Re: [PATCH v2] drm/i915: Sleep and retry a GPU reset if at first we don't succeed

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> writes:

> As we declare the GPU wedged if the reset fails, such a failure is quite
> terminal. Before taking that drastic action, let's sleep first and try
> active, in the hope that the hardware has quietened down and is then
> able to reset. After a few such attempts, it is fair to say that the HW
> is truly wedged.
>
> v2: Always print the failure message now, we precheck whether resets are
> disabled.
>
> References: https://bugs.freedesktop.org/show_bug.cgi?id=104007
> Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> Cc: Mika Kuoppala <mika.kuoppala@xxxxxxxxxxxxxxx>
> Cc: Joonas Lahtinen <joonas.lahtinen@xxxxxxxxxxxxxxx>
> ---
>  drivers/gpu/drm/i915/i915_drv.c | 20 +++++++++++++++-----
>  1 file changed, 15 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
> index e0f053f9c186..7faf20aff25a 100644
> --- a/drivers/gpu/drm/i915/i915_drv.c
> +++ b/drivers/gpu/drm/i915/i915_drv.c
> @@ -1877,7 +1877,9 @@ void i915_reset(struct drm_i915_private *i915, unsigned int flags)
>  {
>  	struct i915_gpu_error *error = &i915->gpu_error;
>  	int ret;
> +	int i;
>  
> +	might_sleep();
>  	lockdep_assert_held(&i915->drm.struct_mutex);
>  	GEM_BUG_ON(!test_bit(I915_RESET_BACKOFF, &error->flags));
>  
> @@ -1900,12 +1902,20 @@ void i915_reset(struct drm_i915_private *i915, unsigned int flags)
>  		goto error;
>  	}
>  
> -	ret = intel_gpu_reset(i915, ALL_ENGINES);
> +	if (!intel_has_gpu_reset(i915)) {
> +		DRM_DEBUG_DRIVER("GPU reset disabled\n");
> +		goto error;
> +	}
> +
> +	for (i = 0; i < 3; i++) {
> +		ret = intel_gpu_reset(i915, ALL_ENGINES);
> +		if (ret == 0)
> +			break;
> +
> +		msleep(100);

Seems reasonable to try few times and pause between defibrillate
attempts instead of throwing dirt on top of coffin right
off the bat.

Also I have been pondering that should we add a minicheck
to intel_gpu_reset to poke that the gpu is really there.
Like doing few nops in (render)ringbuffer and see if head
moves before declaring it as a reset success?

Not that we would not see it in init right after but just
to have more precise location of failure instead of
initing a dead gpu.

Reviewed-by: Mika Kuoppala <mika.kuoppala@xxxxxxxxxxxxxxx>

-Mika


> +	}
>  	if (ret) {
> -		if (ret != -ENODEV)
> -			DRM_ERROR("Failed to reset chip: %i\n", ret);
> -		else
> -			DRM_DEBUG_DRIVER("GPU reset disabled\n");
> +		dev_err(i915->drm.dev, "Failed to reset chip\n");
>  		goto error;
>  	}
>  
> -- 
> 2.15.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx




[Index of Archives]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]
  Powered by Linux