From: Arnd Bergmann <[email protected]>
Naresh and Antonio ran into a build failure with latest Debian
armhf compilers, with lots of output like
tmp/ccY3nOAs.s:2215: Error: selected processor does not support `cpsid i' in ARM mode
As it turns out, $(cc-option) fails early here when the FPU is not
selected before CPU architecture is selected, as the compiler
option check runs before enabling -msoft-float, which causes
a problem when testing a target architecture level without an FPU:
cc1: error: '-mfloat-abi=hard': selected architecture lacks an FPU
Passing e.g. -march=armv6k+fp in place of -march=armv6k would avoid this
issue, but the fallback logic is already broken because all supported
compilers (gcc-5 and higher) are much more recent than these options,
and building with -march=armv5t as a fallback no longer works.
The best way forward that I see is to just remove all the checks, which
also has the nice side-effect of slightly improving the startup time for
'make'.
The -mtune=marvell-f option was apparently never supported by any mainline
compiler, and the custom Codesourcery gcc build that did support is
now too old to build kernels, so just use -mtune=xscale unconditionally
for those.
This should be safe to apply on all stable kernels, and will be required
in order to keep building them with gcc-11 and higher.
Reported-by: Antonio Terceiro <[email protected]>
Reported-by: Naresh Kamboju <[email protected]>
Reported-by: Sebastian Andrzej Siewior <[email protected]>
Link: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=996419
Cc: Matthias Klose <[email protected]>
Cc: [email protected]
Signed-off-by: Arnd Bergmann <[email protected]>
---
arch/arm/Makefile | 22 +++++++++++-----------
1 file changed, 11 insertions(+), 11 deletions(-)
diff --git a/arch/arm/Makefile b/arch/arm/Makefile
index 847c31e7c368..fa45837b8065 100644
--- a/arch/arm/Makefile
+++ b/arch/arm/Makefile
@@ -60,15 +60,15 @@ KBUILD_CFLAGS += $(call cc-option,-fno-ipa-sra)
# Note that GCC does not numerically define an architecture version
# macro, but instead defines a whole series of macros which makes
# testing for a specific architecture or later rather impossible.
-arch-$(CONFIG_CPU_32v7M) =-D__LINUX_ARM_ARCH__=7 -march=armv7-m -Wa,-march=armv7-m
-arch-$(CONFIG_CPU_32v7) =-D__LINUX_ARM_ARCH__=7 $(call cc-option,-march=armv7-a,-march=armv5t -Wa$(comma)-march=armv7-a)
-arch-$(CONFIG_CPU_32v6) =-D__LINUX_ARM_ARCH__=6 $(call cc-option,-march=armv6,-march=armv5t -Wa$(comma)-march=armv6)
+arch-$(CONFIG_CPU_32v7M) =-D__LINUX_ARM_ARCH__=7 -march=armv7-m
+arch-$(CONFIG_CPU_32v7) =-D__LINUX_ARM_ARCH__=7 -march=armv7-a
+arch-$(CONFIG_CPU_32v6) =-D__LINUX_ARM_ARCH__=6 -march=armv6
# Only override the compiler option if ARMv6. The ARMv6K extensions are
# always available in ARMv7
ifeq ($(CONFIG_CPU_32v6),y)
-arch-$(CONFIG_CPU_32v6K) =-D__LINUX_ARM_ARCH__=6 $(call cc-option,-march=armv6k,-march=armv5t -Wa$(comma)-march=armv6k)
+arch-$(CONFIG_CPU_32v6K) =-D__LINUX_ARM_ARCH__=6 -march=armv6k
endif
-arch-$(CONFIG_CPU_32v5) =-D__LINUX_ARM_ARCH__=5 $(call cc-option,-march=armv5te,-march=armv4t)
+arch-$(CONFIG_CPU_32v5) =-D__LINUX_ARM_ARCH__=5 -march=armv5te
arch-$(CONFIG_CPU_32v4T) =-D__LINUX_ARM_ARCH__=4 -march=armv4t
arch-$(CONFIG_CPU_32v4) =-D__LINUX_ARM_ARCH__=4 -march=armv4
arch-$(CONFIG_CPU_32v3) =-D__LINUX_ARM_ARCH__=3 -march=armv3m
@@ -82,7 +82,7 @@ tune-$(CONFIG_CPU_ARM720T) =-mtune=arm7tdmi
