January 2024 - Wine-commits

Alexandre Julliard : configure: Reset host flags in all cross-compiler error paths.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: wine Branch: master Commit: e5becbf6a8528a08049c3575ff3447b863085ded URL: https://gitlab.winehq.org/wine/wine/-/commit/e5becbf6a8528a08049c3575ff3447… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Sun Jan 28 15:29:08 2024 +0100 configure: Reset host flags in all cross-compiler error paths. --- configure | 21 ++++++++------------- configure.ac | 21 ++++++++------------- 2 files changed, 16 insertions(+), 26 deletions(-) diff --git a/configure b/configure index ae5def6a6b1..165cd9c73e6 100755 --- a/configure +++ b/configure @@ -10340,6 +10340,10 @@ else $as_nop fi fi +saved_CC=$CC +saved_CFLAGS=$CFLAGS +saved_LDFLAGS=$LDFLAGS + for wine_arch in $cross_archs do case "x$with_mingw" in @@ -10603,10 +10607,6 @@ test -n "$x86_64_CC" || x86_64_CC="false" esac fi - saved_CC=$CC - saved_CFLAGS=$CFLAGS - saved_LDFLAGS=$LDFLAGS - CFLAGS=${CROSSCFLAGS:-"-g -O2"} LDFLAGS=$CROSSLDFLAGS eval CC=\$${wine_arch}_CC @@ -10768,9 +10768,6 @@ then : as_fn_error $? "MinGW $wine_arch compiler not found. This is an error since --enable-archs=$wine_arch was requested." "$LINENO" 5 fi - CC=$saved_CC - CFLAGS=$saved_CFLAGS - LDFLAGS=$saved_LDFLAGS continue fi @@ -10814,9 +10811,6 @@ fi printf "%s\n" "none needed" >&6; } ;; xno) { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5 printf "%s\n" "unsupported" >&6; } - CC=$saved_CC - CFLAGS=$saved_CFLAGS - LDFLAGS=$saved_LDFLAGS if test ${enable_archs+y} then : as_fn_error $? "MinGW $wine_arch compiler supporting C99 not found. @@ -11964,11 +11958,12 @@ then : as_fn_append ${wine_arch}_LDFLAGS " -Wl,--build-id" fi } - CC=$saved_CC - CFLAGS=$saved_CFLAGS - LDFLAGS=$saved_LDFLAGS done +CC=$saved_CC +CFLAGS=$saved_CFLAGS +LDFLAGS=$saved_LDFLAGS + if test $HOST_ARCH = aarch64 then test "x$PE_ARCHS" != x || as_fn_error $? "PE cross-compilation is required for ARM64, please install clang/llvm-dlltool/lld, or llvm-mingw." "$LINENO" 5 diff --git a/configure.ac b/configure.ac index 475743bc121..b2ec7c4c7c8 100644 --- a/configure.ac +++ b/configure.ac @@ -833,6 +833,10 @@ AS_VAR_SET_IF([enable_archs], test $HOST_ARCH = unknown || cross_archs=$HOST_ARCH fi]) +saved_CC=$CC +saved_CFLAGS=$CFLAGS +saved_LDFLAGS=$LDFLAGS + for wine_arch in $cross_archs do case "x$with_mingw" in @@ -859,10 +863,6 @@ do ;; esac]) - saved_CC=$CC - saved_CFLAGS=$CFLAGS - saved_LDFLAGS=$LDFLAGS - CFLAGS=${CROSSCFLAGS:-"-g -O2"} LDFLAGS=$CROSSLDFLAGS AS_VAR_COPY([CC],[${wine_arch}_CC]) @@ -924,9 +924,6 @@ do AS_VAR_SET_IF([enable_archs], [AC_MSG_ERROR([MinGW $wine_arch compiler not found. This is an error since --enable-archs=$wine_arch was requested.])]) - CC=$saved_CC - CFLAGS=$saved_CFLAGS - LDFLAGS=$saved_LDFLAGS continue fi @@ -948,9 +945,6 @@ This is an error since --enable-archs=$wine_arch was requested.])]) case "x$res" in x) AC_MSG_RESULT([none needed]) ;; xno) AC_MSG_RESULT([unsupported]) - CC=$saved_CC - CFLAGS=$saved_CFLAGS - LDFLAGS=$saved_LDFLAGS AS_VAR_SET_IF([enable_archs], [AC_MSG_ERROR([MinGW $wine_arch compiler supporting C99 not found. This is an error since --enable-archs=$wine_arch was requested.])]) @@ -1040,11 +1034,12 @@ This is an error since --enable-archs=$wine_arch was requested.])]) [AS_VAR_APPEND([${wine_arch}_CFLAGS],[" -Wl,--build-id"]) AS_VAR_APPEND([${wine_arch}_LDFLAGS],[" -Wl,--build-id"])]) - CC=$saved_CC - CFLAGS=$saved_CFLAGS - LDFLAGS=$saved_LDFLAGS done +CC=$saved_CC +CFLAGS=$saved_CFLAGS +LDFLAGS=$saved_LDFLAGS + if test $HOST_ARCH = aarch64 then test "x$PE_ARCHS" != x || AC_MSG_ERROR([PE cross-compilation is required for ARM64, please install clang/llvm-dlltool/lld, or llvm-mingw.])

