Re: [PATCH] drm/xe/ufence: Flush xe ordered_wq in case of ufence timeout

[Nirmoy Das <nirmoy.das@intel.com>]

On 10/24/2024 12:39 AM, Matthew Brost wrote:
> On Wed, Oct 23, 2024 at 05:58:17PM +0200, Nirmoy Das wrote:
>> Flush xe ordered_wq in case of ufence timeout which is observed
>> on LNL and that points recent scheduling issue with E-cores.
>>
>> This is similar to the recent fix:
>> commit e51527233804 ("drm/xe/guc/ct: Flush g2h worker in case of g2h
>> response timeout") and should be removed once there is E core
>> scheduling fix.
>>
>> Cc: Badal Nilawar <badal.nilawar@intel.com>
>> Cc: Matthew Auld <matthew.auld@intel.com>
>> Cc: John Harrison <John.C.Harrison@Intel.com>
>> Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
>> Cc: Lucas De Marchi <lucas.demarchi@intel.com>
> Seems like a reasonable W/A for now which I think we want in 6.12 too.
> I'd follow up with the maintainer who is doing our 6.12 PR to make sure
> this gets in.


Make sense, I will reach out over #xe-chat. Adding a fixes tag that should be useful too.

Fixes: 9c57bc08652a ("drm/xe/lnl: Drop force_probe requirement")


>
> With that:
> Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Thanks!


Nirmoy

>
>> Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2754
>> Suggested-by: Matthew Brost <matthew.brost@intel.com>
>> Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
>> ---
>>  drivers/gpu/drm/xe/xe_wait_user_fence.c | 11 +++++++++++
>>  1 file changed, 11 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/xe/xe_wait_user_fence.c b/drivers/gpu/drm/xe/xe_wait_user_fence.c
>> index f5deb81eba01..7f034871345b 100644
>> --- a/drivers/gpu/drm/xe/xe_wait_user_fence.c
>> +++ b/drivers/gpu/drm/xe/xe_wait_user_fence.c
>> @@ -155,6 +155,17 @@ int xe_wait_user_fence_ioctl(struct drm_device *dev, void *data,
>>  		}
>>  
>>  		if (!timeout) {
>> +			/*
>> +			 * This is analogous to e51527233804 ("drm/xe/guc/ct: Flush g2h worker
>> +			 * in case of g2h response timeout")
>> +			 *
>> +			 * TODO: Drop this change once workqueue scheduling delay issue is
>> +			 * fixed on LNL Hybrid CPU.
>> +			 */
>> +			__flush_workqueue(xe->ordered_wq);
>> +			err = do_compare(addr, args->value, args->mask, args->op);
>> +			if (err <= 0)
>> +				break;
>>  			err = -ETIME;
>>  			break;
>>  		}
>> -- 
>> 2.46.0
>>