diff options
author | Miklos Szeredi <mszeredi@suse.cz> | 2011-11-21 06:11:31 -0500 |
---|---|---|
committer | Al Viro <viro@zeniv.linux.org.uk> | 2012-01-06 23:20:12 -0500 |
commit | 4ed5e82fe77f4147cf386327c9a63a2dd7eff518 (patch) | |
tree | f4eaeefaf5d293014457892ac31f878eece07331 /fs/namespace.c | |
parent | 39f7c4db1d2d9e2e2a90abdf34811783089d217d (diff) |
vfs: protect remounting superblock read-only
Currently remouting superblock read-only is racy in a major way.
With the per mount read-only infrastructure it is now possible to
prevent most races, which this patch attempts.
Before starting the remount read-only, iterate through all mounts
belonging to the superblock and if none of them have any pending
writes, set sb->s_readonly_remount. This indicates that remount is in
progress and no further write requests are allowed. If the remount
succeeds set MS_RDONLY and reset s_readonly_remount.
If the remounting is unsuccessful just reset s_readonly_remount.
This can result in transient EROFS errors, despite the fact the
remount failed. Unfortunately hodling off writes is difficult as
remount itself may touch the filesystem (e.g. through load_nls())
which would deadlock.
A later patch deals with delayed writes due to nlink going to zero.
Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Tested-by: Toshiyuki Okajima <toshi.okajima@jp.fujitsu.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Diffstat (limited to 'fs/namespace.c')
-rw-r--r-- | fs/namespace.c | 40 |
1 files changed, 39 insertions, 1 deletions
diff --git a/fs/namespace.c b/fs/namespace.c index 145217b088d1..98ebc78b21ab 100644 --- a/fs/namespace.c +++ b/fs/namespace.c | |||
@@ -273,6 +273,15 @@ static unsigned int mnt_get_writers(struct mount *mnt) | |||
273 | #endif | 273 | #endif |
274 | } | 274 | } |
275 | 275 | ||
276 | static int mnt_is_readonly(struct vfsmount *mnt) | ||
277 | { | ||
278 | if (mnt->mnt_sb->s_readonly_remount) | ||
279 | return 1; | ||
280 | /* Order wrt setting s_flags/s_readonly_remount in do_remount() */ | ||
281 | smp_rmb(); | ||
282 | return __mnt_is_readonly(mnt); | ||
283 | } | ||
284 | |||
276 | /* | 285 | /* |
277 | * Most r/o checks on a fs are for operations that take | 286 | * Most r/o checks on a fs are for operations that take |
278 | * discrete amounts of time, like a write() or unlink(). | 287 | * discrete amounts of time, like a write() or unlink(). |
@@ -312,7 +321,7 @@ int mnt_want_write(struct vfsmount *m) | |||
312 | * MNT_WRITE_HOLD is cleared. | 321 | * MNT_WRITE_HOLD is cleared. |
313 | */ | 322 | */ |
314 | smp_rmb(); | 323 | smp_rmb(); |
315 | if (__mnt_is_readonly(m)) { | 324 | if (mnt_is_readonly(m)) { |
316 | mnt_dec_writers(mnt); | 325 | mnt_dec_writers(mnt); |
317 | ret = -EROFS; | 326 | ret = -EROFS; |
318 | goto out; | 327 | goto out; |
@@ -435,6 +444,35 @@ static void __mnt_unmake_readonly(struct mount *mnt) | |||
435 | br_write_unlock(vfsmount_lock); | 444 | br_write_unlock(vfsmount_lock); |
436 | } | 445 | } |
437 | 446 | ||
447 | int sb_prepare_remount_readonly(struct super_block *sb) | ||
448 | { | ||
449 | struct mount *mnt; | ||
450 | int err = 0; | ||
451 | |||
452 | br_write_lock(vfsmount_lock); | ||
453 | list_for_each_entry(mnt, &sb->s_mounts, mnt_instance) { | ||
454 | if (!(mnt->mnt.mnt_flags & MNT_READONLY)) { | ||
455 | mnt->mnt.mnt_flags |= MNT_WRITE_HOLD; | ||
456 | smp_mb(); | ||
457 | if (mnt_get_writers(mnt) > 0) { | ||
458 | err = -EBUSY; | ||
459 | break; | ||
460 | } | ||
461 | } | ||
462 | } | ||
463 | if (!err) { | ||
464 | sb->s_readonly_remount = 1; | ||
465 | smp_wmb(); | ||
466 | } | ||
467 | list_for_each_entry(mnt, &sb->s_mounts, mnt_instance) { | ||
468 | if (mnt->mnt.mnt_flags & MNT_WRITE_HOLD) | ||
469 | mnt->mnt.mnt_flags &= ~MNT_WRITE_HOLD; | ||
470 | } | ||
471 | br_write_unlock(vfsmount_lock); | ||
472 | |||
473 | return err; | ||
474 | } | ||
475 | |||
438 | static void free_vfsmnt(struct mount *mnt) | 476 | static void free_vfsmnt(struct mount *mnt) |
439 | { | 477 | { |
440 | kfree(mnt->mnt_devname); | 478 | kfree(mnt->mnt_devname); |