Page MenuHomeFreeBSD
Files F169201678

No OneTemporary
Actions

View Options

This document is not UTF8. It was detected as ISO-8859-1 (Latin 1) and converted to UTF8 for display.
Index: head/sys/compat/pecoff/imgact_pecoff.c
===================================================================
--- head/sys/compat/pecoff/imgact_pecoff.c (revision 130100)
+++ head/sys/compat/pecoff/imgact_pecoff.c (revision 130101)
@@ -1,677 +1,677 @@
/*
* Copyright (c) 2000 Masaru OKI
* Copyright (c) 1994, 1995, 1998 Scott Bartram
* Copyright (c) 1994 Adam Glass
* Copyright (c) 1993, 1994 Christopher G. Demetriou
*
* originally from NetBSD kern/exec_ecoff.c
*
* Copyright (c) 2000 Takanori Watanabe
* Copyright (c) 2000 KUROSAWA Takahiro
* Copyright (c) 1995-1996 Sen Schmidt
* Copyright (c) 1996 Peter Wemm
* All rights reserved.
*
* originally from FreeBSD kern/imgact_elf.c
*
* 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 Masaru OKI.
* 4. The name of the author 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 ``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/systm.h>
#include <sys/imgact.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mman.h>
#include <sys/mutex.h>
#include <sys/namei.h>
#include <sys/proc.h>
#include <sys/signalvar.h>
#include <sys/sysent.h>
#include <sys/vnode.h>
#include <machine/reg.h>
#include <vm/vm.h>
#include <vm/vm_kern.h>
#include <vm/vm_param.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <vm/vm_extern.h>
#include <sys/user.h>
#include <sys/exec.h>
#include <sys/kernel.h>
#include <sys/module.h>
#include <machine/cpu.h>
#include <sys/syscall.h>
#include <sys/sysent.h>
#include <machine/md_var.h>
#include <machine/pecoff_machdep.h>
#include <compat/pecoff/imgact_pecoff.h>
#include "opt_pecoff.h"
#define PECOFF_PE_SIGNATURE "PE\0\0"
static int pecoff_fixup(register_t **, struct image_params *);
static int
pecoff_coredump(register struct thread *, register struct vnode *,
off_t);
#ifndef PECOFF_DEBUG
#define DPRINTF(a)
#else
#define DPRINTF(a) printf a
#endif
static struct sysentvec pecoff_sysvec = {
SYS_MAXSYSCALL,
sysent,
0,
0,
NULL,
0,
NULL,
NULL,
pecoff_fixup,
sendsig,
sigcode,
&szsigcode,
0,
"FreeBSD PECoff",
pecoff_coredump,
NULL,
MINSIGSTKSZ,
PAGE_SIZE,
VM_MIN_ADDRESS,
VM_MAXUSER_ADDRESS,
USRSTACK,
PS_STRINGS,
VM_PROT_ALL,
exec_copyout_strings,
exec_setregs,
NULL
};
static const char signature[] = PECOFF_PE_SIGNATURE;
static int
exec_pecoff_coff_prep_omagic(struct image_params *,
struct coff_filehdr *,
struct coff_aouthdr *, int peoffs);
static int
exec_pecoff_coff_prep_nmagic(struct image_params *,
struct coff_filehdr *,
struct coff_aouthdr *, int peoffs);
static int
exec_pecoff_coff_prep_zmagic(struct image_params *,
struct coff_filehdr *,
struct coff_aouthdr *, int peoffs);
static int
exec_pecoff_coff_makecmds(struct image_params *,
struct coff_filehdr *, int);
static int pecoff_signature(struct thread *, struct vnode *, const struct pecoff_dos_filehdr *);
static int pecoff_read_from(struct thread *, struct vnode *, int, caddr_t, int);
static int
pecoff_load_section(struct thread * td,
struct vmspace * vmspace, struct vnode * vp,
vm_offset_t offset, caddr_t vmaddr, size_t memsz, size_t filsz,
vm_prot_t prot);
static int
pecoff_fixup(register_t ** stack_base, struct image_params * imgp)
{
int len = sizeof(struct pecoff_args);
struct pecoff_imghdr *ap;
register_t *pos;
pos = *stack_base + (imgp->argc + imgp->envc + 2);
ap = (struct pecoff_imghdr *) imgp->auxargs;
if (copyout(ap, pos, len)) {
return 0;
}
free(ap, M_TEMP);
imgp->auxargs = NULL;
(*stack_base)--;
suword(*stack_base, (long) imgp->argc);
return 0;
}
static int
pecoff_coredump(register struct thread * td, register struct vnode * vp,
off_t limit)
{
register struct ucred *cred = td->td_ucred;
struct proc *p = td->td_proc;
register struct vmspace *vm = p->p_vmspace;
char *tempuser;
int error;
#ifdef PECOFF_DEBUG
struct vm_map *map;
struct vm_map_entry *ent;
struct reg regs;
#endif
if (ctob((uarea_pages + kstack_pages) + vm->vm_dsize + vm->vm_ssize) >=
limit)
return (EFAULT);
tempuser = malloc(ctob(uarea_pages + kstack_pages), M_TEMP,
M_WAITOK | M_ZERO);
if (tempuser == NULL)
return (ENOMEM);
PROC_LOCK(p);
fill_kinfo_proc(p, &p->p_uarea->u_kproc);
PROC_UNLOCK(p);
bcopy(p->p_uarea, tempuser, sizeof(struct user));
bcopy(td->td_frame,
tempuser + ctob(uarea_pages) +
((caddr_t)td->td_frame - (caddr_t)td->td_kstack),
sizeof(struct trapframe));
#if PECOFF_DEBUG
fill_regs(td, &regs);
printf("EIP%x\n", regs.r_eip);
printf("EAX%x EBX%x ECX%x EDI%x\n",
regs.r_eax, regs.r_ebx, regs.r_ecx, regs.r_edi);
map = &vm->vm_map;
ent = &map->header;
printf("%p %p %p\n", ent, ent->prev, ent->next);
#endif
error = vn_rdwr(UIO_WRITE, vp, (caddr_t)tempuser,
ctob(uarea_pages + kstack_pages),
(off_t)0, UIO_SYSSPACE, IO_UNIT, cred, NOCRED,
(int *)NULL, td);
free(tempuser, M_TEMP);
if (error == 0)
error = vn_rdwr_inchunks(UIO_WRITE, vp, vm->vm_daddr,
(int)ctob(vm->vm_dsize),
(off_t)ctob((uarea_pages + kstack_pages)),
- UIO_USERSPACE, IO_UNIT, cred, NOCRED, (int *)NULL, td);
+ UIO_USERSPACE, IO_UNIT, cred, NOCRED, NULL, td);
if (error == 0)
error = vn_rdwr_inchunks(UIO_WRITE, vp,
(caddr_t)trunc_page(USRSTACK - ctob(vm->vm_ssize)),
round_page(ctob(vm->vm_ssize)),
(off_t)ctob((uarea_pages + kstack_pages)) +
ctob(vm->vm_dsize),
- UIO_USERSPACE, IO_UNIT, cred, NOCRED, (int *)NULL, td);
+ UIO_USERSPACE, IO_UNIT, cred, NOCRED, NULL, td);
return (error);
}
static int
pecoff_load_section(struct thread * td, struct vmspace * vmspace, struct vnode * vp, vm_offset_t offset, caddr_t vmaddr, size_t memsz, size_t filsz, vm_prot_t prot)
{
size_t map_len;
vm_offset_t map_addr;
int error, rv;
size_t copy_len;
size_t copy_map_len;
size_t copy_start;
vm_object_t object;
vm_offset_t copy_map_offset;
vm_offset_t file_addr;
vm_offset_t data_buf = 0;
object = vp->v_object;
error = 0;
map_addr = trunc_page((vm_offset_t) vmaddr);
file_addr = trunc_page(offset);
DPRINTF(("SECARG:%x %p %x %x\n", offset, vmaddr, memsz, filsz));
if (file_addr != offset) {
/*
* The section is not on page boundary. We can't use
* vm_map_insert(). Use copyin instead.
*/
map_len = round_page(memsz);
copy_len = filsz;
copy_map_offset = file_addr;
copy_map_len = round_page(offset + filsz) - file_addr;
copy_start = offset - file_addr;
DPRINTF(("offset=%x vmaddr=%lx filsz=%x memsz=%x\n",
offset, (long)vmaddr, filsz, memsz));
DPRINTF(("map_len=%x copy_len=%x copy_map_offset=%x"
" copy_map_len=%x copy_start=%x\n",
map_len, copy_len, copy_map_offset,
copy_map_len, copy_start));
} else {
map_len = trunc_page(filsz);
if (map_len != 0) {
vm_object_reference(object);
vm_map_lock(&vmspace->vm_map);
rv = vm_map_insert(&vmspace->vm_map,
object,
file_addr, /* file offset */
map_addr, /* virtual start */
map_addr + map_len, /* virtual end */
prot,
VM_PROT_ALL,
MAP_COPY_ON_WRITE | MAP_PREFAULT);
vm_map_unlock(&vmspace->vm_map);
if (rv != KERN_SUCCESS) {
vm_object_deallocate(object);
return EINVAL;
}
/* we can stop now if we've covered it all */
if (memsz == filsz)
return 0;
}
copy_map_offset = trunc_page(offset + filsz);
copy_map_len = PAGE_SIZE;
copy_start = 0;
copy_len = (offset + filsz) - trunc_page(offset + filsz);
map_addr = trunc_page((vm_offset_t) vmaddr + filsz);
map_len = round_page((vm_offset_t) vmaddr + memsz) - map_addr;
}
if (map_len != 0) {
vm_map_lock(&vmspace->vm_map);
rv = vm_map_insert(&vmspace->vm_map, NULL, 0,
map_addr, map_addr + map_len,
VM_PROT_ALL, VM_PROT_ALL, 0);
vm_map_unlock(&vmspace->vm_map);
DPRINTF(("EMP-rv:%d,%x %x\n", rv, map_addr, map_addr + map_len));
if (rv != KERN_SUCCESS) {
return EINVAL;
}
}
DPRINTF(("COPYARG %x %x\n", map_addr, copy_len));
if (copy_len != 0) {
vm_object_reference(object);
rv = vm_map_find(exec_map,
object,
copy_map_offset,
&data_buf,
copy_map_len,
TRUE,
VM_PROT_READ,
VM_PROT_ALL,
MAP_COPY_ON_WRITE | MAP_PREFAULT_PARTIAL);
if (rv != KERN_SUCCESS) {
vm_object_deallocate(object);
return EINVAL;
}
/* send the page fragment to user space */
error = copyout((caddr_t) data_buf + copy_start,
(caddr_t) map_addr, copy_len);
vm_map_remove(exec_map, data_buf, data_buf + copy_map_len);
DPRINTF(("%d\n", error));
if (error)
return (error);
}
/*
* set it to the specified protection
*/
vm_map_protect(&vmspace->vm_map, map_addr,
map_addr + map_len, prot,
FALSE);
return error;
}
static int
pecoff_load_file(struct thread * td, const char *file, u_long * addr, u_long * entry, u_long * ldexport)
{
struct nameidata nd;
struct pecoff_dos_filehdr dh;
struct coff_filehdr *fp = 0;
struct coff_aouthdr *ap;
struct pecoff_opthdr *wp;
struct coff_scnhdr *sh = 0;
struct vmspace *vmspace = td->td_proc->p_vmspace;
struct vattr attr;
struct image_params image_params, *imgp;
int peofs;
int error, i, scnsiz;
imgp = &image_params;
/*
* Initialize part of the common data
*/
imgp->proc = td->td_proc;
imgp->userspace_argv = NULL;
imgp->userspace_envv = NULL;
imgp->execlabel = NULL;
imgp->attr = &attr;
imgp->firstpage = NULL;
NDINIT(&nd, LOOKUP, LOCKLEAF | FOLLOW, UIO_SYSSPACE, file, td);
if ((error = namei(&nd)) != 0) {
nd.ni_vp = NULL;
goto fail;
}
NDFREE(&nd, NDF_ONLY_PNBUF);
imgp->vp = nd.ni_vp;
/*
* Check permissions, modes, uid, etc on the file, and "open" it.
*/
error = exec_check_permissions(imgp);
if (error) {
VOP_UNLOCK(nd.ni_vp, 0, td);
goto fail;
}
VOP_UNLOCK(nd.ni_vp, 0, td);
if (error)
goto fail;
if ((error = pecoff_read_from(td, imgp->vp, 0, (caddr_t) & dh, sizeof(dh))) != 0)
goto fail;
if ((error = pecoff_signature(td, imgp->vp, &dh) != 0))
goto fail;
fp = malloc(PECOFF_HDR_SIZE, M_TEMP, M_WAITOK);
peofs = dh.d_peofs + sizeof(signature) - 1;
if ((error = pecoff_read_from(td, imgp->vp, peofs, (caddr_t) fp, PECOFF_HDR_SIZE) != 0))
goto fail;
if (COFF_BADMAG(fp)) {
error = ENOEXEC;
goto fail;
}
ap = (void *) ((char *) fp + sizeof(struct coff_filehdr));
wp = (void *) ((char *) ap + sizeof(struct coff_aouthdr));
/* read section header */
scnsiz = sizeof(struct coff_scnhdr) * fp->f_nscns;
sh = malloc(scnsiz, M_TEMP, M_WAITOK);
if ((error = pecoff_read_from(td, imgp->vp, peofs + PECOFF_HDR_SIZE,
(caddr_t) sh, scnsiz)) != 0)
goto fail;
/*
* Read Section infomation and map sections.
*/
for (i = 0; i < fp->f_nscns; i++) {
int prot = 0;
if (sh[i].s_flags & COFF_STYP_DISCARD)
continue;
/* XXX ? */
if ((sh[i].s_flags & COFF_STYP_TEXT) &&
(sh[i].s_flags & COFF_STYP_EXEC) == 0)
continue;
if ((sh[i].s_flags & (COFF_STYP_TEXT | COFF_STYP_DATA | COFF_STYP_BSS)) == 0)
continue;
prot |= (sh[i].s_flags & COFF_STYP_READ) ? VM_PROT_READ : 0;
prot |= (sh[i].s_flags & COFF_STYP_WRITE) ? VM_PROT_WRITE : 0;
prot |= (sh[i].s_flags & COFF_STYP_EXEC) ? VM_PROT_EXECUTE : 0;
sh[i].s_vaddr += wp->w_base; /* RVA --> VA */
if ((error = pecoff_load_section(td, vmspace, imgp->vp, sh[i].s_scnptr
,(caddr_t) sh[i].s_vaddr,
sh[i].s_paddr, sh[i].s_size
,prot)) != 0)
goto fail;
}
*entry = wp->w_base + ap->a_entry;
*addr = wp->w_base;
*ldexport = wp->w_imghdr[0].i_vaddr + wp->w_base;
fail:
if (fp)
free(fp, M_TEMP);
if (sh)
free(sh, M_TEMP);
if (nd.ni_vp)
vrele(nd.ni_vp);
return error;
}
static int
exec_pecoff_coff_prep_omagic(struct image_params * imgp,
struct coff_filehdr * fp,
struct coff_aouthdr * ap, int peofs)
{
return ENOEXEC;
}
static int
exec_pecoff_coff_prep_nmagic(struct image_params * imgp,
struct coff_filehdr * fp,
struct coff_aouthdr * ap, int peofs)
{
return ENOEXEC;
}
static int
exec_pecoff_coff_prep_zmagic(struct image_params * imgp,
struct coff_filehdr * fp,
struct coff_aouthdr * ap, int peofs)
{
int scnsiz = sizeof(struct coff_scnhdr) * fp->f_nscns;
int error = ENOEXEC, i;
int prot;
u_long text_size = 0, data_size = 0, dsize;
u_long text_addr = 0, data_addr = VM_MAXUSER_ADDRESS;
u_long ldexport, ldbase;
struct pecoff_opthdr *wp;
struct coff_scnhdr *sh;
struct vmspace *vmspace;
struct pecoff_args *argp = NULL;
sh = malloc(scnsiz, M_TEMP, M_WAITOK);
wp = (void *) ((char *) ap + sizeof(struct coff_aouthdr));
error = pecoff_read_from(FIRST_THREAD_IN_PROC(imgp->proc), imgp->vp,
peofs + PECOFF_HDR_SIZE, (caddr_t) sh, scnsiz);
if ((error = exec_extract_strings(imgp)) != 0)
goto fail;
exec_new_vmspace(imgp, &pecoff_sysvec);
vmspace = imgp->proc->p_vmspace;
for (i = 0; i < fp->f_nscns; i++) {
prot = VM_PROT_WRITE; /* XXX for relocation? */
prot |= (sh[i].s_flags & COFF_STYP_READ) ? VM_PROT_READ : 0;
prot |= (sh[i].s_flags & COFF_STYP_WRITE) ? VM_PROT_WRITE : 0;
prot |= (sh[i].s_flags & COFF_STYP_EXEC) ? VM_PROT_EXECUTE : 0;
sh[i].s_vaddr += wp->w_base;
if (sh[i].s_flags & COFF_STYP_DISCARD)
continue;
if ((sh[i].s_flags & COFF_STYP_TEXT) != 0) {
error = pecoff_load_section(
FIRST_THREAD_IN_PROC(imgp->proc),
vmspace, imgp->vp, sh[i].s_scnptr,
(caddr_t) sh[i].s_vaddr, sh[i].s_paddr,
sh[i].s_size ,prot);
DPRINTF(("ERROR%d\n", error));
if (error)
goto fail;
text_addr = trunc_page(sh[i].s_vaddr);
text_size = trunc_page(sh[i].s_size + sh[i].s_vaddr - text_addr);
}
if ((sh[i].s_flags & (COFF_STYP_DATA|COFF_STYP_BSS)) != 0) {
if (pecoff_load_section(
FIRST_THREAD_IN_PROC(imgp->proc), vmspace,
imgp->vp, sh[i].s_scnptr, (caddr_t) sh[i].s_vaddr,
sh[i].s_paddr, sh[i].s_size, prot) != 0)
goto fail;
data_addr = min(trunc_page(sh[i].s_vaddr), data_addr);
dsize = round_page(sh[i].s_vaddr + sh[i].s_paddr)
- data_addr;
data_size = max(dsize, data_size);
}
}
vmspace->vm_tsize = text_size >> PAGE_SHIFT;
vmspace->vm_taddr = (caddr_t) (uintptr_t) text_addr;
vmspace->vm_dsize = data_size >> PAGE_SHIFT;
vmspace->vm_daddr = (caddr_t) (uintptr_t) data_addr;
argp = malloc(sizeof(struct pecoff_args), M_TEMP, M_WAITOK);
if (argp == NULL) {
error = ENOMEM;
goto fail;
}
argp->a_base = wp->w_base;
argp->a_entry = wp->w_base + ap->a_entry;
argp->a_end = data_addr + data_size;
argp->a_subsystem = wp->w_subvers;
error = pecoff_load_file(FIRST_THREAD_IN_PROC(imgp->proc),
"/usr/libexec/ld.so.dll", &ldbase, &imgp->entry_addr, &ldexport);
if (error)
goto fail;
argp->a_ldbase = ldbase;
argp->a_ldexport = ldexport;
memcpy(argp->a_imghdr, wp->w_imghdr, sizeof(struct pecoff_imghdr) * 16);
for (i = 0; i < 16; i++) {
argp->a_imghdr[i].i_vaddr += wp->w_base;
}
imgp->proc->p_sysent = &pecoff_sysvec;
if (error)
goto fail;
imgp->auxargs = argp;
imgp->auxarg_size = sizeof(struct pecoff_args);
imgp->interpreted = 0;
if (sh != NULL)
free(sh, M_TEMP);
return 0;
fail:
error = (error) ? error : ENOEXEC;
if (sh != NULL)
free(sh, M_TEMP);
if (argp != NULL)
free(argp, M_TEMP);
return error;
}
int
exec_pecoff_coff_makecmds(struct image_params * imgp,
struct coff_filehdr * fp, int peofs)
{
struct coff_aouthdr *ap;
int error;
if (COFF_BADMAG(fp)) {
return ENOEXEC;
}
ap = (void *) ((char *) fp + sizeof(struct coff_filehdr));
switch (ap->a_magic) {
case COFF_OMAGIC:
error = exec_pecoff_coff_prep_omagic(imgp, fp, ap, peofs);
break;
case COFF_NMAGIC:
error = exec_pecoff_coff_prep_nmagic(imgp, fp, ap, peofs);
break;
case COFF_ZMAGIC:
error = exec_pecoff_coff_prep_zmagic(imgp, fp, ap, peofs);
break;
default:
return ENOEXEC;
}
return error;
}
static int
pecoff_signature(td, vp, dp)
struct thread *td;
struct vnode *vp;
const struct pecoff_dos_filehdr *dp;
{
int error;
char buf[512];
char *pesig;
if (DOS_BADMAG(dp)) {
return ENOEXEC;
}
error = pecoff_read_from(td, vp, dp->d_peofs, buf, sizeof(buf));
if (error) {
return error;
}
pesig = buf;
if (memcmp(pesig, signature, sizeof(signature) - 1) == 0) {
return 0;
}
return EFTYPE;
}
int
pecoff_read_from(td, vp, pos, buf, siz)
struct thread *td;
struct vnode *vp;
int pos;
caddr_t buf;
int siz;
{
int error;
size_t resid;
error = vn_rdwr(UIO_READ, vp, buf, siz, pos,
UIO_SYSSPACE, IO_NODELOCKED, td->td_ucred, NOCRED,
&resid, td);
if (error)
return error;
if (resid != 0) {
return ENOEXEC;
}
return 0;
}
static int
imgact_pecoff(struct image_params * imgp)
{
const struct pecoff_dos_filehdr *dp = (const struct pecoff_dos_filehdr *)
imgp->image_header;
struct coff_filehdr *fp;
int error, peofs;
struct thread *td = curthread;
error = pecoff_signature(FIRST_THREAD_IN_PROC(imgp->proc),
imgp->vp, dp);
if (error) {
return -1;
}
VOP_UNLOCK(imgp->vp, 0, td);
peofs = dp->d_peofs + sizeof(signature) - 1;
fp = malloc(PECOFF_HDR_SIZE, M_TEMP, M_WAITOK);
error = pecoff_read_from(FIRST_THREAD_IN_PROC(imgp->proc),
imgp->vp, peofs, (caddr_t) fp, PECOFF_HDR_SIZE);
if (error)
goto fail;
error = exec_pecoff_coff_makecmds(imgp, fp, peofs);
fail:
free(fp, M_TEMP);
vn_lock(imgp->vp, LK_EXCLUSIVE | LK_RETRY, td);
return error;
}
static struct execsw pecoff_execsw = {imgact_pecoff, "FreeBSD PEcoff"};
EXEC_SET(pecoff, pecoff_execsw);
Index: head/sys/i386/linux/linux_sysvec.c
===================================================================
--- head/sys/i386/linux/linux_sysvec.c (revision 130100)
+++ head/sys/i386/linux/linux_sysvec.c (revision 130101)
@@ -1,974 +1,974 @@
/*-
* Copyright (c) 1994-1996 Søren Schmidt
* 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
* in this position and unchanged.
