[mlir] IntegerRangeAnalysis: handle vector types in getDestWidth() #114898
+12
−1
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
PR llvm#112292 added support for vectors to the integer range inference interface and analysis, but didn't update the getDestWidth() method. This caused crashes when trying to infer the ranges of `arith.extsi` with vector inputs, as the code would try to sign-extend a N-bit value to a 0-bit one, which would assert and crash. This commit fixes the issue by adding a getElementTypeOrSelf().
Member
|
@llvm/pr-subscribers-mlir-vector Author: Krzysztof Drewniak (krzysz00) ChangesPR #112292 added support for vectors to the integer range inference interface and analysis, but didn't update the getDestWidth() method. This caused crashes when trying to infer the ranges of This commit fixes the issue by adding a getElementTypeOrSelf(). Full diff: https://github.com/llvm/llvm-project/pull/114898.diff 2 Files Affected:
diff --git a/mlir/lib/Interfaces/InferIntRangeInterface.cpp b/mlir/lib/Interfaces/InferIntRangeInterface.cpp
index d879b93586899b..63658518dd4a3b 100644
--- a/mlir/lib/Interfaces/InferIntRangeInterface.cpp
+++ b/mlir/lib/Interfaces/InferIntRangeInterface.cpp
@@ -8,6 +8,7 @@
#include "mlir/Interfaces/InferIntRangeInterface.h"
#include "mlir/IR/BuiltinTypes.h"
+#include "mlir/IR/TypeUtilities.h"
#include "mlir/Interfaces/InferIntRangeInterface.cpp.inc"
#include <optional>
@@ -28,6 +29,7 @@ const APInt &ConstantIntRanges::smin() const { return sminVal; }
const APInt &ConstantIntRanges::smax() const { return smaxVal; }
unsigned ConstantIntRanges::getStorageBitwidth(Type type) {
+ type = getElementTypeOrSelf(type);
if (type.isIndex())
return IndexType::kInternalStorageBitWidth;
if (auto integerType = dyn_cast<IntegerType>(type))
diff --git a/mlir/test/Dialect/Vector/int-range-interface.mlir b/mlir/test/Dialect/Vector/int-range-interface.mlir
index 29282423089ba6..09dfe932a52323 100644
--- a/mlir/test/Dialect/Vector/int-range-interface.mlir
+++ b/mlir/test/Dialect/Vector/int-range-interface.mlir
@@ -96,7 +96,7 @@ func.func @vector_insertelement() -> vector<4xindex> {
// CHECK-LABEL: func @test_loaded_vector_extract
// No bounds
-// CHECK: test.reflect_bounds %{{.*}} : i32
+// CHECK: test.reflect_bounds {smax = 2147483647 : si32, smin = -2147483648 : si32, umax = 4294967295 : ui32, umin = 0 : ui32} %{{.*}} : i32
func.func @test_loaded_vector_extract(%memref : memref<16xi32>) -> i32 {
%c0 = arith.constant 0 : index
%v = vector.load %memref[%c0] : memref<16xi32>, vector<4xi32>
@@ -104,3 +104,12 @@ func.func @test_loaded_vector_extract(%memref : memref<16xi32>) -> i32 {
%bounds = test.reflect_bounds %e : i32
func.return %bounds : i32
}
+
+// CHECK-LABEL: func @test_vector_extsi
+// CHECK: test.reflect_bounds {smax = 5 : si32, smin = 1 : si32, umax = 5 : ui32, umin = 1 : ui32}
+func.func @test_vector_extsi() -> vector<2xi32> {
+ %0 = test.with_bounds {smax = 5 : si8, smin = 1 : si8, umax = 5 : ui8, umin = 1 : ui8 } : vector<2xi8>
+ %1 = arith.extsi %0 : vector<2xi8> to vector<2xi32>
+ %2 = test.reflect_bounds %1 : vector<2xi32>
