November 2022 - Wine-gitlab

[PATCH v2 0/7] MR1312: ntdll: Implement heap user flags and fix GlobalHandle / LocalHandle with heap pointers.
by Rémi Bernon 10 Nov '22

10 Nov '22

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=53741 -- v2: ntdll: Implement RtlSetUserFlagsHeap. ntdll: Check block user flags in RtlGetUserInfoHeap. ntdll: Check block user flags in RtlSetUserValueHeap. ntdll: Store the user flags in the heap block flags. ntdll: Introduce a new heap block_set_flags helper. kernel32/tests: Test (Global|Local)Handle with heap allocated pointers. kernel32/tests: Test more Global/Local functions with invalid pointers. https://gitlab.winehq.org/wine/wine/-/merge_requests/1312

2 9

[PATCH 0/3] MR1322: opengl32: Convert to PE.
by Rémi Bernon 10 Nov '22

10 Nov '22

Performance hit measured on a few games (with esync and wined3d), CPU% is the time spent in the syscall dispatcher as measured with `perf top -K` while the game is running, the measured scene is usually right after starting a new game: * Brink (32-bit, GL, id Tech 4): 550 fps, 3% CPU -> 480 fps, 12% CPU * Shadow Warrior 2 (64-bit, d3d11): 75 fps, 2% CPU -> 50 fps, 20% CPU * Hat in Time (64-bit, d3d9, UE3): 220 fps, 5% CPU -> 215 fps, 10% CPU * Control (64-bit, d3d11): 25 fps, 0% CPU -> 25 fps, 3% CPU * Obduction (64-bit, d3d11, UE4): 85 fps, 30% CPU -> 80 fps, 30% CPU -- https://gitlab.winehq.org/wine/wine/-/merge_requests/1322

3 8

Re: [PATCH v2 0/1] MR1304: light.msstyles: Add nonclient metrics.
by Rémi Bernon 10 Nov '22

10 Nov '22

On Thu Nov 10 14:15:17 2022 +0000, Zhiyi Zhang wrote: > It's hard to tell if it's caused by this commit. I can not reproduce > this with Ubuntu 22.10 GNOME 43. Then I switched to FVWM and > successfully reproduced the failures. However, just as when I tried to > bisect, all the failures are gone. I also tried to create a new prefix > with f7e1bb9(right before this commit), then apply this patch and rerun > the tests, but still no failures. After applying the patch, I manually > copied the new light.msstyles to the old prefix, no failures. Then > toggling the theming status or recreating the prefix with 9f3138b can > not reproduce those failures either. > Those failures also look unrelated to the theming changes. Reverting that single commit fixes the Gitlab pipeline, see https://gitlab.winehq.org/rbernon/wine/-/pipelines/4066 vs https://gitlab.winehq.org/rbernon/wine/-/pipelines/4064. It's perhaps something with missing fonts? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/1304#note_13944

1 0

Re: [PATCH v2 0/1] MR1304: light.msstyles: Add nonclient metrics.
by Zhiyi Zhang (＠zhiyi) 10 Nov '22

10 Nov '22

On Thu Nov 10 14:15:18 2022 +0000, Rémi Bernon wrote: > I believe this broke Gitlab CI, this is the only pipeline that failed > from yesterday, in user32, and it now fails with all MRs. It's hard to tell if it's caused by this commit. I can not reproduce this with Ubuntu 22.10 GNOME 43. Then I switched to FVWM and successfully reproduced the failures. However, just as when I tried to bisect, all the failures are gone. I also tried to create a new prefix with f7e1bb9(right before this commit), then apply this patch and rerun the tests, but still no failures. After applying the patch, I manually copied the new light.msstyles to the old prefix, no failures. Then toggling the theming status or recreating the prefix with 9f3138b can not reproduce those failures either. Those failures also look unrelated to the theming changes. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/1304#note_13939

1 0

Re: [PATCH v2 0/1] MR1304: light.msstyles: Add nonclient metrics.
by Rémi Bernon 10 Nov '22

10 Nov '22

I believe this broke Gitlab CI, this is the only pipeline that failed from yesterday, in user32, and it now fails with all MRs. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/1304#note_13938

1 0

[PATCH 0/2] MR1324: Draft: ntdll: Use a faster code path for unix calls.
by Rémi Bernon 10 Nov '22

10 Nov '22

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/1324

1 2

[PATCH v3 0/2] MR1307: wined3d: Use the chunk allocator for GL element array buffers.
by Anton Baskanov (＠baskanov) 10 Nov '22

10 Nov '22

v2: Specify the maximum allowed feature level when creating the device. v3: Rebased. -- v3: wined3d: Use the chunk allocator for GL element array buffers. wined3d: Specify the maximum allowed feature level when creating the device. https://gitlab.winehq.org/wine/wine/-/merge_requests/1307

4 4

[PATCH 0/8] MR1323: winevulkan: Get rid of *_host types.
by Jacek Caban (＠jacek) 10 Nov '22

10 Nov '22

This MR also contains a few fixes and missing conversions support. We no longer need 32-bit types outside vulkan_thunks.c, so we can default to host struct layouts and use `WINE_VK_HOST` in unixlib. This essentially means that vulkan.c is ready for wow64 (vulkan_thunks.c still needs more patches, mostly applying conversions to more types). Instead of defining *_host types, use local *32 types in generated thunks. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/1323

2 8

[PATCH v6 0/2] MR1268: wininet: Partially implement InternetSetFilePointer
by Jason Kuo (＠j20001970) 10 Nov '22

10 Nov '22

The function is implemented by setting cache file pointer to fill read_buf. Consequently, when HTTPREQ_ReadFile is called, it will try reading data from cache after read_buf is depleted, before continuing reading from http stream. Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=26570 -- v6: wininet: Partially implement InternetSetFilePointer https://gitlab.winehq.org/wine/wine/-/merge_requests/1268

3 7

[PATCH v2 0/2] MR1307: wined3d: Use the chunk allocator for GL element array buffers.
by Anton Baskanov (＠baskanov) 10 Nov '22

10 Nov '22

v2: Specify the maximum allowed feature level when creating the device. -- v2: wined3d: Use the chunk allocator for GL element array buffers. wined3d: Specify the maximum allowed feature level when creating the device. https://gitlab.winehq.org/wine/wine/-/merge_requests/1307

3 3