April 2016 - Wine-Devel - list.winehq.org

Re: [PATCH 05/10] wined3d: Try to avoid glGetTexImage() for 2D array textures.

by Józef Kucia

On Wed, Apr 20, 2016 at 10:34 AM, Henri Verbeet <hverbeet(a)gmail.com> wrote: > Well, yes. But it's not up to surface_download_data() to decide > whether to use GetTexImage() or ReadPixels(). surface_load_sysmem() > should decide if it's more appropriate to use surface_download_data() > or read_from_framebuffer(). > > That said, I'm not sure read_from_framebuffer() is necessarily an > improvement in terms of (overall) performance. True, GetTexImage() > reads more data than it needs to, but I don't think it's a given that > that's much more expensive than reading a smaller amount of data > unless the texture is particularly large. In cases where you're going > to access the other sub-resources afterwards anyway it may even be an > improvement, provided resource locations etc. are properly updated. On > the other hand, read_from_framebuffer() calls > context_apply_blit_state(), which does a lot of state invalidation. If > multiple of those are batched together that may be ok (but in that > case perhaps surface_download_data() would be too), but if they're > interleaved with draws it may not be an improvement. > > I.e., it would be a lot easier to justify this if you could show an > application where it helps. I don't have any application to show that this helps. I'll drop the patch. Is patch 121528 ok in its current form? It doesn't feel like a good idea messing with other surfaces in surface_download_data().

April 20, 2016

Re: [PATCH 05/10] wined3d: Try to avoid glGetTexImage() for 2D array textures.

by Henri Verbeet

On 19 April 2016 at 12:17, Józef Kucia <joseph.kucia(a)gmail.com> wrote: > The issue is if we do not have ARB_get_texture_sub_image we use > glGet[Compressed]TexImage() to download the data from texture. This > type of download is implemented in the patch 6. I want to avoid > glGet[Compressed]TexImage() for array textures because these functions > don't allow to download the data for a single wined3d surface. They > allow to download data for the whole miplevel slice. In the patch 6 a > block of memory is allocated, the whole miplevel slice is downloaded > and the data for a single layer is copied to the current wined3d > surface. > > The idea in this patch is to simulate glGetTextureSubImage() using > reads from the framebuffer. An array array texture is attached to > framebuffer using glFramebufferTextureLayer() and the data for a > single wined3d surface is downloaded using glReadPixels(). This works > in my d3d11 tests which I plan to submit in the following patchset. Well, yes. But it's not up to surface_download_data() to decide whether to use GetTexImage() or ReadPixels(). surface_load_sysmem() should decide if it's more appropriate to use surface_download_data() or read_from_framebuffer(). That said, I'm not sure read_from_framebuffer() is necessarily an improvement in terms of (overall) performance. True, GetTexImage() reads more data than it needs to, but I don't think it's a given that that's much more expensive than reading a smaller amount of data unless the texture is particularly large. In cases where you're going to access the other sub-resources afterwards anyway it may even be an improvement, provided resource locations etc. are properly updated. On the other hand, read_from_framebuffer() calls context_apply_blit_state(), which does a lot of state invalidation. If multiple of those are batched together that may be ok (but in that case perhaps surface_download_data() would be too), but if they're interleaved with draws it may not be an improvement. I.e., it would be a lot easier to justify this if you could show an application where it helps.

April 20, 2016

Re: [3/3] services: Track services and processes separately.

by Sebastian Lackner

On 19.04.2016 15:38, Sebastian Lackner wrote: > On 19.04.2016 14:25, Alexandre Julliard wrote: >> Sebastian Lackner <sebastian(a)fds-team.de> writes: >> >>> Signed-off-by: Sebastian Lackner <sebastian(a)fds-team.de> >>> --- >>> programs/services/rpc.c | 37 ++++--- >>> programs/services/services.c | 199 ++++++++++++++++++++++++++----------------- >>> programs/services/services.h | 4 >>> 3 files changed, 147 insertions(+), 93 deletions(-) >> >> It breaks the tests: >> >> ../../../tools/runtest -q -P wine -T ../../.. -M taskschd.dll -p taskschd_test.exe.so scheduler && touch scheduler.ok >> scheduler.c:78: Test failed: Connect error 0x80041315 >> scheduler.c:101: Test failed: Connect error 0x80041315 >> scheduler.c:151: Test failed: Connect error 0x80041315 >> scheduler.c:162: Test failed: GetFolder error 0x800704e3 >> wine: Unhandled page fault on read access to 0x00000000 at address 0x7eff3a5e (thread 0009), starting debugger... >> > > Hm, it didn't fail here, but I'll take a look. Thanks. > I found the issue, will send a fixed version soon.

April 20, 2016

Re: [PATCH 1/5] dwrite/tests: Added GetFontSignature() test (resend)

by Nikolay Sivov

On 20.04.2016 10:12, Sebastian Lackner wrote: > On 19.04.2016 19:10, Nikolay Sivov wrote: >> + ok(fontsig.fsUsb[3] == expected_signature.fsUsb[3], "%s: fsUsb[3] %#x, expected %#x\n", wine_dbgstr_w(nameW), >> + fontsig.fsUsb[0], expected_signature.fsUsb[3]); > ^^^ > > There is a small typo here. > Thanks, both problems are fixed now.

