Index: head/sys/compat/cloudabi/cloudabi_fd.c =================================================================== --- head/sys/compat/cloudabi/cloudabi_fd.c (revision 286020) +++ head/sys/compat/cloudabi/cloudabi_fd.c (revision 286021) @@ -1,496 +1,510 @@ /*- * Copyright (c) 2015 Nuxi, https://nuxi.nl/ * * 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 __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include /* Translation between CloudABI and Capsicum rights. */ #define RIGHTS_MAPPINGS \ MAPPING(CLOUDABI_RIGHT_FD_DATASYNC, CAP_FSYNC) \ MAPPING(CLOUDABI_RIGHT_FD_READ, CAP_READ) \ MAPPING(CLOUDABI_RIGHT_FD_SEEK, CAP_SEEK) \ MAPPING(CLOUDABI_RIGHT_FD_STAT_PUT_FLAGS, CAP_FCNTL) \ MAPPING(CLOUDABI_RIGHT_FD_SYNC, CAP_FSYNC) \ MAPPING(CLOUDABI_RIGHT_FD_TELL, CAP_SEEK_TELL) \ MAPPING(CLOUDABI_RIGHT_FD_WRITE, CAP_WRITE) \ MAPPING(CLOUDABI_RIGHT_FILE_ADVISE) \ MAPPING(CLOUDABI_RIGHT_FILE_ALLOCATE, CAP_WRITE) \ MAPPING(CLOUDABI_RIGHT_FILE_CREATE_DIRECTORY, CAP_MKDIRAT) \ MAPPING(CLOUDABI_RIGHT_FILE_CREATE_FILE, CAP_CREATE) \ MAPPING(CLOUDABI_RIGHT_FILE_CREATE_FIFO, CAP_MKFIFOAT) \ MAPPING(CLOUDABI_RIGHT_FILE_LINK_SOURCE, CAP_LOOKUP) \ MAPPING(CLOUDABI_RIGHT_FILE_LINK_TARGET, CAP_LINKAT) \ MAPPING(CLOUDABI_RIGHT_FILE_OPEN, CAP_LOOKUP) \ MAPPING(CLOUDABI_RIGHT_FILE_READDIR, CAP_READ) \ MAPPING(CLOUDABI_RIGHT_FILE_READLINK, CAP_LOOKUP) \ MAPPING(CLOUDABI_RIGHT_FILE_RENAME_SOURCE, CAP_RENAMEAT) \ MAPPING(CLOUDABI_RIGHT_FILE_RENAME_TARGET, CAP_LINKAT) \ MAPPING(CLOUDABI_RIGHT_FILE_STAT_FGET, CAP_FSTAT) \ MAPPING(CLOUDABI_RIGHT_FILE_STAT_FPUT_SIZE, CAP_FTRUNCATE) \ MAPPING(CLOUDABI_RIGHT_FILE_STAT_FPUT_TIMES, CAP_FUTIMES) \ MAPPING(CLOUDABI_RIGHT_FILE_STAT_GET, CAP_FSTATAT) \ MAPPING(CLOUDABI_RIGHT_FILE_STAT_PUT_TIMES, CAP_FUTIMESAT) \ MAPPING(CLOUDABI_RIGHT_FILE_SYMLINK, CAP_SYMLINKAT) \ MAPPING(CLOUDABI_RIGHT_FILE_UNLINK, CAP_UNLINKAT) \ MAPPING(CLOUDABI_RIGHT_MEM_MAP, CAP_MMAP) \ MAPPING(CLOUDABI_RIGHT_MEM_MAP_EXEC, CAP_MMAP_X) \ MAPPING(CLOUDABI_RIGHT_POLL_FD_READWRITE, CAP_EVENT) \ MAPPING(CLOUDABI_RIGHT_POLL_MODIFY, CAP_KQUEUE_CHANGE) \ MAPPING(CLOUDABI_RIGHT_POLL_PROC_TERMINATE, CAP_PDWAIT) \ MAPPING(CLOUDABI_RIGHT_POLL_WAIT, CAP_KQUEUE_EVENT) \ MAPPING(CLOUDABI_RIGHT_PROC_EXEC, CAP_FEXECVE) \ MAPPING(CLOUDABI_RIGHT_SOCK_ACCEPT, CAP_ACCEPT) \ MAPPING(CLOUDABI_RIGHT_SOCK_BIND_DIRECTORY, CAP_BINDAT) \ MAPPING(CLOUDABI_RIGHT_SOCK_BIND_SOCKET, CAP_BIND) \ MAPPING(CLOUDABI_RIGHT_SOCK_CONNECT_DIRECTORY, CAP_CONNECTAT) \ MAPPING(CLOUDABI_RIGHT_SOCK_CONNECT_SOCKET, CAP_CONNECT) \ MAPPING(CLOUDABI_RIGHT_SOCK_LISTEN, CAP_LISTEN) \ MAPPING(CLOUDABI_RIGHT_SOCK_SHUTDOWN, CAP_SHUTDOWN) \ MAPPING(CLOUDABI_RIGHT_SOCK_STAT_GET, CAP_GETPEERNAME, \ CAP_GETSOCKNAME, CAP_GETSOCKOPT) int cloudabi_sys_fd_close(struct thread *td, struct cloudabi_sys_fd_close_args *uap) { return (kern_close(td, uap->fd)); } int cloudabi_sys_fd_create1(struct thread *td, struct cloudabi_sys_fd_create1_args *uap) { struct socket_args socket_args = { .domain = AF_UNIX, }; switch (uap->type) { case CLOUDABI_FILETYPE_SOCKET_DGRAM: socket_args.type = SOCK_DGRAM; return (sys_socket(td, &socket_args)); case CLOUDABI_FILETYPE_SOCKET_SEQPACKET: socket_args.type = SOCK_SEQPACKET; return (sys_socket(td, &socket_args)); case CLOUDABI_FILETYPE_SOCKET_STREAM: socket_args.type = SOCK_STREAM; return (sys_socket(td, &socket_args)); default: return (EINVAL); } } int cloudabi_sys_fd_create2(struct thread *td, struct cloudabi_sys_fd_create2_args *uap) { + struct filecaps fcaps1 = {}, fcaps2 = {}; int fds[2]; int error; switch (uap->type) { + case CLOUDABI_FILETYPE_FIFO: + /* + * CloudABI pipes are unidirectional. Restrict rights on + * the pipe to simulate this. + */ + cap_rights_init(&fcaps1.fc_rights, CAP_EVENT, CAP_FCNTL, + CAP_FSTAT, CAP_READ); + fcaps1.fc_fcntls = CAP_FCNTL_SETFL; + cap_rights_init(&fcaps2.fc_rights, CAP_EVENT, CAP_FCNTL, + CAP_FSTAT, CAP_WRITE); + fcaps2.fc_fcntls = CAP_FCNTL_SETFL; + error = kern_pipe(td, fds, 0, &fcaps1, &fcaps2); + break; case CLOUDABI_FILETYPE_SOCKET_DGRAM: error = kern_socketpair(td, AF_UNIX, SOCK_DGRAM, 0, fds); break; case CLOUDABI_FILETYPE_SOCKET_SEQPACKET: error = kern_socketpair(td, AF_UNIX, SOCK_SEQPACKET, 0, fds); break; case CLOUDABI_FILETYPE_SOCKET_STREAM: error = kern_socketpair(td, AF_UNIX, SOCK_STREAM, 0, fds); break; default: return (EINVAL); } if (error == 0) { td->td_retval[0] = fds[0]; td->td_retval[1] = fds[1]; } return (0); } int cloudabi_sys_fd_datasync(struct thread *td, struct cloudabi_sys_fd_datasync_args *uap) { struct fsync_args fsync_args = { .fd = uap->fd }; /* Call into fsync(), as FreeBSD lacks fdatasync(). */ return (sys_fsync(td, &fsync_args)); } int cloudabi_sys_fd_dup(struct thread *td, struct cloudabi_sys_fd_dup_args *uap) { return (kern_dup(td, FDDUP_NORMAL, 0, uap->from, 0)); } int cloudabi_sys_fd_replace(struct thread *td, struct cloudabi_sys_fd_replace_args *uap) { int error; /* * CloudABI's equivalent to dup2(). CloudABI processes should * not depend on hardcoded file descriptor layouts, but simply * use the file descriptor numbers that are allocated by the * kernel. Duplicating file descriptors to arbitrary numbers * should not be done. * * Invoke kern_dup() with FDDUP_MUSTREPLACE, so that we return * EBADF when duplicating to a nonexistent file descriptor. Also * clear the return value, as this system call yields no return * value. */ error = kern_dup(td, FDDUP_MUSTREPLACE, 0, uap->from, uap->to); td->td_retval[0] = 0; return (error); } int cloudabi_sys_fd_seek(struct thread *td, struct cloudabi_sys_fd_seek_args *uap) { struct lseek_args lseek_args = { .fd = uap->fd, .offset = uap->offset }; switch (uap->whence) { case CLOUDABI_WHENCE_CUR: lseek_args.whence = SEEK_CUR; break; case CLOUDABI_WHENCE_END: lseek_args.whence = SEEK_END; break; case CLOUDABI_WHENCE_SET: lseek_args.whence = SEEK_SET; break; default: return (EINVAL); } return (sys_lseek(td, &lseek_args)); } /* Converts a file descriptor to a CloudABI file descriptor type. */ cloudabi_filetype_t cloudabi_convert_filetype(const struct file *fp) { struct socket *so; struct vnode *vp; switch (fp->f_type) { case DTYPE_FIFO: return (CLOUDABI_FILETYPE_FIFO); case DTYPE_KQUEUE: return (CLOUDABI_FILETYPE_POLL); case DTYPE_PIPE: return (CLOUDABI_FILETYPE_FIFO); case DTYPE_PROCDESC: return (CLOUDABI_FILETYPE_PROCESS); case DTYPE_SHM: return (CLOUDABI_FILETYPE_SHARED_MEMORY); case DTYPE_SOCKET: so = fp->f_data; switch (so->so_type) { case SOCK_DGRAM: return (CLOUDABI_FILETYPE_SOCKET_DGRAM); case SOCK_SEQPACKET: return (CLOUDABI_FILETYPE_SOCKET_SEQPACKET); case SOCK_STREAM: return (CLOUDABI_FILETYPE_SOCKET_STREAM); default: return (CLOUDABI_FILETYPE_UNKNOWN); } case DTYPE_VNODE: vp = fp->f_vnode; switch (vp->v_type) { case VBLK: return (CLOUDABI_FILETYPE_BLOCK_DEVICE); case VCHR: return (CLOUDABI_FILETYPE_CHARACTER_DEVICE); case VDIR: return (CLOUDABI_FILETYPE_DIRECTORY); case VFIFO: return (CLOUDABI_FILETYPE_FIFO); case VLNK: return (CLOUDABI_FILETYPE_SYMBOLIC_LINK); case VREG: return (CLOUDABI_FILETYPE_REGULAR_FILE); case VSOCK: return (CLOUDABI_FILETYPE_SOCKET_STREAM); default: return (CLOUDABI_FILETYPE_UNKNOWN); } default: return (CLOUDABI_FILETYPE_UNKNOWN); } } /* Removes rights that conflict with the file descriptor type. */ static void cloudabi_remove_conflicting_rights(cloudabi_filetype_t filetype, cloudabi_rights_t *base, cloudabi_rights_t *inheriting) { /* * CloudABI has a small number of additional rights bits to * disambiguate between multiple purposes. Remove the bits that * don't apply to the type of the file descriptor. * * As file descriptor access modes (O_ACCMODE) has been fully * replaced by rights bits, CloudABI distinguishes between * rights that apply to the file descriptor itself (base) versus * rights of new file descriptors derived from them * (inheriting). The code below approximates the pair by * decomposing depending on the file descriptor type. * * We need to be somewhat accurate about which actions can * actually be performed on the file descriptor, as functions * like fcntl(fd, F_GETFL) are emulated on top of this. */ switch (filetype) { case CLOUDABI_FILETYPE_DIRECTORY: *base &= CLOUDABI_RIGHT_FD_STAT_PUT_FLAGS | CLOUDABI_RIGHT_FD_SYNC | CLOUDABI_RIGHT_FILE_ADVISE | CLOUDABI_RIGHT_FILE_CREATE_DIRECTORY | CLOUDABI_RIGHT_FILE_CREATE_FILE | CLOUDABI_RIGHT_FILE_CREATE_FIFO | CLOUDABI_RIGHT_FILE_LINK_SOURCE | CLOUDABI_RIGHT_FILE_LINK_TARGET | CLOUDABI_RIGHT_FILE_OPEN | CLOUDABI_RIGHT_FILE_READDIR | CLOUDABI_RIGHT_FILE_READLINK | CLOUDABI_RIGHT_FILE_RENAME_SOURCE | CLOUDABI_RIGHT_FILE_RENAME_TARGET | CLOUDABI_RIGHT_FILE_STAT_FGET | CLOUDABI_RIGHT_FILE_STAT_FPUT_TIMES | CLOUDABI_RIGHT_FILE_STAT_GET | CLOUDABI_RIGHT_FILE_STAT_PUT_TIMES | CLOUDABI_RIGHT_FILE_SYMLINK | CLOUDABI_RIGHT_FILE_UNLINK | CLOUDABI_RIGHT_POLL_FD_READWRITE | CLOUDABI_RIGHT_SOCK_BIND_DIRECTORY | CLOUDABI_RIGHT_SOCK_CONNECT_DIRECTORY; *inheriting &= CLOUDABI_RIGHT_FD_DATASYNC | CLOUDABI_RIGHT_FD_READ | CLOUDABI_RIGHT_FD_SEEK | CLOUDABI_RIGHT_FD_STAT_PUT_FLAGS | CLOUDABI_RIGHT_FD_SYNC | CLOUDABI_RIGHT_FD_TELL | CLOUDABI_RIGHT_FD_WRITE | CLOUDABI_RIGHT_FILE_ADVISE | CLOUDABI_RIGHT_FILE_ALLOCATE | CLOUDABI_RIGHT_FILE_CREATE_DIRECTORY | CLOUDABI_RIGHT_FILE_CREATE_FILE | CLOUDABI_RIGHT_FILE_CREATE_FIFO | CLOUDABI_RIGHT_FILE_LINK_SOURCE | CLOUDABI_RIGHT_FILE_LINK_TARGET | CLOUDABI_RIGHT_FILE_OPEN | CLOUDABI_RIGHT_FILE_READDIR | CLOUDABI_RIGHT_FILE_READLINK | CLOUDABI_RIGHT_FILE_RENAME_SOURCE | CLOUDABI_RIGHT_FILE_RENAME_TARGET | CLOUDABI_RIGHT_FILE_STAT_FGET | CLOUDABI_RIGHT_FILE_STAT_FPUT_SIZE | CLOUDABI_RIGHT_FILE_STAT_FPUT_TIMES | CLOUDABI_RIGHT_FILE_STAT_GET | CLOUDABI_RIGHT_FILE_STAT_PUT_TIMES | CLOUDABI_RIGHT_FILE_SYMLINK | CLOUDABI_RIGHT_FILE_UNLINK | CLOUDABI_RIGHT_MEM_MAP | CLOUDABI_RIGHT_MEM_MAP_EXEC | CLOUDABI_RIGHT_POLL_FD_READWRITE | CLOUDABI_RIGHT_PROC_EXEC | CLOUDABI_RIGHT_SOCK_BIND_DIRECTORY | CLOUDABI_RIGHT_SOCK_CONNECT_DIRECTORY; break; case CLOUDABI_FILETYPE_FIFO: *base &= CLOUDABI_RIGHT_FD_READ | CLOUDABI_RIGHT_FD_STAT_PUT_FLAGS | CLOUDABI_RIGHT_FD_WRITE | CLOUDABI_RIGHT_FILE_STAT_FGET | CLOUDABI_RIGHT_POLL_FD_READWRITE; *inheriting = 0; break; case CLOUDABI_FILETYPE_POLL: *base &= ~CLOUDABI_RIGHT_FILE_ADVISE; *inheriting = 0; break; case CLOUDABI_FILETYPE_PROCESS: *base &= ~CLOUDABI_RIGHT_FILE_ADVISE; *inheriting = 0; break; case CLOUDABI_FILETYPE_REGULAR_FILE: *base &= CLOUDABI_RIGHT_FD_DATASYNC | CLOUDABI_RIGHT_FD_READ | CLOUDABI_RIGHT_FD_SEEK | CLOUDABI_RIGHT_FD_STAT_PUT_FLAGS | CLOUDABI_RIGHT_FD_SYNC | CLOUDABI_RIGHT_FD_TELL | CLOUDABI_RIGHT_FD_WRITE | CLOUDABI_RIGHT_FILE_ADVISE | CLOUDABI_RIGHT_FILE_ALLOCATE | CLOUDABI_RIGHT_FILE_STAT_FGET | CLOUDABI_RIGHT_FILE_STAT_FPUT_SIZE | CLOUDABI_RIGHT_FILE_STAT_FPUT_TIMES | CLOUDABI_RIGHT_MEM_MAP | CLOUDABI_RIGHT_MEM_MAP_EXEC | CLOUDABI_RIGHT_POLL_FD_READWRITE | CLOUDABI_RIGHT_PROC_EXEC; *inheriting = 0; break; case CLOUDABI_FILETYPE_SHARED_MEMORY: *base &= ~(CLOUDABI_RIGHT_FD_SEEK | CLOUDABI_RIGHT_FD_TELL | CLOUDABI_RIGHT_FILE_ADVISE | CLOUDABI_RIGHT_FILE_ALLOCATE | CLOUDABI_RIGHT_FILE_READDIR); *inheriting = 0; break; case CLOUDABI_FILETYPE_SOCKET_DGRAM: case CLOUDABI_FILETYPE_SOCKET_SEQPACKET: case CLOUDABI_FILETYPE_SOCKET_STREAM: *base &= CLOUDABI_RIGHT_FD_READ | CLOUDABI_RIGHT_FD_STAT_PUT_FLAGS | CLOUDABI_RIGHT_FD_WRITE | CLOUDABI_RIGHT_FILE_STAT_FGET | CLOUDABI_RIGHT_POLL_FD_READWRITE | CLOUDABI_RIGHT_SOCK_ACCEPT | CLOUDABI_RIGHT_SOCK_BIND_SOCKET | CLOUDABI_RIGHT_SOCK_CONNECT_SOCKET | CLOUDABI_RIGHT_SOCK_LISTEN | CLOUDABI_RIGHT_SOCK_SHUTDOWN | CLOUDABI_RIGHT_SOCK_STAT_GET; break; default: *inheriting = 0; break; } } /* Converts FreeBSD's Capsicum rights to CloudABI's set of rights. */ static void convert_capabilities(const cap_rights_t *capabilities, cloudabi_filetype_t filetype, cloudabi_rights_t *base, cloudabi_rights_t *inheriting) { cloudabi_rights_t rights; /* Convert FreeBSD bits to CloudABI bits. */ rights = 0; #define MAPPING(cloudabi, ...) do { \ if (cap_rights_is_set(capabilities, ##__VA_ARGS__)) \ rights |= (cloudabi); \ } while (0); RIGHTS_MAPPINGS #undef MAPPING *base = rights; *inheriting = rights; cloudabi_remove_conflicting_rights(filetype, base, inheriting); } int cloudabi_sys_fd_stat_get(struct thread *td, struct cloudabi_sys_fd_stat_get_args *uap) { cloudabi_fdstat_t fsb = {}; struct filedesc *fdp; struct file *fp; seq_t seq; cap_rights_t rights; int error, oflags; bool modified; /* Obtain file descriptor properties. */ fdp = td->td_proc->p_fd; do { error = fget_unlocked(fdp, uap->fd, cap_rights_init(&rights), &fp, &seq); if (error != 0) return (error); if (fp->f_ops == &badfileops) { fdrop(fp, td); return (EBADF); } rights = *cap_rights(fdp, uap->fd); oflags = OFLAGS(fp->f_flag); fsb.fs_filetype = cloudabi_convert_filetype(fp); modified = fd_modified(fdp, uap->fd, seq); fdrop(fp, td); } while (modified); /* Convert file descriptor flags. */ if (oflags & O_APPEND) fsb.fs_flags |= CLOUDABI_FDFLAG_APPEND; if (oflags & O_NONBLOCK) fsb.fs_flags |= CLOUDABI_FDFLAG_NONBLOCK; if (oflags & O_SYNC) fsb.fs_flags |= CLOUDABI_FDFLAG_SYNC; /* Convert capabilities to CloudABI rights. */ convert_capabilities(&rights, fsb.fs_filetype, &fsb.fs_rights_base, &fsb.fs_rights_inheriting); return (copyout(&fsb, (void *)uap->buf, sizeof(fsb))); } int cloudabi_sys_fd_stat_put(struct thread *td, struct cloudabi_sys_fd_stat_put_args *uap) { /* Not implemented. */ return (ENOSYS); } int cloudabi_sys_fd_sync(struct thread *td, struct cloudabi_sys_fd_sync_args *uap) { struct fsync_args fsync_args = { .fd = uap->fd }; return (sys_fsync(td, &fsync_args)); } Index: head/sys/compat/linux/linux_file.c =================================================================== --- head/sys/compat/linux/linux_file.c (revision 286020) +++ head/sys/compat/linux/linux_file.c (revision 286021) @@ -1,1652 +1,1652 @@ /*- * Copyright (c) 1994-1995 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 __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef COMPAT_LINUX32 #include #include #else #include #include #endif #include #include #include int linux_creat(struct thread *td, struct linux_creat_args *args) { char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(creat)) printf(ARGS(creat, "%s, %d"), path, args->mode); #endif error = kern_openat(td, AT_FDCWD, path, UIO_SYSSPACE, O_WRONLY | O_CREAT | O_TRUNC, args->mode); LFREEPATH(path); return (error); } static int linux_common_open(struct thread *td, int dirfd, char *path, int l_flags, int mode) { cap_rights_t rights; struct proc *p = td->td_proc; struct file *fp; int fd; int bsd_flags, error; bsd_flags = 0; switch (l_flags & LINUX_O_ACCMODE) { case LINUX_O_WRONLY: bsd_flags |= O_WRONLY; break; case LINUX_O_RDWR: bsd_flags |= O_RDWR; break; default: bsd_flags |= O_RDONLY; } if (l_flags & LINUX_O_NDELAY) bsd_flags |= O_NONBLOCK; if (l_flags & LINUX_O_APPEND) bsd_flags |= O_APPEND; if (l_flags & LINUX_O_SYNC) bsd_flags |= O_FSYNC; if (l_flags & LINUX_O_NONBLOCK) bsd_flags |= O_NONBLOCK; if (l_flags & LINUX_FASYNC) bsd_flags |= O_ASYNC; if (l_flags & LINUX_O_CREAT) bsd_flags |= O_CREAT; if (l_flags & LINUX_O_TRUNC) bsd_flags |= O_TRUNC; if (l_flags & LINUX_O_EXCL) bsd_flags |= O_EXCL; if (l_flags & LINUX_O_NOCTTY) bsd_flags |= O_NOCTTY; if (l_flags & LINUX_O_DIRECT) bsd_flags |= O_DIRECT; if (l_flags & LINUX_O_NOFOLLOW) bsd_flags |= O_NOFOLLOW; if (l_flags & LINUX_O_DIRECTORY) bsd_flags |= O_DIRECTORY; /* XXX LINUX_O_NOATIME: unable to be easily implemented. */ error = kern_openat(td, dirfd, path, UIO_SYSSPACE, bsd_flags, mode); if (error != 0) goto done; if (bsd_flags & O_NOCTTY) goto done; /* * XXX In between kern_open() and fget(), another process * having the same filedesc could use that fd without * checking below. */ fd = td->td_retval[0]; if (fget(td, fd, cap_rights_init(&rights, CAP_IOCTL), &fp) == 0) { if (fp->f_type != DTYPE_VNODE) { fdrop(fp, td); goto done; } sx_slock(&proctree_lock); PROC_LOCK(p); if (SESS_LEADER(p) && !(p->p_flag & P_CONTROLT)) { PROC_UNLOCK(p); sx_sunlock(&proctree_lock); /* XXXPJD: Verify if TIOCSCTTY is allowed. */ (void) fo_ioctl(fp, TIOCSCTTY, (caddr_t) 0, td->td_ucred, td); } else { PROC_UNLOCK(p); sx_sunlock(&proctree_lock); } fdrop(fp, td); } done: #ifdef DEBUG if (ldebug(open)) printf(LMSG("open returns error %d"), error); #endif LFREEPATH(path); return (error); } int linux_openat(struct thread *td, struct linux_openat_args *args) { char *path; int dfd; dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; if (args->flags & LINUX_O_CREAT) LCONVPATH_AT(td, args->filename, &path, 1, dfd); else LCONVPATH_AT(td, args->filename, &path, 0, dfd); #ifdef DEBUG if (ldebug(openat)) printf(ARGS(openat, "%i, %s, 0x%x, 0x%x"), args->dfd, path, args->flags, args->mode); #endif return (linux_common_open(td, dfd, path, args->flags, args->mode)); } int linux_open(struct thread *td, struct linux_open_args *args) { char *path; if (args->flags & LINUX_O_CREAT) LCONVPATHCREAT(td, args->path, &path); else LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(open)) printf(ARGS(open, "%s, 0x%x, 0x%x"), path, args->flags, args->mode); #endif return (linux_common_open(td, AT_FDCWD, path, args->flags, args->mode)); } int linux_lseek(struct thread *td, struct linux_lseek_args *args) { struct lseek_args /* { int fd; int pad; off_t offset; int whence; } */ tmp_args; int error; #ifdef DEBUG if (ldebug(lseek)) printf(ARGS(lseek, "%d, %ld, %d"), args->fdes, (long)args->off, args->whence); #endif tmp_args.fd = args->fdes; tmp_args.offset = (off_t)args->off; tmp_args.whence = args->whence; error = sys_lseek(td, &tmp_args); return error; } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) int linux_llseek(struct thread *td, struct linux_llseek_args *args) { struct lseek_args bsd_args; int error; off_t off; #ifdef DEBUG if (ldebug(llseek)) printf(ARGS(llseek, "%d, %d:%d, %d"), args->fd, args->ohigh, args->olow, args->whence); #endif off = (args->olow) | (((off_t) args->ohigh) << 32); bsd_args.fd = args->fd; bsd_args.offset = off; bsd_args.whence = args->whence; if ((error = sys_lseek(td, &bsd_args))) return error; if ((error = copyout(td->td_retval, args->res, sizeof (off_t)))) return error; td->td_retval[0] = 0; return 0; } int linux_readdir(struct thread *td, struct linux_readdir_args *args) { struct linux_getdents_args lda; lda.fd = args->fd; lda.dent = args->dent; lda.count = 1; return linux_getdents(td, &lda); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ /* * Note that linux_getdents(2) and linux_getdents64(2) have the same * arguments. They only differ in the definition of struct dirent they * operate on. We use this to common the code, with the exception of * accessing struct dirent. Note that linux_readdir(2) is implemented * by means of linux_getdents(2). In this case we never operate on * struct dirent64 and thus don't need to handle it... */ struct l_dirent { l_ulong d_ino; l_off_t d_off; l_ushort d_reclen; char d_name[LINUX_NAME_MAX + 1]; }; struct l_dirent64 { uint64_t d_ino; int64_t d_off; l_ushort d_reclen; u_char d_type; char d_name[LINUX_NAME_MAX + 1]; }; /* * Linux uses the last byte in the dirent buffer to store d_type, * at least glibc-2.7 requires it. That is why l_dirent is padded with 2 bytes. */ #define LINUX_RECLEN(namlen) \ roundup((offsetof(struct l_dirent, d_name) + (namlen) + 2), \ sizeof(l_ulong)) #define LINUX_RECLEN64(namlen) \ roundup((offsetof(struct l_dirent64, d_name) + (namlen) + 1), \ sizeof(uint64_t)) #define LINUX_MAXRECLEN max(LINUX_RECLEN(LINUX_NAME_MAX), \ LINUX_RECLEN64(LINUX_NAME_MAX)) #define LINUX_DIRBLKSIZ 512 static int getdents_common(struct thread *td, struct linux_getdents64_args *args, int is64bit) { struct dirent *bdp; struct vnode *vp; caddr_t inp, buf; /* BSD-format */ int len, reclen; /* BSD-format */ caddr_t outp; /* Linux-format */ int resid, linuxreclen=0; /* Linux-format */ caddr_t lbuf; /* Linux-format */ cap_rights_t rights; struct file *fp; struct uio auio; struct iovec aiov; off_t off; struct l_dirent *linux_dirent; struct l_dirent64 *linux_dirent64; int buflen, error, eofflag, nbytes, justone; u_long *cookies = NULL, *cookiep; int ncookies; nbytes = args->count; if (nbytes == 1) { /* readdir(2) case. Always struct dirent. */ if (is64bit) return (EINVAL); nbytes = sizeof(*linux_dirent); justone = 1; } else justone = 0; error = getvnode(td, args->fd, cap_rights_init(&rights, CAP_READ), &fp); if (error != 0) return (error); if ((fp->f_flag & FREAD) == 0) { fdrop(fp, td); return (EBADF); } off = foffset_lock(fp, 0); vp = fp->f_vnode; if (vp->v_type != VDIR) { foffset_unlock(fp, off, 0); fdrop(fp, td); return (EINVAL); } buflen = max(LINUX_DIRBLKSIZ, nbytes); buflen = min(buflen, MAXBSIZE); buf = malloc(buflen, M_LINUX, M_WAITOK); lbuf = malloc(LINUX_MAXRECLEN, M_LINUX, M_WAITOK | M_ZERO); vn_lock(vp, LK_SHARED | LK_RETRY); aiov.iov_base = buf; aiov.iov_len = buflen; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_rw = UIO_READ; auio.uio_segflg = UIO_SYSSPACE; auio.uio_td = td; auio.uio_resid = buflen; auio.uio_offset = off; #ifdef MAC /* * Do directory search MAC check using non-cached credentials. */ if ((error = mac_vnode_check_readdir(td->td_ucred, vp))) goto out; #endif /* MAC */ if ((error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &ncookies, &cookies))) goto out; inp = buf; outp = (caddr_t)args->dirent; resid = nbytes; if ((len = buflen - auio.uio_resid) <= 0) goto eof; cookiep = cookies; if (cookies) { /* * When using cookies, the vfs has the option of reading from * a different offset than that supplied (UFS truncates the * offset to a block boundary to make sure that it never reads * partway through a directory entry, even if the directory * has been compacted). */ while (len > 0 && ncookies > 0 && *cookiep <= off) { bdp = (struct dirent *) inp; len -= bdp->d_reclen; inp += bdp->d_reclen; cookiep++; ncookies--; } } while (len > 0) { if (cookiep && ncookies == 0) break; bdp = (struct dirent *) inp; reclen = bdp->d_reclen; if (reclen & 3) { error = EFAULT; goto out; } if (bdp->d_fileno == 0) { inp += reclen; if (cookiep) { off = *cookiep++; ncookies--; } else off += reclen; len -= reclen; continue; } linuxreclen = (is64bit) ? LINUX_RECLEN64(bdp->d_namlen) : LINUX_RECLEN(bdp->d_namlen); if (reclen > len || resid < linuxreclen) { outp++; break; } if (justone) { /* readdir(2) case. */ linux_dirent = (struct l_dirent*)lbuf; linux_dirent->d_ino = bdp->d_fileno; linux_dirent->d_off = (l_off_t)linuxreclen; linux_dirent->d_reclen = (l_ushort)bdp->d_namlen; strlcpy(linux_dirent->d_name, bdp->d_name, linuxreclen - offsetof(struct l_dirent, d_name)); error = copyout(linux_dirent, outp, linuxreclen); } if (is64bit) { linux_dirent64 = (struct l_dirent64*)lbuf; linux_dirent64->d_ino = bdp->d_fileno; linux_dirent64->d_off = (cookiep) ? (l_off_t)*cookiep : (l_off_t)(off + reclen); linux_dirent64->d_reclen = (l_ushort)linuxreclen; linux_dirent64->d_type = bdp->d_type; strlcpy(linux_dirent64->d_name, bdp->d_name, linuxreclen - offsetof(struct l_dirent64, d_name)); error = copyout(linux_dirent64, outp, linuxreclen); } else if (!justone) { linux_dirent = (struct l_dirent*)lbuf; linux_dirent->d_ino = bdp->d_fileno; linux_dirent->d_off = (cookiep) ? (l_off_t)*cookiep : (l_off_t)(off + reclen); linux_dirent->d_reclen = (l_ushort)linuxreclen; /* * Copy d_type to last byte of l_dirent buffer */ lbuf[linuxreclen-1] = bdp->d_type; strlcpy(linux_dirent->d_name, bdp->d_name, linuxreclen - offsetof(struct l_dirent, d_name)-1); error = copyout(linux_dirent, outp, linuxreclen); } if (error) goto out; inp += reclen; if (cookiep) { off = *cookiep++; ncookies--; } else off += reclen; outp += linuxreclen; resid -= linuxreclen; len -= reclen; if (justone) break; } if (outp == (caddr_t)args->dirent) { nbytes = resid; goto eof; } if (justone) nbytes = resid + linuxreclen; eof: td->td_retval[0] = nbytes - resid; out: free(cookies, M_TEMP); VOP_UNLOCK(vp, 0); foffset_unlock(fp, off, 0); fdrop(fp, td); free(buf, M_LINUX); free(lbuf, M_LINUX); return (error); } int linux_getdents(struct thread *td, struct linux_getdents_args *args) { #ifdef DEBUG if (ldebug(getdents)) printf(ARGS(getdents, "%d, *, %d"), args->fd, args->count); #endif return (getdents_common(td, (struct linux_getdents64_args*)args, 0)); } int linux_getdents64(struct thread *td, struct linux_getdents64_args *args) { #ifdef DEBUG if (ldebug(getdents64)) printf(ARGS(getdents64, "%d, *, %d"), args->fd, args->count); #endif return (getdents_common(td, args, 1)); } /* * These exist mainly for hooks for doing /compat/linux translation. */ int linux_access(struct thread *td, struct linux_access_args *args) { char *path; int error; /* linux convention */ if (args->amode & ~(F_OK | X_OK | W_OK | R_OK)) return (EINVAL); LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(access)) printf(ARGS(access, "%s, %d"), path, args->amode); #endif error = kern_accessat(td, AT_FDCWD, path, UIO_SYSSPACE, 0, args->amode); LFREEPATH(path); return (error); } int linux_faccessat(struct thread *td, struct linux_faccessat_args *args) { char *path; int error, dfd; /* linux convention */ if (args->amode & ~(F_OK | X_OK | W_OK | R_OK)) return (EINVAL); dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHEXIST_AT(td, args->filename, &path, dfd); #ifdef DEBUG if (ldebug(access)) printf(ARGS(access, "%s, %d"), path, args->amode); #endif error = kern_accessat(td, dfd, path, UIO_SYSSPACE, 0, args->amode); LFREEPATH(path); return (error); } int linux_unlink(struct thread *td, struct linux_unlink_args *args) { char *path; int error; struct stat st; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(unlink)) printf(ARGS(unlink, "%s"), path); #endif error = kern_unlinkat(td, AT_FDCWD, path, UIO_SYSSPACE, 0); if (error == EPERM) { /* Introduce POSIX noncompliant behaviour of Linux */ if (kern_statat(td, 0, AT_FDCWD, path, UIO_SYSSPACE, &st, NULL) == 0) { if (S_ISDIR(st.st_mode)) error = EISDIR; } } LFREEPATH(path); return (error); } int linux_unlinkat(struct thread *td, struct linux_unlinkat_args *args) { char *path; int error, dfd; struct stat st; if (args->flag & ~LINUX_AT_REMOVEDIR) return (EINVAL); dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHEXIST_AT(td, args->pathname, &path, dfd); #ifdef DEBUG if (ldebug(unlinkat)) printf(ARGS(unlinkat, "%s"), path); #endif if (args->flag & LINUX_AT_REMOVEDIR) error = kern_rmdirat(td, dfd, path, UIO_SYSSPACE); else error = kern_unlinkat(td, dfd, path, UIO_SYSSPACE, 0); if (error == EPERM && !