[Intel-gfx] [PATCH] drm/i915: Perform link quality check unconditionally during long pulse

[Ville Syrjälä]

On Wed, Feb 22, 2017 at 08:00:38PM -0800, Palmer Dabbelt wrote:
> On Thu, Feb 16, 2017 at 9:46 AM, Ville Syrjälä
> <ville.syrjala at linux.intel.com> wrote:
> > On Thu, Feb 16, 2017 at 09:24:09AM -0800, Manasi Navare wrote:
> >> On Thu, Feb 16, 2017 at 07:18:57PM +0200, Ville Syrjälä wrote:
> >> > On Thu, Feb 16, 2017 at 09:07:53AM -0800, Manasi Navare wrote:
> >> > > On Thu, Feb 16, 2017 at 05:30:07PM +0200, ville.syrjala at linux.intel.com wrote:
> >> > > > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> >> > > >
> >> > > > Apparently some DP sinks are a little nuts and cause HPD to drop
> >> > > > intermittently during modesets. This happens eg. on an ASUS PB287Q.
> >> > > > In oder to recover from this we can't really use the previous
> >> > > > connector status to determine if the link needs retraining, so let's
> >> > > > just ignore that piece of information and do the retrain
> >> > > > unconditionally. We do of course still check whether the link is
> >> > > > supposed to be running or not.
> >> > > >
> >> > > > Cc: stable at vger.kernel.org
> >> > > > Cc: Palmer Dabbelt <palmer at dabbelt.com>
> >> > > > Reported-by: Palmer Dabbelt <palmer at dabbelt.com>
> >> > > > References: https://lists.freedesktop.org/archives/intel-gfx/2017-February/119779.html
> >> > > > Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> >> > > > ---
> >> > > >  drivers/gpu/drm/i915/intel_dp.c | 15 +++++++++++----
> >> > > >  1 file changed, 11 insertions(+), 4 deletions(-)
> >> > > >
> >> > > > diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
> >> > > > index 024798a9c016..37a746f7fbc3 100644
> >> > > > --- a/drivers/gpu/drm/i915/intel_dp.c
> >> > > > +++ b/drivers/gpu/drm/i915/intel_dp.c
> >> > > > @@ -4648,11 +4648,18 @@ intel_dp_long_pulse(struct intel_connector *intel_connector)
> >> > > >                  */
> >> > > >                 status = connector_status_disconnected;
> >> > > >                 goto out;
> >> > > > -       } else if (connector->status == connector_status_connected) {
> >> > > > +       } else {
> >> > > >                 /*
> >> > > > -                * If display was connected already and is still connected
> >> > > > -                * check links status, there has been known issues of
> >> > > > -                * link loss triggerring long pulse!!!!
> >> > > > +                * If display is now connected check links status,
> >> > > > +                * there has been known issues of link loss triggerring
> >> > > > +                * long pulse.
> >> > > > +                *
> >> > > > +                * Some sinks (eg. ASUS PB287Q) seem to perform some
> >> > > > +                * weird HPD ping pong during modesets. So we can apparely
> >> > > > +                * end up with HPD going low during a modeset, and then
> >> > > > +                * going back up soon after. And once that happens we must
> >> > > > +                * retrain the link to get a picture. That's in case no
> >> > > > +                * userspace component reacted to intermittent HPD dip.
> >> > > >                  */
> >> > > >                 drm_modeset_lock(&dev->mode_config.connection_mutex, NULL);
> >> > > >                 intel_dp_check_link_status(intel_dp);
> >> > > > --
> >> > >
> >> > > So here we basically just ignore the connector status and retrain irrespectively.
> >> >
> >> > We ignore the _previous_ connector status.
> >> >
> >> > > But that means even if we have newer values now for max link rate/lane count from
> >> > > DPCD, during this retrain we are just using the stale value of intel_dp->link_rate
> >> > > and intel_dp->lane_count. I think intel_dp->link_rate and lane count values
> >> > > should be set to 0 on HPD pulse, they would be set only during a modeset.
> >> >
> >> > The DPCD has already been parsed by this time.
> >> >
> >> > --
> >> > Ville Syrjälä
> >> > Intel OTC
> >>
> >> Yes, we have parsed the DPCD but we dont write to intel_dp->link-rate and intel_dp->lane_count
> >> until we do a modeset (these get written during pre_enable) and these values get used
> >> during the retraining of the link. So at this point we will still use stale values.
> >
> > They're not stale. They're exactly what we used when we set up the
> > mode that's still being drive out. This is just retraining the link
> > at the same parameters.
> >
> >> Shouldnt we set the link status to BAD and send a hotplug event, like the solution we implemneted
> >> for retraining?
> >>
> >> Regards
> >> Manasi
> >>
> >> Regards
> >> Manasi
> >
> > --
> > Ville Syrjälä
> > Intel OTC
> 
> Sorry, I forgot about this.  A log is attached of my machine booting
> with the monitor attached, me detaching the monitor, and then plugging
> it back in.

I was actually able to reproduce this yesterday with one particular
kernel. It looked like it wasn't even trying to do the EDID read. Not
sure why. And I wasn't able reproduce it anymore with the latest
drm-tip. I should probably try to figure out what changed...

-- 
Ville Syrjälä
Intel OTC