linux-kernel - Re: [Intel-gfx] [PATCH v2] drm/i915/gen9bc: Handle TGP PCH during suspend/resume

Open Source and information security mailing list archives

Date:   Tue, 16 Feb 2021 21:36:01 -0500
From:   Lyude Paul <lyude@...hat.com>
To:     imre.deak@...el.com
Cc:     intel-gfx@...ts.freedesktop.org, David Airlie <airlied@...ux.ie>,
        open list <linux-kernel@...r.kernel.org>,
        "open list:DRM DRIVERS" <dri-devel@...ts.freedesktop.org>
Subject: Re: [Intel-gfx] [PATCH v2] drm/i915/gen9bc: Handle TGP PCH during
 suspend/resume

On Tue, 2021-02-16 at 20:08 +0200, Imre Deak wrote:
> Hi,
> 
> thanks for respinning this patchset, some comments below.
> 
> On Fri, Feb 12, 2021 at 01:50:53PM -0500, Lyude Paul wrote:
> > From: Tejas Upadhyay <tejaskumarx.surendrakumar.upadhyay@...el.com>
> > 
> > For Legacy S3 suspend/resume GEN9 BC needs to enable and
> > setup TGP PCH.
> > 
> > v2:
> > * Move Wa_14010685332 into it's own function - vsyrjala
> > * Add TODO comment about figuring out if we can move this workaround - imre
> > 
> > Cc: Matt Roper <matthew.d.roper@...el.com>
> > Signed-off-by: Tejas Upadhyay <tejaskumarx.surendrakumar.upadhyay@...el.com>
> > Signed-off-by: Lyude Paul <lyude@...hat.com>
> > ---
> >  drivers/gpu/drm/i915/i915_irq.c | 53 ++++++++++++++++++++++-----------
> >  1 file changed, 36 insertions(+), 17 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/i915_irq.c
> > b/drivers/gpu/drm/i915/i915_irq.c
> > index 98145a7f28a4..7d912aa950ee 100644
> > --- a/drivers/gpu/drm/i915/i915_irq.c
> > +++ b/drivers/gpu/drm/i915/i915_irq.c
> > @@ -3040,6 +3040,19 @@ static void valleyview_irq_reset(struct
> > drm_i915_private *dev_priv)
> >         spin_unlock_irq(&dev_priv->irq_lock);
> >  }
> >  
> > +static void cnp_irq_post_reset(struct drm_i915_private *dev_priv)
> 
> Maybe a better name is cnp_display_clock_wa.
> 
> > +{
> > +       struct intel_uncore *uncore = &dev_priv->uncore;
> > +
> > +       /*
> > +        * Wa_14010685332:cnp/cmp,tgp,adp
> 
> Bspec says this WA applies ICL onwards and it's not PCH specific, for
> instance I haven't found the GEN9/CNP/CMP WA entries for it. Please also
> add a 'clarify platforms where this applies' todo item.
> 
> > +        * TODO: Figure out if this workaround can be applied in the s0ix
> > suspend/resume handlers as
> > +        * on earlier platforms and whether the workaround is also needed
> > for runtime suspend/resume
> > +        */
> > +       intel_uncore_rmw(uncore, SOUTH_CHICKEN1, SBCLK_RUN_REFCLK_DIS,
> > SBCLK_RUN_REFCLK_DIS);
> > +       intel_uncore_rmw(uncore, SOUTH_CHICKEN1, SBCLK_RUN_REFCLK_DIS, 0);
> > +}
> > +
> >  static void gen8_irq_reset(struct drm_i915_private *dev_priv)
> >  {
> >         struct intel_uncore *uncore = &dev_priv->uncore;
> > @@ -3061,8 +3074,14 @@ static void gen8_irq_reset(struct drm_i915_private
> > *dev_priv)
> >         GEN3_IRQ_RESET(uncore, GEN8_DE_MISC_);
> >         GEN3_IRQ_RESET(uncore, GEN8_PCU_);
> >  
> > -       if (HAS_PCH_SPLIT(dev_priv))
> > +       if (INTEL_PCH_TYPE(dev_priv) >= PCH_ICP)
> 
> It was mentioned already earlier, why is this check necessary and can't we
> just call ibx_irq_reset() for all PCHs?
> 
> > +               GEN3_IRQ_RESET(uncore, SDE);
> > +       else if (HAS_PCH_SPLIT(dev_priv))
> >                 ibx_irq_reset(dev_priv);
> > +
> > +       if (INTEL_PCH_TYPE(dev_priv) == PCH_CNP ||
> > +           (INTEL_PCH_TYPE(dev_priv) >= PCH_TGP && INTEL_PCH_TYPE(dev_priv)
> > < PCH_DG1))
> 
> The check could be also moved to the helper.
> 
> > +               cnp_irq_post_reset(dev_priv);
> >  }
> >  
> >  static void gen11_display_irq_reset(struct drm_i915_private *dev_priv)
> > @@ -3104,15 +3123,9 @@ static void gen11_display_irq_reset(struct
> > drm_i915_private *dev_priv)
> >         if (INTEL_PCH_TYPE(dev_priv) >= PCH_ICP)
> >                 GEN3_IRQ_RESET(uncore, SDE);
> >  
> > -       /* Wa_14010685332:cnp/cmp,tgp,adp */
> >         if (INTEL_PCH_TYPE(dev_priv) == PCH_CNP ||
> > -           (INTEL_PCH_TYPE(dev_priv) >= PCH_TGP &&
> > -            INTEL_PCH_TYPE(dev_priv) < PCH_DG1)) {
> > -               intel_uncore_rmw(uncore, SOUTH_CHICKEN1,
> > -                                SBCLK_RUN_REFCLK_DIS,
> > SBCLK_RUN_REFCLK_DIS);
> > -               intel_uncore_rmw(uncore, SOUTH_CHICKEN1,
> > -                                SBCLK_RUN_REFCLK_DIS, 0);
> > -       }
> > +           (INTEL_PCH_TYPE(dev_priv) >= PCH_TGP && INTEL_PCH_TYPE(dev_priv)
> > < PCH_DG1))
> > +               cnp_irq_post_reset(dev_priv);
> >  }
> >  
> >  static void gen11_irq_reset(struct drm_i915_private *dev_priv)
> > @@ -3474,6 +3487,9 @@ static void spt_hpd_irq_setup(struct drm_i915_private
> > *dev_priv)
> >         ibx_display_interrupt_update(dev_priv, hotplug_irqs, enabled_irqs);
> >  
> >         spt_hpd_detection_setup(dev_priv);
> > +
> > +       if (INTEL_PCH_TYPE(dev_priv) >= PCH_ICP)
> > +               icp_hpd_irq_setup(dev_priv);
> 
> This doesn't look correct, icp_hpd_irq_setup() redoes the interrupt
> setup done already earlier in this function and
> spt_hpd_detection_setup() is probably also not correct on ICP+. Looks
> like for ICP+ we need to call icp_hpd_irq_setup() instead of
> spt_hpd_irq_setup(), but haven't checked in detail.

