Re: [PATCH] drm: Check for connector->state NULL in drm_atomic_add_affected_connectors

From: Daniel Vetter <daniel@ffwll.ch>
To: Douglas Anderson <dianders@chromium.org>
Cc: David Airlie <airlied@linux.ie>,
	linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org
Subject: Re: [PATCH] drm: Check for connector->state NULL in drm_atomic_add_affected_connectors
Date: Sun, 6 Mar 2016 17:29:51 +0100	[thread overview]
Message-ID: <20160306162951.GA14170@phenom.ffwll.local> (raw)
In-Reply-To: <1457135033-11791-1-git-send-email-dianders@chromium.org>

On Fri, Mar 04, 2016 at 03:43:53PM -0800, Douglas Anderson wrote:
> On a system I'm doing development on I found a crash.  The crawl looked
> like:
> 
>   PC is at drm_atomic_add_affected_connectors+0x98/0xe8
>   ...
>   drm_atomic_add_affected_connectors from __drm_atomic_helper_set_config+0x218/0x344
>   __drm_atomic_helper_set_config from restore_fbdev_mode+0x108/0x250
>   restore_fbdev_mode from drm_fb_helper_restore_fbdev_mode_unlocked+0x3c/0x80
>   drm_fb_helper_restore_fbdev_mode_unlocked from rockchip_drm_lastclose+0x1c/0x20
>   rockchip_drm_lastclose from drm_lastclose+0x4c/0x104
>   drm_lastclose from drm_release+0x424/0x47c
>   drm_release from __fput+0xf8/0x1d4
>   __fput from ____fput+0x18/0x1c
>   ____fput from task_work_run+0xa8/0xbc
>   task_work_run from do_exit+0x448/0x91c
>   do_exit from do_group_exit+0x5c/0xcc
>   do_group_exit from get_signal+0x4dc/0x57c
>   get_signal from do_signal+0x9c/0x3b4
>   do_signal from do_work_pending+0x60/0xb8
>   do_work_pending from slow_work_pending+0xc/0x20
> 
> I found that I could fix the crash by checking connector->state against
> NULL.  This isn't code I'm familiar with and I didn't dig too deep, so
> I'd appreciate any opinions about whether this is a sane thing to do.
> Note that I don't actually have graphics up on my system at the moment,
> so perhaps this is all just a symptom of the strange state I'm in.
> 
> Signed-off-by: Douglas Anderson <dianders@chromium.org>

This is a driver bug - under atomic the assumption is that there is
_always_ a current software state. Most driver set up the initial
"everything off" state in the ->reset functions.
-Daniel

> ---
>  drivers/gpu/drm/drm_atomic.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/drm_atomic.c b/drivers/gpu/drm/drm_atomic.c
> index 8fb469c4e4b8..3377d7ddc6d1 100644
> --- a/drivers/gpu/drm/drm_atomic.c
> +++ b/drivers/gpu/drm/drm_atomic.c
> @@ -1147,7 +1147,7 @@ drm_atomic_add_affected_connectors(struct drm_atomic_state *state,
>  	 * current configuration.
>  	 */
>  	drm_for_each_connector(connector, state->dev) {
> -		if (connector->state->crtc != crtc)
> +		if (!connector->state || connector->state->crtc != crtc)
>  			continue;
>  
>  		conn_state = drm_atomic_get_connector_state(state, connector);
> -- 
> 2.7.0.rc3.207.g0ac5344
> 
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

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