[Intel-wired-lan] [PATCH v1] e1000e: Replace schedule_work with delayed workqueue for watchdog.
Przemek Kitszel
przemyslaw.kitszel at intel.com
Thu May 8 09:34:08 UTC 2025
On 5/8/25 08:14, Jagadeesh Yalapalli wrote:
> From: Jagadeesh <jagadeesh.yalapalli at einfochips.com>
>
> Replace direct schedule_work() usage with queue_delayed_work() to allow
> better timing control for the watchdog task. This resolves potential
> race conditions during interface reset operations.
>
> - Added watchdog_wq workqueue_struct and watchdog_dq delayed_work
> - Updated e1000_watchdog() to use queue_delayed_work()
> - Removed obsolete TODO comment about delayed workqueue
>
> Tested in Qemu :
> / # for i in {1..1000}; do
> > echo 1 > /sys/class/net/eth0/device/reset
> > sleep 0.1
> > done
> [ 726.357499] e1000e 0000:00:02.0: resetting
> [ 726.390737] e1000e 0000:00:02.0: reset done
>
> Signed-off-by: Jagadeesh <jagadeesh.yalapalli at einfochips.com>
Thank you for your contribution.
Please use at least two parts for the author and signoff
> ---
> drivers/net/ethernet/intel/e1000e/e1000.h | 2 ++
> drivers/net/ethernet/intel/e1000e/netdev.c | 3 +--
> 2 files changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/ethernet/intel/e1000e/e1000.h b/drivers/net/ethernet/intel/e1000e/e1000.h
> index ba9c19e6994c..1e7b365c4f31 100644
> --- a/drivers/net/ethernet/intel/e1000e/e1000.h
> +++ b/drivers/net/ethernet/intel/e1000e/e1000.h
> @@ -194,6 +194,8 @@ struct e1000_adapter {
> struct timer_list blink_timer;
>
> struct work_struct reset_task;
> + struct workqueue_struct *watchdog_wq;
> + struct delayed_work watchdog_dq;
> struct work_struct watchdog_task;
>
> const struct e1000_info *ei;
> diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c
> index 8ebcb6a7d608..87a915d09f4e 100644
> --- a/drivers/net/ethernet/intel/e1000e/netdev.c
> +++ b/drivers/net/ethernet/intel/e1000e/netdev.c
> @@ -5178,9 +5178,8 @@ static void e1000_watchdog(struct timer_list *t)
> struct e1000_adapter *adapter = from_timer(adapter, t, watchdog_timer);
>
> /* Do the rest outside of interrupt context */
> - schedule_work(&adapter->watchdog_task);
> + queue_delayed_work(adapter->watchdog_wq, &adapter->watchdog_dq, 0);
there is no much point in using queue_delayed_work() with the delay of 0
please also note that the last commit touching this function was
essentially just reverting the use of delayed work, see commit
d5ad7a6a7f3c ("e1000e: Revert "e1000e: Make watchdog use delayed work"")
would be best to understand what was wrong with the reverted commit
>
> - /* TODO: make this use queue_delayed_work() */
> }
>
> static void e1000_watchdog_task(struct work_struct *work)
More information about the Intel-wired-lan
mailing list