On 19 November 2012 07:20, Brendan Jones <brendan.jones.it@xxxxxxxxx> wrote: > On 11/17/2012 09:42 AM, Ian Malone wrote: >> >> On 17 November 2012 03:09, Orcan Ogetbil <oget.fedora@xxxxxxxxx> wrote: >>> >>> On Fri, Nov 16, 2012 at 5:24 PM, Ian Malone wrote: >>>> >>>> On 15 November 2012 23:20, Ian Malone wrote: >>>>> >>>>> On 15 November 2012 04:40, Orcan Ogetbil wrote: >>>>>> >>>>>> Hi all, >>>>>> >>>>>> A few months ago earlier in the F-17 release cycle, we had a problem >>>>>> with jackd turning verbose, flooding the stdout, even when its >>>>>> verbosity was turned off. This was a bug [1] in the compiler, >>>>>> specifically in the optimized builds, for which we filed a bug against >>>>>> upstream gcc [2]. As a workaround we rebuilt jack unoptimized, and we >>>>>> have been using the unoptimized build since. >>>>>> >>>>>> Late in September the issue got resolved in upstream gcc (see [2]). >>>> >>>> >>>>>> [1] https://bugzilla.redhat.com/show_bug.cgi?id=827748 >>>>>> [2] http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53663 >>>>>> [3] http://koji.fedoraproject.org/koji/buildinfo?buildID=362840 >>>>>> [4] http://oget.fedorapeople.org/jack/ >>>> >>>> >>>> >>>> Further the test in >>>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53663#c3 >>>> still fails on gcc-4.7.2-8.fc18.x86_64 for me (at -O1). >>>> >>>> You weren't running a patched gcc or something? >>>> >>> >>> No, it is the official gcc-4.7.2-2.fc17. You can see tho root.log in >>> the above link [4] for my buildroot. >>> >> >> One build is -9.1 and the other is -10, jack-realtime-compat.patch >> differs between the two. If its a compiler optimisation problem then >> the difference in initial variables may be causing a difference in >> behaviour? Just random speculation, on the basis that if the build >> systems are the same then something else must be the cause. Preferable >> to get the gcc fix though. >> Yes verbose for 1.9.8-12, no verbose for 1.9.8-11. For F17 I tried swapping back to the previous realtime-compat.patch and rebuilding 1.9.8-10 and it doesn't do verbose, so I think that does explain what Orcan was seeing, though of course not actually a fix. Maybe need to go back to a -O0 build before F18 release if the gcc fix isn't going to be applied in time. -- imalone http://ibmalone.blogspot.co.uk -- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