tune-$(CONFIG_CPU_ARM740T) =-mtune=arm7tdmi
tune-$(CONFIG_CPU_ARM9TDMI) =-mtune=arm9tdmi
tune-$(CONFIG_CPU_ARM940T) =-mtune=arm9tdmi
-tune-$(CONFIG_CPU_ARM946E) =$(call cc-option,-mtune=arm9e,-mtune=arm9tdmi)
+tune-$(CONFIG_CPU_ARM946E) =-mtune=arm9e
tune-$(CONFIG_CPU_ARM920T) =-mtune=arm9tdmi
tune-$(CONFIG_CPU_ARM922T) =-mtune=arm9tdmi
tune-$(CONFIG_CPU_ARM925T) =-mtune=arm9tdmi
@@ -90,11 +90,11 @@ tune-$(CONFIG_CPU_ARM926T) =-mtune=arm9tdmi
tune-$(CONFIG_CPU_FA526) =-mtune=arm9tdmi
tune-$(CONFIG_CPU_SA110) =-mtune=strongarm110
tune-$(CONFIG_CPU_SA1100) =-mtune=strongarm1100
-tune-$(CONFIG_CPU_XSCALE) =$(call cc-option,-mtune=xscale,-mtune=strongarm110) -Wa,-mcpu=xscale
-tune-$(CONFIG_CPU_XSC3) =$(call cc-option,-mtune=xscale,-mtune=strongarm110) -Wa,-mcpu=xscale
-tune-$(CONFIG_CPU_FEROCEON) =$(call cc-option,-mtune=marvell-f,-mtune=xscale)
-tune-$(CONFIG_CPU_V6) =$(call cc-option,-mtune=arm1136j-s,-mtune=strongarm)
-tune-$(CONFIG_CPU_V6K) =$(call cc-option,-mtune=arm1136j-s,-mtune=strongarm)
+tune-$(CONFIG_CPU_XSCALE) =-mtune=xscale
+tune-$(CONFIG_CPU_XSC3) =-mtune=xscale
+tune-$(CONFIG_CPU_FEROCEON) =-mtune=xscale
+tune-$(CONFIG_CPU_V6) =-mtune=arm1136j-s
+tune-$(CONFIG_CPU_V6K) =-mtune=arm1136j-s
# Evaluate tune cc-option calls now
tune-y := $(tune-y)
--
2.29.2
On 2021-10-18 16:07:12 [+0200], Arnd Bergmann wrote:
…
> Passing e.g. -march=armv6k+fp in place of -march=armv6k would avoid this
> issue, but the fallback logic is already broken because all supported
> compilers (gcc-5 and higher) are much more recent than these options,
> and building with -march=armv5t as a fallback no longer works.
>
> The best way forward that I see is to just remove all the checks, which
> also has the nice side-effect of slightly improving the startup time for
> 'make'.
>
…
> This should be safe to apply on all stable kernels, and will be required
> in order to keep building them with gcc-11 and higher.
Yes, please.
> Reported-by: Antonio Terceiro <[email protected]>
> Reported-by: Naresh Kamboju <[email protected]>
> Reported-by: Sebastian Andrzej Siewior <[email protected]>
> Link: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=996419
> Cc: Matthias Klose <[email protected]>
> Cc: [email protected]
> Signed-off-by: Arnd Bergmann <[email protected]>
Reviewed-by: Sebastian Andrzej Siewior <[email protected]>
Just booted Debian 9/ Stretch which ships
gcc version 6.3.0 20170516 (Debian 6.3.0-18)
to confirm that it fails to compile with the armv5t fallback.
Sebastian
On 18.10.21 16:07, Arnd Bergmann wrote:
> From: Arnd Bergmann <[email protected]>
>
> Naresh and Antonio ran into a build failure with latest Debian
> armhf compilers, with lots of output like
>
> tmp/ccY3nOAs.s:2215: Error: selected processor does not support `cpsid i' in ARM mode
I just had exactly the same problem, and this patch fixes it for me as well.
Compile- and run-tested with gcc-11 (Debian 11.2.0-10) armhf & Linux 5.15.0
Tested-by: Klaus Kudielka <[email protected]>
On 2021-10-18 16:07:12 [+0200], Arnd Bergmann wrote:
…
>
> This should be safe to apply on all stable kernels, and will be required
> in order to keep building them with gcc-11 and higher.
Could we get this applied, please?
Matthias Klose asked gcc-upstream if the change, causing the problem,
was expected/ intended. I don't think this matters given that the
provided fallback does not work.
