Implement strip.invariant.group

Summary:
This patch introduce new intrinsic -
strip.invariant.group that was described in the
RFC: Devirtualization v2

Reviewers: rsmith, hfinkel, nlopes, sanjoy, amharc, kuhar

Subscribers: arsenm, nhaehnle, JDevlieghere, hiraditya, xbolva00, llvm-commits

Differential Revision: https://reviews.llvm.org/D47103

Co-authored-by: Krzysztof Pszeniczny <krzysztof.pszeniczny@gmail.com>
llvm-svn: 336073

7 files changed, 145 insertions, 18 deletions

diff --git a/llvm/test/Transforms/CodeGenPrepare/invariant.group.ll b/llvm/test/Transforms/CodeGenPrepare/invariant.group.ll
index b506378615a..29ff724f036 100644
--- a/llvm/test/Transforms/CodeGenPrepare/invariant.group.ll
+++ b/llvm/test/Transforms/CodeGenPrepare/invariant.group.ll
@@ -7,8 +7,8 @@ define void @foo() {
 enter:
   ; CHECK-NOT: !invariant.group
   ; CHECK-NOT: @llvm.launder.invariant.group.p0i8(
-  ; CHECK: %val = load i8, i8* @tmp, !tbaa
-  %val = load i8, i8* @tmp, !invariant.group !0, !tbaa !{!1, !1, i64 0}
+  ; CHECK: %val = load i8, i8* @tmp{{$}}
+  %val = load i8, i8* @tmp, !invariant.group !0
   %ptr = call i8* @llvm.launder.invariant.group.p0i8(i8* @tmp)
   
   ; CHECK: store i8 42, i8* @tmp{{$}}
@@ -18,7 +18,23 @@ enter:
 }
 ; CHECK-LABEL: }
 
-declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+; CHECK-LABEL: define void @foo2() {
+define void @foo2() {
+enter:
+  ; CHECK-NOT: !invariant.group
+  ; CHECK-NOT: @llvm.strip.invariant.group.p0i8(
+  ; CHECK: %val = load i8, i8* @tmp{{$}}
+  %val = load i8, i8* @tmp, !invariant.group !0
+  %ptr = call i8* @llvm.strip.invariant.group.p0i8(i8* @tmp)
+
+  ; CHECK: store i8 42, i8* @tmp{{$}}
+  store i8 42, i8* %ptr, !invariant.group !0
 
+  ret void
+}
+; CHECK-LABEL: }
+
+
+declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
 !0 = !{}
-!1 = !{!"x", !0}
diff --git a/llvm/test/Transforms/DeadStoreElimination/launder.invariant.group.ll b/llvm/test/Transforms/DeadStoreElimination/launder.invariant.group.ll
index dfa316e4dd6..725ab6e5e34 100644
--- a/llvm/test/Transforms/DeadStoreElimination/launder.invariant.group.ll
+++ b/llvm/test/Transforms/DeadStoreElimination/launder.invariant.group.ll
@@ -27,4 +27,39 @@ define void @skip2Barriers(i8* %ptr) {
   ret void
 }
 
+; CHECK-LABEL: void @skip3Barriers(i8* %ptr)
+define void @skip3Barriers(i8* %ptr) {
+; CHECK-NOT: store i8 42
+  store i8 42, i8* %ptr
+; CHECK: %ptr2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr)
+  %ptr2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr)
+; CHECK-NOT: store i8 43
+  store i8 43, i8* %ptr2
+  %ptr3 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr2)
+  %ptr4 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr3)
+
+; CHECK: store i8 44
+  store i8 44, i8* %ptr4
+  ret void
+}
+
+; CHECK-LABEL: void @skip4Barriers(i8* %ptr)
+define void @skip4Barriers(i8* %ptr) {
+; CHECK-NOT: store i8 42
+  store i8 42, i8* %ptr
+; CHECK: %ptr2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr)
+  %ptr2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr)
+; CHECK-NOT: store i8 43
+  store i8 43, i8* %ptr2
+  %ptr3 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr2)
+  %ptr4 = call i8* @llvm.strip.invariant.group.p0i8(i8* %ptr3)
+  %ptr5 = call i8* @llvm.launder.invariant.group.p0i8(i8* %ptr3)
+
+; CHECK: store i8 44
+  store i8 44, i8* %ptr5
+  ret void
+}
+
+
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
\ No newline at end of file
diff --git a/llvm/test/Transforms/FunctionAttrs/nocapture.ll b/llvm/test/Transforms/FunctionAttrs/nocapture.ll
index dedec09c53e..0c3fc0af725 100644
--- a/llvm/test/Transforms/FunctionAttrs/nocapture.ll
+++ b/llvm/test/Transforms/FunctionAttrs/nocapture.ll
@@ -237,4 +237,21 @@ define void @captureLaunder(i8* %p) {
   ret void
 }
 
+; CHECK: @nocaptureStrip(i8* nocapture %p)
+define void @nocaptureStrip(i8* %p) {
+entry:
+  %b = call i8* @llvm.strip.invariant.group.p0i8(i8* %p)
+  store i8 42, i8* %b
+  ret void
+}
+
+@g3 = global i8* null
+; CHECK: define void @captureStrip(i8* %p)
+define void @captureStrip(i8* %p) {
+  %b = call i8* @llvm.strip.invariant.group.p0i8(i8* %p)
+  store i8* %b, i8** @g3
+  ret void
+}
+
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
diff --git a/llvm/test/Transforms/GVN/invariant.group.ll b/llvm/test/Transforms/GVN/invariant.group.ll
index 3875fc2886e..6cdd44d16c0 100644
--- a/llvm/test/Transforms/GVN/invariant.group.ll
+++ b/llvm/test/Transforms/GVN/invariant.group.ll
@@ -51,6 +51,18 @@ entry:
     ret i8 %b
 }
 
+; CHECK-LABEL: define i1 @proveEqualityForStrip(
+define i1 @proveEqualityForStrip(i8* %a) {
+; FIXME: The first call could be also removed by GVN. Right now
+; DCE removes it. The second call is CSE'd with the first one.
+; CHECK: %b1 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %b1 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+; CHECK-NOT: llvm.strip.invariant.group
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %r = icmp eq i8* %b1, %b2
+; CHECK: ret i1 true
+  ret i1 %r
+}
 ; CHECK-LABEL: define i8 @unoptimizable1() {
 define i8 @unoptimizable1() {
 entry:
@@ -437,10 +449,10 @@ declare void @_ZN1AC1Ev(%struct.A*)
 declare void @fooBit(i1*, i1)
 
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
+
 
-; Function Attrs: nounwind
-declare void @llvm.assume(i1 %cmp.vtables) #0
+declare void @llvm.assume(i1 %cmp.vtables)
 
 
-attributes #0 = { nounwind }
 !0 = !{}
\ No newline at end of file
diff --git a/llvm/test/Transforms/GlobalOpt/invariant.group.barrier.ll b/llvm/test/Transforms/GlobalOpt/invariant.group.ll
index 9ae7e78b785..8a090cbd0d3 100644
--- a/llvm/test/Transforms/GlobalOpt/invariant.group.barrier.ll
+++ b/llvm/test/Transforms/GlobalOpt/invariant.group.ll
@@ -27,15 +27,15 @@ enter:
 define void @_optimizable() {
 enter:
   %valptr = alloca i32
-  
+
   %val = call i32 @TheAnswerToLifeTheUniverseAndEverything()
   store i32 %val, i32* @tmp
   store i32 %val, i32* %valptr
-  
+
   %0 = bitcast i32* %valptr to i8*
   %barr = call i8* @llvm.launder.invariant.group(i8* %0)
   %1 = bitcast i8* %barr to i32*
-  
+
   %val2 = load i32, i32* %1
   store i32 %val2, i32* @tmp2
   ret void
@@ -43,30 +43,30 @@ enter:
 
 ; We can't step through launder.invariant.group here, because that would change
 ; this load in @usage_of_globals()
-; val = load i32, i32* %ptrVal, !invariant.group !0 
-; into 
+; val = load i32, i32* %ptrVal, !invariant.group !0
+; into
 ; %val = load i32, i32* @tmp3, !invariant.group !0
-; and then we could assume that %val and %val2 to be the same, which coud be 
+; and then we could assume that %val and %val2 to be the same, which coud be
 ; false, because @changeTmp3ValAndCallBarrierInside() may change the value
 ; of @tmp3.
 define void @_not_optimizable() {
 enter:
   store i32 13, i32* @tmp3, !invariant.group !0
-  
+
   %0 = bitcast i32* @tmp3 to i8*
   %barr = call i8* @llvm.launder.invariant.group(i8* %0)
   %1 = bitcast i8* %barr to i32*
-  
+
   store i32* %1, i32** @ptrToTmp3
   store i32 42, i32* %1, !invariant.group !0
-  
+
   ret void
 }
 define void @usage_of_globals() {
 entry:
   %ptrVal = load i32*, i32** @ptrToTmp3
   %val = load i32, i32* %ptrVal, !invariant.group !0
-  
+
   call void @changeTmp3ValAndCallBarrierInside()
   %val2 = load i32, i32* @tmp3, !invariant.group !0
   ret void;
diff --git a/llvm/test/Transforms/InstCombine/invariant.group.ll b/llvm/test/Transforms/InstCombine/invariant.group.ll
index 697ed5995c7..ada45e3417c 100644
--- a/llvm/test/Transforms/InstCombine/invariant.group.ll
+++ b/llvm/test/Transforms/InstCombine/invariant.group.ll
@@ -1,5 +1,6 @@
 ; RUN: opt -instcombine -S < %s | FileCheck %s
 
+
 ; CHECK-LABEL: define i8* @simplifyNullLaunder()
 define i8* @simplifyNullLaunder() {
 ; CHECK-NEXT: ret i8* null
@@ -29,6 +30,39 @@ define i8 addrspace(42)* @simplifyUndefLaunder2() {
   ret i8 addrspace(42)* %b2
 }
 
-
 declare i8* @llvm.launder.invariant.group.p0i8(i8*)
 declare i8 addrspace(42)* @llvm.launder.invariant.group.p42i8(i8 addrspace(42)*)
+
+
+; CHECK-LABEL: define i8* @simplifyNullStrip()
+define i8* @simplifyNullStrip() {
+; CHECK-NEXT: ret i8* null
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* null)
+  ret i8* %b2
+}
+
+; CHECK-LABEL: define i8 addrspace(42)* @dontsimplifyNullStripForDifferentAddrspace()
+define i8 addrspace(42)* @dontsimplifyNullStripForDifferentAddrspace() {
+; CHECK: %b2 = call i8 addrspace(42)* @llvm.strip.invariant.group.p42i8(i8 addrspace(42)* null)
+; CHECK: ret i8 addrspace(42)* %b2
+  %b2 = call i8 addrspace(42)* @llvm.strip.invariant.group.p42i8(i8 addrspace(42)* null)
+  ret i8 addrspace(42)* %b2
+}
+
+; CHECK-LABEL: define i8* @simplifyUndefStrip()
+define i8* @simplifyUndefStrip() {
+; CHECK-NEXT: ret i8* undef
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* undef)
+  ret i8* %b2
+}
+
+; CHECK-LABEL: define i8 addrspace(42)* @simplifyUndefStrip2()
+define i8 addrspace(42)* @simplifyUndefStrip2() {
+; CHECK-NEXT: ret i8 addrspace(42)* undef
+  %b2 = call i8 addrspace(42)* @llvm.strip.invariant.group.p42i8(i8 addrspace(42)* undef)
+  ret i8 addrspace(42)* %b2
+}
+
+declare i8* @llvm.strip.invariant.group.p0i8(i8*)
+declare i8 addrspace(42)* @llvm.strip.invariant.group.p42i8(i8 addrspace(42)*)
+
diff --git a/llvm/test/Transforms/NewGVN/invariant.group.ll b/llvm/test/Transforms/NewGVN/invariant.group.ll
index 5e09c962a0a..78e61a2f2b6 100644
--- a/llvm/test/Transforms/NewGVN/invariant.group.ll
+++ b/llvm/test/Transforms/NewGVN/invariant.group.ll
@@ -52,6 +52,19 @@ entry:
     ret i8 %b
 }
 
+; CHECK-LABEL: define i1 @proveEqualityForStrip(
+define i1 @proveEqualityForStrip(i8* %a) {
+; FIXME: The first call could be also removed by GVN. Right now
+; DCE removes it. The second call is CSE'd with the first one.
+; CHECK: %b1 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %b1 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+; CHECK-NOT: llvm.strip.invariant.group
+  %b2 = call i8* @llvm.strip.invariant.group.p0i8(i8* %a)
+  %r = icmp eq i8* %b1, %b2
+; CHECK: ret i1 true
+  ret i1 %r
+}
+
 ; CHECK-LABEL: define i8 @unoptimizable1() {
 define i8 @unoptimizable1() {
 entry: