[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1d6a10ca-3d86-452a-bdba-03a1204dd232@app.fastmail.com>
Date: Tue, 29 Oct 2024 22:05:10 +0000
From: "Arnd Bergmann" <arnd@...db.de>
To: "Thomas Gleixner" <tglx@...utronix.de>,
"kernel test robot" <lkp@...el.com>,
"Vincenzo Frascino" <vincenzo.frascino@....com>
Cc: oe-kbuild-all@...ts.linux.dev, linux-kernel@...r.kernel.org,
x86@...nel.org, "Geert Uytterhoeven" <geert@...ux-m68k.org>
Subject: Re: [tip:timers/vdso 11/13] include/vdso/page.h:15:39: error:
'CONFIG_PAGE_SHIFT' undeclared here (not in a function); did you mean
'CONFIG_LOG_BUF_SHIFT'?
On Tue, Oct 29, 2024, at 21:18, Thomas Gleixner wrote:
> On Tue, Oct 29 2024 at 06:38, kernel test robot wrote:
>
> That's actually not the culprit. The problem exists in mainline too. The
> actual root cause is:
>
> 5394f1e9b687 ("arch: define CONFIG_PAGE_SIZE_*KB on all architectures")
>
> The config has no CONFIG_PAGE_SHIFT because nothing selects a page size. The
> the original code had:
>
> if defined(CONFIG_SUN3) || defined(CONFIG_COLDFIRE)
> #define PAGE_SHIFT 13
> #else
> #define PAGE_SHIFT 12
> #endif
>
> But the Kconfig magic changes made this depend on various other things.
>
> The patch below cures it, but it needs an eyeball from Geert.
I have previously suggested a similar but more extensive patch
to address the problem:
https://lore.kernel.org/all/CAMuHMdVC-FUSA5C9aNrvP3=RaRWrchhUEC5UYcSGMz_ep1PEhg@mail.gmail.com/
I can rebase that one if Geert still wants it. I think I sent
an updated version somewhere but can't find it now.
> --- a/arch/m68k/Kconfig.cpu
> +++ b/arch/m68k/Kconfig.cpu
> @@ -21,6 +21,8 @@ choice
> config M68KCLASSIC
> bool "Classic M68K CPU family support"
> select HAVE_ARCH_PFN_VALID
> + select HAVE_PAGE_SIZE_4KB if !SUN3
> + select HAVE_PAGE_SIZE_8KB if SUN3
>
> config COLDFIRE
> bool "Coldfire CPU family support"
> @@ -30,7 +32,7 @@ config COLDFIRE
> select GENERIC_CSUM
> select GPIOLIB
> select HAVE_LEGACY_CLK
> - select HAVE_PAGE_SIZE_8KB if !MMU
> + select HAVE_PAGE_SIZE_8KB
>
> endchoice
>
> @@ -46,7 +48,6 @@ config M68000
> select GENERIC_CSUM
> select CPU_NO_EFFICIENT_FFS
> select HAVE_ARCH_HASH
> - select HAVE_PAGE_SIZE_4KB
> select LEGACY_TIMER_TICK
> help
> The Freescale (was Motorola) 68000 CPU is the first generation of
Part of the problem here is that this simpler patch
then just causes different build failures for the
same configurations, since m68k kernels cannot successfully
build when no platform is selected.
Arnd
Powered by blists - more mailing lists