[PATCH blktests v2] loop/010, common/rc: drain udev events after test

[PATCH blktests v2] loop/010, common/rc: drain udev events after test

[Shin'ichiro Kawasaki]

The test case repeats creating and deleting a loop device. This
generates many udev events and makes following test cases fail. To avoid
the unexpected test case failures, drain the udev events. For that
purpose, introduce the helper function _drain_udev_events(). When
systemd-udevd service is running, restart it to discard the events
quickly. When systemd-udevd service is not available, call
"udevadm settle", which takes longer time to drain the events.

Link: https://github.com/linux-blktests/blktests/issues/181
Reported-by: Yi Zhang <yi.zhang@redhat.com>
Suggested-by: Bart Van Assche <bvanassche@acm.org>
Suggested-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
---
Changes from v1:
* Added "udevadm settle" in case systemd-udevd.service is not available
* Introduced _drain_udev_events()

 common/rc      | 9 +++++++++
 tests/loop/010 | 5 +++++
 2 files changed, 14 insertions(+)

diff --git a/common/rc b/common/rc
index 72441ab..dfc389f 100644
--- a/common/rc
+++ b/common/rc
@@ -544,6 +544,15 @@ _systemctl_stop() {
 	done
 }
 
+_drain_udev_events() {
+	if command -v systemctl &>/dev/null &&
+			systemctl is-active --quiet systemd-udevd; then
+		systemctl restart systemd-udevd.service
+	else
+		udevadm settle --timeout=900
+	fi
+}
+
 # Run the given command as NORMAL_USER
 _run_user() {
 	su "$NORMAL_USER" -c "$1"
diff --git a/tests/loop/010 b/tests/loop/010
index 309fd8a..b1a4926 100755
--- a/tests/loop/010
+++ b/tests/loop/010
@@ -78,5 +78,10 @@ test() {
 	if _dmesg_since_test_start | grep --quiet "$grep_str"; then
 		echo "Fail"
 	fi
+
+	# The repeated loop device creations and deletions generated so many
+	# udev events. Drain the events to not influence following test cases.
+	_drain_udev_events
+
 	echo "Test complete"
 }
-- 
2.50.1

Re: [PATCH blktests v2] loop/010, common/rc: drain udev events after test

[Daniel Wagner]

On Tue, Jul 15, 2025 at 01:32:02PM +0900, Shin'ichiro Kawasaki wrote:
> The test case repeats creating and deleting a loop device. This
> generates many udev events and makes following test cases fail. To avoid
> the unexpected test case failures, drain the udev events. For that
> purpose, introduce the helper function _drain_udev_events(). When
> systemd-udevd service is running, restart it to discard the events
> quickly. When systemd-udevd service is not available, call
> "udevadm settle", which takes longer time to drain the events.
> 
> Link: https://github.com/linux-blktests/blktests/issues/181
> Reported-by: Yi Zhang <yi.zhang@redhat.com>
> Suggested-by: Bart Van Assche <bvanassche@acm.org>
> Suggested-by: Daniel Wagner <dwagner@suse.de>
> Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>

Looks good to me. Thanks!

Reviewed-by: Daniel Wagner <dwagner@suse.de>

Re: [PATCH blktests v2] loop/010, common/rc: drain udev events after test

[Bart Van Assche]

On 7/14/25 9:32 PM, Shin'ichiro Kawasaki wrote:
> The test case repeats creating and deleting a loop device. This
> generates many udev events and makes following test cases fail. To avoid
> the unexpected test case failures, drain the udev events. For that
> purpose, introduce the helper function _drain_udev_events(). When
> systemd-udevd service is running, restart it to discard the events
> quickly. When systemd-udevd service is not available, call
> "udevadm settle", which takes longer time to drain the events.
> 
> Link: https://github.com/linux-blktests/blktests/issues/181
> Reported-by: Yi Zhang <yi.zhang@redhat.com>
> Suggested-by: Bart Van Assche <bvanassche@acm.org>
> Suggested-by: Daniel Wagner <dwagner@suse.de>
> Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
> ---
> Changes from v1:
> * Added "udevadm settle" in case systemd-udevd.service is not available
> * Introduced _drain_udev_events()
> 
>   common/rc      | 9 +++++++++
>   tests/loop/010 | 5 +++++
>   2 files changed, 14 insertions(+)
> 
> diff --git a/common/rc b/common/rc
> index 72441ab..dfc389f 100644
> --- a/common/rc
> +++ b/common/rc
> @@ -544,6 +544,15 @@ _systemctl_stop() {
>   	done
>   }
>   
> +_drain_udev_events() {
> +	if command -v systemctl &>/dev/null &&
> +			systemctl is-active --quiet systemd-udevd; then
> +		systemctl restart systemd-udevd.service
> +	else
> +		udevadm settle --timeout=900
> +	fi
> +}
> +
>   # Run the given command as NORMAL_USER
>   _run_user() {
>   	su "$NORMAL_USER" -c "$1"
> diff --git a/tests/loop/010 b/tests/loop/010
> index 309fd8a..b1a4926 100755
> --- a/tests/loop/010
> +++ b/tests/loop/010
> @@ -78,5 +78,10 @@ test() {
>   	if _dmesg_since_test_start | grep --quiet "$grep_str"; then
>   		echo "Fail"
>   	fi
> +
> +	# The repeated loop device creations and deletions generated so many
> +	# udev events. Drain the events to not influence following test cases.
> +	_drain_udev_events
> +

How about changing the above comment into the following to make it more 
clear? "This test generates udev events faster than the rate at which
udevd can process events. Drain udev events to prevent that future test
cases fail." Anyway:

Reviewed-by: Bart Van Assche <bvanassche@acm.org>

Re: [PATCH blktests v2] loop/010, common/rc: drain udev events after test

[Shinichiro Kawasaki]

On Jul 15, 2025 / 13:32, Shin'ichiro Kawasaki wrote:
> The test case repeats creating and deleting a loop device. This
> generates many udev events and makes following test cases fail. To avoid
> the unexpected test case failures, drain the udev events. For that
> purpose, introduce the helper function _drain_udev_events(). When
> systemd-udevd service is running, restart it to discard the events
> quickly. When systemd-udevd service is not available, call
> "udevadm settle", which takes longer time to drain the events.
> 
> Link: https://github.com/linux-blktests/blktests/issues/181
> Reported-by: Yi Zhang <yi.zhang@redhat.com>
> Suggested-by: Bart Van Assche <bvanassche@acm.org>
> Suggested-by: Daniel Wagner <dwagner@suse.de>
> Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>

FYI, I have applied this patch with the comment improvement Bart suggested.
Thanks!