
/*
* 2: quadratic rehash
*/
for (i = 1; i < fs->fs_ncg; i *= 2) {
cg += i;
if (cg >= fs->fs_ncg)
cg -= fs->fs_ncg;
result = (*allocator)(ip, cg, 0, size);
if (result)
return (result);
}
/*
* 3: brute force search
* Note that we start at i == 2, since 0 was checked initially,
* and 1 is always checked in the quadratic rehash.
*/
cg = (icg + 2) % fs->fs_ncg;
for (i = 2; i < fs->fs_ncg; i++) {
result = (*allocator)(ip, cg, 0, size);
if (result)
return (result);
cg++;
if (cg == fs->fs_ncg)
cg = 0;
}
return (0);
}
/*
* Determine whether a block can be allocated.
*
* Check to see if a block of the appropriate size is available,
* and if it is, allocate it.
*/
static daddr_t
ffs_alloccg(struct inode *ip, int cg, daddr_t bpref, int size)
{
struct cg *cgp;
- struct buf *bp;
+ struct m_buf *bp;
daddr_t bno, blkno;
int error, frags, allocsiz, i;
struct fs *fs = ip->i_fs;
const int needswap = UFS_FSNEEDSWAP(fs);
if (fs->fs_cs(fs, cg).cs_nbfree == 0 && size == fs->fs_bsize)
return (0);
- error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize,
- NULL, &bp);
+ error = bread((void *)ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
+ (int)fs->fs_cgsize, NULL, &bp);
if (error) {
brelse(bp);
return (0);
}
cgp = (struct cg *)bp->b_data;
if (!cg_chkmagic_swap(cgp, needswap) ||
(cgp->cg_cs.cs_nbfree == 0 && size == fs->fs_bsize)) {
brelse(bp);
return (0);
}
if (size == fs->fs_bsize) {
bno = ffs_alloccgblk(ip, bp, bpref);
bdwrite(bp);
return (bno);
}
/*
* check to see if any fragments are already available
* allocsiz is the size which will be allocated, hacking
* it down to a smaller size if necessary
*/
frags = numfrags(fs, size);
for (allocsiz = frags; allocsiz < fs->fs_frag; allocsiz++)
if (cgp->cg_frsum[allocsiz] != 0)
break;
if (allocsiz == fs->fs_frag) {
/*
* no fragments were available, so a block will be
* allocated, and hacked up
*/
if (cgp->cg_cs.cs_nbfree == 0) {
brelse(bp);
return (0);
}
bno = ffs_alloccgblk(ip, bp, bpref);
bpref = dtogd(fs, bno);
for (i = frags; i < fs->fs_frag; i++)
setbit(cg_blksfree_swap(cgp, needswap), bpref + i);
i = fs->fs_frag - frags;
ufs_add32(cgp->cg_cs.cs_nffree, i, needswap);
fs->fs_cstotal.cs_nffree += i;
fs->fs_cs(fs, cg).cs_nffree += i;
fs->fs_fmod = 1;
ufs_add32(cgp->cg_frsum[i], 1, needswap);
bdwrite(bp);
return (bno);
}
bno = ffs_mapsearch(fs, cgp, bpref, allocsiz);
for (i = 0; i < frags; i++)
clrbit(cg_blksfree_swap(cgp, needswap), bno + i);
ufs_add32(cgp->cg_cs.cs_nffree, -frags, needswap);
fs->fs_cstotal.cs_nffree -= frags;
fs->fs_cs(fs, cg).cs_nffree -= frags;
fs->fs_fmod = 1;
ufs_add32(cgp->cg_frsum[allocsiz], -1, needswap);
if (frags != allocsiz)
ufs_add32(cgp->cg_frsum[allocsiz - frags], 1, needswap);
blkno = cg * fs->fs_fpg + bno;
bdwrite(bp);
return blkno;
}
/*
* Allocate a block in a cylinder group.
*
* This algorithm implements the following policy:
* 1) allocate the requested block.
* 2) allocate a rotationally optimal block in the same cylinder.
* 3) allocate the next available block on the block rotor for the
* specified cylinder group.
* Note that this routine only allocates fs_bsize blocks; these
* blocks may be fragmented by the routine that allocates them.
*/
static daddr_t
-ffs_alloccgblk(struct inode *ip, struct buf *bp, daddr_t bpref)
+ffs_alloccgblk(struct inode *ip, struct m_buf *bp, daddr_t bpref)
{
struct cg *cgp;
daddr_t blkno;
int32_t bno;
struct fs *fs = ip->i_fs;
const int needswap = UFS_FSNEEDSWAP(fs);
u_int8_t *blksfree_swap;
cgp = (struct cg *)bp->b_data;
blksfree_swap = cg_blksfree_swap(cgp, needswap);
if (bpref == 0 || (uint32_t)dtog(fs, bpref) != ufs_rw32(cgp->cg_cgx, needswap)) {
bpref = ufs_rw32(cgp->cg_rotor, needswap);
} else {
bpref = blknum(fs, bpref);
bno = dtogd(fs, bpref);
/*
* if the requested block is available, use it
*/
if (ffs_isblock(fs, blksfree_swap, fragstoblks(fs, bno)))
goto gotit;
}
/*
* Take the next available one in this cylinder group.
*/
bno = ffs_mapsearch(fs, cgp, bpref, (int)fs->fs_frag);
if (bno < 0)
return (0);
cgp->cg_rotor = ufs_rw32(bno, needswap);
gotit:
blkno = fragstoblks(fs, bno);
ffs_clrblock(fs, blksfree_swap, (long)blkno);
ffs_clusteracct(fs, cgp, blkno, -1);
ufs_add32(cgp->cg_cs.cs_nbfree, -1, needswap);
fs->fs_cstotal.cs_nbfree--;
fs->fs_cs(fs, ufs_rw32(cgp->cg_cgx, needswap)).cs_nbfree--;
fs->fs_fmod = 1;
blkno = ufs_rw32(cgp->cg_cgx, needswap) * fs->fs_fpg + bno;
return (blkno);
}
/*
* Free a block or fragment.
*
* The specified block or fragment is placed back in the
* free map. If a fragment is deallocated, a possible
* block reassembly is checked.
*/
void
ffs_blkfree(struct inode *ip, daddr_t bno, long size)
{
struct cg *cgp;
- struct buf *bp;
+ struct m_buf *bp;
int32_t fragno, cgbno;
int i, error, cg, blk, frags, bbase;
struct fs *fs = ip->i_fs;
const int needswap = UFS_FSNEEDSWAP(fs);
if (size > fs->fs_bsize || fragoff(fs, size) != 0 ||
fragnum(fs, bno) + numfrags(fs, size) > fs->fs_frag) {
errx(1, "blkfree: bad size: bno %lld bsize %d size %ld",
(long long)bno, fs->fs_bsize, size);
}
cg = dtog(fs, bno);
if (bno >= fs->fs_size) {
warnx("bad block %lld, ino %ju", (long long)bno,
(uintmax_t)ip->i_number);
return;
}
- error = bread(ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)), (int)fs->fs_cgsize,
- NULL, &bp);
+ error = bread((void *)ip->i_devvp, fsbtodb(fs, cgtod(fs, cg)),
+ (int)fs->fs_cgsize, NULL, &bp);
if (error) {
brelse(bp);
return;
}
cgp = (struct cg *)bp->b_data;
if (!cg_chkmagic_swap(cgp, needswap)) {
brelse(bp);
return;
}
cgbno = dtogd(fs, bno);
if (size == fs->fs_bsize) {
fragno = fragstoblks(fs, cgbno);
if (!ffs_isfreeblock(fs, cg_blksfree_swap(cgp, needswap), fragno)) {
errx(1, "blkfree: freeing free block %lld",
(long long)bno);
}
ffs_setblock(fs, cg_blksfree_swap(cgp, needswap), fragno);
ffs_clusteracct(fs, cgp, fragno, 1);
ufs_add32(cgp->cg_cs.cs_nbfree, 1, needswap);
fs->fs_cstotal.cs_nbfree++;
fs->fs_cs(fs, cg).cs_nbfree++;
} else {
bbase = cgbno - fragnum(fs, cgbno);
/*
* decrement the counts associated with the old frags
*/
blk = blkmap(fs, cg_blksfree_swap(cgp, needswap), bbase);
ffs_fragacct_swap(fs, blk, cgp->cg_frsum, -1, needswap);
/*
* deallocate the fragment
*/
frags = numfrags(fs, size);
for (i = 0; i < frags; i++) {
if (isset(cg_blksfree_swap(cgp, needswap), cgbno + i)) {
errx(1, "blkfree: freeing free frag: block %lld",
(long long)(cgbno + i));
}
setbit(cg_blksfree_swap(cgp, needswap), cgbno + i);
}
ufs_add32(cgp->cg_cs.cs_nffree, i, needswap);
fs->fs_cstotal.cs_nffree += i;
fs->fs_cs(fs, cg).cs_nffree += i;
/*
* add back in counts associated with the new frags
*/
blk = blkmap(fs, cg_blksfree_swap(cgp, needswap), bbase);
ffs_fragacct_swap(fs, blk, cgp->cg_frsum, 1, needswap);
/*
* if a complete block has been reassembled, account for it
*/
fragno = fragstoblks(fs, bbase);
if (ffs_isblock(fs, cg_blksfree_swap(cgp, needswap), fragno)) {
ufs_add32(cgp->cg_cs.cs_nffree, -fs->fs_frag, needswap);
fs->fs_cstotal.cs_nffree -= fs->fs_frag;
fs->fs_cs(fs, cg).cs_nffree -= fs->fs_frag;
ffs_clusteracct(fs, cgp, fragno, 1);
ufs_add32(cgp->cg_cs.cs_nbfree, 1, needswap);
fs->fs_cstotal.cs_nbfree++;
fs->fs_cs(fs, cg).cs_nbfree++;
}
}
fs->fs_fmod = 1;
bdwrite(bp);
}
static int
scanc(u_int size, const u_char *cp, const u_char table[], int mask)
{
const u_char *end = &cp[size];
while (cp < end && (table[*cp] & mask) == 0)
cp++;
return (end - cp);
}
/*
* Find a block of the specified size in the specified cylinder group.
*
* It is a panic if a request is made to find a block if none are
* available.
*/
static int32_t
ffs_mapsearch(struct fs *fs, struct cg *cgp, daddr_t bpref, int allocsiz)
{
int32_t bno;
int start, len, loc, i;
int blk, field, subfield, pos;
int ostart, olen;
const int needswap = UFS_FSNEEDSWAP(fs);
/*
* find the fragment by searching through the free block
* map for an appropriate bit pattern
*/
if (bpref)
start = dtogd(fs, bpref) / NBBY;
else
start = ufs_rw32(cgp->cg_frotor, needswap) / NBBY;
len = howmany(fs->fs_fpg, NBBY) - start;
ostart = start;
olen = len;
loc = scanc((u_int)len,
(const u_char *)&cg_blksfree_swap(cgp, needswap)[start],
(const u_char *)fragtbl[fs->fs_frag],
(1 << (allocsiz - 1 + (fs->fs_frag % NBBY))));
if (loc == 0) {
len = start + 1;
start = 0;
loc = scanc((u_int)len,
(const u_char *)&cg_blksfree_swap(cgp, needswap)[0],
(const u_char *)fragtbl[fs->fs_frag],
(1 << (allocsiz - 1 + (fs->fs_frag % NBBY))));
if (loc == 0) {
errx(1,
"ffs_alloccg: map corrupted: start %d len %d offset %d %ld",
ostart, olen,
ufs_rw32(cgp->cg_freeoff, needswap),
(long)cg_blksfree_swap(cgp, needswap) - (long)cgp);
/* NOTREACHED */
}
}
bno = (start + len - loc) * NBBY;
cgp->cg_frotor = ufs_rw32(bno, needswap);
/*
* found the byte in the map
* sift through the bits to find the selected frag
*/
for (i = bno + NBBY; bno < i; bno += fs->fs_frag) {
blk = blkmap(fs, cg_blksfree_swap(cgp, needswap), bno);
blk <<= 1;
field = around[allocsiz];
subfield = inside[allocsiz];
for (pos = 0; pos <= fs->fs_frag - allocsiz; pos++) {
if ((blk & field) == subfield)
return (bno + pos);
field <<= 1;
subfield <<= 1;
}
}
errx(1, "ffs_alloccg: block not in map: bno %lld", (long long)bno);
return (-1);
}
/*
* Update the cluster map because of an allocation or free.
*
* Cnt == 1 means free; cnt == -1 means allocating.
*/
void
ffs_clusteracct(struct fs *fs, struct cg *cgp, int32_t blkno, int cnt)
{
int32_t *sump;
int32_t *lp;
u_char *freemapp, *mapp;
int i, start, end, forw, back, map, bit;
const int needswap = UFS_FSNEEDSWAP(fs);
if (fs->fs_contigsumsize <= 0)
return;
freemapp = cg_clustersfree_swap(cgp, needswap);
sump = cg_clustersum_swap(cgp, needswap);
/*
* Allocate or clear the actual block.
*/
if (cnt > 0)
setbit(freemapp, blkno);
else
clrbit(freemapp, blkno);
/*
* Find the size of the cluster going forward.
*/
start = blkno + 1;
end = start + fs->fs_contigsumsize;
if ((unsigned)end >= ufs_rw32(cgp->cg_nclusterblks, needswap))
end = ufs_rw32(cgp->cg_nclusterblks, needswap);
mapp = &freemapp[start / NBBY];
map = *mapp++;
bit = 1 << (start % NBBY);
for (i = start; i < end; i++) {
if ((map & bit) == 0)
break;
if ((i & (NBBY - 1)) != (NBBY - 1)) {
bit <<= 1;
} else {
map = *mapp++;
bit = 1;
}
}
forw = i - start;
/*
* Find the size of the cluster going backward.
*/
start = blkno - 1;
end = start - fs->fs_contigsumsize;
if (end < 0)
end = -1;
mapp = &freemapp[start / NBBY];
map = *mapp--;
bit = 1 << (start % NBBY);
for (i = start; i > end; i--) {
if ((map & bit) == 0)
break;
if ((i & (NBBY - 1)) != 0) {
bit >>= 1;
} else {
map = *mapp--;
bit = 1 << (NBBY - 1);
}
}
back = start - i;
/*
* Account for old cluster and the possibly new forward and
* back clusters.
*/
i = back + forw + 1;
if (i > fs->fs_contigsumsize)
i = fs->fs_contigsumsize;
ufs_add32(sump[i], cnt, needswap);
if (back > 0)
ufs_add32(sump[back], -cnt, needswap);
if (forw > 0)
ufs_add32(sump[forw], -cnt, needswap);
/*
* Update cluster summary information.
*/
lp = &sump[fs->fs_contigsumsize];
for (i = fs->fs_contigsumsize; i > 0; i--)
if (ufs_rw32(*lp--, needswap) > 0)
break;
fs->fs_maxcluster[ufs_rw32(cgp->cg_cgx, needswap)] = i;
}
diff --git a/usr.sbin/makefs/ffs/ffs_balloc.c b/usr.sbin/makefs/ffs/ffs_balloc.c
index d56ed061457b..275ec4c04471 100644
--- a/usr.sbin/makefs/ffs/ffs_balloc.c
+++ b/usr.sbin/makefs/ffs/ffs_balloc.c
@@ -1,580 +1,592 @@
/* $NetBSD: ffs_balloc.c,v 1.13 2004/06/20 22:20:18 jmc Exp $ */
/* From NetBSD: ffs_balloc.c,v 1.25 2001/08/08 08:36:36 lukem Exp */
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)ffs_balloc.c 8.8 (Berkeley) 6/16/95
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/time.h>
#include <assert.h>
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "makefs.h"
#include <ufs/ufs/dinode.h>
#include <ufs/ffs/fs.h>
#include "ffs/ufs_bswap.h"
#include "ffs/buf.h"
#include "ffs/ufs_inode.h"
#include "ffs/ffs_extern.h"
-static int ffs_balloc_ufs1(struct inode *, off_t, int, struct buf **);
-static int ffs_balloc_ufs2(struct inode *, off_t, int, struct buf **);
+static int ffs_balloc_ufs1(struct inode *, off_t, int, struct m_buf **);
+static int ffs_balloc_ufs2(struct inode *, off_t, int, struct m_buf **);
/*
* Balloc defines the structure of file system storage
* by allocating the physical blocks on a device given
* the inode and the logical block number in a file.
