Re: [PATCH] drm/stm: dsi: relax mode_valid clock tolerance

From: Sean Nyekjaer <sean@geanix.com>
To: Maxime Ripard <mripard@kernel.org>
Cc: Yannick Fertre <yannick.fertre@foss.st.com>,
	 Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>,
	Philippe Cornu <philippe.cornu@foss.st.com>,
	 Maarten Lankhorst <maarten.lankhorst@linux.intel.com>,
	Thomas Zimmermann <tzimmermann@suse.de>,
	 David Airlie <airlied@gmail.com>,
	Daniel Vetter <daniel@ffwll.ch>,
	 Maxime Coquelin <mcoquelin.stm32@gmail.com>,
	Alexandre Torgue <alexandre.torgue@foss.st.com>,
	 Robert Foss <rfoss@kernel.org>,
	Antonio Borneo <antonio.borneo@foss.st.com>,
	 dri-devel@lists.freedesktop.org,
	linux-stm32@st-md-mailman.stormreply.com,
	 linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH] drm/stm: dsi: relax mode_valid clock tolerance
Date: Mon, 29 Apr 2024 10:17:45 +0200	[thread overview]
Message-ID: <cvzecixldubeq7pwn77cggs7tcwc5on3arlnboj4fbpqgdygtu@hat3r6afzu7y> (raw)
In-Reply-To: <20240424-famous-fascinating-hyena-8fb3a7@houat>

Hi,

On Wed, Apr 24, 2024 at 09:21:17AM UTC, Maxime Ripard wrote:
> Hi,
> 
> Sorry, my previous review didn't go through.
> 
> On Fri, Mar 22, 2024 at 11:47:31AM +0100, Sean Nyekjaer wrote:
> > When using the DSI interface via DSI2LVDS bridge, it seems a bit harsh
> > to reguire the requested and the actual px clock to be within
> > 50Hz. A typical LVDS display requires the px clock to be within +-10%.
> > 
> > In case for HDMI .5% tolerance is required.
> > 
> > Fixes: e01356d18273 ("drm/stm: dsi: provide the implementation of mode_valid()")
> > Signed-off-by: Sean Nyekjaer <sean@geanix.com>
> > ---
> >  drivers/gpu/drm/stm/dw_mipi_dsi-stm.c | 7 +++----
> >  1 file changed, 3 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/stm/dw_mipi_dsi-stm.c b/drivers/gpu/drm/stm/dw_mipi_dsi-stm.c
> > index d5f8c923d7bc..97936b0ef702 100644
> > --- a/drivers/gpu/drm/stm/dw_mipi_dsi-stm.c
> > +++ b/drivers/gpu/drm/stm/dw_mipi_dsi-stm.c
> > @@ -322,8 +322,6 @@ dw_mipi_dsi_phy_get_timing(void *priv_data, unsigned int lane_mbps,
> >  	return 0;
> >  }
> >  
> > -#define CLK_TOLERANCE_HZ 50
> > -
> >  static enum drm_mode_status
> >  dw_mipi_dsi_stm_mode_valid(void *priv_data,
> >  			   const struct drm_display_mode *mode,
> > @@ -375,9 +373,10 @@ dw_mipi_dsi_stm_mode_valid(void *priv_data,
> >  		/*
> >  		 * Filter modes according to the clock value, particularly useful for
> >  		 * hdmi modes that require precise pixel clocks.
> > +		 * Check that px_clock is within .5% tolerance.
> >  		 */
> > -		if (px_clock_hz < target_px_clock_hz - CLK_TOLERANCE_HZ ||
> > -		    px_clock_hz > target_px_clock_hz + CLK_TOLERANCE_HZ)
> > +		if (px_clock_hz < mult_frac(target_px_clock_hz, 995, 1000) ||
> > +		    px_clock_hz > mult_frac(target_px_clock_hz, 1005, 1000))
> >  			return MODE_CLOCK_RANGE;
> 
> I wonder if it's not something that should be made into a helper. We
> have a couple of drivers doing it already, so it might be worth creating
> a function that checks for a given struct clk pointer and pixel clock if
> it's within parameters.
> 
> Maxime

Yes agree, if the same calculation is happening other places.
I can't identify where it happens though.

Would that helper function exist in drivers/gpu/drm/drm_hdmi_helper.c ?

I will give the implementation a shot.

/Sean