2 * linux/fs/9p/vfs_file.c
4 * This file contians vfs file ops for 9P2000.
6 * Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
7 * Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License version 2
11 * as published by the Free Software Foundation.
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to:
20 * Free Software Foundation
21 * 51 Franklin Street, Fifth Floor
22 * Boston, MA 02111-1301 USA
26 #include <linux/module.h>
27 #include <linux/errno.h>
29 #include <linux/sched.h>
30 #include <linux/file.h>
31 #include <linux/stat.h>
32 #include <linux/string.h>
33 #include <linux/inet.h>
34 #include <linux/list.h>
35 #include <linux/pagemap.h>
36 #include <linux/utsname.h>
37 #include <asm/uaccess.h>
38 #include <linux/idr.h>
39 #include <linux/uio.h>
40 #include <linux/slab.h>
41 #include <net/9p/9p.h>
42 #include <net/9p/client.h>
49 static const struct vm_operations_struct v9fs_file_vm_ops
;
50 static const struct vm_operations_struct v9fs_mmap_file_vm_ops
;
53 * v9fs_file_open - open a file (or directory)
54 * @inode: inode to be opened
55 * @file: file being opened
59 int v9fs_file_open(struct inode
*inode
, struct file
*file
)
62 struct v9fs_inode
*v9inode
;
63 struct v9fs_session_info
*v9ses
;
67 p9_debug(P9_DEBUG_VFS
, "inode: %p file: %p\n", inode
, file
);
68 v9inode
= V9FS_I(inode
);
69 v9ses
= v9fs_inode2v9ses(inode
);
70 if (v9fs_proto_dotl(v9ses
))
71 omode
= v9fs_open_to_dotl_flags(file
->f_flags
);
73 omode
= v9fs_uflags2omode(file
->f_flags
,
74 v9fs_proto_dotu(v9ses
));
75 fid
= file
->private_data
;
77 fid
= v9fs_fid_clone(file_dentry(file
));
81 err
= p9_client_open(fid
, omode
);
86 if ((file
->f_flags
& O_APPEND
) &&
87 (!v9fs_proto_dotu(v9ses
) && !v9fs_proto_dotl(v9ses
)))
88 generic_file_llseek(file
, 0, SEEK_END
);
91 file
->private_data
= fid
;
92 mutex_lock(&v9inode
->v_mutex
);
93 if ((v9ses
->cache
== CACHE_LOOSE
|| v9ses
->cache
== CACHE_FSCACHE
) &&
94 !v9inode
->writeback_fid
&&
95 ((file
->f_flags
& O_ACCMODE
) != O_RDONLY
)) {
97 * clone a fid and add it to writeback_fid
98 * we do it during open time instead of
99 * page dirty time via write_begin/page_mkwrite
100 * because we want write after unlink usecase
103 fid
= v9fs_writeback_fid(file_dentry(file
));
106 mutex_unlock(&v9inode
->v_mutex
);
109 v9inode
->writeback_fid
= (void *) fid
;
111 mutex_unlock(&v9inode
->v_mutex
);
112 if (v9ses
->cache
== CACHE_LOOSE
|| v9ses
->cache
== CACHE_FSCACHE
)
113 v9fs_cache_inode_set_cookie(inode
, file
);
116 p9_client_clunk(file
->private_data
);
117 file
->private_data
= NULL
;
122 * v9fs_file_lock - lock a file (or directory)
123 * @filp: file to be locked
125 * @fl: file lock structure
127 * Bugs: this looks like a local only lock, we should extend into 9P
128 * by using open exclusive
131 static int v9fs_file_lock(struct file
*filp
, int cmd
, struct file_lock
*fl
)
134 struct inode
*inode
= file_inode(filp
);
136 p9_debug(P9_DEBUG_VFS
, "filp: %p lock: %p\n", filp
, fl
);
138 /* No mandatory locks */
139 if (__mandatory_lock(inode
) && fl
->fl_type
!= F_UNLCK
)
142 if ((IS_SETLK(cmd
) || IS_SETLKW(cmd
)) && fl
->fl_type
!= F_UNLCK
) {
143 filemap_write_and_wait(inode
->i_mapping