*
* Assume: flags == B_SYNC | B_CLRBUF
*/
int
-ffs_balloc(struct inode *ip, off_t offset, int bufsize, struct buf **bpp)
+ffs_balloc(struct inode *ip, off_t offset, int bufsize, struct m_buf **bpp)
{
if (ip->i_fs->fs_magic == FS_UFS2_MAGIC)
return ffs_balloc_ufs2(ip, offset, bufsize, bpp);
else
return ffs_balloc_ufs1(ip, offset, bufsize, bpp);
}
static int
-ffs_balloc_ufs1(struct inode *ip, off_t offset, int bufsize, struct buf **bpp)
+ffs_balloc_ufs1(struct inode *ip, off_t offset, int bufsize,
+ struct m_buf **bpp)
{
daddr_t lbn, lastlbn;
int size;
int32_t nb;
- struct buf *bp, *nbp;
+ struct m_buf *bp, *nbp;
struct fs *fs = ip->i_fs;
struct indir indirs[UFS_NIADDR + 2];
daddr_t newb, pref;
int32_t *bap;
int osize, nsize, num, i, error;
int32_t *allocblk, allociblk[UFS_NIADDR + 1];
int32_t *allocib;
const int needswap = UFS_FSNEEDSWAP(fs);
lbn = lblkno(fs, offset);
size = blkoff(fs, offset) + bufsize;
if (bpp != NULL) {
*bpp = NULL;
}
assert(size <= fs->fs_bsize);
if (lbn < 0)
return (EFBIG);
/*
* If the next write will extend the file into a new block,
* and the file is currently composed of a fragment
* this fragment has to be extended to be a full block.
*/
lastlbn = lblkno(fs, ip->i_ffs1_size);
if (lastlbn < UFS_NDADDR && lastlbn < lbn) {
nb = lastlbn;
osize = blksize(fs, ip, nb);
if (osize < fs->fs_bsize && osize > 0) {
warnx("need to ffs_realloccg; not supported!");
abort();
}
}
/*
* The first UFS_NDADDR blocks are direct blocks
*/
if (lbn < UFS_NDADDR) {
nb = ufs_rw32(ip->i_ffs1_db[lbn], needswap);
if (nb != 0 && ip->i_ffs1_size >=
(uint64_t)lblktosize(fs, lbn + 1)) {
/*
* The block is an already-allocated direct block
* and the file already extends past this block,
* thus this must be a whole block.
* Just read the block (if requested).
*/
if (bpp != NULL) {
- error = bread(ip->i_devvp, lbn, fs->fs_bsize,
- NULL, bpp);
+ error = bread((void *)ip->i_devvp, lbn,
+ fs->fs_bsize, NULL, bpp);
if (error) {
brelse(*bpp);
return (error);
}
}
return (0);
}
if (nb != 0) {
/*
* Consider need to reallocate a fragment.
*/
osize = fragroundup(fs, blkoff(fs, ip->i_ffs1_size));
nsize = fragroundup(fs, size);
if (nsize <= osize) {
/*
* The existing block is already
* at least as big as we want.
* Just read the block (if requested).
*/
if (bpp != NULL) {
- error = bread(ip->i_devvp, lbn, osize,
- NULL, bpp);
+ error = bread((void *)ip->i_devvp, lbn,
+ osize, NULL, bpp);
if (error) {
brelse(*bpp);
return (error);
}
}
return 0;
} else {
warnx("need to ffs_realloccg; not supported!");
abort();
}
} else {
/*
* the block was not previously allocated,
* allocate a new block or fragment.
*/
if (ip->i_ffs1_size < (uint64_t)lblktosize(fs, lbn + 1))
nsize = fragroundup(fs, size);
else
nsize = fs->fs_bsize;
error = ffs_alloc(ip, lbn,
ffs_blkpref_ufs1(ip, lbn, (int)lbn,
&ip->i_ffs1_db[0]),
nsize, &newb);
if (error)
return (error);
if (bpp != NULL) {
- bp = getblk(ip->i_devvp, lbn, nsize, 0, 0, 0);
+ bp = getblk((void *)ip->i_devvp, lbn, nsize,
+ 0, 0, 0);
bp->b_blkno = fsbtodb(fs, newb);
clrbuf(bp);
*bpp = bp;
}
}
ip->i_ffs1_db[lbn] = ufs_rw32((int32_t)newb, needswap);
return (0);
}
/*
* Determine the number of levels of indirection.
*/
pref = 0;
if ((error = ufs_getlbns(ip, lbn, indirs, &num)) != 0)
return (error);
if (num < 1) {
warnx("ffs_balloc: ufs_getlbns returned indirect block");
abort();
}
/*
* Fetch the first indirect block allocating if necessary.
*/
--num;
nb = ufs_rw32(ip->i_ffs1_ib[indirs[0].in_off], needswap);
allocib = NULL;
allocblk = allociblk;
if (nb == 0) {
pref = ffs_blkpref_ufs1(ip, lbn, 0, (int32_t *)0);
error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, &newb);
if (error)
return error;
nb = newb;
*allocblk++ = nb;
- bp = getblk(ip->i_devvp, indirs[1].in_lbn, fs->fs_bsize, 0, 0, 0);
+ bp = getblk((void *)ip->i_devvp, indirs[1].in_lbn,
+ fs->fs_bsize, 0, 0, 0);
bp->b_blkno = fsbtodb(fs, nb);
clrbuf(bp);
/*
* Write synchronously so that indirect blocks
* never point at garbage.
*/
if ((error = bwrite(bp)) != 0)
return error;
allocib = &ip->i_ffs1_ib[indirs[0].in_off];
*allocib = ufs_rw32((int32_t)nb, needswap);
}
/*
* Fetch through the indirect blocks, allocating as necessary.
*/
for (i = 1;;) {
- error = bread(ip->i_devvp, indirs[i].in_lbn, fs->fs_bsize,
- NULL, &bp);
+ error = bread((void *)ip->i_devvp, indirs[i].in_lbn,
+ fs->fs_bsize, NULL, &bp);
if (error) {
brelse(bp);
return error;
}
bap = (int32_t *)bp->b_data;
nb = ufs_rw32(bap[indirs[i].in_off], needswap);
if (i == num)
break;
i++;
if (nb != 0) {
brelse(bp);
continue;
}
if (pref == 0)
pref = ffs_blkpref_ufs1(ip, lbn, 0, (int32_t *)0);
error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, &newb);
if (error) {
brelse(bp);
return error;
}
nb = newb;
*allocblk++ = nb;
- nbp = getblk(ip->i_devvp, indirs[i].in_lbn, fs->fs_bsize, 0, 0, 0);
+ nbp = getblk((void *)ip->i_devvp, indirs[i].in_lbn,
+ fs->fs_bsize, 0, 0, 0);
nbp->b_blkno = fsbtodb(fs, nb);
clrbuf(nbp);
/*
* Write synchronously so that indirect blocks
* never point at garbage.
*/
if ((error = bwrite(nbp)) != 0) {
brelse(bp);
return error;
}
bap[indirs[i - 1].in_off] = ufs_rw32(nb, needswap);
bwrite(bp);
}
/*
* Get the data block, allocating if necessary.
*/
if (nb == 0) {
pref = ffs_blkpref_ufs1(ip, lbn, indirs[num].in_off, &bap[0]);
error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, &newb);
if (error) {
brelse(bp);
return error;
}
nb = newb;
*allocblk++ = nb;
if (bpp != NULL) {
- nbp = getblk(ip->i_devvp, lbn, fs->fs_bsize, 0, 0, 0);
+ nbp = getblk((void *)ip->i_devvp, lbn, fs->fs_bsize,
+ 0, 0, 0);
nbp->b_blkno = fsbtodb(fs, nb);
clrbuf(nbp);
*bpp = nbp;
}
bap[indirs[num].in_off] = ufs_rw32(nb, needswap);
/*
* If required, write synchronously, otherwise use
* delayed write.
*/
bwrite(bp);
return (0);
}
brelse(bp);
if (bpp != NULL) {
- error = bread(ip->i_devvp, lbn, (int)fs->fs_bsize, NULL, &nbp);
+ error = bread((void *)ip->i_devvp, lbn, (int)fs->fs_bsize,
+ NULL, &nbp);
if (error) {
brelse(nbp);
return error;
}
*bpp = nbp;
}
return (0);
}
static int
-ffs_balloc_ufs2(struct inode *ip, off_t offset, int bufsize, struct buf **bpp)
+ffs_balloc_ufs2(struct inode *ip, off_t offset, int bufsize,
+ struct m_buf **bpp)
{
daddr_t lbn, lastlbn;
int size;
- struct buf *bp, *nbp;
+ struct m_buf *bp, *nbp;
struct fs *fs = ip->i_fs;
struct indir indirs[UFS_NIADDR + 2];
daddr_t newb, pref, nb;
int64_t *bap;
int osize, nsize, num, i, error;
int64_t *allocblk, allociblk[UFS_NIADDR + 1];
int64_t *allocib;
const int needswap = UFS_FSNEEDSWAP(fs);
lbn = lblkno(fs, offset);
size = blkoff(fs, offset) + bufsize;
if (bpp != NULL) {
*bpp = NULL;
}
assert(size <= fs->fs_bsize);
if (lbn < 0)
return (EFBIG);
/*
* If the next write will extend the file into a new block,
* and the file is currently composed of a fragment
* this fragment has to be extended to be a full block.
*/
lastlbn = lblkno(fs, ip->i_ffs2_size);
if (lastlbn < UFS_NDADDR && lastlbn < lbn) {
nb = lastlbn;
osize = blksize(fs, ip, nb);
if (osize < fs->fs_bsize && osize > 0) {
warnx("need to ffs_realloccg; not supported!");
abort();
}
}
/*
* The first UFS_NDADDR blocks are direct blocks
*/
if (lbn < UFS_NDADDR) {
nb = ufs_rw64(ip->i_ffs2_db[lbn], needswap);
if (nb != 0 && ip->i_ffs2_size >=
(uint64_t)lblktosize(fs, lbn + 1)) {
/*
* The block is an already-allocated direct block
* and the file already extends past this block,
* thus this must be a whole block.
* Just read the block (if requested).
*/
if (bpp != NULL) {
- error = bread(ip->i_devvp, lbn, fs->fs_bsize,
- NULL, bpp);
+ error = bread((void *)ip->i_devvp, lbn,
+ fs->fs_bsize, NULL, bpp);
if (error) {
brelse(*bpp);
return (error);
}
}
return (0);
}
if (nb != 0) {
/*
* Consider need to reallocate a fragment.
*/
osize = fragroundup(fs, blkoff(fs, ip->i_ffs2_size));
nsize = fragroundup(fs, size);
if (nsize <= osize) {
/*
* The existing block is already
* at least as big as we want.
* Just read the block (if requested).
*/
if (bpp != NULL) {
- error = bread(ip->i_devvp, lbn, osize,
- NULL, bpp);
+ error = bread((void *)ip->i_devvp, lbn,
+ osize, NULL, bpp);
if (error) {
brelse(*bpp);
return (error);
}
}
return 0;
} else {
warnx("need to ffs_realloccg; not supported!");
abort();
}
} else {
/*
* the block was not previously allocated,
* allocate a new block or fragment.
*/
if (ip->i_ffs2_size < (uint64_t)lblktosize(fs, lbn + 1))
nsize = fragroundup(fs, size);
else
nsize = fs->fs_bsize;
error = ffs_alloc(ip, lbn,
ffs_blkpref_ufs2(ip, lbn, (int)lbn,
&ip->i_ffs2_db[0]),
nsize, &newb);
if (error)
return (error);
if (bpp != NULL) {
- bp = getblk(ip->i_devvp, lbn, nsize, 0, 0, 0);
+ bp = getblk((void *)ip->i_devvp, lbn, nsize,
+ 0, 0, 0);
bp->b_blkno = fsbtodb(fs, newb);
clrbuf(bp);
*bpp = bp;
}
}
ip->i_ffs2_db[lbn] = ufs_rw64(newb, needswap);
return (0);
}
/*
* Determine the number of levels of indirection.
*/
pref = 0;
if ((error = ufs_getlbns(ip, lbn, indirs, &num)) != 0)
return (error);
if (num < 1) {
warnx("ffs_balloc: ufs_getlbns returned indirect block");
abort();
}
/*
* Fetch the first indirect block allocating if necessary.
*/
--num;
nb = ufs_rw64(ip->i_ffs2_ib[indirs[0].in_off], needswap);
allocib = NULL;
allocblk = allociblk;
if (nb == 0) {
pref = ffs_blkpref_ufs2(ip, lbn, 0, (int64_t *)0);
error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, &newb);
if (error)
return error;
nb = newb;
*allocblk++ = nb;
- bp = getblk(ip->i_devvp, indirs[1].in_lbn, fs->fs_bsize, 0, 0, 0);
+ bp = getblk((void *)ip->i_devvp, indirs[1].in_lbn,
+ fs->fs_bsize, 0, 0, 0);
bp->b_blkno = fsbtodb(fs, nb);
clrbuf(bp);
/*
* Write synchronously so that indirect blocks
* never point at garbage.
*/
if ((error = bwrite(bp)) != 0)
return error;
allocib = &ip->i_ffs2_ib[indirs[0].in_off];
*allocib = ufs_rw64(nb, needswap);
}
/*
* Fetch through the indirect blocks, allocating as necessary.
*/
for (i = 1;;) {
- error = bread(ip->i_devvp, indirs[i].in_lbn, fs->fs_bsize,
- NULL, &bp);
+ error = bread((void *)ip->i_devvp, indirs[i].in_lbn,
+ fs->fs_bsize, NULL, &bp);
if (error) {
brelse(bp);
return error;
}
bap = (int64_t *)bp->b_data;
nb = ufs_rw64(bap[indirs[i].in_off], needswap);
if (i == num)
break;
i++;
if (nb != 0) {
brelse(bp);
continue;
}
if (pref == 0)
pref = ffs_blkpref_ufs2(ip, lbn, 0, (int64_t *)0);
error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, &newb);
if (error) {
brelse(bp);
return error;
}
nb = newb;
*allocblk++ = nb;
- nbp = getblk(ip->i_devvp, indirs[i].in_lbn, fs->fs_bsize, 0, 0, 0);
+ nbp = getblk((void *)ip->i_devvp, indirs[i].in_lbn,
+ fs->fs_bsize, 0, 0, 0);
nbp->b_blkno = fsbtodb(fs, nb);
clrbuf(nbp);
/*
* Write synchronously so that indirect blocks
* never point at garbage.
*/
if ((error = bwrite(nbp)) != 0) {
brelse(bp);
return error;
}
bap[indirs[i - 1].in_off] = ufs_rw64(nb, needswap);
bwrite(bp);
}
/*
* Get the data block, allocating if necessary.
*/
if (nb == 0) {
pref = ffs_blkpref_ufs2(ip, lbn, indirs[num].in_off, &bap[0]);
error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, &newb);
if (error) {
brelse(bp);
return error;
}
nb = newb;
*allocblk++ = nb;
if (bpp != NULL) {
- nbp = getblk(ip->i_devvp, lbn, fs->fs_bsize, 0, 0, 0);
+ nbp = getblk((void *)ip->i_devvp, lbn, fs->fs_bsize,
+ 0, 0, 0);
nbp->b_blkno = fsbtodb(fs, nb);
clrbuf(nbp);
*bpp = nbp;
}
bap[indirs[num].in_off] = ufs_rw64(nb, needswap);
/*
* If required, write synchronously, otherwise use
* delayed write.
*/
bwrite(bp);
return (0);
}
brelse(bp);
if (bpp != NULL) {
- error = bread(ip->i_devvp, lbn, (int)fs->fs_bsize, NULL, &nbp);
+ error = bread((void *)ip->i_devvp, lbn, (int)fs->fs_bsize,
+ NULL, &nbp);
if (error) {
brelse(nbp);
return error;
}
*bpp = nbp;
}
return (0);
}
diff --git a/usr.sbin/makefs/ffs/ffs_extern.h b/usr.sbin/makefs/ffs/ffs_extern.h
index b4d4bbaef6e9..12ba0b77989c 100644
--- a/usr.sbin/makefs/ffs/ffs_extern.h
+++ b/usr.sbin/makefs/ffs/ffs_extern.h
@@ -1,79 +1,79 @@
/* $NetBSD: ffs_extern.h,v 1.6 2003/08/07 11:25:33 agc Exp $ */
/* From: NetBSD: ffs_extern.h,v 1.19 2001/08/17 02:18:48 lukem Exp */
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1991, 1993, 1994
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)ffs_extern.h 8.6 (Berkeley) 3/30/95
* $FreeBSD$
*/
#include "ffs/buf.h"
struct inode;
/*
* Structure used to pass around logical block paths generated by
* ufs_getlbns and used by truncate and bmap code.
