Instcombine: destructor loads of structs that do not contains padding

For non padded structs, we can just proceed and deaggregate them.
We don't want ot do this when there is padding in the struct as to not
lose information about this padding (the subsequents passes would then
try hard to preserve the padding, which is undesirable).

Also update extractvalue.ll and cast.ll so that they use structs with padding.

Remove the FIXME in the extractvalue of laod case as the non padded case is
handled when processing the load, and we don't want to do it on the padded
case.

Patch by: Amaury SECHET <deadalnix@gmail.com>

Differential Revision: http://reviews.llvm.org/D14483

From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 255600

3 files changed, 108 insertions, 84 deletions

diff --git a/llvm/test/Transforms/InstCombine/cast.ll b/llvm/test/Transforms/InstCombine/cast.ll
index f8a94edfb7f..016b6aa6455 100644
--- a/llvm/test/Transforms/InstCombine/cast.ll
+++ b/llvm/test/Transforms/InstCombine/cast.ll
@@ -722,7 +722,7 @@ define i1 @test67(i1 %a, i32 %b) {
 ; CHECK: ret i1 false
 }
 
-%s = type { i32, i32, i32 }
+%s = type { i32, i32, i16 }
 
 define %s @test68(%s *%p, i64 %i) {
 ; CHECK-LABEL: @test68(
diff --git a/llvm/test/Transforms/InstCombine/extractvalue.ll b/llvm/test/Transforms/InstCombine/extractvalue.ll
index 6319590873a..9c293581a06 100644
--- a/llvm/test/Transforms/InstCombine/extractvalue.ll
+++ b/llvm/test/Transforms/InstCombine/extractvalue.ll
@@ -48,16 +48,16 @@ define i32 @foo(i32 %a, i32 %b) {
 ; CHECK: call {{.*}}(i32 [[LOAD]])
 ; CHECK-NOT: extractvalue
 ; CHECK: ret i32 [[LOAD]]
-define i32 @extract2gep({i32, i32}* %pair, i32* %P) {
+define i32 @extract2gep({i16, i32}* %pair, i32* %P) {
         ; The load + extractvalue should be converted
         ; to an inbounds gep + smaller load.
         ; The new load should be in the same spot as the old load.
-        %L = load {i32, i32}, {i32, i32}* %pair
+        %L = load {i16, i32}, {i16, i32}* %pair
         store i32 0, i32* %P
         br label %loop
 
 loop:
-        %E = extractvalue {i32, i32} %L, 1
+        %E = extractvalue {i16, i32} %L, 1
         %C = call i32 @baz(i32 %E)
         store i32 %C, i32* %P
         %cond = icmp eq i32 %C, 0
@@ -67,17 +67,17 @@ end:
         ret i32 %E
 }
 
-; CHECK-LABEL: define i32 @doubleextract2gep(
+; CHECK-LABEL: define i16 @doubleextract2gep(
 ; CHECK-NEXT: [[GEP:%[a-z0-9]+]] = getelementptr inbounds {{.*}}, {{.*}}* %arg, i64 0, i32 1, i32 1
-; CHECK-NEXT: [[LOAD:%[A-Za-z0-9]+]] = load i32, i32* [[GEP]]
-; CHECK-NEXT: ret i32 [[LOAD]]
-define i32 @doubleextract2gep({i32, {i32, i32}}* %arg) {
+; CHECK-NEXT: [[LOAD:%[A-Za-z0-9]+]] = load i16, i16* [[GEP]]
+; CHECK-NEXT: ret i16 [[LOAD]]
+define i16 @doubleextract2gep({i16, {i32, i16}}* %arg) {
         ; The load + extractvalues should be converted
         ; to a 3-index inbounds gep + smaller load.
-        %L = load {i32, {i32, i32}}, {i32, {i32, i32}}* %arg
-        %E1 = extractvalue {i32, {i32, i32}} %L, 1
-        %E2 = extractvalue {i32, i32} %E1, 1
-        ret i32 %E2
+        %L = load {i16, {i32, i16}}, {i16, {i32, i16}}* %arg
+        %E1 = extractvalue {i16, {i32, i16}} %L, 1
+        %E2 = extractvalue {i32, i16} %E1, 1
+        ret i16 %E2
 }
 
