[Tests] Remove unnecessary bitcasts from opaque pointer tests (NFC)

Previously left these behind due to the required instruction renumbering, drop them now. This more accurately represents opaque pointer input IR. Also drop duplicate opaque pointer check lines in one SROA test.
llvm · Jun 22, 2022 · 60a3215 · 60a3215
1 parent 615efc3
commit 60a3215
Show file tree

Hide file tree

Showing 23 changed files with 162 additions and 730 deletions.
diff --git a/llvm/test/Transforms/EarlyCSE/AArch64/intrinsics.ll b/llvm/test/Transforms/EarlyCSE/AArch64/intrinsics.ll
@@ -19,14 +19,12 @@ for.cond:                                         ; preds = %for.body, %entry
   br i1 %cmp, label %for.body, label %for.end
 
 for.body:                                         ; preds = %for.cond
-  %0 = bitcast ptr %a to ptr
-  %1 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
-  %2 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %0 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
+  %1 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %2 = bitcast <16 x i8> %0 to <4 x i32>
   %3 = bitcast <16 x i8> %1 to <4 x i32>
-  %4 = bitcast <16 x i8> %2 to <4 x i32>
-  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %3, <4 x i32> %4, ptr %0)
-  %5 = bitcast ptr %a to ptr
-  %vld2 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %5)
+  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %2, <4 x i32> %3, ptr %a)
+  %vld2 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %a)
   %vld2.fca.0.extract = extractvalue { <4 x i32>, <4 x i32> } %vld2, 0
   %vld2.fca.1.extract = extractvalue { <4 x i32>, <4 x i32> } %vld2, 1
   %call = call <4 x i32> @vaddq_s32(<4 x i32> %vld2.fca.0.extract, <4 x i32> %vld2.fca.0.extract)
@@ -54,15 +52,13 @@ for.cond:                                         ; preds = %for.body, %entry
   br i1 %cmp, label %for.body, label %for.end
 
 for.body:                                         ; preds = %for.cond
-  %0 = bitcast ptr %a to ptr
-  %1 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
-  %2 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %0 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
+  %1 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %2 = bitcast <16 x i8> %0 to <4 x i32>
   %3 = bitcast <16 x i8> %1 to <4 x i32>
-  %4 = bitcast <16 x i8> %2 to <4 x i32>
-  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %3, <4 x i32> %3, ptr %0)
-  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %3, <4 x i32> %4, ptr %0)
-  %5 = bitcast ptr %a to ptr
-  %vld2 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %5)
+  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %2, <4 x i32> %2, ptr %a)
+  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %2, <4 x i32> %3, ptr %a)
+  %vld2 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %a)
   %vld2.fca.0.extract = extractvalue { <4 x i32>, <4 x i32> } %vld2, 0
   %vld2.fca.1.extract = extractvalue { <4 x i32>, <4 x i32> } %vld2, 1
   %call = call <4 x i32> @vaddq_s32(<4 x i32> %vld2.fca.0.extract, <4 x i32> %vld2.fca.0.extract)
@@ -90,12 +86,10 @@ for.cond:                                         ; preds = %for.body, %entry
   br i1 %cmp, label %for.body, label %for.end
 
 for.body:                                         ; preds = %for.cond
-  %0 = bitcast ptr %a to ptr
-  %vld2 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %0)
+  %vld2 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %a)
   %vld2.fca.0.extract = extractvalue { <4 x i32>, <4 x i32> } %vld2, 0
   %vld2.fca.1.extract = extractvalue { <4 x i32>, <4 x i32> } %vld2, 1
-  %1 = bitcast ptr %a to ptr
-  %vld22 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %1)
+  %vld22 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %a)
   %vld22.fca.0.extract = extractvalue { <4 x i32>, <4 x i32> } %vld22, 0
   %vld22.fca.1.extract = extractvalue { <4 x i32>, <4 x i32> } %vld22, 1
   %call = call <4 x i32> @vaddq_s32(<4 x i32> %vld2.fca.0.extract, <4 x i32> %vld22.fca.0.extract)
@@ -124,15 +118,13 @@ for.cond:                                         ; preds = %for.body, %entry
   br i1 %cmp, label %for.body, label %for.end
 
