New build failures in sparc.

New build failures in sparc.

[Paul Gortmaker]

Just a heads up that these turned up today in linux-next:

------------
arch/sparc/lib/mul.S:134: Error: Hardware capability "mul32" not
enabled for "smul".
make[2]: *** [arch/sparc/lib/mul.o] Error 1
arch/sparc/lib/rem.S:378: Error: Hardware capability "div32" not
enabled for "sdivcc".
arch/sparc/lib/rem.S:381: Error: Hardware capability "mul32" not
enabled for "smul".
arch/sparc/lib/sdiv.S:377: Error: Hardware capability "div32" not
enabled for "sdivcc".
make[2]: *** [arch/sparc/lib/rem.o] Error 1
make[2]: *** [arch/sparc/lib/sdiv.o] Error 1
arch/sparc/lib/udiv.S:356: Error: Hardware capability "div32" not
enabled for "udiv".
arch/sparc/lib/umul.S:168: Error: Hardware capability "mul32" not
enabled for "umul".
make[2]: *** [arch/sparc/lib/udiv.o] Error 1
arch/sparc/lib/urem.S:354: Error: Hardware capability "div32" not
enabled for "udiv".
arch/sparc/lib/urem.S:355: Error: Hardware capability "mul32" not
enabled for "umul".
make[2]: *** [arch/sparc/lib/umul.o] Error 1
make[2]: *** [arch/sparc/lib/urem.o] Error 1
make[1]: *** [arch/sparc/lib] Error 2
make: *** [sub-make] Error 2
-------------

http://kisskb.ellerman.id.au/kisskb/buildresult/5869520/
http://kisskb.ellerman.id.au/kisskb/buildresult/5869567/

I've not done any investigation into the root cause (yet).

Paul.

Re: New build failures in sparc.

[David Miller]

From: Paul Gortmaker <paul.gortmaker@windriver.com>
Date: Tue, 13 Mar 2012 18:28:18 -0400

> Just a heads up that these turned up today in linux-next:

You must have updated your sparc binutils recently, which is now more
strict about making sure the proper options are passed to enable v8
instructions.

See also:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50979

Re: New build failures in sparc.

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 696 bytes --]

Hi Dave,

On Tue, 13 Mar 2012 15:48:48 -0700 (PDT) David Miller <davem@davemloft.net> wrote:
>
> From: Paul Gortmaker <paul.gortmaker@windriver.com>
> Date: Tue, 13 Mar 2012 18:28:18 -0400
> 
> > Just a heads up that these turned up today in linux-next:
> 
> You must have updated your sparc binutils recently, which is now more
> strict about making sure the proper options are passed to enable v8
> instructions.
> 
> See also:
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50979

Yeah, we updated our sparc64 toolchain yesterday ... Tony says he will try
applying the patch(es) from the bugzilla.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: New build failures in sparc.

[David Miller]

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Wed, 14 Mar 2012 10:28:30 +1100

> Yeah, we updated our sparc64 toolchain yesterday ... Tony says he
> will try applying the patch(es) from the bugzilla.

Meanwhile I'll add a workaround to the sparc makefiles, and
push that to Linus.  This way it will work no matter what
combination of tools someone has.

Re: New build failures in sparc.

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 581 bytes --]

Hi Dave,

On Tue, 13 Mar 2012 17:37:41 -0700 (PDT) David Miller <davem@davemloft.net> wrote:
>
> From: Stephen Rothwell <sfr@canb.auug.org.au>
> Date: Wed, 14 Mar 2012 10:28:30 +1100
> 
> > Yeah, we updated our sparc64 toolchain yesterday ... Tony says he
> > will try applying the patch(es) from the bugzilla.
> 
> Meanwhile I'll add a workaround to the sparc makefiles, and
> push that to Linus.  This way it will work no matter what
> combination of tools someone has.

Excellent, thanks.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: New build failures in sparc.

