[Intel-gfx] [PATCH] drm/i915: Take care of ring->last_context on ctx destroy

[Daniel Vetter]

On Fri, Dec 04, 2015 at 05:52:43PM +0200, Mika Kuoppala wrote:
> If the context being destroyed have been last in the ring,
> the ring->last_context will be left dangling.
> 
> Later, the unpinning will happen for last_context, and as it
> was already freed, we corrupt memory by doing so.
> 
> This regression introduced in
> commit 6d65ba943a2d1e4292a07ca7ddb6c5138b9efa5d
> Author: Nick Hoath <nicholas.hoath at intel.com>
> Date:   Tue Dec 1 14:48:57 2015 +0000
> 
>     drm/i915: Extend LRC pinning to cover GPU context writeback
> 
> Fix this by clearing the ring->last_context if it is the
> context being destroyed.
> 
> Cc: Nick Hoath <nicholas.hoath at intel.com>
> Cc: Daniel Vetter <daniel.vetter at ffwll.ch>
> Cc: David Gordon <david.s.gordon at intel.com>
> Cc: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Alex Dai <yu.dai at intel.com>
> Signed-off-by: Mika Kuoppala <mika.kuoppala at intel.com>
> ---
>  drivers/gpu/drm/i915/intel_lrc.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
> index c3504a0..5c26fde 100644
> --- a/drivers/gpu/drm/i915/intel_lrc.c
> +++ b/drivers/gpu/drm/i915/intel_lrc.c
> @@ -2432,6 +2432,9 @@ intel_lr_context_clean_ring(struct intel_context *ctx,
>  		}
>  	}
>  
> +	if (ring->last_context == ctx)
> +		ring->last_context = NULL;

This should be plainly impossible. ring->last_context better have a
reference on it, and if that's not the case we need to fix that. Nick, can
you please take a look into this asap? It's causing a BAT regression so
meanwhile I'll revert your patch.

Thanks, Daniel

> +
>  	WARN_ON(ctx->engine[ring->id].pin_count);
>  	intel_ringbuffer_free(ringbuf);
>  	drm_gem_object_unreference(&ctx_obj->base);
> -- 
> 2.5.0
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch