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

Alexandre Julliard : rpcrt4: Generate the parameter extension data for typelibs on ARM platforms.
by Alexandre Julliard 24 Apr '24

24 Apr '24

Module: wine Branch: master Commit: 00756efcfbf07c80373b7a82fdcc6253e48e555e URL: https://gitlab.winehq.org/wine/wine/-/commit/00756efcfbf07c80373b7a82fdcc62… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Tue Apr 23 18:01:42 2024 +0200 rpcrt4: Generate the parameter extension data for typelibs on ARM platforms. --- dlls/rpcrt4/ndr_typelib.c | 167 ++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 153 insertions(+), 14 deletions(-) diff --git a/dlls/rpcrt4/ndr_typelib.c b/dlls/rpcrt4/ndr_typelib.c index 094de78e36d..1006bf1d3a0 100644 --- a/dlls/rpcrt4/ndr_typelib.c +++ b/dlls/rpcrt4/ndr_typelib.c @@ -1115,14 +1115,128 @@ static HRESULT write_param_fs(ITypeInfo *typeinfo, unsigned char *type, return hr; } +#if defined __arm__ || defined __aarch64__ + +/* replace consecutive params code by a repeat sequence: 0x9d code<1> repeat_count<2> */ +static unsigned int compress_params_array( unsigned char *params, unsigned int count ) +{ + unsigned int i, j; + + for (i = 0; i + 4 <= count; i++) + { + for (j = 1; i + j < count; j++) if (params[i + j] != params[i]) break; + if (j < 4) continue; + params[i] = 0x9d; + params[i + 2] = j & 0xff; + params[i + 3] = j >> 8; + memmove( params + i + 4, params + i + j, count - (i + j) ); + count -= j - 4; + i += 3; + } + return count; +} + +/* fill the parameters array for the procedure extra data on ARM platforms */ +static unsigned int fill_params_array( ITypeInfo *typeinfo, FUNCDESC *desc, + unsigned char *params, unsigned int count ) +{ + static const unsigned int pointer_size = sizeof(void *); + unsigned int reg_count = 0, float_count = 0, double_count = 0, stack_pos = 0, offset = 0; + unsigned int i, size, pos, align; + + memset( params, 0x9f /* padding */, count ); + + /* This pointer */ + params[0] = 0x80 + reg_count++; + offset += pointer_size; + + for (i = 0; i < desc->cParams; i++) + { + unsigned char basetype = get_basetype( typeinfo, &desc->lprgelemdescParam[i].tdesc ); + + size = get_stack_size( typeinfo, &desc->lprgelemdescParam[i].tdesc, &align, NULL ); + offset = ROUND_SIZE( offset, align ); + pos = offset / pointer_size; + +#ifdef __aarch64__ + switch (basetype) + { + case FC_FLOAT: + case FC_DOUBLE: + if (double_count >= 8) break; + params[pos] = 0x88 + double_count++; + offset += size; + continue; + + default: + reg_count = ROUND_SIZE( reg_count, align / pointer_size ); + if (reg_count > 8 - size / pointer_size) break; + while (size) + { + params[pos++] = 0x80 + reg_count++; + offset += pointer_size; + size -= pointer_size; + } + continue; + } + (void)float_count; /* unused on arm64 */ +#else + switch (basetype) + { + case FC_FLOAT: + if (!(float_count % 2)) float_count = max( float_count, double_count * 2 ); + if (float_count >= 16) + { + stack_pos = ROUND_SIZE( stack_pos, align ); + params[pos] = 0x100 - (offset - stack_pos) / pointer_size; + stack_pos += size; + } + else + { + params[pos] = 0x84 + float_count++; + } + offset += size; + continue; + + case FC_DOUBLE: + double_count = max( double_count, (float_count + 1) / 2 ); + if (double_count >= 8) break; + params[pos] = 0x84 + 2 * double_count; + params[pos + 1] = 0x84 + 2 * double_count + 1; + double_count++; + offset += size; + continue; + + default: + reg_count = ROUND_SIZE( reg_count, align / pointer_size ); + if (reg_count <= 4 - size / pointer_size || !stack_pos) + { + while (size && reg_count < 4) + { + params[pos++] = 0x80 + reg_count++; + offset += pointer_size; + size -= pointer_size; + } + } + break; + } +#endif + stack_pos = ROUND_SIZE( stack_pos, align ); + memset( params + pos, 0x100 - (offset - stack_pos) / pointer_size, size / pointer_size ); + stack_pos += size; + offset += size; + } + + while (count && params[count - 1] == 0x9f) count--; + return count; +} + +#endif /* __arm__ || __aarch64__ */ + static void write_proc_func_header(ITypeInfo *typeinfo, FUNCDESC *desc, WORD proc_idx, unsigned char *proc, size_t *proclen) { unsigned int i, align, size, stack_size = sizeof(void *); /* This */ -#ifdef __x86_64__ - unsigned short float_mask = 0; - unsigned char basetype; -#endif for (i = 0; i < desc->cParams; i++) { @@ -1141,23 +1255,48 @@ static void write_proc_func_header(ITypeInfo *typeinfo, FUNCDESC *desc, WRITE_SHORT(proc, *proclen, 0); /* constant_server_buffer_size */ WRITE_CHAR (proc, *proclen, 0x47); /* HasExtensions | HasReturn | ClientMustSize | ServerMustSize */ WRITE_CHAR (proc, *proclen, desc->cParams + 1); /* incl. return value */ -#ifdef __x86_64__ - WRITE_CHAR (proc, *proclen, 10); /* extension size */ -#else + +#ifdef __i386__ WRITE_CHAR (proc, *proclen, 8); /* extension size */ -#endif WRITE_CHAR (proc, *proclen, 1); /* HasNewCorrDesc */ WRITE_SHORT(proc, *proclen, 0); /* ClientCorrHint */ WRITE_SHORT(proc, *proclen, 0); /* ServerCorrHint */ WRITE_SHORT(proc, *proclen, 0); /* NotifyIndex */ -#ifdef __x86_64__ - for (i = 0; i < desc->cParams && i < 3; i++) +#elif defined __x86_64__ + { + unsigned short float_mask = 0; + + for (i = 0; i < desc->cParams && i < 3; i++) + { + unsigned char basetype = get_basetype(typeinfo, &desc->lprgelemdescParam[i].tdesc); + if (basetype == FC_FLOAT) float_mask |= (1 << ((i + 1) * 2)); + else if (basetype == FC_DOUBLE) float_mask |= (2 << ((i + 1) * 2)); + } + WRITE_CHAR (proc, *proclen, 10); /* extension size */ + WRITE_CHAR (proc, *proclen, 1); /* HasNewCorrDesc */ + WRITE_SHORT(proc, *proclen, 0); /* ClientCorrHint */ + WRITE_SHORT(proc, *proclen, 0); /* ServerCorrHint */ + WRITE_SHORT(proc, *proclen, 0); /* NotifyIndex */ + WRITE_SHORT(proc, *proclen, float_mask); + } +#else { - basetype = get_basetype(typeinfo, &desc->lprgelemdescParam[i].tdesc); - if (basetype == FC_FLOAT) float_mask |= (1 << ((i + 1) * 2)); - else if (basetype == FC_DOUBLE) float_mask |= (2 << ((i + 1) * 2)); + unsigned int len, count = stack_size / sizeof(void *); + unsigned char *params = malloc( count ); + + count = fill_params_array( typeinfo, desc, params, count ); + len = compress_params_array( params, count ); + WRITE_CHAR (proc, *proclen, 8 + 3 + len + !(len % 2) ); /* extension size */ + WRITE_CHAR (proc, *proclen, 1); /* HasNewCorrDesc */ + WRITE_SHORT(proc, *proclen, 0); /* ClientCorrHint */ + WRITE_SHORT(proc, *proclen, 0); /* ServerCorrHint */ + WRITE_SHORT(proc, *proclen, 0); /* NotifyIndex */ + WRITE_SHORT(proc, *proclen, count); + WRITE_CHAR (proc, *proclen, len); + for (i = 0; i < len; i++) WRITE_CHAR (proc, *proclen, params[i]); + if (!(len % 2)) WRITE_CHAR (proc, *proclen, 0); + free( params ); } - WRITE_SHORT(proc, *proclen, float_mask); #endif }

1 0

Alexandre Julliard : rpcrt4: Extend 8- and 16-bit parameters on ARM.
by Alexandre Julliard 24 Apr '24

24 Apr '24

Module: wine Branch: master Commit: 28cb539ef9b2ea4a262d3994a0a077c4b4a26296 URL: https://gitlab.winehq.org/wine/wine/-/commit/28cb539ef9b2ea4a262d3994a0a077… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Wed Apr 24 16:27:56 2024 +0200 rpcrt4: Extend 8- and 16-bit parameters on ARM. --- dlls/rpcrt4/ndr_stubless.c | 77 ++++++++++++++++++++++++++++++---------------- 1 file changed, 51 insertions(+), 26 deletions(-) diff --git a/dlls/rpcrt4/ndr_stubless.c b/dlls/rpcrt4/ndr_stubless.c index f4b253b7351..5c71ad3c359 100644 --- a/dlls/rpcrt4/ndr_stubless.c +++ b/dlls/rpcrt4/ndr_stubless.c @@ -1062,12 +1062,41 @@ __ASM_GLOBAL_FUNC( NdrClientCall2, #if defined(__aarch64__) || defined(__arm__) static void __attribute__((used)) args_stack_to_regs( void **args, void **regs, void **stack, - const NDR_PROC_HEADER_EXTS *ext ) + const NDR_PROC_PARTIAL_OIF_HEADER *header ) { + const NDR_PROC_HEADER_EXTS *ext = (const NDR_PROC_HEADER_EXTS *)(header + 1); unsigned int i, size, count, pos; unsigned char *data; - if (!ext) return; +#ifdef __arm__ + const NDR_PARAM_OIF *params = (const NDR_PARAM_OIF *)((const char *)ext + ext->Size); + + for (i = 0; i < header->number_of_params; i++) + if (params[i].attr.IsIn && params[i].attr.IsBasetype) + { + int *arg = (int *)((char *)args + params[i].stack_offset); + + switch (params[i].u.type_format_char) + { + case FC_BYTE: + case FC_USMALL: + *arg = (unsigned char)*arg; + break; + case FC_CHAR: + case FC_SMALL: + *arg = (signed char)*arg; + break; + case FC_WCHAR: + case FC_USHORT: + *arg = (unsigned short)*arg; + break; + case FC_SHORT: + *arg = (short)*arg; + break; + } + } +#endif + if (ext->Size < sizeof(*ext) + 3) return; data = (unsigned char *)(ext + 1); size = min( ext->Size - sizeof(*ext) - 3, data[2] ); @@ -1089,9 +1118,10 @@ static void __attribute__((used)) args_stack_to_regs( void **args, void **regs, } } -static void **args_regs_to_stack( void **regs, void **fpu_regs, const NDR_PROC_HEADER_EXTS *ext ) +static void **args_regs_to_stack( void **regs, void **fpu_regs, const NDR_PROC_PARTIAL_OIF_HEADER *header ) { static const unsigned int nb_gpregs = sizeof(void *); /* 4 gpregs on arm32, 8 on arm64 */ + const NDR_PROC_HEADER_EXTS *ext = (const NDR_PROC_HEADER_EXTS *)(header + 1); unsigned int i, size, count, pos, params; unsigned char *data; void **stack, **args = regs + nb_gpregs; @@ -1127,7 +1157,7 @@ static void **args_regs_to_stack( void **regs, void **fpu_regs, const NDR_PROC_H * function */ #if defined __i386__ && defined _MSC_VER __declspec(naked) LONG_PTR __cdecl call_server_func(SERVER_ROUTINE func, unsigned char * args, unsigned int stack_size, - const NDR_PROC_HEADER_EXTS *ext) + const NDR_PROC_PARTIAL_OIF_HEADER *header ) { __asm { @@ -1154,7 +1184,7 @@ __declspec(naked) LONG_PTR __cdecl call_server_func(SERVER_ROUTINE func, unsigne } #elif defined __i386__ && defined __GNUC__ LONG_PTR __cdecl call_server_func(SERVER_ROUTINE func, unsigned char * args, unsigned int stack_size, - const NDR_PROC_HEADER_EXTS *ext); + const NDR_PROC_PARTIAL_OIF_HEADER *header); __ASM_GLOBAL_FUNC(call_server_func, "pushl %ebp\n\t" __ASM_CFI(".cfi_adjust_cfa_offset 4\n\t") @@ -1186,7 +1216,7 @@ __ASM_GLOBAL_FUNC(call_server_func, "ret" ) #elif defined __x86_64__ LONG_PTR __cdecl call_server_func(SERVER_ROUTINE func, unsigned char * args, unsigned int stack_size, - const NDR_PROC_HEADER_EXTS *ext); + const NDR_PROC_PARTIAL_OIF_HEADER *header); __ASM_GLOBAL_FUNC( call_server_func, "pushq %rbp\n\t" __ASM_SEH(".seh_pushreg %rbp\n\t") @@ -1234,7 +1264,7 @@ __ASM_GLOBAL_FUNC( call_server_func, "ret") #elif defined __arm__ LONG_PTR __cdecl call_server_func(SERVER_ROUTINE func, unsigned char *args, unsigned int stack_size, - const NDR_PROC_HEADER_EXTS *ext); + const NDR_PROC_PARTIAL_OIF_HEADER *header); __ASM_GLOBAL_FUNC( call_server_func, "push {r4,r5,fp,lr}\n\t" ".seh_save_regs_w {r4,r5,fp,lr}\n\t" @@ -1257,7 +1287,7 @@ __ASM_GLOBAL_FUNC( call_server_func, "pop {r4,r5,fp,pc}" ) #elif defined __aarch64__ LONG_PTR __cdecl call_server_func(SERVER_ROUTINE func, unsigned char *args, unsigned int stack_size, - const NDR_PROC_HEADER_EXTS *ext); + const NDR_PROC_PARTIAL_OIF_HEADER *header); __ASM_GLOBAL_FUNC( call_server_func, "stp x29, x30, [sp, #-0x20]!