January 2022 - Wine-Devel - list.winehq.org

[PATCH v3 1/6] msvcp140_atomic_wait/tests: Add test for threadpool_work functions.

by Piotr Caban

From: Daniel Lehman <dlehman(a)esri.com> Signed-off-by: Daniel Lehman <dlehman(a)esri.com> Signed-off-by: Piotr Caban <piotr(a)codeweavers.com> --- v3: - fix compilation warnings .../tests/msvcp140_atomic_wait.c | 78 +++++++++++++++++++ 1 file changed, 78 insertions(+)

Jan. 28, 2022

[PATCH vkd3d 2/2] vkd3d: Replace atomic_add_fetch() by InterlockedAdd().

by Alexandre Julliard

By analogy with InterlockedIncrement. It avoids the need for a configure check on Windows platforms. Signed-off-by: Alexandre Julliard <julliard(a)winehq.org> --- include/private/vkd3d_common.h | 13 ++++--------- libs/vkd3d/command.c | 6 +++--- 2 files changed, 7 insertions(+), 12 deletions(-) diff --git a/include/private/vkd3d_common.h b/include/private/vkd3d_common.h index d22c26c098c7..1e19758abc40 100644 --- a/include/private/vkd3d_common.h +++ b/include/private/vkd3d_common.h @@ -201,6 +201,10 @@ static inline LONG InterlockedIncrement(LONG volatile *x) { return __sync_add_and_fetch(x, 1); } +static inline LONG InterlockedAdd(LONG volatile *x, LONG val) +{ + return __sync_add_and_fetch(x, val); +} # else # error "InterlockedIncrement() not implemented for this platform" # endif /* HAVE_SYNC_ADD_AND_FETCH */ @@ -215,15 +219,6 @@ static inline LONG InterlockedDecrement(LONG volatile *x) # endif #endif /* _WIN32 */ -#if HAVE_SYNC_ADD_AND_FETCH -# define atomic_add_fetch(ptr, val) __sync_add_and_fetch(ptr, val) -#elif defined(_MSC_VER) -/* InterlockedAdd returns value after increment, like add_and_fetch. */ -# define atomic_add_fetch(ptr, val) InterlockedAdd(ptr, val) -#else -# error "atomic_add_fetch() not implemented for this platform" -#endif /* HAVE_SYNC_ADD_AND_FETCH */ - static inline void vkd3d_parse_version(const char *version, int *major, int *minor) { *major = atoi(version); diff --git a/libs/vkd3d/command.c b/libs/vkd3d/command.c index 2cf1eba23d94..f1ec6be3fd20 100644 --- a/libs/vkd3d/command.c +++ b/libs/vkd3d/command.c @@ -288,7 +288,7 @@ static void vkd3d_fence_worker_remove_fence(struct vkd3d_fence_worker *worker, s LONG count; int rc; - if (!(count = atomic_add_fetch(&fence->pending_worker_operation_count, 0))) + if (!(count = InterlockedAdd(&fence->pending_worker_operation_count, 0))) return; WARN("Waiting for %u pending fence operations (fence %p).\n", count, fence); @@ -299,7 +299,7 @@ static void vkd3d_fence_worker_remove_fence(struct vkd3d_fence_worker *worker, s return; } - while ((count = atomic_add_fetch(&fence->pending_worker_operation_count, 0))) + while ((count = InterlockedAdd(&fence->pending_worker_operation_count, 0))) { TRACE("Still waiting for %u pending fence operations (fence %p).\n", count, fence); @@ -410,7 +410,7 @@ static void *vkd3d_fence_worker_main(void *arg) { vkd3d_wait_for_gpu_fences(worker); - if (!worker->fence_count || atomic_add_fetch(&worker->enqueued_fence_count, 0)) + if (!worker->fence_count || InterlockedAdd(&worker->enqueued_fence_count, 0)) { if ((rc = pthread_mutex_lock(&worker->mutex))) { -- 2.34.1

Jan. 28, 2022

[PATCH vkd3d 1/2] vkd3d: Always use _BitScanReverse on Windows.

