July 2023 - wine-gitlab

[PATCH v4 0/5] MR280: vkd3d-shader: sm1 semantic matching.
by Zebediah Figura (＠zfigura) 31 Jul '23

31 Jul '23

This is probably the most ugly and controversial bit of API. I don't really know if this is the right approach to solving this. sm4 registers match by register index, such that shaders can mostly be compiled in isolation. sm1 does not—registers may be specified in any order in the vertex and pixel shaders, and will be matched by usage and usage index. By itself this is not much of a problem. Where it gets hairy is that we want to do some degree of caching, as well as pre-compilation, and avoid recompiling either shader every time it's matched with a new one. Wine currently deals with this problem, for GLSL, by generating a "main" GLSL shader for the vertex shader, and then an extra function setup_vs_output(), and linking the two together every time a new pixel shader is used. This could in theory be used for SPIR-V, but it requires the use of extra, probably external, code to link SPIR-V shaders together, which I do not particularly anticipate being well-received. (I'm not sure how Wine deals with this problem in the ARB backend. It seems to take the pixel shader signature into account when generating the vertex shader— cf. init_output_registers()—but it doesn't take it into account when looking up a vertex shader variant? I didn't look too closely at the code, so maybe I'm missing something.) -- The vkd3d parts of this patch are quite straightforward, and looking at them, I think the design is quite intuitive in isolation. There may be some room for internal refactoring (in particular with an eye to not so much overloading the "register_index" field of struct shader_signature_element) but I'm relatively happy with the way it turned out. In isolation, that is. The Wine part is worse. I've uploaded branches for vkd3d and Wine that use this API, and correctly handle shaders with some nontrivial reordering: https://gitlab.winehq.org/zfigura/vkd3d/-/commits/himavant5 https://gitlab.winehq.org/zfigura/wine/-/commits/himavant_cb The test can be run, as before, like so: make tests/shader_runner.exe && WINE_D3D_CONFIG=renderer=vulkan wine tests/shader_runner.exe ../vkd3d/tests/hlsl/sm1-interstage-interface.shader_test The interesting Wine parts are concentrated in a single patch, 5cfb9d930f11e. The patch takes a few shortcuts, partly because I wanted not to block the vkd3d API design questions, but also because while writing it I came up with a couple of problems that I wasn't sure how to fix. There are two main problems I see: (1) This patch has the user pass the signature from the pixel shader when compiling the vertex shader, and looks up register indices already arbitrarily allocated by the pixel shader. This is problematic when trying to use this signature as a cache key, by virtue of it not being clear (or even defined) which fields are key elements and which aren't. It's also not particularly kind to lookup, on account of not being directly comparable with memcmp(). There are a few options I see: (a) Provide an internal function to compare key elements. This feels... odd, like a very special-purpose function, but perhaps workable. (b) Just make the user deal with it, and assert that all fields are key elements. (c) Use some alternative, perhaps shortened structure as a field of vkd3d_shader_next_stage_info. This has the disadvantage that it is not as simple for a hypothetical user to retrieve from the pixel shader, but we would presumably provide a function to generate one from a shader signature. This would probably also be kinder to cache lookup if it's shorter. (d) Make caching vkd3d's responsibility, to some degree. This seems daunting, but the more we optimize, the more difficult it may be to design API that allows for nice caching. (2) Assuming we use signatures, there is a memory management problem that 5cfb9d930f11e spells out. This is probably a matter of "just fix it", but I suppose another option is to take the GLSL or ARB architecture. -- April is the cruellest month, breeding Lilacs out of the dead land, mixing Memory and desire, stirring Dull roots with spring rain. -- v4: vkd3d-shader: Introduce a function to build a varying map between sm1 stages. https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/280

3 7

[PATCH v46 0/7] MR2786: Add support for AF_UNIX sockets
by Ally Sommers (＠dropbear) 31 Jul '23

