Re: [PATCH] drm/bridge/synopsys: avoid field overflow warning

["Heiko Stübner" <heiko@sntech.de>]

Am Donnerstag, 10. April 2025, 14:36:45 Mitteleuropäische Sommerzeit schrieb Arnd Bergmann:
> On Wed, Apr 9, 2025, at 09:07, Arnd Bergmann wrote:
> > On Tue, Apr 8, 2025, at 19:51, Arnd Bergmann wrote:
> >> From: Arnd Bergmann <arnd@arndb.de>
> >>
> >> clang-16 and earlier complain about what it thinks might be an out of
> >> range number:
> >>
> >> drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi2.c:348:8: error: call to 
> >> __compiletime_assert_579 declared with 'error' attribute: FIELD_PREP: 
> >> value too large for the field
> >>                      PHY_SYS_RATIO(tmp));
> >>                      ^
> >> drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi2.c:90:27: note: expanded 
> >> from macro 'PHY_SYS_RATIO'
> >>  #define PHY_SYS_RATIO(x)                FIELD_PREP(GENMASK(16, 0), x)
> >>
> >
> > I still see the same build failure in some other configurations even
> > with this patch. Please disregard this version, I'll try to come
> > up with a better one.
> 
> I couldn't come up with anything that actually worked, other than
> the hack below, which just works around the compiletime error
> in FIELD_PREP(), but doesn't look like a proper fix.
> 
> If anyone else has any ideas, I can test their patch.
> 
>        Arnd
> 
> diff --git a/drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi2.c b/drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi2.c
> index c76f5f2e74d1..8ba528462ede 100644
> --- a/drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi2.c
> +++ b/drivers/gpu/drm/bridge/synopsys/dw-mipi-dsi2.c
> @@ -344,7 +344,7 @@ static void dw_mipi_dsi2_phy_ratio_cfg(struct dw_mipi_dsi2 *dsi2)
>          */
>         tmp = DIV_ROUND_CLOSEST_ULL(phy_hsclk << 16, sys_clk);
>         regmap_write(dsi2->regmap, DSI2_PHY_SYS_RATIO_MAN_CFG,
> -                    PHY_SYS_RATIO(tmp));
> +                    PHY_SYS_RATIO(tmp & GENMASK(16, 0)));

so should we take this one for the time being?

While FIELD_PREP does _some_ type checking for mask and value, they are
"only" build-time checks for when the value is a builtin constant.

It doesn't do checks at runtime, so it seems this check would fall into
the submitting code's realm?


>  }
>  
>  static void dw_mipi_dsi2_lp2hs_or_hs2lp_cfg(struct dw_mipi_dsi2 *dsi2)
>