Import compiler-rt r259194. Primary changes are better support for

128bit long double and a number of refinements in the profiling backend.
This commit is contained in:
joerg 2016-02-27 18:59:29 +00:00
parent 17feded366
commit ef84fd3bd8
335 changed files with 7613 additions and 1131 deletions

View File

@ -14,7 +14,7 @@ Full text of the relevant licenses is included below.
University of Illinois/NCSA
Open Source License
Copyright (c) 2009-2014 by the contributors listed in CREDITS.TXT
Copyright (c) 2009-2015 by the contributors listed in CREDITS.TXT
All rights reserved.
@ -55,7 +55,7 @@ SOFTWARE.
==============================================================================
Copyright (c) 2009-2014 by the contributors listed in CREDITS.TXT
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

View File

@ -220,7 +220,9 @@ _Unwind_Reason_Code __gcc_personality_v0(int version, _Unwind_Action actions,
// for use with some implementations of assert() in <assert.h>
void __eprintf(const char* format, const char* assertion_expression,
const char* line, const char* file);
// for systems with emulated thread local storage
void* __emutls_get_address(struct __emutls_control*);
// Power PC specific functions

View File

@ -21,7 +21,7 @@
COMPILER_RT_ABI di_int
__addvdi3(di_int a, di_int b)
{
di_int s = a + b;
di_int s = (du_int) a + (du_int) b;
if (b >= 0)
{
if (s < a)

View File

@ -21,7 +21,7 @@
COMPILER_RT_ABI si_int
__addvsi3(si_int a, si_int b)
{
si_int s = a + b;
si_int s = (su_int) a + (su_int) b;
if (b >= 0)
{
if (s < a)

View File

@ -23,7 +23,7 @@
COMPILER_RT_ABI ti_int
__addvti3(ti_int a, ti_int b)
{
ti_int s = a + b;
ti_int s = (tu_int) a + (tu_int) b;
if (b >= 0)
{
if (s < a)

View File

@ -0,0 +1,96 @@
//===-- aeabi_cdcmp.S - EABI cdcmp* 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"
#if __BYTE_ORDER__ != __ORDER_LITTLE_ENDIAN__
#error big endian support not implemented
#endif
#define APSR_Z (1 << 30)
#define APSR_C (1 << 29)
// void __aeabi_cdcmpeq(double a, double b) {
// if (isnan(a) || isnan(b)) {
// Z = 0; C = 1;
// } else {
// __aeabi_cdcmple(a, b);
// }
// }
.syntax unified
.p2align 2
DEFINE_COMPILERRT_FUNCTION(__aeabi_cdcmpeq)
push {r0-r3, lr}
bl __aeabi_cdcmpeq_check_nan
cmp r0, #1
pop {r0-r3, lr}
// NaN has been ruled out, so __aeabi_cdcmple can't trap
bne __aeabi_cdcmple
msr CPSR_f, #APSR_C
JMP(lr)
END_COMPILERRT_FUNCTION(__aeabi_cdcmpeq)
// void __aeabi_cdcmple(double a, double b) {
// if (__aeabi_dcmplt(a, b)) {
// Z = 0; C = 0;
// } else if (__aeabi_dcmpeq(a, b)) {
// Z = 1; C = 1;
// } else {
// Z = 0; C = 1;
// }
// }
.syntax unified
.p2align 2
DEFINE_COMPILERRT_FUNCTION(__aeabi_cdcmple)
// Per the RTABI, this function must preserve r0-r11.
// Save lr in the same instruction for compactness
push {r0-r3, lr}
bl __aeabi_dcmplt
cmp r0, #1
moveq ip, #0
beq 1f
ldm sp, {r0-r3}
bl __aeabi_dcmpeq
cmp r0, #1
moveq ip, #(APSR_C | APSR_Z)
movne ip, #(APSR_C)
1:
msr CPSR_f, ip
pop {r0-r3}
POP_PC()
END_COMPILERRT_FUNCTION(__aeabi_cdcmple)
// int __aeabi_cdrcmple(double a, double b) {
// return __aeabi_cdcmple(b, a);
// }
.syntax unified
.p2align 2
DEFINE_COMPILERRT_FUNCTION(__aeabi_cdrcmple)
// Swap r0 and r2
mov ip, r0
mov r0, r2
mov r2, ip
// Swap r1 and r3
mov ip, r1
mov r1, r3
mov r3, ip
b __aeabi_cdcmple
END_COMPILERRT_FUNCTION(__aeabi_cdrcmple)

View File

@ -0,0 +1,16 @@
//===-- lib/arm/aeabi_cdcmpeq_helper.c - Helper for cdcmpeq ---------------===//
//
// 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 <stdint.h>
__attribute__((pcs("aapcs")))
__attribute__((visibility("hidden")))
int __aeabi_cdcmpeq_check_nan(double a, double b) {
return __builtin_isnan(a) || __builtin_isnan(b);
}

View File

@ -0,0 +1,91 @@
//===-- aeabi_cfcmp.S - EABI cfcmp* 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"
#if __BYTE_ORDER__ != __ORDER_LITTLE_ENDIAN__
#error big endian support not implemented
#endif
#define APSR_Z (1 << 30)
#define APSR_C (1 << 29)
// void __aeabi_cfcmpeq(float a, float b) {
// if (isnan(a) || isnan(b)) {
// Z = 0; C = 1;
// } else {
// __aeabi_cfcmple(a, b);
// }
// }
.syntax unified
.p2align 2
DEFINE_COMPILERRT_FUNCTION(__aeabi_cfcmpeq)
push {r0-r3, lr}
bl __aeabi_cfcmpeq_check_nan
cmp r0, #1
pop {r0-r3, lr}
// NaN has been ruled out, so __aeabi_cfcmple can't trap
bne __aeabi_cfcmple
msr CPSR_f, #APSR_C
JMP(lr)
END_COMPILERRT_FUNCTION(__aeabi_cfcmpeq)
// void __aeabi_cfcmple(float a, float b) {
// if (__aeabi_fcmplt(a, b)) {
// Z = 0; C = 0;
// } else if (__aeabi_fcmpeq(a, b)) {
// Z = 1; C = 1;
// } else {
// Z = 0; C = 1;
// }
// }
.syntax unified
.p2align 2
DEFINE_COMPILERRT_FUNCTION(__aeabi_cfcmple)
// Per the RTABI, this function must preserve r0-r11.
// Save lr in the same instruction for compactness
push {r0-r3, lr}
bl __aeabi_fcmplt
cmp r0, #1
moveq ip, #0
beq 1f
ldm sp, {r0-r3}
bl __aeabi_fcmpeq
cmp r0, #1
moveq ip, #(APSR_C | APSR_Z)
movne ip, #(APSR_C)
1:
msr CPSR_f, ip
pop {r0-r3}
POP_PC()
END_COMPILERRT_FUNCTION(__aeabi_cfcmple)
// int __aeabi_cfrcmple(float a, float b) {
// return __aeabi_cfcmple(b, a);
// }
.syntax unified
.p2align 2
DEFINE_COMPILERRT_FUNCTION(__aeabi_cfrcmple)
// Swap r0 and r1
mov ip, r0
mov r0, r1
mov r1, ip
b __aeabi_cfcmple
END_COMPILERRT_FUNCTION(__aeabi_cfrcmple)

View File

@ -0,0 +1,16 @@
//===-- lib/arm/aeabi_cfcmpeq_helper.c - Helper for cdcmpeq ---------------===//
//
// 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 <stdint.h>
__attribute__((pcs("aapcs")))
__attribute__((visibility("hidden")))
int __aeabi_cfcmpeq_check_nan(float a, float b) {
return __builtin_isnan(a) || __builtin_isnan(b);
}

View File

@ -0,0 +1,43 @@
/* ===-- aeabi_div0.c - ARM Runtime ABI support routines 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 implements the division by zero helper routines as specified by the
* Run-time ABI for the ARM Architecture.
*
* ===----------------------------------------------------------------------===
*/
/*
* RTABI 4.3.2 - Division by zero
*
* The *div0 functions:
* - Return the value passed to them as a parameter
* - Or, return a fixed value defined by the execution environment (such as 0)
* - Or, raise a signal (often SIGFPE) or throw an exception, and do not return
*
* An application may provide its own implementations of the *div0 functions to
* for a particular behaviour from the *div and *divmod functions called out of
* line.
*/
/* provide an unused declaration to pacify pendantic compilation */
extern unsigned char declaration;
#if defined(__ARM_EABI__)
int __attribute__((weak)) __attribute__((visibility("hidden")))
__aeabi_idiv0(int return_value) {
return return_value;
}
long long __attribute__((weak)) __attribute__((visibility("hidden")))
__aeabi_ldiv0(long long return_value) {
return return_value;
}
#endif

View File

@ -0,0 +1,19 @@
//===-- lib/arm/aeabi_drsub.c - Double-precision subtraction --------------===//
//
// 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"
COMPILER_RT_ABI fp_t
__aeabi_dsub(fp_t, fp_t);
COMPILER_RT_ABI fp_t
__aeabi_drsub(fp_t a, fp_t b) {
return __aeabi_dsub(b, a);
}

View File

@ -0,0 +1,19 @@
//===-- lib/arm/aeabi_frsub.c - Single-precision subtraction --------------===//
//
// 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"
COMPILER_RT_ABI fp_t
__aeabi_fsub(fp_t, fp_t);
COMPILER_RT_ABI fp_t
__aeabi_frsub(fp_t a, fp_t b) {
return __aeabi_fsub(b, a);
}

View File

@ -21,7 +21,11 @@
// Reverse all the bytes in a 64-bit integer.
//
.p2align 2
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__bswapdi2)
#else
DEFINE_COMPILERRT_FUNCTION(__bswapdi2)
#endif
#if __ARM_ARCH < 6
// before armv6 does not have "rev" instruction
// r2 = rev(r0)

View File

@ -21,7 +21,11 @@
// Reverse all the bytes in a 32-bit integer.
//
.p2align 2
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__bswapsi2)
#else
DEFINE_COMPILERRT_FUNCTION(__bswapsi2)
#endif
#if __ARM_ARCH < 6
// before armv6 does not have "rev" instruction
eor r1, r0, r0, ror #16

View File

@ -21,7 +21,11 @@
.p2align 2
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__clzdi2)
#else
DEFINE_COMPILERRT_FUNCTION(__clzdi2)
#endif
#ifdef __ARM_FEATURE_CLZ
#ifdef __ARMEB__
cmp r0, 0

View File

@ -20,7 +20,11 @@
#endif
.p2align 2
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__clzsi2)
#else
DEFINE_COMPILERRT_FUNCTION(__clzsi2)
#endif
#ifdef __ARM_FEATURE_CLZ
clz r0, r0
JMP(lr)

View File

@ -32,7 +32,11 @@
@ value is the quotient, the remainder is placed in the variable.
.p2align 3
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__divmodsi4)
#else
DEFINE_COMPILERRT_FUNCTION(__divmodsi4)
#endif
#if __ARM_ARCH_EXT_IDIV__
tst r1, r1
beq LOCAL_LABEL(divzero)

View File

@ -33,7 +33,11 @@ DEFINE_AEABI_FUNCTION_ALIAS(__aeabi_idiv, __divsi3)
@ int __divsi3(int divident, int divisor)
@ Calculate and return the quotient of the (signed) division.
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__divsi3)
#else
DEFINE_COMPILERRT_FUNCTION(__divsi3)
#endif
#if __ARM_ARCH_EXT_IDIV__
tst r1,r1
beq LOCAL_LABEL(divzero)

