- Wine-GitLab - list.winehq.org

Re: [PATCH v4 0/1] MR9423: ntdll: Warn when environment block is too large, instead of removing it
by Martino Fontana (＠tinozzo123) Nov. 13, 2025

Nov. 13, 2025

I'm aware that this is a lose-lose scenario, due to EA not fixing their issue on the end, since it's either: - Workaround the issue, but cause breaking changes outside of it - Do nothing, but EA app remains not working without manual intervention And IMO, the second choice is better. And many things can still be done, like having better documentation about the issue. For instance, Wine's wiki has a [Known Issues page](https://gitlab.winehq.org/wine/wine/-/wikis/Known-Issues), where this definitely belongs. Or maybe it could be added to the Arch Wiki as well. Or [make changes to systems that set huge environment variables to be less wasteful](https://github.com/NixOS/nixpkgs/pull/451031) (which is nice even outside of this). (And if there is still intention of doing a workaround for the issue... would it be possible to unset environment variables only if the executable `EADesktop.exe` is launched, to minimize the impact of breaking changes elsewhere?) -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9423#note_121852

1 0

[PATCH 0/3] MR9427: combobox: Adjust MEASUREITEMSTRUCT.itemHeight by 2 instead of 6.
by Paul Gofman (＠gofman) Nov. 13, 2025

Nov. 13, 2025

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=58932 -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9427

3 6

Re: [PATCH v4 0/1] MR9423: ntdll: Warn when environment block is too large, instead of removing it
by Alexandre Julliard (＠julliard) Nov. 13, 2025

Nov. 13, 2025

> It's a change that breaks configurations for little gain Obviously the gain is that a commonly-used app no longer crashes. How would you suggest to achieve that otherwise? Printing a warning may make it easier to determine the cause of the crash, but it doesn't improve the user's experience. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9423#note_121847

1 0

Re: [PATCH v4 0/1] MR9423: ntdll: Warn when environment block is too large, instead of removing it
by Martino Fontana (＠tinozzo123) Nov. 13, 2025

Nov. 13, 2025

Your change would be nice even with the warning I'm proposing, considering that it helps with trimming the environment block. I don't understand why it should still not import the environment block at all. I'll reiterate why I think conditionally not importing it is a bad idea: - If modern Windows allows unlimited environment block size, Wine should allow it too - It's a change that breaks configurations for little gain - Because it's conditional, documentation becomes much more cumbersome. Think of all the resources online saying something like "set this environment variable", which will now need to be updated with "unless your environment block is too big, you'll also have to trim the environment block too" (which is more complicated than this sentence). And that's assuming it gets updated at all. Add in users why follow guides and then don't notice the log (either because they don't use the terminal or because they are newbies and can't see that this log is related to why the guide is not working), and you get a mess. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9423#note_121846

1 0

[PATCH v3 0/5] MR9394: winex11: Lock and flush display in X11DRV_ChangeDisplaySettings.
by Rémi Bernon Nov. 13, 2025

Nov. 13, 2025

-- v3: winex11: Lock and flush display in X11DRV_ChangeDisplaySettings. winex11: Retrieve full modes before applying new display settings. winex11: Introduce a x11drv_mode structure for full modes. winex11: Always return full modes from settings handlers. winex11: Copy XF86VidModeModeInfo to the DEVMODEW driver data. https://gitlab.winehq.org/wine/wine/-/merge_requests/9394

2 9

Re: [PATCH v4 0/1] MR9423: ntdll: Warn when environment block is too large, instead of removing it
by Alexandre Julliard (＠julliard) Nov. 13, 2025

Nov. 13, 2025

I'd suggest something like !9450. You'd then be able to set WINE_HOST_XDG_DATA_DIRS to hide the distro value and keep the environment to a reasonable size. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9423#note_121838

1 0

[PATCH 0/3] MR9450: Don't replace WINE_HOST_ variables if they already exist in the environment.
by Alexandre Julliard (＠julliard) Nov. 13, 2025

Nov. 13, 2025

This makes it possible to override the host value seen by Wine of e.g. XDG_DATA_DIRS by setting WINE_HOST_XDG_DATA_DIRS directly. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9450

2 3

[PATCH 0/1] MR9439: mf: Return E_NOINTERFACE if service is missing
by Aric Stewart (＠aricstewart) Nov. 13, 2025

Nov. 13, 2025

Signed-off-by: Aric Stewart <aric(a)codeweavers.com> -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9439