(args->flag & LINUX_AT_REMOVEDIR)) { /* Introduce POSIX noncompliant behaviour of Linux */ if (kern_statat(td, AT_SYMLINK_NOFOLLOW, dfd, path, UIO_SYSSPACE, &st, NULL) == 0 && S_ISDIR(st.st_mode)) error = EISDIR; } LFREEPATH(path); return (error); } int linux_chdir(struct thread *td, struct linux_chdir_args *args) { char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(chdir)) printf(ARGS(chdir, "%s"), path); #endif error = kern_chdir(td, path, UIO_SYSSPACE); LFREEPATH(path); return (error); } int linux_chmod(struct thread *td, struct linux_chmod_args *args) { char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(chmod)) printf(ARGS(chmod, "%s, %d"), path, args->mode); #endif error = kern_fchmodat(td, AT_FDCWD, path, UIO_SYSSPACE, args->mode, 0); LFREEPATH(path); return (error); } int linux_fchmodat(struct thread *td, struct linux_fchmodat_args *args) { char *path; int error, dfd; dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHEXIST_AT(td, args->filename, &path, dfd); #ifdef DEBUG if (ldebug(fchmodat)) printf(ARGS(fchmodat, "%s, %d"), path, args->mode); #endif error = kern_fchmodat(td, dfd, path, UIO_SYSSPACE, args->mode, 0); LFREEPATH(path); return (error); } int linux_mkdir(struct thread *td, struct linux_mkdir_args *args) { char *path; int error; LCONVPATHCREAT(td, args->path, &path); #ifdef DEBUG if (ldebug(mkdir)) printf(ARGS(mkdir, "%s, %d"), path, args->mode); #endif error = kern_mkdirat(td, AT_FDCWD, path, UIO_SYSSPACE, args->mode); LFREEPATH(path); return (error); } int linux_mkdirat(struct thread *td, struct linux_mkdirat_args *args) { char *path; int error, dfd; dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHCREAT_AT(td, args->pathname, &path, dfd); #ifdef DEBUG if (ldebug(mkdirat)) printf(ARGS(mkdirat, "%s, %d"), path, args->mode); #endif error = kern_mkdirat(td, dfd, path, UIO_SYSSPACE, args->mode); LFREEPATH(path); return (error); } int linux_rmdir(struct thread *td, struct linux_rmdir_args *args) { char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(rmdir)) printf(ARGS(rmdir, "%s"), path); #endif error = kern_rmdirat(td, AT_FDCWD, path, UIO_SYSSPACE); LFREEPATH(path); return (error); } int linux_rename(struct thread *td, struct linux_rename_args *args) { char *from, *to; int error; LCONVPATHEXIST(td, args->from, &from); /* Expand LCONVPATHCREATE so that `from' can be freed on errors */ error = linux_emul_convpath(td, args->to, UIO_USERSPACE, &to, 1, AT_FDCWD); if (to == NULL) { LFREEPATH(from); return (error); } #ifdef DEBUG if (ldebug(rename)) printf(ARGS(rename, "%s, %s"), from, to); #endif error = kern_renameat(td, AT_FDCWD, from, AT_FDCWD, to, UIO_SYSSPACE); LFREEPATH(from); LFREEPATH(to); return (error); } int linux_renameat(struct thread *td, struct linux_renameat_args *args) { char *from, *to; int error, olddfd, newdfd; olddfd = (args->olddfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->olddfd; newdfd = (args->newdfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->newdfd; LCONVPATHEXIST_AT(td, args->oldname, &from, olddfd); /* Expand LCONVPATHCREATE so that `from' can be freed on errors */ error = linux_emul_convpath(td, args->newname, UIO_USERSPACE, &to, 1, newdfd); if (to == NULL) { LFREEPATH(from); return (error); } #ifdef DEBUG if (ldebug(renameat)) printf(ARGS(renameat, "%s, %s"), from, to); #endif error = kern_renameat(td, olddfd, from, newdfd, to, UIO_SYSSPACE); LFREEPATH(from); LFREEPATH(to); return (error); } int linux_symlink(struct thread *td, struct linux_symlink_args *args) { char *path, *to; int error; LCONVPATHEXIST(td, args->path, &path); /* Expand LCONVPATHCREATE so that `path' can be freed on errors */ error = linux_emul_convpath(td, args->to, UIO_USERSPACE, &to, 1, AT_FDCWD); if (to == NULL) { LFREEPATH(path); return (error); } #ifdef DEBUG if (ldebug(symlink)) printf(ARGS(symlink, "%s, %s"), path, to); #endif error = kern_symlinkat(td, path, AT_FDCWD, to, UIO_SYSSPACE); LFREEPATH(path); LFREEPATH(to); return (error); } int linux_symlinkat(struct thread *td, struct linux_symlinkat_args *args) { char *path, *to; int error, dfd; dfd = (args->newdfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->newdfd; LCONVPATHEXIST_AT(td, args->oldname, &path, dfd); /* Expand LCONVPATHCREATE so that `path' can be freed on errors */ error = linux_emul_convpath(td, args->newname, UIO_USERSPACE, &to, 1, dfd); if (to == NULL) { LFREEPATH(path); return (error); } #ifdef DEBUG if (ldebug(symlinkat)) printf(ARGS(symlinkat, "%s, %s"), path, to); #endif error = kern_symlinkat(td, path, dfd, to, UIO_SYSSPACE); LFREEPATH(path); LFREEPATH(to); return (error); } int linux_readlink(struct thread *td, struct linux_readlink_args *args) { char *name; int error; LCONVPATHEXIST(td, args->name, &name); #ifdef DEBUG if (ldebug(readlink)) printf(ARGS(readlink, "%s, %p, %d"), name, (void *)args->buf, args->count); #endif error = kern_readlinkat(td, AT_FDCWD, name, UIO_SYSSPACE, args->buf, UIO_USERSPACE, args->count); LFREEPATH(name); return (error); } int linux_readlinkat(struct thread *td, struct linux_readlinkat_args *args) { char *name; int error, dfd; dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHEXIST_AT(td, args->path, &name, dfd); #ifdef DEBUG if (ldebug(readlinkat)) printf(ARGS(readlinkat, "%s, %p, %d"), name, (void *)args->buf, args->bufsiz); #endif error = kern_readlinkat(td, dfd, name, UIO_SYSSPACE, args->buf, UIO_USERSPACE, args->bufsiz); LFREEPATH(name); return (error); } int linux_truncate(struct thread *td, struct linux_truncate_args *args) { char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(truncate)) printf(ARGS(truncate, "%s, %ld"), path, (long)args->length); #endif error = kern_truncate(td, path, UIO_SYSSPACE, args->length); LFREEPATH(path); return (error); } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) int linux_truncate64(struct thread *td, struct linux_truncate64_args *args) { char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(truncate64)) printf(ARGS(truncate64, "%s, %jd"), path, args->length); #endif error = kern_truncate(td, path, UIO_SYSSPACE, args->length); LFREEPATH(path); return (error); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ int linux_ftruncate(struct thread *td, struct linux_ftruncate_args *args) { struct ftruncate_args /* { int fd; int pad; off_t length; } */ nuap; nuap.fd = args->fd; nuap.length = args->length; return (sys_ftruncate(td, &nuap)); } int linux_link(struct thread *td, struct linux_link_args *args) { char *path, *to; int error; LCONVPATHEXIST(td, args->path, &path); /* Expand LCONVPATHCREATE so that `path' can be freed on errors */ error = linux_emul_convpath(td, args->to, UIO_USERSPACE, &to, 1, AT_FDCWD); if (to == NULL) { LFREEPATH(path); return (error); } #ifdef DEBUG if (ldebug(link)) printf(ARGS(link, "%s, %s"), path, to); #endif error = kern_linkat(td, AT_FDCWD, AT_FDCWD, path, to, UIO_SYSSPACE, FOLLOW); LFREEPATH(path); LFREEPATH(to); return (error); } int linux_linkat(struct thread *td, struct linux_linkat_args *args) { char *path, *to; int error, olddfd, newdfd, follow; if (args->flag & ~LINUX_AT_SYMLINK_FOLLOW) return (EINVAL); olddfd = (args->olddfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->olddfd; newdfd = (args->newdfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->newdfd; LCONVPATHEXIST_AT(td, args->oldname, &path, olddfd); /* Expand LCONVPATHCREATE so that `path' can be freed on errors */ error = linux_emul_convpath(td, args->newname, UIO_USERSPACE, &to, 1, newdfd); if (to == NULL) { LFREEPATH(path); return (error); } #ifdef DEBUG if (ldebug(linkat)) printf(ARGS(linkat, "%i, %s, %i, %s, %i"), args->olddfd, path, args->newdfd, to, args->flag); #endif follow = (args->flag & LINUX_AT_SYMLINK_FOLLOW) == 0 ? NOFOLLOW : FOLLOW; error = kern_linkat(td, olddfd, newdfd, path, to, UIO_SYSSPACE, follow); LFREEPATH(path); LFREEPATH(to); return (error); } int linux_fdatasync(td, uap) struct thread *td; struct linux_fdatasync_args *uap; { struct fsync_args bsd; bsd.fd = uap->fd; return sys_fsync(td, &bsd); } int linux_pread(td, uap) struct thread *td; struct linux_pread_args *uap; { struct pread_args bsd; cap_rights_t rights; struct vnode *vp; int error; bsd.fd = uap->fd; bsd.buf = uap->buf; bsd.nbyte = uap->nbyte; bsd.offset = uap->offset; error = sys_pread(td, &bsd); if (error == 0) { /* This seems to violate POSIX but linux does it */ error = fgetvp(td, uap->fd, cap_rights_init(&rights, CAP_PREAD), &vp); if (error != 0) return (error); if (vp->v_type == VDIR) { vrele(vp); return (EISDIR); } vrele(vp); } return (error); } int linux_pwrite(td, uap) struct thread *td; struct linux_pwrite_args *uap; { struct pwrite_args bsd; bsd.fd = uap->fd; bsd.buf = uap->buf; bsd.nbyte = uap->nbyte; bsd.offset = uap->offset; return sys_pwrite(td, &bsd); } int linux_mount(struct thread *td, struct linux_mount_args *args) { char fstypename[MFSNAMELEN]; char mntonname[MNAMELEN], mntfromname[MNAMELEN]; int error; int fsflags; error = copyinstr(args->filesystemtype, fstypename, MFSNAMELEN - 1, NULL); if (error) return (error); error = copyinstr(args->specialfile, mntfromname, MNAMELEN - 1, NULL); if (error) return (error); error = copyinstr(args->dir, mntonname, MNAMELEN - 1, NULL); if (error) return (error); #ifdef DEBUG if (ldebug(mount)) printf(ARGS(mount, "%s, %s, %s"), fstypename, mntfromname, mntonname); #endif if (strcmp(fstypename, "ext2") == 0) { strcpy(fstypename, "ext2fs"); } else if (strcmp(fstypename, "proc") == 0) { strcpy(fstypename, "linprocfs"); } else if (strcmp(fstypename, "vfat") == 0) { strcpy(fstypename, "msdosfs"); } fsflags = 0; if ((args->rwflag & 0xffff0000) == 0xc0ed0000) { /* * Linux SYNC flag is not included; the closest equivalent * FreeBSD has is !ASYNC, which is our default. */ if (args->rwflag & LINUX_MS_RDONLY) fsflags |= MNT_RDONLY; if (args->rwflag & LINUX_MS_NOSUID) fsflags |= MNT_NOSUID; if (args->rwflag & LINUX_MS_NOEXEC) fsflags |= MNT_NOEXEC; if (args->rwflag & LINUX_MS_REMOUNT) fsflags |= MNT_UPDATE; } error = kernel_vmount(fsflags, "fstype", fstypename, "fspath", mntonname, "from", mntfromname, NULL); return (error); } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) int linux_oldumount(struct thread *td, struct linux_oldumount_args *args) { struct linux_umount_args args2; args2.path = args->path; args2.flags = 0; return (linux_umount(td, &args2)); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ int linux_umount(struct thread *td, struct linux_umount_args *args) { struct unmount_args bsd; bsd.path = args->path; bsd.flags = args->flags; /* XXX correct? */ return (sys_unmount(td, &bsd)); } /* * fcntl family of syscalls */ struct l_flock { l_short l_type; l_short l_whence; l_off_t l_start; l_off_t l_len; l_pid_t l_pid; } #if defined(__amd64__) && defined(COMPAT_LINUX32) __packed #endif ; static void linux_to_bsd_flock(struct l_flock *linux_flock, struct flock *bsd_flock) { switch (linux_flock->l_type) { case LINUX_F_RDLCK: bsd_flock->l_type = F_RDLCK; break; case LINUX_F_WRLCK: bsd_flock->l_type = F_WRLCK; break; case LINUX_F_UNLCK: bsd_flock->l_type = F_UNLCK; break; default: bsd_flock->l_type = -1; break; } bsd_flock->l_whence = linux_flock->l_whence; bsd_flock->l_start = (off_t)linux_flock->l_start; bsd_flock->l_len = (off_t)linux_flock->l_len; bsd_flock->l_pid = (pid_t)linux_flock->l_pid; bsd_flock->l_sysid = 0; } static void bsd_to_linux_flock(struct flock *bsd_flock, struct l_flock *linux_flock) { switch (bsd_flock->l_type) { case F_RDLCK: linux_flock->l_type = LINUX_F_RDLCK; break; case F_WRLCK: linux_flock->l_type = LINUX_F_WRLCK; break; case F_UNLCK: linux_flock->l_type = LINUX_F_UNLCK; break; } linux_flock->l_whence = bsd_flock->l_whence; linux_flock->l_start = (l_off_t)bsd_flock->l_start; linux_flock->l_len = (l_off_t)bsd_flock->l_len; linux_flock->l_pid = (l_pid_t)bsd_flock->l_pid; } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) struct l_flock64 { l_short l_type; l_short l_whence; l_loff_t l_start; l_loff_t l_len; l_pid_t l_pid; } #if defined(__amd64__) && defined(COMPAT_LINUX32) __packed #endif ; static void linux_to_bsd_flock64(struct l_flock64 *linux_flock, struct flock *bsd_flock) { switch (linux_flock->l_type) { case LINUX_F_RDLCK: bsd_flock->l_type = F_RDLCK; break; case LINUX_F_WRLCK: bsd_flock->l_type = F_WRLCK; break; case LINUX_F_UNLCK: bsd_flock->l_type = F_UNLCK; break; default: bsd_flock->l_type = -1; break; } bsd_flock->l_whence = linux_flock->l_whence; bsd_flock->l_start = (off_t)linux_flock->l_start; bsd_flock->l_len = (off_t)linux_flock->l_len; bsd_flock->l_pid = (pid_t)linux_flock->l_pid; bsd_flock->l_sysid = 0; } static void bsd_to_linux_flock64(struct flock *bsd_flock, struct l_flock64 *linux_flock) { switch (bsd_flock->l_type) { case F_RDLCK: linux_flock->l_type = LINUX_F_RDLCK; break; case F_WRLCK: linux_flock->l_type = LINUX_F_WRLCK; break; case F_UNLCK: linux_flock->l_type = LINUX_F_UNLCK; break; } linux_flock->l_whence = bsd_flock->l_whence; linux_flock->l_start = (l_loff_t)bsd_flock->l_start; linux_flock->l_len = (l_loff_t)bsd_flock->l_len; linux_flock->l_pid = (l_pid_t)bsd_flock->l_pid; } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ static int fcntl_common(struct thread *td, struct linux_fcntl_args *args) { struct l_flock linux_flock; struct flock bsd_flock; cap_rights_t rights; struct file *fp; long arg; int error, result; switch (args->cmd) { case LINUX_F_DUPFD: return (kern_fcntl(td, args->fd, F_DUPFD, args->arg)); case LINUX_F_GETFD: return (kern_fcntl(td, args->fd, F_GETFD, 0)); case LINUX_F_SETFD: return (kern_fcntl(td, args->fd, F_SETFD, args->arg)); case LINUX_F_GETFL: error = kern_fcntl(td, args->fd, F_GETFL, 0); result = td->td_retval[0]; td->td_retval[0] = 0; if (result & O_RDONLY) td->td_retval[0] |= LINUX_O_RDONLY; if (result & O_WRONLY) td->td_retval[0] |= LINUX_O_WRONLY; if (result & O_RDWR) td->td_retval[0] |= LINUX_O_RDWR; if (result & O_NDELAY) td->td_retval[0] |= LINUX_O_NONBLOCK; if (result & O_APPEND) td->td_retval[0] |= LINUX_O_APPEND; if (result & O_FSYNC) td->td_retval[0] |= LINUX_O_SYNC; if (result & O_ASYNC) td->td_retval[0] |= LINUX_FASYNC; #ifdef LINUX_O_NOFOLLOW if (result & O_NOFOLLOW) td->td_retval[0] |= LINUX_O_NOFOLLOW; #endif #ifdef LINUX_O_DIRECT if (result & O_DIRECT) td->td_retval[0] |= LINUX_O_DIRECT; #endif return (error); case LINUX_F_SETFL: arg = 0; if (args->arg & LINUX_O_NDELAY) arg |= O_NONBLOCK; if (args->arg & LINUX_O_APPEND) arg |= O_APPEND; if (args->arg & LINUX_O_SYNC) arg |= O_FSYNC; if (args->arg & LINUX_FASYNC) arg |= O_ASYNC; #ifdef LINUX_O_NOFOLLOW if (args->arg & LINUX_O_NOFOLLOW) arg |= O_NOFOLLOW; #endif #ifdef LINUX_O_DIRECT if (args->arg & LINUX_O_DIRECT) arg |= O_DIRECT; #endif return (kern_fcntl(td, args->fd, F_SETFL, arg)); case LINUX_F_GETLK: error = copyin((void *)args->arg, &linux_flock, sizeof(linux_flock)); if (error) return (error); linux_to_bsd_flock(&linux_flock, &bsd_flock); error = kern_fcntl(td, args->fd, F_GETLK, (intptr_t)&bsd_flock); if (error) return (error); bsd_to_linux_flock(&bsd_flock, &linux_flock); return (copyout(&linux_flock, (void *)args->arg, sizeof(linux_flock))); case LINUX_F_SETLK: error = copyin((void *)args->arg, &linux_flock, sizeof(linux_flock)); if (error) return (error); linux_to_bsd_flock(&linux_flock, &bsd_flock); return (kern_fcntl(td, args->fd, F_SETLK, (intptr_t)&bsd_flock)); case LINUX_F_SETLKW: error = copyin((void *)args->arg, &linux_flock, sizeof(linux_flock)); if (error) return (error); linux_to_bsd_flock(&linux_flock, &bsd_flock); return (kern_fcntl(td, args->fd, F_SETLKW, (intptr_t)&bsd_flock)); case LINUX_F_GETOWN: return (kern_fcntl(td, args->fd, F_GETOWN, 0)); case LINUX_F_SETOWN: /* * XXX some Linux applications depend on F_SETOWN having no * significant effect for pipes (SIGIO is not delivered for * pipes under Linux-2.2.35 at least). */ error = fget(td, args->fd, cap_rights_init(&rights, CAP_FCNTL), &fp); if (error) return (error); if (fp->f_type == DTYPE_PIPE) { fdrop(fp, td); return (EINVAL); } fdrop(fp, td); return (kern_fcntl(td, args->fd, F_SETOWN, args->arg)); case LINUX_F_DUPFD_CLOEXEC: return (kern_fcntl(td, args->fd, F_DUPFD_CLOEXEC, args->arg)); } return (EINVAL); } int linux_fcntl(struct thread *td, struct linux_fcntl_args *args) { #ifdef DEBUG if (ldebug(fcntl)) printf(ARGS(fcntl, "%d, %08x, *"), args->fd, args->cmd); #endif return (fcntl_common(td, args)); } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) int linux_fcntl64(struct thread *td, struct linux_fcntl64_args *args) { struct l_flock64 linux_flock; struct flock bsd_flock; struct linux_fcntl_args fcntl_args; int error; #ifdef DEBUG if (ldebug(fcntl64)) printf(ARGS(fcntl64, "%d, %08x, *"), args->fd, args->cmd); #endif switch (args->cmd) { case LINUX_F_GETLK64: error = copyin((void *)args->arg, &linux_flock, sizeof(linux_flock)); if (error) return (error); linux_to_bsd_flock64(&linux_flock, &bsd_flock); error = kern_fcntl(td, args->fd, F_GETLK, (intptr_t)&bsd_flock); if (error) return (error); bsd_to_linux_flock64(&bsd_flock, &linux_flock); return (copyout(&linux_flock, (void *)args->arg, sizeof(linux_flock))); case LINUX_F_SETLK64: error = copyin((void *)args->arg, &linux_flock, sizeof(linux_flock)); if (error) return (error); linux_to_bsd_flock64(&linux_flock, &bsd_flock); return (kern_fcntl(td, args->fd, F_SETLK, (intptr_t)&bsd_flock)); case LINUX_F_SETLKW64: error = copyin((void *)args->arg, &linux_flock, sizeof(linux_flock)); if (error) return (error); linux_to_bsd_flock64(&linux_flock, &bsd_flock); return (kern_fcntl(td, args->fd, F_SETLKW, (intptr_t)&bsd_flock)); } fcntl_args.fd = args->fd; fcntl_args.cmd = args->cmd; fcntl_args.arg = args->arg; return (fcntl_common(td, &fcntl_args)); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ int linux_chown(struct thread *td, struct linux_chown_args *args) { char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(chown)) printf(ARGS(chown, "%s, %d, %d"), path, args->uid, args->gid); #endif error = kern_fchownat(td, AT_FDCWD, path, UIO_SYSSPACE, args->uid, args->gid, 0); LFREEPATH(path); return (error); } int linux_fchownat(struct thread *td, struct linux_fchownat_args *args) { char *path; int error, dfd, flag; if (args->flag & ~LINUX_AT_SYMLINK_NOFOLLOW) return (EINVAL); dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHEXIST_AT(td, args->filename, &path, dfd); #ifdef DEBUG if (ldebug(fchownat)) printf(ARGS(fchownat, "%s, %d, %d"), path, args->uid, args->gid); #endif flag = (args->flag & LINUX_AT_SYMLINK_NOFOLLOW) == 0 ? 0 : AT_SYMLINK_NOFOLLOW; error = kern_fchownat(td, dfd, path, UIO_SYSSPACE, args->uid, args->gid, flag); LFREEPATH(path); return (error); } int linux_lchown(struct thread *td, struct linux_lchown_args *args) { char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(lchown)) printf(ARGS(lchown, "%s, %d, %d"), path, args->uid, args->gid); #endif error = kern_fchownat(td, AT_FDCWD, path, UIO_SYSSPACE, args->uid, args->gid, AT_SYMLINK_NOFOLLOW); LFREEPATH(path); return (error); } static int convert_fadvice(int advice) { switch (advice) { case LINUX_POSIX_FADV_NORMAL: return (POSIX_FADV_NORMAL); case LINUX_POSIX_FADV_RANDOM: return (POSIX_FADV_RANDOM); case LINUX_POSIX_FADV_SEQUENTIAL: return (POSIX_FADV_SEQUENTIAL); case LINUX_POSIX_FADV_WILLNEED: return (POSIX_FADV_WILLNEED); case LINUX_POSIX_FADV_DONTNEED: return (POSIX_FADV_DONTNEED); case LINUX_POSIX_FADV_NOREUSE: return (POSIX_FADV_NOREUSE); default: return (-1); } } int linux_fadvise64(struct thread *td, struct linux_fadvise64_args *args) { int advice; advice = convert_fadvice(args->advice); if (advice == -1) return (EINVAL); return (kern_posix_fadvise(td, args->fd, args->offset, args->len, advice)); } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) int linux_fadvise64_64(struct thread *td, struct linux_fadvise64_64_args *args) { int advice; advice = convert_fadvice(args->advice); if (advice == -1) return (EINVAL); return (kern_posix_fadvise(td, args->fd, args->offset, args->len, advice)); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ int linux_pipe(struct thread *td, struct linux_pipe_args *args) { int fildes[2]; int error; #ifdef DEBUG if (ldebug(pipe)) printf(ARGS(pipe, "*")); #endif - error = kern_pipe2(td, fildes, 0); + error = kern_pipe(td, fildes, 0, NULL, NULL); if (error) return (error); /* XXX: Close descriptors on error. */ return (copyout(fildes, args->pipefds, sizeof(fildes))); } int linux_pipe2(struct thread *td, struct linux_pipe2_args *args) { int fildes[2]; int error, flags; #ifdef DEBUG if (ldebug(pipe2)) printf(ARGS(pipe2, "*, %d"), args->flags); #endif if ((args->flags & ~(LINUX_O_NONBLOCK | LINUX_O_CLOEXEC)) != 0) return (EINVAL); flags = 0; if ((args->flags & LINUX_O_NONBLOCK) != 0) flags |= O_NONBLOCK; if ((args->flags & LINUX_O_CLOEXEC) != 0) flags |= O_CLOEXEC; - error = kern_pipe2(td, fildes, flags); + error = kern_pipe(td, fildes, flags, NULL, NULL); if (error) return (error); /* XXX: Close descriptors on error. */ return (copyout(fildes, args->pipefds, sizeof(fildes))); } int linux_dup3(struct thread *td, struct linux_dup3_args *args) { int cmd; intptr_t newfd; if (args->oldfd == args->newfd) return (EINVAL); if ((args->flags & ~LINUX_O_CLOEXEC) != 0) return (EINVAL); if (args->flags & LINUX_O_CLOEXEC) cmd = F_DUP2FD_CLOEXEC; else cmd = F_DUP2FD; newfd = args->newfd; return (kern_fcntl(td, args->oldfd, cmd, newfd)); } int linux_fallocate(struct thread *td, struct linux_fallocate_args *args) { /* * We emulate only posix_fallocate system call for which * mode should be 0. */ if (args->mode != 0) return (ENOSYS); return (kern_posix_fallocate(td, args->fd, args->offset, args->len)); } Index: head/sys/kern/sys_pipe.c =================================================================== --- head/sys/kern/sys_pipe.c (revision 286020) +++ head/sys/kern/sys_pipe.c (revision 286021) @@ -1,1843 +1,1837 @@ /*- * Copyright (c) 1996 John S. Dyson * Copyright (c) 2012 Giovanni Trematerra * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice immediately at the beginning of the file, without modification, * this list of conditions, and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Absolutely no warranty of function or purpose is made by the author * John S. Dyson. * 4. Modifications may be freely made to this file if the above conditions * are met. */ /* * This file contains a high-performance replacement for the socket-based * pipes scheme originally used in FreeBSD/4.4Lite. It does not support * all features of sockets, but does do everything that pipes normally * do. */ /* * This code has two modes of operation, a small write mode and a large * write mode. The small write mode acts like conventional pipes with * a kernel buffer. If the buffer is less than PIPE_MINDIRECT, then the * "normal" pipe buffering is done. If the buffer is between PIPE_MINDIRECT * and PIPE_SIZE in size, the sending process pins the underlying pages in * memory, and the receiving process copies directly from these pinned pages * in the sending process. * * If the sending process receives a signal, it is possible that it will * go away, and certainly its address space can change, because control * is returned back to the user-mode side. In that case, the pipe code * arranges to copy the buffer supplied by the user process, to a pageable * kernel buffer, and the receiving process will grab the data from the * pageable kernel buffer. Since signals don't happen all that often, * the copy operation is normally eliminated. * * The constant PIPE_MINDIRECT is chosen to make sure that buffering will * happen for small transfers so that the system will not spend all of * its time context switching. * * In order to limit the resource use of pipes, two sysctls exist: * * kern.ipc.maxpipekva - This is a hard limit on the amount of pageable * address space available to us in pipe_map. This value is normally * autotuned, but may also be loader tuned. * * kern.ipc.pipekva - This read-only sysctl tracks the current amount of * memory in use by pipes. * * Based on how large pipekva is relative to maxpipekva, the following * will happen: * * 0% - 50%: * New pipes are given 16K of memory backing, pipes may dynamically * grow to as large as 64K where needed. * 50% - 75%: * New pipes are given 4K (or PAGE_SIZE) of memory backing, * existing pipes may NOT grow. * 75% - 100%: * New pipes are given 4K (or PAGE_SIZE) of memory backing, * existing pipes will be shrunk down to 4K whenever possible. * * Resizing may be disabled by setting kern.ipc.piperesizeallowed=0. If * that is set, the only resize that will occur is the 0 -> SMALL_PIPE_SIZE * resize which MUST occur for reverse-direction pipes when they are * first used. * * Additional information about the current state of pipes may be obtained * from kern.ipc.pipes, kern.ipc.pipefragretry, kern.ipc.pipeallocfail, * and kern.ipc.piperesizefail. * * Locking rules: There are two locks present here: A mutex, used via * PIPE_LOCK, and a flag, used via pipelock(). All locking is done via * the flag, as mutexes can not persist over uiomove. The mutex * exists only to guard access to the flag, and is not in itself a * locking mechanism. Also note that there is only a single mutex for * both directions of a pipe. * * As pipelock() may have to sleep before it can acquire the flag, it * is important to reread all data after a call to pipelock(); everything * in the structure may have changed. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * Use this define if you want to disable *fancy* VM things. Expect an * approx 30% decrease in transfer rate. This could be useful for * NetBSD or OpenBSD. */ /* #define PIPE_NODIRECT */ #define PIPE_PEER(pipe) \ (((pipe)->pipe_state & PIPE_NAMED) ? (pipe) : ((pipe)->pipe_peer)) /* * interfaces to the outside world */ static fo_rdwr_t pipe_read; static fo_rdwr_t pipe_write; static fo_truncate_t pipe_truncate; static fo_ioctl_t pipe_ioctl; static fo_poll_t pipe_poll; static fo_kqfilter_t pipe_kqfilter; static fo_stat_t pipe_stat; static fo_close_t pipe_close; static fo_chmod_t pipe_chmod; static fo_chown_t pipe_chown; static fo_fill_kinfo_t pipe_fill_kinfo; struct fileops pipeops = { .fo_read = pipe_read, .fo_write = pipe_write, .fo_truncate = pipe_truncate, .fo_ioctl = pipe_ioctl, .fo_poll = pipe_poll, .fo_kqfilter = pipe_kqfilter, .fo_stat = pipe_stat, .fo_close = pipe_close, .fo_chmod = pipe_chmod, .fo_chown = pipe_chown, .fo_sendfile = invfo_sendfile, .fo_fill_kinfo = pipe_fill_kinfo, .fo_flags = DFLAG_PASSABLE }; static void filt_pipedetach(struct knote *kn); static void filt_pipedetach_notsup(struct knote *kn); static int filt_pipenotsup(struct knote *kn, long hint); static int filt_piperead(struct knote *kn, long hint); static int filt_pipewrite(struct knote *kn, long hint); static struct filterops pipe_nfiltops = { .f_isfd = 1, .f_detach = filt_pipedetach_notsup, .f_event = filt_pipenotsup }; static struct filterops pipe_rfiltops = { .f_isfd = 1, .f_detach = filt_pipedetach, .f_event = filt_piperead }; static struct filterops pipe_wfiltops = { .f_isfd = 1, .f_detach = filt_pipedetach, .f_event = filt_pipewrite }; /* * Default pipe buffer size(s), this can be kind-of large now because pipe * space is pageable. The pipe code will try to maintain locality of * reference for performance reasons, so small amounts of outstanding I/O * will not wipe the cache. */ #define MINPIPESIZE (PIPE_SIZE/3) #define MAXPIPESIZE (2*PIPE_SIZE/3) static long amountpipekva; static int pipefragretry; static int pipeallocfail; static int piperesizefail; static int piperesizeallowed = 1; SYSCTL_LONG(_kern_ipc, OID_AUTO, maxpipekva, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &maxpipekva, 0, "Pipe KVA limit"); SYSCTL_LONG(_kern_ipc, OID_AUTO, pipekva, CTLFLAG_RD, &amountpipekva, 0, "Pipe KVA usage"); SYSCTL_INT(_kern_ipc, OID_AUTO, pipefragretry, CTLFLAG_RD, &pipefragretry, 0, "Pipe allocation retries due to fragmentation"); SYSCTL_INT(_kern_ipc, OID_AUTO, pipeallocfail, CTLFLAG_RD, &pipeallocfail, 0, "Pipe allocation failures"); SYSCTL_INT(_kern_ipc, OID_AUTO, piperesizefail, CTLFLAG_RD, &piperesizefail, 0, "Pipe resize failures"); SYSCTL_INT(_kern_ipc, OID_AUTO, piperesizeallowed, CTLFLAG_RW, &piperesizeallowed, 0, "Pipe resizing allowed"); static void pipeinit(void *dummy __unused); static void pipeclose(struct pipe *cpipe); static void pipe_free_kmem(struct pipe *cpipe); static void pipe_create(struct pipe *pipe, int backing); static void pipe_paircreate(struct thread *td, struct pipepair **p_pp); static __inline int pipelock(struct pipe *cpipe, int catch); static __inline void pipeunlock(struct pipe *cpipe); #ifndef PIPE_NODIRECT static int pipe_build_write_buffer(struct pipe *wpipe, struct uio *uio); static void pipe_destroy_write_buffer(struct pipe *wpipe); static int pipe_direct_write(struct pipe *wpipe, struct uio *uio); static void pipe_clone_write_buffer(struct pipe *wpipe); #endif static int pipespace(struct pipe *cpipe, int size); static int pipespace_new(struct pipe *cpipe, int size); static int pipe_zone_ctor(void *mem, int size, void *arg, int flags); static int pipe_zone_init(void *mem, int size, int flags); static void pipe_zone_fini(void *mem, int size); static uma_zone_t pipe_zone; static struct unrhdr *pipeino_unr; static dev_t pipedev_ino; SYSINIT(vfs, SI_SUB_VFS, SI_ORDER_ANY, pipeinit, NULL); static void pipeinit(void *dummy __unused) { pipe_zone = uma_zcreate("pipe", sizeof(struct pipepair), pipe_zone_ctor, NULL, pipe_zone_init, pipe_zone_fini, UMA_ALIGN_PTR, 0); KASSERT(pipe_zone != NULL, ("pipe_zone not initialized")); pipeino_unr = new_unrhdr(1, INT32_MAX, NULL); KASSERT(pipeino_unr != NULL, ("pipe fake inodes not initialized")); pipedev_ino = devfs_alloc_cdp_inode(); KASSERT(pipedev_ino > 0, ("pipe dev inode not initialized")); } static int pipe_zone_ctor(void *mem, int size, void *arg, int flags) { struct pipepair *pp; struct pipe *rpipe, *wpipe; KASSERT(size == sizeof(*pp), ("pipe_zone_ctor: wrong size")); pp = (struct pipepair *)mem; /* * We zero both pipe endpoints to make sure all the kmem pointers * are NULL, flag fields are zero'd, etc. We timestamp both * endpoints with the same time. */ rpipe = &pp->pp_rpipe; bzero(rpipe, sizeof(*rpipe)); vfs_timestamp(&rpipe->pipe_ctime); rpipe->pipe_atime = rpipe->pipe_mtime = rpipe->pipe_ctime; wpipe = &pp->pp_wpipe; bzero(wpipe, sizeof(*wpipe)); wpipe->pipe_ctime = rpipe->pipe_ctime; wpipe->pipe_atime = wpipe->pipe_mtime = rpipe->pipe_ctime; rpipe->pipe_peer = wpipe; rpipe->pipe_pair = pp; wpipe->pipe_peer = rpipe; wpipe->pipe_pair = pp; /* * Mark both endpoints as present; they will later get free'd * one at a time. When both are free'd, then the whole pair * is released. */ rpipe->pipe_present = PIPE_ACTIVE; wpipe->pipe_present = PIPE_ACTIVE; /* * Eventually, the MAC Framework may initialize the label * in ctor or init, but for now we do it elswhere to avoid * blocking in ctor or init. */ pp->pp_label = NULL; return (0); } static int pipe_zone_init(void *mem, int size, int flags) { struct pipepair *pp; KASSERT(size == sizeof(*pp), ("pipe_zone_init: wrong size")); pp = (struct pipepair *)mem; mtx_init(&pp->pp_mtx, "pipe mutex", NULL, MTX_DEF | MTX_NEW); return (0); } static void pipe_zone_fini(void *mem, int size) { struct pipepair *pp; KASSERT(size == sizeof(*pp), ("pipe_zone_fini: wrong size")); pp = (struct pipepair *)mem; mtx_destroy(&pp->pp_mtx); } static void pipe_paircreate(struct thread *td, struct pipepair **p_pp) { struct pipepair *pp; struct pipe *rpipe, *wpipe; *p_pp = pp = uma_zalloc(pipe_zone, M_WAITOK); #ifdef MAC /* * The MAC label is shared between the connected endpoints. As a * result mac_pipe_init() and mac_pipe_create() are called once * for the pair, and not on the endpoints. */ mac_pipe_init(pp); mac_pipe_create(td->td_ucred, pp); #endif rpipe = &pp->pp_rpipe; wpipe = &pp->pp_wpipe; knlist_init_mtx(&rpipe->pipe_sel.si_note, PIPE_MTX(rpipe)); knlist_init_mtx(&wpipe->pipe_sel.si_note, PIPE_MTX(wpipe)); /* Only the forward direction pipe is backed by default */ pipe_create(rpipe, 1); pipe_create(wpipe, 0); rpipe->pipe_state |= PIPE_DIRECTOK; wpipe->pipe_state |= PIPE_DIRECTOK; } void pipe_named_ctor(struct pipe **ppipe, struct thread *td) { struct pipepair *pp; pipe_paircreate(td, &pp); pp->pp_rpipe.pipe_state |= PIPE_NAMED; *ppipe = &pp->pp_rpipe; } void pipe_dtor(struct pipe *dpipe) { struct pipe *peer; ino_t ino; ino = dpipe->pipe_ino; peer = (dpipe->pipe_state & PIPE_NAMED) != 0 ? dpipe->pipe_peer : NULL; funsetown(&dpipe->pipe_sigio); pipeclose(dpipe); if (peer != NULL) { funsetown(&peer->pipe_sigio); pipeclose(peer); } if (ino != 0 && ino != (ino_t)-1) free_unr(pipeino_unr, ino); } /* * The pipe system call for the DTYPE_PIPE type of pipes. If we fail, let * the zone pick up the pieces via pipeclose(). */ int -kern_pipe(struct thread *td, int fildes[2]) +kern_pipe(struct thread *td, int fildes[2], int flags, struct filecaps *fcaps1, + struct filecaps *fcaps2) { - - return (kern_pipe2(td, fildes, 0)); -} - -int -kern_pipe2(struct thread *td, int fildes[2], int flags) -{ struct file *rf, *wf; struct pipe *rpipe, *wpipe; struct pipepair *pp; int fd, fflags, error; pipe_paircreate(td, &pp); rpipe = &pp->pp_rpipe; wpipe = &pp->pp_wpipe; - error = falloc(td, &rf, &fd, flags); + error = falloc_caps(td, &rf, &fd, flags, fcaps1); if (error) { pipeclose(rpipe); pipeclose(wpipe); return (error); } - /* An extra reference on `rf' has been held for us by falloc(). */ + /* An extra reference on `rf' has been held for us by falloc_caps(). */ fildes[0] = fd; fflags = FREAD | FWRITE; if ((flags & O_NONBLOCK) != 0) fflags |= FNONBLOCK; /* * Warning: once we've gotten past allocation of the fd for the * read-side, we can only drop the read side via fdrop() in order * to avoid races against processes which manage to dup() the read * side while we are blocked trying to allocate the write side. */ finit(rf, fflags, DTYPE_PIPE, rpipe, &pipeops); - error = falloc(td, &wf, &fd, flags); + error = falloc_caps(td, &wf, &fd, flags, fcaps2); if (error) { fdclose(td, rf, fildes[0]); fdrop(rf, td); /* rpipe has been closed by fdrop(). */ pipeclose(wpipe); return (error); } - /* An extra reference on `wf' has been held for us by falloc(). */ + /* An extra reference on `wf' has been held for us by falloc_caps(). */ finit(wf, fflags, DTYPE_PIPE, wpipe, &pipeops); fdrop(wf, td); fildes[1] = fd; fdrop(rf, td); return (0); } /* ARGSUSED */ int sys_pipe(struct thread *td, struct pipe_args *uap) { int error; int fildes[2]; - error = kern_pipe(td, fildes); + error = kern_pipe(td, fildes, 0, NULL, NULL); if (error) return (error); td->td_retval[0] = fildes[0]; td->td_retval[1] = fildes[1]; return (0); } int sys_pipe2(struct thread *td, struct pipe2_args *uap) { int error, fildes[2]; if (uap->flags & ~(O_CLOEXEC | O_NONBLOCK)) return (EINVAL); - error = kern_pipe2(td, fildes, uap->flags); + error = kern_pipe(td, fildes, uap->flags, NULL, NULL); if (error) return (error); error = copyout(fildes, uap->fildes, 2 * sizeof(int)); if (error) { (void)kern_close(td, fildes[0]); (void)kern_close(td, fildes[1]); } return (error); } /* * Allocate kva for pipe circular buffer, the space is pageable * This routine will 'realloc' the size of a pipe safely, if it fails * it will retain the old buffer. * If it fails it will return ENOMEM. */ static int pipespace_new(cpipe, size) struct pipe *cpipe; int size; { caddr_t buffer; int error, cnt, firstseg; static int curfail = 0; static struct timeval lastfail; KASSERT(!mtx_owned(PIPE_MTX(cpipe)), ("pipespace: pipe mutex locked")); KASSERT(!