RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-22 18:54:02 +01:00
Code Issues Projects Releases Wiki Activity
bedf13b1a7
llvm-mirror/test/Transforms/InstCombine/memcpy-from-global.ll
Andy Kaylor 8be74d232c Fixing an infinite loop problem in InstCombine 
Patch by Mohammad Fawaz

This issues started happening after
b373b5990d
Basically, if the memcpy is volatile, the collectUsers() function should
return false, just like we do for volatile loads.

Differential Revision: https://reviews.llvm.org/D106950

(cherry picked from commit b4d945bacdaf2c60dd5fdb119b90cced73c41beb)
2021-08-04 16:51:40 -07:00

364 lines
19 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -instcombine -S | FileCheck %s
target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64"
@C.0.1248 = internal constant [128 x float] [ float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00 ], align 32 ; <[128 x float]*> [#uses=1]
define float @test1(i32 %hash, float %x, float %y, float %z, float %w) {
; CHECK-LABEL: @test1(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP3:%.*]] = shl i32 [[HASH:%.*]], 2
; CHECK-NEXT: [[TMP5:%.*]] = and i32 [[TMP3]], 124
; CHECK-NEXT: [[TMP0:%.*]] = zext i32 [[TMP5]] to i64
; CHECK-NEXT: [[TMP753:%.*]] = getelementptr [128 x float], [128 x float]* @C.0.1248, i64 0, i64 [[TMP0]]
; CHECK-NEXT: [[TMP9:%.*]] = load float, float* [[TMP753]], align 16
; CHECK-NEXT: [[TMP11:%.*]] = fmul float [[TMP9]], [[X:%.*]]
; CHECK-NEXT: [[TMP13:%.*]] = fadd float [[TMP11]], 0.000000e+00
; CHECK-NEXT: [[TMP17_SUM52:%.*]] = or i32 [[TMP5]], 1
; CHECK-NEXT: [[TMP1:%.*]] = zext i32 [[TMP17_SUM52]] to i64
; CHECK-NEXT: [[TMP1851:%.*]] = getelementptr [128 x float], [128 x float]* @C.0.1248, i64 0, i64 [[TMP1]]
; CHECK-NEXT: [[TMP19:%.*]] = load float, float* [[TMP1851]], align 4
; CHECK-NEXT: [[TMP21:%.*]] = fmul float [[TMP19]], [[Y:%.*]]
; CHECK-NEXT: [[TMP23:%.*]] = fadd float [[TMP21]], [[TMP13]]
; CHECK-NEXT: [[TMP27_SUM50:%.*]] = or i32 [[TMP5]], 2
; CHECK-NEXT: [[TMP2:%.*]] = zext i32 [[TMP27_SUM50]] to i64
; CHECK-NEXT: [[TMP2849:%.*]] = getelementptr [128 x float], [128 x float]* @C.0.1248, i64 0, i64 [[TMP2]]
; CHECK-NEXT: [[TMP29:%.*]] = load float, float* [[TMP2849]], align 8
; CHECK-NEXT: [[TMP31:%.*]] = fmul float [[TMP29]], [[Z:%.*]]
; CHECK-NEXT: [[TMP33:%.*]] = fadd float [[TMP31]], [[TMP23]]
; CHECK-NEXT: [[TMP37_SUM48:%.*]] = or i32 [[TMP5]], 3
; CHECK-NEXT: [[TMP3:%.*]] = zext i32 [[TMP37_SUM48]] to i64
; CHECK-NEXT: [[TMP3847:%.*]] = getelementptr [128 x float], [128 x float]* @C.0.1248, i64 0, i64 [[TMP3]]
; CHECK-NEXT: [[TMP39:%.*]] = load float, float* [[TMP3847]], align 4
; CHECK-NEXT: [[TMP41:%.*]] = fmul float [[TMP39]], [[W:%.*]]
; CHECK-NEXT: [[TMP43:%.*]] = fadd float [[TMP41]], [[TMP33]]
; CHECK-NEXT: ret float [[TMP43]]
;
entry:
%lookupTable = alloca [128 x float], align 16 ; <[128 x float]*> [#uses=5]
%lookupTable1 = bitcast [128 x float]* %lookupTable to i8* ; <i8*> [#uses=1]
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 %lookupTable1, i8* align 16 bitcast ([128 x float]* @C.0.1248 to i8*), i64 512, i1 false)
%tmp3 = shl i32 %hash, 2 ; <i32> [#uses=1]
%tmp5 = and i32 %tmp3, 124 ; <i32> [#uses=4]
