Merge pull request #12 from RadeonOpenCompute/addrspacecast-fixes

Addrspacecast fixes

Author: tstellarAMD

lib/Analysis/Loads.cpp

@@ -91,7 +91,10 @@ static bool isDereferenceableAndAlignedPointer(
     // then the GEP (== Base + Offset == k_0 * Align + k_1 * Align) is also
     // aligned to Align bytes.
 
-    return isDereferenceableAndAlignedPointer(Base, Align, Offset + Size, DL,
+    // Offset and Size may have different bit widths if we have visited an
+    // addrspacecast, so we can't do arithmetic directly on the APInt values.
+    return isDereferenceableAndAlignedPointer(Base, Align,
+                                              Offset + Size.getSExtValue(), DL,
                                               CtxI, DT, Visited);
   }
 

lib/Analysis/ValueTracking.cpp

@@ -2833,11 +2833,17 @@ Value *llvm::GetPointerBaseWithConstantOffset(Value *Ptr, int64_t &Offset,
       break;
 
     if (GEPOperator *GEP = dyn_cast<GEPOperator>(Ptr)) {
-      APInt GEPOffset(BitWidth, 0);
+      // If one of the values we have visited is an addrspacecast, then
+      // the pointer type of this GEP may be different from the type
+      // of the Ptr parameter which was passed to this function.  This
+      // means when we construct GEPOffset, we need to use the size
+      // of GEP's pointer type rather than the size of the original
+      // pointer type.
+      APInt GEPOffset(DL.getPointerTypeSizeInBits(Ptr->getType()), 0);
       if (!GEP->accumulateConstantOffset(DL, GEPOffset))
         break;
 
-      ByteOffset += GEPOffset;
+      ByteOffset += GEPOffset.getSExtValue();
 
       Ptr = GEP->getPointerOperand();
     } else if (Operator::getOpcode(Ptr) == Instruction::BitCast ||

test/Analysis/ValueTracking/dereferenceable-and-aligned.ll

@@ -0,0 +1,21 @@
+; RUN: opt < %s -licm -S | FileCheck %s
+
+target datalayout = "e-p:32:32-p1:64:64-p4:64:64"
+
+; Make sure isDereferenceableAndAlignePointer() doesn't crash when looking
+; walking pointer defs with an addrspacecast that changes pointer size.
+; CHECK-LABEL: @addrspacecast_crash
+define void @addrspacecast_crash() {
+bb:
+  %tmp = alloca [256 x i32]
+  br label %bb1
+
+bb1:
+  %tmp2 = getelementptr inbounds [256 x i32], [256 x i32]* %tmp, i32 0, i32 36
+  %tmp3 = bitcast i32* %tmp2 to <4 x i32>*
+  %tmp4 = addrspacecast <4 x i32>* %tmp3 to <4 x i32> addrspace(4)*
+  %tmp5 = load <4 x i32>, <4 x i32> addrspace(4)* %tmp4
+  %tmp6 = xor <4 x i32> %tmp5, undef
+  store <4 x i32> %tmp6, <4 x i32> addrspace(1)* undef
+  br label %bb1
+}

test/Analysis/ValueTracking/get-pointer-base-with-const-off.ll

@@ -0,0 +1,26 @@
+; RUN: opt -gvn -S < %s | FileCheck %s
+
+; Make sure we don't crash when analyzing an addrspacecast in
+; GetPointerBaseWithConstantOffset()
+
+target datalayout = "e-p:32:32-p4:64:64"
+
+define i32 @addrspacecast-crash() {
+; CHECK-LABEL: @addrspacecast-crash
+; CHECK: %tmp = alloca [25 x i64]
+; CHECK: %tmp1 = getelementptr inbounds [25 x i64], [25 x i64]* %tmp, i32 0, i32 0
+; CHECK: %tmp2 = addrspacecast i64* %tmp1 to <8 x i64> addrspace(4)*
+; CHECK: store <8 x i64> zeroinitializer, <8 x i64> addrspace(4)* %tmp2
+; CHECK-NOT: load
+bb:
+  %tmp = alloca [25 x i64]
+  %tmp1 = getelementptr inbounds [25 x i64], [25 x i64]* %tmp, i32 0, i32 0
+  %tmp2 = addrspacecast i64* %tmp1 to <8 x i64> addrspace(4)*
+  %tmp3 = getelementptr inbounds <8 x i64>, <8 x i64> addrspace(4)* %tmp2, i64 0
+  store <8 x i64> zeroinitializer, <8 x i64> addrspace(4)* %tmp3
+  %tmp4 = getelementptr inbounds [25 x i64], [25 x i64]* %tmp, i32 0, i32 0
+  %tmp5 = addrspacecast i64* %tmp4 to i32 addrspace(4)*
+  %tmp6 = getelementptr inbounds i32, i32 addrspace(4)* %tmp5, i64 10
+  %tmp7 = load i32, i32 addrspace(4)* %tmp6, align 4
+  ret i32 %tmp7
+}