[mlir][arith] adding addition regression tests

[pingshiyu]

arith addition regression tests, a component of the large #92272

[pingshiyu]

@kuhar @banach-space

[llvmbot]

@llvm/pr-subscribers-mlir

@llvm/pr-subscribers-mlir-arith

Author: Jacob Yu (pingshiyu)

Changes

arith addition regression tests, a component of the large #92272

---

Full diff: https://github.com/llvm/llvm-project/pull/96973.diff

1 Files Affected:

• (added) mlir/test/Integration/Dialect/Arith/CPU/addition.mlir (+57)

diff --git a/mlir/test/Integration/Dialect/Arith/CPU/addition.mlir b/mlir/test/Integration/Dialect/Arith/CPU/addition.mlir
new file mode 100644
index 0000000000000..0833ebf1c8ecc
--- /dev/null
+++ b/mlir/test/Integration/Dialect/Arith/CPU/addition.mlir
@@ -0,0 +1,57 @@
+// Tests arith operations on i1 type.
+// These tests are intended to be target agnostic: they should yield the same results 
+// regardless of the target platform.
+
+// RUN: mlir-opt %s --convert-scf-to-cf --convert-cf-to-llvm --convert-vector-to-llvm \
+// RUN:             --convert-func-to-llvm --convert-arith-to-llvm | \
+// RUN:   mlir-cpu-runner -e entry -entry-point-result=void \
+// RUN:                   --shared-libs=%mlir_c_runner_utils | \
+// RUN:   FileCheck %s --match-full-lines
+
+func.func @zero_plus_one_on_i1() {
+    // addi on i1
+    // addi(0, 1) : i1 = 1 : i1; addi(0, -1) : i1 = 1
+    // CHECK:      1
+    // CHECK-NEXT: 1
+    // CHECK-NEXT: 1
+    %false = arith.constant 0 : i1
+    %true = arith.constant 1 : i1
+    %true_0 = arith.constant -1 : i1
+    vector.print %true_0 : i1
+    %0 = arith.addi %false, %true : i1
+    vector.print %0 : i1
+    %1 = arith.addi %false, %true_0 : i1
+    vector.print %1 : i1
+    return
+}
+
+func.func @addui_extended_i1() {
+    // addui_extended on i1
+    // addui_extended 1 1 : i1 = 0, 1
+    // CHECK-NEXT: 0
+    // CHECK-NEXT: 1
+    %true = arith.constant 1 : i1
+    %sum, %overflow = arith.addui_extended %true, %true : i1, i1
+    vector.print %sum : i1
+    vector.print %overflow : i1
+    return
+}
+
+func.func @addui_extended_overflow_bit_is_n1() {
+    // addui_extended overflow bit is treated as -1
+    // addui_extended -1633386 -1643386 = ... 1 (overflow because negative numbers are large positive numbers)
+    // CHECK-NEXT: 0
+    %c-16433886_i64 = arith.constant -16433886 : i64
+    %sum, %overflow = arith.addui_extended %c-16433886_i64, %c-16433886_i64 : i64, i1
+    %false = arith.constant false
+    %0 = arith.cmpi sge, %overflow, %false : i1
+    vector.print %0 : i1 // but prints as "1"
+    return
+}
+
+func.func @entry() {
+    func.call @zero_plus_one_on_i1() : () -> ()
+    func.call @addui_extended_i1() : () -> ()
+    func.call @addui_extended_overflow_bit_is_n1() : () -> ()
+    return
+}

[banach-space]

LGTM, thank you for addressing my comments and for pushing on this!

[pingshiyu]

@banach-space thank you for taking your time over the last week (and 8 weeks ago!) going through these PRs!

I'll add the rest of the tests in the same format, more to come tomorrow - but hopefully should be quicker to review since we have the format established :)

@kuhar, could you please take a look sometime & give a tick if all LGTY? (along with #96975 and #96974)

[pingshiyu]

sorry I don't have

@banach-space thank you for taking your time over the last week (and 8 weeks ago!) going through these PRs!

I'll add the rest of the tests in the same format, more to come tomorrow - but hopefully should be quicker to review since we have the format established :)

@kuhar, could you please take a look sometime & give a tick if all LGTY? (along with #96975 and #96974)

Thank you @banach-space and @kuhar for going through this tranche of changes! I don't have write rights to LLVM unfortunately so would be great if you could also push the merge button on these PRs too :)

[kuhar]

@pingshiyu merged

[pingshiyu]

@kuhar thank you!!