[PATCH v16 9/9] drm/i915: Compute CMRR and calculate vtotal

Thu Jun 20 10:05:37 UTC 2024

Hi @Nathan Chancellor,

Yes, with do_div, we are expecting the remainder value. Regarding the warning related to the adjusted_pixel_rate type cast, I haven't been able to reproduce this locally, possibly due to differences in the cross-compiler. We should consider typecasting adjusted_pixel_rate or treating it as unsigned ?

Adding @Nikula, Jani to suggest.

Regards,
Mitul
> -----Original Message-----
> From: Nathan Chancellor <nathan at kernel.org>
> Sent: Wednesday, June 19, 2024 11:56 PM
> To: Golani, Mitulkumar Ajitkumar <mitulkumar.ajitkumar.golani at intel.com>
> Cc: intel-gfx at lists.freedesktop.org; Nautiyal, Ankit K
> <ankit.k.nautiyal at intel.com>; intel-xe at lists.freedesktop.org
> Subject: Re: [PATCH v16 9/9] drm/i915: Compute CMRR and calculate vtotal
> 
> On Wed, Jun 19, 2024 at 06:10:34PM +0000, Golani, Mitulkumar Ajitkumar
> wrote:
> > Hi @Nathan Chancellor
> >
> > Probably fix is merged in drm-intel-next related patch:
> > https://patchwork.freedesktop.org/series/134860/
> >
> > Can you please check and suggest if this patch is merged ?
> 
> This is still reproducible at commit 851de367dede ("drm/i915: Enable
> plane/pipeDMC ATS fault interrupts on mtl") in drm-intel-next, which includes
> that change as commit e2dc7cb72b25 ("drm/i915/display: Update calculation
> to avoid overflow"). The issue is the dividend in do_div() is required to be an
> unsigned 64-bit type but you used a signed type.
> Updating adjusted_pixel_rate to be a u64 should resolve the issue and match
> the return type of mul_u32_u32(). I just wasn't sure if that was the only fix
> this code would need, as do_div() is not typically used with an assignment.
> 
> Cheers,
> Nathan
> 
> > > -----Original Message-----
> > > From: Nathan Chancellor <nathan at kernel.org>
> > > Sent: Wednesday, June 19, 2024 9:12 PM
> > > To: Golani, Mitulkumar Ajitkumar
> > > <mitulkumar.ajitkumar.golani at intel.com>
> > > Cc: intel-gfx at lists.freedesktop.org; Nautiyal, Ankit K
> > > <ankit.k.nautiyal at intel.com>; intel-xe at lists.freedesktop.org
> > > Subject: Re: [PATCH v16 9/9] drm/i915: Compute CMRR and calculate
> > > vtotal
> > >
> > > Hi Mitul,
> > >
> > > On Mon, Jun 10, 2024 at 12:52:02PM +0530, Mitul Golani wrote:
> > > ...
> > > > diff --git a/drivers/gpu/drm/i915/display/intel_vrr.c
> > > > b/drivers/gpu/drm/i915/display/intel_vrr.c
> > > > index 4ad99a54aa83..05f67dc9d98d 100644
> > > > --- a/drivers/gpu/drm/i915/display/intel_vrr.c
> > > > +++ b/drivers/gpu/drm/i915/display/intel_vrr.c
> > > > @@ -12,6 +12,9 @@
> > > >  #include "intel_vrr_regs.h"
> > > >  #include "intel_dp.h"
> > > >
> > > > +#define FIXED_POINT_PRECISION		100
> > > > +#define CMRR_PRECISION_TOLERANCE	10
> > > > +
> > > >  bool intel_vrr_is_capable(struct intel_connector *connector)  {
> > > >  	const struct drm_display_info *info =
> > > > &connector->base.display_info; @@ -107,6 +110,52 @@ int
> > > > intel_vrr_vmax_vblank_start(const struct
> > > intel_crtc_state *crtc_state)
> > > >  	return crtc_state->vrr.vmax -
> > > > intel_vrr_vblank_exit_length(crtc_state);
> > > >  }
> > > >
> > > > +static bool
> > > > +is_cmrr_frac_required(struct intel_crtc_state *crtc_state) {
> > > > +	int calculated_refresh_k, actual_refresh_k, pixel_clock_per_line;
> > > > +	struct drm_display_mode *adjusted_mode = &crtc_state-
> > > >hw.adjusted_mode;
> > > > +	struct drm_i915_private *i915 =
> > > > +to_i915(crtc_state->uapi.crtc->dev);
