February 2024 - Wine-Commits

Alexandre Julliard : ntdll: Update the image information when loading a builtin dll.
by Alexandre Julliard Feb. 7, 2024

Feb. 7, 2024

Module: wine Branch: master Commit: 0d37cc1d0637551ceae42773f0e1f059b380fa12 URL: https://gitlab.winehq.org/wine/wine/-/commit/0d37cc1d0637551ceae42773f0e1f0… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Wed Feb 7 19:04:29 2024 +0100 ntdll: Update the image information when loading a builtin dll. --- dlls/ntdll/unix/loader.c | 8 ++++---- dlls/ntdll/unix/unix_private.h | 3 ++- dlls/ntdll/unix/virtual.c | 12 ++++++++---- 3 files changed, 14 insertions(+), 9 deletions(-) diff --git a/dlls/ntdll/unix/loader.c b/dlls/ntdll/unix/loader.c index 803d8079213..662ec5ef7a8 100644 --- a/dlls/ntdll/unix/loader.c +++ b/dlls/ntdll/unix/loader.c @@ -1277,11 +1277,11 @@ done: * Return STATUS_IMAGE_ALREADY_LOADED if we should keep the native one that we have found. */ NTSTATUS load_builtin( const pe_image_info_t *image_info, WCHAR *filename, USHORT machine, - void **module, SIZE_T *size, ULONG_PTR limit_low, ULONG_PTR limit_high ) + SECTION_IMAGE_INFORMATION *info, void **module, SIZE_T *size, + ULONG_PTR limit_low, ULONG_PTR limit_high ) { NTSTATUS status; UNICODE_STRING nt_name; - SECTION_IMAGE_INFORMATION info; enum loadorder loadorder; init_unicode_string( &nt_name, filename ); @@ -1307,10 +1307,10 @@ NTSTATUS load_builtin( const pe_image_info_t *image_info, WCHAR *filename, USHOR case LO_NATIVE_BUILTIN: return STATUS_IMAGE_ALREADY_LOADED; case LO_BUILTIN: - return find_builtin_dll( &nt_name, module, size, &info, limit_low, limit_high, + return find_builtin_dll( &nt_name, module, size, info, limit_low, limit_high, image_info->machine, machine, FALSE ); default: - status = find_builtin_dll( &nt_name, module, size, &info, limit_low, limit_high, + status = find_builtin_dll( &nt_name, module, size, info, limit_low, limit_high, image_info->machine, machine, (loadorder == LO_DEFAULT) ); if (status == STATUS_DLL_NOT_FOUND || status == STATUS_NOT_SUPPORTED) return STATUS_IMAGE_ALREADY_LOADED; diff --git a/dlls/ntdll/unix/unix_private.h b/dlls/ntdll/unix/unix_private.h index 07f2724eac7..af746b220e8 100644 --- a/dlls/ntdll/unix/unix_private.h +++ b/dlls/ntdll/unix/unix_private.h @@ -185,7 +185,8 @@ extern char **build_envp( const WCHAR *envW ); extern char *get_alternate_wineloader( WORD machine ); extern NTSTATUS exec_wineloader( char **argv, int socketfd, const pe_image_info_t *pe_info ); extern NTSTATUS load_builtin( const pe_image_info_t *image_info, WCHAR *filename, USHORT machine, - void **addr_ptr, SIZE_T *size_ptr, ULONG_PTR limit_low, ULONG_PTR limit_high ); + SECTION_IMAGE_INFORMATION *info, void **module, SIZE_T *size, + ULONG_PTR limit_low, ULONG_PTR limit_high ); extern BOOL is_builtin_path( const UNICODE_STRING *path, WORD *machine ); extern NTSTATUS load_main_exe( const WCHAR *name, const char *unix_name, const WCHAR *curdir, USHORT load_machine, WCHAR **image, void **module ); diff --git a/dlls/ntdll/unix/virtual.c b/dlls/ntdll/unix/virtual.c index 2a53f56879e..cfe19eae89d 100644 --- a/dlls/ntdll/unix/virtual.c +++ b/dlls/ntdll/unix/virtual.c @@ -3161,9 +3161,12 @@ static unsigned int virtual_map_section( HANDLE handle, PVOID *addr_ptr, ULONG_P if (image_info) { + SECTION_IMAGE_INFORMATION info; + filename = (WCHAR *)(image_info + 1); /* check if we can replace that mapping with the builtin */ - res = load_builtin( image_info, filename, machine, addr_ptr, size_ptr, limit_low, limit_high ); + res = load_builtin( image_info, filename, machine, &info, + addr_ptr, size_ptr, limit_low, limit_high ); if (res == STATUS_IMAGE_ALREADY_LOADED) res = virtual_map_image( handle, addr_ptr, size_ptr, shared_file, limit_low, limit_high, alloc_type, machine, image_info, filename, FALSE ); @@ -3462,12 +3465,13 @@ NTSTATUS virtual_map_module( HANDLE mapping, void **module, SIZE_T *size, SECTIO filename = (WCHAR *)(image_info + 1); /* check if we can replace that mapping with the builtin */ - status = load_builtin( image_info, filename, machine, module, size, limit_low, limit_high ); + status = load_builtin( image_info, filename, machine, info, module, size, limit_low, limit_high ); if (status == STATUS_IMAGE_ALREADY_LOADED) + { status = virtual_map_image( mapping, module, size, shared_file, limit_low, limit_high, 0, machine, image_info, filename, FALSE ); - - virtual_fill_image_information( image_info, info ); + virtual_fill_image_information( image_info, info ); + } if (shared_file) NtClose( shared_file ); free( image_info ); return status;

