January 2025 - Wine-gitlab - list.winehq.org

[PATCH v2 0/5] MR7137: sapi: Implement most ISpStream methods.
by Shaun Ren (＠shaunren) 13 Jan '25

13 Jan '25

Needed for Diablo 4's screen reader. -- v2: sapi: Implement IStream methods for SpStream. sapi: Implement ISpStream::GetFormat. sapi: Implement ISpStream::Close. sapi: Implement ISpStream::Set/GetBaseStream. https://gitlab.winehq.org/wine/wine/-/merge_requests/7137

2 5

[PATCH 0/5] MR7137: sapi: Implement most ISpStream methods.
by Shaun Ren (＠shaunren) 13 Jan '25

13 Jan '25

Needed for Diablo 4's screen reader. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7137

2 5

Re: [PATCH v2 0/1] MR7124: mfmediaengine: Fix maths in scaling check.
by Brendan McGrath (＠redmcg) 12 Jan '25

12 Jan '25

On Sun Jan 12 20:25:27 2025 +0000, Nikolay Sivov wrote: > Do we need this intersection check at all? I believe we do. The code in `media_engine_transfer_d3d11` performs a `CopySubresourceRegion` directly to the application supplied texture, which only supports copy; it doesn't support any stretch. So `media_engine_transfer_d3d11` is used as an optimisation for when the application supplied texture has the same (or larger) dimensions as the output sample from the MF pipeline. This check ensures we fallback when stretching is required. The fallback writes to a media engine supplied texture, that is then applied to a quad that can be stretched (before being written to the application texture). -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7124#note_91816

1 0

[PATCH 0/1] MR7107: gdi32: Avoid crashing when tracing NULL xform.
by Rémi Bernon 12 Jan '25

12 Jan '25

Fixes 8375e6c86e0a01f670e48a666088fafff53e39d8 Fixing gdiplus tests crashes in the nightly CI run. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7107

2 2

[PATCH 0/1] MR7136: winex11.drv: Fix display name in X11DRV_UpdateDisplayDevices().
by Zhiyi Zhang (＠zhiyi) 12 Jan '25

12 Jan '25

For example, when GPU 1 has adapter 1 and GPU 2 has adapter 2, without this fix, the `adapter` variable in the for loop is always zero when forming GDI display names \\.\DISPLAY* because these two adapters are on different GPUs. Thus "\\.\DISPLAY1" is incorrectly passed to xrandr14_get_id() for the second adapter and eventually gets the wrong current display mode for the second adapter. Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=57610 -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7136

4 5

[PATCH 0/1] MR7126: winex11: Fixup monitor / display mode inconsistencies.
by Rémi Bernon 12 Jan '25

12 Jan '25

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=57610 This is a simple workaround to fix the regression. The issue seems to be residing in inconsistent values returned by XRandR, and arguably a host driver bug, but it probably could be better fixed by enumerating the XRandR tree only on startup and when it is updated, but this would be a much larger change. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7126

2 3

[PATCH 0/2] MR7110: Draft: win32u: Update current modes relative to the primary source.
by Rémi Bernon 12 Jan '25

12 Jan '25

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=57610 This attempt to fix a regression which is caused by XRandR enumerated output/crtc changing during the display device initial enumeration. The user setup uses an external display and somehow XRandR enumerates the first (non-primary) monitor and its mode without being able to find a primary output/crtc (and uses that first monitor position as primary as fallback). Then, when the second (primary) monitor and its modes are enumerated it actually finds it and uses its as it should. This results in both monitor modes being positioned at 0x0, which breaks various things later on. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7110

2 4

[PATCH 0/1] MR7135: gdiplus/tests: Fix use-after-free of a graphics object (ASan).
by Nikolay Sivov (＠nsivov) 12 Jan '25

12 Jan '25

Signed-off-by: Nikolay Sivov <nsivov(a)codeweavers.com> -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7135

2 2

[PATCH 0/1] MR7134: kernel32/tests: Avoid stack-buffer-overflow in get_com_dir_size. (ASan)
by Nikolay Sivov (＠nsivov) 12 Jan '25

12 Jan '25

Function gets called with Magic == 0xdead, which causes the test to use the 64-bit path, which results in this buffer overflow. Signed-off-by: Nikolay Sivov <nsivov(a)codeweavers.com> -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7134

2 1

[PATCH 0/2] MR7133: gdiplus: Fix some path functions not starting a new figure.
by Esme Povirk (＠madewokherd) 11 Jan '25

11 Jan '25

Found by Mono's System.Drawing tests. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7133

3 3