 for.body:                                         ; preds = %for.cond
-  %0 = bitcast ptr %a to ptr
-  %1 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
-  %2 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %0 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
+  %1 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %2 = bitcast <16 x i8> %0 to <4 x i32>
   %3 = bitcast <16 x i8> %1 to <4 x i32>
-  %4 = bitcast <16 x i8> %2 to <4 x i32>
-  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %3, <4 x i32> %4, ptr %0)
+  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %2, <4 x i32> %3, ptr %a)
   store i32 0, ptr %b, align 4
-  %5 = bitcast ptr %a to ptr
-  %vld2 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %5)
+  %vld2 = call { <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld2.v4i32.p0(ptr %a)
   %vld2.fca.0.extract = extractvalue { <4 x i32>, <4 x i32> } %vld2, 0
   %vld2.fca.1.extract = extractvalue { <4 x i32>, <4 x i32> } %vld2, 1
   %call = call <4 x i32> @vaddq_s32(<4 x i32> %vld2.fca.0.extract, <4 x i32> %vld2.fca.0.extract)
@@ -160,14 +152,12 @@ for.cond:                                         ; preds = %for.body, %entry
   br i1 %cmp, label %for.body, label %for.end
 
 for.body:                                         ; preds = %for.cond
-  %0 = bitcast ptr %a to ptr
-  %1 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
-  %2 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %0 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
+  %1 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %2 = bitcast <16 x i8> %0 to <4 x i32>
   %3 = bitcast <16 x i8> %1 to <4 x i32>
-  %4 = bitcast <16 x i8> %2 to <4 x i32>
-  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %3, <4 x i32> %4, ptr %0)
-  %5 = bitcast ptr %a to ptr
-  %vld3 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld3.v4i32.p0(ptr %5)
+  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %2, <4 x i32> %3, ptr %a)
+  %vld3 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld3.v4i32.p0(ptr %a)
   %vld3.fca.0.extract = extractvalue { <4 x i32>, <4 x i32>, <4 x i32> } %vld3, 0
   %vld3.fca.2.extract = extractvalue { <4 x i32>, <4 x i32>, <4 x i32> } %vld3, 2
   %call = call <4 x i32> @vaddq_s32(<4 x i32> %vld3.fca.0.extract, <4 x i32> %vld3.fca.2.extract)
@@ -196,15 +186,13 @@ for.cond:                                         ; preds = %for.body, %entry
   br i1 %cmp, label %for.body, label %for.end
 
 for.body:                                         ; preds = %for.cond
-  %0 = bitcast ptr %a to ptr
-  %1 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
-  %2 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %0 = bitcast <4 x i32> %s.coerce.fca.0.extract to <16 x i8>
+  %1 = bitcast <4 x i32> %s.coerce.fca.1.extract to <16 x i8>
+  %2 = bitcast <16 x i8> %0 to <4 x i32>
   %3 = bitcast <16 x i8> %1 to <4 x i32>
-  %4 = bitcast <16 x i8> %2 to <4 x i32>
-  call void @llvm.aarch64.neon.st3.v4i32.p0(<4 x i32> %4, <4 x i32> %3, <4 x i32> %3, ptr %0)
-  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %3, <4 x i32> %3, ptr %0)
-  %5 = bitcast ptr %a to ptr
-  %vld3 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld3.v4i32.p0(ptr %5)
+  call void @llvm.aarch64.neon.st3.v4i32.p0(<4 x i32> %3, <4 x i32> %2, <4 x i32> %2, ptr %a)
+  call void @llvm.aarch64.neon.st2.v4i32.p0(<4 x i32> %2, <4 x i32> %2, ptr %a)
+  %vld3 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld3.v4i32.p0(ptr %a)
   %vld3.fca.0.extract = extractvalue { <4 x i32>, <4 x i32>, <4 x i32> } %vld3, 0
   %vld3.fca.1.extract = extractvalue { <4 x i32>, <4 x i32>, <4 x i32> } %vld3, 1
   %call = call <4 x i32> @vaddq_s32(<4 x i32> %vld3.fca.0.extract, <4 x i32> %vld3.fca.0.extract)

diff --git a/llvm/test/Transforms/EarlyCSE/X86/preserve_memoryssa.ll b/llvm/test/Transforms/EarlyCSE/X86/preserve_memoryssa.ll
@@ -95,20 +95,17 @@ entry:
   br i1 undef, label %if.then274, label %for.cond404
 
 if.then274:                                       ; preds = %if.end256
-  %0 = bitcast ptr %exl.i to ptr
   %arrayidx.i = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 1
   %arrayidx1.i = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 2
   %arrayidx2.i = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 3
   %arrayidx3.i = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 4
-  %1 = bitcast ptr %exl.i to ptr
   %arrayidx.i1034 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 1
   %arrayidx1.i1035 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 2
   %arrayidx2.i1036 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 3
   %arrayidx3.i1037 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 4
   unreachable
 
 for.cond404:                                      ; preds = %if.end256
-  %2 = bitcast ptr %exl.i to ptr
   %arrayidx.i960 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 1
   %arrayidx1.i961 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 2
   %arrayidx2.i962 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 3
@@ -117,14 +114,13 @@ for.cond404:                                      ; preds = %if.end256
   %arrayidx3.i963 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 4
 
   ; MemoryUse(LoE)
-  %3 = load ptr, ptr @gnodeArray, align 8
-  %arrayidx6.i968 = getelementptr inbounds ptr, ptr %3, i64 undef
+  %0 = load ptr, ptr @gnodeArray, align 8
+  %arrayidx6.i968 = getelementptr inbounds ptr, ptr %0, i64 undef
   ; MemoryUse(1) MayAlias
-  %4 = load ptr, ptr %arrayidx6.i968, align 8
+  %1 = load ptr, ptr %arrayidx6.i968, align 8
   br i1 undef, label %for.cond26.preheader.i974, label %if.then20.for.body_crit_edge.i999
 
 for.cond26.preheader.i974:                        ; preds = %if.then20.i996
-  %5 = bitcast ptr %exl.i to ptr
   %arrayidx.i924 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 1
   %arrayidx1.i925 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 2
   %arrayidx2.i926 = getelementptr inbounds [5 x i32], ptr %exl.i, i64 0, i64 3

diff --git a/llvm/test/Transforms/InstSimplify/ConstProp/timeout.ll b/llvm/test/Transforms/InstSimplify/ConstProp/timeout.ll
@@ -23,50 +23,47 @@ for.cond:                                         ; preds = %for.inc, %entry
 
 for.body:                                         ; preds = %for.cond
   %1 = load ptr, ptr %s, align 4
-  %2 = bitcast ptr %1 to ptr
-  %add.ptr = getelementptr inbounds i8, ptr %2, i32 4
-  %3 = ptrtoint ptr %add.ptr to i32
-  %4 = load ptr, ptr %s, align 4
-  %5 = bitcast ptr %4 to ptr
-  %add.ptr1 = getelementptr inbounds i8, ptr %5, i32 4
-  %6 = ptrtoint ptr %add.ptr1 to i32
-  %rem = urem i32 %6, 2
+  %add.ptr = getelementptr inbounds i8, ptr %1, i32 4
+  %2 = ptrtoint ptr %add.ptr to i32
+  %3 = load ptr, ptr %s, align 4
+  %add.ptr1 = getelementptr inbounds i8, ptr %3, i32 4
+  %4 = ptrtoint ptr %add.ptr1 to i32
+  %rem = urem i32 %4, 2
   %cmp2 = icmp eq i32 %rem, 0
   br i1 %cmp2, label %cond.true, label %cond.false
 
 cond.true:                                        ; preds = %for.body
   br label %cond.end
 
 cond.false:                                       ; preds = %for.body
-  %7 = load ptr, ptr %s, align 4
-  %8 = bitcast ptr %7 to ptr
-  %add.ptr3 = getelementptr inbounds i8, ptr %8, i32 4
-  %9 = ptrtoint ptr %add.ptr3 to i32
-  %rem4 = urem i32 %9, 2
+  %5 = load ptr, ptr %s, align 4
+  %add.ptr3 = getelementptr inbounds i8, ptr %5, i32 4
+  %6 = ptrtoint ptr %add.ptr3 to i32
+  %rem4 = urem i32 %6, 2
   br label %cond.end
 
 cond.end:                                         ; preds = %cond.false, %cond.true
   %cond = phi i32 [ 0, %cond.true ], [ %rem4, %cond.false ]
-  %add = add i32 %3, %cond
-  %10 = inttoptr i32 %add to ptr
-  %11 = load ptr, ptr %s, align 4
-  %next = getelementptr inbounds %struct.ST, ptr %11, i32 0, i32 0
-  store ptr %10, ptr %next, align 4
-  %12 = load ptr, ptr %s, align 4
-  %next5 = getelementptr inbounds %struct.ST, ptr %12, i32 0, i32 0
-  %13 = load ptr, ptr %next5, align 4
-  store ptr %13, ptr %s, align 4
+  %add = add i32 %2, %cond
+  %7 = inttoptr i32 %add to ptr
+  %8 = load ptr, ptr %s, align 4
+  %next = getelementptr inbounds %struct.ST, ptr %8, i32 0, i32 0
+  store ptr %7, ptr %next, align 4
+  %9 = load ptr, ptr %s, align 4
+  %next5 = getelementptr inbounds %struct.ST, ptr %9, i32 0, i32 0
+  %10 = load ptr, ptr %next5, align 4
+  store ptr %10, ptr %s, align 4
   br label %for.inc
 
