[PATCH 0/2] remove deadlock risk with loop-back mounted NFS filesystems

[NeilBrown <neilb@suse.de>]

Now that the new wait_on_bit code is in the kernel (in 3.17-rc1) these
last two patches to support deadlock-free loop-back NFS mounts can be
applied (hopefully to go upstream for 3.18-rc1).

A deadlock can happen if nfsd tries to allocate memory, calls
->releasepage() on a page in an NFS filesystem, and nfs_release_page()
blocks waiting for the nfsd to confirm the COMMIT.

With this patch nfs_release_page() will not wait more than 100ms
for COMMIT to a non-remote nfs fileserver.

Thanks,
NeilBrown

---

NeilBrown (2):
      SUNRPC: track when a client connection is routed to the local host.
      NFS: avoid deadlocks with loop-back mounted NFS filesystems.


 fs/nfs/file.c               |    2 +
 fs/nfs/write.c              |   72 ++++++++++++++++++++++++++++++++++++++++---
 include/linux/freezer.h     |   10 ++++++
 include/linux/sunrpc/clnt.h |    1 +
 include/linux/sunrpc/xprt.h |    1 +
 include/uapi/linux/nfs_fs.h |    3 ++
 net/sunrpc/clnt.c           |   25 +++++++++++++++
 net/sunrpc/xprtsock.c       |    9 +++++
 8 files changed, 117 insertions(+), 6 deletions(-)

-- 
Signature