February 2025 - Wine-GitLab

[PATCH 0/25] MR7226: ntdll: Implement in-process synchronization via the Linux "ntsync" driver.
by Elizabeth Figura (＠zfigura) Feb. 5, 2025

Feb. 5, 2025

This introduces a faster implementation of signal and wait operations on NT events, semaphores, and mutexes, which improves performance to native levels for a wide variety of games and other applications. The goal here is similar to the long-standing out-of-tree "esync" and "fsync" patch sets, but without the flaws that make those patch sets not upstreamable. The Linux "ntsync" driver is not currently released. It has been accepted into the trunk Linux tree for 6.14, so barring any extraordinary circumstances, the API is frozen and it will be released in its current form in about 2 months. Since it has passed all relevant reviewers on the kernel side, and the API is all but released, it seems there is no reason any more not to submit the Wine side to match. Some important notes: * This patch set does *not* include any way to disable ntsync support, since that kind of configuration often seems to be dispreferred where not necessary. In essence, ntsync should just work everywhere. Probably the easiest way to effectively disable ntsync, for the purposes of testing, is to chmod the /dev/ntsync device to prevent its being opened. Regardless, a Wine switch to disable ntsync can be added simply enough. Note that it should probably not take the form of a registry key, however, since it needs to be easily accessible from the server itself. * It is, generally speaking, not possible for only some objects, or some processes, to have backing ntsync objects, while others use the old server path. The esync/fsync patch sets explicitly protected against this by making sure every process had a consistent view of whether esync was enabled. This is not provided here, since no switch is provided to toggle ntsync, and it should not be possible to get into such an inconsistent state without gross misconfiguration. * Similarly, no diagnostic messages are provided to note that ntsync is in use, or not in use. These messages are part of esync/fsync, as well as part of ntsync "testing" trees unofficially distributed. However, if ntsync is working correctly, no message should be necessary. The basic structure is: * Each type of server object which can be waited on by the client (including events, semaphores, mutexes, but also other types such as processes, files) must store an "inproc_sync" object. This "inproc_sync" object is a full server object (note that this differs from esync/fsync). A vector and server request is introduced to retrieve an NT handle to this object from an arbitrary NT handle. Since the actual ntsync objects are simply distinct file descriptions, we then call get_handle_fd from the client to retrieve an fd to the object, and then perform ioctls on it. * Objects signaled by the server (processes, files, etc) perform ntsync ioctls on that object. The backing object in all such cases is simply an event. * Signal and wait operations on the client side attempt to defer to an "inproc_*" function, falling back to the server implementation if it returns STATUS_NOT_IMPLEMENTED. This mirrors how in-process synchronization objects (critical sections, SRW locks, etc) used to be implemented—attempting to use an architecture-specific "fast_*" function and falling back if it returned STATUS_NOT_IMPLEMENTED. * The inproc_sync handles, once retrieved, are cached per-process. This caching takes a similar form to the fd cache. It does not reuse the same infrastructure, however. The primary reason for this is that the fd cache is designed to fit within a 64-bit value and uses 64-bit atomic operations to ensure consistency. However, we need to store more than 64 bits of information. [We also need to modify them after caching, in order to correctly implement handle closing—see below.] The secondary reason is that retrieving the ntsync fd from the inproc_sync handle itself uses the fd cache. * In order to keep the Linux driver simple, it does not implement access flags (EVENT_MODIFY_STATE etc.) Instead, the flags are cached locally and validated there. This too mirrors the fd cache. Note that this means that a malicious process can now modify objects it should not be able modify—which is less true than it is with wineserver—but this is no different from the way other objects (notably fds) are handled, and would require manual syscalls. * In order to achieve correct behaviour related to closing objects while they are used, this patch set essentially relies on refcounting. This is broadly true of the server as well, but because we need to avoid server calls when performing object operations, significantly more care must be taken. In particular, because waits need to be interruptable by signals and then be restarted, we need the backing ntsync object to remain valid until all users are done with it. On a process level, this is achieved by letting multiple processes own handles to the underlying inproc_sync server object. On a thread level, multiple simultaneous calls need to refcount the process's local handle. This refcount is stored in the sync object cache. When it reaches zero, the cache is cleared. Punting this behaviour to the Linux driver would have introduced a great deal more complexity, which is best kept in userspace and out of the kernel. * The cache is, as such, treated as a cache. The penultimate commit, which introduces client support but does not yet cache the objects, effectively illustrates this by never actually caching anything, and retrieving a new NT handle and fd every time. * Certain waits, on internal handles (such as async, startup_info, completion), are delegated to the server even when ntsync is used. Those server objects do not create an underlying ntsync object. -- This merge request has too many patches to be relayed via email. Please visit the URL below to see the contents of the merge request. https://gitlab.winehq.org/wine/wine/-/merge_requests/7226

8 34

Re: [PATCH v3 0/3] MR7256: ntdll: monotonic_counter and NtQuerySystemTime performance improvements on macOS.
by Brendan Shanks (＠bshanks) Feb. 5, 2025

Feb. 5, 2025

On Tue Feb 4 22:24:37 2025 +0000, Marc-Aurel Zent wrote: > Indeed, but this commit essentially inlines the call to > `__commpage_gettimeofday()` inside `gettimeofday()`, wheres the latter > still incurs of overhead of 23 CPU cycles before calling into > `__commpage_gettimeofday()`, if I counted correctly. > It actually does more than inlining though, `gettimeofday()`, will do a > syscall if `__commpage_gettimeofday()` fails, and use that value instead. > This can happen rarely (about 1 in a million in my testing), probably > when page time values that are updated asynchronously from the kernel > are read simultaneously. > As the commit is right now, it effectively tries to read from the > commpage *twice*, making that chance of doing a syscall astronomically low. > Also it is still measurably faster to use `__commpage_gettimeofday()` > directly (benchmarked at 10000000 iterations through Rosetta): > ``` > clock_gettime() Duration: 0.290506 seconds > gettimeofday() Duration: 0.233024 seconds > __commpage_gettimeofday() Duration: 0.217468 seconds > ``` IMO, this feels like a micro-optimization that isn't worth the complexity and risk to save 23 cycles per call. If it makes a meaningful difference in a game benchmark it could be worth it, but otherwise it feels better to chase bigger gains. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7256#note_93758

1 0

[PATCH v23 0/2] MR7131: DIR command improvements: Make /o by itself behave equal to how it does in Windows
by Joe Souza (＠JoeS209) Feb. 5, 2025

Feb. 5, 2025

"DIR /o" (no order specifier after /o) hasn't functioned the same as it does on Windows. This change makes the functionality of /o by itself equivalent to behavior on Windows. -- v23: programs/cmd: Cleanup DIR /O logic and remove @todo_wine@ from working tests. programs/cmd/tests: Add tests for command DIR /O. https://gitlab.winehq.org/wine/wine/-/merge_requests/7131

2 3

[PATCH 0/1] MR7271: wineboot: Add dummy entry for SystemBiosDate
by Sebastian Scheibner (＠sebastian) Feb. 5, 2025

Feb. 5, 2025

The software Helicon Filter 5 crashes at startup when it tries to query the registry entry `SystemBiosDate` under `HKEY_LOCAL_MACHINE\Hardware\Description\System`, which doesn't exist in the wine registry. That part of the registry is volatile, so the workaround of just adding that key is short lived. In this commit I just added a dummy date (01/01/70) for simplicity, not sure if that's ok? There's code in wineboot.c which sets the actual BIOS date into the `BIOSReleaseDate` key. That could potentially be used, but it has a different format: 01/01/1970 (four digits for the year instead of two) I've found no official documentation of the `SystemBiosDate` key. The only place I've found is in an example on this page: https://learn.microsoft.com/en-us/windows-hardware/drivers/debuggercmds/-reg <details><summary>Wine log from startup crash when querying SystemBiosDate</summary> <pre> 0024:Call advapi32.RegOpenKeyExA(80000002,00810cdc "HARDWARE\\DESCRIPTION\\System",00000000,00020019,00e6fb68) ret=00440607 0024:Call ntdll.RtlRunOnceExecuteOnce(7b69ec60,7b65bad0,00000000,00000000) ret=7b6106d9 0024:Ret ntdll.RtlRunOnceExecuteOnce() retval=00000000 ret=7b6106d9 0024:Call ntdll.wcscpy(00e6f9ac,7b69ec94 L"") ret=7b60ce6b 0024:Ret ntdll.wcscpy() retval=00e6f9ac ret=7b60ce6b 0024:Call ntdll.RtlInitAnsiString(00e6fadc,00810cdc "HARDWARE\\DESCRIPTION\\System") ret=7b62f0bc 0024:Ret ntdll.RtlInitAnsiString() retval=0000001c ret=7b62f0bc 0024:Call ntdll.RtlAnsiStringToUnicodeString(7ffc2bf8,00e6fadc,00000000) ret=7b62f0e1 0024:Ret ntdll.RtlAnsiStringToUnicodeString() retval=00000000 ret=7b62f0e1 0024:Call ntdll.NtOpenKeyEx(00e6fb68,00020019,00e6f9a8,00000000) ret=7b65e875 0024:Ret ntdll.NtOpenKeyEx() retval=00000000 ret=7b65e875 0024:Call ntdll.RtlNtStatusToDosError(00000000) ret=7b62f0f1 0024:Ret ntdll.RtlNtStatusToDosError() retval=00000000 ret=7b62f0f1 0024:Ret advapi32.RegOpenKeyExA() retval=00000000 ret=00440607 0024:Call advapi32.RegQueryValueExA(000000f8,00810cc4 "SystemBiosDate",00000000,00e6fb58,00000000,00e6fb74) ret=00440c43 0024:Call ntdll.RtlInitAnsiString(00e6f9b0,00810cc4 "SystemBiosDate") ret=7b6305f6 0024:Ret ntdll.RtlInitAnsiString() retval=0000000f ret=7b6305f6 0024:Call ntdll.RtlAnsiStringToUnicodeString(00e6f9b8,00e6f9b0,00000001) ret=7b630614 0024:Ret ntdll.RtlAnsiStringToUnicodeString() retval=00000000 ret=7b630614 0024:Call ntdll.NtQueryValueKey(000000f8,00e6f9b8,00000002,00e6f9c0,00000100,00e6f9a8) ret=7b630695 0024:Ret ntdll.NtQueryValueKey() retval=c0000034 ret=7b630695 0024:Call ntdll.RtlFreeUnicodeString(00e6f9b8) ret=7b6306c0 0024:Ret ntdll.RtlFreeUnicodeString() retval=00000001 ret=7b6306c0 0024:Call ntdll.RtlNtStatusToDosError(c0000034) ret=7b6306d2 0024:Ret ntdll.RtlNtStatusToDosError() retval=00000002 ret=7b6306d2 0024:Ret advapi32.RegQueryValueExA() retval=00000002 ret=00440c43 0024:Call advapi32.RegCloseKey(000000f8) ret=00440433 0024:Call ntdll.NtClose(000000f8) ret=7b628979 0024:Ret ntdll.NtClose() retval=00000000 ret=7b628979 0024:Call ntdll.RtlNtStatusToDosError(00000000) ret=7bcebd50 0024:Ret ntdll.RtlNtStatusToDosError() retval=00000000 ret=7bcebd50 0024:Ret advapi32.RegCloseKey() retval=00000000 ret=00440433 0024:Call user32.LoadStringA(00400000,0000fff0,00e6eb54,00001000) ret=0040879f 0024:Call ntdll.LdrFindResource_U(00400000,00e6ea4c,00000003,00e6ea38) ret=7b5fbb3b 0024:Ret ntdll.LdrFindResource_U() retval=00000000 ret=7b5fbb3b 0024:Call ntdll.LdrAccessResource(00400000,00b033e0,00e6eac0,00000000) ret=7b6148f6 0024:Ret ntdll.LdrAccessResource() retval=00000000 ret=7b6148f6 0024:Ret user32.LoadStringA() retval=00000021 ret=0040879f 0024:Call KERNEL32.RaiseException(0eedfade,00000001,00000007,00e6fb88) ret=00417362 0024:trace:seh:dispatch_exception code=eedfade (unknown) flags=1 addr=7B627448 0024:trace:seh:dispatch_exception info[0]=00417362 0024:trace:seh:dispatch_exception info[1]=02B80430 0024:trace:seh:dispatch_exception info[2]=0041625C 0024:trace:seh:dispatch_exception info[3]=00E6FBB8 0024:trace:seh:dispatch_exception info[4]=00000000 0024:trace:seh:dispatch_exception info[5]=00E6FDF8 0024:trace:seh:dispatch_exception info[6]=00E6FBA4 0024:trace:seh:dispatch_exception eip=7b627448 esp=00e6fae0 ebp=00e6fb38 eflags=00000246 0024:trace:seh:dispatch_exception eax=00e6fae0 ebx=00000018 ecx=00e6faf4 edx=00e6fb88 0024:trace:seh:dispatch_exception esi=00e6fdf8 edi=00e6fb50 cs=0023 ds=002b es=002b fs=0063 gs=006b ss=002b 0024:trace:seh:call_seh_handlers calling handler at 00810CAC code=eedfade flags=1 0024:trace:seh:call_seh_handlers handler at 00810CAC returned 1 0024:trace:seh:call_seh_handlers calling handler at 00874F3A code=eedfade flags=1 0024:trace:seh:call_seh_handlers handler at 00874F3A returned 1 0024:trace:seh:call_seh_handlers calling handler at 0096B5C7 code=eedfade flags=1 0024:trace:seh:call_seh_handlers handler at 0096B5C7 returned 1 0024:trace:seh:call_seh_handlers calling handler at 004059FC code=eedfade flags=1 0024:Call KERNEL32.UnhandledExceptionFilter(00e6ef18) ret=00405a20 0024:Call ntdll.NtQueryInformationProcess(ffffffff,0000000c,00e6ee84,00000004,00000000) ret=7b602304 0024:Ret ntdll.NtQueryInformationProcess() retval=00000000 ret=7b602304 0024:Call ntdll.NtCreateEvent(00e6eeac,001f0003,00e6eeb0,00000000,00000000) ret=7b642f85 0024:Ret ntdll.NtCreateEvent() retval=00000000 ret=7b642f85 wine: Unhandled exception 0x0eedfade in thread 24 at address 7B627448 (thread 0024), starting debugger... </pre> </details> -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7271

2 1

[PATCH 0/1] MR7270: ntdll: Zero terminate return string for NtQueryInformationProcess( ProcessImageFileName[Win32] ).
by Paul Gofman (＠gofman) Feb. 5, 2025

Feb. 5, 2025

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/7270

2 1

[PATCH v3 0/4] MR7251: TLS slots alignment.
by eric pouech (＠epo) Feb. 5, 2025

Feb. 5, 2025

This MR enforces alignment of TLS slots as described in PE file. -- v3: ntdll: Let layout of TLS vector be closer to native. ntdll: Enforce the alignment of TLS directory entries. kernel32: Add test for TLS memory layout. kernel32/tests: Add a test about TLS slot alignment. https://gitlab.winehq.org/wine/wine/-/merge_requests/7251

3 5

[PATCH 0/1] MR7269: imm32: Always validate the IME UI window when painting.
by Tim Clem (＠tclem) Feb. 5, 2025

Feb. 5, 2025

Even when there is no context. Prevents a storm of WM_PAINTs if the window is somehow shown in that case. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7269

3 2

Re: [PATCH v17 0/4] MR4551: Server-side thread priorities implementation (Part 1)
by Marc-Aurel Zent (＠mzent) Feb. 5, 2025

Feb. 5, 2025

Rebased and seems to be still working fine. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/4551#note_93740

1 0

[PATCH v2 0/1] MR7268: ntdll: Fix syscall_cfa offset in user_mode_abort_thread for ARM64.
by Jinoh Kang (＠iamahuman) Feb. 5, 2025

Feb. 5, 2025

Otherwise sending ^C to a running ARM64 process crashes it randomly. Note that 0x110 is prev_frame, the field that immediately precedes syscall_cfa. Fixes: ca79629b3e436c8799281319dd313249bf05b9b5 -- v2: ntdll: Fix syscall_cfa offset in user_mode_abort_thread for ARM64. https://gitlab.winehq.org/wine/wine/-/merge_requests/7268

2 1

[PATCH 0/1] MR7268: ntdll: Fix syscall_cfa offset in user_mode_abort_thread.
by Jinoh Kang (＠iamahuman) Feb. 5, 2025

Feb. 5, 2025

Otherwise sending ^C to a running ARM64 process crashes it randomly. Note that 0x110 is prev_frame, the field that immediately precedes syscall_cfa. Fixes: ca79629b3e436c8799281319dd313249bf05b9b5 -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7268

2 1