September 2019 - Wine-Devel

Re: [PATCH 5/6] vkd3d: Optimize GPU VA allocator.

by Henri Verbeet

On Mon, 30 Sep 2019 at 17:48, Hans-Kristian Arntzen <post(a)arntzen-software.no> wrote: > The GPU VA allocator was allocating memory in a way where dereferencing > GPU VA required a lock + bsearch to find the right VA range. > > Rather than going this route, we turn the common case into O(1) and > lockless by creating a slab allocator which allows us to lookup a ptr > directly from GPU VA with (VA - Base) / PageSize. > > The number of allocations in the fast path must be limited since we > cannot trivially grow the allocator while remaining lock-free for > dereferences. > I need to review this in more detail, but on first sight it makes sense.

Sept. 30, 2019

Re: [PATCH 4/6] vkd3d: Do not use RESET_COMMAND_BUFFERS_BIT.

by Henri Verbeet

On Mon, 30 Sep 2019 at 17:48, Hans-Kristian Arntzen <post(a)arntzen-software.no> wrote: > Allow the Vulkan command pools to behave like D3D12 command pools, where > memory is owned by the pool rather than individual command buffers. > > Also, do not release resources on vkResetCommandPool, as the intention > from D3D12 docs is to reuse memory rather than freeing it. > > Signed-off-by: Hans-Kristian Arntzen <post(a)arntzen-software.no> > --- > libs/vkd3d/command.c | 9 ++++++--- > 1 file changed, 6 insertions(+), 3 deletions(-) > This one seems fine as well, but it also sounds like this could be split in two individual commits.

Sept. 30, 2019

Re: [PATCH 3/6] vkd3d: Deal with Map out parameter being NULL.

by Henri Verbeet

On Mon, 30 Sep 2019 at 17:48, Hans-Kristian Arntzen <post(a)arntzen-software.no> wrote: > It is possible to map a resource, but not disclose the VA to caller. > This is used for WriteToSubresource. > > Signed-off-by: Hans-Kristian Arntzen <post(a)arntzen-software.no> > --- > libs/vkd3d/resource.c | 13 +++++++++---- > 1 file changed, 9 insertions(+), 4 deletions(-) > This seems fine. Do you have a regression test to go along with this as well?

Sept. 30, 2019

Re: [PATCH 2/6] vkd3d: Implement support for persistently mapped HOST_VISIBLE heaps.

by Henri Verbeet

On Mon, 30 Sep 2019 at 17:48, Hans-Kristian Arntzen <post(a)arntzen-software.no> wrote: > Greatly reduces number of maps for applications which do not do this > themselves. Also add a config option to use older map-on-demand > behavior, which might be nice for debugging tools. > I don't want to just say no, but this does seem to go against the spirit of an "explicit" API like d3d12 somewhat. My other concern with this is that while on 64-bit we typically have plenty of address space, things can be a bit more tight on 32-bit. I assume the main justification for this patch is performance; do you have some numbers?

Sept. 30, 2019

Re: [PATCH 1/6] vkd3d: Enable MSVC to compile.

by Henri Verbeet

Hi Hans-Kristian, >From a cursory look, most of this seems fine, but see what Chip said about e.g. vkd3d_log2i(). Could you please split this into individual patches though? Both in order to make it more reviewable and to get more useful bisects. Wrt. the pthread wrappers, it seems tempting to introduce internal functions along the lines of vkd3d_create_thread(), vkd3d_join_thread(), etc. Henri

Sept. 30, 2019

Re: [PATCH] scrrun: Add support for TristateUseDefault text stream format.

by Marvin

Hi, While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=57159 Your paranoid android. === w2003std (32 bit report) === scrrun: filesystem.c:1900: Test failed: got 0x00000000 filesystem.c:1923: Test failed: got 0x00000000 filesystem.c:1945: Test failed: got 0x00000000 filesystem.c:1963: Test failed: got 0x00000000 filesystem.c:1984: Test failed: got 0x00000000 filesystem.c:2006: Test failed: got 0x00000000

Sept. 30, 2019

[PATCH v2] dmime: Improved tracing of the DMUS_OBJECTDESC structure.

