Re: [PATCH] lib: fix kconfig dependency on ARCH_WANT_FRAME_POINTERS

[Geert Uytterhoeven]

Hi Julian,

On Thu, Mar 25, 2021 at 1:09 AM Julian Braha <julianbraha@xxxxxxxxx> wrote:
> On Wednesday, March 24, 2021 4:12:34 AM EDT you wrote:

> > On Wed, Mar 24, 2021 at 7:48 AM Julian Braha <julianbraha@xxxxxxxxx> wrote:
> > > On Monday, March 22, 2021 3:43:41 AM EDT you wrote:
> > > > On Sun, Mar 21, 2021 at 11:40 PM Julian Braha <julianbraha@xxxxxxxxx> wrote:
> > > > > On Sunday, March 21, 2021 2:28:43 PM EDT you wrote:
> > > > > > On Sat, Mar 20, 2021 at 1:17 AM Julian Braha <julianbraha@xxxxxxxxx> wrote:
> > > > > > > When LATENCYTOP is enabled and ARCH_WANT_FRAME_POINTERS
> > > > > > > is disabled, Kbuild gives the following warning:
> > > > > > >
> > > > > > > WARNING: unmet direct dependencies detected for FRAME_POINTER
> > > > > > >   Depends on [n]: DEBUG_KERNEL [=y] && (M68K || UML || SUPERH) || ARCH_WANT_FRAME_POINTERS [=n] || MCOUNT [=n]
> > > > > > >   Selected by [y]:
> > > > > > >   - LATENCYTOP [=y] && DEBUG_KERNEL [=y] && STACKTRACE_SUPPORT [=y] && PROC_FS [=y] && !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
> > > > > > >
> > > > > > > This is because LATENCYTOP selects FRAME_POINTER,
> > > > > > > without selecting or depending on ARCH_WANT_FRAME_POINTERS,
> > > > > > > despite FRAME_POINTER depending on ARCH_WANT_FRAME_POINTERS.
> > > > > > >
> > > > > > > Signed-off-by: Julian Braha <julianbraha@xxxxxxxxx>
> > > > > >
> > > > > > Thanks for your patch!
> > > > > >
> > > > > > > --- a/lib/Kconfig.debug
> > > > > > > +++ b/lib/Kconfig.debug
> > > > > > > @@ -1675,6 +1675,7 @@ config LATENCYTOP
> > > > > > >         depends on DEBUG_KERNEL
> > > > > > >         depends on STACKTRACE_SUPPORT
> > > > > > >         depends on PROC_FS
> > > > > > > +       select ARCH_WANT_FRAME_POINTERS if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
> > > > > >
> > > > > > ARCH_WANT_FRAME_POINTERS is a symbol that is only to be selected by
> > > > > > architecture-specific configuration, and must not be overridden:
> > > > > >
> > > > > >     # Select this config option from the architecture Kconfig, if it
> > > > > >     # is preferred to always offer frame pointers as a config
> > > > > >     # option on the architecture (regardless of KERNEL_DEBUG):
> > > > > >
> > > > > > Probably this should be turned into a depends instead?
> > > > > >
> > > > > > >         select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
> > > > > > >         select KALLSYMS
> > > > > > >         select KALLSYMS_ALL
> > > >
> > > > > Making this a 'depends' causes a recursive dependency error.
> > > > > Any other ideas?
> > > >
> > > > What about
> > > >
> > > >     -select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE &&
> > > > !ARM && !ARC && !X86
> > > >     +depends on FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE
> > > > && !ARM && !ARC && !X86
> > > >
> > > > ?
> > >
> > > Sadly, this won't work either. In Kconfig, 'depends' cannot have an 'if' after it (only 'select' can.)
> > > Kbuild gives an error for this.
> >
> > Oops
> >
> >     select FRAME_POINTER || MIPS || PPC || S390 || MICROBLAZE || ARM
> > || ARC || X86
> >
> > of course.

> I think it's a typo, but if you meant:
> select FRAME_POINTER if MIPS || PPC || S390 || MICROBLAZE || ARM || ARC || X86
> Then that works.

(Bummer, I shouldn't reply to emails before my morning coffee)
Yes, it is a typo.  I meant:

    depends on FRAME_POINTER || MIPS || PPC || S390 || MICROBLAZE ||
ARM || ARC || X86

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds