// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// RUN: %clang_cc1 -no-opaque-pointers -ffreestanding -triple armv8a-none-eabi -target-feature +crc -target-feature +dsp -O0 -disable-O0-optnone -S -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s -check-prefixes=ARM,AArch32
// RUN: %clang_cc1 -no-opaque-pointers -ffreestanding -Wno-error=implicit-function-declaration -triple aarch64-none-eabi -target-feature +neon -target-feature +crc -target-feature +crypto -O0 -disable-O0-optnone -S -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s -check-prefixes=ARM,AArch64
// RUN: %clang_cc1 -no-opaque-pointers -ffreestanding -triple aarch64-none-eabi -target-feature +v8.3a -O0 -disable-O0-optnone -S -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s -check-prefixes=ARM,AArch64,AArch6483
// RUN: %clang_cc1 -no-opaque-pointers -ffreestanding -triple aarch64-none-eabi -target-feature +v8.5a -target-feature +rand -O0 -disable-O0-optnone -S -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s -check-prefixes=ARM,AArch64,AArch6483,AArch6485

#include <arm_acle.h>

// REQUIRES: arm-registered-target,aarch64-registered-target

/* 8 SYNCHRONIZATION, BARRIER AND HINT INTRINSICS */
/* 8.3 Memory Barriers */

// AArch32-LABEL: @test_dmb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.dmb(i32 1)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_dmb(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.dmb(i32 1)
// AArch64-NEXT:    ret void
//
void test_dmb(void) {
  __dmb(1);
}

// AArch32-LABEL: @test_dsb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.dsb(i32 2)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_dsb(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.dsb(i32 2)
// AArch64-NEXT:    ret void
//
void test_dsb(void) {
  __dsb(2);
}

// AArch32-LABEL: @test_isb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.isb(i32 3)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_isb(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.isb(i32 3)
// AArch64-NEXT:    ret void
//
void test_isb(void) {
  __isb(3);
}

/* 8.4 Hints */
// AArch32-LABEL: @test_yield(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.hint(i32 1)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_yield(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.hint(i32 1)
// AArch64-NEXT:    ret void
//
void test_yield(void) {
  __yield();
}

// AArch32-LABEL: @test_wfe(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.hint(i32 2)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_wfe(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.hint(i32 2)
// AArch64-NEXT:    ret void
//
void test_wfe(void) {
  __wfe();
}

// AArch32-LABEL: @test_wfi(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.hint(i32 3)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_wfi(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.hint(i32 3)
// AArch64-NEXT:    ret void
//
void test_wfi(void) {
  __wfi();
}

// AArch32-LABEL: @test_sev(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.hint(i32 4)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_sev(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.hint(i32 4)
// AArch64-NEXT:    ret void
//
void test_sev(void) {
  __sev();
}

// AArch32-LABEL: @test_sevl(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.hint(i32 5)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_sevl(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.hint(i32 5)
// AArch64-NEXT:    ret void
//
void test_sevl(void) {
  __sevl();
}

#if __ARM_32BIT_STATE
// AArch32-LABEL: @test_dbg(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.dbg(i32 0)
// AArch32-NEXT:    ret void
//
void test_dbg(void) {
  __dbg(0);
}
#endif

/* 8.5 Swap */
// AArch32-LABEL: @test_swp(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = bitcast i8* [[P:%.*]] to i32*
// AArch32-NEXT:    br label [[DO_BODY_I:%.*]]
// AArch32:       do.body.i:
// AArch32-NEXT:    [[LDREX_I:%.*]] = call i32 @llvm.arm.ldrex.p0i32(i32* elementtype(i32) [[TMP0]])
// AArch32-NEXT:    [[STREX_I:%.*]] = call i32 @llvm.arm.strex.p0i32(i32 [[X:%.*]], i32* elementtype(i32) [[TMP0]])
// AArch32-NEXT:    [[TOBOOL_I:%.*]] = icmp ne i32 [[STREX_I]], 0
// AArch32-NEXT:    br i1 [[TOBOOL_I]], label [[DO_BODY_I]], label [[__SWP_EXIT:%.*]], !llvm.loop [[LOOP7:![0-9]+]]
// AArch32:       __swp.exit:
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_swp(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = bitcast i8* [[P:%.*]] to i32*
// AArch64-NEXT:    br label [[DO_BODY_I:%.*]]
// AArch64:       do.body.i:
// AArch64-NEXT:    [[LDXR_I:%.*]] = call i64 @llvm.aarch64.ldxr.p0i32(i32* elementtype(i32) [[TMP0]])
// AArch64-NEXT:    [[TMP1:%.*]] = trunc i64 [[LDXR_I]] to i32
// AArch64-NEXT:    [[TMP2:%.*]] = zext i32 [[X:%.*]] to i64
// AArch64-NEXT:    [[STXR_I:%.*]] = call i32 @llvm.aarch64.stxr.p0i32(i64 [[TMP2]], i32* elementtype(i32) [[TMP0]])
// AArch64-NEXT:    [[TOBOOL_I:%.*]] = icmp ne i32 [[STXR_I]], 0
// AArch64-NEXT:    br i1 [[TOBOOL_I]], label [[DO_BODY_I]], label [[__SWP_EXIT:%.*]], !llvm.loop [[LOOP6:![0-9]+]]
// AArch64:       __swp.exit:
// AArch64-NEXT:    ret void
//
void test_swp(uint32_t x, volatile void *p) {
  __swp(x, p);
}

/* 8.6 Memory prefetch intrinsics */
/* 8.6.1 Data prefetch */
// ARM-LABEL: @test_pld(
// ARM-NEXT:  entry:
// ARM-NEXT:    call void @llvm.prefetch.p0i8(i8* null, i32 0, i32 3, i32 1)
// ARM-NEXT:    ret void
//
void test_pld() {
  __pld(0);
}

// AArch32-LABEL: @test_pldx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.prefetch.p0i8(i8* null, i32 1, i32 3, i32 1)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_pldx(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.prefetch.p0i8(i8* null, i32 1, i32 1, i32 1)
// AArch64-NEXT:    ret void
//
void test_pldx() {
  __pldx(1, 2, 0, 0);
}

/* 8.6.2 Instruction prefetch */
// ARM-LABEL: @test_pli(
// ARM-NEXT:  entry:
// ARM-NEXT:    call void @llvm.prefetch.p0i8(i8* null, i32 0, i32 3, i32 0)
// ARM-NEXT:    ret void
//
void test_pli() {
  __pli(0);
}

// AArch32-LABEL: @test_plix(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.prefetch.p0i8(i8* null, i32 0, i32 3, i32 0)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_plix(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.prefetch.p0i8(i8* null, i32 0, i32 1, i32 0)
// AArch64-NEXT:    ret void
//
void test_plix() {
  __plix(2, 0, 0);
}

/* 8.7 NOP */
// AArch32-LABEL: @test_nop(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.arm.hint(i32 0)
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_nop(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.aarch64.hint(i32 0)
// AArch64-NEXT:    ret void
//
void test_nop(void) {
  __nop();
}

/* 9 DATA-PROCESSING INTRINSICS */

/* 9.2 Miscellaneous data-processing intrinsics */
// ARM-LABEL: @test_ror(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[REM_I:%.*]] = urem i32 [[Y:%.*]], 32
// ARM-NEXT:    [[CMP_I:%.*]] = icmp eq i32 [[REM_I]], 0
// ARM-NEXT:    br i1 [[CMP_I]], label [[IF_THEN_I:%.*]], label [[IF_END_I:%.*]]
// ARM:       if.then.i:
// ARM-NEXT:    br label [[__ROR_EXIT:%.*]]
// ARM:       if.end.i:
// ARM-NEXT:    [[SHR_I:%.*]] = lshr i32 [[X:%.*]], [[REM_I]]
// ARM-NEXT:    [[SUB_I:%.*]] = sub i32 32, [[REM_I]]
// ARM-NEXT:    [[SHL_I:%.*]] = shl i32 [[X]], [[SUB_I]]
// ARM-NEXT:    [[OR_I:%.*]] = or i32 [[SHR_I]], [[SHL_I]]
// ARM-NEXT:    br label [[__ROR_EXIT]]
// ARM:       __ror.exit:
// ARM-NEXT:    [[RETVAL_I_0:%.*]] = phi i32 [ [[X]], [[IF_THEN_I]] ], [ [[OR_I]], [[IF_END_I]] ]
// ARM-NEXT:    ret i32 [[RETVAL_I_0]]
//
uint32_t test_ror(uint32_t x, uint32_t y) {
  return __ror(x, y);
}

