Implement the OpMatrixTimesScalar linear algebra case (#3450)

This PR implements the OpMatrixTimesScalar case for the
replace linear algebra instruction transformation.
diff --git a/source/fuzz/fuzzer_pass_replace_linear_algebra_instructions.cpp b/source/fuzz/fuzzer_pass_replace_linear_algebra_instructions.cpp
index 1c7b828..7f9b848 100644
--- a/source/fuzz/fuzzer_pass_replace_linear_algebra_instructions.cpp
+++ b/source/fuzz/fuzzer_pass_replace_linear_algebra_instructions.cpp
@@ -42,6 +42,7 @@
     // addressed the following conditional can use the function
     // |spvOpcodeIsLinearAlgebra|.
     if (instruction->opcode() != SpvOpVectorTimesScalar &&
+        instruction->opcode() != SpvOpMatrixTimesScalar &&
         instruction->opcode() != SpvOpDot) {
       return;
     }
diff --git a/source/fuzz/protobufs/spvtoolsfuzz.proto b/source/fuzz/protobufs/spvtoolsfuzz.proto
index b8d07cf..7b8adc0 100644
--- a/source/fuzz/protobufs/spvtoolsfuzz.proto
+++ b/source/fuzz/protobufs/spvtoolsfuzz.proto
@@ -1134,13 +1134,13 @@
   // This transformation is only applicable if the described instruction has one of the following opcodes.
   // Supported:
   //   OpVectorTimesScalar
+  //   OpMatrixTimesScalar
   //   OpDot
   // TODO(https://github.com/KhronosGroup/SPIRV-Tools/issues/3354):
   // Right now we only support certain operations. When this issue is addressed
   // the supporting comments can be removed.
   // To be supported in the future:
   //   OpTranspose
-  //   OpMatrixTimesScalar
   //   OpVectorTimesMatrix
   //   OpMatrixTimesVector
   //   OpMatrixTimesMatrix
diff --git a/source/fuzz/transformation_replace_linear_algebra_instruction.cpp b/source/fuzz/transformation_replace_linear_algebra_instruction.cpp
index 14ed502..1c7d0c9 100644
--- a/source/fuzz/transformation_replace_linear_algebra_instruction.cpp
+++ b/source/fuzz/transformation_replace_linear_algebra_instruction.cpp
@@ -46,6 +46,7 @@
   // the following conditional can use the function |spvOpcodeIsLinearAlgebra|.
   // It must be a supported linear algebra instruction.
   if (instruction->opcode() != SpvOpVectorTimesScalar &&
+      instruction->opcode() != SpvOpMatrixTimesScalar &&
       instruction->opcode() != SpvOpDot) {
     return false;
   }
@@ -77,6 +78,9 @@
     case SpvOpVectorTimesScalar:
       ReplaceOpVectorTimesScalar(ir_context, linear_algebra_instruction);
       break;
+    case SpvOpMatrixTimesScalar:
+      ReplaceOpMatrixTimesScalar(ir_context, linear_algebra_instruction);
+      break;
     case SpvOpDot:
       ReplaceOpDot(ir_context, linear_algebra_instruction);
       break;
@@ -110,7 +114,21 @@
                                ->type_id())
                  ->AsVector()
                  ->element_count();
-    case SpvOpDot: {
+    case SpvOpMatrixTimesScalar: {
+      // For each matrix column, |1 + column.size| OpCompositeExtract,
+      // |column.size| OpFMul and 1 OpCompositeConstruct instructions will be
+      // inserted.
+      auto matrix_instruction = ir_context->get_def_use_mgr()->GetDef(
+          instruction->GetSingleWordInOperand(0));
+      auto matrix_type =
+          ir_context->get_type_mgr()->GetType(matrix_instruction->type_id());
+      return 2 * matrix_type->AsMatrix()->element_count() *
+             (1 + matrix_type->AsMatrix()
+                      ->element_type()
+                      ->AsVector()
+                      ->element_count());
+    }
+    case SpvOpDot:
       // For each pair of vector components, 2 OpCompositeExtract and 1 OpFMul
       // will be inserted. The first two OpFMul instructions will result the
       // first OpFAdd instruction to be inserted. For each remaining OpFMul, 1
@@ -124,7 +142,6 @@
                      ->AsVector()
                      ->element_count() -
              2;
-    }
     default:
       assert(false && "Unsupported linear algebra instruction.");
       return 0;
@@ -179,6 +196,90 @@
   }
 }
 
+void TransformationReplaceLinearAlgebraInstruction::ReplaceOpMatrixTimesScalar(
+    opt::IRContext* ir_context,
+    opt::Instruction* linear_algebra_instruction) const {
+  // Gets OpMatrixTimesScalar in operands.
+  auto matrix_instruction = ir_context->get_def_use_mgr()->GetDef(
+      linear_algebra_instruction->GetSingleWordInOperand(0));
+  auto scalar_instruction = ir_context->get_def_use_mgr()->GetDef(
+      linear_algebra_instruction->GetSingleWordInOperand(1));
+
+  // Gets matrix information.
+  uint32_t matrix_column_count = ir_context->get_type_mgr()
+                                     ->GetType(matrix_instruction->type_id())
+                                     ->AsMatrix()
+                                     ->element_count();
+  auto matrix_column_type = ir_context->get_type_mgr()
+                                ->GetType(matrix_instruction->type_id())
+                                ->AsMatrix()
+                                ->element_type();
+  uint32_t matrix_column_size = matrix_column_type->AsVector()->element_count();
+
+  std::vector<uint32_t> composite_construct_ids(matrix_column_count);
+  uint32_t fresh_id_index = 0;
+
+  for (uint32_t i = 0; i < matrix_column_count; i++) {
+    // Extracts |matrix| column.
+    uint32_t matrix_extract_id = message_.fresh_ids(fresh_id_index++);
+    fuzzerutil::UpdateModuleIdBound(ir_context, matrix_extract_id);
+    linear_algebra_instruction->InsertBefore(MakeUnique<opt::Instruction>(
+        ir_context, SpvOpCompositeExtract,
+        ir_context->get_type_mgr()->GetId(matrix_column_type),
+        matrix_extract_id,
+        opt::Instruction::OperandList(
+            {{SPV_OPERAND_TYPE_ID, {matrix_instruction->result_id()}},
+             {SPV_OPERAND_TYPE_LITERAL_INTEGER, {i}}})));
+
+    std::vector<uint32_t> float_multiplication_ids(matrix_column_size);
+
+    for (uint32_t j = 0; j < matrix_column_size; j++) {
+      // Extracts |column| component.
+      uint32_t column_extract_id = message_.fresh_ids(fresh_id_index++);
+      fuzzerutil::UpdateModuleIdBound(ir_context, column_extract_id);
+      linear_algebra_instruction->InsertBefore(MakeUnique<opt::Instruction>(
+          ir_context, SpvOpCompositeExtract, scalar_instruction->type_id(),
+          column_extract_id,
+          opt::Instruction::OperandList(
+              {{SPV_OPERAND_TYPE_ID, {matrix_extract_id}},
+               {SPV_OPERAND_TYPE_LITERAL_INTEGER, {j}}})));
+
+      // Multiplies the |column| component with the |scalar|.
+      float_multiplication_ids[j] = message_.fresh_ids(fresh_id_index++);
+      fuzzerutil::UpdateModuleIdBound(ir_context, float_multiplication_ids[j]);
+      linear_algebra_instruction->InsertBefore(MakeUnique<opt::Instruction>(
+          ir_context, SpvOpFMul, scalar_instruction->type_id(),
+          float_multiplication_ids[j],
+          opt::Instruction::OperandList(
+              {{SPV_OPERAND_TYPE_ID, {column_extract_id}},
+               {SPV_OPERAND_TYPE_ID, {scalar_instruction->result_id()}}})));
+    }
+
+    // Constructs a new column multiplied by |scalar|.
+    opt::Instruction::OperandList composite_construct_in_operands;
+    for (uint32_t& float_multiplication_id : float_multiplication_ids) {
+      composite_construct_in_operands.push_back(
+          {SPV_OPERAND_TYPE_ID, {float_multiplication_id}});
+    }
+    composite_construct_ids[i] = message_.fresh_ids(fresh_id_index++);
+    fuzzerutil::UpdateModuleIdBound(ir_context, composite_construct_ids[i]);
+    linear_algebra_instruction->InsertBefore(MakeUnique<opt::Instruction>(
+        ir_context, SpvOpCompositeConstruct,
+        ir_context->get_type_mgr()->GetId(matrix_column_type),
+        composite_construct_ids[i], composite_construct_in_operands));
+  }
+
+  // The OpMatrixTimesScalar instruction is changed to an OpCompositeConstruct
+  // instruction.
+  linear_algebra_instruction->SetOpcode(SpvOpCompositeConstruct);
+  linear_algebra_instruction->SetInOperand(0, {composite_construct_ids[0]});
+  linear_algebra_instruction->SetInOperand(1, {composite_construct_ids[1]});
+  for (uint32_t i = 2; i < composite_construct_ids.size(); i++) {
+    linear_algebra_instruction->AddOperand(
+        {SPV_OPERAND_TYPE_ID, {composite_construct_ids[i]}});
+  }
+}
+
 void TransformationReplaceLinearAlgebraInstruction::ReplaceOpDot(
     opt::IRContext* ir_context,
     opt::Instruction* linear_algebra_instruction) const {
diff --git a/source/fuzz/transformation_replace_linear_algebra_instruction.h b/source/fuzz/transformation_replace_linear_algebra_instruction.h
index de280e4..45b1262 100644
--- a/source/fuzz/transformation_replace_linear_algebra_instruction.h
+++ b/source/fuzz/transformation_replace_linear_algebra_instruction.h
@@ -56,6 +56,10 @@
   void ReplaceOpVectorTimesScalar(opt::IRContext* ir_context,
                                   opt::Instruction* instruction) const;
 
+  // Replaces an OpMatrixTimesScalar instruction.
+  void ReplaceOpMatrixTimesScalar(opt::IRContext* ir_context,
+                                  opt::Instruction* instruction) const;
+
   // Replaces an OpDot instruction.
   void ReplaceOpDot(opt::IRContext* ir_context,
                     opt::Instruction* instruction) const;
diff --git a/test/fuzz/transformation_replace_linear_algebra_instruction_test.cpp b/test/fuzz/transformation_replace_linear_algebra_instruction_test.cpp
index 42906d0..c9a1aee 100644
--- a/test/fuzz/transformation_replace_linear_algebra_instruction_test.cpp
+++ b/test/fuzz/transformation_replace_linear_algebra_instruction_test.cpp
@@ -250,6 +250,280 @@
   ASSERT_TRUE(IsEqual(env, variant_shader, context.get()));
 }
 
+TEST(TransformationReplaceLinearAlgebraInstructionTest,
+     ReplaceOpMatrixTimesScalar) {
+  std::string reference_shader = R"(
+               OpCapability Shader
+          %1 = OpExtInstImport "GLSL.std.450"
+               OpMemoryModel Logical GLSL450
+               OpEntryPoint Fragment %54 "main"
+               OpExecutionMode %54 OriginUpperLeft
+
+; Types
+          %2 = OpTypeVoid
+          %3 = OpTypeFunction %2
+          %4 = OpTypeFloat 32
+          %5 = OpTypeVector %4 2
+          %6 = OpTypeVector %4 3
+          %7 = OpTypeVector %4 4
+          %8 = OpTypeMatrix %5 2
+          %9 = OpTypeMatrix %5 3
+         %10 = OpTypeMatrix %5 4
+         %11 = OpTypeMatrix %6 2
