March 2025 - wine-gitlab

Re: [PATCH v5 0/6] MR6866: ntdll: On macOS x86_64, swap GSBASE between the TEB and macOS TSD when entering/leaving PE code.
by Brendan Shanks (＠bshanks) 18 Mar '25

18 Mar '25

On Tue Mar 18 23:36:40 2025 +0000, Jinoh Kang wrote: > Yeah, I think so. FWIW, even with two MOVs, should be faster than one > MOV from GS indirect. Thanks, I suspect the performance impact is negligible but the patch is certainly cleaner. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/6866#note_98354

1 0

[PATCH v4 0/6] MR6866: ntdll: On macOS x86_64, swap GSBASE between the TEB and macOS TSD when entering/leaving PE code.
by Brendan Shanks (＠bshanks) 18 Mar '25

18 Mar '25

x86_64 Windows and macOS both use `%gs` to access thread-specific data (Windows TEB, macOS TSD). To date, Wine has worked around this conflict by filling the most important TEB fields (`0x30`/`Self`, `0x58`/`ThreadLocalStorage`) in the macOS TSD structure (Apple reserved the fields for our use). This was sufficient for most Windows apps. CrossOver's Wine had an additional hack to handle `0x60`/`ProcessEnvironmentBlock`, and binary patches for certain CEF binaries which directly accessed `0x8`/`StackBase`. Additionally, Apple's libd3dshared could activate a special mode in Rosetta 2 where code executing in certain regions would use the Windows TEB when accessing `%gs`. Now that the PE separation is complete, GSBASE can be swapped when entering/exiting PE code. This is done in the syscall dispatcher, unix-call dispatcher, and for user-mode callbacks. GSBASE also needs to be set to the macOS TSD when entering signal handlers (in `init_handler()`), and then restored to the Windows TEB when exiting (in `leave_handler()`). Some changes to the syscall dispatcher were needed to ensure that the TEB is not accessed through `%gs` while on the kernel stack (since a SIGUSR1 while on the kernel stack will result in GSBASE being set to the TSD). --- I've tested this successfully on macOS 15 (Apple Silicon and Intel) and macOS 10.13 with several apps and games, including the `cefclient.exe` CEF sample. Encouragingly, in some simple tests I didn't see a noticeable performance regression from this MR. There are drawbacks though: - libraries which jump directly from PE code into Unix code (expecting that %gs is always pointing to the macOS TSD) will crash. Notable examples are D3DMetal and DXMT. These will need to be changed to use Unix calls. - If Windows code uses the `syscall` instruction directly, the stack pointer likely needs to be valid (which is probably not true on Windows). This is due to the syscall dispatcher saving registers onto the user stack and having to call `_thread_set_tsd_base`. I can't say I've ever seen direct syscalls done with an invalid `%rsp`, but it seems like something anticheat code might do. --- macOS does not have a public API for setting GSBASE, but the private `_thread_set_tsd_base()` works and was added in macOS 10.12. `_thread_set_tsd_base()` is a small thunk that sets `%esi`, `%eax`, and does the `syscall`: https://github.com/apple-oss-distributions/xnu/blob/8d741a5de7ff4191bf97d57…. The syscall instruction itself clobbers `%rcx` and `%r11`. I've tried to save as few registers as possible when calling `_thread_set_tsd_base()`, but there may be room for improvement there. --- I also tested an alternate implementation strategy for this which took advantage of the expanded "full" thread state which is passed to signal handlers when a process has set a user LDT. The full thread state includes GSBASE, so GSBASE is set back to whatever is in the sigcontext on return (like every other field in the context). This would avoid needing to explicitly reset GSBASE in `leave_handler()`. This strategy was simpler, but I'm not using it for 2 reasons: - the "full" thread state is only available starting with macOS 10.15, and we still support 10.13. - more crucially, Rosetta 2 doesn't seem to correctly implement the GS.base field of the full thread state. It's set to 0 on entry, and isn't read on exit. -- v4: ntdll: Remove x86_64 Mac-specific TEB access workarounds that are no longer needed. ntdll: On macOS x86_64, swap GSBASE between the TEB and macOS TSD when entering/leaving PE code. ntdll: Leave kernel stack before accessing %gs in x86_64 syscall dispatcher. ntdll: Don't access the TEB through %gs when using the kernel stack in x86_64 syscall dispatcher. ntdll: Ensure init_handler runs in signal handlers before any compiler-generated memset calls. ntdll: Remove ugly fallback method for getting a thread's GSBASE on macOS. https://gitlab.winehq.org/wine/wine/-/merge_requests/6866