1 0

Alexandre Julliard : ntdll/tests: Add some process machine tests for ARM64X.
by Alexandre Julliard Feb. 7, 2024

Feb. 7, 2024

Module: wine Branch: master Commit: 818b5777273ce793573dad1f2fbaa0db26bbac02 URL: https://gitlab.winehq.org/wine/wine/-/commit/818b5777273ce793573dad1f2fbaa0… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Wed Feb 7 13:17:45 2024 +0100 ntdll/tests: Add some process machine tests for ARM64X. --- dlls/ntdll/tests/wow64.c | 140 +++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 134 insertions(+), 6 deletions(-)

1 0

Alexandre Julliard : server: Don't report alternate 64-bit machines as supported.
by Alexandre Julliard Feb. 7, 2024

Feb. 7, 2024

Module: wine Branch: master Commit: 4022fb9ba3eacd6b09d35b457754a7e17c67564e URL: https://gitlab.winehq.org/wine/wine/-/commit/4022fb9ba3eacd6b09d35b457754a7… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Wed Feb 7 15:46:11 2024 +0100 server: Don't report alternate 64-bit machines as supported. --- dlls/ntdll/unix/system.c | 26 +++++++++++++------------- server/registry.c | 12 ++---------- 2 files changed, 15 insertions(+), 23 deletions(-) diff --git a/dlls/ntdll/unix/system.c b/dlls/ntdll/unix/system.c index d2df1dd50c2..cb99b7c2cc3 100644 --- a/dlls/ntdll/unix/system.c +++ b/dlls/ntdll/unix/system.c @@ -3385,24 +3385,24 @@ NTSTATUS WINAPI NtQuerySystemInformationEx( SYSTEM_INFORMATION_CLASS class, ret = STATUS_BUFFER_TOO_SMALL; break; } - for (i = 0; i < supported_machines_count; i++) + memset( machines, 0, len ); + + /* native machine */ + machines[0].Machine = supported_machines[0]; + machines[0].UserMode = 1; + machines[0].KernelMode = 1; + machines[0].Native = 1; + machines[0].Process = (supported_machines[0] == machine || is_machine_64bit( machine )); + machines[0].WoW64Container = 0; + machines[0].ReservedZero0 = 0; + /* wow64 machines */ + for (i = 1; i < supported_machines_count; i++) { machines[i].Machine = supported_machines[i]; machines[i].UserMode = 1; - machines[i].KernelMode = machines[i].Native = i == 0; machines[i].Process = supported_machines[i] == machine; - machines[i].WoW64Container = 0; - machines[i].ReservedZero0 = 0; + machines[i].WoW64Container = 1; } - - machines[i].Machine = 0; - machines[i].KernelMode = 0; - machines[i].UserMode = 0; - machines[i].Native = 0; - machines[i].Process = 0; - machines[i].WoW64Container = 0; - machines[i].ReservedZero0 = 0; - ret = STATUS_SUCCESS; break; } diff --git a/server/registry.c b/server/registry.c index da6a6d0982e..e88bed1e72e 100644 --- a/server/registry.c +++ b/server/registry.c @@ -1857,10 +1857,9 @@ static void init_supported_machines(void) if (prefix_type == PREFIX_64BIT) { supported_machines[count++] = IMAGE_FILE_MACHINE_ARM64; - supported_machines[count++] = IMAGE_FILE_MACHINE_AMD64; supported_machines[count++] = IMAGE_FILE_MACHINE_I386; + /* supported_machines[count++] = IMAGE_FILE_MACHINE_ARMNT; not supported yet */ } - supported_machines[count++] = IMAGE_FILE_MACHINE_ARMNT; #else #error Unsupported machine #endif @@ -1877,15 +1876,9 @@ void init_registry(void) static const WCHAR classes_i386[] = {'S','o','f','t','w','a','r','e','\\', 'C','l','a','s','s','e','s','\\', 'W','o','w','6','4','3','2','N','o','d','e'}; - static const WCHAR classes_amd64[] = {'S','o','f','t','w','a','r','e','\\', - 'C','l','a','s','s','e','s','\\', - 'W','o','w','6','4','6','4','N','o','d','e'}; static const WCHAR classes_arm[] = {'S','o','f','t','w','a','r','e','\\', 'C','l','a','s','s','e','s','\\', 'W','o','w','A','A','3','2','N','o','d','e'}; - static const WCHAR classes_arm64[] = {'S','o','f','t','w','a','r','e','\\', - 'C','l','a','s','s','e','s','\\', - 'W','o','w','A','A','6','4','N','o','d','e'}; static const WCHAR perflib[] = {'S','o','f','t','w','a','r','e','\\', 'M','i','c','r','o','s','o','f','t','\\', 'W','i','n','d','o','w','s',' ','N','T','\\', @@ -1956,8 +1949,7 @@ void init_registry(void) { case IMAGE_FILE_MACHINE_I386: name.str = classes_i386; name.len = sizeof(classes_i386); break; case IMAGE_FILE_MACHINE_ARMNT: name.str = classes_arm; name.len = sizeof(classes_arm); break; - case IMAGE_FILE_MACHINE_AMD64: name.str = classes_amd64; name.len = sizeof(classes_amd64); break; - case IMAGE_FILE_MACHINE_ARM64: name.str = classes_arm64; name.len = sizeof(classes_arm64); break; + default: continue; } if ((key = create_key_recursive( hklm, &name, current_time ))) {

1 0

Alexandre Julliard : ntdll: Fix stack layout for ARM syscalls.
by Alexandre Julliard Feb. 7, 2024

