intel · Fznamznon · Oct 1, 2021
@@ -1670,6 +1670,29 @@ static bool replacingOperandWithVariableIsCheap(const Instruction *I,
   return !isa<IntrinsicInst>(I);
 }
 
+static bool moduleHasSpirTarget(const Module *M) {
+  if (StringRef(M->getTargetTriple()).startswith("spir"))
+    return true;
+  return false;
+}
+
+static bool isPointerToOCLOrSPIRVImageOrSampler(const llvm::Type *Ty) {
+  if (auto PT = dyn_cast<PointerType>(Ty)) {
+    if (auto ST = dyn_cast<StructType>(PT->getElementType())) {
+      if (ST->isOpaque()) {
+        StringRef TypeName = ST->getName();
+        if (TypeName.startswith("opencl.sampler_t") ||
+            TypeName.startswith("opencl.image") ||
+            TypeName.startswith("spirv.SampledImage") ||
+            TypeName.startswith("spirv.Image") ||
+            TypeName.startswith("spirv.Sampler"))
+          return true;
+      }
+    }
+  }
+  return false;
+}
+
 // All instructions in Insts belong to different blocks that all unconditionally
 // branch to a common successor. Analyze each instruction and return true if it
 // would be possible to sink them into their successor, creating one common
@@ -1688,6 +1711,15 @@ static bool canSinkInstructions(
         I->getType()->isTokenTy())
       return false;
 
+    // In SPIR-V neither image nor sampler types cannot go through phi and
+    // select instructions. Lowering of phi nodes and select instructions with
+    // image/sampler result type by SPIR-V translator tool is very complicated
+    // in common case, so the transformation is disabled for image/sampler types
+    // here.
+    if (moduleHasSpirTarget(I->getModule()) &&
+        isPointerToOCLOrSPIRVImageOrSampler(I->getType()))
+      return false;
+
     // Do not try to sink an instruction in an infinite loop - it can cause
     // this algorithm to infinite loop.
     if (I->getParent()->getSingleSuccessor() == I->getParent())
@@ -1784,6 +1816,15 @@ static bool canSinkInstructions(
       // Don't touch any operand of token type.
       return false;
 
+    // In SPIR-V neither image nor sampler types cannot go through phi and
+    // select instructions. Lowering of phi nodes and select instructions with
+    // image/sampler result type by SPIR-V translator tool is very complicated
+    // in common case, so the transformation is disabled for image/sampler types
+    // here.
+    if (moduleHasSpirTarget(I0->getModule()) &&
+        isPointerToOCLOrSPIRVImageOrSampler(Op->getType()))
+      return false;
+
     auto SameAsI0 = [&I0, OI](const Instruction *I) {
       assert(I->getNumOperands() == I0->getNumOperands());
       return I->getOperand(OI) == I0->getOperand(OI);

