May 2024 - Wine-commits - list.winehq.org

Rémi Bernon : win32u: Read / write source modes as a single registry blob.
by Alexandre Julliard 07 May '24

07 May '24

Module: wine Branch: master Commit: 64a7ca75475d00af269c203349bc6fe386f16a22 URL: https://gitlab.winehq.org/wine/wine/-/commit/64a7ca75475d00af269c203349bc6f… Author: Rémi Bernon <rbernon(a)codeweavers.com> Date: Thu Apr 11 15:30:53 2024 +0200 win32u: Read / write source modes as a single registry blob. --- dlls/win32u/sysparams.c | 53 ++++++++++++++++++++++++++----------------------- 1 file changed, 28 insertions(+), 25 deletions(-) diff --git a/dlls/win32u/sysparams.c b/dlls/win32u/sysparams.c index e66ad488486..3211d4e7d8b 100644 --- a/dlls/win32u/sysparams.c +++ b/dlls/win32u/sysparams.c @@ -67,6 +67,7 @@ static const WCHAR device_descW[] = {'D','e','v','i','c','e','D','e','s','c',0}; static const WCHAR driver_descW[] = {'D','r','i','v','e','r','D','e','s','c',0}; static const WCHAR yesW[] = {'Y','e','s',0}; static const WCHAR noW[] = {'N','o',0}; +static const WCHAR modesW[] = {'M','o','d','e','s',0}; static const WCHAR mode_countW[] = {'M','o','d','e','C','o','u','n','t',0}; static const char guid_devclass_displayA[] = "{4D36E968-E325-11CE-BFC1-08002BE10318}"; @@ -431,10 +432,13 @@ static const char *debugstr_devmodew( const DEVMODEW *devmode ) static BOOL write_source_mode( HKEY hkey, UINT index, const DEVMODEW *mode ) { WCHAR bufferW[MAX_PATH] = {0}; - char buffer[MAX_PATH]; - snprintf( buffer, sizeof(buffer), "Modes\\%08X", index ); - asciiz_to_unicode( bufferW, buffer ); + assert( index == ENUM_CURRENT_SETTINGS || index == ENUM_REGISTRY_SETTINGS ); + + if (index == ENUM_CURRENT_SETTINGS) asciiz_to_unicode( bufferW, "Current" ); + else if (index == ENUM_REGISTRY_SETTINGS) asciiz_to_unicode( bufferW, "Registry" ); + else return FALSE; + return set_reg_value( hkey, bufferW, REG_BINARY, &mode->dmFields, sizeof(*mode) - offsetof(DEVMODEW, dmFields) ); } @@ -442,11 +446,15 @@ static BOOL read_source_mode( HKEY hkey, UINT index, DEVMODEW *mode ) { char value_buf[offsetof(KEY_VALUE_PARTIAL_INFORMATION, Data[sizeof(*mode)])]; KEY_VALUE_PARTIAL_INFORMATION *value = (void *)value_buf; - char buffer[MAX_PATH]; + const char *key; - snprintf( buffer, sizeof(buffer), "Modes\\%08X", index ); - if (!query_reg_ascii_value( hkey, buffer, value, sizeof(value_buf) )) return FALSE; + assert( index == ENUM_CURRENT_SETTINGS || index == ENUM_REGISTRY_SETTINGS ); + if (index == ENUM_CURRENT_SETTINGS) key = "Current"; + else if (index == ENUM_REGISTRY_SETTINGS) key = "Registry"; + else return FALSE; + + if (!query_reg_ascii_value( hkey, key, value, sizeof(value_buf) )) return FALSE; memcpy( &mode->dmFields, value->Data, sizeof(*mode) - offsetof(DEVMODEW, dmFields) ); return TRUE; } @@ -620,7 +628,6 @@ static BOOL reade_source_from_registry( unsigned int index, struct source *sourc char buffer[4096]; KEY_VALUE_PARTIAL_INFORMATION *value = (void *)buffer; WCHAR *value_str = (WCHAR *)value->Data; - DEVMODEW *mode; DWORD i, size; HKEY hkey; @@ -646,19 +653,18 @@ static BOOL reade_source_from_registry( unsigned int index, struct source *sourc if (query_reg_ascii_value( hkey, "ModeCount", value, sizeof(buffer) ) && value->Type == REG_DWORD) source->mode_count = *(const DWORD *)value->Data; - /* Modes, allocate an extra mode for easier iteration */ - if ((source->modes = calloc( source->mode_count + 1, sizeof(DEVMODEW) ))) + /* Modes */ + size = offsetof( KEY_VALUE_PARTIAL_INFORMATION, Data[(source->mode_count + 1) * sizeof(*source->modes)] ); + if (!(value = malloc( size )) || !query_reg_ascii_value( hkey, "Modes", value, size )) free( value ); + else { - for (i = 0, mode = source->modes; i < source->mode_count; i++) - { - mode->dmSize = offsetof(DEVMODEW, dmICMMethod); - if (!read_source_mode( hkey, i, mode )) break; - mode = NEXT_DEVMODEW(mode); - } - source->mode_count = i; - - qsort(source->modes, source->mode_count, sizeof(*source->modes) + source->modes->dmDriverExtra, mode_compare); + source->modes = (DEVMODEW *)value; + source->mode_count = value->DataLength / sizeof(*source->modes); + memmove( source->modes, value->Data, value->DataLength ); + memset( source->modes + source->mode_count, 0, sizeof(*source->modes) ); /* extra empty mode for easier iteration */ + qsort( source->modes, source->mode_count, sizeof(*source->modes), mode_compare ); } + value = (void *)buffer; /* DeviceID */ size = query_reg_ascii_value( hkey, "GPUID", value, sizeof(buffer) ); @@ -1451,7 +1457,6 @@ static void add_monitor( const struct gdi_monitor *gdi_monitor, void *param ) static void add_modes( const DEVMODEW *current, UINT modes_count, const DEVMODEW *modes, void *param ) { struct device_manager_ctx *ctx = param; - const DEVMODEW *mode; DEVMODEW dummy, detached = *current; TRACE( "current %s, modes_count %u, modes %p, param %p\n", debugstr_devmodew( current ), modes_count, modes, param ); @@ -1466,12 +1471,10 @@ static void add_modes( const DEVMODEW *current, UINT modes_count, const DEVMODEW write_source_mode( ctx->source_key, ENUM_REGISTRY_SETTINGS, current ); write_source_mode( ctx->source_key, ENUM_CURRENT_SETTINGS, current ); - for (mode = modes; modes_count; mode = NEXT_DEVMODEW(mode), modes_count--) - { - TRACE( "mode: %s\n", debugstr_devmodew( mode ) ); - if (write_source_mode( ctx->source_key, ctx->source.mode_count, mode )) ctx->source.mode_count++; - } - set_reg_value( ctx->source_key, mode_countW, REG_DWORD, &ctx->source.mode_count, sizeof(ctx->source.mode_count) ); + assert( !modes_count || modes->dmDriverExtra == 0 ); + set_reg_value( ctx->source_key, modesW, REG_BINARY, modes, modes_count * sizeof(*modes) ); + set_reg_value( ctx->source_key, mode_countW, REG_DWORD, &modes_count, sizeof(modes_count) ); + ctx->source.mode_count = modes_count; } static const struct gdi_device_manager device_manager =

1 0

Rémi Bernon : win32u: Remove now unnecessary reset_display_manager_ctx.
by Alexandre Julliard 07 May '24

07 May '24

Module: wine Branch: master Commit: 79f3573d8feb638c383c4dff4573ccf711d571af URL: https://gitlab.winehq.org/wine/wine/-/commit/79f3573d8feb638c383c4dff4573cc… Author: Rémi Bernon <rbernon(a)codeweavers.com> Date: Mon Apr 22 17:10:22 2024 +0200 win32u: Remove now unnecessary reset_display_manager_ctx. --- dlls/win32u/sysparams.c | 23 +++++++---------------- 1 file changed, 7 insertions(+), 16 deletions(-) diff --git a/dlls/win32u/sysparams.c b/dlls/win32u/sysparams.c index 3211d4e7d8b..9ebd6e26fc4 100644 --- a/dlls/win32u/sysparams.c +++ b/dlls/win32u/sysparams.c @@ -1485,21 +1485,6 @@ static const struct gdi_device_manager device_manager = add_modes, }; -static void reset_display_manager_ctx( struct device_manager_ctx *ctx ) -{ - HANDLE mutex = ctx->mutex; - - if (ctx->source_key) - { - NtClose( ctx->source_key ); - last_query_display_time = 0; - } - if (ctx->gpu_count) cleanup_devices(); - - memset( ctx, 0, sizeof(*ctx) ); - if ((ctx->mutex = mutex)) prepare_devices(); -} - static void release_display_manager_ctx( struct device_manager_ctx *ctx ) { if (ctx->mutex) @@ -1508,7 +1493,13 @@ static void release_display_manager_ctx( struct device_manager_ctx *ctx ) release_display_device_init_mutex( ctx->mutex ); ctx->mutex = 0; } - reset_display_manager_ctx( ctx ); + + if (ctx->source_key) + { + NtClose( ctx->source_key ); + last_query_display_time = 0; + } + if (ctx->gpu_count) cleanup_devices(); } static void clear_display_devices(void)

1 0

Rémi Bernon : win32u: Use struct pci_id in struct gdi_gpu.
by Alexandre Julliard 07 May '24

07 May '24

Module: wine Branch: master Commit: 62c60a9e10cdd73bda2b27eb048f4295bae7babb URL: https://gitlab.winehq.org/wine/wine/-/commit/62c60a9e10cdd73bda2b27eb048f42… Author: Rémi Bernon <rbernon(a)codeweavers.com> Date: Mon Apr 22 18:16:07 2024 +0200 win32u: Use struct pci_id in struct gdi_gpu. --- dlls/win32u/sysparams.c | 24 +++++------------------- dlls/winemac.drv/display.c | 11 +++++++---- dlls/winex11.drv/xrandr.c | 4 ++-- include/wine/gdi_driver.h | 15 ++++++++++----- 4 files changed, 24 insertions(+), 30 deletions(-) diff --git a/dlls/win32u/sysparams.c b/dlls/win32u/sysparams.c index 9ebd6e26fc4..8bda17f2892 100644 --- a/dlls/win32u/sysparams.c +++ b/dlls/win32u/sysparams.c @@ -85,14 +85,6 @@ static const char guid_devinterface_monitorA[] = "{E6F07B5F-EE97-4A90-B076-33F57 #define NEXT_DEVMODEW(mode) ((DEVMODEW *)((char *)((mode) + 1) + (mode)->dmDriverExtra)) -struct pci_id -{ - UINT16 vendor; - UINT16 device; - UINT16 subsystem; - UINT16 revision; -}; - struct gpu { LONG refcount; @@ -1193,21 +1185,15 @@ static BOOL write_gpu_to_registry( const struct gpu *gpu, const struct pci_id *p static void add_gpu( const struct gdi_gpu *gpu, void *param ) { - const struct pci_id pci_id = - { - .vendor = gpu->vendor_id, - .device = gpu->device_id, - .subsystem = gpu->subsys_id, - .revision = gpu->revision_id, - }; + const struct pci_id *pci_id = &gpu->pci_id; struct device_manager_ctx *ctx = param; char buffer[4096]; KEY_VALUE_PARTIAL_INFORMATION *value = (void *)buffer; unsigned int i; HKEY hkey, subkey; - TRACE( "%s %04X %04X %08X %02X\n", debugstr_w(gpu->name), - gpu->vendor_id, gpu->device_id, gpu->subsys_id, gpu->revision_id ); + TRACE( "%s %04X %04X %08X %02X\n", debugstr_w( gpu->name ), pci_id->vendor, pci_id->device, + pci_id->subsystem, pci_id->revision ); if (!enum_key && !(enum_key = reg_create_ascii_key( NULL, enum_keyA, 0, NULL ))) return; @@ -1225,7 +1211,7 @@ static void add_gpu( const struct gdi_gpu *gpu, void *param ) ctx->gpu.vulkan_uuid = gpu->vulkan_uuid; snprintf( ctx->gpu.path, sizeof(ctx->gpu.path), "PCI\\VEN_%04X&DEV_%04X&SUBSYS_%08X&REV_%02X\\%08X", - gpu->vendor_id, gpu->device_id, gpu->subsys_id, gpu->revision_id, ctx->gpu.index ); + pci_id->vendor, pci_id->device, pci_id->subsystem, pci_id->revision, ctx->gpu.index ); if (!(hkey = reg_create_ascii_key( enum_key, ctx->gpu.path, 0, NULL ))) return; if ((subkey = reg_create_ascii_key( hkey, "Device Parameters", 0, NULL ))) @@ -1265,7 +1251,7 @@ static void add_gpu( const struct gdi_gpu *gpu, void *param ) NtClose( hkey ); - if (!write_gpu_to_registry( &ctx->gpu, &pci_id, gpu->memory_size )) + if (!write_gpu_to_registry( &ctx->gpu, pci_id, gpu->memory_size )) WARN( "Failed to write gpu to registry\n" ); else ctx->gpu_count++; diff --git a/dlls/winemac.drv/display.c b/dlls/winemac.drv/display.c index 5221f4402fa..b52bb39c0f6 100644 --- a/dlls/winemac.drv/display.c +++ b/dlls/winemac.drv/display.c @@ -1142,10 +1142,13 @@ BOOL macdrv_UpdateDisplayDevices( const struct gdi_device_manager *device_manage struct gdi_gpu gdi_gpu = { .id = gpu->id, - .vendor_id = gpu->vendor_id, - .device_id = gpu->device_id, - .subsys_id = gpu->subsys_id, - .revision_id = gpu->revision_id, + .pci_id = + { + .vendor = gpu->vendor_id, + .device = gpu->device_id, + .subsystem = gpu->subsys_id, + .revision = gpu->revision_id, + }, }; RtlUTF8ToUnicodeN(gdi_gpu.name, sizeof(gdi_gpu.name), &len, gpu->name, strlen(gpu->name)); device_manager->add_gpu(&gdi_gpu, param); diff --git a/dlls/winex11.drv/xrandr.c b/dlls/winex11.drv/xrandr.c index 89ce2a6a263..e183501b23f 100644 --- a/dlls/winex11.drv/xrandr.c +++ b/dlls/winex11.drv/xrandr.c @@ -737,8 +737,8 @@ static BOOL get_gpu_properties_from_vulkan( struct gdi_gpu *gpu, const XRRProvid /* Ignore Khronos vendor IDs */ if (properties2.properties.vendorID < 0x10000) { - gpu->vendor_id = properties2.properties.vendorID; - gpu->device_id = properties2.properties.deviceID; + gpu->pci_id.vendor = properties2.properties.vendorID; + gpu->pci_id.device = properties2.properties.deviceID; } RtlUTF8ToUnicodeN( gpu->name, sizeof(gpu->name), &len, properties2.properties.deviceName, strlen( properties2.properties.deviceName ) + 1 ); diff --git a/include/wine/gdi_driver.h b/include/wine/gdi_driver.h index 61c9342d50f..447b5a86988 100644 --- a/include/wine/gdi_driver.h +++ b/include/wine/gdi_driver.h @@ -237,14 +237,19 @@ static inline ULONG window_surface_release( struct window_surface *surface ) /* display manager interface, used to initialize display device registry data */ +struct pci_id +{ + UINT16 vendor; + UINT16 device; + UINT16 subsystem; + UINT16 revision; +}; + struct gdi_gpu { ULONG_PTR id; - WCHAR name[128]; /* name */ - UINT vendor_id; /* PCI ID */ - UINT device_id; - UINT subsys_id; - UINT revision_id; + WCHAR name[128]; + struct pci_id pci_id; GUID vulkan_uuid; /* Vulkan device UUID */ ULONGLONG memory_size; };

