RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-22 04:22:57 +02:00
Code Issues Projects Releases Wiki Activity

[Hexagon] Remove incorrect pattern for swiz

Browse Source 
The pattern had a couple of problems:
- It was checking for loads of bytes in the reverse order to what it
  should have been looking for.
- It would replace loads of bytes with a load of a word without making
  sure that the alignment was correct.

Thanks to Eli Friedman for pointing it out.

llvm-svn: 343514
This commit is contained in:
Krzysztof Parzyszek 2018-10-01 18:24:40 +00:00
parent 42e8418c19
commit d6bbcef5f6
2 changed files with 0 additions and 59 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
lib/Target/Hexagon/HexagonPatterns.td
View File

@ -1119,14 +1119,6 @@ def: Pat<(or (or (or (shl (Zext64 (and I32:$b, (i32 65535))), (i32 16)),
(Combinew (A2_combine_ll I32:$d, I32:$c),
(A2_combine_ll I32:$b, I32:$a))>;
def: Pat<(or (or (shl (or (shl (i32 (extloadi8 (add I32:$b, 3))),
(i32 8)),
(i32 (zextloadi8 (add I32:$b, 2)))),
(i32 16)),
(shl (i32 (zextloadi8 (add I32:$b, 1))), (i32 8))),
(zextloadi8 I32:$b)),
(A2_swiz (L2_loadri_io IntRegs:$b, 0))>;
let AddedComplexity = 200 in {
def: Pat<(or (shl I32:$Rt, (i32 16)), (and I32:$Rs, (i32 65535))),
(A2_combine_ll I32:$Rt, I32:$Rs)>;

51
test/CodeGen/Hexagon/swiz.ll
View File

@ -1,51 +0,0 @@
; RUN: llc -march=hexagon < %s | FileCheck %s
; CHECK: swiz
target triple = "hexagon"
%s.0 = type { [4 x i32], [2 x i32], [64 x i8] }
; Function Attrs: nounwind
define void @f0(%s.0* nocapture %a0, i8* nocapture %a1, i32 %a2) #0 {
b0:
%v0 = getelementptr inbounds %s.0, %s.0* %a0, i32 0, i32 2, i32 0
br label %b1
b1: ; preds = %b1, %b0
%v1 = phi i16 [ 16, %b0 ], [ %v22, %b1 ]
%v2 = phi i8* [ %v0, %b0 ], [ %v21, %b1 ]
%v3 = getelementptr inbounds i8, i8* %v2, i32 3
%v4 = load i8, i8* %v3, align 1, !tbaa !0
%v5 = zext i8 %v4 to i32
%v6 = shl nuw nsw i32 %v5, 8
%v7 = getelementptr inbounds i8, i8* %v2, i32 2
%v8 = load i8, i8* %v7, align 1, !tbaa !0
%v9 = zext i8 %v8 to i32
%v10 = or i32 %v6, %v9
%v11 = shl nuw i32 %v10, 16
%v12 = getelementptr inbounds i8, i8* %v2, i32 1
%v13 = load i8, i8* %v12, align 1, !tbaa !0
%v14 = zext i8 %v13 to i32
%v15 = shl nuw nsw i32 %v14, 8
%v16 = load i8, i8* %v2, align 1, !tbaa !0
%v17 = zext i8 %v16 to i32
%v18 = or i32 %v11, %v15
%v19 = or i32 %v18, %v17
%v20 = bitcast i8* %v2 to i32*
store i32 %v19, i32* %v20, align 4, !tbaa !3
%v21 = getelementptr inbounds i8, i8* %v2, i32 4
%v22 = add i16 %v1, -1
%v23 = icmp eq i16 %v22, 0
br i1 %v23, label %b2, label %b1
b2: ; preds = %b1
ret void
}
attributes #0 = { nounwind }
!0 = !{!1, !1, i64 0}
!1 = !{!"omnipotent char", !2}
!2 = !{!"Simple C/C++ TBAA"}
!3 = !{!4, !4, i64 0}
!4 = !{!"int", !1}