Fixed syntax issues && handled EH label &&TODO: fix computation tuple

9Tempest · Nov 20, 2023 · cc2fad5 · cc2fad5
1 parent 1e874a4
commit cc2fad5
Show file tree

Hide file tree

Showing 7 changed files with 138,471 additions and 25 deletions.
diff --git a/gematria/basic_block/basic_block.h b/gematria/basic_block/basic_block.h
@@ -284,6 +284,9 @@ struct Instruction {
   uint64_t address = 0;
   // The size of the instruction.
   size_t size = 0;
+
+  // The instruction is valid or not
+  bool is_valid = true;
 };
 
 std::ostream& operator<<(std::ostream& os, const Instruction& instruction);

diff --git a/gematria/datasets/bhive_importer.cc b/gematria/datasets/bhive_importer.cc
@@ -174,29 +174,21 @@ absl::StatusOr<BasicBlockProto> BHiveImporter::BasicBlockProtoFromMBBName(
   LOG("MBB is " << *MBB);
   for (llvm::MachineInstr& MI : *MBB){
     // if MI is a control instruction(ret,branch,jmp), skip it
-    if (MI.isInlineAsm() || MI.isTerminator()) {
+    if (MI.isInlineAsm() || MI.isTerminator() || MI.isEHLabel()) {
       LOG("MI is a control instruction, skipping it " << MI);
       continue;
     }
 
     // Assert MI cannot be a CALL instruction
     assert(!MI.isCall() && "MI is a CALL instruction, bad dataset");
-    *basic_block_proto.add_canonicalized_instructions() = ProtoFromInstruction(
-         canonicalizer_.InstructionFromMachineInstr(MI));
+    auto I = canonicalizer_.InstructionFromMachineInstr(MI);
+    if (!I.is_valid) {
+      LOG("MI is not valid, skipping it " << MI);
+      return absl::InvalidArgumentError(
+        absl::StrCat("Could not parse MachineInstr "));
+    }
+    *basic_block_proto.add_canonicalized_instructions() = ProtoFromInstruction(I);
   }
-
-  // for (DisassembledInstruction& instruction : *instructions) {
-  // NOT VERY IMPORTANT THESE 3 LINES
-  //   MachineInstructionProto& machine_instruction =
-  //       *basic_block_proto.add_machine_instructions();
-  //   machine_instruction.set_address(instruction.address);
-  //   machine_instruction.set_assembly(instruction.assembly);
-  //   machine_instruction.set_machine_code(instruction.machine_code);
-  // VERY IMPORTANT!!! Do this first TODO: change this to use the unique name to get the MBB
-  //   *basic_block_proto.add_canonicalized_instructions() = ProtoFromInstruction(
-  //       canonicalizer_.InstructionFromMCInst(instruction.mc_inst));
-  // }
-
   return basic_block_proto;
 }
 
@@ -219,16 +211,16 @@ absl::StatusOr<BasicBlockWithThroughputProto> BHiveImporter::ParseMIRCsvLine(
         "different, but were both %d: %s",
         BB_name_index, line));
   }
-  // const std::string_view BB_unique_name =
-  //     columns[BB_name_index];
+  const std::string_view BB_unique_name =
+      columns[BB_name_index];
   const std::string_view throughput_str = columns[throughput_column_index];
 
   BasicBlockWithThroughputProto proto;
 
-  // absl::StatusOr<BasicBlockProto> block_proto_or_status =
-  //     BasicBlockProtoFromMBBName(BB_unique_name, base_address);
-  // if (!block_proto_or_status.ok()) return block_proto_or_status.status();
-  // *proto.mutable_basic_block() = std::move(block_proto_or_status).value();
+  absl::StatusOr<BasicBlockProto> block_proto_or_status =
+      BasicBlockProtoFromMBBName(BB_unique_name, base_address);
+  if (!block_proto_or_status.ok()) return block_proto_or_status.status();
+  *proto.mutable_basic_block() = std::move(block_proto_or_status).value();
 
   double throughput_cycles = 0.0;
   if (!absl::SimpleAtod(throughput_str, &throughput_cycles)) {

diff --git a/gematria/datasets/bhive_importer_test.cc b/gematria/datasets/bhive_importer_test.cc
@@ -235,7 +235,10 @@ TEST_F(BHiveImporterTest, MIRDatasetBasicTest) {
 }
 
 TEST_F(BHiveImporterTest, MIRDatasetTest2) {
-  EXPECT_THAT(x86_bhive_importer_->LoadMIRModule("mir_input/output/BatchRulesReduceOps.mir"),
+  EXPECT_THAT(x86_bhive_importer_->LoadMIRModule("sample_dataset/native_test.mir"),
+              IsOk());
+  EXPECT_THAT(x86_bhive_importer_->ParseMIRCsvLine(kSourceName, "a,b,BB_13,2.37", 2,
+                                                     3, kScaling),
               IsOk());
 }
 

diff --git a/gematria/datasets/python/import_from_mir.py b/gematria/datasets/python/import_from_mir.py
@@ -160,7 +160,7 @@ def main(argv: Sequence[str]) -> None:
                                 throughput_scaling=_THROUGHPUT_SCALING.value,
                             )
                             writer.write(block_proto.SerializeToString())
-                        except bhive_importer.BasicBlockNotFoundError:
+                        except:
                             num_skipped_blocks += 1
             except:
                 logging.exception('Could not load file "%s"', mir_file)

diff --git a/gematria/llvm/canonicalizer.cc b/gematria/llvm/canonicalizer.cc
@@ -495,7 +495,7 @@ void X86Canonicalizer::AddOperand(const llvm::MachineInstr& mi, int operand_inde
     llvm::errs() << "Unsupported operand type: ";
     operand.print(llvm::errs());
     llvm::errs() << "\n";
-    assert(false);
+    instruction.is_valid = false;
   }
 }