*/
struct indir {
daddr_t in_lbn; /* Logical block number. */
int in_off; /* Offset in buffer. */
};
/* ffs.c */
_Noreturn void panic(const char *, ...) __printflike(1, 2);
/* ffs_alloc.c */
int ffs_alloc(struct inode *, daddr_t, daddr_t, int, daddr_t *);
daddr_t ffs_blkpref_ufs1(struct inode *, daddr_t, int, int32_t *);
daddr_t ffs_blkpref_ufs2(struct inode *, daddr_t, int, int64_t *);
void ffs_blkfree(struct inode *, daddr_t, long);
void ffs_clusteracct(struct fs *, struct cg *, int32_t, int);
/* ffs_balloc.c */
-int ffs_balloc(struct inode *, off_t, int, struct buf **);
+int ffs_balloc(struct inode *, off_t, int, struct m_buf **);
/* ffs_bswap.c */
void ffs_sb_swap(struct fs*, struct fs *);
void ffs_dinode1_swap(struct ufs1_dinode *, struct ufs1_dinode *);
void ffs_dinode2_swap(struct ufs2_dinode *, struct ufs2_dinode *);
void ffs_csum_swap(struct csum *, struct csum *, int);
void ffs_cg_swap(struct cg *, struct cg *, struct fs *);
/* ffs_subr.c */
void ffs_fragacct(struct fs *, int, int32_t[], int, int);
int ffs_isblock(struct fs *, u_char *, int32_t);
int ffs_isfreeblock(struct fs *, u_char *, int32_t);
void ffs_clrblock(struct fs *, u_char *, int32_t);
void ffs_setblock(struct fs *, u_char *, int32_t);
/* ufs_bmap.c */
int ufs_getlbns(struct inode *, daddr_t, struct indir *, int *);
diff --git a/usr.sbin/makefs/msdos.c b/usr.sbin/makefs/msdos.c
index a0e0f7174f25..567122c1db7e 100644
--- a/usr.sbin/makefs/msdos.c
+++ b/usr.sbin/makefs/msdos.c
@@ -1,272 +1,273 @@
/* $NetBSD: msdos.c,v 1.20 2017/04/14 15:40:35 christos Exp $ */
/*-
* Copyright (c) 2013 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Christos Zoulas.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#if HAVE_NBTOOL_CONFIG_H
#include "nbtool_config.h"
#endif
#include <sys/cdefs.h>
#if defined(__RCSID) && !defined(__lint)
__FBSDID("$FreeBSD$");
#endif /* !__lint */
#include <sys/param.h>
#if !HAVE_NBTOOL_CONFIG_H
#include <sys/mount.h>
#endif
#include <assert.h>
#include <errno.h>
#include <fcntl.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdint.h>
#include <unistd.h>
#include <dirent.h>
#include <util.h>
-#include "ffs/buf.h"
-#include "makefs.h"
-#include "msdos.h"
-
#include <mkfs_msdos.h>
#include <fs/msdosfs/bpb.h>
#include "msdos/direntry.h"
#include <fs/msdosfs/denode.h>
#include <fs/msdosfs/msdosfsmount.h>
+#undef clrbuf
+#include "ffs/buf.h"
+#include "makefs.h"
+#include "msdos.h"
+
static int msdos_populate_dir(const char *, struct denode *, fsnode *,
fsnode *, fsinfo_t *);
struct msdos_options_ex {
struct msdos_options options;
};
void
msdos_prep_opts(fsinfo_t *fsopts)
{
struct msdos_options_ex *msdos_opt = ecalloc(1, sizeof(*msdos_opt));
const option_t msdos_options[] = {
#define AOPT(_opt, _type, _name, _min, _desc) { \
.letter = _opt, \
.name = # _name, \
.type = _min == -1 ? OPT_STRPTR : \
(_min == -2 ? OPT_BOOL : \
(sizeof(_type) == 1 ? OPT_INT8 : \
(sizeof(_type) == 2 ? OPT_INT16 : \
(sizeof(_type) == 4 ? OPT_INT32 : OPT_INT64)))), \
.value = &msdos_opt->options._name, \
.minimum = _min, \
.maximum = sizeof(_type) == 1 ? UINT8_MAX : \
(sizeof(_type) == 2 ? UINT16_MAX : \
(sizeof(_type) == 4 ? UINT32_MAX : INT64_MAX)), \
.desc = _desc, \
},
ALLOPTS
#undef AOPT
{ .name = NULL }
};
fsopts->fs_specific = msdos_opt;
fsopts->fs_options = copy_opts(msdos_options);
}
void
msdos_cleanup_opts(fsinfo_t *fsopts)
{
free(fsopts->fs_specific);
free(fsopts->fs_options);
}
int
msdos_parse_opts(const char *option, fsinfo_t *fsopts)
{
struct msdos_options *msdos_opt = fsopts->fs_specific;
option_t *msdos_options = fsopts->fs_options;
int rv;
assert(option != NULL);
assert(fsopts != NULL);
assert(msdos_opt != NULL);
if (debug & DEBUG_FS_PARSE_OPTS)
printf("msdos_parse_opts: got `%s'\n", option);
rv = set_option(msdos_options, option, NULL, 0);
if (rv == -1)
return rv;
if (strcmp(msdos_options[rv].name, "volume_id") == 0)
msdos_opt->volume_id_set = 1;
else if (strcmp(msdos_options[rv].name, "media_descriptor") == 0)
msdos_opt->media_descriptor_set = 1;
else if (strcmp(msdos_options[rv].name, "hidden_sectors") == 0)
msdos_opt->hidden_sectors_set = 1;
if (stampst.st_ino) {
msdos_opt->timestamp_set = 1;
msdos_opt->timestamp = stampst.st_mtime;
}
return 1;
}
void
msdos_makefs(const char *image, const char *dir, fsnode *root, fsinfo_t *fsopts)
{
struct msdos_options_ex *msdos_opt = fsopts->fs_specific;
- struct vnode vp, rootvp;
+ struct m_vnode vp, rootvp;
struct timeval start;
struct msdosfsmount *pmp;
uint32_t flags;
assert(image != NULL);
assert(dir != NULL);
assert(root != NULL);
assert(fsopts != NULL);
fsopts->size = fsopts->maxsize;
msdos_opt->options.create_size = MAX(msdos_opt->options.create_size,
fsopts->offset + fsopts->size);
if (fsopts->offset > 0)
msdos_opt->options.offset = fsopts->offset;
if (msdos_opt->options.bytes_per_sector == 0) {
if (fsopts->sectorsize == -1)
fsopts->sectorsize = 512;
msdos_opt->options.bytes_per_sector = fsopts->sectorsize;
} else if (fsopts->sectorsize == -1) {
fsopts->sectorsize = msdos_opt->options.bytes_per_sector;
} else if (fsopts->sectorsize != msdos_opt->options.bytes_per_sector) {
err(1, "inconsistent sectorsize -S %u"
"!= -o bytes_per_sector %u",
fsopts->sectorsize, msdos_opt->options.bytes_per_sector);
}
/* create image */
printf("Creating `%s'\n", image);
TIMER_START(start);
if (mkfs_msdos(image, NULL, &msdos_opt->options) == -1)
return;
TIMER_RESULTS(start, "mkfs_msdos");
fsopts->fd = open(image, O_RDWR);
vp.fs = fsopts;
flags = 0;
- if ((pmp = msdosfs_mount(&vp)) == NULL)
+ if ((pmp = m_msdosfs_mount(&vp)) == NULL)
err(1, "msdosfs_mount");
if (msdosfs_root(pmp, &rootvp) != 0)
err(1, "msdosfs_root");
if (debug & DEBUG_FS_MAKEFS)
printf("msdos_makefs: image %s directory %s root %p\n",
image, dir, root);
/* populate image */
printf("Populating `%s'\n", image);
TIMER_START(start);
if (msdos_populate_dir(dir, VTODE(&rootvp), root, root, fsopts) == -1)
errx(1, "Image file `%s' not created.", image);
TIMER_RESULTS(start, "msdos_populate_dir");
if (msdosfs_fsiflush(pmp) != 0)
errx(1, "Unable to update FSInfo block.");
if (debug & DEBUG_FS_MAKEFS)
putchar('\n');
/* ensure no outstanding buffers remain */
if (debug & DEBUG_FS_MAKEFS)
bcleanup();
printf("Image `%s' complete\n", image);
}
static int
msdos_populate_dir(const char *path, struct denode *dir, fsnode *root,
fsnode *parent, fsinfo_t *fsopts)
{
fsnode *cur;
char pbuf[MAXPATHLEN];
assert(dir != NULL);
assert(root != NULL);
assert(fsopts != NULL);
for (cur = root->next; cur != NULL; cur = cur->next) {
if ((size_t)snprintf(pbuf, sizeof(pbuf), "%s/%s", path,
cur->name) >= sizeof(pbuf)) {
warnx("path %s too long", pbuf);
return -1;
}
if ((cur->inode->flags & FI_ALLOCATED) == 0) {
cur->inode->flags |= FI_ALLOCATED;
if (cur != root) {
fsopts->curinode++;
cur->inode->ino = fsopts->curinode;
cur->parent = parent;
}
}
if (cur->inode->flags & FI_WRITTEN) {
continue; // hard link
}
cur->inode->flags |= FI_WRITTEN;
if (cur->child) {
struct denode *de;
if ((de = msdosfs_mkdire(pbuf, dir, cur)) == NULL) {
warn("msdosfs_mkdire %s", pbuf);
return -1;
}
if (msdos_populate_dir(pbuf, de, cur->child, cur,
fsopts) == -1) {
warn("msdos_populate_dir %s", pbuf);
return -1;
}
continue;
} else if (!S_ISREG(cur->type)) {
warnx("skipping non-regular file %s/%s", cur->path,
cur->name);
continue;
}
if (msdosfs_mkfile(cur->contents ? cur->contents : pbuf, dir,
cur) == NULL) {
warn("msdosfs_mkfile %s", pbuf);
return -1;
}
}
return 0;
}
diff --git a/usr.sbin/makefs/msdos.h b/usr.sbin/makefs/msdos.h
index d9f1cc4cdb11..a51420de76e5 100644
--- a/usr.sbin/makefs/msdos.h
+++ b/usr.sbin/makefs/msdos.h
@@ -1,61 +1,71 @@
/* $FreeBSD$ */
/* $NetBSD: msdos.h,v 1.3 2015/10/16 16:40:02 christos Exp $ */
/*-
* Copyright (c) 2013 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Christos Zoulas.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef _MAKEFS_MSDOS_H
#define _MAKEFS_MSDOS_H
#define NOCRED NULL
#define MSDOSFS_DPRINTF(args) do { \
if (debug & DEBUG_MSDOSFS) \
printf args; \
} while (0);
struct vnode;
struct denode;
struct fsnode;
struct msdosfsmount;
struct componentname {
char *cn_nameptr;
size_t cn_namelen;
};
+struct m_vnode;
+struct m_buf;
+
int msdosfs_fsiflush(struct msdosfsmount *);
-struct msdosfsmount *msdosfs_mount(struct vnode *);
-int msdosfs_root(struct msdosfsmount *, struct vnode *);
+struct msdosfsmount *msdosfs_mount(struct m_vnode *);
+int msdosfs_root(struct msdosfsmount *, struct m_vnode *);
struct denode *msdosfs_mkfile(const char *, struct denode *, fsnode *);
struct denode *msdosfs_mkdire(const char *, struct denode *, fsnode *);
+int m_readde(struct denode *dep, struct m_buf **bpp, struct direntry **epp);
+int m_readep(struct msdosfsmount *pmp, u_long dirclust, u_long diroffset,
+ struct m_buf **bpp, struct direntry **epp);
+int m_extendfile(struct denode *dep, u_long count, struct m_buf **bpp,
+ u_long *ncp, int flags);
+
+struct msdosfsmount *m_msdosfs_mount(struct m_vnode *devvp);
#endif
diff --git a/usr.sbin/makefs/msdos/msdosfs_denode.c b/usr.sbin/makefs/msdos/msdosfs_denode.c
index 283ef2e83e0b..f2faed234228 100644
--- a/usr.sbin/makefs/msdos/msdosfs_denode.c
+++ b/usr.sbin/makefs/msdos/msdosfs_denode.c
@@ -1,378 +1,378 @@
/* $NetBSD: msdosfs_denode.c,v 1.7 2015/03/29 05:52:59 agc Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1994, 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*-
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
* This software is provided "as is".
*
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
* October 1992
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/errno.h>
#include <stdbool.h>
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <util.h>
-#include "ffs/buf.h"
-
#include <fs/msdosfs/bpb.h>
#include <fs/msdosfs/direntry.h>
#include <fs/msdosfs/denode.h>
#include <fs/msdosfs/fat.h>
#include <fs/msdosfs/msdosfsmount.h>
+#undef clrbuf
+#include "ffs/buf.h"
#include "makefs.h"
#include "msdos.h"
/*
* If deget() succeeds it returns with the gotten denode locked().
*
* pmp - address of msdosfsmount structure of the filesystem containing
* the denode of interest. The pm_dev field and the address of
* the msdosfsmount structure are used.
* dirclust - which cluster bp contains, if dirclust is 0 (root directory)
* diroffset is relative to the beginning of the root directory,
* otherwise it is cluster relative.
* diroffset - offset past begin of cluster of denode we want
* depp - returns the address of the gotten denode.
*/
int
deget(struct msdosfsmount *pmp, u_long dirclust, u_long diroffset,
struct denode **depp)
{
int error;
uint64_t inode;
struct direntry *direntptr;
struct denode *ldep;
- struct buf *bp;
+ struct m_buf *bp;
MSDOSFS_DPRINTF(("deget(pmp %p, dirclust %lu, diroffset %lx, depp %p)\n",
pmp, dirclust, diroffset, depp));
/*
* On FAT32 filesystems, root is a (more or less) normal
* directory
*/
if (FAT32(pmp) && dirclust == MSDOSFSROOT)
dirclust = pmp->pm_rootdirblk;
inode = (uint64_t)pmp->pm_bpcluster * dirclust + diroffset;
ldep = ecalloc(1, sizeof(*ldep));
ldep->de_vnode = NULL;
ldep->de_flag = 0;
ldep->de_dirclust = dirclust;
ldep->de_diroffset = diroffset;
ldep->de_inode = inode;
ldep->de_pmp = pmp;
ldep->de_refcnt = 1;
fc_purge(ldep, 0); /* init the FAT cache for this denode */
/*
* Copy the directory entry into the denode area of the vnode.
*/
if ((dirclust == MSDOSFSROOT
|| (FAT32(pmp) && dirclust == pmp->pm_rootdirblk))
&& diroffset == MSDOSFSROOT_OFS) {
/*
* Directory entry for the root directory. There isn't one,
* so we manufacture one. We should probably rummage
* through the root directory and find a label entry (if it
* exists), and then use the time and date from that entry
* as the time and date for the root denode.
*/
ldep->de_vnode = (struct vnode *)-1;
ldep->de_Attributes = ATTR_DIRECTORY;
ldep->de_LowerCase = 0;
if (FAT32(pmp))
ldep->de_StartCluster = pmp->pm_rootdirblk;
/* de_FileSize will be filled in further down */
else {
ldep->de_StartCluster = MSDOSFSROOT;
ldep->de_FileSize = pmp->pm_rootdirsize * DEV_BSIZE;
}
/*
* fill in time and date so that dos2unixtime() doesn't
* spit up when called from msdosfs_getattr() with root
* denode
*/
ldep->de_CHun = 0;
ldep->de_CTime = 0x0000; /* 00:00:00 */
ldep->de_CDate = (0 << DD_YEAR_SHIFT) | (1 << DD_MONTH_SHIFT)
| (1 << DD_DAY_SHIFT);
/* Jan 1, 1980 */
ldep->de_ADate = ldep->de_CDate;
ldep->de_MTime = ldep->de_CTime;
ldep->de_MDate = ldep->de_CDate;
/* leave the other fields as garbage */
} else {
- error = readep(pmp, dirclust, diroffset, &bp, &direntptr);
+ error = m_readep(pmp, dirclust, diroffset, &bp, &direntptr);
if (error) {
ldep->de_Name[0] = SLOT_DELETED;
*depp = NULL;
return (error);
}
(void)DE_INTERNALIZE(ldep, direntptr);
brelse(bp);
}
/*
* Fill in a few fields of the vnode and finish filling in the
* denode. Then return the address of the found denode.