3 4

[PATCH v3 0/8] MR9438: dsound: Various
by Matteo Bruni (＠Mystral) Nov. 13, 2025

Nov. 13, 2025

Some small dsound fixes I came up with while looking into Wine bug 30639 -- v3: dsound/tests: Mark weird DescriptionW propset failure on the AMD testbot machine. dsound: Return the mmdevapi endpoint ID as module. dsound: Fill all the DSPROPERTY_DIRECTSOUNDDEVICE_ENUMERATE data. dsound: Check data size in propset methods. dsound: Consistently trace locked byte count as an unsigned value. dsound: Print time as an unsigned value. dsound: Simplify check in DSOUND_RecalcFormat(). dsound/tests: Disable a crashing test. https://gitlab.winehq.org/wine/wine/-/merge_requests/9438

3 10

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Also, as Huw mentioned above, it looks like you still didn't fully address the comments from @nsivov that were pointed out [here](https://www.winehq.org/mailman3/hyperkitty/list/wine-devel(a)winehq.o…. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121819

1 0

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Sven Baars (@sbaars) commented about dlls/shell32/shlview_cmenu.c: > + > + if (!SUCCEEDED(IShellLinkW_QueryInterface(shelllink, &IID_IPersistFile, (LPVOID*)&persistfile))) > + { > + ERR("couldn't get IPersistFile interface\n"); > + IShellLinkW_Release(shelllink); > + return; > + } > + > + do { > + WCHAR *link_filename; > + > + if (counter == 1) > + { > + static const WCHAR *fmt = L"%s - %s.lnk"; > + int length = wcslen(filename) + wcslen(shortcutW) + 7; /* length of the format*/ > + link_filename = malloc((length + 1) * sizeof(WCHAR)); Should you check if malloc returns null? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121810

1 0

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Sven Baars (@sbaars) commented about dlls/shell32/shlview_cmenu.c: > + WCHAR *link_filename; > + > + if (counter == 1) > + { > + static const WCHAR *fmt = L"%s - %s.lnk"; > + int length = wcslen(filename) + wcslen(shortcutW) + 7; /* length of the format*/ > + link_filename = malloc((length + 1) * sizeof(WCHAR)); > + wsprintfW(link_filename, fmt, filename, shortcutW); > + } > + else > + { > + static const WCHAR *fmt = L"%s - %s (%s).lnk"; > + int length = wcslen(filename) + wcslen(shortcutW) + 10; /* length of the format*/ > + WCHAR counter_string[34]; > + wsprintfW(counter_string, L"%i", counter); > + length += wcslen(counter_string); `wsprintfW` also returns the length of the string by the way. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121809

1 0

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Sven Baars (@sbaars) commented about dlls/shell32/shlview_cmenu.c: > + { > + ERR("SetIDList failed\n"); > + return; > + } > + _ILSimpleGetTextW(This->apidl[0], (LPVOID)filename, MAX_PATH); > + length = wcslen(filename) + wcslen(shortcutW); > + if (length > MAX_PATH) > + { > + ERR("Path maximum length exceeded (%i>%i)\n", length, MAX_PATH); > + IShellLinkW_Release(shelllink); > + return; > + } > + > + SHGetPathFromIDListW(This->pidl, root); > + > + if (!SUCCEEDED(IShellLinkW_QueryInterface(shelllink, &IID_IPersistFile, (LPVOID*)&persistfile))) In these failure cases, it's probably possible to write some tests to see what `HRESULT` should be returned. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121808

1 0

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Sven Baars (@sbaars) commented about dlls/shell32/shlview_cmenu.c: > + do { > + WCHAR *link_filename; > + > + if (counter == 1) > + { > + static const WCHAR *fmt = L"%s - %s.lnk"; > + int length = wcslen(filename) + wcslen(shortcutW) + 7; /* length of the format*/ > + link_filename = malloc((length + 1) * sizeof(WCHAR)); > + wsprintfW(link_filename, fmt, filename, shortcutW); > + } > + else > + { > + static const WCHAR *fmt = L"%s - %s (%s).lnk"; > + int length = wcslen(filename) + wcslen(shortcutW) + 10; /* length of the format*/ > + WCHAR counter_string[34]; > + wsprintfW(counter_string, L"%i", counter); I think we can just allocate a sufficient amount for `link_filename` and add the counter directly in the `wsprintfW` call below. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121805

