Linux MIPS Architecture development
 help / color / mirror / Atom feed
From: Nathan Chancellor <nathan@kernel.org>
To: Jiaxun Yang <jiaxun.yang@flygoat.com>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>,
	"linux-mips@vger.kernel.org" <linux-mips@vger.kernel.org>,
	llvm@lists.linux.dev, patches@lists.linux.dev
Subject: Re: [PATCH] MIPS: generic: Do not select CPUs that are unsupported in clang
Date: Thu, 6 Apr 2023 14:39:42 -0700	[thread overview]
Message-ID: <20230406213942.GA2148894@dev-arch.thelio-3990X> (raw)
In-Reply-To: <6B77736C-1015-45A7-A48A-92E4CBE95ACA@flygoat.com>

On Thu, Apr 06, 2023 at 10:30:22PM +0100, Jiaxun Yang wrote:
> 
> 
> > 2023年4月6日 22:19,Jiaxun Yang <jiaxun.yang@flygoat.com> 写道:
> > 
> > 
> > 
> >> 2023年4月6日 22:09,Nathan Chancellor <nathan@kernel.org> 写道:
> >> 
> >> On Thu, Apr 06, 2023 at 09:59:45PM +0100, Jiaxun Yang wrote:
> >>> 
> >>> 
> >>>> 2023年4月6日 21:09,Nathan Chancellor <nathan@kernel.org> 写道:
> >>>> 
> >>>> When building allnoconfig with clang after commit de34007751aa ("MIPS:
> >>>> generic: Enable all CPUs supported by virt board in Kconfig"), the
> >>>> following error occurs:
> >>>> 
> >>>> error: unknown target CPU 'r4600'
> >>>> note: valid target CPU values are: mips1, mips2, mips3, mips4, mips5, mips32, mips32r2, mips32r3, mips32r5, mips32r6, mips64, mips64r2, mips64r3, mips64r5, mips64r6, octeon, octeon+, p5600
> >>>> 
> >>>> Working around that, there are similar errors for 'loongson2e' and
> >>>> 'loongson2f'.
> >>>> 
> >>>> These CPUs are not supported in clang/LLVM, so do not select support for
> >>>> them in MIPS_GENERIC_KERNEL when building with clang.
> >>>> 
> >>>> Fixes: de34007751aa ("MIPS: generic: Enable all CPUs supported by virt board in Kconfig")
> >>>> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> >>> 
> >>> Thanks for the patch.
> >>> 
> >>> Actually there is no need for any special support for those CPUs.
> >>> We should be able to build those kernels with -march=mips3.
> >>> 
> >>> I think something like:
> >>> 
> >>> --- a/arch/mips/Makefile
> >>> +++ b/arch/mips/Makefile
> >>> @@ -150,7 +150,7 @@ cflags-y += $(call cc-option,-Wa$(comma)-mno-fix-loongson3-llsc,)
> >>> #
> >>> cflags-$(CONFIG_CPU_R3000)     += -march=r3000
> >>> cflags-$(CONFIG_CPU_R4300)     += -march=r4300 -Wa,--trap
> >>> -cflags-$(CONFIG_CPU_R4X00)     += -march=r4600 -Wa,--trap
> >>> +cflags-$(CONFIG_CPU_R4X00)     += $(call cc-option,-march=r4600,-march=mips3) -Wa,--trap
> >>> cflags-$(CONFIG_CPU_TX49XX)    += -march=r4600 -Wa,--trap
> >>> cflags-$(CONFIG_CPU_MIPS32_R1) += -march=mips32 -Wa,--trap
> >>> cflags-$(CONFIG_CPU_MIPS32_R2) += -march=mips32r2 -Wa,--trap
> >>> @@ -181,8 +181,8 @@ endif
> >>> cflags-$(CONFIG_CAVIUM_CN63XXP1) += -Wa,-mfix-cn63xxp1
> >>> cflags-$(CONFIG_CPU_BMIPS)     += -march=mips32 -Wa,-mips32 -Wa,--trap
> >>> 
> >>> -cflags-$(CONFIG_CPU_LOONGSON2E) += -march=loongson2e -Wa,--trap
> >>> -cflags-$(CONFIG_CPU_LOONGSON2F) += -march=loongson2f -Wa,--trap
> >>> +cflags-$(CONFIG_CPU_LOONGSON2E) += $(call cc-option,-march=loongson2e,-march=mips3) -Wa,--trap
> >>> +cflags-$(CONFIG_CPU_LOONGSON2F) += $(call cc-option,-march=loongson2f,-march=mips3) -Wa,--trap
> >>> # Some -march= flags enable MMI instructions, and GCC complains about that
> >>> # support being enabled alongside -msoft-float. Thus explicitly disable MMI.
> >>> cflags-$(CONFIG_CPU_LOONGSON2EF) += $(call cc-option,-mno-loongson-mmi)
> >>> 
> >>> Will make them build.
> >> 
> >> Thank you for taking a look and the suggestion! I applied it and tried
> >> to build allnoconfig but I get
> >> 
> >> error: ABI 'o32' is not supported on CPU 'mips3'
> >> 
> >> immediately, which could certainly be a bug in clang...
> > 
> > There is a clang patch:
> > https://reviews.llvm.org/D146269

Aha, good to know that issue is finally getting resolved :)

> > I’ve seen this warning with CONFIG_MIPS32_O32 enabled, but I think that’s not the case for allnoconfig...
> 
> 
> Oh I see, it’s trying to build a o32 kernel.

