xfs: use KM_NOFS for allocations during attribute list operations
When listing attributes, we are doiing memory allocations under the
inode ilock using only KM_SLEEP. This allows memory allocation to
recurse back into the filesystem and do writeback, which may the
ilock we already hold on the current inode. THis will deadlock.
Hence use KM_NOFS for such allocations outside of transaction
context to ensure that reclaim recursion does not occur.
Reported-by: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
diff --git a/fs/xfs/xfs_attr_leaf.c b/fs/xfs/xfs_attr_leaf.c
index a6cff8e..71e90dc2 100644
--- a/fs/xfs/xfs_attr_leaf.c
+++ b/fs/xfs/xfs_attr_leaf.c
@@ -637,7 +637,7 @@
* It didn't all fit, so we have to sort everything on hashval.
*/
sbsize = sf->hdr.count * sizeof(*sbuf);
- sbp = sbuf = kmem_alloc(sbsize, KM_SLEEP);
+ sbp = sbuf = kmem_alloc(sbsize, KM_SLEEP | KM_NOFS);
/*
* Scan the attribute list for the rest of the entries, storing
@@ -2386,7 +2386,7 @@
args.dp = context->dp;
args.whichfork = XFS_ATTR_FORK;
args.valuelen = valuelen;
- args.value = kmem_alloc(valuelen, KM_SLEEP);
+ args.value = kmem_alloc(valuelen, KM_SLEEP | KM_NOFS);
args.rmtblkno = be32_to_cpu(name_rmt->valueblk);
args.rmtblkcnt = XFS_B_TO_FSB(args.dp->i_mount, valuelen);
retval = xfs_attr_rmtval_get(&args);