31 Jul '23

Windows 10 [received support](https://devblogs.microsoft.com/commandline/af_unix-comes-to-window… for AF_UNIX sockets in Insider Build 17063. This merge request adds basic support for AF_UNIX sockets to ws2_32 and wineserver. Of particular note is the difficulty in handling `sun_path`. Most of the functions that allow for translating Windows paths to Unix paths are not accessible from ws2_32. I considered the following options: * Pass the Windows path to wineserver and do the conversion there. * This is, as far as I can tell, not possible without major rearchitecting. wineserver does not have functions to translate Windows paths to Unix paths, for obvious reasons. * Obtain the current working directory of the requesting process and temporarily change directories to there. * This only handles relative paths and fails for absolute paths, UNC paths, etc. * Conditionally change directories based on whether the path is relative or not. * This is error-prone and wineserver does not have the requisite functions to do this cleanly. I ultimately decided to pass the translated Unix path to wineserver, which changes directories to `dirname(path)`. It then provides `bind` and `connect` with `basename(path)`. This is not threadsafe, but wineserver is not (currently) multithreaded. Abstract sockets are not fully supported by this patch, matching the behavior of Windows. -- v46: ws2_32/tests: Add test for AF_UNIX sockets fix. server: Fix getsockname() and accept() on AF_UNIX sockets. server: Introduce error when attempting to create a SOCK_DGRAM AF_UNIX socket. server: Allow for deletion of socket files. ws2_32: Add support for AF_UNIX sockets. ntdll/unix: Add support for AF_UNIX sockets to multiple functions. ws2_32: Add afunix.h header. https://gitlab.winehq.org/wine/wine/-/merge_requests/2786

3 8

[PATCH v3 0/5] MR280: vkd3d-shader: sm1 semantic matching.
by Zebediah Figura (＠zfigura) 31 Jul '23

31 Jul '23

This is probably the most ugly and controversial bit of API. I don't really know if this is the right approach to solving this. sm4 registers match by register index, such that shaders can mostly be compiled in isolation. sm1 does not—registers may be specified in any order in the vertex and pixel shaders, and will be matched by usage and usage index. By itself this is not much of a problem. Where it gets hairy is that we want to do some degree of caching, as well as pre-compilation, and avoid recompiling either shader every time it's matched with a new one. Wine currently deals with this problem, for GLSL, by generating a "main" GLSL shader for the vertex shader, and then an extra function setup_vs_output(), and linking the two together every time a new pixel shader is used. This could in theory be used for SPIR-V, but it requires the use of extra, probably external, code to link SPIR-V shaders together, which I do not particularly anticipate being well-received. (I'm not sure how Wine deals with this problem in the ARB backend. It seems to take the pixel shader signature into account when generating the vertex shader— cf. init_output_registers()—but it doesn't take it into account when looking up a vertex shader variant? I didn't look too closely at the code, so maybe I'm missing something.) -- The vkd3d parts of this patch are quite straightforward, and looking at them, I think the design is quite intuitive in isolation. There may be some room for internal refactoring (in particular with an eye to not so much overloading the "register_index" field of struct shader_signature_element) but I'm relatively happy with the way it turned out. In isolation, that is. The Wine part is worse. I've uploaded branches for vkd3d and Wine that use this API, and correctly handle shaders with some nontrivial reordering: https://gitlab.winehq.org/zfigura/vkd3d/-/commits/himavant5 https://gitlab.winehq.org/zfigura/wine/-/commits/himavant_cb The test can be run, as before, like so: make tests/shader_runner.exe && WINE_D3D_CONFIG=renderer=vulkan wine tests/shader_runner.exe ../vkd3d/tests/hlsl/sm1-interstage-interface.shader_test The interesting Wine parts are concentrated in a single patch, 5cfb9d930f11e. The patch takes a few shortcuts, partly because I wanted not to block the vkd3d API design questions, but also because while writing it I came up with a couple of problems that I wasn't sure how to fix. There are two main problems I see: (1) This patch has the user pass the signature from the pixel shader when compiling the vertex shader, and looks up register indices already arbitrarily allocated by the pixel shader. This is problematic when trying to use this signature as a cache key, by virtue of it not being clear (or even defined) which fields are key elements and which aren't. It's also not particularly kind to lookup, on account of not being directly comparable with memcmp(). There are a few options I see: (a) Provide an internal function to compare key elements. This feels... odd, like a very special-purpose function, but perhaps workable. (b) Just make the user deal with it, and assert that all fields are key elements. (c) Use some alternative, perhaps shortened structure as a field of vkd3d_shader_next_stage_info. This has the disadvantage that it is not as simple for a hypothetical user to retrieve from the pixel shader, but we would presumably provide a function to generate one from a shader signature. This would probably also be kinder to cache lookup if it's shorter. (d) Make caching vkd3d's responsibility, to some degree. This seems daunting, but the more we optimize, the more difficult it may be to design API that allows for nice caching. (2) Assuming we use signatures, there is a memory management problem that 5cfb9d930f11e spells out. This is probably a matter of "just fix it", but I suppose another option is to take the GLSL or ARB architecture. -- April is the cruellest month, breeding Lilacs out of the dead land, mixing Memory and desire, stirring Dull roots with spring rain. -- v3: vkd3d-shader: Introduce a function to build a varying map between sm1 stages. vkd3d-shader/spirv: Make output varyings not consumed by the next stage private variables. vkd3d-shader: Implement remapping shader output registers to match the next shader's semantics. vkd3d-shader: Add a separate field for the target location of a signature element. vkd3d-shader/d3dbc: Make sure all inter-stage varyings have a unique register index. https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/280