April 20, 2016

Re: msi/tests: Spelling fixes for some action and package ok() messages.

by Marvin

Hi, While running your changed tests on Windows, 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=22354 Your paranoid android. === wvistau64 (32 bit action) === The test timed out

April 20, 2016

Re: [PATCH 8/8] d3d11/tests: Add a geometry shader test.

by Józef Kucia

On Tue, Apr 19, 2016 at 6:56 PM, Matteo Bruni <mbruni(a)codeweavers.com> wrote: > Signed-off-by: Matteo Bruni <mbruni(a)codeweavers.com> > --- > dlls/d3d11/tests/d3d11.c | 232 +++++++++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 232 insertions(+) > > diff --git a/dlls/d3d11/tests/d3d11.c b/dlls/d3d11/tests/d3d11.c > index 713c9cf..0fafaaf 100644 > --- a/dlls/d3d11/tests/d3d11.c > +++ b/dlls/d3d11/tests/d3d11.c > @@ -6569,6 +6569,237 @@ static void test_getdc(void) > ok(!refcount, "Device has %u references left.\n", refcount); > } > > +static void test_geometry_shader(void) > +{ > + static const struct > + { > + struct vec4 position; > + unsigned int color; > + } > + vertex[] = > + { > + {{0.0f, 0.0f, 1.0f, 1.0f}, 0xffffff00}, > + }; > + static const D3D11_INPUT_ELEMENT_DESC layout_desc[] = > + { > + {"SV_POSITION", 0, DXGI_FORMAT_R32G32B32A32_FLOAT, 0, 0, D3D10_INPUT_PER_VERTEX_DATA, 0}, > + {"COLOR", 0, DXGI_FORMAT_R8G8B8A8_UNORM, 0, 16, D3D10_INPUT_PER_VERTEX_DATA, 0}, It would be better to use D3D11 constants. > + static const D3D_FEATURE_LEVEL feature_level = D3D_FEATURE_LEVEL_10_0; Why should we limit this test to feature level 10_0? > + buffer_desc.ByteWidth = sizeof(vertex); > + buffer_desc.Usage = D3D11_USAGE_DEFAULT; > + buffer_desc.BindFlags = D3D11_BIND_VERTEX_BUFFER; > + buffer_desc.CPUAccessFlags = 0; > + buffer_desc.MiscFlags = 0; The "StructureByteStride" field is uninitialized. I suppose it doesn't matter in practice because this field is used only for structured buffers.

April 20, 2016

Re: [PATCH 2/5] dwrite/tests: Some basic tests for fontface references

by Sebastian Lackner

On 19.04.2016 19:10, Nikolay Sivov wrote: > + hr = IDWriteFactory3_CreateFontFaceReference_(factory3, file, 0, DWRITE_FONT_SIMULATIONS_NONE, &ref); > + ok(hr == S_OK, "got 0x%08x\n", hr); > + > + hr = IDWriteFontFaceReference_GetFontFile(ref, &file1); > + ok(hr == S_OK, "got 0x%08x\n", hr); > + ok(file != file1, "got %p, previous file %p\n", file1, file); Also not really critical, but you might be leaking a reference to "ref" here. > + > + IDWriteFontFile_Release(file); > + IDWriteFontFile_Release(file1); > +} > IDWriteFactory3_Release(factory3); > DELETE_FONTFILE(path); > }

April 20, 2016

Re: [PATCH 1/5] dwrite/tests: Added GetFontSignature() test (resend)

by Sebastian Lackner

On 19.04.2016 19:10, Nikolay Sivov wrote: > + ok(fontsig.fsUsb[3] == expected_signature.fsUsb[3], "%s: fsUsb[3] %#x, expected %#x\n", wine_dbgstr_w(nameW), > + fontsig.fsUsb[0], expected_signature.fsUsb[3]); ^^^ There is a small typo here.

April 20, 2016

Re: taskschd/tests: Fix a use-after-free in scheduler tests.

by Sebastian Lackner

On 19.04.2016 23:41, Marvin wrote: > Hi, > > While running your changed tests on Windows, 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=22349 > > Your paranoid android. > > > === w1064 (32 bit scheduler) === > scheduler.c:159: Test failed: expected ERROR_INVALID_NAME, got 0x80070002 > scheduler.c:204: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 > scheduler.c:354: Test failed: expected ERROR_FILE_NOT_FOUND, got 0 > scheduler.c:767: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 > scheduler.c:773: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 > scheduler.c:941: Test failed: expected ERROR_FILE_NOT_FOUND, got 0 > > === w1064 (64 bit scheduler) === > scheduler.c:159: Test failed: expected ERROR_INVALID_NAME, got 0x80070002 > scheduler.c:204: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 > scheduler.c:354: Test failed: expected ERROR_FILE_NOT_FOUND, got 0 > scheduler.c:767: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 > scheduler.c:773: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 > scheduler.c:941: Test failed: expected ERROR_FILE_NOT_FOUND, got 0 > Those are existing failures.

April 19, 2016

Re: taskschd/tests: Fix a use-after-free in scheduler tests.

by Marvin

Hi, While running your changed tests on Windows, 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=22349 Your paranoid android. === w1064 (32 bit scheduler) === scheduler.c:159: Test failed: expected ERROR_INVALID_NAME, got 0x80070002 scheduler.c:204: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 scheduler.c:354: Test failed: expected ERROR_FILE_NOT_FOUND, got 0 scheduler.c:767: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 scheduler.c:773: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 scheduler.c:941: Test failed: expected ERROR_FILE_NOT_FOUND, got 0 === w1064 (64 bit scheduler) === scheduler.c:159: Test failed: expected ERROR_INVALID_NAME, got 0x80070002 scheduler.c:204: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 scheduler.c:354: Test failed: expected ERROR_FILE_NOT_FOUND, got 0 scheduler.c:767: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 scheduler.c:773: Test failed: expected ERROR_PATH_NOT_FOUND, got 0x80070002 scheduler.c:941: Test failed: expected ERROR_FILE_NOT_FOUND, got 0

April 19, 2016