* 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 name of the author 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 ``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("$FreeBSD$");
/* XXX we use functions that might not exist. */
#include "opt_compat.h"
#ifndef COMPAT_43
#error "Unable to compile Linux-emulator due to missing COMPAT_43 option!"
#endif
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/exec.h>
#include <sys/imgact.h>
#include <sys/imgact_aout.h>
#include <sys/imgact_elf.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/module.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/signalvar.h>
#include <sys/syscallsubr.h>
#include <sys/sysent.h>
#include <sys/sysproto.h>
#include <sys/user.h>
#include <sys/vnode.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <vm/vm_extern.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_param.h>
#include <machine/cpu.h>
#include <machine/md_var.h>
#include <i386/linux/linux.h>
#include <i386/linux/linux_proto.h>
#include <compat/linux/linux_mib.h>
#include <compat/linux/linux_signal.h>
#include <compat/linux/linux_util.h>
MODULE_VERSION(linux, 1);
MODULE_DEPEND(linux, sysvmsg, 1, 1, 1);
MODULE_DEPEND(linux, sysvsem, 1, 1, 1);
MODULE_DEPEND(linux, sysvshm, 1, 1, 1);
MALLOC_DEFINE(M_LINUX, "linux", "Linux mode structures");
#if BYTE_ORDER == LITTLE_ENDIAN
#define SHELLMAGIC 0x2123 /* #! */
#else
#define SHELLMAGIC 0x2321
#endif
/*
* Allow the sendsig functions to use the ldebug() facility
* even though they are not syscalls themselves. Map them
* to syscall 0. This is slightly less bogus than using
* ldebug(sigreturn).
*/
#define LINUX_SYS_linux_rt_sendsig 0
#define LINUX_SYS_linux_sendsig 0
extern char linux_sigcode[];
extern int linux_szsigcode;
extern struct sysent linux_sysent[LINUX_SYS_MAXSYSCALL];
SET_DECLARE(linux_ioctl_handler_set, struct linux_ioctl_handler);
static int linux_fixup(register_t **stack_base,
struct image_params *iparams);
static int elf_linux_fixup(register_t **stack_base,
struct image_params *iparams);
static void linux_prepsyscall(struct trapframe *tf, int *args, u_int *code,
caddr_t *params);
static void linux_sendsig(sig_t catcher, int sig, sigset_t *mask,
u_long code);
static void exec_linux_setregs(struct thread *td, u_long entry,
u_long stack, u_long ps_strings);
/*
* Linux syscalls return negative errno's, we do positive and map them
*/
static int bsd_to_linux_errno[ELAST + 1] = {
-0, -1, -2, -3, -4, -5, -6, -7, -8, -9,
-10, -35, -12, -13, -14, -15, -16, -17, -18, -19,
-20, -21, -22, -23, -24, -25, -26, -27, -28, -29,
-30, -31, -32, -33, -34, -11,-115,-114, -88, -89,
-90, -91, -92, -93, -94, -95, -96, -97, -98, -99,
-100,-101,-102,-103,-104,-105,-106,-107,-108,-109,
-110,-111, -40, -36,-112,-113, -39, -11, -87,-122,
-116, -66, -6, -6, -6, -6, -6, -37, -38, -9,
-6, -6, -43, -42, -75, -6, -84
};
int bsd_to_linux_signal[LINUX_SIGTBLSZ] = {
LINUX_SIGHUP, LINUX_SIGINT, LINUX_SIGQUIT, LINUX_SIGILL,
LINUX_SIGTRAP, LINUX_SIGABRT, 0, LINUX_SIGFPE,
LINUX_SIGKILL, LINUX_SIGBUS, LINUX_SIGSEGV, LINUX_SIGSYS,
LINUX_SIGPIPE, LINUX_SIGALRM, LINUX_SIGTERM, LINUX_SIGURG,
LINUX_SIGSTOP, LINUX_SIGTSTP, LINUX_SIGCONT, LINUX_SIGCHLD,
LINUX_SIGTTIN, LINUX_SIGTTOU, LINUX_SIGIO, LINUX_SIGXCPU,
LINUX_SIGXFSZ, LINUX_SIGVTALRM, LINUX_SIGPROF, LINUX_SIGWINCH,
0, LINUX_SIGUSR1, LINUX_SIGUSR2
};
int linux_to_bsd_signal[LINUX_SIGTBLSZ] = {
SIGHUP, SIGINT, SIGQUIT, SIGILL,
SIGTRAP, SIGABRT, SIGBUS, SIGFPE,
SIGKILL, SIGUSR1, SIGSEGV, SIGUSR2,
SIGPIPE, SIGALRM, SIGTERM, SIGBUS,
SIGCHLD, SIGCONT, SIGSTOP, SIGTSTP,
SIGTTIN, SIGTTOU, SIGURG, SIGXCPU,
SIGXFSZ, SIGVTALRM, SIGPROF, SIGWINCH,
SIGIO, SIGURG, SIGSYS
};
#define LINUX_T_UNKNOWN 255
static int _bsd_to_linux_trapcode[] = {
LINUX_T_UNKNOWN, /* 0 */
6, /* 1 T_PRIVINFLT */
LINUX_T_UNKNOWN, /* 2 */
3, /* 3 T_BPTFLT */
LINUX_T_UNKNOWN, /* 4 */
LINUX_T_UNKNOWN, /* 5 */
16, /* 6 T_ARITHTRAP */
254, /* 7 T_ASTFLT */
LINUX_T_UNKNOWN, /* 8 */
13, /* 9 T_PROTFLT */
1, /* 10 T_TRCTRAP */
LINUX_T_UNKNOWN, /* 11 */
14, /* 12 T_PAGEFLT */
LINUX_T_UNKNOWN, /* 13 */
17, /* 14 T_ALIGNFLT */
LINUX_T_UNKNOWN, /* 15 */
LINUX_T_UNKNOWN, /* 16 */
LINUX_T_UNKNOWN, /* 17 */
0, /* 18 T_DIVIDE */
2, /* 19 T_NMI */
4, /* 20 T_OFLOW */
5, /* 21 T_BOUND */
7, /* 22 T_DNA */
8, /* 23 T_DOUBLEFLT */
9, /* 24 T_FPOPFLT */
10, /* 25 T_TSSFLT */
11, /* 26 T_SEGNPFLT */
12, /* 27 T_STKFLT */
18, /* 28 T_MCHK */
19, /* 29 T_XMMFLT */
15 /* 30 T_RESERVED */
};
#define bsd_to_linux_trapcode(code) \
((code)<sizeof(_bsd_to_linux_trapcode)/sizeof(*_bsd_to_linux_trapcode)? \
_bsd_to_linux_trapcode[(code)]: \
LINUX_T_UNKNOWN)
/*
* If FreeBSD & Linux have a difference of opinion about what a trap
* means, deal with it here.
*
* MPSAFE
*/
static int
translate_traps(int signal, int trap_code)
{
if (signal != SIGBUS)
return signal;
switch (trap_code) {
case T_PROTFLT:
case T_TSSFLT:
case T_DOUBLEFLT:
case T_PAGEFLT:
return SIGSEGV;
default:
return signal;
}
}
static int
linux_fixup(register_t **stack_base, struct image_params *imgp)
{
register_t *argv, *envp;
argv = *stack_base;
envp = *stack_base + (imgp->argc + 1);
(*stack_base)--;
**stack_base = (intptr_t)(void *)envp;
(*stack_base)--;
**stack_base = (intptr_t)(void *)argv;
(*stack_base)--;
**stack_base = imgp->argc;
return 0;
}
static int
elf_linux_fixup(register_t **stack_base, struct image_params *imgp)
{
Elf32_Auxargs *args;
register_t *pos;
KASSERT(curthread->td_proc == imgp->proc &&
(curthread->td_proc->p_flag & P_SA) == 0,
("unsafe elf_linux_fixup(), should be curproc"));
args = (Elf32_Auxargs *)imgp->auxargs;
pos = *stack_base + (imgp->argc + imgp->envc + 2);
if (args->trace)
AUXARGS_ENTRY(pos, AT_DEBUG, 1);
if (args->execfd != -1)
AUXARGS_ENTRY(pos, AT_EXECFD, args->execfd);
AUXARGS_ENTRY(pos, AT_PHDR, args->phdr);
AUXARGS_ENTRY(pos, AT_PHENT, args->phent);
AUXARGS_ENTRY(pos, AT_PHNUM, args->phnum);
AUXARGS_ENTRY(pos, AT_PAGESZ, args->pagesz);
AUXARGS_ENTRY(pos, AT_FLAGS, args->flags);
AUXARGS_ENTRY(pos, AT_ENTRY, args->entry);
AUXARGS_ENTRY(pos, AT_BASE, args->base);
AUXARGS_ENTRY(pos, AT_UID, imgp->proc->p_ucred->cr_ruid);
AUXARGS_ENTRY(pos, AT_EUID, imgp->proc->p_ucred->cr_svuid);
AUXARGS_ENTRY(pos, AT_GID, imgp->proc->p_ucred->cr_rgid);
AUXARGS_ENTRY(pos, AT_EGID, imgp->proc->p_ucred->cr_svgid);
AUXARGS_ENTRY(pos, AT_NULL, 0);
free(imgp->auxargs, M_TEMP);
imgp->auxargs = NULL;
(*stack_base)--;
**stack_base = (register_t)imgp->argc;
return 0;
}
extern int _ucodesel, _udatasel;
extern unsigned long linux_sznonrtsigcode;
static void
linux_rt_sendsig(sig_t catcher, int sig, sigset_t *mask, u_long code)
{
struct thread *td = curthread;
struct proc *p = td->td_proc;
struct sigacts *psp;
struct trapframe *regs;
struct l_rt_sigframe *fp, frame;
int oonstack;
PROC_LOCK_ASSERT(p, MA_OWNED);
psp = p->p_sigacts;
mtx_assert(&psp->ps_mtx, MA_OWNED);
regs = td->td_frame;
oonstack = sigonstack(regs->tf_esp);
#ifdef DEBUG
if (ldebug(rt_sendsig))
printf(ARGS(rt_sendsig, "%p, %d, %p, %lu"),
catcher, sig, (void*)mask, code);
#endif
/*
* Allocate space for the signal handler context.
*/
if ((td->td_pflags & TDP_ALTSTACK) && !oonstack &&
SIGISMEMBER(psp->ps_sigonstack, sig)) {
fp = (struct l_rt_sigframe *)(td->td_sigstk.ss_sp +
td->td_sigstk.ss_size - sizeof(struct l_rt_sigframe));
} else
fp = (struct l_rt_sigframe *)regs->tf_esp - 1;
mtx_unlock(&psp->ps_mtx);
/*
* Build the argument list for the signal handler.
*/
if (p->p_sysent->sv_sigtbl)
if (sig <= p->p_sysent->sv_sigsize)
sig = p->p_sysent->sv_sigtbl[_SIG_IDX(sig)];
bzero(&frame, sizeof(frame));
frame.sf_handler = catcher;
frame.sf_sig = sig;
frame.sf_siginfo = &fp->sf_si;
frame.sf_ucontext = &fp->sf_sc;
/* Fill in POSIX parts */
frame.sf_si.lsi_signo = sig;
frame.sf_si.lsi_code = code;
frame.sf_si.lsi_addr = (void *)regs->tf_err;
/*
* Build the signal context to be used by sigreturn.
*/
frame.sf_sc.uc_flags = 0; /* XXX ??? */
frame.sf_sc.uc_link = NULL; /* XXX ??? */
frame.sf_sc.uc_stack.ss_sp = td->td_sigstk.ss_sp;
frame.sf_sc.uc_stack.ss_size = td->td_sigstk.ss_size;
frame.sf_sc.uc_stack.ss_flags = (td->td_pflags & TDP_ALTSTACK)
? ((oonstack) ? LINUX_SS_ONSTACK : 0) : LINUX_SS_DISABLE;
PROC_UNLOCK(p);
bsd_to_linux_sigset(mask, &frame.sf_sc.uc_sigmask);
frame.sf_sc.uc_mcontext.sc_mask = frame.sf_sc.uc_sigmask.__bits[0];
frame.sf_sc.uc_mcontext.sc_gs = rgs();
frame.sf_sc.uc_mcontext.sc_fs = regs->tf_fs;
frame.sf_sc.uc_mcontext.sc_es = regs->tf_es;
frame.sf_sc.uc_mcontext.sc_ds = regs->tf_ds;
frame.sf_sc.uc_mcontext.sc_edi = regs->tf_edi;
frame.sf_sc.uc_mcontext.sc_esi = regs->tf_esi;
frame.sf_sc.uc_mcontext.sc_ebp = regs->tf_ebp;
frame.sf_sc.uc_mcontext.sc_ebx = regs->tf_ebx;
frame.sf_sc.uc_mcontext.sc_edx = regs->tf_edx;
frame.sf_sc.uc_mcontext.sc_ecx = regs->tf_ecx;
frame.sf_sc.uc_mcontext.sc_eax = regs->tf_eax;
frame.sf_sc.uc_mcontext.sc_eip = regs->tf_eip;
frame.sf_sc.uc_mcontext.sc_cs = regs->tf_cs;
frame.sf_sc.uc_mcontext.sc_eflags = regs->tf_eflags;
frame.sf_sc.uc_mcontext.sc_esp_at_signal = regs->tf_esp;
frame.sf_sc.uc_mcontext.sc_ss = regs->tf_ss;
frame.sf_sc.uc_mcontext.sc_err = regs->tf_err;
frame.sf_sc.uc_mcontext.sc_trapno = bsd_to_linux_trapcode(code);
#ifdef DEBUG
if (ldebug(rt_sendsig))
printf(LMSG("rt_sendsig flags: 0x%x, sp: %p, ss: 0x%x, mask: 0x%x"),
frame.sf_sc.uc_stack.ss_flags, td->td_sigstk.ss_sp,
td->td_sigstk.ss_size, frame.sf_sc.uc_mcontext.sc_mask);
#endif
if (copyout(&frame, fp, sizeof(frame)) != 0) {
/*
* Process has trashed its stack; give it an illegal
* instruction to halt it in its tracks.
*/
#ifdef DEBUG
if (ldebug(rt_sendsig))
printf(LMSG("rt_sendsig: bad stack %p, oonstack=%x"),
fp, oonstack);
#endif
PROC_LOCK(p);
sigexit(td, SIGILL);
}
/*
* Build context to run handler in.
*/
regs->tf_esp = (int)fp;
regs->tf_eip = PS_STRINGS - *(p->p_sysent->sv_szsigcode) +
linux_sznonrtsigcode;
regs->tf_eflags &= ~(PSL_T | PSL_VM);
regs->tf_cs = _ucodesel;
regs->tf_ds = _udatasel;
regs->tf_es = _udatasel;
regs->tf_fs = _udatasel;
regs->tf_ss = _udatasel;
PROC_LOCK(p);
mtx_lock(&psp->ps_mtx);
}
/*
* Send an interrupt to process.
*
* Stack is set up to allow sigcode stored
* in u. to call routine, followed by kcall
* to sigreturn routine below. After sigreturn
* resets the signal mask, the stack, and the
* frame pointer, it returns to the user
* specified pc, psl.
*/
static void
linux_sendsig(sig_t catcher, int sig, sigset_t *mask, u_long code)
{
struct thread *td = curthread;
struct proc *p = td->td_proc;
struct sigacts *psp;
struct trapframe *regs;
struct l_sigframe *fp, frame;
l_sigset_t lmask;
int oonstack, i;
PROC_LOCK_ASSERT(p, MA_OWNED);
psp = p->p_sigacts;
mtx_assert(&psp->ps_mtx, MA_OWNED);
if (SIGISMEMBER(psp->ps_siginfo, sig)) {
/* Signal handler installed with SA_SIGINFO. */
linux_rt_sendsig(catcher, sig, mask, code);
return;
}
regs = td->td_frame;
oonstack = sigonstack(regs->tf_esp);
#ifdef DEBUG
if (ldebug(sendsig))
printf(ARGS(sendsig, "%p, %d, %p, %lu"),
catcher, sig, (void*)mask, code);
#endif
/*
* Allocate space for the signal handler context.
*/
if ((td->td_pflags & TDP_ALTSTACK) && !oonstack &&
SIGISMEMBER(psp->ps_sigonstack, sig)) {
fp = (struct l_sigframe *)(td->td_sigstk.ss_sp +
td->td_sigstk.ss_size - sizeof(struct l_sigframe));
} else
fp = (struct l_sigframe *)regs->tf_esp - 1;
mtx_unlock(&psp->ps_mtx);
PROC_UNLOCK(p);
/*
* Build the argument list for the signal handler.
*/
if (p->p_sysent->sv_sigtbl)
if (sig <= p->p_sysent->sv_sigsize)
sig = p->p_sysent->sv_sigtbl[_SIG_IDX(sig)];
bzero(&frame, sizeof(frame));
frame.sf_handler = catcher;
frame.sf_sig = sig;
bsd_to_linux_sigset(mask, &lmask);
/*
* Build the signal context to be used by sigreturn.
*/
frame.sf_sc.sc_mask = lmask.__bits[0];
frame.sf_sc.sc_gs = rgs();
frame.sf_sc.sc_fs = regs->tf_fs;
frame.sf_sc.sc_es = regs->tf_es;
frame.sf_sc.sc_ds = regs->tf_ds;
frame.sf_sc.sc_edi = regs->tf_edi;
frame.sf_sc.sc_esi = regs->tf_esi;
frame.sf_sc.sc_ebp = regs->tf_ebp;
frame.sf_sc.sc_ebx = regs->tf_ebx;
frame.sf_sc.sc_edx = regs->tf_edx;
frame.sf_sc.sc_ecx = regs->tf_ecx;
frame.sf_sc.sc_eax = regs->tf_eax;
frame.sf_sc.sc_eip = regs->tf_eip;
frame.sf_sc.sc_cs = regs->tf_cs;
frame.sf_sc.sc_eflags = regs->tf_eflags;
frame.sf_sc.sc_esp_at_signal = regs->tf_esp;
frame.sf_sc.sc_ss = regs->tf_ss;
frame.sf_sc.sc_err = regs->tf_err;
frame.sf_sc.sc_trapno = bsd_to_linux_trapcode(code);
for (i = 0; i < (LINUX_NSIG_WORDS-1); i++)
frame.sf_extramask[i] = lmask.__bits[i+1];
if (copyout(&frame, fp, sizeof(frame)) != 0) {
/*
* Process has trashed its stack; give it an illegal
* instruction to halt it in its tracks.
*/
PROC_LOCK(p);
sigexit(td, SIGILL);
}
/*
* Build context to run handler in.