+         %12 = OpTypeMatrix %6 3
+         %13 = OpTypeMatrix %6 4
+         %14 = OpTypeMatrix %7 2
+         %15 = OpTypeMatrix %7 3
+         %16 = OpTypeMatrix %7 4
+
+; Constant scalars
+         %17 = OpConstant %4 1
+         %18 = OpConstant %4 2
+         %19 = OpConstant %4 3
+         %20 = OpConstant %4 4
+         %21 = OpConstant %4 5
+         %22 = OpConstant %4 6
+         %23 = OpConstant %4 7
+         %24 = OpConstant %4 8
+         %25 = OpConstant %4 9
+         %26 = OpConstant %4 10
+         %27 = OpConstant %4 11
+         %28 = OpConstant %4 12
+         %29 = OpConstant %4 13
+         %30 = OpConstant %4 14
+         %31 = OpConstant %4 15
+         %32 = OpConstant %4 16
+
+; Constant vectors
+         %33 = OpConstantComposite %5 %17 %18
+         %34 = OpConstantComposite %5 %19 %20
+         %35 = OpConstantComposite %5 %21 %22
+         %36 = OpConstantComposite %5 %23 %24
+         %37 = OpConstantComposite %6 %17 %18 %19
+         %38 = OpConstantComposite %6 %20 %21 %22
+         %39 = OpConstantComposite %6 %23 %24 %25
+         %40 = OpConstantComposite %6 %26 %27 %28
+         %41 = OpConstantComposite %7 %17 %18 %19 %20
+         %42 = OpConstantComposite %7 %21 %22 %23 %24
+         %43 = OpConstantComposite %7 %25 %26 %27 %28
+         %44 = OpConstantComposite %7 %29 %30 %31 %32
+
+; Constant matrices
+         %45 = OpConstantComposite %8 %33 %34
+         %46 = OpConstantComposite %9 %33 %34 %35
+         %47 = OpConstantComposite %10 %33 %34 %35 %36
+         %48 = OpConstantComposite %11 %37 %38
+         %49 = OpConstantComposite %12 %37 %38 %39
+         %50 = OpConstantComposite %13 %37 %38 %39 %40
+         %51 = OpConstantComposite %14 %41 %42
+         %52 = OpConstantComposite %15 %41 %42 %43
+         %53 = OpConstantComposite %16 %41 %42 %43 %44
+
+; main function
+         %54 = OpFunction %2 None %3
+         %55 = OpLabel
+
+; Multiplying 2-row matrices by scalar
+         %56 = OpMatrixTimesScalar %8 %45 %17
+         %57 = OpMatrixTimesScalar %9 %46 %18
+         %58 = OpMatrixTimesScalar %10 %47 %19
+
+; Multiplying 3-row matrices by scalar
+         %59 = OpMatrixTimesScalar %11 %48 %21
+         %60 = OpMatrixTimesScalar %12 %49 %22
+         %61 = OpMatrixTimesScalar %13 %50 %23
+
+; Multiplying 4-row matrices by scalar
+         %62 = OpMatrixTimesScalar %14 %51 %24
+         %63 = OpMatrixTimesScalar %15 %52 %25
+         %64 = OpMatrixTimesScalar %16 %53 %26
+               OpReturn
+               OpFunctionEnd
+  )";
+
+  const auto env = SPV_ENV_UNIVERSAL_1_5;
+  const auto consumer = nullptr;
+  const auto context =
+      BuildModule(env, consumer, reference_shader, kFuzzAssembleOption);
+  ASSERT_TRUE(IsValid(env, context.get()));
+
+  FactManager fact_manager;
+  spvtools::ValidatorOptions validator_options;
+  TransformationContext transformation_context(&fact_manager,
+                                               validator_options);
+
+  auto instruction_descriptor =
+      MakeInstructionDescriptor(56, SpvOpMatrixTimesScalar, 0);
+  auto transformation = TransformationReplaceLinearAlgebraInstruction(
+      {65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76}, instruction_descriptor);