*/
if (ldep->de_Attributes & ATTR_DIRECTORY) {
/*
* Since DOS directory entries that describe directories
* have 0 in the filesize field, we take this opportunity
* to find out the length of the directory and plug it into
* the denode structure.
*/
u_long size;
/*
* XXX it sometimes happens that the "." entry has cluster
* number 0 when it shouldn't. Use the actual cluster number
* instead of what is written in directory entry.
*/
if (diroffset == 0 && ldep->de_StartCluster != dirclust) {
MSDOSFS_DPRINTF(("deget(): \".\" entry at clust %lu != %lu\n",
dirclust, ldep->de_StartCluster));
ldep->de_StartCluster = dirclust;
}
if (ldep->de_StartCluster != MSDOSFSROOT) {
error = pcbmap(ldep, 0xffff, 0, &size, 0);
if (error == E2BIG) {
ldep->de_FileSize = de_cn2off(pmp, size);
error = 0;
} else {
MSDOSFS_DPRINTF(("deget(): pcbmap returned %d\n",
error));
}
}
}
*depp = ldep;
return (0);
}
/*
* Truncate the file described by dep to the length specified by length.
*/
int
detrunc(struct denode *dep, u_long length, int flags, struct ucred *cred)
{
int error;
int allerror;
u_long eofentry;
u_long chaintofree;
daddr_t bn;
int boff;
int isadir = dep->de_Attributes & ATTR_DIRECTORY;
- struct buf *bp;
+ struct m_buf *bp;
struct msdosfsmount *pmp = dep->de_pmp;
MSDOSFS_DPRINTF(("detrunc(): file %s, length %lu, flags %x\n",
dep->de_Name, length, flags));
/*
* Disallow attempts to truncate the root directory since it is of
* fixed size. That's just the way dos filesystems are. We use
* the VROOT bit in the vnode because checking for the directory
* bit and a startcluster of 0 in the denode is not adequate to
* recognize the root directory at this point in a file or
* directory's life.
*/
if (dep->de_vnode != NULL && !FAT32(pmp)) {
MSDOSFS_DPRINTF(("detrunc(): can't truncate root directory, "
"clust %ld, offset %ld\n",
dep->de_dirclust, dep->de_diroffset));
return (EINVAL);
}
if (dep->de_FileSize < length)
return deextend(dep, length, cred);
/*
* If the desired length is 0 then remember the starting cluster of
* the file and set the StartCluster field in the directory entry
* to 0. If the desired length is not zero, then get the number of
* the last cluster in the shortened file. Then get the number of
* the first cluster in the part of the file that is to be freed.
* Then set the next cluster pointer in the last cluster of the
* file to CLUST_EOFE.
*/
if (length == 0) {
chaintofree = dep->de_StartCluster;
dep->de_StartCluster = 0;
eofentry = ~0;
} else {
error = pcbmap(dep, de_clcount(pmp, length) - 1, 0,
&eofentry, 0);
if (error) {
MSDOSFS_DPRINTF(("detrunc(): pcbmap fails %d\n",
error));
return (error);
}
}
fc_purge(dep, de_clcount(pmp, length));
/*
* If the new length is not a multiple of the cluster size then we
* must zero the tail end of the new last cluster in case it
* becomes part of the file again because of a seek.
*/
if ((boff = length & pmp->pm_crbomask) != 0) {
if (isadir) {
bn = cntobn(pmp, eofentry);
- error = bread(pmp->pm_devvp, bn, pmp->pm_bpcluster,
- 0, &bp);
+ error = bread((void *)pmp->pm_devvp, bn,
+ pmp->pm_bpcluster, 0, &bp);
if (error) {
brelse(bp);
MSDOSFS_DPRINTF(("detrunc(): bread fails %d\n",
error));
return (error);
}
memset(bp->b_data + boff, 0, pmp->pm_bpcluster - boff);
bwrite(bp);
}
}
/*
* Write out the updated directory entry. Even if the update fails
* we free the trailing clusters.
*/
dep->de_FileSize = length;
if (!isadir)
dep->de_flag |= DE_UPDATE|DE_MODIFIED;
MSDOSFS_DPRINTF(("detrunc(): allerror %d, eofentry %lu\n",
allerror, eofentry));
/*
* If we need to break the cluster chain for the file then do it
* now.
*/
if (eofentry != ~0) {
error = fatentry(FAT_GET_AND_SET, pmp, eofentry,
&chaintofree, CLUST_EOFE);
if (error) {
MSDOSFS_DPRINTF(("detrunc(): fatentry errors %d\n",
error));
return (error);
}
fc_setcache(dep, FC_LASTFC, de_cluster(pmp, length - 1),
eofentry);
}
/*
* Now free the clusters removed from the file because of the
* truncation.
*/
if (chaintofree != 0 && !MSDOSFSEOF(pmp, chaintofree))
freeclusterchain(pmp, chaintofree);
return (allerror);
}
/*
* Extend the file described by dep to length specified by length.
*/
int
deextend(struct denode *dep, u_long length, struct ucred *cred)
{
struct msdosfsmount *pmp = dep->de_pmp;
u_long count;
int error;
/*
* The root of a DOS filesystem cannot be extended.
*/
if (dep->de_vnode != NULL && !FAT32(pmp))
return (EINVAL);
/*
* Directories cannot be extended.
*/
if (dep->de_Attributes & ATTR_DIRECTORY)
return (EISDIR);
if (length <= dep->de_FileSize)
return (E2BIG);
/*
* Compute the number of clusters to allocate.
*/
count = de_clcount(pmp, length) - de_clcount(pmp, dep->de_FileSize);
if (count > 0) {
if (count > pmp->pm_freeclustercount)
return (ENOSPC);
- error = extendfile(dep, count, NULL, NULL, DE_CLEAR);
+ error = m_extendfile(dep, count, NULL, NULL, DE_CLEAR);
if (error) {
/* truncate the added clusters away again */
(void) detrunc(dep, dep->de_FileSize, 0, cred);
return (error);
}
}
/*
* Zero extend file range; ubc_zerorange() uses ubc_alloc() and a
* memset(); we set the write size so ubc won't read in file data that
* is zero'd later.
*/
dep->de_FileSize = length;
dep->de_flag |= DE_UPDATE | DE_MODIFIED;
return 0;
}
diff --git a/usr.sbin/makefs/msdos/msdosfs_fat.c b/usr.sbin/makefs/msdos/msdosfs_fat.c
index eacc448b09de..a70b5741aa78 100644
--- a/usr.sbin/makefs/msdos/msdosfs_fat.c
+++ b/usr.sbin/makefs/msdos/msdosfs_fat.c
@@ -1,1057 +1,1060 @@
/* $FreeBSD$ */
/* $NetBSD: msdosfs_fat.c,v 1.28 1997/11/17 15:36:49 ws Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1994, 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*-
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
* This software is provided "as is".
*
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
* October 1992
*/
#include <sys/param.h>
#include <sys/errno.h>
#include <assert.h>
#include <stdbool.h>
#include <stdio.h>
#include <string.h>
#include <strings.h>
-#include "ffs/buf.h"
-
#include <fs/msdosfs/bpb.h>
#include "msdos/direntry.h"
#include <fs/msdosfs/denode.h>
#include <fs/msdosfs/fat.h>
#include <fs/msdosfs/msdosfsmount.h>
+#undef clrbuf
+#include "ffs/buf.h"
#include "makefs.h"
#include "msdos.h"
#define FULL_RUN ((u_int)0xffffffff)
#define SYNCHRONOUS_WRITES(pmp) 1
static int chainalloc(struct msdosfsmount *pmp, u_long start,
u_long count, u_long fillwith, u_long *retcluster,
u_long *got);
static int chainlength(struct msdosfsmount *pmp, u_long start,
u_long count);
static void fatblock(struct msdosfsmount *pmp, u_long ofs, u_long *bnp,
u_long *sizep, u_long *bop);
static int fatchain(struct msdosfsmount *pmp, u_long start, u_long count,
u_long fillwith);
static void fc_lookup(struct denode *dep, u_long findcn, u_long *frcnp,
u_long *fsrcnp);
-static void updatefats(struct msdosfsmount *pmp, struct buf *bp,
+static void updatefats(struct msdosfsmount *pmp, struct m_buf *bp,
u_long fatbn);
static __inline void
usemap_alloc(struct msdosfsmount *pmp, u_long cn);
static __inline void
usemap_free(struct msdosfsmount *pmp, u_long cn);
static int clusteralloc1(struct msdosfsmount *pmp, u_long start,
u_long count, u_long fillwith, u_long *retcluster,
u_long *got);
static void
fatblock(struct msdosfsmount *pmp, u_long ofs, u_long *bnp, u_long *sizep,
u_long *bop)
{
u_long bn, size;
bn = ofs / pmp->pm_fatblocksize * pmp->pm_fatblocksec;
size = MIN(pmp->pm_fatblocksec, pmp->pm_FATsecs - bn)
* DEV_BSIZE;
bn += pmp->pm_fatblk + pmp->pm_curfat * pmp->pm_FATsecs;
if (bnp)
*bnp = bn;
if (sizep)
*sizep = size;
if (bop)
*bop = ofs % pmp->pm_fatblocksize;
}
/*
* Map the logical cluster number of a file into a physical disk sector
* that is filesystem relative.
*
* dep - address of denode representing the file of interest
* findcn - file relative cluster whose filesystem relative cluster number
* and/or block number are/is to be found
* bnp - address of where to place the filesystem relative block number.
* If this pointer is null then don't return this quantity.
* cnp - address of where to place the filesystem relative cluster number.
* If this pointer is null then don't return this quantity.
* sp - pointer to returned block size
*
* NOTE: Either bnp or cnp must be non-null.
* This function has one side effect. If the requested file relative cluster
* is beyond the end of file, then the actual number of clusters in the file
* is returned in *cnp. This is useful for determining how long a directory is.
* If cnp is null, nothing is returned.
*/
int
pcbmap(struct denode *dep, u_long findcn, daddr_t *bnp, u_long *cnp, int *sp)
{
int error;
u_long i;
u_long cn;
u_long prevcn = 0; /* XXX: prevcn could be used unititialized */
u_long byteoffset;
u_long bn;
u_long bo;
- struct buf *bp = NULL;
+ struct m_buf *bp = NULL;
u_long bp_bn = -1;
struct msdosfsmount *pmp = dep->de_pmp;
u_long bsize;
assert(bnp != NULL || cnp != NULL || sp != NULL);
cn = dep->de_StartCluster;
/*
* The "file" that makes up the root directory is contiguous,
* permanently allocated, of fixed size, and is not made up of
* clusters. If the cluster number is beyond the end of the root
* directory, then return the number of clusters in the file.
*/
if (cn == MSDOSFSROOT) {
if (dep->de_Attributes & ATTR_DIRECTORY) {
if (de_cn2off(pmp, findcn) >= dep->de_FileSize) {
if (cnp)
*cnp = de_bn2cn(pmp, pmp->pm_rootdirsize);
return (E2BIG);
}
if (bnp)
*bnp = pmp->pm_rootdirblk + de_cn2bn(pmp, findcn);
if (cnp)
*cnp = MSDOSFSROOT;
if (sp)
*sp = MIN(pmp->pm_bpcluster,
dep->de_FileSize - de_cn2off(pmp, findcn));
return (0);
} else { /* just an empty file */
if (cnp)
*cnp = 0;
return (E2BIG);
}
}
/*
* All other files do I/O in cluster sized blocks
*/
if (sp)
*sp = pmp->pm_bpcluster;
/*
* Rummage around in the FAT cache, maybe we can avoid tromping
* through every FAT entry for the file. And, keep track of how far
* off the cache was from where we wanted to be.
*/
i = 0;
fc_lookup(dep, findcn, &i, &cn);
/*
* Handle all other files or directories the normal way.
*/
for (; i < findcn; i++) {
/*
* Stop with all reserved clusters, not just with EOF.
*/
if ((cn | ~pmp->pm_fatmask) >= CLUST_RSRVD)
goto hiteof;
byteoffset = FATOFS(pmp, cn);
fatblock(pmp, byteoffset, &bn, &bsize, &bo);
if (bn != bp_bn) {
if (bp)
brelse(bp);
- error = bread(pmp->pm_devvp, bn, bsize, NOCRED, &bp);
+ error = bread((void *)pmp->pm_devvp, bn, bsize,
+ NOCRED, &bp);
if (error) {
brelse(bp);
return (error);
}
bp_bn = bn;
}
prevcn = cn;
if (bo >= bsize) {
if (bp)
brelse(bp);
return (EIO);
}
if (FAT32(pmp))
cn = getulong(bp->b_data + bo);
else
cn = getushort(bp->b_data + bo);
if (FAT12(pmp) && (prevcn & 1))
cn >>= 4;
cn &= pmp->pm_fatmask;
/*
* Force the special cluster numbers
* to be the same for all cluster sizes
* to let the rest of msdosfs handle
* all cases the same.
*/
if ((cn | ~pmp->pm_fatmask) >= CLUST_RSRVD)
cn |= ~pmp->pm_fatmask;
}
if (!MSDOSFSEOF(pmp, cn)) {
if (bp)
brelse(bp);
if (bnp)
*bnp = cntobn(pmp, cn);
if (cnp)
*cnp = cn;
fc_setcache(dep, FC_LASTMAP, i, cn);
return (0);
}
hiteof:;
if (cnp)
*cnp = i;
if (bp)
brelse(bp);
/* update last file cluster entry in the FAT cache */
fc_setcache(dep, FC_LASTFC, i - 1, prevcn);
return (E2BIG);
}
/*
* Find the closest entry in the FAT cache to the cluster we are looking
* for.
*/
static void
fc_lookup(struct denode *dep, u_long findcn, u_long *frcnp, u_long *fsrcnp)
{
int i;
u_long cn;
struct fatcache *closest = NULL;
for (i = 0; i < FC_SIZE; i++) {
cn = dep->de_fc[i].fc_frcn;
if (cn != FCE_EMPTY && cn <= findcn) {
if (closest == NULL || cn > closest->fc_frcn)
closest = &dep->de_fc[i];
}
}
if (closest) {
*frcnp = closest->fc_frcn;
*fsrcnp = closest->fc_fsrcn;
}
}
/*
* Purge the FAT cache in denode dep of all entries relating to file
* relative cluster frcn and beyond.
*/
void
fc_purge(struct denode *dep, u_int frcn)
{
int i;
struct fatcache *fcp;
fcp = dep->de_fc;
for (i = 0; i < FC_SIZE; i++, fcp++) {
if (fcp->fc_frcn >= frcn)
fcp->fc_frcn = FCE_EMPTY;
}
}
/*
* Update the FAT.
* If mirroring the FAT, update all copies, with the first copy as last.
* Else update only the current FAT (ignoring the others).
*
* pmp - msdosfsmount structure for filesystem to update
* bp - addr of modified FAT block
* fatbn - block number relative to begin of filesystem of the modified FAT block.
*/
static void
-updatefats(struct msdosfsmount *pmp, struct buf *bp, u_long fatbn)
+updatefats(struct msdosfsmount *pmp, struct m_buf *bp, u_long fatbn)
{
- struct buf *bpn;
+ struct m_buf *bpn;
int cleanfat, i;
#ifdef MSDOSFS_DEBUG
printf("updatefats(pmp %p, bp %p, fatbn %lu)\n", pmp, bp, fatbn);
#endif
if (pmp->pm_flags & MSDOSFS_FATMIRROR) {
/*
* Now copy the block(s) of the modified FAT to the other copies of
* the FAT and write them out. This is faster than reading in the
* other FATs and then writing them back out. This could tie up
* the FAT for quite a while. Preventing others from accessing it.
* To prevent us from going after the FAT quite so much we use
* delayed writes, unless they specified "synchronous" when the
* filesystem was mounted. If synch is asked for then use
* bwrite()'s and really slow things down.