*/
regs->tf_esp = (int)fp;
regs->tf_eip = PS_STRINGS - *(p->p_sysent->sv_szsigcode);
regs->tf_eflags &= ~(PSL_T | PSL_VM);
regs->tf_cs = _ucodesel;
regs->tf_ds = _udatasel;
regs->tf_es = _udatasel;
regs->tf_fs = _udatasel;
regs->tf_ss = _udatasel;
PROC_LOCK(p);
mtx_lock(&psp->ps_mtx);
}
/*
* System call to cleanup state after a signal
* has been taken. Reset signal mask and
* stack state from context left by sendsig (above).
* Return to previous pc and psl as specified by
* context left by sendsig. Check carefully to
* make sure that the user has not modified the
* psl to gain improper privileges or to cause
* a machine fault.
*/
int
linux_sigreturn(struct thread *td, struct linux_sigreturn_args *args)
{
struct proc *p = td->td_proc;
struct l_sigframe frame;
struct trapframe *regs;
l_sigset_t lmask;
int eflags, i;
regs = td->td_frame;
#ifdef DEBUG
if (ldebug(sigreturn))
printf(ARGS(sigreturn, "%p"), (void *)args->sfp);
#endif
/*
* The trampoline code hands us the sigframe.
* It is unsafe to keep track of it ourselves, in the event that a
* program jumps out of a signal handler.
*/
if (copyin(args->sfp, &frame, sizeof(frame)) != 0)
return (EFAULT);
/*
* Check for security violations.
*/
#define EFLAGS_SECURE(ef, oef) ((((ef) ^ (oef)) & ~PSL_USERCHANGE) == 0)
eflags = frame.sf_sc.sc_eflags;
/*
* XXX do allow users to change the privileged flag PSL_RF. The
* cpu sets PSL_RF in tf_eflags for faults. Debuggers should
* sometimes set it there too. tf_eflags is kept in the signal
* context during signal handling and there is no other place
* to remember it, so the PSL_RF bit may be corrupted by the
* signal handler without us knowing. Corruption of the PSL_RF
* bit at worst causes one more or one less debugger trap, so
* allowing it is fairly harmless.
*/
if (!EFLAGS_SECURE(eflags & ~PSL_RF, regs->tf_eflags & ~PSL_RF))
return(EINVAL);
/*
* Don't allow users to load a valid privileged %cs. Let the
* hardware check for invalid selectors, excess privilege in
* other selectors, invalid %eip's and invalid %esp's.
*/
#define CS_SECURE(cs) (ISPL(cs) == SEL_UPL)
if (!CS_SECURE(frame.sf_sc.sc_cs)) {
trapsignal(td, SIGBUS, T_PROTFLT);
return(EINVAL);
}
lmask.__bits[0] = frame.sf_sc.sc_mask;
for (i = 0; i < (LINUX_NSIG_WORDS-1); i++)
lmask.__bits[i+1] = frame.sf_extramask[i];
PROC_LOCK(p);
linux_to_bsd_sigset(&lmask, &td->td_sigmask);
SIG_CANTMASK(td->td_sigmask);
signotify(td);
PROC_UNLOCK(p);
/*
* Restore signal context.
*/
/* %gs was restored by the trampoline. */
regs->tf_fs = frame.sf_sc.sc_fs;
regs->tf_es = frame.sf_sc.sc_es;
regs->tf_ds = frame.sf_sc.sc_ds;
regs->tf_edi = frame.sf_sc.sc_edi;
regs->tf_esi = frame.sf_sc.sc_esi;
regs->tf_ebp = frame.sf_sc.sc_ebp;
regs->tf_ebx = frame.sf_sc.sc_ebx;
regs->tf_edx = frame.sf_sc.sc_edx;
regs->tf_ecx = frame.sf_sc.sc_ecx;
regs->tf_eax = frame.sf_sc.sc_eax;
regs->tf_eip = frame.sf_sc.sc_eip;
regs->tf_cs = frame.sf_sc.sc_cs;
regs->tf_eflags = eflags;
regs->tf_esp = frame.sf_sc.sc_esp_at_signal;
regs->tf_ss = frame.sf_sc.sc_ss;
return (EJUSTRETURN);
}
/*
* System call to cleanup state after a signal
* has been taken. Reset signal mask and
* stack state from context left by rt_sendsig (above).
* Return to previous pc and psl as specified by
* context left by sendsig. Check carefully to
* make sure that the user has not modified the
* psl to gain improper privileges or to cause
* a machine fault.
*/
int
linux_rt_sigreturn(struct thread *td, struct linux_rt_sigreturn_args *args)
{
struct proc *p = td->td_proc;
struct l_ucontext uc;
struct l_sigcontext *context;
l_stack_t *lss;
stack_t ss;
struct trapframe *regs;
int eflags;
regs = td->td_frame;
#ifdef DEBUG
if (ldebug(rt_sigreturn))
printf(ARGS(rt_sigreturn, "%p"), (void *)args->ucp);
#endif
/*
* The trampoline code hands us the ucontext.
* It is unsafe to keep track of it ourselves, in the event that a
* program jumps out of a signal handler.
*/
if (copyin(args->ucp, &uc, sizeof(uc)) != 0)
return (EFAULT);
context = &uc.uc_mcontext;
/*
* Check for security violations.
*/
#define EFLAGS_SECURE(ef, oef) ((((ef) ^ (oef)) & ~PSL_USERCHANGE) == 0)
eflags = context->sc_eflags;
/*
* XXX do allow users to change the privileged flag PSL_RF. The
* cpu sets PSL_RF in tf_eflags for faults. Debuggers should
* sometimes set it there too. tf_eflags is kept in the signal
* context during signal handling and there is no other place
* to remember it, so the PSL_RF bit may be corrupted by the
* signal handler without us knowing. Corruption of the PSL_RF
* bit at worst causes one more or one less debugger trap, so
* allowing it is fairly harmless.
*/
if (!EFLAGS_SECURE(eflags & ~PSL_RF, regs->tf_eflags & ~PSL_RF))
return(EINVAL);
/*
* Don't allow users to load a valid privileged %cs. Let the
* hardware check for invalid selectors, excess privilege in
* other selectors, invalid %eip's and invalid %esp's.
*/
#define CS_SECURE(cs) (ISPL(cs) == SEL_UPL)
if (!CS_SECURE(context->sc_cs)) {
trapsignal(td, SIGBUS, T_PROTFLT);
return(EINVAL);
}
PROC_LOCK(p);
linux_to_bsd_sigset(&uc.uc_sigmask, &td->td_sigmask);
SIG_CANTMASK(td->td_sigmask);
signotify(td);
PROC_UNLOCK(p);
/*
* Restore signal context
*/
/* %gs was restored by the trampoline. */
regs->tf_fs = context->sc_fs;
regs->tf_es = context->sc_es;
regs->tf_ds = context->sc_ds;
regs->tf_edi = context->sc_edi;
regs->tf_esi = context->sc_esi;
regs->tf_ebp = context->sc_ebp;
regs->tf_ebx = context->sc_ebx;
regs->tf_edx = context->sc_edx;
regs->tf_ecx = context->sc_ecx;
regs->tf_eax = context->sc_eax;
regs->tf_eip = context->sc_eip;
regs->tf_cs = context->sc_cs;
regs->tf_eflags = eflags;
regs->tf_esp = context->sc_esp_at_signal;
regs->tf_ss = context->sc_ss;
/*
* call sigaltstack & ignore results..
*/
lss = &uc.uc_stack;
ss.ss_sp = lss->ss_sp;
ss.ss_size = lss->ss_size;
ss.ss_flags = linux_to_bsd_sigaltstack(lss->ss_flags);
#ifdef DEBUG
if (ldebug(rt_sigreturn))
printf(LMSG("rt_sigret flags: 0x%x, sp: %p, ss: 0x%x, mask: 0x%x"),
ss.ss_flags, ss.ss_sp, ss.ss_size, context->sc_mask);
#endif
(void)kern_sigaltstack(td, &ss, NULL);
return (EJUSTRETURN);
}
/*
* MPSAFE
*/
static void
linux_prepsyscall(struct trapframe *tf, int *args, u_int *code, caddr_t *params)
{
args[0] = tf->tf_ebx;
args[1] = tf->tf_ecx;
args[2] = tf->tf_edx;
args[3] = tf->tf_esi;
args[4] = tf->tf_edi;
args[5] = tf->tf_ebp; /* Unconfirmed */
*params = NULL; /* no copyin */
}
/*
* Dump core, into a file named as described in the comments for
* expand_name(), unless the process was setuid/setgid.
*/
static int
linux_aout_coredump(struct thread *td, struct vnode *vp, off_t limit)
{
struct proc *p = td->td_proc;
struct ucred *cred = td->td_ucred;
struct vmspace *vm = p->p_vmspace;
char *tempuser;
int error;
if (ctob((uarea_pages + kstack_pages) +
vm->vm_dsize + vm->vm_ssize) >= limit)
return (EFAULT);
tempuser = malloc(ctob(uarea_pages + kstack_pages), M_TEMP,
M_WAITOK | M_ZERO);
if (tempuser == NULL)
return (ENOMEM);
PROC_LOCK(p);
fill_kinfo_proc(p, &p->p_uarea->u_kproc);
PROC_UNLOCK(p);
bcopy(p->p_uarea, tempuser, sizeof(struct user));
bcopy(td->td_frame,
tempuser + ctob(uarea_pages) +
((caddr_t)td->td_frame - (caddr_t)td->td_kstack),
sizeof(struct trapframe));
error = vn_rdwr(UIO_WRITE, vp, (caddr_t)tempuser,
ctob(uarea_pages + kstack_pages),
(off_t)0, UIO_SYSSPACE, IO_UNIT, cred, NOCRED,
(int *)NULL, td);
free(tempuser, M_TEMP);
if (error == 0)
error = vn_rdwr(UIO_WRITE, vp, vm->vm_daddr,
(int)ctob(vm->vm_dsize),
(off_t)ctob(uarea_pages + kstack_pages), UIO_USERSPACE,
IO_UNIT | IO_DIRECT, cred, NOCRED, (int *) NULL, td);
if (error == 0)
error = vn_rdwr_inchunks(UIO_WRITE, vp,
(caddr_t)trunc_page(USRSTACK - ctob(vm->vm_ssize)),
round_page(ctob(vm->vm_ssize)),
(off_t)ctob(uarea_pages + kstack_pages) +
ctob(vm->vm_dsize), UIO_USERSPACE,
- IO_UNIT | IO_DIRECT, cred, NOCRED, (int *) NULL, td);
+ IO_UNIT | IO_DIRECT, cred, NOCRED, NULL, td);
return (error);
}
/*
* If a linux binary is exec'ing something, try this image activator
* first. We override standard shell script execution in order to
* be able to modify the interpreter path. We only do this if a linux
* binary is doing the exec, so we do not create an EXEC module for it.
*/
static int exec_linux_imgact_try(struct image_params *iparams);
static int
exec_linux_imgact_try(struct image_params *imgp)
{
const char *head = (const char *)imgp->image_header;
int error = -1;
/*
* The interpreter for shell scripts run from a linux binary needs
* to be located in /compat/linux if possible in order to recursively
* maintain linux path emulation.
*/
if (((const short *)head)[0] == SHELLMAGIC) {
/*
* Run our normal shell image activator. If it succeeds attempt
* to use the alternate path for the interpreter. If an alternate
* path is found, use our stringspace to store it.
*/
if ((error = exec_shell_imgact(imgp)) == 0) {
char *rpath = NULL;
linux_emul_find(FIRST_THREAD_IN_PROC(imgp->proc), NULL,
imgp->interpreter_name, &rpath, 0);
if (rpath != imgp->interpreter_name) {
int len = strlen(rpath) + 1;
if (len <= MAXSHELLCMDLEN) {
memcpy(imgp->interpreter_name, rpath, len);
}
free(rpath, M_TEMP);
}
}
}
return(error);
}
/*
* exec_setregs may initialize some registers differently than Linux
* does, thus potentially confusing Linux binaries. If necessary, we
* override the exec_setregs default(s) here.
*/
static void
exec_linux_setregs(struct thread *td, u_long entry,
u_long stack, u_long ps_strings)
{
struct pcb *pcb = td->td_pcb;
exec_setregs(td, entry, stack, ps_strings);
/* Linux sets %gs to 0, we default to _udatasel */
pcb->pcb_gs = 0; load_gs(0);
}
struct sysentvec linux_sysvec = {
LINUX_SYS_MAXSYSCALL,
linux_sysent,
0xff,
LINUX_SIGTBLSZ,
bsd_to_linux_signal,
ELAST + 1,
bsd_to_linux_errno,
translate_traps,
linux_fixup,
linux_sendsig,
linux_sigcode,
&linux_szsigcode,
linux_prepsyscall,
"Linux a.out",
linux_aout_coredump,
exec_linux_imgact_try,
LINUX_MINSIGSTKSZ,
PAGE_SIZE,
VM_MIN_ADDRESS,
VM_MAXUSER_ADDRESS,
USRSTACK,
PS_STRINGS,
VM_PROT_ALL,
exec_copyout_strings,
exec_linux_setregs,
NULL
};
struct sysentvec elf_linux_sysvec = {
LINUX_SYS_MAXSYSCALL,
linux_sysent,
0xff,
LINUX_SIGTBLSZ,
bsd_to_linux_signal,
ELAST + 1,
bsd_to_linux_errno,
translate_traps,
elf_linux_fixup,
linux_sendsig,
linux_sigcode,
&linux_szsigcode,
linux_prepsyscall,
"Linux ELF",
elf32_coredump,
exec_linux_imgact_try,
LINUX_MINSIGSTKSZ,
PAGE_SIZE,
VM_MIN_ADDRESS,
VM_MAXUSER_ADDRESS,
USRSTACK,
PS_STRINGS,
VM_PROT_ALL,
exec_copyout_strings,
exec_linux_setregs,
NULL
};
static Elf32_Brandinfo linux_brand = {
ELFOSABI_LINUX,
EM_386,
"Linux",
"/compat/linux",
"/lib/ld-linux.so.1",
&elf_linux_sysvec,
NULL,
};
static Elf32_Brandinfo linux_glibc2brand = {
ELFOSABI_LINUX,
EM_386,
"Linux",
"/compat/linux",
"/lib/ld-linux.so.2",
&elf_linux_sysvec,
NULL,
};
Elf32_Brandinfo *linux_brandlist[] = {
&linux_brand,
&linux_glibc2brand,
NULL
};
static int
linux_elf_modevent(module_t mod, int type, void *data)
{
Elf32_Brandinfo **brandinfo;
int error;
struct linux_ioctl_handler **lihp;
error = 0;
switch(type) {
case MOD_LOAD:
for (brandinfo = &linux_brandlist[0]; *brandinfo != NULL;
++brandinfo)
if (elf32_insert_brand_entry(*brandinfo) < 0)
error = EINVAL;
if (error == 0) {
SET_FOREACH(lihp, linux_ioctl_handler_set)
linux_ioctl_register_handler(*lihp);
if (bootverbose)
printf("Linux ELF exec handler installed\n");
} else
printf("cannot insert Linux ELF brand handler\n");
break;
case MOD_UNLOAD:
for (brandinfo = &linux_brandlist[0]; *brandinfo != NULL;
++brandinfo)
if (elf32_brand_inuse(*brandinfo))
error = EBUSY;
if (error == 0) {
for (brandinfo = &linux_brandlist[0];
*brandinfo != NULL; ++brandinfo)
if (elf32_remove_brand_entry(*brandinfo) < 0)
error = EINVAL;
}
if (error == 0) {
SET_FOREACH(lihp, linux_ioctl_handler_set)
linux_ioctl_unregister_handler(*lihp);
if (bootverbose)
printf("Linux ELF exec handler removed\n");
linux_mib_destroy();
} else
printf("Could not deinstall ELF interpreter entry\n");
break;
default:
break;
}
return error;
}
static moduledata_t linux_elf_mod = {
"linuxelf",
linux_elf_modevent,
0
};
DECLARE_MODULE(linuxelf, linux_elf_mod, SI_SUB_EXEC, SI_ORDER_ANY);
Index: head/sys/kern/imgact_aout.c
===================================================================
--- head/sys/kern/imgact_aout.c (revision 130100)
+++ head/sys/kern/imgact_aout.c (revision 130101)
@@ -1,318 +1,318 @@
/*
* Copyright (c) 1993, David Greenman
* 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 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/exec.h>
#include <sys/imgact.h>
#include <sys/imgact_aout.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/resourcevar.h>
#include <sys/signalvar.h>
#include <sys/syscall.h>
#include <sys/sysent.h>
#include <sys/systm.h>
#include <sys/vnode.h>
#include <sys/user.h>
#include <machine/frame.h>
#include <machine/md_var.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <vm/vm_param.h>
static int exec_aout_imgact(struct image_params *imgp);
static int aout_fixup(register_t **stack_base, struct image_params *imgp);
struct sysentvec aout_sysvec = {
SYS_MAXSYSCALL,
sysent,
0,
0,
NULL,
0,
NULL,
NULL,
aout_fixup,
sendsig,
sigcode,
&szsigcode,
NULL,
"FreeBSD a.out",
aout_coredump,
NULL,
MINSIGSTKSZ,
PAGE_SIZE,
VM_MIN_ADDRESS,
VM_MAXUSER_ADDRESS,
USRSTACK,
PS_STRINGS,
VM_PROT_ALL,
exec_copyout_strings,
exec_setregs,
NULL
};
static int
aout_fixup(stack_base, imgp)
register_t **stack_base;
struct image_params *imgp;
{
return (suword(--(*stack_base), imgp->argc));
}
static int
exec_aout_imgact(imgp)
struct image_params *imgp;
{
const struct exec *a_out = (const struct exec *) imgp->image_header;
struct vmspace *vmspace;
struct vnode *vp;
vm_map_t map;
vm_object_t object;
vm_offset_t text_end, data_end;
unsigned long virtual_offset;
unsigned long file_offset;
unsigned long bss_size;
int error;
GIANT_REQUIRED;
/*
* Linux and *BSD binaries look very much alike,
* only the machine id is different:
* 0x64 for Linux, 0x86 for *BSD, 0x00 for BSDI.
* NetBSD is in network byte order.. ugh.
*/
if (((a_out->a_magic >> 16) & 0xff) != 0x86 &&
((a_out->a_magic >> 16) & 0xff) != 0 &&
((((int)ntohl(a_out->a_magic)) >> 16) & 0xff) != 0x86)
return -1;
/*
* Set file/virtual offset based on a.out variant.
* We do two cases: host byte order and network byte order
* (for NetBSD compatibility)
*/
switch ((int)(a_out->a_magic & 0xffff)) {
case ZMAGIC:
virtual_offset = 0;
if (a_out->a_text) {
file_offset = PAGE_SIZE;
} else {
/* Bill's "screwball mode" */
file_offset = 0;
}
break;
case QMAGIC:
virtual_offset = PAGE_SIZE;
file_offset = 0;
/* Pass PS_STRINGS for BSD/OS binaries only. */
if (N_GETMID(*a_out) == MID_ZERO)
imgp->ps_strings = aout_sysvec.sv_psstrings;
break;
default:
/* NetBSD compatibility */
switch ((int)(ntohl(a_out->a_magic) & 0xffff)) {
case ZMAGIC:
case QMAGIC:
virtual_offset = PAGE_SIZE;
file_offset = 0;
break;
default:
return (-1);
}
}
bss_size = roundup(a_out->a_bss, PAGE_SIZE);
/*
* Check various fields in header for validity/bounds.
*/
if (/* entry point must lay with text region */
a_out->a_entry < virtual_offset ||
a_out->a_entry >= virtual_offset + a_out->a_text ||
/* text and data size must each be page rounded */
a_out->a_text & PAGE_MASK || a_out->a_data & PAGE_MASK)
return (-1);
/* text + data can't exceed file size */
if (a_out->a_data + a_out->a_text > imgp->attr->va_size)
return (EFAULT);
/*
* text/data/bss must not exceed limits
*/
PROC_LOCK(imgp->proc);
if (/* text can't exceed maximum text size */
a_out->a_text > maxtsiz ||
/* data + bss can't exceed rlimit */
a_out->a_data + bss_size > lim_cur(imgp->proc, RLIMIT_DATA)) {
PROC_UNLOCK(imgp->proc);
return (ENOMEM);
}
PROC_UNLOCK(imgp->proc);
/* copy in arguments and/or environment from old process */
error = exec_extract_strings(imgp);
if (error)
return (error);
/*
* Destroy old process VM and create a new one (with a new stack)
*/
exec_new_vmspace(imgp, &aout_sysvec);
/*
* The vm space can be changed by exec_new_vmspace
*/
vmspace = imgp->proc->p_vmspace;
vp = imgp->vp;
object = imgp->object;
map = &vmspace->vm_map;
vm_map_lock(map);
vm_object_reference(object);
text_end = virtual_offset + a_out->a_text;
error = vm_map_insert(map, object,
file_offset,
virtual_offset, text_end,
VM_PROT_READ | VM_PROT_EXECUTE, VM_PROT_ALL,
MAP_COPY_ON_WRITE | MAP_PREFAULT);
if (error) {
vm_map_unlock(map);
return (error);
}
data_end = text_end + a_out->a_data;
if (a_out->a_data) {
vm_object_reference(object);
error = vm_map_insert(map, object,
file_offset + a_out->a_text,
text_end, data_end,
VM_PROT_ALL, VM_PROT_ALL,
MAP_COPY_ON_WRITE | MAP_PREFAULT);
if (error) {
vm_map_unlock(map);
return (error);
}
}
if (bss_size) {
error = vm_map_insert(map, NULL, 0,
data_end, data_end + bss_size,
VM_PROT_ALL, VM_PROT_ALL, 0);
if (error) {
vm_map_unlock(map);
return (error);
}
}
vm_map_unlock(map);
/* Fill in process VM information */
vmspace->vm_tsize = a_out->a_text >> PAGE_SHIFT;
vmspace->vm_dsize = (a_out->a_data + bss_size) >> PAGE_SHIFT;
vmspace->vm_taddr = (caddr_t) (uintptr_t) virtual_offset;
vmspace->vm_daddr = (caddr_t) (uintptr_t)
(virtual_offset + a_out->a_text);
/* Fill in image_params */
imgp->interpreted = 0;
imgp->entry_addr = a_out->a_entry;
imgp->proc->p_sysent = &aout_sysvec;
return (0);
}
/*
* Dump core, into a file named as described in the comments for
* expand_name(), unless the process was setuid/setgid.