1 0

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Sven Baars (@sbaars) commented about dlls/shell32/shlview_cmenu.c: > + ILFree(full_pidl); > + if (FAILED(hr)) > + { > + ERR("SetIDList failed\n"); > + return; > + } > + _ILSimpleGetTextW(This->apidl[0], (LPVOID)filename, MAX_PATH); > + length = wcslen(filename) + wcslen(shortcutW); > + if (length > MAX_PATH) > + { > + ERR("Path maximum length exceeded (%i>%i)\n", length, MAX_PATH); > + IShellLinkW_Release(shelllink); > + return; > + } > + > + SHGetPathFromIDListW(This->pidl, root); In other places, we check the return value of this function. Not sure if we should do that here as well. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121807

1 0

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Sven Baars (@sbaars) commented about dlls/shell32/shlview_cmenu.c: > + ITEMIDLIST *full_pidl; > + HRESULT hr; > + int length; > + int counter = 1; > + > + static const WCHAR *shortcutW = L"Shortcut"; > + > + if (FAILED(IShellLink_Constructor(NULL, &IID_IShellLinkW, (LPVOID*)&shelllink))) > + { > + ERR("couldn't create ShellLink object\n"); > + return; > + } > + full_pidl = ILCombine(This->pidl, This->apidl[0]); > + if (!full_pidl) > + { > + ERR("Out of Memory\n"); Why don't we just return an `HRESULT` here and return that from `ItemMenu_InvokeCommand`, instead of the verbose ERRs? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121806

1 0

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Sven Baars (@sbaars) commented about dlls/shell32/shlview_cmenu.c: > + return; > + } > + > + PathCombineW(lnkfile, root, link_filename); > + free(link_filename); > + counter++; > + > + if (PathFileExistsW(lnkfile)) > + { > + hr = HRESULT_FROM_WIN32(ERROR_FILE_EXISTS); > + } > + else > + { > + ERR("attempting to write link to %s\n", wine_dbgstr_w(lnkfile)); > + hr = IPersistFile_Save(persistfile, lnkfile, FALSE); > + ERR("ARIC %#lx\n",hr); Debug statements -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121804

1 0

Re: [PATCH v8 0/1] MR5373: shell32: Implement FCIDM_SHVIEW_CREATELINK
by Sven Baars (＠sbaars) Nov. 13, 2025

Nov. 13, 2025

Sven Baars (@sbaars) commented about dlls/shell32/shlview_cmenu.c: > + full_pidl = ILCombine(This->pidl, This->apidl[0]); > + if (!full_pidl) > + { > + ERR("Out of Memory\n"); > + return; > + } > + > + hr = IShellLinkW_SetIDList(shelllink, full_pidl); > + ILFree(full_pidl); > + if (FAILED(hr)) > + { > + ERR("SetIDList failed\n"); > + return; > + } > + _ILSimpleGetTextW(This->apidl[0], (LPVOID)filename, MAX_PATH); > + length = wcslen(filename) + wcslen(shortcutW); This looks like a duplicate check. There is another one in the loop below. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/5373#note_121803

1 0

[PATCH 0/4] MR9429: Avoid a few compiler warnings.
by Bernhard Übelacker Nov. 13, 2025