Feb. 7, 2024

Module: wine Branch: master Commit: b41566fb31b257d04f6044b01d996f71d2859493 URL: https://gitlab.winehq.org/wine/wine/-/commit/b41566fb31b257d04f6044b01d996f… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Wed Feb 7 11:52:30 2024 +0100 ntdll: Fix stack layout for ARM syscalls. --- dlls/ntdll/unix/signal_arm.c | 9 +++++---- include/wine/asm.h | 23 ++++------------------- 2 files changed, 9 insertions(+), 23 deletions(-) diff --git a/dlls/ntdll/unix/signal_arm.c b/dlls/ntdll/unix/signal_arm.c index 9ca0569bd62..0f6f81d2ee1 100644 --- a/dlls/ntdll/unix/signal_arm.c +++ b/dlls/ntdll/unix/signal_arm.c @@ -1052,7 +1052,10 @@ NTSTATUS call_user_apc_dispatcher( CONTEXT *context, ULONG_PTR arg1, ULONG_PTR a */ void call_raise_user_exception_dispatcher(void) { - arm_thread_data()->syscall_frame->pc = (DWORD)pKiRaiseUserExceptionDispatcher; + struct syscall_frame *frame = arm_thread_data()->syscall_frame; + + frame->sp += 16; + frame->pc = (DWORD)pKiRaiseUserExceptionDispatcher; } @@ -1608,8 +1611,7 @@ __ASM_GLOBAL_FUNC( __wine_syscall_dispatcher, "ldr r1, [r2, #0x1d8]\n\t" /* arm_thread_data()->syscall_frame */ "add r0, r1, #0x10\n\t" "stm r0, {r4-r12,lr}\n\t" - "add r0, sp, #0x10\n\t" - "str r0, [r1, #0x38]\n\t" + "str sp, [r1, #0x38]\n\t" "str r3, [r1, #0x3c]\n\t" "mrs r0, CPSR\n\t" "bfi r0, lr, #5, #1\n\t" /* set thumb bit */ @@ -1682,7 +1684,6 @@ __ASM_GLOBAL_FUNC( __wine_syscall_dispatcher, "5:\tmovw r0, #0x000d\n\t" /* STATUS_INVALID_PARAMETER */ "movt r0, #0xc000\n\t" - "add sp, sp, #0x10\n\t" "b " __ASM_LOCAL_LABEL("__wine_syscall_dispatcher_return") "\n\t" ".globl " __ASM_NAME("__wine_syscall_dispatcher_return") "\n" __ASM_NAME("__wine_syscall_dispatcher_return") ":\n\t" diff --git a/include/wine/asm.h b/include/wine/asm.h index 0fa2dfcd1b7..908c7d7d19a 100644 --- a/include/wine/asm.h +++ b/include/wine/asm.h @@ -268,34 +268,19 @@ # define __ASM_SYSCALL_FUNC(id,name,args) \ __ASM_GLOBAL_FUNC( name, \ "push {r0-r3}\n\t" \ + ".seh_save_regs {r0-r3}\n\t" \ + ".seh_endprologue\n\t" \ "movw ip, #(" #id ")\n\t" \ "mov r3, lr\n\t" \ "bl " __ASM_NAME("__wine_syscall") "\n\t" \ + "add sp, #16\n\t" \ "bx lr" ) -# ifndef __PIC__ -# define DEFINE_SYSCALL_HELPER32() \ +# define DEFINE_SYSCALL_HELPER32() \ __ASM_GLOBAL_FUNC( __wine_syscall, \ "movw r0, :lower16:" __ASM_NAME("__wine_syscall_dispatcher") "\n\t" \ "movt r0, :upper16:" __ASM_NAME("__wine_syscall_dispatcher") "\n\t" \ "ldr r0, [r0]\n\t" \ "bx r0" ) -# elif defined __thumb__ -# define DEFINE_SYSCALL_HELPER32() \ - __ASM_GLOBAL_FUNC( __wine_syscall, \ - "ldr r0, 2f\n" \ - "1:\tadd r0, pc\n\t" \ - "ldr r0, [r0]\n\t" \ - "bx r0\n" \ - "2:\t.long " __ASM_NAME("__wine_syscall_dispatcher") "-1b-4" ) -# else -# define DEFINE_SYSCALL_HELPER32() \ - __ASM_GLOBAL_FUNC( __wine_syscall, \ - "ldr r0, 2f\n" \ - "1:\tadd r0, pc\n\t" \ - "ldr r0, [r0]\n\t" \ - "bx r0\n" \ - "2:\t.long " __ASM_NAME("__wine_syscall_dispatcher") "-1b-8" ) -# endif #endif #endif /* __WINE_WINE_ASM_H */

1 0

Conor McCarthy : vkd3d-shader/spirv: Handle the ACOS, ASIN and ATAN instructions in spirv_compiler_emit_ext_glsl_instruction().
by Alexandre Julliard Feb. 6, 2024

