[SROA] pr37267: fix assertion failure in integer widening

inouehrs · inouehrs · commit f5c0e6c285d2 · 2018-05-17T06:32:17.000Z
The current integer widening does not support rewriting partial split slices in rewriteIntegerStore (and rewriteIntegerLoad). This patch adds explicit checks for this case in isIntegerWideningViableForSlice. Before r322533, splitting is allowed only for the whole-alloca slice and hence the above case is implicitly rejected by another check `if (DL.getTypeStoreSize(ValueTy) > Size)` because whole-alloca slice is larger than the partition. Differential Revision: https://reviews.llvm.org/D46750 llvm-svn: 332575
diff --git a/llvm/lib/Transforms/Scalar/SROA.cpp b/llvm/lib/Transforms/Scalar/SROA.cpp
@@ -1968,6 +1968,10 @@ static bool isIntegerWideningViableForSlice(const Slice &S,
     // We can't handle loads that extend past the allocated memory.
     if (DL.getTypeStoreSize(LI->getType()) > Size)
       return false;
+    // So far, AllocaSliceRewriter does not support widening split slice tails
+    // in rewriteIntegerLoad.
+    if (S.beginOffset() < AllocBeginOffset)
+      return false;
     // Note that we don't count vector loads or stores as whole-alloca
     // operations which enable integer widening because we would prefer to use
     // vector widening instead.
@@ -1989,6 +1993,10 @@ static bool isIntegerWideningViableForSlice(const Slice &S,
     // We can't handle stores that extend past the allocated memory.
     if (DL.getTypeStoreSize(ValueTy) > Size)
       return false;
+    // So far, AllocaSliceRewriter does not support widening split slice tails
+    // in rewriteIntegerStore.
+    if (S.beginOffset() < AllocBeginOffset)
+      return false;
     // Note that we don't count vector loads or stores as whole-alloca
     // operations which enable integer widening because we would prefer to use
     // vector widening instead.
diff --git a/llvm/test/Transforms/SROA/pr37267.ll b/llvm/test/Transforms/SROA/pr37267.ll
@@ -0,0 +1,74 @@
+; RUN: opt < %s -sroa -S | FileCheck %s
+target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-n32:64-S128"
+target triple = "sparcv9-sun-solaris"
+
+; PR37267
+; Check that we don't crash on this test.
+
+define i16 @f1() {
+; CHECK-LABEL: @f1
+; CHECK: %[[retval:.*]] = add i16 2, 2
+; CHECK: ret i16 %[[retval]]
+
+bb1:
+; This 12-byte alloca is split into partitions as [0,2), [2,4), [4,8), [8,10), [10, 12).
+; The reported error happened when rewriteIntegerStore try to widen a split tail of slice 1 for [4, 8) partition.
+; alloca  012345678901
+; slice 1:  WWWW
+; slice 2:        WWWW
+; slice 3:        RR
+; slice 4:  RR
+
+  %a.3 = alloca [6 x i16], align 1
+; slice 1: [2,6)
+  %_tmp3 = getelementptr inbounds [6 x i16], [6 x i16]* %a.3, i16 0, i16 1
+  %_tmp5 = bitcast i16* %_tmp3 to i32*
+  store i32 131074, i32* %_tmp5, align 1
+; slice 2: [8,12)
+  %_tmp8 = getelementptr inbounds [6 x i16], [6 x i16]* %a.3, i16 0, i16 4
+  %_tmp10 = bitcast i16* %_tmp8 to i32*
+  store i32 131074, i32* %_tmp10, align 1
+; slice 3: [8,10)
+  %_tmp12 = getelementptr inbounds [6 x i16], [6 x i16]* %a.3, i16 0, i16 4
+  %_tmp13 = load i16, i16* %_tmp12, align 1
+; slice 4: [2,4)
+  %_tmp15 = getelementptr inbounds [6 x i16], [6 x i16]* %a.3, i16 0, i16 1
+  %_tmp16 = load i16, i16* %_tmp15, align 1
+
+  %rc = add i16 %_tmp13, %_tmp16
+  ret i16 %rc
+}
+
+define i16 @f2() {
+; CHECK-LABEL: @f2
+; CHECK: %[[retval:.*]] = add i16 2, undef
+; CHECK: ret i16 %[[retval]]
+
+bb1:
+; This 12-byte alloca is split into partitions as [0,2), [2,4), [4,8), [8,10), [10, 12).
+; The reported error happened when visitLoadInst rewrites a split tail of slice 1 for [4, 8) partition.
+; alloca  012345678901
+; slice 1:  RRRR
+; slice 2:        WWWW
+; slice 3:        RR
+; slice 4:  RR
+
+  %a.3 = alloca [6 x i16], align 1
+; slice 1: [2,6)
+  %_tmp3 = getelementptr inbounds [6 x i16], [6 x i16]* %a.3, i16 0, i16 1
+  %_tmp5 = bitcast i16* %_tmp3 to i32*
+  %_tmp6 = load i32, i32* %_tmp5, align 1
+; slice 2: [8,12)
+  %_tmp8 = getelementptr inbounds [6 x i16], [6 x i16]* %a.3, i16 0, i16 4
+  %_tmp10 = bitcast i16* %_tmp8 to i32*
+  store i32 131074, i32* %_tmp10, align 1
+; slice 3: [8,10)
+  %_tmp12 = getelementptr inbounds [6 x i16], [6 x i16]* %a.3, i16 0, i16 4
+  %_tmp13 = load i16, i16* %_tmp12, align 1
+; slice 4: [2,4)
+  %_tmp15 = getelementptr inbounds [6 x i16], [6 x i16]* %a.3, i16 0, i16 1
+  %_tmp16 = load i16, i16* %_tmp15, align 1
+
+  %rc = add i16 %_tmp13, %_tmp16
+  ret i16 %rc
+}