[Tony Breeds]

[-- Attachment #1: Type: text/plain, Size: 575 bytes --]

On Tue, Mar 13, 2012 at 05:37:41PM -0700, David Miller wrote:
> From: Stephen Rothwell <sfr@canb.auug.org.au>
> Date: Wed, 14 Mar 2012 10:28:30 +1100
> 
> > Yeah, we updated our sparc64 toolchain yesterday ... Tony says he
> > will try applying the patch(es) from the bugzilla.
> 
> Meanwhile I'll add a workaround to the sparc makefiles, and
> push that to Linus.  This way it will work no matter what
> combination of tools someone has.

Thanks Dave, we may be hitting a related but different issue as gcc
4.6.3 seems to have that patch already.

Yours Tony

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: New build failures in sparc.

[David Miller]

From: Tony Breeds <tony@bakeyournoodle.com>
Date: Wed, 14 Mar 2012 15:01:02 +1100

> On Tue, Mar 13, 2012 at 05:37:41PM -0700, David Miller wrote:
>> From: Stephen Rothwell <sfr@canb.auug.org.au>
>> Date: Wed, 14 Mar 2012 10:28:30 +1100
>> 
>> > Yeah, we updated our sparc64 toolchain yesterday ... Tony says he
>> > will try applying the patch(es) from the bugzilla.
>> 
>> Meanwhile I'll add a workaround to the sparc makefiles, and
>> push that to Linus.  This way it will work no matter what
>> combination of tools someone has.
> 
> Thanks Dave, we may be hitting a related but different issue as gcc
> 4.6.3 seems to have that patch already.

See the patch I sent to Linus which landed in his tree an hour or so
ago:

--------------------
[PATCH] sparc32: Add -Av8 to assembler command line.

Newer version of binutils are more strict about specifying the
correct options to enable certain classes of instructions.

The sparc32 build is done for v7 in order to support sun4c systems
which lack hardware integer multiply and divide instructions.

So we have to pass -Av8 when building the assembler routines that
use these instructions and get patched into the kernel when we find
out that we have a v8 capable cpu.

Reported-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc/Makefile |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/sparc/Makefile b/arch/sparc/Makefile
index ad1fb5d..eddcfb3 100644
--- a/arch/sparc/Makefile
+++ b/arch/sparc/Makefile
@@ -31,7 +31,7 @@ UTS_MACHINE    := sparc
 
 #KBUILD_CFLAGS += -g -pipe -fcall-used-g5 -fcall-used-g7
 KBUILD_CFLAGS += -m32 -pipe -mno-fpu -fcall-used-g5 -fcall-used-g7
-KBUILD_AFLAGS += -m32
+KBUILD_AFLAGS += -m32 -Wa,-Av8
 
 #LDFLAGS_vmlinux = -N -Ttext 0xf0004000
 #  Since 2.5.40, the first stage is left not btfix-ed.
-- 
1.7.6.5

Re: New build failures in sparc.

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 398 bytes --]

Hi Dave,

On Tue, 13 Mar 2012 21:46:09 -0700 (PDT) David Miller <davem@davemloft.net> wrote:
>
> See the patch I sent to Linus which landed in his tree an hour or so
> ago:
> 
> --------------------
> [PATCH] sparc32: Add -Av8 to assembler command line.

I will cherry-pick that onto the end of linux-next today.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: New build failures in sparc.

[Paul Gortmaker]

On Tue, Mar 13, 2012 at 6:48 PM, David Miller <davem@davemloft.net> wrote:
> From: Paul Gortmaker <paul.gortmaker@windriver.com>
> Date: Tue, 13 Mar 2012 18:28:18 -0400
>
>> Just a heads up that these turned up today in linux-next:
>
> You must have updated your sparc binutils recently, which is now more
> strict about making sure the proper options are passed to enable v8
> instructions.

Thanks Dave, you are absolutely correct.  Stephen indicated some
toolchain updates were pending when I'd asked about mips but
it never crossed my mind again until now.

I went and looked, and last weeks build (which was OK) was using
the gcc-4.4.3 (+ whatever binutils) and the failure today is now on
gcc-4.6.3 (+ some newer binutils).

http://kisskb.ellerman.id.au/kisskb/buildresult/5826617/

vs:

http://kisskb.ellerman.id.au/kisskb/buildresult/5869520/

Paul.

---