+ func.return %2 : vector<2xi32>
+}
|
Member
|
@llvm/pr-subscribers-mlir Author: Krzysztof Drewniak (krzysz00) ChangesPR #112292 added support for vectors to the integer range inference interface and analysis, but didn't update the getDestWidth() method. This caused crashes when trying to infer the ranges of This commit fixes the issue by adding a getElementTypeOrSelf(). Full diff: https://github.com/llvm/llvm-project/pull/114898.diff 2 Files Affected:
diff --git a/mlir/lib/Interfaces/InferIntRangeInterface.cpp b/mlir/lib/Interfaces/InferIntRangeInterface.cpp
index d879b93586899b..63658518dd4a3b 100644
--- a/mlir/lib/Interfaces/InferIntRangeInterface.cpp
+++ b/mlir/lib/Interfaces/InferIntRangeInterface.cpp
@@ -8,6 +8,7 @@
#include "mlir/Interfaces/InferIntRangeInterface.h"
#include "mlir/IR/BuiltinTypes.h"
+#include "mlir/IR/TypeUtilities.h"
#include "mlir/Interfaces/InferIntRangeInterface.cpp.inc"
#include <optional>
@@ -28,6 +29,7 @@ const APInt &ConstantIntRanges::smin() const { return sminVal; }
const APInt &ConstantIntRanges::smax() const { return smaxVal; }
unsigned ConstantIntRanges::getStorageBitwidth(Type type) {
+ type = getElementTypeOrSelf(type);
if (type.isIndex())
return IndexType::kInternalStorageBitWidth;
if (auto integerType = dyn_cast<IntegerType>(type))
diff --git a/mlir/test/Dialect/Vector/int-range-interface.mlir b/mlir/test/Dialect/Vector/int-range-interface.mlir
index 29282423089ba6..09dfe932a52323 100644
--- a/mlir/test/Dialect/Vector/int-range-interface.mlir
+++ b/mlir/test/Dialect/Vector/int-range-interface.mlir
@@ -96,7 +96,7 @@ func.func @vector_insertelement() -> vector<4xindex> {
// CHECK-LABEL: func @test_loaded_vector_extract
// No bounds
-// CHECK: test.reflect_bounds %{{.*}} : i32
+// CHECK: test.reflect_bounds {smax = 2147483647 : si32, smin = -2147483648 : si32, umax = 4294967295 : ui32, umin = 0 : ui32} %{{.*}} : i32
func.func @test_loaded_vector_extract(%memref : memref<16xi32>) -> i32 {
%c0 = arith.constant 0 : index
%v = vector.load %memref[%c0] : memref<16xi32>, vector<4xi32>
@@ -104,3 +104,12 @@ func.func @test_loaded_vector_extract(%memref : memref<16xi32>) -> i32 {
%bounds = test.reflect_bounds %e : i32
func.return %bounds : i32
}
+
+// CHECK-LABEL: func @test_vector_extsi
+// CHECK: test.reflect_bounds {smax = 5 : si32, smin = 1 : si32, umax = 5 : ui32, umin = 1 : ui32}
+func.func @test_vector_extsi() -> vector<2xi32> {
+ %0 = test.with_bounds {smax = 5 : si8, smin = 1 : si8, umax = 5 : ui8, umin = 1 : ui8 } : vector<2xi8>
+ %1 = arith.extsi %0 : vector<2xi8> to vector<2xi32>
+ %2 = test.reflect_bounds %1 : vector<2xi32>
+ func.return %2 : vector<2xi32>
+}
|
Hardcode84
approved these changes
Nov 5, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR #112292 added support for vectors to the integer range inference interface and analysis, but didn't update the getDestWidth() method. This caused crashes when trying to infer the ranges of
arith.extsiwith vector inputs, as the code would try to sign-extend a N-bit value to a 0-bit one, which would assert and crash.This commit fixes the issue by adding a getElementTypeOrSelf().