2 5

[PATCH 0/2] MR3456: comctl32: Add support for MSAA events on tab controls.
by Esme Povirk (＠madewokherd) 31 Jul '23

31 Jul '23

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/3456

4 7

Re: [PATCH v2 0/2] MR3431: wined3d: Fix clears of the current render target in Vulkan
by Henri Verbeet (＠hverbeet) 31 Jul '23

31 Jul '23

Mostly just a couple of style comments; feel free to disregard if you like: The subject line, comments, and ok messages are missing full stops. Personally, I'm not a fan of the switch/case style of tests. My preference would be to either just explicitly write out the tests, or to encode them using the "tests[]" construction we use in a couple of other tests. > ```diff > + static const D3D11_VIEWPORT viewport = {0, 0, 2, 1, 0, 1}; > ... > + texture_desc.Width = (UINT)viewport.Width; > + texture_desc.Height = (UINT)viewport.Height; > ... > + ID3D11DeviceContext_RSSetViewports(context, 1, &viewport); > ``` That works, but more typically we'd initialise "texture_desc", and then call "set_viewport(..., texture_desc.Width, texture_desc.Height, ...)". > ```diff > + int test; > ``` "test" can't be negative, right? > ```diff > + unsigned int i; > + static const UINT vertex_stride = sizeof(depth_quarter_quad_data[0]); > + static const UINT vertex_offset = 0; > + static const struct vec4 initial_color = {0, 0, 0, 1}; > + static const float clear_color[] = {0, 1, 0, 1}; > + DWORD expected_color[] = {0xff000000, 0xff000000}; > + FLOAT expected_depth[] = {0.5f, 0.5f}; > ``` The choice for "FLOAT" seems odd, given that get_readback_float() returns a float. We don't tend to use UINT and DWORD a lot either. > ```diff > + for (i = 0; i < ARRAYSIZE(expected_color); i++) > ``` It's not wrong, but all the existing code uses ARRAY_SIZE... > ```diff > + (!damavand && test == 3 && i == 0) /* Draw with no PS doesn't write color on D3D11 but does on OpenGL */ > ``` I don't think the comment is terribly clear, but it sounds like we're getting fixed-function with the GL backend, and things just happen to work with the Vulkan backend because we don't yet have a fixed-function implementation there. That likely won't last forever though. For comparison, vkd3d handles this explicitly in d3d12_pipeline_state_init_graphics(). > ```diff > + ok(depth == expected_depth[i], "Test %d: Got unexpected depth %.2f at %u\n", test, depth, i); > ``` I don't think %.2f is quite sufficient to show what's going on if the result is only off by a couple of ulps. That's perhaps not terribly likely for clears with carefully chosen depth values, but it still looks a little sketchy to me. > ```diff > +static void wined3d_context_prepare_used_framebuffer(struct wined3d_rendertarget_view *rtv, struct wined3d_context *context) > +{ > + if (wined3d_rendertarget_view_get_locations(rtv) & WINED3D_LOCATION_CLEARED) > + { > + /* Need to rebind framebuffer or the clear won't happen */ > + context_invalidate_state(context, STATE_FRAMEBUFFER); > + wined3d_rendertarget_view_prepare_location(rtv, context, rtv->resource->draw_binding); > + } > + else > + { > + wined3d_rendertarget_view_load_location(rtv, context, rtv->resource->draw_binding); > + } > +} > ``` The comment doesn't seem quite as helpful as it perhaps could be; if I understand the issue correctly, we don't actually care about rebinding the framebuffer, but instead about starting a new render pass, because that's where we'll clear these. That's also what the commit message seems to suggest. By convention, "context" would be the first parameter for wined3d_context_*() functions. The function name is also slightly odd, because it's preparing/loading individual views, not complete framebuffers. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/3431#note_40852

