diff options
| author | Trond Myklebust <Trond.Myklebust@netapp.com> | 2012-05-09 14:30:35 -0400 |
|---|---|---|
| committer | Trond Myklebust <Trond.Myklebust@netapp.com> | 2012-05-09 15:16:07 -0400 |
| commit | 3aff4ebb95b20ad8db2c1447e8c52097d89af5a7 (patch) | |
| tree | ad56aceafbcbd0650e71e029ffa138eedffd33ba /fs/nfs/write.c | |
| parent | dc327ed4cd320be689596365372a3683208c3ba0 (diff) | |
| download | linux-3aff4ebb95b20ad8db2c1447e8c52097d89af5a7.tar.bz2 | |
NFS: Prevent a deadlock in the new writeback code
We have to unlock the nfs_page before we call nfs_end_page_writeback
to avoid races with functions that expect the page to be unlocked
when PG_locked and PG_writeback are not set.
The problem is that nfs_unlock_request also releases the nfs_page,
causing a deadlock if the release of the nfs_open_context
triggers an iput() while the PG_writeback flag is still set...
The solution is to separate the unlocking and release of the nfs_page,
so that we can do the former before nfs_end_page_writeback and the
latter after.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Fred Isaman <iisaman@netapp.com>
Diffstat (limited to 'fs/nfs/write.c')
| -rw-r--r-- | fs/nfs/write.c | 6 |
1 files changed, 4 insertions, 2 deletions
diff --git a/fs/nfs/write.c b/fs/nfs/write.c index 6f263daac748..fd36b31ee72e 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -628,8 +628,9 @@ static void nfs_write_completion(struct nfs_pgio_header *hdr) remove_req: nfs_inode_remove_request(req); next: - nfs_unlock_request(req); + nfs_unlock_request_dont_release(req); nfs_end_page_writeback(page); + nfs_release_request(req); } out: hdr->release(hdr); @@ -1042,8 +1043,9 @@ static void nfs_redirty_request(struct nfs_page *req) struct page *page = req->wb_page; nfs_mark_request_dirty(req); - nfs_unlock_request(req); + nfs_unlock_request_dont_release(req); nfs_end_page_writeback(page); + nfs_release_request(req); } static void nfs_async_write_error(struct list_head *head) |