- Wine-GitLab - list.winehq.org

[PATCH v4 0/4] MR9060: ntdll: Fix crashes in syscall dispatcher due to attempt to restore unitialized data.
by Paul Gofman (＠gofman) Sept. 30, 2025

Sept. 30, 2025

-- v4: ntdll: Handle invalid FP state in usr1_handler() on i386. https://gitlab.winehq.org/wine/wine/-/merge_requests/9060

3 6

[PATCH 0/1] MR9077: imm32: ensure HIMC is unlocked in ImmGenerateMessage
by zhengyong chen (＠chenzhengyong) Sept. 30, 2025

Sept. 30, 2025

Previously, ImmGenerateMessage could return early when ImmLockIMCC(ctx-\>hMsgBuf) failed, without unlocking the HIMC obtained via ImmLockIMC. This could leave the HIMC locked and lead to resource leaks. This patch adds calls to ImmUnlockIMC(himc) at all early return points, and also unlocks the HIMC before returning TRUE at the end, ensuring that every ImmLockIMC call is properly balanced with a corresponding ImmUnlockIMC. Tested on: Ubuntu 25.10 & deepin Linux V25, Wine compiled from master branch. No functional behavior of message dispatch is changed. Signed-off-by: chenzhengyong chenzhengyong(a)uniontech.com -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9077

3 3

[PATCH 0/1] MR9074: imm32: ensure HIMC is unlocked in ImmGenerateMessage
by zhengyong chen (＠chenzhengyong) Sept. 30, 2025

Sept. 30, 2025

Previously, ImmGenerateMessage could return early when ImmLockIMCC(ctx-\>hMsgBuf) failed, without unlocking the HIMC obtained via ImmLockIMC. This could leave the HIMC locked and lead to resource leaks. This patch adds calls to ImmUnlockIMC(himc) at all early return points, and also unlocks the HIMC before returning TRUE at the end, ensuring that every ImmLockIMC call is properly balanced with a corresponding ImmUnlockIMC. Tested on: Ubuntu 25.10 & deepin Linux V25, Wine compiled from master branch. No functional behavior of message dispatch is changed. Signed-off-by: chenzhengyong chenzhengyong(a)uniontech.com -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9074

3 4

[PATCH 0/1] MR9063: ntoskrnl.exe: Create keys for devices in HKEY_DYN_DATA\Config Manager\Enum.
by Elizabeth Figura (＠zfigura) Sept. 30, 2025

Sept. 30, 2025

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/9063

4 6

[PATCH 0/2] MR9014: server: Create an inproc sync for user APC signaling.
by Rémi Bernon Sept. 30, 2025

Sept. 30, 2025

Unlike the message queue for which I believe we can get rid of any special handling, I don't see a way to avoid a dedicated inproc sync for user APCs. This will make sure that the sync gets created on thread creation. It could arguably be lazily allocated instead, but I don't see a reason to delay the failure to a later time and end up with successfully created but unalertable thread when they are supposed to be. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9014

4 8

[PATCH v4 0/7] MR9034: d3dx10: Add support for creating textures from images using shared code.
by Connor McAdams (＠cmcadams) Sept. 30, 2025

Sept. 30, 2025

These patches implement the existing functionality of `load_texture_data()` using shared code. -- v4: d3dx10: Exclusively use shared code in load_texture_data(). d3dx10: Exclusively use shared code to load DDS files in load_texture_data(). d3dx10: Add support for compressing BC4 and BC5 formats. d3dx10: Add support for decompressing BC4 and BC5 formats. d3dx10: Ignore alpha channel values for WICPixelFormat32bppBGR BMP images. d3dx10: Use shared code in load_texture_data() when possible. d3dx{9,10}: Add a tweak to stb_dxt to more closely match native output. https://gitlab.winehq.org/wine/wine/-/merge_requests/9034

3 8

[PATCH v2 0/1] MR9052: win32u: Handle errors when creating EGL context
by Vasiliy Stelmachenok (＠ventureoo) Sept. 30, 2025

Sept. 30, 2025

Some background: I wanted to test EGL backend on my old laptop with Intel graphics limited with OpenGL 2.1 / GLES 2.0. When trying to run any game this led to a crash with NULL GL_RENDERER, but not with GLX. Tracing has shown that wined3d tries to create a context with high `WGL_CONTEXT_MAJOR_VERSION_ARB`/`WGL_CONTEXT_MINOR_VERSION`, which is not supported by hardware, resulting in the creation of nil context in EGL, but due to lack of error handling it's always assumed to be successful. In GLX this results in an error and creating a context does not lead to success, after which it tries again to create a context with a lower `WGL_CONTEXT_MAJOR_VERSION_ARB` / `WGL_CONTEXT_MINOR_VERSION` until the actual context is created. Proper error handling when creating EGL context as in this patch also allows wined3d to work on such hardware. -- v2: win32u: Handle errors when creating EGL context https://gitlab.winehq.org/wine/wine/-/merge_requests/9052

3 2

[PATCH v3 0/7] MR9034: d3dx10: Add support for creating textures from images using shared code.
by Connor McAdams (＠cmcadams) Sept. 30, 2025

Sept. 30, 2025

These patches implement the existing functionality of `load_texture_data()` using shared code. -- v3: d3dx10: Exclusively use shared code in load_texture_data(). d3dx10: Exclusively use shared code to load DDS files in load_texture_data(). d3dx10: Add support for compressing BC4 and BC5 formats. d3dx10: Add support for decompressing BC4 and BC5 formats. d3dx10: Ignore alpha channel values for WICPixelFormat32bppBGR BMP images. d3dx10: Use shared code in load_texture_data() when possible. https://gitlab.winehq.org/wine/wine/-/merge_requests/9034

3 10

[PATCH v10 0/4] MR9010: cfgmgr32: Device query API part 5: Implement support for logical boolean operators in DEVPROP_FILTER_EXPRESSION.
by Vibhav Pant (＠vibhavp) Sept. 30, 2025

Sept. 30, 2025

Additionally, add explicit rules to the AQS parser for resolving shift/reduce conflicts involving NOT/OR clauses. -- v10: cfgmgr32: Implement support for logical operators in DEVPROP_FILTER_EXPRESSION. cfgmgr32/tests: Add additional tests for device query filters. windows.devices.enumeration: Ensure that all AQS logical operators bind left to right. windows.devices.enumeration/tests: Allow timing out on native while waiting for the Enumerated event on a DeviceWatcher. https://gitlab.winehq.org/wine/wine/-/merge_requests/9010

3 7

[PATCH 0/5] MR9075: cfgmgr32: Implement support for logical operators in DEVPROP_FILTER_EXPRESSION.
by Rémi Bernon Sept. 30, 2025

Sept. 30, 2025

What about doing this for !9010? I think it'd be simpler. I left the test timeout change aside, as commented on the other MR. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9075

3 7

[PATCH 0/4] MR9076: xmllite/tests: Add more tests for name validation when writing.
by Nikolay Sivov (＠nsivov) Sept. 30, 2025

Sept. 30, 2025

Signed-off-by: Nikolay Sivov <nsivov(a)codeweavers.com> -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9076

3 5

[PATCH 0/1] MR9052: win32u: Handle errors when creating EGL context
by Vasiliy Stelmachenok (＠ventureoo) Sept. 30, 2025

Sept. 30, 2025

Some background: I wanted to test EGL backend on my old laptop with Intel graphics limited with OpenGL 2.1 / GLES 2.0. When trying to run any game this led to a crash with NULL GL_RENDERER, but not with GLX. Tracing has shown that wined3d tries to create a context with high `WGL_CONTEXT_MAJOR_VERSION_ARB`/`WGL_CONTEXT_MINOR_VERSION`, which is not supported by hardware, resulting in the creation of nil context in EGL, but due to lack of error handling it's always assumed to be successful. In GLX this results in an error and creating a context does not lead to success, after which it tries again to create a context with a lower `WGL_CONTEXT_MAJOR_VERSION_ARB` / `WGL_CONTEXT_MINOR_VERSION` until the actual context is created. Proper error handling when creating EGL context as in this patch also allows wined3d to work on such hardware. I think the real issue is somewhere in wined3d/wgl and passing wrong version of GL for context at first, but I don’t have proper code base knowledge to talk about it exactly. Anyway, I think that error handling when creating a context like with GLX is necessary in case of such errors. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9052

3 3

Re: [PATCH v2 0/7] MR9034: d3dx10: Add support for creating textures from images using shared code.
by Connor McAdams (＠cmcadams) Sept. 30, 2025

Sept. 30, 2025

On Mon Sep 29 21:59:33 2025 +0000, Matteo Bruni wrote: > Just a question in passing, unrelated to the patch: do we know if > `D3DX10_FILTER_POINT` is actually the default filter? I've yet to find any explicit mention of the default filter type for d3dx10/d3dx11 in the documentation, but I doubt that it's a point filter :) I intend on writing some tests to try and figure it out now that I have some understanding of the different filter types. Since we don't handle the `Filter` member in the passed in `D3DX10_IMAGE_LOAD_INFO` and the default type is unknown, I've just defaulted to point here. I can print a `FIXME()` somewhere around here if that'd help clarify, or maybe a code comment. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9034#note_117155

1 0

Re: [PATCH v2 0/7] MR9034: d3dx10: Add support for creating textures from images using shared code.
by Connor McAdams (＠cmcadams) Sept. 30, 2025

Sept. 30, 2025

On Mon Sep 29 21:59:33 2025 +0000, Matteo Bruni wrote: > If I understand correctly, this affects the output. In which case I'd > avoid calling it an optimization, both in the comment and the commit subject. > It's a bit of a pet peeve of mine, please bear with me... I wrote this patch awhile ago, I _think_ I used the term optimization to bolster my argument for changing this behavior. It technically should be faster as it lets us get out of this function earlier. The decompressed values don't change with this "optimization". The actual compressed data values shouldn't really matter, what should matter is the decompressed values. But, in `check_texture{2d,3d}_data()`, we check for compressed values. What would be a better term to use here? Just "Wine specific change"? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9034#note_117153

1 0

[PATCH 0/5] MR9073: Cmd engine rewrite (part XXXVII)
by eric pouech (＠epo) Sept. 30, 2025

Sept. 30, 2025

This MR finishes the eradication of WCMD_ReadFile in favor of WCMD_fgets started in previous MR. A couple of enhancements to MORE command. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9073

2 5

[PATCH 0/6] MR9072: win32u: Implement D3DKMT object sharing through NT handles.
by Rémi Bernon Sept. 30, 2025

Sept. 30, 2025

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/9072

1 6

[PATCH 0/4] MR9071: server: Introduce an internal queue bits field.
by Rémi Bernon Sept. 30, 2025

Sept. 30, 2025

And set a QS_DRIVER bit when the queue fd is ready for reading, or clear it if it isn't, before waiting on it. Clear the bit after waiting as we expect the client to process events before calling the server again and before we can poll it again. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9071

1 4

[PATCH 0/1] MR9070: ddraw/tests: Use ANSI versions of user32 functions.
by Elizabeth Figura (＠zfigura) Sept. 30, 2025

Sept. 30, 2025

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/9070

2 1

[PATCH v2 0/1] MR9058: d3dx9: Add an initial implementation of UpdateSkinnedMesh().
by Elizabeth Figura (＠zfigura) Sept. 29, 2025

Sept. 29, 2025

-- v2: d3dx9: Add an initial implementation of UpdateSkinnedMesh(). https://gitlab.winehq.org/wine/wine/-/merge_requests/9058

3 2

Re: [PATCH v2 0/7] MR9034: d3dx10: Add support for creating textures from images using shared code.
by Matteo Bruni (＠Mystral) Sept. 29, 2025

Sept. 29, 2025

Very minor comments, looks good. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9034#note_117123