);
144 invalidate_mapping_pages(&inode
->i_data
, 0, -1);
150 static int v9fs_file_do_lock(struct file
*filp
, int cmd
, struct file_lock
*fl
)
152 struct p9_flock flock
;
154 uint8_t status
= P9_LOCK_ERROR
;
156 unsigned char fl_type
;
157 struct v9fs_session_info
*v9ses
;
159 fid
= filp
->private_data
;
162 if ((fl
->fl_flags
& FL_POSIX
) != FL_POSIX
)
165 res
= locks_lock_file_wait(filp
, fl
);
169 /* convert posix lock to p9 tlock args */
170 memset(&flock
, 0, sizeof(flock
));
171 /* map the lock type */
172 switch (fl
->fl_type
) {
174 flock
.type
= P9_LOCK_TYPE_RDLCK
;
177 flock
.type
= P9_LOCK_TYPE_WRLCK
;
180 flock
.type
= P9_LOCK_TYPE_UNLCK
;
183 flock
.start
= fl
->fl_start
;
184 if (fl
->fl_end
== OFFSET_MAX
)
187 flock
.length
= fl
->fl_end
- fl
->fl_start
+ 1;
188 flock
.proc_id
= fl
->fl_pid
;
189 flock
.client_id
= fid
->clnt
->name
;
191 flock
.flags
= P9_LOCK_FLAGS_BLOCK
;
193 v9ses
= v9fs_inode2v9ses(file_inode(filp
));
196 * if its a blocked request and we get P9_LOCK_BLOCKED as the status
197 * for lock request, keep on trying
200 res
= p9_client_lock_dotl(fid
, &flock
, &status
);
204 if (status
!= P9_LOCK_BLOCKED
)
206 if (status
== P9_LOCK_BLOCKED
&& !IS_SETLKW(cmd
))
208 if (schedule_timeout_interruptible(v9ses
->session_lock_timeout
)
212 * p9_client_lock_dotl overwrites flock.client_id with the
213 * server message, free and reuse the client name
215 if (flock
.client_id
!= fid
->clnt
->name
) {
216 kfree(flock
.client_id
);
217 flock
.client_id
= fid
->clnt
->name
;
221 /* map 9p status to VFS status */
223 case P9_LOCK_SUCCESS
:
226 case P9_LOCK_BLOCKED
:
230 WARN_ONCE(1, "unknown lock status code: %d\n", status
);
240 * incase server returned error for lock request, revert
243 if (res
< 0 && fl
->fl_type
!= F_UNLCK
) {
244 fl_type
= fl
->fl_type
;
245 fl
->fl_type
= F_UNLCK
;
246 /* Even if this fails we want to return the remote error */
247 locks_lock_file_wait(filp
, fl
);
248 fl
->fl_type
= fl_type
;
250 if (flock
.client_id
!= fid
->clnt
->name
)
251 kfree(flock
.client_id
);
256 static int v9fs_file_getlock(struct file
*filp
, struct file_lock
*fl
)
258 struct p9_getlock glock
;
262 fid
= filp
->private_data
;
265 posix_test_lock(filp
, fl
);
267 * if we have a conflicting lock locally, no need to validate
270 if (fl
->fl_type
!= F_UNLCK
)
273 /* convert posix lock to p9 tgetlock args */
274 memset(&glock
, 0, sizeof(glock
));
275 glock
.type
= P9_LOCK_TYPE_UNLCK
;
276 glock
.start
= fl
->fl_start
;
277 if (fl
->fl_end
== OFFSET_MAX
)
280 glock
.length
= fl
->fl_end
- fl
->fl_start
+ 1;
281 glock
.proc_id
= fl
->fl_pid
;
282 glock
.client_id
= fid
->clnt
->name
;
284 res
= p9_client_getlock_dotl(fid
, &glock
);
287 /* map 9p lock type to os lock type */
288 switch (glock
.type
) {
289 case P9_LOCK_TYPE_RDLCK
:
290 fl
->fl_type
= F_RDLCK
;
292 case P9_LOCK_TYPE_WRLCK
:
293 fl
->fl_type
= F_WRLCK
;
295 case P9_LOCK_TYPE_UNLCK
:
296 fl
->fl_type
= F_UNLCK
;
299 if (glock
.type
!= P9_LOCK_TYPE_UNLCK
) {
300 fl
->fl_start
= glock
.start
;
301 if (glock
.length
== 0)
302 fl
->fl_end
= OFFSET_MAX
;
304 fl
->fl_end
= glock
.start
+ glock
.length
- 1;
305 fl
->fl_pid
= glock
.proc_id
;
308 if (glock