Nov. 13, 2025

Gitlab CI uses `--enable-werror` for its builds, but to improve backtraces with my recent address sanitizer experiments I am using these flags `-g -O1 -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -fno-optimize-sibling-calls -fno-inline-functions` with both, gcc at the unix side and llvm-mingw at the pe side. Therefore these warnings would turn into errors. This four patches are helping gcc at the unix side. Would these patches be acceptable? Should they be separate merge requests? <details> <summary>Details</summary> ``` ccache gcc -m64 -c -o tools/winebuild/utils.o .../wine/tools/winebuild/utils.c \ -Itools/winebuild -I.../wine/tools/winebuild -Iinclude \ -I.../wine/include -D__WINESRC__ -DWINE_UNIX_LIB -Wall -pipe \ -fcf-protection=none -fvisibility=hidden -fno-stack-protector -fno-strict-aliasing \ -Wdeclaration-after-statement -Wempty-body -Wignored-qualifiers -Winit-self -Wlogical-op \ -Wno-packed-not-aligned -Wpointer-arith -Wshift-overflow=2 -Wstrict-prototypes -Wtype-limits \ -Wunused-but-set-parameter -Wvla -Wwrite-strings -Wno-misleading-indentation -gdwarf-4 -Werror \ -g -O1 -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -fno-optimize-sibling-calls -fno-inline-functions .../wine/tools/winebuild/utils.c: In function 'find_tool': .../wine/tools/winebuild/utils.c:230:8: error: 'file' may be used uninitialized [-Werror=maybe-uninitialized] 230 | if (!file && cc_command.count) file = find_clang_tool( cc_command, name ); | ^ .../wine/tools/winebuild/utils.c:214:17: note: 'file' was declared here 214 | const char *file; | ^~~~ cc1: all warnings being treated as errors make: *** [Makefile:610992: tools/winebuild/utils.o] Error 1 ccache gcc -m64 -c -o dlls/ntdll/unix/file.o .../wine/dlls/ntdll/unix/file.c -Idlls/ntdll \ -I.../wine/dlls/ntdll -Iinclude \ -I.../wine/include -D__WINESRC__ -DLTC_NO_PROTOTYPES \ -DLTC_SOURCE -D_NTSYSTEM_ -D_ACRTIMP= -DWINBASEAPI= -DWINE_UNIX_LIB \ -I/usr/include/x86_64-linux-gnu -Wall -pipe -fcf-protection=none -fvisibility=hidden \ -fno-stack-protector -fno-strict-aliasing -Wdeclaration-after-statement -Wempty-body \ -Wignored-qualifiers -Winit-self -Wlogical-op -Wno-packed-not-aligned -Wpointer-arith \ -Wshift-overflow=2 -Wstrict-prototypes -Wtype-limits -Wunused-but-set-parameter -Wvla \ -Wwrite-strings -Wno-misleading-indentation -gdwarf-4 -Werror -fPIC -fasynchronous-unwind-tables \ -g -O1 -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -fno-optimize-sibling-calls -fno-inline-functions .../wine/dlls/ntdll/unix/file.c: In function 'NtReadFile': .../wine/dlls/ntdll/unix/file.c:5758:23: error: 'avail_mode' may be used uninitialized [-Werror=maybe-uninitialized] 5758 | if (total && avail_mode) | ~~~~~~^~~~~~~~~~~~~ .../wine/dlls/ntdll/unix/file.c:5755:18: note: 'avail_mode' was declared here 5755 | BOOL avail_mode; | ^~~~~~~~~~ cc1: all warnings being treated as errors make: *** [Makefile:318598: dlls/ntdll/unix/file.o] Error 1 ccache gcc -m64 -c -o dlls/sane.ds/unixlib.o .../wine/dlls/sane.ds/unixlib.c -Idlls/sane.ds \ -I.../wine/dlls/sane.ds -Iinclude \ -I.../wine/include -D__WINESRC__ -DWINE_UNIX_LIB -Wall -pipe \ -fcf-protection=none -fvisibility=hidden -fno-stack-protector -fno-strict-aliasing \ -Wdeclaration-after-statement -Wempty-body -Wignored-qualifiers -Winit-self -Wlogical-op \ -Wno-packed-not-aligned -Wpointer-arith -Wshift-overflow=2 -Wstrict-prototypes -Wtype-limits \ -Wunused-but-set-parameter -Wvla -Wwrite-strings -Wno-misleading-indentation -gdwarf-4 -Werror \ -fPIC -fasynchronous-unwind-tables -g -O1 -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -fno-optimize-sibling-calls -fno-inline-functions .../wine/dlls/sane.ds/unixlib.c: In function 'read_data': .../wine/dlls/sane.ds/unixlib.c:332:8: error: 'status' may be used uninitialized [-Werror=maybe-uninitialized] 332 | if (status == SANE_STATUS_EOF) return TWCC_SUCCESS; | ^ .../wine/dlls/sane.ds/unixlib.c:321:17: note: 'status' was declared here 321 | SANE_Status status; | ^~~~~~ cc1: all warnings being treated as errors make: *** [Makefile:366624: dlls/sane.ds/unixlib.o] Error 1 ccache gcc -m64 -c -o server/directory.o .../wine/server/directory.c -Iserver \ -I.../wine/server -Iinclude \ -I.../wine/include -D__WINESRC__ -DWINE_UNIX_LIB -Wall -pipe \ -fcf-protection=none -fvisibility=hidden -fno-stack-protector -fno-strict-aliasing \ -Wdeclaration-after-statement -Wempty-body -Wignored-qualifiers -Winit-self -Wlogical-op \ -Wno-packed-not-aligned -Wpointer-arith -Wshift-overflow=2 -Wstrict-prototypes -Wtype-limits \ -Wunused-but-set-parameter -Wvla -Wwrite-strings -Wno-misleading-indentation -gdwarf-4 -Werror \ -g -O1 -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -fno-optimize-sibling-calls -fno-inline-functions .../wine/server/directory.c: In function 'create_session': .../wine/server/directory.c:327:44: error: 'snprintf' output may be truncated before the last format character [-Werror=format-truncation=] 327 | snprintf( id_strA, sizeof(id_strA), "%u", id ); | ^ .../wine/server/directory.c:327:5: note: 'snprintf' output between 2 and 11 bytes into a destination of size 10 327 | snprintf( id_strA, sizeof(id_strA), "%u", id ); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ cc1: all warnings being treated as errors make: *** [Makefile:607457: server/directory.o] Error 1 ``` </details> -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9429

