Le jeudi 11 juin 2009 18:10:24, Bryan Kadzban a écrit : > Bruce Dubbs wrote: > > Bryan Kadzban wrote: > >> Bryan Kadzban wrote: > >>> Poking around in glibc, this *looks* like a bug in it (at least in > >>> 2.10.1). > >> > >> Filed this bug: > >> > >> http://sources.redhat.com/bugzilla/show_bug.cgi?id=10262 > >> > >> Does the patch mentioned there (to glibc) fix the testsuite? It needs > >> to be done whether it fixes the tests or not, though (for 32-bit > >> builds)... > > Thanks for the patch, Bryan. I was about to start a new build, but have tried your patch before. It fixes this futex looping bug (for this 32-bit build). Now, no processes are left looping after the tests :) There are some other tests, still failing: stdio-common/bug22.out posix/wordexp-test.out c++-types-check.out and also the posix/annexc.out I'm going to re-build using the new 2.6.30 kernel, atm :) > > Wow. That's really crawling around the innards Bryan. > > Heh. And I don't even know for sure if this was the issue causing the > testsuite failures. :-P > > It does really look like a bug, though. > > (Re: "crawling around the innards": That's what happens when my most > recent coding was trying to get signalfd to work on a really old glibc > but a fairly new kernel. :-) ) > > > My observation to their responsiveness to outside bugs is that they will > > probably ignore it. Let's hope I'm wrong. > > Somebody from SuSE seems to have commented. *crossing fingers*... > > It seems fairly obvious from comparing 32-bit and 64-bit glibc what's wrong > with the constants being used, as long as someone actually does that. > > (I'm just hoping it doesn't get closed as a duplicate of bug 333: "build > failures aren't to be reported here".)
-- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