Feb. 6, 2024

Module: vkd3d Branch: master Commit: 7f87a3e5fcfd940eea352083dfb15bbb6da96258 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/7f87a3e5fcfd940eea352083dfb15… Author: Conor McCarthy <cmccarthy(a)codeweavers.com> Date: Mon Jan 22 16:00:47 2024 +1000 vkd3d-shader/spirv: Handle the ACOS, ASIN and ATAN instructions in spirv_compiler_emit_ext_glsl_instruction(). --- libs/vkd3d-shader/spirv.c | 6 ++++ tests/hlsl/inverse-trig.shader_test | 60 ++++++++++++++++++------------------- 2 files changed, 36 insertions(+), 30 deletions(-) diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index e9437c6f..5f7a33da 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -6997,6 +6997,9 @@ static enum GLSLstd450 spirv_compiler_map_ext_glsl_instruction( } glsl_insts[] = { + {VKD3DSIH_ACOS, GLSLstd450Acos}, + {VKD3DSIH_ASIN, GLSLstd450Asin}, + {VKD3DSIH_ATAN, GLSLstd450Atan}, {VKD3DSIH_DFMA, GLSLstd450Fma}, {VKD3DSIH_DMAX, GLSLstd450NMax}, {VKD3DSIH_DMIN, GLSLstd450NMin}, @@ -9581,6 +9584,9 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, case VKD3DSIH_ISFINITE: spirv_compiler_emit_isfinite(compiler, instruction); break; + case VKD3DSIH_ACOS: + case VKD3DSIH_ASIN: + case VKD3DSIH_ATAN: case VKD3DSIH_DFMA: case VKD3DSIH_DMAX: case VKD3DSIH_DMIN: diff --git a/tests/hlsl/inverse-trig.shader_test b/tests/hlsl/inverse-trig.shader_test index dd069fef..0c1fdc01 100644 --- a/tests/hlsl/inverse-trig.shader_test +++ b/tests/hlsl/inverse-trig.shader_test @@ -13,23 +13,23 @@ float4 main() : sv_target [test] uniform 0 float4 -1.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (3.14159274, 0.0, 0.0, 0.0) 128 uniform 0 float4 -0.5 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (2.094441441, 0.0, 0.0, 0.0) 256 uniform 0 float4 0.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (1.57072878, 0.0, 0.0, 0.0) 1024 uniform 0 float4 0.5 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (1.04715133, 0.0, 0.0, 0.0) 512 uniform 0 float4 1.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) 128 [pixel shader todo] @@ -44,7 +44,7 @@ float4 main() : sv_target [test] uniform 0 float4 -1.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (-31416.0, 0.0, 0.0, 0.0) [require] @@ -72,15 +72,15 @@ shader model >= 6.0 % RADV are a bit lower than these, hence the large max ulp difference. [test] uniform 0 float4 -0.5 0.0 0.0 0.0 -todo draw quad +draw quad probe all rgba (-10472.0, 0.0, 0.0, 0.0) 4096 uniform 0 float4 0.0 0.0 0.0 0.0 -todo draw quad +draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) uniform 0 float4 0.5 0.0 0.0 0.0 -todo draw quad +draw quad probe all rgba (10472.0, 0.0, 0.0, 0.0) 4096 [require] @@ -88,7 +88,7 @@ probe all rgba (10472.0, 0.0, 0.0, 0.0) 4096 [test] uniform 0 float4 1.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (31416.0, 0.0, 0.0, 0.0) @@ -102,23 +102,23 @@ float4 main() : sv_target [test] uniform 0 float4 -1.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (-0.785409629, 0.0, 0.0, 0.0) 512 uniform 0 float4 -0.5 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (-0.4636476, 0.0, 0.0, 0.0) 256 uniform 0 float4 0.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) 256 uniform 0 float4 0.5 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (0.4636476, 0.0, 0.0, 0.0) 256 uniform 0 float4 1.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (0.785409629, 0.0, 0.0, 0.0) 512 [pixel shader todo] @@ -133,64 +133,64 @@ float4 main() : sv_target [test] % Non-degenerate cases uniform 0 float4 1.0 1.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (0.785385, 0.0, 0.0, 0.0) 512 uniform 0 float4 5.0 -5.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (2.356194, 0.0, 0.0, 0.0) 256 uniform 0 float4 -3.0 -3.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (-2.356194, 0.0, 0.0, 0.0) 256 uniform 0 float4 1.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (1.570796, 0.0, 0.0, 0.0) 256 uniform 0 float4 -1.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (-1.570796, 0.0, 0.0, 0.0) 256 uniform 0 float4 0.0 1.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (0.0, 0.0, 0.0, 0.0) 256 uniform 0 float4 0.0 -1.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256 % Degenerate cases uniform 0 float4 0.00001 0.00002 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (0.463647, 0.0, 0.0, 0.0) 256 uniform 0 float4 0.00001 -0.00002 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (2.677945, 0.0, 0.0, 0.0) 256 uniform 0 float4 -0.00001 100000.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (-0.000000000099986595, 0.0, 0.0, 0.0) 2048 uniform 0 float4 10000000.0 0.00000001 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (1.570796, 0.0, 0.0, 0.0) 256 % Negative zero behavior should be to treat it the % same as normal zero. uniform 0 float4 1000000000.0 0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (1.570796, 0.0, 0.0, 0.0) 256 uniform 0 float4 1000000000.0 -0.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (1.570796, 0.0, 0.0, 0.0) 256 uniform 0 float4 0.0 -1.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256 uniform 0 float4 -0.0 -1.0 0.0 0.0 -todo draw quad +todo(sm<6) draw quad probe all rgba (3.1415927, 0.0, 0.0, 0.0) 256

1 0

Conor McCarthy : vkd3d-shader/dxil: Handle inverse trigonometric functions in sm6_parser_emit_dx_unary().
by Alexandre Julliard Feb. 6, 2024