by Alexandre Julliard

Use ULONG instead of unsigned long for non-Mingw msvcrt builds. Signed-off-by: Alexandre Julliard <julliard(a)winehq.org> --- include/private/vkd3d_common.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/include/private/vkd3d_common.h b/include/private/vkd3d_common.h index b7ce9ae4046a..d22c26c098c7 100644 --- a/include/private/vkd3d_common.h +++ b/include/private/vkd3d_common.h @@ -88,10 +88,10 @@ static inline bool vkd3d_bitmask_is_contiguous(unsigned int mask) /* Undefined for x == 0. */ static inline unsigned int vkd3d_log2i(unsigned int x) { -#ifdef _MSC_VER +#ifdef _WIN32 /* _BitScanReverse returns the index of the highest set bit, * unlike clz which is 31 - index. */ - unsigned long result; + ULONG result; _BitScanReverse(&result, x); return (unsigned int)result; #elif defined(HAVE_BUILTIN_CLZ) -- 2.34.1

Jan. 28, 2022

Re: [PATCH vkd3d v4 2/6] tests: Test a number of simple HLSL operations.

by Francisco Casas

Hi, I found something that I am not sure if it may be a problem: January 28, 2022 5:03 AM, "Giovanni Mascellani" <gmascellani(a)codeweavers.com> wrote: > +[test] > +draw quad > +probe all rgba (20.0, 4294967300.0, 75.0, 0.0) > + > +[pixel shader] > +float4 main() : SV_TARGET > +{ > + uint x = 5; > + uint y = 15; > + > + return float4(x % y, +x, -x, y / x); > +} > + > +[test] > +draw quad > +probe all rgba (5.0, 5.0, 4294967300.0, 3.0) The native compiler output is --- ps_4_0 dcl_output o0.xyzw mov o0.xyzw, l(20.000000,4294967296.000000,75.000000,0) ret --- Probably because float doesn't have enough precision to represent 4294967300. > +[pixel shader] > +float4 main() : SV_TARGET > +{ > + uint x = 5; > + uint y = 15; > + > + return float4(x == y, x != y, x < y, x <= y); > +} Same for the next test: > +[test] > +draw quad > +probe all rgba (0.0, 1.0, 1.0, 1.0) > + > +[pixel shader] > +float4 main() : SV_TARGET > +{ > + uint x = 5; > + uint y = 15; > + uint zero = 0; > + > + return float4(x > y, x >= y, !x, !zero); > +} > + And this one: > +[pixel shader] > +float4 main() : SV_TARGET > +{ > + uint x = 5; > + uint y = 15; > + > + return float4(x & y, x | y, x ^ y, ~x); > +} > + > +[test] > +draw quad > +probe all rgba (5.0, 15.0, 10.0, 4294967300.0) Everything else seems okay.

Jan. 28, 2022

Re: [PATCH 3/3] wined3d: Introduce WINED3D_LOCATION_CLEARED and use it for the initial state of buffers.

by Henri Verbeet

Signed-off-by: Henri Verbeet <hverbeet(a)codeweavers.com>

Jan. 28, 2022

Re: [PATCH 2/3] wined3d: Prefer mapping a BO if the buffer has WINED3D_BUFFER_USE_BO set.

by Henri Verbeet

Signed-off-by: Henri Verbeet <hverbeet(a)codeweavers.com>

Jan. 28, 2022

[PATCH 6/6] wined3d: Use wined3d_buffer_load_location() in wined3d_buffer_get_memory().

by Henri Verbeet

From: Zebediah Figura <zfigura(a)codeweavers.com> wined3d_buffer_load_location() can handle loading from WINED3D_LOCATION_DISCARDED just fine, so use it instead of duplicating the functionality here. Signed-off-by: Zebediah Figura <zfigura(a)codeweavers.com> Signed-off-by: Henri Verbeet <hverbeet(a)codeweavers.com> --- This supersedes patch 224835. v2: Add a commit messages. Otherwise unmodified. dlls/wined3d/buffer.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/dlls/wined3d/buffer.c b/dlls/wined3d/buffer.c index b42c7bd919a..3d66af3e5e6 100644 --- a/dlls/wined3d/buffer.c +++ b/dlls/wined3d/buffer.c @@ -653,14 +653,12 @@ DWORD wined3d_buffer_get_memory(struct wined3d_buffer *buffer, struct wined3d_co if (locations & WINED3D_LOCATION_DISCARDED) { locations = ((buffer->flags & WINED3D_BUFFER_USE_BO) ? WINED3D_LOCATION_BUFFER : WINED3D_LOCATION_SYSMEM); - if (!wined3d_buffer_prepare_location(buffer, context, locations)) + if (!wined3d_buffer_load_location(buffer, context, locations)) { data->buffer_object = 0; data->addr = NULL; return 0; } - wined3d_buffer_validate_location(buffer, locations); - wined3d_buffer_invalidate_location(buffer, WINED3D_LOCATION_DISCARDED); } if (locations & WINED3D_LOCATION_BUFFER) { -- 2.30.2

Jan. 28, 2022

[PATCH 5/6] wined3d: Use wined3d_bit_scan() in shader_arb_load_np2fixup_constants().

by Henri Verbeet

Signed-off-by: Henri Verbeet <hverbeet(a)codeweavers.com> --- dlls/wined3d/arb_program_shader.c | 21 ++++++++++----------- 1 file changed, 10 insertions(+), 11 deletions(-) diff --git a/dlls/wined3d/arb_program_shader.c b/dlls/wined3d/arb_program_shader.c index 820ba0199e8..54a90a690ac 100644 --- a/dlls/wined3d/arb_program_shader.c +++ b/dlls/wined3d/arb_program_shader.c @@ -497,27 +497,26 @@ static void shader_arb_load_np2fixup_constants(const struct arb_ps_np2fixup_info const struct wined3d_gl_info *gl_info, const struct wined3d_state *state) { GLfloat np2fixup_constants[4 * WINED3D_MAX_FRAGMENT_SAMPLERS]; - WORD active = fixup->super.active; - UINT i; + uint32_t active = fixup->super.active; + const struct wined3d_texture *tex; + unsigned char idx; + GLfloat *tex_dim; + unsigned int i; if (!active) return; - for (i = 0; active; active >>= 1, ++i) + while (active) { - const struct wined3d_texture *tex = state->textures[i]; - unsigned char idx = fixup->super.idx[i]; - GLfloat *tex_dim = &np2fixup_constants[(idx >> 1) * 4]; - - if (!(active & 1)) - continue; - - if (!tex) + i = wined3d_bit_scan(&active); + if (!(tex = state->textures[i])) { ERR("Nonexistent texture is flagged for NP2 texcoord fixup.\n"); continue; } + idx = fixup->super.idx[i]; + tex_dim = &np2fixup_constants[(idx >> 1) * 4]; if (idx % 2) { tex_dim[2] = tex->pow2_matrix[0]; -- 2.30.2

Jan. 28, 2022

[PATCH 4/6] wined3d: Use wined3d_bit_scan() in shader_glsl_generate_fragment_shader().

by Henri Verbeet

Signed-off-by: Henri Verbeet <hverbeet(a)codeweavers.com> --- dlls/wined3d/glsl_shader.c | 33 +++++++++++++++------------------ 1 file changed, 15 insertions(+), 18 deletions(-) diff --git a/dlls/wined3d/glsl_shader.c b/dlls/wined3d/glsl_shader.c index 5a2111e46a7..d10b84a9e05 100644 --- a/dlls/wined3d/glsl_shader.c +++ b/dlls/wined3d/glsl_shader.c @@ -7643,7 +7643,7 @@ static GLuint shader_glsl_generate_fragment_shader(const struct wined3d_context_ unsigned int i, extra_constants_needed = 0; struct shader_glsl_ctx_priv priv_ctx; GLuint shader_id; - DWORD map; + uint32_t map; memset(&priv_ctx, 0, sizeof(priv_ctx)); priv_ctx.gl_info = gl_info; @@ -7762,11 +7762,10 @@ static GLuint shader_glsl_generate_fragment_shader(const struct wined3d_context_ shader_addline(buffer, "vec4 %s_in[%u];\n", prefix, in_count); } - for (i = 0, map = reg_maps->bumpmat; map; map >>= 1, ++i) + map = reg_maps->bumpmat; + while (map) { - if (!(map & 1)) - continue; - + i = wined3d_bit_scan(&map); shader_addline(buffer, "uniform mat2 bumpenv_mat%u;\n", i); if (reg_maps->luminanceparams & (1u << i)) @@ -7903,7 +7902,6 @@ static GLuint shader_glsl_generate_fragment_shader(const struct wined3d_context_ if (reg_maps->shader_version.major < 3 || args->vp_mode != WINED3D_VP_MODE_SHADER) { unsigned int i; - WORD map = reg_maps->texcoord; if (legacy_syntax) { @@ -7913,19 +7911,18 @@ static GLuint shader_glsl_generate_fragment_shader(const struct wined3d_context_ shader_addline(buffer, "ffp_varying_specular = gl_SecondaryColor;\n"); } - for (i = 0; map; map >>= 1, ++i) + map = reg_maps->texcoord; + while (map) { - if (map & 1) - { - if (args->pointsprite) - shader_addline(buffer, "ffp_texcoord[%u] = vec4(gl_PointCoord.xy, 0.0, 0.0);\n", i); - else if (args->texcoords_initialized & (1u << i)) - shader_addline(buffer, "ffp_texcoord[%u] = %s[%u];\n", i, - legacy_syntax ? "gl_TexCoord" : "ffp_varying_texcoord", i); - else - shader_addline(buffer, "ffp_texcoord[%u] = vec4(0.0);\n", i); - shader_addline(buffer, "vec4 T%u = ffp_texcoord[%u];\n", i, i); - } + i = wined3d_bit_scan(&map); + if (args->pointsprite) + shader_addline(buffer, "ffp_texcoord[%u] = vec4(gl_PointCoord.xy, 0.0, 0.0);\n", i); + else if (args->texcoords_initialized & (1u << i)) + shader_addline(buffer, "ffp_texcoord[%u] = %s[%u];\n", i, + legacy_syntax ? "gl_TexCoord" : "ffp_varying_texcoord", i); + else + shader_addline(buffer, "ffp_texcoord[%u] = vec4(0.0);\n", i); + shader_addline(buffer, "vec4 T%u = ffp_texcoord[%u];\n", i, i); } if (legacy_syntax) -- 2.30.2