%tmp753 = getelementptr [128 x float], [128 x float]* %lookupTable, i32 0, i32 %tmp5 ; <float*> [#uses=1]
%tmp9 = load float, float* %tmp753 ; <float> [#uses=1]
%tmp11 = fmul float %tmp9, %x ; <float> [#uses=1]
%tmp13 = fadd float %tmp11, 0.000000e+00 ; <float> [#uses=1]
%tmp17.sum52 = or i32 %tmp5, 1 ; <i32> [#uses=1]
%tmp1851 = getelementptr [128 x float], [128 x float]* %lookupTable, i32 0, i32 %tmp17.sum52 ; <float*> [#uses=1]
%tmp19 = load float, float* %tmp1851 ; <float> [#uses=1]
%tmp21 = fmul float %tmp19, %y ; <float> [#uses=1]
%tmp23 = fadd float %tmp21, %tmp13 ; <float> [#uses=1]
%tmp27.sum50 = or i32 %tmp5, 2 ; <i32> [#uses=1]
%tmp2849 = getelementptr [128 x float], [128 x float]* %lookupTable, i32 0, i32 %tmp27.sum50 ; <float*> [#uses=1]
%tmp29 = load float, float* %tmp2849 ; <float> [#uses=1]
%tmp31 = fmul float %tmp29, %z ; <float> [#uses=1]
%tmp33 = fadd float %tmp31, %tmp23 ; <float> [#uses=1]
%tmp37.sum48 = or i32 %tmp5, 3 ; <i32> [#uses=1]
%tmp3847 = getelementptr [128 x float], [128 x float]* %lookupTable, i32 0, i32 %tmp37.sum48 ; <float*> [#uses=1]
%tmp39 = load float, float* %tmp3847 ; <float> [#uses=1]
%tmp41 = fmul float %tmp39, %w ; <float> [#uses=1]
%tmp43 = fadd float %tmp41, %tmp33 ; <float> [#uses=1]
ret float %tmp43
}
declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i1) nounwind
declare void @llvm.memcpy.p1i8.p0i8.i64(i8 addrspace(1)* nocapture, i8* nocapture, i64, i1) nounwind
declare void @llvm.memcpy.p0i8.p1i8.i64(i8* nocapture, i8 addrspace(1)* nocapture, i64, i1) nounwind
declare void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* nocapture, i8 addrspace(1)* nocapture, i64, i1) nounwind
%T = type { i8, [123 x i8] }
%U = type { i32, i32, i32, i32, i32 }
@G = constant %T {i8 1, [123 x i8] zeroinitializer }
@H = constant [2 x %U] zeroinitializer, align 16
@I = internal addrspace(1) constant [4 x float] zeroinitializer , align 4
define void @test2() {
; CHECK-LABEL: @test2(
; CHECK-NEXT: [[B1:%.*]] = alloca [124 x i8], align 8
; CHECK-NEXT: [[B1_SUB:%.*]] = getelementptr inbounds [124 x i8], [124 x i8]* [[B1]], i64 0, i64 0
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 8 dereferenceable(124) [[B1_SUB]], i8* noundef nonnull align 16 dereferenceable(124) getelementptr inbounds (%T, %T* @G, i64 0, i32 0), i64 124, i1 false)
; CHECK-NEXT: call void @bar(i8* nonnull [[B1_SUB]])
; CHECK-NEXT: ret void
;
%A = alloca %T
%B = alloca %T
%a = bitcast %T* %A to i8*
%b = bitcast %T* %B to i8*
; %A alloca is deleted
; use @G instead of %A
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false)
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %b, i8* align 4 %a, i64 124, i1 false)
call void @bar(i8* %b)
ret void
}
define void @test2_no_null_opt() #0 {
; CHECK-LABEL: @test2_no_null_opt(
; CHECK-NEXT: [[B1:%.*]] = alloca [124 x i8], align 8
; CHECK-NEXT: [[B1_SUB:%.*]] = getelementptr inbounds [124 x i8], [124 x i8]* [[B1]], i64 0, i64 0
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef align 8 dereferenceable(124) [[B1_SUB]], i8* noundef align 16 dereferenceable(124) getelementptr inbounds (%T, %T* @G, i64 0, i32 0), i64 124, i1 false)
; CHECK-NEXT: call void @bar(i8* [[B1_SUB]])