*/
int
aout_coredump(td, vp, limit)
register struct thread *td;
register struct vnode *vp;
off_t limit;
{
struct proc *p = td->td_proc;
register struct ucred *cred = td->td_ucred;
register struct vmspace *vm = p->p_vmspace;
char *tempuser;
int error;
if (ctob((uarea_pages + kstack_pages)
+ vm->vm_dsize + vm->vm_ssize) >= limit)
return (EFAULT);
tempuser = malloc(ctob(uarea_pages + kstack_pages), M_TEMP,
M_WAITOK | M_ZERO);
if (tempuser == NULL)
return (ENOMEM);
PROC_LOCK(p);
fill_kinfo_proc(p, &p->p_uarea->u_kproc);
PROC_UNLOCK(p);
bcopy(p->p_uarea, tempuser, sizeof(struct user));
bcopy(td->td_frame,
tempuser + ctob(uarea_pages) +
((caddr_t)td->td_frame - (caddr_t)td->td_kstack),
sizeof(struct trapframe));
error = vn_rdwr(UIO_WRITE, vp, (caddr_t)tempuser,
ctob(uarea_pages + kstack_pages),
(off_t)0, UIO_SYSSPACE, IO_UNIT, cred, NOCRED,
(int *)NULL, td);
free(tempuser, M_TEMP);
if (error == 0)
error = vn_rdwr(UIO_WRITE, vp, vm->vm_daddr,
(int)ctob(vm->vm_dsize),
(off_t)ctob(uarea_pages + kstack_pages), UIO_USERSPACE,
IO_UNIT | IO_DIRECT, cred, NOCRED, (int *) NULL, td);
if (error == 0)
error = vn_rdwr_inchunks(UIO_WRITE, vp,
(caddr_t)trunc_page(p->p_sysent->sv_usrstack -
ctob(vm->vm_ssize)), round_page(ctob(vm->vm_ssize)),
(off_t)ctob(uarea_pages + kstack_pages) +
ctob(vm->vm_dsize), UIO_USERSPACE,
- IO_UNIT | IO_DIRECT, cred, NOCRED, (int *) NULL, td);
+ IO_UNIT | IO_DIRECT, cred, NOCRED, NULL, td);
return (error);
}
/*
* Tell kern_execve.c about it, with a little help from the linker.
*/
static struct execsw aout_execsw = { exec_aout_imgact, "a.out" };
EXEC_SET(aout, aout_execsw);
Index: head/sys/kern/imgact_elf.c
===================================================================
--- head/sys/kern/imgact_elf.c (revision 130100)
+++ head/sys/kern/imgact_elf.c (revision 130101)
@@ -1,1300 +1,1300 @@
/*-
* Copyright (c) 2000 David O'Brien
* Copyright (c) 1995-1996 Søren Schmidt
* Copyright (c) 1996 Peter Wemm
* 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
* in this position and unchanged.
* 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 name of the author 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 ``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/exec.h>
#include <sys/fcntl.h>
#include <sys/imgact.h>
#include <sys/imgact_elf.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/mman.h>
#include <sys/namei.h>
#include <sys/pioctl.h>
#include <sys/proc.h>
#include <sys/procfs.h>
#include <sys/resourcevar.h>
#include <sys/systm.h>
#include <sys/signalvar.h>
#include <sys/stat.h>
#include <sys/sx.h>
#include <sys/syscall.h>
#include <sys/sysctl.h>
#include <sys/sysent.h>
#include <sys/vnode.h>
#include <vm/vm.h>
#include <vm/vm_kern.h>
#include <vm/vm_param.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <vm/vm_extern.h>
#include <machine/elf.h>
#include <machine/md_var.h>
#define OLD_EI_BRAND 8
static int __elfN(check_header)(const Elf_Ehdr *hdr);
static Elf_Brandinfo *__elfN(get_brandinfo)(const Elf_Ehdr *hdr,
const char *interp);
static int __elfN(load_file)(struct proc *p, const char *file, u_long *addr,
u_long *entry, size_t pagesize);
static int __elfN(load_section)(struct proc *p,
struct vmspace *vmspace, struct vnode *vp, vm_object_t object,
vm_offset_t offset, caddr_t vmaddr, size_t memsz, size_t filsz,
vm_prot_t prot, size_t pagesize);
static int __CONCAT(exec_, __elfN(imgact))(struct image_params *imgp);
SYSCTL_NODE(_kern, OID_AUTO, __CONCAT(elf, __ELF_WORD_SIZE), CTLFLAG_RW, 0,
"");
int __elfN(fallback_brand) = -1;
SYSCTL_INT(__CONCAT(_kern_elf, __ELF_WORD_SIZE), OID_AUTO,
fallback_brand, CTLFLAG_RW, &__elfN(fallback_brand), 0,
__XSTRING(__CONCAT(ELF, __ELF_WORD_SIZE)) " brand of last resort");
TUNABLE_INT("kern.elf" __XSTRING(__ELF_WORD_SIZE) ".fallback_brand",
&__elfN(fallback_brand));
static int elf_trace = 0;
SYSCTL_INT(_debug, OID_AUTO, __elfN(trace), CTLFLAG_RW, &elf_trace, 0, "");
static int elf_legacy_coredump = 0;
SYSCTL_INT(_debug, OID_AUTO, __elfN(legacy_coredump), CTLFLAG_RW,
&elf_legacy_coredump, 0, "");
static Elf_Brandinfo *elf_brand_list[MAX_BRANDS];
int
__elfN(insert_brand_entry)(Elf_Brandinfo *entry)
{
int i;
for (i = 0; i < MAX_BRANDS; i++) {
if (elf_brand_list[i] == NULL) {
elf_brand_list[i] = entry;
break;
}
}
if (i == MAX_BRANDS)
return (-1);
return (0);
}
int
__elfN(remove_brand_entry)(Elf_Brandinfo *entry)
{
int i;
for (i = 0; i < MAX_BRANDS; i++) {
if (elf_brand_list[i] == entry) {
elf_brand_list[i] = NULL;
break;
}
}
if (i == MAX_BRANDS)
return (-1);
return (0);
}
int
__elfN(brand_inuse)(Elf_Brandinfo *entry)
{
struct proc *p;
int rval = FALSE;
sx_slock(&allproc_lock);
LIST_FOREACH(p, &allproc, p_list) {
if (p->p_sysent == entry->sysvec) {
rval = TRUE;
break;
}
}
sx_sunlock(&allproc_lock);
return (rval);
}
static Elf_Brandinfo *
__elfN(get_brandinfo)(const Elf_Ehdr *hdr, const char *interp)
{
Elf_Brandinfo *bi;
int i;
/*
* We support three types of branding -- (1) the ELF EI_OSABI field
* that SCO added to the ELF spec, (2) FreeBSD 3.x's traditional string
* branding w/in the ELF header, and (3) path of the `interp_path'
* field. We should also look for an ".note.ABI-tag" ELF section now
* in all Linux ELF binaries, FreeBSD 4.1+, and some NetBSD ones.
*/
/* If the executable has a brand, search for it in the brand list. */
for (i = 0; i < MAX_BRANDS; i++) {
bi = elf_brand_list[i];
if (bi != NULL && hdr->e_machine == bi->machine &&
(hdr->e_ident[EI_OSABI] == bi->brand ||
strncmp((const char *)&hdr->e_ident[OLD_EI_BRAND],
bi->compat_3_brand, strlen(bi->compat_3_brand)) == 0))
return (bi);
}
/* Lacking a known brand, search for a recognized interpreter. */
if (interp != NULL) {
for (i = 0; i < MAX_BRANDS; i++) {
bi = elf_brand_list[i];
if (bi != NULL && hdr->e_machine == bi->machine &&
strcmp(interp, bi->interp_path) == 0)
return (bi);
}
}
/* Lacking a recognized interpreter, try the default brand */
for (i = 0; i < MAX_BRANDS; i++) {
bi = elf_brand_list[i];
if (bi != NULL && hdr->e_machine == bi->machine &&
__elfN(fallback_brand) == bi->brand)
return (bi);
}
return (NULL);
}
static int
__elfN(check_header)(const Elf_Ehdr *hdr)
{
Elf_Brandinfo *bi;
int i;
if (!IS_ELF(*hdr) ||
hdr->e_ident[EI_CLASS] != ELF_TARG_CLASS ||
hdr->e_ident[EI_DATA] != ELF_TARG_DATA ||
hdr->e_ident[EI_VERSION] != EV_CURRENT ||
hdr->e_phentsize != sizeof(Elf_Phdr) ||
hdr->e_version != ELF_TARG_VER)
return (ENOEXEC);
/*
* Make sure we have at least one brand for this machine.
*/
for (i = 0; i < MAX_BRANDS; i++) {
bi = elf_brand_list[i];
if (bi != NULL && bi->machine == hdr->e_machine)
break;
}
if (i == MAX_BRANDS)
return (ENOEXEC);
return (0);
}
static int
__elfN(map_partial)(vm_map_t map, vm_object_t object, vm_ooffset_t offset,
vm_offset_t start, vm_offset_t end, vm_prot_t prot,
vm_prot_t max)
{
int error, rv;
vm_offset_t off;
vm_offset_t data_buf = 0;
/*
* Create the page if it doesn't exist yet. Ignore errors.
*/
vm_map_lock(map);
vm_map_insert(map, NULL, 0, trunc_page(start), round_page(end), max,
max, 0);
vm_map_unlock(map);
/*
* Find the page from the underlying object.
*/
if (object) {
vm_object_reference(object);
rv = vm_map_find(exec_map,
object,
trunc_page(offset),
&data_buf,
PAGE_SIZE,
TRUE,
VM_PROT_READ,
VM_PROT_ALL,
MAP_COPY_ON_WRITE | MAP_PREFAULT_PARTIAL);
if (rv != KERN_SUCCESS) {
vm_object_deallocate(object);
return (rv);
}
off = offset - trunc_page(offset);
error = copyout((caddr_t)data_buf + off, (caddr_t)start,
end - start);
vm_map_remove(exec_map, data_buf, data_buf + PAGE_SIZE);
if (error) {
return (KERN_FAILURE);
}
}
return (KERN_SUCCESS);
}
static int
__elfN(map_insert)(vm_map_t map, vm_object_t object, vm_ooffset_t offset,
vm_offset_t start, vm_offset_t end, vm_prot_t prot,
vm_prot_t max, int cow)
{
vm_offset_t data_buf, off;
vm_size_t sz;
int error, rv;
if (start != trunc_page(start)) {
rv = __elfN(map_partial)(map, object, offset, start,
round_page(start), prot, max);
if (rv)
return (rv);
offset += round_page(start) - start;
start = round_page(start);
}
if (end != round_page(end)) {
rv = __elfN(map_partial)(map, object, offset +
trunc_page(end) - start, trunc_page(end), end, prot, max);
if (rv)
return (rv);
end = trunc_page(end);
}
if (end > start) {
if (offset & PAGE_MASK) {
/*
* The mapping is not page aligned. This means we have
* to copy the data. Sigh.
*/
rv = vm_map_find(map, 0, 0, &start, end - start,
FALSE, prot, max, 0);
if (rv)
return (rv);
data_buf = 0;
while (start < end) {
vm_object_reference(object);
rv = vm_map_find(exec_map,
object,
trunc_page(offset),
&data_buf,
2 * PAGE_SIZE,
TRUE,
VM_PROT_READ,
VM_PROT_ALL,
(MAP_COPY_ON_WRITE
| MAP_PREFAULT_PARTIAL));
if (rv != KERN_SUCCESS) {
vm_object_deallocate(object);
return (rv);
}
off = offset - trunc_page(offset);
sz = end - start;
if (sz > PAGE_SIZE)
sz = PAGE_SIZE;
error = copyout((caddr_t)data_buf + off,
(caddr_t)start, sz);
vm_map_remove(exec_map, data_buf,
data_buf + 2 * PAGE_SIZE);
if (error) {
return (KERN_FAILURE);
}
start += sz;
}
rv = KERN_SUCCESS;
} else {
vm_map_lock(map);
rv = vm_map_insert(map, object, offset, start, end,
prot, max, cow);
vm_map_unlock(map);
}
return (rv);
} else {
return (KERN_SUCCESS);
}
}
static int
__elfN(load_section)(struct proc *p, struct vmspace *vmspace,
struct vnode *vp, vm_object_t object, vm_offset_t offset,
caddr_t vmaddr, size_t memsz, size_t filsz, vm_prot_t prot,
size_t pagesize)
{
size_t map_len;
vm_offset_t map_addr;
int error, rv, cow;
size_t copy_len;
vm_offset_t file_addr;
vm_offset_t data_buf = 0;
GIANT_REQUIRED;
error = 0;
/*
* It's necessary to fail if the filsz + offset taken from the
* header is greater than the actual file pager object's size.
* If we were to allow this, then the vm_map_find() below would
* walk right off the end of the file object and into the ether.
*
* While I'm here, might as well check for something else that
* is invalid: filsz cannot be greater than memsz.
*/
if ((off_t)filsz + offset > object->un_pager.vnp.vnp_size ||
filsz > memsz) {
uprintf("elf_load_section: truncated ELF file\n");
return (ENOEXEC);
}
#define trunc_page_ps(va, ps) ((va) & ~(ps - 1))
#define round_page_ps(va, ps) (((va) + (ps - 1)) & ~(ps - 1))
map_addr = trunc_page_ps((vm_offset_t)vmaddr, pagesize);
file_addr = trunc_page_ps(offset, pagesize);
/*
* We have two choices. We can either clear the data in the last page
* of an oversized mapping, or we can start the anon mapping a page
* early and copy the initialized data into that first page. We
* choose the second..
*/
if (memsz > filsz)
map_len = trunc_page_ps(offset + filsz, pagesize) - file_addr;
else
map_len = round_page_ps(offset + filsz, pagesize) - file_addr;
if (map_len != 0) {
vm_object_reference(object);
/* cow flags: don't dump readonly sections in core */
cow = MAP_COPY_ON_WRITE | MAP_PREFAULT |
(prot & VM_PROT_WRITE ? 0 : MAP_DISABLE_COREDUMP);
rv = __elfN(map_insert)(&vmspace->vm_map,
object,
file_addr, /* file offset */
map_addr, /* virtual start */
map_addr + map_len,/* virtual end */
prot,
VM_PROT_ALL,
cow);
if (rv != KERN_SUCCESS) {
vm_object_deallocate(object);
return (EINVAL);
}
/* we can stop now if we've covered it all */
if (memsz == filsz) {
return (0);
}
}
/*
* We have to get the remaining bit of the file into the first part
* of the oversized map segment. This is normally because the .data
* segment in the file is extended to provide bss. It's a neat idea
* to try and save a page, but it's a pain in the behind to implement.
*/
copy_len = (offset + filsz) - trunc_page_ps(offset + filsz, pagesize);
map_addr = trunc_page_ps((vm_offset_t)vmaddr + filsz, pagesize);
map_len = round_page_ps((vm_offset_t)vmaddr + memsz, pagesize) -
map_addr;
/* This had damn well better be true! */
if (map_len != 0) {
rv = __elfN(map_insert)(&vmspace->vm_map, NULL, 0, map_addr,
map_addr + map_len, VM_PROT_ALL, VM_PROT_ALL, 0);
if (rv != KERN_SUCCESS) {
return (EINVAL);
}
}
if (copy_len != 0) {
vm_offset_t off;
vm_object_reference(object);
rv = vm_map_find(exec_map,
object,
trunc_page(offset + filsz),
&data_buf,
PAGE_SIZE,
TRUE,
VM_PROT_READ,
VM_PROT_ALL,
MAP_COPY_ON_WRITE | MAP_PREFAULT_PARTIAL);
if (rv != KERN_SUCCESS) {
vm_object_deallocate(object);
return (EINVAL);
}
/* send the page fragment to user space */
off = trunc_page_ps(offset + filsz, pagesize) -
trunc_page(offset + filsz);
error = copyout((caddr_t)data_buf + off, (caddr_t)map_addr,
copy_len);
vm_map_remove(exec_map, data_buf, data_buf + PAGE_SIZE);
if (error) {
return (error);
}
}
/*
* set it to the specified protection.
* XXX had better undo the damage from pasting over the cracks here!
*/
vm_map_protect(&vmspace->vm_map, trunc_page(map_addr),
round_page(map_addr + map_len), prot, FALSE);
return (error);
}
/*
* Load the file "file" into memory. It may be either a shared object
* or an executable.
*
* The "addr" reference parameter is in/out. On entry, it specifies
* the address where a shared object should be loaded. If the file is
* an executable, this value is ignored. On exit, "addr" specifies
* where the file was actually loaded.
*
* The "entry" reference parameter is out only. On exit, it specifies
* the entry point for the loaded file.
*/
static int
__elfN(load_file)(struct proc *p, const char *file, u_long *addr,
u_long *entry, size_t pagesize)
{
struct {
struct nameidata nd;
struct vattr attr;
struct image_params image_params;
} *tempdata;
const Elf_Ehdr *hdr = NULL;
const Elf_Phdr *phdr = NULL;
struct nameidata *nd;
struct vmspace *vmspace = p->p_vmspace;
struct vattr *attr;
struct image_params *imgp;
vm_prot_t prot;
u_long rbase;
u_long base_addr = 0;
int error, i, numsegs;
if (curthread->td_proc != p)
panic("elf_load_file - thread"); /* XXXKSE DIAGNOSTIC */
tempdata = malloc(sizeof(*tempdata), M_TEMP, M_WAITOK);
nd = &tempdata->nd;
attr = &tempdata->attr;
imgp = &tempdata->image_params;
/*
* Initialize part of the common data
*/
imgp->proc = p;
imgp->userspace_argv = NULL;
imgp->userspace_envv = NULL;
imgp->attr = attr;
imgp->firstpage = NULL;
imgp->image_header = NULL;
imgp->object = NULL;
imgp->execlabel = NULL;
/* XXXKSE */
NDINIT(nd, LOOKUP, LOCKLEAF|FOLLOW, UIO_SYSSPACE, file, curthread);
if ((error = namei(nd)) != 0) {
nd->ni_vp = NULL;
goto fail;
}
NDFREE(nd, NDF_ONLY_PNBUF);
imgp->vp = nd->ni_vp;
/*
* Check permissions, modes, uid, etc on the file, and "open" it.
*/
error = exec_check_permissions(imgp);
if (error) {
VOP_UNLOCK(nd->ni_vp, 0, curthread); /* XXXKSE */
goto fail;
}
error = exec_map_first_page(imgp);
/*
* Also make certain that the interpreter stays the same, so set
* its VV_TEXT flag, too.
*/
if (error == 0)
nd->ni_vp->v_vflag |= VV_TEXT;
VOP_GETVOBJECT(nd->ni_vp, &imgp->object);
vm_object_reference(imgp->object);
VOP_UNLOCK(nd->ni_vp, 0, curthread); /* XXXKSE */
if (error)
goto fail;
hdr = (const Elf_Ehdr *)imgp->image_header;
if ((error = __elfN(check_header)(hdr)) != 0)
goto fail;
if (hdr->e_type == ET_DYN)
rbase = *addr;
else if (hdr->e_type == ET_EXEC)
rbase = 0;
else {
error = ENOEXEC;
goto fail;
}
/* Only support headers that fit within first page for now */
/* (multiplication of two Elf_Half fields will not overflow) */
if ((hdr->e_phoff > PAGE_SIZE) ||
(hdr->e_phentsize * hdr->e_phnum) > PAGE_SIZE - hdr->e_phoff) {
error = ENOEXEC;
goto fail;
}
phdr = (const Elf_Phdr *)(imgp->image_header + hdr->e_phoff);
for (i = 0, numsegs = 0; i < hdr->e_phnum; i++) {
if (phdr[i].p_type == PT_LOAD) { /* Loadable segment */
prot = 0;
if (phdr[i].p_flags & PF_X)
prot |= VM_PROT_EXECUTE;
if (phdr[i].p_flags & PF_W)
prot |= VM_PROT_WRITE;
if (phdr[i].p_flags & PF_R)
prot |= VM_PROT_READ;
if ((error = __elfN(load_section)(p, vmspace,
nd->ni_vp, imgp->object, phdr[i].p_offset,
(caddr_t)(uintptr_t)phdr[i].p_vaddr + rbase,
phdr[i].p_memsz, phdr[i].p_filesz, prot,
pagesize)) != 0)
goto fail;
/*
* Establish the base address if this is the
* first segment.
