Re: Arch-specific LTO problems
On Tue, Aug 4, 2020 at 9:06 AM Jeff Law wrote: > On Tue, 2020-08-04 at 08:05 -0700, Tom Stellard wrote: > > I think this is also an OOM problem. I've seen similar error messages > > before when a compiler process gets killed while it is in the middle of > > piping assembly into the assembler. > Agreed. Okay, I'll check into that possibility, too. Thanks for the feedback, Tom and Jeff. -- Jerry James http://www.jamezone.org/ ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Re: Arch-specific LTO problems
On Tue, 2020-08-04 at 08:05 -0700, Tom Stellard wrote: > On 8/4/20 11:02 AM, Jerry James wrote: > > On Tue, Aug 4, 2020 at 12:33 AM Jeff Law wrote: > > > If we're blowing up memory on the builder, then we should probably > > > disable LTO on > > > the 32bit platforms. This is something that was expected, though I > > > didn't trip > > > over any in my i686 testing (I have a theory for why, but it's not > > > terribly > > > important). > > > > I will try to come up with a definitive answer today on whether memory > > exhaustion is or is not the problem. However, the ppc64le symbol > > problem with the primecount package appears to be something else > > entirely. See https://bugzilla.redhat.com/show_bug.cgi?id=1862181. > > > > I think this is also an OOM problem. I've seen similar error messages > before when a compiler process gets killed while it is in the middle of > piping assembly into the assembler. Agreed. jeff ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Re: Arch-specific LTO problems
On 8/4/20 11:02 AM, Jerry James wrote: On Tue, Aug 4, 2020 at 12:33 AM Jeff Law wrote: If we're blowing up memory on the builder, then we should probably disable LTO on the 32bit platforms. This is something that was expected, though I didn't trip over any in my i686 testing (I have a theory for why, but it's not terribly important). I will try to come up with a definitive answer today on whether memory exhaustion is or is not the problem. However, the ppc64le symbol problem with the primecount package appears to be something else entirely. See https://bugzilla.redhat.com/show_bug.cgi?id=1862181. I think this is also an OOM problem. I've seen similar error messages before when a compiler process gets killed while it is in the middle of piping assembly into the assembler. -Tom ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Re: Arch-specific LTO problems
On Tue, Aug 4, 2020 at 12:33 AM Jeff Law wrote: > If we're blowing up memory on the builder, then we should probably disable > LTO on > the 32bit platforms. This is something that was expected, though I didn't > trip > over any in my i686 testing (I have a theory for why, but it's not terribly > important). I will try to come up with a definitive answer today on whether memory exhaustion is or is not the problem. However, the ppc64le symbol problem with the primecount package appears to be something else entirely. See https://bugzilla.redhat.com/show_bug.cgi?id=1862181. -- Jerry James http://www.jamezone.org/ ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Re: Arch-specific LTO problems
On Mon, 2020-08-03 at 14:06 -0600, Jerry James wrote: > On Mon, Aug 3, 2020 at 11:33 AM Tom Stellard wrote: > > These are all likely caused by the linker running out of memory and > > getting killed by the OOM killer. > > I see. In that case, I'll try resubmitting each build once and see if > the same thing happens. Thanks, Tom. If we're blowing up memory on the builder, then we should probably disable LTO on the 32bit platforms. This is something that was expected, though I didn't trip over any in my i686 testing (I have a theory for why, but it's not terribly important). jeff > ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Re: Arch-specific LTO problems
On Mon, Aug 3, 2020 at 11:33 AM Tom Stellard wrote: > These are all likely caused by the linker running out of memory and > getting killed by the OOM killer. I see. In that case, I'll try resubmitting each build once and see if the same thing happens. Thanks, Tom. -- Jerry James http://www.jamezone.org/ ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Re: Arch-specific LTO problems
On 8/3/20 1:28 PM, Jerry James wrote: I've been going through mass rebuild failures. Several of these look like they might be LTO problems that only manifest on one architecture. libfplll: arm https://koji.fedoraproject.org/koji/buildinfo?buildID=1575475 Link step fails: /usr/bin/ld: error: lto-wrapper failed collect2: error: ld returned 1 exit status linbox: arm https://koji.fedoraproject.org/koji/buildinfo?buildID=1575610 Link step fails: /usr/bin/ld: error: lto-wrapper failed collect2: error: ld returned 1 exit status mathicgb: arm https://koji.fedoraproject.org/koji/buildinfo?buildID=1575673 Link step fails: /usr/bin/ld: error: lto-wrapper failed collect2: error: ld returned 1 exit status primecount: ppc64le https://koji.fedoraproject.org/koji/buildinfo?buildID=1576863 Bad assembly generated on ppc64le only when LTO is active. See https://bugzilla.redhat.com/show_bug.cgi?id=1862181 These are all likely caused by the linker running out of memory and getting killed by the OOM killer. -Tom ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
Arch-specific LTO problems
I've been going through mass rebuild failures. Several of these look like they might be LTO problems that only manifest on one architecture. libfplll: arm https://koji.fedoraproject.org/koji/buildinfo?buildID=1575475 Link step fails: /usr/bin/ld: error: lto-wrapper failed collect2: error: ld returned 1 exit status linbox: arm https://koji.fedoraproject.org/koji/buildinfo?buildID=1575610 Link step fails: /usr/bin/ld: error: lto-wrapper failed collect2: error: ld returned 1 exit status mathicgb: arm https://koji.fedoraproject.org/koji/buildinfo?buildID=1575673 Link step fails: /usr/bin/ld: error: lto-wrapper failed collect2: error: ld returned 1 exit status primecount: ppc64le https://koji.fedoraproject.org/koji/buildinfo?buildID=1576863 Bad assembly generated on ppc64le only when LTO is active. See https://bugzilla.redhat.com/show_bug.cgi?id=1862181 -- Jerry James http://www.jamezone.org/ ___ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org