On 7 Apr 2017, at 8:22, Jeff Layton wrote:
On Thu, 2017-04-06 at 07:23 -0400, Benjamin Coddington wrote:
NFS attempts to wait for read and write completion before unlocking
in
order to ensure that the data returned was protected by the lock.
When
this waiting is interrupted by a signal, the unlock may be skipped,
and
messages similar to the following are seen in the kernel ring buffer:
[20.167876] Leaked locks on dev=0x0:0x2b ino=0x8dd4c3:
[20.168286] POSIX: fl_owner=ffff880078b06940 fl_flags=0x1 fl_type=0x0
fl_pid=20183
[20.168727] POSIX: fl_owner=ffff880078b06680 fl_flags=0x1 fl_type=0x0
fl_pid=20185
For NFSv3, the missing unlock will cause the server to refuse
conflicting
locks indefinitely. For NFSv4, the leftover lock will be removed by
the
server after the lease timeout.
This patch fixes this issue by skipping the usual wait in
nfs_iocounter_wait if the FL_CLOSE flag is set when signaled.
Instead, the
wait happens in the unlock RPC task on the NFS UOC rpc_waitqueue.
For NFSv3, use lockd's new nlmclnt_operations along with
nfs_async_iocounter_wait to defer NLM's unlock task until the lock
context's iocounter reaches zero.
For NFSv4, call nfs_async_iocounter_wait() directly from unlock's
current rpc_call_prepare.
Dare I ask about v2? :)
Hmm actually, it looks like v2 could use the same operations as v3. I
don't see anything protocol-specific there.
Signed-off-by: Benjamin Coddington <bcodding@xxxxxxxxxx>
---
fs/nfs/file.c | 10 +++++-----
fs/nfs/nfs3proc.c | 38 +++++++++++++++++++++++++++++++++++++-
fs/nfs/nfs4proc.c | 10 +++++++---
3 files changed, 49 insertions(+), 9 deletions(-)
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index a490f45df4db..df695f52bb9d 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -697,14 +697,14 @@ do_unlk(struct file *filp, int cmd, struct
file_lock *fl, int is_local)
if (!IS_ERR(l_ctx)) {
status = nfs_iocounter_wait(l_ctx);
nfs_put_lock_context(l_ctx);
- if (status < 0)
+ /* NOTE: special case
+ * If we're signalled while cleaning up locks on process exit, we
+ * still need to complete the unlock.
+ */
+ if (status < 0 && !(fl->fl_flags & FL_CLOSE))
return status;
}
- /* NOTE: special case
- * If we're signalled while cleaning up locks on process exit, we
- * still need to complete the unlock.
- */
/*
* Use local locking if mounted with "-onolock" or with appropriate
* "-olocal_lock="
diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c
index 086623ab07b1..0e21306bfed9 100644
--- a/fs/nfs/nfs3proc.c
+++ b/fs/nfs/nfs3proc.c
@@ -865,12 +865,48 @@ static void nfs3_proc_commit_setup(struct
nfs_commit_data *data, struct rpc_mess
msg->rpc_proc = &nfs3_procedures[NFS3PROC_COMMIT];
}
+void nfs3_nlm_alloc_call(void *data)
+{
+ struct nfs_lock_context *l_ctx = data;
+ nfs_get_lock_context(l_ctx->open_context);
+}
+
+void nfs3_nlm_release_call(void *data)
+{
+ struct nfs_lock_context *l_ctx = data;
+ nfs_put_lock_context(l_ctx);
+}
+
+const struct nlmclnt_operations nlmclnt_fl_close_lock_ops = {
+ .nlmclnt_alloc_call = nfs3_nlm_alloc_call,
+ .nlmclnt_unlock_prepare = nfs_async_iocounter_wait,
+ .nlmclnt_release_call = nfs3_nlm_release_call,
+};
+
static int
nfs3_proc_lock(struct file *filp, int cmd, struct file_lock *fl)
{
struct inode *inode = file_inode(filp);
+ struct nfs_lock_context *l_ctx = NULL;
+ const struct nlmclnt_operations *nlmclnt_ops = NULL;
+ int status;
- return nlmclnt_proc(NFS_SERVER(inode)->nlm_host, cmd, fl, NULL,
NULL);
+ if (fl->fl_flags & FL_CLOSE) {
+ l_ctx = nfs_get_lock_context(nfs_file_open_context(filp));
+ if (IS_ERR(l_ctx)) {
+ l_ctx = NULL;
+ } else {
+ set_bit(NFS_CONTEXT_UNLOCK, &l_ctx->open_context->flags);
+ nlmclnt_ops = &nlmclnt_fl_close_lock_ops;
FWIW, I'm not a huge fan of using the pointer to indicate whether to
run
the operations or not. IMO, it'd be cleaner to:
1) store the pointer to the operations struct in the nlm_host, pass it
a
pointer to it in the nlmclnt_initdata.
2) make the decision to do the operations or not based on the setting
of
NFS_CONTEXT_UNLOCK in the callbacks themselves. If it's not set, just
return quickly without doing anything.
Yes, I can see how that could be a bit neater. And I can add in the
same
behavior for v2 while making this change.
I'll work it up that way and send it again.
Ben
--
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
