[llvm] r280807 - [InstCombine][SSE4a] Fix assertion failure in the insertq/insertqi combining logic.

[Andrea Di Biagio via llvm-commits]

Author: adibiagio
Date: Wed Sep  7 07:47:53 2016
New Revision: 280807

URL: http://llvm.org/viewvc/llvm-project?rev=280807&view=rev
Log:
[InstCombine][SSE4a] Fix assertion failure in the insertq/insertqi combining logic.

This fixes a similar issue to the one already fixed by r280804
(revieved in D24256). Revision 280804 fixed the problem with unsafe dyn_casts
in the extrq/extrqi combining logic. However, it turns out that even the
insertq/insertqi logic was affected by the same problem.

Modified:
    llvm/trunk/lib/Transforms/InstCombine/InstCombineCalls.cpp
    llvm/trunk/test/Transforms/InstCombine/x86-sse4a.ll

Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineCalls.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineCalls.cpp?rev=280807&r1=280806&r2=280807&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/InstCombine/InstCombineCalls.cpp (original)
+++ llvm/trunk/lib/Transforms/InstCombine/InstCombineCalls.cpp Wed Sep  7 07:47:53 2016
@@ -722,10 +722,10 @@ static Value *simplifyX86insertq(Intrins
   Constant *C0 = dyn_cast<Constant>(Op0);
   Constant *C1 = dyn_cast<Constant>(Op1);
   ConstantInt *CI00 =
-      C0 ? dyn_cast<ConstantInt>(C0->getAggregateElement((unsigned)0))
+      C0 ? dyn_cast_or_null<ConstantInt>(C0->getAggregateElement((unsigned)0))
          : nullptr;
   ConstantInt *CI10 =
-      C1 ? dyn_cast<ConstantInt>(C1->getAggregateElement((unsigned)0))
+      C1 ? dyn_cast_or_null<ConstantInt>(C1->getAggregateElement((unsigned)0))
          : nullptr;
 
   // Constant Fold - insert bottom Length bits starting at the Index'th bit.
@@ -1919,7 +1919,7 @@ Instruction *InstCombiner::visitCallInst
     // See if we're dealing with constant values.
     Constant *C1 = dyn_cast<Constant>(Op1);
     ConstantInt *CI11 =
-        C1 ? dyn_cast<ConstantInt>(C1->getAggregateElement((unsigned)1))
+        C1 ? dyn_cast_or_null<ConstantInt>(C1->getAggregateElement((unsigned)1))
            : nullptr;
 
     // Attempt to simplify to a constant, shuffle vector or INSERTQI call.

Modified: llvm/trunk/test/Transforms/InstCombine/x86-sse4a.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstCombine/x86-sse4a.ll?rev=280807&r1=280806&r2=280807&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/InstCombine/x86-sse4a.ll (original)
+++ llvm/trunk/test/Transforms/InstCombine/x86-sse4a.ll Wed Sep  7 07:47:53 2016
@@ -177,6 +177,15 @@ define <2 x i64> @test_insertq_constant_
   ret <2 x i64> %1
 }
 
+define <2 x i64> @test_insertq_call_constexpr(<2 x i64> %x) {
+; CHECK-LABEL: @test_insertq_call_constexpr(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <2 x i64> @llvm.x86.sse4a.insertq(<2 x i64> %x, <2 x i64> bitcast (<16 x i8> trunc (<16 x i16> bitcast (<4 x i64> <i64 0, i64 undef, i64 2, i64 undef> to <16 x i16>) to <16 x i8>) to <2 x i64>))
+; CHECK-NEXT:    ret <2 x i64> [[TMP1]]
+;
+  %1 = tail call <2 x i64> @llvm.x86.sse4a.insertq(<2 x i64> %x, <2 x i64> bitcast (<16 x i8> trunc (<16 x i16> bitcast (<4 x i64> <i64 0, i64 undef, i64 2, i64 undef> to <16 x i16>) to <16 x i8>) to <2 x i64>))
+  ret <2 x i64> %1
+}
+
 ;
 ; INSERTQI
 ;
@@ -213,6 +222,15 @@ define <2 x i64> @test_insertqi_constant
   ret <2 x i64> %1
 }
 
+define <2 x i64> @test_insertqi_call_constexpr(<2 x i64> %x) {
+; CHECK-LABEL: @test_insertqi_call_constexpr(
+; CHECK-NEXT:    [[TMP1:%.*]] = tail call <2 x i64> @llvm.x86.sse4a.insertqi(<2 x i64> %x, <2 x i64> bitcast (<16 x i8> trunc (<16 x i16> bitcast (<4 x i64> <i64 0, i64 undef, i64 2, i64 undef> to <16 x i16>) to <16 x i8>) to <2 x i64>), i8 48, i8 3)
+; CHECK-NEXT:    ret <2 x i64> [[TMP1]]
+;
+  %1 = tail call <2 x i64> @llvm.x86.sse4a.insertqi(<2 x i64> %x, <2 x i64> bitcast (<16 x i8> trunc (<16 x i16> bitcast (<4 x i64> <i64 0, i64 undef, i64 2, i64 undef> to <16 x i16>) to <16 x i8>) to <2 x i64>), i8 48, i8 3)
+  ret <2 x i64> %1
+}
+
 ; The result of this insert is the second arg, since the top 64 bits of
 ; the result are undefined, and we copy the bottom 64 bits from the
 ; second arg