\n\t" ".seh_save_fplr_x 0x20\n\t" @@ -1290,7 +1320,7 @@ __ASM_GLOBAL_FUNC( call_server_func, #else #warning call_server_func not implemented for your architecture LONG_PTR __cdecl call_server_func(SERVER_ROUTINE func, unsigned char * args, unsigned short stack_size, - const NDR_PROC_HEADER_EXTS *ext) + const NDR_PROC_PARTIAL_OIF_HEADER *header) { FIXME("Not implemented for your architecture\n"); return 0; @@ -1322,7 +1352,7 @@ LONG_PTR WINAPI ndr_stubless_client_call( unsigned int index, void **args, void unsigned short fpu_mask = *(unsigned short *)(ext + 1); for (int i = 0; i < 4; i++, fpu_mask >>= 2) if (fpu_mask & 3) args[i] = fpu_regs[i]; #else - stack_top = args_regs_to_stack( args, fpu_regs, ext ); + stack_top = args_regs_to_stack( args, fpu_regs, hdr ); #endif } } @@ -1448,7 +1478,7 @@ LONG WINAPI NdrStubCall2( enum stubless_phase phase; /* header for procedure string */ const NDR_PROC_HEADER *pProcHeader; - const NDR_PROC_HEADER_EXTS *extensions = NULL; + const NDR_PROC_PARTIAL_OIF_HEADER *pOIFHeader = NULL; /* location to put retval into */ LONG_PTR *retval_ptr = NULL; /* correlation cache */ @@ -1546,8 +1576,7 @@ LONG WINAPI NdrStubCall2( if (is_oicf_stubdesc(pStubDesc)) { - const NDR_PROC_PARTIAL_OIF_HEADER *pOIFHeader = (const NDR_PROC_PARTIAL_OIF_HEADER *)pFormat; - + pOIFHeader = (const NDR_PROC_PARTIAL_OIF_HEADER *)pFormat; Oif_flags = pOIFHeader->Oi2Flags; number_of_params = pOIFHeader->number_of_params; @@ -1557,7 +1586,7 @@ LONG WINAPI NdrStubCall2( if (Oif_flags.HasExtensions) { - extensions = (const NDR_PROC_HEADER_EXTS *)pFormat; + const NDR_PROC_HEADER_EXTS *extensions = (const NDR_PROC_HEADER_EXTS *)pFormat; ext_flags = extensions->Flags2; pFormat += extensions->Size; } @@ -1612,7 +1641,7 @@ LONG WINAPI NdrStubCall2( else func = pServerInfo->DispatchTable[pRpcMsg->ProcNum]; - retval = call_server_func(func, args, stack_size, extensions); + retval = call_server_func(func, args, stack_size, pOIFHeader); if (retval_ptr) { @@ -2111,7 +2140,7 @@ void RPC_ENTRY NdrAsyncServerCall(PRPC_MESSAGE pRpcMsg) unsigned char *args; /* header for procedure string */ const NDR_PROC_HEADER *pProcHeader; - const NDR_PROC_HEADER_EXTS *extensions = NULL; + const NDR_PROC_PARTIAL_OIF_HEADER *pOIFHeader = NULL; struct async_call_data *async_call_data; PRPC_ASYNC_STATE pAsync; RPC_STATUS status; @@ -2219,27 +2248,23 @@ void RPC_ENTRY NdrAsyncServerCall(PRPC_MESSAGE pRpcMsg) if (is_oicf_stubdesc(pStubDesc)) { - const NDR_PROC_PARTIAL_OIF_HEADER *pOIFHeader = (const NDR_PROC_PARTIAL_OIF_HEADER *)pFormat; - /* cache of Oif_flags from v2 procedure header */ - INTERPRETER_OPT_FLAGS Oif_flags; - /* cache of extension flags from NDR_PROC_HEADER_EXTS */ INTERPRETER_OPT_FLAGS2 ext_flags = { 0 }; - Oif_flags = pOIFHeader->Oi2Flags; + pOIFHeader = (const NDR_PROC_PARTIAL_OIF_HEADER *)pFormat; async_call_data->number_of_params = pOIFHeader->number_of_params; pFormat += sizeof(NDR_PROC_PARTIAL_OIF_HEADER); - TRACE("Oif_flags = %s\n", debugstr_INTERPRETER_OPT_FLAGS(Oif_flags) ); + TRACE("Oif_flags = %s\n", debugstr_INTERPRETER_OPT_FLAGS(pOIFHeader->Oi2Flags) ); - if (Oif_flags.HasExtensions) + if (pOIFHeader->Oi2Flags.HasExtensions) { - extensions = (const NDR_PROC_HEADER_EXTS *)pFormat; + const NDR_PROC_HEADER_EXTS *extensions = (const NDR_PROC_HEADER_EXTS *)pFormat; ext_flags = extensions->Flags2; pFormat += extensions->Size; } - if (Oif_flags.HasPipes) + if (pOIFHeader->Oi2Flags.HasPipes) { FIXME("pipes not supported yet\n"); RpcRaiseException(RPC_X_WRONG_STUB_VERSION); /* FIXME: remove when implemented */ @@ -2283,7 +2308,7 @@ void RPC_ENTRY NdrAsyncServerCall(PRPC_MESSAGE pRpcMsg) pServerInfo->ThunkTable[pRpcMsg->ProcNum](async_call_data->pStubMsg); else call_server_func(pServerInfo->DispatchTable[pRpcMsg->ProcNum], args, async_call_data->stack_size, - extensions); + pOIFHeader); } RPC_STATUS NdrpCompleteAsyncServerCall(RPC_ASYNC_STATE *pAsync, void *Reply)