; CHECK-NEXT: ret void
;
%A = alloca %T
%B = alloca %T
%a = bitcast %T* %A to i8*
%b = bitcast %T* %B to i8*
; %A alloca is deleted
; use @G instead of %A
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false)
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %b, i8* align 4 %a, i64 124, i1 false)
call void @bar(i8* %b)
ret void
}
define void @test2_addrspacecast() {
; CHECK-LABEL: @test2_addrspacecast(
; CHECK-NEXT: [[B1:%.*]] = alloca [124 x i8], align 8
; CHECK-NEXT: [[B1_SUB:%.*]] = getelementptr inbounds [124 x i8], [124 x i8]* [[B1]], i64 0, i64 0
; CHECK-NEXT: [[B:%.*]] = addrspacecast i8* [[B1_SUB]] to i8 addrspace(1)*
; CHECK-NEXT: call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* noundef align 4 dereferenceable(124) [[B]], i8 addrspace(1)* noundef align 4 dereferenceable(124) addrspacecast (i8* getelementptr inbounds (%T, %T* @G, i64 0, i32 0) to i8 addrspace(1)*), i64 124, i1 false)
; CHECK-NEXT: call void @bar_as1(i8 addrspace(1)* [[B]])
; CHECK-NEXT: ret void
;
%A = alloca %T
%B = alloca %T
%a = addrspacecast %T* %A to i8 addrspace(1)*
%b = addrspacecast %T* %B to i8 addrspace(1)*
; %A alloca is deleted
; This doesn't exactly match what test2 does, because folding the type
; cast into the alloca doesn't work for the addrspacecast yet.
; use @G instead of %A
call void @llvm.memcpy.p1i8.p0i8.i64(i8 addrspace(1)* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false)
call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* align 4 %b, i8 addrspace(1)* align 4 %a, i64 124, i1 false)
call void @bar_as1(i8 addrspace(1)* %b)
ret void
}
declare void @bar(i8*)
declare void @bar_as1(i8 addrspace(1)*)
;; Should be able to eliminate the alloca.
define void @test3() {
; CHECK-LABEL: @test3(
; CHECK-NEXT: call void @bar(i8* getelementptr inbounds (%T, %T* @G, i64 0, i32 0)) [[ATTR2:#.*]]
; CHECK-NEXT: ret void
;
%A = alloca %T
%a = bitcast %T* %A to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false)
call void @bar(i8* %a) readonly
ret void
}
define void @test3_addrspacecast() {
; CHECK-LABEL: @test3_addrspacecast(
; CHECK-NEXT: call void @bar(i8* getelementptr inbounds (%T, %T* @G, i64 0, i32 0)) [[ATTR2]]
; CHECK-NEXT: ret void
;
%A = alloca %T
%a = bitcast %T* %A to i8*
call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %a, i8 addrspace(1)* align 4 addrspacecast (%T* @G to i8 addrspace(1)*), i64 124, i1 false)
call void @bar(i8* %a) readonly
ret void
}
define void @test4() {
; CHECK-LABEL: @test4(
; CHECK-NEXT: call void @baz(i8* byval(i8) getelementptr inbounds (%T, %T* @G, i64 0, i32 0))
; CHECK-NEXT: ret void
;
%A = alloca %T
%a = bitcast %T* %A to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false)
call void @baz(i8* byval(i8) %a)
ret void
}
declare void @llvm.lifetime.start.p0i8(i64, i8*)
define void @test5() {
; CHECK-LABEL: @test5(
; CHECK-NEXT: call void @baz(i8* byval(i8) getelementptr inbounds (%T, %T* @G, i64 0, i32 0))
; CHECK-NEXT: ret void
;
%A = alloca %T
%a = bitcast %T* %A to i8*
call void @llvm.lifetime.start.p0i8(i64 -1, i8* %a)
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false)
call void @baz(i8* byval(i8) %a)
ret void
}
declare void @baz(i8* byval(i8))
define void @test6() {
; CHECK-LABEL: @test6(
; CHECK-NEXT: call void @bar(i8* bitcast ([2 x %U]* @H to i8*)) [[ATTR2]]
; CHECK-NEXT: ret void
;
%A = alloca %U, align 16
%a = bitcast %U* %A to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 %a, i8* align 16 bitcast ([2 x %U]* @H to i8*), i64 20, i1 false)
call void @bar(i8* %a) readonly
ret void
}
define void @test7() {
; CHECK-LABEL: @test7(
; CHECK-NEXT: call void @bar(i8* bitcast ([2 x %U]* @H to i8*)) [[ATTR2]]
; CHECK-NEXT: ret void
;
%A = alloca %U, align 16
%a = bitcast %U* %A to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 0) to i8*), i64 20, i1 false)
call void @bar(i8* %a) readonly
ret void
}
define void @test8() {
; CHECK-LABEL: @test8(
; CHECK-NEXT: [[AL:%.*]] = alloca [[U:%.*]], align 16
; CHECK-NEXT: [[A:%.*]] = bitcast %U* [[AL]] to i8*
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 16 dereferenceable(20) [[A]], i8* noundef nonnull align 4 dereferenceable(20) bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*), i64 20, i1 false)
; CHECK-NEXT: call void @bar(i8* nonnull [[A]]) [[ATTR2]]
; CHECK-NEXT: ret void
;
%al = alloca %U, align 16
%a = bitcast %U* %al to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8*), i64 20, i1 false)
call void @bar(i8* %a) readonly
ret void
}
define void @test8_addrspacecast() {
; CHECK-LABEL: @test8_addrspacecast(
; CHECK-NEXT: [[AL:%.*]] = alloca [[U:%.*]], align 16
; CHECK-NEXT: [[A:%.*]] = bitcast %U* [[AL]] to i8*
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p1i8.i64(i8* noundef nonnull align 16 dereferenceable(20) [[A]], i8 addrspace(1)* noundef align 4 dereferenceable(20) addrspacecast (i8* bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*) to i8 addrspace(1)*), i64 20, i1 false)
; CHECK-NEXT: call void @bar(i8* nonnull [[A]]) [[ATTR2]]
; CHECK-NEXT: ret void
;
%Al = alloca %U, align 16
%a = bitcast %U* %Al to i8*
call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %a, i8 addrspace(1)* align 4 addrspacecast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8 addrspace(1)*), i64 20, i1 false)
call void @bar(i8* %a) readonly
ret void
}
define void @test9() {
; CHECK-LABEL: @test9(
; CHECK-NEXT: call void @bar(i8* bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*)) [[ATTR2]]
; CHECK-NEXT: ret void
;
%A = alloca %U, align 4
%a = bitcast %U* %A to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8*), i64 20, i1 false)
call void @bar(i8* %a) readonly
ret void
}
define void @test9_addrspacecast() {
; CHECK-LABEL: @test9_addrspacecast(
; CHECK-NEXT: call void @bar(i8* bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*)) [[ATTR2]]
; CHECK-NEXT: ret void
;
%A = alloca %U, align 4
%a = bitcast %U* %A to i8*
call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %a, i8 addrspace(1)* align 4 addrspacecast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8 addrspace(1)*), i64 20, i1 false)
call void @bar(i8* %a) readonly
ret void
}
@bbb = local_unnamed_addr global [1000000 x i8] zeroinitializer, align 16
@_ZL3KKK = internal unnamed_addr constant [3 x i8] c"\01\01\02", align 1
; Should not replace alloca with global because of size mismatch.