View File

@ -30,7 +30,11 @@
@ Calculate and return the remainder of the (signed) division.
.p2align 3
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__modsi3)
#else
DEFINE_COMPILERRT_FUNCTION(__modsi3)
#endif
#if __ARM_ARCH_EXT_IDIV__
tst r1, r1
beq LOCAL_LABEL(divzero)

View File

@ -19,7 +19,7 @@
.p2align 2 ; \
.thumb ; \
.syntax unified ; \
DEFINE_COMPILERRT_FUNCTION(__sync_fetch_and_ ## op) \
DEFINE_COMPILERRT_THUMB_FUNCTION(__sync_fetch_and_ ## op) \
dmb ; \
mov r12, r0 ; \
LOCAL_LABEL(tryatomic_ ## op): \
@ -35,7 +35,7 @@
.p2align 2 ; \
.thumb ; \
.syntax unified ; \
DEFINE_COMPILERRT_FUNCTION(__sync_fetch_and_ ## op) \
DEFINE_COMPILERRT_THUMB_FUNCTION(__sync_fetch_and_ ## op) \
push {r4, r5, r6, lr} ; \
dmb ; \
mov r12, r0 ; \

View File

@ -14,9 +14,11 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define add_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) \
adds rD_LO, rN_LO, rM_LO ; \
adc rD_HI, rN_HI, rM_HI
SYNC_OP_8(add_8)
#endif

View File

@ -14,8 +14,10 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define and_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) \
and rD_LO, rN_LO, rM_LO ; \
and rD_HI, rN_HI, rM_HI
SYNC_OP_8(and_8)
#endif

View File

@ -14,6 +14,8 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define max_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) MINMAX_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI, gt)
SYNC_OP_8(max_8)
#endif

View File

@ -14,6 +14,8 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define min_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) MINMAX_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI, lt)
SYNC_OP_8(min_8)
#endif

View File

@ -14,9 +14,11 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define nand_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) \
bic rD_LO, rN_LO, rM_LO ; \
bic rD_HI, rN_HI, rM_HI
SYNC_OP_8(nand_8)
#endif

View File

@ -14,9 +14,11 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define or_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) \
orr rD_LO, rN_LO, rM_LO ; \
orr rD_HI, rN_HI, rM_HI
SYNC_OP_8(or_8)
#endif

View File

@ -14,9 +14,11 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define sub_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) \
subs rD_LO, rN_LO, rM_LO ; \
sbc rD_HI, rN_HI, rM_HI
SYNC_OP_8(sub_8)
#endif

View File

@ -14,6 +14,8 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define umax_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) MINMAX_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI, hi)
SYNC_OP_8(umax_8)
#endif

View File

@ -14,6 +14,8 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define umin_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) MINMAX_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI, lo)
SYNC_OP_8(umin_8)
#endif

View File

@ -14,9 +14,11 @@
#include "sync-ops.h"
#if __ARM_ARCH_PROFILE != 'M'
#define xor_8(rD_LO, rD_HI, rN_LO, rN_HI, rM_LO, rM_HI) \
eor rD_LO, rN_LO, rM_LO ; \
eor rD_HI, rN_HI, rM_HI
SYNC_OP_8(xor_8)
#endif

View File

@ -27,7 +27,11 @@
@ value is the quotient, the remainder is placed in the variable.
.p2align 2
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__udivmodsi4)
#else
DEFINE_COMPILERRT_FUNCTION(__udivmodsi4)
#endif
#if __ARM_ARCH_EXT_IDIV__
tst r1, r1
beq LOCAL_LABEL(divby0)

View File

@ -27,7 +27,11 @@ DEFINE_AEABI_FUNCTION_ALIAS(__aeabi_uidiv, __udivsi3)
@ unsigned int __udivsi3(unsigned int divident, unsigned int divisor)
@ Calculate and return the quotient of the (unsigned) division.
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__udivsi3)
#else
DEFINE_COMPILERRT_FUNCTION(__udivsi3)
#endif
#if __ARM_ARCH_EXT_IDIV__
tst r1, r1
beq LOCAL_LABEL(divby0)

View File

@ -24,7 +24,11 @@
@ Calculate and return the remainder of the (unsigned) division.
.p2align 2
#if __ARM_ARCH_ISA_THUMB == 2
DEFINE_COMPILERRT_THUMB_FUNCTION(__umodsi3)
#else
DEFINE_COMPILERRT_FUNCTION(__umodsi3)
#endif
#if __ARM_ARCH_EXT_IDIV__
tst r1, r1
beq LOCAL_LABEL(divby0)

View File

@ -28,7 +28,10 @@
// 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
#elif defined(__ELF__)
#define HIDDEN(name) .hidden name
#define LOCAL_LABEL(name) .L_##name
#define FILE_LEVEL_DIRECTIVE
@ -37,45 +40,23 @@
#else
#define SYMBOL_IS_FUNC(name) .type name,@function
#endif
#else
#define HIDDEN_DIRECTIVE(name)
#define CONST_SECTION .section .rodata
#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 FILE_LEVEL_DIRECTIVE
#define CONST_SECTION .section .rdata,"rd"
#endif
#if defined(__arm__)
#ifndef __ARM_ARCH
#if defined(__ARM_ARCH_7__) || defined(__ARM_ARCH_7A__) || \
defined(__ARM_ARCH_7R__) || defined(__ARM_ARCH_7M__) || \
defined(__ARM_ARCH_7EM__)
#define __ARM_ARCH 7
#endif
#endif
#ifndef __ARM_ARCH
#if defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) || \
defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6Z__) || \
defined(__ARM_ARCH_6ZK__) || defined(__ARM_ARCH_6ZM__)
#define __ARM_ARCH 6
#endif
#endif
#ifndef __ARM_ARCH
#if defined(__ARM_ARCH_5__) || defined(__ARM_ARCH_5T__) || \
defined(__ARM_ARCH_5TE__) || defined(__ARM_ARCH_5TEJ__)
#define __ARM_ARCH 5
#endif
#endif
#ifndef __ARM_ARCH
#define __ARM_ARCH 4
#endif
#if defined(__ARM_ARCH_4T__) || __ARM_ARCH >= 5
#define ARM_HAS_BX
#endif
@ -92,6 +73,15 @@
#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 __ARM_ARCH_ISA_THUMB == 2
#define IT(cond) it cond
#define ITT(cond) itt cond
@ -125,6 +115,14 @@
DECLARE_SYMBOL_VISIBILITY(name) \
SYMBOL_NAME(name):
#define DEFINE_COMPILERRT_THUMB_FUNCTION(name) \
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) \
FILE_LEVEL_DIRECTIVE SEPARATOR \
.globl SYMBOL_NAME(name) SEPARATOR \

View File

@ -28,20 +28,14 @@
#include <stdint.h>
#include <string.h>
#include "assembly.h"
// Clang objects if you redefine a builtin. This little hack allows us to
// define a function with the same name as an intrinsic.
#if __APPLE__
// mach-o has extra leading underscore
#pragma redefine_extname __atomic_load_c ___atomic_load
#pragma redefine_extname __atomic_store_c ___atomic_store
#pragma redefine_extname __atomic_exchange_c ___atomic_exchange
#pragma redefine_extname __atomic_compare_exchange_c ___atomic_compare_exchange
#else
#pragma redefine_extname __atomic_load_c __atomic_load
#pragma redefine_extname __atomic_store_c __atomic_store
#pragma redefine_extname __atomic_exchange_c __atomic_exchange
#pragma redefine_extname __atomic_compare_exchange_c __atomic_compare_exchange
#endif
#pragma redefine_extname __atomic_load_c SYMBOL_NAME(__atomic_load)
#pragma redefine_extname __atomic_store_c SYMBOL_NAME(__atomic_store)
#pragma redefine_extname __atomic_exchange_c SYMBOL_NAME(__atomic_exchange)
#pragma redefine_extname __atomic_compare_exchange_c SYMBOL_NAME(__atomic_compare_exchange)
/// Number of locks. This allocates one page on 32-bit platforms, two on
/// 64-bit. This can be specified externally if a different trade between
@ -62,13 +56,13 @@ static const long SPINLOCK_MASK = SPINLOCK_COUNT - 1;
#include <machine/atomic.h>
#include <sys/umtx.h>
typedef struct _usem Lock;
inline static void unlock(Lock *l) {
__inline static void unlock(Lock *l) {
__c11_atomic_store((_Atomic(uint32_t)*)&l->_count, 1, __ATOMIC_RELEASE);
__c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
if (l->_has_waiters)
_umtx_op(l, UMTX_OP_SEM_WAKE, 1, 0, 0);
}
inline static void lock(Lock *l) {
__inline static void lock(Lock *l) {
uint32_t old = 1;
while (!__c11_atomic_compare_exchange_weak((_Atomic(uint32_t)*)&l->_count, &old,
0, __ATOMIC_ACQUIRE, __ATOMIC_RELAXED)) {
@ -82,12 +76,12 @@ static Lock locks[SPINLOCK_COUNT] = { [0 ... SPINLOCK_COUNT-1] = {0,1,0} };
#elif defined(__APPLE__)
#include <libkern/OSAtomic.h>
typedef OSSpinLock Lock;
inline static void unlock(Lock *l) {
__inline static void unlock(Lock *l) {
OSSpinLockUnlock(l);
}
/// Locks a lock. In the current implementation, this is potentially
/// unbounded in the contended case.
inline static void lock(Lock *l) {
__inline static void lock(Lock *l) {
OSSpinLockLock(l);
}
static Lock locks[SPINLOCK_COUNT]; // initialized to OS_SPINLOCK_INIT which is 0
@ -95,12 +89,12 @@ static Lock locks[SPINLOCK_COUNT]; // initialized to OS_SPINLOCK_INIT which is 0
#else
typedef _Atomic(uintptr_t) Lock;
/// Unlock a lock. This is a release operation.
inline static void unlock(Lock *l) {
__inline static void unlock(Lock *l) {
__c11_atomic_store(l, 0, __ATOMIC_RELEASE);
}
/// Locks a lock. In the current implementation, this is potentially
/// unbounded in the contended case.
inline static void lock(Lock *l) {
__inline static void lock(Lock *l) {
uintptr_t old = 0;
while (!__c11_atomic_compare_exchange_weak(l, &old, 1, __ATOMIC_ACQUIRE,
__ATOMIC_RELAXED))
@ -112,7 +106,7 @@ static Lock locks[SPINLOCK_COUNT];
/// Returns a lock to use for a given pointer.
static inline Lock *lock_for_pointer(void *ptr) {
static __inline Lock *lock_for_pointer(void *ptr) {
intptr_t hash = (intptr_t)ptr;
// Disregard the lowest 4 bits. We want all values that may be part of the
// same memory operation to hash to the same value and therefore use the same

View File

@ -0,0 +1,27 @@
/*===-- atomic_flag_clear.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 atomic_flag_clear from C11's stdatomic.h.
*
*===------------------------------------------------------------------------===
*/
#ifndef __has_include
#define __has_include(inc) 0
#endif
#if __has_include(<stdatomic.h>)
#include <stdatomic.h>
#undef atomic_flag_clear
void atomic_flag_clear(volatile atomic_flag *object) {
__c11_atomic_store(&(object)->_Value, 0, __ATOMIC_SEQ_CST);
}
#endif

