[InstCombine] Fold IntToPtr and PtrToInt into preceding loads.

[philip.pfaffe]

Summary

Currently we only fold a BitCast into a Load when the BitCast is its only user. This patch enables doing the same for IntToPtr and PtrToInt in the case they are noops.

[philip.pfaffe]

This change also affects a clang test-case. I created another patch to fix this: {D9153: [clang] Update a test-case affected by IntToPtr/PtrToInt being folded into Loads}

[chandlerc]

Looks good.

[majnemer]

lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp

487–488

Your comment has mismatched parenthesis.

490–496

Please fix the formatting here

[philip.pfaffe]

Rebased and addressed majnemer's comments.

[majnemer]

LGTM

lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp

if (std::all_of(LI.user_begin(), LI.user_end(), [&LI](User *U) {
       }
       assert(LI.use_empty() && "Failed to remove all users of the load!");
       // Return the old load so the combiner can delete it safely.
       return &LI;
     }
   }
 
   // Fold away bit casts of the loaded value by loading the desired type.
+  // We can do this for BitCastInsts as well as casts from and to pointer types,
+  // as long as those are noops (i.e., the source or dest type have the same
+  // bitwidth as the target's pointers).
   if (LI.hasOneUse())
-    if (auto *BC = dyn_cast<BitCastInst>(LI.user_back())) {
-      LoadInst *NewLoad = combineLoadToNewType(IC, LI, BC->getDestTy());
-      BC->replaceAllUsesWith(NewLoad);
-      IC.EraseInstFromFunction(*BC);
+    if (auto* CI = dyn_cast<CastInst>(LI.user_back())) {
+      if (CI->isNoopCast(DL)) {
+        LoadInst *NewLoad = combineLoadToNewType(IC, LI, CI->getDestTy());
+        CI->replaceAllUsesWith(NewLoad);
+        IC.EraseInstFromFunction(*CI);
         return &LI;
       }
+    }
 
   // FIXME: We should also canonicalize loads of vectors when their elements are
   // cast to other types.
   return nullptr;
 }
 
 static Instruction *unpackLoadToAggregate(InstCombiner &IC, LoadInst &LI) {
   // FIXME: We could probably with some care handle both volatile and atomic

test/Transforms/InstCombine/load-bitcast32.ll

+; RUN: opt -instcombine -S < %s | FileCheck %s
+
+target datalayout = "p:32:32:32"
+
+
+define i64* @test1(i8* %x) {
+entry:
+; CHECK-LABEL: @test1(
+; CHECK: load i64, i64*
+; CHECK: ret
+  %a = bitcast i8* %x to i64*
+  %b = load i64, i64* %a
+  %c = inttoptr i64 %b to i64*
+
+  ret i64* %c
+}
+
+define i32* @test2(i8* %x) {
+entry:
+; CHECK-LABEL: @test2(
+; CHECK: load i32*, i32**
+; CHECK: ret
+  %a = bitcast i8* %x to i32*
+  %b = load i32, i32* %a
+  %c = inttoptr i32 %b to i32*
+
+  ret i32* %c
+}
+
+define i64* @test3(i8* %x) {
+entry:
+; CHECK-LABEL: @test3(
+; CHECK: load i64*, i64**
+; CHECK: ret
+  %a = bitcast i8* %x to i32*
+  %b = load i32, i32* %a
+  %c = inttoptr i32 %b to i64*
+
+  ret i64* %c
+}
+
+define i64 @test4(i8* %x) {
+entry:
+; CHECK-LABEL: @test4(
+; CHECK: load i32, i32*
+; CHECK: zext
+; CHECK: ret
+  %a = bitcast i8* %x to i64**
+  %b = load i64*, i64** %a
+  %c = ptrtoint i64* %b to i64
+
+  ret i64 %c
+}
+
+define i32 @test5(i8* %x) {
+entry:
+; CHECK-LABEL: @test5(
+; CHECK: load i32, i32*
+; CHECK: ret
+  %a = bitcast i8* %x to i32**
+  %b = load i32*, i32** %a
+  %c = ptrtoint i32* %b to i32
+
+  ret i32 %c
+}
+
+define i64 @test6(i8* %x) {
+entry:
+; CHECK-LABEL: @test6(
+; CHECK: load i32, i32*
+; CHECK: zext
+; CHECK: ret
+  %a = bitcast i8* %x to i32**
+  %b = load i32*, i32** %a
+  %c = ptrtoint i32* %b to i64
+
+  ret i64 %c
+}
+

test/Transforms/InstCombine/load-bitcast64.ll

+; RUN: opt -instcombine -S < %s | FileCheck %s
+
+target datalayout = "p:64:64:64"
+
+
+define i64* @test1(i8* %x) {
+entry:
+; CHECK-LABEL: @test1(
+; CHECK: load i64*, i64**
+; CHECK: ret
+  %a = bitcast i8* %x to i64*
+  %b = load i64, i64* %a
+  %c = inttoptr i64 %b to i64*
+
+  ret i64* %c
+}
+
+define i32* @test2(i8* %x) {
+entry:
+; CHECK-LABEL: @test2(
+; CHECK: load i32, i32*
+; CHECK: ret
+  %a = bitcast i8* %x to i32*
+  %b = load i32, i32* %a
+  %c = inttoptr i32 %b to i32*
+
+  ret i32* %c
+}
+
+define i64* @test3(i8* %x) {
+entry:
+; CHECK-LABEL: @test3(
+; CHECK: load i32, i32*
+; CHECK: ret
+  %a = bitcast i8* %x to i32*
+  %b = load i32, i32* %a
+  %c = inttoptr i32 %b to i64*
+
+  ret i64* %c
+}
+
+define i64 @test4(i8* %x) {
+entry:
+; CHECK-LABEL: @test4(
+; CHECK: load i64, i64*
+; CHECK: ret
+  %a = bitcast i8* %x to i64**
+  %b = load i64*, i64** %a
+  %c = ptrtoint i64* %b to i64
+
+  ret i64 %c
+}
+
+define i32 @test5(i8* %x) {
+entry:
+; CHECK-LABEL: @test5(
+; CHECK: load i64, i64*
+; CHECK: trunc
+; CHECK: ret
+  %a = bitcast i8* %x to i32**
+  %b = load i32*, i32** %a
+  %c = ptrtoint i32* %b to i32
+
+  ret i32 %c
+}
+
+define i64 @test6(i8* %x) {
+entry:
+; CHECK-LABEL: @test6(
+; CHECK: load i64, i64*
+; CHECK: ret
+  %a = bitcast i8* %x to i32**
+  %b = load i32*, i32** %a
+  %c = ptrtoint i32* %b to i64
+
+  ret i64 %c
+}
+