#asahi-dev · 2023-03-13 10:37
marcanI think this is another gdb bug...#asahi-dev · 2023-03-13 10:32
marcansigh#asahi-dev · 2023-03-13 10:32
marcanbut I just can't get the gdb watchpoint stuff to not suck#asahi-dev · 2023-03-13 10:32
marcanwhat bugs me here is this *could* be a random memory corruption issue#asahi-dev · 2023-03-13 10:31
marcandoes not repro on amd64 :/#asahi-dev · 2023-03-13 10:22
marcanlet me see if I can repro this on a random Arch amd64 system#asahi-dev · 2023-03-13 10:21
marcan(though it might be a Qt issue)#asahi-dev · 2023-03-13 10:21
marcanI'll file a KDE bug#asahi-dev · 2023-03-13 10:21
marcanthis is just broken everywhere depending on when GC runs#asahi-dev · 2023-03-13 10:21
marcanok yeah, repro'd on M1 Pro#asahi-dev · 2023-03-13 10:16
marcanjannau: consistently?#asahi-dev · 2023-03-13 10:16
marcanI might just punt this to upstream#asahi-dev · 2023-03-13 10:16
marcanand the bad value is in the JS stack#asahi-dev · 2023-03-13 10:14
marcan(and software is too slow here...)#asahi-dev · 2023-03-13 10:13
marcansigh, and now hardware watchpoints don't work properly?#asahi-dev · 2023-03-13 10:06
marcanthis is some messed up javascript type confusion thing or something...#asahi-dev · 2023-03-13 10:06
marcanjannau: QV4_MM_AGGRESSIVE_GC=1 *may* help repro the issue#asahi-dev · 2023-03-13 09:15
marcan... and there is a _TLS_MODULE_BASE symbol#asahi-dev · 2023-03-13 09:15
marcanwhich is things that start with _T#asahi-dev · 2023-03-13 09:14
marcan... which should only happen for things that look like Swift symbols??#asahi-dev · 2023-03-13 09:13
marcanand this all happens because it tries to dlsym... swift_demangle??#asahi-dev · 2023-03-13 09:08
marcanwhich is illegal#asahi-dev · 2023-03-13 09:08
marcanthe asan hooks end up with it doing a rlock -> (asan) -> wlock -> wunlock -> (out of asan) -> runlock sequence I think#asahi-dev · 2023-03-13 09:08
marcanokay no it's just asan#asahi-dev · 2023-03-13 09:04
marcanwait, not sure#asahi-dev · 2023-03-13 09:03
marcangettext bug?#asahi-dev · 2023-03-13 09:03
marcannow it's starting to look like someone unlocks an already unlocked lock#asahi-dev · 2023-03-13 08:47
marcangood question#asahi-dev · 2023-03-13 08:45
marcanthis is going great.#asahi-dev · 2023-03-13 08:45
marcan... and I just realized ALARM is shipping glibc 2.35 while their PKGBUILD is at 2.37-2#asahi-dev · 2023-03-13 08:34
marcanis glibc broken now too?#asahi-dev · 2023-03-13 08:34
marcanok wtaf this is a pthread mutex under the hood and as far as I can tell it is unlocked and locking it deadlocks#asahi-dev · 2023-03-13 08:21
marcanthis is getting ridiculous#asahi-dev · 2023-03-13 08:21
marcanmaybe, or maybe gnulib locks are broken on M1 because reasons#asahi-dev · 2023-03-13 08:17
marcanhow the hell is gettext deadlocking, who is holding this lock?#asahi-dev · 2023-03-13 08:14
marcanif it's a race it's probably going to depend on nonsense like how big my taskbar is and what widgets I have on it#asahi-dev · 2023-03-13 08:12
marcanwhy is everything so broken#asahi-dev · 2023-03-13 08:12
marcanand gettext#asahi-dev · 2023-03-13 08:11
marcansomehow pipewire is involved in the deadlock with asan#asahi-dev · 2023-03-13 08:11
marcanand I think it is a race of some sort#asahi-dev · 2023-03-13 08:11
marcanthis is alarm#asahi-dev · 2023-03-13 08:07
marcansigh... valgrind doesn't repro, libasan just deadlocks (??)#asahi-dev · 2023-03-13 07:58
marcanmaybe it only happens without plasmashell running (which is the state I'm in right now)#asahi-dev · 2023-03-13 07:52
marcanit's not consistent#asahi-dev · 2023-03-13 07:52
marcanit does that on random surfaces for me, including title bars sometimes#asahi-dev · 2023-03-13 07:50
marcannah, the pink boxes are probably the gpu driver still missing sync#asahi-dev · 2023-03-13 07:49
marcanif this is another page size thing...#asahi-dev · 2023-03-13 07:49
marcanthis is some... heap memory manager... collecting stuff from the javascript stack?#asahi-dev · 2023-03-13 07:49
marcanthis=0x3f sigh#asahi-dev · 2023-03-13 07:48
marcannow cursor redraws leave garbage behind