diff options
author | Trond Myklebust <Trond.Myklebust@netapp.com> | 2006-03-20 13:44:51 -0500 |
---|---|---|
committer | Trond Myklebust <Trond.Myklebust@netapp.com> | 2006-03-20 13:44:51 -0500 |
commit | c42de9dd67250fe984e0e31c9b542d721af6454b (patch) | |
tree | a4079b25a044f7124837f572e5342dc7614ca75d /fs/nfs | |
parent | 7d46a49f512e8d10b23353781a8ba85edd4fa640 (diff) |
NFS: Fix a race in nfs_sync_inode()
Kudos to Neil Brown for spotting the problem:
"in nfs_sync_inode, there is effectively the sequence:
nfs_wait_on_requests
nfs_flush_inode
nfs_commit_inode
This seems a bit racy to me as if the only requests are on the
->commit list, and nfs_commit_inode is called separately after
nfs_wait_on_requests completes, and before nfs_commit_inode start
(say: by nfs_write_inode) then none of these function will return
>0, yet there will be some pending request that aren't waited for."
The solution is to search for requests to wait upon, search for dirty
requests, and search for uncommitted requests while holding the
nfsi->req_lock
The patch also cleans up nfs_sync_inode(), getting rid of the redundant
FLUSH_WAIT flag. It turns out that we were always setting it.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Diffstat (limited to 'fs/nfs')
-rw-r--r-- | fs/nfs/inode.c | 4 | ||||
-rw-r--r-- | fs/nfs/write.c | 72 |
2 files changed, 52 insertions, 24 deletions
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index a0cda53461b3..60aac58270a8 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c | |||
@@ -137,7 +137,7 @@ nfs_fattr_to_ino_t(struct nfs_fattr *fattr) | |||
137 | static int | 137 | static int |
138 | nfs_write_inode(struct inode *inode, int sync) | 138 | nfs_write_inode(struct inode *inode, int sync) |
139 | { | 139 | { |
140 | int flags = sync ? FLUSH_WAIT : 0; | 140 | int flags = sync ? FLUSH_SYNC : 0; |
141 | int ret; | 141 | int ret; |
142 | 142 | ||
143 | ret = nfs_commit_inode(inode, flags); | 143 | ret = nfs_commit_inode(inode, flags); |
@@ -1051,7 +1051,7 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat) | |||
1051 | int err; | 1051 | int err; |
1052 | 1052 | ||
1053 | /* Flush out writes to the server in order to update c/mtime */ | 1053 | /* Flush out writes to the server in order to update c/mtime */ |
1054 | nfs_sync_inode(inode, 0, 0, FLUSH_WAIT|FLUSH_NOCOMMIT); | 1054 | nfs_sync_inode_wait(inode, 0, 0, FLUSH_NOCOMMIT); |
1055 | 1055 | ||
1056 | /* | 1056 | /* |
1057 | * We may force a getattr if the user cares about atime. | 1057 | * We may force a getattr if the user cares about atime. |
diff --git a/fs/nfs/write.c b/fs/nfs/write.c index 2beb1268bb1b..3f5225404c97 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c | |||
@@ -539,8 +539,7 @@ nfs_mark_request_commit(struct nfs_page *req) | |||
539 | * | 539 | * |
540 | * Interruptible by signals only if mounted with intr flag. | 540 | * Interruptible by signals only if mounted with intr flag. |
541 | */ | 541 | */ |
542 | static int | 542 | static int nfs_wait_on_requests_locked(struct inode *inode, unsigned long idx_start, unsigned int npages) |
543 | nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, unsigned int npages) | ||
544 | { | 543 | { |
545 | struct nfs_inode *nfsi = NFS_I(inode); | 544 | struct nfs_inode *nfsi = NFS_I(inode); |
546 | struct nfs_page *req; | 545 | struct nfs_page *req; |
@@ -553,7 +552,6 @@ nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, unsigned int | |||
553 | else | 552 | else |
554 | idx_end = idx_start + npages - 1; | 553 | idx_end = idx_start + npages - 1; |
555 | 554 | ||
556 | spin_lock(&nfsi->req_lock); | ||
557 | next = idx_start; | 555 | next = idx_start; |
558 | while (radix_tree_gang_lookup_tag(&nfsi->nfs_page_tree, (void **)&req, next, 1, NFS_PAGE_TAG_WRITEBACK)) { | 556 | while (radix_tree_gang_lookup_tag(&nfsi->nfs_page_tree, (void **)&req, next, 1, NFS_PAGE_TAG_WRITEBACK)) { |
559 | if (req->wb_index > idx_end) | 557 | if (req->wb_index > idx_end) |
@@ -566,15 +564,25 @@ nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, unsigned int | |||
566 | spin_unlock(&nfsi->req_lock); | 564 | spin_unlock(&nfsi->req_lock); |
567 | error = nfs_wait_on_request(req); | 565 | error = nfs_wait_on_request(req); |
568 | nfs_release_request(req); | 566 | nfs_release_request(req); |