2 7

[PATCH v3 0/1] MR9449: winex11.drv: Fix xinerama_get_fullscreen_monitors() not working correctly with...
by Zhiyi Zhang (＠zhiyi) Nov. 13, 2025

Nov. 13, 2025

winex11.drv: Fix xinerama_get_fullscreen_monitors() not working correctly with non-fullscreen rects. xinerama_get_fullscreen_monitors() was supposed to be called with fullscreen rects. Thus, we could assume the monitor indices are all zero when there is only one monitor as an optimization. However, after baf65867, xinerama_get_fullscreen_monitors() can be called with non-fullscreen rects. So we need to remove the optimization so that it works correctly for non-fullscreen rects as well. Fix the Guild Wars 2 game window having an offset in fullscreen mode. -- v3: winex11.drv: Fix xinerama_get_fullscreen_monitors() not working correctly with non-fullscreen rects. https://gitlab.winehq.org/wine/wine/-/merge_requests/9449

3 6

[PATCH 0/2] MR9446: dmsynth: Fix instrument selection
by Anton Baskanov (＠baskanov) Nov. 13, 2025

Nov. 13, 2025

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/9446

3 4

[PATCH v2 0/1] MR9449: winex11.drv: Fix xinerama_get_fullscreen_monitors() not working correctly with...
by Zhiyi Zhang (＠zhiyi) Nov. 13, 2025

Nov. 13, 2025

winex11.drv: Fix xinerama_get_fullscreen_monitors() not working correctly with non-fullscreen rects. xinerama_get_fullscreen_monitors() was supposed to be called with fullscreen rects. Thus, we could assume the monitor indices are all zero when there is only one monitor as an optimization. However, after baf65867, xinerama_get_fullscreen_monitors() can be called with non-fullscreen rects. So we need to remove the optimization so that it works correctly for non-fullscreen rects as well. Fix the Guild Wars 2 game window having an offset in fullscreen mode. -- v2: winex11.drv: Fix xinerama_get_fullscreen_monitors() not working correctly with non-fullscreen rects. https://gitlab.winehq.org/wine/wine/-/merge_requests/9449

3 2

Re: [PATCH v4 0/1] MR9423: ntdll: Warn when environment block is too large, instead of removing it
by Alexandre Julliard (＠julliard) Nov. 13, 2025

Nov. 13, 2025

No, I don't think it addresses the issue at all. I'll try to come up with something. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9423#note_121789

1 0

[PATCH 0/1] MR9449: winex11.drv: Fix xinerama_get_fullscreen_monitors() not working correctly with...
by Zhiyi Zhang (＠zhiyi) Nov. 13, 2025

Nov. 13, 2025

winex11.drv: Fix xinerama_get_fullscreen_monitors() not working correctly with non-fullscreen rects. xinerama_get_fullscreen_monitors() was supposed to be called with fullscreen rects. Thus, we could assume the monitor indices are all zero when there is only one monitor as an optimization. However, after baf65867, xinerama_get_fullscreen_monitors() can be called with non-fullscreen rects. So we need to remove the optimization so that it works correctly for non-fullscreen rects as well. Fix the Guild Wars 2 game window having an offset in fullscreen mode. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9449

2 2

[PATCH 0/2] MR9447: bcrypt: Complete the remaining four RSA parameters when importing the private key.
by Haoyang Chen (＠chenhaoyang) Nov. 13, 2025

Nov. 13, 2025

Some versions of libgnutls require passing all parameters when importing private keys, otherwise they will crash. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9447

3 9