| 1 | /* |
| 2 | * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> |
| 3 | * |
| 4 | * This file is part of FFmpeg. |
| 5 | * |
| 6 | * FFmpeg is free software; you can redistribute it and/or |
| 7 | * modify it under the terms of the GNU Lesser General Public |
| 8 | * License as published by the Free Software Foundation; either |
| 9 | * version 2.1 of the License, or (at your option) any later version. |
| 10 | * |
| 11 | * FFmpeg is distributed in the hope that it will be useful, |
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 14 | * Lesser General Public License for more details. |
| 15 | * |
| 16 | * You should have received a copy of the GNU Lesser General Public |
| 17 | * License along with FFmpeg; if not, write to the Free Software |
| 18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 19 | */ |
| 20 | |
| 21 | #ifndef AVUTIL_X86_ASM_H |
| 22 | #define AVUTIL_X86_ASM_H |
| 23 | |
| 24 | #include <stdint.h> |
| 25 | #include "config.h" |
| 26 | |
| 27 | typedef struct xmm_reg { uint64_t a, b; } xmm_reg; |
| 28 | typedef struct ymm_reg { uint64_t a, b, c, d; } ymm_reg; |
| 29 | |
| 30 | #if ARCH_X86_64 |
| 31 | # define OPSIZE "q" |
| 32 | # define REG_a "rax" |
| 33 | # define REG_b "rbx" |
| 34 | # define REG_c "rcx" |
| 35 | # define REG_d "rdx" |
| 36 | # define REG_D "rdi" |
| 37 | # define REG_S "rsi" |
| 38 | # define PTR_SIZE "8" |
| 39 | typedef int64_t x86_reg; |
| 40 | |
| 41 | # define REG_SP "rsp" |
| 42 | # define REG_BP "rbp" |
| 43 | # define REGBP rbp |
| 44 | # define REGa rax |
| 45 | # define REGb rbx |
| 46 | # define REGc rcx |
| 47 | # define REGd rdx |
| 48 | # define REGSP rsp |
| 49 | |
| 50 | #elif ARCH_X86_32 |
| 51 | |
| 52 | # define OPSIZE "l" |
| 53 | # define REG_a "eax" |
| 54 | # define REG_b "ebx" |
| 55 | # define REG_c "ecx" |
| 56 | # define REG_d "edx" |
| 57 | # define REG_D "edi" |
| 58 | # define REG_S "esi" |
| 59 | # define PTR_SIZE "4" |
| 60 | typedef int32_t x86_reg; |
| 61 | |
| 62 | # define REG_SP "esp" |
| 63 | # define REG_BP "ebp" |
| 64 | # define REGBP ebp |
| 65 | # define REGa eax |
| 66 | # define REGb ebx |
| 67 | # define REGc ecx |
| 68 | # define REGd edx |
| 69 | # define REGSP esp |
| 70 | #else |
| 71 | typedef int x86_reg; |
| 72 | #endif |
| 73 | |
| 74 | #define HAVE_7REGS (ARCH_X86_64 || (HAVE_EBX_AVAILABLE && HAVE_EBP_AVAILABLE)) |
| 75 | #define HAVE_6REGS (ARCH_X86_64 || (HAVE_EBX_AVAILABLE || HAVE_EBP_AVAILABLE)) |
| 76 | |
| 77 | #if ARCH_X86_64 && defined(PIC) |
| 78 | # define BROKEN_RELOCATIONS 1 |
| 79 | #endif |
| 80 | |
| 81 | /* |
| 82 | * If gcc is not set to support sse (-msse) it will not accept xmm registers |
| 83 | * in the clobber list for inline asm. XMM_CLOBBERS takes a list of xmm |
| 84 | * registers to be marked as clobbered and evaluates to nothing if they are |
| 85 | * not supported, or to the list itself if they are supported. Since a clobber |
| 86 | * list may not be empty, XMM_CLOBBERS_ONLY should be used if the xmm |
| 87 | * registers are the only in the clobber list. |
| 88 | * For example a list with "eax" and "xmm0" as clobbers should become: |
| 89 | * : XMM_CLOBBERS("xmm0",) "eax" |
| 90 | * and a list with only "xmm0" should become: |
| 91 | * XMM_CLOBBERS_ONLY("xmm0") |
| 92 | */ |
| 93 | #if HAVE_XMM_CLOBBERS |
| 94 | # define XMM_CLOBBERS(...) __VA_ARGS__ |
| 95 | # define XMM_CLOBBERS_ONLY(...) : __VA_ARGS__ |
| 96 | #else |
| 97 | # define XMM_CLOBBERS(...) |
| 98 | # define XMM_CLOBBERS_ONLY(...) |
| 99 | #endif |
| 100 | |
| 101 | /* Use to export labels from asm. */ |
| 102 | #define LABEL_MANGLE(a) EXTERN_PREFIX #a |
| 103 | |
| 104 | // Use rip-relative addressing if compiling PIC code on x86-64. |
| 105 | #if ARCH_X86_64 && defined(PIC) |
| 106 | # define LOCAL_MANGLE(a) #a "(%%rip)" |
| 107 | #else |
| 108 | # define LOCAL_MANGLE(a) #a |
| 109 | #endif |
| 110 | |
| 111 | #if HAVE_INLINE_ASM_DIRECT_SYMBOL_REFS |
| 112 | # define MANGLE(a) EXTERN_PREFIX LOCAL_MANGLE(a) |
| 113 | # define NAMED_CONSTRAINTS_ADD(...) |
| 114 | # define NAMED_CONSTRAINTS(...) |
| 115 | # define NAMED_CONSTRAINTS_ARRAY_ADD(...) |
| 116 | # define NAMED_CONSTRAINTS_ARRAY(...) |
| 117 | #else |
| 118 | /* When direct symbol references are used in code passed to a compiler that does not support them |
| 119 | * then these references need to be converted to named asm constraints instead. |
| 120 | * Instead of returning a direct symbol MANGLE now returns a named constraint for that specific symbol. |
| 121 | * In order for this to work there must also be a corresponding entry in the asm-interface. To add this |
| 122 | * entry use the macro NAMED_CONSTRAINTS() and pass in a list of each symbol reference used in the |
| 123 | * corresponding block of code. (e.g. NAMED_CONSTRAINTS(var1,var2,var3) where var1 is the first symbol etc. ). |
| 124 | * If there are already existing constraints then use NAMED_CONSTRAINTS_ADD to add to the existing constraint list. |
| 125 | */ |
| 126 | # define MANGLE(a) "%["#a"]" |
| 127 | // Intel/MSVC does not correctly expand va-args so we need a rather ugly hack in order to get it to work |
| 128 | # define FE_0(P,X) P(X) |
| 129 | # define FE_1(P,X,X1) P(X), FE_0(P,X1) |
| 130 | # define FE_2(P,X,X1,X2) P(X), FE_1(P,X1,X2) |
| 131 | # define FE_3(P,X,X1,X2,X3) P(X), FE_2(P,X1,X2,X3) |
| 132 | # define FE_4(P,X,X1,X2,X3,X4) P(X), FE_3(P,X1,X2,X3,X4) |
| 133 | # define FE_5(P,X,X1,X2,X3,X4,X5) P(X), FE_4(P,X1,X2,X3,X4,X5) |
| 134 | # define FE_6(P,X,X1,X2,X3,X4,X5,X6) P(X), FE_5(P,X1,X2,X3,X4,X5,X6) |
| 135 | # define FE_7(P,X,X1,X2,X3,X4,X5,X6,X7) P(X), FE_6(P,X1,X2,X3,X4,X5,X6,X7) |
| 136 | # define FE_8(P,X,X1,X2,X3,X4,X5,X6,X7,X8) P(X), FE_7(P,X1,X2,X3,X4,X5,X6,X7,X8) |
| 137 | # define FE_9(P,X,X1,X2,X3,X4,X5,X6,X7,X8,X9) P(X), FE_8(P,X1,X2,X3,X4,X5,X6,X7,X8,X9) |
| 138 | # define GET_FE_IMPL(_0,_1,_2,_3,_4,_5,_6,_7,_8,_9,NAME,...) NAME |
| 139 | # define GET_FE(A) GET_FE_IMPL A |
| 140 | # define GET_FE_GLUE(x, y) x y |
| 141 | # define FOR_EACH_VA(P,...) GET_FE_GLUE(GET_FE((__VA_ARGS__,FE_9,FE_8,FE_7,FE_6,FE_5,FE_4,FE_3,FE_2,FE_1,FE_0)), (P,__VA_ARGS__)) |
| 142 | # define NAME_CONSTRAINT(x) [x] "m"(x) |
| 143 | // Parameters are a list of each symbol reference required |
| 144 | # define NAMED_CONSTRAINTS_ADD(...) , FOR_EACH_VA(NAME_CONSTRAINT,__VA_ARGS__) |
| 145 | // Same but without comma for when there are no previously defined constraints |
| 146 | # define NAMED_CONSTRAINTS(...) FOR_EACH_VA(NAME_CONSTRAINT,__VA_ARGS__) |
| 147 | // Same as above NAMED_CONSTRAINTS except used for passing arrays/pointers instead of normal variables |
| 148 | # define NAME_CONSTRAINT_ARRAY(x) [x] "m"(*x) |
| 149 | # define NAMED_CONSTRAINTS_ARRAY_ADD(...) , FOR_EACH_VA(NAME_CONSTRAINT_ARRAY,__VA_ARGS__) |
| 150 | # define NAMED_CONSTRAINTS_ARRAY(...) FOR_EACH_VA(NAME_CONSTRAINT_ARRAY,__VA_ARGS__) |
| 151 | #endif |
| 152 | |
| 153 | #endif /* AVUTIL_X86_ASM_H */ |