1 0

Re: [PATCH v2 0/7] MR9034: d3dx10: Add support for creating textures from images using shared code.
by Matteo Bruni (＠Mystral) Sept. 29, 2025

Sept. 29, 2025

Matteo Bruni (@Mystral) commented about dlls/d3dx10_43/texture.c: > + struct d3dx_pixels src_pixels, dst_pixels; > + uint32_t dst_row_pitch, dst_slice_pitch; > + > + hr = d3dx_image_get_pixels(&image, i, j, &src_pixels); > + if (FAILED(hr)) > + break; > + > + hr = d3dx_calculate_pixels_size(fmt_desc->format, dst_size.width, dst_size.height, &dst_row_pitch, > + &dst_slice_pitch); > + if (FAILED(hr)) > + break; > + > + set_d3dx_pixels(&dst_pixels, pixels_buffer, dst_row_pitch, dst_slice_pitch, NULL, dst_size.width, > + dst_size.height, dst_size.depth, &unaligned_rect); > + > + hr = d3dx_load_pixels_from_pixels(&dst_pixels, fmt_desc, &src_pixels, src_desc, D3DX10_FILTER_POINT, 0); Just a question in passing, unrelated to the patch: do we know if `D3DX10_FILTER_POINT` is actually the default filter? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9034#note_117122

1 0

Re: [PATCH v2 0/7] MR9034: d3dx10: Add support for creating textures from images using shared code.
by Matteo Bruni (＠Mystral) Sept. 29, 2025

Sept. 29, 2025

Matteo Bruni (@Mystral) commented about dlls/d3dx9_36/d3dx_helpers.c: > expected_src_data_size = (image->layer_pitch * image->layer_count) + header_size; > if (src_data_size < expected_src_data_size) > { > + const uint32_t dxt10_info_only_flags = D3DX_IMAGE_INFO_ONLY | D3DX_IMAGE_SUPPORT_DXT10; > + > WARN("File is too short %u, expected at least %u bytes.\n", src_data_size, expected_src_data_size); > - /* D3DX10/D3DX11 do not validate the size of the pixels, only the header. */ > - if (!(flags & D3DX_IMAGE_SUPPORT_DXT10)) > + /* > + * D3DX10/D3DX11 do not validate the size of the pixels, only the header. > + * However, if we're loading the image data, don't match native > + * behavior. > + */ I'd mention explicitly that we're doing it to be safe and until we find reason to do otherwise. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9034#note_117121

1 0

Re: [PATCH v2 0/7] MR9034: d3dx10: Add support for creating textures from images using shared code.
by Matteo Bruni (＠Mystral) Sept. 29, 2025

Sept. 29, 2025

Matteo Bruni (@Mystral) commented about dlls/d3dx9_36/stb_dxt.h: > dest[1] = (unsigned char)mn; > dest += 2; > > + /* > + * Wine specific optimization to more closely match Windows behavior: If > + * max is equal to minimum, just set all bits to 0 (which means the value > + * is the value of max in this case). > + */ > + if (mx == mn) { > + memset(dest, 0, 6); > + return; > + } If I understand correctly, this affects the output. In which case I'd avoid calling it an optimization, both in the comment and the commit subject. It's a bit of a pet peeve of mine, please bear with me... -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9034#note_117120

1 0

[PATCH v2 0/1] MR9048: libs/fluidsynth: Fix data race in g_get_monotonic_time().
by Jinoh Kang (＠iamahuman) Sept. 29, 2025

Sept. 29, 2025

Fixes: f768d6b31bebc35fbaf751d0cd57c8bd302a8d60 -- v2: libs/fluidsynth: Fix data race in g_get_monotonic_time(). https://gitlab.winehq.org/wine/wine/-/merge_requests/9048

3 3

Re: [PATCH v2 0/1] MR9055: gdiplus: Handle PixelFormatPAlpha separately in alpha_blend_hdc_pixels(). - approved
by Alexandre Julliard (＠julliard) Sept. 29, 2025

Sept. 29, 2025

This merge request was approved by Alexandre Julliard. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9055

1 0