1 0

Alexandre Julliard : oleaut32: Extend 8- and 16-bit parameters on ARM.
by Alexandre Julliard 24 Apr '24

24 Apr '24

Module: wine Branch: master Commit: fffed9b3b198b9f26157e903291af0ee6de33867 URL: https://gitlab.winehq.org/wine/wine/-/commit/fffed9b3b198b9f26157e903291af0… Author: Alexandre Julliard <julliard(a)winehq.org> Date: Wed Apr 24 16:54:48 2024 +0200 oleaut32: Extend 8- and 16-bit parameters on ARM. And update tests to show that Windows does the same. --- dlls/oleaut32/tests/typelib.c | 31 +++++++++++++++-------------- dlls/oleaut32/typelib.c | 46 +++++++++++++++++++++---------------------- 2 files changed, 39 insertions(+), 38 deletions(-) diff --git a/dlls/oleaut32/tests/typelib.c b/dlls/oleaut32/tests/typelib.c index d755729d9e9..f8606bcfe68 100644 --- a/dlls/oleaut32/tests/typelib.c +++ b/dlls/oleaut32/tests/typelib.c @@ -1188,7 +1188,11 @@ static int WINAPI int_func( int a0, int a1, int a2, int a3, int a4 ) { ok( a0 == 1, "wrong arg0 %x\n", a0 ); ok( a1 == -1, "wrong arg1 %x\n", a1 ); +#ifdef __arm__ /* parameters are extended on arm */ + ok( a2 == 1234, "wrong arg2 %x\n", a2 ); +#else ok( a2 == (0x55550000 | 1234), "wrong arg2 %x\n", a2 ); +#endif ok( a3 == 0xdeadbeef, "wrong arg3 %x\n", a3 ); ok( a4 == 0x555555fd, "wrong arg4 %x\n", a4 ); return 4321; @@ -1226,16 +1230,13 @@ static VARIANT WINAPI variant_func( int a0, BOOL a1, DECIMAL a2, VARIANT a3 ) static int CDECL void_func( int a0, int a1 ) { - if (is_win64) /* VT_EMPTY is passed as real arg on win64 */ - { - ok( a0 == 0x55555555, "wrong arg0 %x\n", a0 ); - ok( a1 == 1111, "wrong arg1 %x\n", a1 ); - } - else - { - ok( a0 == 1111, "wrong arg0 %x\n", a0 ); - ok( a1 == 0, "wrong arg1 %x\n", a1 ); - } +#ifdef __i386__ + ok( a0 == 1111, "wrong arg0 %x\n", a0 ); + ok( a1 == 0, "wrong arg1 %x\n", a1 ); +#else /* VT_EMPTY is passed as real arg on other platforms */ + ok( a0 == 0x55555555, "wrong arg0 %x\n", a0 ); + ok( a1 == 1111, "wrong arg1 %x\n", a1 ); +#endif return 12; } @@ -1433,11 +1434,11 @@ static void test_DispCallFunc(void) res = DispCallFunc( NULL, (ULONG_PTR)void_func, CC_CDECL, VT_EMPTY, 5, types, pargs, &result ); ok( res == S_OK, "DispCallFunc failed %lx\n", res ); ok( V_VT(&result) == VT_EMPTY, "wrong result type %d\n", V_VT(&result) ); - if (is_win64) - ok( V_UI4(&result) == 12, "wrong result %08lx\n", V_UI4(&result) ); - else - ok( V_UI4(&result) == 0xcccccccc, "wrong result %08lx\n", V_UI4(&result) ); - +#ifdef __i386__ + ok( V_UI4(&result) == 0xcccccccc, "wrong result %08lx\n", V_UI4(&result) ); +#else + ok( V_UI4(&result) == 12, "wrong result %08lx\n", V_UI4(&result) ); +#endif memset( args, 0x55, sizeof(args) ); types[0] = VT_I4; V_I4(&args[0]) = 3; diff --git a/dlls/oleaut32/typelib.c b/dlls/oleaut32/typelib.c index d920ee7c0cc..670938fb81c 100644 --- a/dlls/oleaut32/typelib.c +++ b/dlls/oleaut32/typelib.c @@ -6618,9 +6618,7 @@ __ASM_GLOBAL_FUNC( call_method, "bgt 2b\n\t" /* Loop till done */ "1:\n\t" -#ifndef __SOFTFP__ "vldm r3!, {s0-s15}\n\t" /* Load the s0-s15/d0-d7 arguments */ -#endif "mov ip, r0\n\t" /* Save the function call address to ip before we nuke r0 with arguments to pass */ "ldm r3, {r0-r3}\n\t" /* Load the r0-r3 arguments */ @@ -6642,19 +6640,15 @@ HRESULT WINAPI DispCallFunc( void* pvInstance, ULONG_PTR oVft, CALLCONV cc, VART UINT i; DWORD *args; struct { -#ifndef __SOFTFP__ union { float s[16]; double d[8]; } sd; -#endif DWORD r[4]; } regs; int rcount; /* 32-bit register index count */ -#ifndef __SOFTFP__ int scount = 0; /* single-precision float register index count */ int dcount = 0; /* double-precision float register index count */ -#endif TRACE("(%p, %Id, %d, %d, %d, %p, %p, %p (vt=%d))\n", pvInstance, oVft, cc, vtReturn, cActuals, prgvt, prgpvarg, pvargResult, V_VT(pvargResult)); @@ -6702,11 +6696,8 @@ HRESULT WINAPI DispCallFunc( void* pvInstance, ULONG_PTR oVft, CALLCONV cc, VART switch (prgvt[i]) { - case VT_EMPTY: - break; case VT_R8: /* these must be 8-byte aligned, and put in 'd' regs or stack, as they are double-floats */ case VT_DATE: -#ifndef __SOFTFP__ dcount = max( (scount + 1) / 2, dcount ); if (dcount < 8) { @@ -6719,7 +6710,6 @@ HRESULT WINAPI DispCallFunc( void* pvInstance, ULONG_PTR oVft, CALLCONV cc, VART argspos += sizeof(V_R8(arg)) / sizeof(DWORD); } break; -#endif case VT_I8: /* these must be 8-byte aligned, and put in 'r' regs or stack, as they are long-longs */ case VT_UI8: case VT_CY: @@ -6757,26 +6747,37 @@ HRESULT WINAPI DispCallFunc( void* pvInstance, ULONG_PTR oVft, CALLCONV cc, VART --ntemp; } break; - case VT_BOOL: /* VT_BOOL is 16-bit but BOOL is 32-bit, needs to be extended */ - if (rcount < 4) - regs.r[rcount++] = V_BOOL(arg); - else - args[argspos++] = V_BOOL(arg); - break; case VT_R4: /* these must be 4-byte aligned, and put in 's' regs or stack, as they are single-floats */ -#ifndef __SOFTFP__ if (!(scount % 2)) scount = max( scount, dcount * 2 ); if (scount < 16) regs.sd.s[scount++] = V_R4(arg); else args[argspos++] = V_UI4(arg); break; -#endif + /* extend parameters to 32 bits */ + case VT_I1: + if (rcount < 4) regs.r[rcount++] = V_I1(arg); + else args[argspos++] = V_I1(arg); + break; + case VT_UI1: + if (rcount < 4) regs.r[rcount++] = V_UI1(arg); + else args[argspos++] = V_UI1(arg); + break; + case VT_I2: + if (rcount < 4) regs.r[rcount++] = V_I2(arg); + else args[argspos++] = V_I2(arg); + break; + case VT_UI2: + if (rcount < 4) regs.r[rcount++] = V_UI2(arg); + else args[argspos++] = V_UI2(arg); + break; + case VT_BOOL: + if (rcount < 4) regs.r[rcount++] = V_BOOL(arg); + else args[argspos++] = V_BOOL(arg); + break; default: - if (rcount < 4) - regs.r[rcount++] = V_UI4(arg); - else - args[argspos++] = V_UI4(arg); + if (rcount < 4) regs.r[rcount++] = V_UI4(arg); + else args[argspos++] = V_UI4(arg); break; } TRACE("arg %u: type %s %s\n", i, debugstr_vt(prgvt[i]), debugstr_variant(arg)); @@ -6786,7 +6787,6 @@ HRESULT WINAPI DispCallFunc( void* pvInstance, ULONG_PTR oVft, CALLCONV cc, VART switch (vtReturn) { - case VT_EMPTY: /* EMPTY = no return value */ case VT_DECIMAL: /* DECIMAL and VARIANT already have a pointer argument passed (see above) */ case VT_VARIANT: call_method( func, argspos, args, (DWORD*)&regs );

1 0

Zebediah Figura : vkd3d-shader/hlsl: Make HLSL_TYPE_DEPTHSTENCILVIEW into a separate class.
by Alexandre Julliard 23 Apr '24