1 0

Zebediah Figura : wined3d: Retrieve the VkCommandBuffer from wined3d_context_vk after executing RTV barriers.
by Alexandre Julliard 07 May '24

07 May '24

Module: wine Branch: master Commit: e7db99d85f8214eafc299e75f644ce222ea655fc URL: https://gitlab.winehq.org/wine/wine/-/commit/e7db99d85f8214eafc299e75f644ce… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Wed Apr 3 17:05:17 2024 -0500 wined3d: Retrieve the VkCommandBuffer from wined3d_context_vk after executing RTV barriers. Part of beginning a render pass involves executing an RTV barrier, which itself needs to call wined3d_context_vk_get_command_buffer(). However, that function may decide to submit the command buffer, in order to prevent resource buildup, or [in the future] because it has been some length of time since the last submission. Therefore we cannot retrieve and store a VkCommandBuffer pointer before executing an RTV barrier and then use it later. --- dlls/wined3d/context_vk.c | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-) diff --git a/dlls/wined3d/context_vk.c b/dlls/wined3d/context_vk.c index dc793a839bb..377c437ee09 100644 --- a/dlls/wined3d/context_vk.c +++ b/dlls/wined3d/context_vk.c @@ -2699,7 +2699,7 @@ static bool wined3d_context_vk_update_graphics_pipeline_key(struct wined3d_conte } static bool wined3d_context_vk_begin_render_pass(struct wined3d_context_vk *context_vk, - VkCommandBuffer vk_command_buffer, const struct wined3d_state *state, const struct wined3d_vk_info *vk_info) + const struct wined3d_state *state, const struct wined3d_vk_info *vk_info) { struct wined3d_device_vk *device_vk = wined3d_device_vk(context_vk->c.device); VkClearValue clear_values[WINED3D_MAX_RENDER_TARGETS + 1]; @@ -2709,6 +2709,7 @@ static bool wined3d_context_vk_begin_render_pass(struct wined3d_context_vk *cont struct wined3d_rendertarget_view *view; const VkPhysicalDeviceLimits *limits; struct wined3d_query_vk *query_vk; + VkCommandBuffer vk_command_buffer; VkRenderPassBeginInfo begin_info; unsigned int attachment_count, i; struct wined3d_texture *texture; @@ -2814,6 +2815,12 @@ static bool wined3d_context_vk_begin_render_pass(struct wined3d_context_vk *cont ++attachment_count; } + if (!(vk_command_buffer = wined3d_context_vk_get_command_buffer(context_vk))) + { + ERR("Failed to get command buffer.\n"); + return false; + } + if (!(context_vk->vk_render_pass = wined3d_context_vk_get_render_pass(context_vk, &state->fb, ARRAY_SIZE(state->fb.render_targets), !!state->fb.depth_stencil, 0))) { @@ -3772,19 +3779,15 @@ VkCommandBuffer wined3d_context_vk_apply_draw_state(struct wined3d_context_vk *c wined3d_context_vk_load_buffers(context_vk, state, indirect_vk, indexed); - if (!(vk_command_buffer = wined3d_context_vk_get_command_buffer(context_vk))) - { - ERR("Failed to get command buffer.\n"); - return VK_NULL_HANDLE; - } - if (wined3d_context_is_graphics_state_dirty(&context_vk->c, STATE_FRAMEBUFFER)) wined3d_context_vk_end_current_render_pass(context_vk); - if (!wined3d_context_vk_begin_render_pass(context_vk, vk_command_buffer, state, vk_info)) + + if (!wined3d_context_vk_begin_render_pass(context_vk, state, vk_info)) { ERR("Failed to begin render pass.\n"); return VK_NULL_HANDLE; } + vk_command_buffer = context_vk->current_command_buffer.vk_command_buffer; while (invalidate_rt) {

1 0

Zebediah Figura : wined3d: Submit command buffers after 512 draw or dispatch commands.
by Alexandre Julliard 07 May '24

