January 2024 - Wine-gitlab - list.winehq.org

[PATCH v2 0/1] MR4914: ntdll: Avoid sending context in wait_suspend() when not required.
by Paul Gofman (＠gofman) 08 Feb '24

08 Feb '24

Currently usr1_handler always delivers the context to server. With AVX enabled the single context size is around 1k (while there are two of those for wow64). I am now working on a more generic xstate support in contexts (mostly motivated by AVX512), with AVX512 the single context to be transferred to server is ~1k bigger. The context is needed to be passed to the server from usr1_handler only for NtGetContextThread, NtSetContextThread and NtSuspendThread handling (e. g., when stop_thread is called on the server). The vast majority of usr1_handler usage is for kernel APCs (e. g., APC_ASYNC_IO involved in every async operation) that don't need the thread context transferred to the server and back. My measurements of single SERVER_START_REQ( select ) in server_select() shows that the turnaround time of the request with the context (on native x64 without wow context) is almost two times bigger on average when currently supported AVX context is involved (that is, on every usr1_handle invocation on a machine supporting AVX). So, this patch is expected to increase the time of relatively rare calls which actually need contexts by roughly 50% but decrease the turnaround time of frequent calls involving system APCs by 50%. The difference will be more in favour of this patch once huge AVX512 contexts are added. -- v2: ntdll: Avoid sending context in wait_suspend() when not required. https://gitlab.winehq.org/wine/wine/-/merge_requests/4914

5 23

[PATCH 0/4] MR3422: ddraw tests for Pick and GetPickRecords
by Jeffrey Smith (＠whydoubt) 08 Feb '24

08 Feb '24

Basically, this is the tests from https://gitlab.winehq.org/wine/wine/-/merge_requests/3420 with fixes applied where they originate, rather than later in the series. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/3422

4 6

[PATCH v3 0/1] MR590: vkd3d-shader/dxil: Support forward-referenced value ids.
by Conor McCarthy (＠cmccarthy) 07 Feb '24

07 Feb '24

This is relevant to the structuriser because branched DXIL can sometimes contain forward refs. It doesn't need to be implemented in the backend until necessary, but it adds a parameter to `register_init_ssa_vector()` and `register_init_ssa_scalar()` so it's convenient to add before further use of those functions. -- v3: vkd3d-shader/dxil: Support forward-referenced value ids. https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/590

4 6

[PATCH v4 0/2] MR4786: win32u: Query virtual desktop status from the registry.
by Zhiyi Zhang (＠zhiyi) 07 Feb '24

07 Feb '24

Checking flags of the thread desktop to determine whether virtual desktop is on is unreliable. For example, CEF applications create their own desktop and so is_virtual_desktop() could incorrectly report that virtual desktop is off. Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=55810 -- v4: server: inherit wine internal desktop flags when creating desktops. https://gitlab.winehq.org/wine/wine/-/merge_requests/4786

3 5

[PATCH v3 0/3] MR4915: d3d9: Implement Direct3DCreate9On12().
by Mohamad Al-Jaf (＠maljaf) 07 Feb '24

07 Feb '24

Needed by Orbiter Space Flight Simulator and Train Simulator Classic. -- v3: d3d9/tests: Add Direct3DCreate9On12() tests. d3d9: Implement Direct3DCreate9On12(). https://gitlab.winehq.org/wine/wine/-/merge_requests/4915

3 11

[PATCH 0/1] MR4737: configure: On aarch64 align PE binaries to 64k pages
by Fabian Maurer (＠DarkShadow44) 06 Feb '24

06 Feb '24

It still works using lower page sizes while also allowing page sizes > 4k Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52715 -- https://gitlab.winehq.org/wine/wine/-/merge_requests/4737

5 10

[PATCH v2 0/2] MR618: vkd3d-shader/dxil: Hyperbolic trigonometry.
by Conor McCarthy (＠cmccarthy) 06 Feb '24

06 Feb '24

-- v2: vkd3d-shader/dxil: Handle hyperbolic trigonometric functions in sm6_parser_emit_dx_unary(). tests/shader-runner: Add hyperbolic trigonometry tests. https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/618

3 3

[PATCH v2 0/1] MR4854: ntdll: Reimplement set_native_thread_name() for macOS
by Marc-Aurel Zent (＠mzent) 05 Feb '24

05 Feb '24

This reworks `set_native_thread_name()` for macOS to not rely on mach thread ports returned server side (which were potentially colliding cross-process anyways) and re-implements what `pthread_setname_np()` would do on wines side, with additionally allowing setting threadnames cross-process. On the kernel side `bsd_setthreadname()` is apparently "known to race with regards to the contents of the thread name", so that is only exposed for the thread calling it on itself. Since the pthread struct is private, the location of the thread name is determined at runtime (although in practice, it is constant (80 bytes) from at least 10.15 up to 14.0). This has an effect on both `pthread_getname_np()` and the `NSThread` implementation on top a native library would use. Cross-thread naming is only not visible to external debuggers, which have acquired a mach port right and calling `thread_info()` with `THREAD_EXTENDED_INFO` -- v2: ntdll: Reimplement set_native_thread_name() for macOS. https://gitlab.winehq.org/wine/wine/-/merge_requests/4854

3 3

[PATCH 0/4] MR4613: Stg handles' flags when inheriting.
by eric pouech (＠epo) 05 Feb '24

05 Feb '24

This series' main purpose is to fix the handles' attributes of standard I/O handles in CreateProcess. Needed behavior in CreateProcess is implemented by supporting the DUPLICATE_SAME_ATTRIBUTES option in NtDuplicateObject() and DuplicateHandle(). This option is not documented in SDK, but is in DDK. Manifest constant has been added to DDK header (maybe now Wine's DUPLICATE_MAKE_GLOBAL could be moved here to). -- https://gitlab.winehq.org/wine/wine/-/merge_requests/4613

3 8

[PATCH 0/2] MR4967: winsock: Stop racings send forcing async results and reporting socket as not write ready in select().
by Paul Gofman (＠gofman) 05 Feb '24

05 Feb '24

When two sends are performed in parallel on a socket the send which came second to server's (socked_send) will always considered blocking if the first send hasn't performed set_async_direct_result() in client's ntdll yet. Similarly, if select() is coming shortly after send() it may report socket as not write ready. We don't care for avoiding a race here because if those two sends or send / select are not synced by app it is unlikely possible for an app to depend on which gets processed first, and the one which was called earlier in wall time now also is not guaranteed to be actually processed first. The patches fix Tekken 8 which gets frequently disconnected from multiplayer matches. The game treats WSAEWOULDBLOCK from send() on its non-blocking UDP socket, as well as the same socket not reported as write ready from select as a disconnection condition. Those conditions normally (outside of actual network problem / contention) only happen due to the races described above (while the game performs send() concurrently from different threads or is racing send with select). -- https://gitlab.winehq.org/wine/wine/-/merge_requests/4967

5 9