summaryrefslogtreecommitdiff
path: root/include
diff options
context:
space:
mode:
authorTrond Myklebust <trond.myklebust@hammerspace.com>2018-06-19 02:07:24 +0300
committerTrond Myklebust <trond.myklebust@hammerspace.com>2018-06-19 16:25:27 +0300
commit42f86b44a4d356edba626171dfe0be061fc695af (patch)
tree664b44aa67881d0ad2371d9d281babf45f69838c /include
parent0dae72d581dfe795aedaf5523c1faeb18958b1a7 (diff)
downloadlinux-42f86b44a4d356edba626171dfe0be061fc695af.tar.xz
pNFS/flexfiles: Don't tie up all the rpciod threads in resends
We do not want to have rpciod threads perform recursive calls into the RPC layer since that can deadlock. In particular, having to wait for a layoutget can be nasty... We want rather to defer scheduling those retries until we're in the rpc_release() callback, since that is called from the nfsiod workqueue. Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Diffstat (limited to 'include')
-rw-r--r--include/linux/nfs_xdr.h2
1 files changed, 2 insertions, 0 deletions
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index 9dee3c23895d..712eed156d09 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -1438,6 +1438,8 @@ enum {
NFS_IOHDR_EOF,
NFS_IOHDR_REDO,
NFS_IOHDR_STAT,
+ NFS_IOHDR_RESEND_PNFS,
+ NFS_IOHDR_RESEND_MDS,
};
struct nfs_io_completion;