[Intel-gfx] [PATCH 2/3] drm/i915: Fix some invalid requests cancellations

[Tvrtko Ursulin]

On 29/01/16 16:49, Chris Wilson wrote:
> As we add the VMA to the request early, it may be cancelled during
> execbuf reservation. This will leave the context object pointing to a

I don't get it, request is created after the reservation.

> dangling request; i915_wait_request() simply skips the wait and so we
> may unbind the object whilst it is still active.
>
> However, if at any point we make a change to the hardware (and equally
> importantly our bookkeeping in the driver), we cannot cancel the request
> as what has already been written must be submitted. Submitting a partial
> request is far easier than trying to unwind the incomplete change.
>
> Unfortunately this patch undoes the excess breadcrumb usage that olr
> prevented, e.g. if we interrupt batchbuffer submission then we submit
> the requests along with the memory writes and interrupt (even though we
> do no real work). Disassociating requests from breadcrumbs (and
> semaphores) is a topic for a past/future series, but now much more
> important.
 >
> v2: Rebase
>
> Note that igt/gem_concurrent_blit tiggers both misrendering and a GPF
> that is fixed by this patch.
>
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93907
> Testcase: igt/gem_concurrent_blit
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Daniel Vetter <daniel.vetter at ffwll.ch>
> Cc: stable at vger.kernel.org
> ---
>   drivers/gpu/drm/i915/i915_drv.h            |  1 -
>   drivers/gpu/drm/i915/i915_gem.c            |  7 ++-----
>   drivers/gpu/drm/i915/i915_gem_context.c    | 21 +++++++++------------
>   drivers/gpu/drm/i915/i915_gem_execbuffer.c | 16 +++++-----------
>   drivers/gpu/drm/i915/intel_display.c       |  2 +-
>   drivers/gpu/drm/i915/intel_lrc.c           |  1 -
>   6 files changed, 17 insertions(+), 31 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> index a2d2f08b7515..f828a7ffed37 100644
> --- a/drivers/gpu/drm/i915/i915_drv.h
> +++ b/drivers/gpu/drm/i915/i915_drv.h
> @@ -2823,7 +2823,6 @@ int i915_gem_sw_finish_ioctl(struct drm_device *dev, void *data,
>   			     struct drm_file *file_priv);
>   void i915_gem_execbuffer_move_to_active(struct list_head *vmas,
>   					struct drm_i915_gem_request *req);
> -void i915_gem_execbuffer_retire_commands(struct i915_execbuffer_params *params);
>   int i915_gem_ringbuffer_submission(struct i915_execbuffer_params *params,
>   				   struct drm_i915_gem_execbuffer2 *args,
>   				   struct list_head *vmas);
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index e9b19bca1383..f764f33580fc 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -3407,12 +3407,9 @@ int i915_gpu_idle(struct drm_device *dev)
>   				return PTR_ERR(req);
>
>   			ret = i915_switch_context(req);
> -			if (ret) {
> -				i915_gem_request_cancel(req);
> -				return ret;
> -			}
> -
>   			i915_add_request_no_flush(req);
> +			if (ret)
> +				return ret;
>   		}
>
>   		ret = intel_ring_idle(ring);
> diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
> index 83a097c94911..5da7adc3f7b2 100644
> --- a/drivers/gpu/drm/i915/i915_gem_context.c
> +++ b/drivers/gpu/drm/i915/i915_gem_context.c
> @@ -652,7 +652,6 @@ static int do_switch(struct drm_i915_gem_request *req)
>   	struct drm_i915_private *dev_priv = ring->dev->dev_private;
>   	struct intel_context *from = ring->last_context;
>   	u32 hw_flags = 0;
> -	bool uninitialized = false;
>   	int ret, i;
>
>   	if (from != NULL && ring == &dev_priv->ring[RCS]) {
> @@ -759,6 +758,15 @@ static int do_switch(struct drm_i915_gem_request *req)
>   			to->remap_slice &= ~(1<<i);
>   	}
>
> +	if (!to->legacy_hw_ctx.initialized) {
> +		if (ring->init_context) {
> +			ret = ring->init_context(req);
> +			if (ret)
> +				goto unpin_out;
> +		}
> +		to->legacy_hw_ctx.initialized = true;
> +	}
> +
>   	/* The backing object for the context is done after switching to the
>   	 * *next* context. Therefore we cannot retire the previous context until
>   	 * the next context has already started running. In fact, the below code
> @@ -782,21 +790,10 @@ static int do_switch(struct drm_i915_gem_request *req)
>   		i915_gem_context_unreference(from);
>   	}
>
> -	uninitialized = !to->legacy_hw_ctx.initialized;
> -	to->legacy_hw_ctx.initialized = true;
> -
>   done:
>   	i915_gem_context_reference(to);
>   	ring->last_context = to;
>
> -	if (uninitialized) {
> -		if (ring->init_context) {
> -			ret = ring->init_context(req);
> -			if (ret)
> -				DRM_ERROR("ring init context: %d\n", ret);
> -		}
> -	}
> -
>   	return 0;
>
>   unpin_out:
> diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> index 8fd00d279447..61bb15507b30 100644
> --- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> +++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> @@ -1133,7 +1133,7 @@ i915_gem_execbuffer_move_to_active(struct list_head *vmas,
>   	}
>   }
>
> -void
> +static void
>   i915_gem_execbuffer_retire_commands(struct i915_execbuffer_params *params)
>   {
>   	/* Unconditionally force add_request to emit a full flush. */
> @@ -1318,7 +1318,6 @@ i915_gem_ringbuffer_submission(struct i915_execbuffer_params *params,
>   	trace_i915_gem_ring_dispatch(params->request, params->dispatch_flags);
>
>   	i915_gem_execbuffer_move_to_active(vmas, params->request);
> -	i915_gem_execbuffer_retire_commands(params);

Hm this whole block block from trace_... to 
i915_gem_execbuffer_retire_commands could be moved to do_execbuffer as 
and independent cleanup patch if under "if (ret == 0)".

So the concern is that something has been written to the ringbuffer but 
not all of it?

Why do we have to submit that, I am assuming whatever was written is not 
interesting to be executed unless the whole bb went in.

So could we just rewind the pointers? Store them at the beginning and 
rewind if something failed.

>
>   	return 0;
>   }
> @@ -1616,8 +1615,10 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
>   	}
>
>   	ret = i915_gem_request_add_to_client(req, file);
> -	if (ret)
> +	if (ret) {
> +		i915_gem_request_cancel(req);
>   		goto err_batch_unpin;
> +	}

Doesn't look like this can fail so a side cleanup only?

Regards,

Tvrtko