(cpipe->pipe_state & PIPE_DIRECTW), ("pipespace: resize of direct writes not allowed")); retry: cnt = cpipe->pipe_buffer.cnt; if (cnt > size) size = cnt; size = round_page(size); buffer = (caddr_t) vm_map_min(pipe_map); error = vm_map_find(pipe_map, NULL, 0, (vm_offset_t *) &buffer, size, 0, VMFS_ANY_SPACE, VM_PROT_ALL, VM_PROT_ALL, 0); if (error != KERN_SUCCESS) { if ((cpipe->pipe_buffer.buffer == NULL) && (size > SMALL_PIPE_SIZE)) { size = SMALL_PIPE_SIZE; pipefragretry++; goto retry; } if (cpipe->pipe_buffer.buffer == NULL) { pipeallocfail++; if (ppsratecheck(&lastfail, &curfail, 1)) printf("kern.ipc.maxpipekva exceeded; see tuning(7)\n"); } else { piperesizefail++; } return (ENOMEM); } /* copy data, then free old resources if we're resizing */ if (cnt > 0) { if (cpipe->pipe_buffer.in <= cpipe->pipe_buffer.out) { firstseg = cpipe->pipe_buffer.size - cpipe->pipe_buffer.out; bcopy(&cpipe->pipe_buffer.buffer[cpipe->pipe_buffer.out], buffer, firstseg); if ((cnt - firstseg) > 0) bcopy(cpipe->pipe_buffer.buffer, &buffer[firstseg], cpipe->pipe_buffer.in); } else { bcopy(&cpipe->pipe_buffer.buffer[cpipe->pipe_buffer.out], buffer, cnt); } } pipe_free_kmem(cpipe); cpipe->pipe_buffer.buffer = buffer; cpipe->pipe_buffer.size = size; cpipe->pipe_buffer.in = cnt; cpipe->pipe_buffer.out = 0; cpipe->pipe_buffer.cnt = cnt; atomic_add_long(&amountpipekva, cpipe->pipe_buffer.size); return (0); } /* * Wrapper for pipespace_new() that performs locking assertions. */ static int pipespace(cpipe, size) struct pipe *cpipe; int size; { KASSERT(cpipe->pipe_state & PIPE_LOCKFL, ("Unlocked pipe passed to pipespace")); return (pipespace_new(cpipe, size)); } /* * lock a pipe for I/O, blocking other access */ static __inline int pipelock(cpipe, catch) struct pipe *cpipe; int catch; { int error; PIPE_LOCK_ASSERT(cpipe, MA_OWNED); while (cpipe->pipe_state & PIPE_LOCKFL) { cpipe->pipe_state |= PIPE_LWANT; error = msleep(cpipe, PIPE_MTX(cpipe), catch ? (PRIBIO | PCATCH) : PRIBIO, "pipelk", 0); if (error != 0) return (error); } cpipe->pipe_state |= PIPE_LOCKFL; return (0); } /* * unlock a pipe I/O lock */ static __inline void pipeunlock(cpipe) struct pipe *cpipe; { PIPE_LOCK_ASSERT(cpipe, MA_OWNED); KASSERT(cpipe->pipe_state & PIPE_LOCKFL, ("Unlocked pipe passed to pipeunlock")); cpipe->pipe_state &= ~PIPE_LOCKFL; if (cpipe->pipe_state & PIPE_LWANT) { cpipe->pipe_state &= ~PIPE_LWANT; wakeup(cpipe); } } void pipeselwakeup(cpipe) struct pipe *cpipe; { PIPE_LOCK_ASSERT(cpipe, MA_OWNED); if (cpipe->pipe_state & PIPE_SEL) { selwakeuppri(&cpipe->pipe_sel, PSOCK); if (!SEL_WAITING(&cpipe->pipe_sel)) cpipe->pipe_state &= ~PIPE_SEL; } if ((cpipe->pipe_state & PIPE_ASYNC) && cpipe->pipe_sigio) pgsigio(&cpipe->pipe_sigio, SIGIO, 0); KNOTE_LOCKED(&cpipe->pipe_sel.si_note, 0); } /* * Initialize and allocate VM and memory for pipe. The structure * will start out zero'd from the ctor, so we just manage the kmem. */ static void pipe_create(pipe, backing) struct pipe *pipe; int backing; { if (backing) { /* * Note that these functions can fail if pipe map is exhausted * (as a result of too many pipes created), but we ignore the * error as it is not fatal and could be provoked by * unprivileged users. The only consequence is worse performance * with given pipe. */ if (amountpipekva > maxpipekva / 2) (void)pipespace_new(pipe, SMALL_PIPE_SIZE); else (void)pipespace_new(pipe, PIPE_SIZE); } pipe->pipe_ino = -1; } /* ARGSUSED */ static int pipe_read(fp, uio, active_cred, flags, td) struct file *fp; struct uio *uio; struct ucred *active_cred; struct thread *td; int flags; { struct pipe *rpipe; int error; int nread = 0; int size; rpipe = fp->f_data; PIPE_LOCK(rpipe); ++rpipe->pipe_busy; error = pipelock(rpipe, 1); if (error) goto unlocked_error; #ifdef MAC error = mac_pipe_check_read(active_cred, rpipe->pipe_pair); if (error) goto locked_error; #endif if (amountpipekva > (3 * maxpipekva) / 4) { if (!(rpipe->pipe_state & PIPE_DIRECTW) && (rpipe->pipe_buffer.size > SMALL_PIPE_SIZE) && (rpipe->pipe_buffer.cnt <= SMALL_PIPE_SIZE) && (piperesizeallowed == 1)) { PIPE_UNLOCK(rpipe); pipespace(rpipe, SMALL_PIPE_SIZE); PIPE_LOCK(rpipe); } } while (uio->uio_resid) { /* * normal pipe buffer receive */ if (rpipe->pipe_buffer.cnt > 0) { size = rpipe->pipe_buffer.size - rpipe->pipe_buffer.out; if (size > rpipe->pipe_buffer.cnt) size = rpipe->pipe_buffer.cnt; if (size > uio->uio_resid) size = uio->uio_resid; PIPE_UNLOCK(rpipe); error = uiomove( &rpipe->pipe_buffer.buffer[rpipe->pipe_buffer.out], size, uio); PIPE_LOCK(rpipe); if (error) break; rpipe->pipe_buffer.out += size; if (rpipe->pipe_buffer.out >= rpipe->pipe_buffer.size) rpipe->pipe_buffer.out = 0; rpipe->pipe_buffer.cnt -= size; /* * If there is no more to read in the pipe, reset * its pointers to the beginning. This improves * cache hit stats. */ if (rpipe->pipe_buffer.cnt == 0) { rpipe->pipe_buffer.in = 0; rpipe->pipe_buffer.out = 0; } nread += size; #ifndef PIPE_NODIRECT /* * Direct copy, bypassing a kernel buffer. */ } else if ((size = rpipe->pipe_map.cnt) && (rpipe->pipe_state & PIPE_DIRECTW)) { if (size > uio->uio_resid) size = (u_int) uio->uio_resid; PIPE_UNLOCK(rpipe); error = uiomove_fromphys(rpipe->pipe_map.ms, rpipe->pipe_map.pos, size, uio); PIPE_LOCK(rpipe); if (error) break; nread += size; rpipe->pipe_map.pos += size; rpipe->pipe_map.cnt -= size; if (rpipe->pipe_map.cnt == 0) { rpipe->pipe_state &= ~(PIPE_DIRECTW|PIPE_WANTW); wakeup(rpipe); } #endif } else { /* * detect EOF condition * read returns 0 on EOF, no need to set error */ if (rpipe->pipe_state & PIPE_EOF) break; /* * If the "write-side" has been blocked, wake it up now. */ if (rpipe->pipe_state & PIPE_WANTW) { rpipe->pipe_state &= ~PIPE_WANTW; wakeup(rpipe); } /* * Break if some data was read. */ if (nread > 0) break; /* * Unlock the pipe buffer for our remaining processing. * We will either break out with an error or we will * sleep and relock to loop. */ pipeunlock(rpipe); /* * Handle non-blocking mode operation or * wait for more data. */ if (fp->f_flag & FNONBLOCK) { error = EAGAIN; } else { rpipe->pipe_state |= PIPE_WANTR; if ((error = msleep(rpipe, PIPE_MTX(rpipe), PRIBIO | PCATCH, "piperd", 0)) == 0) error = pipelock(rpipe, 1); } if (error) goto unlocked_error; } } #ifdef MAC locked_error: #endif pipeunlock(rpipe); /* XXX: should probably do this before getting any locks. */ if (error == 0) vfs_timestamp(&rpipe->pipe_atime); unlocked_error: --rpipe->pipe_busy; /* * PIPE_WANT processing only makes sense if pipe_busy is 0. */ if ((rpipe->pipe_busy == 0) && (rpipe->pipe_state & PIPE_WANT)) { rpipe->pipe_state &= ~(PIPE_WANT|PIPE_WANTW); wakeup(rpipe); } else if (rpipe->pipe_buffer.cnt < MINPIPESIZE) { /* * Handle write blocking hysteresis. */ if (rpipe->pipe_state & PIPE_WANTW) { rpipe->pipe_state &= ~PIPE_WANTW; wakeup(rpipe); } } if ((rpipe->pipe_buffer.size - rpipe->pipe_buffer.cnt) >= PIPE_BUF) pipeselwakeup(rpipe); PIPE_UNLOCK(rpipe); return (error); } #ifndef PIPE_NODIRECT /* * Map the sending processes' buffer into kernel space and wire it. * This is similar to a physical write operation. */ static int pipe_build_write_buffer(wpipe, uio) struct pipe *wpipe; struct uio *uio; { u_int size; int i; PIPE_LOCK_ASSERT(wpipe, MA_NOTOWNED); KASSERT(wpipe->pipe_state & PIPE_DIRECTW, ("Clone attempt on non-direct write pipe!")); if (uio->uio_iov->iov_len > wpipe->pipe_buffer.size) size = wpipe->pipe_buffer.size; else size = uio->uio_iov->iov_len; if ((i = vm_fault_quick_hold_pages(&curproc->p_vmspace->vm_map, (vm_offset_t)uio->uio_iov->iov_base, size, VM_PROT_READ, wpipe->pipe_map.ms, PIPENPAGES)) < 0) return (EFAULT); /* * set up the control block */ wpipe->pipe_map.npages = i; wpipe->pipe_map.pos = ((vm_offset_t) uio->uio_iov->iov_base) & PAGE_MASK; wpipe->pipe_map.cnt = size; /* * and update the uio data */ uio->uio_iov->iov_len -= size; uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + size; if (uio->uio_iov->iov_len == 0) uio->uio_iov++; uio->uio_resid -= size; uio->uio_offset += size; return (0); } /* * unmap and unwire the process buffer */ static void pipe_destroy_write_buffer(wpipe) struct pipe *wpipe; { PIPE_LOCK_ASSERT(wpipe, MA_OWNED); vm_page_unhold_pages(wpipe->pipe_map.ms, wpipe->pipe_map.npages); wpipe->pipe_map.npages = 0; } /* * In the case of a signal, the writing process might go away. This * code copies the data into the circular buffer so that the source * pages can be freed without loss of data. */ static void pipe_clone_write_buffer(wpipe) struct pipe *wpipe; { struct uio uio; struct iovec iov; int size; int pos; PIPE_LOCK_ASSERT(wpipe, MA_OWNED); size = wpipe->pipe_map.cnt; pos = wpipe->pipe_map.pos; wpipe->pipe_buffer.in = size; wpipe->pipe_buffer.out = 0; wpipe->pipe_buffer.cnt = size; wpipe->pipe_state &= ~PIPE_DIRECTW; PIPE_UNLOCK(wpipe); iov.iov_base = wpipe->pipe_buffer.buffer; iov.iov_len = size; uio.uio_iov = &iov; uio.uio_iovcnt = 1; uio.uio_offset = 0; uio.uio_resid = size; uio.uio_segflg = UIO_SYSSPACE; uio.uio_rw = UIO_READ; uio.uio_td = curthread; uiomove_fromphys(wpipe->pipe_map.ms, pos, size, &uio); PIPE_LOCK(wpipe); pipe_destroy_write_buffer(wpipe); } /* * This implements the pipe buffer write mechanism. Note that only * a direct write OR a normal pipe write can be pending at any given time. * If there are any characters in the pipe buffer, the direct write will * be deferred until the receiving process grabs all of the bytes from * the pipe buffer. Then the direct mapping write is set-up. */ static int pipe_direct_write(wpipe, uio) struct pipe *wpipe; struct uio *uio; { int error; retry: PIPE_LOCK_ASSERT(wpipe, MA_OWNED); error = pipelock(wpipe, 1); if (error != 0) goto error1; if ((wpipe->pipe_state & PIPE_EOF) != 0) { error = EPIPE; pipeunlock(wpipe); goto error1; } while (wpipe->pipe_state & PIPE_DIRECTW) { if (wpipe->pipe_state & PIPE_WANTR) { wpipe->pipe_state &= ~PIPE_WANTR; wakeup(wpipe); } pipeselwakeup(wpipe); wpipe->pipe_state |= PIPE_WANTW; pipeunlock(wpipe); error = msleep(wpipe, PIPE_MTX(wpipe), PRIBIO | PCATCH, "pipdww", 0); if (error) goto error1; else goto retry; } wpipe->pipe_map.cnt = 0; /* transfer not ready yet */ if (wpipe->pipe_buffer.cnt > 0) { if (wpipe->pipe_state & PIPE_WANTR) { wpipe->pipe_state &= ~PIPE_WANTR; wakeup(wpipe); } pipeselwakeup(wpipe); wpipe->pipe_state |= PIPE_WANTW; pipeunlock(wpipe); error = msleep(wpipe, PIPE_MTX(wpipe), PRIBIO | PCATCH, "pipdwc", 0); if (error) goto error1; else goto retry; } wpipe->pipe_state |= PIPE_DIRECTW; PIPE_UNLOCK(wpipe); error = pipe_build_write_buffer(wpipe, uio); PIPE_LOCK(wpipe); if (error) { wpipe->pipe_state &= ~PIPE_DIRECTW; pipeunlock(wpipe); goto error1; } error = 0; while (!error && (wpipe->pipe_state & PIPE_DIRECTW)) { if (wpipe->pipe_state & PIPE_EOF) { pipe_destroy_write_buffer(wpipe); pipeselwakeup(wpipe); pipeunlock(wpipe); error = EPIPE; goto error1; } if (wpipe->pipe_state & PIPE_WANTR) { wpipe->pipe_state &= ~PIPE_WANTR; wakeup(wpipe); } pipeselwakeup(wpipe); wpipe->pipe_state |= PIPE_WANTW; pipeunlock(wpipe); error = msleep(wpipe, PIPE_MTX(wpipe), PRIBIO | PCATCH, "pipdwt", 0); pipelock(wpipe, 0); } if (wpipe->pipe_state & PIPE_EOF) error = EPIPE; if (wpipe->pipe_state & PIPE_DIRECTW) { /* * this bit of trickery substitutes a kernel buffer for * the process that might be going away. */ pipe_clone_write_buffer(wpipe); } else { pipe_destroy_write_buffer(wpipe); } pipeunlock(wpipe); return (error); error1: wakeup(wpipe); return (error); } #endif static int pipe_write(fp, uio, active_cred, flags, td) struct file *fp; struct uio *uio; struct ucred *active_cred; struct thread *td; int flags; { int error = 0; int desiredsize; ssize_t orig_resid; struct pipe *wpipe, *rpipe; rpipe = fp->f_data; wpipe = PIPE_PEER(rpipe); PIPE_LOCK(rpipe); error = pipelock(wpipe, 1); if (error) { PIPE_UNLOCK(rpipe); return (error); } /* * detect loss of pipe read side, issue SIGPIPE if lost. */ if (wpipe->pipe_present != PIPE_ACTIVE || (wpipe->pipe_state & PIPE_EOF)) { pipeunlock(wpipe); PIPE_UNLOCK(rpipe); return (EPIPE); } #ifdef MAC error = mac_pipe_check_write(active_cred, wpipe->pipe_pair); if (error) { pipeunlock(wpipe); PIPE_UNLOCK(rpipe); return (error); } #endif ++wpipe->pipe_busy; /* Choose a larger size if it's advantageous */ desiredsize = max(SMALL_PIPE_SIZE, wpipe->pipe_buffer.size); while (desiredsize < wpipe->pipe_buffer.cnt + uio->uio_resid) { if (piperesizeallowed != 1) break; if (amountpipekva > maxpipekva / 2) break; if (desiredsize == BIG_PIPE_SIZE) break; desiredsize = desiredsize * 2; } /* Choose a smaller size if we're in a OOM situation */ if ((amountpipekva > (3 * maxpipekva) / 4) && (wpipe->pipe_buffer.size > SMALL_PIPE_SIZE) && (wpipe->pipe_buffer.cnt <= SMALL_PIPE_SIZE) && (piperesizeallowed == 1)) desiredsize = SMALL_PIPE_SIZE; /* Resize if the above determined that a new size was necessary */ if ((desiredsize != wpipe->pipe_buffer.size) && ((wpipe->pipe_state & PIPE_DIRECTW) == 0)) { PIPE_UNLOCK(wpipe); pipespace(wpipe, desiredsize); PIPE_LOCK(wpipe); } if (wpipe->pipe_buffer.size == 0) { /* * This can only happen for reverse direction use of pipes * in a complete OOM situation. */ error = ENOMEM; --wpipe->pipe_busy; pipeunlock(wpipe); PIPE_UNLOCK(wpipe); return (error); } pipeunlock(wpipe); orig_resid = uio->uio_resid; while (uio->uio_resid) { int space; pipelock(wpipe, 0); if (wpipe->pipe_state & PIPE_EOF) { pipeunlock(wpipe); error = EPIPE; break; } #ifndef PIPE_NODIRECT /* * If the transfer is large, we can gain performance if * we do process-to-process copies directly. * If the write is non-blocking, we don't use the * direct write mechanism. * * The direct write mechanism will detect the reader going * away on us. */ if (uio->uio_segflg == UIO_USERSPACE && uio->uio_iov->iov_len >= PIPE_MINDIRECT && wpipe->pipe_buffer.size >= PIPE_MINDIRECT && (fp->f_flag & FNONBLOCK) == 0) { pipeunlock(wpipe); error = pipe_direct_write(wpipe, uio); if (error) break; continue; } #endif /* * Pipe buffered writes cannot be coincidental with * direct writes. We wait until the currently executing * direct write is completed before we start filling the * pipe buffer. We break out if a signal occurs or the * reader goes away. */ if (wpipe->pipe_state & PIPE_DIRECTW) { if (wpipe->pipe_state & PIPE_WANTR) { wpipe->pipe_state &= ~PIPE_WANTR; wakeup(wpipe); } pipeselwakeup(wpipe); wpipe->pipe_state |= PIPE_WANTW; pipeunlock(wpipe); error = msleep(wpipe, PIPE_MTX(rpipe), PRIBIO | PCATCH, "pipbww", 0); if (error) break; else continue; } space = wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt; /* Writes of size <= PIPE_BUF must be atomic. */ if ((space < uio->uio_resid) && (orig_resid <= PIPE_BUF)) space = 0; if (space > 0) { int size; /* Transfer size */ int segsize; /* first segment to transfer */ /* * Transfer size is minimum of uio transfer * and free space in pipe buffer. */ if (space > uio->uio_resid) size = uio->uio_resid; else size = space; /* * First segment to transfer is minimum of * transfer size and contiguous space in * pipe buffer. If first segment to transfer * is less than the transfer size, we've got * a wraparound in the buffer. */ segsize = wpipe->pipe_buffer.size - wpipe->pipe_buffer.in; if (segsize > size) segsize = size; /* Transfer first segment */ PIPE_UNLOCK(rpipe); error = uiomove(&wpipe->pipe_buffer.buffer[wpipe->pipe_buffer.in], segsize, uio); PIPE_LOCK(rpipe); if (error == 0 && segsize < size) { KASSERT(wpipe->pipe_buffer.in + segsize == wpipe->pipe_buffer.size, ("Pipe buffer wraparound disappeared")); /* * Transfer remaining part now, to * support atomic writes. Wraparound * happened. */ PIPE_UNLOCK(rpipe); error = uiomove( &wpipe->pipe_buffer.buffer[0], size - segsize, uio); PIPE_LOCK(rpipe); } if (error == 0) { wpipe->pipe_buffer.in += size; if (wpipe->pipe_buffer.in >= wpipe->pipe_buffer.size) { KASSERT(wpipe->pipe_buffer.in == size - segsize + wpipe->pipe_buffer.size, ("Expected wraparound bad")); wpipe->pipe_buffer.in = size - segsize; } wpipe->pipe_buffer.cnt += size; KASSERT(wpipe->pipe_buffer.cnt <= wpipe->pipe_buffer.size, ("Pipe buffer overflow")); } pipeunlock(wpipe); if (error != 0) break; } else { /* * If the "read-side" has been blocked, wake it up now. */ if (wpipe->pipe_state & PIPE_WANTR) { wpipe->pipe_state &= ~PIPE_WANTR; wakeup(wpipe); } /* * don't block on non-blocking I/O */ if (fp->f_flag & FNONBLOCK) { error = EAGAIN; pipeunlock(wpipe); break; } /* * We have no more space and have something to offer, * wake up select/poll. */ pipeselwakeup(wpipe); wpipe->pipe_state |= PIPE_WANTW; pipeunlock(wpipe); error = msleep(wpipe, PIPE_MTX(rpipe), PRIBIO | PCATCH, "pipewr", 0); if (error != 0) break; } } pipelock(wpipe, 0); --wpipe->pipe_busy; if ((wpipe->pipe_busy == 0) && (wpipe->pipe_state & PIPE_WANT)) { wpipe->pipe_state &= ~(PIPE_WANT | PIPE_WANTR); wakeup(wpipe); } else if (wpipe->pipe_buffer.cnt > 0) { /* * If we have put any characters in the buffer, we wake up * the reader. */ if (wpipe->pipe_state & PIPE_WANTR) { wpipe->pipe_state &= ~PIPE_WANTR; wakeup(wpipe); } } /* * Don't return EPIPE if any byte was written. * EINTR and other interrupts are handled by generic I/O layer. * Do not pretend that I/O succeeded for obvious user error * like EFAULT. */ if (uio->uio_resid != orig_resid && error == EPIPE) error = 0; if (error == 0) vfs_timestamp(&wpipe->pipe_mtime); /* * We have something to offer, * wake up select/poll. */ if (wpipe->pipe_buffer.cnt) pipeselwakeup(wpipe); pipeunlock(wpipe); PIPE_UNLOCK(rpipe); return (error); } /* ARGSUSED */ static int pipe_truncate(fp, length, active_cred, td) struct file *fp; off_t length; struct ucred *active_cred; struct thread *td; { struct pipe *cpipe; int error; cpipe = fp->f_data; if (cpipe->pipe_state & PIPE_NAMED) error = vnops.fo_truncate(fp, length, active_cred, td); else error = invfo_truncate(fp, length, active_cred, td); return (error); } /* * we implement a very minimal set of ioctls for compatibility with sockets. */ static int pipe_ioctl(fp, cmd, data, active_cred, td) struct file *fp; u_long cmd; void *data; struct ucred *active_cred; struct thread *td; { struct pipe *mpipe = fp->f_data; int error; PIPE_LOCK(mpipe); #ifdef MAC error = mac_pipe_check_ioctl(active_cred, mpipe->pipe_pair, cmd, data); if (error) { PIPE_UNLOCK(mpipe); return (error); } #endif error = 0; switch (cmd) { case FIONBIO: break; case FIOASYNC: if (*(int *)data) { mpipe->pipe_state |= PIPE_ASYNC; } else { mpipe->pipe_state &= ~PIPE_ASYNC; } break; case FIONREAD: if (!(fp->f_flag & FREAD)) { *(int *)data = 0; PIPE_UNLOCK(mpipe); return (0); } if (mpipe->pipe_state & PIPE_DIRECTW) *(int *)data = mpipe->pipe_map.cnt; else *(int *)data = mpipe->pipe_buffer.cnt; break; case FIOSETOWN: PIPE_UNLOCK(mpipe); error = fsetown(*(int *)data, &mpipe->pipe_sigio); goto out_unlocked; case FIOGETOWN: *(int *)data = fgetown(&mpipe->pipe_sigio); break; /* This is deprecated, FIOSETOWN should be used instead. */ case TIOCSPGRP: PIPE_UNLOCK(mpipe); error = fsetown(-(*(int *)data), &mpipe->pipe_sigio); goto out_unlocked; /* This is deprecated, FIOGETOWN should be used instead. */ case TIOCGPGRP: *(int *)data = -fgetown(&mpipe->pipe_sigio); break; default: error = ENOTTY; break; } PIPE_UNLOCK(mpipe); out_unlocked: return (error); } static int pipe_poll(fp, events, active_cred, td) struct file *fp; int events; struct ucred *active_cred; struct thread *td; { struct pipe *rpipe; struct pipe *wpipe; int levents, revents; #ifdef MAC int error; #endif revents = 0; rpipe = fp->f_data; wpipe = PIPE_PEER(rpipe); PIPE_LOCK(rpipe); #ifdef MAC error = mac_pipe_check_poll(active_cred, rpipe->pipe_pair); if (error) goto locked_error; #endif if (fp->f_flag & FREAD && events & (POLLIN | POLLRDNORM)) if ((rpipe->pipe_state & PIPE_DIRECTW) || (rpipe->pipe_buffer.cnt > 0)) revents |= events & (POLLIN | POLLRDNORM); if (fp->f_flag & FWRITE && events & (POLLOUT | POLLWRNORM)) if (wpipe->pipe_present != PIPE_ACTIVE || (wpipe->pipe_state & PIPE_EOF) || (((wpipe->pipe_state & PIPE_DIRECTW) == 0) && ((wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt) >= PIPE_BUF || wpipe->pipe_buffer.size == 0))) revents |= events & (POLLOUT | POLLWRNORM); levents = events & (POLLIN | POLLINIGNEOF | POLLPRI | POLLRDNORM | POLLRDBAND); if (rpipe->pipe_state & PIPE_NAMED && fp->f_flag & FREAD && levents && fp->f_seqcount == rpipe->pipe_wgen) events |= POLLINIGNEOF; if ((events & POLLINIGNEOF) == 0) { if (rpipe->pipe_state & PIPE_EOF) { revents |= (events & (POLLIN | POLLRDNORM)); if (wpipe->pipe_present != PIPE_ACTIVE || (wpipe->pipe_state & PIPE_EOF)) revents |= POLLHUP; } } if (revents == 0) { if (fp->f_flag & FREAD && events & (POLLIN | POLLRDNORM)) { selrecord(td, &rpipe->pipe_sel); if (SEL_WAITING(&rpipe->pipe_sel)) rpipe->pipe_state |= PIPE_SEL; } if (fp->f_flag & FWRITE && events & (POLLOUT | POLLWRNORM)) { selrecord(td, &wpipe->pipe_sel); if (SEL_WAITING(&wpipe->pipe_sel)) wpipe->pipe_state |= PIPE_SEL; } } #ifdef MAC locked_error: #endif PIPE_UNLOCK(rpipe); return (revents); } /* * We shouldn't need locks here as we're doing a read and this should * be a natural race. */ static int pipe_stat(fp, ub, active_cred, td) struct file *fp; struct stat *ub; struct ucred *active_cred; struct thread *td; { struct pipe *pipe; int new_unr; #ifdef MAC int error; #endif pipe = fp->f_data; PIPE_LOCK(pipe); #ifdef MAC error = mac_pipe_check_stat(active_cred, pipe->pipe_pair); if (error) { PIPE_UNLOCK(pipe); return (error); } #endif /* For named pipes ask the underlying filesystem. */ if (pipe->pipe_state & PIPE_NAMED) { PIPE_UNLOCK(pipe); return (vnops.fo_stat(fp, ub, active_cred, td)); } /* * Lazily allocate an inode number for the pipe. Most pipe * users do not call fstat(2) on the pipe, which means that * postponing the inode allocation until it is must be * returned to userland is useful. If alloc_unr failed, * assign st_ino zero instead of returning an error. * Special pipe_ino values: * -1 - not yet initialized; * 0 - alloc_unr failed, return 0 as st_ino forever. */ if (pipe->pipe_ino == (ino_t)-1) { new_unr = alloc_unr(pipeino_unr); if (new_unr != -1) pipe->pipe_ino = new_unr; else pipe->pipe_ino = 0; } PIPE_UNLOCK(pipe); bzero(ub, sizeof(*ub)); ub->st_mode = S_IFIFO; ub->st_blksize = PAGE_SIZE; if (pipe->pipe_state & PIPE_DIRECTW) ub->st_size = pipe->pipe_map.cnt; else ub->st_size = pipe->pipe_buffer.cnt; ub->st_blocks = (ub->st_size + ub->st_blksize - 1) / ub->st_blksize; ub->st_atim = pipe->pipe_atime; ub->st_mtim = pipe->pipe_mtime; ub->st_ctim = pipe->pipe_ctime; ub->st_uid = fp->f_cred->cr_uid; ub->st_gid = fp->f_cred->cr_gid; ub->st_dev = pipedev_ino; ub->st_ino = pipe->pipe_ino; /* * Left as 0: st_nlink, st_rdev, st_flags, st_gen. */ return (0); } /* ARGSUSED */ static int pipe_close(fp, td) struct file *fp; struct thread *td; { if (fp->f_vnode != NULL) return vnops.fo_close(fp, td); fp->f_ops = &badfileops; pipe_dtor(fp->f_data); fp->f_data = NULL; return (0); } static int pipe_chmod(struct file *fp, mode_t mode, struct ucred *active_cred, struct thread *td) { struct pipe *cpipe; int error; cpipe = fp->f_data; if (cpipe->pipe_state & PIPE_NAMED) error = vn_chmod(fp, mode, active_cred, td); else error = invfo_chmod(fp, mode, active_cred, td); return (error); } static int pipe_chown(fp, uid, gid, active_cred, td) struct file *fp; uid_t uid; gid_t gid; struct ucred *active_cred; struct thread *td; { struct pipe *cpipe; int error; cpipe = fp->f_data; if (cpipe->pipe_state & PIPE_NAMED) error = vn_chown(fp, uid, gid, active_cred, td); else error = invfo_chown(fp, uid, gid, active_cred, td); return (error); } static int pipe_fill_kinfo(struct file *fp, struct kinfo_file *kif, struct filedesc *fdp) { struct pipe *pi; if (fp->f_type == DTYPE_FIFO) return (vn_fill_kinfo(fp, kif, fdp)); kif->kf_type = KF_TYPE_PIPE; pi = fp->f_data; kif->kf_un.kf_pipe.kf_pipe_addr = (uintptr_t)pi; kif->kf_un.kf_pipe.kf_pipe_peer = (uintptr_t)pi->pipe_peer; kif->kf_un.kf_pipe.kf_pipe_buffer_cnt = pi->pipe_buffer.cnt; return (0); } static void pipe_free_kmem(cpipe) struct pipe *cpipe; { KASSERT(!mtx_owned(PIPE_MTX(cpipe)), ("pipe_free_kmem: pipe mutex locked")); if (cpipe->pipe_buffer.buffer != NULL) { atomic_subtract_long(&amountpipekva, cpipe->pipe_buffer.size); vm_map_remove(pipe_map, (vm_offset_t)cpipe->pipe_buffer.buffer, (vm_offset_t)cpipe->pipe_buffer.buffer + cpipe->pipe_buffer.size); cpipe->pipe_buffer.buffer = NULL; } #ifndef PIPE_NODIRECT { cpipe->pipe_map.cnt = 0; cpipe->pipe_map.pos = 0; cpipe->pipe_map.npages = 0; } #endif } /* * shutdown the pipe */ static void pipeclose(cpipe) struct pipe *cpipe; { struct pipepair *pp; struct pipe *ppipe; KASSERT(cpipe != NULL, ("pipeclose: cpipe == NULL")); PIPE_LOCK(cpipe); pipelock(cpipe, 0); pp = cpipe->pipe_pair; pipeselwakeup(cpipe); /* * If the other side is blocked, wake it up saying that * we want to close it down. */ cpipe->pipe_state |= PIPE_EOF; while (cpipe->pipe_busy) { wakeup(cpipe); cpipe->pipe_state |= PIPE_WANT; pipeunlock(cpipe); msleep(cpipe, PIPE_MTX(cpipe), PRIBIO, "pipecl", 0); pipelock(cpipe, 0); } /* * Disconnect from peer, if any. */ ppipe = cpipe->pipe_peer; if (ppipe->pipe_present == PIPE_ACTIVE) { pipeselwakeup(ppipe); ppipe->pipe_state |= PIPE_EOF; wakeup(ppipe); KNOTE_LOCKED(&ppipe->pipe_sel.si_note, 0); } /* * Mark this endpoint as free. Release kmem resources. We * don't mark this endpoint as unused until we've finished * doing that, or the pipe might disappear out from under * us. */ PIPE_UNLOCK(cpipe); pipe_free_kmem(cpipe); PIPE_LOCK(cpipe); cpipe->pipe_present = PIPE_CLOSING; pipeunlock(cpipe); /* * knlist_clear() may sleep dropping the PIPE_MTX. Set the * PIPE_FINALIZED, that allows other end to free the * pipe_pair, only after the knotes are completely dismantled. */ knlist_clear(&cpipe->pipe_sel.si_note, 1); cpipe->pipe_present = PIPE_FINALIZED; seldrain(&cpipe->pipe_sel); knlist_destroy(&cpipe->pipe_sel.si_note); /* * If both endpoints are now closed, release the memory for the * pipe pair. If not, unlock. */ if (ppipe->pipe_present == PIPE_FINALIZED) { PIPE_UNLOCK(cpipe); #ifdef MAC mac_pipe_destroy(pp); #endif uma_zfree(pipe_zone, cpipe->pipe_pair); } else PIPE_UNLOCK(cpipe); } /*ARGSUSED*/ static int pipe_kqfilter(struct file *fp, struct knote *kn) { struct pipe *cpipe; /* * If a filter is requested that is not supported by this file * descriptor, don't return an error, but also don't ever generate an * event. */ if ((kn->kn_filter == EVFILT_READ) && !(fp->f_flag & FREAD)) { kn->kn_fop = &pipe_nfiltops; return (0); } if ((kn->kn_filter == EVFILT_WRITE) && !