 ; CHECK: define i32 @nogep-multiuse
diff --git a/llvm/test/Transforms/InstCombine/unpack-fca.ll b/llvm/test/Transforms/InstCombine/unpack-fca.ll
index 48bb157956a..9b8d1045749 100644
--- a/llvm/test/Transforms/InstCombine/unpack-fca.ll
+++ b/llvm/test/Transforms/InstCombine/unpack-fca.ll
@@ -5,110 +5,134 @@ target triple = "x86_64-unknown-linux-gnu"
 
 %A__vtbl = type { i8*, i32 (%A*)* }
 %A = type { %A__vtbl* }
+%B = type { i8*, i64 }
 
 @A__vtblZ = constant %A__vtbl { i8* null, i32 (%A*)* @A.foo }
 
 declare i32 @A.foo(%A* nocapture %this)
 
-declare i8* @allocmemory(i64)
-
-define void @storeA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to %A*
+define void @storeA(%A* %a.ptr) {
 ; CHECK-LABEL: storeA
-; CHECK: store %A__vtbl* @A__vtblZ
-  store %A { %A__vtbl* @A__vtblZ }, %A* %1, align 8
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds %A, %A* %a.ptr, i64 0, i32 0
+; CHECK-NEXT: store %A__vtbl* @A__vtblZ, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: ret void
+  store %A { %A__vtbl* @A__vtblZ }, %A* %a.ptr, align 8
+  ret void
+}
+
+define void @storeB(%B* %b.ptr) {
+; CHECK-LABEL: storeB
+; CHECK-NEXT: [[GEP1:%[a-z0-9\.]+]] = getelementptr inbounds %B, %B* %b.ptr, i64 0, i32 0
+; CHECK-NEXT: store i8* null, i8** [[GEP1]], align 8
+; CHECK-NEXT: [[GEP2:%[a-z0-9\.]+]] = getelementptr inbounds %B, %B* %b.ptr, i64 0, i32 1
+; CHECK-NEXT: store i64 42, i64* [[GEP2]], align 8
+; CHECK-NEXT: ret void
+  store %B { i8* null, i64 42 }, %B* %b.ptr, align 8
   ret void
 }
 
-define void @storeStructOfA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to { %A }*
+define void @storeStructOfA({ %A }* %sa.ptr) {
 ; CHECK-LABEL: storeStructOfA
-; CHECK: store %A__vtbl* @A__vtblZ
-  store { %A } { %A { %A__vtbl* @A__vtblZ } }, { %A }* %1, align 8
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds { %A }, { %A }* %sa.ptr, i64 0, i32 0, i32 0
+; CHECK-NEXT: store %A__vtbl* @A__vtblZ, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: ret void
+  store { %A } { %A { %A__vtbl* @A__vtblZ } }, { %A }* %sa.ptr, align 8
   ret void
 }
 
-define void @storeArrayOfA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to [1 x %A]*
+define void @storeArrayOfA([1 x %A]* %aa.ptr) {
 ; CHECK-LABEL: storeArrayOfA
-; CHECK: store %A__vtbl* @A__vtblZ
-  store [1 x %A] [%A { %A__vtbl* @A__vtblZ }], [1 x %A]* %1, align 8
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds [1 x %A], [1 x %A]* %aa.ptr, i64 0, i64 0, i32 0
+; CHECK-NEXT: store %A__vtbl* @A__vtblZ, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: ret void
+  store [1 x %A] [%A { %A__vtbl* @A__vtblZ }], [1 x %A]* %aa.ptr, align 8
   ret void
 }
 
-define void @storeStructOfArrayOfA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to { [1 x %A] }*
+define void @storeStructOfArrayOfA({ [1 x %A] }* %saa.ptr) {
 ; CHECK-LABEL: storeStructOfArrayOfA
-; CHECK: store %A__vtbl* @A__vtblZ
-  store { [1 x %A] } { [1 x %A] [%A { %A__vtbl* @A__vtblZ }] }, { [1 x %A] }* %1, align 8
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds { [1 x %A] }, { [1 x %A] }* %saa.ptr, i64 0, i32 0, i64 0, i32 0
+; CHECK-NEXT: store %A__vtbl* @A__vtblZ, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: ret void
+  store { [1 x %A] } { [1 x %A] [%A { %A__vtbl* @A__vtblZ }] }, { [1 x %A] }* %saa.ptr, align 8
   ret void
 }
 
-define %A @loadA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to %A*
+define %A @loadA(%A* %a.ptr) {
 ; CHECK-LABEL: loadA
-; CHECK: load %A__vtbl*,
-; CHECK: insertvalue %A undef, %A__vtbl* {{.*}}, 0
-  %2 = load %A, %A* %1, align 8
-  ret %A %2
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds %A, %A* %a.ptr, i64 0, i32 0
+; CHECK-NEXT: [[LOAD:%[a-z0-9\.]+]] = load %A__vtbl*, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: [[IV:%[a-z0-9\.]+]] = insertvalue %A undef, %A__vtbl* [[LOAD]], 0
+; CHECK-NEXT: ret %A [[IV]]
+  %1 = load %A, %A* %a.ptr, align 8
+  ret %A %1
 }
 
-define { %A } @loadStructOfA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to { %A }*
+define %B @loadB(%B* %b.ptr) {
+; CHECK-LABEL: loadB
+; CHECK-NEXT: [[GEP1:%[a-z0-9\.]+]] = getelementptr inbounds %B, %B* %b.ptr, i64 0, i32 0
+; CHECK-NEXT: [[LOAD1:%[a-z0-9\.]+]] = load i8*, i8** [[GEP1]], align 8
+; CHECK-NEXT: [[IV1:%[a-z0-9\.]+]] = insertvalue %B undef, i8* [[LOAD1]], 0
+; CHECK-NEXT: [[GEP2:%[a-z0-9\.]+]] = getelementptr inbounds %B, %B* %b.ptr, i64 0, i32 1
+; CHECK-NEXT: [[LOAD2:%[a-z0-9\.]+]] = load i64, i64* [[GEP2]], align 8
+; CHECK-NEXT: [[IV2:%[a-z0-9\.]+]] = insertvalue %B [[IV1]], i64 [[LOAD2]], 1
+; CHECK-NEXT: ret %B [[IV2]]
+  %1 = load %B, %B* %b.ptr, align 8
+  ret %B %1
+}
+
+define { %A } @loadStructOfA({ %A }* %sa.ptr) {
 ; CHECK-LABEL: loadStructOfA
-; CHECK: load %A__vtbl*,
-; CHECK: insertvalue %A undef, %A__vtbl* {{.*}}, 0
-; CHECK: insertvalue { %A } undef, %A {{.*}}, 0
-  %2 = load { %A }, { %A }* %1, align 8
-  ret { %A } %2
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds { %A }, { %A }* %sa.ptr, i64 0, i32 0, i32 0
+; CHECK-NEXT: [[LOAD:%[a-z0-9\.]+]] = load %A__vtbl*, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: [[IV1:%[a-z0-9\.]+]] = insertvalue %A undef, %A__vtbl* [[LOAD]], 0
+; CHECK-NEXT: [[IV2:%[a-z0-9\.]+]] = insertvalue { %A } undef, %A [[IV1]], 0
+; CHECK-NEXT: ret { %A } [[IV2]]
+  %1 = load { %A }, { %A }* %sa.ptr, align 8
+  ret { %A } %1
 }
 
-define [1 x %A] @loadArrayOfA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to [1 x %A]*
+define [1 x %A] @loadArrayOfA([1 x %A]* %aa.ptr) {
 ; CHECK-LABEL: loadArrayOfA
-; CHECK: load %A__vtbl*,
-; CHECK: insertvalue %A undef, %A__vtbl* {{.*}}, 0
-; CHECK: insertvalue [1 x %A] undef, %A {{.*}}, 0
-  %2 = load [1 x %A], [1 x %A]* %1, align 8
-  ret [1 x %A] %2
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds [1 x %A], [1 x %A]* %aa.ptr, i64 0, i64 0, i32 0
+; CHECK-NEXT: [[LOAD:%[a-z0-9\.]+]] = load %A__vtbl*, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: [[IV1:%[a-z0-9\.]+]] = insertvalue %A undef, %A__vtbl* [[LOAD]], 0
+; CHECK-NEXT: [[IV2:%[a-z0-9\.]+]] = insertvalue [1 x %A] undef, %A [[IV1]], 0
+; CHECK-NEXT: ret [1 x %A] [[IV2]]
+  %1 = load [1 x %A], [1 x %A]* %aa.ptr, align 8
+  ret [1 x %A] %1
 }
 
-define { [1 x %A] } @loadStructOfArrayOfA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to { [1 x %A] }*
+define { [1 x %A] } @loadStructOfArrayOfA({ [1 x %A] }* %saa.ptr) {
 ; CHECK-LABEL: loadStructOfArrayOfA
-; CHECK: load %A__vtbl*,
-; CHECK: insertvalue %A undef, %A__vtbl* {{.*}}, 0
-; CHECK: insertvalue [1 x %A] undef, %A {{.*}}, 0
-; CHECK: insertvalue { [1 x %A] } undef, [1 x %A] {{.*}}, 0
-  %2 = load { [1 x %A] }, { [1 x %A] }* %1, align 8
-  ret { [1 x %A] } %2
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds { [1 x %A] }, { [1 x %A] }* %saa.ptr, i64 0, i32 0, i64 0, i32 0
+; CHECK-NEXT: [[LOAD:%[a-z0-9\.]+]] = load %A__vtbl*, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: [[IV1:%[a-z0-9\.]+]] = insertvalue %A undef, %A__vtbl* [[LOAD]], 0
+; CHECK-NEXT: [[IV2:%[a-z0-9\.]+]] = insertvalue [1 x %A] undef, %A [[IV1]], 0
+; CHECK-NEXT: [[IV3:%[a-z0-9\.]+]] = insertvalue { [1 x %A] } undef, [1 x %A] [[IV2]], 0
+; CHECK-NEXT: ret { [1 x %A] } [[IV3]]
+  %1 = load { [1 x %A] }, { [1 x %A] }* %saa.ptr, align 8
+  ret { [1 x %A] } %1
 }
 
-define { %A } @structOfA() {
-body:
-  %0 = tail call i8* @allocmemory(i64 32)
-  %1 = bitcast i8* %0 to { %A }*
+define { %A } @structOfA({ %A }* %sa.ptr) {
 ; CHECK-LABEL: structOfA
-; CHECK: store %A__vtbl* @A__vtblZ
-  store { %A } { %A { %A__vtbl* @A__vtblZ } }, { %A }* %1, align 8
-  %2 = load { %A }, { %A }* %1, align 8
-; CHECK-NOT: load
-; CHECK: ret { %A } { %A { %A__vtbl* @A__vtblZ } }
-  ret { %A } %2
+; CHECK-NEXT: [[GEP:%[a-z0-9\.]+]] = getelementptr inbounds { %A }, { %A }* %sa.ptr, i64 0, i32 0, i32 0
+; CHECK-NEXT: store %A__vtbl* @A__vtblZ, %A__vtbl** [[GEP]], align 8
+; CHECK-NEXT: ret { %A } { %A { %A__vtbl* @A__vtblZ } }
+  store { %A } { %A { %A__vtbl* @A__vtblZ } }, { %A }* %sa.ptr, align 8
+  %1 = load { %A }, { %A }* %sa.ptr, align 8
+  ret { %A } %1
+}
+
+define %B @structB(%B* %b.ptr) {
+; CHECK-LABEL: structB
+; CHECK-NEXT: [[GEP1:%[a-z0-9\.]+]] = getelementptr inbounds %B, %B* %b.ptr, i64 0, i32 0
+; CHECK-NEXT: store i8* null, i8** [[GEP1]], align 8
+; CHECK-NEXT: [[GEP2:%[a-z0-9\.]+]] = getelementptr inbounds %B, %B* %b.ptr, i64 0, i32 1
+; CHECK-NEXT: store i64 42, i64* [[GEP2]], align 8
+; CHECK-NEXT: ret %B { i8* null, i64 42 }
+  store %B { i8* null, i64 42 }, %B* %b.ptr, align 8
+  %1 = load %B, %B* %b.ptr, align 8
+  ret %B %1
 }