*/
if (fatbn != pmp->pm_fatblk || FAT12(pmp))
cleanfat = 0;
else if (FAT16(pmp))
cleanfat = 16;
else
cleanfat = 32;
for (i = 1; i < pmp->pm_FATs; i++) {
fatbn += pmp->pm_FATsecs;
/* getblk() never fails */
- bpn = getblk(pmp->pm_devvp, fatbn, bp->b_bcount,
- 0, 0, 0);
+ bpn = getblk((void *)pmp->pm_devvp, fatbn,
+ bp->b_bcount, 0, 0, 0);
memcpy(bpn->b_data, bp->b_data, bp->b_bcount);
/* Force the clean bit on in the other copies. */
if (cleanfat == 16)
((uint8_t *)bpn->b_data)[3] |= 0x80;
else if (cleanfat == 32)
((uint8_t *)bpn->b_data)[7] |= 0x08;
if (SYNCHRONOUS_WRITES(pmp))
bwrite(bpn);
else
bdwrite(bpn);
}
}
/*
* Write out the first (or current) FAT last.
*/
if (SYNCHRONOUS_WRITES(pmp))
bwrite(bp);
else
bdwrite(bp);
}
/*
* Updating entries in 12 bit FATs is a pain in the butt.
*
* The following picture shows where nibbles go when moving from a 12 bit
* cluster number into the appropriate bytes in the FAT.
*
* byte m byte m+1 byte m+2
* +----+----+ +----+----+ +----+----+
* | 0 1 | | 2 3 | | 4 5 | FAT bytes
* +----+----+ +----+----+ +----+----+
*
* +----+----+----+ +----+----+----+
* | 3 0 1 | | 4 5 2 |
* +----+----+----+ +----+----+----+
* cluster n cluster n+1
*
* Where n is even. m = n + (n >> 2)
*
*/
static __inline void
usemap_alloc(struct msdosfsmount *pmp, u_long cn)
{
assert(cn <= pmp->pm_maxcluster);
assert((pmp->pm_flags & MSDOSFSMNT_RONLY) == 0);
assert((pmp->pm_inusemap[cn / N_INUSEBITS] & (1 << (cn % N_INUSEBITS)))
== 0);
assert(pmp->pm_freeclustercount > 0);
pmp->pm_inusemap[cn / N_INUSEBITS] |= 1U << (cn % N_INUSEBITS);
pmp->pm_freeclustercount--;
pmp->pm_flags |= MSDOSFS_FSIMOD;
}
static __inline void
usemap_free(struct msdosfsmount *pmp, u_long cn)
{
assert(cn <= pmp->pm_maxcluster);
assert((pmp->pm_flags & MSDOSFSMNT_RONLY) == 0);
assert((pmp->pm_inusemap[cn / N_INUSEBITS] & (1 << (cn % N_INUSEBITS)))
!= 0);
pmp->pm_freeclustercount++;
pmp->pm_flags |= MSDOSFS_FSIMOD;
pmp->pm_inusemap[cn / N_INUSEBITS] &= ~(1U << (cn % N_INUSEBITS));
}
int
clusterfree(struct msdosfsmount *pmp, u_long cluster, u_long *oldcnp)
{
int error;
u_long oldcn;
error = fatentry(FAT_GET_AND_SET, pmp, cluster, &oldcn, MSDOSFSFREE);
if (error)
return (error);
/*
* If the cluster was successfully marked free, then update
* the count of free clusters, and turn off the "allocated"
* bit in the "in use" cluster bit map.
*/
usemap_free(pmp, cluster);
if (oldcnp)
*oldcnp = oldcn;
return (0);
}
/*
* Get or Set or 'Get and Set' the cluster'th entry in the FAT.
*
* function - whether to get or set a FAT entry
* pmp - address of the msdosfsmount structure for the filesystem
* whose FAT is to be manipulated.
* cn - which cluster is of interest
* oldcontents - address of a word that is to receive the contents of the
* cluster'th entry if this is a get function
* newcontents - the new value to be written into the cluster'th element of
* the FAT if this is a set function.
*
* This function can also be used to free a cluster by setting the FAT entry
* for a cluster to 0.
*
* All copies of the FAT are updated if this is a set function. NOTE: If
* fatentry() marks a cluster as free it does not update the inusemap in
* the msdosfsmount structure. This is left to the caller.
*/
int
fatentry(int function, struct msdosfsmount *pmp, u_long cn, u_long *oldcontents,
u_long newcontents)
{
int error;
u_long readcn;
u_long bn, bo, bsize, byteoffset;
- struct buf *bp;
+ struct m_buf *bp;
#ifdef MSDOSFS_DEBUG
printf("fatentry(func %d, pmp %p, clust %lu, oldcon %p, newcon %lx)\n",
function, pmp, cn, oldcontents, newcontents);
#endif
#ifdef DIAGNOSTIC
/*
* Be sure they asked us to do something.
*/
if ((function & (FAT_SET | FAT_GET)) == 0) {
#ifdef MSDOSFS_DEBUG
printf("fatentry(): function code doesn't specify get or set\n");
#endif
return (EINVAL);
}
/*
* If they asked us to return a cluster number but didn't tell us
* where to put it, give them an error.
*/
if ((function & FAT_GET) && oldcontents == NULL) {
#ifdef MSDOSFS_DEBUG
printf("fatentry(): get function with no place to put result\n");
#endif
return (EINVAL);
}
#endif
/*
* Be sure the requested cluster is in the filesystem.
*/
if (cn < CLUST_FIRST || cn > pmp->pm_maxcluster)
return (EINVAL);
byteoffset = FATOFS(pmp, cn);
fatblock(pmp, byteoffset, &bn, &bsize, &bo);
- error = bread(pmp->pm_devvp, bn, bsize, NOCRED, &bp);
+ error = bread((void *)pmp->pm_devvp, bn, bsize, NOCRED, &bp);
if (error) {
brelse(bp);
return (error);
}
if (function & FAT_GET) {
if (FAT32(pmp))
readcn = getulong(bp->b_data + bo);
else
readcn = getushort(bp->b_data + bo);
if (FAT12(pmp) & (cn & 1))
readcn >>= 4;
readcn &= pmp->pm_fatmask;
/* map reserved FAT entries to same values for all FATs */
if ((readcn | ~pmp->pm_fatmask) >= CLUST_RSRVD)
readcn |= ~pmp->pm_fatmask;
*oldcontents = readcn;
}
if (function & FAT_SET) {
switch (pmp->pm_fatmask) {
case FAT12_MASK:
readcn = getushort(bp->b_data + bo);
if (cn & 1) {
readcn &= 0x000f;
readcn |= newcontents << 4;
} else {
readcn &= 0xf000;
readcn |= newcontents & 0xfff;
}
putushort(bp->b_data + bo, readcn);
break;
case FAT16_MASK:
putushort(bp->b_data + bo, newcontents);
break;
case FAT32_MASK:
/*
* According to spec we have to retain the
* high order bits of the FAT entry.
*/
readcn = getulong(bp->b_data + bo);
readcn &= ~FAT32_MASK;
readcn |= newcontents & FAT32_MASK;
putulong(bp->b_data + bo, readcn);
break;
}
updatefats(pmp, bp, bn);
bp = NULL;
pmp->pm_fmod = 1;
}
if (bp)
brelse(bp);
return (0);
}
/*
* Update a contiguous cluster chain
*
* pmp - mount point
* start - first cluster of chain
* count - number of clusters in chain
* fillwith - what to write into FAT entry of last cluster
*/
static int
fatchain(struct msdosfsmount *pmp, u_long start, u_long count, u_long fillwith)
{
int error;
u_long bn, bo, bsize, byteoffset, readcn, newc;
- struct buf *bp;
+ struct m_buf *bp;
#ifdef MSDOSFS_DEBUG
printf("fatchain(pmp %p, start %lu, count %lu, fillwith %lx)\n",
pmp, start, count, fillwith);
#endif
/*
* Be sure the clusters are in the filesystem.
*/
if (start < CLUST_FIRST || start + count - 1 > pmp->pm_maxcluster)
return (EINVAL);
while (count > 0) {
byteoffset = FATOFS(pmp, start);
fatblock(pmp, byteoffset, &bn, &bsize, &bo);
- error = bread(pmp->pm_devvp, bn, bsize, NOCRED, &bp);
+ error = bread((void *)pmp->pm_devvp, bn, bsize, NOCRED, &bp);
if (error) {
brelse(bp);
return (error);
}
while (count > 0) {
start++;
newc = --count > 0 ? start : fillwith;
switch (pmp->pm_fatmask) {
case FAT12_MASK:
readcn = getushort(bp->b_data + bo);
if (start & 1) {
readcn &= 0xf000;
readcn |= newc & 0xfff;
} else {
readcn &= 0x000f;
readcn |= newc << 4;
}
putushort(bp->b_data + bo, readcn);
bo++;
if (!(start & 1))
bo++;
break;
case FAT16_MASK:
putushort(bp->b_data + bo, newc);
bo += 2;
break;
case FAT32_MASK:
readcn = getulong(bp->b_data + bo);
readcn &= ~pmp->pm_fatmask;
readcn |= newc & pmp->pm_fatmask;
putulong(bp->b_data + bo, readcn);
bo += 4;
break;
}
if (bo >= bsize)
break;
}
updatefats(pmp, bp, bn);
}
pmp->pm_fmod = 1;
return (0);
}
/*
* Check the length of a free cluster chain starting at start.
*
* pmp - mount point
* start - start of chain
* count - maximum interesting length
*/
static int
chainlength(struct msdosfsmount *pmp, u_long start, u_long count)
{
u_long idx, max_idx;
u_int map;
u_long len;
if (start > pmp->pm_maxcluster)
return (0);
max_idx = pmp->pm_maxcluster / N_INUSEBITS;
idx = start / N_INUSEBITS;
start %= N_INUSEBITS;
map = pmp->pm_inusemap[idx];
map &= ~((1 << start) - 1);
if (map) {
len = ffs(map) - 1 - start;
len = MIN(len, count);
if (start + len > pmp->pm_maxcluster)
len = pmp->pm_maxcluster - start + 1;
return (len);
}
len = N_INUSEBITS - start;
if (len >= count) {
len = count;
if (start + len > pmp->pm_maxcluster)
len = pmp->pm_maxcluster - start + 1;
return (len);
}
while (++idx <= max_idx) {
if (len >= count)
break;
map = pmp->pm_inusemap[idx];
if (map) {
len += ffs(map) - 1;
break;
}
len += N_INUSEBITS;
}
len = MIN(len, count);
if (start + len > pmp->pm_maxcluster)
len = pmp->pm_maxcluster - start + 1;
return (len);
}
/*
* Allocate contigous free clusters.
*
* pmp - mount point.
* start - start of cluster chain.
* count - number of clusters to allocate.
* fillwith - put this value into the FAT entry for the
* last allocated cluster.
* retcluster - put the first allocated cluster's number here.
* got - how many clusters were actually allocated.
*/
static int
chainalloc(struct msdosfsmount *pmp, u_long start, u_long count,
u_long fillwith, u_long *retcluster, u_long *got)
{
int error;
u_long cl, n;
assert((pmp->pm_flags & MSDOSFSMNT_RONLY) == 0);
for (cl = start, n = count; n-- > 0;)
usemap_alloc(pmp, cl++);
pmp->pm_nxtfree = start + count;
if (pmp->pm_nxtfree > pmp->pm_maxcluster)
pmp->pm_nxtfree = CLUST_FIRST;
pmp->pm_flags |= MSDOSFS_FSIMOD;
error = fatchain(pmp, start, count, fillwith);
if (error != 0) {
for (cl = start, n = count; n-- > 0;)
usemap_free(pmp, cl++);
return (error);
}
#ifdef MSDOSFS_DEBUG
printf("clusteralloc(): allocated cluster chain at %lu (%lu clusters)\n",
start, count);
#endif
if (retcluster)
*retcluster = start;
if (got)
*got = count;
return (0);
}
/*
* Allocate contiguous free clusters.
*
* pmp - mount point.
* start - preferred start of cluster chain.
* count - number of clusters requested.
* fillwith - put this value into the FAT entry for the
* last allocated cluster.
* retcluster - put the first allocated cluster's number here.
* got - how many clusters were actually allocated.
*/
int
clusteralloc(struct msdosfsmount *pmp, u_long start, u_long count,
u_long fillwith, u_long *retcluster, u_long *got)
{
int error;
error = clusteralloc1(pmp, start, count, fillwith, retcluster, got);
return (error);
}
static int
clusteralloc1(struct msdosfsmount *pmp, u_long start, u_long count,
u_long fillwith, u_long *retcluster, u_long *got)
{
u_long idx;
u_long len, newst, foundl, cn, l;
u_long foundcn = 0; /* XXX: foundcn could be used unititialized */
u_int map;
MSDOSFS_DPRINTF(("clusteralloc(): find %lu clusters\n", count));
if (start) {
if ((len = chainlength(pmp, start, count)) >= count)
return (chainalloc(pmp, start, count, fillwith, retcluster, got));
} else
len = 0;
newst = pmp->pm_nxtfree;
foundl = 0;
for (cn = newst; cn <= pmp->pm_maxcluster;) {
idx = cn / N_INUSEBITS;
map = pmp->pm_inusemap[idx];
map |= (1U << (cn % N_INUSEBITS)) - 1;
if (map != FULL_RUN) {
cn = idx * N_INUSEBITS + ffs(map ^ FULL_RUN) - 1;
if ((l = chainlength(pmp, cn, count)) >= count)
return (chainalloc(pmp, cn, count, fillwith, retcluster, got));
if (l > foundl) {
foundcn = cn;
foundl = l;
}
cn += l + 1;
continue;
}
cn += N_INUSEBITS - cn % N_INUSEBITS;
}
for (cn = 0; cn < newst;) {
idx = cn / N_INUSEBITS;
map = pmp->pm_inusemap[idx];
map |= (1U << (cn % N_INUSEBITS)) - 1;
if (map != FULL_RUN) {
cn = idx * N_INUSEBITS + ffs(map ^ FULL_RUN) - 1;
if ((l = chainlength(pmp, cn, count)) >= count)
return (chainalloc(pmp, cn, count, fillwith, retcluster, got));
if (l > foundl) {
foundcn = cn;
foundl = l;
}
cn += l + 1;
continue;
}
cn += N_INUSEBITS - cn % N_INUSEBITS;
}
if (!foundl)
return (ENOSPC);
if (len)
return (chainalloc(pmp, start, len, fillwith, retcluster, got));
else
return (chainalloc(pmp, foundcn, foundl, fillwith, retcluster, got));
}
/*
* Free a chain of clusters.
*
* pmp - address of the msdosfs mount structure for the filesystem
* containing the cluster chain to be freed.
* startcluster - number of the 1st cluster in the chain of clusters to be
* freed.
*/
int
freeclusterchain(struct msdosfsmount *pmp, u_long cluster)
{
int error;
- struct buf *bp = NULL;
+ struct m_buf *bp = NULL;
u_long bn, bo, bsize, byteoffset;
u_long readcn, lbn = -1;
while (cluster >= CLUST_FIRST && cluster <= pmp->pm_maxcluster) {
byteoffset = FATOFS(pmp, cluster);
fatblock(pmp, byteoffset, &bn, &bsize, &bo);
if (lbn != bn) {
if (bp)
updatefats(pmp, bp, lbn);
- error = bread(pmp->pm_devvp, bn, bsize, NOCRED, &bp);
+ error = bread((void *)pmp->pm_devvp, bn, bsize,
+ NOCRED, &bp);
if (error) {
brelse(bp);
return (error);
}
lbn = bn;
}
usemap_free(pmp, cluster);
switch (pmp->pm_fatmask) {
case FAT12_MASK:
readcn = getushort(bp->b_data + bo);
if (cluster & 1) {
cluster = readcn >> 4;
readcn &= 0x000f;
readcn |= MSDOSFSFREE << 4;
} else {
cluster = readcn;
readcn &= 0xf000;
readcn |= MSDOSFSFREE & 0xfff;
}
putushort(bp->b_data + bo, readcn);
break;
case FAT16_MASK:
cluster = getushort(bp->b_data + bo);
putushort(bp->b_data + bo, MSDOSFSFREE);
break;
case FAT32_MASK:
cluster = getulong(bp->b_data + bo);
putulong(bp->b_data + bo,
(MSDOSFSFREE & FAT32_MASK) | (cluster & ~FAT32_MASK));
break;
}
cluster &= pmp->pm_fatmask;
if ((cluster | ~pmp->pm_fatmask) >= CLUST_RSRVD)
cluster |= pmp->pm_fatmask;
}
if (bp)
updatefats(pmp, bp, bn);
return (0);
}
/*
* Read in FAT blocks looking for free clusters. For every free cluster
* found turn off its corresponding bit in the pm_inusemap.
