On Fri, Nov 19, 2010 at 10:26 AM, Rob Herring robherring2@gmail.com wrote:

Michael,

On 11/18/2010 03:08 PM, Michael Hope wrote:

...

On Fri, Nov 19, 2010 at 6:26 AM, Rob Herringrobherring2@gmail.com  wrote:

...

Doing an allmodconfig build on the kernel, I get the following:

CC      arch/arm/kernel/asm-offsets.s In file included from /home/rob/proj/git/linux-2.6-dt/include/linux/kernel.h:12,                 from /home/rob/proj/git/linux-2.6-dt/include/linux/sched.h:54,                 from /home/rob/proj/git/linux-2.6-dt/arch/arm/kernel/asm-offsets.c:13: /usr/lib/gcc/arm-linux-gnueabi/4.4.5/include/stdarg.h:40: internal compiler error: Segmentation fault

It occurs on Maverick 4.4, 4.5 and CodeSourcery 2009Q1 cross toolchains. It's confirmed by Codesourcery here:

http://www.codesourcery.com/archives/arm-gnu/msg03719.html

What's the status on this issue? I didn't see anything in Linaro gcc bugs that looks related.

Hi Rob.  This is LP: #623945 'ICE using va_list with -g -femit-struct-debug-baseonly' which was fixed in the recently released Linaro GCC 4.5-2010.11. See:  https://bugs.launchpad.net/gcc-linaro/+bug/623945 for more information.  Note that Linaro GCC 4.4 is in maintenance and doesn't include the fix.

Out of interest, where do you pick up the binary build of your toolchain from?  Self built?  From Ubuntu?  Some other way?  I'm wondering about the best way you can pick up this fix.

It is from ubuntu maverick. This was my next question. As Linaro 4.5 releases continue, does ubuntu pick them up as updates for maverick?

I'm not certain, but I don't think so. The Linaro changes are being picked up as we go along in Ubuntu Natty, but released versions are another story.

Marcin is working on binary packages, including backports to the Ubuntu Lucid and Maverick. You can follow his progress here: https://blueprints.launchpad.net/ubuntu/+spec/other-linaro-n-cross-compilers

Just knowing the work-around is good enough for me on this particular issue. Turning off CONFIG_DEBUG_INFO_REDUCED in the kernel turns off this option.

Good to know you have a work-around.

-- Michael

On Fri, Nov 19, 2010, Christian Robottom Reis wrote:

That's right -- I think justifying an Ubuntu SRU for a compiler is unlikely, though Steve L. or Loïc may have experience in cases where that did happen (I don't).

While a SRU would be possible if it's a not too large patch, I don't think there is interest in spending too much time in fixing 4.4 bugs when the focus moved to 4.5. We should put our efforts in providing backports, and that will make everybody happier :)