1 0

[PATCH v3 0/2] MR3461: gdiplus: Reject zero-width/zero-height rectangles
by David Kahurani (＠kahurani) 31 Jul '23

31 Jul '23

Works around the issue manifesting here : https://bugs.winehq.org/show_bug.cgi?id=55351 The MR also contains a patch that refactors the related tests a bit. Signed-off-by: David Kahurani k.kahurani(a)gmail.com -- v3: gdiplus: Move Flatten() tests into the same function gdiplus: Reject zero-width/zero-height rectangles https://gitlab.winehq.org/wine/wine/-/merge_requests/3461

2 3

[PATCH v2 0/2] MR3461: gdiplus: Reject zero-width/zero-height rectangles
by David Kahurani (＠kahurani) 31 Jul '23

31 Jul '23

Works around the issue manifesting here : https://bugs.winehq.org/show_bug.cgi?id=55351 The MR also contains a patch that refactors the related tests a bit. Signed-off-by: David Kahurani k.kahurani(a)gmail.com -- v2: gdiplus: Move Flatten() tests into the same function gdiplus: Reject zero-width/zero-height rectangles https://gitlab.winehq.org/wine/wine/-/merge_requests/3461

4 11

[PATCH 0/8] MR3464: win32u: Use user message packing for more messages. (part 4)
by Jacek Caban (＠jacek) 31 Jul '23

31 Jul '23

More of !3388. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/3464

3 9

[PATCH v8 0/1] MR3392: comctl32/listview: Reset bNoItemMetrics to TRUE to make sure that listview items can be arranged correctly..
by Zhao Yi (＠Zhaoyi) 31 Jul '23

31 Jul '23

Same as MR !2394 Signed-off-by: Zhao Yi zhaoyi(a)uniontech.com -- v8: comctl32/listview: Reset bNoItemMetrics to TRUE to make sure that listview items can be arranged correctly. https://gitlab.winehq.org/wine/wine/-/merge_requests/3392

4 5

[PATCH 0/1] MR3443: jscript: Call the getter with the proper 'this' in invoke_prop_func.
by Gabriel Ivăncescu (＠insn) 31 Jul '23

31 Jul '23

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/3443

3 4

2025

2024

2023

2022

wine-gitlab July 2023