Chris Wilson <[email protected]> writes:

> We can not require that the system process a tasklet in reasonable time
> (thanks be to ksoftirqd), but we can insist that having waited
> sufficiently for the error interrupt to have been raised and having
> kicked the tasklet, the reset has begun and the request will be marked
> as in error (if not already completed).
>
> Signed-off-by: Chris Wilson <[email protected]>
> ---
>  drivers/gpu/drm/i915/gt/selftest_lrc.c | 15 +++++++++++----
>  1 file changed, 11 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/gt/selftest_lrc.c 
> b/drivers/gpu/drm/i915/gt/selftest_lrc.c
> index 8b7383f6d9b3..ccd4cd2c202d 100644
> --- a/drivers/gpu/drm/i915/gt/selftest_lrc.c
> +++ b/drivers/gpu/drm/i915/gt/selftest_lrc.c
> @@ -503,14 +503,21 @@ static int live_error_interrupt(void *arg)
>                       }
>  
>                       for (i = 0; i < ARRAY_SIZE(client); i++) {
> -                             if (i915_request_wait(client[i], 0, HZ / 5) < 
> 0) {
> -                                     pr_err("%s: %s request still 
> executing!\n",
> -                                            engine->name,
> -                                            error_repr(p->error[i]));
> +                             if (i915_request_wait(client[i], 0, HZ / 5) < 0)
> +                                     pr_debug("%s: %s request incomplete!\n",
> +                                              engine->name,
> +                                              error_repr(p->error[i]));
> +
> +                             if (!i915_request_started(client[i])) {
> +                                     pr_debug("%s: %s request not stated!\n",
> +                                              engine->name,
> +                                              error_repr(p->error[i]));
>                                       err = -ETIME;
>                                       goto out;
>                               }
>  
> +                             /* Kick the tasklet to process the error */
> +                             intel_engine_flush_submission(engine);

This pattern of usage in selftests makes me want to add mb(); to
intel_engine_flush_submission(), as it does not seem to do it
implicitly.

We want to flush submission and observe the effects, thus
it seems like a good place.

Not in a scope of this patch tho,
Reviewed-by: Mika Kuoppala <[email protected]>

-Mika

>                               if (client[i]->fence.error != p->error[i]) {
>                                       pr_err("%s: %s request completed with 
> wrong error code: %d\n",
>                                              engine->name,
> -- 
> 2.25.0
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Reply via email to