Since 9b8409fce4da18e3d9a914a9e5831d10eb9052de, a PE compiler is required for 32 bit arm.
That can either be supplied by using plain Clang (from a distro), or llvm-mingw. However when using plain Clang (in MSVC mode), we're lacking compiler builtin functions that either would be provided by MSVC libraries or by compiler-rt libraries bundled in llvm-mingw.
Vendor a copy of the relevant files from compiler-rt and include them when building for arm in MSVC mode.
This allows building a functional wine for 32 bit arm without requiring third party tools such as llvm-mingw.
From: Martin Storsjö martin@martin.st
When building for ARM targets, these helper functions are needed. When building with a llvm-mingw provided compiler, the default compiler-rt builtins from that distribution get linked in implicitly, but if building with a plain Clang (in MSVC mode), Wine needs to supply these routines (in the same way as Wine supplies e.g. __chkstk).
These are taken from the compiler-rt release 8.0.1; this was the last version which was dual licensed under the MIT and University of Illinois licenses. (Newer versions are under Apache 2.0 with LLVM Exceptions.) The files are taken from the compiler-rt/lib/builtins directory. --- libs/compiler-rt/LICENSE.TXT | 91 +++++++ libs/compiler-rt/arm/aeabi_idivmod.S | 51 ++++ libs/compiler-rt/arm/aeabi_ldivmod.S | 46 ++++ libs/compiler-rt/arm/aeabi_uidivmod.S | 58 +++++ libs/compiler-rt/arm/aeabi_uldivmod.S | 46 ++++ libs/compiler-rt/arm/divmodsi4.S | 71 ++++++ libs/compiler-rt/arm/udivmodsi4.S | 180 ++++++++++++++ libs/compiler-rt/assembly.h | 204 ++++++++++++++++ libs/compiler-rt/divdi3.c | 29 +++ libs/compiler-rt/divmoddi4.c | 25 ++ libs/compiler-rt/fixdfdi.c | 55 +++++ libs/compiler-rt/fixsfdi.c | 55 +++++ libs/compiler-rt/fixunsdfdi.c | 52 ++++ libs/compiler-rt/fixunssfdi.c | 53 +++++ libs/compiler-rt/floatdidf.c | 115 +++++++++ libs/compiler-rt/floatdisf.c | 88 +++++++ libs/compiler-rt/floatundidf.c | 114 +++++++++ libs/compiler-rt/floatundisf.c | 85 +++++++ libs/compiler-rt/fp_fixint_impl.inc | 41 ++++ libs/compiler-rt/fp_fixuint_impl.inc | 39 +++ libs/compiler-rt/fp_lib.h | 327 ++++++++++++++++++++++++++ libs/compiler-rt/int_endianness.h | 116 +++++++++ libs/compiler-rt/int_lib.h | 134 +++++++++++ libs/compiler-rt/int_math.h | 110 +++++++++ libs/compiler-rt/int_types.h | 185 +++++++++++++++ libs/compiler-rt/int_util.h | 33 +++ libs/compiler-rt/mingw_fixfloat.c | 36 +++ libs/compiler-rt/udivmoddi4.c | 231 ++++++++++++++++++ 28 files changed, 2670 insertions(+) create mode 100644 libs/compiler-rt/LICENSE.TXT create mode 100644 libs/compiler-rt/arm/aeabi_idivmod.S create mode 100644 libs/compiler-rt/arm/aeabi_ldivmod.S create mode 100644 libs/compiler-rt/arm/aeabi_uidivmod.S create mode 100644 libs/compiler-rt/arm/aeabi_uldivmod.S create mode 100644 libs/compiler-rt/arm/divmodsi4.S create mode 100644 libs/compiler-rt/arm/udivmodsi4.S create mode 100644 libs/compiler-rt/assembly.h create mode 100644 libs/compiler-rt/divdi3.c create mode 100644 libs/compiler-rt/divmoddi4.c create mode 100644 libs/compiler-rt/fixdfdi.c create mode 100644 libs/compiler-rt/fixsfdi.c create mode 100644 libs/compiler-rt/fixunsdfdi.c create mode 100644 libs/compiler-rt/fixunssfdi.c create mode 100644 libs/compiler-rt/floatdidf.c create mode 100644 libs/compiler-rt/floatdisf.c create mode 100644 libs/compiler-rt/floatundidf.c create mode 100644 libs/compiler-rt/floatundisf.c create mode 100644 libs/compiler-rt/fp_fixint_impl.inc create mode 100644 libs/compiler-rt/fp_fixuint_impl.inc create mode 100644 libs/compiler-rt/fp_lib.h create mode 100644 libs/compiler-rt/int_endianness.h create mode 100644 libs/compiler-rt/int_lib.h create mode 100644 libs/compiler-rt/int_math.h create mode 100644 libs/compiler-rt/int_types.h create mode 100644 libs/compiler-rt/int_util.h create mode 100644 libs/compiler-rt/mingw_fixfloat.c create mode 100644 libs/compiler-rt/udivmoddi4.c
diff --git a/libs/compiler-rt/LICENSE.TXT b/libs/compiler-rt/LICENSE.TXT new file mode 100644 index 00000000000..1c94ad5d891 --- /dev/null +++ b/libs/compiler-rt/LICENSE.TXT @@ -0,0 +1,91 @@ +============================================================================== +compiler_rt License +============================================================================== + +The compiler_rt library is dual licensed under both the University of Illinois +"BSD-Like" license and the MIT license. As a user of this code you may choose +to use it under either license. As a contributor, you agree to allow your code +to be used under both. + +Full text of the relevant licenses is included below. + +============================================================================== + +University of Illinois/NCSA +Open Source License + +Copyright (c) 2009-2019 by the contributors listed in CREDITS.TXT + +All rights reserved. + +Developed by: + + LLVM Team + + University of Illinois at Urbana-Champaign + + http://llvm.org + +Permission is hereby granted, free of charge, to any person obtaining a copy of +this software and associated documentation files (the "Software"), to deal with +the Software without restriction, including without limitation the rights to +use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies +of the Software, and to permit persons to whom the Software is furnished to do +so, subject to the following conditions: + + * Redistributions of source code must retain the above copyright notice, + this list of conditions and the following disclaimers. + + * Redistributions in binary form must reproduce the above copyright notice, + this list of conditions and the following disclaimers in the + documentation and/or other materials provided with the distribution. + + * Neither the names of the LLVM Team, University of Illinois at + Urbana-Champaign, nor the names of its contributors may be used to + endorse or promote products derived from this Software without specific + prior written permission. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS +FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +CONTRIBUTORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS WITH THE +SOFTWARE. + +============================================================================== + +Copyright (c) 2009-2015 by the contributors listed in CREDITS.TXT + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in +all copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +THE SOFTWARE. + +============================================================================== +Copyrights and Licenses for Third Party Software Distributed with LLVM: +============================================================================== +The LLVM software contains code written by third parties. Such software will +have its own individual LICENSE.TXT file in the directory in which it appears. +This file will describe the copyrights, license, and restrictions which apply +to that code. + +The disclaimer of warranty in the University of Illinois Open Source License +applies to all code in the LLVM Distribution, and nothing in any of the +other licenses gives permission to use the names of the LLVM Team or the +University of Illinois to endorse or promote products derived from this +Software. + diff --git a/libs/compiler-rt/arm/aeabi_idivmod.S b/libs/compiler-rt/arm/aeabi_idivmod.S new file mode 100644 index 00000000000..9c9c80ab5a7 --- /dev/null +++ b/libs/compiler-rt/arm/aeabi_idivmod.S @@ -0,0 +1,51 @@ +//===-- aeabi_idivmod.S - EABI idivmod implementation ---------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is dual licensed under the MIT and the University of Illinois Open +// Source Licenses. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "../assembly.h" + +// struct { int quot, int rem} __aeabi_idivmod(int numerator, int denominator) { +// int rem, quot; +// quot = __divmodsi4(numerator, denominator, &rem); +// return {quot, rem}; +// } + +#if defined(__MINGW32__) +#define __aeabi_idivmod __rt_sdiv +#endif + + .syntax unified + .text + DEFINE_CODE_STATE + .p2align 2 +DEFINE_COMPILERRT_FUNCTION(__aeabi_idivmod) +#if defined(USE_THUMB_1) + push {r0, r1, lr} + bl SYMBOL_NAME(__divsi3) + pop {r1, r2, r3} // now r0 = quot, r1 = num, r2 = denom + muls r2, r0, r2 // r2 = quot * denom + subs r1, r1, r2 + JMP (r3) +#else // defined(USE_THUMB_1) + push { lr } + sub sp, sp, #4 + mov r2, sp +#if defined(__MINGW32__) + mov r3, r0 + mov r0, r1 + mov r1, r3 +#endif + bl SYMBOL_NAME(__divmodsi4) + ldr r1, [sp] + add sp, sp, #4 + pop { pc } +#endif // defined(USE_THUMB_1) +END_COMPILERRT_FUNCTION(__aeabi_idivmod) + +NO_EXEC_STACK_DIRECTIVE + diff --git a/libs/compiler-rt/arm/aeabi_ldivmod.S b/libs/compiler-rt/arm/aeabi_ldivmod.S new file mode 100644 index 00000000000..038ae5d723a --- /dev/null +++ b/libs/compiler-rt/arm/aeabi_ldivmod.S @@ -0,0 +1,46 @@ +//===-- aeabi_ldivmod.S - EABI ldivmod implementation ---------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is dual licensed under the MIT and the University of Illinois Open +// Source Licenses. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "../assembly.h" + +// struct { int64_t quot, int64_t rem} +// __aeabi_ldivmod(int64_t numerator, int64_t denominator) { +// int64_t rem, quot; +// quot = __divmoddi4(numerator, denominator, &rem); +// return {quot, rem}; +// } + +#if defined(__MINGW32__) +#define __aeabi_ldivmod __rt_sdiv64 +#endif + + .syntax unified + .p2align 2 +DEFINE_COMPILERRT_FUNCTION(__aeabi_ldivmod) + push {r6, lr} + sub sp, sp, #16 + add r6, sp, #8 + str r6, [sp] +#if defined(__MINGW32__) + movs r6, r0 + movs r0, r2 + movs r2, r6 + movs r6, r1 + movs r1, r3 + movs r3, r6 +#endif + bl SYMBOL_NAME(__divmoddi4) + ldr r2, [sp, #8] + ldr r3, [sp, #12] + add sp, sp, #16 + pop {r6, pc} +END_COMPILERRT_FUNCTION(__aeabi_ldivmod) + +NO_EXEC_STACK_DIRECTIVE + diff --git a/libs/compiler-rt/arm/aeabi_uidivmod.S b/libs/compiler-rt/arm/aeabi_uidivmod.S new file mode 100644 index 00000000000..88a4a6d8bc1 --- /dev/null +++ b/libs/compiler-rt/arm/aeabi_uidivmod.S @@ -0,0 +1,58 @@ +//===-- aeabi_uidivmod.S - EABI uidivmod implementation -------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is dual licensed under the MIT and the University of Illinois Open +// Source Licenses. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "../assembly.h" + +// struct { unsigned quot, unsigned rem} +// __aeabi_uidivmod(unsigned numerator, unsigned denominator) { +// unsigned rem, quot; +// quot = __udivmodsi4(numerator, denominator, &rem); +// return {quot, rem}; +// } + +#if defined(__MINGW32__) +#define __aeabi_uidivmod __rt_udiv +#endif + + .syntax unified + .text + DEFINE_CODE_STATE + .p2align 2 +DEFINE_COMPILERRT_FUNCTION(__aeabi_uidivmod) +#if defined(USE_THUMB_1) + cmp r0, r1 + bcc LOCAL_LABEL(case_denom_larger) + push {r0, r1, lr} + bl SYMBOL_NAME(__aeabi_uidiv) + pop {r1, r2, r3} + muls r2, r0, r2 // r2 = quot * denom + subs r1, r1, r2 + JMP (r3) +LOCAL_LABEL(case_denom_larger): + movs r1, r0 + movs r0, #0 + JMP (lr) +#else // defined(USE_THUMB_1) + push { lr } + sub sp, sp, #4 + mov r2, sp +#if defined(__MINGW32__) + mov r3, r0 + mov r0, r1 + mov r1, r3 +#endif + bl SYMBOL_NAME(__udivmodsi4) + ldr r1, [sp] + add sp, sp, #4 + pop { pc } +#endif +END_COMPILERRT_FUNCTION(__aeabi_uidivmod) + +NO_EXEC_STACK_DIRECTIVE + diff --git a/libs/compiler-rt/arm/aeabi_uldivmod.S b/libs/compiler-rt/arm/aeabi_uldivmod.S new file mode 100644 index 00000000000..be343b6bc82 --- /dev/null +++ b/libs/compiler-rt/arm/aeabi_uldivmod.S @@ -0,0 +1,46 @@ +//===-- aeabi_uldivmod.S - EABI uldivmod implementation -------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is dual licensed under the MIT and the University of Illinois Open +// Source Licenses. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "../assembly.h" + +// struct { uint64_t quot, uint64_t rem} +// __aeabi_uldivmod(uint64_t numerator, uint64_t denominator) { +// uint64_t rem, quot; +// quot = __udivmoddi4(numerator, denominator, &rem); +// return {quot, rem}; +// } + +#if defined(__MINGW32__) +#define __aeabi_uldivmod __rt_udiv64 +#endif + + .syntax unified + .p2align 2 +DEFINE_COMPILERRT_FUNCTION(__aeabi_uldivmod) + push {r6, lr} + sub sp, sp, #16 + add r6, sp, #8 + str r6, [sp] +#if defined(__MINGW32__) + movs r6, r0 + movs r0, r2 + movs r2, r6 + movs r6, r1 + movs r1, r3 + movs r3, r6 +#endif + bl SYMBOL_NAME(__udivmoddi4) + ldr r2, [sp, #8] + ldr r3, [sp, #12] + add sp, sp, #16 + pop {r6, pc} +END_COMPILERRT_FUNCTION(__aeabi_uldivmod) + +NO_EXEC_STACK_DIRECTIVE + diff --git a/libs/compiler-rt/arm/divmodsi4.S b/libs/compiler-rt/arm/divmodsi4.S new file mode 100644 index 00000000000..8a027b741ef --- /dev/null +++ b/libs/compiler-rt/arm/divmodsi4.S @@ -0,0 +1,71 @@ +/*===-- divmodsi4.S - 32-bit signed integer divide and modulus ------------===// + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + *===----------------------------------------------------------------------===// + * + * This file implements the __divmodsi4 (32-bit signed integer divide and + * modulus) function for the ARM architecture. A naive digit-by-digit + * computation is employed for simplicity. + * + *===----------------------------------------------------------------------===*/ + +#include "../assembly.h" + +#define ESTABLISH_FRAME \ + push {r4-r7, lr} ;\ + add r7, sp, #12 +#define CLEAR_FRAME_AND_RETURN \ + pop {r4-r7, pc} + + .syntax unified + .text + DEFINE_CODE_STATE + +@ int __divmodsi4(int divident, int divisor, int *remainder) +@ Calculate the quotient and remainder of the (signed) division. The return +@ value is the quotient, the remainder is placed in the variable. + + .p2align 3 +DEFINE_COMPILERRT_FUNCTION(__divmodsi4) +#if __ARM_ARCH_EXT_IDIV__ + tst r1, r1 + beq LOCAL_LABEL(divzero) + mov r3, r0 + sdiv r0, r3, r1 + mls r1, r0, r1, r3 + str r1, [r2] + bx lr +LOCAL_LABEL(divzero): + mov r0, #0 + bx lr +#else + ESTABLISH_FRAME +// Set aside the sign of the quotient and modulus, and the address for the +// modulus. + eor r4, r0, r1 + mov r5, r0 + mov r6, r2 +// Take the absolute value of a and b via abs(x) = (x^(x >> 31)) - (x >> 31). + eor ip, r0, r0, asr #31 + eor lr, r1, r1, asr #31 + sub r0, ip, r0, asr #31 + sub r1, lr, r1, asr #31 +// Unsigned divmod: + bl SYMBOL_NAME(__udivmodsi4) +// Apply the sign of quotient and modulus + ldr r1, [r6] + eor r0, r0, r4, asr #31 + eor r1, r1, r5, asr #31 + sub r0, r0, r4, asr #31 + sub r1, r1, r5, asr #31 + str r1, [r6] + CLEAR_FRAME_AND_RETURN +#endif +END_COMPILERRT_FUNCTION(__divmodsi4) + +NO_EXEC_STACK_DIRECTIVE + diff --git a/libs/compiler-rt/arm/udivmodsi4.S b/libs/compiler-rt/arm/udivmodsi4.S new file mode 100644 index 00000000000..ee3950c9b0e --- /dev/null +++ b/libs/compiler-rt/arm/udivmodsi4.S @@ -0,0 +1,180 @@ +/*===-- udivmodsi4.S - 32-bit unsigned integer divide and modulus ---------===// + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + *===----------------------------------------------------------------------===// + * + * This file implements the __udivmodsi4 (32-bit unsigned integer divide and + * modulus) function for the ARM 32-bit architecture. + * + *===----------------------------------------------------------------------===*/ + +#include "../assembly.h" + + .syntax unified + .text + DEFINE_CODE_STATE + +@ unsigned int __udivmodsi4(unsigned int divident, unsigned int divisor, +@ unsigned int *remainder) +@ Calculate the quotient and remainder of the (unsigned) division. The return +@ value is the quotient, the remainder is placed in the variable. + + .p2align 2 +DEFINE_COMPILERRT_FUNCTION(__udivmodsi4) +#if __ARM_ARCH_EXT_IDIV__ + tst r1, r1 + beq LOCAL_LABEL(divby0) + mov r3, r0 + udiv r0, r3, r1 + mls r1, r0, r1, r3 + str r1, [r2] + bx lr +#else + cmp r1, #1 + bcc LOCAL_LABEL(divby0) + beq LOCAL_LABEL(divby1) + cmp r0, r1 + bcc LOCAL_LABEL(quotient0) + /* + * Implement division using binary long division algorithm. + * + * r0 is the numerator, r1 the denominator. + * + * The code before JMP computes the correct shift I, so that + * r0 and (r1 << I) have the highest bit set in the same position. + * At the time of JMP, ip := .Ldiv0block - 12 * I. + * This depends on the fixed instruction size of block. + * For ARM mode, this is 12 Bytes, for THUMB mode 14 Bytes. + * + * block(shift) implements the test-and-update-quotient core. + * It assumes (r0 << shift) can be computed without overflow and + * that (r0 << shift) < 2 * r1. The quotient is stored in r3. + */ + +# ifdef __ARM_FEATURE_CLZ + clz ip, r0 + clz r3, r1 + /* r0 >= r1 implies clz(r0) <= clz(r1), so ip <= r3. */ + sub r3, r3, ip +# if defined(USE_THUMB_2) + adr ip, LOCAL_LABEL(div0block) + 1 + sub ip, ip, r3, lsl #1 +# else + adr ip, LOCAL_LABEL(div0block) +# endif + sub ip, ip, r3, lsl #2 + sub ip, ip, r3, lsl #3 + mov r3, #0 + bx ip +# else +# if defined(USE_THUMB_2) +# error THUMB mode requires CLZ or UDIV +# endif + str r4, [sp, #-8]! + + mov r4, r0 + adr ip, LOCAL_LABEL(div0block) + + lsr r3, r4, #16 + cmp r3, r1 + movhs r4, r3 + subhs ip, ip, #(16 * 12) + + lsr r3, r4, #8 + cmp r3, r1 + movhs r4, r3 + subhs ip, ip, #(8 * 12) + + lsr r3, r4, #4 + cmp r3, r1 + movhs r4, r3 + subhs ip, #(4 * 12) + + lsr r3, r4, #2 + cmp r3, r1 + movhs r4, r3 + subhs ip, ip, #(2 * 12) + + /* Last block, no need to update r3 or r4. */ + cmp r1, r4, lsr #1 + subls ip, ip, #(1 * 12) + + ldr r4, [sp], #8 /* restore r4, we are done with it. */ + mov r3, #0 + + JMP(ip) +# endif + +#define IMM # + +#define block(shift) \ + cmp r0, r1, lsl IMM shift; \ + ITT(hs); \ + WIDE(addhs) r3, r3, IMM (1 << shift); \ + WIDE(subhs) r0, r0, r1, lsl IMM shift + + block(31) + block(30) + block(29) + block(28) + block(27) + block(26) + block(25) + block(24) + block(23) + block(22) + block(21) + block(20) + block(19) + block(18) + block(17) + block(16) + block(15) + block(14) + block(13) + block(12) + block(11) + block(10) + block(9) + block(8) + block(7) + block(6) + block(5) + block(4) + block(3) + block(2) + block(1) +LOCAL_LABEL(div0block): + block(0) + + str r0, [r2] + mov r0, r3 + JMP(lr) + +LOCAL_LABEL(quotient0): + str r0, [r2] + mov r0, #0 + JMP(lr) + +LOCAL_LABEL(divby1): + mov r3, #0 + str r3, [r2] + JMP(lr) +#endif /* __ARM_ARCH_EXT_IDIV__ */ + +LOCAL_LABEL(divby0): + mov r0, #0 +#ifdef __ARM_EABI__ + b __aeabi_idiv0 +#else + JMP(lr) +#endif + +END_COMPILERRT_FUNCTION(__udivmodsi4) + +NO_EXEC_STACK_DIRECTIVE + diff --git a/libs/compiler-rt/assembly.h b/libs/compiler-rt/assembly.h new file mode 100644 index 00000000000..3f5e59b2544 --- /dev/null +++ b/libs/compiler-rt/assembly.h @@ -0,0 +1,204 @@ +/* ===-- assembly.h - compiler-rt assembler support macros -----------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file defines macros for use in compiler-rt assembler source. + * This file is not part of the interface of this library. + * + * ===----------------------------------------------------------------------=== + */ + +#ifndef COMPILERRT_ASSEMBLY_H +#define COMPILERRT_ASSEMBLY_H + +#if defined(__POWERPC__) || defined(__powerpc__) || defined(__ppc__) +#define SEPARATOR @ +#else +#define SEPARATOR ; +#endif + +#if defined(__APPLE__) +#define HIDDEN(name) .private_extern name +#define LOCAL_LABEL(name) L_##name +// tell linker it can break up file at label boundaries +#define FILE_LEVEL_DIRECTIVE .subsections_via_symbols +#define SYMBOL_IS_FUNC(name) +#define CONST_SECTION .const + +#define NO_EXEC_STACK_DIRECTIVE + +#elif defined(__ELF__) + +#define HIDDEN(name) .hidden name +#define LOCAL_LABEL(name) .L_##name +#define FILE_LEVEL_DIRECTIVE +#if defined(__arm__) +#define SYMBOL_IS_FUNC(name) .type name,%function +#else +#define SYMBOL_IS_FUNC(name) .type name,@function +#endif +#define CONST_SECTION .section .rodata + +#if defined(__GNU__) || defined(__FreeBSD__) || defined(__Fuchsia__) || \ + defined(__linux__) +#define NO_EXEC_STACK_DIRECTIVE .section .note.GNU-stack,"",%progbits +#else +#define NO_EXEC_STACK_DIRECTIVE +#endif + +#else // !__APPLE__ && !__ELF__ + +#define HIDDEN(name) +#define LOCAL_LABEL(name) .L ## name +#define FILE_LEVEL_DIRECTIVE +#define SYMBOL_IS_FUNC(name) \ + .def name SEPARATOR \ + .scl 2 SEPARATOR \ + .type 32 SEPARATOR \ + .endef +#define CONST_SECTION .section .rdata,"rd" + +#define NO_EXEC_STACK_DIRECTIVE + +#endif + +#if defined(__arm__) + +/* + * Determine actual [ARM][THUMB[1][2]] ISA using compiler predefined macros: + * - for '-mthumb -march=armv6' compiler defines '__thumb__' + * - for '-mthumb -march=armv7' compiler defines '__thumb__' and '__thumb2__' + */ +#if defined(__thumb2__) || defined(__thumb__) +#define DEFINE_CODE_STATE .thumb SEPARATOR +#define DECLARE_FUNC_ENCODING .thumb_func SEPARATOR +#if defined(__thumb2__) +#define USE_THUMB_2 +#define IT(cond) it cond +#define ITT(cond) itt cond +#define ITE(cond) ite cond +#else +#define USE_THUMB_1 +#define IT(cond) +#define ITT(cond) +#define ITE(cond) +#endif // defined(__thumb__2) +#else // !defined(__thumb2__) && !defined(__thumb__) +#define DEFINE_CODE_STATE .arm SEPARATOR +#define DECLARE_FUNC_ENCODING +#define IT(cond) +#define ITT(cond) +#define ITE(cond) +#endif + +#if defined(USE_THUMB_1) && defined(USE_THUMB_2) +#error "USE_THUMB_1 and USE_THUMB_2 can't be defined together." +#endif + +#if defined(__ARM_ARCH_4T__) || __ARM_ARCH >= 5 +#define ARM_HAS_BX +#endif +#if !defined(__ARM_FEATURE_CLZ) && !defined(USE_THUMB_1) && \ + (__ARM_ARCH >= 6 || (__ARM_ARCH == 5 && !defined(__ARM_ARCH_5__))) +#define __ARM_FEATURE_CLZ +#endif + +#ifdef ARM_HAS_BX +#define JMP(r) bx r +#define JMPc(r, c) bx##c r +#else +#define JMP(r) mov pc, r +#define JMPc(r, c) mov##c pc, r +#endif + +// pop {pc} can't switch Thumb mode on ARMv4T +#if __ARM_ARCH >= 5 +#define POP_PC() pop {pc} +#else +#define POP_PC() \ + pop {ip}; \ + JMP(ip) +#endif + +#if defined(USE_THUMB_2) +#define WIDE(op) op.w +#else +#define WIDE(op) op +#endif +#else // !defined(__arm) +#define DECLARE_FUNC_ENCODING +#define DEFINE_CODE_STATE +#endif + +#define GLUE2(a, b) a##b +#define GLUE(a, b) GLUE2(a, b) +#define SYMBOL_NAME(name) GLUE(__USER_LABEL_PREFIX__, name) + +#ifdef VISIBILITY_HIDDEN +#define DECLARE_SYMBOL_VISIBILITY(name) \ + HIDDEN(SYMBOL_NAME(name)) SEPARATOR +#else +#define DECLARE_SYMBOL_VISIBILITY(name) +#endif + +#define DEFINE_COMPILERRT_FUNCTION(name) \ + DEFINE_CODE_STATE \ + FILE_LEVEL_DIRECTIVE SEPARATOR \ + .globl SYMBOL_NAME(name) SEPARATOR \ + SYMBOL_IS_FUNC(SYMBOL_NAME(name)) SEPARATOR \ + DECLARE_SYMBOL_VISIBILITY(name) \ + DECLARE_FUNC_ENCODING \ + SYMBOL_NAME(name): + +#define DEFINE_COMPILERRT_THUMB_FUNCTION(name) \ + DEFINE_CODE_STATE \ + FILE_LEVEL_DIRECTIVE SEPARATOR \ + .globl SYMBOL_NAME(name) SEPARATOR \ + SYMBOL_IS_FUNC(SYMBOL_NAME(name)) SEPARATOR \ + DECLARE_SYMBOL_VISIBILITY(name) SEPARATOR \ + .thumb_func SEPARATOR \ + SYMBOL_NAME(name): + +#define DEFINE_COMPILERRT_PRIVATE_FUNCTION(name) \ + DEFINE_CODE_STATE \ + FILE_LEVEL_DIRECTIVE SEPARATOR \ + .globl SYMBOL_NAME(name) SEPARATOR \ + SYMBOL_IS_FUNC(SYMBOL_NAME(name)) SEPARATOR \ + HIDDEN(SYMBOL_NAME(name)) SEPARATOR \ + DECLARE_FUNC_ENCODING \ + SYMBOL_NAME(name): + +#define DEFINE_COMPILERRT_PRIVATE_FUNCTION_UNMANGLED(name) \ + DEFINE_CODE_STATE \ + .globl name SEPARATOR \ + SYMBOL_IS_FUNC(name) SEPARATOR \ + HIDDEN(name) SEPARATOR \ + DECLARE_FUNC_ENCODING \ + name: + +#define DEFINE_COMPILERRT_FUNCTION_ALIAS(name, target) \ + .globl SYMBOL_NAME(name) SEPARATOR \ + SYMBOL_IS_FUNC(SYMBOL_NAME(name)) SEPARATOR \ + DECLARE_SYMBOL_VISIBILITY(SYMBOL_NAME(name)) SEPARATOR \ + .set SYMBOL_NAME(name), SYMBOL_NAME(target) SEPARATOR + +#if defined(__ARM_EABI__) +#define DEFINE_AEABI_FUNCTION_ALIAS(aeabi_name, name) \ + DEFINE_COMPILERRT_FUNCTION_ALIAS(aeabi_name, name) +#else +#define DEFINE_AEABI_FUNCTION_ALIAS(aeabi_name, name) +#endif + +#ifdef __ELF__ +#define END_COMPILERRT_FUNCTION(name) \ + .size SYMBOL_NAME(name), . - SYMBOL_NAME(name) +#else +#define END_COMPILERRT_FUNCTION(name) +#endif + +#endif /* COMPILERRT_ASSEMBLY_H */ diff --git a/libs/compiler-rt/divdi3.c b/libs/compiler-rt/divdi3.c new file mode 100644 index 00000000000..b8eebcb2046 --- /dev/null +++ b/libs/compiler-rt/divdi3.c @@ -0,0 +1,29 @@ +/* ===-- divdi3.c - Implement __divdi3 -------------------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file implements __divdi3 for the compiler_rt library. + * + * ===----------------------------------------------------------------------=== + */ + +#include "int_lib.h" + +/* Returns: a / b */ + +COMPILER_RT_ABI di_int +__divdi3(di_int a, di_int b) +{ + const int bits_in_dword_m1 = (int)(sizeof(di_int) * CHAR_BIT) - 1; + di_int s_a = a >> bits_in_dword_m1; /* s_a = a < 0 ? -1 : 0 */ + di_int s_b = b >> bits_in_dword_m1; /* s_b = b < 0 ? -1 : 0 */ + a = (a ^ s_a) - s_a; /* negate if s_a == -1 */ + b = (b ^ s_b) - s_b; /* negate if s_b == -1 */ + s_a ^= s_b; /*sign of quotient */ + return (__udivmoddi4(a, b, (du_int*)0) ^ s_a) - s_a; /* negate if s_a == -1 */ +} diff --git a/libs/compiler-rt/divmoddi4.c b/libs/compiler-rt/divmoddi4.c new file mode 100644 index 00000000000..0d4df67a63e --- /dev/null +++ b/libs/compiler-rt/divmoddi4.c @@ -0,0 +1,25 @@ +/*===-- divmoddi4.c - Implement __divmoddi4 --------------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file implements __divmoddi4 for the compiler_rt library. + * + * ===----------------------------------------------------------------------=== + */ + +#include "int_lib.h" + +/* Returns: a / b, *rem = a % b */ + +COMPILER_RT_ABI di_int +__divmoddi4(di_int a, di_int b, di_int* rem) +{ + di_int d = __divdi3(a,b); + *rem = a - (d*b); + return d; +} diff --git a/libs/compiler-rt/fixdfdi.c b/libs/compiler-rt/fixdfdi.c new file mode 100644 index 00000000000..54e312d3c8f --- /dev/null +++ b/libs/compiler-rt/fixdfdi.c @@ -0,0 +1,55 @@ +/* ===-- fixdfdi.c - Implement __fixdfdi -----------------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + */ + +#define DOUBLE_PRECISION +#include "fp_lib.h" + +#ifndef __SOFT_FP__ +/* Support for systems that have hardware floating-point; can set the invalid + * flag as a side-effect of computation. + */ + +COMPILER_RT_ABI du_int __fixunsdfdi(double a); + +COMPILER_RT_ABI di_int +__fixdfdi(double a) +{ + if (a < 0.0) { + return -__fixunsdfdi(-a); + } + return __fixunsdfdi(a); +} + +#else +/* Support for systems that don't have hardware floating-point; there are no + * flags to set, and we don't want to code-gen to an unknown soft-float + * implementation. + */ + +typedef di_int fixint_t; +typedef du_int fixuint_t; +#include "fp_fixint_impl.inc" + +COMPILER_RT_ABI di_int +__fixdfdi(fp_t a) { + return __fixint(a); +} + +#endif + +#if defined(__ARM_EABI__) +#if defined(COMPILER_RT_ARMHF_TARGET) +AEABI_RTABI di_int __aeabi_d2lz(fp_t a) { + return __fixdfdi(a); +} +#else +AEABI_RTABI di_int __aeabi_d2lz(fp_t a) COMPILER_RT_ALIAS(__fixdfdi); +#endif +#endif diff --git a/libs/compiler-rt/fixsfdi.c b/libs/compiler-rt/fixsfdi.c new file mode 100644 index 00000000000..32e87c60889 --- /dev/null +++ b/libs/compiler-rt/fixsfdi.c @@ -0,0 +1,55 @@ +/* ===-- fixsfdi.c - Implement __fixsfdi -----------------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + */ + +#define SINGLE_PRECISION +#include "fp_lib.h" + +#ifndef __SOFT_FP__ +/* Support for systems that have hardware floating-point; can set the invalid + * flag as a side-effect of computation. + */ + +COMPILER_RT_ABI du_int __fixunssfdi(float a); + +COMPILER_RT_ABI di_int +__fixsfdi(float a) +{ + if (a < 0.0f) { + return -__fixunssfdi(-a); + } + return __fixunssfdi(a); +} + +#else +/* Support for systems that don't have hardware floating-point; there are no + * flags to set, and we don't want to code-gen to an unknown soft-float + * implementation. + */ + +typedef di_int fixint_t; +typedef du_int fixuint_t; +#include "fp_fixint_impl.inc" + +COMPILER_RT_ABI di_int +__fixsfdi(fp_t a) { + return __fixint(a); +} + +#endif + +#if defined(__ARM_EABI__) +#if defined(COMPILER_RT_ARMHF_TARGET) +AEABI_RTABI di_int __aeabi_f2lz(fp_t a) { + return __fixsfdi(a); +} +#else +AEABI_RTABI di_int __aeabi_f2lz(fp_t a) COMPILER_RT_ALIAS(__fixsfdi); +#endif +#endif diff --git a/libs/compiler-rt/fixunsdfdi.c b/libs/compiler-rt/fixunsdfdi.c new file mode 100644 index 00000000000..bfe4dbb2565 --- /dev/null +++ b/libs/compiler-rt/fixunsdfdi.c @@ -0,0 +1,52 @@ +/* ===-- fixunsdfdi.c - Implement __fixunsdfdi -----------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + */ + +#define DOUBLE_PRECISION +#include "fp_lib.h" + +#ifndef __SOFT_FP__ +/* Support for systems that have hardware floating-point; can set the invalid + * flag as a side-effect of computation. + */ + +COMPILER_RT_ABI du_int +__fixunsdfdi(double a) +{ + if (a <= 0.0) return 0; + su_int high = a / 4294967296.f; /* a / 0x1p32f; */ + su_int low = a - (double)high * 4294967296.f; /* high * 0x1p32f; */ + return ((du_int)high << 32) | low; +} + +#else +/* Support for systems that don't have hardware floating-point; there are no + * flags to set, and we don't want to code-gen to an unknown soft-float + * implementation. + */ + +typedef du_int fixuint_t; +#include "fp_fixuint_impl.inc" + +COMPILER_RT_ABI du_int +__fixunsdfdi(fp_t a) { + return __fixuint(a); +} + +#endif + +#if defined(__ARM_EABI__) +#if defined(COMPILER_RT_ARMHF_TARGET) +AEABI_RTABI du_int __aeabi_d2ulz(fp_t a) { + return __fixunsdfdi(a); +} +#else +AEABI_RTABI du_int __aeabi_d2ulz(fp_t a) COMPILER_RT_ALIAS(__fixunsdfdi); +#endif +#endif diff --git a/libs/compiler-rt/fixunssfdi.c b/libs/compiler-rt/fixunssfdi.c new file mode 100644 index 00000000000..080a25bb1e9 --- /dev/null +++ b/libs/compiler-rt/fixunssfdi.c @@ -0,0 +1,53 @@ +/* ===-- fixunssfdi.c - Implement __fixunssfdi -----------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + */ + +#define SINGLE_PRECISION +#include "fp_lib.h" + +#ifndef __SOFT_FP__ +/* Support for systems that have hardware floating-point; can set the invalid + * flag as a side-effect of computation. + */ + +COMPILER_RT_ABI du_int +__fixunssfdi(float a) +{ + if (a <= 0.0f) return 0; + double da = a; + su_int high = da / 4294967296.f; /* da / 0x1p32f; */ + su_int low = da - (double)high * 4294967296.f; /* high * 0x1p32f; */ + return ((du_int)high << 32) | low; +} + +#else +/* Support for systems that don't have hardware floating-point; there are no + * flags to set, and we don't want to code-gen to an unknown soft-float + * implementation. + */ + +typedef du_int fixuint_t; +#include "fp_fixuint_impl.inc" + +COMPILER_RT_ABI du_int +__fixunssfdi(fp_t a) { + return __fixuint(a); +} + +#endif + +#if defined(__ARM_EABI__) +#if defined(COMPILER_RT_ARMHF_TARGET) +AEABI_RTABI du_int __aeabi_f2ulz(fp_t a) { + return __fixunssfdi(a); +} +#else +AEABI_RTABI du_int __aeabi_f2ulz(fp_t a) COMPILER_RT_ALIAS(__fixunssfdi); +#endif +#endif diff --git a/libs/compiler-rt/floatdidf.c b/libs/compiler-rt/floatdidf.c new file mode 100644 index 00000000000..36b856e078d --- /dev/null +++ b/libs/compiler-rt/floatdidf.c @@ -0,0 +1,115 @@ +/*===-- floatdidf.c - Implement __floatdidf -------------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + *===----------------------------------------------------------------------=== + * + * This file implements __floatdidf for the compiler_rt library. + * + *===----------------------------------------------------------------------=== + */ + +#include "int_lib.h" + +/* Returns: convert a to a double, rounding toward even. */ + +/* Assumption: double is a IEEE 64 bit floating point type + * di_int is a 64 bit integral type + */ + +/* seee eeee eeee mmmm mmmm mmmm mmmm mmmm | mmmm mmmm mmmm mmmm mmmm mmmm mmmm mmmm */ + +#ifndef __SOFT_FP__ +/* Support for systems that have hardware floating-point; we'll set the inexact flag + * as a side-effect of this computation. + */ + +COMPILER_RT_ABI double +__floatdidf(di_int a) +{ + static const double twop52 = 4503599627370496.0; // 0x1.0p52 + static const double twop32 = 4294967296.0; // 0x1.0p32 + + union { int64_t x; double d; } low = { .d = twop52 }; + + const double high = (int32_t)(a >> 32) * twop32; + low.x |= a & INT64_C(0x00000000ffffffff); + + const double result = (high - twop52) + low.d; + return result; +} + +#else +/* Support for systems that don't have hardware floating-point; there are no flags to + * set, and we don't want to code-gen to an unknown soft-float implementation. + */ + +COMPILER_RT_ABI double +__floatdidf(di_int a) +{ + if (a == 0) + return 0.0; + const unsigned N = sizeof(di_int) * CHAR_BIT; + const di_int s = a >> (N-1); + a = (a ^ s) - s; + int sd = N - __builtin_clzll(a); /* number of significant digits */ + int e = sd - 1; /* exponent */ + if (sd > DBL_MANT_DIG) + { + /* start: 0000000000000000000001xxxxxxxxxxxxxxxxxxxxxxPQxxxxxxxxxxxxxxxxxx + * finish: 000000000000000000000000000000000000001xxxxxxxxxxxxxxxxxxxxxxPQR + * 12345678901234567890123456 + * 1 = msb 1 bit + * P = bit DBL_MANT_DIG-1 bits to the right of 1 + * Q = bit DBL_MANT_DIG bits to the right of 1 + * R = "or" of all bits to the right of Q + */ + switch (sd) + { + case DBL_MANT_DIG + 1: + a <<= 1; + break; + case DBL_MANT_DIG + 2: + break; + default: + a = ((du_int)a >> (sd - (DBL_MANT_DIG+2))) | + ((a & ((du_int)(-1) >> ((N + DBL_MANT_DIG+2) - sd))) != 0); + }; + /* finish: */ + a |= (a & 4) != 0; /* Or P into R */ + ++a; /* round - this step may add a significant bit */ + a >>= 2; /* dump Q and R */ + /* a is now rounded to DBL_MANT_DIG or DBL_MANT_DIG+1 bits */ + if (a & ((du_int)1 << DBL_MANT_DIG)) + { + a >>= 1; + ++e; + } + /* a is now rounded to DBL_MANT_DIG bits */ + } + else + { + a <<= (DBL_MANT_DIG - sd); + /* a is now rounded to DBL_MANT_DIG bits */ + } + double_bits fb; + fb.u.s.high = ((su_int)s & 0x80000000) | /* sign */ + ((e + 1023) << 20) | /* exponent */ + ((su_int)(a >> 32) & 0x000FFFFF); /* mantissa-high */ + fb.u.s.low = (su_int)a; /* mantissa-low */ + return fb.f; +} +#endif + +#if defined(__ARM_EABI__) +#if defined(COMPILER_RT_ARMHF_TARGET) +AEABI_RTABI double __aeabi_l2d(di_int a) { + return __floatdidf(a); +} +#else +AEABI_RTABI double __aeabi_l2d(di_int a) COMPILER_RT_ALIAS(__floatdidf); +#endif +#endif diff --git a/libs/compiler-rt/floatdisf.c b/libs/compiler-rt/floatdisf.c new file mode 100644 index 00000000000..a2f09eb2ed2 --- /dev/null +++ b/libs/compiler-rt/floatdisf.c @@ -0,0 +1,88 @@ +/*===-- floatdisf.c - Implement __floatdisf -------------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + *===----------------------------------------------------------------------=== + * + * This file implements __floatdisf for the compiler_rt library. + * + *===----------------------------------------------------------------------=== + */ + +/* Returns: convert a to a float, rounding toward even.*/ + +/* Assumption: float is a IEEE 32 bit floating point type + * di_int is a 64 bit integral type + */ + +/* seee eeee emmm mmmm mmmm mmmm mmmm mmmm */ + +#include "int_lib.h" + +COMPILER_RT_ABI float +__floatdisf(di_int a) +{ + if (a == 0) + return 0.0F; + const unsigned N = sizeof(di_int) * CHAR_BIT; + const di_int s = a >> (N-1); + a = (a ^ s) - s; + int sd = N - __builtin_clzll(a); /* number of significant digits */ + int e = sd - 1; /* exponent */ + if (sd > FLT_MANT_DIG) + { + /* start: 0000000000000000000001xxxxxxxxxxxxxxxxxxxxxxPQxxxxxxxxxxxxxxxxxx + * finish: 000000000000000000000000000000000000001xxxxxxxxxxxxxxxxxxxxxxPQR + * 12345678901234567890123456 + * 1 = msb 1 bit + * P = bit FLT_MANT_DIG-1 bits to the right of 1 + * Q = bit FLT_MANT_DIG bits to the right of 1 + * R = "or" of all bits to the right of Q + */ + switch (sd) + { + case FLT_MANT_DIG + 1: + a <<= 1; + break; + case FLT_MANT_DIG + 2: + break; + default: + a = ((du_int)a >> (sd - (FLT_MANT_DIG+2))) | + ((a & ((du_int)(-1) >> ((N + FLT_MANT_DIG+2) - sd))) != 0); + }; + /* finish: */ + a |= (a & 4) != 0; /* Or P into R */ + ++a; /* round - this step may add a significant bit */ + a >>= 2; /* dump Q and R */ + /* a is now rounded to FLT_MANT_DIG or FLT_MANT_DIG+1 bits */ + if (a & ((du_int)1 << FLT_MANT_DIG)) + { + a >>= 1; + ++e; + } + /* a is now rounded to FLT_MANT_DIG bits */ + } + else + { + a <<= (FLT_MANT_DIG - sd); + /* a is now rounded to FLT_MANT_DIG bits */ + } + float_bits fb; + fb.u = ((su_int)s & 0x80000000) | /* sign */ + ((e + 127) << 23) | /* exponent */ + ((su_int)a & 0x007FFFFF); /* mantissa */ + return fb.f; +} + +#if defined(__ARM_EABI__) +#if defined(COMPILER_RT_ARMHF_TARGET) +AEABI_RTABI float __aeabi_l2f(di_int a) { + return __floatdisf(a); +} +#else +AEABI_RTABI float __aeabi_l2f(di_int a) COMPILER_RT_ALIAS(__floatdisf); +#endif +#endif diff --git a/libs/compiler-rt/floatundidf.c b/libs/compiler-rt/floatundidf.c new file mode 100644 index 00000000000..8bc2a096324 --- /dev/null +++ b/libs/compiler-rt/floatundidf.c @@ -0,0 +1,114 @@ +/* ===-- floatundidf.c - Implement __floatundidf ---------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file implements __floatundidf for the compiler_rt library. + * + * ===----------------------------------------------------------------------=== + */ + +/* Returns: convert a to a double, rounding toward even. */ + +/* Assumption: double is a IEEE 64 bit floating point type + * du_int is a 64 bit integral type + */ + +/* seee eeee eeee mmmm mmmm mmmm mmmm mmmm | mmmm mmmm mmmm mmmm mmmm mmmm mmmm mmmm */ + +#include "int_lib.h" + +#ifndef __SOFT_FP__ +/* Support for systems that have hardware floating-point; we'll set the inexact flag + * as a side-effect of this computation. + */ + +COMPILER_RT_ABI double +__floatundidf(du_int a) +{ + static const double twop52 = 4503599627370496.0; // 0x1.0p52 + static const double twop84 = 19342813113834066795298816.0; // 0x1.0p84 + static const double twop84_plus_twop52 = 19342813118337666422669312.0; // 0x1.00000001p84 + + union { uint64_t x; double d; } high = { .d = twop84 }; + union { uint64_t x; double d; } low = { .d = twop52 }; + + high.x |= a >> 32; + low.x |= a & UINT64_C(0x00000000ffffffff); + + const double result = (high.d - twop84_plus_twop52) + low.d; + return result; +} + +#else +/* Support for systems that don't have hardware floating-point; there are no flags to + * set, and we don't want to code-gen to an unknown soft-float implementation. + */ + +COMPILER_RT_ABI double +__floatundidf(du_int a) +{ + if (a == 0) + return 0.0; + const unsigned N = sizeof(du_int) * CHAR_BIT; + int sd = N - __builtin_clzll(a); /* number of significant digits */ + int e = sd - 1; /* exponent */ + if (sd > DBL_MANT_DIG) + { + /* start: 0000000000000000000001xxxxxxxxxxxxxxxxxxxxxxPQxxxxxxxxxxxxxxxxxx + * finish: 000000000000000000000000000000000000001xxxxxxxxxxxxxxxxxxxxxxPQR + * 12345678901234567890123456 + * 1 = msb 1 bit + * P = bit DBL_MANT_DIG-1 bits to the right of 1 + * Q = bit DBL_MANT_DIG bits to the right of 1 + * R = "or" of all bits to the right of Q + */ + switch (sd) + { + case DBL_MANT_DIG + 1: + a <<= 1; + break; + case DBL_MANT_DIG + 2: + break; + default: + a = (a >> (sd - (DBL_MANT_DIG+2))) | + ((a & ((du_int)(-1) >> ((N + DBL_MANT_DIG+2) - sd))) != 0); + }; + /* finish: */ + a |= (a & 4) != 0; /* Or P into R */ + ++a; /* round - this step may add a significant bit */ + a >>= 2; /* dump Q and R */ + /* a is now rounded to DBL_MANT_DIG or DBL_MANT_DIG+1 bits */ + if (a & ((du_int)1 << DBL_MANT_DIG)) + { + a >>= 1; + ++e; + } + /* a is now rounded to DBL_MANT_DIG bits */ + } + else + { + a <<= (DBL_MANT_DIG - sd); + /* a is now rounded to DBL_MANT_DIG bits */ + } + double_bits fb; + fb.u.s.high = ((e + 1023) << 20) | /* exponent */ + ((su_int)(a >> 32) & 0x000FFFFF); /* mantissa-high */ + fb.u.s.low = (su_int)a; /* mantissa-low */ + return fb.f; +} +#endif + +#if defined(__ARM_EABI__) +#if defined(COMPILER_RT_ARMHF_TARGET) +AEABI_RTABI double __aeabi_ul2d(du_int a) { + return __floatundidf(a); +} +#else +AEABI_RTABI double __aeabi_ul2d(du_int a) COMPILER_RT_ALIAS(__floatundidf); +#endif +#endif diff --git a/libs/compiler-rt/floatundisf.c b/libs/compiler-rt/floatundisf.c new file mode 100644 index 00000000000..844786ea777 --- /dev/null +++ b/libs/compiler-rt/floatundisf.c @@ -0,0 +1,85 @@ +/*===-- floatundisf.c - Implement __floatundisf ---------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file implements __floatundisf for the compiler_rt library. + * + *===----------------------------------------------------------------------=== + */ + +/* Returns: convert a to a float, rounding toward even. */ + +/* Assumption: float is a IEEE 32 bit floating point type + * du_int is a 64 bit integral type + */ + +/* seee eeee emmm mmmm mmmm mmmm mmmm mmmm */ + +#include "int_lib.h" + +COMPILER_RT_ABI float +__floatundisf(du_int a) +{ + if (a == 0) + return 0.0F; + const unsigned N = sizeof(du_int) * CHAR_BIT; + int sd = N - __builtin_clzll(a); /* number of significant digits */ + int e = sd - 1; /* 8 exponent */ + if (sd > FLT_MANT_DIG) + { + /* start: 0000000000000000000001xxxxxxxxxxxxxxxxxxxxxxPQxxxxxxxxxxxxxxxxxx + * finish: 000000000000000000000000000000000000001xxxxxxxxxxxxxxxxxxxxxxPQR + * 12345678901234567890123456 + * 1 = msb 1 bit + * P = bit FLT_MANT_DIG-1 bits to the right of 1 + * Q = bit FLT_MANT_DIG bits to the right of 1 + * R = "or" of all bits to the right of Q + */ + switch (sd) + { + case FLT_MANT_DIG + 1: + a <<= 1; + break; + case FLT_MANT_DIG + 2: + break; + default: + a = (a >> (sd - (FLT_MANT_DIG+2))) | + ((a & ((du_int)(-1) >> ((N + FLT_MANT_DIG+2) - sd))) != 0); + }; + /* finish: */ + a |= (a & 4) != 0; /* Or P into R */ + ++a; /* round - this step may add a significant bit */ + a >>= 2; /* dump Q and R */ + /* a is now rounded to FLT_MANT_DIG or FLT_MANT_DIG+1 bits */ + if (a & ((du_int)1 << FLT_MANT_DIG)) + { + a >>= 1; + ++e; + } + /* a is now rounded to FLT_MANT_DIG bits */ + } + else + { + a <<= (FLT_MANT_DIG - sd); + /* a is now rounded to FLT_MANT_DIG bits */ + } + float_bits fb; + fb.u = ((e + 127) << 23) | /* exponent */ + ((su_int)a & 0x007FFFFF); /* mantissa */ + return fb.f; +} + +#if defined(__ARM_EABI__) +#if defined(COMPILER_RT_ARMHF_TARGET) +AEABI_RTABI float __aeabi_ul2f(du_int a) { + return __floatundisf(a); +} +#else +AEABI_RTABI float __aeabi_ul2f(du_int a) COMPILER_RT_ALIAS(__floatundisf); +#endif +#endif diff --git a/libs/compiler-rt/fp_fixint_impl.inc b/libs/compiler-rt/fp_fixint_impl.inc new file mode 100644 index 00000000000..da70d4d3930 --- /dev/null +++ b/libs/compiler-rt/fp_fixint_impl.inc @@ -0,0 +1,41 @@ +//===-- lib/fixdfsi.c - Double-precision -> integer conversion ----*- C -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is dual licensed under the MIT and the University of Illinois Open +// Source Licenses. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements float to integer conversion for the +// compiler-rt library. +// +//===----------------------------------------------------------------------===// + +#include "fp_lib.h" + +static __inline fixint_t __fixint(fp_t a) { + const fixint_t fixint_max = (fixint_t)((~(fixuint_t)0) / 2); + const fixint_t fixint_min = -fixint_max - 1; + // Break a into sign, exponent, significand + const rep_t aRep = toRep(a); + const rep_t aAbs = aRep & absMask; + const fixint_t sign = aRep & signBit ? -1 : 1; + const int exponent = (aAbs >> significandBits) - exponentBias; + const rep_t significand = (aAbs & significandMask) | implicitBit; + + // If exponent is negative, the result is zero. + if (exponent < 0) + return 0; + + // If the value is too large for the integer type, saturate. + if ((unsigned)exponent >= sizeof(fixint_t) * CHAR_BIT) + return sign == 1 ? fixint_max : fixint_min; + + // If 0 <= exponent < significandBits, right shift to get the result. + // Otherwise, shift left. + if (exponent < significandBits) + return sign * (significand >> (significandBits - exponent)); + else + return sign * ((fixint_t)significand << (exponent - significandBits)); +} diff --git a/libs/compiler-rt/fp_fixuint_impl.inc b/libs/compiler-rt/fp_fixuint_impl.inc new file mode 100644 index 00000000000..d68ccf27a79 --- /dev/null +++ b/libs/compiler-rt/fp_fixuint_impl.inc @@ -0,0 +1,39 @@ +//===-- lib/fixdfsi.c - Double-precision -> integer conversion ----*- C -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is dual licensed under the MIT and the University of Illinois Open +// Source Licenses. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file implements float to unsigned integer conversion for the +// compiler-rt library. +// +//===----------------------------------------------------------------------===// + +#include "fp_lib.h" + +static __inline fixuint_t __fixuint(fp_t a) { + // Break a into sign, exponent, significand + const rep_t aRep = toRep(a); + const rep_t aAbs = aRep & absMask; + const int sign = aRep & signBit ? -1 : 1; + const int exponent = (aAbs >> significandBits) - exponentBias; + const rep_t significand = (aAbs & significandMask) | implicitBit; + + // If either the value or the exponent is negative, the result is zero. + if (sign == -1 || exponent < 0) + return 0; + + // If the value is too large for the integer type, saturate. + if ((unsigned)exponent >= sizeof(fixuint_t) * CHAR_BIT) + return ~(fixuint_t)0; + + // If 0 <= exponent < significandBits, right shift to get the result. + // Otherwise, shift left. + if (exponent < significandBits) + return significand >> (significandBits - exponent); + else + return (fixuint_t)significand << (exponent - significandBits); +} diff --git a/libs/compiler-rt/fp_lib.h b/libs/compiler-rt/fp_lib.h new file mode 100644 index 00000000000..a0e19ab6a8f --- /dev/null +++ b/libs/compiler-rt/fp_lib.h @@ -0,0 +1,327 @@ +//===-- lib/fp_lib.h - Floating-point utilities -------------------*- C -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is dual licensed under the MIT and the University of Illinois Open +// Source Licenses. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file is a configuration header for soft-float routines in compiler-rt. +// This file does not provide any part of the compiler-rt interface, but defines +// many useful constants and utility routines that are used in the +// implementation of the soft-float routines in compiler-rt. +// +// Assumes that float, double and long double correspond to the IEEE-754 +// binary32, binary64 and binary 128 types, respectively, and that integer +// endianness matches floating point endianness on the target platform. +// +//===----------------------------------------------------------------------===// + +#ifndef FP_LIB_HEADER +#define FP_LIB_HEADER + +#include <stdint.h> +#include <stdbool.h> +#include <limits.h> +#include "int_lib.h" +#include "int_math.h" + +// x86_64 FreeBSD prior v9.3 define fixed-width types incorrectly in +// 32-bit mode. +#if defined(__FreeBSD__) && defined(__i386__) +# include <sys/param.h> +# if __FreeBSD_version < 903000 // v9.3 +# define uint64_t unsigned long long +# define int64_t long long +# undef UINT64_C +# define UINT64_C(c) (c ## ULL) +# endif +#endif + +#if defined SINGLE_PRECISION + +typedef uint32_t rep_t; +typedef int32_t srep_t; +typedef float fp_t; +#define REP_C UINT32_C +#define significandBits 23 + +static __inline int rep_clz(rep_t a) { + return __builtin_clz(a); +} + +// 32x32 --> 64 bit multiply +static __inline void wideMultiply(rep_t a, rep_t b, rep_t *hi, rep_t *lo) { + const uint64_t product = (uint64_t)a*b; + *hi = product >> 32; + *lo = product; +} +COMPILER_RT_ABI fp_t __addsf3(fp_t a, fp_t b); + +#elif defined DOUBLE_PRECISION + +typedef uint64_t rep_t; +typedef int64_t srep_t; +typedef double fp_t; +#define REP_C UINT64_C +#define significandBits 52 + +static __inline int rep_clz(rep_t a) { +#if defined __LP64__ + return __builtin_clzl(a); +#else + if (a & REP_C(0xffffffff00000000)) + return __builtin_clz(a >> 32); + else + return 32 + __builtin_clz(a & REP_C(0xffffffff)); +#endif +} + +#define loWord(a) (a & 0xffffffffU) +#define hiWord(a) (a >> 32) + +// 64x64 -> 128 wide multiply for platforms that don't have such an operation; +// many 64-bit platforms have this operation, but they tend to have hardware +// floating-point, so we don't bother with a special case for them here. +static __inline void wideMultiply(rep_t a, rep_t b, rep_t *hi, rep_t *lo) { + // Each of the component 32x32 -> 64 products + const uint64_t plolo = loWord(a) * loWord(b); + const uint64_t plohi = loWord(a) * hiWord(b); + const uint64_t philo = hiWord(a) * loWord(b); + const uint64_t phihi = hiWord(a) * hiWord(b); + // Sum terms that contribute to lo in a way that allows us to get the carry + const uint64_t r0 = loWord(plolo); + const uint64_t r1 = hiWord(plolo) + loWord(plohi) + loWord(philo); + *lo = r0 + (r1 << 32); + // Sum terms contributing to hi with the carry from lo + *hi = hiWord(plohi) + hiWord(philo) + hiWord(r1) + phihi; +} +#undef loWord +#undef hiWord + +COMPILER_RT_ABI fp_t __adddf3(fp_t a, fp_t b); + +#elif defined QUAD_PRECISION +#if __LDBL_MANT_DIG__ == 113 +#define CRT_LDBL_128BIT +typedef __uint128_t rep_t; +typedef __int128_t srep_t; +typedef long double fp_t; +#define REP_C (__uint128_t) +// Note: Since there is no explicit way to tell compiler the constant is a +// 128-bit integer, we let the constant be casted to 128-bit integer +#define significandBits 112 + +static __inline int rep_clz(rep_t a) { + const union + { + __uint128_t ll; +#if _YUGA_BIG_ENDIAN + struct { uint64_t high, low; } s; +#else + struct { uint64_t low, high; } s; +#endif + } uu = { .ll = a }; + + uint64_t word; + uint64_t add; + + if (uu.s.high){ + word = uu.s.high; + add = 0; + } + else{ + word = uu.s.low; + add = 64; + } + return __builtin_clzll(word) + add; +} + +#define Word_LoMask UINT64_C(0x00000000ffffffff) +#define Word_HiMask UINT64_C(0xffffffff00000000) +#define Word_FullMask UINT64_C(0xffffffffffffffff) +#define Word_1(a) (uint64_t)((a >> 96) & Word_LoMask) +#define Word_2(a) (uint64_t)((a >> 64) & Word_LoMask) +#define Word_3(a) (uint64_t)((a >> 32) & Word_LoMask) +#define Word_4(a) (uint64_t)(a & Word_LoMask) + +// 128x128 -> 256 wide multiply for platforms that don't have such an operation; +// many 64-bit platforms have this operation, but they tend to have hardware +// floating-point, so we don't bother with a special case for them here. +static __inline void wideMultiply(rep_t a, rep_t b, rep_t *hi, rep_t *lo) { + + const uint64_t product11 = Word_1(a) * Word_1(b); + const uint64_t product12 = Word_1(a) * Word_2(b); + const uint64_t product13 = Word_1(a) * Word_3(b); + const uint64_t product14 = Word_1(a) * Word_4(b); + const uint64_t product21 = Word_2(a) * Word_1(b); + const uint64_t product22 = Word_2(a) * Word_2(b); + const uint64_t product23 = Word_2(a) * Word_3(b); + const uint64_t product24 = Word_2(a) * Word_4(b); + const uint64_t product31 = Word_3(a) * Word_1(b); + const uint64_t product32 = Word_3(a) * Word_2(b); + const uint64_t product33 = Word_3(a) * Word_3(b); + const uint64_t product34 = Word_3(a) * Word_4(b); + const uint64_t product41 = Word_4(a) * Word_1(b); + const uint64_t product42 = Word_4(a) * Word_2(b); + const uint64_t product43 = Word_4(a) * Word_3(b); + const uint64_t product44 = Word_4(a) * Word_4(b); + + const __uint128_t sum0 = (__uint128_t)product44; + const __uint128_t sum1 = (__uint128_t)product34 + + (__uint128_t)product43; + const __uint128_t sum2 = (__uint128_t)product24 + + (__uint128_t)product33 + + (__uint128_t)product42; + const __uint128_t sum3 = (__uint128_t)product14 + + (__uint128_t)product23 + + (__uint128_t)product32 + + (__uint128_t)product41; + const __uint128_t sum4 = (__uint128_t)product13 + + (__uint128_t)product22 + + (__uint128_t)product31; + const __uint128_t sum5 = (__uint128_t)product12 + + (__uint128_t)product21; + const __uint128_t sum6 = (__uint128_t)product11; + + const __uint128_t r0 = (sum0 & Word_FullMask) + + ((sum1 & Word_LoMask) << 32); + const __uint128_t r1 = (sum0 >> 64) + + ((sum1 >> 32) & Word_FullMask) + + (sum2 & Word_FullMask) + + ((sum3 << 32) & Word_HiMask); + + *lo = r0 + (r1 << 64); + *hi = (r1 >> 64) + + (sum1 >> 96) + + (sum2 >> 64) + + (sum3 >> 32) + + sum4 + + (sum5 << 32) + + (sum6 << 64); +} +#undef Word_1 +#undef Word_2 +#undef Word_3 +#undef Word_4 +#undef Word_HiMask +#undef Word_LoMask +#undef Word_FullMask +#endif // __LDBL_MANT_DIG__ == 113 +#else +#error SINGLE_PRECISION, DOUBLE_PRECISION or QUAD_PRECISION must be defined. +#endif + +#if defined(SINGLE_PRECISION) || defined(DOUBLE_PRECISION) || defined(CRT_LDBL_128BIT) +#define typeWidth (sizeof(rep_t)*CHAR_BIT) +#define exponentBits (typeWidth - significandBits - 1) +#define maxExponent ((1 << exponentBits) - 1) +#define exponentBias (maxExponent >> 1) + +#define implicitBit (REP_C(1) << significandBits) +#define significandMask (implicitBit - 1U) +#define signBit (REP_C(1) << (significandBits + exponentBits)) +#define absMask (signBit - 1U) +#define exponentMask (absMask ^ significandMask) +#define oneRep ((rep_t)exponentBias << significandBits) +#define infRep exponentMask +#define quietBit (implicitBit >> 1) +#define qnanRep (exponentMask | quietBit) + +static __inline rep_t toRep(fp_t x) { + const union { fp_t f; rep_t i; } rep = {.f = x}; + return rep.i; +} + +static __inline fp_t fromRep(rep_t x) { + const union { fp_t f; rep_t i; } rep = {.i = x}; + return rep.f; +} + +static __inline int normalize(rep_t *significand) { + const int shift = rep_clz(*significand) - rep_clz(implicitBit); + *significand <<= shift; + return 1 - shift; +} + +static __inline void wideLeftShift(rep_t *hi, rep_t *lo, int count) { + *hi = *hi << count | *lo >> (typeWidth - count); + *lo = *lo << count; +} + +static __inline void wideRightShiftWithSticky(rep_t *hi, rep_t *lo, unsigned int count) { + if (count < typeWidth) { + const bool sticky = *lo << (typeWidth - count); + *lo = *hi << (typeWidth - count) | *lo >> count | sticky; + *hi = *hi >> count; + } + else if (count < 2*typeWidth) { + const bool sticky = *hi << (2*typeWidth - count) | *lo; + *lo = *hi >> (count - typeWidth) | sticky; + *hi = 0; + } else { + const bool sticky = *hi | *lo; + *lo = sticky; + *hi = 0; + } +} + +// Implements logb methods (logb, logbf, logbl) for IEEE-754. This avoids +// pulling in a libm dependency from compiler-rt, but is not meant to replace +// it (i.e. code calling logb() should get the one from libm, not this), hence +// the __compiler_rt prefix. +static __inline fp_t __compiler_rt_logbX(fp_t x) { + rep_t rep = toRep(x); + int exp = (rep & exponentMask) >> significandBits; + + // Abnormal cases: + // 1) +/- inf returns +inf; NaN returns NaN + // 2) 0.0 returns -inf + if (exp == maxExponent) { + if (((rep & signBit) == 0) || (x != x)) { + return x; // NaN or +inf: return x + } else { + return -x; // -inf: return -x + } + } else if (x == 0.0) { + // 0.0: return -inf + return fromRep(infRep | signBit); + } + + if (exp != 0) { + // Normal number + return exp - exponentBias; // Unbias exponent + } else { + // Subnormal number; normalize and repeat + rep &= absMask; + const int shift = 1 - normalize(&rep); + exp = (rep & exponentMask) >> significandBits; + return exp - exponentBias - shift; // Unbias exponent + } +} +#endif + +#if defined(SINGLE_PRECISION) +static __inline fp_t __compiler_rt_logbf(fp_t x) { + return __compiler_rt_logbX(x); +} +#elif defined(DOUBLE_PRECISION) +static __inline fp_t __compiler_rt_logb(fp_t x) { + return __compiler_rt_logbX(x); +} +#elif defined(QUAD_PRECISION) + #if defined(CRT_LDBL_128BIT) +static __inline fp_t __compiler_rt_logbl(fp_t x) { + return __compiler_rt_logbX(x); +} + #else +// The generic implementation only works for ieee754 floating point. For other +// floating point types, continue to rely on the libm implementation for now. +static __inline long double __compiler_rt_logbl(long double x) { + return crt_logbl(x); +} + #endif +#endif + +#endif // FP_LIB_HEADER diff --git a/libs/compiler-rt/int_endianness.h b/libs/compiler-rt/int_endianness.h new file mode 100644 index 00000000000..e2586c56bac --- /dev/null +++ b/libs/compiler-rt/int_endianness.h @@ -0,0 +1,116 @@ +/* ===-- int_endianness.h - configuration header for compiler-rt ------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file is a configuration header for compiler-rt. + * This file is not part of the interface of this library. + * + * ===----------------------------------------------------------------------=== + */ + +#ifndef INT_ENDIANNESS_H +#define INT_ENDIANNESS_H + +#if defined(__BYTE_ORDER__) && defined(__ORDER_BIG_ENDIAN__) && \ + defined(__ORDER_LITTLE_ENDIAN__) + +/* Clang and GCC provide built-in endianness definitions. */ +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ +#define _YUGA_LITTLE_ENDIAN 0 +#define _YUGA_BIG_ENDIAN 1 +#elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ +#define _YUGA_LITTLE_ENDIAN 1 +#define _YUGA_BIG_ENDIAN 0 +#endif /* __BYTE_ORDER__ */ + +#else /* Compilers other than Clang or GCC. */ + +#if defined(__SVR4) && defined(__sun) +#include <sys/byteorder.h> + +#if defined(_BIG_ENDIAN) +#define _YUGA_LITTLE_ENDIAN 0 +#define _YUGA_BIG_ENDIAN 1 +#elif defined(_LITTLE_ENDIAN) +#define _YUGA_LITTLE_ENDIAN 1 +#define _YUGA_BIG_ENDIAN 0 +#else /* !_LITTLE_ENDIAN */ +#error "unknown endianness" +#endif /* !_LITTLE_ENDIAN */ + +#endif /* Solaris and AuroraUX. */ + +/* .. */ + +#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__DragonFly__) || \ + defined(__minix) +#include <sys/endian.h> + +#if _BYTE_ORDER == _BIG_ENDIAN +#define _YUGA_LITTLE_ENDIAN 0 +#define _YUGA_BIG_ENDIAN 1 +#elif _BYTE_ORDER == _LITTLE_ENDIAN +#define _YUGA_LITTLE_ENDIAN 1 +#define _YUGA_BIG_ENDIAN 0 +#endif /* _BYTE_ORDER */ + +#endif /* *BSD */ + +#if defined(__OpenBSD__) +#include <machine/endian.h> + +#if _BYTE_ORDER == _BIG_ENDIAN +#define _YUGA_LITTLE_ENDIAN 0 +#define _YUGA_BIG_ENDIAN 1 +#elif _BYTE_ORDER == _LITTLE_ENDIAN +#define _YUGA_LITTLE_ENDIAN 1 +#define _YUGA_BIG_ENDIAN 0 +#endif /* _BYTE_ORDER */ + +#endif /* OpenBSD */ + +/* .. */ + +/* Mac OSX has __BIG_ENDIAN__ or __LITTLE_ENDIAN__ automatically set by the + * compiler (at least with GCC) */ +#if defined(__APPLE__) || defined(__ellcc__ ) + +#ifdef __BIG_ENDIAN__ +#if __BIG_ENDIAN__ +#define _YUGA_LITTLE_ENDIAN 0 +#define _YUGA_BIG_ENDIAN 1 +#endif +#endif /* __BIG_ENDIAN__ */ + +#ifdef __LITTLE_ENDIAN__ +#if __LITTLE_ENDIAN__ +#define _YUGA_LITTLE_ENDIAN 1 +#define _YUGA_BIG_ENDIAN 0 +#endif +#endif /* __LITTLE_ENDIAN__ */ + +#endif /* Mac OSX */ + +/* .. */ + +#if defined(_WIN32) + +#define _YUGA_LITTLE_ENDIAN 1 +#define _YUGA_BIG_ENDIAN 0 + +#endif /* Windows */ + +#endif /* Clang or GCC. */ + +/* . */ + +#if !defined(_YUGA_LITTLE_ENDIAN) || !defined(_YUGA_BIG_ENDIAN) +#error Unable to determine endian +#endif /* Check we found an endianness correctly. */ + +#endif /* INT_ENDIANNESS_H */ diff --git a/libs/compiler-rt/int_lib.h b/libs/compiler-rt/int_lib.h new file mode 100644 index 00000000000..fe8a3bdedc0 --- /dev/null +++ b/libs/compiler-rt/int_lib.h @@ -0,0 +1,134 @@ +/* ===-- int_lib.h - configuration header for compiler-rt -----------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file is a configuration header for compiler-rt. + * This file is not part of the interface of this library. + * + * ===----------------------------------------------------------------------=== + */ + +#ifndef INT_LIB_H +#define INT_LIB_H + +/* Assumption: Signed integral is 2's complement. */ +/* Assumption: Right shift of signed negative is arithmetic shift. */ +/* Assumption: Endianness is little or big (not mixed). */ + +#if defined(__ELF__) +#define FNALIAS(alias_name, original_name) \ + void alias_name() __attribute__((__alias__(#original_name))) +#define COMPILER_RT_ALIAS(aliasee) __attribute__((__alias__(#aliasee))) +#else +#define FNALIAS(alias, name) _Pragma("GCC error("alias unsupported on this file format")") +#define COMPILER_RT_ALIAS(aliasee) _Pragma("GCC error("alias unsupported on this file format")") +#endif + +/* ABI macro definitions */ + +#if __ARM_EABI__ +# ifdef COMPILER_RT_ARMHF_TARGET +# define COMPILER_RT_ABI +# else +# define COMPILER_RT_ABI __attribute__((__pcs__("aapcs"))) +# endif +#else +# define COMPILER_RT_ABI +#endif + +#define AEABI_RTABI __attribute__((__pcs__("aapcs"))) + +#if defined(_MSC_VER) && !defined(__clang__) +#define ALWAYS_INLINE __forceinline +#define NOINLINE __declspec(noinline) +#define NORETURN __declspec(noreturn) +#define UNUSED +#else +#define ALWAYS_INLINE __attribute__((always_inline)) +#define NOINLINE __attribute__((noinline)) +#define NORETURN __attribute__((noreturn)) +#define UNUSED __attribute__((unused)) +#endif + +#if defined(__NetBSD__) && (defined(_KERNEL) || defined(_STANDALONE)) +/* + * Kernel and boot environment can't use normal headers, + * so use the equivalent system headers. + */ +# include <machine/limits.h> +# include <sys/stdint.h> +# include <sys/types.h> +#else +/* Include the standard compiler builtin headers we use functionality from. */ +# include <limits.h> +# include <stdint.h> +# include <stdbool.h> +# include <float.h> +#endif + +/* Include the commonly used internal type definitions. */ +#include "int_types.h" + +/* Include internal utility function declarations. */ +#include "int_util.h" + +COMPILER_RT_ABI si_int __paritysi2(si_int a); +COMPILER_RT_ABI si_int __paritydi2(di_int a); + +COMPILER_RT_ABI di_int __divdi3(di_int a, di_int b); +COMPILER_RT_ABI si_int __divsi3(si_int a, si_int b); +COMPILER_RT_ABI su_int __udivsi3(su_int n, su_int d); + +COMPILER_RT_ABI su_int __udivmodsi4(su_int a, su_int b, su_int* rem); +COMPILER_RT_ABI du_int __udivmoddi4(du_int a, du_int b, du_int* rem); +#ifdef CRT_HAS_128BIT +COMPILER_RT_ABI si_int __clzti2(ti_int a); +COMPILER_RT_ABI tu_int __udivmodti4(tu_int a, tu_int b, tu_int* rem); +#endif + +/* Definitions for builtins unavailable on MSVC */ +#if defined(_MSC_VER) && !defined(__clang__) +#include <intrin.h> + +uint32_t __inline __builtin_ctz(uint32_t value) { + unsigned long trailing_zero = 0; + if (_BitScanForward(&trailing_zero, value)) + return trailing_zero; + return 32; +} + +uint32_t __inline __builtin_clz(uint32_t value) { + unsigned long leading_zero = 0; + if (_BitScanReverse(&leading_zero, value)) + return 31 - leading_zero; + return 32; +} + +#if defined(_M_ARM) || defined(_M_X64) +uint32_t __inline __builtin_clzll(uint64_t value) { + unsigned long leading_zero = 0; + if (_BitScanReverse64(&leading_zero, value)) + return 63 - leading_zero; + return 64; +} +#else +uint32_t __inline __builtin_clzll(uint64_t value) { + if (value == 0) + return 64; + uint32_t msh = (uint32_t)(value >> 32); + uint32_t lsh = (uint32_t)(value & 0xFFFFFFFF); + if (msh != 0) + return __builtin_clz(msh); + return 32 + __builtin_clz(lsh); +} +#endif + +#define __builtin_clzl __builtin_clzll +#endif /* defined(_MSC_VER) && !defined(__clang__) */ + +#endif /* INT_LIB_H */ diff --git a/libs/compiler-rt/int_math.h b/libs/compiler-rt/int_math.h new file mode 100644 index 00000000000..aa3d0721a8a --- /dev/null +++ b/libs/compiler-rt/int_math.h @@ -0,0 +1,110 @@ +/* ===-- int_math.h - internal math inlines ---------------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===-----------------------------------------------------------------------=== + * + * This file is not part of the interface of this library. + * + * This file defines substitutes for the libm functions used in some of the + * compiler-rt implementations, defined in such a way that there is not a direct + * dependency on libm or math.h. Instead, we use the compiler builtin versions + * where available. This reduces our dependencies on the system SDK by foisting + * the responsibility onto the compiler. + * + * ===-----------------------------------------------------------------------=== + */ + +#ifndef INT_MATH_H +#define INT_MATH_H + +#ifndef __has_builtin +# define __has_builtin(x) 0 +#endif + +#if defined(_MSC_VER) && !defined(__clang__) +#include <math.h> +#include <stdlib.h> +#include <ymath.h> +#endif + +#if defined(_MSC_VER) && !defined(__clang__) +#define CRT_INFINITY INFINITY +#else +#define CRT_INFINITY __builtin_huge_valf() +#endif + +#if defined(_MSC_VER) && !defined(__clang__) +#define crt_isfinite(x) _finite((x)) +#define crt_isinf(x) !_finite((x)) +#define crt_isnan(x) _isnan((x)) +#else +/* Define crt_isfinite in terms of the builtin if available, otherwise provide + * an alternate version in terms of our other functions. This supports some + * versions of GCC which didn't have __builtin_isfinite. + */ +#if __has_builtin(__builtin_isfinite) +# define crt_isfinite(x) __builtin_isfinite((x)) +#elif defined(__GNUC__) +# define crt_isfinite(x) \ + __extension__(({ \ + __typeof((x)) x_ = (x); \ + !crt_isinf(x_) && !crt_isnan(x_); \ + })) +#else +# error "Do not know how to check for infinity" +#endif /* __has_builtin(__builtin_isfinite) */ +#define crt_isinf(x) __builtin_isinf((x)) +#define crt_isnan(x) __builtin_isnan((x)) +#endif /* _MSC_VER */ + +#if defined(_MSC_VER) && !defined(__clang__) +#define crt_copysign(x, y) copysign((x), (y)) +#define crt_copysignf(x, y) copysignf((x), (y)) +#define crt_copysignl(x, y) copysignl((x), (y)) +#else +#define crt_copysign(x, y) __builtin_copysign((x), (y)) +#define crt_copysignf(x, y) __builtin_copysignf((x), (y)) +#define crt_copysignl(x, y) __builtin_copysignl((x), (y)) +#endif + +#if defined(_MSC_VER) && !defined(__clang__) +#define crt_fabs(x) fabs((x)) +#define crt_fabsf(x) fabsf((x)) +#define crt_fabsl(x) fabs((x)) +#else +#define crt_fabs(x) __builtin_fabs((x)) +#define crt_fabsf(x) __builtin_fabsf((x)) +#define crt_fabsl(x) __builtin_fabsl((x)) +#endif + +#if defined(_MSC_VER) && !defined(__clang__) +#define crt_fmax(x, y) __max((x), (y)) +#define crt_fmaxf(x, y) __max((x), (y)) +#define crt_fmaxl(x, y) __max((x), (y)) +#else +#define crt_fmax(x, y) __builtin_fmax((x), (y)) +#define crt_fmaxf(x, y) __builtin_fmaxf((x), (y)) +#define crt_fmaxl(x, y) __builtin_fmaxl((x), (y)) +#endif + +#if defined(_MSC_VER) && !defined(__clang__) +#define crt_logbl(x) logbl((x)) +#else +#define crt_logbl(x) __builtin_logbl((x)) +#endif + +#if defined(_MSC_VER) && !defined(__clang__) +#define crt_scalbn(x, y) scalbn((x), (y)) +#define crt_scalbnf(x, y) scalbnf((x), (y)) +#define crt_scalbnl(x, y) scalbnl((x), (y)) +#else +#define crt_scalbn(x, y) __builtin_scalbn((x), (y)) +#define crt_scalbnf(x, y) __builtin_scalbnf((x), (y)) +#define crt_scalbnl(x, y) __builtin_scalbnl((x), (y)) +#endif + +#endif /* INT_MATH_H */ diff --git a/libs/compiler-rt/int_types.h b/libs/compiler-rt/int_types.h new file mode 100644 index 00000000000..9f8da56cb77 --- /dev/null +++ b/libs/compiler-rt/int_types.h @@ -0,0 +1,185 @@ +/* ===-- int_lib.h - configuration header for compiler-rt -----------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file is not part of the interface of this library. + * + * This file defines various standard types, most importantly a number of unions + * used to access parts of larger types. + * + * ===----------------------------------------------------------------------=== + */ + +#ifndef INT_TYPES_H +#define INT_TYPES_H + +#include "int_endianness.h" + +/* si_int is defined in Linux sysroot's asm-generic/siginfo.h */ +#ifdef si_int +#undef si_int +#endif +typedef int si_int; +typedef unsigned su_int; + +typedef long long di_int; +typedef unsigned long long du_int; + +typedef union +{ + di_int all; + struct + { +#if _YUGA_LITTLE_ENDIAN + su_int low; + si_int high; +#else + si_int high; + su_int low; +#endif /* _YUGA_LITTLE_ENDIAN */ + }s; +} dwords; + +typedef union +{ + du_int all; + struct + { +#if _YUGA_LITTLE_ENDIAN + su_int low; + su_int high; +#else + su_int high; + su_int low; +#endif /* _YUGA_LITTLE_ENDIAN */ + }s; +} udwords; + +#if defined(__LP64__) || defined(__wasm__) || defined(__mips64) || \ + defined(__riscv) || defined(_WIN64) +#define CRT_HAS_128BIT +#endif + +/* MSVC doesn't have a working 128bit integer type. Users should really compile + * compiler-rt with clang, but if they happen to be doing a standalone build for + * asan or something else, disable the 128 bit parts so things sort of work. + */ +#if defined(_MSC_VER) && !defined(__clang__) +#undef CRT_HAS_128BIT +#endif + +#ifdef CRT_HAS_128BIT +typedef int ti_int __attribute__ ((mode (TI))); +typedef unsigned tu_int __attribute__ ((mode (TI))); + +typedef union +{ + ti_int all; + struct + { +#if _YUGA_LITTLE_ENDIAN + du_int low; + di_int high; +#else + di_int high; + du_int low; +#endif /* _YUGA_LITTLE_ENDIAN */ + }s; +} twords; + +typedef union +{ + tu_int all; + struct + { +#if _YUGA_LITTLE_ENDIAN + du_int low; + du_int high; +#else + du_int high; + du_int low; +#endif /* _YUGA_LITTLE_ENDIAN */ + }s; +} utwords; + +static __inline ti_int make_ti(di_int h, di_int l) { + twords r; + r.s.high = h; + r.s.low = l; + return r.all; +} + +static __inline tu_int make_tu(du_int h, du_int l) { + utwords r; + r.s.high = h; + r.s.low = l; + return r.all; +} + +#endif /* CRT_HAS_128BIT */ + +typedef union +{ + su_int u; + float f; +} float_bits; + +typedef union +{ + udwords u; + double f; +} double_bits; + +typedef struct +{ +#if _YUGA_LITTLE_ENDIAN + udwords low; + udwords high; +#else + udwords high; + udwords low; +#endif /* _YUGA_LITTLE_ENDIAN */ +} uqwords; + +/* Check if the target supports 80 bit extended precision long doubles. + * Notably, on x86 Windows, MSVC only provides a 64-bit long double, but GCC + * still makes it 80 bits. Clang will match whatever compiler it is trying to + * be compatible with. + */ +#if ((defined(__i386__) || defined(__x86_64__)) && !defined(_MSC_VER)) || \ + defined(__m68k__) || defined(__ia64__) +#define HAS_80_BIT_LONG_DOUBLE 1 +#else +#define HAS_80_BIT_LONG_DOUBLE 0 +#endif + +typedef union +{ + uqwords u; + long double f; +} long_double_bits; + +#if __STDC_VERSION__ >= 199901L +typedef float _Complex Fcomplex; +typedef double _Complex Dcomplex; +typedef long double _Complex Lcomplex; + +#define COMPLEX_REAL(x) __real__(x) +#define COMPLEX_IMAGINARY(x) __imag__(x) +#else +typedef struct { float real, imaginary; } Fcomplex; + +typedef struct { double real, imaginary; } Dcomplex; + +typedef struct { long double real, imaginary; } Lcomplex; + +#define COMPLEX_REAL(x) (x).real +#define COMPLEX_IMAGINARY(x) (x).imaginary +#endif +#endif /* INT_TYPES_H */ + diff --git a/libs/compiler-rt/int_util.h b/libs/compiler-rt/int_util.h new file mode 100644 index 00000000000..c3c87381ad8 --- /dev/null +++ b/libs/compiler-rt/int_util.h @@ -0,0 +1,33 @@ +/* ===-- int_util.h - internal utility functions ----------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===-----------------------------------------------------------------------=== + * + * This file is not part of the interface of this library. + * + * This file defines non-inline utilities which are available for use in the + * library. The function definitions themselves are all contained in int_util.c + * which will always be compiled into any compiler-rt library. + * + * ===-----------------------------------------------------------------------=== + */ + +#ifndef INT_UTIL_H +#define INT_UTIL_H + +/** \brief Trigger a program abort (or panic for kernel code). */ +#define compilerrt_abort() __compilerrt_abort_impl(__FILE__, __LINE__, __func__) + +NORETURN void __compilerrt_abort_impl(const char *file, int line, + const char *function); + +#define COMPILE_TIME_ASSERT(expr) COMPILE_TIME_ASSERT1(expr, __COUNTER__) +#define COMPILE_TIME_ASSERT1(expr, cnt) COMPILE_TIME_ASSERT2(expr, cnt) +#define COMPILE_TIME_ASSERT2(expr, cnt) \ + typedef char ct_assert_##cnt[(expr) ? 1 : -1] UNUSED + +#endif /* INT_UTIL_H */ diff --git a/libs/compiler-rt/mingw_fixfloat.c b/libs/compiler-rt/mingw_fixfloat.c new file mode 100644 index 00000000000..c462e0dbf65 --- /dev/null +++ b/libs/compiler-rt/mingw_fixfloat.c @@ -0,0 +1,36 @@ +/* ===-- mingw_fixfloat.c - Wrap int/float conversions for arm/windows -----=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + */ + +#include "int_lib.h" + +COMPILER_RT_ABI di_int __fixdfdi(double a); +COMPILER_RT_ABI di_int __fixsfdi(float a); +COMPILER_RT_ABI du_int __fixunsdfdi(double a); +COMPILER_RT_ABI du_int __fixunssfdi(float a); +COMPILER_RT_ABI double __floatdidf(di_int a); +COMPILER_RT_ABI float __floatdisf(di_int a); +COMPILER_RT_ABI double __floatundidf(du_int a); +COMPILER_RT_ABI float __floatundisf(du_int a); + +COMPILER_RT_ABI di_int __dtoi64(double a) { return __fixdfdi(a); } + +COMPILER_RT_ABI di_int __stoi64(float a) { return __fixsfdi(a); } + +COMPILER_RT_ABI du_int __dtou64(double a) { return __fixunsdfdi(a); } + +COMPILER_RT_ABI du_int __stou64(float a) { return __fixunssfdi(a); } + +COMPILER_RT_ABI double __i64tod(di_int a) { return __floatdidf(a); } + +COMPILER_RT_ABI float __i64tos(di_int a) { return __floatdisf(a); } + +COMPILER_RT_ABI double __u64tod(du_int a) { return __floatundidf(a); } + +COMPILER_RT_ABI float __u64tos(du_int a) { return __floatundisf(a); } diff --git a/libs/compiler-rt/udivmoddi4.c b/libs/compiler-rt/udivmoddi4.c new file mode 100644 index 00000000000..0c8b4ff4647 --- /dev/null +++ b/libs/compiler-rt/udivmoddi4.c @@ -0,0 +1,231 @@ +/* ===-- udivmoddi4.c - Implement __udivmoddi4 -----------------------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is dual licensed under the MIT and the University of Illinois Open + * Source Licenses. See LICENSE.TXT for details. + * + * ===----------------------------------------------------------------------=== + * + * This file implements __udivmoddi4 for the compiler_rt library. + * + * ===----------------------------------------------------------------------=== + */ + +#include "int_lib.h" + +/* Effects: if rem != 0, *rem = a % b + * Returns: a / b + */ + +/* Translated from Figure 3-40 of The PowerPC Compiler Writer's Guide */ + +COMPILER_RT_ABI du_int +__udivmoddi4(du_int a, du_int b, du_int* rem) +{ + const unsigned n_uword_bits = sizeof(su_int) * CHAR_BIT; + const unsigned n_udword_bits = sizeof(du_int) * CHAR_BIT; + udwords n; + n.all = a; + udwords d; + d.all = b; + udwords q; + udwords r; + unsigned sr; + /* special cases, X is unknown, K != 0 */ + if (n.s.high == 0) + { + if (d.s.high == 0) + { + /* 0 X + * --- + * 0 X + */ + if (rem) + *rem = n.s.low % d.s.low; + return n.s.low / d.s.low; + } + /* 0 X + * --- + * K X + */ + if (rem) + *rem = n.s.low; + return 0; + } + /* n.s.high != 0 */ + if (d.s.low == 0) + { + if (d.s.high == 0) + { + /* K X + * --- + * 0 0 + */ + if (rem) + *rem = n.s.high % d.s.low; + return n.s.high / d.s.low; + } + /* d.s.high != 0 */ + if (n.s.low == 0) + { + /* K 0 + * --- + * K 0 + */ + if (rem) + { + r.s.high = n.s.high % d.s.high; + r.s.low = 0; + *rem = r.all; + } + return n.s.high / d.s.high; + } + /* K K + * --- + * K 0 + */ + if ((d.s.high & (d.s.high - 1)) == 0) /* if d is a power of 2 */ + { + if (rem) + { + r.s.low = n.s.low; + r.s.high = n.s.high & (d.s.high - 1); + *rem = r.all; + } + return n.s.high >> __builtin_ctz(d.s.high); + } + /* K K + * --- + * K 0 + */ + sr = __builtin_clz(d.s.high) - __builtin_clz(n.s.high); + /* 0 <= sr <= n_uword_bits - 2 or sr large */ + if (sr > n_uword_bits - 2) + { + if (rem) + *rem = n.all; + return 0; + } + ++sr; + /* 1 <= sr <= n_uword_bits - 1 */ + /* q.all = n.all << (n_udword_bits - sr); */ + q.s.low = 0; + q.s.high = n.s.low << (n_uword_bits - sr); + /* r.all = n.all >> sr; */ + r.s.high = n.s.high >> sr; + r.s.low = (n.s.high << (n_uword_bits - sr)) | (n.s.low >> sr); + } + else /* d.s.low != 0 */ + { + if (d.s.high == 0) + { + /* K X + * --- + * 0 K + */ + if ((d.s.low & (d.s.low - 1)) == 0) /* if d is a power of 2 */ + { + if (rem) + *rem = n.s.low & (d.s.low - 1); + if (d.s.low == 1) + return n.all; + sr = __builtin_ctz(d.s.low); + q.s.high = n.s.high >> sr; + q.s.low = (n.s.high << (n_uword_bits - sr)) | (n.s.low >> sr); + return q.all; + } + /* K X + * --- + * 0 K + */ + sr = 1 + n_uword_bits + __builtin_clz(d.s.low) - __builtin_clz(n.s.high); + /* 2 <= sr <= n_udword_bits - 1 + * q.all = n.all << (n_udword_bits - sr); + * r.all = n.all >> sr; + */ + if (sr == n_uword_bits) + { + q.s.low = 0; + q.s.high = n.s.low; + r.s.high = 0; + r.s.low = n.s.high; + } + else if (sr < n_uword_bits) // 2 <= sr <= n_uword_bits - 1 + { + q.s.low = 0; + q.s.high = n.s.low << (n_uword_bits - sr); + r.s.high = n.s.high >> sr; + r.s.low = (n.s.high << (n_uword_bits - sr)) | (n.s.low >> sr); + } + else // n_uword_bits + 1 <= sr <= n_udword_bits - 1 + { + q.s.low = n.s.low << (n_udword_bits - sr); + q.s.high = (n.s.high << (n_udword_bits - sr)) | + (n.s.low >> (sr - n_uword_bits)); + r.s.high = 0; + r.s.low = n.s.high >> (sr - n_uword_bits); + } + } + else + { + /* K X + * --- + * K K + */ + sr = __builtin_clz(d.s.high) - __builtin_clz(n.s.high); + /* 0 <= sr <= n_uword_bits - 1 or sr large */ + if (sr > n_uword_bits - 1) + { + if (rem) + *rem = n.all; + return 0; + } + ++sr; + /* 1 <= sr <= n_uword_bits */ + /* q.all = n.all << (n_udword_bits - sr); */ + q.s.low = 0; + if (sr == n_uword_bits) + { + q.s.high = n.s.low; + r.s.high = 0; + r.s.low = n.s.high; + } + else + { + q.s.high = n.s.low << (n_uword_bits - sr); + r.s.high = n.s.high >> sr; + r.s.low = (n.s.high << (n_uword_bits - sr)) | (n.s.low >> sr); + } + } + } + /* Not a special case + * q and r are initialized with: + * q.all = n.all << (n_udword_bits - sr); + * r.all = n.all >> sr; + * 1 <= sr <= n_udword_bits - 1 + */ + su_int carry = 0; + for (; sr > 0; --sr) + { + /* r:q = ((r:q) << 1) | carry */ + r.s.high = (r.s.high << 1) | (r.s.low >> (n_uword_bits - 1)); + r.s.low = (r.s.low << 1) | (q.s.high >> (n_uword_bits - 1)); + q.s.high = (q.s.high << 1) | (q.s.low >> (n_uword_bits - 1)); + q.s.low = (q.s.low << 1) | carry; + /* carry = 0; + * if (r.all >= d.all) + * { + * r.all -= d.all; + * carry = 1; + * } + */ + const di_int s = (di_int)(d.all - r.all - 1) >> (n_udword_bits - 1); + carry = s & 1; + r.all -= d.all & s; + } + q.all = (q.all << 1) | carry; + if (rem) + *rem = r.all; + return q.all; +}
From: Martin Storsjö martin@martin.st
Change __MINGW32__ ifdefs to _WIN32; in compiler-rt builtins, these functions are only needed for mingw targets, as the functions are expected to be provided by MSVC libraries in MSVC builds.
