mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-30 23:42:52 +01:00
6164109851
Aggressive anti-dependency breaking is enabled by default for all PPC cores. This provides a general speedup on the P7 and other platforms (among other factors, the instruction group formation for the non-embedded PPC cores is done during post-RA scheduling). In order to do this safely, the incompatibility between uses of the MFOCRF instruction and anti-dependency breaking are resolved by marking MFOCRF with hasExtraSrcRegAllocReq. As noted in the removed FIXME, the problem was that MFOCRF's output is sensitive to the identify of the source register, and always paired with a shift to undo this effect. Because anti-dependency breaking is unaware of this hidden dependency of the shift amount on the source register of the MFOCRF instruction, changing that register must be inhibited. Two test cases were adjusted: The SjLj test was made more insensitive to register choices and scheduling; the saveCR test disabled anti-dependency breaking because part of what it is testing is proper register reuse. llvm-svn: 190587
42 lines
1.6 KiB
LLVM
42 lines
1.6 KiB
LLVM
; RUN: llc < %s -mtriple=powerpc-apple-darwin -mcpu=g4 -break-anti-dependencies=none | FileCheck %s
|
|
; ModuleID = 'hh.c'
|
|
target datalayout = "E-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f128:64:128-n32"
|
|
target triple = "powerpc-apple-darwin9.6"
|
|
|
|
define void @foo() nounwind {
|
|
entry:
|
|
; Note that part of what is being checked here is proper register reuse.
|
|
; CHECK: mfcr [[T1:r[0-9]+]] ; cr2
|
|
; CHECK: lis [[T2:r[0-9]+]], 1
|
|
; CHECK: addi r3, r1, 72
|
|
; CHECK: rlwinm [[T1]], [[T1]], 8, 0, 31
|
|
; CHECK: ori [[T2]], [[T2]], 34540
|
|
; CHECK: stwx [[T1]], r1, [[T2]]
|
|
; CHECK: lis [[T3:r[0-9]+]], 1
|
|
; CHECK: mfcr [[T4:r[0-9]+]] ; cr3
|
|
; CHECK: ori [[T3]], [[T3]], 34536
|
|
; CHECK: rlwinm [[T4]], [[T4]], 12, 0, 31
|
|
; CHECK: stwx [[T4]], r1, [[T3]]
|
|
%x = alloca [100000 x i8] ; <[100000 x i8]*> [#uses=1]
|
|
%"alloca point" = bitcast i32 0 to i32 ; <i32> [#uses=0]
|
|
%x1 = bitcast [100000 x i8]* %x to i8* ; <i8*> [#uses=1]
|
|
call void @bar(i8* %x1) nounwind
|
|
call void asm sideeffect "", "~{cr2},~{cr3}"() nounwind
|
|
br label %return
|
|
|
|
return: ; preds = %entry
|
|
; CHECK: lis [[T1:r[0-9]+]], 1
|
|
; CHECK: ori [[T1]], [[T1]], 34536
|
|
; CHECK: lwzx [[T1]], r1, [[T1]]
|
|
; CHECK: rlwinm [[T1]], [[T1]], 20, 0, 31
|
|
; CHECK: mtcrf 16, [[T1]]
|
|
; CHECK: lis [[T1]], 1
|
|
; CHECK: ori [[T1]], [[T1]], 34540
|
|
; CHECK: lwzx [[T1]], r1, [[T1]]
|
|
; CHECK: rlwinm [[T1]], [[T1]], 24, 0, 31
|
|
; CHECK: mtcrf 32, [[T1]]
|
|
ret void
|
|
}
|
|
|
|
declare void @bar(i8*)
|