*/
if (numsegs == 0)
base_addr = trunc_page(phdr[i].p_vaddr +
rbase);
numsegs++;
}
}
*addr = base_addr;
*entry = (unsigned long)hdr->e_entry + rbase;
fail:
if (imgp->firstpage)
exec_unmap_first_page(imgp);
if (imgp->object)
vm_object_deallocate(imgp->object);
if (nd->ni_vp)
vrele(nd->ni_vp);
free(tempdata, M_TEMP);
return (error);
}
static int
__CONCAT(exec_, __elfN(imgact))(struct image_params *imgp)
{
const Elf_Ehdr *hdr = (const Elf_Ehdr *)imgp->image_header;
const Elf_Phdr *phdr;
Elf_Auxargs *elf_auxargs = NULL;
struct vmspace *vmspace;
vm_prot_t prot;
u_long text_size = 0, data_size = 0, total_size = 0;
u_long text_addr = 0, data_addr = 0;
u_long seg_size, seg_addr;
u_long addr, entry = 0, proghdr = 0;
int error, i;
const char *interp = NULL;
Elf_Brandinfo *brand_info;
char *path;
struct thread *td = curthread;
struct sysentvec *sv;
GIANT_REQUIRED;
/*
* Do we have a valid ELF header ?
*/
if (__elfN(check_header)(hdr) != 0 || hdr->e_type != ET_EXEC)
return (-1);
/*
* From here on down, we return an errno, not -1, as we've
* detected an ELF file.
*/
if ((hdr->e_phoff > PAGE_SIZE) ||
(hdr->e_phoff + hdr->e_phentsize * hdr->e_phnum) > PAGE_SIZE) {
/* Only support headers in first page for now */
return (ENOEXEC);
}
phdr = (const Elf_Phdr *)(imgp->image_header + hdr->e_phoff);
/*
* From this point on, we may have resources that need to be freed.
*/
VOP_UNLOCK(imgp->vp, 0, td);
for (i = 0; i < hdr->e_phnum; i++) {
switch (phdr[i].p_type) {
case PT_INTERP: /* Path to interpreter */
if (phdr[i].p_filesz > MAXPATHLEN ||
phdr[i].p_offset + phdr[i].p_filesz > PAGE_SIZE) {
error = ENOEXEC;
goto fail;
}
interp = imgp->image_header + phdr[i].p_offset;
break;
default:
break;
}
}
brand_info = __elfN(get_brandinfo)(hdr, interp);
if (brand_info == NULL) {
uprintf("ELF binary type \"%u\" not known.\n",
hdr->e_ident[EI_OSABI]);
error = ENOEXEC;
goto fail;
}
sv = brand_info->sysvec;
if (interp != NULL && brand_info->interp_newpath != NULL)
interp = brand_info->interp_newpath;
if ((error = exec_extract_strings(imgp)) != 0)
goto fail;
exec_new_vmspace(imgp, sv);
vmspace = imgp->proc->p_vmspace;
for (i = 0; i < hdr->e_phnum; i++) {
switch (phdr[i].p_type) {
case PT_LOAD: /* Loadable segment */
prot = 0;
if (phdr[i].p_flags & PF_X)
prot |= VM_PROT_EXECUTE;
if (phdr[i].p_flags & PF_W)
prot |= VM_PROT_WRITE;
if (phdr[i].p_flags & PF_R)
prot |= VM_PROT_READ;
#if defined(__ia64__) && __ELF_WORD_SIZE == 32 && defined(IA32_ME_HARDER)
/*
* Some x86 binaries assume read == executable,
* notably the M3 runtime and therefore cvsup
*/
if (prot & VM_PROT_READ)
prot |= VM_PROT_EXECUTE;
#endif
if ((error = __elfN(load_section)(imgp->proc, vmspace,
imgp->vp, imgp->object, phdr[i].p_offset,
(caddr_t)(uintptr_t)phdr[i].p_vaddr,
phdr[i].p_memsz, phdr[i].p_filesz, prot,
sv->sv_pagesize)) != 0)
goto fail;
seg_addr = trunc_page(phdr[i].p_vaddr);
seg_size = round_page(phdr[i].p_memsz +
phdr[i].p_vaddr - seg_addr);
/*
* Is this .text or .data? We can't use
* VM_PROT_WRITE or VM_PROT_EXEC, it breaks the
* alpha terribly and possibly does other bad
* things so we stick to the old way of figuring
* it out: If the segment contains the program
* entry point, it's a text segment, otherwise it
* is a data segment.
*
* Note that obreak() assumes that data_addr +
* data_size == end of data load area, and the ELF
* file format expects segments to be sorted by
* address. If multiple data segments exist, the
* last one will be used.
*/
if (hdr->e_entry >= phdr[i].p_vaddr &&
hdr->e_entry < (phdr[i].p_vaddr +
phdr[i].p_memsz)) {
text_size = seg_size;
text_addr = seg_addr;
entry = (u_long)hdr->e_entry;
} else {
data_size = seg_size;
data_addr = seg_addr;
}
total_size += seg_size;
break;
case PT_PHDR: /* Program header table info */
proghdr = phdr[i].p_vaddr;
break;
default:
break;
}
}
if (data_addr == 0 && data_size == 0) {
data_addr = text_addr;
data_size = text_size;
}
/*
* Check limits. It should be safe to check the
* limits after loading the segments since we do
* not actually fault in all the segments pages.
*/
PROC_LOCK(imgp->proc);
if (data_size > lim_cur(imgp->proc, RLIMIT_DATA) ||
text_size > maxtsiz ||
total_size > lim_cur(imgp->proc, RLIMIT_VMEM)) {
PROC_UNLOCK(imgp->proc);
error = ENOMEM;
goto fail;
}
vmspace->vm_tsize = text_size >> PAGE_SHIFT;
vmspace->vm_taddr = (caddr_t)(uintptr_t)text_addr;
vmspace->vm_dsize = data_size >> PAGE_SHIFT;
vmspace->vm_daddr = (caddr_t)(uintptr_t)data_addr;
/*
* We load the dynamic linker where a userland call
* to mmap(0, ...) would put it. The rationale behind this
* calculation is that it leaves room for the heap to grow to
* its maximum allowed size.
*/
addr = round_page((vm_offset_t)imgp->proc->p_vmspace->vm_daddr +
lim_max(imgp->proc, RLIMIT_DATA));
PROC_UNLOCK(imgp->proc);
imgp->entry_addr = entry;
imgp->proc->p_sysent = sv;
if (interp != NULL && brand_info->emul_path != NULL &&
brand_info->emul_path[0] != '\0') {
path = malloc(MAXPATHLEN, M_TEMP, M_WAITOK);
snprintf(path, MAXPATHLEN, "%s%s", brand_info->emul_path,
interp);
error = __elfN(load_file)(imgp->proc, path, &addr,
&imgp->entry_addr, sv->sv_pagesize);
free(path, M_TEMP);
if (error == 0)
interp = NULL;
}
if (interp != NULL) {
error = __elfN(load_file)(imgp->proc, interp, &addr,
&imgp->entry_addr, sv->sv_pagesize);
if (error != 0) {
uprintf("ELF interpreter %s not found\n", interp);
goto fail;
}
}
/*
* Construct auxargs table (used by the fixup routine)
*/
elf_auxargs = malloc(sizeof(Elf_Auxargs), M_TEMP, M_WAITOK);
elf_auxargs->execfd = -1;
elf_auxargs->phdr = proghdr;
elf_auxargs->phent = hdr->e_phentsize;
elf_auxargs->phnum = hdr->e_phnum;
elf_auxargs->pagesz = PAGE_SIZE;
elf_auxargs->base = addr;
elf_auxargs->flags = 0;
elf_auxargs->entry = entry;
elf_auxargs->trace = elf_trace;
imgp->auxargs = elf_auxargs;
imgp->interpreted = 0;
fail:
vn_lock(imgp->vp, LK_EXCLUSIVE | LK_RETRY, td);
return (error);
}
#define suword __CONCAT(suword, __ELF_WORD_SIZE)
int
__elfN(freebsd_fixup)(register_t **stack_base, struct image_params *imgp)
{
Elf_Auxargs *args = (Elf_Auxargs *)imgp->auxargs;
Elf_Addr *base;
Elf_Addr *pos;
base = (Elf_Addr *)*stack_base;
pos = base + (imgp->argc + imgp->envc + 2);
if (args->trace) {
AUXARGS_ENTRY(pos, AT_DEBUG, 1);
}
if (args->execfd != -1) {
AUXARGS_ENTRY(pos, AT_EXECFD, args->execfd);
}
AUXARGS_ENTRY(pos, AT_PHDR, args->phdr);
AUXARGS_ENTRY(pos, AT_PHENT, args->phent);
AUXARGS_ENTRY(pos, AT_PHNUM, args->phnum);
AUXARGS_ENTRY(pos, AT_PAGESZ, args->pagesz);
AUXARGS_ENTRY(pos, AT_FLAGS, args->flags);
AUXARGS_ENTRY(pos, AT_ENTRY, args->entry);
AUXARGS_ENTRY(pos, AT_BASE, args->base);
AUXARGS_ENTRY(pos, AT_NULL, 0);
free(imgp->auxargs, M_TEMP);
imgp->auxargs = NULL;
base--;
suword(base, (long)imgp->argc);
*stack_base = (register_t *)base;
return (0);
}
/*
* Code for generating ELF core dumps.
*/
typedef void (*segment_callback)(vm_map_entry_t, void *);
/* Closure for cb_put_phdr(). */
struct phdr_closure {
Elf_Phdr *phdr; /* Program header to fill in */
Elf_Off offset; /* Offset of segment in core file */
};
/* Closure for cb_size_segment(). */
struct sseg_closure {
int count; /* Count of writable segments. */
size_t size; /* Total size of all writable segments. */
};
static void cb_put_phdr(vm_map_entry_t, void *);
static void cb_size_segment(vm_map_entry_t, void *);
static void each_writable_segment(struct proc *, segment_callback, void *);
static int __elfN(corehdr)(struct thread *, struct vnode *, struct ucred *,
int, void *, size_t);
static void __elfN(puthdr)(struct proc *, void *, size_t *, int);
static void __elfN(putnote)(void *, size_t *, const char *, int,
const void *, size_t);
extern int osreldate;
int
__elfN(coredump)(td, vp, limit)
struct thread *td;
register struct vnode *vp;
off_t limit;
{
register struct proc *p = td->td_proc;
register struct ucred *cred = td->td_ucred;
int error = 0;
struct sseg_closure seginfo;
void *hdr;
size_t hdrsize;
/* Size the program segments. */
seginfo.count = 0;
seginfo.size = 0;
each_writable_segment(p, cb_size_segment, &seginfo);
/*
* Calculate the size of the core file header area by making
* a dry run of generating it. Nothing is written, but the
* size is calculated.
*/
hdrsize = 0;
__elfN(puthdr)(p, (void *)NULL, &hdrsize, seginfo.count);
if (hdrsize + seginfo.size >= limit)
return (EFAULT);
/*
* Allocate memory for building the header, fill it up,
* and write it out.
*/
hdr = malloc(hdrsize, M_TEMP, M_WAITOK);
if (hdr == NULL) {
return (EINVAL);
}
error = __elfN(corehdr)(td, vp, cred, seginfo.count, hdr, hdrsize);
/* Write the contents of all of the writable segments. */
if (error == 0) {
Elf_Phdr *php;
off_t offset;
int i;
php = (Elf_Phdr *)((char *)hdr + sizeof(Elf_Ehdr)) + 1;
offset = hdrsize;
for (i = 0; i < seginfo.count; i++) {
error = vn_rdwr_inchunks(UIO_WRITE, vp,
(caddr_t)(uintptr_t)php->p_vaddr,
php->p_filesz, offset, UIO_USERSPACE,
- IO_UNIT | IO_DIRECT, cred, NOCRED, (int *)NULL,
+ IO_UNIT | IO_DIRECT, cred, NOCRED, NULL,
curthread); /* XXXKSE */
if (error != 0)
break;
offset += php->p_filesz;
php++;
}
}
free(hdr, M_TEMP);
return (error);
}
/*
* A callback for each_writable_segment() to write out the segment's
* program header entry.
*/
static void
cb_put_phdr(entry, closure)
vm_map_entry_t entry;
void *closure;
{
struct phdr_closure *phc = (struct phdr_closure *)closure;
Elf_Phdr *phdr = phc->phdr;
phc->offset = round_page(phc->offset);
phdr->p_type = PT_LOAD;
phdr->p_offset = phc->offset;
phdr->p_vaddr = entry->start;
phdr->p_paddr = 0;
phdr->p_filesz = phdr->p_memsz = entry->end - entry->start;
phdr->p_align = PAGE_SIZE;
phdr->p_flags = 0;
if (entry->protection & VM_PROT_READ)
phdr->p_flags |= PF_R;
if (entry->protection & VM_PROT_WRITE)
phdr->p_flags |= PF_W;
if (entry->protection & VM_PROT_EXECUTE)
phdr->p_flags |= PF_X;
phc->offset += phdr->p_filesz;
phc->phdr++;
}
/*
* A callback for each_writable_segment() to gather information about
* the number of segments and their total size.
*/
static void
cb_size_segment(entry, closure)
vm_map_entry_t entry;
void *closure;
{
struct sseg_closure *ssc = (struct sseg_closure *)closure;
ssc->count++;
ssc->size += entry->end - entry->start;
}
/*
* For each writable segment in the process's memory map, call the given
* function with a pointer to the map entry and some arbitrary
* caller-supplied data.
*/
static void
each_writable_segment(p, func, closure)
struct proc *p;
segment_callback func;
void *closure;
{
vm_map_t map = &p->p_vmspace->vm_map;
vm_map_entry_t entry;
for (entry = map->header.next; entry != &map->header;
entry = entry->next) {
vm_object_t obj;
/*
* Don't dump inaccessible mappings, deal with legacy
* coredump mode.
*
* Note that read-only segments related to the elf binary
* are marked MAP_ENTRY_NOCOREDUMP now so we no longer
* need to arbitrarily ignore such segments.
*/
if (elf_legacy_coredump) {
if ((entry->protection & VM_PROT_RW) != VM_PROT_RW)
continue;
} else {
if ((entry->protection & VM_PROT_ALL) == 0)
continue;
}
/*
* Dont include memory segment in the coredump if
* MAP_NOCORE is set in mmap(2) or MADV_NOCORE in
* madvise(2). Do not dump submaps (i.e. parts of the
* kernel map).
*/
if (entry->eflags & (MAP_ENTRY_NOCOREDUMP|MAP_ENTRY_IS_SUB_MAP))
continue;
if ((obj = entry->object.vm_object) == NULL)
continue;
/* Find the deepest backing object. */
while (obj->backing_object != NULL)
obj = obj->backing_object;
/* Ignore memory-mapped devices and such things. */
if (obj->type != OBJT_DEFAULT &&
obj->type != OBJT_SWAP &&
obj->type != OBJT_VNODE)
continue;
(*func)(entry, closure);
}
}
/*
* Write the core file header to the file, including padding up to
* the page boundary.
*/
static int
__elfN(corehdr)(td, vp, cred, numsegs, hdr, hdrsize)
struct thread *td;
struct vnode *vp;
struct ucred *cred;
int numsegs;
size_t hdrsize;
void *hdr;
{
struct proc *p = td->td_proc;
size_t off;
/* Fill in the header. */
bzero(hdr, hdrsize);
off = 0;
__elfN(puthdr)(p, hdr, &off, numsegs);
/* Write it to the core file. */
return (vn_rdwr_inchunks(UIO_WRITE, vp, hdr, hdrsize, (off_t)0,
UIO_SYSSPACE, IO_UNIT | IO_DIRECT, cred, NOCRED, NULL,
td)); /* XXXKSE */
}
static void
__elfN(puthdr)(struct proc *p, void *dst, size_t *off, int numsegs)
{
struct {
prstatus_t status;
prfpregset_t fpregset;
prpsinfo_t psinfo;
} *tempdata;
prstatus_t *status;
prfpregset_t *fpregset;
prpsinfo_t *psinfo;
struct thread *first, *thr;
size_t ehoff, noteoff, notesz, phoff;
ehoff = *off;
*off += sizeof(Elf_Ehdr);
phoff = *off;
*off += (numsegs + 1) * sizeof(Elf_Phdr);
noteoff = *off;
/*
* Don't allocate space for the notes if we're just calculating
* the size of the header. We also don't collect the data.
*/
if (dst != NULL) {
tempdata = malloc(sizeof(*tempdata), M_TEMP, M_ZERO|M_WAITOK);
status = &tempdata->status;
fpregset = &tempdata->fpregset;
psinfo = &tempdata->psinfo;
} else {
tempdata = NULL;
status = NULL;
fpregset = NULL;
psinfo = NULL;
}
if (dst != NULL) {
psinfo->pr_version = PRPSINFO_VERSION;
psinfo->pr_psinfosz = sizeof(prpsinfo_t);
strlcpy(psinfo->pr_fname, p->p_comm, sizeof(psinfo->pr_fname));
/*
* XXX - We don't fill in the command line arguments properly
* yet.
*/
strlcpy(psinfo->pr_psargs, p->p_comm,
sizeof(psinfo->pr_psargs));
}
__elfN(putnote)(dst, off, "FreeBSD", NT_PRPSINFO, psinfo,
sizeof *psinfo);
/*
* We want to start with the registers of the initial thread in the
* process so that the .reg and .reg2 pseudo-sections created by bfd
* will be identical to the .reg/$PID and .reg2/$PID pseudo-sections.
* This makes sure that any tool that only looks for .reg and .reg2
* and not for .reg/$PID and .reg2/$PID will behave the same as
* before. The first thread is the thread with an ID equal to the
* process' ID.
* Note that the initial thread may already be gone. In that case
* 'first' is NULL.
*/
thr = first = TAILQ_FIRST(&p->p_threads);
while (first != NULL && first->td_tid > PID_MAX)
first = TAILQ_NEXT(first, td_plist);
if (first != NULL)
thr = first;
do {
if (dst != NULL) {
status->pr_version = PRSTATUS_VERSION;
status->pr_statussz = sizeof(prstatus_t);
status->pr_gregsetsz = sizeof(gregset_t);
status->pr_fpregsetsz = sizeof(fpregset_t);
status->pr_osreldate = osreldate;
status->pr_cursig = p->p_sig;
status->pr_pid = thr->td_tid;
fill_regs(thr, &status->pr_reg);
fill_fpregs(thr, fpregset);
}
__elfN(putnote)(dst, off, "FreeBSD", NT_PRSTATUS, status,
sizeof *status);
__elfN(putnote)(dst, off, "FreeBSD", NT_FPREGSET, fpregset,
sizeof *fpregset);
/* XXX allow for MD specific notes. */
thr = (thr == first) ? TAILQ_FIRST(&p->p_threads) :
TAILQ_NEXT(thr, td_plist);
if (thr == first && thr != NULL)
thr = TAILQ_NEXT(thr, td_plist);
} while (thr != NULL);
notesz = *off - noteoff;
if (dst != NULL)
free(tempdata, M_TEMP);
/* Align up to a page boundary for the program segments. */
*off = round_page(*off);
if (dst != NULL) {
Elf_Ehdr *ehdr;
Elf_Phdr *phdr;
struct phdr_closure phc;
/*
* Fill in the ELF header.
*/
ehdr = (Elf_Ehdr *)((char *)dst + ehoff);
ehdr->e_ident[EI_MAG0] = ELFMAG0;
ehdr->e_ident[EI_MAG1] = ELFMAG1;
ehdr->e_ident[EI_MAG2] = ELFMAG2;
ehdr->e_ident[EI_MAG3] = ELFMAG3;
ehdr->e_ident[EI_CLASS] = ELF_CLASS;
ehdr->e_ident[EI_DATA] = ELF_DATA;
ehdr->e_ident[EI_VERSION] = EV_CURRENT;
ehdr->e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
ehdr->e_ident[EI_ABIVERSION] = 0;
ehdr->e_ident[EI_PAD] = 0;
ehdr->e_type = ET_CORE;
ehdr->e_machine = ELF_ARCH;
ehdr->e_version = EV_CURRENT;
ehdr->e_entry = 0;
ehdr->e_phoff = phoff;
ehdr->e_flags = 0;
ehdr->e_ehsize = sizeof(Elf_Ehdr);
ehdr->e_phentsize = sizeof(Elf_Phdr);
ehdr->e_phnum = numsegs + 1;
ehdr->e_shentsize = sizeof(Elf_Shdr);
ehdr->e_shnum = 0;
ehdr->e_shstrndx = SHN_UNDEF;
/*
* Fill in the program header entries.