Could you please check :)? I don't work at Intel so you have far more access to
this information then I do. 

FWIW the code -looks- mostly equivalent to me (SHOTPLUG_CTL_DDI and
SHOTPLUG_CTL_TC seem to be equivalent registers to what's set in
spt_hpd_irq_setup()), but the icelake point codepath enables an additional port,
and changes an additional register called SHPD_FILTER_CNT. I'll update it to use
this in the next patch, but please definitely confirm this. I would very much
like to avoid potentially breaking unrelated ICP systems with this.

> 
> >  }
> >  
> >  static u32 ilk_hotplug_enables(struct drm_i915_private *i915,
> > @@ -3764,9 +3780,19 @@ static void gen8_de_irq_postinstall(struct
> > drm_i915_private *dev_priv)
> >         }
> >  }
> >  
> > +static void icp_irq_postinstall(struct drm_i915_private *dev_priv)
> > +{
> > +       struct intel_uncore *uncore = &dev_priv->uncore;
> > +       u32 mask = SDE_GMBUS_ICP;
> > +
> > +       GEN3_IRQ_INIT(uncore, SDE, ~mask, 0xffffffff);
> > +}
> > +
> >  static void gen8_irq_postinstall(struct drm_i915_private *dev_priv)
> >  {
> > -       if (HAS_PCH_SPLIT(dev_priv))
> > +       if (INTEL_PCH_TYPE(dev_priv) >= PCH_ICP)
> > +               icp_irq_postinstall(dev_priv);
> > +       else if (HAS_PCH_SPLIT(dev_priv))
> >                 ibx_irq_postinstall(dev_priv);
> >  
> >         gen8_gt_irq_postinstall(&dev_priv->gt);
> > @@ -3775,13 +3801,6 @@ static void gen8_irq_postinstall(struct
> > drm_i915_private *dev_priv)
> >         gen8_master_intr_enable(dev_priv->uncore.regs);
> >  }
> >  
> > -static void icp_irq_postinstall(struct drm_i915_private *dev_priv)
> > -{
> > -       struct intel_uncore *uncore = &dev_priv->uncore;
> > -       u32 mask = SDE_GMBUS_ICP;
> > -
> > -       GEN3_IRQ_INIT(uncore, SDE, ~mask, 0xffffffff);
> > -}
> >  
> >  static void gen11_irq_postinstall(struct drm_i915_private *dev_priv)
> >  {
> > -- 
> > 2.29.2
> > 
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx@...ts.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
> 

-- 
Sincerely,
   Lyude Paul (she/her)
   Software Engineer at Red Hat

Note: I deal with a lot of emails and have a lot of bugs on my plate. If you've
asked me a question, are waiting for a review/merge on a patch, etc. and I
haven't responded in a while, please feel free to send me another email to check
on my status. I don't bite!

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives