11:19 LiquidAcid: greeting, i updated my whole wine environment (on gentoo) yesterday, and realized that we no longer have wine-d3d9; so i migrated to wine-vanilla and the standalone gallium-nine -- however ninewinecfg segfaults, and i'm not sure where to begin looking
11:19 LiquidAcid: might be related to me using iris for this, or maybe it's also a wine problem?
11:20 LiquidAcid: this is with wine-5.4, and gallium-nine 0.5, everything inside a 32bit chroot
11:40 orbea: LiquidAcid: what compiler? just guessing but winecfg crashes with clang for me
11:40 orbea: you do have nine-wine-standalone, right?
11:41 LiquidAcid: orbea, gcc-9.2.0
11:41 LiquidAcid: the package on gentoo is called app-emulation/gallium-nine-standalone
11:42 LiquidAcid: and wine works, so i rule out a compiler problem
11:42 orbea: i looked, that is the same thing. Why doesn't gentoo just use the upstream name to avoid confusion...
11:43 orbea: so winecfg only crashes after enabling gallium nine and stops crashing when you disable it again?
11:44 orbea: i mean enabling/disabling with ninewinecfg/ninewinecfg64
11:44 LiquidAcid: so, winecfg does not crash
11:45 LiquidAcid: *no
11:45 orbea: oh sorry, I read wrong
11:45 LiquidAcid: it's ninewinecfg crashing
11:45 orbea: never seen it crash here...what does stdout say?
11:46 LiquidAcid: https://www.math.uni-bielefeld.de/~tjakobi/archive/backtrace.txt
11:46 LiquidAcid: that's the backtrace
11:46 LiquidAcid: and this is stdout:
11:46 LiquidAcid: wine: Unhandled page fault on read access to 00000010 at address 78F9B0A6 (thread 0009), starting debugger...
11:49 orbea: does it happen if you use the llvm pipe?
11:50 orbea: LIBGL_ALWAYS_SOFTWARE=true
11:50 LiquidAcid: same thing
11:51 orbea: so its probably not iris
11:51 orbea: i dont really know either, stick around and one of the actual devs have a better idea :)
11:52 LiquidAcid: launching this with LIBGL_DEBUG=verbose doesn't even show any dri driver being loaded
11:52 LiquidAcid: i doubt it has to do with the gfx stack itself
11:53 orbea: just to be sure, have you tried a fresh wine prefix?
11:54 LiquidAcid: just did, same thing
11:58 LiquidAcid: i'm wondering if the instructions are even correct, because it says i should be launching /usr/lib/ninewinecfg.exe.so
11:58 LiquidAcid: which is an ELF, how is that supposed work?
11:59 LiquidAcid: there is also a /usr/lib/ninewinecfg.exe.fake, which is a PE32, but launching that one does nothing
14:58 LiquidAcid: i did a bit of tracing, and i'm more and more convinced that the instructions are wrong
14:58 LiquidAcid: 75448.863:0008:0009:trace:module:load_dll L"\\??\\Z:\\usr\\lib\\ninewinecfg.exe.fake" is a fake Wine dll
14:59 LiquidAcid: that's when executing the exe.fake (the PE32)
14:59 LiquidAcid: but that fails a bit later:
14:59 LiquidAcid: 75448.863:0008:0009:warn:module:find_builtin_dll cannot find builtin library for L"ninewinecfg.exe.fake"
14:59 LiquidAcid: which results in:
14:59 LiquidAcid: 75448.863:0008:0009:err:module:__wine_process_init L"Z:\\usr\\lib\\ninewinecfg.exe.fake" not found
15:00 LiquidAcid: and then cleanup happens
15:30 LiquidAcid: ok, so i installed it manually, had to edit release.sh though (it assumes a x86-64 system)
15:30 LiquidAcid: wine ninewinecfg.exe -H <- works
15:30 LiquidAcid: wine ninewinecfg.exe <- segfaults
15:31 LiquidAcid: wine ninewinecfg.exe -e < works
15:31 LiquidAcid: wine ninewinecfg.exe -p <- segfaults
15:31 LiquidAcid: wtf?!
16:00 iive: LiquidAcid, ninewinecfg is executable, not library
16:28 LiquidAcid: iive, and?
17:01 iive: i'm puzzled that it is loaded as dynamic library (dll)
17:03 LiquidAcid: apparantly some internal wine magic
17:04 LiquidAcid: the thing is, the build done by portage works the same: wine /usr/lib/ninewinecfg.exe.so -H
17:04 LiquidAcid: works, but no arguments: segfault
17:04 LiquidAcid: probing, segfault
17:04 LiquidAcid: the backtrace is even (nearly) identical