1 0

Alexandre Julliard : winsta: Start time is an input parameter in WinStationGetProcessSid.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: wine Branch: master Commit: 2e9da0ba24ab7f9c5a6b5547c5a20602210db043 URL: https://gitlab.winehq.org/wine/wine/-/commit/2e9da0ba24ab7f9c5a6b5547c5a206… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Sun Jan 28 15:52:52 2024 +0100 winsta: Start time is an input parameter in WinStationGetProcessSid. Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=56256 --- dlls/winsta/main.c | 6 ++++-- dlls/winsta/winsta.h | 2 +- dlls/winsta/winsta.spec | 2 +- 3 files changed, 6 insertions(+), 4 deletions(-) diff --git a/dlls/winsta/main.c b/dlls/winsta/main.c index 8e054f14314..80588392ca4 100644 --- a/dlls/winsta/main.c +++ b/dlls/winsta/main.c @@ -66,10 +66,12 @@ BOOLEAN WINAPI WinStationGetAllProcesses( HANDLE server, ULONG level, return FALSE; } -BOOLEAN WINAPI WinStationGetProcessSid( HANDLE server, ULONG process_id, FILETIME *process_start_time, +BOOLEAN WINAPI WinStationGetProcessSid( HANDLE server, ULONG process_id, FILETIME process_start_time, PVOID process_user_sid, PULONG sid_size ) { - FIXME( "(%p, %ld, %p, %p, %p): stub\n", server, process_id, process_start_time, process_user_sid, sid_size); + FIXME( "(%p, %ld, %I64x, %p, %p): stub\n", server, process_id, + ((UINT64)process_start_time.dwHighDateTime << 32) | process_start_time.dwLowDateTime, + process_user_sid, sid_size); SetLastError( ERROR_CALL_NOT_IMPLEMENTED ); return FALSE; } diff --git a/dlls/winsta/winsta.h b/dlls/winsta/winsta.h index c4f5e986f16..bbe84a03cb7 100644 --- a/dlls/winsta/winsta.h +++ b/dlls/winsta/winsta.h @@ -87,7 +87,7 @@ typedef struct _TS_ALL_PROCESSES_INFO BOOLEAN WINAPI WinStationEnumerateW(HANDLE,PSESSIONIDW *,PULONG); BOOLEAN WINAPI WinStationGetAllProcesses(HANDLE,ULONG,PULONG,PTS_ALL_PROCESSES_INFO *); -BOOLEAN WINAPI WinStationGetProcessSid(HANDLE,ULONG,FILETIME *,PVOID,PULONG); +BOOLEAN WINAPI WinStationGetProcessSid(HANDLE,ULONG,FILETIME,PVOID,PULONG); BOOLEAN WINAPI WinStationQueryInformationW(HANDLE,ULONG,WINSTATIONINFOCLASS,PVOID,ULONG,PULONG); BOOLEAN WINAPI WinStationRegisterConsoleNotification(HANDLE,HWND,ULONG); BOOLEAN WINAPI WinStationUnRegisterConsoleNotification(HANDLE,HWND); diff --git a/dlls/winsta/winsta.spec b/dlls/winsta/winsta.spec index 1f38a238f0c..a582a7336bb 100644 --- a/dlls/winsta/winsta.spec +++ b/dlls/winsta/winsta.spec @@ -39,7 +39,7 @@ @ stub WinStationGetLanAdapterNameA @ stub WinStationGetLanAdapterNameW @ stub WinStationGetMachinePolicy -@ stdcall WinStationGetProcessSid(ptr long ptr ptr ptr) +@ stdcall WinStationGetProcessSid(ptr long int64 ptr ptr) @ stub WinStationGetTermSrvCountersValue @ stub WinStationInstallLicense @ stub WinStationIsHelpAssistantSession

1 0

Conor McCarthy : vkd3d-shader/dxil: Implement DX intrinsic Tertiary.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: vkd3d Branch: master Commit: 3bbe374ca8b9bb046a7e7cc06de257c0660d2605 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/3bbe374ca8b9bb046a7e7cc06de25… Author: Conor McCarthy <cmccarthy(a)codeweavers.com> Date: Tue Jan 23 12:32:39 2024 +1000 vkd3d-shader/dxil: Implement DX intrinsic Tertiary. IBFE and UBFE are not emitted for HLSL sources which perform bitfield extractions, e.g. loading a value from a struct containing bitfields, or the equivalent done with bit shifts. These instructions are probably only emitted by the TPF -> DXIL converter, which makes them hard to test. --- libs/vkd3d-shader/dxil.c | 32 ++++++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+) diff --git a/libs/vkd3d-shader/dxil.c b/libs/vkd3d-shader/dxil.c index ebe600e1..7a4e2ede 100644 --- a/libs/vkd3d-shader/dxil.c +++ b/libs/vkd3d-shader/dxil.c @@ -354,6 +354,8 @@ enum dx_intrinsic_opcode DX_IMIN = 38, DX_UMAX = 39, DX_UMIN = 40, + DX_IBFE = 51, + DX_UBFE = 52, DX_CREATE_HANDLE = 57, DX_CBUFFER_LOAD_LEGACY = 59, DX_BUFFER_LOAD = 68, @@ -3701,6 +3703,34 @@ static void sm6_parser_emit_dx_create_handle(struct sm6_parser *sm6, enum dx_int ins->handler_idx = VKD3DSIH_NOP; } +static enum vkd3d_shader_opcode sm6_dx_map_tertiary_op(enum dx_intrinsic_opcode op) +{ + switch (op) + { + case DX_IBFE: + return VKD3DSIH_IBFE; + case DX_UBFE: + return VKD3DSIH_UBFE; + default: + vkd3d_unreachable(); + } +} + +static void sm6_parser_emit_dx_tertiary(struct sm6_parser *sm6, enum dx_intrinsic_opcode op, + const struct sm6_value **operands, struct function_emission_state *state) +{ + struct vkd3d_shader_instruction *ins = state->ins; + struct vkd3d_shader_src_param *src_params; + unsigned int i; + + vsir_instruction_init(ins, &sm6->p.location, sm6_dx_map_tertiary_op(op)); + src_params = instruction_src_params_alloc(ins, 3, sm6); + for (i = 0; i < 3; ++i) + src_param_init_from_value(&src_params[i], operands[i]); + + instruction_dst_param_init_ssa_scalar(ins, sm6); +} + static void sm6_parser_emit_dx_load_input(struct sm6_parser *sm6, enum dx_intrinsic_opcode op, const struct sm6_value **operands, struct function_emission_state *state) { @@ -3936,6 +3966,7 @@ static const struct sm6_dx_opcode_info sm6_dx_op_table[] = [DX_FMAX ] = {"g", "RR", sm6_parser_emit_dx_binary}, [DX_FMIN ] = {"g", "RR", sm6_parser_emit_dx_binary}, [DX_FRC ] = {"g", "R", sm6_parser_emit_dx_unary}, + [DX_IBFE ] = {"m", "iiR", sm6_parser_emit_dx_tertiary}, [DX_IMAX ] = {"m", "RR", sm6_parser_emit_dx_binary}, [DX_IMIN ] = {"m", "RR", sm6_parser_emit_dx_binary}, [DX_ISFINITE ] = {"1", "g", sm6_parser_emit_dx_unary}, @@ -3955,6 +3986,7 @@ static const struct sm6_dx_opcode_info sm6_dx_op_table[] = [DX_SQRT ] = {"g", "R", sm6_parser_emit_dx_unary}, [DX_STORE_OUTPUT ] = {"v", "ii8o", sm6_parser_emit_dx_store_output}, [DX_TAN ] = {"g", "R", sm6_parser_emit_dx_unary}, + [DX_UBFE ] = {"m", "iiR", sm6_parser_emit_dx_tertiary}, [DX_UMAX ] = {"m", "RR", sm6_parser_emit_dx_binary}, [DX_UMIN ] = {"m", "RR", sm6_parser_emit_dx_binary}, };

1 0

Conor McCarthy : vkd3d-shader/spirv: Support 64-bit source value for bitfield instructions.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: vkd3d Branch: master Commit: 92f0b371331cf1d0fa54ea2f1343f7bca4a00601 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/92f0b371331cf1d0fa54ea2f1343f… Author: Conor McCarthy <cmccarthy(a)codeweavers.com> Date: Tue Jan 23 12:48:22 2024 +1000 vkd3d-shader/spirv: Support 64-bit source value for bitfield instructions. --- libs/vkd3d-shader/spirv.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/libs/vkd3d-shader/spirv.c b/libs/vkd3d-shader/spirv.c index 91411bf6..ef12d32b 100644 --- a/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d-shader/spirv.c @@ -7472,7 +7472,7 @@ static void spirv_compiler_emit_bitfield_instruction(struct spirv_compiler *comp const struct vkd3d_shader_dst_param *dst = instruction->dst; const struct vkd3d_shader_src_param *src = instruction->src; enum vkd3d_shader_component_type component_type; - unsigned int i, j, k, src_count; + unsigned int i, j, k, src_count, size; uint32_t write_mask; SpvOp op; @@ -7481,8 +7481,9 @@ static void spirv_compiler_emit_bitfield_instruction(struct spirv_compiler *comp component_type = vkd3d_component_type_from_data_type(dst->reg.data_type); type_id = vkd3d_spirv_get_type_id(builder, component_type, 1); - mask_id = spirv_compiler_get_constant_uint(compiler, 0x1f); - size_id = spirv_compiler_get_constant_uint(compiler, 0x20); + size = (src[src_count - 1].reg.data_type == VKD3D_DATA_UINT64) ? 0x40 : 0x20; + mask_id = spirv_compiler_get_constant_uint(compiler, size - 1); + size_id = spirv_compiler_get_constant_uint(compiler, size); switch (instruction->handler_idx) {

1 0

Francisco Casas : tests: Rename register-reservations.shader_test to register-reservations-resources.shader_test.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: vkd3d Branch: master Commit: d7d4c9bc909ab0cbe0e3f78357785071a741c147 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/d7d4c9bc909ab0cbe0e3f78357785… Author: Francisco Casas <fcasas(a)codeweavers.com> Date: Fri Nov 10 17:06:16 2023 -0300 tests: Rename register-reservations.shader_test to register-reservations-resources.shader_test. --- Makefile.am | 2 +- ...rvations.shader_test => register-reservations-resources.shader_test} | 0 2 files changed, 1 insertion(+), 1 deletion(-) diff --git a/Makefile.am b/Makefile.am index bfd11fdb..0cb97b66 100644 --- a/Makefile.am +++ b/Makefile.am @@ -152,7 +152,7 @@ vkd3d_shader_tests = \ tests/hlsl/object-references.shader_test \ tests/hlsl/pow.shader_test \ tests/hlsl/reflect.shader_test \ - tests/hlsl/register-reservations.shader_test \ + tests/hlsl/register-reservations-resources.shader_test \ tests/hlsl/return-implicit-conversion.shader_test \ tests/hlsl/return.shader_test \ tests/hlsl/round.shader_test \ diff --git a/tests/hlsl/register-reservations.shader_test b/tests/hlsl/register-reservations-resources.shader_test similarity index 100% rename from tests/hlsl/register-reservations.shader_test rename to tests/hlsl/register-reservations-resources.shader_test