.client_id
!= fid
->clnt
->name
)
309 kfree(glock
.client_id
);
314 * v9fs_file_lock_dotl - lock a file (or directory)
315 * @filp: file to be locked
317 * @fl: file lock structure
321 static int v9fs_file_lock_dotl(struct file
*filp
, int cmd
, struct file_lock
*fl
)
323 struct inode
*inode
= file_inode(filp
);
326 p9_debug(P9_DEBUG_VFS
, "filp: %p cmd:%d lock: %p name: %pD\n",
327 filp
, cmd
, fl
, filp
);
329 /* No mandatory locks */
330 if (__mandatory_lock(inode
) && fl
->fl_type
!= F_UNLCK
)
333 if ((IS_SETLK(cmd
) || IS_SETLKW(cmd
)) && fl
->fl_type
!= F_UNLCK
) {
334 filemap_write_and_wait(inode
->i_mapping
);
335 invalidate_mapping_pages(&inode
->i_data
, 0, -1);
338 if (IS_SETLK(cmd
) || IS_SETLKW(cmd
))
339 ret
= v9fs_file_do_lock(filp
, cmd
, fl
);
340 else if (IS_GETLK(cmd
))
341 ret
= v9fs_file_getlock(filp
, fl
);
349 * v9fs_file_flock_dotl - lock a file
350 * @filp: file to be locked
352 * @fl: file lock structure
356 static int v9fs_file_flock_dotl(struct file
*filp
, int cmd
,
357 struct file_lock
*fl
)
359 struct inode
*inode
= file_inode(filp
);
362 p9_debug(P9_DEBUG_VFS
, "filp: %p cmd:%d lock: %p name: %pD\n",
363 filp
, cmd
, fl
, filp
);
365 /* No mandatory locks */
366 if (__mandatory_lock(inode
) && fl
->fl_type
!= F_UNLCK
)
369 if (!(fl
->fl_flags
& FL_FLOCK
))
372 if ((IS_SETLK(cmd
) || IS_SETLKW(cmd
)) && fl
->fl_type
!= F_UNLCK
) {
373 filemap_write_and_wait(inode
->i_mapping
);
374 invalidate_mapping_pages(&inode
->i_data
, 0, -1);
376 /* Convert flock to posix lock */
377 fl
->fl_flags
|= FL_POSIX
;
378 fl
->fl_flags
^= FL_FLOCK
;
380 if (IS_SETLK(cmd
) | IS_SETLKW(cmd
))
381 ret
= v9fs_file_do_lock(filp
, cmd
, fl
);
389 * v9fs_file_read - read from a file
390 * @filp: file pointer to read
391 * @udata: user data buffer to read data into
392 * @count: size of buffer
393 * @offset: offset at which to read data
398 v9fs_file_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
)
400 struct p9_fid
*fid
= iocb
->ki_filp
->private_data
;
403 p9_debug(P9_DEBUG_VFS
, "count %zu offset %lld\n",
404 iov_iter_count(to
), iocb
->ki_pos
);
406 ret
= p9_client_read(fid
, iocb
->ki_pos
, to
, &err
);
415 * v9fs_file_write - write to a file
416 * @filp: file pointer to write
417 * @data: data buffer to write data from
418 * @count: size of buffer
419 * @offset: offset at which to write data
423 v9fs_file_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
)
425 struct file
*file
= iocb
->ki_filp
;
430 retval
= generic_write_checks(iocb
, from
);
434 origin
= iocb
->ki_pos
;
435 retval
= p9_client_write(file
->private_data
, iocb
->ki_pos
, from
, &err
);
437 struct inode
*inode
= file_inode(file
);
439 unsigned long pg_start
, pg_end
;
440 pg_start
= origin
>> PAGE_SHIFT
;
441 pg_end
= (origin
+ retval
- 1) >> PAGE_SHIFT
;
442 if (inode
->i_mapping
&& inode
->i_mapping
->nrpages
)
443 invalidate_inode_pages2_range(inode
->i_mapping
,
445 iocb
->ki_pos
+= retval
;
446 i_size
= i_size_read(inode
);
447 if (iocb
->ki_pos
> i_size
) {
448 inode_add_bytes(inode
, iocb
->ki_pos
- i_size
);
450 * Need to serialize against i_size_write() in
453 v9fs_i_size_write(inode
, iocb
->ki_pos
);