// AArch32-LABEL: @test_rorl(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[REM_I_I:%.*]] = urem i32 [[Y:%.*]], 32
// AArch32-NEXT:    [[CMP_I_I:%.*]] = icmp eq i32 [[REM_I_I]], 0
// AArch32-NEXT:    br i1 [[CMP_I_I]], label [[IF_THEN_I_I:%.*]], label [[IF_END_I_I:%.*]]
// AArch32:       if.then.i.i:
// AArch32-NEXT:    br label [[__RORL_EXIT:%.*]]
// AArch32:       if.end.i.i:
// AArch32-NEXT:    [[SHR_I_I:%.*]] = lshr i32 [[X:%.*]], [[REM_I_I]]
// AArch32-NEXT:    [[SUB_I_I:%.*]] = sub i32 32, [[REM_I_I]]
// AArch32-NEXT:    [[SHL_I_I:%.*]] = shl i32 [[X]], [[SUB_I_I]]
// AArch32-NEXT:    [[OR_I_I:%.*]] = or i32 [[SHR_I_I]], [[SHL_I_I]]
// AArch32-NEXT:    br label [[__RORL_EXIT]]
// AArch32:       __rorl.exit:
// AArch32-NEXT:    [[RETVAL_I_I_0:%.*]] = phi i32 [ [[X]], [[IF_THEN_I_I]] ], [ [[OR_I_I]], [[IF_END_I_I]] ]
// AArch32-NEXT:    ret i32 [[RETVAL_I_I_0]]
//
// AArch64-LABEL: @test_rorl(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[REM_I:%.*]] = urem i32 [[Y:%.*]], 64
// AArch64-NEXT:    [[CMP_I:%.*]] = icmp eq i32 [[REM_I]], 0
// AArch64-NEXT:    br i1 [[CMP_I]], label [[IF_THEN_I:%.*]], label [[IF_END_I:%.*]]
// AArch64:       if.then.i:
// AArch64-NEXT:    br label [[__RORLL_EXIT:%.*]]
// AArch64:       if.end.i:
// AArch64-NEXT:    [[SH_PROM_I:%.*]] = zext i32 [[REM_I]] to i64
// AArch64-NEXT:    [[SHR_I:%.*]] = lshr i64 [[X:%.*]], [[SH_PROM_I]]
// AArch64-NEXT:    [[SUB_I:%.*]] = sub i32 64, [[REM_I]]
// AArch64-NEXT:    [[SH_PROM1_I:%.*]] = zext i32 [[SUB_I]] to i64
// AArch64-NEXT:    [[SHL_I:%.*]] = shl i64 [[X]], [[SH_PROM1_I]]
// AArch64-NEXT:    [[OR_I:%.*]] = or i64 [[SHR_I]], [[SHL_I]]
// AArch64-NEXT:    br label [[__RORLL_EXIT]]
// AArch64:       __rorll.exit:
// AArch64-NEXT:    [[RETVAL_I_0:%.*]] = phi i64 [ [[X]], [[IF_THEN_I]] ], [ [[OR_I]], [[IF_END_I]] ]
// AArch64-NEXT:    ret i64 [[RETVAL_I_0]]
//
unsigned long test_rorl(unsigned long x, uint32_t y) {
  return __rorl(x, y);
}

// ARM-LABEL: @test_rorll(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[REM_I:%.*]] = urem i32 [[Y:%.*]], 64
// ARM-NEXT:    [[CMP_I:%.*]] = icmp eq i32 [[REM_I]], 0
// ARM-NEXT:    br i1 [[CMP_I]], label [[IF_THEN_I:%.*]], label [[IF_END_I:%.*]]
// ARM:       if.then.i:
// ARM-NEXT:    br label [[__RORLL_EXIT:%.*]]
// ARM:       if.end.i:
// ARM-NEXT:    [[SH_PROM_I:%.*]] = zext i32 [[REM_I]] to i64
// ARM-NEXT:    [[SHR_I:%.*]] = lshr i64 [[X:%.*]], [[SH_PROM_I]]
// ARM-NEXT:    [[SUB_I:%.*]] = sub i32 64, [[REM_I]]
// ARM-NEXT:    [[SH_PROM1_I:%.*]] = zext i32 [[SUB_I]] to i64
// ARM-NEXT:    [[SHL_I:%.*]] = shl i64 [[X]], [[SH_PROM1_I]]
// ARM-NEXT:    [[OR_I:%.*]] = or i64 [[SHR_I]], [[SHL_I]]
// ARM-NEXT:    br label [[__RORLL_EXIT]]
// ARM:       __rorll.exit:
// ARM-NEXT:    [[RETVAL_I_0:%.*]] = phi i64 [ [[X]], [[IF_THEN_I]] ], [ [[OR_I]], [[IF_END_I]] ]
// ARM-NEXT:    ret i64 [[RETVAL_I_0]]
//
uint64_t test_rorll(uint64_t x, uint32_t y) {
  return __rorll(x, y);
}

// ARM-LABEL: @test_clz(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[TMP0:%.*]] = call i32 @llvm.ctlz.i32(i32 [[T:%.*]], i1 false)
// ARM-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_clz(uint32_t t) {
  return __clz(t);
}

// AArch32-LABEL: @test_clzl(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.ctlz.i32(i32 [[T:%.*]], i1 false)
// AArch32-NEXT:    ret i32 [[TMP0]]
//
// AArch64-LABEL: @test_clzl(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.ctlz.i64(i64 [[T:%.*]], i1 false)
// AArch64-NEXT:    [[CAST_I:%.*]] = trunc i64 [[TMP0]] to i32
// AArch64-NEXT:    [[CONV_I:%.*]] = sext i32 [[CAST_I]] to i64
// AArch64-NEXT:    ret i64 [[CONV_I]]
//
long test_clzl(long t) {
  return __clzl(t);
}

// ARM-LABEL: @test_clzll(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[TMP0:%.*]] = call i64 @llvm.ctlz.i64(i64 [[T:%.*]], i1 false)
// ARM-NEXT:    [[CAST_I:%.*]] = trunc i64 [[TMP0]] to i32
// ARM-NEXT:    [[CONV_I:%.*]] = sext i32 [[CAST_I]] to i64
// ARM-NEXT:    ret i64 [[CONV_I]]
//
uint64_t test_clzll(uint64_t t) {
  return __clzll(t);
}

// AArch32-LABEL: @test_cls(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[CLS_I:%.*]] = call i32 @llvm.arm.cls(i32 [[T:%.*]])
// AArch32-NEXT:    ret i32 [[CLS_I]]
//
// AArch64-LABEL: @test_cls(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[CLS_I:%.*]] = call i32 @llvm.aarch64.cls(i32 [[T:%.*]])
// AArch64-NEXT:    ret i32 [[CLS_I]]
//
unsigned test_cls(uint32_t t) {
  return __cls(t);
}

// AArch32-LABEL: @test_clsl(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[CLS_I:%.*]] = call i32 @llvm.arm.cls(i32 [[T:%.*]])
// AArch32-NEXT:    ret i32 [[CLS_I]]
//
// AArch64-LABEL: @test_clsl(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[CLS_I:%.*]] = call i32 @llvm.aarch64.cls64(i64 [[T:%.*]])
// AArch64-NEXT:    ret i32 [[CLS_I]]
//
unsigned test_clsl(unsigned long t) {
  return __clsl(t);
}