*/
phdr = (Elf_Phdr *)((char *)dst + phoff);
/* The note segement. */
phdr->p_type = PT_NOTE;
phdr->p_offset = noteoff;
phdr->p_vaddr = 0;
phdr->p_paddr = 0;
phdr->p_filesz = notesz;
phdr->p_memsz = 0;
phdr->p_flags = 0;
phdr->p_align = 0;
phdr++;
/* All the writable segments from the program. */
phc.phdr = phdr;
phc.offset = *off;
each_writable_segment(p, cb_put_phdr, &phc);
}
}
static void
__elfN(putnote)(void *dst, size_t *off, const char *name, int type,
const void *desc, size_t descsz)
{
Elf_Note note;
note.n_namesz = strlen(name) + 1;
note.n_descsz = descsz;
note.n_type = type;
if (dst != NULL)
bcopy(&note, (char *)dst + *off, sizeof note);
*off += sizeof note;
if (dst != NULL)
bcopy(name, (char *)dst + *off, note.n_namesz);
*off += roundup2(note.n_namesz, sizeof(Elf_Size));
if (dst != NULL)
bcopy(desc, (char *)dst + *off, note.n_descsz);
*off += roundup2(note.n_descsz, sizeof(Elf_Size));
}
/*
* Tell kern_execve.c about it, with a little help from the linker.
*/
static struct execsw __elfN(execsw) = {
__CONCAT(exec_, __elfN(imgact)),
__XSTRING(__CONCAT(ELF, __ELF_WORD_SIZE))
};
EXEC_SET(__CONCAT(elf, __ELF_WORD_SIZE), __elfN(execsw));
Index: head/sys/kern/vfs_vnops.c
===================================================================
--- head/sys/kern/vfs_vnops.c (revision 130100)
+++ head/sys/kern/vfs_vnops.c (revision 130101)
@@ -1,1194 +1,1196 @@
/*
* 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.
* 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.
*
* @(#)vfs_vnops.c 8.2 (Berkeley) 1/21/94
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_mac.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/fcntl.h>
#include <sys/file.h>
#include <sys/stat.h>
#include <sys/proc.h>
#include <sys/limits.h>
#include <sys/lock.h>
#include <sys/mac.h>
#include <sys/mount.h>
#include <sys/mutex.h>
#include <sys/namei.h>
#include <sys/vnode.h>
#include <sys/bio.h>
#include <sys/buf.h>
#include <sys/filio.h>
#include <sys/sx.h>
#include <sys/ttycom.h>
#include <sys/conf.h>
#include <sys/syslog.h>
#include <sys/unistd.h>
static fo_rdwr_t vn_read;
static fo_rdwr_t vn_write;
static fo_ioctl_t vn_ioctl;
static fo_poll_t vn_poll;
static fo_kqfilter_t vn_kqfilter;
static fo_stat_t vn_statfile;
static fo_close_t vn_closefile;
struct fileops vnops = {
.fo_read = vn_read,
.fo_write = vn_write,
.fo_ioctl = vn_ioctl,
.fo_poll = vn_poll,
.fo_kqfilter = vn_kqfilter,
.fo_stat = vn_statfile,
.fo_close = vn_closefile,
.fo_flags = DFLAG_PASSABLE | DFLAG_SEEKABLE
};
int
vn_open(ndp, flagp, cmode, fdidx)
struct nameidata *ndp;
int *flagp, cmode, fdidx;
{
struct thread *td = ndp->ni_cnd.cn_thread;
return (vn_open_cred(ndp, flagp, cmode, td->td_ucred, fdidx));
}
/*
* Common code for vnode open operations.
* Check permissions, and call the VOP_OPEN or VOP_CREATE routine.
*
* Note that this does NOT free nameidata for the successful case,
* due to the NDINIT being done elsewhere.
*/
int
vn_open_cred(ndp, flagp, cmode, cred, fdidx)
struct nameidata *ndp;
int *flagp, cmode;
struct ucred *cred;
int fdidx;
{
struct vnode *vp;
struct mount *mp;
struct thread *td = ndp->ni_cnd.cn_thread;
struct vattr vat;
struct vattr *vap = &vat;
int mode, fmode, error;
#ifdef LOOKUP_SHARED
int exclusive; /* The current intended lock state */
exclusive = 0;
#endif
restart:
fmode = *flagp;
if (fmode & O_CREAT) {
ndp->ni_cnd.cn_nameiop = CREATE;
ndp->ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF;
if ((fmode & O_EXCL) == 0 && (fmode & O_NOFOLLOW) == 0)
ndp->ni_cnd.cn_flags |= FOLLOW;
bwillwrite();
if ((error = namei(ndp)) != 0)
return (error);
if (ndp->ni_vp == NULL) {
VATTR_NULL(vap);
vap->va_type = VREG;
vap->va_mode = cmode;
if (fmode & O_EXCL)
vap->va_vaflags |= VA_EXCLUSIVE;
if (vn_start_write(ndp->ni_dvp, &mp, V_NOWAIT) != 0) {
NDFREE(ndp, NDF_ONLY_PNBUF);
vput(ndp->ni_dvp);
if ((error = vn_start_write(NULL, &mp,
V_XSLEEP | PCATCH)) != 0)
return (error);
goto restart;
}
#ifdef MAC
error = mac_check_vnode_create(cred, ndp->ni_dvp,
&ndp->ni_cnd, vap);
if (error == 0) {
#endif
VOP_LEASE(ndp->ni_dvp, td, cred, LEASE_WRITE);
error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp,
&ndp->ni_cnd, vap);
#ifdef MAC
}
#endif
vput(ndp->ni_dvp);
vn_finished_write(mp);
if (error) {
NDFREE(ndp, NDF_ONLY_PNBUF);
return (error);
}
ASSERT_VOP_UNLOCKED(ndp->ni_dvp, "create");
ASSERT_VOP_LOCKED(ndp->ni_vp, "create");
fmode &= ~O_TRUNC;
vp = ndp->ni_vp;
#ifdef LOOKUP_SHARED
exclusive = 1;
#endif
} else {
if (ndp->ni_dvp == ndp->ni_vp)
vrele(ndp->ni_dvp);
else
vput(ndp->ni_dvp);
ndp->ni_dvp = NULL;
vp = ndp->ni_vp;
if (fmode & O_EXCL) {
error = EEXIST;
goto bad;
}
fmode &= ~O_CREAT;
}
} else {
ndp->ni_cnd.cn_nameiop = LOOKUP;
#ifdef LOOKUP_SHARED
ndp->ni_cnd.cn_flags =
((fmode & O_NOFOLLOW) ? NOFOLLOW : FOLLOW) |
LOCKSHARED | LOCKLEAF;
#else
ndp->ni_cnd.cn_flags =
((fmode & O_NOFOLLOW) ? NOFOLLOW : FOLLOW) | LOCKLEAF;
#endif
if ((error = namei(ndp)) != 0)
return (error);
vp = ndp->ni_vp;
}
if (vp->v_type == VLNK) {
error = EMLINK;
goto bad;
}
if (vp->v_type == VSOCK) {
error = EOPNOTSUPP;
goto bad;
}
mode = 0;
if (fmode & (FWRITE | O_TRUNC)) {
if (vp->v_type == VDIR) {
error = EISDIR;
goto bad;
}
mode |= VWRITE;
}
if (fmode & FREAD)
mode |= VREAD;
if (fmode & O_APPEND)
mode |= VAPPEND;
#ifdef MAC
error = mac_check_vnode_open(cred, vp, mode);
if (error)
goto bad;
#endif
if ((fmode & O_CREAT) == 0) {
if (mode & VWRITE) {
error = vn_writechk(vp);
if (error)
goto bad;
}
if (mode) {
error = VOP_ACCESS(vp, mode, cred, td);
if (error)
goto bad;
}
}
if ((error = VOP_GETATTR(vp, vap, cred, td)) == 0) {
vp->v_cachedfs = vap->va_fsid;
vp->v_cachedid = vap->va_fileid;
}
if ((error = VOP_OPEN(vp, fmode, cred, td, fdidx)) != 0)
goto bad;
/*
* Make sure that a VM object is created for VMIO support.
*/
if (vn_canvmio(vp) == TRUE) {
#ifdef LOOKUP_SHARED
int flock;
if (!exclusive && VOP_GETVOBJECT(vp, NULL) != 0)
VOP_LOCK(vp, LK_UPGRADE, td);
/*
* In cases where the object is marked as dead object_create
* will unlock and relock exclusive. It is safe to call in
* here with a shared lock because we only examine fields that
* the shared lock guarantees will be stable. In the UPGRADE
* case it is not likely that anyone has used this vnode yet
* so there will be no contention. The logic after this call
* restores the requested locking state.
*/
#endif
if ((error = vfs_object_create(vp, td, cred)) != 0) {
VOP_UNLOCK(vp, 0, td);
VOP_CLOSE(vp, fmode, cred, td);
NDFREE(ndp, NDF_ONLY_PNBUF);
vrele(vp);
*flagp = fmode;
return (error);
}
#ifdef LOOKUP_SHARED
flock = VOP_ISLOCKED(vp, td);
if (!exclusive && flock == LK_EXCLUSIVE)
VOP_LOCK(vp, LK_DOWNGRADE, td);
#endif
}
if (fmode & FWRITE)
vp->v_writecount++;
*flagp = fmode;
ASSERT_VOP_LOCKED(vp, "vn_open_cred");
return (0);
bad:
NDFREE(ndp, NDF_ONLY_PNBUF);
vput(vp);
*flagp = fmode;
ndp->ni_vp = NULL;
return (error);
}
/*
* Check for write permissions on the specified vnode.
* Prototype text segments cannot be written.
*/
int
vn_writechk(vp)
register struct vnode *vp;
{
ASSERT_VOP_LOCKED(vp, "vn_writechk");
/*
* If there's shared text associated with
* the vnode, try to free it up once. If
* we fail, we can't allow writing.
*/
if (vp->v_vflag & VV_TEXT)
return (ETXTBSY);
return (0);
}
/*
* Vnode close call
*/
int
vn_close(vp, flags, file_cred, td)
register struct vnode *vp;
int flags;
struct ucred *file_cred;
struct thread *td;
{
int error;
if (flags & FWRITE)
vp->v_writecount--;
error = VOP_CLOSE(vp, flags, file_cred, td);
/*
* XXX - In certain instances VOP_CLOSE has to do the vrele
* itself. If the vrele has been done, it will return EAGAIN
* to indicate that the vrele should not be done again. When
* this happens, we just return success. The correct thing to
* do would be to have all VOP_CLOSE instances do the vrele.
*/
if (error == EAGAIN)
return (0);
vrele(vp);
return (error);
}
/*
* Sequential heuristic - detect sequential operation
*/
static __inline
int
sequential_heuristic(struct uio *uio, struct file *fp)
{
if ((uio->uio_offset == 0 && fp->f_seqcount > 0) ||
uio->uio_offset == fp->f_nextoff) {
/*
* XXX we assume that the filesystem block size is
* the default. Not true, but still gives us a pretty
* good indicator of how sequential the read operations
* are.
*/
fp->f_seqcount += (uio->uio_resid + BKVASIZE - 1) / BKVASIZE;
if (fp->f_seqcount > IO_SEQMAX)
fp->f_seqcount = IO_SEQMAX;
return(fp->f_seqcount << IO_SEQSHIFT);
}
/*
* Not sequential, quick draw-down of seqcount
*/
if (fp->f_seqcount > 1)
fp->f_seqcount = 1;
else
fp->f_seqcount = 0;
return(0);
}
/*
* Package up an I/O request on a vnode into a uio and do it.
*/
int
vn_rdwr(rw, vp, base, len, offset, segflg, ioflg, active_cred, file_cred,
aresid, td)
enum uio_rw rw;
struct vnode *vp;
caddr_t base;
int len;
off_t offset;
enum uio_seg segflg;
int ioflg;
struct ucred *active_cred;
struct ucred *file_cred;
int *aresid;
struct thread *td;
{
struct uio auio;
struct iovec aiov;
struct mount *mp;
struct ucred *cred;
int error;
if ((ioflg & IO_NODELOCKED) == 0) {
mp = NULL;
if (rw == UIO_WRITE) {
if (vp->v_type != VCHR &&
(error = vn_start_write(vp, &mp, V_WAIT | PCATCH))
!= 0)
return (error);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
} else {
/*
* XXX This should be LK_SHARED but I don't trust VFS
* enough to leave it like that until it has been
* reviewed further.
*/
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
}
}
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
aiov.iov_base = base;
aiov.iov_len = len;
auio.uio_resid = len;
auio.uio_offset = offset;
auio.uio_segflg = segflg;
auio.uio_rw = rw;
auio.uio_td = td;
error = 0;
#ifdef MAC
if ((ioflg & IO_NOMACCHECK) == 0) {
if (rw == UIO_READ)
error = mac_check_vnode_read(active_cred, file_cred,
vp);
else
error = mac_check_vnode_write(active_cred, file_cred,
vp);
}
#endif
if (error == 0) {
if (file_cred)
cred = file_cred;
else
cred = active_cred;
if (rw == UIO_READ)
error = VOP_READ(vp, &auio, ioflg, cred);
else
error = VOP_WRITE(vp, &auio, ioflg, cred);
}
if (aresid)
*aresid = auio.uio_resid;
else
if (auio.uio_resid && error == 0)
error = EIO;
if ((ioflg & IO_NODELOCKED) == 0) {
if (rw == UIO_WRITE)
vn_finished_write(mp);
VOP_UNLOCK(vp, 0, td);
}
return (error);
}
/*
* Package up an I/O request on a vnode into a uio and do it. The I/O
* request is split up into smaller chunks and we try to avoid saturating
* the buffer cache while potentially holding a vnode locked, so we
* check bwillwrite() before calling vn_rdwr(). We also call uio_yield()
* to give other processes a chance to lock the vnode (either other processes
* core'ing the same binary, or unrelated processes scanning the directory).
*/
int
vn_rdwr_inchunks(rw, vp, base, len, offset, segflg, ioflg, active_cred,
file_cred, aresid, td)
enum uio_rw rw;
struct vnode *vp;
caddr_t base;
- int len;
+ size_t len;
off_t offset;
enum uio_seg segflg;
int ioflg;
struct ucred *active_cred;
struct ucred *file_cred;
- int *aresid;
+ size_t *aresid;
struct thread *td;
{
int error = 0;
+ int iaresid;
do {
int chunk;
/*
* Force `offset' to a multiple of MAXBSIZE except possibly
* for the first chunk, so that filesystems only need to
* write full blocks except possibly for the first and last
* chunks.
*/
chunk = MAXBSIZE - (uoff_t)offset % MAXBSIZE;
if (chunk > len)
chunk = len;
if (rw != UIO_READ && vp->v_type == VREG)
bwillwrite();
+ iaresid = 0;
error = vn_rdwr(rw, vp, base, chunk, offset, segflg,
- ioflg, active_cred, file_cred, aresid, td);
+ ioflg, active_cred, file_cred, &iaresid, td);
len -= chunk; /* aresid calc already includes length */
if (error)
break;
offset += chunk;
base += chunk;
uio_yield();
} while (len);
if (aresid)
- *aresid += len;
+ *aresid = len + iaresid;
return (error);
}
/*
* File table vnode read routine.
*/
static int
vn_read(fp, uio, active_cred, flags, td)
struct file *fp;
struct uio *uio;
struct ucred *active_cred;
struct thread *td;
int flags;
{
struct vnode *vp;
int error, ioflag;
mtx_lock(&Giant);
KASSERT(uio->uio_td == td, ("uio_td %p is not td %p",
uio->uio_td, td));
vp = fp->f_vnode;
ioflag = 0;
if (fp->f_flag & FNONBLOCK)
ioflag |= IO_NDELAY;
if (fp->f_flag & O_DIRECT)
ioflag |= IO_DIRECT;
VOP_LEASE(vp, td, fp->f_cred, LEASE_READ);
/*
* According to McKusick the vn lock is protecting f_offset here.
* Once this field has it's own lock we can acquire this shared.
*/
if ((flags & FOF_OFFSET) == 0) {
vn_lock(vp, LK_EXCLUSIVE | LK_NOPAUSE | LK_RETRY, td);
uio->uio_offset = fp->f_offset;
} else
vn_lock(vp, LK_SHARED | LK_NOPAUSE | LK_RETRY, td);
ioflag |= sequential_heuristic(uio, fp);
#ifdef MAC
error = mac_check_vnode_read(active_cred, fp->f_cred, vp);
if (error == 0)
#endif
error = VOP_READ(vp, uio, ioflag, fp->f_cred);
if ((flags & FOF_OFFSET) == 0)
fp->f_offset = uio->uio_offset;
fp->f_nextoff = uio->uio_offset;
VOP_UNLOCK(vp, 0, td);
mtx_unlock(&Giant);
return (error);
}
/*
* File table vnode write routine.
*/
static int
vn_write(fp, uio, active_cred, flags, td)
struct file *fp;
struct uio *uio;
struct ucred *active_cred;
struct thread *td;
int flags;
{
struct vnode *vp;
struct mount *mp;
int error, ioflag;
mtx_lock(&Giant);
KASSERT(uio->uio_td == td, ("uio_td %p is not td %p",
uio->uio_td, td));
vp = fp->f_vnode;
if (vp->v_type == VREG)
bwillwrite();
ioflag = IO_UNIT;
if (vp->v_type == VREG && (fp->f_flag & O_APPEND))
ioflag |= IO_APPEND;
if (fp->f_flag & FNONBLOCK)
ioflag |= IO_NDELAY;
if (fp->f_flag & O_DIRECT)
ioflag |= IO_DIRECT;
if ((fp->f_flag & O_FSYNC) ||
(vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS)))
ioflag |= IO_SYNC;
mp = NULL;
if (vp->v_type != VCHR &&
(error = vn_start_write(vp, &mp, V_WAIT | PCATCH)) != 0) {
mtx_unlock(&Giant);
return (error);
}
VOP_LEASE(vp, td, fp->f_cred, LEASE_WRITE);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
if ((flags & FOF_OFFSET) == 0)
uio->uio_offset = fp->f_offset;
ioflag |= sequential_heuristic(uio, fp);
#ifdef MAC
error = mac_check_vnode_write(active_cred, fp->f_cred, vp);
if (error == 0)
#endif
error = VOP_WRITE(vp, uio, ioflag, fp->f_cred);
if ((flags & FOF_OFFSET) == 0)
fp->f_offset = uio->uio_offset;
fp->f_nextoff = uio->uio_offset;
VOP_UNLOCK(vp, 0, td);
vn_finished_write(mp);
mtx_unlock(&Giant);
return (error);
}
/*
* File table vnode stat routine.
*/
static int
vn_statfile(fp, sb, active_cred, td)
struct file *fp;
struct stat *sb;
struct ucred *active_cred;
struct thread *td;
{
struct vnode *vp = fp->f_vnode;
int error;
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
error = vn_stat(vp, sb, active_cred, fp->f_cred, td);
VOP_UNLOCK(vp, 0, td);
return (error);
}
/*
* Stat a vnode; implementation for the stat syscall
*/
int
vn_stat(vp, sb, active_cred, file_cred, td)
struct vnode *vp;
register struct stat *sb;
struct ucred *active_cred;
struct ucred *file_cred;
struct thread *td;
{
struct vattr vattr;
register struct vattr *vap;
int error;
u_short mode;
#ifdef MAC
error = mac_check_vnode_stat(active_cred, file_cred, vp);
if (error)
return (error);
#endif
vap = &vattr;
error = VOP_GETATTR(vp, vap, active_cred, td);
if (error)
return (error);
vp->v_cachedfs = vap->va_fsid;
vp->v_cachedid = vap->va_fileid;
/*
* Zero the spare stat fields
*/
bzero(sb, sizeof *sb);
/*
* Copy from vattr table
*/
if (vap->va_fsid != VNOVAL)
sb->st_dev = vap->va_fsid;
else
sb->st_dev = vp->v_mount->mnt_stat.f_fsid.val[0];
sb->st_ino = vap->va_fileid;
mode = vap->va_mode;
switch (vap->va_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;
/* This is a cosmetic change, symlinks do not have a mode. */
if (vp->v_mount->mnt_flag & MNT_NOSYMFOLLOW)
sb->st_mode &= ~ACCESSPERMS; /* 0000 */
else
sb->st_mode |= ACCESSPERMS; /* 0777 */
break;
case VSOCK:
mode |= S_IFSOCK;
break;
case VFIFO:
mode |= S_IFIFO;
break;
default:
return (EBADF);
};
sb->st_mode = mode;
sb->st_nlink = vap->va_nlink;
sb->st_uid = vap->va_uid;
sb->st_gid = vap->va_gid;
sb->st_rdev = vap->va_rdev;
if (vap->va_size > OFF_MAX)
return (EOVERFLOW);
sb->st_size = vap->va_size;
sb->st_atimespec = vap->va_atime;
sb->st_mtimespec = vap->va_mtime;
sb->st_ctimespec = vap->va_ctime;
sb->st_birthtimespec = vap->va_birthtime;
/*
* According to www.opengroup.org, the meaning of st_blksize is
* "a filesystem-specific preferred I/O block size for this
* object. In some filesystem types, this may vary from file
* to file"
* Default to PAGE_SIZE after much discussion.