@@ -0,0 +1,64 @@
+; RUN: opt < %s -simplifycfg -sink-common-insts -S | FileCheck %s
+target datalayout = "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-n8:16:32:64"
+target triple = "spir-unknown-unknown"
+
+%opencl.image3d_ro_t = type opaque
+%opencl.sampler_t = type opaque
+
+; CHECK-LABEL: @K(
+; CHECK-NOT: select i1 {{%.+}}, %opencl.sampler_t
+; CHECK-NOT: select i1 {{%.+}}, %opencl.image
+
+; Function Attrs: convergent mustprogress nofree norecurse nounwind willreturn
+define dso_local spir_kernel void @K(i32 addrspace(1)* nocapture readonly %A, %opencl.sampler_t addrspace(2)* %S, %opencl.image3d_ro_t addrspace(1)* %image, %opencl.image3d_ro_t addrspace(1)* nocapture readnone %image1, %opencl.sampler_t addrspace(2)* %S1, <4 x float> addrspace(1)* nocapture %tap) local_unnamed_addr #0 !kernel_arg_addr_space !3 !kernel_arg_access_qual !4 !kernel_arg_type !5 !kernel_arg_base_type !6 !kernel_arg_type_qual !7 {
+entry:
+  %call = call spir_func i32 @_Z13get_global_idj(i32 0) #3
+  %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %A, i32 %call
+  %0 = load i32, i32 addrspace(1)* %arrayidx, align 4
+  %conv = sitofp i32 %0 to float
+  %splat.splatinsert = insertelement <4 x float> poison, float %conv, i32 0
+  %splat.splat = shufflevector <4 x float> %splat.splatinsert, <4 x float> poison, <4 x i32> zeroinitializer
+  %1 = and i32 %0, 1
+  %tobool.not = icmp eq i32 %1, 0
+  br i1 %tobool.not, label %cond.false, label %cond.true
+
+cond.true:                                        ; preds = %entry
+  %call2 = call spir_func <4 x float> @_Z11read_imagef14ocl_image3d_ro11ocl_samplerDv4_f(%opencl.image3d_ro_t addrspace(1)* %image, %opencl.sampler_t addrspace(2)* %S, <4 x float> %splat.splat) #4
+  br label %cond.end
+
+cond.false:                                       ; preds = %entry
+  %call3 = call spir_func <4 x float> @_Z11read_imagef14ocl_image3d_ro11ocl_samplerDv4_f(%opencl.image3d_ro_t addrspace(1)* %image, %opencl.sampler_t addrspace(2)* %S1, <4 x float> %splat.splat) #4
+  br label %cond.end
+
+cond.end:                                         ; preds = %cond.false, %cond.true
+  %cond = phi <4 x float> [ %call2, %cond.true ], [ %call3, %cond.false ]
+  %arrayidx4 = getelementptr inbounds <4 x float>, <4 x float> addrspace(1)* %tap, i32 %call
+  store <4 x float> %cond, <4 x float> addrspace(1)* %arrayidx4, align 16
+  ret void
+}
+
+; Function Attrs: convergent nounwind readnone willreturn
+declare spir_func i32 @_Z13get_global_idj(i32) #1
+
+; Function Attrs: convergent nounwind readonly willreturn
+declare spir_func <4 x float> @_Z11read_imagef14ocl_image3d_ro11ocl_samplerDv4_f(%opencl.image3d_ro_t addrspace(1)*, %opencl.sampler_t addrspace(2)*, <4 x float>) #2
+
+attributes #0 = { convergent norecurse nounwind "frame-pointer"="none" "min-legal-vector-width"="128" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "uniform-work-group-size"="false" }
+attributes #1 = { convergent nounwind readnone willreturn "frame-pointer"="none" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
+attributes #2 = { convergent nounwind readonly willreturn "frame-pointer"="none" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
+attributes #3 = { convergent nounwind readnone willreturn }
+attributes #4 = { convergent nounwind readonly willreturn }
+
+!llvm.module.flags = !{!0}
+!opencl.ocl.version = !{!1}
+!opencl.spir.version = !{!1}
+!llvm.ident = !{!2}
+
+!0 = !{i32 1, !"wchar_size", i32 4}
+!1 = !{i32 2, i32 0}
+!2 = !{!"clang version 14.0.0 (https://github.com/intel/llvm.git 183caf354e603f8935163874374df716a25cf33d)"}
+!3 = !{i32 1, i32 0, i32 1, i32 1, i32 0, i32 1}
+!4 = !{!"none", !"none", !"read_only", !"read_only", !"none", !"none"}
+!5 = !{!"int*", !"sampler_t", !"image3d_t", !"image3d_t", !"sampler_t", !"float4*"}
+!6 = !{!"int*", !"sampler_t", !"image3d_t", !"image3d_t", !"sampler_t", !"float __attribute__((ext_vector_type(4)))*"}
+!7 = !{!"", !"", !"", !"", !"", !""}