public inbox for intel-gfx@lists.freedesktop.org
 help / color / mirror / Atom feed
From: Imre Deak <imre.deak@intel.com>
To: "Souza, Jose" <jose.souza@intel.com>
Cc: "intel-gfx@lists.freedesktop.org" <intel-gfx@lists.freedesktop.org>
Subject: Re: [Intel-gfx] [PATCH] drm/i915: Apply CMTG clock disabling WA while DPLL0 is enabled
Date: Tue, 3 Aug 2021 00:07:54 +0300	[thread overview]
Message-ID: <20210802210754.GA2115791@ideak-desk.fi.intel.com> (raw)
In-Reply-To: <70c089841d159b322f6c6749fad8e416766676dc.camel@intel.com>

On Mon, Aug 02, 2021 at 11:52:41PM +0300, Souza, Jose wrote:
> On Mon, 2021-08-02 at 22:01 +0300, Imre Deak wrote:
> > CI test results/further experiments show that the workaround added in
> > 
> > commit 573d7ce4f69a ("drm/i915/adlp: Add workaround to disable CMTG clock gating")
> > 
> > can be applied only while DPLL0 is enabled. If it's disabled the
> > TRANS_CMTG_CHICKEN register is not accessible. Accordingly move the WA
> > to DPLL0 HW state sanitization and enabling.
> > 
> > This fixes an issue where the WA won't get applied (and a WARN is thrown
> > due to an unexpected value in TRANS_CMTG_CHICKEN) if the driver is
> > loaded without DPLL0 being enabled: booting without BIOS enabling an
> > output with this PLL, or reloading the driver.
> > 
> > While at it also add a debug print for the unexpected register value.
> 
> Workaround do not mention nothing about this DPLL0 dependency, maybe
> would be nice to comment in HSD about this.

Ok, can add comment.

> Have you tried to check if the workaround applies if DPLL1 is enabled?
> We could comment DPLL0 out from the adlp_plls table.

No, only DPLL0 makes it work, DPLL1 being enabled is not enough. You can
experiment with this by unloading the driver and simply enable/disable
DPLL0/1 (both power and enable flag) and try to read/modify/re-read the
CMTG_CHICKEN reg.