Jan. 28, 2022

[PATCH 3/6] wined3d: Use wined3d_bit_scan() in shader_generate_glsl_declarations().

by Henri Verbeet

Signed-off-by: Henri Verbeet <hverbeet(a)codeweavers.com> --- dlls/wined3d/glsl_shader.c | 21 +++++++++++++-------- 1 file changed, 13 insertions(+), 8 deletions(-) diff --git a/dlls/wined3d/glsl_shader.c b/dlls/wined3d/glsl_shader.c index aa3c8780269..5a2111e46a7 100644 --- a/dlls/wined3d/glsl_shader.c +++ b/dlls/wined3d/glsl_shader.c @@ -2168,7 +2168,7 @@ static void shader_generate_glsl_declarations(const struct wined3d_context_gl *c const struct wined3d_shader_lconst *lconst; const char *prefix; unsigned int i; - DWORD map; + uint32_t map; if (wined3d_settings.strict_shader_math) shader_addline(buffer, "#pragma optionNV(fastmath off)\n"); @@ -2176,9 +2176,11 @@ static void shader_generate_glsl_declarations(const struct wined3d_context_gl *c prefix = shader_glsl_get_prefix(version->type); /* Prototype the subroutines */ - for (i = 0, map = reg_maps->labels; map; map >>= 1, ++i) + map = reg_maps->labels; + while (map) { - if (map & 1) shader_addline(buffer, "void subroutine%u();\n", i); + i = wined3d_bit_scan(&map); + shader_addline(buffer, "void subroutine%u();\n", i); } if (version->type != WINED3D_SHADER_TYPE_PIXEL && version->type != WINED3D_SHADER_TYPE_COMPUTE) @@ -2500,9 +2502,11 @@ static void shader_generate_glsl_declarations(const struct wined3d_context_gl *c } /* Declare address variables */ - for (i = 0, map = reg_maps->address; map; map >>= 1, ++i) + map = reg_maps->address; + while (map) { - if (map & 1) shader_addline(buffer, "ivec4 A%u;\n", i); + i = wined3d_bit_scan(&map); + shader_addline(buffer, "ivec4 A%u;\n", i); } /* Declare output register temporaries */ @@ -2517,10 +2521,11 @@ static void shader_generate_glsl_declarations(const struct wined3d_context_gl *c } else if (version->major < 4) { - for (i = 0, map = reg_maps->temporary; map; map >>= 1, ++i) + map = reg_maps->temporary; + while (map) { - if (map & 1) - shader_addline(buffer, "vec4 R%u;\n", i); + i = wined3d_bit_scan(&map); + shader_addline(buffer, "vec4 R%u;\n", i); } } -- 2.30.2

Jan. 28, 2022