The check before alone is not enough for a case where there is another bug introduced so that

context->stream_count is not in sync with actual number of streams across entire resource_context.

At least assert indeed should be there.

Andrey

On 03/12/2018 07:06 PM, Li, Roman wrote:

There is a check just before for-loop that should ensure pipe_ctx is not null:

/* Only supports single display */ if (context->stream_count != 1) return false;

To remove the subject warning - we can rather add an assert: assert(pipe_ctx);

Thanks, Roman

-----Original Message----- From: Grodzovsky, Andrey Sent: Monday, March 12, 2018 6:23 PM To: David Binderman dcb314@hotmail.com; Deucher, Alexander Alexander.Deucher@amd.com; Koenig, Christian Christian.Koenig@amd.com; Zhou, David(ChunMing) David1.Zhou@amd.com; airlied@linux.ie; amd-gfx@lists.freedesktop.org; dri-devel@lists.freedesktop.org; linux-kernel@vger.kernel.org; Wentland, Harry Harry.Wentland@amd.com; Li, Roman Roman.Li@amd.com Subject: Re: linux-4.16-rc5/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c:1723]: (warning) Possible null pointer dereference: pipe_ctx

On 03/12/2018 06:22 AM, David Binderman wrote:

...

hello there,

Source code is

for (i = 0; i < dc->res_pool->pipe_count; i++) {         if (res_ctx->pipe_ctx[i].stream) {             pipe_ctx = &res_ctx->pipe_ctx[i];             *pipe_idx = i;             break;         }     }

Indeed something like

if (!pipe_ctx)               return false;

seems the right thing here.

Harry, Roman , your opinion ?

Andrey

...

/* Only supports eDP */     if (pipe_ctx->stream->sink->link->connector_signal != SIGNAL_TYPE_EDP)         return false;

Suggest add some code to deal with the case that the for loop doesn't find what it is looking for and so pipe_ctx is NULL.

Regards

David Binderman

---

amd-gfx mailing list amd-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/amd-gfx