*/
int
fillinusemap(struct msdosfsmount *pmp)
{
- struct buf *bp;
+ struct m_buf *bp;
u_long bn, bo, bsize, byteoffset, cn, readcn;
int error;
bp = NULL;
/*
* Mark all clusters in use, we mark the free ones in the FAT scan
* loop further down.
*/
for (cn = 0; cn < (pmp->pm_maxcluster + N_INUSEBITS) / N_INUSEBITS; cn++)
pmp->pm_inusemap[cn] = FULL_RUN;
/*
* Figure how many free clusters are in the filesystem by ripping
* through the FAT counting the number of entries whose content is
* zero. These represent free clusters.
*/
pmp->pm_freeclustercount = 0;
for (cn = 0; cn <= pmp->pm_maxcluster; cn++) {
byteoffset = FATOFS(pmp, cn);
bo = byteoffset % pmp->pm_fatblocksize;
if (bo == 0) {
/* Read new FAT block */
if (bp != NULL)
brelse(bp);
fatblock(pmp, byteoffset, &bn, &bsize, NULL);
- error = bread(pmp->pm_devvp, bn, bsize, NOCRED, &bp);
+ error = bread((void *)pmp->pm_devvp, bn, bsize,
+ NOCRED, &bp);
if (error != 0)
return (error);
}
if (FAT32(pmp))
readcn = getulong(bp->b_data + bo);
else
readcn = getushort(bp->b_data + bo);
if (FAT12(pmp) && (cn & 1))
readcn >>= 4;
readcn &= pmp->pm_fatmask;
/*
* Check if the FAT ID matches the BPB's media descriptor and
* all other bits are set to 1.
*/
if (cn == 0 && readcn != ((pmp->pm_fatmask & 0xffffff00) |
pmp->pm_bpb.bpbMedia)) {
#ifdef MSDOSFS_DEBUG
printf("mountmsdosfs(): Media descriptor in BPB"
"does not match FAT ID\n");
#endif
brelse(bp);
return (EINVAL);
} else if (readcn == CLUST_FREE)
usemap_free(pmp, cn);
}
if (bp != NULL)
brelse(bp);
for (cn = pmp->pm_maxcluster + 1; cn < (pmp->pm_maxcluster +
N_INUSEBITS) / N_INUSEBITS; cn++)
pmp->pm_inusemap[cn / N_INUSEBITS] |= 1U << (cn % N_INUSEBITS);
return (0);
}
/*
* Allocate a new cluster and chain it onto the end of the file.
*
* dep - the file to extend
* count - number of clusters to allocate
* bpp - where to return the address of the buf header for the first new
* file block
* ncp - where to put cluster number of the first newly allocated cluster
* If this pointer is 0, do not return the cluster number.
* flags - see fat.h
*
* NOTE: This function is not responsible for turning on the DE_UPDATE bit of
* the de_flag field of the denode and it does not change the de_FileSize
* field. This is left for the caller to do.
*/
int
-extendfile(struct denode *dep, u_long count, struct buf **bpp, u_long *ncp,
+m_extendfile(struct denode *dep, u_long count, struct m_buf **bpp, u_long *ncp,
int flags)
{
int error;
u_long frcn;
u_long cn, got;
struct msdosfsmount *pmp = dep->de_pmp;
- struct buf *bp;
+ struct m_buf *bp;
/*
* Don't try to extend the root directory
*/
if (dep->de_StartCluster == MSDOSFSROOT
&& (dep->de_Attributes & ATTR_DIRECTORY)) {
#ifdef MSDOSFS_DEBUG
printf("extendfile(): attempt to extend root directory\n");
#endif
return (ENOSPC);
}
/*
* If the "file's last cluster" cache entry is empty, and the file
* is not empty, then fill the cache entry by calling pcbmap().
*/
if (dep->de_fc[FC_LASTFC].fc_frcn == FCE_EMPTY &&
dep->de_StartCluster != 0) {
error = pcbmap(dep, 0xffff, 0, &cn, 0);
/* we expect it to return E2BIG */
if (error != E2BIG)
return (error);
}
dep->de_fc[FC_NEXTTOLASTFC].fc_frcn =
dep->de_fc[FC_LASTFC].fc_frcn;
dep->de_fc[FC_NEXTTOLASTFC].fc_fsrcn =
dep->de_fc[FC_LASTFC].fc_fsrcn;
while (count > 0) {
/*
* Allocate a new cluster chain and cat onto the end of the
* file. If the file is empty we make de_StartCluster point
* to the new block. Note that de_StartCluster being 0 is
* sufficient to be sure the file is empty since we exclude
* attempts to extend the root directory above, and the root
* dir is the only file with a startcluster of 0 that has
* blocks allocated (sort of).
*/
if (dep->de_StartCluster == 0)
cn = 0;
else
cn = dep->de_fc[FC_LASTFC].fc_fsrcn + 1;
error = clusteralloc(pmp, cn, count, CLUST_EOFE, &cn, &got);
if (error)
return (error);
count -= got;
/*
* Give them the filesystem relative cluster number if they want
* it.
*/
if (ncp) {
*ncp = cn;
ncp = NULL;
}
if (dep->de_StartCluster == 0) {
dep->de_StartCluster = cn;
frcn = 0;
} else {
error = fatentry(FAT_SET, pmp,
dep->de_fc[FC_LASTFC].fc_fsrcn,
0, cn);
if (error) {
clusterfree(pmp, cn, NULL);
return (error);
}
frcn = dep->de_fc[FC_LASTFC].fc_frcn + 1;
}
/*
* Update the "last cluster of the file" entry in the
* denode's FAT cache.
*/
fc_setcache(dep, FC_LASTFC, frcn + got - 1, cn + got - 1);
if ((flags & DE_CLEAR) &&
(dep->de_Attributes & ATTR_DIRECTORY)) {
while (got-- > 0) {
- bp = getblk(pmp->pm_devvp,
+ bp = getblk((void *)pmp->pm_devvp,
cntobn(pmp, cn++),
pmp->pm_bpcluster, 0, 0, 0);
clrbuf(bp);
if (bpp) {
*bpp = bp;
bpp = NULL;
} else {
bdwrite(bp);
}
}
}
}
return (0);
}
diff --git a/usr.sbin/makefs/msdos/msdosfs_lookup.c b/usr.sbin/makefs/msdos/msdosfs_lookup.c
index 27ce216a488f..3fca0532468b 100644
--- a/usr.sbin/makefs/msdos/msdosfs_lookup.c
+++ b/usr.sbin/makefs/msdos/msdosfs_lookup.c
@@ -1,302 +1,306 @@
/* $FreeBSD$ */
/* $NetBSD: msdosfs_lookup.c,v 1.37 1997/11/17 15:36:54 ws Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1994, 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*-
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
* This software is provided "as is".
*
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
* October 1992
*/
#include <sys/param.h>
#include <sys/errno.h>
#include <stdbool.h>
#include <stdio.h>
#include <string.h>
-#include "ffs/buf.h"
#include <fs/msdosfs/bpb.h>
#include "msdos/direntry.h"
#include <fs/msdosfs/denode.h>
#include <fs/msdosfs/fat.h>
#include <fs/msdosfs/msdosfsmount.h>
+#undef clrbuf
+#include "ffs/buf.h"
#include "makefs.h"
#include "msdos.h"
/*
* dep - directory entry to copy into the directory
* ddep - directory to add to
* depp - return the address of the denode for the created directory entry
* if depp != 0
* cnp - componentname needed for Win95 long filenames
*/
int
createde(struct denode *dep, struct denode *ddep, struct denode **depp,
struct componentname *cnp)
{
int error;
u_long dirclust, diroffset;
struct direntry *ndep;
struct msdosfsmount *pmp = ddep->de_pmp;
- struct buf *bp;
+ struct m_buf *bp;
daddr_t bn;
int blsize;
MSDOSFS_DPRINTF(("createde(dep %p, ddep %p, depp %p, cnp %p)\n",
dep, ddep, depp, cnp));
/*
* If no space left in the directory then allocate another cluster
* and chain it onto the end of the file. There is one exception
* to this. That is, if the root directory has no more space it
* can NOT be expanded. extendfile() checks for and fails attempts
* to extend the root directory. We just return an error in that
* case.
*/
if (ddep->de_fndoffset >= ddep->de_FileSize) {
diroffset = ddep->de_fndoffset + sizeof(struct direntry)
- ddep->de_FileSize;
dirclust = de_clcount(pmp, diroffset);
- error = extendfile(ddep, dirclust, 0, 0, DE_CLEAR);
+ error = m_extendfile(ddep, dirclust, 0, 0, DE_CLEAR);
if (error) {
(void)detrunc(ddep, ddep->de_FileSize, 0, NULL);
return error;
}
/*
* Update the size of the directory
*/
ddep->de_FileSize += de_cn2off(pmp, dirclust);
}
/*
* We just read in the cluster with space. Copy the new directory
* entry in. Then write it to disk. NOTE: DOS directories
* do not get smaller as clusters are emptied.
*/
error = pcbmap(ddep, de_cluster(pmp, ddep->de_fndoffset),
&bn, &dirclust, &blsize);
if (error)
return error;
diroffset = ddep->de_fndoffset;
if (dirclust != MSDOSFSROOT)
diroffset &= pmp->pm_crbomask;
- if ((error = bread(pmp->pm_devvp, bn, blsize, NOCRED, &bp)) != 0) {
+ if ((error = bread((void *)pmp->pm_devvp, bn, blsize, NOCRED,
+ &bp)) != 0) {
brelse(bp);
return error;
}
ndep = bptoep(pmp, bp, ddep->de_fndoffset);
DE_EXTERNALIZE(ndep, dep);
/*
* Now write the Win95 long name
*/
if (ddep->de_fndcnt > 0) {
uint8_t chksum = winChksum(ndep->deName);
const u_char *un = (const u_char *)cnp->cn_nameptr;
int unlen = cnp->cn_namelen;
int cnt = 1;
while (--ddep->de_fndcnt >= 0) {
if (!(ddep->de_fndoffset & pmp->pm_crbomask)) {
if ((error = bwrite(bp)) != 0)
return error;
ddep->de_fndoffset -= sizeof(struct direntry);
error = pcbmap(ddep,
de_cluster(pmp,
ddep->de_fndoffset),
&bn, 0, &blsize);
if (error)
return error;
- error = bread(pmp->pm_devvp, bn, blsize,
+ error = bread((void *)pmp->pm_devvp, bn, blsize,
NOCRED, &bp);
if (error) {
brelse(bp);
return error;
}
ndep = bptoep(pmp, bp, ddep->de_fndoffset);
} else {
ndep--;
ddep->de_fndoffset -= sizeof(struct direntry);
}
if (!unix2winfn(un, unlen, (struct winentry *)ndep,
cnt++, chksum))
break;
}
}
if ((error = bwrite(bp)) != 0)
return error;
/*
* If they want us to return with the denode gotten.
*/
if (depp) {
if (dep->de_Attributes & ATTR_DIRECTORY) {
dirclust = dep->de_StartCluster;
if (FAT32(pmp) && dirclust == pmp->pm_rootdirblk)
dirclust = MSDOSFSROOT;
if (dirclust == MSDOSFSROOT)
diroffset = MSDOSFSROOT_OFS;
else
diroffset = 0;
}
return deget(pmp, dirclust, diroffset, depp);
}
return 0;
}
/*
* Read in the disk block containing the directory entry (dirclu, dirofs)
* and return the address of the buf header, and the address of the
* directory entry within the block.
*/
int
-readep(struct msdosfsmount *pmp, u_long dirclust, u_long diroffset,
- struct buf **bpp, struct direntry **epp)
+m_readep(struct msdosfsmount *pmp, u_long dirclust, u_long diroffset,
+ struct m_buf **bpp, struct direntry **epp)
{
int error;
daddr_t bn;
int blsize;
blsize = pmp->pm_bpcluster;
if (dirclust == MSDOSFSROOT
&& de_blk(pmp, diroffset + blsize) > pmp->pm_rootdirsize)
blsize = de_bn2off(pmp, pmp->pm_rootdirsize) & pmp->pm_crbomask;
bn = detobn(pmp, dirclust, diroffset);
- if ((error = bread(pmp->pm_devvp, bn, blsize, NOCRED, bpp)) != 0) {
+ if ((error = bread((void *)pmp->pm_devvp, bn, blsize, NOCRED,
+ bpp)) != 0) {
brelse(*bpp);
*bpp = NULL;
return (error);
}
if (epp)
*epp = bptoep(pmp, *bpp, diroffset);
return (0);
}
/*
* Read in the disk block containing the directory entry dep came from and
* return the address of the buf header, and the address of the directory
* entry within the block.
*/
int
-readde(struct denode *dep, struct buf **bpp, struct direntry **epp)
+m_readde(struct denode *dep, struct m_buf **bpp, struct direntry **epp)
{
- return (readep(dep->de_pmp, dep->de_dirclust, dep->de_diroffset,
+ return (m_readep(dep->de_pmp, dep->de_dirclust, dep->de_diroffset,
bpp, epp));
}
/*
* Create a unique DOS name in dvp
*/
int
uniqdosname(struct denode *dep, struct componentname *cnp, u_char *cp)
{
struct msdosfsmount *pmp = dep->de_pmp;
struct direntry *dentp;
int gen;
int blsize;
u_long cn;
daddr_t bn;
- struct buf *bp;
+ struct m_buf *bp;
int error;
if (pmp->pm_flags & MSDOSFSMNT_SHORTNAME)
return (unix2dosfn((const u_char *)cnp->cn_nameptr, cp,
cnp->cn_namelen, 0) ? 0 : EINVAL);
for (gen = 1;; gen++) {
/*
* Generate DOS name with generation number
*/
if (!unix2dosfn((const u_char *)cnp->cn_nameptr, cp,
cnp->cn_namelen, gen))
return gen == 1 ? EINVAL : EEXIST;
/*
* Now look for a dir entry with this exact name
*/
for (cn = error = 0; !error; cn++) {
if ((error = pcbmap(dep, cn, &bn, 0, &blsize)) != 0) {
if (error == E2BIG) /* EOF reached and not found */
return 0;
return error;
}
- error = bread(pmp->pm_devvp, bn, blsize, NOCRED, &bp);
+ error = bread((void *)pmp->pm_devvp, bn, blsize,
+ NOCRED, &bp);
if (error) {
brelse(bp);
return error;
}
for (dentp = (struct direntry *)bp->b_data;
(char *)dentp < bp->b_data + blsize;
dentp++) {
if (dentp->deName[0] == SLOT_EMPTY) {
/*
* Last used entry and not found
*/
brelse(bp);
return 0;
}
/*
* Ignore volume labels and Win95 entries
*/
if (dentp->deAttributes & ATTR_VOLUME)
continue;
if (!bcmp(dentp->deName, cp, 11)) {
error = EEXIST;
break;
}
}
brelse(bp);
}
}
}
diff --git a/usr.sbin/makefs/msdos/msdosfs_vfsops.c b/usr.sbin/makefs/msdos/msdosfs_vfsops.c
index 07fc91cccd4a..b6788f6876f9 100644
--- a/usr.sbin/makefs/msdos/msdosfs_vfsops.c
+++ b/usr.sbin/makefs/msdos/msdosfs_vfsops.c
@@ -1,392 +1,391 @@
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1994, 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*-
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
* This software is provided "as is".