by Michael Stefaniuc

From: Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> sprintf corupted the output string thus making the output useless. Use TRACE to print each of the member variables. One place used debugstr_DMUS_OBJECTDESC but never printed it. Signed-off-by: Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com> Signed-off-by: Michael Stefaniuc <mstefani(a)winehq.org> --- Some cleanups (whitespace, LPJUNK). dlls/dmime/audiopath.c | 2 +- dlls/dmime/dmutils.c | 54 ++++++++++++++++++++++++------------------ dlls/dmime/dmutils.h | 2 +- dlls/dmime/graph.c | 2 +- dlls/dmime/segment.c | 2 +- 5 files changed, 35 insertions(+), 27 deletions(-) diff --git a/dlls/dmime/audiopath.c b/dlls/dmime/audiopath.c index 21aaa4d412..ac3121c0e0 100644 --- a/dlls/dmime/audiopath.c +++ b/dlls/dmime/audiopath.c @@ -285,7 +285,7 @@ static HRESULT WINAPI path_IDirectMusicObject_ParseDescriptor(IDirectMusicObject desc->guidClass = CLSID_DirectMusicAudioPathConfig; desc->dwValidData |= DMUS_OBJ_CLASS; - TRACE("returning descriptor:\n%s\n", debugstr_DMUS_OBJECTDESC (desc)); + dump_DMUS_OBJECTDESC(desc); return S_OK; } diff --git a/dlls/dmime/dmutils.c b/dlls/dmime/dmutils.c index 121e6dc5a4..7b3bcec079 100644 --- a/dlls/dmime/dmutils.c +++ b/dlls/dmime/dmutils.c @@ -41,6 +41,7 @@ #include "dmutils.h" WINE_DEFAULT_DEBUG_CHANNEL(dmfile); +WINE_DECLARE_DEBUG_CHANNEL(dmime); static HRESULT IDirectMusicUtils_IPersistStream_ParseDescGeneric (DMUS_PRIVATE_CHUNK* pChunk, IStream* pStm, LPDMUS_OBJECTDESC pDesc) { @@ -143,7 +144,7 @@ HRESULT IDirectMusicUtils_IPersistStream_ParseReference (LPPERSISTSTREAM iface, TRACE("** DM Reference Begin of Load ***\n"); TRACE("With Desc:\n"); - debugstr_DMUS_OBJECTDESC(&ref_desc); + dump_DMUS_OBJECTDESC(&ref_desc); { LPDIRECTMUSICGETLOADER pGetLoader = NULL; @@ -430,26 +431,33 @@ static const char *debugstr_DMUS_OBJ_FLAGS (DWORD flagmask) { return debugstr_flags(flagmask, flags, ARRAY_SIZE(flags)); } -const char *debugstr_DMUS_OBJECTDESC (LPDMUS_OBJECTDESC pDesc) { - if (pDesc) { - char buffer[1024], *ptr = buffer; - - ptr += sprintf(ptr, "DMUS_OBJECTDESC (%p):\n", pDesc); - ptr += sprintf(ptr, " - dwSize = 0x%08X\n", pDesc->dwSize); - ptr += sprintf(ptr, " - dwValidData = 0x%08X ( %s)\n", pDesc->dwValidData, debugstr_DMUS_OBJ_FLAGS (pDesc->dwValidData)); - if (pDesc->dwValidData & DMUS_OBJ_CLASS) ptr += sprintf(ptr, " - guidClass = %s\n", debugstr_dmguid(&pDesc->guidClass)); - if (pDesc->dwValidData & DMUS_OBJ_OBJECT) ptr += sprintf(ptr, " - guidObject = %s\n", debugstr_guid(&pDesc->guidObject)); - if (pDesc->dwValidData & DMUS_OBJ_DATE) ptr += sprintf(ptr, " - ftDate = %s\n", debugstr_filetime (&pDesc->ftDate)); - if (pDesc->dwValidData & DMUS_OBJ_VERSION) ptr += sprintf(ptr, " - vVersion = %s\n", debugstr_dmversion(&pDesc->vVersion)); - if (pDesc->dwValidData & DMUS_OBJ_NAME) ptr += sprintf(ptr, " - wszName = %s\n", debugstr_w(pDesc->wszName)); - if (pDesc->dwValidData & DMUS_OBJ_CATEGORY) ptr += sprintf(ptr, " - wszCategory = %s\n", debugstr_w(pDesc->wszCategory)); - if (pDesc->dwValidData & DMUS_OBJ_FILENAME) ptr += sprintf(ptr, " - wszFileName = %s\n", debugstr_w(pDesc->wszFileName)); - if (pDesc->dwValidData & DMUS_OBJ_MEMORY) ptr += sprintf(ptr, " - llMemLength = 0x%s\n - pbMemData = %p\n", - wine_dbgstr_longlong(pDesc->llMemLength), pDesc->pbMemData); - if (pDesc->dwValidData & DMUS_OBJ_STREAM) ptr += sprintf(ptr, " - pStream = %p\n", pDesc->pStream); - - return wine_dbg_sprintf("%s", buffer); - } else { - return wine_dbg_sprintf("(NULL)"); - } +void dump_DMUS_OBJECTDESC(DMUS_OBJECTDESC *pDesc) +{ + if (!TRACE_ON(dmime)) + return; + + if (pDesc) { + TRACE_(dmime)("DMUS_OBJECTDESC (%p):", pDesc); + TRACE_(dmime)(" - dwSize = 0x%08X\n", pDesc->dwSize); + TRACE_(dmime)(" - dwValidData = 0x%08X ( %s)\n", pDesc->dwValidData, debugstr_DMUS_OBJ_FLAGS (pDesc->dwValidData)); + if (pDesc->dwValidData & DMUS_OBJ_CLASS) + TRACE_(dmime)(" - guidClass = %s\n", debugstr_dmguid(&pDesc->guidClass)); + if (pDesc->dwValidData & DMUS_OBJ_OBJECT) + TRACE_(dmime)(" - guidObject = %s\n", debugstr_guid(&pDesc->guidObject)); + if (pDesc->dwValidData & DMUS_OBJ_DATE) + TRACE_(dmime)(" - ftDate = %s\n", debugstr_filetime (&pDesc->ftDate)); + if (pDesc->dwValidData & DMUS_OBJ_VERSION) + TRACE_(dmime)(" - vVersion = %s\n", debugstr_dmversion(&pDesc->vVersion)); + if (pDesc->dwValidData & DMUS_OBJ_NAME) + TRACE_(dmime)(" - wszName = %s\n", debugstr_w(pDesc->wszName)); + if (pDesc->dwValidData & DMUS_OBJ_CATEGORY) + TRACE_(dmime)(" - wszCategory = %s\n", debugstr_w(pDesc->wszCategory)); + if (pDesc->dwValidData & DMUS_OBJ_FILENAME) + TRACE_(dmime)(" - wszFileName = %s\n", debugstr_w(pDesc->wszFileName)); + if (pDesc->dwValidData & DMUS_OBJ_MEMORY) + TRACE_(dmime)(" - llMemLength = 0x%s\n - pbMemData = %p\n", wine_dbgstr_longlong(pDesc->llMemLength), + pDesc->pbMemData); + if (pDesc->dwValidData & DMUS_OBJ_STREAM) + TRACE_(dmime)(" - pStream = %p\n", pDesc->pStream); + } } diff --git a/dlls/dmime/dmutils.h b/dlls/dmime/dmutils.h index 187c9ed3d4..7242746031 100644 --- a/dlls/dmime/dmutils.h +++ b/dlls/dmime/dmutils.h @@ -56,6 +56,6 @@ extern const char *debugstr_fourcc (DWORD fourcc) DECLSPEC_HIDDEN; extern const char *debugstr_dmguid (const GUID *id) DECLSPEC_HIDDEN; /* dump whole DMUS_OBJECTDESC struct */ -extern const char *debugstr_DMUS_OBJECTDESC (LPDMUS_OBJECTDESC pDesc) DECLSPEC_HIDDEN; +void dump_DMUS_OBJECTDESC(DMUS_OBJECTDESC *pDesc) DECLSPEC_HIDDEN; #endif /* __WINE_DMUTILS_H */ diff --git a/dlls/dmime/graph.c b/dlls/dmime/graph.c index fafd4ecfff..eebebef4d1 100644 --- a/dlls/dmime/graph.c +++ b/dlls/dmime/graph.c @@ -216,7 +216,7 @@ static HRESULT WINAPI graph_IDirectMusicObject_ParseDescriptor(IDirectMusicObjec desc->guidClass = CLSID_DirectMusicGraph; desc->dwValidData |= DMUS_OBJ_CLASS; - TRACE("returning descriptor:\n%s\n", debugstr_DMUS_OBJECTDESC (desc)); + dump_DMUS_OBJECTDESC(desc); return S_OK; } diff --git a/dlls/dmime/segment.c b/dlls/dmime/segment.c index 9a9872bd4c..a0e6191864 100644 --- a/dlls/dmime/segment.c +++ b/dlls/dmime/segment.c @@ -607,7 +607,7 @@ static HRESULT WINAPI seg_IDirectMusicObject_ParseDescriptor(IDirectMusicObject desc->guidClass = CLSID_DirectMusicSegment; desc->dwValidData |= DMUS_OBJ_CLASS; - TRACE("returning descriptor:\n%s\n", debugstr_DMUS_OBJECTDESC (desc)); + dump_DMUS_OBJECTDESC(desc); return S_OK; } -- 2.20.1

Sept. 30, 2019

Re: [PATCH] msvcrt: Implement %R format for strftime

by Marvin

Hi, While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=57158 Your paranoid android. === w2003std (32 bit report) === msvcrt: time.c:900: Test failed: clock() = 937, expected 1093

Sept. 30, 2019

[PATCH 6/6] vbscript: Add Second function implementation.

by Jacek Caban

Signed-off-by: Jacek Caban <jacek(a)codeweavers.com> --- dlls/vbscript/global.c | 9 +++++++-- dlls/vbscript/tests/api.vbs | 2 ++ 2 files changed, 9 insertions(+), 2 deletions(-)

Sept. 30, 2019

[PATCH 5/6] vbscript: Add Minute function implementation.

by Jacek Caban

Signed-off-by: Jacek Caban <jacek(a)codeweavers.com> --- dlls/vbscript/global.c | 9 +++++++-- dlls/vbscript/tests/api.vbs | 3 +++ 2 files changed, 10 insertions(+), 2 deletions(-)

Sept. 30, 2019