Feb. 6, 2024

Module: vkd3d Branch: master Commit: 19d23fa696523ec4f46c5cde75009a5d76f533b3 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/19d23fa696523ec4f46c5cde75009… Author: Conor McCarthy <cmccarthy(a)codeweavers.com> Date: Mon Jan 22 15:26:16 2024 +1000 vkd3d-shader/dxil: Handle inverse trigonometric functions in sm6_parser_emit_dx_unary(). --- libs/vkd3d-shader/d3d_asm.c | 3 +++ libs/vkd3d-shader/dxil.c | 12 ++++++++++++ libs/vkd3d-shader/vkd3d_shader_private.h | 3 +++ 3 files changed, 18 insertions(+) diff --git a/libs/vkd3d-shader/d3d_asm.c b/libs/vkd3d-shader/d3d_asm.c index dd96b7fa..3ae736a8 100644 --- a/libs/vkd3d-shader/d3d_asm.c +++ b/libs/vkd3d-shader/d3d_asm.c @@ -30,8 +30,11 @@ static const char * const shader_opcode_names[] = { [VKD3DSIH_ABS ] = "abs", + [VKD3DSIH_ACOS ] = "acos", [VKD3DSIH_ADD ] = "add", [VKD3DSIH_AND ] = "and", + [VKD3DSIH_ASIN ] = "asin", + [VKD3DSIH_ATAN ] = "atan", [VKD3DSIH_ATOMIC_AND ] = "atomic_and", [VKD3DSIH_ATOMIC_CMP_STORE ] = "atomic_cmp_store", [VKD3DSIH_ATOMIC_IADD ] = "atomic_iadd", diff --git a/libs/vkd3d-shader/dxil.c b/libs/vkd3d-shader/dxil.c index 291012d1..e718ed84 100644 --- a/libs/vkd3d-shader/dxil.c +++ b/libs/vkd3d-shader/dxil.c @@ -339,6 +339,9 @@ enum dx_intrinsic_opcode DX_COS = 12, DX_SIN = 13, DX_TAN = 14, + DX_ACOS = 15, + DX_ASIN = 16, + DX_ATAN = 17, DX_EXP = 21, DX_FRC = 22, DX_LOG = 23, @@ -3625,6 +3628,12 @@ static enum vkd3d_shader_opcode map_dx_unary_op(enum dx_intrinsic_opcode op) return VKD3DSIH_ISFINITE; case DX_TAN: return VKD3DSIH_TAN; + case DX_ACOS: + return VKD3DSIH_ACOS; + case DX_ASIN: + return VKD3DSIH_ASIN; + case DX_ATAN: + return VKD3DSIH_ATAN; case DX_EXP: return VKD3DSIH_EXP; case DX_FRC: @@ -4127,6 +4136,9 @@ struct sm6_dx_opcode_info */ static const struct sm6_dx_opcode_info sm6_dx_op_table[] = { + [DX_ACOS ] = {"g", "R", sm6_parser_emit_dx_unary}, + [DX_ASIN ] = {"g", "R", sm6_parser_emit_dx_unary}, + [DX_ATAN ] = {"g", "R", sm6_parser_emit_dx_unary}, [DX_BFREV ] = {"m", "R", sm6_parser_emit_dx_unary}, [DX_BUFFER_LOAD ] = {"o", "Hii", sm6_parser_emit_dx_buffer_load}, [DX_CBUFFER_LOAD_LEGACY ] = {"o", "Hi", sm6_parser_emit_dx_cbuffer_load}, diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index 51320ad0..3d043c24 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -225,8 +225,11 @@ enum vkd3d_shader_error enum vkd3d_shader_opcode { VKD3DSIH_ABS, + VKD3DSIH_ACOS, VKD3DSIH_ADD, VKD3DSIH_AND, + VKD3DSIH_ASIN, + VKD3DSIH_ATAN, VKD3DSIH_ATOMIC_AND, VKD3DSIH_ATOMIC_CMP_STORE, VKD3DSIH_ATOMIC_IADD,

1 0

Henri Verbeet : vkd3d-common: Get rid of InterlockedIncrement().
by Alexandre Julliard Feb. 6, 2024

