[PATCH 1/2] drm/i915: Prevent HW access during init from SDVO TV get_modes hook

Hogander, Jouni jouni.hogander at intel.com
Wed Feb 7 15:35:53 UTC 2024


On Wed, 2024-02-07 at 16:16 +0200, Imre Deak wrote:
> On Wed, Feb 07, 2024 at 03:45:15PM +0200, Hogander, Jouni wrote:
> > On Wed, 2024-02-07 at 15:26 +0200, Hogander, Jouni wrote:
> > > On Tue, 2024-02-06 at 17:39 +0200, Imre Deak wrote:
> > > > Prevent accessing the HW from the SDVO/TV get_modes connector
> > > > hook.
> > > > Returning 0 from the hook will make the caller -
> > > > drm_helper_probe_single_connector_modes() - keep the previously
> > > > detected
> > > > mode list of the connector.
> > > 
> > > I don't see where this is done? Not sure if looking at wrong
> > > place,
> > > but
> > > I see it tries using some override edid and in case that fails as
> > > well
> > > uses drm_add_modes_noedid?
> > > 
> > > > 
> > > > Signed-off-by: Imre Deak <imre.deak at intel.com>
> > > > ---
> > > >  drivers/gpu/drm/i915/display/intel_sdvo.c | 4 ++++
> > > >  1 file changed, 4 insertions(+)
> > > > 
> > > > diff --git a/drivers/gpu/drm/i915/display/intel_sdvo.c
> > > > b/drivers/gpu/drm/i915/display/intel_sdvo.c
> > > > index 2571ef5a1b211..ccea0efbd136f 100644
> > > > --- a/drivers/gpu/drm/i915/display/intel_sdvo.c
> > > > +++ b/drivers/gpu/drm/i915/display/intel_sdvo.c
> > > > @@ -2287,6 +2287,7 @@ static const struct drm_display_mode
> > > > sdvo_tv_modes[] = {
> > > >  static int intel_sdvo_get_tv_modes(struct drm_connector
> > > > *connector)
> > > 
> > > I see intel_sdvo_get_tv_modes is called from
> > > intel_sdvo_get_modes.
> > > Why there is no need to do the same in intel_sdvo_get_lvds_modes
> > > and
> > > intel_sdvo_get_ddc_modes as well?
> > 
> > It seems you are taking care of intel_svdo_get_ddc_modes in next
> > patch.
> > Intel_svdo_get_lvds_modes doesn't touch the hw -> nothing to do
> > there.
> 
> Yes, all the connectors that read out an EDID - vs. the TV connector
> which doesn't - is handled by the next patch. And yes, the connectors
> which don't access the HW in the get_modes hook doesn't need this
> check.
> Btw, I think all the connectors should work - eventually - in this
> way:
> access the HW only in the encoder detect hook and from the get_modes
> hook only return the detected mode list w/o accessing the HW. DP-SST
> and
> HDMI do already the correct thing wrt. this.
> 

Thank you for the clarification.

Reviewed-by: Jouni Högander <jouni.hogander at intel.com>

> > 
> > > 
> > > BR,
> > > 
> > > Jouni Högander
> > > 
> > > >  {
> > > >         struct intel_sdvo *intel_sdvo =
> > > > intel_attached_sdvo(to_intel_connector(connector));
> > > > +       struct drm_i915_private *i915 = to_i915(intel_sdvo-
> > > > > base.base.dev);
> > > >         struct intel_sdvo_connector *intel_sdvo_connector =
> > > >                 to_intel_sdvo_connector(connector);
> > > >         const struct drm_connector_state *conn_state =
> > > > connector-
> > > > > state;
> > > > @@ -2298,6 +2299,9 @@ static int intel_sdvo_get_tv_modes(struct
> > > > drm_connector *connector)
> > > >         DRM_DEBUG_KMS("[CONNECTOR:%d:%s]\n",
> > > >                       connector->base.id, connector->name);
> > > > 
> > > > +       if (!intel_display_driver_check_access(i915))
> > > > +               return 0;
> > > > +
> > > >         /*
> > > >          * Read the list of supported input resolutions for the
> > > > selected TV
> > > >          * format.
> > > 
> > 



More information about the Intel-gfx mailing list