Re: [PATCH] clk: rockchip: don't COMPILE_TEST builds on M68K

[Heiko Stuebner <heiko@sntech.de>]

Hi Geert,

Am Donnerstag, 4. Juni 2026, 09:56:04 Mitteleuropäische Sommerzeit schrieb Geert Uytterhoeven:
> On Wed, 3 Jun 2026 at 23:37, Heiko Stuebner <heiko@sntech.de> wrote:
> > Rockchip clock drivers use hash-tables with enums as inputs.
> >
> > M68K does interesting things in its __hash_32() implementation, casting
> > that u32 input to an u16 and therefore triggering warnings like:
> >
> >    drivers/clk/rockchip/clk-rk3528.c: note: in included file (through include/linux/hash.h, include/linux/slab.h):
> >      arch/m68k/include/asm/hash.h:57:24: sparse: sparse: cast truncates bits from constant value (18720 becomes 8720)
> >      arch/m68k/include/asm/hash.h:57:24: sparse: sparse: cast truncates bits from constant value (1e8e8 becomes e8e8)
> >
> > As M68K should never ever need Rockchip clock drivers, simply disable
> > compile-tests for M68K.
> >
> > Fixes: 7edfb7fb58ee ("clk: rockchip: allow COMPILE_TEST builds")
> > Reported-by: kernel test robot <lkp@intel.com>
> > Closes: https://lore.kernel.org/oe-kbuild-all/202605191434.PQkj2Rki-lkp@intel.com/
> > Signed-off-by: Heiko Stuebner <heiko@sntech.de>
> 
> Thanks for your patch!
> 
> > --- a/drivers/clk/rockchip/Kconfig
> > +++ b/drivers/clk/rockchip/Kconfig
> > @@ -4,6 +4,7 @@
> >  config COMMON_CLK_ROCKCHIP
> >         bool "Rockchip clock controller common support"
> >         depends on ARCH_ROCKCHIP || COMPILE_TEST
> > +       depends on !M68K
> >         default ARCH_ROCKCHIP
> >         help
> >           Say y here to enable common clock controller for Rockchip platforms.
> 
> Obviously this is the wrong fix ;-)
> 
> I have sent a better one, fixing the issue for every user:
> "[PATCH] m68k: hash: Use lower_16_bits() helper"
> https://lore.kernel.org/b55e9bd0532c0cad519809c86e0a8400060d75a1.1780559561.git.geert@linux-m68k.org

oh nice - fixing the actual problem instead of working around the
issue. I had no clue about m68k and thus no idea of how to test
anything, so went for the easx "fix" :-)


So thanks for looking into that
Heiko