[Mlir-commits] [mlir] b91ce7b - [Tosa] : Fix integer overflow for computing intmax+1 in tosa.cast to linalg. (#112455)

[llvmlistbot at llvm.org]

Author: Sayan Saha
Date: 2024-10-25T10:51:09+01:00
New Revision: b91ce7bc0eda8ec91c7ba724fbe78607843e0e06

URL: https://github.com/llvm/llvm-project/commit/b91ce7bc0eda8ec91c7ba724fbe78607843e0e06
DIFF: https://github.com/llvm/llvm-project/commit/b91ce7bc0eda8ec91c7ba724fbe78607843e0e06.diff

LOG: [Tosa] : Fix integer overflow for computing intmax+1 in tosa.cast to linalg. (#112455)

This PR fixes an issue related to integer overflow when computing
`(intmax+1)` for `i64` during `tosa-to-linalg` pass for `tosa.cast`.

Found this issue while debugging a numerical mismatch for `deeplabv3`
model from `torchvision` represented in `tosa` dialect using the
`TorchToTosa` pipeline in `torch-mlir` repository. `torch.aten.to.dtype`
is converted to `tosa.cast` that casts `f32` to `i64` type. Technically
by the specification, `tosa.cast` doesn't handle casting `f32` to `i64`.
So it's possible to add a verifier to error out for such tosa ops
instead of producing incorrect code. However, I chose to fix the
overflow issue to still be able to represent the `deeplabv3` model with
`tosa` ops in the above-mentioned pipeline. Open to suggestions if
adding the verifier is more appropriate instead.

Added: 
    

Modified: 
    mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp
    mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir

Removed: 
    


################################################################################
diff  --git a/mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp b/mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp
index 495f1b4f10b028..251c48859d2de0 100644
--- a/mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp
+++ b/mlir/lib/Conversion/TosaToLinalg/TosaToLinalg.cpp
@@ -555,9 +555,10 @@ static Value createLinalgBodyCalculationForElementwiseOp(
       auto intMaxPlusOneFP = rewriter.create<arith::ConstantOp>(
           loc, rewriter.getFloatAttr(
                    getElementTypeOrSelf(srcTy),
-                   APInt::getSignedMaxValue(dstTy.getIntOrFloatBitWidth())
-                           .getSExtValue() +
-                       1));
+                   static_cast<double>(
+                       APInt::getSignedMaxValue(dstTy.getIntOrFloatBitWidth())
+                           .getSExtValue()) +
+                       1.0f));
 
       auto intMax = rewriter.create<arith::ConstantOp>(
           loc, rewriter.getIntegerAttr(

diff  --git a/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir b/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
index f9d37f9427d4f4..1a29d3f9f3507c 100644
--- a/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
+++ b/mlir/test/Conversion/TosaToLinalg/tosa-to-linalg.mlir
@@ -1929,3 +1929,30 @@ func.func @test_dynamic_fft2d(%arg0: tensor<?x?x?xf32>, %arg1: tensor<?x?x?xf32>
   %output_real, %output_imag = "tosa.fft2d"(%arg0, %arg1) {inverse = true} : (tensor<?x?x?xf32>, tensor<?x?x?xf32>) -> (tensor<?x?x?xf32>, tensor<?x?x?xf32>)
   return %output_real, %output_imag : tensor<?x?x?xf32>, tensor<?x?x?xf32>
 }
+
+
+// -----
+
+// CHECK: #[[$MAP0:.+]] = affine_map<(d0) -> (0)>
+// CHECK: #[[$MAP1:.+]] = affine_map<(d0) -> (d0)>
+
+// CHECK-LABEL:   func.func @test_cast_fp32_i64(
+// CHECK-SAME:                                  %[[ARG0:.*]]: tensor<1xf32>) -> tensor<1xi64> {
+// CHECK:           %[[EMPTY_TENSOR:.*]] = tensor.empty() : tensor<1xi64>
+// CHECK:           %[[RESULT:.*]] = linalg.generic {indexing_maps = [#[[$MAP0]], #[[$MAP1]]], iterator_types = ["parallel"]} ins(%[[ARG0]] : tensor<1xf32>) outs(%[[EMPTY_TENSOR]] : tensor<1xi64>) {
+// CHECK:           ^bb0(%[[IN:.*]]: f32, %[[OUT:.*]]: i64):
+// CHECK:             %[[ROUND_EVEN:.*]] = math.roundeven %[[IN]] : f32
+// CHECK:             %[[FP_INT_MIN:.*]] = arith.constant -9.22337203E+18 : f32
+// CHECK:             %[[FP_INT_MAX_PLUS_ONE:.*]] = arith.constant 9.22337203E+18 : f32
+// CHECK:             %[[INT_MAX:.*]] = arith.constant 9223372036854775807 : i64
+// CHECK:             %[[MAX:.*]] = arith.maximumf %[[ROUND_EVEN]], %[[FP_INT_MIN]] : f32
+// CHECK:             %[[FPTOSI:.*]] = arith.fptosi %[[MAX]] : f32 to i64
+// CHECK:             %[[CMPF:.*]] = arith.cmpf uge, %[[ROUND_EVEN]], %[[FP_INT_MAX_PLUS_ONE]] : f32
+// CHECK:             %[[SELECT:.*]] = arith.select %[[CMPF]], %[[INT_MAX]], %[[FPTOSI]] : i64
+// CHECK:             linalg.yield %[[SELECT]] : i64
+// CHECK:           } -> tensor<1xi64>
+// CHECK:           return %[[RESULT]] : tensor<1xi64>
+func.func @test_cast_fp32_i64(%arg0: tensor<1xf32>) -> (tensor<1xi64>) {
+  %0 = tosa.cast %arg0 : (tensor<1xf32>) -> tensor<1xi64>
+  return %0: tensor<1xi64>
+}