07 May '24

Module: wine Branch: master Commit: cb723c6dd3baf13583b0bde2fc71191b86100229 URL: https://gitlab.winehq.org/wine/wine/-/commit/cb723c6dd3baf13583b0bde2fc7119… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Wed Apr 24 18:30:08 2024 -0500 wined3d: Submit command buffers after 512 draw or dispatch commands. This improves performance for the game "Grounded", on a AMD Radeon RX 6700 XT, with radv from Mesa 22.3.6. Testing was done with the "cb_access_map_w" option enabled, which also improves performance with the game by itself. Grounded generally makes about 4000 draw calls per frame, which seems not atypical. This change makes it submit at most an extra 8 times per frame, but in practice due to WINED3D_PERIODIC_SUBMIT_MAX_BUFFERS it submits less (usually only 2-3). The most demanding game I've seen made about 20,000 draw calls per frame, at which point the overhead of adapter_vk_draw_primitive() itself becomes a serious bottleneck. For such a game we would submit 40 more times per frame with these settings, although WINED3D_PERIODIC_SUBMIT_MAX_BUFFERS means we would likely submit less than that. In any case if submission itself becomes a bottleneck, we should offload it to a separate thread. Credit goes to Philip Rebohle and his work on DXVK for helping me to notice that periodic submission might make a difference. --- dlls/wined3d/adapter_vk.c | 4 ++++ dlls/wined3d/context_vk.c | 35 ++++++++++++++++++++++++++++++++++- dlls/wined3d/wined3d_vk.h | 3 +++ 3 files changed, 41 insertions(+), 1 deletion(-) diff --git a/dlls/wined3d/adapter_vk.c b/dlls/wined3d/adapter_vk.c index 418b67de8b6..39b263872ed 100644 --- a/dlls/wined3d/adapter_vk.c +++ b/dlls/wined3d/adapter_vk.c @@ -1807,6 +1807,8 @@ static void adapter_vk_draw_primitive(struct wined3d_device *device, context_vk->c.transform_feedback_active = 0; } + ++context_vk->command_buffer_work_count; + context_release(&context_vk->c); } @@ -1851,6 +1853,8 @@ static void adapter_vk_dispatch_compute(struct wined3d_device *device, VK_CALL(vkCmdPipelineBarrier(vk_command_buffer, VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, VK_PIPELINE_STAGE_ALL_GRAPHICS_BIT, 0, 0, NULL, 0, NULL, 0, NULL)); + ++context_vk->command_buffer_work_count; + context_release(&context_vk->c); } diff --git a/dlls/wined3d/context_vk.c b/dlls/wined3d/context_vk.c index 377c437ee09..200dcb57b7f 100644 --- a/dlls/wined3d/context_vk.c +++ b/dlls/wined3d/context_vk.c @@ -1771,6 +1771,37 @@ void wined3d_context_vk_cleanup(struct wined3d_context_vk *context_vk) wined3d_context_cleanup(&context_vk->c); } +/* In general we only submit when necessary or when a frame ends. However, + * applications which do a lot of work per frame can end up with the GPU idle + * for long periods of time while the CPU is building commands, and drivers may + * choose to reclock the GPU to a lower power level if they detect it being idle + * for that long. + * + * This may also help performance simply by virtue of allowing more parallelism + * between the GPU and CPU, although no clear evidence of that has been seen + * yet. */ + +#define WINED3D_PERIODIC_SUBMIT_WORK_COUNT 512 +#define WINED3D_PERIODIC_SUBMIT_MAX_BUFFERS 3 + +static bool should_periodic_submit(struct wined3d_context_vk *context_vk) +{ + uint64_t busy_count; + + if (context_vk->command_buffer_work_count < WINED3D_PERIODIC_SUBMIT_WORK_COUNT) + return false; + + /* The point of periodic submit is to keep the GPU busy, so if it's already + * busy with 4 or more command buffers, don't submit another one now. */ + busy_count = context_vk->current_command_buffer.id - context_vk->completed_command_buffer_id - 1; + if (busy_count > WINED3D_PERIODIC_SUBMIT_MAX_BUFFERS) + return false; + + TRACE("Periodically submitting command buffer, %u draw/dispatch commands since last buffer, %I64u currently busy.\n", + context_vk->command_buffer_work_count, busy_count); + return true; +} + VkCommandBuffer wined3d_context_vk_get_command_buffer(struct wined3d_context_vk *context_vk) { struct wined3d_device_vk *device_vk = wined3d_device_vk(context_vk->c.device); @@ -1785,7 +1816,7 @@ VkCommandBuffer wined3d_context_vk_get_command_buffer(struct wined3d_context_vk buffer = &context_vk->current_command_buffer; if (buffer->vk_command_buffer) { - if (context_vk->retired_bo_size > WINED3D_RETIRED_BO_SIZE_THRESHOLD) + if (context_vk->retired_bo_size > WINED3D_RETIRED_BO_SIZE_THRESHOLD || should_periodic_submit(context_vk)) wined3d_context_vk_submit_command_buffer(context_vk, 0, NULL, NULL, 0, NULL); else { @@ -1854,6 +1885,8 @@ VkCommandBuffer wined3d_context_vk_get_command_buffer(struct wined3d_context_vk wined3d_query_vk_resume(query_vk, context_vk); } + context_vk->command_buffer_work_count = 0; + TRACE("Created new command buffer %p with id 0x%s.\n", buffer->vk_command_buffer, wine_dbgstr_longlong(buffer->id)); diff --git a/dlls/wined3d/wined3d_vk.h b/dlls/wined3d/wined3d_vk.h index 94a6b6c0c5e..ad8eb2453f5 100644 --- a/dlls/wined3d/wined3d_vk.h +++ b/dlls/wined3d/wined3d_vk.h @@ -614,6 +614,9 @@ struct wined3d_context_vk struct wined3d_command_buffer_vk current_command_buffer; uint64_t completed_command_buffer_id; VkDeviceSize retired_bo_size; + /* Number of draw or dispatch calls that have been recorded into the + * current command buffer. */ + unsigned int command_buffer_work_count; struct {

1 0

Zebediah Figura : vkd3d-utils: Implement ID3D12ShaderReflection::GetResourceBindingDesc().
by Alexandre Julliard 06 May '24