// AArch32-LABEL: @test_clsll(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[CLS_I:%.*]] = call i32 @llvm.arm.cls64(i64 [[T:%.*]])
// AArch32-NEXT:    ret i32 [[CLS_I]]
//
// AArch64-LABEL: @test_clsll(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[CLS_I:%.*]] = call i32 @llvm.aarch64.cls64(i64 [[T:%.*]])
// AArch64-NEXT:    ret i32 [[CLS_I]]
//
unsigned test_clsll(uint64_t t) {
  return __clsll(t);
}

// ARM-LABEL: @test_rev(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[TMP0:%.*]] = call i32 @llvm.bswap.i32(i32 [[T:%.*]])
// ARM-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_rev(uint32_t t) {
  return __rev(t);
}

// AArch32-LABEL: @test_revl(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.bswap.i32(i32 [[T:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
// AArch64-LABEL: @test_revl(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.bswap.i64(i64 [[T:%.*]])
// AArch64-NEXT:    ret i64 [[TMP0]]
//
long test_revl(long t) {
  return __revl(t);
}

// ARM-LABEL: @test_revll(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[TMP0:%.*]] = call i64 @llvm.bswap.i64(i64 [[T:%.*]])
// ARM-NEXT:    ret i64 [[TMP0]]
//
uint64_t test_revll(uint64_t t) {
  return __revll(t);
}

// ARM-LABEL: @test_rev16(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[TMP0:%.*]] = call i32 @llvm.bswap.i32(i32 [[T:%.*]])
// ARM-NEXT:    [[REM_I_I:%.*]] = urem i32 16, 32
// ARM-NEXT:    [[CMP_I_I:%.*]] = icmp eq i32 [[REM_I_I]], 0
// ARM-NEXT:    br i1 [[CMP_I_I]], label [[IF_THEN_I_I:%.*]], label [[IF_END_I_I:%.*]]
// ARM:       if.then.i.i:
// ARM-NEXT:    br label [[__REV16_EXIT:%.*]]
// ARM:       if.end.i.i:
// ARM-NEXT:    [[SHR_I_I:%.*]] = lshr i32 [[TMP0]], [[REM_I_I]]
// ARM-NEXT:    [[SUB_I_I:%.*]] = sub i32 32, [[REM_I_I]]
// ARM-NEXT:    [[SHL_I_I:%.*]] = shl i32 [[TMP0]], [[SUB_I_I]]
// ARM-NEXT:    [[OR_I_I:%.*]] = or i32 [[SHR_I_I]], [[SHL_I_I]]
// ARM-NEXT:    br label [[__REV16_EXIT]]
// ARM:       __rev16.exit:
// ARM-NEXT:    [[RETVAL_I_I_0:%.*]] = phi i32 [ [[TMP0]], [[IF_THEN_I_I]] ], [ [[OR_I_I]], [[IF_END_I_I]] ]
// ARM-NEXT:    ret i32 [[RETVAL_I_I_0]]
//
uint32_t test_rev16(uint32_t t) {
  return __rev16(t);
}

// AArch32-LABEL: @test_rev16l(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.bswap.i32(i32 [[T:%.*]])
// AArch32-NEXT:    [[REM_I_I_I:%.*]] = urem i32 16, 32
// AArch32-NEXT:    [[CMP_I_I_I:%.*]] = icmp eq i32 [[REM_I_I_I]], 0
// AArch32-NEXT:    br i1 [[CMP_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[IF_END_I_I_I:%.*]]
// AArch32:       if.then.i.i.i:
// AArch32-NEXT:    br label [[__REV16L_EXIT:%.*]]
// AArch32:       if.end.i.i.i:
// AArch32-NEXT:    [[SHR_I_I_I:%.*]] = lshr i32 [[TMP0]], [[REM_I_I_I]]
// AArch32-NEXT:    [[SUB_I_I_I:%.*]] = sub i32 32, [[REM_I_I_I]]
// AArch32-NEXT:    [[SHL_I_I_I:%.*]] = shl i32 [[TMP0]], [[SUB_I_I_I]]
// AArch32-NEXT:    [[OR_I_I_I:%.*]] = or i32 [[SHR_I_I_I]], [[SHL_I_I_I]]
// AArch32-NEXT:    br label [[__REV16L_EXIT]]
// AArch32:       __rev16l.exit:
// AArch32-NEXT:    [[RETVAL_I_I_I_0:%.*]] = phi i32 [ [[TMP0]], [[IF_THEN_I_I_I]] ], [ [[OR_I_I_I]], [[IF_END_I_I_I]] ]
// AArch32-NEXT:    ret i32 [[RETVAL_I_I_I_0]]
//
// AArch64-LABEL: @test_rev16l(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[SHR_I:%.*]] = lshr i64 [[T:%.*]], 32
// AArch64-NEXT:    [[CONV_I:%.*]] = trunc i64 [[SHR_I]] to i32
// AArch64-NEXT:    [[TMP0:%.*]] = call i32 @llvm.bswap.i32(i32 [[CONV_I]])
// AArch64-NEXT:    [[REM_I_I10_I:%.*]] = urem i32 16, 32
// AArch64-NEXT:    [[CMP_I_I11_I:%.*]] = icmp eq i32 [[REM_I_I10_I]], 0
// AArch64-NEXT:    br i1 [[CMP_I_I11_I]], label [[IF_THEN_I_I12_I:%.*]], label [[IF_END_I_I17_I:%.*]]
// AArch64:       if.then.i.i12.i:
// AArch64-NEXT:    br label [[__REV16_EXIT18_I:%.*]]
// AArch64:       if.end.i.i17.i:
// AArch64-NEXT:    [[SHR_I_I13_I:%.*]] = lshr i32 [[TMP0]], [[REM_I_I10_I]]
// AArch64-NEXT:    [[SUB_I_I14_I:%.*]] = sub i32 32, [[REM_I_I10_I]]
// AArch64-NEXT:    [[SHL_I_I15_I:%.*]] = shl i32 [[TMP0]], [[SUB_I_I14_I]]
// AArch64-NEXT:    [[OR_I_I16_I:%.*]] = or i32 [[SHR_I_I13_I]], [[SHL_I_I15_I]]
// AArch64-NEXT:    br label [[__REV16_EXIT18_I]]
// AArch64:       __rev16.exit18.i:
// AArch64-NEXT:    [[RETVAL_I_I6_I_0:%.*]] = phi i32 [ [[TMP0]], [[IF_THEN_I_I12_I]] ], [ [[OR_I_I16_I]], [[IF_END_I_I17_I]] ]
// AArch64-NEXT:    [[CONV1_I:%.*]] = zext i32 [[RETVAL_I_I6_I_0]] to i64
// AArch64-NEXT:    [[SHL_I:%.*]] = shl i64 [[CONV1_I]], 32
// AArch64-NEXT:    [[CONV2_I:%.*]] = trunc i64 [[T]] to i32
// AArch64-NEXT:    [[TMP1:%.*]] = call i32 @llvm.bswap.i32(i32 [[CONV2_I]])
// AArch64-NEXT:    [[REM_I_I_I:%.*]] = urem i32 16, 32
// AArch64-NEXT:    [[CMP_I_I_I:%.*]] = icmp eq i32 [[REM_I_I_I]], 0
// AArch64-NEXT:    br i1 [[CMP_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[IF_END_I_I_I:%.*]]
// AArch64:       if.then.i.i.i:
// AArch64-NEXT:    br label [[__REV16LL_EXIT:%.*]]
// AArch64:       if.end.i.i.i:
// AArch64-NEXT:    [[SHR_I_I_I:%.*]] = lshr i32 [[TMP1]], [[REM_I_I_I]]
// AArch64-NEXT:    [[SUB_I_I_I:%.*]] = sub i32 32, [[REM_I_I_I]]
// AArch64-NEXT:    [[SHL_I_I_I:%.*]] = shl i32 [[TMP1]], [[SUB_I_I_I]]
// AArch64-NEXT:    [[OR_I_I_I:%.*]] = or i32 [[SHR_I_I_I]], [[SHL_I_I_I]]
// AArch64-NEXT:    br label [[__REV16LL_EXIT]]
// AArch64:       __rev16ll.exit:
// AArch64-NEXT:    [[RETVAL_I_I_I_0:%.*]] = phi i32 [ [[TMP1]], [[IF_THEN_I_I_I]] ], [ [[OR_I_I_I]], [[IF_END_I_I_I]] ]
// AArch64-NEXT:    [[CONV4_I:%.*]] = zext i32 [[RETVAL_I_I_I_0]] to i64
// AArch64-NEXT:    [[OR_I:%.*]] = or i64 [[SHL_I]], [[CONV4_I]]
// AArch64-NEXT:    ret i64 [[OR_I]]
//
long test_rev16l(long t) {
  return __rev16l(t);
}

// ARM-LABEL: @test_rev16ll(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[SHR_I:%.*]] = lshr i64 [[T:%.*]], 32
// ARM-NEXT:    [[CONV_I:%.*]] = trunc i64 [[SHR_I]] to i32
// ARM-NEXT:    [[TMP0:%.*]] = call i32 @llvm.bswap.i32(i32 [[CONV_I]])
// ARM-NEXT:    [[REM_I_I10_I:%.*]] = urem i32 16, 32
// ARM-NEXT:    [[CMP_I_I11_I:%.*]] = icmp eq i32 [[REM_I_I10_I]], 0
// ARM-NEXT:    br i1 [[CMP_I_I11_I]], label [[IF_THEN_I_I12_I:%.*]], label [[IF_END_I_I17_I:%.*]]
// ARM:       if.then.i.i12.i:
// ARM-NEXT:    br label [[__REV16_EXIT18_I:%.*]]
// ARM:       if.end.i.i17.i:
// ARM-NEXT:    [[SHR_I_I13_I:%.*]] = lshr i32 [[TMP0]], [[REM_I_I10_I]]
// ARM-NEXT:    [[SUB_I_I14_I:%.*]] = sub i32 32, [[REM_I_I10_I]]
// ARM-NEXT:    [[SHL_I_I15_I:%.*]] = shl i32 [[TMP0]], [[SUB_I_I14_I]]
// ARM-NEXT:    [[OR_I_I16_I:%.*]] = or i32 [[SHR_I_I13_I]], [[SHL_I_I15_I]]
// ARM-NEXT:    br label [[__REV16_EXIT18_I]]
// ARM:       __rev16.exit18.i:
// ARM-NEXT:    [[RETVAL_I_I6_I_0:%.*]] = phi i32 [ [[TMP0]], [[IF_THEN_I_I12_I]] ], [ [[OR_I_I16_I]], [[IF_END_I_I17_I]] ]
// ARM-NEXT:    [[CONV1_I:%.*]] = zext i32 [[RETVAL_I_I6_I_0]] to i64
// ARM-NEXT:    [[SHL_I:%.*]] = shl i64 [[CONV1_I]], 32
// ARM-NEXT:    [[CONV2_I:%.*]] = trunc i64 [[T]] to i32
// ARM-NEXT:    [[TMP1:%.*]] = call i32 @llvm.bswap.i32(i32 [[CONV2_I]])
// ARM-NEXT:    [[REM_I_I_I:%.*]] = urem i32 16, 32
// ARM-NEXT:    [[CMP_I_I_I:%.*]] = icmp eq i32 [[REM_I_I_I]], 0
// ARM-NEXT:    br i1 [[CMP_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[IF_END_I_I_I:%.*]]
// ARM:       if.then.i.i.i:
// ARM-NEXT:    br label [[__REV16LL_EXIT:%.*]]
// ARM:       if.end.i.i.i:
// ARM-NEXT:    [[SHR_I_I_I:%.*]] = lshr i32 [[TMP1]], [[REM_I_I_I]]
// ARM-NEXT:    [[SUB_I_I_I:%.*]] = sub i32 32, [[REM_I_I_I]]
// ARM-NEXT:    [[SHL_I_I_I:%.*]] = shl i32 [[TMP1]], [[SUB_I_I_I]]
// ARM-NEXT:    [[OR_I_I_I:%.*]] = or i32 [[SHR_I_I_I]], [[SHL_I_I_I]]
// ARM-NEXT:    br label [[__REV16LL_EXIT]]
// ARM:       __rev16ll.exit:
// ARM-NEXT:    [[RETVAL_I_I_I_0:%.*]] = phi i32 [ [[TMP1]], [[IF_THEN_I_I_I]] ], [ [[OR_I_I_I]], [[IF_END_I_I_I]] ]
// ARM-NEXT:    [[CONV4_I:%.*]] = zext i32 [[RETVAL_I_I_I_0]] to i64
// ARM-NEXT:    [[OR_I:%.*]] = or i64 [[SHL_I]], [[CONV4_I]]
// ARM-NEXT:    ret i64 [[OR_I]]
//
uint64_t test_rev16ll(uint64_t t) {
  return __rev16ll(t);
}

// ARM-LABEL: @test_revsh(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[TMP0:%.*]] = call i16 @llvm.bswap.i16(i16 [[T:%.*]])
// ARM-NEXT:    ret i16 [[TMP0]]
//
int16_t test_revsh(int16_t t) {
  return __revsh(t);
}

// ARM-LABEL: @test_rbit(
// ARM-NEXT:  entry:
// ARM-NEXT:    [[RBIT_I:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[T:%.*]])
// ARM-NEXT:    ret i32 [[RBIT_I]]
//
uint32_t test_rbit(uint32_t t) {
  return __rbit(t);
}

// AArch32-LABEL: @test_rbitl(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[RBIT_I_I:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[T:%.*]])
// AArch32-NEXT:    ret i32 [[RBIT_I_I]]
//
// AArch64-LABEL: @test_rbitl(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[RBIT_I:%.*]] = call i64 @llvm.bitreverse.i64(i64 [[T:%.*]])
// AArch64-NEXT:    ret i64 [[RBIT_I]]
//
long test_rbitl(long t) {
  return __rbitl(t);
}

// AArch32-LABEL: @test_rbitll(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[CONV_I:%.*]] = trunc i64 [[T:%.*]] to i32
// AArch32-NEXT:    [[RBIT_I:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[CONV_I]])
// AArch32-NEXT:    [[CONV1_I:%.*]] = zext i32 [[RBIT_I]] to i64
// AArch32-NEXT:    [[SHL_I:%.*]] = shl i64 [[CONV1_I]], 32
// AArch32-NEXT:    [[SHR_I:%.*]] = lshr i64 [[T]], 32
// AArch32-NEXT:    [[CONV2_I:%.*]] = trunc i64 [[SHR_I]] to i32
// AArch32-NEXT:    [[RBIT3_I:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[CONV2_I]])
// AArch32-NEXT:    [[CONV4_I:%.*]] = zext i32 [[RBIT3_I]] to i64
// AArch32-NEXT:    [[OR_I:%.*]] = or i64 [[SHL_I]], [[CONV4_I]]
// AArch32-NEXT:    ret i64 [[OR_I]]
//
// AArch64-LABEL: @test_rbitll(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[RBIT_I:%.*]] = call i64 @llvm.bitreverse.i64(i64 [[T:%.*]])
// AArch64-NEXT:    ret i64 [[RBIT_I]]
//
uint64_t test_rbitll(uint64_t t) {
  return __rbitll(t);
}

/* 9.4 Saturating intrinsics */
#ifdef __ARM_FEATURE_SAT
/* 9.4.1 Width-specified saturation intrinsics */
// AArch32-LABEL: @test_ssat(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.ssat(i32 [[T:%.*]], i32 1)
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_ssat(int32_t t) {
  return __ssat(t, 1);
}

// AArch32-LABEL: @test_usat(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.usat(i32 [[T:%.*]], i32 2)
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_usat(int32_t t) {
  return __usat(t, 2);
}
#endif

/* 9.4.2 Saturating addition and subtraction intrinsics */
#ifdef __ARM_FEATURE_DSP
// AArch32-LABEL: @test_qadd(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qadd(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_qadd(int32_t a, int32_t b) {
  return __qadd(a, b);
}

// AArch32-LABEL: @test_qsub(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qsub(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_qsub(int32_t a, int32_t b) {
  return __qsub(a, b);
}

extern int32_t f();
// AArch32-LABEL: @test_qdbl(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[CALL:%.*]] = call i32 bitcast (i32 (...)* @f to i32 ()*)() #[[ATTR7:[0-9]+]]
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qadd(i32 [[CALL]], i32 [[CALL]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_qdbl() {
  return __qdbl(f());
}
#endif

/*
 * 9.3 16-bit multiplications
 */
#if __ARM_FEATURE_DSP
// AArch32-LABEL: @test_smulbb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smulbb(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smulbb(int32_t a, int32_t b) {
  return __smulbb(a, b);
}

// AArch32-LABEL: @test_smulbt(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smulbt(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smulbt(int32_t a, int32_t b) {
  return __smulbt(a, b);
}

// AArch32-LABEL: @test_smultb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smultb(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smultb(int32_t a, int32_t b) {
  return __smultb(a, b);
}

// AArch32-LABEL: @test_smultt(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smultt(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smultt(int32_t a, int32_t b) {
  return __smultt(a, b);
}

// AArch32-LABEL: @test_smulwb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smulwb(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smulwb(int32_t a, int32_t b) {
  return __smulwb(a, b);
}

// AArch32-LABEL: @test_smulwt(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smulwt(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smulwt(int32_t a, int32_t b) {
  return __smulwt(a, b);
}
#endif

/* 9.4.3 Accumultating multiplications */
#if __ARM_FEATURE_DSP
// AArch32-LABEL: @test_smlabb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlabb(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlabb(int32_t a, int32_t b, int32_t c) {
  return __smlabb(a, b, c);
}

// AArch32-LABEL: @test_smlabt(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlabt(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlabt(int32_t a, int32_t b, int32_t c) {
  return __smlabt(a, b, c);
}

// AArch32-LABEL: @test_smlatb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlatb(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlatb(int32_t a, int32_t b, int32_t c) {
  return __smlatb(a, b, c);
}

// AArch32-LABEL: @test_smlatt(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlatt(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlatt(int32_t a, int32_t b, int32_t c) {
  return __smlatt(a, b, c);
}

// AArch32-LABEL: @test_smlawb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlawb(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlawb(int32_t a, int32_t b, int32_t c) {
  return __smlawb(a, b, c);
}

// AArch32-LABEL: @test_smlawt(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlawt(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlawt(int32_t a, int32_t b, int32_t c) {
  return __smlawt(a, b, c);
}
#endif

/* 9.5.4 Parallel 16-bit saturation */
#if __ARM_FEATURE_SIMD32
// AArch32-LABEL: @test_ssat16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.ssat16(i32 [[A:%.*]], i32 15)
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_ssat16(int16x2_t a) {
  return __ssat16(a, 15);
}

// AArch32-LABEL: @test_usat16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.usat16(i32 [[A:%.*]], i32 15)
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_usat16(int16x2_t a) {
  return __usat16(a, 15);
}
#endif

/* 9.5.5 Packing and unpacking */
#if __ARM_FEATURE_SIMD32
// AArch32-LABEL: @test_sxtab16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.sxtab16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_sxtab16(int16x2_t a, int8x4_t b) {
  return __sxtab16(a, b);
}

// AArch32-LABEL: @test_sxtb16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.sxtb16(i32 [[A:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_sxtb16(int8x4_t a) {
  return __sxtb16(a);
}

// AArch32-LABEL: @test_uxtab16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uxtab16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_uxtab16(int16x2_t a, int8x4_t b) {
  return __uxtab16(a, b);
}

// AArch32-LABEL: @test_uxtb16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uxtb16(i32 [[A:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_uxtb16(int8x4_t a) {
  return __uxtb16(a);
}
#endif

/* 9.5.6 Parallel selection */
#if __ARM_FEATURE_SIMD32
// AArch32-LABEL: @test_sel(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.sel(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint8x4_t test_sel(uint8x4_t a, uint8x4_t b) {
  return __sel(a, b);
}
#endif

/* 9.5.7 Parallel 8-bit addition and subtraction */
#if __ARM_FEATURE_SIMD32
// AArch32-LABEL: @test_qadd8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qadd8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_qadd8(int8x4_t a, int8x4_t b) {
  return __qadd8(a, b);
}

// AArch32-LABEL: @test_qsub8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qsub8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int8x4_t test_qsub8(int8x4_t a, int8x4_t b) {
  return __qsub8(a, b);
}

// AArch32-LABEL: @test_sadd8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.sadd8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int8x4_t test_sadd8(int8x4_t a, int8x4_t b) {
  return __sadd8(a, b);
}

// AArch32-LABEL: @test_shadd8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.shadd8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int8x4_t test_shadd8(int8x4_t a, int8x4_t b) {
  return __shadd8(a, b);
}

// AArch32-LABEL: @test_shsub8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.shsub8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int8x4_t test_shsub8(int8x4_t a, int8x4_t b) {
  return __shsub8(a, b);
}

// AArch32-LABEL: @test_ssub8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.ssub8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int8x4_t test_ssub8(int8x4_t a, int8x4_t b) {
  return __ssub8(a, b);
}

// AArch32-LABEL: @test_uadd8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uadd8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint8x4_t test_uadd8(uint8x4_t a, uint8x4_t b) {
  return __uadd8(a, b);
}

// AArch32-LABEL: @test_uhadd8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uhadd8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint8x4_t test_uhadd8(uint8x4_t a, uint8x4_t b) {
  return __uhadd8(a, b);
}

// AArch32-LABEL: @test_uhsub8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uhsub8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint8x4_t test_uhsub8(uint8x4_t a, uint8x4_t b) {
  return __uhsub8(a, b);
}

// AArch32-LABEL: @test_uqadd8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uqadd8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint8x4_t test_uqadd8(uint8x4_t a, uint8x4_t b) {
  return __uqadd8(a, b);
}

// AArch32-LABEL: @test_uqsub8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uqsub8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint8x4_t test_uqsub8(uint8x4_t a, uint8x4_t b) {
  return __uqsub8(a, b);
}

// AArch32-LABEL: @test_usub8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.usub8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint8x4_t test_usub8(uint8x4_t a, uint8x4_t b) {
  return __usub8(a, b);
}
#endif

/* 9.5.8 Sum of 8-bit absolute differences */
#if __ARM_FEATURE_SIMD32
// AArch32-LABEL: @test_usad8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.usad8(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_usad8(uint8x4_t a, uint8x4_t b) {
  return __usad8(a, b);
}

// AArch32-LABEL: @test_usada8(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[CONV:%.*]] = zext i8 [[A:%.*]] to i32
// AArch32-NEXT:    [[CONV1:%.*]] = zext i8 [[B:%.*]] to i32
// AArch32-NEXT:    [[CONV2:%.*]] = zext i8 [[C:%.*]] to i32
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.usada8(i32 [[CONV]], i32 [[CONV1]], i32 [[CONV2]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_usada8(uint8_t a, uint8_t b, uint8_t c) {
  return __usada8(a, b, c);
}
#endif

/* 9.5.9 Parallel 16-bit addition and subtraction */
#if __ARM_FEATURE_SIMD32
// AArch32-LABEL: @test_qadd16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qadd16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_qadd16(int16x2_t a, int16x2_t b) {
  return __qadd16(a, b);
}

// AArch32-LABEL: @test_qasx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qasx(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_qasx(int16x2_t a, int16x2_t b) {
  return __qasx(a, b);
}

// AArch32-LABEL: @test_qsax(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qsax(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_qsax(int16x2_t a, int16x2_t b) {
  return __qsax(a, b);
}

// AArch32-LABEL: @test_qsub16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.qsub16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_qsub16(int16x2_t a, int16x2_t b) {
  return __qsub16(a, b);
}

// AArch32-LABEL: @test_sadd16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.sadd16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_sadd16(int16x2_t a, int16x2_t b) {
  return __sadd16(a, b);
}

// AArch32-LABEL: @test_sasx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.sasx(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_sasx(int16x2_t a, int16x2_t b) {
  return __sasx(a, b);
}

// AArch32-LABEL: @test_shadd16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.shadd16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_shadd16(int16x2_t a, int16x2_t b) {
  return __shadd16(a, b);
}

// AArch32-LABEL: @test_shasx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.shasx(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_shasx(int16x2_t a, int16x2_t b) {
  return __shasx(a, b);
}

// AArch32-LABEL: @test_shsax(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.shsax(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_shsax(int16x2_t a, int16x2_t b) {
  return __shsax(a, b);
}

// AArch32-LABEL: @test_shsub16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.shsub16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_shsub16(int16x2_t a, int16x2_t b) {
  return __shsub16(a, b);
}

// AArch32-LABEL: @test_ssax(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.ssax(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_ssax(int16x2_t a, int16x2_t b) {
  return __ssax(a, b);
}

// AArch32-LABEL: @test_ssub16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.ssub16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int16x2_t test_ssub16(int16x2_t a, int16x2_t b) {
  return __ssub16(a, b);
}

// AArch32-LABEL: @test_uadd16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uadd16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uadd16(uint16x2_t a, uint16x2_t b) {
  return __uadd16(a, b);
}

// AArch32-LABEL: @test_uasx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uasx(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uasx(uint16x2_t a, uint16x2_t b) {
  return __uasx(a, b);
}

// AArch32-LABEL: @test_uhadd16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uhadd16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uhadd16(uint16x2_t a, uint16x2_t b) {
  return __uhadd16(a, b);
}

// AArch32-LABEL: @test_uhasx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uhasx(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uhasx(uint16x2_t a, uint16x2_t b) {
  return __uhasx(a, b);
}

// AArch32-LABEL: @test_uhsax(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uhsax(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uhsax(uint16x2_t a, uint16x2_t b) {
  return __uhsax(a, b);
}

// AArch32-LABEL: @test_uhsub16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uhsub16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uhsub16(uint16x2_t a, uint16x2_t b) {
  return __uhsub16(a, b);
}

// AArch32-LABEL: @test_uqadd16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uqadd16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uqadd16(uint16x2_t a, uint16x2_t b) {
  return __uqadd16(a, b);
}

// AArch32-LABEL: @test_uqasx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uqasx(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uqasx(uint16x2_t a, uint16x2_t b) {
  return __uqasx(a, b);
}

// AArch32-LABEL: @test_uqsax(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uqsax(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uqsax(uint16x2_t a, uint16x2_t b) {
  return __uqsax(a, b);
}

// AArch32-LABEL: @test_uqsub16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.uqsub16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_uqsub16(uint16x2_t a, uint16x2_t b) {
  return __uqsub16(a, b);
}

// AArch32-LABEL: @test_usax(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.usax(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_usax(uint16x2_t a, uint16x2_t b) {
  return __usax(a, b);
}

// AArch32-LABEL: @test_usub16(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.usub16(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
uint16x2_t test_usub16(uint16x2_t a, uint16x2_t b) {
  return __usub16(a, b);
}
#endif

/* 9.5.10 Parallel 16-bit multiplications */
#if __ARM_FEATURE_SIMD32
// AArch32-LABEL: @test_smlad(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlad(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlad(int16x2_t a, int16x2_t b, int32_t c) {
  return __smlad(a, b, c);
}

// AArch32-LABEL: @test_smladx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smladx(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smladx(int16x2_t a, int16x2_t b, int32_t c) {
  return __smladx(a, b, c);
}

// AArch32-LABEL: @test_smlald(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i64 @llvm.arm.smlald(i32 [[A:%.*]], i32 [[B:%.*]], i64 [[C:%.*]])
// AArch32-NEXT:    ret i64 [[TMP0]]
//
int64_t test_smlald(int16x2_t a, int16x2_t b, int64_t c) {
  return __smlald(a, b, c);
}

// AArch32-LABEL: @test_smlaldx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i64 @llvm.arm.smlaldx(i32 [[A:%.*]], i32 [[B:%.*]], i64 [[C:%.*]])
// AArch32-NEXT:    ret i64 [[TMP0]]
//
int64_t test_smlaldx(int16x2_t a, int16x2_t b, int64_t c) {
  return __smlaldx(a, b, c);
}

// AArch32-LABEL: @test_smlsd(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlsd(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlsd(int16x2_t a, int16x2_t b, int32_t c) {
  return __smlsd(a, b, c);
}

// AArch32-LABEL: @test_smlsdx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smlsdx(i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smlsdx(int16x2_t a, int16x2_t b, int32_t c) {
  return __smlsdx(a, b, c);
}

// AArch32-LABEL: @test_smlsld(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i64 @llvm.arm.smlsld(i32 [[A:%.*]], i32 [[B:%.*]], i64 [[C:%.*]])
// AArch32-NEXT:    ret i64 [[TMP0]]
//
int64_t test_smlsld(int16x2_t a, int16x2_t b, int64_t c) {
  return __smlsld(a, b, c);
}

// AArch32-LABEL: @test_smlsldx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i64 @llvm.arm.smlsldx(i32 [[A:%.*]], i32 [[B:%.*]], i64 [[C:%.*]])
// AArch32-NEXT:    ret i64 [[TMP0]]
//
int64_t test_smlsldx(int16x2_t a, int16x2_t b, int64_t c) {
  return __smlsldx(a, b, c);
}

// AArch32-LABEL: @test_smuad(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smuad(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smuad(int16x2_t a, int16x2_t b) {
  return __smuad(a, b);
}

// AArch32-LABEL: @test_smuadx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smuadx(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smuadx(int16x2_t a, int16x2_t b) {
  return __smuadx(a, b);
}

// AArch32-LABEL: @test_smusd(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smusd(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smusd(int16x2_t a, int16x2_t b) {
  return __smusd(a, b);
}

// AArch32-LABEL: @test_smusdx(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.smusdx(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
int32_t test_smusdx(int16x2_t a, int16x2_t b) {
  return __smusdx(a, b);
}
#endif

/* 9.7 CRC32 intrinsics */
// AArch32-LABEL: @test_crc32b(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = zext i8 [[B:%.*]] to i32
// AArch32-NEXT:    [[TMP1:%.*]] = call i32 @llvm.arm.crc32b(i32 [[A:%.*]], i32 [[TMP0]])
// AArch32-NEXT:    ret i32 [[TMP1]]
//
// AArch64-LABEL: @test_crc32b(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = zext i8 [[B:%.*]] to i32
// AArch64-NEXT:    [[TMP1:%.*]] = call i32 @llvm.aarch64.crc32b(i32 [[A:%.*]], i32 [[TMP0]])
// AArch64-NEXT:    ret i32 [[TMP1]]
//
uint32_t test_crc32b(uint32_t a, uint8_t b) {
  return __crc32b(a, b);
}

// AArch32-LABEL: @test_crc32h(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = zext i16 [[B:%.*]] to i32
// AArch32-NEXT:    [[TMP1:%.*]] = call i32 @llvm.arm.crc32h(i32 [[A:%.*]], i32 [[TMP0]])
// AArch32-NEXT:    ret i32 [[TMP1]]
//
// AArch64-LABEL: @test_crc32h(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = zext i16 [[B:%.*]] to i32
// AArch64-NEXT:    [[TMP1:%.*]] = call i32 @llvm.aarch64.crc32h(i32 [[A:%.*]], i32 [[TMP0]])
// AArch64-NEXT:    ret i32 [[TMP1]]
//
uint32_t test_crc32h(uint32_t a, uint16_t b) {
  return __crc32h(a, b);
}