View File

@ -0,0 +1,28 @@
/*===-- atomic_flag_clear_explicit.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 atomic_flag_clear_explicit from C11's stdatomic.h.
*
*===------------------------------------------------------------------------===
*/
#ifndef __has_include
#define __has_include(inc) 0
#endif
#if __has_include(<stdatomic.h>)
#include <stdatomic.h>
#undef atomic_flag_clear_explicit
void atomic_flag_clear_explicit(volatile atomic_flag *object,
memory_order order) {
__c11_atomic_store(&(object)->_Value, 0, order);
}
#endif

View File

@ -0,0 +1,27 @@
/*===-- atomic_flag_test_and_set.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 atomic_flag_test_and_set from C11's stdatomic.h.
*
*===------------------------------------------------------------------------===
*/
#ifndef __has_include
#define __has_include(inc) 0
#endif
#if __has_include(<stdatomic.h>)
#include <stdatomic.h>
#undef atomic_flag_test_and_set
_Bool atomic_flag_test_and_set(volatile atomic_flag *object) {
return __c11_atomic_exchange(&(object)->_Value, 1, __ATOMIC_SEQ_CST);
}
#endif

View File

@ -0,0 +1,28 @@
/*===-- atomic_flag_test_and_set_explicit.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 atomic_flag_test_and_set_explicit from C11's stdatomic.h
*
*===------------------------------------------------------------------------===
*/
#ifndef __has_include
#define __has_include(inc) 0
#endif
#if __has_include(<stdatomic.h>)
#include <stdatomic.h>
#undef atomic_flag_test_and_set_explicit
_Bool atomic_flag_test_and_set_explicit(volatile atomic_flag *object,
memory_order order) {
return __c11_atomic_exchange(&(object)->_Value, 1, order);
}
#endif

View File

@ -0,0 +1,27 @@
/*===-- atomic_signal_fence.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 atomic_signal_fence from C11's stdatomic.h.
*
*===------------------------------------------------------------------------===
*/
#ifndef __has_include
#define __has_include(inc) 0
#endif
#if __has_include(<stdatomic.h>)
#include <stdatomic.h>
#undef atomic_signal_fence
void atomic_signal_fence(memory_order order) {
__c11_atomic_signal_fence(order);
}
#endif

View File

@ -0,0 +1,27 @@
/*===-- atomic_thread_fence.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 atomic_thread_fence from C11's stdatomic.h.
*
*===------------------------------------------------------------------------===
*/
#ifndef __has_include
#define __has_include(inc) 0
#endif
#if __has_include(<stdatomic.h>)
#include <stdatomic.h>
#undef atomic_thread_fence
void atomic_thread_fence(memory_order order) {
__c11_atomic_thread_fence(order);
}
#endif

View File

@ -9,19 +9,71 @@
*/
#include "int_lib.h"
#include <stddef.h>
#if __APPLE__
#include <libkern/OSCacheControl.h>
#endif
#if (defined(__FreeBSD__) || defined(__Bitrig__)) && defined(__arm__)
#include <sys/types.h>
#include <machine/sysarch.h>
#endif
#if defined(__NetBSD__) && defined(__arm__)
#include <machine/sysarch.h>
#endif
#if defined(ANDROID) && defined(__mips__)
#if defined(__mips__)
#include <sys/cachectl.h>
#include <sys/syscall.h>
#include <unistd.h>
#if defined(__ANDROID__) && defined(__LP64__)
/*
* clear_mips_cache - Invalidates instruction cache for Mips.
*/
static void clear_mips_cache(const void* Addr, size_t Size) {
asm volatile (
".set push\n"
".set noreorder\n"
".set noat\n"
"beq %[Size], $zero, 20f\n" /* If size == 0, branch around. */
"nop\n"
"daddu %[Size], %[Addr], %[Size]\n" /* Calculate end address + 1 */
"rdhwr $v0, $1\n" /* Get step size for SYNCI.
$1 is $HW_SYNCI_Step */
"beq $v0, $zero, 20f\n" /* If no caches require
synchronization, branch
around. */
"nop\n"
"10:\n"
"synci 0(%[Addr])\n" /* Synchronize all caches around
address. */
"daddu %[Addr], %[Addr], $v0\n" /* Add step size. */
"sltu $at, %[Addr], %[Size]\n" /* Compare current with end
address. */
"bne $at, $zero, 10b\n" /* Branch if more to do. */
"nop\n"
"sync\n" /* Clear memory hazards. */
"20:\n"
"bal 30f\n"
"nop\n"
"30:\n"
"daddiu $ra, $ra, 12\n" /* $ra has a value of $pc here.
Add offset of 12 to point to the
instruction after the last nop.
*/
"jr.hb $ra\n" /* Return, clearing instruction
hazards. */
"nop\n"
".set pop\n"
: [Addr] "+r"(Addr), [Size] "+r"(Size)
:: "at", "ra", "v0", "memory"
);
}
#endif
#endif
#if defined(ANDROID) && defined(__arm__)
#if defined(__ANDROID__) && defined(__arm__)
#include <asm/unistd.h>
#endif
@ -32,24 +84,22 @@
* specified range.
*/
COMPILER_RT_EXPORT void
__clear_cache(void* start, void* end)
{
void __clear_cache(void *start, void *end) {
#if __i386__ || __x86_64__
/*
* Intel processors have a unified instruction and data cache
* so there is nothing to do
*/
#elif defined(__arm__) && !defined(__APPLE__)
#if defined(__NetBSD__)
#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__Bitrig__)
struct arm_sync_icache_args arg;
arg.addr = (uintptr_t)start;
arg.len = (uintptr_t)end - (uintptr_t)start;
sysarch(ARM_SYNC_ICACHE, &arg);
#elif defined(ANDROID)
const register int start_reg __asm("r0") = (int) (intptr_t) start;
#elif defined(__ANDROID__)
register int start_reg __asm("r0") = (int) (intptr_t) start;
const register int end_reg __asm("r1") = (int) (intptr_t) end;
const register int flags __asm("r2") = 0;
const register int syscall_nr __asm("r7") = __ARM_NR_cacheflush;
@ -61,13 +111,24 @@ __clear_cache(void* start, void* end)
#else
compilerrt_abort();
#endif
#elif defined(ANDROID) && defined(__mips__)
#elif defined(__mips__)
const uintptr_t start_int = (uintptr_t) start;
const uintptr_t end_int = (uintptr_t) end;
_flush_cache(start, (end_int - start_int), BCACHE);
#if defined(__ANDROID__) && defined(__LP64__)
// Call synci implementation for short address range.
const uintptr_t address_range_limit = 256;
if ((end_int - start_int) <= address_range_limit) {
clear_mips_cache(start, (end_int - start_int));
} else {
syscall(__NR_cacheflush, start, (end_int - start_int), BCACHE);
}
#else
syscall(__NR_cacheflush, start, (end_int - start_int), BCACHE);
#endif
#elif defined(__aarch64__) && !defined(__APPLE__)
uint64_t xstart = (uint64_t)(uintptr_t) start;
uint64_t xend = (uint64_t)(uintptr_t) end;
uint64_t addr;
// Get Cache Type Info
uint64_t ctr_el0;
@ -78,12 +139,12 @@ __clear_cache(void* start, void* end)
* uintptr_t in case this runs in an IPL32 environment.
*/
const size_t dcache_line_size = 4 << ((ctr_el0 >> 16) & 15);
for (uint64_t addr = xstart; addr < xend; addr += dcache_line_size)
for (addr = xstart; addr < xend; addr += dcache_line_size)
__asm __volatile("dc cvau, %0" :: "r"(addr));
__asm __volatile("dsb ish");
const size_t icache_line_size = 4 << ((ctr_el0 >> 0) & 15);
for (uint64_t addr = xstart; addr < xend; addr += icache_line_size)
for (addr = xstart; addr < xend; addr += icache_line_size)
__asm __volatile("ic ivau, %0" :: "r"(addr));
__asm __volatile("isb sy");
#else

View File

@ -80,6 +80,11 @@ __ledf2(fp_t a, fp_t b) {
}
}
#if defined(__ELF__)
// Alias for libgcc compatibility
FNALIAS(__cmpdf2, __ledf2);
#endif
enum GE_RESULT {
GE_LESS = -1,
GE_EQUAL = 0,

View File

@ -80,6 +80,11 @@ __lesf2(fp_t a, fp_t b) {
}
}
#if defined(__ELF__)
// Alias for libgcc compatibility
FNALIAS(__cmpsf2, __lesf2);
#endif
enum GE_RESULT {
GE_LESS = -1,
GE_EQUAL = 0,

View File

@ -79,6 +79,11 @@ COMPILER_RT_ABI enum LE_RESULT __letf2(fp_t a, fp_t b) {
}
}
#if defined(__ELF__)
// Alias for libgcc compatibility
FNALIAS(__cmptf2, __letf2);
#endif
enum GE_RESULT {
GE_LESS = -1,
GE_EQUAL = 0,

View File

@ -17,7 +17,7 @@
/* Returns: the quotient of (a + ib) / (c + id) */
COMPILER_RT_ABI double _Complex
COMPILER_RT_ABI Dcomplex
__divdc3(double __a, double __b, double __c, double __d)
{
int __ilogbw = 0;
@ -29,31 +29,31 @@ __divdc3(double __a, double __b, double __c, double __d)
__d = crt_scalbn(__d, -__ilogbw);
}
double __denom = __c * __c + __d * __d;
double _Complex z;
__real__ z = crt_scalbn((__a * __c + __b * __d) / __denom, -__ilogbw);
__imag__ z = crt_scalbn((__b * __c - __a * __d) / __denom, -__ilogbw);
if (crt_isnan(__real__ z) && crt_isnan(__imag__ z))
Dcomplex z;
COMPLEX_REAL(z) = crt_scalbn((__a * __c + __b * __d) / __denom, -__ilogbw);
COMPLEX_IMAGINARY(z) = crt_scalbn((__b * __c - __a * __d) / __denom, -__ilogbw);
if (crt_isnan(COMPLEX_REAL(z)) && crt_isnan(COMPLEX_IMAGINARY(z)))
{
if ((__denom == 0.0) && (!crt_isnan(__a) || !crt_isnan(__b)))
{
__real__ z = crt_copysign(CRT_INFINITY, __c) * __a;
__imag__ z = crt_copysign(CRT_INFINITY, __c) * __b;
COMPLEX_REAL(z) = crt_copysign(CRT_INFINITY, __c) * __a;
COMPLEX_IMAGINARY(z) = crt_copysign(CRT_INFINITY, __c) * __b;
}
else if ((crt_isinf(__a) || crt_isinf(__b)) &&
crt_isfinite(__c) && crt_isfinite(__d))
{
__a = crt_copysign(crt_isinf(__a) ? 1.0 : 0.0, __a);
__b = crt_copysign(crt_isinf(__b) ? 1.0 : 0.0, __b);
__real__ z = CRT_INFINITY * (__a * __c + __b * __d);
__imag__ z = CRT_INFINITY * (__b * __c - __a * __d);
COMPLEX_REAL(z) = CRT_INFINITY * (__a * __c + __b * __d);
COMPLEX_IMAGINARY(z) = CRT_INFINITY * (__b * __c - __a * __d);
}
else if (crt_isinf(__logbw) && __logbw > 0.0 &&
crt_isfinite(__a) && crt_isfinite(__b))
{
__c = crt_copysign(crt_isinf(__c) ? 1.0 : 0.0, __c);
__d = crt_copysign(crt_isinf(__d) ? 1.0 : 0.0, __d);
__real__ z = 0.0 * (__a * __c + __b * __d);
__imag__ z = 0.0 * (__b * __c - __a * __d);
COMPLEX_REAL(z) = 0.0 * (__a * __c + __b * __d);
COMPLEX_IMAGINARY(z) = 0.0 * (__b * __c - __a * __d);
}
}
return z;