define void @test9_small_global() {
; CHECK-LABEL: @test9_small_global(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[CC:%.*]] = alloca [1000000 x i8], align 16
; CHECK-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1000000 x i8], [1000000 x i8]* [[CC]], i64 0, i64 0
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 16 dereferenceable(3) [[ARRAYDECAY]], i8* noundef nonnull align 16 dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i64 0, i64 0), i64 3, i1 false)
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 16 dereferenceable(1000000) getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i64 0, i64 0), i8* noundef nonnull align 16 dereferenceable(1000000) [[ARRAYDECAY]], i64 1000000, i1 false)
; CHECK-NEXT: ret void
;
entry:
%cc = alloca [1000000 x i8], align 16
%cc.0..sroa_idx = getelementptr inbounds [1000000 x i8], [1000000 x i8]* %cc, i64 0, i64 0
%arraydecay = getelementptr inbounds [1000000 x i8], [1000000 x i8]* %cc, i32 0, i32 0
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %arraydecay, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i32 0, i32 0), i64 3, i1 false)
call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i32 0, i32 0), i8* align 16 %arraydecay, i64 1000000, i1 false)
ret void
}
; Should replace alloca with global as they have exactly the same size.
define void @test10_same_global() {
; CHECK-LABEL: @test10_same_global(
; CHECK-NEXT: entry:
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 16 dereferenceable(3) getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i64 0, i64 0), i8* noundef nonnull align 16 dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i64 0, i64 0), i64 3, i1 false)
; CHECK-NEXT: ret void
;
entry:
%cc = alloca [3 x i8], align 1
%cc.0..sroa_idx = getelementptr inbounds [3 x i8], [3 x i8]* %cc, i64 0, i64 0
%arraydecay = getelementptr inbounds [3 x i8], [3 x i8]* %cc, i32 0, i32 0
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %arraydecay, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i32 0, i32 0), i64 3, i1 false)
call void @llvm.memcpy.p0i8.p0i8.i64(i8* getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i32 0, i32 0), i8* %arraydecay, i64 3, i1 false)
ret void
}
; Should replace alloca with global even when the global is in a different address space
define float @test11(i64 %i) {
; CHECK-LABEL: @test11(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[GEP:%.*]] = getelementptr [4 x float], [4 x float] addrspace(1)* @I, i64 0, i64 %i
; CHECK-NEXT: [[LD:%.*]] = load float, float addrspace(1)* [[GEP]]
; CHECK-NEXT: ret float [[LD]]
entry:
%a = alloca [4 x float], align 4
%b = bitcast [4 x float]* %a to i8*
call void @llvm.lifetime.start.p0i8(i64 16, i8* %b)
call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %b, i8 addrspace(1)* align 4 bitcast ([4 x float] addrspace(1)* @I to i8 addrspace(1)*), i64 16, i1 false)
%g = getelementptr inbounds [4 x float], [4 x float]* %a, i64 0, i64 %i
%r = load float, float* %g, align 4
ret float %r
}
; If the memcpy is volatile, it should not be removed
define float @test11_volatile(i64 %i) {
; CHECK-LABEL: @test11_volatile(
; CHECK-NEXT: entry:
; CHECK-NEXT: alloca
; CHECK: call void @llvm.lifetime.start.p0i8
; CHECK: call void @llvm.memcpy
entry:
%a = alloca [4 x float], align 4
%b = bitcast [4 x float]* %a to i8*
call void @llvm.lifetime.start.p0i8(i64 16, i8* %b)
call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %b, i8 addrspace(1)* align 4 bitcast ([4 x float] addrspace(1)* @I to i8 addrspace(1)*), i64 16, i1 true)
%g = getelementptr inbounds [4 x float], [4 x float]* %a, i64 0, i64 %i
%r = load float, float* %g, align 4
ret float %r
}
attributes #0 = { null_pointer_is_valid }
Reference in New Issue View Git Blame Copy Permalink