Sebastian
Hi,
On Mon, Oct 18, 2021 at 04:07:12PM +0200, Arnd Bergmann wrote:
> From: Arnd Bergmann <[email protected]>
>
> Naresh and Antonio ran into a build failure with latest Debian
> armhf compilers, with lots of output like
>
> tmp/ccY3nOAs.s:2215: Error: selected processor does not support `cpsid i' in ARM mode
>
> As it turns out, $(cc-option) fails early here when the FPU is not
> selected before CPU architecture is selected, as the compiler
> option check runs before enabling -msoft-float, which causes
> a problem when testing a target architecture level without an FPU:
>
> cc1: error: '-mfloat-abi=hard': selected architecture lacks an FPU
>
> Passing e.g. -march=armv6k+fp in place of -march=armv6k would avoid this
> issue, but the fallback logic is already broken because all supported
> compilers (gcc-5 and higher) are much more recent than these options,
> and building with -march=armv5t as a fallback no longer works.
>
> The best way forward that I see is to just remove all the checks, which
> also has the nice side-effect of slightly improving the startup time for
> 'make'.
>
> The -mtune=marvell-f option was apparently never supported by any mainline
> compiler, and the custom Codesourcery gcc build that did support is
> now too old to build kernels, so just use -mtune=xscale unconditionally
> for those.
>
> This should be safe to apply on all stable kernels, and will be required
> in order to keep building them with gcc-11 and higher.
>
> Reported-by: Antonio Terceiro <[email protected]>
> Reported-by: Naresh Kamboju <[email protected]>
> Reported-by: Sebastian Andrzej Siewior <[email protected]>
> Link: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=996419
> Cc: Matthias Klose <[email protected]>
> Cc: [email protected]
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
Thanks, I ran into this issue after affected gcc release migrated to
Debian testing. The patch makes the kernel compile again:
Tested-by: Sebastian Reichel <[email protected]>
Would be great if this could become part of 5.16-rc1, which is
usually used as base by subsystem maintainers.
Thanks,
-- Sebastian
> arch/arm/Makefile | 22 +++++++++++-----------
> 1 file changed, 11 insertions(+), 11 deletions(-)
>
> diff --git a/arch/arm/Makefile b/arch/arm/Makefile
> index 847c31e7c368..fa45837b8065 100644
> --- a/arch/arm/Makefile
> +++ b/arch/arm/Makefile
> @@ -60,15 +60,15 @@ KBUILD_CFLAGS += $(call cc-option,-fno-ipa-sra)
> # Note that GCC does not numerically define an architecture version
> # macro, but instead defines a whole series of macros which makes
> # testing for a specific architecture or later rather impossible.
> -arch-$(CONFIG_CPU_32v7M) =-D__LINUX_ARM_ARCH__=7 -march=armv7-m -Wa,-march=armv7-m
> -arch-$(CONFIG_CPU_32v7) =-D__LINUX_ARM_ARCH__=7 $(call cc-option,-march=armv7-a,-march=armv5t -Wa$(comma)-march=armv7-a)
> -arch-$(CONFIG_CPU_32v6) =-D__LINUX_ARM_ARCH__=6 $(call cc-option,-march=armv6,-march=armv5t -Wa$(comma)-march=armv6)
> +arch-$(CONFIG_CPU_32v7M) =-D__LINUX_ARM_ARCH__=7 -march=armv7-m
> +arch-$(CONFIG_CPU_32v7) =-D__LINUX_ARM_ARCH__=7 -march=armv7-a
> +arch-$(CONFIG_CPU_32v6) =-D__LINUX_ARM_ARCH__=6 -march=armv6
> # Only override the compiler option if ARMv6. The ARMv6K extensions are
> # always available in ARMv7
