Yes, I reported to Christophe yesterday, he confirmed this comes from his linker commit and he's looking at it. Laurent

On Thu, 12 Sept 2024 at 04:38, Maxim Kuvyrkov maxim.kuvyrkov@linaro.org wrote:

Hi Mark,

I believe this was caused by Christophe's binutils patch, which has now been fixed.

Christophe, please correct me if I'm wrong.

Thanks!

-- Maxim Kuvyrkov https://www.linaro.org

...

On Sep 12, 2024, at 00:34, Mark Wielaard mark@klomp.org wrote:

Hi,

ci_notify@linaro.org is sending test results to gcc-restresults@gcc.gnu.org but recently starting to fail so many tests that they don't get accepted anymore:

From: ci_notify@linaro.org on Tue Sep 10 02:17:32 2024 Subject: Results for 15.0.0 20240906 (experimental) [master revision gcc-15-3512-ge8378231bb8] (GCC) testsuite on arm-unknown-eabi Cause: Message body is too big: 6548864 bytes with a limit of 1000 KB

From: ci_notify@linaro.org on Tue Sep 10 05:28:38 2024 Subject: Results for 15.0.0 20240906 (experimental) [master revision gcc-15-3512-ge8378231bb88] (GCC) testsuite on arm-unknown-eabi Cause: Message body is too big: 6559111 bytes with a limit of 1000 KB

From: ci_notify@linaro.org on Tue Sep 10 10:55:10 2024 Subject: Results for 15.0.0 20240906 (experimental) [master revision gcc-15-3512-ge8378231bb88] (GCC) testsuite on arm-unknown-eabi Cause: Message body is too big: 6548861 bytes with a limit of 1000 KB

So that is 6.5MB of results. Almost all look like:

FAIL: 17_intro/freestanding.cc -std=gnu++17 (test for excess errors) UNRESOLVED: 17_intro/freestanding.cc -std=gnu++17 compilation failed to produce executable

Could someone take a look? It seems that the compiler is just crashing on every test.

Thanks,

Mark _______________________________________________ linaro-toolchain mailing list -- linaro-toolchain@lists.linaro.org To unsubscribe send an email to linaro-toolchain-leave@lists.linaro.org

---

linaro-toolchain mailing list -- linaro-toolchain@lists.linaro.org To unsubscribe send an email to linaro-toolchain-leave@lists.linaro.org

On Thu, Sep 12, 2024 at 09:25:08AM +0200, Christophe Lyon wrote:

Right, sorry for the breakage. This should now be fixed by https://sourceware.org/pipermail/binutils/2024-September/136743.html

Thanks, the latest testresults look a lot more reasonable.

Cheers,

Mark

On Fri, 20 Sept 2024 at 14:47, Mark Wielaard mark@klomp.org wrote:

Hi,

I thought this was resolved, but...

me too :-)

On Thu, 2024-09-12 at 21:47 +0200, Mark Wielaard wrote:

...

On Thu, Sep 12, 2024 at 09:25:08AM +0200, Christophe Lyon wrote:

...

Right, sorry for the breakage. This should now be fixed by https://sourceware.org/pipermail/binutils/2024-September/136743.html

Thanks, the latest testresults look a lot more reasonable.

There are still some results coming in that look totally broken. Last one was:

Subject: Results for 15.0.0 20240910 (experimental) [master revision gcc-15-3571-ge783a4a68376] (GCC) testsuite on arm-unknown-eabi Size: 6565846 bytes Reason: Message body is too big: 6560798 bytes with a limit of 1000 KB Received: Fri Sep 20 11:57:18 2024

# From https://ci.linaro.org/job/tcwg_gnu_embed_check_gcc--master-thumb_m55_hard_ea... : LAST_UPDATED: 2024-09-20T10:14:36+00:00 (master revision gcc-15-3571- ge783a4a68376) arm-eabi {-mthumb/-march=armv8.1-m.main+mve.fp+fp.dp/- mtune=cortex-m55/-mfloat-abi=hard/-mfpu=auto}

Could someone take a look?

Looks like our build queue is full (well more than full ~60 builds of this type pending), bisecting regressions introduced before I pushed my fix in the linker.

I think that it should be OK within a few more days. Is that causing problems on your side?

Thanks,

Christophe

Thanks,

Mark

On Fri, 20 Sept 2024 at 15:34, Mark Wielaard mark@klomp.org wrote:

Hi Christophe,

On Fri, 2024-09-20 at 15:30 +0200, Christophe Lyon wrote:

...

Looks like our build queue is full (well more than full ~60 builds of this type pending), bisecting regressions introduced before I pushed my fix in the linker.

I think that it should be OK within a few more days. Is that causing problems on your side?

That should be fine, but the messages to gcc-testresults will be blocked because they are 6MB each (and don't look very useful).

Sure, I think you can discard them safely.

Christophe

Thanks,

Mark