23 Apr '24

Module: vkd3d Branch: master Commit: ee2bde3aba8d85e7a15560d3f17355f11f2a8baa URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/ee2bde3aba8d85e7a15560d3f1735… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Tue Feb 27 13:31:20 2024 -0600 vkd3d-shader/hlsl: Make HLSL_TYPE_DEPTHSTENCILVIEW into a separate class. --- libs/vkd3d-shader/d3dbc.c | 2 ++ libs/vkd3d-shader/fx.c | 40 ++++++++++++++++++++++++++-------------- 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, 37 insertions(+), 16 deletions(-) diff --git a/libs/vkd3d-shader/d3dbc.c b/libs/vkd3d-shader/d3dbc.c index 6299239b..88f5630f 100644 --- a/libs/vkd3d-shader/d3dbc.c +++ b/libs/vkd3d-shader/d3dbc.c @@ -1517,6 +1517,7 @@ D3DXPARAMETER_CLASS hlsl_sm1_class(const struct hlsl_type *type) case HLSL_CLASS_STRING: case HLSL_CLASS_TEXTURE: return D3DXPC_OBJECT; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_PASS: case HLSL_CLASS_TECHNIQUE: @@ -1617,6 +1618,7 @@ D3DXPARAMETER_TYPE hlsl_sm1_base_type(const struct hlsl_type *type) case HLSL_CLASS_STRING: return D3DXPT_STRING; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_PASS: case HLSL_CLASS_TECHNIQUE: diff --git a/libs/vkd3d-shader/fx.c b/libs/vkd3d-shader/fx.c index 73160b63..1ed2c4b7 100644 --- a/libs/vkd3d-shader/fx.c +++ b/libs/vkd3d-shader/fx.c @@ -354,7 +354,6 @@ static const char * get_fx_4_type_name(const struct hlsl_type *type) [HLSL_TYPE_PIXELSHADER] = "PixelShader", [HLSL_TYPE_VERTEXSHADER] = "VertexShader", [HLSL_TYPE_RENDERTARGETVIEW] = "RenderTargetView", - [HLSL_TYPE_DEPTHSTENCILVIEW] = "DepthStencilView", }; static const char * const texture_type_names[] = { @@ -380,19 +379,28 @@ static const char * get_fx_4_type_name(const struct hlsl_type *type) [HLSL_SAMPLER_DIM_STRUCTURED_BUFFER] = "RWStructuredBuffer", }; - if (type->class == HLSL_CLASS_TEXTURE) - return texture_type_names[type->sampler_dim]; + switch (type->class) + { + case HLSL_CLASS_TEXTURE: + return texture_type_names[type->sampler_dim]; - if (type->class == HLSL_CLASS_UAV) - return uav_type_names[type->sampler_dim]; + case HLSL_CLASS_UAV: + return uav_type_names[type->sampler_dim]; + + case HLSL_CLASS_DEPTH_STENCIL_VIEW: + return "DepthStencilView"; + + case HLSL_CLASS_OBJECT: + switch (type->base_type) + { + case HLSL_TYPE_PIXELSHADER: + case HLSL_TYPE_VERTEXSHADER: + case HLSL_TYPE_RENDERTARGETVIEW: + return object_type_names[type->base_type]; + default: + return type->name; + } - switch (type->base_type) - { - case HLSL_TYPE_PIXELSHADER: - case HLSL_TYPE_VERTEXSHADER: - case HLSL_TYPE_RENDERTARGETVIEW: - case HLSL_TYPE_DEPTHSTENCILVIEW: - return object_type_names[type->base_type]; default: return type->name; } @@ -426,6 +434,7 @@ static uint32_t write_fx_4_type(const struct hlsl_type *type, struct fx_write_co put_u32_unaligned(buffer, 1); break; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_OBJECT: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: @@ -513,6 +522,10 @@ static uint32_t write_fx_4_type(const struct hlsl_type *type, struct fx_write_co put_u32_unaligned(buffer, uav_type[type->sampler_dim]); } + else if (type->class == HLSL_CLASS_DEPTH_STENCIL_VIEW) + { + put_u32_unaligned(buffer, 20); + } else if (type->class == HLSL_CLASS_OBJECT) { static const uint32_t object_type[] = @@ -520,12 +533,10 @@ static uint32_t write_fx_4_type(const struct hlsl_type *type, struct fx_write_co [HLSL_TYPE_PIXELSHADER] = 5, [HLSL_TYPE_VERTEXSHADER] = 6, [HLSL_TYPE_RENDERTARGETVIEW] = 19, - [HLSL_TYPE_DEPTHSTENCILVIEW] = 20, }; switch (type->base_type) { - case HLSL_TYPE_DEPTHSTENCILVIEW: case HLSL_TYPE_PIXELSHADER: case HLSL_TYPE_RENDERTARGETVIEW: case HLSL_TYPE_VERTEXSHADER: @@ -834,6 +845,7 @@ static bool is_type_supported_fx_2(struct hlsl_ctx *ctx, const struct hlsl_type hlsl_fixme(ctx, loc, "Write fx 2.0 parameter class %#x.", type->class); return false; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_UAV: case HLSL_CLASS_VOID: return false; diff --git a/libs/vkd3d-shader/hlsl.c b/libs/vkd3d-shader/hlsl.c index b5bb72a9..ba297153 100644 --- a/libs/vkd3d-shader/hlsl.c +++ b/libs/vkd3d-shader/hlsl.c @@ -363,6 +363,7 @@ static void hlsl_type_calculate_reg_size(struct hlsl_ctx *ctx, struct hlsl_type type->reg_size[HLSL_REGSET_UAVS] = 1; break; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_OBJECT: case HLSL_CLASS_PASS: @@ -428,6 +429,7 @@ static bool type_is_single_component(const struct hlsl_type *type) { switch (type->class) { + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_SCALAR: case HLSL_CLASS_OBJECT: case HLSL_CLASS_SAMPLER: @@ -567,6 +569,7 @@ unsigned int hlsl_type_get_component_offset(struct hlsl_ctx *ctx, struct hlsl_ty } break; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_OBJECT: case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: @@ -943,6 +946,7 @@ unsigned int hlsl_type_component_count(const struct hlsl_type *type) case HLSL_CLASS_ARRAY: return hlsl_type_component_count(type->e.array.type) * type->e.array.elements_count; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_OBJECT: case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: @@ -2366,6 +2370,7 @@ struct vkd3d_string_buffer *hlsl_type_to_string(struct hlsl_ctx *ctx, const stru } return string; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_OBJECT: case HLSL_CLASS_PASS: @@ -3550,7 +3555,6 @@ static void declare_predefined_types(struct hlsl_ctx *ctx) {"pixelshader", HLSL_CLASS_OBJECT, HLSL_TYPE_PIXELSHADER, 1, 1}, {"vertexshader", HLSL_CLASS_OBJECT, HLSL_TYPE_VERTEXSHADER, 1, 1}, {"RenderTargetView",HLSL_CLASS_OBJECT, HLSL_TYPE_RENDERTARGETVIEW, 1, 1}, - {"DepthStencilView",HLSL_CLASS_OBJECT, HLSL_TYPE_DEPTHSTENCILVIEW, 1, 1}, }; static const struct @@ -3662,6 +3666,7 @@ static void declare_predefined_types(struct hlsl_ctx *ctx) } ctx->builtin_types.Void = hlsl_new_simple_type(ctx, "void", HLSL_CLASS_VOID); + hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "DepthStencilView", HLSL_CLASS_DEPTH_STENCIL_VIEW)); hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "fxgroup", HLSL_CLASS_EFFECT_GROUP)); hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "pass", HLSL_CLASS_PASS)); hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "STRING", HLSL_CLASS_STRING)); diff --git a/libs/vkd3d-shader/hlsl.h b/libs/vkd3d-shader/hlsl.h index a9e47c6f..452aec16 100644 --- a/libs/vkd3d-shader/hlsl.h +++ b/libs/vkd3d-shader/hlsl.h @@ -79,6 +79,7 @@ enum hlsl_type_class HLSL_CLASS_STRUCT, HLSL_CLASS_ARRAY, HLSL_CLASS_OBJECT, + HLSL_CLASS_DEPTH_STENCIL_VIEW, HLSL_CLASS_EFFECT_GROUP, HLSL_CLASS_PASS, HLSL_CLASS_SAMPLER, @@ -101,7 +102,6 @@ enum hlsl_base_type HLSL_TYPE_PIXELSHADER, HLSL_TYPE_VERTEXSHADER, HLSL_TYPE_RENDERTARGETVIEW, - HLSL_TYPE_DEPTHSTENCILVIEW, }; enum hlsl_sampler_dim diff --git a/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d-shader/hlsl_codegen.c index 5681761c..0669ff40 100644 --- a/libs/vkd3d-shader/hlsl_codegen.c +++ b/libs/vkd3d-shader/hlsl_codegen.c @@ -1638,6 +1638,7 @@ static bool copy_propagation_transform_load(struct hlsl_ctx *ctx, * matrices yet. */ return false; + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_PASS: case HLSL_CLASS_STRING: diff --git a/libs/vkd3d-shader/tpf.c b/libs/vkd3d-shader/tpf.c index 52e6eac8..e01d8d91 100644 --- a/libs/vkd3d-shader/tpf.c +++ b/libs/vkd3d-shader/tpf.c @@ -3006,6 +3006,7 @@ static D3D_SHADER_VARIABLE_CLASS sm4_class(const struct hlsl_type *type) return D3D_SVC_VECTOR; case HLSL_CLASS_ARRAY: + case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_STRUCT: case HLSL_CLASS_OBJECT:

1 0

Zebediah Figura : vkd3d-shader/hlsl: Make HLSL_TYPE_RENDERTARGETVIEW into a separate class.
by Alexandre Julliard 23 Apr '24

23 Apr '24

Module: vkd3d Branch: master Commit: d7da2786938cd986a7cd4f7378c51f7c9c1846fb URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/d7da2786938cd986a7cd4f7378c51… Author: Zebediah Figura <zfigura(a)codeweavers.com> Date: Tue Feb 27 13:34:52 2024 -0600 vkd3d-shader/hlsl: Make HLSL_TYPE_RENDERTARGETVIEW into a separate class. --- libs/vkd3d-shader/d3dbc.c | 2 ++ libs/vkd3d-shader/fx.c | 26 ++++++++++++++++++-------- 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, 29 insertions(+), 10 deletions(-) diff --git a/libs/vkd3d-shader/d3dbc.c b/libs/vkd3d-shader/d3dbc.c index 88f5630f..f7160f12 100644 --- a/libs/vkd3d-shader/d3dbc.c +++ b/libs/vkd3d-shader/d3dbc.c @@ -1520,6 +1520,7 @@ D3DXPARAMETER_CLASS hlsl_sm1_class(const struct hlsl_type *type) case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_PASS: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_TECHNIQUE: case HLSL_CLASS_UAV: case HLSL_CLASS_VOID: @@ -1621,6 +1622,7 @@ D3DXPARAMETER_TYPE hlsl_sm1_base_type(const struct hlsl_type *type) case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_PASS: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_TECHNIQUE: case HLSL_CLASS_UAV: case HLSL_CLASS_VOID: diff --git a/libs/vkd3d-shader/fx.c b/libs/vkd3d-shader/fx.c index 1ed2c4b7..168378e6 100644 --- a/libs/vkd3d-shader/fx.c +++ b/libs/vkd3d-shader/fx.c @@ -353,7 +353,6 @@ static const char * get_fx_4_type_name(const struct hlsl_type *type) { [HLSL_TYPE_PIXELSHADER] = "PixelShader", [HLSL_TYPE_VERTEXSHADER] = "VertexShader", - [HLSL_TYPE_RENDERTARGETVIEW] = "RenderTargetView", }; static const char * const texture_type_names[] = { @@ -390,12 +389,14 @@ static const char * get_fx_4_type_name(const struct hlsl_type *type) case HLSL_CLASS_DEPTH_STENCIL_VIEW: return "DepthStencilView"; + case HLSL_CLASS_RENDER_TARGET_VIEW: + return "RenderTargetView"; + case HLSL_CLASS_OBJECT: switch (type->base_type) { case HLSL_TYPE_PIXELSHADER: case HLSL_TYPE_VERTEXSHADER: - case HLSL_TYPE_RENDERTARGETVIEW: return object_type_names[type->base_type]; default: return type->name; @@ -436,6 +437,7 @@ static uint32_t write_fx_4_type(const struct hlsl_type *type, struct fx_write_co case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_OBJECT: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: put_u32_unaligned(buffer, 2); @@ -526,19 +528,21 @@ static uint32_t write_fx_4_type(const struct hlsl_type *type, struct fx_write_co { put_u32_unaligned(buffer, 20); } + else if (type->class == HLSL_CLASS_RENDER_TARGET_VIEW) + { + put_u32_unaligned(buffer, 19); + } else if (type->class == HLSL_CLASS_OBJECT) { static const uint32_t object_type[] = { [HLSL_TYPE_PIXELSHADER] = 5, [HLSL_TYPE_VERTEXSHADER] = 6, - [HLSL_TYPE_RENDERTARGETVIEW] = 19, }; switch (type->base_type) { case HLSL_TYPE_PIXELSHADER: - case HLSL_TYPE_RENDERTARGETVIEW: case HLSL_TYPE_VERTEXSHADER: put_u32_unaligned(buffer, object_type[type->base_type]); break; @@ -547,11 +551,17 @@ static uint32_t write_fx_4_type(const struct hlsl_type *type, struct fx_write_co return 0; } } - else /* Numeric type */ + else if (hlsl_is_numeric_type(type)) { numeric_desc = get_fx_4_numeric_type_description(type, fx); put_u32_unaligned(buffer, numeric_desc); } + else + { + FIXME("Type %u is not supported.\n", type->class); + set_status(fx, VKD3D_ERROR_NOT_IMPLEMENTED); + return 0; + } return offset; } @@ -847,6 +857,7 @@ static bool is_type_supported_fx_2(struct hlsl_ctx *ctx, const struct hlsl_type case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_UAV: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_VOID: return false; @@ -1032,6 +1043,7 @@ static void write_fx_4_object_variable(struct hlsl_ir_var *var, struct fx_write_ /* Initializer */ switch (type->class) { + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: break; @@ -1039,8 +1051,6 @@ static void write_fx_4_object_variable(struct hlsl_ir_var *var, struct fx_write_ case HLSL_CLASS_OBJECT: switch (type->base_type) { - case HLSL_TYPE_RENDERTARGETVIEW: - break; case HLSL_TYPE_PIXELSHADER: case HLSL_TYPE_VERTEXSHADER: /* FIXME: write shader blobs, once parser support works. */ @@ -1139,6 +1149,7 @@ static bool is_object_variable(const struct hlsl_ir_var *var) switch (type->class) { + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_SAMPLER: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: @@ -1149,7 +1160,6 @@ static bool is_object_variable(const struct hlsl_ir_var *var) { case HLSL_TYPE_PIXELSHADER: case HLSL_TYPE_VERTEXSHADER: - case HLSL_TYPE_RENDERTARGETVIEW: return true; default: return false; diff --git a/libs/vkd3d-shader/hlsl.c b/libs/vkd3d-shader/hlsl.c index ba297153..4fc1493b 100644 --- a/libs/vkd3d-shader/hlsl.c +++ b/libs/vkd3d-shader/hlsl.c @@ -367,6 +367,7 @@ static void hlsl_type_calculate_reg_size(struct hlsl_ctx *ctx, struct hlsl_type case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_OBJECT: case HLSL_CLASS_PASS: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_STRING: case HLSL_CLASS_TECHNIQUE: case HLSL_CLASS_VOID: @@ -434,6 +435,7 @@ static bool type_is_single_component(const struct hlsl_type *type) case HLSL_CLASS_OBJECT: case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_TEXTURE: case HLSL_CLASS_UAV: return true; @@ -571,6 +573,7 @@ unsigned int hlsl_type_get_component_offset(struct hlsl_ctx *ctx, struct hlsl_ty case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_OBJECT: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: case HLSL_CLASS_TEXTURE: @@ -948,6 +951,7 @@ unsigned int hlsl_type_component_count(const struct hlsl_type *type) case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_OBJECT: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: case HLSL_CLASS_TEXTURE: @@ -2374,6 +2378,7 @@ struct vkd3d_string_buffer *hlsl_type_to_string(struct hlsl_ctx *ctx, const stru case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_OBJECT: case HLSL_CLASS_PASS: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: case HLSL_CLASS_TECHNIQUE: @@ -3554,7 +3559,6 @@ static void declare_predefined_types(struct hlsl_ctx *ctx) {"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}, - {"RenderTargetView",HLSL_CLASS_OBJECT, HLSL_TYPE_RENDERTARGETVIEW, 1, 1}, }; static const struct @@ -3669,6 +3673,7 @@ static void declare_predefined_types(struct hlsl_ctx *ctx) hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "DepthStencilView", HLSL_CLASS_DEPTH_STENCIL_VIEW)); hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "fxgroup", HLSL_CLASS_EFFECT_GROUP)); hlsl_scope_add_type(ctx->globals, hlsl_new_simple_type(ctx, "pass", HLSL_CLASS_PASS)); + 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)); diff --git a/libs/vkd3d-shader/hlsl.h b/libs/vkd3d-shader/hlsl.h index 452aec16..a1e3aacd 100644 --- a/libs/vkd3d-shader/hlsl.h +++ b/libs/vkd3d-shader/hlsl.h @@ -82,6 +82,7 @@ enum hlsl_type_class HLSL_CLASS_DEPTH_STENCIL_VIEW, HLSL_CLASS_EFFECT_GROUP, HLSL_CLASS_PASS, + HLSL_CLASS_RENDER_TARGET_VIEW, HLSL_CLASS_SAMPLER, HLSL_CLASS_STRING, HLSL_CLASS_TECHNIQUE, @@ -101,7 +102,6 @@ enum hlsl_base_type HLSL_TYPE_LAST_SCALAR = HLSL_TYPE_BOOL, HLSL_TYPE_PIXELSHADER, HLSL_TYPE_VERTEXSHADER, - HLSL_TYPE_RENDERTARGETVIEW, }; enum hlsl_sampler_dim diff --git a/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d-shader/hlsl_codegen.c index 0669ff40..a6d6b336 100644 --- a/libs/vkd3d-shader/hlsl_codegen.c +++ b/libs/vkd3d-shader/hlsl_codegen.c @@ -1641,6 +1641,7 @@ static bool copy_propagation_transform_load(struct hlsl_ctx *ctx, case HLSL_CLASS_DEPTH_STENCIL_VIEW: case HLSL_CLASS_EFFECT_GROUP: case HLSL_CLASS_PASS: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_STRING: case HLSL_CLASS_TECHNIQUE: case HLSL_CLASS_VOID: diff --git a/libs/vkd3d-shader/tpf.c b/libs/vkd3d-shader/tpf.c index e01d8d91..708ab626 100644 --- a/libs/vkd3d-shader/tpf.c +++ b/libs/vkd3d-shader/tpf.c @@ -3011,6 +3011,7 @@ static D3D_SHADER_VARIABLE_CLASS sm4_class(const struct hlsl_type *type) case HLSL_CLASS_STRUCT: case HLSL_CLASS_OBJECT: case HLSL_CLASS_PASS: + case HLSL_CLASS_RENDER_TARGET_VIEW: case HLSL_CLASS_SAMPLER: case HLSL_CLASS_STRING: case HLSL_CLASS_TECHNIQUE:

1 0

Nikolay Sivov : tests: Add some tests for string variables.
by Alexandre Julliard 23 Apr '24

23 Apr '24

Module: vkd3d Branch: master Commit: 836fe0591602a729556da5ecf83e26cdbe139dd5 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/836fe0591602a729556da5ecf83e2… Author: Nikolay Sivov <nsivov(a)codeweavers.com> Date: Mon Apr 22 14:18:18 2024 +0200 tests: Add some tests for string variables. Signed-off-by: Nikolay Sivov <nsivov(a)codeweavers.com> --- Makefile.am | 1 + tests/hlsl/strings.shader_test | 51 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 52 insertions(+) diff --git a/Makefile.am b/Makefile.am index 432f4f27..50a5e8b3 100644 --- a/Makefile.am +++ b/Makefile.am @@ -199,6 +199,7 @@ vkd3d_shader_tests = \ tests/hlsl/static-initializer.shader_test \ tests/hlsl/step.shader_test \ tests/hlsl/storage-qualifiers.shader_test \ + tests/hlsl/strings.shader_test \ tests/hlsl/struct-array.shader_test \ tests/hlsl/struct-assignment.shader_test \ tests/hlsl/struct-semantics.shader_test \ diff --git a/tests/hlsl/strings.shader_test b/tests/hlsl/strings.shader_test new file mode 100644 index 00000000..a0cdcd48 --- /dev/null +++ b/tests/hlsl/strings.shader_test @@ -0,0 +1,51 @@ +[pixel shader todo fail(sm>=6)] +String s = "text"; +string s2 = "text2"; +sTring s3 = "text3"; + +float4 main() : sv_target +{ + return 0; +} + +[pixel shader todo] +typedef float4 sTring; +sTring var = float4(1, 2, 3, 4); + +float4 main() : sv_target +{ + return 0; +} + +[pixel shader todo fail(sm>=6)] +// Uninitialized strings are allowed for non-fx profiles +String s; + +float4 main() : sv_target +{ + return 0; +} + +[pixel shader fail] +typedef float4 string; + +float4 main() : sv_target +{ + return 0; +} + +[pixel shader fail(sm<6) todo] +typedef float4 String; + +float4 main() : sv_target +{ + return 0; +} + +[pixel shader fail(sm<6) todo] +float4 String; + +float4 main() : sv_target +{ + return 0; +}

1 0

Conor McCarthy : vkd3d-shader/ir: Accept undefined PHI incomings as valid.
by Alexandre Julliard 23 Apr '24

23 Apr '24

Module: vkd3d Branch: master Commit: e3dde09bb99e3173c447d509734469a2b702ce6a URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/e3dde09bb99e3173c447d50973446… Author: Conor McCarthy <cmccarthy(a)codeweavers.com> Date: Wed Apr 24 00:23:12 2024 +1000 vkd3d-shader/ir: Accept undefined PHI incomings as valid. An incoming can be undefined if the PHI result is not used if the incoming's path was taken, or if the undefined value is consumed by an instruction (dxc will compile such code without errors). --- libs/vkd3d-shader/ir.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/libs/vkd3d-shader/ir.c b/libs/vkd3d-shader/ir.c index 62292f16..552de311 100644 --- a/libs/vkd3d-shader/ir.c +++ b/libs/vkd3d-shader/ir.c @@ -5971,7 +5971,8 @@ static void vsir_validate_instruction(struct validation_context *ctx) unsigned int value_idx = 2 * i; unsigned int label_idx = 2 * i + 1; - if (!register_is_constant(&instruction->src[value_idx].reg) && !register_is_ssa(&instruction->src[value_idx].reg)) + if (!register_is_constant_or_undef(&instruction->src[value_idx].reg) + && !register_is_ssa(&instruction->src[value_idx].reg)) validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_REGISTER_TYPE, "Invalid value register for incoming %zu of type %#x in PHI instruction, " "expected SSA, IMMCONST or IMMCONST64.", i, instruction->src[value_idx].reg.type);

1 0

Francisco Casas : vkd3d-shader/hlsl: Cleanup parse_attribute_list structs (valgrind).
by Alexandre Julliard 23 Apr '24

23 Apr '24

Module: vkd3d Branch: master Commit: aa35b2cecb63085581dbc06f9b074153cb99b209 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/aa35b2cecb63085581dbc06f9b074… Author: Francisco Casas <fcasas(a)codeweavers.com> Date: Tue Apr 23 01:19:54 2024 -0400 vkd3d-shader/hlsl: Cleanup parse_attribute_list structs (valgrind). --- libs/vkd3d-shader/hlsl.y | 32 ++++++++++++++++++++++---------- 1 file changed, 22 insertions(+), 10 deletions(-) diff --git a/libs/vkd3d-shader/hlsl.y b/libs/vkd3d-shader/hlsl.y index f99f322d..0eed15c5 100644 --- a/libs/vkd3d-shader/hlsl.y +++ b/libs/vkd3d-shader/hlsl.y @@ -656,6 +656,16 @@ static unsigned int initializer_size(const struct parse_initializer *initializer return count; } +static void cleanup_parse_attribute_list(struct parse_attribute_list *attr_list) +{ + unsigned int i = 0; + + assert(attr_list); + for (i = 0; i < attr_list->count; ++i) + hlsl_free_attribute((struct hlsl_attribute *) attr_list->attrs[i]); + vkd3d_free(attr_list->attrs); +} + static void free_parse_initializer(struct parse_initializer *initializer) { destroy_block(initializer->instrs); @@ -6033,11 +6043,7 @@ attribute_list: $$ = $1; if (!(new_array = vkd3d_realloc($$.attrs, ($$.count + 1) * sizeof(*$$.attrs)))) { - unsigned int i; - - for (i = 0; i < $$.count; ++i) - hlsl_free_attribute((void *)$$.attrs[i]); - vkd3d_free($$.attrs); + cleanup_parse_attribute_list(&$$); YYABORT; } $$.attrs = new_array; @@ -6243,11 +6249,7 @@ func_prototype: } else { - unsigned int i; - - for (i = 0; i < $1.count; ++i) - hlsl_free_attribute((void *)$1.attrs[i]); - vkd3d_free($1.attrs); + cleanup_parse_attribute_list(&$1); } $$ = $2; } @@ -7358,6 +7360,7 @@ selection_statement: { destroy_block($6.then_block); destroy_block($6.else_block); + cleanup_parse_attribute_list(&$1); YYABORT; } @@ -7365,10 +7368,12 @@ selection_statement: { destroy_block($6.then_block); destroy_block($6.else_block); + cleanup_parse_attribute_list(&$1); YYABORT; } destroy_block($6.then_block); destroy_block($6.else_block); + cleanup_parse_attribute_list(&$1); $$ = $4; hlsl_block_add_instr($$, instr); @@ -7391,21 +7396,25 @@ loop_statement: { $$ = create_loop(ctx, LOOP_WHILE, &$1, NULL, $5, NULL, $7, &@3); hlsl_pop_scope(ctx); + cleanup_parse_attribute_list(&$1); } | attribute_list_optional loop_scope_start KW_DO statement KW_WHILE '(' expr ')' ';' { $$ = create_loop(ctx, LOOP_DO_WHILE, &$1, NULL, $7, NULL, $4, &@3); hlsl_pop_scope(ctx); + cleanup_parse_attribute_list(&$1); } | attribute_list_optional loop_scope_start KW_FOR '(' expr_statement expr_statement expr_optional ')' statement { $$ = create_loop(ctx, LOOP_FOR, &$1, $5, $6, $7, $9, &@3); hlsl_pop_scope(ctx); + cleanup_parse_attribute_list(&$1); } | attribute_list_optional loop_scope_start KW_FOR '(' declaration expr_statement expr_optional ')' statement { $$ = create_loop(ctx, LOOP_FOR, &$1, $5, $6, $7, $9, &@3); hlsl_pop_scope(ctx); + cleanup_parse_attribute_list(&$1); } switch_statement: @@ -7418,6 +7427,7 @@ switch_statement: { destroy_switch_cases($8); destroy_block($5); + cleanup_parse_attribute_list(&$1); YYABORT; } @@ -7428,6 +7438,7 @@ switch_statement: if (!s) { destroy_block($5); + cleanup_parse_attribute_list(&$1); YYABORT; } @@ -7435,6 +7446,7 @@ switch_statement: hlsl_block_add_instr($$, s); hlsl_pop_scope(ctx); + cleanup_parse_attribute_list(&$1); } switch_case:

1 0

Henri Verbeet : vkd3d-common: Introduce VKD3D_BITMAP_SIZE.
by Alexandre Julliard 23 Apr '24

23 Apr '24

Module: vkd3d Branch: master Commit: 46fca3f9f4a9b47b32e9dfbacda0f3d19536c02c URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/46fca3f9f4a9b47b32e9dfbacda0f… Author: Henri Verbeet <hverbeet(a)codeweavers.com> Date: Wed Apr 17 13:12:13 2024 +0200 vkd3d-common: Introduce VKD3D_BITMAP_SIZE. --- include/private/vkd3d_common.h | 2 ++ libs/vkd3d-shader/d3dbc.c | 2 +- libs/vkd3d-shader/dxil.c | 2 +- libs/vkd3d-shader/ir.c | 3 +-- tests/shader_runner.h | 2 +- 5 files changed, 6 insertions(+), 5 deletions(-) diff --git a/include/private/vkd3d_common.h b/include/private/vkd3d_common.h index c03ac72e..de6878a4 100644 --- a/include/private/vkd3d_common.h +++ b/include/private/vkd3d_common.h @@ -340,6 +340,8 @@ static inline int vkd3d_u32_compare(uint32_t x, uint32_t y) return (x > y) - (x < y); } +#define VKD3D_BITMAP_SIZE(x) (((x) + 0x1f) >> 5) + static inline bool bitmap_clear(uint32_t *map, unsigned int idx) { return map[idx >> 5] &= ~(1u << (idx & 0x1f)); diff --git a/libs/vkd3d-shader/d3dbc.c b/libs/vkd3d-shader/d3dbc.c index f7160f12..55d9ecf7 100644 --- a/libs/vkd3d-shader/d3dbc.c +++ b/libs/vkd3d-shader/d3dbc.c @@ -216,7 +216,7 @@ struct vkd3d_shader_sm1_parser struct vkd3d_shader_parser p; #define MAX_CONSTANT_COUNT 8192 - uint32_t constant_def_mask[3][MAX_CONSTANT_COUNT / 32]; + uint32_t constant_def_mask[3][VKD3D_BITMAP_SIZE(MAX_CONSTANT_COUNT)]; }; /* This table is not order or position dependent. */ diff --git a/libs/vkd3d-shader/dxil.c b/libs/vkd3d-shader/dxil.c index d07d5ade..6a1fb6bd 100644 --- a/libs/vkd3d-shader/dxil.c +++ b/libs/vkd3d-shader/dxil.c @@ -805,7 +805,7 @@ struct sm6_parser struct vkd3d_shader_dst_param *output_params; struct vkd3d_shader_dst_param *input_params; struct vkd3d_shader_dst_param *patch_constant_params; - uint32_t io_regs_declared[(VKD3DSPR_COUNT + 0x1f) / 0x20]; + uint32_t io_regs_declared[VKD3D_BITMAP_SIZE(VKD3DSPR_COUNT)]; struct sm6_function *functions; size_t function_count; diff --git a/libs/vkd3d-shader/ir.c b/libs/vkd3d-shader/ir.c index 552de311..59b74c06 100644 --- a/libs/vkd3d-shader/ir.c +++ b/libs/vkd3d-shader/ir.c @@ -2917,8 +2917,7 @@ static enum vkd3d_result vsir_block_init(struct vsir_block *block, unsigned int if (block_count > SIZE_MAX - (sizeof(*block->dominates) * CHAR_BIT - 1)) return VKD3D_ERROR_OUT_OF_MEMORY; - block_count = align(block_count, sizeof(*block->dominates) * CHAR_BIT); - byte_count = block_count / CHAR_BIT; + byte_count = VKD3D_BITMAP_SIZE(block_count) * sizeof(*block->dominates); assert(label); memset(block, 0, sizeof(*block)); diff --git a/tests/shader_runner.h b/tests/shader_runner.h index ef0b753b..5f061fa7 100644 --- a/tests/shader_runner.h +++ b/tests/shader_runner.h @@ -177,7 +177,7 @@ struct shader_runner struct resource *resources[MAX_RESOURCES]; size_t resource_count; - uint32_t failed_resources[RESOURCE_TYPE_VERTEX_BUFFER + 1][(MAX_RESOURCES + 0x1f) / 0x20]; + uint32_t failed_resources[RESOURCE_TYPE_VERTEX_BUFFER + 1][VKD3D_BITMAP_SIZE(MAX_RESOURCES)]; unsigned int failed_resource_count; struct sampler samplers[MAX_SAMPLERS];

1 0

Giovanni Mascellani : vkd3d-shader/ir: Skip invalid blocks when materializing undominated SSAs.
by Alexandre Julliard 23 Apr '24

23 Apr '24

Module: vkd3d Branch: master Commit: d8e4b19cdcd7917d82a70fc2f1cb31166fa52742 URL: https://gitlab.winehq.org/wine/vkd3d/-/commit/d8e4b19cdcd7917d82a70fc2f1cb3… Author: Giovanni Mascellani <gmascellani(a)codeweavers.com> Date: Fri Apr 19 13:13:15 2024 +0200 vkd3d-shader/ir: Skip invalid blocks when materializing undominated SSAs. --- libs/vkd3d-shader/ir.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/libs/vkd3d-shader/ir.c b/libs/vkd3d-shader/ir.c index 1f8d60c6..974cfb91 100644 --- a/libs/vkd3d-shader/ir.c +++ b/libs/vkd3d-shader/ir.c @@ -5001,6 +5001,9 @@ static enum vkd3d_result vsir_cfg_materialize_undominated_ssas_to_temps(struct v struct vsir_block *block = &cfg->blocks[i]; struct vkd3d_shader_instruction *ins; + if (block->label == 0) + continue; + for (ins = block->begin; ins <= block->end; ++ins) { for (j = 0; j < ins->dst_count; ++j) @@ -5016,6 +5019,9 @@ static enum vkd3d_result vsir_cfg_materialize_undominated_ssas_to_temps(struct v struct vsir_block *block = &cfg->blocks[i]; struct vkd3d_shader_instruction *ins; + if (block->label == 0) + continue; + for (ins = block->begin; ins <= block->end; ++ins) { for (j = 0; j < ins->src_count; ++j)

1 0