06 May '24

Module: vkd3d Branch: master Commit: cefd6f9de6e278c45f6579c5ec342a7977767a7c URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/cefd6f9de6e278c45f6579c5ec342… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Thu May 2 23:43:38 2024 -0500 vkd3d-utils: Implement ID3D12ShaderReflection::GetResourceBindingDesc(). --- libs/vkd3d-utils/reflection.c | 84 +++++++++++++++++++++++++++++++++++++++++-- tests/hlsl_d3d12.c | 7 ++-- 2 files changed, 85 insertions(+), 6 deletions(-) diff --git a/libs/vkd3d-utils/reflection.c b/libs/vkd3d-utils/reflection.c index 0c9017a4..e8363ae4 100644 --- a/libs/vkd3d-utils/reflection.c +++ b/libs/vkd3d-utils/reflection.c @@ -61,6 +61,8 @@ struct d3d12_reflection D3D12_SHADER_DESC desc; struct d3d12_buffer *buffers; + + D3D12_SHADER_INPUT_BIND_DESC *bindings; }; static struct d3d12_buffer null_buffer; @@ -397,6 +399,10 @@ static ULONG STDMETHODCALLTYPE d3d12_reflection_Release(ID3D12ShaderReflection * } vkd3d_free(reflection->buffers); + for (UINT i = 0; i < reflection->desc.BoundResources; ++i) + vkd3d_free((void *)reflection->bindings[i].Name); + vkd3d_free(reflection->bindings); + vkd3d_shader_free_scan_signature_info(&reflection->signature_info); free(reflection); } @@ -444,9 +450,18 @@ static struct ID3D12ShaderReflectionConstantBuffer * STDMETHODCALLTYPE d3d12_ref static HRESULT STDMETHODCALLTYPE d3d12_reflection_GetResourceBindingDesc( ID3D12ShaderReflection *iface, UINT index, D3D12_SHADER_INPUT_BIND_DESC *desc) { - FIXME("iface %p, index %u, desc %p stub!\n", iface, index, desc); + struct d3d12_reflection *reflection = impl_from_ID3D12ShaderReflection(iface); - return E_NOTIMPL; + TRACE("iface %p, index %u, desc %p.\n", iface, index, desc); + + if (index >= reflection->desc.BoundResources) + { + WARN("Invalid index %u.\n", index); + return E_INVALIDARG; + } + + *desc = reflection->bindings[index]; + return S_OK; } static HRESULT get_signature_parameter(const struct vkd3d_shader_signature *signature, @@ -730,6 +745,20 @@ struct rdef_field uint32_t offset; }; +struct rdef_binding +{ + uint32_t name_offset; + uint32_t type; + uint32_t resource_format; + uint32_t dimension; + uint32_t multisample_count; + uint32_t index; + uint32_t count; + uint32_t flags; + uint32_t space; + uint32_t id; +}; + static HRESULT d3d12_type_init(struct d3d12_type *type, uint32_t type_offset, uint32_t type_size, const struct vkd3d_shader_code *section, uint32_t field_offset) { @@ -856,6 +885,7 @@ static HRESULT d3d12_buffer_init(struct d3d12_buffer *buffer, const struct rdef_ static HRESULT parse_rdef(struct d3d12_reflection *reflection, const struct vkd3d_shader_code *section) { uint32_t variable_size = offsetof(struct rdef_variable, resource_binding); + uint32_t binding_size = offsetof(struct rdef_binding, space); uint32_t type_size = offsetof(struct rdef_type, unknown); const struct rdef_header *header; const struct rdef_rd11 *rd11; @@ -897,6 +927,14 @@ static HRESULT parse_rdef(struct d3d12_reflection *reflection, const struct vkd3 } variable_size = rd11->variable_size; + if (rd11->binding_size != sizeof(struct rdef_binding) + && rd11->binding_size != offsetof(struct rdef_binding, space)) + { + FIXME("Unexpected binding size %#x.\n", rd11->binding_size); + return E_INVALIDARG; + } + binding_size = rd11->binding_size; + if (rd11->type_size != sizeof(struct rdef_type)) { FIXME("Unexpected type size %#x.\n", rd11->type_size); @@ -931,6 +969,48 @@ static HRESULT parse_rdef(struct d3d12_reflection *reflection, const struct vkd3 } } + reflection->desc.BoundResources = header->binding_count; + + if (header->binding_count) + { + if (!(reflection->bindings = vkd3d_calloc(header->binding_count, sizeof(*reflection->bindings)))) + return E_OUTOFMEMORY; + + for (uint32_t i = 0; i < header->binding_count; ++i) + { + const struct rdef_binding *rdef_binding; + D3D12_SHADER_INPUT_BIND_DESC *binding; + char *name; + + if (!(rdef_binding = get_data_ptr(section, header->bindings_offset + (i * binding_size), 1, binding_size))) + return E_INVALIDARG; + + if (FAILED(hr = get_string(section, rdef_binding->name_offset, &name))) + return hr; + + binding = &reflection->bindings[i]; + + binding->Name = name; + binding->Type = rdef_binding->type; + binding->BindPoint = rdef_binding->index; + binding->BindCount = rdef_binding->count; + binding->uFlags = rdef_binding->flags; + binding->ReturnType = rdef_binding->resource_format; + binding->Dimension = rdef_binding->dimension; + binding->NumSamples = rdef_binding->multisample_count; + if (binding_size == sizeof(*rdef_binding)) + { + binding->Space = rdef_binding->space; + binding->uID = rdef_binding->id; + } + else + { + binding->Space = 0; + binding->uID = rdef_binding->index; + } + } + } + return S_OK; } diff --git a/tests/hlsl_d3d12.c b/tests/hlsl_d3d12.c index ffab9fdd..8d5d8d90 100644 --- a/tests/hlsl_d3d12.c +++ b/tests/hlsl_d3d12.c @@ -1573,7 +1573,7 @@ static void test_reflection(void) hr = ID3D12ShaderReflection_GetDesc(reflection, &shader_desc); ok(hr == S_OK, "Got unexpected hr %#x.\n", hr); ok(shader_desc.ConstantBuffers == tests[t].buffer_count, "Got %u buffers.\n", shader_desc.ConstantBuffers); - todo ok(shader_desc.BoundResources == tests[t].binding_count, "Got %u resources.\n", shader_desc.BoundResources); + ok(shader_desc.BoundResources == tests[t].binding_count, "Got %u resources.\n", shader_desc.BoundResources); for (unsigned int i = 0; i < shader_desc.ConstantBuffers; ++i) { @@ -1669,8 +1669,7 @@ static void test_reflection(void) ok(binding_desc.Type == expect->Type, "Got type %#x.\n", binding_desc.Type); ok(binding_desc.BindPoint == expect->BindPoint, "Got bind point %u.\n", binding_desc.BindPoint); ok(binding_desc.BindCount == expect->BindCount, "Got bind count %u.\n", binding_desc.BindCount); - todo_if ((expect->uFlags & D3D_SIF_USERPACKED) && expect->Type != D3D_SIT_CBUFFER) - ok(binding_desc.uFlags == expect->uFlags, "Got flags %#x.\n", binding_desc.uFlags); + ok(binding_desc.uFlags == expect->uFlags, "Got flags %#x.\n", binding_desc.uFlags); ok(binding_desc.ReturnType == expect->ReturnType, "Got return type %#x.\n", binding_desc.ReturnType); ok(binding_desc.Dimension == expect->Dimension, "Got dimension %#x.\n", binding_desc.Dimension); ok(binding_desc.NumSamples == expect->NumSamples, "Got multisample count %u.\n", binding_desc.NumSamples); @@ -1679,7 +1678,7 @@ static void test_reflection(void) } hr = ID3D12ShaderReflection_GetResourceBindingDesc(reflection, shader_desc.BoundResources, &binding_desc); - todo ok(hr == E_INVALIDARG, "Got hr %#x.\n", hr); + ok(hr == E_INVALIDARG, "Got hr %#x.\n", hr); ID3D10Blob_Release(code); refcount = ID3D12ShaderReflection_Release(reflection);

1 0

Zebediah Figura : vkd3d-shader/hlsl: Ensure that the type is numeric before calling expr_common_base_type().
by Alexandre Julliard 06 May '24

