This set of patches is yet another attempt to fix a couple of issues with renewd.
Patch 1/6 makes sure that renewd is synchronously killed before sending a destroy_session to the server.
Patch 2/6 schedules new renewd request only if a previous one returns (error or success).
Patch 3-4/6 fix the umount race by forcing renewd to take/put a client reference, so access to nfs_free_client is serialized.
Patch 5/6 sets a timeout (similar to timeo) for the nfs_client->rpcclient that equals the server advertised lease_time. This ensures that NFS requests fail (so they can be retried) within a lease period.
Patch 6/6 makes the internal timeouts of the RPC layer sensitive to the major timeout value; IOW there is no point in sleeping waiting for an event for longer than the remaining time before a major timeout.
Alexandros Batsakis (6):
nfs: kill renewd before clearing client minor version
nfs: prevent backlogging of renewd requests
nfs41: fix race between umount and renewd sequence operations
nfs4: fix race between umount and renewd renew operations
nfs4: adjust rpc timeout for nfs_client rpc client based on the
lease_time
RPC: adjust timeout for connect, bind, restablish so that they
sensitive to the major time out value
fs/nfs/client.c | 45 +++++++++++++++++++++----------------------
fs/nfs/nfs4proc.c | 50 +++++++++++++++++++++++++++++++++++++++++++++++-
fs/nfs/nfs4renewd.c | 19 ++++++-----------
net/sunrpc/clnt.c | 2 +-
net/sunrpc/sunrpc.h | 2 +
net/sunrpc/xprt.c | 18 ++++++++++++++--
net/sunrpc/xprtsock.c | 12 +++++++++-
7 files changed, 105 insertions(+), 43 deletions(-)
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