(fp->f_flag & FWRITE)) { kn->kn_fop = &pipe_nfiltops; return (0); } cpipe = fp->f_data; PIPE_LOCK(cpipe); switch (kn->kn_filter) { case EVFILT_READ: kn->kn_fop = &pipe_rfiltops; break; case EVFILT_WRITE: kn->kn_fop = &pipe_wfiltops; if (cpipe->pipe_peer->pipe_present != PIPE_ACTIVE) { /* other end of pipe has been closed */ PIPE_UNLOCK(cpipe); return (EPIPE); } cpipe = PIPE_PEER(cpipe); break; default: PIPE_UNLOCK(cpipe); return (EINVAL); } kn->kn_hook = cpipe; knlist_add(&cpipe->pipe_sel.si_note, kn, 1); PIPE_UNLOCK(cpipe); return (0); } static void filt_pipedetach(struct knote *kn) { struct pipe *cpipe = kn->kn_hook; PIPE_LOCK(cpipe); knlist_remove(&cpipe->pipe_sel.si_note, kn, 1); PIPE_UNLOCK(cpipe); } /*ARGSUSED*/ static int filt_piperead(struct knote *kn, long hint) { struct pipe *rpipe = kn->kn_hook; struct pipe *wpipe = rpipe->pipe_peer; int ret; PIPE_LOCK_ASSERT(rpipe, MA_OWNED); kn->kn_data = rpipe->pipe_buffer.cnt; if ((kn->kn_data == 0) && (rpipe->pipe_state & PIPE_DIRECTW)) kn->kn_data = rpipe->pipe_map.cnt; if ((rpipe->pipe_state & PIPE_EOF) || wpipe->pipe_present != PIPE_ACTIVE || (wpipe->pipe_state & PIPE_EOF)) { kn->kn_flags |= EV_EOF; return (1); } ret = kn->kn_data > 0; return ret; } /*ARGSUSED*/ static int filt_pipewrite(struct knote *kn, long hint) { struct pipe *wpipe; wpipe = kn->kn_hook; PIPE_LOCK_ASSERT(wpipe, MA_OWNED); if (wpipe->pipe_present != PIPE_ACTIVE || (wpipe->pipe_state & PIPE_EOF)) { kn->kn_data = 0; kn->kn_flags |= EV_EOF; return (1); } kn->kn_data = (wpipe->pipe_buffer.size > 0) ? (wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt) : PIPE_BUF; if (wpipe->pipe_state & PIPE_DIRECTW) kn->kn_data = 0; return (kn->kn_data >= PIPE_BUF); } static void filt_pipedetach_notsup(struct knote *kn) { } static int filt_pipenotsup(struct knote *kn, long hint) { return (0); } Index: head/sys/sys/syscallsubr.h =================================================================== --- head/sys/sys/syscallsubr.h (revision 286020) +++ head/sys/sys/syscallsubr.h (revision 286021) @@ -1,258 +1,259 @@ /*- * Copyright (c) 2002 Ian Dowse. 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. * * $FreeBSD$ */ #ifndef _SYS_SYSCALLSUBR_H_ #define _SYS_SYSCALLSUBR_H_ #include #include #include #include #include struct file; +struct filecaps; enum idtype; struct itimerval; struct image_args; struct jail; struct kevent; struct kevent_copyops; struct kld_file_stat; struct ksiginfo; struct mbuf; struct msghdr; struct msqid_ds; struct pollfd; struct ogetdirentries_args; struct rlimit; struct rusage; union semun; struct sendfile_args; struct sockaddr; struct stat; struct thr_param; struct sched_param; struct __wrusage; int kern___getcwd(struct thread *td, char *buf, enum uio_seg bufseg, u_int buflen, u_int path_max); int kern_accept(struct thread *td, int s, struct sockaddr **name, socklen_t *namelen, struct file **fp); int kern_accept4(struct thread *td, int s, struct sockaddr **name, socklen_t *namelen, int flags, struct file **fp); int kern_accessat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int flags, int mode); int kern_adjtime(struct thread *td, struct timeval *delta, struct timeval *olddelta); int kern_alternate_path(struct thread *td, const char *prefix, const char *path, enum uio_seg pathseg, char **pathbuf, int create, int dirfd); int kern_bindat(struct thread *td, int dirfd, int fd, struct sockaddr *sa); int kern_cap_ioctls_limit(struct thread *td, int fd, u_long *cmds, size_t ncmds); int kern_chdir(struct thread *td, char *path, enum uio_seg pathseg); int kern_clock_getcpuclockid2(struct thread *td, id_t id, int which, clockid_t *clk_id); int kern_clock_getres(struct thread *td, clockid_t clock_id, struct timespec *ts); int kern_clock_gettime(struct thread *td, clockid_t clock_id, struct timespec *ats); int kern_clock_settime(struct thread *td, clockid_t clock_id, struct timespec *ats); int kern_close(struct thread *td, int fd); int kern_connectat(struct thread *td, int dirfd, int fd, struct sockaddr *sa); int kern_dup(struct thread *td, u_int mode, int flags, int old, int new); int kern_execve(struct thread *td, struct image_args *args, struct mac *mac_p); int kern_fchmodat(struct thread *td, int fd, char *path, enum uio_seg pathseg, mode_t mode, int flag); int kern_fchownat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int uid, int gid, int flag); int kern_fcntl(struct thread *td, int fd, int cmd, intptr_t arg); int kern_fcntl_freebsd(struct thread *td, int fd, int cmd, long arg); int kern_fhstat(struct thread *td, fhandle_t fh, struct stat *buf); int kern_fhstatfs(struct thread *td, fhandle_t fh, struct statfs *buf); int kern_fstat(struct thread *td, int fd, struct stat *sbp); int kern_fstatfs(struct thread *td, int fd, struct statfs *buf); int kern_ftruncate(struct thread *td, int fd, off_t length); int kern_futimes(struct thread *td, int fd, struct timeval *tptr, enum uio_seg tptrseg); int kern_futimens(struct thread *td, int fd, struct timespec *tptr, enum uio_seg tptrseg); int kern_getdirentries(struct thread *td, int fd, char *buf, u_int count, long *basep, ssize_t *residp, enum uio_seg bufseg); int kern_getfsstat(struct thread *td, struct statfs **buf, size_t bufsize, size_t *countp, enum uio_seg bufseg, int flags); int kern_getitimer(struct thread *, u_int, struct itimerval *); int kern_getppid(struct thread *); int kern_getpeername(struct thread *td, int fd, struct sockaddr **sa, socklen_t *alen); int kern_getrusage(struct thread *td, int who, struct rusage *rup); int kern_getsockname(struct thread *td, int fd, struct sockaddr **sa, socklen_t *alen); int kern_getsockopt(struct thread *td, int s, int level, int name, void *optval, enum uio_seg valseg, socklen_t *valsize); int kern_ioctl(struct thread *td, int fd, u_long com, caddr_t data); int kern_jail(struct thread *td, struct jail *j); int kern_jail_get(struct thread *td, struct uio *options, int flags); int kern_jail_set(struct thread *td, struct uio *options, int flags); int kern_kevent(struct thread *td, int fd, int nchanges, int nevents, struct kevent_copyops *k_ops, const struct timespec *timeout); int kern_kevent_fp(struct thread *td, struct file *fp, int nchanges, int nevents, struct kevent_copyops *k_ops, const struct timespec *timeout); int kern_kqueue(struct thread *td, int flags); int kern_kldload(struct thread *td, const char *file, int *fileid); int kern_kldstat(struct thread *td, int fileid, struct kld_file_stat *stat); int kern_kldunload(struct thread *td, int fileid, int flags); int kern_linkat(struct thread *td, int fd1, int fd2, char *path1, char *path2, enum uio_seg segflg, int follow); int kern_lutimes(struct thread *td, char *path, enum uio_seg pathseg, struct timeval *tptr, enum uio_seg tptrseg); int kern_mkdirat(struct thread *td, int fd, char *path, enum uio_seg segflg, int mode); int kern_mkfifoat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int mode); int kern_mknodat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int mode, int dev); int kern_msgctl(struct thread *, int, int, struct msqid_ds *); int kern_msgsnd(struct thread *, int, const void *, size_t, int, long); int kern_msgrcv(struct thread *, int, void *, size_t, long, int, long *); int kern_nanosleep(struct thread *td, struct timespec *rqt, struct timespec *rmt); int kern_ogetdirentries(struct thread *td, struct ogetdirentries_args *uap, long *ploff); int kern_openat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int flags, int mode); int kern_pathconf(struct thread *td, char *path, enum uio_seg pathseg, int name, u_long flags); -int kern_pipe(struct thread *td, int fildes[2]); -int kern_pipe2(struct thread *td, int fildes[2], int flags); +int kern_pipe(struct thread *td, int fildes[2], int flags, + struct filecaps *fcaps1, struct filecaps *fcaps2); int kern_poll(struct thread *td, struct pollfd *fds, u_int nfds, struct timespec *tsp, sigset_t *uset); int kern_posix_fadvise(struct thread *td, int fd, off_t offset, off_t len, int advice); int kern_posix_fallocate(struct thread *td, int fd, off_t offset, off_t len); int kern_procctl(struct thread *td, enum idtype idtype, id_t id, int com, void *data); int kern_preadv(struct thread *td, int fd, struct uio *auio, off_t offset); int kern_pselect(struct thread *td, int nd, fd_set *in, fd_set *ou, fd_set *ex, struct timeval *tvp, sigset_t *uset, int abi_nfdbits); int kern_ptrace(struct thread *td, int req, pid_t pid, void *addr, int data); int kern_pwritev(struct thread *td, int fd, struct uio *auio, off_t offset); int kern_readlinkat(struct thread *td, int fd, char *path, enum uio_seg pathseg, char *buf, enum uio_seg bufseg, size_t count); int kern_readv(struct thread *td, int fd, struct uio *auio); int kern_recvit(struct thread *td, int s, struct msghdr *mp, enum uio_seg fromseg, struct mbuf **controlp); int kern_renameat(struct thread *td, int oldfd, char *old, int newfd, char *new, enum uio_seg pathseg); int kern_rmdirat(struct thread *td, int fd, char *path, enum uio_seg pathseg); int kern_sched_getparam(struct thread *td, struct thread *targettd, struct sched_param *param); int kern_sched_getscheduler(struct thread *td, struct thread *targettd, int *policy); int kern_sched_setparam(struct thread *td, struct thread *targettd, struct sched_param *param); int kern_sched_setscheduler(struct thread *td, struct thread *targettd, int policy, struct sched_param *param); int kern_sched_rr_get_interval(struct thread *td, pid_t pid, struct timespec *ts); int kern_sched_rr_get_interval_td(struct thread *td, struct thread *targettd, struct timespec *ts); int kern_semctl(struct thread *td, int semid, int semnum, int cmd, union semun *arg, register_t *rval); int kern_select(struct thread *td, int nd, fd_set *fd_in, fd_set *fd_ou, fd_set *fd_ex, struct timeval *tvp, int abi_nfdbits); int kern_sendfile(struct thread *td, struct sendfile_args *uap, struct uio *hdr_uio, struct uio *trl_uio, int compat); int kern_sendit(struct thread *td, int s, struct msghdr *mp, int flags, struct mbuf *control, enum uio_seg segflg); int kern_setgroups(struct thread *td, u_int ngrp, gid_t *groups); int kern_setitimer(struct thread *, u_int, struct itimerval *, struct itimerval *); int kern_setrlimit(struct thread *, u_int, struct rlimit *); int kern_setsockopt(struct thread *td, int s, int level, int name, void *optval, enum uio_seg valseg, socklen_t valsize); int kern_settimeofday(struct thread *td, struct timeval *tv, struct timezone *tzp); int kern_shmat(struct thread *td, int shmid, const void *shmaddr, int shmflg); int kern_shmctl(struct thread *td, int shmid, int cmd, void *buf, size_t *bufsz); int kern_sigaction(struct thread *td, int sig, const struct sigaction *act, struct sigaction *oact, int flags); int kern_sigaltstack(struct thread *td, stack_t *ss, stack_t *oss); int kern_sigprocmask(struct thread *td, int how, sigset_t *set, sigset_t *oset, int flags); int kern_sigsuspend(struct thread *td, sigset_t mask); int kern_sigtimedwait(struct thread *td, sigset_t waitset, struct ksiginfo *ksi, struct timespec *timeout); int kern_statat(struct thread *td, int flag, int fd, char *path, enum uio_seg pathseg, struct stat *sbp, void (*hook)(struct vnode *vp, struct stat *sbp)); int kern_statfs(struct thread *td, char *path, enum uio_seg pathseg, struct statfs *buf); int kern_symlinkat(struct thread *td, char *path1, int fd, char *path2, enum uio_seg segflg); int kern_ktimer_create(struct thread *td, clockid_t clock_id, struct sigevent *evp, int *timerid, int preset_id); int kern_ktimer_delete(struct thread *, int); int kern_ktimer_settime(struct thread *td, int timer_id, int flags, struct itimerspec *val, struct itimerspec *oval); int kern_ktimer_gettime(struct thread *td, int timer_id, struct itimerspec *val); int kern_ktimer_getoverrun(struct thread *td, int timer_id); int kern_thr_alloc(struct proc *, int pages, struct thread **); int kern_thr_exit(struct thread *td); int kern_thr_new(struct thread *td, struct thr_param *param); int kern_thr_suspend(struct thread *td, struct timespec *tsp); int kern_truncate(struct thread *td, char *path, enum uio_seg pathseg, off_t length); int kern_unlinkat(struct thread *td, int fd, char *path, enum uio_seg pathseg, ino_t oldinum); int kern_utimesat(struct thread *td, int fd, char *path, enum uio_seg pathseg, struct timeval *tptr, enum uio_seg tptrseg); int kern_utimensat(struct thread *td, int fd, char *path, enum uio_seg pathseg, struct timespec *tptr, enum uio_seg tptrseg, int follow); int kern_wait(struct thread *td, pid_t pid, int *status, int options, struct rusage *rup); int kern_wait6(struct thread *td, enum idtype idtype, id_t id, int *status, int options, struct __wrusage *wrup, siginfo_t *sip); int kern_writev(struct thread *td, int fd, struct uio *auio); int kern_socketpair(struct thread *td, int domain, int type, int protocol, int *rsv); /* flags for kern_sigaction */ #define KSA_OSIGSET 0x0001 /* uses osigact_t */ #define KSA_FREEBSD4 0x0002 /* uses ucontext4 */ #endif /* !_SYS_SYSCALLSUBR_H_ */