*/
if (vap->va_type == VREG) {
sb->st_blksize = vap->va_blocksize;
} else if (vn_isdisk(vp, NULL)) {
sb->st_blksize = vp->v_rdev->si_bsize_best;
if (sb->st_blksize < vp->v_rdev->si_bsize_phys)
sb->st_blksize = vp->v_rdev->si_bsize_phys;
if (sb->st_blksize < BLKDEV_IOSIZE)
sb->st_blksize = BLKDEV_IOSIZE;
} else {
sb->st_blksize = PAGE_SIZE;
}
sb->st_flags = vap->va_flags;
if (suser(td))
sb->st_gen = 0;
else
sb->st_gen = vap->va_gen;
#if (S_BLKSIZE == 512)
/* Optimize this case */
sb->st_blocks = vap->va_bytes >> 9;
#else
sb->st_blocks = vap->va_bytes / S_BLKSIZE;
#endif
return (0);
}
/*
* File table vnode ioctl routine.
*/
static int
vn_ioctl(fp, com, data, active_cred, td)
struct file *fp;
u_long com;
void *data;
struct ucred *active_cred;
struct thread *td;
{
struct vnode *vp = fp->f_vnode;
struct vnode *vpold;
struct vattr vattr;
int error;
switch (vp->v_type) {
case VREG:
case VDIR:
if (com == FIONREAD) {
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
error = VOP_GETATTR(vp, &vattr, active_cred, td);
VOP_UNLOCK(vp, 0, td);
if (error)
return (error);
*(int *)data = vattr.va_size - fp->f_offset;
return (0);
}
if (com == FIONBIO || com == FIOASYNC) /* XXX */
return (0); /* XXX */
/* FALLTHROUGH */
default:
#if 0
return (ENOTTY);
#endif
case VFIFO:
case VCHR:
case VBLK:
if (com == FIODTYPE) {
if (vp->v_type != VCHR && vp->v_type != VBLK)
return (ENOTTY);
*(int *)data = devsw(vp->v_rdev)->d_flags & D_TYPEMASK;
return (0);
}
error = VOP_IOCTL(vp, com, data, fp->f_flag, active_cred, td);
if (error == ENOIOCTL) {
#ifdef DIAGNOSTIC
Debugger("ENOIOCTL leaked through");
#endif
error = ENOTTY;
}
if (error == 0 && com == TIOCSCTTY) {
/* Do nothing if reassigning same control tty */
sx_slock(&proctree_lock);
if (td->td_proc->p_session->s_ttyvp == vp) {
sx_sunlock(&proctree_lock);
return (0);
}
vpold = td->td_proc->p_session->s_ttyvp;
VREF(vp);
SESS_LOCK(td->td_proc->p_session);
td->td_proc->p_session->s_ttyvp = vp;
SESS_UNLOCK(td->td_proc->p_session);
sx_sunlock(&proctree_lock);
/* Get rid of reference to old control tty */
if (vpold)
vrele(vpold);
}
return (error);
}
}
/*
* File table vnode poll routine.
*/
static int
vn_poll(fp, events, active_cred, td)
struct file *fp;
int events;
struct ucred *active_cred;
struct thread *td;
{
struct vnode *vp;
#ifdef MAC
int error;
#endif
vp = fp->f_vnode;
#ifdef MAC
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
error = mac_check_vnode_poll(active_cred, fp->f_cred, vp);
VOP_UNLOCK(vp, 0, td);
if (error)
return (error);
#endif
return (VOP_POLL(vp, events, fp->f_cred, td));
}
/*
* Check that the vnode is still valid, and if so
* acquire requested lock.
*/
int
#ifndef DEBUG_LOCKS
vn_lock(vp, flags, td)
#else
debug_vn_lock(vp, flags, td, filename, line)
#endif
struct vnode *vp;
int flags;
struct thread *td;
#ifdef DEBUG_LOCKS
const char *filename;
int line;
#endif
{
int error;
do {
if ((flags & LK_INTERLOCK) == 0)
VI_LOCK(vp);
if ((vp->v_iflag & VI_XLOCK) && vp->v_vxthread != curthread) {
if ((flags & LK_NOWAIT) != 0) {
VI_UNLOCK(vp);
return (ENOENT);
}
vp->v_iflag |= VI_XWANT;
msleep(vp, VI_MTX(vp), PINOD, "vn_lock", 0);
if ((flags & LK_RETRY) == 0) {
VI_UNLOCK(vp);
return (ENOENT);
}
}
#ifdef DEBUG_LOCKS
vp->filename = filename;
vp->line = line;
#endif
/*
* lockmgr drops interlock before it will return for
* any reason. So force the code above to relock it.
*/
error = VOP_LOCK(vp, flags | LK_NOPAUSE | LK_INTERLOCK, td);
flags &= ~LK_INTERLOCK;
} while (flags & LK_RETRY && error != 0);
return (error);
}
/*
* File table vnode close routine.
*/
static int
vn_closefile(fp, td)
struct file *fp;
struct thread *td;
{
struct vnode *vp;
struct flock lf;
GIANT_REQUIRED;
vp = fp->f_vnode;
if (fp->f_type == DTYPE_VNODE && fp->f_flag & FHASLOCK) {
lf.l_whence = SEEK_SET;
lf.l_start = 0;
lf.l_len = 0;
lf.l_type = F_UNLCK;
(void) VOP_ADVLOCK(vp, (caddr_t)fp, F_UNLCK, &lf, F_FLOCK);
}
fp->f_ops = &badfileops;
return (vn_close(vp, fp->f_flag, fp->f_cred, td));
}
/*
* Preparing to start a filesystem write operation. If the operation is
* permitted, then we bump the count of operations in progress and
* proceed. If a suspend request is in progress, we wait until the
* suspension is over, and then proceed.
*/
int
vn_start_write(vp, mpp, flags)
struct vnode *vp;
struct mount **mpp;
int flags;
{
struct mount *mp;
int error;
GIANT_REQUIRED;
/*
* If a vnode is provided, get and return the mount point that
* to which it will write.
*/
if (vp != NULL) {
if ((error = VOP_GETWRITEMOUNT(vp, mpp)) != 0) {
*mpp = NULL;
if (error != EOPNOTSUPP)
return (error);
return (0);
}
}
if ((mp = *mpp) == NULL)
return (0);
/*
* Check on status of suspension.
*/
while ((mp->mnt_kern_flag & MNTK_SUSPEND) != 0) {
if (flags & V_NOWAIT)
return (EWOULDBLOCK);
error = tsleep(&mp->mnt_flag, (PUSER - 1) | (flags & PCATCH),
"suspfs", 0);
if (error)
return (error);
}
if (flags & V_XSLEEP)
return (0);
mp->mnt_writeopcount++;
return (0);
}
/*
* Secondary suspension. Used by operations such as vop_inactive
* routines that are needed by the higher level functions. These
* are allowed to proceed until all the higher level functions have
* completed (indicated by mnt_writeopcount dropping to zero). At that
* time, these operations are halted until the suspension is over.
*/
int
vn_write_suspend_wait(vp, mp, flags)
struct vnode *vp;
struct mount *mp;
int flags;
{
int error;
if (vp != NULL) {
if ((error = VOP_GETWRITEMOUNT(vp, &mp)) != 0) {
if (error != EOPNOTSUPP)
return (error);
return (0);
}
}
/*
* If we are not suspended or have not yet reached suspended
* mode, then let the operation proceed.
*/
if (mp == NULL || (mp->mnt_kern_flag & MNTK_SUSPENDED) == 0)
return (0);
if (flags & V_NOWAIT)
return (EWOULDBLOCK);
/*
* Wait for the suspension to finish.
*/
return (tsleep(&mp->mnt_flag, (PUSER - 1) | (flags & PCATCH),
"suspfs", 0));
}
/*
* Filesystem write operation has completed. If we are suspending and this
* operation is the last one, notify the suspender that the suspension is
* now in effect.
*/
void
vn_finished_write(mp)
struct mount *mp;
{
GIANT_REQUIRED;
if (mp == NULL)
return;
mp->mnt_writeopcount--;
if (mp->mnt_writeopcount < 0)
panic("vn_finished_write: neg cnt");
if ((mp->mnt_kern_flag & MNTK_SUSPEND) != 0 &&
mp->mnt_writeopcount <= 0)
wakeup(&mp->mnt_writeopcount);
}
/*
* Request a filesystem to suspend write operations.
*/
int
vfs_write_suspend(mp)
struct mount *mp;
{
struct thread *td = curthread;
int error;
if (mp->mnt_kern_flag & MNTK_SUSPEND)
return (0);
mp->mnt_kern_flag |= MNTK_SUSPEND;
if (mp->mnt_writeopcount > 0)
(void) tsleep(&mp->mnt_writeopcount, PUSER - 1, "suspwt", 0);
if ((error = VFS_SYNC(mp, MNT_WAIT, td->td_ucred, td)) != 0) {
vfs_write_resume(mp);
return (error);
}
mp->mnt_kern_flag |= MNTK_SUSPENDED;
return (0);
}
/*
* Request a filesystem to resume write operations.
*/
void
vfs_write_resume(mp)
struct mount *mp;
{
if ((mp->mnt_kern_flag & MNTK_SUSPEND) == 0)
return;
mp->mnt_kern_flag &= ~(MNTK_SUSPEND | MNTK_SUSPENDED);
wakeup(&mp->mnt_writeopcount);
wakeup(&mp->mnt_flag);
}
/*
* Implement kqueues for files by translating it to vnode operation.
*/
static int
vn_kqfilter(struct file *fp, struct knote *kn)
{
return (VOP_KQFILTER(fp->f_vnode, kn));
}
/*
* Simplified in-kernel wrapper calls for extended attribute access.
* Both calls pass in a NULL credential, authorizing as "kernel" access.
* Set IO_NODELOCKED in ioflg if the vnode is already locked.
*/
int
vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
const char *attrname, int *buflen, char *buf, struct thread *td)
{
struct uio auio;
struct iovec iov;
int error;
iov.iov_len = *buflen;
iov.iov_base = buf;
auio.uio_iov = &iov;
auio.uio_iovcnt = 1;
auio.uio_rw = UIO_READ;
auio.uio_segflg = UIO_SYSSPACE;
auio.uio_td = td;
auio.uio_offset = 0;
auio.uio_resid = *buflen;
if ((ioflg & IO_NODELOCKED) == 0)
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
/* authorize attribute retrieval as kernel */
error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, NULL,
td);
if ((ioflg & IO_NODELOCKED) == 0)
VOP_UNLOCK(vp, 0, td);
if (error == 0) {
*buflen = *buflen - auio.uio_resid;
}
return (error);
}
/*
* XXX failure mode if partially written?
*/
int
vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
const char *attrname, int buflen, char *buf, struct thread *td)
{
struct uio auio;
struct iovec iov;
struct mount *mp;
int error;
iov.iov_len = buflen;
iov.iov_base = buf;
auio.uio_iov = &iov;
auio.uio_iovcnt = 1;
auio.uio_rw = UIO_WRITE;
auio.uio_segflg = UIO_SYSSPACE;
auio.uio_td = td;
auio.uio_offset = 0;
auio.uio_resid = buflen;
if ((ioflg & IO_NODELOCKED) == 0) {
if ((error = vn_start_write(vp, &mp, V_WAIT)) != 0)
return (error);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
}
/* authorize attribute setting as kernel */
error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio, NULL, td);
if ((ioflg & IO_NODELOCKED) == 0) {
vn_finished_write(mp);
VOP_UNLOCK(vp, 0, td);
}
return (error);
}
int
vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
const char *attrname, struct thread *td)
{
struct mount *mp;
int error;
if ((ioflg & IO_NODELOCKED) == 0) {
if ((error = vn_start_write(vp, &mp, V_WAIT)) != 0)
return (error);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
}
/* authorize attribute removal as kernel */
error = VOP_DELETEEXTATTR(vp, attrnamespace, attrname, NULL, td);
if (error == EOPNOTSUPP)
error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL,
NULL, td);
if ((ioflg & IO_NODELOCKED) == 0) {
vn_finished_write(mp);
VOP_UNLOCK(vp, 0, td);
}
return (error);
}
Index: head/sys/sys/vnode.h
===================================================================
--- head/sys/sys/vnode.h (revision 130100)
+++ head/sys/sys/vnode.h (revision 130101)
@@ -1,725 +1,725 @@
/*
* Copyright (c) 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.
* 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.
*
* @(#)vnode.h 8.7 (Berkeley) 2/4/94
* $FreeBSD$
*/
#ifndef _SYS_VNODE_H_
#define _SYS_VNODE_H_
/*
* XXX - compatability until lockmgr() goes away or all the #includes are
* updated.
*/
#include <sys/lockmgr.h>
#include <sys/queue.h>
#include <sys/_lock.h>
#include <sys/lock.h>
#include <sys/_mutex.h>
#include <sys/mutex.h>
#include <sys/selinfo.h>
#include <sys/uio.h>
#include <sys/acl.h>
#include <sys/ktr.h>
/*
* The vnode is the focus of all file activity in UNIX. There is a
* unique vnode allocated for each active file, each current directory,
* each mounted-on file, text file, and the root.
*/
/*
* Vnode types. VNON means no type.
*/
enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD };
/*
* Each underlying filesystem allocates its own private area and hangs
* it from v_data. If non-null, this area is freed in getnewvnode().
*/
TAILQ_HEAD(buflists, buf);
typedef int vop_t(void *);
struct namecache;
struct vpollinfo {
struct mtx vpi_lock; /* lock to protect below */
struct selinfo vpi_selinfo; /* identity of poller(s) */
short vpi_events; /* what they are looking for */
short vpi_revents; /* what has happened */
};
/*
* Reading or writing any of these items requires holding the appropriate lock.
*
* Lock reference:
* c - namecache mutex
* f - freelist mutex
* i - interlock
* m - mntvnodes mutex
* p - pollinfo lock
* s - spechash mutex
* S - syncer mutex
* u - Only a reference to the vnode is needed to read.
* v - vnode lock
*
* Vnodes may be found on many lists. The general way to deal with operating
* on a vnode that is on a list is:
* 1) Lock the list and find the vnode.
* 2) Lock interlock so that the vnode does not go away.
* 3) Unlock the list to avoid lock order reversals.
* 4) vget with LK_INTERLOCK and check for ENOENT, or
* 5) Check for XLOCK if the vnode lock is not required.
* 6) Perform your operation, then vput().
*
* XXX Not all fields are locked yet and some fields that are marked are not
* locked consistently. This is a work in progress. Requires Giant!
*/
struct vnode {
struct mtx v_interlock; /* lock for "i" things */
u_long v_iflag; /* i vnode flags (see below) */
int v_usecount; /* i ref count of users */
long v_numoutput; /* i writes in progress */
struct thread *v_vxthread; /* i thread owning VXLOCK */
int v_holdcnt; /* i page & buffer references */
struct buflists v_cleanblkhd; /* i SORTED clean blocklist */
struct buf *v_cleanblkroot; /* i clean buf splay tree */
int v_cleanbufcnt; /* i number of clean buffers */
struct buflists v_dirtyblkhd; /* i SORTED dirty blocklist */
struct buf *v_dirtyblkroot; /* i dirty buf splay tree */
int v_dirtybufcnt; /* i number of dirty buffers */
u_long v_vflag; /* v vnode flags */
int v_writecount; /* v ref count of writers */
struct vm_object *v_object; /* v Place to store VM object */
daddr_t v_lastw; /* v last write (write cluster) */
daddr_t v_cstart; /* v start block of cluster */
daddr_t v_lasta; /* v last allocation (cluster) */
int v_clen; /* v length of current cluster */
union {
struct mount *vu_mountedhere;/* v ptr to mounted vfs (VDIR) */
struct socket *vu_socket; /* v unix ipc (VSOCK) */
struct {
struct cdev *vu_cdev; /* v device (VCHR, VBLK) */
SLIST_ENTRY(vnode) vu_specnext; /* s device aliases */
} vu_spec;
struct fifoinfo *vu_fifoinfo; /* v fifo (VFIFO) */
} v_un;
TAILQ_ENTRY(vnode) v_freelist; /* f vnode freelist */
TAILQ_ENTRY(vnode) v_nmntvnodes; /* m vnodes for mount point */
LIST_ENTRY(vnode) v_synclist; /* S dirty vnode list */
enum vtype v_type; /* u vnode type */
const char *v_tag; /* u type of underlying data */
void *v_data; /* u private data for fs */
struct lock v_lock; /* u used if fs don't have one */
struct lock *v_vnlock; /* u pointer to vnode lock */
vop_t **v_op; /* u vnode operations vector */
struct mount *v_mount; /* u ptr to vfs we are in */
LIST_HEAD(, namecache) v_cache_src; /* c Cache entries from us */
TAILQ_HEAD(, namecache) v_cache_dst; /* c Cache entries to us */
u_long v_id; /* c capability identifier */
struct vnode *v_dd; /* c .. vnode */
u_long v_ddid; /* c .. capability identifier */
struct vpollinfo *v_pollinfo; /* p Poll events */
struct label *v_label; /* MAC label for vnode */
#ifdef DEBUG_LOCKS
const char *filename; /* Source file doing locking */
int line; /* Line number doing locking */
#endif
udev_t v_cachedfs; /* cached fs id */
ino_t v_cachedid; /* cached file id */
};
#define v_mountedhere v_un.vu_mountedhere
#define v_socket v_un.vu_socket
#define v_rdev v_un.vu_spec.vu_cdev
#define v_specnext v_un.vu_spec.vu_specnext
#define v_fifoinfo v_un.vu_fifoinfo
/*
* Userland version of struct vnode, for sysctl.
*/
struct xvnode {
size_t xv_size; /* sizeof(struct xvnode) */
void *xv_vnode; /* address of real vnode */
u_long xv_flag; /* vnode vflags */
int xv_usecount; /* reference count of users */
int xv_writecount; /* reference count of writers */
int xv_holdcnt; /* page & buffer references */
u_long xv_id; /* capability identifier */
void *xv_mount; /* address of parent mount */
long xv_numoutput; /* num of writes in progress */
enum vtype xv_type; /* vnode type */
union {
void *xvu_socket; /* socket, if VSOCK */
void *xvu_fifo; /* fifo, if VFIFO */
udev_t xvu_rdev; /* maj/min, if VBLK/VCHR */
struct {
udev_t xvu_dev; /* device, if VDIR/VREG/VLNK */
ino_t xvu_ino; /* id, if VDIR/VREG/VLNK */
} xv_uns;
} xv_un;
};
#define xv_socket xv_un.xvu_socket
#define xv_fifo xv_un.xvu_fifo
#define xv_rdev xv_un.xvu_rdev
#define xv_dev xv_un.xv_uns.xvu_dev
#define xv_ino xv_un.xv_uns.xvu_ino
#define VN_POLLEVENT(vp, events) \
do { \
if ((vp)->v_pollinfo != NULL && \
(vp)->v_pollinfo->vpi_events & (events)) \
vn_pollevent((vp), (events)); \
} while (0)
#define VN_KNOTE(vp, b) \
do { \
if ((vp)->v_pollinfo != NULL) \
KNOTE(&vp->v_pollinfo->vpi_selinfo.si_note, (b)); \
} while (0)
/*
* Vnode flags.
* VI flags are protected by interlock and live in v_iflag
* VV flags are protected by the vnode lock and live in v_vflag
*/
#define VI_XLOCK 0x0001 /* vnode is locked to change vtype */
#define VI_XWANT 0x0002 /* thread is waiting for vnode */
#define VI_BWAIT 0x0004 /* waiting for output to complete */
#define VI_OLOCK 0x0008 /* vnode is locked waiting for an object */
#define VI_OWANT 0x0010 /* a thread is waiting for VOLOCK */
#define VI_MOUNT 0x0020 /* Mount in progress */
#define VI_AGE 0x0040 /* Insert vnode at head of free list */
#define VI_DOOMED 0x0080 /* This vnode is being recycled */
#define VI_FREE 0x0100 /* This vnode is on the freelist */
#define VI_OBJDIRTY 0x0400 /* object might be dirty */
#define VI_DOINGINACT 0x0800 /* VOP_INACTIVE is in progress */
/*
* XXX VI_ONWORKLST could be replaced with a check for NULL list elements
* in v_synclist.