 for.inc:                                          ; preds = %cond.end
-  %14 = load i32, ptr %j, align 4
-  %inc = add nsw i32 %14, 1
+  %11 = load i32, ptr %j, align 4
+  %inc = add nsw i32 %11, 1
   store i32 %inc, ptr %j, align 4
   br label %for.cond
 
 for.end:                                          ; preds = %for.cond
-  %15 = load ptr, ptr %s, align 4
-  %next6 = getelementptr inbounds %struct.ST, ptr %15, i32 0, i32 0
+  %12 = load ptr, ptr %s, align 4
+  %next6 = getelementptr inbounds %struct.ST, ptr %12, i32 0, i32 0
   store ptr null, ptr %next6, align 4
   ret void
 }

diff --git a/llvm/test/Transforms/InstSimplify/require-dominator.ll b/llvm/test/Transforms/InstSimplify/require-dominator.ll
@@ -12,14 +12,13 @@ define void @foo(ptr) #1 align 2 {
 
 ; <label>:2:
   %3 = tail call ptr @_Znwm(i64 56) #10
-  %4 = bitcast ptr %3 to ptr
   %p = load ptr, ptr undef, align 8
-  %5 = icmp eq ptr %p, %4
-  br i1 %5, label %exit, label %6
+  %4 = icmp eq ptr %p, %3
+  br i1 %4, label %exit, label %5
 
 ; <label>:6:
-  %7 = icmp eq ptr %p, null
-  br i1 %7, label %exit, label %8
+  %6 = icmp eq ptr %p, null
+  br i1 %6, label %exit, label %7
 
 ; <label>:8:
   br label %exit

diff --git a/llvm/test/Transforms/InstSimplify/simplify-nested-bitcast.ll b/llvm/test/Transforms/InstSimplify/simplify-nested-bitcast.ll
@@ -19,10 +19,8 @@ entry:
 define i32 @f0(ptr addrspace(1), ptr addrspace(4)) #1 {
 entry:
   %2 = getelementptr inbounds %struct.__block_literal_generic, ptr addrspace(4) %1, i64 0, i32 3
-  %3 = bitcast ptr addrspace(4) %1 to ptr addrspace(4)
-  %4 = bitcast ptr addrspace(4) %2 to ptr addrspace(4)
-  %5 = load ptr, ptr addrspace(4) %4, align 8
-  %call = tail call i32 %5(ptr addrspace(4) %3, ptr addrspace(1) %0) #2
+  %3 = load ptr, ptr addrspace(4) %2, align 8
+  %call = tail call i32 %3(ptr addrspace(4) %1, ptr addrspace(1) %0) #2
   ret i32 %call
 }
 

diff --git a/llvm/test/Transforms/SROA/address-spaces.ll b/llvm/test/Transforms/SROA/address-spaces.ll
@@ -65,8 +65,7 @@ define void @copy_struct([5 x i64] %in.coerce, ptr addrspace(1) align 4 %ptr) {
 ;
 for.end:
   %in = alloca %struct.struct_test_27.0.13, align 8
-  %0 = bitcast ptr %in to ptr
-  store [5 x i64] %in.coerce, ptr %0, align 8
+  store [5 x i64] %in.coerce, ptr %in, align 8
   %scevgep9 = getelementptr %struct.struct_test_27.0.13, ptr %in, i32 0, i32 4, i32 0
   call void @llvm.memcpy.p1.p0.i32(ptr addrspace(1) align 4 %ptr, ptr align 4 %scevgep9, i32 16, i1 false)
   ret void
@@ -87,10 +86,8 @@ define void @pr27557() {
 ; CHECK-NEXT:    ret void
 ;
   %1 = alloca %union.anon, align 8
-  %2 = bitcast ptr %1 to ptr
-  store ptr @g, ptr %2, align 8
-  %3 = bitcast ptr %1 to ptr
-  store ptr addrspace(3) @l, ptr %3, align 8
+  store ptr @g, ptr %1, align 8
+  store ptr addrspace(3) @l, ptr %1, align 8
   ret void
 }
 
@@ -103,11 +100,9 @@ define ptr @pr27557.alt() {
 ; CHECK-NEXT:    ret ptr inttoptr (i64 ptrtoint (ptr addrspace(2) @l2 to i64) to ptr)
 ;
   %1 = alloca %union.anon, align 8
-  %2 = bitcast ptr %1 to ptr
-  store ptr addrspace(2) @l2, ptr %2, align 8
-  %3 = bitcast ptr %1 to ptr
-  %4 = load ptr, ptr %3, align 8
-  ret ptr %4
+  store ptr addrspace(2) @l2, ptr %1, align 8
+  %2 = load ptr, ptr %1, align 8
+  ret ptr %2
 }
 
 ; Make sure pre-splitting doesn't try to introduce an illegal bitcast

diff --git a/llvm/test/Transforms/SROA/alignment.ll b/llvm/test/Transforms/SROA/alignment.ll
@@ -219,8 +219,7 @@ define dso_local i32 @pr45010(ptr %A) {
   %B = alloca %struct, align 4
   %1 = load i32, ptr %A, align 4
   store atomic volatile i32 %1, ptr %B release, align 4
-  %2 = bitcast ptr %B to ptr
-  %x = load atomic volatile i32, ptr %2 acquire, align 4
+  %x = load atomic volatile i32, ptr %B acquire, align 4
   ret i32 %x
 }
 

diff --git a/llvm/test/Transforms/SROA/alloca-address-space.ll b/llvm/test/Transforms/SROA/alloca-address-space.ll
@@ -64,8 +64,7 @@ define void @copy_struct([5 x i64] %in.coerce, ptr addrspace(1) align 4 %ptr) {
 ;
 for.end:
   %in = alloca %struct.struct_test_27.0.13, align 8, addrspace(2)
-  %0 = bitcast ptr addrspace(2) %in to ptr addrspace(2)
-  store [5 x i64] %in.coerce, ptr addrspace(2) %0, align 8
+  store [5 x i64] %in.coerce, ptr addrspace(2) %in, align 8
   %scevgep9 = getelementptr %struct.struct_test_27.0.13, ptr addrspace(2) %in, i32 0, i32 4, i32 0
   call void @llvm.memcpy.p1.p2.i32(ptr addrspace(1) align 4 %ptr, ptr addrspace(2) align 4 %scevgep9, i32 16, i1 false)
   ret void
@@ -86,10 +85,8 @@ define void @pr27557() {
 ; CHECK-NEXT:    ret void
 ;
   %1 = alloca %union.anon, align 8, addrspace(2)
-  %2 = bitcast ptr addrspace(2) %1 to ptr addrspace(2)
-  store ptr @g, ptr addrspace(2) %2, align 8
-  %3 = bitcast ptr addrspace(2) %1 to ptr addrspace(2)
-  store ptr addrspace(3) @l, ptr addrspace(2) %3, align 8
+  store ptr @g, ptr addrspace(2) %1, align 8
+  store ptr addrspace(3) @l, ptr addrspace(2) %1, align 8
   ret void
 }
 
@@ -102,11 +99,9 @@ define ptr @pr27557.alt() {
 ; CHECK-NEXT:    ret ptr inttoptr (i64 ptrtoint (ptr addrspace(4) @l4 to i64) to ptr)
 ;
   %1 = alloca %union.anon, align 8, addrspace(2)
-  %2 = bitcast ptr addrspace(2) %1 to ptr addrspace(2)
-  store ptr addrspace(4) @l4, ptr addrspace(2) %2, align 8
-  %3 = bitcast ptr addrspace(2) %1 to ptr addrspace(2)
-  %4 = load ptr, ptr addrspace(2) %3, align 8
-  ret ptr %4
+  store ptr addrspace(4) @l4, ptr addrspace(2) %1, align 8
+  %2 = load ptr, ptr addrspace(2) %1, align 8
+  ret ptr %2
 }
 
 ; Test load from and store to non-zero address space.

diff --git a/llvm/test/Transforms/SROA/assume.ll b/llvm/test/Transforms/SROA/assume.ll
@@ -19,9 +19,7 @@ define void @_ZN2bg2baIiEC2ES_(i64 %v) {
 entry:
   %y = alloca i64, align 8
   store i64 %v, ptr %y, align 8
-  %0 = bitcast ptr %y to ptr
-  %1 = bitcast ptr %0 to ptr
-  call void @llvm.assume(i1 true) [ "nonnull"(ptr %1) ]
+  call void @llvm.assume(i1 true) [ "nonnull"(ptr %y) ]
   ret void
 }