1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-24 19:52:54 +01:00
llvm-mirror/test/CodeGen/X86/xchg-nofold.ll
Simon Pilgrim d654e7d40c [X86] Handle COPYs of physregs better (regalloc hints)
Enable enableMultipleCopyHints() on X86.

Original Patch by @jonpa:

While enabling the mischeduler for SystemZ, it was discovered that for some reason a test needed one extra seemingly needless COPY (test/CodeGen/SystemZ/call-03.ll). The handling for that is resulted in this patch, which improves the register coalescing by providing not just one copy hint, but a sorted list of copy hints. On SystemZ, this gives ~12500 less register moves on SPEC, as well as marginally less spilling.

Instead of improving just the SystemZ backend, the improvement has been implemented in common-code (calculateSpillWeightAndHint(). This gives a lot of test failures, but since this should be a general improvement I hope that the involved targets will help and review the test updates.

Differential Revision: https://reviews.llvm.org/D38128

llvm-svn: 342578
2018-09-19 18:59:08 +00:00

61 lines
2.0 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=x86_64-linux-gnu < %s | FileCheck %s
%"struct.std::atomic" = type { %"struct.std::atomic_bool" }
%"struct.std::atomic_bool" = type { %"struct.std::__atomic_base" }
%"struct.std::__atomic_base" = type { i8 }
; CHECK-LABEL: _Z3fooRSt6atomicIbEb
define zeroext i1 @_Z3fooRSt6atomicIbEb(%"struct.std::atomic"* nocapture dereferenceable(1) %a, i1 returned zeroext %b) nounwind {
; CHECK-LABEL: _Z3fooRSt6atomicIbEb:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: movl %esi, %eax
; CHECK-NEXT: movq %rdi, %rcx
; CHECK-NEXT: shrq $3, %rcx
; CHECK-NEXT: movb 2147450880(%rcx), %cl
; CHECK-NEXT: testb %cl, %cl
; CHECK-NEXT: je .LBB0_3
; CHECK-NEXT: # %bb.1:
; CHECK-NEXT: movl %edi, %edx
; CHECK-NEXT: andl $7, %edx
; CHECK-NEXT: cmpb %cl, %dl
; CHECK-NEXT: jge .LBB0_2
; CHECK-NEXT: .LBB0_3:
; CHECK-NEXT: movl %eax, %ecx
; CHECK-NEXT: xchgb %cl, (%rdi)
; CHECK-NEXT: # kill: def $al killed $al killed $eax
; CHECK-NEXT: retq
; CHECK-NEXT: .LBB0_2:
; CHECK-NEXT: pushq %rax
; CHECK-NEXT: callq __asan_report_store1
; CHECK-NEXT: #APP
; CHECK-NEXT: #NO_APP
entry:
%frombool.i.i = zext i1 %b to i8
%_M_i.i.i = getelementptr inbounds %"struct.std::atomic", %"struct.std::atomic"* %a, i64 0, i32 0, i32 0, i32 0
%0 = ptrtoint i8* %_M_i.i.i to i64
%1 = lshr i64 %0, 3
%2 = add i64 %1, 2147450880
%3 = inttoptr i64 %2 to i8*
%4 = load i8, i8* %3
%5 = icmp ne i8 %4, 0
br i1 %5, label %6, label %11
; <label>:6: ; preds = %entry
%7 = and i64 %0, 7
%8 = trunc i64 %7 to i8
%9 = icmp sge i8 %8, %4
br i1 %9, label %10, label %11
; <label>:10: ; preds = %6
call void @__asan_report_store1(i64 %0)
call void asm sideeffect "", ""()
unreachable
; <label>:11: ; preds = %6, %entry
store atomic i8 %frombool.i.i, i8* %_M_i.i.i seq_cst, align 1
ret i1 %b
}
declare void @__asan_report_store1(i64)