[Intel-gfx] [PATCH v8] drm/i915/dsc: Add Per connector debugfs node for DSC support/enable

[Manasi Navare]

On Wed, Dec 19, 2018 at 08:08:51AM +0000, Chris Wilson wrote:
> Quoting Manasi Navare (2018-12-06 00:54:07)
> > DSC can be supported per DP connector. This patch adds a per connector
> > debugfs node to expose DSC support capability by the kernel.
> > The same node can be used from userspace to force DSC enable.
> 
> So this has a nice deadlock that is killing icl sporadically...
> 
> > +static int i915_dsc_fec_support_show(struct seq_file *m, void *data)
> > +{
> > +       struct drm_connector *connector = m->private;
> > +       struct drm_device *dev = connector->dev;
> > +       struct drm_crtc *crtc;
> > +       struct intel_dp *intel_dp;
> > +       struct drm_modeset_acquire_ctx ctx;
> > +       struct intel_crtc_state *crtc_state = NULL;
> > +       int ret = 0;
> > +       bool try_again = false;
> > +
> > +       drm_modeset_acquire_init(&ctx, DRM_MODESET_ACQUIRE_INTERRUPTIBLE);
> > +
> > +       do {
> > +               ret = drm_modeset_lock(&dev->mode_config.connection_mutex,
> > +                                      &ctx);
> > +               if (ret) {
> > +                       ret = -EINTR;
> > +                       break;
> > +               }
> > +               crtc = connector->state->crtc;
> > +               if (connector->status != connector_status_connected || !crtc) {
> > +                       ret = -ENODEV;
> > +                       break;
> > +               }
> > +               ret = drm_modeset_lock(&crtc->mutex, &ctx);
> > +               if (ret == -EDEADLK) {
> > +                       ret = drm_modeset_backoff(&ctx);
> > +                       if (!ret) {
> > +                               try_again = true;
> 
> try_again is never cleared on the next loop, so after you hit the
> backoff, we never escape on success.

Thanks for catching this, teh CI results came green before this patch was merged so
never noticed this.
I will submit a patch to fix this.

Thanks again Chris and Tomi for catching this bug

Manasi

> 
> Reported-by: Tomi Sarvela <tomi.p.sarvela at intel.com>
> -Chris