When used in Wine, these files are needed specifically to fulfill that role in MSVC mode builds, so generalize the ifdefs. --- libs/compiler-rt/arm/aeabi_idivmod.S | 4 ++-- libs/compiler-rt/arm/aeabi_ldivmod.S | 4 ++-- libs/compiler-rt/arm/aeabi_uidivmod.S | 4 ++-- libs/compiler-rt/arm/aeabi_uldivmod.S | 4 ++-- 4 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/libs/compiler-rt/arm/aeabi_idivmod.S b/libs/compiler-rt/arm/aeabi_idivmod.S index 9c9c80ab5a7..dadffb3ca87 100644 --- a/libs/compiler-rt/arm/aeabi_idivmod.S +++ b/libs/compiler-rt/arm/aeabi_idivmod.S @@ -15,7 +15,7 @@ // return {quot, rem}; // }
-#if defined(__MINGW32__) +#if defined(_WIN32) #define __aeabi_idivmod __rt_sdiv #endif
@@ -35,7 +35,7 @@ DEFINE_COMPILERRT_FUNCTION(__aeabi_idivmod) push { lr } sub sp, sp, #4 mov r2, sp -#if defined(__MINGW32__) +#if defined(_WIN32) mov r3, r0 mov r0, r1 mov r1, r3 diff --git a/libs/compiler-rt/arm/aeabi_ldivmod.S b/libs/compiler-rt/arm/aeabi_ldivmod.S index 038ae5d723a..ca47e42cf15 100644 --- a/libs/compiler-rt/arm/aeabi_ldivmod.S +++ b/libs/compiler-rt/arm/aeabi_ldivmod.S @@ -16,7 +16,7 @@ // return {quot, rem}; // }
-#if defined(__MINGW32__) +#if defined(_WIN32) #define __aeabi_ldivmod __rt_sdiv64 #endif
@@ -27,7 +27,7 @@ DEFINE_COMPILERRT_FUNCTION(__aeabi_ldivmod) sub sp, sp, #16 add r6, sp, #8 str r6, [sp] -#if defined(__MINGW32__) +#if defined(_WIN32) movs r6, r0 movs r0, r2 movs r2, r6 diff --git a/libs/compiler-rt/arm/aeabi_uidivmod.S b/libs/compiler-rt/arm/aeabi_uidivmod.S index 88a4a6d8bc1..06462c9cf35 100644 --- a/libs/compiler-rt/arm/aeabi_uidivmod.S +++ b/libs/compiler-rt/arm/aeabi_uidivmod.S @@ -16,7 +16,7 @@ // return {quot, rem}; // }
-#if defined(__MINGW32__) +#if defined(_WIN32) #define __aeabi_uidivmod __rt_udiv #endif
@@ -42,7 +42,7 @@ LOCAL_LABEL(case_denom_larger): push { lr } sub sp, sp, #4 mov r2, sp -#if defined(__MINGW32__) +#if defined(_WIN32) mov r3, r0 mov r0, r1 mov r1, r3 diff --git a/libs/compiler-rt/arm/aeabi_uldivmod.S b/libs/compiler-rt/arm/aeabi_uldivmod.S index be343b6bc82..689a54ce1ee 100644 --- a/libs/compiler-rt/arm/aeabi_uldivmod.S +++ b/libs/compiler-rt/arm/aeabi_uldivmod.S @@ -16,7 +16,7 @@ // return {quot, rem}; // }
-#if defined(__MINGW32__) +#if defined(_WIN32) #define __aeabi_uldivmod __rt_udiv64 #endif
@@ -27,7 +27,7 @@ DEFINE_COMPILERRT_FUNCTION(__aeabi_uldivmod) sub sp, sp, #16 add r6, sp, #8 str r6, [sp] -#if defined(__MINGW32__) +#if defined(_WIN32) movs r6, r0 movs r0, r2 movs r2, r6
From: Martin Storsjö martin@martin.st
--- libs/compiler-rt/fp_lib.h | 2 ++ libs/compiler-rt/int_lib.h | 2 ++ 2 files changed, 4 insertions(+)
diff --git a/libs/compiler-rt/fp_lib.h b/libs/compiler-rt/fp_lib.h index a0e19ab6a8f..71c722774fa 100644 --- a/libs/compiler-rt/fp_lib.h +++ b/libs/compiler-rt/fp_lib.h @@ -27,6 +27,8 @@ #include "int_lib.h" #include "int_math.h"
+#pragma GCC diagnostic ignored "-Wdeclaration-after-statement" + // x86_64 FreeBSD prior v9.3 define fixed-width types incorrectly in // 32-bit mode. #if defined(__FreeBSD__) && defined(__i386__) diff --git a/libs/compiler-rt/int_lib.h b/libs/compiler-rt/int_lib.h index fe8a3bdedc0..d5ebcfc8ea8 100644 --- a/libs/compiler-rt/int_lib.h +++ b/libs/compiler-rt/int_lib.h @@ -16,6 +16,8 @@ #ifndef INT_LIB_H #define INT_LIB_H
+#pragma GCC diagnostic ignored "-Wdeclaration-after-statement" + /* Assumption: Signed integral is 2's complement. */ /* Assumption: Right shift of signed negative is arithmetic shift. */ /* Assumption: Endianness is little or big (not mixed). */
From: Martin Storsjö martin@martin.st
The arm assembly only can be built on arm. The functions aren't needed in other configurations, so only include them for the arm/msvc configuration by wrapping them in "#if defined(__arm__) && defined(_MSC_VER)" conditions. --- libs/compiler-rt/arm/aeabi_idivmod.S | 3 ++- libs/compiler-rt/arm/aeabi_ldivmod.S | 3 ++- libs/compiler-rt/arm/aeabi_uidivmod.S | 3 ++- libs/compiler-rt/arm/aeabi_uldivmod.S | 3 ++- libs/compiler-rt/arm/divmodsi4.S | 3 ++- libs/compiler-rt/arm/udivmodsi4.S | 3 ++- libs/compiler-rt/divdi3.c | 2 ++ libs/compiler-rt/divmoddi4.c | 2 ++ libs/compiler-rt/fixdfdi.c | 2 ++ libs/compiler-rt/fixsfdi.c | 2 ++ libs/compiler-rt/fixunsdfdi.c | 2 ++ libs/compiler-rt/fixunssfdi.c | 2 ++ libs/compiler-rt/floatdidf.c | 2 ++ libs/compiler-rt/floatdisf.c | 2 ++ libs/compiler-rt/floatundidf.c | 2 ++ libs/compiler-rt/floatundisf.c | 2 ++ libs/compiler-rt/mingw_fixfloat.c | 2 ++ libs/compiler-rt/udivmoddi4.c | 2 ++ 18 files changed, 36 insertions(+), 6 deletions(-)
diff --git a/libs/compiler-rt/arm/aeabi_idivmod.S b/libs/compiler-rt/arm/aeabi_idivmod.S index dadffb3ca87..fd464c2dfca 100644 --- a/libs/compiler-rt/arm/aeabi_idivmod.S +++ b/libs/compiler-rt/arm/aeabi_idivmod.S @@ -7,6 +7,7 @@ // //===----------------------------------------------------------------------===//
+#if defined(__arm__) && defined(_MSC_VER) #include "../assembly.h"
// struct { int quot, int rem} __aeabi_idivmod(int numerator, int denominator) { @@ -48,4 +49,4 @@ DEFINE_COMPILERRT_FUNCTION(__aeabi_idivmod) END_COMPILERRT_FUNCTION(__aeabi_idivmod)
NO_EXEC_STACK_DIRECTIVE - +#endif diff --git a/libs/compiler-rt/arm/aeabi_ldivmod.S b/libs/compiler-rt/arm/aeabi_ldivmod.S index ca47e42cf15..dbe77f7439f 100644 --- a/libs/compiler-rt/arm/aeabi_ldivmod.S +++ b/libs/compiler-rt/arm/aeabi_ldivmod.S @@ -7,6 +7,7 @@ // //===----------------------------------------------------------------------===//
+#if defined(__arm__) && defined(_MSC_VER) #include "../assembly.h"
// struct { int64_t quot, int64_t rem} @@ -43,4 +44,4 @@ DEFINE_COMPILERRT_FUNCTION(__aeabi_ldivmod) END_COMPILERRT_FUNCTION(__aeabi_ldivmod)
NO_EXEC_STACK_DIRECTIVE - +#endif diff --git a/libs/compiler-rt/arm/aeabi_uidivmod.S b/libs/compiler-rt/arm/aeabi_uidivmod.S index 06462c9cf35..095807c2ae7 100644 --- a/libs/compiler-rt/arm/aeabi_uidivmod.S +++ b/libs/compiler-rt/arm/aeabi_uidivmod.S @@ -7,6 +7,7 @@ // //===----------------------------------------------------------------------===//
+#if defined(__arm__) && defined(_MSC_VER) #include "../assembly.h"
// struct { unsigned quot, unsigned rem} @@ -55,4 +56,4 @@ LOCAL_LABEL(case_denom_larger): END_COMPILERRT_FUNCTION(__aeabi_uidivmod)
NO_EXEC_STACK_DIRECTIVE - +#endif diff --git a/libs/compiler-rt/arm/aeabi_uldivmod.S b/libs/compiler-rt/arm/aeabi_uldivmod.S index 689a54ce1ee..501fa3a112e 100644 --- a/libs/compiler-rt/arm/aeabi_uldivmod.S +++ b/libs/compiler-rt/arm/aeabi_uldivmod.S @@ -7,6 +7,7 @@ // //===----------------------------------------------------------------------===//
+#if defined(__arm__) && defined(_MSC_VER) #include "../assembly.h"
// struct { uint64_t quot, uint64_t rem} @@ -43,4 +44,4 @@ DEFINE_COMPILERRT_FUNCTION(__aeabi_uldivmod) END_COMPILERRT_FUNCTION(__aeabi_uldivmod)
NO_EXEC_STACK_DIRECTIVE - +#endif diff --git a/libs/compiler-rt/arm/divmodsi4.S b/libs/compiler-rt/arm/divmodsi4.S index 8a027b741ef..9aadd91aef9 100644 --- a/libs/compiler-rt/arm/divmodsi4.S +++ b/libs/compiler-rt/arm/divmodsi4.S @@ -13,6 +13,7 @@ * *===----------------------------------------------------------------------===*/
+#if defined(__arm__) && defined(_MSC_VER) #include "../assembly.h"
#define ESTABLISH_FRAME \ @@ -68,4 +69,4 @@ LOCAL_LABEL(divzero): END_COMPILERRT_FUNCTION(__divmodsi4)
NO_EXEC_STACK_DIRECTIVE - +#endif diff --git a/libs/compiler-rt/arm/udivmodsi4.S b/libs/compiler-rt/arm/udivmodsi4.S index ee3950c9b0e..7b8a2566bf3 100644 --- a/libs/compiler-rt/arm/udivmodsi4.S +++ b/libs/compiler-rt/arm/udivmodsi4.S @@ -12,6 +12,7 @@ * *===----------------------------------------------------------------------===*/
+#if defined(__arm__) && defined(_MSC_VER) #include "../assembly.h"
.syntax unified @@ -177,4 +178,4 @@ LOCAL_LABEL(divby0): END_COMPILERRT_FUNCTION(__udivmodsi4)
NO_EXEC_STACK_DIRECTIVE - +#endif diff --git a/libs/compiler-rt/divdi3.c b/libs/compiler-rt/divdi3.c index b8eebcb2046..1ab4d4ae780 100644 --- a/libs/compiler-rt/divdi3.c +++ b/libs/compiler-rt/divdi3.c @@ -12,6 +12,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #include "int_lib.h"
/* Returns: a / b */ @@ -27,3 +28,4 @@ __divdi3(di_int a, di_int b) s_a ^= s_b; /*sign of quotient */ return (__udivmoddi4(a, b, (du_int*)0) ^ s_a) - s_a; /* negate if s_a == -1 */ } +#endif diff --git a/libs/compiler-rt/divmoddi4.c b/libs/compiler-rt/divmoddi4.c index 0d4df67a63e..60433ce9c22 100644 --- a/libs/compiler-rt/divmoddi4.c +++ b/libs/compiler-rt/divmoddi4.c @@ -12,6 +12,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #include "int_lib.h"
/* Returns: a / b, *rem = a % b */ @@ -23,3 +24,4 @@ __divmoddi4(di_int a, di_int b, di_int* rem) *rem = a - (d*b); return d; } +#endif diff --git a/libs/compiler-rt/fixdfdi.c b/libs/compiler-rt/fixdfdi.c index 54e312d3c8f..a13d798823c 100644 --- a/libs/compiler-rt/fixdfdi.c +++ b/libs/compiler-rt/fixdfdi.c @@ -8,6 +8,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #define DOUBLE_PRECISION #include "fp_lib.h"
@@ -53,3 +54,4 @@ AEABI_RTABI di_int __aeabi_d2lz(fp_t a) { AEABI_RTABI di_int __aeabi_d2lz(fp_t a) COMPILER_RT_ALIAS(__fixdfdi); #endif #endif +#endif diff --git a/libs/compiler-rt/fixsfdi.c b/libs/compiler-rt/fixsfdi.c index 32e87c60889..7b85d993076 100644 --- a/libs/compiler-rt/fixsfdi.c +++ b/libs/compiler-rt/fixsfdi.c @@ -8,6 +8,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #define SINGLE_PRECISION #include "fp_lib.h"
@@ -53,3 +54,4 @@ AEABI_RTABI di_int __aeabi_f2lz(fp_t a) { AEABI_RTABI di_int __aeabi_f2lz(fp_t a) COMPILER_RT_ALIAS(__fixsfdi); #endif #endif +#endif diff --git a/libs/compiler-rt/fixunsdfdi.c b/libs/compiler-rt/fixunsdfdi.c index bfe4dbb2565..e4f75dd8507 100644 --- a/libs/compiler-rt/fixunsdfdi.c +++ b/libs/compiler-rt/fixunsdfdi.c @@ -8,6 +8,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #define DOUBLE_PRECISION #include "fp_lib.h"
@@ -50,3 +51,4 @@ AEABI_RTABI du_int __aeabi_d2ulz(fp_t a) { AEABI_RTABI du_int __aeabi_d2ulz(fp_t a) COMPILER_RT_ALIAS(__fixunsdfdi); #endif #endif +#endif diff --git a/libs/compiler-rt/fixunssfdi.c b/libs/compiler-rt/fixunssfdi.c index 080a25bb1e9..97687342bec 100644 --- a/libs/compiler-rt/fixunssfdi.c +++ b/libs/compiler-rt/fixunssfdi.c @@ -8,6 +8,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #define SINGLE_PRECISION #include "fp_lib.h"
@@ -51,3 +52,4 @@ AEABI_RTABI du_int __aeabi_f2ulz(fp_t a) { AEABI_RTABI du_int __aeabi_f2ulz(fp_t a) COMPILER_RT_ALIAS(__fixunssfdi); #endif #endif +#endif diff --git a/libs/compiler-rt/floatdidf.c b/libs/compiler-rt/floatdidf.c index 36b856e078d..4022f72857a 100644 --- a/libs/compiler-rt/floatdidf.c +++ b/libs/compiler-rt/floatdidf.c @@ -12,6 +12,7 @@ *===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #include "int_lib.h"
/* Returns: convert a to a double, rounding toward even. */ @@ -113,3 +114,4 @@ AEABI_RTABI double __aeabi_l2d(di_int a) { AEABI_RTABI double __aeabi_l2d(di_int a) COMPILER_RT_ALIAS(__floatdidf); #endif #endif +#endif diff --git a/libs/compiler-rt/floatdisf.c b/libs/compiler-rt/floatdisf.c index a2f09eb2ed2..e71fcb9e5ad 100644 --- a/libs/compiler-rt/floatdisf.c +++ b/libs/compiler-rt/floatdisf.c @@ -12,6 +12,7 @@ *===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) /* Returns: convert a to a float, rounding toward even.*/
/* Assumption: float is a IEEE 32 bit floating point type @@ -86,3 +87,4 @@ AEABI_RTABI float __aeabi_l2f(di_int a) { AEABI_RTABI float __aeabi_l2f(di_int a) COMPILER_RT_ALIAS(__floatdisf); #endif #endif +#endif diff --git a/libs/compiler-rt/floatundidf.c b/libs/compiler-rt/floatundidf.c index 8bc2a096324..a28c9c39f4c 100644 --- a/libs/compiler-rt/floatundidf.c +++ b/libs/compiler-rt/floatundidf.c @@ -12,6 +12,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) /* Returns: convert a to a double, rounding toward even. */
/* Assumption: double is a IEEE 64 bit floating point type @@ -112,3 +113,4 @@ AEABI_RTABI double __aeabi_ul2d(du_int a) { AEABI_RTABI double __aeabi_ul2d(du_int a) COMPILER_RT_ALIAS(__floatundidf); #endif #endif +#endif diff --git a/libs/compiler-rt/floatundisf.c b/libs/compiler-rt/floatundisf.c index 844786ea777..7d8f7c3d977 100644 --- a/libs/compiler-rt/floatundisf.c +++ b/libs/compiler-rt/floatundisf.c @@ -12,6 +12,7 @@ *===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) /* Returns: convert a to a float, rounding toward even. */
/* Assumption: float is a IEEE 32 bit floating point type @@ -83,3 +84,4 @@ AEABI_RTABI float __aeabi_ul2f(du_int a) { AEABI_RTABI float __aeabi_ul2f(du_int a) COMPILER_RT_ALIAS(__floatundisf); #endif #endif +#endif diff --git a/libs/compiler-rt/mingw_fixfloat.c b/libs/compiler-rt/mingw_fixfloat.c index c462e0dbf65..d94bcbd4262 100644 --- a/libs/compiler-rt/mingw_fixfloat.c +++ b/libs/compiler-rt/mingw_fixfloat.c @@ -8,6 +8,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #include "int_lib.h"
COMPILER_RT_ABI di_int __fixdfdi(double a); @@ -34,3 +35,4 @@ COMPILER_RT_ABI float __i64tos(di_int a) { return __floatdisf(a); } COMPILER_RT_ABI double __u64tod(du_int a) { return __floatundidf(a); }
COMPILER_RT_ABI float __u64tos(du_int a) { return __floatundisf(a); } +#endif diff --git a/libs/compiler-rt/udivmoddi4.c b/libs/compiler-rt/udivmoddi4.c index 0c8b4ff4647..5338287ecf0 100644 --- a/libs/compiler-rt/udivmoddi4.c +++ b/libs/compiler-rt/udivmoddi4.c @@ -12,6 +12,7 @@ * ===----------------------------------------------------------------------=== */
+#if defined(__arm__) && defined(_MSC_VER) #include "int_lib.h"
/* Effects: if rem != 0, *rem = a % b @@ -229,3 +230,4 @@ __udivmoddi4(du_int a, du_int b, du_int* rem) *rem = r.all; return q.all; } +#endif
From: Martin Storsjö martin@martin.st
--- dlls/winecrt0/Makefile.in | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+)
diff --git a/dlls/winecrt0/Makefile.in b/dlls/winecrt0/Makefile.in index 863e5ed4190..cf4a83b33b7 100644 --- a/dlls/winecrt0/Makefile.in +++ b/dlls/winecrt0/Makefile.in @@ -1,11 +1,19 @@ STATICLIB = libwinecrt0.a +PARENTSRC = ../../libs/compiler-rt
SOURCES = \ + arm/aeabi_idivmod.S \ + arm/aeabi_ldivmod.S \ + arm/aeabi_uidivmod.S \ + arm/aeabi_uldivmod.S \ arm64ec.c \ crt_dllmain.c \ crt_fltused.c \ debug.c \ delay_load.c \ + divdi3.c \ + divmoddi4.c \ + arm/divmodsi4.S \ dll_canunload.c \ dll_entry.c \ dll_main.c \ @@ -17,7 +25,18 @@ SOURCES = \ exe_main.c \ exe_wentry.c \ exe_wmain.c \ + fixdfdi.c \ + fixunsdfdi.c \ + floatdidf.c \ + floatundidf.c \ + fixsfdi.c \ + fixunssfdi.c \ + floatdisf.c \ + floatundisf.c \ + mingw_fixfloat.c \ register.c \ setjmp.c \ stub.c \ + udivmoddi4.c \ + arm/udivmodsi4.S \ unix_lib.c
So can Wine actually be compiled with MSVC Clang without requiring proprietary Microsoft libraries/headers? Could that infrastructure be used to compile regular C Windows programs with full MSVC compatibility (and possibly C++ ones if libc++ can be built) using a fully open-source toolchain?
So can Wine actually be compiled with MSVC Clang
Yes. In fact Clang MSVC mode is currently the only supported mode for building Wine ARM64.
without requiring proprietary Microsoft libraries/headers?
Wine does not require Windows SDK libraries or headers for building under any configuration.
The Clang MSVC mode does not specifially require Microsoft proprietary headers or libraries to work.
So can Wine actually be compiled with MSVC Clang without requiring proprietary Microsoft libraries/headers?
Yes, this is being done all the time. When probing for a suitable PE compiler, Wine checks for `<arch>-w64-mingw32-{gcc,clang}` - but it also tries `clang -target <arch>-windows`. In the latter mode, with a `<arch>-windows` target triple, Clang runs in MSVC mode. This works for i686, x86_64 and aarch64 just fine and has been done for a couple of years already, but for arm we run into the issue that we need a handful of extra compiler helper functions (for divisions, and float/int conversions) that Wine didn't provide so far. (And when using a mingw compiler setup, Wine would use the corresponding libgcc or compiler-rt builtins from the mingw toolchain.)
Could that infrastructure be used to compile regular C Windows programs with full MSVC compatibility (and possibly C++ ones if libc++ can be built) using a fully open-source toolchain?
You'd need suitable headers; maybe the Wine headers are usable enough for this case - maybe? (The mingw headers don't quite work in MSVC mode.)
You'd also need a bunch of startup/glue libraries. In the case of MSVC, this is vcruntime and UCRT, where there's a bunch of statically linked routines. Wine has got winecrt0 which I guess supplies some parts of this, enough for Wine's purposes, but I'm not sure if it's enough to replicate a full regular user toolchain with all the details it needs.
As for C++, libc++ in MSVC environments relies on a bunch of lower level C++ ABI stuff from the MS STL/vcruntime. (In the case of mingw environments, it uses libstdc++/libsupc++ or libcxxabi, plus libunwind or libgcc, for those bits.)
In fact Clang MSVC mode is currently the only supported mode for building Wine ARM64.
This is not true, it can be built and works just fine with Clang in either mingw or MSVC mode. If you have llvm-mingw available in path, this is what gets picked by default.
This is not true, it can be built and works just fine with Clang in either mingw or MSVC mode. If you have llvm-mingw available in path, this is what gets picked by default.
MinGW arm32 works indeed. I was talking about ARM64 specifically.
MinGW arm32 works indeed. I was talking about ARM64 specifically.
Yes - ARM64 works just fine with Clang in mingw mode. I test this build mode every day.
Can you point to any specific details to the contrary?
On Mon Jan 27 11:29:21 2025 +0000, Martin Storsjö wrote:
MinGW arm32 works indeed. I was talking about ARM64 specifically.
Yes - ARM64 works just fine with Clang in mingw mode. I test this build mode every day. Can you point to any specific details to the contrary?
I stand corrected. Last time I checked, aarch64-w64-mingw-gcc wasn't a thing. I guess things have changed since then.
On Mon Jan 27 11:36:18 2025 +0000, Jinoh Kang wrote:
I stand corrected. Last time I checked, aarch64-w64-mingw-gcc wasn't a thing. I guess things have changed since then. Edit: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108678 is still NEW. I guess MinGW went its own way using Clang, without GCC counterpart.
Now you are totally mixing things up.
Clang can operate in either mingw mode or MSVC mode. This has been the case since essentially forever. If you invoke Clang as e.g. `clang -target aarch64-w64-mingw32` or `clang -target aarch64-windows-gnu`, this is what you get. With specific mingw toolchains such as llvm-mingw, there are also frontend wrappers like `aarch64-w64-mingw32-clang` which implicitly sets the target triple. The same thing goes for any other architecture as well.
llvm-mingw also provide convenience frontends named `aarch64-w64-mingw32-gcc`, that despite the name also just invoke Clang (in order to ease use with e.g. autotools build systems, so it's enough to specify just `--host=` to use it).
All of these cases work just fine with Wine, and have done this also for as long as Wine has supported building DLLs as native PE modules.
Then separately: For a long time, GCC did not support targeting aarch64 on Windows - but there's a recent effort by a number of people to implement this, so there's an ongoing effort to implement this and upstream it. With current GCC from git, you can build a somewhat working C compiler for aarch64 mingw targets. It's not quite yet ABI compatible with the defacto aarch64 mingw environments (e.g. it uses the incorrect ABI for variable argument functions, and for long doubles) - these are being worked on by the GCC/aarch64-mingw developers though.
I guess MinGW went its own way using Clang, without GCC counterpart.
It's not that "mingw" itself goes any specific way at all. Mingw is just the headers and libraries (mostly import libraries). Since forever, you can choose to assemble a full mingw toolchain with either GCC or Clang. You can use Clang+binutils, or GCC+LLD, or whichever combination you like. For c++ library you can use libstdc++ or libc++, for base library you can use libgcc or compiler-rt. Any toolchain vendor can decide to build almost whichever combination of them that they like.
llvm-mingw happens to be one such toolchain setup, which uses Clang, LLD, compiler-rt, libunwind, libc++abi, libc++, without any of the corresponding GNU counterparts. These toolchains support both i686, x86_64, armv7 and aarch64. (GCC doesn't support targeting armv7 Windows either, only i686 and x86_64, and the in-progress aarch64 target. There was some sort of support for older ARM for Windows CE though.)
It's not that "mingw" itself goes any specific way at all. Mingw is just the headers and libraries (mostly import libraries). Since forever, you can choose to assemble a full mingw toolchain with either GCC or Clang. You can use Clang+binutils, or GCC+LLD, or whichever combination you like. For c++ library you can use libstdc++ or libc++, for base library you can use libgcc or compiler-rt. Any toolchain vendor can decide to build almost whichever combination of them that they like.
FWIW, within msys2, there are also similarly a couple of different environments, on x86_64. The default environment (mingw64) uses GCC, binutils, libgcc and libstdc++, while they also provide a separate environment (clang64) which uses Clang, LLD, compiler-rt and libc++ as default components - similar to llvm-mingw.
In neither case it was "mingw" which decided to go any specific way at all, it's just that different vendors can set up different toolchains.
As for C++, libc++ in MSVC environments relies on a bunch of lower level C++ ABI stuff from the MS STL/vcruntime. `
So is there no way to compile C++ stuff with Clang MSVC without using Microsoft's proprietary C++ runtime?
So is there no way to compile C++ stuff with Clang MSVC without using Microsoft's proprietary C++ runtime?
Not that I'm aware of, no. Feel free to work on reimplementing what's missing though :-)
Microsoft did offer some parts of what libc++ needs under a suitable license, so that libc++ could reduce its reliance on MS STL (parts about the exception classes, IIRC), but that bit hasn't been integrated yet, and I would think that more parts are required to make it entirely standalone.
On Mon Jan 27 12:18:01 2025 +0000, Martin Storsjö wrote:
So is there no way to compile C++ stuff with Clang MSVC without using
Microsoft's proprietary C++ runtime? Not that I'm aware of, no. Feel free to work on reimplementing what's missing though :-) Microsoft did offer some parts of what libc++ needs under a suitable license, so that libc++ could reduce its reliance on MS STL (parts about the exception classes, IIRC), but that bit hasn't been integrated yet, and I would think that more parts are required to make it entirely standalone.
Admittedly my terminology was befuddled, but I believe I wasn't intentionally "mixing things up."
MinGW(-w64) demands constructs like `__MINGW32__`, `__attribute__((ms_struct))`, and the Itanium C++ ABI (for x86_64) from the compiler. Initially only GCC supported them, and then Clang had to follow suit by matching GCC's semantics. LLVM eventually had to gain two separate modes (MinGW and MSVC) just for Windows PE target because MinGW has its own set of "C(++) extensions" incompatible with MSVC (which defines `_MSC_VER` instead, has ms struct semantics by default, and has its own C++ ABI).
On ARM64, the role is reversed: LLVM got to decide first what the aforementioned MinGW constructs mean on Windows ARM. Now, if GCC is to ever support ARM64, it has to be compatible with whatever semantics Clang has already decided along while GCC's support was "lagging behind." IMO it doesn't make sense for MinGW to distinguish between "GCC MinGW mode" and "Clang MinGW mode." The former has to emulate the later.
In that sense, MinGW (while ostensibly being just "headers and libraries") actually introduces a lot of stuff into the toolchain landscape like the choice of C++ ABI (incompatible with MSVC) as well as its own set of macros and extensions. This is what I meant by "MinGW going its own way with Clang, without GCC counterpart."
On Mon Jan 27 13:08:13 2025 +0000, Jinoh Kang wrote:
Admittedly my terminology was befuddled, but I believe I wasn't intentionally "mixing things up." MinGW(-w64) demands constructs like `__MINGW32__`, `__attribute__((ms_struct))`, and the Itanium C++ ABI (for x86_64) from the compiler. Initially only GCC supported them, and then Clang had to follow suit by matching GCC's semantics. LLVM eventually had to gain two separate modes (MinGW and MSVC) just for Windows PE target because MinGW has its own set of "C(++) extensions" incompatible with MSVC (which defines `_MSC_VER` instead, has ms struct semantics by default, and has its own C++ ABI). On ARM64, the role is reversed: LLVM got to decide first what the aforementioned MinGW constructs mean on Windows ARM. Now, if GCC is to ever support ARM64, it has to be compatible with whatever semantics Clang has already decided along while GCC's support was "lagging behind." IMO it doesn't make sense for MinGW to distinguish between "GCC MinGW mode" and "Clang MinGW mode." The former has to emulate the later. In that sense, MinGW (while ostensibly being just "headers and libraries") actually introduces a lot of stuff into the toolchain landscape like the choice of C++ ABI (incompatible with MSVC) as well as its own set of macros and extensions. This is what I meant by "MinGW going its own way with Clang, without GCC counterpart."
Feel free to correct me though; in any case I'm not an active contributor to MSYS2/MinGW and it's very possible that I got details completely wrong. I just wanted to clarify what was arguably initially vague from my side. In any case, thanks for your patience.
(Please excuse me if my replies are late; I'm currently on vacation and my timezone GMT+9 is quite distant from some other Wine contributors.)
On Mon Jan 27 13:11:58 2025 +0000, Jinoh Kang wrote:
Feel free to correct me though; in any case I'm not an active contributor to MSYS2/MinGW and it's very possible that I got details completely wrong. I just wanted to clarify what was arguably initially vague from my side. In any case, thanks for your patience. (Please excuse me if my replies are late; I'm currently on vacation and my timezone GMT+9 is quite distant from some other Wine contributors.)
Thanks for the clarifications. Yes, your understanding sounds mostly correct.
In practice, the vast majority of the "mingw-ism" are quite platform independent, so most of them (like the Itanium C++ ABI etc, prebuilt defines like `__MINGW32__` etc) are just brought over in the exact same form as on x86. So the amount of "LLVM got to decide first what the MinGW constructs mean on Windows ARM" was, in practice, 95% just copying the existing practices on x86.
(Also, as a practical detail, surprisingly little of the actual details of what a mingw toolchain needs to do are encoded in the actual mingw-w64 source; most of it is conventions within GCC/binutils and within LLVM/Clang.)
There are, practically, essentially only 2 architecture specific cases where we in LLVM did take the lead and settle on ABI details that we'd want GCC to follow now when they're catching up:
- The kind of `long double`. On x86, mingw uses 80 bit `long double`, while MSVC uses 64 bit bit ones. As x86 does have hardware for 80 bit floats, this somewhat seems sensible, but it causes an endless source of extra work within the mingw ecosystem (we can't ever use anything from UCRT/msvcrt regarding long doubles). Aarch64 doesn't have any >64 bit floats in hardware. There, MSVC uses 64 bit long doubles, Darwin does the same. Linux uses 128 bit softfloats for `long double` though. To avoid unnecessary divergence, unnecessary extra work, and avoiding relying on slow softfloats, we matched the MSVC ABI and went with 64 bit `long double`s. The in-progress GCC port for this target uses 128 bit long doubles, currently - diverging from the already established ABI. This ABI detail is visible quite a lot within the mingw-w64 code though; a lot of the math functions for aarch64 do assume that `long double` is equal to regular `double`, so currently the GCC port has to patch a lot of that code. They have signaled that they will switch to 64 bit long doubles to match LLVM though.
- The name of the stack probe function. On MSVC, the stack probe function is called `__chkstk`, on all architectures. Within mingw environments on x86, it's called `_alloca` on i386 and `___chkstk_ms` on x86_64, while it retains the same (custom) calling convention as MSVC. For arm and aarch64, we didn't see any reason to deviate from what MSVC does, so we use the same name, `__chkstk`. For the GCC port, they don't provide any stack probe function at all yet, and just expand the stack probing inline in function prologues. (This would break Clang-built code if linked against the GCC port's libgcc though.) This aspect is not visible at all in the mingw-w64 repo itself; it's in practice a contract between the code generation code (in GCC/Clang) and the runtime library (in libgcc or compiler-rt).
-
Aida Jonikienė (＠DodoGTA) -
Jinoh Kang (＠iamahuman)
-
Martin Storsjö
-
Martin Storsjö (＠mstorsjo)