I can test this diff tomorrow. Sounds like you have a better grasp on
this problem than I do though so it would probably be better if you sent
the patch but I do not mind doing it if you are unable to.

For the record, with allnoconfig I end up with:

$ rg 64BIT .config
292:CONFIG_SYS_SUPPORTS_64BIT_KERNEL=y
294:CONFIG_CPU_SUPPORTS_64BIT_KERNEL=y
302:# CONFIG_64BIT is not set

$ rg O32 .config
345:# CONFIG_MIPS_O32_FP64_SUPPORT is not set

> --- a/arch/mips/Kconfig
> +++ b/arch/mips/Kconfig
> @@ -1579,7 +1579,7 @@ config CPU_R4300
> config CPU_R4X00
> bool "R4x00"
> depends on SYS_HAS_CPU_R4X00
> - select CPU_SUPPORTS_32BIT_KERNEL
> + select CPU_SUPPORTS_32BIT_KERNEL if CC_SUPPORTS_O32_WITH_64BIT_ARCH
> select CPU_SUPPORTS_64BIT_KERNEL
> select CPU_SUPPORTS_HUGEPAGES
> help
> @@ -1743,9 +1743,11 @@ config CPU_MIPS32_R5_XPA
> if CPU_LOONGSON2F
> config CPU_NOP_WORKAROUNDS
> bool
> + depends on $(cc-option,-Wa,-mfix-loongson2f-nop)
> config CPU_JUMP_WORKAROUNDS
> bool
> + depends on $(cc-option,-Wa,-mfix-loongson2f-jump)
> config CPU_LOONGSON2F_WORKAROUNDS
> bool "Loongson 2F Workarounds"
> @@ -2497,6 +2499,7 @@ config CPU_HAS_SYNC
> # "MIPS R4400MC Errata, Processor Revision 1.0", erratum #5
> config CPU_DADDI_WORKAROUNDS
> bool
> + depends on $(cc-option,-mno-daddi)
> # Work around certain R4000 CPU errata (as implemented by GCC):
> #
> @@ -2518,6 +2521,7 @@ config CPU_DADDI_WORKAROUNDS
> # erratum #52
> config CPU_R4000_WORKAROUNDS
> bool
> + depends on $(cc-option,-mfix-r4000)
> select CPU_R4400_WORKAROUNDS
> # Work around certain R4400 CPU errata (as implemented by GCC):
> @@ -2528,6 +2532,7 @@ config CPU_R4000_WORKAROUNDS
> # "MIPS R4400MC Errata, Processor Revision 2.0 & 3.0", erratum #4
> config CPU_R4400_WORKAROUNDS
> bool
> + depends on $(cc-option,-mfix-r4400)
> config CPU_R4X00_BUGS64
> bool
> @@ -3174,6 +3179,7 @@ config COMPAT
> config MIPS32_O32
> bool "Kernel support for o32 binaries"
> depends on 64BIT
> + depends on CC_SUPPORTS_O32_WITH_64BIT_ARCH
> select ARCH_WANT_OLD_COMPAT_IPC
> select COMPAT
> select MIPS32_COMPAT
> @@ -3206,6 +3212,9 @@ config CC_HAS_MNO_BRANCH_LIKELY
> config CC_HAS_BROKEN_INLINE_COMPAT_BRANCH
> def_bool y if CC_IS_CLANG
> +config CC_SUPPORTS_O32_WITH_64BIT_ARCH
> + def_bool y if $(cc-option,-march=mips3 -mabi=o32)
> +
> menu "Power management options"
> config ARCH_HIBERNATION_POSSIBLE
> 
> This will prevent clang build 32bit kernel in case it doesn’t support o32 ABI with 64bit CPU.

Cheers,
Nathan

      reply	other threads:[~2023-04-06 21:39 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-04-06 20:09 [PATCH] MIPS: generic: Do not select CPUs that are unsupported in clang Nathan Chancellor
2023-04-06 20:59 ` Jiaxun Yang
2023-04-06 21:09   ` Nathan Chancellor
2023-04-06 21:19     ` Jiaxun Yang
2023-04-06 21:30       ` Jiaxun Yang
2023-04-06 21:39         ` Nathan Chancellor [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230406213942.GA2148894@dev-arch.thelio-3990X \
    --to=nathan@kernel.org \
    --cc=jiaxun.yang@flygoat.com \
    --cc=linux-mips@vger.kernel.org \
    --cc=llvm@lists.linux.dev \
    --cc=patches@lists.linux.dev \
    --cc=tsbogend@alpha.franken.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox