August 2020 - Wine-devel - list.winehq.org

[PATCH] winevulkan: Fix vkEnumerateInstanceExtensionProperties() behavior when no Vulkan driver is present.
by Brendan Shanks 15 Aug '20

15 Aug '20

Signed-off-by: Brendan Shanks <bshanks(a)codeweavers.com> --- This is compliant with the spec and matches the behavior of the official Vulkan loader after a recent change, see <https://github.com/KhronosGroup/Vulkan-Loader/pull/423>. dlls/winevulkan/vulkan.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/dlls/winevulkan/vulkan.c b/dlls/winevulkan/vulkan.c index 1b359d22c8c..9f8e2451ad9 100644 --- a/dlls/winevulkan/vulkan.c +++ b/dlls/winevulkan/vulkan.c @@ -793,7 +793,10 @@ VkResult WINAPI wine_vkEnumerateInstanceExtensionProperties(const char *layer_na wine_vk_init_once(); if (!vk_funcs) - return VK_ERROR_INITIALIZATION_FAILED; + { + *count = 0; + return VK_SUCCESS; + } res = vk_funcs->p_vkEnumerateInstanceExtensionProperties(NULL, &num_host_properties, NULL); if (res != VK_SUCCESS) -- 2.26.2

2 1

[PATCH] winmm: Default to 1ms resolution like we used to
by Arkadiusz Hiler 15 Aug '20

15 Aug '20

Wine's winmm is built on an assumption that we default to 1ms resolution without ever calling timeBeginPeriod(1). For simplicity timeGetTime() was backed by GetTickCount() which recently has changed its implementation, and in worst case scenario returns a new value each ~16ms. This breaks our assumption. This patch introduces a proper timeGetTime() that uses QueryPerformanceCounter(). timeGetTime() is now also used instead of GetTickCount() internally by winmm. The research comment was updated with the latest findings. Fixes: cd215bb49bc2 ("kernel32: Use the user shared data to implement GetTickCount().") Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=49564 Signed-off-by: Arkadiusz Hiler <ahiler(a)codeweavers.com> --- dlls/winmm/time.c | 44 +++++++++++++++++++++++++++++++++---------- dlls/winmm/winmm.c | 12 ++++++------ dlls/winmm/winmm.spec | 2 +- 3 files changed, 41 insertions(+), 17 deletions(-) diff --git a/dlls/winmm/time.c b/dlls/winmm/time.c index 22c89852a2..4d802703f1 100644 --- a/dlls/winmm/time.c +++ b/dlls/winmm/time.c @@ -74,17 +74,30 @@ static inline void link_timer( WINE_TIMERENTRY *timer ) /* * Some observations on the behavior of winmm on Windows. - * First, the call to timeBeginPeriod(xx) can never be used - * to raise the timer resolution, only lower it. + * + * First, the call to timeBeginPeriod(xx) can never be used to + * lower the timer resolution (i.e. increase the update + * interval), only to increase the timer resolution (i.e. lower + * the update interval). * * Second, a brief survey of a variety of Win 2k and Win X * machines showed that a 'standard' (aka default) timer * resolution was 1 ms (Win9x is documented as being 1). However, one * machine had a standard timer resolution of 10 ms. * - * Further, if we set our default resolution to 1, - * the implementation of timeGetTime becomes GetTickCount(), - * and we can optimize the code to reduce overhead. + * Further, timeBeginPeriod(xx) also affects the resolution of + * wait calls such as NtDelayExecution() and + * NtWaitForMultipleObjects() which by default round up their + * timeout to the nearest multiple of 15.625ms across all Windows + * versions. In Wine, all of those currently work with sub-1ms + * accuracy. + * + * Effective time resolution is a global value that is the max + * of the resolutions (i.e. min of update periods) requested by + * all the processes. A lot of programs seem to do + * timeBeginPeriod(1) forcing it onto everyone else. + * + * Defaulting to 1ms accuracy in winmm should be safe. * * Additionally, a survey of Event behaviors shows that * if we request a Periodic event every 50 ms, then Windows @@ -97,6 +110,7 @@ static inline void link_timer( WINE_TIMERENTRY *timer ) * no delays. * * Jeremy White, October 2004 + * Arkadiusz Hiler, August 2020 */ #define MMSYSTIME_MININTERVAL (1) #define MMSYSTIME_MAXINTERVAL (65535) @@ -131,7 +145,7 @@ static int TIME_MMSysTimeCallback(void) } timer = LIST_ENTRY( ptr, WINE_TIMERENTRY, entry ); - delta_time = timer->dwTriggerTime - GetTickCount(); + delta_time = timer->dwTriggerTime - timeGetTime(); if (delta_time > 0) break; list_remove( &timer->entry ); @@ -242,16 +256,26 @@ void TIME_MMTimeStop(void) */ MMRESULT WINAPI timeGetSystemTime(LPMMTIME lpTime, UINT wSize) { - if (wSize >= sizeof(*lpTime)) { - lpTime->wType = TIME_MS; - lpTime->u.ms = GetTickCount(); + lpTime->wType = TIME_MS; + lpTime->u.ms = timeGetTime(); } return 0; } +/************************************************************************** + * timeGetTime [WINMM.@] + */ +DWORD WINAPI timeGetTime(void) +{ + LARGE_INTEGER now; + if (!QueryPerformanceCounter(&now)) + WARN("QueryPerformanceCounter should not fail!\n"); + return now.QuadPart/10000; +} + /************************************************************************** * timeSetEvent [WINMM.@] */ @@ -272,7 +296,7 @@ MMRESULT WINAPI timeSetEvent(UINT wDelay, UINT wResol, LPTIMECALLBACK lpFunc, return 0; lpNewTimer->wDelay = wDelay; - lpNewTimer->dwTriggerTime = GetTickCount() + wDelay; + lpNewTimer->dwTriggerTime = timeGetTime() + wDelay; /* FIXME - wResol is not respected, although it is not clear that we could change our precision meaningfully */ diff --git a/dlls/winmm/winmm.c b/dlls/winmm/winmm.c index 9075a78f34..eea29ea61e 100644 --- a/dlls/winmm/winmm.c +++ b/dlls/winmm/winmm.c @@ -1030,7 +1030,7 @@ static DWORD midistream_get_playing_position(WINE_MIDIStream* lpMidiStrm) case MSM_STATUS_PAUSED: return lpMidiStrm->dwElapsedMS; case MSM_STATUS_PLAYING: - return GetTickCount() - lpMidiStrm->dwStartTicks; + return timeGetTime() - lpMidiStrm->dwStartTicks; default: FIXME("Unknown playing status %hu\n", lpMidiStrm->status); return 0; @@ -1080,7 +1080,7 @@ static BOOL MMSYSTEM_MidiStream_MessageHandler(WINE_MIDIStream* lpMidiStrm, LPWI /* FIXME: send out cc64 0 (turn off sustain pedal) on every channel */ if (lpMidiStrm->status != MSM_STATUS_PLAYING) { EnterCriticalSection(&lpMidiStrm->lock); - lpMidiStrm->dwStartTicks = GetTickCount() - lpMidiStrm->dwElapsedMS; + lpMidiStrm->dwStartTicks = timeGetTime() - lpMidiStrm->dwElapsedMS; lpMidiStrm->status = MSM_STATUS_PLAYING; LeaveCriticalSection(&lpMidiStrm->lock); } @@ -1090,7 +1090,7 @@ static BOOL MMSYSTEM_MidiStream_MessageHandler(WINE_MIDIStream* lpMidiStrm, LPWI /* FIXME: send out cc64 0 (turn off sustain pedal) on every channel */ if (lpMidiStrm->status != MSM_STATUS_PAUSED) { EnterCriticalSection(&lpMidiStrm->lock); - lpMidiStrm->dwElapsedMS = GetTickCount() - lpMidiStrm->dwStartTicks; + lpMidiStrm->dwElapsedMS = timeGetTime() - lpMidiStrm->dwStartTicks; lpMidiStrm->status = MSM_STATUS_PAUSED; LeaveCriticalSection(&lpMidiStrm->lock); } @@ -1126,7 +1126,7 @@ static BOOL MMSYSTEM_MidiStream_MessageHandler(WINE_MIDIStream* lpMidiStrm, LPWI case WINE_MSM_HEADER: /* sets initial tick count for first MIDIHDR */ if (!lpMidiStrm->dwStartTicks) - lpMidiStrm->dwStartTicks = GetTickCount(); + lpMidiStrm->dwStartTicks = timeGetTime(); lpMidiHdr = (LPMIDIHDR)msg->lParam; lpData = (LPBYTE)lpMidiHdr->lpData; TRACE("Adding %s lpMidiHdr=%p [lpData=0x%p dwBytesRecorded=%u/%u dwFlags=0x%08x size=%lu]\n", @@ -1235,8 +1235,8 @@ start_header: dwToGo = lpMidiStrm->dwStartTicks + lpMidiStrm->position_usec / 1000; - TRACE("%u/%u/%u\n", dwToGo, GetTickCount(), me->dwDeltaTime); - while (dwToGo - (dwCurrTC = GetTickCount()) <= MAXLONG) { + TRACE("%u/%u/%u\n", dwToGo, timeGetTime(), me->dwDeltaTime); + while (dwToGo - (dwCurrTC = timeGetTime()) <= MAXLONG) { if (MsgWaitForMultipleObjects(0, NULL, FALSE, dwToGo - dwCurrTC, QS_ALLINPUT) == WAIT_OBJECT_0) { /* got a message, handle it */ while (PeekMessageA(&msg, 0, 0, 0, PM_REMOVE)) { diff --git a/dlls/winmm/winmm.spec b/dlls/winmm/winmm.spec index 9c6ed2dfa8..d7c7379d77 100644 --- a/dlls/winmm/winmm.spec +++ b/dlls/winmm/winmm.spec @@ -147,7 +147,7 @@ @ stdcall timeEndPeriod(long) @ stdcall timeGetDevCaps(ptr long) @ stdcall timeGetSystemTime(ptr long) -@ stdcall timeGetTime() kernel32.GetTickCount +@ stdcall timeGetTime() @ stdcall timeKillEvent(long) @ stdcall timeSetEvent(long long ptr long long) @ stdcall waveInAddBuffer(long ptr long) -- 2.28.0

5 4

msvcrt - memmove/memcpy optimizations
by Fabian Maurer 14 Aug '20

14 Aug '20

Hello, since msvcrt isn't relying on the standard library memmove/memcpy anymore, there's been a pretty bad performance regression. See https://bugs.winehq.org/ show_bug.cgi?id=49663. For the best performance, and since those memory operations are pretty common, we'd presumably like to optimize them as much as possible. You might have seen my patch for an implementation from musl, although Zebediah rightfully pointed out we might want to opt for the best performance we can get... glibc currently offers the best performance, thanks to SSE/AVX implementations and runtime selection of the best supported path. First, would you have any objections adding specialized paths written in assembly for x86? And if we were to add them, would we link against assembly files, or someway transform them into inline assembly? AFAIK, Wine didn't come with pure assembly files yet... If you want, I could set up a few crude benchmarks to see how different versions compare. Regards, Fabian Maurer

5 5

[PATCH v2 4/4] ntdll: Fix arm64 sigcontext access for darwin
by Martin Storsjo 14 Aug '20

14 Aug '20

Signed-off-by: Martin Storsjo <martin(a)martin.st> --- Reduced the patch a bit by defining matching REG_sig/REGn_sig macros. --- dlls/ntdll/unix/signal_arm64.c | 31 +++++++++++++++++++++++++++++++ 1 file changed, 31 insertions(+) diff --git a/dlls/ntdll/unix/signal_arm64.c b/dlls/ntdll/unix/signal_arm64.c index a6ab1dfda7..cb10b000f4 100644 --- a/dlls/ntdll/unix/signal_arm64.c +++ b/dlls/ntdll/unix/signal_arm64.c @@ -110,6 +110,24 @@ static DWORD64 get_fault_esr( ucontext_t *sigcontext ) return 0; } +#elif defined(__APPLE__) + +/* All Registers access - only for local access */ +# define REG_sig(reg_name, context) ((context)->uc_mcontext->__ss.__ ## reg_name) +# define REGn_sig(reg_num, context) ((context)->uc_mcontext->__ss.__x[reg_num]) + +/* Special Registers access */ +# define SP_sig(context) REG_sig(sp, context) /* Stack pointer */ +# define PC_sig(context) REG_sig(pc, context) /* Program counter */ +# define PSTATE_sig(context) REG_sig(cpsr, context) /* Current State Register */ +# define FP_sig(context) REG_sig(fp, context) /* Frame pointer */ +# define LR_sig(context) REG_sig(lr, context) /* Link Register */ + +static DWORD64 get_fault_esr( ucontext_t *sigcontext ) +{ + return sigcontext->uc_mcontext->__es.__esr; +} + #endif /* linux */ static pthread_key_t teb_key; @@ -328,6 +346,7 @@ static void restore_context( const CONTEXT *context, ucontext_t *sigcontext ) */ static void save_fpu( CONTEXT *context, ucontext_t *sigcontext ) { +#ifdef linux struct fpsimd_context *fp = get_fpsimd_context( sigcontext ); if (!fp) return; @@ -335,6 +354,12 @@ static void save_fpu( CONTEXT *context, ucontext_t *sigcontext ) context->Fpcr = fp->fpcr; context->Fpsr = fp->fpsr; memcpy( context->V, fp->vregs, sizeof(context->V) ); +#elif defined(__APPLE__) + context->ContextFlags |= CONTEXT_FLOATING_POINT; + context->Fpcr = sigcontext->uc_mcontext->__ns.__fpcr; + context->Fpsr = sigcontext->uc_mcontext->__ns.__fpsr; + memcpy( context->V, sigcontext->uc_mcontext->__ns.__v, sizeof(context->V) ); +#endif } @@ -345,12 +370,18 @@ static void save_fpu( CONTEXT *context, ucontext_t *sigcontext ) */ static void restore_fpu( CONTEXT *context, ucontext_t *sigcontext ) { +#ifdef linux struct fpsimd_context *fp = get_fpsimd_context( sigcontext ); if (!fp) return; fp->fpcr = context->Fpcr; fp->fpsr = context->Fpsr; memcpy( fp->vregs, context->V, sizeof(fp->vregs) ); +#elif defined(__APPLE__) + sigcontext->uc_mcontext->__ns.__fpcr = context->Fpcr; + sigcontext->uc_mcontext->__ns.__fpsr = context->Fpsr; + memcpy( sigcontext->uc_mcontext->__ns.__v, context->V, sizeof(context->V) ); +#endif } -- 2.24.3 (Apple Git-128)

1 0

[PATCH 0/8] Fix SetForegroundWindow focus inversion.
by Rémi Bernon 14 Aug '20

14 Aug '20

I'm resending these, with the tests also run in cross-process mode (in addition to the same-process/cross-thread calls). The results are the same in both cases, and it exposes an inversion in Wine which can then cause various focus inconsistencies. PATCH 1/8 is not strictly necessary to make the test run well on testbot, and can possibly be dropped, but it helped me when running the test in a VM, especially when multiple processes are involved (the new process windows couldn't get foreground and were just flashing in the taskbar instead). The remaining todo_wine after the series are caused by the same kind of behavior as is fixed by PATCH 8/8, where WM_KILLFOCUS/WM_SETFOCUS messages aren't sent because the window already has focus. As foreground was temporarily stolen by another thread they should still apparently be sent. The logic was a little bit more complicated so I left it aside. The remaining focus issues with Wine are mostly caused by: * asynchronous SetForegroundWindow consequences, with user actions interleaved, causing events coming late to steal foreground that was already lost, * threads not checking their messages for a while, causing a delay between host focus changes, and wineserver focus state updates. My plan, to fix those, would be to: * use timestamps to order SetForegroundWindow requests and their corresponding host events, and then ignore out-of-order events, * track the host focus state globally, for instance from explorer, so that threads that do not check their messages for a while will not make wine focus state inconsistent. This inversion makes the whole thing useless, as threads can steal foreground from each other, as they process the internal WM_WINE_SETACTIVEWINDOW messages in the wrong order, so it has to be fixed first. Cheers, Rémi Bernon (8): user32/tests: Attach a debugger to help SetForegroundWindow suceed. user32/tests: Add concurrency tests for SetForegroundWindow. user32: Do not deactivate if thread is foreground. user32: Call set_active_window from internal handler. server: Allow filtering internal sent messages. user32: Add discard_internal parameter to peek_message. user32: Discard internal messages in set_active_window. user32: Send WM_NCACTIVATE on SetForegroundWindow call. dlls/user32/focus.c | 17 +- dlls/user32/message.c | 34 +++- dlls/user32/tests/win.c | 388 +++++++++++++++++++++++++++++++------ dlls/user32/user_private.h | 2 + server/queue.c | 9 +- 5 files changed, 374 insertions(+), 76 deletions(-) -- 2.28.0

3 17

[PATCH 1/2] windowscodecs: Add tests and initial implementation for DdsFrameDecode_CopyPixels().
by Ziqing Hui 14 Aug '20

14 Aug '20

Signed-off-by: Ziqing Hui <zhui(a)codeweavers.com> --- dlls/windowscodecs/ddsformat.c | 57 ++++++++++++++++- dlls/windowscodecs/tests/ddsformat.c | 91 ++++++++++++++++++++++++++-- 2 files changed, 141 insertions(+), 7 deletions(-)

2 1

[PATCH] mshtml: Add trace to IOleCommandTarget::Exec().
by Dmitry Timoshkov 14 Aug '20

14 Aug '20

Signed-off-by: Dmitry Timoshkov <dmitry(a)baikal.ru> --- dlls/mshtml/olecmd.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/dlls/mshtml/olecmd.c b/dlls/mshtml/olecmd.c index 43f9285e8f..a79f2d000e 100644 --- a/dlls/mshtml/olecmd.c +++ b/dlls/mshtml/olecmd.c @@ -917,6 +917,8 @@ static HRESULT WINAPI OleCommandTarget_Exec(IOleCommandTarget *iface, const GUID { HTMLDocument *This = impl_from_IOleCommandTarget(iface); + TRACE("(%p)->(%s %d %d %s %p)\n", This, debugstr_guid(pguidCmdGroup), nCmdID, nCmdexecopt, wine_dbgstr_variant(pvaIn), pvaOut); + if(!This->doc_node->browser) return E_UNEXPECTED; -- 2.26.2

2 1

[PATCH] ieframe: When activating UI also activate the embedded document.
by Dmitry Timoshkov 14 Aug '20

14 Aug '20

Signed-off-by: Dmitry Timoshkov <dmitry(a)baikal.ru> --- dlls/ieframe/dochost.c | 26 ++++++++++++++++++++++++++ dlls/ieframe/ieframe.h | 1 + dlls/ieframe/oleobject.c | 2 ++ 3 files changed, 29 insertions(+) diff --git a/dlls/ieframe/dochost.c b/dlls/ieframe/dochost.c index 46d5cc5dd1..042417b5b1 100644 --- a/dlls/ieframe/dochost.c +++ b/dlls/ieframe/dochost.c @@ -598,6 +598,32 @@ HRESULT refresh_document(DocHost *This, const VARIANT *level) return S_OK; } +void activate_document(DocHost *This) +{ + IOleObject *oleobj; + IHlinkTarget *hlink; + HRESULT hres; + + if(!This->document) return; + + hres = IUnknown_QueryInterface(This->document, &IID_IHlinkTarget, (void**)&hlink); + if(SUCCEEDED(hres)) { + IHlinkTarget_Navigate(hlink, 0, NULL); + IHlinkTarget_Release(hlink); + } + + hres = IUnknown_QueryInterface(This->document, &IID_IOleObject, (void**)&oleobj); + if(SUCCEEDED(hres)) { + IOleObject_DoVerb(oleobj, OLEIVERB_SHOW, NULL, NULL, -1, 0, NULL); + IOleObject_Release(oleobj); + } + + refresh_document(This, NULL); + + if(This->view) + IOleDocumentView_UIActivate(This->view, TRUE); +} + void release_dochost_client(DocHost *This) { if(This->hwnd) { diff --git a/dlls/ieframe/ieframe.h b/dlls/ieframe/ieframe.h index 24d490c272..a35507e6fb 100644 --- a/dlls/ieframe/ieframe.h +++ b/dlls/ieframe/ieframe.h @@ -283,6 +283,7 @@ HRESULT set_dochost_url(DocHost*,const WCHAR*) DECLSPEC_HIDDEN; void handle_navigation_error(DocHost*,HRESULT,BSTR,IHTMLWindow2*) DECLSPEC_HIDDEN; HRESULT dochost_object_available(DocHost*,IUnknown*) DECLSPEC_HIDDEN; void set_doc_state(DocHost*,READYSTATE) DECLSPEC_HIDDEN; +void activate_document(DocHost*) DECLSPEC_HIDDEN; void deactivate_document(DocHost*) DECLSPEC_HIDDEN; void create_doc_view_hwnd(DocHost*) DECLSPEC_HIDDEN; void on_commandstate_change(DocHost*,LONG,BOOL) DECLSPEC_HIDDEN; diff --git a/dlls/ieframe/oleobject.c b/dlls/ieframe/oleobject.c index a2f87dda2f..be0e6ec5bf 100644 --- a/dlls/ieframe/oleobject.c +++ b/dlls/ieframe/oleobject.c @@ -230,6 +230,8 @@ static HRESULT activate_ui(WebBrowser *This, IOleClientSite *active_site) SetFocus(This->shell_embedding_hwnd); notify_on_focus(This, TRUE); + activate_document(&This->doc_host); + This->ui_activated = TRUE; return S_OK; -- 2.26.2

2 1

[PATCH v2] xactengine3_7: Implement IXACT3Engine PrepareWave
by Alistair Leslie-Hughes 14 Aug '20

14 Aug '20

v2: Deal with FACTAudioEngine_PrepareWave isn't implemented. Wine-bug: https://bugs.winehq.org/show_bug.cgi?id=49689 Signed-off-by: Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> --- dlls/xactengine3_7/xact_dll.c | 32 ++++++++++++++++++++++++++++++-- 1 file changed, 30 insertions(+), 2 deletions(-) diff --git a/dlls/xactengine3_7/xact_dll.c b/dlls/xactengine3_7/xact_dll.c index c5f96decb72..5bcd8a6202c 100644 --- a/dlls/xactengine3_7/xact_dll.c +++ b/dlls/xactengine3_7/xact_dll.c @@ -986,8 +986,36 @@ static HRESULT WINAPI IXACT3EngineImpl_PrepareWave(IXACT3Engine *iface, IXACT3Wave **ppWave) { XACT3EngineImpl *This = impl_from_IXACT3Engine(iface); - FIXME("(%p): stub!\n", This); - return E_NOTIMPL; + XACT3WaveImpl *wave; + FACTWave *fwave = NULL; + UINT ret; + + TRACE("(%p)->(0x%08x, %s, %d, %d, %d, %d, %p)\n", This, dwFlags, debugstr_a(szWavePath), + wStreamingPacketSize, dwAlignment, dwPlayOffset, nLoopCount, ppWave); + + ret = FACTAudioEngine_PrepareWave(This->fact_engine, dwFlags, szWavePath, wStreamingPacketSize, + dwAlignment, dwPlayOffset, nLoopCount, &fwave); + if(ret != 0 || !fwave) + { + ERR("Failed to CreateWave: %d (%p)\n", ret, fwave); + return E_FAIL; + } + + wave = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(*wave)); + if (!wave) + { + FACTWave_Destroy(fwave); + ERR("Failed to allocate XACT3WaveImpl!"); + return E_OUTOFMEMORY; + } + + wave->IXACT3Wave_iface.lpVtbl = &XACT3Wave_Vtbl; + wave->fact_wave = fwave; + *ppWave = &wave->IXACT3Wave_iface; + + TRACE("Created Wave: %p\n", wave); + + return S_OK; } enum { -- 2.28.0

1 0

[PATCH 4/6] kernel32: Use IOCTL_CONDRV_GET_INPUT_INFO in GetConsoleWindow.
by Jacek Caban 14 Aug '20

14 Aug '20

Signed-off-by: Jacek Caban <jacek(a)codeweavers.com> --- dlls/kernel32/console.c | 16 ++++++---------- dlls/kernel32/tests/console.c | 6 ++++++ 2 files changed, 12 insertions(+), 10 deletions(-)

2 1