> > > > +
> > > > +	if (!HAS_CMRR(i915))
> > > > +		return false;
> > > > +
> > > > +	actual_refresh_k =
> > > > +		drm_mode_vrefresh(adjusted_mode) *
> > > FIXED_POINT_PRECISION;
> > > > +	pixel_clock_per_line =
> > > > +		adjusted_mode->crtc_clock * 1000 / adjusted_mode-
> > > >crtc_htotal;
> > > > +	calculated_refresh_k =
> > > > +		pixel_clock_per_line * FIXED_POINT_PRECISION /
> > > > +adjusted_mode->crtc_vtotal;
> > > > +
> > > > +	if ((actual_refresh_k - calculated_refresh_k) <
> > > CMRR_PRECISION_TOLERANCE)
> > > > +		return false;
> > > > +
> > > > +	return true;
> > > > +}
> > > > +
> > > > +static unsigned int
> > > > +cmrr_get_vtotal(struct intel_crtc_state *crtc_state, bool
> > > > +video_mode_required) {
> > > > +	int multiplier_m = 1, multiplier_n = 1, vtotal, desired_refresh_rate;
> > > > +	long long adjusted_pixel_rate;
> > > > +	struct drm_display_mode *adjusted_mode =
> > > > +&crtc_state->hw.adjusted_mode;
> > > > +
> > > > +	desired_refresh_rate = drm_mode_vrefresh(adjusted_mode);
> > > > +
> > > > +	if (video_mode_required) {
> > > > +		multiplier_m = 1001;
> > > > +		multiplier_n = 1000;
> > > > +	}
> > > > +
> > > > +	crtc_state->cmrr.cmrr_n =
> > > > +		desired_refresh_rate * adjusted_mode->crtc_htotal *
> > > multiplier_n;
> > > > +	vtotal = (adjusted_mode->crtc_clock * 1000 * multiplier_n) /
> > > crtc_state->cmrr.cmrr_n;
> > > > +	adjusted_pixel_rate = adjusted_mode->crtc_clock * 1000 *
> > > multiplier_m;
> > > > +	crtc_state->cmrr.cmrr_m = do_div(adjusted_pixel_rate,
> > > > +crtc_state->cmrr.cmrr_n);
> > > > +
> > > > +	return vtotal;
> > > > +}
> > >
> > > This change is now in -next as commit 1676ecd303ac ("drm/i915:
> > > Compute CMRR and calculate vtotal"), where it breaks the xe build
> > > for 32-bit platforms
> > > with:
> > >
> > >   $ make -skj"$(nproc)" ARCH=arm CROSS_COMPILE=arm-linux-gnueabi-
> > > allmodconfig drivers/gpu/drm/xe/i915-display/intel_vrr.o
> > >   In file included from arch/arm/include/asm/div64.h:107,
> > >                    from include/linux/math.h:6,
> > >                    from include/linux/kernel.h:27,
> > >                    from include/linux/cpumask.h:11,
> > >                    from include/linux/smp.h:13,
> > >                    from include/linux/lockdep.h:14,
> > >                    from include/linux/spinlock.h:63,
> > >                    from include/linux/kref.h:16,
> > >                    from include/drm/drm_device.h:5,
> > >                    from include/drm/drm_drv.h:35,
> > >                    from drivers/gpu/drm/xe/compat-i915-headers/i915_drv.h:13,
> > >                    from drivers/gpu/drm/i915/display/intel_vrr.c:7:
> > >   drivers/gpu/drm/i915/display/intel_vrr.c: In function 'cmrr_get_vtotal':
> > >   include/asm-generic/div64.h:222:35: error: comparison of distinct
> > > pointer types lacks a cast [-Werror]
> > >     222 |         (void)(((typeof((n)) *)0) == ((uint64_t *)0));  \
> > >         |                                   ^~
> > >   drivers/gpu/drm/i915/display/intel_vrr.c:155:35: note: in
> > > expansion of macro 'do_div'
> > >     155 |         crtc_state->cmrr.cmrr_m = do_div(adjusted_pixel_rate,
> crtc_state-
> > > >cmrr.cmrr_n);
> > >         |                                   ^~~~~~
> > >   cc1: all warnings being treated as errors
> > >
> > > Also, is do_div() correct here? It is different from the other
> > > div_() macros in that the "return value" is the remainder, not the result of
> the division.
> > >
> > > Cheers,
> > > Nathan