Arch Linux

I tried to rebuild wine package with makepkg but I get the same bug,

I notice that on an old PC ( pentium 4 2.4 Ghz ) I don't have the bug ( i686 version of wine ),

my configuration :

archlinux 64 bits
cpu pentium dual core E6800 3.33 Ghz
ati radeon HD4650 PCIe ( radeon open source driver )
4 Gb ram

the backtrace with winedbg if I execute "notepad" :

Unhandled exception: assertion failed in 32-bit code (0xf7722cf0).
Register dump:
CS:0023 SS:002b DS:002b ES:002b FS:0063 GS:006b
EIP:f7722cf0 ESP:0033e274 EBP:0033e538 EFLAGS:00000296( - -- I S -A-P- )
EAX:00000000 EBX:00000639 ECX:00000639 EDX:00000006
ESI:00000049 EDI:f7522000
Stack dump:
0x0033e274: 0033e538 00000006 00000639 f7399f37
0x0033e284: f7522000 0033e324 f739b579 00000006
0x0033e294: 0033e2a4 00000000 3a303020 30203030
0x0033e2a4: 00000020 00000000 00000000 00000000
0x0033e2b4: 00000000 00000000 00000000 00000000
0x0033e2c4: 00000000 00000000 00000000 00000000
000c: sel=0067 base=00000000 limit=00000000 16-bit r-x
Backtrace:
=>0 0xf7722cf0 __kernel_vsyscall+0x10() in [vdso].so (0x0033e538)
1 0xf7399f37 gsignal+0x46() in libc.so.6 (0x0033e538)
2 0xf739b579 abort+0x148() in libc.so.6 (0x0033e538)
3 0xf73d9cde __libc_message+0x28d() in libc.so.6 (0x0033e538)
4 0xf73df927 malloc_printerr+0xa6() in libc.so.6 (0xf7522840)
5 0xf73e010d _int_free+0x10c() in libc.so.6 (0xf7522840)
6 0x7df4a467 in libxcursor.so.1 (+0x6466) (0x7cc9a7b0)
7 0x7df4a61d XcursorLibraryLoadImages+0x7c() in libxcursor.so.1 (0x00000015)
8 0x7df4a6cc XcursorLibraryLoadCursor+0x3b() in libxcursor.so.1 (0x00000015)
9 0x7df4b124 XcursorTryShapeBitmapCursor+0x103() in libxcursor.so.1 (0x7df4b6bd)
10 0x7dfb5c24 _XTryShapeBitmapCursor+0xc3() in libx11.so.6 (0x0033e800)
11 0x7dfb516e XCreatePixmapCursor+0x4d() in libx11.so.6 (0x0033e800)
12 0x7e14b8f9 in winex11 (+0x2b8f8) (0x0033e828)
13 0x7e14d645 in winex11 (+0x2d644) (0x0033e868)
14 0x7e14d71a in winex11 (+0x2d719) (0x0033e938)
15 0x7e1619d4 in winex11 (+0x419d3) (0x0033f1e8)
16 0x7e1628c1 X11DRV_WindowPosChanging+0x3a0() in winex11 (0x0033f268)
17 0x7e9616f8 in user32 (+0x916f7) (0x0033f3b8)
18 0x7e95d602 in user32 (+0x8d601) (0x0033f578)
19 0x7e957025 CreateWindowExW+0x84() in user32 (0x0033f5f8)
20 0x7edb6d4a WinMain+0xaf9() in notepad (0x0033fd38)
21 0x7edb1ba4 main+0xd3() in notepad (0x0033fdd8)
22 0x7edb755f in notepad (+0x755e) (0x0033fe28)
23 0x7b85c64c call_process_entry+0xb() in kernel32 (0x0033fe48)
24 0x7b85d74a in kernel32 (+0x4d749) (0x0033fe78)
25 0x7bc7c640 call_thread_func_wrapper+0xb() in ntdll (0x0033fe98)
26 0x7bc7f6bf call_thread_func+0xde() in ntdll (0x0033ff88)
27 0x7bc7c61e RtlRaiseException+0x21() in ntdll (0x0033ffb8)
28 0x7bc51451 call_dll_entry_point+0x440() in ntdll (0x0033ffd8)
29 0xf754ba8d wine_call_on_stack+0x1c() in libwine.so.1 (0x00000000)
30 0xf754bc0a wine_switch_to_stack+0x29() in libwine.so.1 (0xffc7d0c8)
31 0x7bc56c05 LdrInitializeThunk+0x1f4() in ntdll (0xffc7d118)
32 0x7b8636f2 __wine_kernel_init+0x871() in kernel32 (0xffc7e288)
33 0x7bc57a63 __wine_process_init+0x162() in ntdll (0xffc7e2f8)
34 0xf7549643 wine_init+0x2f2() in libwine.so.1 (0xffc7e368)
35 0x7bf00db2 main+0x81() in <wine-loader> (0xffc7e7c8)
36 0xf7385e5e __libc_start_main+0xdd() in libc.so.6 (0x00000000)
0xf7722cf0 __kernel_vsyscall+0x10 in [vdso].so: popl %ebp

I want to know if I am alone with this bug,

and if the archlinux developpers who have been assigned to this bug can reproduce this bug ?

maybe my archlinux system is broken ?

I use wine daily and have not run into this. You need to provide a more concrete test case. Does it fail on every Windows program? If you compile a hello world with mingw32 and run it, does that fail, too? Are you using win32 WINEARCH?

I notice If I run "win32 console program" with wine ( like an "hello world", a windows program without GUI ) then I have no bugs,

It fails on every windows program who have a GUI,

my ~/.wine is a 64 bits installation,

I tried to delete ~/.wine, but the bug is still here with a new ~/.wine

If I run this command : "WINEARCH=win32 winecfg" I get the same crash,

I tried to downgrade to wine 1.7.24 but it's the same crash,

could it be a problem with glibc 2.20 ?

I tried also to create a new user ( in order to have a new /home ) but I get the same bug with wine,

it's very frustrating, because 2 months ago I could run wine without problems

Could it be a problem with ldconf cache file ? ( /etc/ld.so.cache )

Frankly I've never seen this and never had a user who couldn't run any program with wine at all. If you don't give me any vectors, there's nothing for me to look at.

I can run programs with wine, but only "windows console mode programs" can run with wine, this is strange,

I tried to reinstall all packages with this command :

# pacman -Qenq | pacman -S -

but it didn't solve my problem

wine is the only package who has problem with my PC, but in the past I have no bugs with wine, I don't know why I have this problem now

I found the solution,

it was a strange bug in X server related to icon theme cursor,

in /usr/share/icons/default/index.theme I have set a custom icon theme for the X server :

[Icon Theme]
Inherits=Oxygen_White

when I saw the word "libXcursor" in the backtrace it gives me an idea : to avoid my setting and set the original icon theme :

[Icon Theme]
Inherits=Adwaita

and now the bug is gone, wine runs without problems !

it's very strange that a valid icon theme ( Oxygen_White ) can trigger such a severe bug