*/
#define VI_ONWORKLST 0x0200 /* On syncer work-list */
#define VV_ROOT 0x0001 /* root of its filesystem */
#define VV_ISTTY 0x0002 /* vnode represents a tty */
#define VV_NOSYNC 0x0004 /* unlinked, stop syncing */
#define VV_OBJBUF 0x0008 /* Allocate buffers in VM object */
#define VV_CACHEDLABEL 0x0010 /* Vnode has valid cached MAC label */
#define VV_TEXT 0x0020 /* vnode is a pure text prototype */
#define VV_COPYONWRITE 0x0040 /* vnode is doing copy-on-write */
#define VV_SYSTEM 0x0080 /* vnode being used by kernel */
#define VV_PROCDEP 0x0100 /* vnode is process dependent */
/*
* 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 */
short va_nlink; /* number of references to file */
uid_t va_uid; /* owner user id */
gid_t va_gid; /* owner group id */
udev_t va_fsid; /* filesystem id */
long 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 */
udev_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 */
};
/*
* Flags for va_vaflags.
*/
#define VA_UTIMES_NULL 0x01 /* utimes argument was NULL */
#define VA_EXCLUSIVE 0x02 /* exclusive create request */
/*
* Flags for ioflag. (high 16 bits used to ask for read-ahead and
* help with write clustering)
*/
#define IO_UNIT 0x0001 /* do I/O as atomic unit */
#define IO_APPEND 0x0002 /* append write to end */
#define IO_SYNC 0x0004 /* do I/O synchronously */
#define IO_NODELOCKED 0x0008 /* underlying node already locked */
#define IO_NDELAY 0x0010 /* FNDELAY flag set in file table */
#define IO_VMIO 0x0020 /* data already in VMIO space */
#define IO_INVAL 0x0040 /* invalidate after I/O */
#define IO_ASYNC 0x0080 /* bawrite rather then bdwrite */
#define IO_DIRECT 0x0100 /* attempt to bypass buffer cache */
#define IO_EXT 0x0400 /* operate on external attributes */
#define IO_NORMAL 0x0800 /* operate on regular data */
#define IO_NOMACCHECK 0x1000 /* MAC checks unnecessary */
#define IO_SEQMAX 0x7F /* seq heuristic max value */
#define IO_SEQSHIFT 16 /* seq heuristic in upper 16 bits */
/*
* Modes. Some values same as Ixxx entries from inode.h for now.
*/
#define VEXEC 000100 /* execute/search permission */
#define VWRITE 000200 /* write permission */
#define VREAD 000400 /* read permission */
#define VSVTX 001000 /* save swapped text even after use */
#define VSGID 002000 /* set group id on execution */
#define VSUID 004000 /* set user id on execution */
#define VADMIN 010000 /* permission to administer */
#define VSTAT 020000 /* permission to retrieve attrs */
#define VAPPEND 040000 /* permission to write/append */
#define VALLPERM (VEXEC | VWRITE | VREAD | VADMIN | VSTAT | VAPPEND)
/*
* Token indicating no attribute value yet assigned.
*/
#define VNOVAL (-1)
/*
* LK_TIMELOCK timeout for vnode locks (used mainly by the pageout daemon)
*/
#define VLKTIMEOUT (hz / 20 + 1)
#ifdef _KERNEL
#ifdef MALLOC_DECLARE
MALLOC_DECLARE(M_VNODE);
#endif
/*
* Convert between vnode types and inode formats (since POSIX.1
* defines mode word of stat structure in terms of inode formats).
*/
extern enum vtype iftovt_tab[];
extern int vttoif_tab[];
#define IFTOVT(mode) (iftovt_tab[((mode) & S_IFMT) >> 12])
#define VTTOIF(indx) (vttoif_tab[(int)(indx)])
#define MAKEIMODE(indx, mode) (int)(VTTOIF(indx) | (mode))
/*
* Flags to various vnode functions.
*/
#define SKIPSYSTEM 0x0001 /* vflush: skip vnodes marked VSYSTEM */
#define FORCECLOSE 0x0002 /* vflush: force file closure */
#define WRITECLOSE 0x0004 /* vflush: only close writable files */
#define DOCLOSE 0x0008 /* vclean: close active files */
#define V_SAVE 0x0001 /* vinvalbuf: sync file first */
#define V_ALT 0x0002 /* vinvalbuf: invalidate only alternate bufs */
#define V_NORMAL 0x0004 /* vinvalbuf: invalidate only regular bufs */
#define REVOKEALL 0x0001 /* vop_revoke: revoke all aliases */
#define V_WAIT 0x0001 /* vn_start_write: sleep for suspend */
#define V_NOWAIT 0x0002 /* vn_start_write: don't sleep for suspend */
#define V_XSLEEP 0x0004 /* vn_start_write: just return after sleep */
#define VREF(vp) vref(vp)
#ifdef DIAGNOSTIC
#define VATTR_NULL(vap) vattr_null(vap)
#else
#define VATTR_NULL(vap) (*(vap) = va_null) /* initialize a vattr */
#endif /* DIAGNOSTIC */
#define NULLVP ((struct vnode *)NULL)
#define VNODEOP_SET(f) \
C_SYSINIT(f##init, SI_SUB_VFS, SI_ORDER_SECOND, vfs_add_vnodeops, &f); \
C_SYSUNINIT(f##uninit, SI_SUB_VFS, SI_ORDER_SECOND, vfs_rm_vnodeops, &f);
/*
* Global vnode data.
*/
extern struct vnode *rootvnode; /* root (i.e. "/") vnode */
extern int async_io_version; /* 0 or POSIX version of AIO i'face */
extern int desiredvnodes; /* number of vnodes desired */
extern struct uma_zone *namei_zone;
extern int prtactive; /* nonzero to call vprint() */
extern struct vattr va_null; /* predefined null vattr structure */
/*
* Macro/function to check for client cache inconsistency w.r.t. leasing.
*/
#define LEASE_READ 0x1 /* Check lease for readers */
#define LEASE_WRITE 0x2 /* Check lease for modifiers */
extern void (*lease_updatetime)(int deltat);
/* Requires interlock. */
#define VSHOULDFREE(vp) \
(!((vp)->v_iflag & (VI_FREE|VI_DOOMED|VI_DOINGINACT)) && \
!(vp)->v_holdcnt && !(vp)->v_usecount && \
(!(vp)->v_object || \
!((vp)->v_object->ref_count || (vp)->v_object->resident_page_count)))
/* Requires interlock. */
#define VMIGHTFREE(vp) \
(!((vp)->v_iflag & (VI_FREE|VI_DOOMED|VI_XLOCK|VI_DOINGINACT)) && \
LIST_EMPTY(&(vp)->v_cache_src) && !(vp)->v_usecount)
/* Requires interlock. */
#define VSHOULDBUSY(vp) \
(((vp)->v_iflag & VI_FREE) && \
((vp)->v_holdcnt || (vp)->v_usecount))
#define VI_LOCK(vp) mtx_lock(&(vp)->v_interlock)
#define VI_TRYLOCK(vp) mtx_trylock(&(vp)->v_interlock)
#define VI_UNLOCK(vp) mtx_unlock(&(vp)->v_interlock)
#define VI_MTX(vp) (&(vp)->v_interlock)
#endif /* _KERNEL */
/*
* Mods for extensibility.
*/
/*
* Flags for vdesc_flags:
*/
#define VDESC_MAX_VPS 16
/* Low order 16 flag bits are reserved for willrele flags for vp arguments. */
#define VDESC_VP0_WILLRELE 0x0001
#define VDESC_VP1_WILLRELE 0x0002
#define VDESC_VP2_WILLRELE 0x0004
#define VDESC_VP3_WILLRELE 0x0008
#define VDESC_NOMAP_VPP 0x0100
#define VDESC_VPP_WILLRELE 0x0200
/*
* VDESC_NO_OFFSET is used to identify the end of the offset list
* and in places where no such field exists.
*/
#define VDESC_NO_OFFSET -1
/*
* This structure describes the vnode operation taking place.
*/
struct vnodeop_desc {
int vdesc_offset; /* offset in vector,first for speed */
char *vdesc_name; /* a readable name for debugging */
int vdesc_flags; /* VDESC_* flags */
/*
* These ops are used by bypass routines to map and locate arguments.
* Creds and procs are not needed in bypass routines, but sometimes
* they are useful to (for example) transport layers.
* Nameidata is useful because it has a cred in it.
*/
int *vdesc_vp_offsets; /* list ended by VDESC_NO_OFFSET */
int vdesc_vpp_offset; /* return vpp location */
int vdesc_cred_offset; /* cred location, if any */
int vdesc_thread_offset; /* thread location, if any */
int vdesc_componentname_offset; /* if any */
/*
* Finally, we've got a list of private data (about each operation)
* for each transport layer. (Support to manage this list is not
* yet part of BSD.)
*/
caddr_t *vdesc_transports;
};
#ifdef _KERNEL
/*
* A list of all the operation descs.
*/
extern struct vnodeop_desc *vnodeop_descs[];
#define VOPARG_OFFSETOF(s_type, field) __offsetof(s_type, field)
#define VOPARG_OFFSETTO(s_type, s_offset, struct_p) \
((s_type)(((char*)(struct_p)) + (s_offset)))
/*
* This structure is used to configure the new vnodeops vector.
*/
struct vnodeopv_entry_desc {
struct vnodeop_desc *opve_op; /* which operation this is */
vop_t *opve_impl; /* code implementing this operation */
};
struct vnodeopv_desc {
/* ptr to the ptr to the vector where op should go */
vop_t ***opv_desc_vector_p;
struct vnodeopv_entry_desc *opv_desc_ops; /* null terminated list */
};
/*
* A generic structure.
* This can be used by bypass routines to identify generic arguments.
*/
struct vop_generic_args {
struct vnodeop_desc *a_desc;
/* other random data follows, presumably */
};
#ifdef DEBUG_VFS_LOCKS
/*
* Support code to aid in debugging VFS locking problems. Not totally
* reliable since if the thread sleeps between changing the lock
* state and checking it with the assert, some other thread could
* change the state. They are good enough for debugging a single
* filesystem using a single-threaded test.
*/
void assert_vi_locked(struct vnode *vp, const char *str);
void assert_vi_unlocked(struct vnode *vp, const char *str);
#if 0
void assert_vop_elocked(struct vnode *vp, const char *str);
void assert_vop_elocked_other(struct vnode *vp, const char *str);
#endif
void assert_vop_locked(struct vnode *vp, const char *str);
#if 0
voi0 assert_vop_slocked(struct vnode *vp, const char *str);
#endif
void assert_vop_unlocked(struct vnode *vp, const char *str);
/* These are called from within the actual VOPS. */
void vop_lock_pre(void *a);
void vop_lock_post(void *a, int rc);
void vop_lookup_post(void *a, int rc);
void vop_lookup_pre(void *a);
void vop_rename_pre(void *a);
void vop_strategy_pre(void *a);
void vop_unlock_post(void *a, int rc);
void vop_unlock_pre(void *a);
#define ASSERT_VI_LOCKED(vp, str) assert_vi_locked((vp), (str))
#define ASSERT_VI_UNLOCKED(vp, str) assert_vi_unlocked((vp), (str))
#if 0
#define ASSERT_VOP_ELOCKED(vp, str) assert_vop_elocked((vp), (str))
#define ASSERT_VOP_ELOCKED_OTHER(vp, str) assert_vop_locked_other((vp), (str))
#endif
#define ASSERT_VOP_LOCKED(vp, str) assert_vop_locked((vp), (str))
#if 0
#define ASSERT_VOP_SLOCKED(vp, str) assert_vop_slocked((vp), (str))
#endif
#define ASSERT_VOP_UNLOCKED(vp, str) assert_vop_unlocked((vp), (str))
#else /* !DEBUG_VFS_LOCKS */
#define ASSERT_VI_LOCKED(vp, str)
#define ASSERT_VI_UNLOCKED(vp, str)
#if 0
#define ASSERT_VOP_ELOCKED(vp, str)
#define ASSERT_VOP_ELOCKED_OTHER(vp, str)
#endif
#define ASSERT_VOP_LOCKED(vp, str)
#if 0
#define ASSERT_VOP_SLOCKED(vp, str)
#endif
#define ASSERT_VOP_UNLOCKED(vp, str)
#endif /* DEBUG_VFS_LOCKS */
/*
* VOCALL calls an op given an ops vector. We break it out because BSD's
* vclean changes the ops vector and then wants to call ops with the old
* vector.
*/
#define VOCALL(OPSV,OFF,AP) (( *((OPSV)[(OFF)])) (AP))
/*
* This call works for vnodes in the kernel.
*/
#define VCALL(VP,OFF,AP) VOCALL((VP)->v_op,(OFF),(AP))
#define VDESC(OP) (& __CONCAT(OP,_desc))
#define VOFFSET(OP) (VDESC(OP)->vdesc_offset)
/*
* VMIO support inline
*/
extern int vmiodirenable;
static __inline int
vn_canvmio(struct vnode *vp)
{
if (vp && (vp->v_type == VREG || (vmiodirenable && vp->v_type == VDIR)))
return(TRUE);
return(FALSE);
}
/*
* Finally, include the default set of vnode operations.
*/
#include "vnode_if.h"
/*
* Public vnode manipulation functions.
*/
struct componentname;
struct file;
struct mount;
struct nameidata;
struct ostat;
struct thread;
struct proc;
struct stat;
struct nstat;
struct ucred;
struct uio;
struct vattr;
struct vnode;
extern int (*lease_check_hook)(struct vop_lease_args *);
extern int (*softdep_fsync_hook)(struct vnode *);
extern int (*softdep_process_worklist_hook)(struct mount *);
struct vnode *addaliasu(struct vnode *vp, udev_t nvp_rdev);
int bdevvp(dev_t dev, struct vnode **vpp);
/* cache_* may belong in namei.h. */
void cache_enter(struct vnode *dvp, struct vnode *vp,
struct componentname *cnp);
int cache_lookup(struct vnode *dvp, struct vnode **vpp,
struct componentname *cnp);
void cache_purge(struct vnode *vp);
void cache_purgevfs(struct mount *mp);
int cache_leaf_test(struct vnode *vp);
int change_dir(struct vnode *vp, struct thread *td);
int change_root(struct vnode *vp, struct thread *td);
void cvtstat(struct stat *st, struct ostat *ost);
void cvtnstat(struct stat *sb, struct nstat *nsb);
int getnewvnode(const char *tag, struct mount *mp, vop_t **vops,
struct vnode **vpp);
int lease_check(struct vop_lease_args *ap);
int spec_vnoperate(struct vop_generic_args *);
int speedup_syncer(void);
#define textvp_fullpath(p, rb, rfb) \
vn_fullpath(FIRST_THREAD_IN_PROC(p), (p)->p_textvp, rb, rfb)
int vn_fullpath(struct thread *td, struct vnode *vn,
char **retbuf, char **freebuf);
int vaccess(enum vtype type, mode_t file_mode, uid_t uid, gid_t gid,
mode_t acc_mode, struct ucred *cred, int *privused);
int vaccess_acl_posix1e(enum vtype type, uid_t file_uid,
gid_t file_gid, struct acl *acl, mode_t acc_mode,
struct ucred *cred, int *privused);
void vattr_null(struct vattr *vap);
int vcount(struct vnode *vp);
void vdrop(struct vnode *);
void vdropl(struct vnode *);
int vfinddev(dev_t dev, struct vnode **vpp);
void vfs_add_vnodeops(const void *);
void vfs_rm_vnodeops(const void *);
int vflush(struct mount *mp, int rootrefs, int flags);
int vget(struct vnode *vp, int lockflag, struct thread *td);
void vgone(struct vnode *vp);
void vgonel(struct vnode *vp, struct thread *td);
void vhold(struct vnode *);
void vholdl(struct vnode *);
int vinvalbuf(struct vnode *vp, int save, struct ucred *cred,
struct thread *td, int slpflag, int slptimeo);
int vtruncbuf(struct vnode *vp, struct ucred *cred, struct thread *td,
off_t length, int blksize);
void vprint(char *label, struct vnode *vp);
int vrecycle(struct vnode *vp, struct mtx *inter_lkp,
struct thread *td);
int vn_close(struct vnode *vp,
int flags, struct ucred *file_cred, struct thread *td);
void vn_finished_write(struct mount *mp);
int vn_isdisk(struct vnode *vp, int *errp);
int vn_lock(struct vnode *vp, int flags, struct thread *td);
#ifdef DEBUG_LOCKS
int debug_vn_lock(struct vnode *vp, int flags, struct thread *p,
const char *filename, int line);
#define vn_lock(vp,flags,p) debug_vn_lock(vp,flags,p,__FILE__,__LINE__)
#endif
int vn_open(struct nameidata *ndp, int *flagp, int cmode, int fdidx);
int vn_open_cred(struct nameidata *ndp, int *flagp, int cmode,
struct ucred *cred, int fdidx);
void vn_pollevent(struct vnode *vp, int events);
void vn_pollgone(struct vnode *vp);
int vn_pollrecord(struct vnode *vp, struct thread *p, int events);
int vn_rdwr(enum uio_rw rw, struct vnode *vp, caddr_t base,
int len, off_t offset, enum uio_seg segflg, int ioflg,
struct ucred *active_cred, struct ucred *file_cred, int *aresid,
struct thread *td);
int vn_rdwr_inchunks(enum uio_rw rw, struct vnode *vp, caddr_t base,
- int len, off_t offset, enum uio_seg segflg, int ioflg,
- struct ucred *active_cred, struct ucred *file_cred, int *aresid,
+ size_t len, off_t offset, enum uio_seg segflg, int ioflg,
+ struct ucred *active_cred, struct ucred *file_cred, size_t *aresid,
struct thread *td);
int vn_stat(struct vnode *vp, struct stat *sb, struct ucred *active_cred,
struct ucred *file_cred, struct thread *td);
int vn_start_write(struct vnode *vp, struct mount **mpp, int flags);
dev_t vn_todev(struct vnode *vp);
int vn_write_suspend_wait(struct vnode *vp, struct mount *mp,
int flags);
int vn_writechk(struct vnode *vp);
int vn_extattr_get(struct vnode *vp, int ioflg, int attrnamespace,
const char *attrname, int *buflen, char *buf, struct thread *td);
int vn_extattr_set(struct vnode *vp, int ioflg, int attrnamespace,
const char *attrname, int buflen, char *buf, struct thread *td);
int vn_extattr_rm(struct vnode *vp, int ioflg, int attrnamespace,
const char *attrname, struct thread *td);
int vfs_cache_lookup(struct vop_lookup_args *ap);
int vfs_object_create(struct vnode *vp, struct thread *td,
struct ucred *cred);
void vfs_timestamp(struct timespec *);
void vfs_write_resume(struct mount *mp);
int vfs_write_suspend(struct mount *mp);
int vop_stdbmap(struct vop_bmap_args *);
int vop_stdfsync(struct vop_fsync_args *);
int vop_stdgetwritemount(struct vop_getwritemount_args *);
int vop_stdgetpages(struct vop_getpages_args *);
int vop_stdinactive(struct vop_inactive_args *);
int vop_stdislocked(struct vop_islocked_args *);
int vop_stdlock(struct vop_lock_args *);
int vop_stdputpages(struct vop_putpages_args *);
int vop_stdunlock(struct vop_unlock_args *);
int vop_nopoll(struct vop_poll_args *);
int vop_stdpathconf(struct vop_pathconf_args *);
int vop_stdpoll(struct vop_poll_args *);
int vop_revoke(struct vop_revoke_args *);
int vop_eopnotsupp(struct vop_generic_args *ap);
int vop_ebadf(struct vop_generic_args *ap);
int vop_einval(struct vop_generic_args *ap);
int vop_enotty(struct vop_generic_args *ap);
int vop_defaultop(struct vop_generic_args *ap);
int vop_null(struct vop_generic_args *ap);
int vop_panic(struct vop_generic_args *ap);
int vop_stdcreatevobject(struct vop_createvobject_args *ap);
int vop_stddestroyvobject(struct vop_destroyvobject_args *ap);
int vop_stdgetvobject(struct vop_getvobject_args *ap);
void vfree(struct vnode *);
void vput(struct vnode *vp);
void vrele(struct vnode *vp);
void vref(struct vnode *vp);
int vrefcnt(struct vnode *vp);
void vbusy(struct vnode *vp);
void v_addpollinfo(struct vnode *vp);
extern vop_t **default_vnodeop_p;
extern vop_t **spec_vnodeop_p;
extern vop_t **dead_vnodeop_p;
#endif /* _KERNEL */
#endif /* !_SYS_VNODE_H_ */

File Metadata

Mime Type
application/octet-stream
Expires
Wed, Sep 2, 7:40 PM (2 d)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
37851447
Default Alt Text
(147 KB)

Event Timeline