diff options
| author | Trond Myklebust <trond.myklebust@primarydata.com> | 2017-05-01 17:06:56 -0400 |
|---|---|---|
| committer | Trond Myklebust <trond.myklebust@primarydata.com> | 2017-05-02 12:35:33 -0400 |
| commit | 61f454e30c18a28924e96be12592c0d5e24bcc81 (patch) | |
| tree | 515adf7a272e0c0f9c57be0c6d7e2ca14f5ba895 /fs/nfs | |
| parent | 5466d21411baf413c838d9a197f6734e94910a65 (diff) | |
| download | linux-61f454e30c18a28924e96be12592c0d5e24bcc81.tar.gz linux-61f454e30c18a28924e96be12592c0d5e24bcc81.tar.bz2 linux-61f454e30c18a28924e96be12592c0d5e24bcc81.zip | |
pNFS: Fix a deadlock when coalescing writes and returning the layout
Consider the following deadlock:
Process P1 Process P2 Process P3
========== ========== ==========
lock_page(page)
lseg = pnfs_update_layout(inode)
lo = NFS_I(inode)->layout
pnfs_error_mark_layout_for_return(lo)
lock_page(page)
lseg = pnfs_update_layout(inode)
In this scenario,
- P1 has declared the layout to be in error, but P2 holds a reference to
a layout segment on that inode, so the layoutreturn is deferred.
- P2 is waiting for a page lock held by P3.
- P3 is asking for a new layout segment, but is blocked waiting
for the layoutreturn.
The fix is to ensure that pnfs_error_mark_layout_for_return() does
not set the NFS_LAYOUT_RETURN flag, which blocks P3. Instead, we allow
the latter to call LAYOUTGET so that it can make progress and unblock
P2.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Diffstat (limited to 'fs/nfs')
| -rw-r--r-- | fs/nfs/pnfs.c | 2 |
1 files changed, 0 insertions, 2 deletions
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index cea1e838efae..adc6ec28d4b5 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -2063,8 +2063,6 @@ void pnfs_error_mark_layout_for_return(struct inode *inode, return; } pnfs_set_plh_return_info(lo, range.iomode, 0); - /* Block LAYOUTGET */ - set_bit(NFS_LAYOUT_RETURN, &lo->plh_flags); /* * mark all matching lsegs so that we are sure to have no live * segments at hand when sending layoutreturn. See pnfs_put_lseg() |