+  transformation.Apply(context.get(), &transformation_context);
+
+  instruction_descriptor =
+      MakeInstructionDescriptor(57, SpvOpMatrixTimesScalar, 0);
+  transformation = TransformationReplaceLinearAlgebraInstruction(
+      {77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94},
+      instruction_descriptor);
+  transformation.Apply(context.get(), &transformation_context);
+
+  instruction_descriptor =
+      MakeInstructionDescriptor(58, SpvOpMatrixTimesScalar, 0);
+  transformation = TransformationReplaceLinearAlgebraInstruction(
+      {95,  96,  97,  98,  99,  100, 101, 102, 103, 104, 105, 106,
+       107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118},
+      instruction_descriptor);
+  transformation.Apply(context.get(), &transformation_context);
+
+  std::string variant_shader = R"(
+               OpCapability Shader
+          %1 = OpExtInstImport "GLSL.std.450"
+               OpMemoryModel Logical GLSL450
+               OpEntryPoint Fragment %54 "main"
+               OpExecutionMode %54 OriginUpperLeft
+
+; Types
+          %2 = OpTypeVoid
+          %3 = OpTypeFunction %2
+          %4 = OpTypeFloat 32
+          %5 = OpTypeVector %4 2
+          %6 = OpTypeVector %4 3
+          %7 = OpTypeVector %4 4
+          %8 = OpTypeMatrix %5 2
+          %9 = OpTypeMatrix %5 3
+         %10 = OpTypeMatrix %5 4
+         %11 = OpTypeMatrix %6 2
+         %12 = OpTypeMatrix %6 3
+         %13 = OpTypeMatrix %6 4
+         %14 = OpTypeMatrix %7 2
+         %15 = OpTypeMatrix %7 3
+         %16 = OpTypeMatrix %7 4
+
+; Constant scalars
+         %17 = OpConstant %4 1
+         %18 = OpConstant %4 2
+         %19 = OpConstant %4 3
+         %20 = OpConstant %4 4
+         %21 = OpConstant %4 5
+         %22 = OpConstant %4 6
+         %23 = OpConstant %4 7
+         %24 = OpConstant %4 8
+         %25 = OpConstant %4 9
+         %26 = OpConstant %4 10
+         %27 = OpConstant %4 11
+         %28 = OpConstant %4 12
+         %29 = OpConstant %4 13
+         %30 = OpConstant %4 14
+         %31 = OpConstant %4 15
+         %32 = OpConstant %4 16
+
+; Constant vectors
+         %33 = OpConstantComposite %5 %17 %18
+         %34 = OpConstantComposite %5 %19 %20
+         %35 = OpConstantComposite %5 %21 %22
+         %36 = OpConstantComposite %5 %23 %24
+         %37 = OpConstantComposite %6 %17 %18 %19
+         %38 = OpConstantComposite %6 %20 %21 %22
+         %39 = OpConstantComposite %6 %23 %24 %25
+         %40 = OpConstantComposite %6 %26 %27 %28
+         %41 = OpConstantComposite %7 %17 %18 %19 %20
+         %42 = OpConstantComposite %7 %21 %22 %23 %24
+         %43 = OpConstantComposite %7 %25 %26 %27 %28
+         %44 = OpConstantComposite %7 %29 %30 %31 %32
+
+; Constant matrices
+         %45 = OpConstantComposite %8 %33 %34
+         %46 = OpConstantComposite %9 %33 %34 %35
+         %47 = OpConstantComposite %10 %33 %34 %35 %36
+         %48 = OpConstantComposite %11 %37 %38
+         %49 = OpConstantComposite %12 %37 %38 %39
+         %50 = OpConstantComposite %13 %37 %38 %39 %40
+         %51 = OpConstantComposite %14 %41 %42
+         %52 = OpConstantComposite %15 %41 %42 %43
+         %53 = OpConstantComposite %16 %41 %42 %43 %44
+
+; main function
+         %54 = OpFunction %2 None %3