06 May '24

Module: vkd3d Branch: master Commit: e904b4596a2f7d0b83d203cd03c5bec14f14c5e0 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/e904b4596a2f7d0b83d203cd03c5b… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Wed Feb 7 13:37:53 2024 -0600 vkd3d-shader/hlsl: Ensure that the type is numeric before calling expr_common_base_type(). --- libs/vkd3d-shader/hlsl.y | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) diff --git a/libs/vkd3d-shader/hlsl.y b/libs/vkd3d-shader/hlsl.y index 0eed15c5..930929ad 100644 --- a/libs/vkd3d-shader/hlsl.y +++ b/libs/vkd3d-shader/hlsl.y @@ -1386,10 +1386,6 @@ static bool expr_compatible_data_types(struct hlsl_type *t1, struct hlsl_type *t static enum hlsl_base_type expr_common_base_type(enum hlsl_base_type t1, enum hlsl_base_type t2) { - if (t1 > HLSL_TYPE_LAST_SCALAR || t2 > HLSL_TYPE_LAST_SCALAR) { - FIXME("Unexpected base type.\n"); - return HLSL_TYPE_FLOAT; - } if (t1 == t2) return t1 == HLSL_TYPE_BOOL ? HLSL_TYPE_INT : t1; if (t1 == HLSL_TYPE_DOUBLE || t2 == HLSL_TYPE_DOUBLE) @@ -1593,13 +1589,13 @@ static struct hlsl_ir_node *add_unary_logical_expr(struct hlsl_ctx *ctx, struct static struct hlsl_type *get_common_numeric_type(struct hlsl_ctx *ctx, const struct hlsl_ir_node *arg1, const struct hlsl_ir_node *arg2, const struct vkd3d_shader_location *loc) { - enum hlsl_base_type base = expr_common_base_type(arg1->data_type->base_type, arg2->data_type->base_type); enum hlsl_type_class type; + enum hlsl_base_type base; unsigned int dimx, dimy; if (!expr_common_shape(ctx, arg1->data_type, arg2->data_type, loc, &type, &dimx, &dimy)) return NULL; - + base = expr_common_base_type(arg1->data_type->base_type, arg2->data_type->base_type); return hlsl_get_numeric_type(ctx, type, base, dimx, dimy); } @@ -1636,14 +1632,15 @@ static struct hlsl_ir_node *add_binary_comparison_expr(struct hlsl_ctx *ctx, str const struct vkd3d_shader_location *loc) { struct hlsl_type *common_type, *return_type; - enum hlsl_base_type base = expr_common_base_type(arg1->data_type->base_type, arg2->data_type->base_type); enum hlsl_type_class type; + enum hlsl_base_type base; unsigned int dimx, dimy; struct hlsl_ir_node *args[HLSL_MAX_OPERANDS] = {0}; if (!expr_common_shape(ctx, arg1->data_type, arg2->data_type, loc, &type, &dimx, &dimy)) return NULL; + base = expr_common_base_type(arg1->data_type->base_type, arg2->data_type->base_type); common_type = hlsl_get_numeric_type(ctx, type, base, dimx, dimy); return_type = hlsl_get_numeric_type(ctx, type, HLSL_TYPE_BOOL, dimx, dimy);

1 0

Zebediah Figura : vkd3d-shader/hlsl: Make HLSL_TYPE_VERTEXSHADER into a separate class.
by Alexandre Julliard 06 May '24

