We need to set TASK_INTERRUPTIBLE before calling kthread_should_stop().
Otherwise, kthread_stop() might see that the pccardd thread is still
in TASK_RUNNING state and fail to wake it up.

Additonally, we only need to set the state back to TASK_RUNNING if
kthread_should_stop() breaks the loop.

Cc: Matthew Wilcox <willy@infradead.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Fixes: d3046ba809ce ("pcmcia: fix a boot time warning in pcmcia cs code")
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
---
 drivers/pcmcia/cs.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/drivers/pcmcia/cs.c b/drivers/pcmcia/cs.c
index e211e2619680..f70197154a36 100644
--- a/drivers/pcmcia/cs.c
+++ b/drivers/pcmcia/cs.c
@@ -666,18 +666,16 @@ static int pccardd(void *__skt)
 		if (events || sysfs_events)
 			continue;
 
+		set_current_state(TASK_INTERRUPTIBLE);
 		if (kthread_should_stop())
 			break;
 
-		set_current_state(TASK_INTERRUPTIBLE);
-
 		schedule();
 
-		/* make sure we are running */
-		__set_current_state(TASK_RUNNING);
-
 		try_to_freeze();
 	}
+	/* make sure we are running before we exit */
+	__set_current_state(TASK_RUNNING);
 
 	/* shut down socket, if a device is still present */
 	if (skt->state & SOCKET_PRESENT) {
-- 
2.34.1

On Sun, Jan 09, 2022 at 09:14:15PM +0100, Dominik Brodowski wrote:
> We need to set TASK_INTERRUPTIBLE before calling kthread_should_stop().
> Otherwise, kthread_stop() might see that the pccardd thread is still
> in TASK_RUNNING state and fail to wake it up.
> 
> Additonally, we only need to set the state back to TASK_RUNNING if
> kthread_should_stop() breaks the loop.
> 
> Cc: Matthew Wilcox <willy@infradead.org>
> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
> Fixes: d3046ba809ce ("pcmcia: fix a boot time warning in pcmcia cs code")
> Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>

Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>