> ifeq ($(CONFIG_CPU_32v6),y)
> -arch-$(CONFIG_CPU_32v6K) =-D__LINUX_ARM_ARCH__=6 $(call cc-option,-march=armv6k,-march=armv5t -Wa$(comma)-march=armv6k)
> +arch-$(CONFIG_CPU_32v6K) =-D__LINUX_ARM_ARCH__=6 -march=armv6k
> endif
> -arch-$(CONFIG_CPU_32v5) =-D__LINUX_ARM_ARCH__=5 $(call cc-option,-march=armv5te,-march=armv4t)
> +arch-$(CONFIG_CPU_32v5) =-D__LINUX_ARM_ARCH__=5 -march=armv5te
> arch-$(CONFIG_CPU_32v4T) =-D__LINUX_ARM_ARCH__=4 -march=armv4t
> arch-$(CONFIG_CPU_32v4) =-D__LINUX_ARM_ARCH__=4 -march=armv4
> arch-$(CONFIG_CPU_32v3) =-D__LINUX_ARM_ARCH__=3 -march=armv3m
> @@ -82,7 +82,7 @@ tune-$(CONFIG_CPU_ARM720T) =-mtune=arm7tdmi
> tune-$(CONFIG_CPU_ARM740T) =-mtune=arm7tdmi
> tune-$(CONFIG_CPU_ARM9TDMI) =-mtune=arm9tdmi
> tune-$(CONFIG_CPU_ARM940T) =-mtune=arm9tdmi
> -tune-$(CONFIG_CPU_ARM946E) =$(call cc-option,-mtune=arm9e,-mtune=arm9tdmi)
> +tune-$(CONFIG_CPU_ARM946E) =-mtune=arm9e
> tune-$(CONFIG_CPU_ARM920T) =-mtune=arm9tdmi
> tune-$(CONFIG_CPU_ARM922T) =-mtune=arm9tdmi
> tune-$(CONFIG_CPU_ARM925T) =-mtune=arm9tdmi
> @@ -90,11 +90,11 @@ tune-$(CONFIG_CPU_ARM926T) =-mtune=arm9tdmi
> tune-$(CONFIG_CPU_FA526) =-mtune=arm9tdmi
> tune-$(CONFIG_CPU_SA110) =-mtune=strongarm110
> tune-$(CONFIG_CPU_SA1100) =-mtune=strongarm1100
> -tune-$(CONFIG_CPU_XSCALE) =$(call cc-option,-mtune=xscale,-mtune=strongarm110) -Wa,-mcpu=xscale
> -tune-$(CONFIG_CPU_XSC3) =$(call cc-option,-mtune=xscale,-mtune=strongarm110) -Wa,-mcpu=xscale
> -tune-$(CONFIG_CPU_FEROCEON) =$(call cc-option,-mtune=marvell-f,-mtune=xscale)
> -tune-$(CONFIG_CPU_V6) =$(call cc-option,-mtune=arm1136j-s,-mtune=strongarm)
> -tune-$(CONFIG_CPU_V6K) =$(call cc-option,-mtune=arm1136j-s,-mtune=strongarm)
> +tune-$(CONFIG_CPU_XSCALE) =-mtune=xscale
> +tune-$(CONFIG_CPU_XSC3) =-mtune=xscale
> +tune-$(CONFIG_CPU_FEROCEON) =-mtune=xscale
> +tune-$(CONFIG_CPU_V6) =-mtune=arm1136j-s
> +tune-$(CONFIG_CPU_V6K) =-mtune=arm1136j-s
>
> # Evaluate tune cc-option calls now
> tune-y := $(tune-y)
> --
> 2.29.2
>
On Sat, Nov 6, 2021 at 5:59 PM Sebastian Reichel
<[email protected]> wrote:
>
> Thanks, I ran into this issue after affected gcc release migrated to
> Debian testing. The patch makes the kernel compile again:
>
> Tested-by: Sebastian Reichel <[email protected]>
>
> Would be great if this could become part of 5.16-rc1, which is
> usually used as base by subsystem maintainers.
I've added it to Russell's patch tracker now as
https://www.armlinux.org.uk/developer/patches/viewpatch.php?id=9156/1
Arnd
Hi,
On Sat, Nov 06, 2021 at 08:04:35PM +0100, Arnd Bergmann wrote:
> On Sat, Nov 6, 2021 at 5:59 PM Sebastian Reichel
> <[email protected]> wrote:
> >
> > Thanks, I ran into this issue after affected gcc release migrated to
> > Debian testing. The patch makes the kernel compile again:
> >
> > Tested-by: Sebastian Reichel <[email protected]>
> >
> > Would be great if this could become part of 5.16-rc1, which is
> > usually used as base by subsystem maintainers.
>
> I've added it to Russell's patch tracker now as
>
> https://www.armlinux.org.uk/developer/patches/viewpatch.php?id=9156/1
Thanks.
FWIW, we uploaded in Debian 5.15.1 to experimental with the patch
applied, and the builds pass now: https://buildd.debian.org/status/logs.php?pkg=linux&arch=armhf
Regards,
Salvatore