06 May '24

Module: vkd3d Branch: master Commit: 874937dab48aedbef1fa99cb2fa60fdcc0d804ac URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/874937dab48aedbef1fa99cb2fa60… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Tue Feb 6 17:33:26 2024 -0600 vkd3d-shader/hlsl: Make HLSL_TYPE_VERTEXSHADER into a separate class. --- libs/vkd3d-shader/d3dbc.c | 6 ++++-- libs/vkd3d-shader/fx.c | 24 +++++++++++++++++------- libs/vkd3d-shader/hlsl.c | 7 ++++++- libs/vkd3d-shader/hlsl.h | 2 +- libs/vkd3d-shader/hlsl_codegen.c | 1 + libs/vkd3d-shader/tpf.c | 1 + 6 files changed, 30 insertions(+), 11 deletions(-) diff --git a/libs/vkd3d-shader/d3dbc.c b/libs/vkd3d-shader/d3dbc.c index aa235844..145ebd85 100644 --- a/libs/vkd3d-shader/d3dbc.c +++ b/libs/vkd3d-shader/d3dbc.c @@ -1518,6 +1518,7 @@ D3DXPARAMETER_CLASS hlsl_sm1_class(const struct hlsl_type *type) case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: case HLSL_CLASS_TEXTURE: + case HLSL_CLASS_VERTEX_SHADER: return D3DXPC_OBJECT; case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: @@ -1605,8 +1606,6 @@ D3DXPARAMETER_TYPE hlsl_sm1_base_type(const struct hlsl_type *type) { case HLSL_TYPE_PIXELSHADER: return D3DXPT_PIXELSHADER; - case HLSL_TYPE_VERTEXSHADER: - return D3DXPT_VERTEXSHADER; default: vkd3d_unreachable(); } @@ -1621,6 +1620,9 @@ D3DXPARAMETER_TYPE hlsl_sm1_base_type(const struct hlsl_type *type) case HLSL_CLASS_STRING: return D3DXPT_STRING; + case HLSL_CLASS_VERTEX_SHADER: + return D3DXPT_VERTEXSHADER; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_PASS: diff --git a/libs/vkd3d-shader/fx.c b/libs/vkd3d-shader/fx.c index 168378e6..4f94f428 100644 --- a/libs/vkd3d-shader/fx.c +++ b/libs/vkd3d-shader/fx.c @@ -352,7 +352,6 @@ static const char * get_fx_4_type_name(const struct hlsl_type *type) static const char * const object_type_names[] = { [HLSL_TYPE_PIXELSHADER] = "PixelShader", - [HLSL_TYPE_VERTEXSHADER] = "VertexShader", }; static const char * const texture_type_names[] = { @@ -392,11 +391,13 @@ static const char * get_fx_4_type_name(const struct hlsl_type *type) case HLSL_CLASS_RENDER_TARGET_VIEW: return "RenderTargetView"; + case HLSL_CLASS_VERTEX_SHADER: + return "VertexShader"; + case HLSL_CLASS_OBJECT: switch (type->base_type) { case HLSL_TYPE_PIXELSHADER: - case HLSL_TYPE_VERTEXSHADER: return object_type_names[type->base_type]; default: return type->name; @@ -440,6 +441,7 @@ static uint32_t write_fx_4_type(const struct hlsl_type *type, struct fx_write_co case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: + case HLSL_CLASS_VERTEX_SHADER: put_u32_unaligned(buffer, 2); break; @@ -532,18 +534,20 @@ static uint32_t write_fx_4_type(const struct hlsl_type *type, struct fx_write_co { put_u32_unaligned(buffer, 19); } + else if (type->class == HLSL_CLASS_VERTEX_SHADER) + { + put_u32_unaligned(buffer, 6); + } else if (type->class == HLSL_CLASS_OBJECT) { static const uint32_t object_type[] = { [HLSL_TYPE_PIXELSHADER] = 5, - [HLSL_TYPE_VERTEXSHADER] = 6, }; switch (type->base_type) { case HLSL_TYPE_PIXELSHADER: - case HLSL_TYPE_VERTEXSHADER: put_u32_unaligned(buffer, object_type[type->base_type]); break; default: @@ -842,7 +846,6 @@ static bool is_type_supported_fx_2(struct hlsl_ctx *ctx, const struct hlsl_type switch (type->base_type) { case HLSL_TYPE_PIXELSHADER: - case HLSL_TYPE_VERTEXSHADER: hlsl_fixme(ctx, loc, "Write fx 2.0 parameter object type %#x.", type->base_type); return false; @@ -852,6 +855,7 @@ static bool is_type_supported_fx_2(struct hlsl_ctx *ctx, const struct hlsl_type case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: + case HLSL_CLASS_VERTEX_SHADER: hlsl_fixme(ctx, loc, "Write fx 2.0 parameter class %#x.", type->class); return false; @@ -1052,7 +1056,6 @@ static void write_fx_4_object_variable(struct hlsl_ir_var *var, struct fx_write_ switch (type->base_type) { case HLSL_TYPE_PIXELSHADER: - case HLSL_TYPE_VERTEXSHADER: /* FIXME: write shader blobs, once parser support works. */ for (i = 0; i < elements_count; ++i) put_u32(buffer, 0); @@ -1064,6 +1067,13 @@ static void write_fx_4_object_variable(struct hlsl_ir_var *var, struct fx_write_ } break; + case HLSL_CLASS_VERTEX_SHADER: + /* FIXME: write shader blobs, once parser support works. */ + for (i = 0; i < elements_count; ++i) + put_u32(buffer, 0); + ++fx->shader_variable_count; + break; + default: hlsl_fixme(ctx, &ctx->location, "Writing initializer for object type %u is not implemented.", type->base_type); @@ -1153,13 +1163,13 @@ static bool is_object_variable(const struct hlsl_ir_var *var) case HLSL_CLASS_SAMPLER: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: + case HLSL_CLASS_VERTEX_SHADER: return true; case HLSL_CLASS_OBJECT: switch (type->base_type) { case HLSL_TYPE_PIXELSHADER: - case HLSL_TYPE_VERTEXSHADER: return true; default: return false; diff --git a/libs/vkd3d-shader/hlsl.c b/libs/vkd3d-shader/hlsl.c index 0b48b17d..20ae3a64 100644 --- a/libs/vkd3d-shader/hlsl.c +++ b/libs/vkd3d-shader/hlsl.c @@ -370,6 +370,7 @@ static void hlsl_type_calculate_reg_size(struct hlsl_ctx *ctx, struct hlsl_type case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_STRING: case HLSL_CLASS_TECHNIQUE: + case HLSL_CLASS_VERTEX_SHADER: case HLSL_CLASS_VOID: break; } @@ -438,6 +439,7 @@ static bool type_is_single_component(const struct hlsl_type *type) case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: + case HLSL_CLASS_VERTEX_SHADER: return true; case HLSL_CLASS_VECTOR: @@ -578,6 +580,7 @@ unsigned int hlsl_type_get_component_offset(struct hlsl_ctx *ctx, struct hlsl_ty case HLSL_CLASS_STRING: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: + case HLSL_CLASS_VERTEX_SHADER: assert(idx == 0); break; @@ -956,6 +959,7 @@ unsigned int hlsl_type_component_count(const struct hlsl_type *type) case HLSL_CLASS_STRING: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: + case HLSL_CLASS_VERTEX_SHADER: return 1; case HLSL_CLASS_EFFECT_GROUP: @@ -2382,6 +2386,7 @@ struct vkd3d_string_buffer *hlsl_type_to_string(struct hlsl_ctx *ctx, const stru case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: case HLSL_CLASS_TECHNIQUE: + case HLSL_CLASS_VERTEX_SHADER: case HLSL_CLASS_VOID: break; } @@ -3558,7 +3563,6 @@ static void declare_predefined_types(struct hlsl_ctx *ctx) {"vector", HLSL_CLASS_VECTOR, HLSL_TYPE_FLOAT, 4, 1}, {"matrix", HLSL_CLASS_MATRIX, HLSL_TYPE_FLOAT, 4, 4}, {"pixelshader", HLSL_CLASS_OBJECT, HLSL_TYPE_PIXELSHADER, 1, 1}, - {"vertexshader", HLSL_CLASS_OBJECT, HLSL_TYPE_VERTEXSHADER, 1, 1}, }; static const struct @@ -3685,6 +3689,7 @@ static void declare_predefined_types(struct hlsl_ctx *ctx) hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "RenderTargetView", HLSL_CLASS_RENDER_TARGET_VIEW)); hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "STRING", HLSL_CLASS_STRING)); hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "texture", HLSL_CLASS_TEXTURE)); + hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "vertexshader", HLSL_CLASS_VERTEX_SHADER)); for (i = 0; i < ARRAY_SIZE(effect_types); ++i) { diff --git a/libs/vkd3d-shader/hlsl.h b/libs/vkd3d-shader/hlsl.h index ec3f5826..ac8ccd61 100644 --- a/libs/vkd3d-shader/hlsl.h +++ b/libs/vkd3d-shader/hlsl.h @@ -88,6 +88,7 @@ enum hlsl_type_class HLSL_CLASS_TECHNIQUE, HLSL_CLASS_TEXTURE, HLSL_CLASS_UAV, + HLSL_CLASS_VERTEX_SHADER, HLSL_CLASS_VOID, }; @@ -101,7 +102,6 @@ enum hlsl_base_type HLSL_TYPE_BOOL, HLSL_TYPE_LAST_SCALAR = HLSL_TYPE_BOOL, HLSL_TYPE_PIXELSHADER, - HLSL_TYPE_VERTEXSHADER, }; enum hlsl_sampler_dim diff --git a/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d-shader/hlsl_codegen.c index 94acb70f..ff114c32 100644 --- a/libs/vkd3d-shader/hlsl_codegen.c +++ b/libs/vkd3d-shader/hlsl_codegen.c @@ -1628,6 +1628,7 @@ static bool copy_propagation_transform_load(struct hlsl_ctx *ctx, case HLSL_CLASS_SAMPLER: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: + case HLSL_CLASS_VERTEX_SHADER: case HLSL_CLASS_OBJECT: break; diff --git a/libs/vkd3d-shader/tpf.c b/libs/vkd3d-shader/tpf.c index 708ab626..ffc0d9d5 100644 --- a/libs/vkd3d-shader/tpf.c +++ b/libs/vkd3d-shader/tpf.c @@ -3017,6 +3017,7 @@ static D3D_SHADER_VARIABLE_CLASS sm4_class(const struct hlsl_type *type) case HLSL_CLASS_TECHNIQUE: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: + case HLSL_CLASS_VERTEX_SHADER: case HLSL_CLASS_VOID: break; }

