Use getTargetConstant for BLENDI, and add a test to catch it.

[saugustine]

Summary

This fixes a crasher introduced by r372338.

[echristo]

The x86isellowering change is great, see if you can't remove some of the branches before committing?

[MaskRay]

llvm/test/CodeGen/AMDGPU/GlobalISel/isel-blendi-gettargetconstant.ll

2

This can be removed.

67

llvm.ident can be removed.

[saugustine]

• Tighten up the test case.

[saugustine]

In D67785#1675848, @echristo wrote:

The x86isellowering change is great, see if you can't remove some of the branches before committing?

Done.

[echristo]

Sweet.

[saugustine]

Committed as r372366.

[MaskRay]

In the description of the git commit:

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67785

Tighten up the test case.     <---- stray comment

(I usually also delete the Subscribers: and Tags: #llvm lines before committing. I feel they just occupy space.)

[RKSimon]

llvm/trunk/test/CodeGen/AMDGPU/GlobalISel/isel-blendi-gettargetconstant.ll

3 ↗

(On Diff #220935)

The test file needs a lot of improvement - its missing a RUN line so isn't actually tested, most of the attributes/metadata can probably be ditched, and surely it should be in test/CodeGen/X86/GlobalIsel ?

[thakis]

Reverted in r372383 for now, to unbreak the bots.

[arsenm]

llvm/trunk/test/CodeGen/AMDGPU/GlobalISel/isel-blendi-gettargetconstant.ll

3 ↗

(On Diff #220935)

This belongs in test/CodeGen/X86 (no GlobalISel)

14 ↗

(On Diff #220935)

Probably don't need most of these

[saugustine]

Trying again in https://reviews.llvm.org/D67850

llvm/lib/Target/X86/X86ISelLowering.cpp

if (EltVT == MVT::f32) {
         // a vector, we prefer to generate a blend with immediate rather
         // than an insertps. Blends are simpler operations in hardware and so
         // will always have equal or better performance than insertps.
         // But if optimizing for size and there's a load folding opportunity,
         // generate insertps because blendps does not have a 32-bit memory
         // operand form.
         N1 = DAG.getNode(ISD::SCALAR_TO_VECTOR, dl, MVT::v4f32, N1);
         return DAG.getNode(X86ISD::BLENDI, dl, VT, N0, N1,
-                           DAG.getConstant(1, dl, MVT::i8));
+                           DAG.getTargetConstant(1, dl, MVT::i8));
       }
       // Create this as a scalar to vector..
       N1 = DAG.getNode(ISD::SCALAR_TO_VECTOR, dl, MVT::v4f32, N1);
       return DAG.getNode(X86ISD::INSERTPS, dl, VT, N0, N1,
                          DAG.getTargetConstant(IdxVal << 4, dl, MVT::i8));
     }
 
     // PINSR* works with constant index.

llvm/test/CodeGen/AMDGPU/GlobalISel/isel-blendi-gettargetconstant.ll

+; ModuleID = 'bugpoint-reduced-simplified.bc'
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-grtev4-linux-gnu"
+
+define dso_local void @csrot_(float* %0) local_unnamed_addr #0 align 32 {
+1:
+  %2 = load float, float* %0, align 4
+  %3 = fsub float -0.000000e+00, %2
+  %4 = insertelement <2 x float> <float undef, float -0.000000e+00>, float %3, i32 0
+  store <2 x float> %4, <2 x float>* undef, align 8
+  ret void
+}
+
+attributes #0 = { "target-features"="+aes,+cx8,+fxsr,+mmx,+pclmul,+popcnt,+sse,+sse2,+sse3,+sse4.1,+sse4.2,+ssse3,+x87" }