View File

@ -17,7 +17,7 @@
/* Returns: the quotient of (a + ib) / (c + id) */
COMPILER_RT_ABI float _Complex
COMPILER_RT_ABI Fcomplex
__divsc3(float __a, float __b, float __c, float __d)
{
int __ilogbw = 0;
@ -29,31 +29,31 @@ __divsc3(float __a, float __b, float __c, float __d)
__d = crt_scalbnf(__d, -__ilogbw);
}
float __denom = __c * __c + __d * __d;
float _Complex z;
__real__ z = crt_scalbnf((__a * __c + __b * __d) / __denom, -__ilogbw);
__imag__ z = crt_scalbnf((__b * __c - __a * __d) / __denom, -__ilogbw);
if (crt_isnan(__real__ z) && crt_isnan(__imag__ z))
Fcomplex z;
COMPLEX_REAL(z) = crt_scalbnf((__a * __c + __b * __d) / __denom, -__ilogbw);
COMPLEX_IMAGINARY(z) = crt_scalbnf((__b * __c - __a * __d) / __denom, -__ilogbw);
if (crt_isnan(COMPLEX_REAL(z)) && crt_isnan(COMPLEX_IMAGINARY(z)))
{
if ((__denom == 0) && (!crt_isnan(__a) || !crt_isnan(__b)))
{
__real__ z = crt_copysignf(CRT_INFINITY, __c) * __a;
__imag__ z = crt_copysignf(CRT_INFINITY, __c) * __b;
COMPLEX_REAL(z) = crt_copysignf(CRT_INFINITY, __c) * __a;
COMPLEX_IMAGINARY(z) = crt_copysignf(CRT_INFINITY, __c) * __b;
}
else if ((crt_isinf(__a) || crt_isinf(__b)) &&
crt_isfinite(__c) && crt_isfinite(__d))
{
__a = crt_copysignf(crt_isinf(__a) ? 1 : 0, __a);
__b = crt_copysignf(crt_isinf(__b) ? 1 : 0, __b);
__real__ z = CRT_INFINITY * (__a * __c + __b * __d);
__imag__ z = CRT_INFINITY * (__b * __c - __a * __d);
COMPLEX_REAL(z) = CRT_INFINITY * (__a * __c + __b * __d);
COMPLEX_IMAGINARY(z) = CRT_INFINITY * (__b * __c - __a * __d);
}
else if (crt_isinf(__logbw) && __logbw > 0 &&
crt_isfinite(__a) && crt_isfinite(__b))
{
__c = crt_copysignf(crt_isinf(__c) ? 1 : 0, __c);
__d = crt_copysignf(crt_isinf(__d) ? 1 : 0, __d);
__real__ z = 0 * (__a * __c + __b * __d);
__imag__ z = 0 * (__b * __c - __a * __d);
COMPLEX_REAL(z) = 0 * (__a * __c + __b * __d);
COMPLEX_IMAGINARY(z) = 0 * (__b * __c - __a * __d);
}
}
return z;

View File

@ -0,0 +1,60 @@
/*===-- divtc3.c - Implement __divtc3 -------------------------------------===
*
* 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 __divtc3 for the compiler_rt library.
*
*===----------------------------------------------------------------------===
*/
#include "int_lib.h"
#include "int_math.h"
/* Returns: the quotient of (a + ib) / (c + id) */
COMPILER_RT_ABI long double _Complex
__divtc3(long double __a, long double __b, long double __c, long double __d)
{
int __ilogbw = 0;
long double __logbw = crt_logbl(crt_fmaxl(crt_fabsl(__c), crt_fabsl(__d)));
if (crt_isfinite(__logbw))
{
__ilogbw = (int)__logbw;
__c = crt_scalbnl(__c, -__ilogbw);
__d = crt_scalbnl(__d, -__ilogbw);
}
long double __denom = __c * __c + __d * __d;
long double _Complex z;
__real__ z = crt_scalbnl((__a * __c + __b * __d) / __denom, -__ilogbw);
__imag__ z = crt_scalbnl((__b * __c - __a * __d) / __denom, -__ilogbw);
if (crt_isnan(__real__ z) && crt_isnan(__imag__ z))
{
if ((__denom == 0.0) && (!crt_isnan(__a) || !crt_isnan(__b)))
{
__real__ z = crt_copysignl(CRT_INFINITY, __c) * __a;
__imag__ z = crt_copysignl(CRT_INFINITY, __c) * __b;
}
else if ((crt_isinf(__a) || crt_isinf(__b)) &&
crt_isfinite(__c) && crt_isfinite(__d))
{
__a = crt_copysignl(crt_isinf(__a) ? 1.0 : 0.0, __a);
__b = crt_copysignl(crt_isinf(__b) ? 1.0 : 0.0, __b);
__real__ z = CRT_INFINITY * (__a * __c + __b * __d);
__imag__ z = CRT_INFINITY * (__b * __c - __a * __d);
}
else if (crt_isinf(__logbw) && __logbw > 0.0 &&
crt_isfinite(__a) && crt_isfinite(__b))
{
__c = crt_copysignl(crt_isinf(__c) ? 1.0 : 0.0, __c);
__d = crt_copysignl(crt_isinf(__d) ? 1.0 : 0.0, __d);
__real__ z = 0.0 * (__a * __c + __b * __d);
__imag__ z = 0.0 * (__b * __c - __a * __d);
}
}
return z;
}

View File

@ -18,7 +18,7 @@
/* Returns: the quotient of (a + ib) / (c + id) */
COMPILER_RT_ABI long double _Complex
COMPILER_RT_ABI Lcomplex
__divxc3(long double __a, long double __b, long double __c, long double __d)
{
int __ilogbw = 0;
@ -30,31 +30,31 @@ __divxc3(long double __a, long double __b, long double __c, long double __d)
__d = crt_scalbnl(__d, -__ilogbw);
}
long double __denom = __c * __c + __d * __d;
long double _Complex z;
__real__ z = crt_scalbnl((__a * __c + __b * __d) / __denom, -__ilogbw);
__imag__ z = crt_scalbnl((__b * __c - __a * __d) / __denom, -__ilogbw);
if (crt_isnan(__real__ z) && crt_isnan(__imag__ z))
Lcomplex z;
COMPLEX_REAL(z) = crt_scalbnl((__a * __c + __b * __d) / __denom, -__ilogbw);
COMPLEX_IMAGINARY(z) = crt_scalbnl((__b * __c - __a * __d) / __denom, -__ilogbw);
if (crt_isnan(COMPLEX_REAL(z)) && crt_isnan(COMPLEX_IMAGINARY(z)))
{
if ((__denom == 0) && (!crt_isnan(__a) || !crt_isnan(__b)))
{
__real__ z = crt_copysignl(CRT_INFINITY, __c) * __a;
__imag__ z = crt_copysignl(CRT_INFINITY, __c) * __b;
COMPLEX_REAL(z) = crt_copysignl(CRT_INFINITY, __c) * __a;
COMPLEX_IMAGINARY(z) = crt_copysignl(CRT_INFINITY, __c) * __b;
}
else if ((crt_isinf(__a) || crt_isinf(__b)) &&
crt_isfinite(__c) && crt_isfinite(__d))
{
__a = crt_copysignl(crt_isinf(__a) ? 1 : 0, __a);
__b = crt_copysignl(crt_isinf(__b) ? 1 : 0, __b);
__real__ z = CRT_INFINITY * (__a * __c + __b * __d);
__imag__ z = CRT_INFINITY * (__b * __c - __a * __d);
COMPLEX_REAL(z) = CRT_INFINITY * (__a * __c + __b * __d);
COMPLEX_IMAGINARY(z) = CRT_INFINITY * (__b * __c - __a * __d);
}
else if (crt_isinf(__logbw) && __logbw > 0 &&
crt_isfinite(__a) && crt_isfinite(__b))
{
__c = crt_copysignl(crt_isinf(__c) ? 1 : 0, __c);
__d = crt_copysignl(crt_isinf(__d) ? 1 : 0, __d);
__real__ z = 0 * (__a * __c + __b * __d);
__imag__ z = 0 * (__b * __c - __a * __d);
COMPLEX_REAL(z) = 0 * (__a * __c + __b * __d);
COMPLEX_IMAGINARY(z) = 0 * (__b * __c - __a * __d);
}
}
return z;

View File

@ -0,0 +1,183 @@
/* ===---------- emutls.c - Implements __emutls_get_address ---------------===
*
* 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 <pthread.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include "int_lib.h"
#include "int_util.h"
/* Default is not to use posix_memalign, so systems like Android
* can use thread local data without heavier POSIX memory allocators.
*/
#ifndef EMUTLS_USE_POSIX_MEMALIGN
#define EMUTLS_USE_POSIX_MEMALIGN 0
#endif
/* For every TLS variable xyz,
* there is one __emutls_control variable named __emutls_v.xyz.
* If xyz has non-zero initial value, __emutls_v.xyz's "value"
* will point to __emutls_t.xyz, which has the initial value.
*/
typedef struct __emutls_control {
size_t size; /* size of the object in bytes */
size_t align; /* alignment of the object in bytes */
union {
uintptr_t index; /* data[index-1] is the object address */
void* address; /* object address, when in single thread env */
} object;
void* value; /* null or non-zero initial value for the object */
} __emutls_control;
static __inline void *emutls_memalign_alloc(size_t align, size_t size) {
void *base;
#if EMUTLS_USE_POSIX_MEMALIGN
if (posix_memalign(&base, align, size) != 0)
abort();
#else
#define EXTRA_ALIGN_PTR_BYTES (align - 1 + sizeof(void*))
char* object;
if ((object = malloc(EXTRA_ALIGN_PTR_BYTES + size)) == NULL)
abort();
base = (void*)(((uintptr_t)(object + EXTRA_ALIGN_PTR_BYTES))
& ~(uintptr_t)(align - 1));
((void**)base)[-1] = object;
#endif
return base;
}
static __inline void emutls_memalign_free(void *base) {
#if EMUTLS_USE_POSIX_MEMALIGN
free(base);
#else
/* The mallocated address is in ((void**)base)[-1] */
free(((void**)base)[-1]);
#endif
}
/* Emulated TLS objects are always allocated at run-time. */
static __inline void *emutls_allocate_object(__emutls_control *control) {
/* Use standard C types, check with gcc's emutls.o. */
typedef unsigned int gcc_word __attribute__((mode(word)));
typedef unsigned int gcc_pointer __attribute__((mode(pointer)));
COMPILE_TIME_ASSERT(sizeof(size_t) == sizeof(gcc_word));
COMPILE_TIME_ASSERT(sizeof(uintptr_t) == sizeof(gcc_pointer));
COMPILE_TIME_ASSERT(sizeof(uintptr_t) == sizeof(void*));
size_t size = control->size;
size_t align = control->align;
if (align < sizeof(void*))
align = sizeof(void*);
/* Make sure that align is power of 2. */
if ((align & (align - 1)) != 0)
abort();
void* base = emutls_memalign_alloc(align, size);
if (control->value)
memcpy(base, control->value, size);
else
memset(base, 0, size);
return base;
}
static pthread_mutex_t emutls_mutex = PTHREAD_MUTEX_INITIALIZER;
static size_t emutls_num_object = 0; /* number of allocated TLS objects */
typedef struct emutls_address_array {
uintptr_t size; /* number of elements in the 'data' array */
void* data[];
} emutls_address_array;
static pthread_key_t emutls_pthread_key;
static void emutls_key_destructor(void* ptr) {
emutls_address_array* array = (emutls_address_array*)ptr;
uintptr_t i;
for (i = 0; i < array->size; ++i) {
if (array->data[i])
emutls_memalign_free(array->data[i]);
}
free(ptr);
}
static void emutls_init(void) {
if (pthread_key_create(&emutls_pthread_key, emutls_key_destructor) != 0)
abort();
}
/* Returns control->object.index; set index if not allocated yet. */
static __inline uintptr_t emutls_get_index(__emutls_control *control) {
uintptr_t index = __atomic_load_n(&control->object.index, __ATOMIC_ACQUIRE);
if (!index) {
static pthread_once_t once = PTHREAD_ONCE_INIT;
pthread_once(&once, emutls_init);
pthread_mutex_lock(&emutls_mutex);
index = control->object.index;
if (!index) {
index = ++emutls_num_object;
__atomic_store_n(&control->object.index, index, __ATOMIC_RELEASE);
}
pthread_mutex_unlock(&emutls_mutex);
}
return index;
}
/* Updates newly allocated thread local emutls_address_array. */
static __inline void emutls_check_array_set_size(emutls_address_array *array,
uintptr_t size) {
if (array == NULL)
abort();
array->size = size;
pthread_setspecific(emutls_pthread_key, (void*)array);
}
/* Returns the new 'data' array size, number of elements,
* which must be no smaller than the given index.
*/
static __inline uintptr_t emutls_new_data_array_size(uintptr_t index) {
/* Need to allocate emutls_address_array with one extra slot
* to store the data array size.
* Round up the emutls_address_array size to multiple of 16.
*/
return ((index + 1 + 15) & ~((uintptr_t)15)) - 1;
}
/* Returns the thread local emutls_address_array.
* Extends its size if necessary to hold address at index.
*/
static __inline emutls_address_array *
emutls_get_address_array(uintptr_t index) {
emutls_address_array* array = pthread_getspecific(emutls_pthread_key);
if (array == NULL) {
uintptr_t new_size = emutls_new_data_array_size(index);
array = calloc(new_size + 1, sizeof(void*));
emutls_check_array_set_size(array, new_size);
} else if (index > array->size) {
uintptr_t orig_size = array->size;
uintptr_t new_size = emutls_new_data_array_size(index);
array = realloc(array, (new_size + 1) * sizeof(void*));
if (array)
memset(array->data + orig_size, 0,
(new_size - orig_size) * sizeof(void*));
emutls_check_array_set_size(array, new_size);
}
return array;
}
void* __emutls_get_address(__emutls_control* control) {
uintptr_t index = emutls_get_index(control);
emutls_address_array* array = emutls_get_address_array(index);
if (array->data[index - 1] == NULL)
array->data[index - 1] = emutls_allocate_object(control);
return array->data[index - 1];
}

View File

@ -10,7 +10,9 @@
#include "int_lib.h"
#ifndef _WIN32
#include <sys/mman.h>
#endif
/* #include "config.h"
* FIXME: CMake - include when cmake system is ready.
@ -18,9 +20,14 @@
*/
#define HAVE_SYSCONF 1
#ifdef _WIN32
#define WIN32_LEAN_AND_MEAN
#include <Windows.h>
#else
#ifndef __APPLE__
#include <unistd.h>
#endif /* __APPLE__ */
#endif /* _WIN32 */
#if __LP64__
#define TRAMPOLINE_SIZE 48
@ -40,6 +47,12 @@ COMPILER_RT_ABI void
__enable_execute_stack(void* addr)
{
#if _WIN32
MEMORY_BASIC_INFORMATION mbi;
if (!VirtualQuery (addr, &mbi, sizeof(mbi)))
return; /* We should probably assert here because there is no return value */
VirtualProtect (mbi.BaseAddress, mbi.RegionSize, PAGE_EXECUTE_READWRITE, &mbi.Protect);
#else
#if __APPLE__
/* On Darwin, pagesize is always 4096 bytes */
const uintptr_t pageSize = 4096;
@ -55,4 +68,5 @@ __enable_execute_stack(void* addr)
unsigned char* endPage = (unsigned char*)((p+TRAMPOLINE_SIZE+pageSize) & pageAlignMask);
size_t length = endPage - startPage;
(void) mprotect((void *)startPage, length, PROT_READ | PROT_WRITE | PROT_EXEC);
#endif
}

View File

@ -0,0 +1,25 @@
//===-- lib/extendhfsf2.c - half -> single 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.
//
//===----------------------------------------------------------------------===//
//
#define SRC_HALF
#define DST_SINGLE
#include "fp_extend_impl.inc"
ARM_EABI_FNALIAS(h2f, extendhfsf2)
// Use a forwarding definition and noinline to implement a poor man's alias,
// as there isn't a good cross-platform way of defining one.
COMPILER_RT_ABI NOINLINE float __extendhfsf2(uint16_t a) {
return __extendXfYf2__(a);
}
COMPILER_RT_ABI float __gnu_h2f_ieee(uint16_t a) {
return __extendhfsf2(a);
}

View File

@ -6,40 +6,41 @@
* Source Licenses. See LICENSE.TXT for details.
*
* ===----------------------------------------------------------------------===
*
* This file implements __fixdfdi for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
/* Returns: convert a to a signed long long, rounding toward zero. */
/* Assumption: double is a IEEE 64 bit floating point type
* su_int is a 32 bit integral type
* value in double is representable in di_int (no range checking performed)
*/
/* seee eeee eeee mmmm mmmm mmmm mmmm mmmm | mmmm mmmm mmmm mmmm mmmm mmmm mmmm mmmm */
#define DOUBLE_PRECISION
#include "fp_lib.h"
ARM_EABI_FNALIAS(d2lz, fixdfdi)
#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)
{
double_bits fb;
fb.f = a;
int e = ((fb.u.s.high & 0x7FF00000) >> 20) - 1023;
if (e < 0)
return 0;
di_int s = (si_int)(fb.u.s.high & 0x80000000) >> 31;
dwords r;
r.s.high = (fb.u.s.high & 0x000FFFFF) | 0x00100000;
r.s.low = fb.u.s.low;
if (e > 52)
r.all <<= (e - 52);
else
r.all >>= (52 - e);
return (r.all ^ s) - s;
}
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

View File

@ -1,50 +1,22 @@
//===-- 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 double-precision to integer conversion for the
// compiler-rt library. No range checking is performed; the behavior of this
// conversion is undefined for out of range values in the C standard.
//
//===----------------------------------------------------------------------===//
/* ===-- fixdfsi.c - Implement __fixdfsi -----------------------------------===
*
* 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"
#include "int_lib.h"
typedef si_int fixint_t;
typedef su_int fixuint_t;
#include "fp_fixint_impl.inc"
ARM_EABI_FNALIAS(d2iz, fixdfsi)
COMPILER_RT_ABI int
COMPILER_RT_ABI si_int
__fixdfsi(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 0 < exponent < significandBits, right shift to get the result.
if ((unsigned int)exponent < significandBits) {
return sign * (significand >> (significandBits - exponent));
}
// If exponent is negative, the result is zero.
else if (exponent < 0) {
return 0;
}
// If significandBits < exponent, left shift to get the result. This shift
// may end up being larger than the type width, which incurs undefined
// behavior, but the conversion itself is undefined in that case, so
// whatever the compiler decides to do is fine.
else {
return sign * (significand << (exponent - significandBits));
}
return __fixint(a);
}

View File

@ -6,40 +6,21 @@
* Source Licenses. See LICENSE.TXT for details.
*
* ===----------------------------------------------------------------------===
*
* This file implements __fixdfti for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
#ifdef CRT_HAS_128BIT
#define DOUBLE_PRECISION
#include "fp_lib.h"
/* Returns: convert a to a signed long long, rounding toward zero. */
/* Assumption: double is a IEEE 64 bit floating point type
* su_int is a 32 bit integral type
* value in double is representable in ti_int (no range checking performed)
*/
/* seee eeee eeee mmmm mmmm mmmm mmmm mmmm | mmmm mmmm mmmm mmmm mmmm mmmm mmmm mmmm */
typedef ti_int fixint_t;
typedef tu_int fixuint_t;
#include "fp_fixint_impl.inc"
COMPILER_RT_ABI ti_int
__fixdfti(double a)
{
double_bits fb;
fb.f = a;
int e = ((fb.u.s.high & 0x7FF00000) >> 20) - 1023;
if (e < 0)
return 0;
ti_int s = (si_int)(fb.u.s.high & 0x80000000) >> 31;
ti_int r = 0x0010000000000000uLL | (0x000FFFFFFFFFFFFFuLL & fb.u.all);
if (e > 52)
r <<= (e - 52);
else
r >>= (52 - e);
return (r ^ s) - s;
__fixdfti(fp_t a) {
return __fixint(a);
}
#endif /* CRT_HAS_128BIT */

View File

@ -1,43 +1,47 @@
/* ===-- fixsfdi.c - Implement __fixsfdi -----------------------------------===
*
* The LLVM Compiler Infrastructure
* 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 __fixsfdi for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
/* Returns: convert a to a signed long long, rounding toward zero. */
/* Assumption: float is a IEEE 32 bit floating point type
* su_int is a 32 bit integral type
* value in float is representable in di_int (no range checking performed)
*/
/* seee eeee emmm mmmm mmmm mmmm mmmm mmmm */
#define SINGLE_PRECISION
#include "fp_lib.h"
ARM_EABI_FNALIAS(f2lz, fixsfdi)
#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)
{
float_bits fb;
fb.f = a;
int e = ((fb.u & 0x7F800000) >> 23) - 127;
if (e < 0)
return 0;
di_int s = (si_int)(fb.u & 0x80000000) >> 31;
di_int r = (fb.u & 0x007FFFFF) | 0x00800000;
if (e > 23)
r <<= (e - 23);
else
r >>= (23 - e);
return (r ^ s) - s;
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

View File

@ -1,47 +1,22 @@
//===-- lib/fixsfsi.c - Single-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 single-precision to integer conversion for the
// compiler-rt library. No range checking is performed; the behavior of this
// conversion is undefined for out of range values in the C standard.
//
//===----------------------------------------------------------------------===//
/* ===-- fixsfsi.c - Implement __fixsfsi -----------------------------------===
*
* 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"
typedef si_int fixint_t;
typedef su_int fixuint_t;
#include "fp_fixint_impl.inc"
ARM_EABI_FNALIAS(f2iz, fixsfsi)
COMPILER_RT_ABI int
COMPILER_RT_ABI si_int
__fixsfsi(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 0 < exponent < significandBits, right shift to get the result.
if ((unsigned int)exponent < significandBits) {
return sign * (significand >> (significandBits - exponent));
}
// If exponent is negative, the result is zero.
else if (exponent < 0) {
return 0;
}
// If significandBits < exponent, left shift to get the result. This shift
// may end up being larger than the type width, which incurs undefined
// behavior, but the conversion itself is undefined in that case, so
// whatever the compiler decides to do is fine.
else {
return sign * (significand << (exponent - significandBits));
}
return __fixint(a);
}

View File

@ -6,40 +6,21 @@
* Source Licenses. See LICENSE.TXT for details.
*
* ===----------------------------------------------------------------------===
*
* This file implements __fixsfti for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
#ifdef CRT_HAS_128BIT
#define SINGLE_PRECISION
#include "fp_lib.h"
/* Returns: convert a to a signed long long, rounding toward zero. */
/* Assumption: float is a IEEE 32 bit floating point type
* su_int is a 32 bit integral type
* value in float is representable in ti_int (no range checking performed)
*/
/* seee eeee emmm mmmm mmmm mmmm mmmm mmmm */
typedef ti_int fixint_t;
typedef tu_int fixuint_t;
#include "fp_fixint_impl.inc"
COMPILER_RT_ABI ti_int
__fixsfti(float a)
{
float_bits fb;
fb.f = a;
int e = ((fb.u & 0x7F800000) >> 23) - 127;
if (e < 0)
return 0;
ti_int s = (si_int)(fb.u & 0x80000000) >> 31;
ti_int r = (fb.u & 0x007FFFFF) | 0x00800000;
if (e > 23)
r <<= (e - 23);
else
r >>= (23 - e);
return (r ^ s) - s;
__fixsfti(fp_t a) {
return __fixint(a);
}
#endif /* CRT_HAS_128BIT */

View File

@ -0,0 +1,23 @@
/* ===-- fixtfdi.c - Implement __fixtfdi -----------------------------------===
*
* 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 QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
typedef di_int fixint_t;
typedef du_int fixuint_t;
#include "fp_fixint_impl.inc"
COMPILER_RT_ABI di_int
__fixtfdi(fp_t a) {
return __fixint(a);
}
#endif

View File

@ -0,0 +1,23 @@
/* ===-- fixtfsi.c - Implement __fixtfsi -----------------------------------===
*
* 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 QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
typedef si_int fixint_t;
typedef su_int fixuint_t;
#include "fp_fixint_impl.inc"
COMPILER_RT_ABI si_int
__fixtfsi(fp_t a) {
return __fixint(a);
}
#endif

View File

@ -0,0 +1,23 @@
/* ===-- fixtfti.c - Implement __fixtfti -----------------------------------===
*
* 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 QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
typedef ti_int fixint_t;
typedef tu_int fixuint_t;
#include "fp_fixint_impl.inc"
COMPILER_RT_ABI ti_int
__fixtfti(fp_t a) {
return __fixint(a);
}
#endif

View File

@ -6,42 +6,39 @@
* Source Licenses. See LICENSE.TXT for details.
*
* ===----------------------------------------------------------------------===
*
* This file implements __fixunsdfdi for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
/* Returns: convert a to a unsigned long long, rounding toward zero.
* Negative values all become zero.
*/
/* Assumption: double is a IEEE 64 bit floating point type
* du_int is a 64 bit integral type
* value in double is representable in du_int or is negative
* (no range checking performed)
*/
/* seee eeee eeee mmmm mmmm mmmm mmmm mmmm | mmmm mmmm mmmm mmmm mmmm mmmm mmmm mmmm */
#define DOUBLE_PRECISION
#include "fp_lib.h"
ARM_EABI_FNALIAS(d2ulz, fixunsdfdi)
#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)
{
double_bits fb;
fb.f = a;
int e = ((fb.u.s.high & 0x7FF00000) >> 20) - 1023;
if (e < 0 || (fb.u.s.high & 0x80000000))
return 0;
udwords r;
r.s.high = (fb.u.s.high & 0x000FFFFF) | 0x00100000;
r.s.low = fb.u.s.low;
if (e > 52)
r.all <<= (e - 52);
else
r.all >>= (52 - e);
return r.all;
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

View File

@ -6,39 +6,16 @@
* Source Licenses. See LICENSE.TXT for details.
*
* ===----------------------------------------------------------------------===
*
* This file implements __fixunsdfsi for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
/* Returns: convert a to a unsigned int, rounding toward zero.
* Negative values all become zero.
*/
/* Assumption: double is a IEEE 64 bit floating point type
* su_int is a 32 bit integral type
* value in double is representable in su_int or is negative
* (no range checking performed)
*/
/* seee eeee eeee mmmm mmmm mmmm mmmm mmmm | mmmm mmmm mmmm mmmm mmmm mmmm mmmm mmmm */
#define DOUBLE_PRECISION
#include "fp_lib.h"
typedef su_int fixuint_t;
#include "fp_fixuint_impl.inc"
ARM_EABI_FNALIAS(d2uiz, fixunsdfsi)
COMPILER_RT_ABI su_int
__fixunsdfsi(double a)
{
double_bits fb;
fb.f = a;
int e = ((fb.u.s.high & 0x7FF00000) >> 20) - 1023;
if (e < 0 || (fb.u.s.high & 0x80000000))
return 0;
return (
0x80000000u |
((fb.u.s.high & 0x000FFFFF) << 11) |
(fb.u.s.low >> 21)
) >> (31 - e);
__fixunsdfsi(fp_t a) {
return __fixuint(a);
}

View File

@ -6,42 +6,18 @@
* Source Licenses. See LICENSE.TXT for details.
*
* ===----------------------------------------------------------------------===
*
* This file implements __fixunsdfti for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
#ifdef CRT_HAS_128BIT
/* Returns: convert a to a unsigned long long, rounding toward zero.
* Negative values all become zero.
*/
/* Assumption: double is a IEEE 64 bit floating point type
* tu_int is a 64 bit integral type
* value in double is representable in tu_int or is negative
* (no range checking performed)
*/
/* seee eeee eeee mmmm mmmm mmmm mmmm mmmm | mmmm mmmm mmmm mmmm mmmm mmmm mmmm mmmm */
#define DOUBLE_PRECISION
#include "fp_lib.h"
typedef tu_int fixuint_t;
#include "fp_fixuint_impl.inc"
COMPILER_RT_ABI tu_int
__fixunsdfti(double a)
{
double_bits fb;
fb.f = a;
int e = ((fb.u.s.high & 0x7FF00000) >> 20) - 1023;
if (e < 0 || (fb.u.s.high & 0x80000000))
return 0;
tu_int r = 0x0010000000000000uLL | (fb.u.all & 0x000FFFFFFFFFFFFFuLL);
if (e > 52)
r <<= (e - 52);
else
r >>= (52 - e);
return r;
__fixunsdfti(fp_t a) {
return __fixuint(a);
}
#endif /* CRT_HAS_128BIT */

View File

@ -6,39 +6,40 @@
* Source Licenses. See LICENSE.TXT for details.
*
* ===----------------------------------------------------------------------===
*
* This file implements __fixunssfdi for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
/* Returns: convert a to a unsigned long long, rounding toward zero.
* Negative values all become zero.
*/
/* Assumption: float is a IEEE 32 bit floating point type
* du_int is a 64 bit integral type
* value in float is representable in du_int or is negative
* (no range checking performed)
*/
/* seee eeee emmm mmmm mmmm mmmm mmmm mmmm */
#define SINGLE_PRECISION
#include "fp_lib.h"
ARM_EABI_FNALIAS(f2ulz, fixunssfdi)
#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)
{
float_bits fb;
fb.f = a;
int e = ((fb.u & 0x7F800000) >> 23) - 127;
if (e < 0 || (fb.u & 0x80000000))
return 0;
du_int r = (fb.u & 0x007FFFFF) | 0x00800000;
if (e > 23)
r <<= (e - 23);
else
r >>= (23 - e);
return r;
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

View File

@ -12,34 +12,14 @@
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
/* Returns: convert a to a unsigned int, rounding toward zero.
* Negative values all become zero.
*/
/* Assumption: float is a IEEE 32 bit floating point type
* su_int is a 32 bit integral type
* value in float is representable in su_int or is negative
* (no range checking performed)
*/
/* seee eeee emmm mmmm mmmm mmmm mmmm mmmm */
#define SINGLE_PRECISION
#include "fp_lib.h"
typedef su_int fixuint_t;
#include "fp_fixuint_impl.inc"
ARM_EABI_FNALIAS(f2uiz, fixunssfsi)
COMPILER_RT_ABI su_int
__fixunssfsi(float a)
{
float_bits fb;
fb.f = a;
int e = ((fb.u & 0x7F800000) >> 23) - 127;
if (e < 0 || (fb.u & 0x80000000))
return 0;
su_int r = (fb.u & 0x007FFFFF) | 0x00800000;
if (e > 23)
r <<= (e - 23);
else
r >>= (23 - e);
return r;
__fixunssfsi(fp_t a) {
return __fixuint(a);
}

View File

@ -12,36 +12,15 @@
* ===----------------------------------------------------------------------===
*/
#include "int_lib.h"
#define SINGLE_PRECISION
#include "fp_lib.h"
#ifdef CRT_HAS_128BIT
/* Returns: convert a to a unsigned long long, rounding toward zero.
* Negative values all become zero.
*/
/* Assumption: float is a IEEE 32 bit floating point type
* tu_int is a 64 bit integral type
* value in float is representable in tu_int or is negative
* (no range checking performed)
*/
/* seee eeee emmm mmmm mmmm mmmm mmmm mmmm */
#if defined(CRT_HAS_128BIT)
typedef tu_int fixuint_t;
#include "fp_fixuint_impl.inc"
COMPILER_RT_ABI tu_int
__fixunssfti(float a)
{
float_bits fb;
fb.f = a;
int e = ((fb.u & 0x7F800000) >> 23) - 127;
if (e < 0 || (fb.u & 0x80000000))
return 0;
tu_int r = (fb.u & 0x007FFFFF) | 0x00800000;
if (e > 23)
r <<= (e - 23);
else
r >>= (23 - e);
return r;
__fixunssfti(fp_t a) {
return __fixuint(a);
}
#endif /* CRT_HAS_128BIT */
#endif

View File

@ -0,0 +1,22 @@
/* ===-- fixunstfdi.c - Implement __fixunstfdi -----------------------------===
*
* 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 QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
typedef du_int fixuint_t;
#include "fp_fixuint_impl.inc"
COMPILER_RT_ABI du_int
__fixunstfdi(fp_t a) {
return __fixuint(a);
}
#endif

View File

@ -0,0 +1,22 @@
/* ===-- fixunstfsi.c - Implement __fixunstfsi -----------------------------===
*
* 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 QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
typedef su_int fixuint_t;
#include "fp_fixuint_impl.inc"
COMPILER_RT_ABI su_int
__fixunstfsi(fp_t a) {
return __fixuint(a);
}
#endif

View File

@ -0,0 +1,22 @@
/* ===-- fixunstfsi.c - Implement __fixunstfsi -----------------------------===
*
* 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 QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
typedef tu_int fixuint_t;
#include "fp_fixuint_impl.inc"
COMPILER_RT_ABI tu_int
__fixunstfti(fp_t a) {
return __fixuint(a);
}
#endif

View File

@ -38,6 +38,8 @@ __fixunsxfdi(long double a)
int e = (fb.u.high.s.low & 0x00007FFF) - 16383;
if (e < 0 || (fb.u.high.s.low & 0x00008000))
return 0;
if ((unsigned)e > sizeof(du_int) * CHAR_BIT)
return ~(du_int)0;
return fb.u.low.all >> (63 - e);
}

View File

@ -23,7 +23,6 @@
/* Assumption: long double is an intel 80 bit floating point type padded with 6 bytes
* su_int is a 32 bit integral type
* value in long double is representable in su_int or is negative
* (no range checking performed)
*/
/* gggg gggg gggg gggg gggg gggg gggg gggg | gggg gggg gggg gggg seee eeee eeee eeee |
@ -38,6 +37,8 @@ __fixunsxfsi(long double a)
int e = (fb.u.high.s.low & 0x00007FFF) - 16383;
if (e < 0 || (fb.u.high.s.low & 0x00008000))
return 0;
if ((unsigned)e > sizeof(su_int) * CHAR_BIT)
return ~(su_int)0;
return fb.u.low.s.high >> (31 - e);
}

View File

@ -21,9 +21,8 @@
*/
/* Assumption: long double is an intel 80 bit floating point type padded with 6 bytes
* tu_int is a 64 bit integral type
* tu_int is a 128 bit integral type
* value in long double is representable in tu_int or is negative
* (no range checking performed)
*/
/* gggg gggg gggg gggg gggg gggg gggg gggg | gggg gggg gggg gggg seee eeee eeee eeee |
@ -38,6 +37,8 @@ __fixunsxfti(long double a)
int e = (fb.u.high.s.low & 0x00007FFF) - 16383;
if (e < 0 || (fb.u.high.s.low & 0x00008000))
return 0;
if ((unsigned)e > sizeof(tu_int) * CHAR_BIT)
return ~(tu_int)0;
tu_int r = fb.u.low.all;
if (e > 63)
r <<= (e - 63);

View File

@ -19,7 +19,7 @@
/* Returns: convert a to a signed long long, rounding toward zero. */
/* Assumption: long double is an intel 80 bit floating point type padded with 6 bytes
* su_int is a 32 bit integral type
* di_int is a 64 bit integral type
* value in long double is representable in di_int (no range checking performed)
*/
@ -30,11 +30,15 @@
COMPILER_RT_ABI di_int
__fixxfdi(long double a)
{
const di_int di_max = (di_int)((~(du_int)0) / 2);
const di_int di_min = -di_max - 1;
long_double_bits fb;
fb.f = a;
int e = (fb.u.high.s.low & 0x00007FFF) - 16383;
if (e < 0)
return 0;
if ((unsigned)e >= sizeof(di_int) * CHAR_BIT)
return a > 0 ? di_max : di_min;
di_int s = -(si_int)((fb.u.high.s.low & 0x00008000) >> 15);
di_int r = fb.u.low.all;
r = (du_int)r >> (63 - e);

View File

@ -19,8 +19,8 @@
/* Returns: convert a to a signed long long, rounding toward zero. */
/* Assumption: long double is an intel 80 bit floating point type padded with 6 bytes
* su_int is a 32 bit integral type
* value in long double is representable in ti_int (no range checking performed)
* ti_int is a 128 bit integral type
* value in long double is representable in ti_int
*/
/* gggg gggg gggg gggg gggg gggg gggg gggg | gggg gggg gggg gggg seee eeee eeee eeee |
@ -30,6 +30,8 @@
COMPILER_RT_ABI ti_int
__fixxfti(long double a)
{
const ti_int ti_max = (ti_int)((~(tu_int)0) / 2);
const ti_int ti_min = -ti_max - 1;
long_double_bits fb;
fb.f = a;
int e = (fb.u.high.s.low & 0x00007FFF) - 16383;
@ -37,6 +39,8 @@ __fixxfti(long double a)
return 0;
ti_int s = -(si_int)((fb.u.high.s.low & 0x00008000) >> 15);
ti_int r = fb.u.low.all;
if ((unsigned)e >= sizeof(ti_int) * CHAR_BIT)
return a > 0 ? ti_max : ti_min;
if (e > 63)
r <<= (e - 63);
else

View File

@ -32,8 +32,8 @@ ARM_EABI_FNALIAS(l2d, floatdidf)
COMPILER_RT_ABI double
__floatdidf(di_int a)
{
static const double twop52 = 0x1.0p52;
static const double twop32 = 0x1.0p32;
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 };

View File

@ -0,0 +1,50 @@
//===-- lib/floatditf.c - integer -> quad-precision 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 di_int to quad-precision conversion for the
// compiler-rt library in the IEEE-754 default round-to-nearest, ties-to-even
// mode.
//
//===----------------------------------------------------------------------===//
#define QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
COMPILER_RT_ABI fp_t __floatditf(di_int a) {
const int aWidth = sizeof a * CHAR_BIT;
// Handle zero as a special case to protect clz
if (a == 0)
return fromRep(0);
// All other cases begin by extracting the sign and absolute value of a
rep_t sign = 0;
du_int aAbs = (du_int)a;
if (a < 0) {
sign = signBit;
aAbs = ~(du_int)a + 1U;
}
// Exponent of (fp_t)a is the width of abs(a).
const int exponent = (aWidth - 1) - __builtin_clzll(aAbs);
rep_t result;
// Shift a into the significand field, rounding if it is a right-shift
const int shift = significandBits - exponent;
result = (rep_t)aAbs << shift ^ implicitBit;
// Insert the exponent
result += (rep_t)(exponent + exponentBias) << significandBits;
// Insert the sign bit and return
return fromRep(result | sign);
}
#endif

View File

@ -0,0 +1,50 @@
//===-- lib/floatsitf.c - integer -> quad-precision 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 integer to quad-precision conversion for the
// compiler-rt library in the IEEE-754 default round-to-nearest, ties-to-even
// mode.
//
//===----------------------------------------------------------------------===//
#define QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
COMPILER_RT_ABI fp_t __floatsitf(int a) {
const int aWidth = sizeof a * CHAR_BIT;
// Handle zero as a special case to protect clz
if (a == 0)
return fromRep(0);
// All other cases begin by extracting the sign and absolute value of a
rep_t sign = 0;
unsigned aAbs = (unsigned)a;
if (a < 0) {
sign = signBit;
aAbs = ~(unsigned)a + 1U;
}
// Exponent of (fp_t)a is the width of abs(a).
const int exponent = (aWidth - 1) - __builtin_clz(aAbs);
rep_t result;
// Shift a into the significand field and clear the implicit bit.
const int shift = significandBits - exponent;
result = (rep_t)aAbs << shift ^ implicitBit;
// Insert the exponent
result += (rep_t)(exponent + exponentBias) << significandBits;
// Insert the sign bit and return
return fromRep(result | sign);
}
#endif

View File

@ -32,9 +32,9 @@ ARM_EABI_FNALIAS(ul2d, floatundidf)
COMPILER_RT_ABI double
__floatundidf(du_int a)
{
static const double twop52 = 0x1.0p52;
static const double twop84 = 0x1.0p84;
static const double twop84_plus_twop52 = 0x1.00000001p84;
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 };

View File

@ -0,0 +1,40 @@
//===-- lib/floatunditf.c - uint -> quad-precision 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 du_int to quad-precision conversion for the
// compiler-rt library in the IEEE-754 default round-to-nearest, ties-to-even
// mode.
//
//===----------------------------------------------------------------------===//
#define QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
COMPILER_RT_ABI fp_t __floatunditf(du_int a) {
const int aWidth = sizeof a * CHAR_BIT;
// Handle zero as a special case to protect clz
if (a == 0) return fromRep(0);
// Exponent of (fp_t)a is the width of abs(a).
const int exponent = (aWidth - 1) - __builtin_clzll(a);
rep_t result;
// Shift a into the significand field and clear the implicit bit.
const int shift = significandBits - exponent;
result = (rep_t)a << shift ^ implicitBit;
// Insert the exponent
result += (rep_t)(exponent + exponentBias) << significandBits;
return fromRep(result);
}
#endif

View File

@ -0,0 +1,40 @@
//===-- lib/floatunsitf.c - uint -> quad-precision 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 unsigned integer to quad-precision conversion for the
// compiler-rt library in the IEEE-754 default round-to-nearest, ties-to-even
// mode.
//
//===----------------------------------------------------------------------===//
#define QUAD_PRECISION
#include "fp_lib.h"
#if defined(CRT_HAS_128BIT) && defined(CRT_LDBL_128BIT)
COMPILER_RT_ABI fp_t __floatunsitf(unsigned int a) {
const int aWidth = sizeof a * CHAR_BIT;
// Handle zero as a special case to protect clz
if (a == 0) return fromRep(0);
// Exponent of (fp_t)a is the width of abs(a).
const int exponent = (aWidth - 1) - __builtin_clz(a);
rep_t result;
// Shift a into the significand field and clear the implicit bit.
const int shift = significandBits - exponent;
result = (rep_t)a << shift ^ implicitBit;
// Insert the exponent
result += (rep_t)(exponent + exponentBias) << significandBits;
return fromRep(result);
}
#endif

View File

@ -14,7 +14,7 @@
#include "fp_lib.h"
static inline fp_t __addXf3__(fp_t a, fp_t b) {
static __inline fp_t __addXf3__(fp_t a, fp_t b) {
rep_t aRep = toRep(a);
rep_t bRep = toRep(b);
const rep_t aAbs = aRep & absMask;

View File

@ -28,7 +28,7 @@ typedef double src_t;
typedef uint64_t src_rep_t;
#define SRC_REP_C UINT64_C
static const int srcSigBits = 52;
static inline int src_rep_t_clz(src_rep_t a) {
static __inline int src_rep_t_clz(src_rep_t a) {
#if defined __LP64__
return __builtin_clzl(a);
#else
@ -39,11 +39,24 @@ static inline int src_rep_t_clz(src_rep_t a) {
#endif
}
#elif defined SRC_HALF
typedef uint16_t src_t;
typedef uint16_t src_rep_t;
#define SRC_REP_C UINT16_C
static const int srcSigBits = 10;
#define src_rep_t_clz __builtin_clz
#else
#error Source should be single precision or double precision!
#error Source should be half, single, or double precision!
#endif //end source precision
#if defined DST_DOUBLE
#if defined DST_SINGLE
typedef float dst_t;
typedef uint32_t dst_rep_t;
#define DST_REP_C UINT32_C
static const int dstSigBits = 23;
#elif defined DST_DOUBLE
typedef double dst_t;
typedef uint64_t dst_rep_t;
#define DST_REP_C UINT64_C
@ -56,18 +69,18 @@ typedef __uint128_t dst_rep_t;
static const int dstSigBits = 112;
#else
#error Destination should be double precision or quad precision!
#error Destination should be single, double, or quad precision!
#endif //end destination precision
// End of specialization parameters. Two helper routines for conversion to and
// from the representation of floating-point data as integer values follow.
static inline src_rep_t srcToRep(src_t x) {
static __inline src_rep_t srcToRep(src_t x) {
const union { src_t f; src_rep_t i; } rep = {.f = x};
return rep.i;
}
static inline dst_t dstFromRep(dst_rep_t x) {
static __inline dst_t dstFromRep(dst_rep_t x) {
const union { dst_t f; dst_rep_t i; } rep = {.i = x};
return rep.f;
}

View File

@ -38,7 +38,7 @@
#include "fp_extend.h"
static inline dst_t __extendXfYf2__(src_t a) {
static __inline dst_t __extendXfYf2__(src_t a) {
// Various constants whose values follow from the type parameters.
// Any reasonable optimizer will fold and propagate all of these.
const int srcBits = sizeof(src_t)*CHAR_BIT;
@ -66,7 +66,9 @@ static inline dst_t __extendXfYf2__(src_t a) {
const src_rep_t sign = aRep & srcSignMask;
dst_rep_t absResult;
if (aAbs - srcMinNormal < srcInfinity - srcMinNormal) {
// If sizeof(src_rep_t) < sizeof(int), the subtraction result is promoted
// to (signed) int. To avoid that, explicitly cast to src_rep_t.
if ((src_rep_t)(aAbs - srcMinNormal) < srcInfinity - srcMinNormal) {
// a is a normal number.
// Extend to the destination type by shifting the significand and
// exponent into the proper position and rebiasing the exponent.

View File

@ -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));
}

View File

@ -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);
}

View File

@ -46,12 +46,12 @@ typedef float fp_t;
#define REP_C UINT32_C
#define significandBits 23
static inline int rep_clz(rep_t a) {
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) {
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;
@ -66,7 +66,7 @@ typedef double fp_t;
#define REP_C UINT64_C
#define significandBits 52
static inline int rep_clz(rep_t a) {
static __inline int rep_clz(rep_t a) {
#if defined __LP64__
return __builtin_clzl(a);
#else
@ -83,7 +83,7 @@ static inline int rep_clz(rep_t a) {
// 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) {
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);
@ -112,7 +112,7 @@ typedef long double fp_t;
// 128-bit integer, we let the constant be casted to 128-bit integer
#define significandBits 112
static inline int rep_clz(rep_t a) {
static __inline int rep_clz(rep_t a) {
const union
{
__uint128_t ll;
@ -148,7 +148,7 @@ static inline int rep_clz(rep_t a) {
// 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) {
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);
@ -228,28 +228,28 @@ static inline void wideMultiply(rep_t a, rep_t b, rep_t *hi, rep_t *lo) {
#define quietBit (implicitBit >> 1)
#define qnanRep (exponentMask | quietBit)
static inline rep_t toRep(fp_t x) {
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) {
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) {
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) {
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) {
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;

View File

@ -14,7 +14,7 @@
#include "fp_lib.h"
static inline fp_t __mulXf3__(fp_t a, fp_t b) {
static __inline fp_t __mulXf3__(fp_t a, fp_t b) {
const unsigned int aExponent = toRep(a) >> significandBits & maxExponent;
const unsigned int bExponent = toRep(b) >> significandBits & maxExponent;
const rep_t productSign = (toRep(a) ^ toRep(b)) & signBit;

View File

@ -16,7 +16,13 @@
#include "int_lib.h"
#if defined SRC_DOUBLE
#if defined SRC_SINGLE
typedef float src_t;
typedef uint32_t src_rep_t;
#define SRC_REP_C UINT32_C
static const int srcSigBits = 23;
#elif defined SRC_DOUBLE
typedef double src_t;
typedef uint64_t src_rep_t;
#define SRC_REP_C UINT64_C
@ -44,6 +50,12 @@ typedef uint32_t dst_rep_t;
#define DST_REP_C UINT32_C
static const int dstSigBits = 23;
#elif defined DST_HALF
typedef uint16_t dst_t;
typedef uint16_t dst_rep_t;
#define DST_REP_C UINT16_C
static const int dstSigBits = 10;
#else
#error Destination should be single precision or double precision!
#endif //end destination precision
@ -51,12 +63,12 @@ static const int dstSigBits = 23;
// End of specialization parameters. Two helper routines for conversion to and
// from the representation of floating-point data as integer values follow.
static inline src_rep_t srcToRep(src_t x) {
static __inline src_rep_t srcToRep(src_t x) {
const union { src_t f; src_rep_t i; } rep = {.f = x};
return rep.i;
}
static inline dst_t dstFromRep(dst_rep_t x) {
static __inline dst_t dstFromRep(dst_rep_t x) {
const union { dst_t f; dst_rep_t i; } rep = {.i = x};
return rep.f;
}

View File

@ -39,7 +39,7 @@
#include "fp_trunc.h"
static inline dst_t __truncXfYf2__(src_t a) {
static __inline dst_t __truncXfYf2__(src_t a) {
// Various constants whose values follow from the type parameters.
// Any reasonable optimizer will fold and propagate all of these.
const int srcBits = sizeof(src_t)*CHAR_BIT;
@ -99,7 +99,7 @@ static inline dst_t __truncXfYf2__(src_t a) {
absResult |= dstQNaN;
absResult |= ((aAbs & srcNaNCode) >> (srcSigBits - dstSigBits)) & dstNaNCode;
}
else if (aAbs > overflow) {
else if (aAbs >= overflow) {
// a overflows to infinity.
absResult = (dst_rep_t)dstInfExp << dstSigBits;
}

View File

@ -0,0 +1,34 @@
// This file is dual licensed under the MIT and the University of Illinois Open
// Source Licenses. See LICENSE.TXT for details.
#include "../assembly.h"
// _chkstk routine
// This routine is windows specific
// http://msdn.microsoft.com/en-us/library/ms648426.aspx
#ifdef __i386__
.text
.balign 4
DEFINE_COMPILERRT_FUNCTION(__chkstk_ms)
push %ecx
push %eax
cmp $0x1000,%eax
lea 12(%esp),%ecx
jb 1f
2:
sub $0x1000,%ecx
test %ecx,(%ecx)
sub $0x1000,%eax
cmp $0x1000,%eax
ja 2b
1:
sub %eax,%ecx
test %ecx,(%ecx)
pop %eax
pop %ecx
ret
END_COMPILERRT_FUNCTION(__chkstk_ms)
#endif // __i386__

View File

@ -0,0 +1,40 @@
// This file is dual licensed under the MIT and the University of Illinois Open
// Source Licenses. See LICENSE.TXT for details.
#include "../assembly.h"
#ifdef __i386__
// _chkstk (_alloca) routine - probe stack between %esp and (%esp-%eax) in 4k increments,
// then decrement %esp by %eax. Preserves all registers except %esp and flags.
// This routine is windows specific
// http://msdn.microsoft.com/en-us/library/ms648426.aspx
.text
.balign 4
DEFINE_COMPILERRT_FUNCTION(_alloca) // _chkstk and _alloca are the same function
DEFINE_COMPILERRT_FUNCTION(__chkstk)
push %ecx
cmp $0x1000,%eax
lea 8(%esp),%ecx // esp before calling this routine -> ecx
jb 1f
2:
sub $0x1000,%ecx
test %ecx,(%ecx)
sub $0x1000,%eax
cmp $0x1000,%eax
ja 2b
1:
sub %eax,%ecx
test %ecx,(%ecx)
lea 4(%esp),%eax // load pointer to the return address into eax
mov %ecx,%esp // install the new top of stack pointer into esp
mov -4(%eax),%ecx // restore ecx
push (%eax) // push return address onto the stack
sub %esp,%eax // restore the original value in eax
ret
END_COMPILERRT_FUNCTION(__chkstk)
END_COMPILERRT_FUNCTION(_alloca)
#endif // __i386__

View File

@ -7,13 +7,7 @@
#ifdef __i386__
#if defined(__APPLE__)
.const
#elif defined(__ELF__)
.section .rodata
#else
.section .rdata,"rd"
#endif
CONST_SECTION
.balign 16
twop52:

View File

@ -17,13 +17,7 @@
#ifdef __i386__
#if defined(__APPLE__)
.const
#elif defined(__ELF__)
.section .rodata
#else
.section .rdata,"rd"
#endif
CONST_SECTION
.balign 16
twop52:

View File

@ -18,7 +18,7 @@
#ifdef __i386__
.const
CONST_SECTION
.balign 3
.quad 0x43f0000000000000
@ -52,13 +52,7 @@ END_COMPILERRT_FUNCTION(__floatundisf)
#ifdef __i386__
#if defined(__APPLE__)
.const
#elif defined(__ELF__)
.section .rodata
#else
.section .rdata,"rd"
#endif
CONST_SECTION
.balign 16
twop52:

View File

@ -7,13 +7,7 @@
#ifdef __i386__
#if defined(__APPLE__)
.const
#elif defined(__ELF__)
.section .rodata
#else
.section .rdata,"rd"
#endif
CONST_SECTION
.balign 16
twop52:

View File

@ -16,6 +16,20 @@
#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>
@ -33,7 +47,8 @@
/* .. */
#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__DragonFly__) || defined(__minix)
#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__DragonFly__) || \
defined(__minix)
#include <sys/endian.h>
#if _BYTE_ORDER == _BIG_ENDIAN
@ -61,7 +76,8 @@
/* .. */
/* Mac OSX has __BIG_ENDIAN__ or __LITTLE_ENDIAN__ automatically set by the compiler (at least with GCC) */
/* 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__
@ -82,19 +98,6 @@
/* .. */
#if defined(__linux__)
#include <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 /* GNU/Linux */
#if defined(_WIN32)
#define _YUGA_LITTLE_ENDIAN 1
@ -102,6 +105,8 @@
#endif /* Windows */
#endif /* Clang or GCC. */
/* . */
#if !defined(_YUGA_LITTLE_ENDIAN) || !defined(_YUGA_BIG_ENDIAN)

View File

@ -20,21 +20,38 @@
/* Assumption: Right shift of signed negative is arithmetic shift. */
/* Assumption: Endianness is little or big (not mixed). */
/* ABI macro definitions */
#if defined(__ELF__)
#define FNALIAS(alias_name, original_name) \
void alias_name() __attribute__((alias(#original_name)))
#else
#define FNALIAS(alias, name) _Pragma("GCC error(\"alias unsupported on this file format\")")
#endif
/*
* TODO define this appropriately for targets that require explicit export
* declarations (i.e. Windows)
*/
#define COMPILER_RT_EXPORT
/* ABI macro definitions */
#if __ARM_EABI__
# define ARM_EABI_FNALIAS(aeabi_name, name) \
void __aeabi_##aeabi_name() __attribute__((alias("__" #name)));
# define COMPILER_RT_ABI COMPILER_RT_EXPORT __attribute__((pcs("aapcs")))
# define COMPILER_RT_ABI __attribute__((pcs("aapcs")))
#else
# define ARM_EABI_FNALIAS(aeabi_name, name)
# define COMPILER_RT_ABI COMPILER_RT_EXPORT
# if defined(__arm__) && defined(_WIN32) && (!defined(_MSC_VER) || defined(__clang__))
# define COMPILER_RT_ABI __attribute__((pcs("aapcs")))
# else
# define COMPILER_RT_ABI
# endif
#endif
#ifdef _MSC_VER
#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))
@ -73,4 +90,44 @@ 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) {
uint32_t trailing_zero = 0;
if (_BitScanForward(&trailing_zero, value))
return trailing_zero;
return 32;
}
uint32_t __inline __builtin_clz(uint32_t value) {
uint32_t 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) {
uint32_t 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 */

View File

@ -25,43 +25,90 @@
# 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
#define crt_isinf(x) __builtin_isinf((x))
#define crt_isnan(x) __builtin_isnan((x))
#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))
#else
#elif defined(__GNUC__)
# define crt_isfinite(x) \
__extension__(({ \
__typeof((x)) x_ = (x); \
!crt_isinf(x_) && !crt_isnan(x_); \
}))
#endif
#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_logb(x) logb((x))
#define crt_logbf(x) logbf((x))
#define crt_logbl(x) logbl((x))
#else
#define crt_logb(x) __builtin_logb((x))
#define crt_logbf(x) __builtin_logbf((x))
#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 */

View File

@ -20,6 +20,10 @@
#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;
@ -56,7 +60,9 @@ typedef union
}s;
} udwords;
#if __LP64__
/* MIPS64 issue: PR 20098 */
#if (defined(__LP64__) || defined(__wasm__)) && \
!(defined(__mips__) && defined(__clang__))
#define CRT_HAS_128BIT
#endif
@ -94,14 +100,14 @@ typedef union
}s;
} utwords;
static inline ti_int make_ti(di_int h, di_int l) {
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) {
static __inline tu_int make_tu(du_int h, du_int l) {
utwords r;
r.s.high = h;
r.s.low = l;
@ -139,5 +145,22 @@ typedef union
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 */

Some files were not shown because too many files have changed in this diff Show More