*
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
* October 1992
*/
#include <sys/cdefs.h>
/* $NetBSD: msdosfs_vfsops.c,v 1.10 2016/01/30 09:59:27 mlelstv Exp $ */
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/mount.h>
#include <errno.h>
#include <stdbool.h>
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <util.h>
-#include "ffs/buf.h"
#include <fs/msdosfs/bootsect.h>
#include <fs/msdosfs/bpb.h>
#include "msdos/direntry.h"
#include <fs/msdosfs/denode.h>
#include <fs/msdosfs/fat.h>
#include <fs/msdosfs/msdosfsmount.h>
#include <mkfs_msdos.h>
+#undef clrbuf
+#include "ffs/buf.h"
#include "makefs.h"
#include "msdos.h"
-
-
struct msdosfsmount *
-msdosfs_mount(struct vnode *devvp)
+m_msdosfs_mount(struct m_vnode *devvp)
{
struct msdosfsmount *pmp = NULL;
- struct buf *bp;
+ struct m_buf *bp;
union bootsector *bsp;
struct byte_bpb33 *b33;
struct byte_bpb50 *b50;
struct byte_bpb710 *b710;
uint8_t SecPerClust;
int ronly = 0, error;
int bsize;
unsigned secsize = 512;
MSDOSFS_DPRINTF(("%s(bread 0)\n", __func__));
- if ((error = bread(devvp, 0, secsize, 0, &bp)) != 0)
+ if ((error = bread((void *)devvp, 0, secsize, 0, &bp)) != 0)
goto error_exit;
bsp = (union bootsector *)bp->b_data;
b33 = (struct byte_bpb33 *)bsp->bs33.bsBPB;
b50 = (struct byte_bpb50 *)bsp->bs50.bsBPB;
b710 = (struct byte_bpb710 *)bsp->bs710.bsBPB;
if (bsp->bs50.bsBootSectSig0 != BOOTSIG0 ||
bsp->bs50.bsBootSectSig1 != BOOTSIG1) {
MSDOSFS_DPRINTF(("bootsig0 %d bootsig1 %d\n",
bsp->bs50.bsBootSectSig0,
bsp->bs50.bsBootSectSig1));
error = EINVAL;
goto error_exit;
}
bsize = 0;
pmp = ecalloc(1, sizeof(*pmp));
/*
* Compute several useful quantities from the bpb in the
* bootsector. Copy in the dos 5 variant of the bpb then fix up
* the fields that are different between dos 5 and dos 3.3.
*/
SecPerClust = b50->bpbSecPerClust;
pmp->pm_BytesPerSec = getushort(b50->bpbBytesPerSec);
pmp->pm_ResSectors = getushort(b50->bpbResSectors);
pmp->pm_FATs = b50->bpbFATs;
pmp->pm_RootDirEnts = getushort(b50->bpbRootDirEnts);
pmp->pm_Sectors = getushort(b50->bpbSectors);
pmp->pm_FATsecs = getushort(b50->bpbFATsecs);
pmp->pm_SecPerTrack = getushort(b50->bpbSecPerTrack);
pmp->pm_Heads = getushort(b50->bpbHeads);
pmp->pm_Media = b50->bpbMedia;
MSDOSFS_DPRINTF(("%s(BytesPerSec=%u, ResSectors=%u, FATs=%d, "
"RootDirEnts=%u, Sectors=%u, FATsecs=%lu, SecPerTrack=%u, "
"Heads=%u, Media=%u)\n",
__func__, pmp->pm_BytesPerSec, pmp->pm_ResSectors,
pmp->pm_FATs, pmp->pm_RootDirEnts, pmp->pm_Sectors,
pmp->pm_FATsecs, pmp->pm_SecPerTrack, pmp->pm_Heads,
pmp->pm_Media));
/* XXX - We should probably check more values here */
if (!pmp->pm_BytesPerSec || !SecPerClust
|| pmp->pm_SecPerTrack > 63) {
MSDOSFS_DPRINTF(("bytespersec %d secperclust %d "
"secpertrack %d\n", pmp->pm_BytesPerSec,
SecPerClust, pmp->pm_SecPerTrack));
error = EINVAL;
goto error_exit;
}
if (pmp->pm_Sectors == 0) {
pmp->pm_HiddenSects = getulong(b50->bpbHiddenSecs);
pmp->pm_HugeSectors = getulong(b50->bpbHugeSectors);
} else {
pmp->pm_HiddenSects = getushort(b33->bpbHiddenSecs);
pmp->pm_HugeSectors = pmp->pm_Sectors;
}
pmp->pm_flags = 0;
if (pmp->pm_RootDirEnts == 0) {
unsigned short vers = getushort(b710->bpbFSVers);
/*
* Some say that bsBootSectSig[23] must be zero, but
* Windows does not require this and some digital cameras
* do not set these to zero. Therefore, do not insist.
*/
if (pmp->pm_Sectors || pmp->pm_FATsecs || vers) {
MSDOSFS_DPRINTF(("sectors %d fatsecs %lu vers %d\n",
pmp->pm_Sectors, pmp->pm_FATsecs, vers));
error = EINVAL;
goto error_exit;
}
pmp->pm_fatmask = FAT32_MASK;
pmp->pm_fatmult = 4;
pmp->pm_fatdiv = 1;
pmp->pm_FATsecs = getulong(b710->bpbBigFATsecs);
/* mirrorring is enabled if the FATMIRROR bit is not set */
if ((getushort(b710->bpbExtFlags) & FATMIRROR) == 0)
pmp->pm_flags |= MSDOSFS_FATMIRROR;
else
pmp->pm_curfat = getushort(b710->bpbExtFlags) & FATNUM;
} else
pmp->pm_flags |= MSDOSFS_FATMIRROR;
/* Check that fs has nonzero FAT size */
if (pmp->pm_FATsecs == 0) {
MSDOSFS_DPRINTF(("FATsecs is 0\n"));
error = EINVAL;
goto error_exit;
}
pmp->pm_fatblk = pmp->pm_ResSectors;
if (FAT32(pmp)) {
pmp->pm_rootdirblk = getulong(b710->bpbRootClust);
pmp->pm_firstcluster = pmp->pm_fatblk
+ (pmp->pm_FATs * pmp->pm_FATsecs);
pmp->pm_fsinfo = getushort(b710->bpbFSInfo);
} else {
pmp->pm_rootdirblk = pmp->pm_fatblk +
(pmp->pm_FATs * pmp->pm_FATsecs);
pmp->pm_rootdirsize = (pmp->pm_RootDirEnts * sizeof(struct direntry)
+ pmp->pm_BytesPerSec - 1)
/ pmp->pm_BytesPerSec;/* in sectors */
pmp->pm_firstcluster = pmp->pm_rootdirblk + pmp->pm_rootdirsize;
}
pmp->pm_maxcluster = ((pmp->pm_HugeSectors - pmp->pm_firstcluster) /
SecPerClust) + 1;
pmp->pm_fatsize = pmp->pm_FATsecs * pmp->pm_BytesPerSec;
if (pmp->pm_fatmask == 0) {
if (pmp->pm_maxcluster
<= ((CLUST_RSRVD - CLUST_FIRST) & FAT12_MASK)) {
/*
* This will usually be a floppy disk. This size makes
* sure that one FAT entry will not be split across
* multiple blocks.
*/
pmp->pm_fatmask = FAT12_MASK;
pmp->pm_fatmult = 3;
pmp->pm_fatdiv = 2;
} else {
pmp->pm_fatmask = FAT16_MASK;
pmp->pm_fatmult = 2;
pmp->pm_fatdiv = 1;
}
}
if (FAT12(pmp))
pmp->pm_fatblocksize = 3 * pmp->pm_BytesPerSec;
else
pmp->pm_fatblocksize = MAXBSIZE;
pmp->pm_fatblocksec = pmp->pm_fatblocksize / pmp->pm_BytesPerSec;
pmp->pm_bnshift = ffs(pmp->pm_BytesPerSec) - 1;
/*
* Compute mask and shift value for isolating cluster relative byte
* offsets and cluster numbers from a file offset.
*/
pmp->pm_bpcluster = SecPerClust * pmp->pm_BytesPerSec;
pmp->pm_crbomask = pmp->pm_bpcluster - 1;
pmp->pm_cnshift = ffs(pmp->pm_bpcluster) - 1;
MSDOSFS_DPRINTF(("%s(fatmask=%lu, fatmult=%u, fatdiv=%u, "
"fatblocksize=%lu, fatblocksec=%lu, bnshift=%lu, pbcluster=%lu, "
"crbomask=%lu, cnshift=%lu)\n",
__func__, (unsigned long)pmp->pm_fatmask, pmp->pm_fatmult,
pmp->pm_fatdiv, pmp->pm_fatblocksize, pmp->pm_fatblocksec,
pmp->pm_bnshift, pmp->pm_bpcluster, pmp->pm_crbomask,
pmp->pm_cnshift));
/*
* Check for valid cluster size
* must be a power of 2
*/
if (pmp->pm_bpcluster ^ (1 << pmp->pm_cnshift)) {
MSDOSFS_DPRINTF(("bpcluster %lu cnshift %lu\n",
pmp->pm_bpcluster, pmp->pm_cnshift));
error = EINVAL;
goto error_exit;
}
/*
* Release the bootsector buffer.
*/
brelse(bp);
bp = NULL;
/*
* Check FSInfo.
*/
if (pmp->pm_fsinfo) {
struct fsinfo *fp;
/*
* XXX If the fsinfo block is stored on media with
* 2KB or larger sectors, is the fsinfo structure
* padded at the end or in the middle?
*/
- if ((error = bread(devvp, pmp->pm_fsinfo, pmp->pm_BytesPerSec,
- 0, &bp)) != 0)
+ if ((error = bread((void *)devvp, pmp->pm_fsinfo,
+ pmp->pm_BytesPerSec, 0, &bp)) != 0)
goto error_exit;
fp = (struct fsinfo *)bp->b_data;
if (!memcmp(fp->fsisig1, "RRaA", 4)
&& !memcmp(fp->fsisig2, "rrAa", 4)
&& !memcmp(fp->fsisig3, "\0\0\125\252", 4))
pmp->pm_nxtfree = getulong(fp->fsinxtfree);
else
pmp->pm_fsinfo = 0;
brelse(bp);
bp = NULL;
}
/*
* Check and validate (or perhaps invalidate?) the fsinfo structure?
* XXX
*/
if (pmp->pm_fsinfo) {
if ((pmp->pm_nxtfree == 0xffffffffUL) ||
(pmp->pm_nxtfree > pmp->pm_maxcluster))
pmp->pm_fsinfo = 0;
}
/*
* Allocate memory for the bitmap of allocated clusters, and then
* fill it in.
*/
pmp->pm_inusemap = ecalloc(sizeof(*pmp->pm_inusemap),
((pmp->pm_maxcluster + N_INUSEBITS) / N_INUSEBITS));
/*
* fillinusemap() needs pm_devvp.
*/
pmp->pm_dev = 0;
- pmp->pm_devvp = devvp;
+ pmp->pm_devvp = (void *)devvp;
/*
* Have the inuse map filled in.
*/
if ((error = fillinusemap(pmp)) != 0) {
MSDOSFS_DPRINTF(("fillinusemap %d\n", error));
goto error_exit;
}
/*
* Finish up.
*/
if (ronly)
pmp->pm_flags |= MSDOSFSMNT_RONLY;
else
pmp->pm_fmod = 1;
/*
* If we ever do quotas for DOS filesystems this would be a place
* to fill in the info in the msdosfsmount structure. You dolt,
* quotas on dos filesystems make no sense because files have no
* owners on dos filesystems. of course there is some empty space
* in the directory entry where we could put uid's and gid's.
*/
return pmp;
error_exit:
if (bp)
brelse(bp);
if (pmp) {
if (pmp->pm_inusemap)
free(pmp->pm_inusemap);
free(pmp);
}
errno = error;
return NULL;
}
int
-msdosfs_root(struct msdosfsmount *pmp, struct vnode *vp) {
+msdosfs_root(struct msdosfsmount *pmp, struct m_vnode *vp) {
struct denode *ndep;
int error;
- *vp = *pmp->pm_devvp;
+ *vp = *(struct m_vnode *)pmp->pm_devvp;
if ((error = deget(pmp, MSDOSFSROOT, MSDOSFSROOT_OFS, &ndep)) != 0) {
errno = error;
return -1;
}
vp->v_data = ndep;
return 0;
}
/*
* If we have an FSInfo block, update it.
*/
int
msdosfs_fsiflush(struct msdosfsmount *pmp)
{
struct fsinfo *fp;
- struct buf *bp;
+ struct m_buf *bp;
int error;
if (pmp->pm_fsinfo == 0 || (pmp->pm_flags & MSDOSFS_FSIMOD) == 0) {
error = 0;
goto out;
}
- error = bread(pmp->pm_devvp, pmp->pm_fsinfo, pmp->pm_BytesPerSec,
- NOCRED, &bp);
+ error = bread((void *)pmp->pm_devvp, pmp->pm_fsinfo,
+ pmp->pm_BytesPerSec, NOCRED, &bp);
if (error != 0) {
brelse(bp);
goto out;
}
fp = (struct fsinfo *)bp->b_data;
putulong(fp->fsinfree, pmp->pm_freeclustercount);
putulong(fp->fsinxtfree, pmp->pm_nxtfree);
pmp->pm_flags &= ~MSDOSFS_FSIMOD;
error = bwrite(bp);
out:
return (error);
}
diff --git a/usr.sbin/makefs/msdos/msdosfs_vnops.c b/usr.sbin/makefs/msdos/msdosfs_vnops.c
index 8bd4d3d498d8..2233c4b35eb2 100644
--- a/usr.sbin/makefs/msdos/msdosfs_vnops.c
+++ b/usr.sbin/makefs/msdos/msdosfs_vnops.c
@@ -1,644 +1,646 @@
/* $NetBSD: msdosfs_vnops.c,v 1.19 2017/04/13 17:10:12 christos Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1994, 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*-
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
* This software is provided "as is".
*
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
* October 1992
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/errno.h>
#include <sys/mman.h>
#include <sys/time.h>
#include <fcntl.h>
#include <stdbool.h>
#include <stdio.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
-#include "ffs/buf.h"
#include <fs/msdosfs/bpb.h>
#include "msdos/direntry.h"
#include <fs/msdosfs/denode.h>
#include <fs/msdosfs/fat.h>
#include <fs/msdosfs/msdosfsmount.h>
+#undef clrbuf
+#include "ffs/buf.h"
#include "makefs.h"
#include "msdos.h"
/*
* Some general notes:
*
* In the ufs filesystem the inodes, superblocks, and indirect blocks are
* read/written using the vnode for the filesystem. Blocks that represent
* the contents of a file are read/written using the vnode for the file
* (including directories when they are read/written as files). This
* presents problems for the dos filesystem because data that should be in
* an inode (if dos had them) resides in the directory itself. Since we
* must update directory entries without the benefit of having the vnode
* for the directory we must use the vnode for the filesystem. This means
* that when a directory is actually read/written (via read, write, or
* readdir, or seek) we must use the vnode for the filesystem instead of
* the vnode for the directory as would happen in ufs. This is to insure we
* retrieve the correct block from the buffer cache since the hash value is
* based upon the vnode address and the desired block number.
*/
static int msdosfs_wfile(const char *, struct denode *, fsnode *);
static void unix2fattime(const struct timespec *tsp, uint16_t *ddp,
uint16_t *dtp);
static void
msdosfs_times(struct denode *dep, const struct stat *st)
{
if (stampst.st_ino)
st = &stampst;
#ifdef HAVE_STRUCT_STAT_BIRTHTIME
unix2fattime(&st->st_birthtim, &dep->de_CDate, &dep->de_CTime);
#else
unix2fattime(&st->st_ctim, &dep->de_CDate, &dep->de_CTime);
#endif
unix2fattime(&st->st_atim, &dep->de_ADate, NULL);
unix2fattime(&st->st_mtim, &dep->de_MDate, &dep->de_MTime);
}
static void
unix2fattime(const struct timespec *tsp, uint16_t *ddp, uint16_t *dtp)
{
time_t t1;
struct tm lt = {0};
t1 = tsp->tv_sec;
localtime_r(&t1, <);
unsigned long fat_time = ((lt.tm_year - 80) << 25) |
((lt.tm_mon + 1) << 21) |
(lt.tm_mday << 16) |
(lt.tm_hour << 11) |
(lt.tm_min << 5) |
(lt.tm_sec >> 1);
if (ddp != NULL)
*ddp = (uint16_t)(fat_time >> 16);
if (dtp != NULL)
*dtp = (uint16_t)fat_time;
}
/*
* When we search a directory the blocks containing directory entries are
* read and examined. The directory entries contain information that would
* normally be in the inode of a unix filesystem. This means that some of
* a directory's contents may also be in memory resident denodes (sort of
* an inode). This can cause problems if we are searching while some other
* process is modifying a directory. To prevent one process from accessing
* incompletely modified directory information we depend upon being the
* sole owner of a directory block. bread/brelse provide this service.