+         %55 = OpLabel
+
+; Multiplying 2x2 matrix by scalar
+         %65 = OpCompositeExtract %5 %45 0
+         %66 = OpCompositeExtract %4 %65 0
+         %67 = OpFMul %4 %66 %17
+         %68 = OpCompositeExtract %4 %65 1
+         %69 = OpFMul %4 %68 %17
+         %70 = OpCompositeConstruct %5 %67 %69
+         %71 = OpCompositeExtract %5 %45 1
+         %72 = OpCompositeExtract %4 %71 0
+         %73 = OpFMul %4 %72 %17
+         %74 = OpCompositeExtract %4 %71 1
+         %75 = OpFMul %4 %74 %17
+         %76 = OpCompositeConstruct %5 %73 %75
+         %56 = OpCompositeConstruct %8 %70 %76
+
+; Multiplying 2x3 matrix by scalar
+         %77 = OpCompositeExtract %5 %46 0
+         %78 = OpCompositeExtract %4 %77 0
+         %79 = OpFMul %4 %78 %18
+         %80 = OpCompositeExtract %4 %77 1
+         %81 = OpFMul %4 %80 %18
+         %82 = OpCompositeConstruct %5 %79 %81
+         %83 = OpCompositeExtract %5 %46 1
+         %84 = OpCompositeExtract %4 %83 0
+         %85 = OpFMul %4 %84 %18
+         %86 = OpCompositeExtract %4 %83 1
+         %87 = OpFMul %4 %86 %18
+         %88 = OpCompositeConstruct %5 %85 %87
+         %89 = OpCompositeExtract %5 %46 2
+         %90 = OpCompositeExtract %4 %89 0
+         %91 = OpFMul %4 %90 %18
+         %92 = OpCompositeExtract %4 %89 1
+         %93 = OpFMul %4 %92 %18
+         %94 = OpCompositeConstruct %5 %91 %93
+         %57 = OpCompositeConstruct %9 %82 %88 %94
+
+; Multiplying 2x4 matrix by scalar
+         %95 = OpCompositeExtract %5 %47 0
+         %96 = OpCompositeExtract %4 %95 0
+         %97 = OpFMul %4 %96 %19
+         %98 = OpCompositeExtract %4 %95 1
+         %99 = OpFMul %4 %98 %19
+        %100 = OpCompositeConstruct %5 %97 %99
+        %101 = OpCompositeExtract %5 %47 1
+        %102 = OpCompositeExtract %4 %101 0
+        %103 = OpFMul %4 %102 %19
+        %104 = OpCompositeExtract %4 %101 1
+        %105 = OpFMul %4 %104 %19
+        %106 = OpCompositeConstruct %5 %103 %105
+        %107 = OpCompositeExtract %5 %47 2
+        %108 = OpCompositeExtract %4 %107 0
+        %109 = OpFMul %4 %108 %19
+        %110 = OpCompositeExtract %4 %107 1
+        %111 = OpFMul %4 %110 %19
+        %112 = OpCompositeConstruct %5 %109 %111
+        %113 = OpCompositeExtract %5 %47 3
+        %114 = OpCompositeExtract %4 %113 0
+        %115 = OpFMul %4 %114 %19
+        %116 = OpCompositeExtract %4 %113 1
+        %117 = OpFMul %4 %116 %19
+        %118 = OpCompositeConstruct %5 %115 %117
+         %58 = OpCompositeConstruct %10 %100 %106 %112 %118
+
+; Multiplying 3-row matrices by scalar
+         %59 = OpMatrixTimesScalar %11 %48 %21
+         %60 = OpMatrixTimesScalar %12 %49 %22
+         %61 = OpMatrixTimesScalar %13 %50 %23
+
+; Multiplying 4-row matrices by scalar
+         %62 = OpMatrixTimesScalar %14 %51 %24
+         %63 = OpMatrixTimesScalar %15 %52 %25
+         %64 = OpMatrixTimesScalar %16 %53 %26
+               OpReturn
+               OpFunctionEnd
+  )";
+
+  ASSERT_TRUE(IsValid(env, context.get()));
+  ASSERT_TRUE(IsEqual(env, variant_shader, context.get()));
+}
+
 TEST(TransformationReplaceLinearAlgebraInstructionTest, ReplaceOpDot) {
   std::string reference_shader = R"(
                OpCapability Shader