Feb. 6, 2024

Module: vkd3d Branch: master Commit: c64921e79b5e26bf0e39cf2c188979db85f8d3b3 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/c64921e79b5e26bf0e39cf2c18897… Author: Henri Verbeet <hverbeet(a)codeweavers.com> Date: Tue Jan 30 12:54:54 2024 +0100 vkd3d-common: Get rid of InterlockedIncrement(). --- include/private/vkd3d_common.h | 5 ----- include/private/vkd3d_test.h | 40 +++++++++++++++++++--------------------- tests/vkd3d_api.c | 6 +++--- 3 files changed, 22 insertions(+), 29 deletions(-) diff --git a/include/private/vkd3d_common.h b/include/private/vkd3d_common.h index 01e219b3..6f878cce 100644 --- a/include/private/vkd3d_common.h +++ b/include/private/vkd3d_common.h @@ -305,11 +305,6 @@ static inline uint32_t vkd3d_atomic_increment_u32(uint32_t volatile *x) } #ifndef _WIN32 -static inline LONG InterlockedIncrement(LONG volatile *x) -{ - return vkd3d_atomic_increment_u32((uint32_t *)x); -} - static inline LONG InterlockedDecrement(LONG volatile *x) { return vkd3d_atomic_decrement_u32((uint32_t *)x); diff --git a/include/private/vkd3d_test.h b/include/private/vkd3d_test.h index 081443c4..a337ac07 100644 --- a/include/private/vkd3d_test.h +++ b/include/private/vkd3d_test.h @@ -101,12 +101,12 @@ static void vkd3d_test_end_todo(void); struct vkd3d_test_state { - LONG success_count; - LONG failure_count; - LONG skip_count; - LONG todo_count; - LONG todo_success_count; - LONG bug_count; + unsigned int success_count; + unsigned int failure_count; + unsigned int skip_count; + unsigned int todo_count; + unsigned int todo_success_count; + unsigned int bug_count; unsigned int debug_level; @@ -150,13 +150,13 @@ vkd3d_test_check_assert_that(unsigned int line, bool result, const char *fmt, va { if (result) { - InterlockedIncrement(&vkd3d_test_state.success_count); + vkd3d_atomic_increment_u32(&vkd3d_test_state.success_count); if (vkd3d_test_state.debug_level > 1) vkd3d_test_printf(line, "Test succeeded.\n"); } else { - InterlockedIncrement(&vkd3d_test_state.failure_count); + vkd3d_atomic_increment_u32(&vkd3d_test_state.failure_count); vkd3d_test_printf(line, "Test failed: "); vprintf(fmt, args); } @@ -180,7 +180,7 @@ vkd3d_test_check_ok(unsigned int line, bool result, const char *fmt, va_list arg if (is_bug && vkd3d_test_state.bug_enabled) { - InterlockedIncrement(&vkd3d_test_state.bug_count); + vkd3d_atomic_increment_u32(&vkd3d_test_state.bug_count); if (is_todo) result = !result; if (result) @@ -193,12 +193,12 @@ vkd3d_test_check_ok(unsigned int line, bool result, const char *fmt, va_list arg { if (result) { - InterlockedIncrement(&vkd3d_test_state.todo_success_count); + vkd3d_atomic_increment_u32(&vkd3d_test_state.todo_success_count); vkd3d_test_printf(line, "Todo succeeded: "); } else { - InterlockedIncrement(&vkd3d_test_state.todo_count); + vkd3d_atomic_increment_u32(&vkd3d_test_state.todo_count); vkd3d_test_printf(line, "Todo: "); } vprintf(fmt, args); @@ -227,7 +227,7 @@ vkd3d_test_skip(unsigned int line, const char *fmt, ...) vkd3d_test_printf(line, "Test skipped: "); vprintf(fmt, args); va_end(args); - InterlockedIncrement(&vkd3d_test_state.skip_count); + vkd3d_atomic_increment_u32(&vkd3d_test_state.skip_count); } static void VKD3D_PRINTF_FUNC(2, 3) VKD3D_UNUSED @@ -293,16 +293,14 @@ int main(int argc, char **argv) vkd3d_test_main(argc, argv); - printf("%s: %lu tests executed (%lu failures, %lu skipped, %lu todo, %lu bugs).\n", + printf("%s: %u tests executed (%u failures, %u skipped, %u todo, %u bugs).\n", vkd3d_test_name, - (unsigned long)(vkd3d_test_state.success_count - + vkd3d_test_state.failure_count + vkd3d_test_state.todo_count - + vkd3d_test_state.todo_success_count), - (unsigned long)(vkd3d_test_state.failure_count - + vkd3d_test_state.todo_success_count), - (unsigned long)vkd3d_test_state.skip_count, - (unsigned long)vkd3d_test_state.todo_count, - (unsigned long)vkd3d_test_state.bug_count); + vkd3d_test_state.success_count + vkd3d_test_state.failure_count + + vkd3d_test_state.todo_count + vkd3d_test_state.todo_success_count, + vkd3d_test_state.failure_count + vkd3d_test_state.todo_success_count, + vkd3d_test_state.skip_count, + vkd3d_test_state.todo_count, + vkd3d_test_state.bug_count); if (test_platform) free(test_platform); diff --git a/tests/vkd3d_api.c b/tests/vkd3d_api.c index 5a585fec..29948783 100644 --- a/tests/vkd3d_api.c +++ b/tests/vkd3d_api.c @@ -625,7 +625,7 @@ static void test_adapter_luid(void) struct parent { IUnknown IUnknown_iface; - LONG refcount; + unsigned int refcount; }; static struct parent *parent_from_IUnknown(IUnknown *iface) @@ -651,14 +651,14 @@ static ULONG STDMETHODCALLTYPE parent_AddRef(IUnknown *iface) { struct parent *parent = parent_from_IUnknown(iface); - return InterlockedIncrement(&parent->refcount); + return vkd3d_atomic_increment_u32(&parent->refcount); } static ULONG STDMETHODCALLTYPE parent_Release(IUnknown *iface) { struct parent *parent = parent_from_IUnknown(iface); - return InterlockedDecrement(&parent->refcount); + return InterlockedDecrement((LONG *)&parent->refcount); } static const struct IUnknownVtbl parent_vtbl =

1 0

Henri Verbeet : vkd3d-common: Get rid of InterlockedDecrement().
by Alexandre Julliard Feb. 6, 2024

Feb. 6, 2024

Module: vkd3d Branch: master Commit: 532e902a5630c8ac606c06bd25b1bcbab14f9dc3 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/532e902a5630c8ac606c06bd25b1b… Author: Henri Verbeet <hverbeet(a)codeweavers.com> Date: Tue Jan 30 12:59:02 2024 +0100 vkd3d-common: Get rid of InterlockedDecrement(). --- include/private/vkd3d_common.h | 7 ------- tests/vkd3d_api.c | 2 +- 2 files changed, 1 insertion(+), 8 deletions(-) diff --git a/include/private/vkd3d_common.h b/include/private/vkd3d_common.h index 6f878cce..8d6e7b37 100644 --- a/include/private/vkd3d_common.h +++ b/include/private/vkd3d_common.h @@ -304,13 +304,6 @@ static inline uint32_t vkd3d_atomic_increment_u32(uint32_t volatile *x) return vkd3d_atomic_add_fetch_u32(x, 1); } -#ifndef _WIN32 -static inline LONG InterlockedDecrement(LONG volatile *x) -{ - return vkd3d_atomic_decrement_u32((uint32_t *)x); -} -#endif /* _WIN32 */ - static inline void vkd3d_parse_version(const char *version, int *major, int *minor) { *major = atoi(version); diff --git a/tests/vkd3d_api.c b/tests/vkd3d_api.c index 29948783..62e30b47 100644 --- a/tests/vkd3d_api.c +++ b/tests/vkd3d_api.c @@ -658,7 +658,7 @@ static ULONG STDMETHODCALLTYPE parent_Release(IUnknown *iface) { struct parent *parent = parent_from_IUnknown(iface); - return InterlockedDecrement((LONG *)&parent->refcount); + return vkd3d_atomic_decrement_u32(&parent->refcount); } static const struct IUnknownVtbl parent_vtbl =

1 0

Petrichor Park : tests/shader-runner: Add tests for acos and asin trig intrinsics.
by Alexandre Julliard Feb. 6, 2024

Feb. 6, 2024

Module: vkd3d Branch: master Commit: 92044d5e4114ae65b77b232fba442f347f21069d URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/92044d5e4114ae65b77b232fba442… Author: Petrichor Park <ppark(a)codeweavers.com> Date: Wed Sep 6 15:14:12 2023 -0500 tests/shader-runner: Add tests for acos and asin trig intrinsics. Extracted by Conor McCarthy from an HLSL patch, and modified to include SM 6 variations. --- Makefile.am | 1 + tests/hlsl/inverse-trig.shader_test | 92 +++++++++++++++++++++++++++++++++++++ 2 files changed, 93 insertions(+) diff --git a/Makefile.am b/Makefile.am index 90e7dcfc..1687d9a0 100644 --- a/Makefile.am +++ b/Makefile.am @@ -121,6 +121,7 @@ vkd3d_shader_tests = \ tests/hlsl/initializer-struct.shader_test \ tests/hlsl/intrinsic-override.shader_test \ tests/hlsl/invalid.shader_test \ + tests/hlsl/inverse-trig.shader_test \ tests/hlsl/is-front-face.shader_test \ tests/hlsl/ldexp.shader_test \ tests/hlsl/length.shader_test \ diff --git a/tests/hlsl/inverse-trig.shader_test b/tests/hlsl/inverse-trig.shader_test new file mode 100644 index 00000000..6077b4b5 --- /dev/null +++ b/tests/hlsl/inverse-trig.shader_test @@ -0,0 +1,92 @@ +% TPF does not define instructions for inverse trig; these intrinsics are +% decomposed into other instructions. FXC emits code which may vary wrt other +% implementations. DXIL defines intrinsics for inverse trig, to be implemented +% by the backend. + +[pixel shader todo] +uniform float4 a; + +float4 main() : sv_target +{ + return float4(acos(a.x), 0.0, 0.0, 0.0); +} + +[test] +uniform 0 float4 -1.0 0.0 0.0 0.0 +todo draw quad +probe all rgba (3.14159274, 0.0, 0.0, 0.0) 128 + +uniform 0 float4 -0.5 0.0 0.0 0.0 +todo draw quad +probe all rgba (2.094441441, 0.0, 0.0, 0.0) 256 + +uniform 0 float4 0.0 0.0 0.0 0.0 +todo draw quad +probe all rgba (1.57072878, 0.0, 0.0, 0.0) 1024 + +uniform 0 float4 0.5 0.0 0.0 0.0 +todo draw quad +probe all rgba (1.04715133, 0.0, 0.0, 0.0) 512 + +uniform 0 float4 1.0 0.0 0.0 0.0 +todo draw quad +probe all rgba (0.0, 0.0, 0.0, 0.0) 128 + +[pixel shader todo] +uniform float4 a; + +float4 main() : sv_target +{ + float4 result = float4(asin(a.x), 0.0, 0.0, 0.0); + // Quantize to cover implementation variations, which are quite large for asin(). + return round(result * 20000.0); +} + +[test] +uniform 0 float4 -1.0 0.0 0.0 0.0 +todo draw quad +probe all rgba (-31416.0, 0.0, 0.0, 0.0) + +[require] +shader model < 6.0 + +[test] +uniform 0 float4 -0.5 0.0 0.0 0.0 +todo draw quad +probe all rgba (-10473.0, 0.0, 0.0, 0.0) + +uniform 0 float4 0.0 0.0 0.0 0.0 +todo draw quad +probe all rgba (1.0, 0.0, 0.0, 0.0) + +uniform 0 float4 0.5 0.0 0.0 0.0 +todo draw quad +probe all rgba (10473.0, 0.0, 0.0, 0.0) + +[require] +shader model >= 6.0 + +% We implement SM 6.0 inverse trig instructions using the native equivalents +% available in the backend. The values below are from the AMD Windows drivers, +% which are very close to those from Ubuntu's calculator app. Results from +% RADV are a bit lower than these, hence the large max ulp difference. +[test] +uniform 0 float4 -0.5 0.0 0.0 0.0 +todo draw quad +probe all rgba (-10472.0, 0.0, 0.0, 0.0) 4096 + +uniform 0 float4 0.0 0.0 0.0 0.0 +todo draw quad +probe all rgba (0.0, 0.0, 0.0, 0.0) + +uniform 0 float4 0.5 0.0 0.0 0.0 +todo draw quad +probe all rgba (10472.0, 0.0, 0.0, 0.0) 4096 + +[require] +% reset requirements + +[test] +uniform 0 float4 1.0 0.0 0.0 0.0 +todo draw quad +probe all rgba (31416.0, 0.0, 0.0, 0.0)

1 0

Henri Verbeet : vkd3d: Introduce a debug helper for GPU descriptor handles.
by Alexandre Julliard Feb. 6, 2024

Feb. 6, 2024

Module: vkd3d Branch: master Commit: 51cdddb96141587a99e17237370194adc14290fe URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/51cdddb96141587a99e1723737019… Author: Henri Verbeet <hverbeet(a)codeweavers.com> Date: Tue Jan 30 15:12:56 2024 +0100 vkd3d: Introduce a debug helper for GPU descriptor handles. --- libs/vkd3d/command.c | 16 ++++++++-------- libs/vkd3d/utils.c | 5 +++++ libs/vkd3d/vkd3d_private.h | 1 + 3 files changed, 14 insertions(+), 8 deletions(-) diff --git a/libs/vkd3d/command.c b/libs/vkd3d/command.c index f9940b3d..609300d3 100644 --- a/libs/vkd3d/command.c +++ b/libs/vkd3d/command.c @@ -4499,8 +4499,8 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootDescriptorTable(I { struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); - TRACE("iface %p, root_parameter_index %u, base_descriptor %#"PRIx64".\n", - iface, root_parameter_index, base_descriptor.ptr); + TRACE("iface %p, root_parameter_index %u, base_descriptor %s.\n", + iface, root_parameter_index, debug_gpu_handle(base_descriptor)); d3d12_command_list_set_descriptor_table(list, VKD3D_PIPELINE_BIND_POINT_COMPUTE, root_parameter_index, base_descriptor); @@ -4511,8 +4511,8 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootDescriptorTable( { struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); - TRACE("iface %p, root_parameter_index %u, base_descriptor %#"PRIx64".\n", - iface, root_parameter_index, base_descriptor.ptr); + TRACE("iface %p, root_parameter_index %u, base_descriptor %s.\n", + iface, root_parameter_index, debug_gpu_handle(base_descriptor)); d3d12_command_list_set_descriptor_table(list, VKD3D_PIPELINE_BIND_POINT_GRAPHICS, root_parameter_index, base_descriptor); @@ -5432,8 +5432,8 @@ static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewUint(ID struct d3d12_resource *resource_impl; VkClearColorValue colour; - TRACE("iface %p, gpu_handle %#"PRIx64", cpu_handle %s, resource %p, values %p, rect_count %u, rects %p.\n", - iface, gpu_handle.ptr, debug_cpu_handle(cpu_handle), resource, values, rect_count, rects); + TRACE("iface %p, gpu_handle %s, cpu_handle %s, resource %p, values %p, rect_count %u, rects %p.\n", + iface, debug_gpu_handle(gpu_handle), debug_cpu_handle(cpu_handle), resource, values, rect_count, rects); resource_impl = unsafe_impl_from_ID3D12Resource(resource); if (!(descriptor = d3d12_desc_from_cpu_handle(cpu_handle)->s.u.view)) @@ -5496,8 +5496,8 @@ static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewFloat(I VkClearColorValue colour; struct vkd3d_view *view; - TRACE("iface %p, gpu_handle %#"PRIx64", cpu_handle %s, resource %p, values %p, rect_count %u, rects %p.\n", - iface, gpu_handle.ptr, debug_cpu_handle(cpu_handle), resource, values, rect_count, rects); + TRACE("iface %p, gpu_handle %s, cpu_handle %s, resource %p, values %p, rect_count %u, rects %p.\n", + iface, debug_gpu_handle(gpu_handle), debug_cpu_handle(cpu_handle), resource, values, rect_count, rects); resource_impl = unsafe_impl_from_ID3D12Resource(resource); if (!(view = d3d12_desc_from_cpu_handle(cpu_handle)->s.u.view)) diff --git a/libs/vkd3d/utils.c b/libs/vkd3d/utils.c index 28913a83..ac79ae5d 100644 --- a/libs/vkd3d/utils.c +++ b/libs/vkd3d/utils.c @@ -678,6 +678,11 @@ const char *debug_d3d12_shader_component_mapping(unsigned int mapping) debug_d3d12_shader_component(D3D12_DECODE_SHADER_4_COMPONENT_MAPPING(3, mapping))); } +const char *debug_gpu_handle(D3D12_GPU_DESCRIPTOR_HANDLE handle) +{ + return vkd3d_dbg_sprintf("{%#"PRIx64"}", handle.ptr); +} + const char *debug_vk_extent_3d(VkExtent3D extent) { return vkd3d_dbg_sprintf("(%u, %u, %u)", diff --git a/libs/vkd3d/vkd3d_private.h b/libs/vkd3d/vkd3d_private.h index bb78ba77..5955e616 100644 --- a/libs/vkd3d/vkd3d_private.h +++ b/libs/vkd3d/vkd3d_private.h @@ -1963,6 +1963,7 @@ HRESULT return_interface(void *iface, REFIID iface_iid, REFIID requested_iid, vo const char *debug_cpu_handle(D3D12_CPU_DESCRIPTOR_HANDLE handle); const char *debug_d3d12_box(const D3D12_BOX *box); const char *debug_d3d12_shader_component_mapping(unsigned int mapping); +const char *debug_gpu_handle(D3D12_GPU_DESCRIPTOR_HANDLE handle); const char *debug_vk_extent_3d(VkExtent3D extent); const char *debug_vk_memory_heap_flags(VkMemoryHeapFlags flags); const char *debug_vk_memory_property_flags(VkMemoryPropertyFlags flags);

1 0