Re: [PATCH] m68k: Define NR_CPUS to 1

["Uwe Kleine-König" <ukleinek@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 1784 bytes --]

On Mon, Aug 03, 2026 at 09:58:14AM +0200, Geert Uytterhoeven wrote:
> Hi Uwe,
> 
> Thanks for your patch!
> 
> On Fri, 31 Jul 2026 at 11:50, Uwe Kleine-König <ukleinek@kernel.org> wrote:
> > This fixes a Kconfig warning
> >
> >         fs/erofs/Kconfig:137:warning: range is invalid
> >
> > which originates from EROFS_FS_ZIP_LZMA_DEFAULT_MAX_STREAMS using
> > NR_CPUS which up to now didn't exist for ARCH=m68k. All other
> > architectures define this symbol, so fix the outlier.
> 
> Apart from the Kconfig warning, this also
>   1. Breaks pressing "ENTER" in "make oldconfig" to apply a sensible
>      default value, just repeating the question ad infinitum instead
>      (the default is 16),
>   2. "make olddefconfig" sets it to literal NR_CPUS, which is invalid, too.
> 
> > Fixes: c9b47e6b2311 ("erofs: cap LZMA stream pool size")
> > Signed-off-by: Uwe Kleine-König <ukleinek@kernel.org>
> 
> > --- a/arch/m68k/Kconfig.cpu
> > +++ b/arch/m68k/Kconfig.cpu
> > @@ -549,6 +549,10 @@ config CACHE_COPYBACK
> >  endchoice
> >  endif # HAVE_CACHE_CB
> >
> > +config NR_CPUS
> > +       int
> > +       default "1"
> 
> While this does fix the issues on m68k, I don't think this is the
> right fix:
>   - NR_CPUS depends on SMP on most architectures, so you do not get
>      the "range is invalid" warning on non-SMP, but NR_CPUS is zero.
>     Hence I managed to reproduce the two issues above (albeit 2 with
>     CONFIG_EROFS_FS_ZIP_LZMA_DEFAULT_MAX_STREAMS=0) with a non-SMP
>     ARM config.
>   - Kconfig logic using NR_CPUS typically includes a check for SMP, too.
> 
> So I think EROFS_FS_ZIP_LZMA_DEFAULT_MAX_STREAMS should be fixed
> instead.

Agreed, probably it should depend on SMP, too.

Thanks for your feedback
Uwe

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]