1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 19:23:23 +01:00
llvm-mirror/test/CodeGen/X86/pr12360.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

59 lines
1.3 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s
define zeroext i1 @f1(i8* %x) {
; CHECK-LABEL: f1:
; CHECK: ## %bb.0: ## %entry
; CHECK-NEXT: movb (%rdi), %al
; CHECK-NEXT: retq
entry:
%0 = load i8, i8* %x, align 1, !range !0
%tobool = trunc i8 %0 to i1
ret i1 %tobool
}
define zeroext i1 @f2(i8* %x) {
; CHECK-LABEL: f2:
; CHECK: ## %bb.0: ## %entry
; CHECK-NEXT: movb (%rdi), %al
; CHECK-NEXT: retq
entry:
%0 = load i8, i8* %x, align 1, !range !0
%tobool = icmp ne i8 %0, 0
ret i1 %tobool
}
!0 = !{i8 0, i8 2}
; check that we don't build a "trunc" from i1 to i1, which would assert.
define zeroext i1 @f3(i1 %x) {
; CHECK-LABEL: f3:
; CHECK: ## %bb.0: ## %entry
; CHECK-NEXT: movl %edi, %eax
; CHECK-NEXT: andb $1, %al
; CHECK-NEXT: ## kill: def $al killed $al killed $eax
; CHECK-NEXT: retq
entry:
%tobool = icmp ne i1 %x, 0
ret i1 %tobool
}
; check that we don't build a trunc when other bits are needed
define zeroext i1 @f4(i32 %x) {
; CHECK-LABEL: f4:
; CHECK: ## %bb.0: ## %entry
; CHECK-NEXT: movzwl %di, %eax
; CHECK-NEXT: shrl $15, %eax
; CHECK-NEXT: ## kill: def $al killed $al killed $eax
; CHECK-NEXT: retq
entry:
%y = and i32 %x, 32768
%z = icmp ne i32 %y, 0
ret i1 %z
}