> > Cc: José Roberto de Souza <jose.souza@intel.com>
> > Signed-off-by: Imre Deak <imre.deak@intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_display.c  | 18 ----------
> >  drivers/gpu/drm/i915/display/intel_dpll_mgr.c | 34 ++++++++++++++++++-
> >  2 files changed, 33 insertions(+), 19 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> > index 4ca354f154215..98f7fbede6226 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -13284,24 +13284,6 @@ static void intel_early_display_was(struct drm_i915_private *dev_priv)
> >  			     KBL_ARB_FILL_SPARE_13 | KBL_ARB_FILL_SPARE_14,
> >  			     KBL_ARB_FILL_SPARE_14);
> >  	}
> > -
> > -	if (IS_ADLP_DISPLAY_STEP(dev_priv, STEP_A0, STEP_B0)) {
> > -		u32 val;
> > -
> > -		/*
> > -		 * Wa_16011069516:adl-p[a0]
> > -		 *
> > -		 * All CMTG regs are unreliable until CMTG clock gating is
> > -		 * disabled, so we can only assume the default CMTG_CHICKEN
> > -		 * reg value and sanity check this assumption with a double
> > -		 * read, which presumably returns the correct value even with
> > -		 * clock gating on.
> > -		 */
> > -		val = intel_de_read(dev_priv, TRANS_CMTG_CHICKEN);
> > -		val = intel_de_read(dev_priv, TRANS_CMTG_CHICKEN);
> > -		intel_de_write(dev_priv, TRANS_CMTG_CHICKEN, DISABLE_DPT_CLK_GATING);
> > -		drm_WARN_ON(&dev_priv->drm, val & ~DISABLE_DPT_CLK_GATING);
> > -	}
> >  }
> >  
> >  static void ibx_sanitize_pch_hdmi_port(struct drm_i915_private *dev_priv,
> > diff --git a/drivers/gpu/drm/i915/display/intel_dpll_mgr.c b/drivers/gpu/drm/i915/display/intel_dpll_mgr.c
> > index 0d72917e5670f..5c91d125a3371 100644
> > --- a/drivers/gpu/drm/i915/display/intel_dpll_mgr.c
> > +++ b/drivers/gpu/drm/i915/display/intel_dpll_mgr.c
> > @@ -3735,6 +3735,31 @@ static void icl_pll_enable(struct drm_i915_private *dev_priv,
> >  		drm_err(&dev_priv->drm, "PLL %d not locked\n", pll->info->id);
> >  }
> >  
> > +static void adlp_cmtg_clock_gating_wa(struct drm_i915_private *i915, struct intel_shared_dpll *pll)
> > +{
> > +	u32 val;
> > +
> > +	if (!IS_ADLP_DISPLAY_STEP(i915, STEP_A0, STEP_B0) ||
> > +	    pll->info->id != DPLL_ID_ICL_DPLL0)
> > +		return;
> > +	/*
> > +	 * Wa_16011069516:adl-p[a0]
> > +	 *
> > +	 * All CMTG regs are unreliable until CMTG clock gating is disabled,
> > +	 * so we can only assume the default TRANS_CMTG_CHICKEN reg value and
> > +	 * sanity check this assumption with a double read, which presumably
> > +	 * returns the correct value even with clock gating on.
> > +	 *
> > +	 * Instead of the usual place for workarounds we apply this one here,
> > +	 * since TRANS_CMTG_CHICKEN is only accessible while DPLL0 is enabled.
> > +	 */
> > +	val = intel_de_read(i915, TRANS_CMTG_CHICKEN);
> > +	val = intel_de_read(i915, TRANS_CMTG_CHICKEN);
> > +	intel_de_write(i915, TRANS_CMTG_CHICKEN, DISABLE_DPT_CLK_GATING);
> > +	if (drm_WARN_ON(&i915->drm, val & ~DISABLE_DPT_CLK_GATING))
> > +		drm_dbg_kms(&i915->drm, "Unexpected flags in TRANS_CMTG_CHICKEN: %08x\n", val);
> > +}
> > +
> >  static void combo_pll_enable(struct drm_i915_private *dev_priv,
> >  			     struct intel_shared_dpll *pll)
> >  {
> > @@ -3764,6 +3789,8 @@ static void combo_pll_enable(struct drm_i915_private *dev_priv,
> >  
> >  	icl_pll_enable(dev_priv, pll, enable_reg);
> >  
> > +	adlp_cmtg_clock_gating_wa(dev_priv, pll);
> > +
> >  	/* DVFS post sequence would be here. See the comment above. */
> >  }
> >  
> > @@ -4273,7 +4300,12 @@ void intel_dpll_readout_hw_state(struct drm_i915_private *i915)
> >  static void sanitize_dpll_state(struct drm_i915_private *i915,
> >  				struct intel_shared_dpll *pll)
> >  {
> > -	if (!pll->on || pll->active_mask)
> > +	if (!pll->on)
> > +		return;
> > +
> > +	adlp_cmtg_clock_gating_wa(i915, pll);
> > +
> > +	if (pll->active_mask)
> >  		return;
> >  
> >  	drm_dbg_kms(&i915->drm,
> 

  reply	other threads:[~2021-08-02 21:08 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-02 19:01 [Intel-gfx] [PATCH] drm/i915: Apply CMTG clock disabling WA while DPLL0 is enabled Imre Deak
2021-08-02 20:01 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for " Patchwork
2021-08-02 20:32 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2021-08-02 20:52 ` [Intel-gfx] [PATCH] " Souza, Jose
2021-08-02 21:07   ` Imre Deak [this message]
2021-08-02 21:16     ` Souza, Jose
2021-08-02 21:20       ` Imre Deak
2021-08-03 12:08 ` [Intel-gfx] ✗ Fi.CI.IGT: failure for " Patchwork
2021-08-03 13:51   ` Imre Deak
2021-08-03 14:38 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2021-08-03 15:32 ` [Intel-gfx] ✓ Fi.CI.IGT: " Patchwork
2021-08-04  5:32 ` Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210802210754.GA2115791@ideak-desk.fi.intel.com \
    --to=imre.deak@intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=jose.souza@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox