maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/X86/atomic16.ll

251 lines
5.7 KiB
LLVM
Raw Normal View History

Fix PR15355 - Clear 'mayStore' flag when loading from the atomic variable before the spin loop - Clear kill flag from one use to multiple use in registers forming the address to that atomic variable - don't use a physical register as live-in register in BB (neither entry nor landing pad.) by copying it into virtual register (patch by Cameron Zwarich) llvm-svn: 176538
2013-03-06 08:17:04 +08:00
; RUN: llc < %s -O0 -mtriple=x86_64-unknown-unknown -mcpu=corei7 -verify-machineinstrs -show-mc-encoding | FileCheck %s --check-prefix X64
; RUN: llc < %s -O0 -mtriple=i386-unknown-unknown -mcpu=corei7 -verify-machineinstrs | FileCheck %s --check-prefix X32
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
@sc16 = external global i16
define void @atomic_fetch_add16() nounwind {
; X64: atomic_fetch_add16
; X32: atomic_fetch_add16
entry:
; 32-bit
%t1 = atomicrmw add i16* @sc16, i16 1 acquire
; X64: lock
; X64: incw
; X32: lock
; X32: incw
%t2 = atomicrmw add i16* @sc16, i16 3 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: addw $3, {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: addw $3
%t3 = atomicrmw add i16* @sc16, i16 5 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: xaddw {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: xaddw
%t4 = atomicrmw add i16* @sc16, i16 %t3 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: addw {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: addw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_sub16() nounwind {
; X64: atomic_fetch_sub16
; X32: atomic_fetch_sub16
%t1 = atomicrmw sub i16* @sc16, i16 1 acquire
; X64: lock
; X64: decw
; X32: lock
; X32: decw
%t2 = atomicrmw sub i16* @sc16, i16 3 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: subw $3, {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: subw $3
%t3 = atomicrmw sub i16* @sc16, i16 5 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: xaddw {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: xaddw
%t4 = atomicrmw sub i16* @sc16, i16 %t3 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: subw {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: subw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_and16() nounwind {
; X64: atomic_fetch_and16
; X32: atomic_fetch_and16
%t1 = atomicrmw and i16* @sc16, i16 3 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: andw $3, {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: andw $3
%t2 = atomicrmw and i16* @sc16, i16 5 acquire
; X64: andw
; X64: lock
; X64: cmpxchgw
; X32: andw
; X32: lock
; X32: cmpxchgw
%t3 = atomicrmw and i16* @sc16, i16 %t2 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: andw {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: andw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_or16() nounwind {
; X64: atomic_fetch_or16
; X32: atomic_fetch_or16
%t1 = atomicrmw or i16* @sc16, i16 3 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: orw $3, {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: orw $3
%t2 = atomicrmw or i16* @sc16, i16 5 acquire
; X64: orw
; X64: lock
; X64: cmpxchgw
; X32: orw
; X32: lock
; X32: cmpxchgw
%t3 = atomicrmw or i16* @sc16, i16 %t2 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: orw {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: orw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_xor16() nounwind {
; X64: atomic_fetch_xor16
; X32: atomic_fetch_xor16
%t1 = atomicrmw xor i16* @sc16, i16 3 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: xorw $3, {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: xorw $3
%t2 = atomicrmw xor i16* @sc16, i16 5 acquire
; X64: xorw
; X64: lock
; X64: cmpxchgw
; X32: xorw
; X32: lock
; X32: cmpxchgw
%t3 = atomicrmw xor i16* @sc16, i16 %t2 acquire
; X64: lock
Revise test to avoid using of 'grep' llvm-svn: 164472
2012-09-23 10:41:47 +08:00
; X64: xorw {{.*}} # encoding: [0xf0,0x66
Re-work X86 code generation of atomic ops with spin-loop - Rewrite/merge pseudo-atomic instruction emitters to address the following issue: * Reduce one unnecessary load in spin-loop previously the spin-loop looks like thisMBB: newMBB: ld t1 = [bitinstr.addr] op t2 = t1, [bitinstr.val] not t3 = t2 (if Invert) mov EAX = t1 lcs dest = [bitinstr.addr], t3 [EAX is implicit] bz newMBB fallthrough -->nextMBB the 'ld' at the beginning of newMBB should be lift out of the loop as lcs (or CMPXCHG on x86) will load the current memory value into EAX. This loop is refined as: thisMBB: EAX = LOAD [MI.addr] mainMBB: t1 = OP [MI.val], EAX LCMPXCHG [MI.addr], t1, [EAX is implicitly used & defined] JNE mainMBB sinkMBB: * Remove immopc as, so far, all pseudo-atomic instructions has all-register form only, there is no immedidate operand. * Remove unnecessary attributes/modifiers in pseudo-atomic instruction td * Fix issues in PR13458 - Add comprehensive tests on atomic ops on various data types. NOTE: Some of them are turned off due to missing functionality. - Revise tests due to the new spin-loop generated. llvm-svn: 164281
2012-09-20 11:06:15 +08:00
; X32: lock
; X32: xorw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_nand16(i16 %x) nounwind {
; X64: atomic_fetch_nand16
; X32: atomic_fetch_nand16
%t1 = atomicrmw nand i16* @sc16, i16 %x acquire
; X64: andw
; X64: notw
; X64: lock
; X64: cmpxchgw
; X32: andw
; X32: notw
; X32: lock
; X32: cmpxchgw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_max16(i16 %x) nounwind {
%t1 = atomicrmw max i16* @sc16, i16 %x acquire
; X64: cmpw
; X64: cmov
; X64: lock
; X64: cmpxchgw
; X32: cmpw
; X32: cmov
; X32: lock
; X32: cmpxchgw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_min16(i16 %x) nounwind {
%t1 = atomicrmw min i16* @sc16, i16 %x acquire
; X64: cmpw
; X64: cmov
; X64: lock
; X64: cmpxchgw
; X32: cmpw
; X32: cmov
; X32: lock
; X32: cmpxchgw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_umax16(i16 %x) nounwind {
%t1 = atomicrmw umax i16* @sc16, i16 %x acquire
; X64: cmpw
; X64: cmov
; X64: lock
; X64: cmpxchgw
; X32: cmpw
; X32: cmov
; X32: lock
; X32: cmpxchgw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_umin16(i16 %x) nounwind {
%t1 = atomicrmw umin i16* @sc16, i16 %x acquire
; X64: cmpw
; X64: cmov
; X64: lock
; X64: cmpxchgw
; X32: cmpw
; X32: cmov
; X32: lock
; X32: cmpxchgw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_cmpxchg16() nounwind {
%t1 = cmpxchg i16* @sc16, i16 0, i16 1 acquire
; X64: lock
; X64: cmpxchgw
; X32: lock
; X32: cmpxchgw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_store16(i16 %x) nounwind {
store atomic i16 %x, i16* @sc16 release, align 4
; X64-NOT: lock
; X64: movw
; X32-NOT: lock
; X32: movw
ret void
; X64: ret
; X32: ret
}
define void @atomic_fetch_swap16(i16 %x) nounwind {
%t1 = atomicrmw xchg i16* @sc16, i16 %x acquire
; X64-NOT: lock
; X64: xchgw
; X32-NOT: lock
; X32: xchgw
ret void
; X64: ret
; X32: ret
}