// AArch32-LABEL: @test_crc32w(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.crc32w(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
// AArch64-LABEL: @test_crc32w(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i32 @llvm.aarch64.crc32w(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch64-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_crc32w(uint32_t a, uint32_t b) {
  return __crc32w(a, b);
}

// AArch32-LABEL: @test_crc32d(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = trunc i64 [[B:%.*]] to i32
// AArch32-NEXT:    [[TMP1:%.*]] = lshr i64 [[B]], 32
// AArch32-NEXT:    [[TMP2:%.*]] = trunc i64 [[TMP1]] to i32
// AArch32-NEXT:    [[TMP3:%.*]] = call i32 @llvm.arm.crc32w(i32 [[A:%.*]], i32 [[TMP0]])
// AArch32-NEXT:    [[TMP4:%.*]] = call i32 @llvm.arm.crc32w(i32 [[TMP3]], i32 [[TMP2]])
// AArch32-NEXT:    ret i32 [[TMP4]]
//
// AArch64-LABEL: @test_crc32d(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i32 @llvm.aarch64.crc32x(i32 [[A:%.*]], i64 [[B:%.*]])
// AArch64-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_crc32d(uint32_t a, uint64_t b) {
  return __crc32d(a, b);
}

// AArch32-LABEL: @test_crc32cb(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = zext i8 [[B:%.*]] to i32
// AArch32-NEXT:    [[TMP1:%.*]] = call i32 @llvm.arm.crc32cb(i32 [[A:%.*]], i32 [[TMP0]])
// AArch32-NEXT:    ret i32 [[TMP1]]
//
// AArch64-LABEL: @test_crc32cb(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = zext i8 [[B:%.*]] to i32
// AArch64-NEXT:    [[TMP1:%.*]] = call i32 @llvm.aarch64.crc32cb(i32 [[A:%.*]], i32 [[TMP0]])
// AArch64-NEXT:    ret i32 [[TMP1]]
//
uint32_t test_crc32cb(uint32_t a, uint8_t b) {
  return __crc32cb(a, b);
}

// AArch32-LABEL: @test_crc32ch(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = zext i16 [[B:%.*]] to i32
// AArch32-NEXT:    [[TMP1:%.*]] = call i32 @llvm.arm.crc32ch(i32 [[A:%.*]], i32 [[TMP0]])
// AArch32-NEXT:    ret i32 [[TMP1]]
//
// AArch64-LABEL: @test_crc32ch(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = zext i16 [[B:%.*]] to i32
// AArch64-NEXT:    [[TMP1:%.*]] = call i32 @llvm.aarch64.crc32ch(i32 [[A:%.*]], i32 [[TMP0]])
// AArch64-NEXT:    ret i32 [[TMP1]]
//
uint32_t test_crc32ch(uint32_t a, uint16_t b) {
  return __crc32ch(a, b);
}

// AArch32-LABEL: @test_crc32cw(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.arm.crc32cw(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
// AArch64-LABEL: @test_crc32cw(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i32 @llvm.aarch64.crc32cw(i32 [[A:%.*]], i32 [[B:%.*]])
// AArch64-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_crc32cw(uint32_t a, uint32_t b) {
  return __crc32cw(a, b);
}

// AArch32-LABEL: @test_crc32cd(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = trunc i64 [[B:%.*]] to i32
// AArch32-NEXT:    [[TMP1:%.*]] = lshr i64 [[B]], 32
// AArch32-NEXT:    [[TMP2:%.*]] = trunc i64 [[TMP1]] to i32
// AArch32-NEXT:    [[TMP3:%.*]] = call i32 @llvm.arm.crc32cw(i32 [[A:%.*]], i32 [[TMP0]])
// AArch32-NEXT:    [[TMP4:%.*]] = call i32 @llvm.arm.crc32cw(i32 [[TMP3]], i32 [[TMP2]])
// AArch32-NEXT:    ret i32 [[TMP4]]
//
// AArch64-LABEL: @test_crc32cd(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i32 @llvm.aarch64.crc32cx(i32 [[A:%.*]], i64 [[B:%.*]])
// AArch64-NEXT:    ret i32 [[TMP0]]
//
uint32_t test_crc32cd(uint32_t a, uint64_t b) {
  return __crc32cd(a, b);
}

/* 10.1 Special register intrinsics */
// AArch32-LABEL: @test_rsr(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.read_volatile_register.i32(metadata [[META9:![0-9]+]])
// AArch32-NEXT:    ret i32 [[TMP0]]
//
// AArch64-LABEL: @test_rsr(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.read_volatile_register.i64(metadata [[META8:![0-9]+]])
// AArch64-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// AArch64-NEXT:    ret i32 [[TMP1]]
//
uint32_t test_rsr() {
#ifdef __ARM_32BIT_STATE
  return __arm_rsr("cp1:2:c3:c4:5");
#else
  return __arm_rsr("1:2:3:4:5");
#endif
}

// AArch32-LABEL: @test_rsr64(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i64 @llvm.read_volatile_register.i64(metadata [[META10:![0-9]+]])
// AArch32-NEXT:    ret i64 [[TMP0]]
//
// AArch64-LABEL: @test_rsr64(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.read_volatile_register.i64(metadata [[META8]])
// AArch64-NEXT:    ret i64 [[TMP0]]
//
uint64_t test_rsr64() {
#ifdef __ARM_32BIT_STATE
  return __arm_rsr64("cp1:2:c3");
#else
  return __arm_rsr64("1:2:3:4:5");
#endif
}

// AArch32-LABEL: @test_rsrp(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.read_volatile_register.i32(metadata [[META11:![0-9]+]])
// AArch32-NEXT:    [[TMP1:%.*]] = inttoptr i32 [[TMP0]] to i8*
// AArch32-NEXT:    ret i8* [[TMP1]]
//
// AArch64-LABEL: @test_rsrp(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.read_volatile_register.i64(metadata [[META9:![0-9]+]])
// AArch64-NEXT:    [[TMP1:%.*]] = inttoptr i64 [[TMP0]] to i8*
// AArch64-NEXT:    ret i8* [[TMP1]]
//
void *test_rsrp() {
  return __arm_rsrp("sysreg");
}

// AArch32-LABEL: @test_wsr(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.write_register.i32(metadata [[META9]], i32 [[V:%.*]])
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_wsr(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = zext i32 [[V:%.*]] to i64
// AArch64-NEXT:    call void @llvm.write_register.i64(metadata [[META8]], i64 [[TMP0]])
// AArch64-NEXT:    ret void
//
void test_wsr(uint32_t v) {
#ifdef __ARM_32BIT_STATE
  __arm_wsr("cp1:2:c3:c4:5", v);
#else
  __arm_wsr("1:2:3:4:5", v);
#endif
}

// AArch32-LABEL: @test_wsr64(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    call void @llvm.write_register.i64(metadata [[META10]], i64 [[V:%.*]])
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_wsr64(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    call void @llvm.write_register.i64(metadata [[META8]], i64 [[V:%.*]])
// AArch64-NEXT:    ret void
//
void test_wsr64(uint64_t v) {
#ifdef __ARM_32BIT_STATE
  __arm_wsr64("cp1:2:c3", v);
#else
  __arm_wsr64("1:2:3:4:5", v);
#endif
}

// AArch32-LABEL: @test_wsrp(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[TMP0:%.*]] = ptrtoint i8* [[V:%.*]] to i32
// AArch32-NEXT:    call void @llvm.write_register.i32(metadata [[META11]], i32 [[TMP0]])
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_wsrp(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[TMP0:%.*]] = ptrtoint i8* [[V:%.*]] to i64
// AArch64-NEXT:    call void @llvm.write_register.i64(metadata [[META9]], i64 [[TMP0]])
// AArch64-NEXT:    ret void
//
void test_wsrp(void *v) {
  __arm_wsrp("sysreg", v);
}