1 0

Francisco Casas : tests: Test register(cX) reservations.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: vkd3d Branch: master Commit: 4791c7c23bda7c98c7e122c8818aeb2000c70d2d URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/4791c7c23bda7c98c7e122c8818ae… Author: Francisco Casas <fcasas(a)codeweavers.com> Date: Wed Nov 8 19:00:27 2023 -0300 tests: Test register(cX) reservations. --- Makefile.am | 1 + .../hlsl/register-reservations-numeric.shader_test | 279 +++++++++++++++++++++ 2 files changed, 280 insertions(+) diff --git a/Makefile.am b/Makefile.am index 0cb97b66..90e7dcfc 100644 --- a/Makefile.am +++ b/Makefile.am @@ -152,6 +152,7 @@ vkd3d_shader_tests = \ tests/hlsl/object-references.shader_test \ tests/hlsl/pow.shader_test \ tests/hlsl/reflect.shader_test \ + tests/hlsl/register-reservations-numeric.shader_test \ tests/hlsl/register-reservations-resources.shader_test \ tests/hlsl/return-implicit-conversion.shader_test \ tests/hlsl/return.shader_test \ diff --git a/tests/hlsl/register-reservations-numeric.shader_test b/tests/hlsl/register-reservations-numeric.shader_test new file mode 100644 index 00000000..f89cb139 --- /dev/null +++ b/tests/hlsl/register-reservations-numeric.shader_test @@ -0,0 +1,279 @@ +[pixel shader fail(sm<6) todo] +// Overlapping register(cX) reservations are not allowed except on SM6, where they are aliased. +// On SM1 this gives hr 0x88760b59. +float a : register(c0); +float b : register(c0); + +float4 main() : sv_target +{ + return a + b; +} + + +[pixel shader] +// It is not required to provide a register(cX) for all elements in the $Globals buffer. +float4 a; // will get register(c1) +float4 b : register(c0); + +float4 main() : sv_target +{ + return float4(a.xw, b.yz); +} + +[test] +uniform 0 float4 0.1 0.2 0.3 0.4 +uniform 4 float4 1.1 1.2 1.3 1.4 +draw quad +todo(sm<6) probe all rgba (1.1, 1.4, 0.2, 0.3) + + +[pixel shader] +float4 a[3]; // will get register(c3) +float4 b[2] : register(c1); + +float4 main() : sv_target +{ + return float4(a[1].xy, b[0].zw); +} + +[test] +uniform 0 float4 0.1 0.2 0.3 0.4 +uniform 4 float4 1.1 1.2 1.3 1.4 +uniform 8 float4 2.1 2.2 2.3 2.4 +uniform 12 float4 3.1 3.2 3.3 3.4 +uniform 16 float4 4.1 4.2 4.3 4.4 +draw quad +todo(sm<6) probe all rgba (4.1, 4.2, 1.3, 1.4) + + +[require] +shader model < 4.0 + +[pixel shader] +float a : register(c2); +float b; // will get register c0 in SM1 + +float4 main() : sv_target +{ + return float4(a, b, 0.0, 0.0); +} + +[test] +uniform 0 float4 0.1 0.2 0.3 0.4 +uniform 4 float4 1.1 1.2 1.3 1.4 +uniform 8 float4 2.1 2.2 2.3 2.4 +uniform 12 float4 3.1 3.2 3.3 3.4 +draw quad +todo probe all rgba (2.1, 0.1, 0.0, 0.0) + + +[require] +shader model >= 4.0 + +[pixel shader] +float a : register(c2); +float b; // will get offset equivalent to c2.y in SM4 and SM6 + +float4 main() : sv_target +{ + return float4(a, b, 0.0, 0.0); +} + +[test] +uniform 0 float4 0.1 0.2 0.3 0.4 +uniform 4 float4 1.1 1.2 1.3 1.4 +uniform 8 float4 2.1 2.2 2.3 2.4 +uniform 12 float4 3.1 3.2 3.3 3.4 +draw quad +todo(sm<6) probe all rgba (2.1, 2.2, 0.0, 0.0) + + +[require] +shader model >= 6.0 + +[pixel shader] +// Variables with overlapping register(cX) reservations are aliased in SM6. +float2 a : register(c2); +float3 b : register(c2); + +float4 main() : sv_target +{ + return float4(a, b.yz); +} + +[test] +uniform 0 float4 0.1 0.2 0.3 0.4 +uniform 4 float4 1.1 1.2 1.3 1.4 +uniform 8 float4 2.1 2.2 2.3 2.4 +draw quad +probe all rgba (2.1, 2.2, 2.2, 2.3) + + +% Results differ between SM1 and SM4 because in the latter variables can share the same register, +% using different writemasks. +[require] +shader model < 4.0 + +[pixel shader] +struct +{ + float2 a; + float b; +} apple : register(c2); + +float4 main() : sv_target +{ + return float4(apple.a, apple.b, 0); +} + +[test] +uniform 0 float4 0.1 0.2 0.3 0.4 +uniform 4 float4 1.1 1.2 1.3 1.4 +uniform 8 float4 2.1 2.2 2.3 2.4 +uniform 12 float4 3.1 3.2 3.3 3.4 +draw quad +todo probe all rgba (2.1, 2.2, 3.1, 0.0) + + +[require] +shader model >= 4.0 + +[pixel shader] +struct +{ + float2 a; + float b; +} apple : register(c2); + +float4 main() : sv_target +{ + return float4(apple.a, apple.b, 0); +} + +[test] +uniform 0 float4 0.1 0.2 0.3 0.4 +uniform 4 float4 1.1 1.2 1.3 1.4 +uniform 8 float4 2.1 2.2 2.3 2.4 +uniform 12 float4 3.1 3.2 3.3 3.4 +draw quad +todo(sm<6) probe all rgba (2.1, 2.2, 2.3, 0.0) + + +[pixel shader] +// On SM4, register(cX) has no effect unless in the $Globals buffer. +cbuffer extra +{ + float a : register(c1); +}; + +float4 main() : sv_target +{ + return a; +} + +[test] +uniform 0 float 100 +uniform 4 float 101 +draw quad +probe all rgba (100, 100, 100, 100) + + +[pixel shader fail(sm>=6)] +// On SM4 register(cX) has no effect unless in the $Globals buffer. +float4 main(uniform float a : register(c1)) : sv_target +{ + return a; +} + +[test] +uniform 0 float 100 +uniform 4 float 101 +draw quad +probe all rgba (100, 100, 100, 100) + +[pixel shader todo] +cbuffer c +{ + float a : packoffset(c1); + float b : packoffset(c2) : register(c1); + // ^ register(c1) ignored for cbuffer that is not $Globals. +} + +float4 main() : sv_target +{ + return float4(a, b, 0, 0); +} + +[test] +uniform 0 float 200 +uniform 4 float 201 +uniform 8 float 202 +todo(sm<6) draw quad +todo(sm<6) probe all rgba (201.0, 202.0, 0.0, 0.0) + + +[pixel shader fail(sm<4)] +int k : register(i0); // register(cX) is also required. + +float4 main() : sv_target +{ + return k; +} + + +[require] +% All shader models. + +% In SM1, most variables are needed in the "c" register group, for float operations. +% If a variable is needed in the "c" register group, register() reservations in other groups can be +% provided only if a register(cX) reservation is also provided. + +[pixel shader fail(sm<4) todo(sm<4)] +int k : register(i0); +// ^^ register(cX) is also required in SM1. + +float4 main() : sv_target +{ + return k; +} + +[pixel shader todo] +int k : register(i0) : register(c1); +// Shader compiles because a "c" register reservation is provided for "k". + +float4 main() : sv_target +{ + return k; +} + + +[require] +shader model >= 3.0 +% model 2.0 doesn't support unrollable loops. + +[pixel shader todo(sm<4)] +int k : register(i0); +// ^^ register(cX) is not required since "k" is just needed in the "i" register group. + +float4 main() : sv_target +{ + float f = 0; + + for (int i = 0; i < k; ++i) + f += i; + return f; +} + + +[pixel shader todo] +int k : register(c0) : register(b0); +// ^^ unlike the "c" register group, a reservation is not required for the "i" group, even though "k" is needed on it. + +float4 main() : sv_target +{ + float f = 0; + + for (int i = 0; i < k; ++i) + f += i; + return f; +}

1 0

Francisco Casas : vkd3d-shader/hlsl: Make register(cX) reservations work for SM1.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: vkd3d Branch: master Commit: c7f76365341b149578392233cf25024f21fcdf14 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/c7f76365341b149578392233cf250… Author: Francisco Casas <fcasas(a)codeweavers.com> Date: Tue Nov 7 21:13:52 2023 -0300 vkd3d-shader/hlsl: Make register(cX) reservations work for SM1. --- libs/vkd3d-shader/hlsl_codegen.c | 41 +++++++++++++++++++--- .../hlsl/register-reservations-numeric.shader_test | 10 +++--- 2 files changed, 42 insertions(+), 9 deletions(-) diff --git a/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d-shader/hlsl_codegen.c index 1fe141a3..2d2e904b 100644 --- a/libs/vkd3d-shader/hlsl_codegen.c +++ b/libs/vkd3d-shader/hlsl_codegen.c @@ -4123,13 +4123,46 @@ static void allocate_const_registers(struct hlsl_ctx *ctx, struct hlsl_ir_functi LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) { - if (var->is_uniform && var->last_read) + unsigned int reg_size = var->data_type->reg_size[HLSL_REGSET_NUMERIC]; + + if (!var->is_uniform || !var->last_read || reg_size == 0) + continue; + + if (var->reg_reservation.reg_type == 'c') { - unsigned int reg_size = var->data_type->reg_size[HLSL_REGSET_NUMERIC]; + unsigned int reg_idx = var->reg_reservation.reg_index; + unsigned int i; - if (reg_size == 0) - continue; + assert(reg_size % 4 == 0); + for (i = 0; i < reg_size / 4; ++i) + { + if (get_available_writemask(&allocator, 1, UINT_MAX, reg_idx + i) != VKD3DSP_WRITEMASK_ALL) + { + hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, + "Overlapping register() reservations on 'c%u'.", reg_idx + i); + } + + record_allocation(ctx, &allocator, reg_idx + i, VKD3DSP_WRITEMASK_ALL, 1, UINT_MAX); + } + var->regs[HLSL_REGSET_NUMERIC].id = reg_idx; + var->regs[HLSL_REGSET_NUMERIC].allocation_size = reg_size / 4; + var->regs[HLSL_REGSET_NUMERIC].writemask = VKD3DSP_WRITEMASK_ALL; + var->regs[HLSL_REGSET_NUMERIC].allocated = true; + TRACE("Allocated reserved %s to %s.\n", var->name, + debug_register('c', var->regs[HLSL_REGSET_NUMERIC], var->data_type)); + } + } + + LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) + { + unsigned int reg_size = var->data_type->reg_size[HLSL_REGSET_NUMERIC]; + + if (!var->is_uniform || !var->last_read || reg_size == 0) + continue; + + if (!var->regs[HLSL_REGSET_NUMERIC].allocated) + { var->regs[HLSL_REGSET_NUMERIC] = allocate_numeric_registers_for_type(ctx, &allocator, 1, UINT_MAX, var->data_type); TRACE("Allocated %s to %s.\n", var->name, diff --git a/tests/hlsl/register-reservations-numeric.shader_test b/tests/hlsl/register-reservations-numeric.shader_test index f89cb139..9b2bae57 100644 --- a/tests/hlsl/register-reservations-numeric.shader_test +++ b/tests/hlsl/register-reservations-numeric.shader_test @@ -1,4 +1,4 @@ -[pixel shader fail(sm<6) todo] +[pixel shader fail(sm<6) todo(sm>=4)] // Overlapping register(cX) reservations are not allowed except on SM6, where they are aliased. // On SM1 this gives hr 0x88760b59. float a : register(c0); @@ -24,7 +24,7 @@ float4 main() : sv_target uniform 0 float4 0.1 0.2 0.3 0.4 uniform 4 float4 1.1 1.2 1.3 1.4 draw quad -todo(sm<6) probe all rgba (1.1, 1.4, 0.2, 0.3) +todo(sm>=4 & sm<6) probe all rgba (1.1, 1.4, 0.2, 0.3) [pixel shader] @@ -43,7 +43,7 @@ uniform 8 float4 2.1 2.2 2.3 2.4 uniform 12 float4 3.1 3.2 3.3 3.4 uniform 16 float4 4.1 4.2 4.3 4.4 draw quad -todo(sm<6) probe all rgba (4.1, 4.2, 1.3, 1.4) +todo(sm>=4 & sm<6) probe all rgba (4.1, 4.2, 1.3, 1.4) [require] @@ -64,7 +64,7 @@ uniform 4 float4 1.1 1.2 1.3 1.4 uniform 8 float4 2.1 2.2 2.3 2.4 uniform 12 float4 3.1 3.2 3.3 3.4 draw quad -todo probe all rgba (2.1, 0.1, 0.0, 0.0) +probe all rgba (2.1, 0.1, 0.0, 0.0) [require] @@ -132,7 +132,7 @@ uniform 4 float4 1.1 1.2 1.3 1.4 uniform 8 float4 2.1 2.2 2.3 2.4 uniform 12 float4 3.1 3.2 3.3 3.4 draw quad -todo probe all rgba (2.1, 2.2, 3.1, 0.0) +probe all rgba (2.1, 2.2, 3.1, 0.0) [require]

1 0

Francisco Casas : vkd3d-shader/hlsl: Turn register(cX) reservations into buffer offset for SM4.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: vkd3d Branch: master Commit: 13f713f74bdd3ab3650cbe72a7db2e872ab41cdf URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/13f713f74bdd3ab3650cbe72a7db2… Author: Francisco Casas <fcasas(a)codeweavers.com> Date: Wed Nov 8 18:45:26 2023 -0300 vkd3d-shader/hlsl: Turn register(cX) reservations into buffer offset for SM4. --- libs/vkd3d-shader/hlsl_codegen.c | 96 ++++++++++++++-------- .../hlsl/register-reservations-numeric.shader_test | 10 +-- 2 files changed, 67 insertions(+), 39 deletions(-) diff --git a/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d-shader/hlsl_codegen.c index 2d2e904b..6ad60e4c 100644 --- a/libs/vkd3d-shader/hlsl_codegen.c +++ b/libs/vkd3d-shader/hlsl_codegen.c @@ -4302,45 +4302,52 @@ static const struct hlsl_buffer *get_reserved_buffer(struct hlsl_ctx *ctx, uint3 return NULL; } -static void calculate_buffer_offset(struct hlsl_ctx *ctx, struct hlsl_ir_var *var) +static void calculate_buffer_offset(struct hlsl_ctx *ctx, struct hlsl_ir_var *var, bool register_reservation) { unsigned int var_reg_size = var->data_type->reg_size[HLSL_REGSET_NUMERIC]; enum hlsl_type_class var_class = var->data_type->class; struct hlsl_buffer *buffer = var->buffer; - if (var->reg_reservation.offset_type == 'c') + if (register_reservation) { - if (var->reg_reservation.offset_index % 4) + var->buffer_offset = 4 * var->reg_reservation.reg_index; + } + else + { + if (var->reg_reservation.offset_type == 'c') { - if (var_class == HLSL_CLASS_MATRIX) - { - hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, - "packoffset() reservations with matrix types must be aligned with the beginning of a register."); - } - else if (var_class == HLSL_CLASS_ARRAY) - { - hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, - "packoffset() reservations with array types must be aligned with the beginning of a register."); - } - else if (var_class == HLSL_CLASS_STRUCT) - { - hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, - "packoffset() reservations with struct types must be aligned with the beginning of a register."); - } - else if (var_class == HLSL_CLASS_VECTOR) + if (var->reg_reservation.offset_index % 4) { - unsigned int aligned_offset = hlsl_type_get_sm4_offset(var->data_type, var->reg_reservation.offset_index); - - if (var->reg_reservation.offset_index != aligned_offset) + if (var_class == HLSL_CLASS_MATRIX) + { hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, - "packoffset() reservations with vector types cannot span multiple registers."); + "packoffset() reservations with matrix types must be aligned with the beginning of a register."); + } + else if (var_class == HLSL_CLASS_ARRAY) + { + hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, + "packoffset() reservations with array types must be aligned with the beginning of a register."); + } + else if (var_class == HLSL_CLASS_STRUCT) + { + hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, + "packoffset() reservations with struct types must be aligned with the beginning of a register."); + } + else if (var_class == HLSL_CLASS_VECTOR) + { + unsigned int aligned_offset = hlsl_type_get_sm4_offset(var->data_type, var->reg_reservation.offset_index); + + if (var->reg_reservation.offset_index != aligned_offset) + hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, + "packoffset() reservations with vector types cannot span multiple registers."); + } } + var->buffer_offset = var->reg_reservation.offset_index; + } + else + { + var->buffer_offset = hlsl_type_get_sm4_offset(var->data_type, buffer->size); } - var->buffer_offset = var->reg_reservation.offset_index; - } - else - { - var->buffer_offset = hlsl_type_get_sm4_offset(var->data_type, buffer->size); } TRACE("Allocated buffer offset %u to %s.\n", var->buffer_offset, var->name); @@ -4409,6 +4416,11 @@ static void validate_buffer_offsets(struct hlsl_ctx *ctx) } } +static bool var_has_buffer_offset_register_reservation(struct hlsl_ctx *ctx, const struct hlsl_ir_var *var) +{ + return var->reg_reservation.reg_type == 'c' && var->buffer == ctx->globals_buffer; +} + static void allocate_buffers(struct hlsl_ctx *ctx) { struct hlsl_buffer *buffer; @@ -4417,13 +4429,29 @@ static void allocate_buffers(struct hlsl_ctx *ctx) LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) { - if (var->is_uniform && !hlsl_type_is_resource(var->data_type)) - { - if (var->is_param) - var->buffer = ctx->params_buffer; + if (!var->is_uniform || hlsl_type_is_resource(var->data_type)) + continue; - calculate_buffer_offset(ctx, var); - } + if (var->is_param) + var->buffer = ctx->params_buffer; + } + + LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) + { + if (!var->is_uniform || hlsl_type_is_resource(var->data_type)) + continue; + + if (var_has_buffer_offset_register_reservation(ctx, var)) + calculate_buffer_offset(ctx, var, true); + } + + LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) + { + if (!var->is_uniform || hlsl_type_is_resource(var->data_type)) + continue; + + if (!var_has_buffer_offset_register_reservation(ctx, var)) + calculate_buffer_offset(ctx, var, false); } validate_buffer_offsets(ctx); diff --git a/tests/hlsl/register-reservations-numeric.shader_test b/tests/hlsl/register-reservations-numeric.shader_test index 9b2bae57..5067d54f 100644 --- a/tests/hlsl/register-reservations-numeric.shader_test +++ b/tests/hlsl/register-reservations-numeric.shader_test @@ -1,4 +1,4 @@ -[pixel shader fail(sm<6) todo(sm>=4)] +[pixel shader fail(sm<6)] // Overlapping register(cX) reservations are not allowed except on SM6, where they are aliased. // On SM1 this gives hr 0x88760b59. float a : register(c0); @@ -24,7 +24,7 @@ float4 main() : sv_target uniform 0 float4 0.1 0.2 0.3 0.4 uniform 4 float4 1.1 1.2 1.3 1.4 draw quad -todo(sm>=4 & sm<6) probe all rgba (1.1, 1.4, 0.2, 0.3) +probe all rgba (1.1, 1.4, 0.2, 0.3) [pixel shader] @@ -43,7 +43,7 @@ uniform 8 float4 2.1 2.2 2.3 2.4 uniform 12 float4 3.1 3.2 3.3 3.4 uniform 16 float4 4.1 4.2 4.3 4.4 draw quad -todo(sm>=4 & sm<6) probe all rgba (4.1, 4.2, 1.3, 1.4) +probe all rgba (4.1, 4.2, 1.3, 1.4) [require] @@ -85,7 +85,7 @@ uniform 4 float4 1.1 1.2 1.3 1.4 uniform 8 float4 2.1 2.2 2.3 2.4 uniform 12 float4 3.1 3.2 3.3 3.4 draw quad -todo(sm<6) probe all rgba (2.1, 2.2, 0.0, 0.0) +probe all rgba (2.1, 2.2, 0.0, 0.0) [require] @@ -156,7 +156,7 @@ uniform 4 float4 1.1 1.2 1.3 1.4 uniform 8 float4 2.1 2.2 2.3 2.4 uniform 12 float4 3.1 3.2 3.3 3.4 draw quad -todo(sm<6) probe all rgba (2.1, 2.2, 2.3, 0.0) +probe all rgba (2.1, 2.2, 2.3, 0.0) [pixel shader]

1 0

Giovanni Mascellani : vkd3d-shader: Convert write masks between 32 and 64 bit by cases.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: vkd3d Branch: master Commit: 2b93aae069c9b6920db08d3fe4a7f68f18671946 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/2b93aae069c9b6920db08d3fe4a7f… Author: Giovanni Mascellani <gmascellani(a)codeweavers.com> Date: Sat Jan 20 23:30:23 2024 +0100 vkd3d-shader: Convert write masks between 32 and 64 bit by cases. There are only three cases, and while the code is longer it is also hopefully easier to read. Moreover, an error message is casted if we're doing something unexpected. --- libs/vkd3d-shader/vkd3d_shader_private.h | 35 +++++++++++++++++++++++++++----- 1 file changed, 30 insertions(+), 5 deletions(-) diff --git a/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d-shader/vkd3d_shader_private.h index f8c4d61e..cbac998a 100644 --- a/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d-shader/vkd3d_shader_private.h @@ -1605,15 +1605,40 @@ static inline unsigned int vkd3d_write_mask_from_component_count(unsigned int co static inline uint32_t vsir_write_mask_64_from_32(uint32_t write_mask32) { - uint32_t write_mask64 = write_mask32 | (write_mask32 >> 1); - return (write_mask64 & VKD3DSP_WRITEMASK_0) | ((write_mask64 & VKD3DSP_WRITEMASK_2) >> 1); + switch (write_mask32) + { + case VKD3DSP_WRITEMASK_0 | VKD3DSP_WRITEMASK_1: + return VKD3DSP_WRITEMASK_0; + + case VKD3DSP_WRITEMASK_2 | VKD3DSP_WRITEMASK_3: + return VKD3DSP_WRITEMASK_1; + + case VKD3DSP_WRITEMASK_0 | VKD3DSP_WRITEMASK_1 | VKD3DSP_WRITEMASK_2 | VKD3DSP_WRITEMASK_3: + return VKD3DSP_WRITEMASK_0 | VKD3DSP_WRITEMASK_1; + + default: + ERR("Invalid 32 bit writemask when converting to 64 bit: %#x.\n", write_mask32); + return VKD3DSP_WRITEMASK_0; + } } static inline uint32_t vsir_write_mask_32_from_64(uint32_t write_mask64) { - uint32_t write_mask32 = (write_mask64 | (write_mask64 << 1)) - & (VKD3DSP_WRITEMASK_0 | VKD3DSP_WRITEMASK_2); - return write_mask32 | (write_mask32 << 1); + switch (write_mask64) + { + case VKD3DSP_WRITEMASK_0: + return VKD3DSP_WRITEMASK_0 | VKD3DSP_WRITEMASK_1; + + case VKD3DSP_WRITEMASK_1: + return VKD3DSP_WRITEMASK_2 | VKD3DSP_WRITEMASK_3; + + case VKD3DSP_WRITEMASK_0 | VKD3DSP_WRITEMASK_1: + return VKD3DSP_WRITEMASK_0 | VKD3DSP_WRITEMASK_1 | VKD3DSP_WRITEMASK_2 | VKD3DSP_WRITEMASK_3; + + default: + ERR("Invalid 64 bit writemask: %#x.\n", write_mask64); + return VKD3DSP_WRITEMASK_0; + } } static inline unsigned int vsir_swizzle_get_component(uint32_t swizzle, unsigned int idx)

1 0

Giovanni Mascellani : include/vkd3d-shader: Generate swizzles without a function call.
by Alexandre Julliard 29 Jan '24

29 Jan '24

Module: vkd3d Branch: master Commit: 5ec1825eb31e68838f5763f17b920e7cd85003cd URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/5ec1825eb31e68838f5763f17b920… Author: Giovanni Mascellani <gmascellani(a)codeweavers.com> Date: Sat Jan 20 23:28:14 2024 +0100 include/vkd3d-shader: Generate swizzles without a function call. So they can be used when a constant expression is expected, for instance on case labels. --- include/vkd3d_shader.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/include/vkd3d_shader.h b/include/vkd3d_shader.h index a8cc3a33..2f4478a7 100644 --- a/include/vkd3d_shader.h +++ b/include/vkd3d_shader.h @@ -1777,10 +1777,10 @@ struct vkd3d_shader_dxbc_desc * \endcode */ #define VKD3D_SHADER_SWIZZLE(x, y, z, w) \ - vkd3d_shader_create_swizzle(VKD3D_SHADER_SWIZZLE_ ## x, \ - VKD3D_SHADER_SWIZZLE_ ## y, \ - VKD3D_SHADER_SWIZZLE_ ## z, \ - VKD3D_SHADER_SWIZZLE_ ## w) + (VKD3D_SHADER_SWIZZLE_ ## x << VKD3D_SHADER_SWIZZLE_SHIFT(0) \ + | VKD3D_SHADER_SWIZZLE_ ## y << VKD3D_SHADER_SWIZZLE_SHIFT(1) \ + | VKD3D_SHADER_SWIZZLE_ ## z << VKD3D_SHADER_SWIZZLE_SHIFT(2) \ + | VKD3D_SHADER_SWIZZLE_ ## w << VKD3D_SHADER_SWIZZLE_SHIFT(3)) /** The identity swizzle ".xyzw". */ #define VKD3D_SHADER_NO_SWIZZLE VKD3D_SHADER_SWIZZLE(X, Y, Z, W)

1 0