* This being the case, when a process modifies a directory it must first
* acquire the disk block that contains the directory entry to be modified.
* Then update the disk block and the denode, and then write the disk block
* out to disk. This way disk blocks containing directory entries and in
* memory denode's will be in synch.
*/
static int
msdosfs_findslot(struct denode *dp, struct componentname *cnp)
{
daddr_t bn;
int error;
int slotcount;
int slotoffset = 0;
int frcn;
u_long cluster;
int blkoff;
u_int diroff;
int blsize;
struct msdosfsmount *pmp;
- struct buf *bp = 0;
+ struct m_buf *bp = 0;
struct direntry *dep;
u_char dosfilename[12];
int wincnt = 1;
int chksum = -1, chksum_ok;
int olddos = 1;
pmp = dp->de_pmp;
switch (unix2dosfn((const u_char *)cnp->cn_nameptr, dosfilename,
cnp->cn_namelen, 0)) {
case 0:
return (EINVAL);
case 1:
break;
case 2:
wincnt = winSlotCnt((const u_char *)cnp->cn_nameptr,
cnp->cn_namelen) + 1;
break;
case 3:
olddos = 0;
wincnt = winSlotCnt((const u_char *)cnp->cn_nameptr,
cnp->cn_namelen) + 1;
break;
}
if (pmp->pm_flags & MSDOSFSMNT_SHORTNAME)
wincnt = 1;
/*
* Suppress search for slots unless creating
* file and at end of pathname, in which case
* we watch for a place to put the new file in
* case it doesn't already exist.
*/
slotcount = 0;
MSDOSFS_DPRINTF(("%s(): dos filename: %s\n", __func__, dosfilename));
/*
* Search the directory pointed at by vdp for the name pointed at
* by cnp->cn_nameptr.
*/
/*
* The outer loop ranges over the clusters that make up the
* directory. Note that the root directory is different from all
* other directories. It has a fixed number of blocks that are not
* part of the pool of allocatable clusters. So, we treat it a
* little differently. The root directory starts at "cluster" 0.
*/
diroff = 0;
for (frcn = 0; diroff < dp->de_FileSize; frcn++) {
if ((error = pcbmap(dp, frcn, &bn, &cluster, &blsize)) != 0) {
if (error == E2BIG)
break;
return (error);
}
- error = bread(pmp->pm_devvp, bn, blsize, 0, &bp);
+ error = bread((void *)pmp->pm_devvp, bn, blsize, 0, &bp);
if (error) {
return (error);
}
for (blkoff = 0; blkoff < blsize;
blkoff += sizeof(struct direntry),
diroff += sizeof(struct direntry)) {
dep = (struct direntry *)(bp->b_data + blkoff);
/*
* If the slot is empty and we are still looking
* for an empty then remember this one. If the
* slot is not empty then check to see if it
* matches what we are looking for. If the slot
* has never been filled with anything, then the
* remainder of the directory has never been used,
* so there is no point in searching it.
*/
if (dep->deName[0] == SLOT_EMPTY ||
dep->deName[0] == SLOT_DELETED) {
/*
* Drop memory of previous long matches
*/
chksum = -1;
if (slotcount < wincnt) {
slotcount++;
slotoffset = diroff;
}
if (dep->deName[0] == SLOT_EMPTY) {
brelse(bp);
goto notfound;
}
} else {
/*
* If there wasn't enough space for our
* winentries, forget about the empty space
*/
if (slotcount < wincnt)
slotcount = 0;
/*
* Check for Win95 long filename entry
*/
if (dep->deAttributes == ATTR_WIN95) {
if (pmp->pm_flags & MSDOSFSMNT_SHORTNAME)
continue;
chksum = winChkName(
(const u_char *)cnp->cn_nameptr,
cnp->cn_namelen,
(struct winentry *)dep, chksum);
continue;
}
/*
* Ignore volume labels (anywhere, not just
* the root directory).
*/
if (dep->deAttributes & ATTR_VOLUME) {
chksum = -1;
continue;
}
/*
* Check for a checksum or name match
*/
chksum_ok = (chksum == winChksum(dep->deName));
if (!chksum_ok
&& (!olddos || memcmp(dosfilename, dep->deName, 11))) {
chksum = -1;
continue;
}
MSDOSFS_DPRINTF(("%s(): match blkoff %d, diroff %u\n",
__func__, blkoff, diroff));
/*
* Remember where this directory
* entry came from for whoever did
* this lookup.
*/
dp->de_fndoffset = diroff;
dp->de_fndcnt = 0;
return EEXIST;
}
} /* for (blkoff = 0; .... */
/*
* Release the buffer holding the directory cluster just
* searched.
*/
brelse(bp);
} /* for (frcn = 0; ; frcn++) */
notfound:
/*
* We hold no disk buffers at this point.
*/
/*
* If we get here we didn't find the entry we were looking for. But
* that's ok if we are creating or renaming and are at the end of
* the pathname and the directory hasn't been removed.
*/
MSDOSFS_DPRINTF(("%s(): refcnt %ld, slotcount %d, slotoffset %d\n",
__func__, dp->de_refcnt, slotcount, slotoffset));
/*
* Fixup the slot description to point to the place where
* we might put the new DOS direntry (putting the Win95
* long name entries before that)
*/
if (!slotcount) {
slotcount = 1;
slotoffset = diroff;
}
if (wincnt > slotcount) {
slotoffset += sizeof(struct direntry) * (wincnt - slotcount);
}
/*
* Return an indication of where the new directory
* entry should be put.
*/
dp->de_fndoffset = slotoffset;
dp->de_fndcnt = wincnt - 1;
/*
* We return with the directory locked, so that
* the parameters we set up above will still be
* valid if we actually decide to do a direnter().
* We return ni_vp == NULL to indicate that the entry
* does not currently exist; we leave a pointer to
* the (locked) directory inode in ndp->ni_dvp.
*
* NB - if the directory is unlocked, then this
* information cannot be used.
*/
return 0;
}
/*
* Create a regular file. On entry the directory to contain the file being
* created is locked. We must release before we return.
*/
struct denode *
msdosfs_mkfile(const char *path, struct denode *pdep, fsnode *node)
{
struct componentname cn;
struct denode ndirent;
struct denode *dep;
int error;
struct stat *st = &node->inode->st;
cn.cn_nameptr = node->name;
cn.cn_namelen = strlen(node->name);
MSDOSFS_DPRINTF(("%s(name %s, mode 0%o size %zu)\n",
__func__, node->name, st->st_mode, (size_t)st->st_size));
/*
* If this is the root directory and there is no space left we
* can't do anything. This is because the root directory can not
* change size.
*/
if (pdep->de_StartCluster == MSDOSFSROOT
&& pdep->de_fndoffset >= pdep->de_FileSize) {
error = ENOSPC;
goto bad;
}
/*
* Create a directory entry for the file, then call createde() to
* have it installed. NOTE: DOS files are always executable. We
* use the absence of the owner write bit to make the file
* readonly.
*/
memset(&ndirent, 0, sizeof(ndirent));
if ((error = uniqdosname(pdep, &cn, ndirent.de_Name)) != 0)
goto bad;
ndirent.de_Attributes = (st->st_mode & S_IWUSR) ?
ATTR_ARCHIVE : ATTR_ARCHIVE | ATTR_READONLY;
ndirent.de_StartCluster = 0;
ndirent.de_FileSize = 0;
ndirent.de_pmp = pdep->de_pmp;
ndirent.de_flag = DE_ACCESS | DE_CREATE | DE_UPDATE;
msdosfs_times(&ndirent, &node->inode->st);
if ((error = msdosfs_findslot(pdep, &cn)) != 0)
goto bad;
if ((error = createde(&ndirent, pdep, &dep, &cn)) != 0)
goto bad;
if ((error = msdosfs_wfile(path, dep, node)) != 0)
goto bad;
return dep;
bad:
errno = error;
return NULL;
}
static int
msdosfs_updatede(struct denode *dep)
{
- struct buf *bp;
+ struct m_buf *bp;
struct direntry *dirp;
int error;
dep->de_flag &= ~DE_MODIFIED;
- error = readde(dep, &bp, &dirp);
+ error = m_readde(dep, &bp, &dirp);
if (error)
return error;
DE_EXTERNALIZE(dirp, dep);
error = bwrite(bp);
return error;
}
/*
* Write data to a file or directory.
*/
static int
msdosfs_wfile(const char *path, struct denode *dep, fsnode *node)
{
int error, fd;
size_t osize = dep->de_FileSize;
struct stat *st = &node->inode->st;
size_t nsize, offs;
struct msdosfsmount *pmp = dep->de_pmp;
- struct buf *bp;
+ struct m_buf *bp;
char *dat;
u_long cn = 0;
error = 0; /* XXX: gcc/vax */
MSDOSFS_DPRINTF(("%s(diroff %lu, dirclust %lu, startcluster %lu)\n",
__func__, dep->de_diroffset, dep->de_dirclust,
dep->de_StartCluster));
if (st->st_size == 0)
return 0;
/* Don't bother to try to write files larger than the fs limit */
if (st->st_size > MSDOSFS_FILESIZE_MAX)
return EFBIG;
nsize = st->st_size;
MSDOSFS_DPRINTF(("%s(nsize=%zu, osize=%zu)\n", __func__, nsize, osize));
if (nsize > osize) {
if ((error = deextend(dep, nsize, NULL)) != 0)
return error;
if ((error = msdosfs_updatede(dep)) != 0)
return error;
}
if ((fd = open(path, O_RDONLY)) == -1) {
error = errno;
fprintf(stderr, "open %s: %s\n", path, strerror(error));
return error;
}
if ((dat = mmap(0, nsize, PROT_READ, MAP_FILE | MAP_PRIVATE, fd, 0))
== MAP_FAILED) {
error = errno;
fprintf(stderr, "%s: mmap %s: %s\n", __func__, node->name,
strerror(error));
close(fd);
goto out;
}
close(fd);
for (offs = 0; offs < nsize;) {
int blsize, cpsize;
daddr_t bn;
u_long on = offs & pmp->pm_crbomask;
if ((error = pcbmap(dep, cn++, &bn, NULL, &blsize)) != 0) {
MSDOSFS_DPRINTF(("%s: pcbmap %lu",
__func__, (unsigned long)bn));
goto out;
}
MSDOSFS_DPRINTF(("%s(cn=%lu, bn=%llu, blsize=%d)\n",
__func__, cn, (unsigned long long)bn, blsize));
- if ((error = bread(pmp->pm_devvp, bn, blsize, 0, &bp)) != 0) {
+ if ((error = bread((void *)pmp->pm_devvp, bn, blsize, 0,
+ &bp)) != 0) {
MSDOSFS_DPRINTF(("bread %d\n", error));
goto out;
}
cpsize = MIN((nsize - offs), blsize - on);
memcpy(bp->b_data + on, dat + offs, cpsize);
bwrite(bp);
offs += cpsize;
}
munmap(dat, nsize);
return 0;
out:
munmap(dat, nsize);
return error;
}
static const struct {
struct direntry dot;
struct direntry dotdot;
} dosdirtemplate = {
{ ". ", /* the . entry */
ATTR_DIRECTORY, /* file attribute */
0, /* reserved */
0, { 0, 0 }, { 0, 0 }, /* create time & date */
{ 0, 0 }, /* access date */
{ 0, 0 }, /* high bits of start cluster */
{ 210, 4 }, { 210, 4 }, /* modify time & date */
{ 0, 0 }, /* startcluster */
{ 0, 0, 0, 0 } /* filesize */
},
{ ".. ", /* the .. entry */
ATTR_DIRECTORY, /* file attribute */
0, /* reserved */
0, { 0, 0 }, { 0, 0 }, /* create time & date */
{ 0, 0 }, /* access date */
{ 0, 0 }, /* high bits of start cluster */
{ 210, 4 }, { 210, 4 }, /* modify time & date */
{ 0, 0 }, /* startcluster */
{ 0, 0, 0, 0 } /* filesize */
}
};
struct denode *
msdosfs_mkdire(const char *path, struct denode *pdep, fsnode *node) {
struct denode ndirent;
struct denode *dep;
struct componentname cn;
struct msdosfsmount *pmp = pdep->de_pmp;
int error;
u_long newcluster, pcl, bn;
struct direntry *denp;
- struct buf *bp;
+ struct m_buf *bp;
cn.cn_nameptr = node->name;
cn.cn_namelen = strlen(node->name);
/*
* If this is the root directory and there is no space left we
* can't do anything. This is because the root directory can not
* change size.
*/
if (pdep->de_StartCluster == MSDOSFSROOT
&& pdep->de_fndoffset >= pdep->de_FileSize) {
error = ENOSPC;
goto bad2;
}
/*
* Allocate a cluster to hold the about to be created directory.
*/
error = clusteralloc(pmp, 0, 1, CLUST_EOFE, &newcluster, NULL);
if (error)
goto bad2;
memset(&ndirent, 0, sizeof(ndirent));
ndirent.de_pmp = pmp;
ndirent.de_flag = DE_ACCESS | DE_CREATE | DE_UPDATE;
msdosfs_times(&ndirent, &node->inode->st);
/*
* Now fill the cluster with the "." and ".." entries. And write
* the cluster to disk. This way it is there for the parent
* directory to be pointing at if there were a crash.
*/
bn = cntobn(pmp, newcluster);
MSDOSFS_DPRINTF(("%s(newcluster %lu, bn=%lu)\n",
__func__, newcluster, bn));
/* always succeeds */
- bp = getblk(pmp->pm_devvp, bn, pmp->pm_bpcluster, 0, 0, 0);
+ bp = getblk((void *)pmp->pm_devvp, bn, pmp->pm_bpcluster, 0, 0, 0);
memset(bp->b_data, 0, pmp->pm_bpcluster);
memcpy(bp->b_data, &dosdirtemplate, sizeof dosdirtemplate);
denp = (struct direntry *)bp->b_data;
putushort(denp[0].deStartCluster, newcluster);
putushort(denp[0].deCDate, ndirent.de_CDate);
putushort(denp[0].deCTime, ndirent.de_CTime);
denp[0].deCHundredth = ndirent.de_CHun;
putushort(denp[0].deADate, ndirent.de_ADate);
putushort(denp[0].deMDate, ndirent.de_MDate);
putushort(denp[0].deMTime, ndirent.de_MTime);
pcl = pdep->de_StartCluster;
MSDOSFS_DPRINTF(("%s(pcl %lu, rootdirblk=%lu)\n", __func__, pcl,
pmp->pm_rootdirblk));
if (FAT32(pmp) && pcl == pmp->pm_rootdirblk)
pcl = 0;
putushort(denp[1].deStartCluster, pcl);
putushort(denp[1].deCDate, ndirent.de_CDate);
putushort(denp[1].deCTime, ndirent.de_CTime);
denp[1].deCHundredth = ndirent.de_CHun;
putushort(denp[1].deADate, ndirent.de_ADate);
putushort(denp[1].deMDate, ndirent.de_MDate);
putushort(denp[1].deMTime, ndirent.de_MTime);
if (FAT32(pmp)) {
putushort(denp[0].deHighClust, newcluster >> 16);
putushort(denp[1].deHighClust, pdep->de_StartCluster >> 16);
} else {
putushort(denp[0].deHighClust, 0);
putushort(denp[1].deHighClust, 0);
}
if ((error = bwrite(bp)) != 0)
goto bad;
/*
* Now build up a directory entry pointing to the newly allocated
* cluster. This will be written to an empty slot in the parent
* directory.
*/
if ((error = uniqdosname(pdep, &cn, ndirent.de_Name)) != 0)
goto bad;
ndirent.de_Attributes = ATTR_DIRECTORY;
ndirent.de_StartCluster = newcluster;
ndirent.de_FileSize = 0;
ndirent.de_pmp = pdep->de_pmp;
if ((error = msdosfs_findslot(pdep, &cn)) != 0)
goto bad;
if ((error = createde(&ndirent, pdep, &dep, &cn)) != 0)
goto bad;
if ((error = msdosfs_updatede(dep)) != 0)
goto bad;
return dep;
bad:
clusterfree(pmp, newcluster, NULL);
bad2:
errno = error;
return NULL;
}