567 | spin_lock(&nfsi->req_lock); | ||
569 | if (error < 0) | 568 | if (error < 0) |
570 | return error; | 569 | return error; |
571 | spin_lock(&nfsi->req_lock); | ||
572 | res++; | 570 | res++; |
573 | } | 571 | } |
574 | spin_unlock(&nfsi->req_lock); | ||
575 | return res; | 572 | return res; |
576 | } | 573 | } |
577 | 574 | ||
575 | static int nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, unsigned int npages) | ||
576 | { | ||
577 | struct nfs_inode *nfsi = NFS_I(inode); | ||
578 | int ret; | ||
579 | |||
580 | spin_lock(&nfsi->req_lock); | ||
581 | ret = nfs_wait_on_requests_locked(inode, idx_start, npages); | ||
582 | spin_unlock(&nfsi->req_lock); | ||
583 | return ret; | ||
584 | } | ||
585 | |||
578 | /* | 586 | /* |
579 | * nfs_scan_dirty - Scan an inode for dirty requests | 587 | * nfs_scan_dirty - Scan an inode for dirty requests |
580 | * @inode: NFS inode to scan | 588 | * @inode: NFS inode to scan |
@@ -626,6 +634,11 @@ nfs_scan_commit(struct inode *inode, struct list_head *dst, unsigned long idx_st | |||
626 | } | 634 | } |
627 | return res; | 635 | return res; |
628 | } | 636 | } |
637 | #else | ||
638 | static inline int nfs_scan_commit(struct inode *inode, struct list_head *dst, unsigned long idx_start, unsigned int npages) | ||
639 | { | ||
640 | return 0; | ||
641 | } | ||
629 | #endif | 642 | #endif |
630 | 643 | ||
631 | static int nfs_wait_on_write_congestion(struct address_space *mapping, int intr) | 644 | static int nfs_wait_on_write_congestion(struct address_space *mapping, int intr) |
@@ -1421,6 +1434,11 @@ static const struct rpc_call_ops nfs_commit_ops = { | |||
1421 | .rpc_call_done = nfs_commit_done, | 1434 | .rpc_call_done = nfs_commit_done, |
1422 | .rpc_release = nfs_commit_release, | 1435 | .rpc_release = nfs_commit_release, |
1423 | }; | 1436 | }; |
1437 | #else | ||
1438 | static inline int nfs_commit_list(struct inode *inode, struct list_head *head, int how) | ||
1439 | { | ||
1440 | return 0; | ||
1441 | } | ||
1424 | #endif | 1442 | #endif |
1425 | 1443 | ||
1426 | static int nfs_flush_inode(struct inode *inode, unsigned long idx_start, | 1444 | static int nfs_flush_inode(struct inode *inode, unsigned long idx_start, |
@@ -1460,28 +1478,38 @@ int nfs_commit_inode(struct inode *inode, int how) | |||
1460 | } | 1478 | } |
1461 | #endif | 1479 | #endif |
1462 | 1480 | ||
1463 | int nfs_sync_inode(struct inode *inode, unsigned long idx_start, | 1481 | int nfs_sync_inode_wait(struct inode *inode, unsigned long idx_start, |
1464 | unsigned int npages, int how) | 1482 | unsigned int npages, int how) |
1465 | { | 1483 | { |
1484 | struct nfs_inode *nfsi = NFS_I(inode); | ||
1485 | LIST_HEAD(head); | ||
1466 | int nocommit = how & FLUSH_NOCOMMIT; | 1486 | int nocommit = how & FLUSH_NOCOMMIT; |
1467 | int wait = how & FLUSH_WAIT; | 1487 | int pages, ret; |
1468 | int error; | ||
1469 | |||
1470 | how &= ~(FLUSH_WAIT|FLUSH_NOCOMMIT); | ||
1471 | 1488 | ||
1489 | how &= ~FLUSH_NOCOMMIT; | ||
1490 | spin_lock(&nfsi->req_lock); | ||
1472 | do { | 1491 | do { |
1473 | if (wait) { | 1492 | ret = nfs_wait_on_requests_locked(inode, idx_start, npages); |
1474 | error = nfs_wait_on_requests(inode, idx_start, npages); | 1493 | if (ret != 0) |
1475 | if (error != 0) | ||
1476 | continue; | ||
1477 | } | ||
1478 | error = nfs_flush_inode(inode, idx_start, npages, how); | ||
1479 | if (error != 0) | ||
1480 | continue; | 1494 | continue; |
1481 | if (!nocommit) | 1495 | pages = nfs_scan_dirty(inode, &head, idx_start, npages); |
1482 | error = nfs_commit_inode(inode, how); | 1496 | if (pages != 0) { |
1483 | } while (error > 0); | 1497 | spin_unlock(&nfsi->req_lock); |
1484 | return error; | 1498 | ret = nfs_flush_list(inode, &head, pages, how); |
1499 | spin_lock(&nfsi->req_lock); | ||
1500 | continue; | ||
1501 | } | ||
1502 | if (nocommit) | ||
1503 | break; | ||
1504 | pages = nfs_scan_commit(inode, &head, 0, 0); | ||
1505 | if (pages == 0) | ||
1506 | break; | ||
1507 | spin_unlock(&nfsi->req_lock); | ||
1508 | ret = nfs_commit_list(inode, &head, how); | ||
1509 | spin_lock(&nfsi->req_lock); | ||
1510 | } while (ret >= 0); | ||
1511 | spin_unlock(&nfsi->req_lock); | ||
1512 | return ret; | ||
1485 | } | 1513 | } |
1486 | 1514 | ||
1487 | int nfs_init_writepagecache(void) | 1515 | int nfs_init_writepagecache(void) |