[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <85b807289ff2400ea5887ced63655862@AcuMS.aculab.com>
Date: Sat, 9 Mar 2024 14:33:26 +0000
From: David Laight <David.Laight@...LAB.COM>
To: 'Maxime Ripard' <mripard@...nel.org>, Arnd Bergmann <arnd@...db.de>
CC: Naresh Kamboju <naresh.kamboju@...aro.org>, open list
<linux-kernel@...r.kernel.org>, Linux ARM
<linux-arm-kernel@...ts.infradead.org>, "linux-sunxi@...ts.linux.dev"
<linux-sunxi@...ts.linux.dev>, "dri-devel@...ts.freedesktop.org"
<dri-devel@...ts.freedesktop.org>, "lkft-triage@...ts.linaro.org"
<lkft-triage@...ts.linaro.org>, Dave Airlie <airlied@...hat.com>, "Dan
Carpenter" <dan.carpenter@...aro.org>, Ard Biesheuvel <ardb@...nel.org>
Subject: RE: arm: ERROR: modpost: "__aeabi_uldivmod"
[drivers/gpu/drm/sun4i/sun4i-drm-hdmi.ko] undefined!
From: Maxime Ripard
> Sent: 04 March 2024 11:46
>
> On Mon, Mar 04, 2024 at 12:11:36PM +0100, Arnd Bergmann wrote:
> > On Mon, Mar 4, 2024, at 09:07, Naresh Kamboju wrote:
> > > The arm defconfig builds failed on today's Linux next tag next-20240304.
> > >
> > > Build log:
> > > ---------
> > > ERROR: modpost: "__aeabi_uldivmod"
> > > [drivers/gpu/drm/sun4i/sun4i-drm-hdmi.ko] undefined!
> > >
> >
> > Apparently caused by the 64-bit division in 358e76fd613a
> > ("drm/sun4i: hdmi: Consolidate atomic_check and mode_valid"):
> >
> >
> > +static enum drm_mode_status
> > +sun4i_hdmi_connector_clock_valid(const struct drm_connector *connector,
> > + const struct drm_display_mode *mode,
> > + unsigned long long clock)
> > {
> > - struct sun4i_hdmi *hdmi = drm_encoder_to_sun4i_hdmi(encoder);
> > - unsigned long rate = mode->clock * 1000;
> > - unsigned long diff = rate / 200; /* +-0.5% allowed by HDMI spec */
> > + const struct sun4i_hdmi *hdmi = drm_connector_to_sun4i_hdmi(connector);
> > + unsigned long diff = clock / 200; /* +-0.5% allowed by HDMI spec */
> > long rounded_rate;
> >
> > This used to be a 32-bit division. If the rate is never more than
> > 4.2GHz, clock could be turned back into 'unsigned long' to avoid
> > the expensive div_u64().
>
> I sent a fix for it this morning:
> https://lore.kernel.org/r/20240304091225.366325-1-mripard@kernel.org
>
> The framework will pass an unsigned long long because HDMI character
> rates can go up to 5.9GHz.
You could do:
/* The max clock is 5.9GHz, split the divide */
u32 diff = (u32)(clock / 8) / (200/8);
The code should really use u32 and u64.
Otherwise the sizes are different on 32bit.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)
Powered by blists - more mailing lists