// AArch32-LABEL: @test_rsrf(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[REF_TMP:%.*]] = alloca i32, align 4
// AArch32-NEXT:    [[TMP0:%.*]] = call i32 @llvm.read_volatile_register.i32(metadata [[META9]])
// AArch32-NEXT:    store i32 [[TMP0]], i32* [[REF_TMP]], align 4
// AArch32-NEXT:    [[TMP1:%.*]] = bitcast i32* [[REF_TMP]] to float*
// AArch32-NEXT:    [[TMP2:%.*]] = load float, float* [[TMP1]], align 4
// AArch32-NEXT:    ret float [[TMP2]]
//
// AArch64-LABEL: @test_rsrf(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[REF_TMP:%.*]] = alloca i32, align 4
// AArch64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.read_volatile_register.i64(metadata [[META8]])
// AArch64-NEXT:    [[TMP1:%.*]] = trunc i64 [[TMP0]] to i32
// AArch64-NEXT:    store i32 [[TMP1]], i32* [[REF_TMP]], align 4
// AArch64-NEXT:    [[TMP2:%.*]] = bitcast i32* [[REF_TMP]] to float*
// AArch64-NEXT:    [[TMP3:%.*]] = load float, float* [[TMP2]], align 4
// AArch64-NEXT:    ret float [[TMP3]]
//
float test_rsrf() {
#ifdef __ARM_32BIT_STATE
  return __arm_rsrf("cp1:2:c3:c4:5");
#else
  return __arm_rsrf("1:2:3:4:5");
#endif
}

// AArch32-LABEL: @test_rsrf64(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[REF_TMP:%.*]] = alloca i64, align 8
// AArch32-NEXT:    [[TMP0:%.*]] = call i64 @llvm.read_volatile_register.i64(metadata [[META10]])
// AArch32-NEXT:    store i64 [[TMP0]], i64* [[REF_TMP]], align 8
// AArch32-NEXT:    [[TMP1:%.*]] = bitcast i64* [[REF_TMP]] to double*
// AArch32-NEXT:    [[TMP2:%.*]] = load double, double* [[TMP1]], align 8
// AArch32-NEXT:    ret double [[TMP2]]
//
// AArch64-LABEL: @test_rsrf64(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[REF_TMP:%.*]] = alloca i64, align 8
// AArch64-NEXT:    [[TMP0:%.*]] = call i64 @llvm.read_volatile_register.i64(metadata [[META8]])
// AArch64-NEXT:    store i64 [[TMP0]], i64* [[REF_TMP]], align 8
// AArch64-NEXT:    [[TMP1:%.*]] = bitcast i64* [[REF_TMP]] to double*
// AArch64-NEXT:    [[TMP2:%.*]] = load double, double* [[TMP1]], align 8
// AArch64-NEXT:    ret double [[TMP2]]
//
double test_rsrf64() {
#ifdef __ARM_32BIT_STATE
  return __arm_rsrf64("cp1:2:c3");
#else
  return __arm_rsrf64("1:2:3:4:5");
#endif
}

// AArch32-LABEL: @test_wsrf(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[V_ADDR:%.*]] = alloca float, align 4
// AArch32-NEXT:    store float [[V:%.*]], float* [[V_ADDR]], align 4
// AArch32-NEXT:    [[TMP0:%.*]] = bitcast float* [[V_ADDR]] to i32*
// AArch32-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// AArch32-NEXT:    call void @llvm.write_register.i32(metadata [[META9]], i32 [[TMP1]])
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_wsrf(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[V_ADDR:%.*]] = alloca float, align 4
// AArch64-NEXT:    store float [[V:%.*]], float* [[V_ADDR]], align 4
// AArch64-NEXT:    [[TMP0:%.*]] = bitcast float* [[V_ADDR]] to i32*
// AArch64-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// AArch64-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// AArch64-NEXT:    call void @llvm.write_register.i64(metadata [[META8]], i64 [[TMP2]])
// AArch64-NEXT:    ret void
//
void test_wsrf(float v) {
#ifdef __ARM_32BIT_STATE
  __arm_wsrf("cp1:2:c3:c4:5", v);
#else
  __arm_wsrf("1:2:3:4:5", v);
#endif
}

// AArch32-LABEL: @test_wsrf64(
// AArch32-NEXT:  entry:
// AArch32-NEXT:    [[V_ADDR:%.*]] = alloca double, align 8
// AArch32-NEXT:    store double [[V:%.*]], double* [[V_ADDR]], align 8
// AArch32-NEXT:    [[TMP0:%.*]] = bitcast double* [[V_ADDR]] to i64*
// AArch32-NEXT:    [[TMP1:%.*]] = load i64, i64* [[TMP0]], align 8
// AArch32-NEXT:    call void @llvm.write_register.i64(metadata [[META10]], i64 [[TMP1]])
// AArch32-NEXT:    ret void
//
// AArch64-LABEL: @test_wsrf64(
// AArch64-NEXT:  entry:
// AArch64-NEXT:    [[V_ADDR:%.*]] = alloca double, align 8
// AArch64-NEXT:    store double [[V:%.*]], double* [[V_ADDR]], align 8
// AArch64-NEXT:    [[TMP0:%.*]] = bitcast double* [[V_ADDR]] to i64*
// AArch64-NEXT:    [[TMP1:%.*]] = load i64, i64* [[TMP0]], align 8
// AArch64-NEXT:    call void @llvm.write_register.i64(metadata [[META8]], i64 [[TMP1]])
// AArch64-NEXT:    ret void
//
void test_wsrf64(double v) {
#ifdef __ARM_32BIT_STATE
  __arm_wsrf64("cp1:2:c3", v);
#else
  __arm_wsrf64("1:2:3:4:5", v);
#endif
}

#ifdef __ARM_64BIT_STATE
// AArch6483-LABEL: @test_jcvt(
// AArch6483-NEXT:  entry:
// AArch6483-NEXT:    [[TMP0:%.*]] = call i32 @llvm.aarch64.fjcvtzs(double [[V:%.*]])
// AArch6483-NEXT:    ret i32 [[TMP0]]
//
int32_t test_jcvt(double v) {
  return __jcvt(v);
}
#endif


#if __ARM_64BIT_STATE && defined(__ARM_FEATURE_RNG)

// AArch6485-LABEL: @test_rndr(
// AArch6485-NEXT:  entry:
// AArch6485-NEXT:    [[TMP0:%.*]] = call { i64, i1 } @llvm.aarch64.rndr()
// AArch6485-NEXT:    [[TMP1:%.*]] = extractvalue { i64, i1 } [[TMP0]], 0
// AArch6485-NEXT:    [[TMP2:%.*]] = extractvalue { i64, i1 } [[TMP0]], 1
// AArch6485-NEXT:    store i64 [[TMP1]], i64* [[__ADDR:%.*]], align 8
// AArch6485-NEXT:    [[TMP3:%.*]] = zext i1 [[TMP2]] to i32
// AArch6485-NEXT:    ret i32 [[TMP3]]
//
int test_rndr(uint64_t *__addr) {
  return __rndr(__addr);
}

// AArch6485-LABEL: @test_rndrrs(
// AArch6485-NEXT:  entry:
// AArch6485-NEXT:    [[TMP0:%.*]] = call { i64, i1 } @llvm.aarch64.rndrrs()
// AArch6485-NEXT:    [[TMP1:%.*]] = extractvalue { i64, i1 } [[TMP0]], 0
// AArch6485-NEXT:    [[TMP2:%.*]] = extractvalue { i64, i1 } [[TMP0]], 1
// AArch6485-NEXT:    store i64 [[TMP1]], i64* [[__ADDR:%.*]], align 8
// AArch6485-NEXT:    [[TMP3:%.*]] = zext i1 [[TMP2]] to i32
// AArch6485-NEXT:    ret i32 [[TMP3]]
//
int test_rndrrs(uint64_t *__addr) {
  return __rndrrs(__addr);
}
#endif