mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-24 03:33:20 +01:00
db3c93e416
This will remove suboptimal branching from the generated ll/sc loops. The extra simplification pass affects a lot of testcases, which have been modified to accommodate this change: either by modifying the test to become immune to the CFG simplification, or (less preferablt) by adding option -hexagon-initial-cfg-clenaup=0. llvm-svn: 338774
47 lines
1.6 KiB
LLVM
47 lines
1.6 KiB
LLVM
; RUN: llc -march=hexagon -enable-pipeliner-opt-size -hexagon-initial-cfg-cleanup=0 < %s | FileCheck %s
|
|
|
|
; Test that we generate the correct names for the phis in the kernel for the
|
|
; incoming values. In this case, the loop contains a phi and has another phi
|
|
; as its loop definition, and the two phis are scheduled in different stages.
|
|
;
|
|
; vreg5 = phi(x, vreg4) is scheduled in stage 1, cycle 0
|
|
; vreg4 = phi(y, z) is scheduled in stage 0, cycle 0
|
|
|
|
; CHECK-DAG: :[[REG0:[0-9]+]]{{.*}} = {{.*}},#17
|
|
; CHECK-DAG: loop0(.LBB0_[[LOOP:.]],
|
|
; CHECK: .LBB0_[[LOOP]]:
|
|
; CHECK: r{{[0-9]+}} = sxth(r[[REG0]])
|
|
; CHECK: endloop0
|
|
|
|
; Function Attrs: nounwind optsize
|
|
define void @f0() #0 {
|
|
b0:
|
|
%v0 = getelementptr [8 x i16], [8 x i16]* undef, i32 0, i32 7
|
|
%v1 = bitcast i16* %v0 to [8 x i16]*
|
|
br label %b2
|
|
|
|
b1: ; preds = %b2
|
|
unreachable
|
|
|
|
b2: ; preds = %b2, %b0
|
|
%v2 = phi i32 [ 7, %b0 ], [ %v11, %b2 ]
|
|
%v3 = phi i16 [ 17, %b0 ], [ %v7, %b2 ]
|
|
%v4 = phi i16 [ 18, %b0 ], [ %v3, %b2 ]
|
|
%v5 = sext i16 %v4 to i32
|
|
%v6 = getelementptr i16, i16* null, i32 -2
|
|
%v7 = load i16, i16* %v6, align 2
|
|
%v8 = sext i16 %v7 to i32
|
|
%v9 = tail call i32 @llvm.hexagon.A2.subsat(i32 %v5, i32 %v8)
|
|
%v10 = trunc i32 %v9 to i16
|
|
store i16 %v10, i16* null, align 2
|
|
%v11 = add nsw i32 %v2, -1
|
|
%v12 = icmp sgt i32 %v11, 1
|
|
br i1 %v12, label %b2, label %b1
|
|
}
|
|
|
|
; Function Attrs: nounwind readnone
|
|
declare i32 @llvm.hexagon.A2.subsat(i32, i32) #1
|
|
|
|
attributes #0 = { nounwind optsize "target-cpu"="hexagonv60" }
|
|
attributes #1 = { nounwind readnone }
|