2 6

[PATCH 0/1] MR7609: ntoskrnl.exe: Implement MmMapLockedPages.
by Alexander Morozov (＠amorozov) 18 Mar '25

18 Mar '25

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/7609

2 1

[PATCH 0/5] MR7608: server: Avoid delegating console waits to other objects.
by Elizabeth Figura (＠zfigura) 18 Mar '25

18 Mar '25

Some console objects currently do several unique things: * Delegate waits onto the queue of another object. This is not really a problem for in-process waits, since we can just return the sync object for the delegate. However, it's also unnecessary, adds to the complexity of the server logic, and is one out of one places where this is done. * Make the wait state dependent on the process. This is difficult to emulate with ntsync and would require creating separate server objects for each process, hacking into duplicate_handle. * Fail a wait entirely in certain circumstances. This is pretty much impossible to emulate with in-process waits. Although ntsync has been in development for some time, I have regrettably failed to notice these problems until now. Fortunately, none of these behaviours happen on modern Windows. Although I/O on unbound handles delegates to the console of the current process, the signaled state does not. As the tests here show, the signaled state of a handle depends on the active console of the process in which the handle was created. If that console no longer exists, the signaled state is no longer updated [with one rather inexplicable exception]. Crucially, in current Windows waits never fail, and the state of an object is the same across all process which hold handles to it. Therefore this patch brings our behaviour to closer match current Windows. In theory these are fds and should use default_fd_signaled(). However, the points at which the handles are signaled are completely different, and I/O does not trigger console handles to become signaled when it normally would. Therefore for the time being I've kept the code using custom signaled ops. There is one other oddity related to consoles, which is the existence of console_add_queue(), which seeks to lazily create an input thread when a console is first waited on. This is one out of two places, after this patch, when the wait process is hijacked (the other being message queues). Fortunately this is easy to handle for in-process synchronization objects, by queueing the ioctl from the callback used to retrieve the in-process synchronization object itself. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7608

2 5

[PATCH v11 0/4] MR6855: server: Fix renaming file to a hardlink of itself.
by Gabriel Ivăncescu (＠insn) 18 Mar '25

18 Mar '25

-- v11: server: Handle hardlinks and casefolding when renaming the same file. server: Handle renames to destinations containing trailing slashes. kernel32/tests: Test renaming a file into a hardlink of itself. kernel32/tests: Use FindClose instead of CloseHandle when closing https://gitlab.winehq.org/wine/wine/-/merge_requests/6855

3 11

Re: [PATCH v5 0/1] MR7550: ntdll: Print name for TOKEN_INFORMATION_CLASS - approved
by Alexandre Julliard (＠julliard) 18 Mar '25

18 Mar '25

This merge request was approved by Alexandre Julliard. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7550

1 0

Re: [PATCH v2 0/1] MR7398: winex11.drv: Don't add MWM_DECOR_BORDER to windows without a caption. - approved
by Alexandre Julliard (＠julliard) 18 Mar '25

18 Mar '25

This merge request was approved by Alexandre Julliard. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7398

1 0

Re: [PATCH v3 0/2] MR7601: bluetoothapis: Implement BluetoothGetDeviceInfo. - approved
by Alexandre Julliard (＠julliard) 18 Mar '25

18 Mar '25

This merge request was approved by Alexandre Julliard. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7601

1 0

Re: [PATCH v3 0/2] MR7595: comctl32/tests: Add a test for LVM_FINDITEM with LVS_OWNERDATA. - approved
by Alexandre Julliard (＠julliard) 18 Mar '25

18 Mar '25

This merge request was approved by Alexandre Julliard. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7595

1 0

Re: [PATCH v4 0/1] MR7594: msvcrt: Add _IONBF flag check in _filbuf to avoid dead loop in application which sets the flag. - approved
by Alexandre Julliard (＠julliard) 18 Mar '25

18 Mar '25

This merge request was approved by Alexandre Julliard. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7594

1 0

2025

2024

2023

2022

wine-gitlab March 2025