1 0

Zebediah Figura : vkd3d-shader/hlsl: Make HLSL_TYPE_PIXELSHADER into a separate class.
by Alexandre Julliard 06 May '24

06 May '24

Module: vkd3d Branch: master Commit: d9f7a883295141a35de9a7e0b4e6c2d2744c525e URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/d9f7a883295141a35de9a7e0b4e6c… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Tue Feb 6 19:41:15 2024 -0600 vkd3d-shader/hlsl: Make HLSL_TYPE_PIXELSHADER into a separate class. --- libs/vkd3d-shader/d3dbc.c | 15 +++----- libs/vkd3d-shader/fx.c | 76 ++++++---------------------------------- libs/vkd3d-shader/hlsl.c | 12 +++---- libs/vkd3d-shader/hlsl.h | 4 +-- libs/vkd3d-shader/hlsl_codegen.c | 2 +- libs/vkd3d-shader/tpf.c | 2 +- 6 files changed, 23 insertions(+), 88 deletions(-)

1 0

Zebediah Figura : vkd3d-shader/hlsl: Use a switch in hlsl_types_are_equal().
by Alexandre Julliard 06 May '24

06 May '24

Module: vkd3d Branch: master Commit: 4fd3f46109f95d6804c6f072416ef7a0c476a8d9 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/4fd3f46109f95d6804c6f072416ef… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Tue Feb 27 17:20:28 2024 -0600 vkd3d-shader/hlsl: Use a switch in hlsl_types_are_equal(). --- libs/vkd3d-shader/hlsl.c | 102 ++++++++++++++++++++++++++++------------------- 1 file changed, 60 insertions(+), 42 deletions(-) diff --git a/libs/vkd3d-shader/hlsl.c b/libs/vkd3d-shader/hlsl.c index da7029c5..97da23c1 100644 --- a/libs/vkd3d-shader/hlsl.c +++ b/libs/vkd3d-shader/hlsl.c @@ -979,55 +979,73 @@ bool hlsl_types_are_equal(const struct hlsl_type *t1, const struct hlsl_type *t2 if (t1->class != t2->class) return false; - if (t1->base_type != t2->base_type) - return false; - if (t1->class == HLSL_CLASS_SAMPLER || t1->class == HLSL_CLASS_TEXTURE || t1->class == HLSL_CLASS_UAV) - { - if (t1->sampler_dim != t2->sampler_dim) - return false; - if ((t1->class == HLSL_CLASS_TEXTURE || t1->class == HLSL_CLASS_UAV) - && t1->sampler_dim != HLSL_SAMPLER_DIM_GENERIC - && !hlsl_types_are_equal(t1->e.resource.format, t2->e.resource.format)) - return false; - if (t1->class == HLSL_CLASS_UAV && t1->e.resource.rasteriser_ordered != t2->e.resource.rasteriser_ordered) - return false; - } - if ((t1->modifiers & HLSL_MODIFIER_ROW_MAJOR) - != (t2->modifiers & HLSL_MODIFIER_ROW_MAJOR)) - return false; - if (t1->dimx != t2->dimx) - return false; - if (t1->dimy != t2->dimy) - return false; - if (t1->class == HLSL_CLASS_STRUCT) - { - size_t i; - - if (t1->e.record.field_count != t2->e.record.field_count) - return false; - for (i = 0; i < t1->e.record.field_count; ++i) - { - const struct hlsl_struct_field *field1 = &t1->e.record.fields[i]; - const struct hlsl_struct_field *field2 = &t2->e.record.fields[i]; + switch (t1->class) + { + case HLSL_CLASS_SCALAR: + case HLSL_CLASS_VECTOR: + case HLSL_CLASS_MATRIX: + if (t1->base_type != t2->base_type) + return false; + if ((t1->modifiers & HLSL_MODIFIER_ROW_MAJOR) + != (t2->modifiers & HLSL_MODIFIER_ROW_MAJOR)) + return false; + if (t1->dimx != t2->dimx) + return false; + if (t1->dimy != t2->dimy) + return false; + return true; - if (!hlsl_types_are_equal(field1->type, field2->type)) + case HLSL_CLASS_UAV: + if (t1->e.resource.rasteriser_ordered != t2->e.resource.rasteriser_ordered) return false; + /* fall through */ + case HLSL_CLASS_TEXTURE: + if (t1->sampler_dim != HLSL_SAMPLER_DIM_GENERIC + && !hlsl_types_are_equal(t1->e.resource.format, t2->e.resource.format)) + return false; + /* fall through */ + case HLSL_CLASS_SAMPLER: + if (t1->sampler_dim != t2->sampler_dim) + return false; + return true; - if (strcmp(field1->name, field2->name)) + case HLSL_CLASS_STRUCT: + if (t1->e.record.field_count != t2->e.record.field_count) return false; - } - } - if (t1->class == HLSL_CLASS_ARRAY) - return t1->e.array.elements_count == t2->e.array.elements_count - && hlsl_types_are_equal(t1->e.array.type, t2->e.array.type); - if (t1->class == HLSL_CLASS_TECHNIQUE) - { - if (t1->e.version != t2->e.version) - return false; + + for (size_t i = 0; i < t1->e.record.field_count; ++i) + { + const struct hlsl_struct_field *field1 = &t1->e.record.fields[i]; + const struct hlsl_struct_field *field2 = &t2->e.record.fields[i]; + + if (!hlsl_types_are_equal(field1->type, field2->type)) + return false; + + if (strcmp(field1->name, field2->name)) + return false; + } + return true; + + case HLSL_CLASS_ARRAY: + return t1->e.array.elements_count == t2->e.array.elements_count + && hlsl_types_are_equal(t1->e.array.type, t2->e.array.type); + + case HLSL_CLASS_TECHNIQUE: + return t1->e.version == t2->e.version; + + case HLSL_CLASS_DEPTH_STENCIL_VIEW: + case HLSL_CLASS_EFFECT_GROUP: + case HLSL_CLASS_PASS: + case HLSL_CLASS_PIXEL_SHADER: + case HLSL_CLASS_RENDER_TARGET_VIEW: + case HLSL_CLASS_STRING: + case HLSL_CLASS_VERTEX_SHADER: + case HLSL_CLASS_VOID: + return true; } - return true; + vkd3d_unreachable(); } struct hlsl_type *hlsl_type_clone(struct hlsl_ctx *ctx, struct hlsl_type *old,

1 0