This is a note to let you know that I've just added the patch titled
NFSv4: Fix a state manager thread deadlock regression
to the 6.5-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
The filename of the patch is:
nfsv4-fix-a-state-manager-thread-deadlock-regression.patch
and it can be found in the queue-6.5 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@xxxxxxxxxxxxxxx> know about it.
>From 956fd46f97d238032cb5fa4771cdaccc6e760f9a Mon Sep 17 00:00:00 2001
From: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx>
Date: Sun, 24 Sep 2023 13:14:15 -0400
Subject: NFSv4: Fix a state manager thread deadlock regression
From: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx>
commit 956fd46f97d238032cb5fa4771cdaccc6e760f9a upstream.
Commit 4dc73c679114 reintroduces the deadlock that was fixed by commit
aeabb3c96186 ("NFSv4: Fix a NFSv4 state manager deadlock") because it
prevents the setup of new threads to handle reboot recovery, while the
older recovery thread is stuck returning delegations.
Fixes: 4dc73c679114 ("NFSv4: keep state manager thread active if swap is enabled")
Cc: stable@xxxxxxxxxxxxxxx
Signed-off-by: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx>
Signed-off-by: Anna Schumaker <Anna.Schumaker@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
fs/nfs/nfs4proc.c | 4 +++-
fs/nfs/nfs4state.c | 38 ++++++++++++++++++++++++++------------
2 files changed, 29 insertions(+), 13 deletions(-)
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -10622,7 +10622,9 @@ static void nfs4_disable_swap(struct ino
*/
struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
- nfs4_schedule_state_manager(clp);
+ set_bit(NFS4CLNT_RUN_MANAGER, &clp->cl_state);
+ clear_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state);
+ wake_up_var(&clp->cl_state);
}
static const struct inode_operations nfs4_dir_inode_operations = {
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -1209,16 +1209,26 @@ void nfs4_schedule_state_manager(struct
{
struct task_struct *task;
char buf[INET6_ADDRSTRLEN + sizeof("-manager") + 1];
+ struct rpc_clnt *clnt = clp->cl_rpcclient;
+ bool swapon = false;
- if (clp->cl_rpcclient->cl_shutdown)
+ if (clnt->cl_shutdown)
return;
set_bit(NFS4CLNT_RUN_MANAGER, &clp->cl_state);
- if (test_and_set_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state) != 0) {
- wake_up_var(&clp->cl_state);
- return;
+
+ if (atomic_read(&clnt->cl_swapper)) {
+ swapon = !test_and_set_bit(NFS4CLNT_MANAGER_AVAILABLE,
+ &clp->cl_state);
+ if (!swapon) {
+ wake_up_var(&clp->cl_state);
+ return;
+ }
}
- set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state);
+
+ if (test_and_set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) != 0)
+ return;
+
__module_get(THIS_MODULE);
refcount_inc(&clp->cl_count);
@@ -1235,8 +1245,9 @@ void nfs4_schedule_state_manager(struct
__func__, PTR_ERR(task));
if (!nfs_client_init_is_complete(clp))
nfs_mark_client_ready(clp, PTR_ERR(task));
+ if (swapon)
+ clear_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state);
nfs4_clear_state_manager_bit(clp);
- clear_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state);
nfs_put_client(clp);
module_put(THIS_MODULE);
}
@@ -2741,22 +2752,25 @@ static int nfs4_run_state_manager(void *
allow_signal(SIGKILL);
again:
- set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state);
nfs4_state_manager(clp);
- if (atomic_read(&cl->cl_swapper)) {
+
+ if (test_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state) &&
+ !test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state)) {
wait_var_event_interruptible(&clp->cl_state,
test_bit(NFS4CLNT_RUN_MANAGER,
&clp->cl_state));
- if (atomic_read(&cl->cl_swapper) &&
- test_bit(NFS4CLNT_RUN_MANAGER, &clp->cl_state))
+ if (!atomic_read(&cl->cl_swapper))
+ clear_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state);
+ if (refcount_read(&clp->cl_count) > 1 && !signalled() &&
+ !test_and_set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state))
goto again;
/* Either no longer a swapper, or were signalled */
+ clear_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state);
}
- clear_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state);
if (refcount_read(&clp->cl_count) > 1 && !signalled() &&
test_bit(NFS4CLNT_RUN_MANAGER, &clp->cl_state) &&
- !test_and_set_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state))
+ !test_and_set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state))
goto again;
nfs_put_client(clp);
Patches currently in stable-queue which might be from trond.myklebust@xxxxxxxxxxxxxxx are
queue-6.5/nfs-fix-o_direct-locking-issues.patch
queue-6.5/sunrpc-mark-the-cred-for-revalidation-if-the-server-.patch
queue-6.5/nfs-pnfs-report-einval-errors-from-connect-to-the-se.patch
queue-6.5/nfs-more-fixes-for-nfs_direct_write_reschedule_io.patch
queue-6.5/nfs-use-the-correct-commit-info-in-nfs_join_page_gro.patch
queue-6.5/nfs-more-o_direct-accounting-fixes-for-error-paths.patch
queue-6.5/nfsv4-fix-a-state-manager-thread-deadlock-regression.patch
queue-6.5/revert-sunrpc-dont-update-timeout-value-on-connection-reset.patch
queue-6.5/nfs-fix-error-handling-for-o_direct-write-scheduling.patch