460 static int v9fs_file_fsync(struct file
*filp
, loff_t start
, loff_t end
,
464 struct inode
*inode
= filp
->f_mapping
->host
;
465 struct p9_wstat wstat
;
468 retval
= filemap_write_and_wait_range(inode
->i_mapping
, start
, end
);
473 p9_debug(P9_DEBUG_VFS
, "filp %p datasync %x\n", filp
, datasync
);
475 fid
= filp
->private_data
;
476 v9fs_blank_wstat(&wstat
);
478 retval
= p9_client_wstat(fid
, &wstat
);
484 int v9fs_file_fsync_dotl(struct file
*filp
, loff_t start
, loff_t end
,
488 struct inode
*inode
= filp
->f_mapping
->host
;
491 retval
= filemap_write_and_wait_range(inode
->i_mapping
, start
, end
);
496 p9_debug(P9_DEBUG_VFS
, "filp %p datasync %x\n", filp
, datasync
);
498 fid
= filp
->private_data
;
500 retval
= p9_client_fsync(fid
, datasync
);
507 v9fs_file_mmap(struct file
*filp
, struct vm_area_struct
*vma
)
512 retval
= generic_file_mmap(filp
, vma
);
514 vma
->vm_ops
= &v9fs_file_vm_ops
;
520 v9fs_mmap_file_mmap(struct file
*filp
, struct vm_area_struct
*vma
)
524 struct v9fs_inode
*v9inode
;
527 inode
= file_inode(filp
);
528 v9inode
= V9FS_I(inode
);
529 mutex_lock(&v9inode
->v_mutex
);
530 if (!v9inode
->writeback_fid
&&
531 (vma
->vm_flags
& VM_SHARED
) &&
532 (vma
->vm_flags
& VM_WRITE
)) {
534 * clone a fid and add it to writeback_fid
535 * we do it during mmap instead of
536 * page dirty time via write_begin/page_mkwrite
537 * because we want write after unlink usecase
540 fid
= v9fs_writeback_fid(file_dentry(filp
));
542 retval
= PTR_ERR(fid
);
543 mutex_unlock(&v9inode
->v_mutex
);
546 v9inode
->writeback_fid
= (void *) fid
;
548 mutex_unlock(&v9inode
->v_mutex
);
550 retval
= generic_file_mmap(filp
, vma
);
552 vma
->vm_ops
= &v9fs_mmap_file_vm_ops
;
558 v9fs_vm_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
)
560 struct v9fs_inode
*v9inode
;
561 struct page
*page
= vmf
->page
;
562 struct file
*filp
= vma
->vm_file
;
563 struct inode
*inode
= file_inode(filp
);
566 p9_debug(P9_DEBUG_VFS
, "page %p fid %lx\n",
567 page
, (unsigned long)filp
->private_data
);
569 /* Update file times before taking page lock */
570 file_update_time(filp
);
572 v9inode
= V9FS_I(inode
);
573 /* make sure the cache has finished storing the page */
574 v9fs_fscache_wait_on_page_write(inode
, page
);
575 BUG_ON(!v9inode
->writeback_fid
);
577 if (page
->mapping
!= inode
->i_mapping
)
579 wait_for_stable_page(page
);
581 return VM_FAULT_LOCKED
;
584 return VM_FAULT_NOPAGE
;
588 * v9fs_mmap_file_read - read from a file
589 * @filp: file pointer to read
590 * @data: user data buffer to read data into
591 * @count: size of buffer
592 * @offset: offset at which to read data
596 v9fs_mmap_file_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
)
598 /* TODO: Check if there are dirty pages */
599 return v9fs_file_read_iter(iocb
, to
);
603 * v9fs_mmap_file_write - write to a file
604 * @filp: file pointer to write
605 * @data: data buffer to write data from
606 * @count: size of buffer
607 * @offset: offset at which to write data
611 v9fs_mmap_file_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
)
614 * TODO: invalidate mmaps on filp's inode between
615 * offset and offset+count
617 return v9fs_file_write_iter(iocb
, from
);
620 static void v9fs_mmap_vm_close(struct vm_area_struct
*vma
)
624 struct writeback_control wbc
= {
625 .nr_to_write
= LONG_MAX
,
626 .sync_mode
= WB_SYNC_ALL
,
627 .range_start
= (loff_t
)vma
->vm_pgoff
* PAGE_SIZE
,
628 /* absolute end, byte at end included */
629 .range_end
= (loff_t
)vma
->vm_pgoff
* PAGE_SIZE
+
630 (vma
->vm_end
- vma
->vm_start
- 1),
633 if (!(vma
->vm_flags
& VM_SHARED
))
636 p9_debug(P9_DEBUG_VFS
, "9p VMA close, %p, flushing", vma
);
638 inode
= file_inode(vma
->vm_file
);
640 if (!mapping_cap_writeback_dirty(inode
->i_mapping
))
644 sync_inode(inode
, &wbc
);
648 static const struct vm_operations_struct v9fs_file_vm_ops
= {
649 .fault
= filemap_fault
,
650 .map_pages
= filemap_map_pages
,
651 .page_mkwrite
= v9fs_vm_page_mkwrite
,
654 static const struct vm_operations_struct v9fs_mmap_file_vm_ops
= {
655 .close
= v9fs_mmap_vm_close
,
656 .fault
= filemap_fault
,
657 .map_pages
= filemap_map_pages
,
658 .page_mkwrite
= v9fs_vm_page_mkwrite
,
662 const struct file_operations v9fs_cached_file_operations
= {
663 .llseek
= generic_file_llseek
,
664 .read_iter
= generic_file_read_iter
,
665 .write_iter
= generic_file_write_iter
,
666 .open
= v9fs_file_open
,
667 .release
= v9fs_dir_release
,
668 .lock
= v9fs_file_lock
,
669 .mmap
= v9fs_file_mmap
,
670 .fsync
= v9fs_file_fsync
,
673 const struct file_operations v9fs_cached_file_operations_dotl
= {
674 .llseek
= generic_file_llseek
,
675 .read_iter
= generic_file_read_iter
,
676 .write_iter
= generic_file_write_iter
,
677 .open
= v9fs_file_open
,
678 .release
= v9fs_dir_release
,
679 .lock
= v9fs_file_lock_dotl
,
680 .flock
= v9fs_file_flock_dotl
,
681 .mmap
= v9fs_file_mmap
,
682 .fsync
= v9fs_file_fsync_dotl
,
685 const struct file_operations v9fs_file_operations
= {
686 .llseek
= generic_file_llseek
,
687 .read_iter
= v9fs_file_read_iter
,
688 .write_iter
= v9fs_file_write_iter
,
689 .open
= v9fs_file_open
,
690 .release
= v9fs_dir_release
,
691 .lock
= v9fs_file_lock
,
692 .mmap
= generic_file_readonly_mmap
,
693 .fsync
= v9fs_file_fsync
,
696 const struct file_operations v9fs_file_operations_dotl
= {
697 .llseek
= generic_file_llseek
,
698 .read_iter
= v9fs_file_read_iter
,
699 .write_iter
= v9fs_file_write_iter
,
700 .open
= v9fs_file_open
,
701 .release
= v9fs_dir_release
,
702 .lock
= v9fs_file_lock_dotl
,
703 .flock
= v9fs_file_flock_dotl
,
704 .mmap
= generic_file_readonly_mmap
,
705 .fsync
= v9fs_file_fsync_dotl
,
708 const struct file_operations v9fs_mmap_file_operations
= {
709 .llseek
= generic_file_llseek
,
710 .read_iter
= v9fs_mmap_file_read_iter
,
711 .write_iter
= v9fs_mmap_file_write_iter
,
712 .open
= v9fs_file_open
,
713 .release
= v9fs_dir_release
,
714 .lock
= v9fs_file_lock
,
715 .mmap
= v9fs_mmap_file_mmap
,
716 .fsync
= v9fs_file_fsync
,
719 const struct file_operations v9fs_mmap_file_operations_dotl
= {
720 .llseek
= generic_file_llseek
,
721 .read_iter
= v9fs_mmap_file_read_iter
,
722 .write_iter
= v9fs_mmap_file_write_iter
,
723 .open
= v9fs_file_open
,
724 .release
= v9fs_dir_release
,
725 .lock
= v9fs_file_lock_dotl
,
726 .flock
= v9fs_file_flock_dotl
,
727 .mmap
= v9fs_mmap_file_mmap
,
728 .fsync
= v9fs_file_fsync_dotl
,