**** BEGIN LOGGING AT Sun Jan 03 02:59:56 2021
Jan 03 05:39:14 <seebs> so someone found a lovely quirk, which is that FD_ZERO blows up under msan with glibc, because glibc is doing inline assembly for "zero out a small array", which is probably not even a good candidate for inline assembly as opposed to trusting the compiler to recognize a loop.
Jan 03 11:03:00 <user23413213> what kind of projects are you guys working on in which you use yocto, perhaps somebody likes to share a story to start 2021 off :)
Jan 03 11:26:55 <RP> seebs: sounds 'fun'
Jan 03 15:27:46 <seebs> turns out that glibc killed the x86 inline assembly implementation of FD_ZERO recently anyway, which is a good choice since it's ridiculous to hand-code "zero this loop" in a thing that's not even used all that often.
Jan 03 22:51:11 <dv> can't find omp.h while building rpm-native with gatesgarth https://pastebin.com/ZDMDB5P0
Jan 03 23:08:22 <RP> dv: that sounds familiar, can't remember the fix though :(
Jan 03 23:08:49 <dv> https://lists.yoctoproject.org/g/poky/message/12048
Jan 03 23:08:55 <RP> seebs: it does sound like an interesting choice of optimisation...
Jan 03 23:09:37 <RP> seebs: sometime this week we need to find out why creating packages which contain a fifo fails but only on arm hosts (pseudo segfault), not x86 ones :/
Jan 03 23:09:48 <dv> some people already faced that https://lists.yoctoproject.org/g/poky/message/12048
Jan 03 23:10:08 <RP> dv: ah, it did look familiar! :)
Jan 03 23:10:44 <RP> dv: sounds like its host gcc version?
Jan 03 23:11:10 <dv> RP, 8.4.0
Jan 03 23:19:02 <RP> dv: give buildtools-tarball a try?
Jan 03 23:19:20 <dv> what is it?
Jan 03 23:20:34 <dv> RP, I found the solution: installed libgomp-devel package and stopped on the different step: https://pastebin.com/X1XmuLv2
Jan 03 23:29:10 <RP> dv: looks related, "librpmbuild.so.9: undefined symbol: omp_get_max_threads" suggests it can't find libomp :/
Jan 03 23:33:41 <dv> RP, yes. thank you
**** ENDING LOGGING AT Mon Jan 04 02:59:57 2021