[mlir][arith] Add truncation and extension integration tests

[pingshiyu]

Trunc/ext operations regression tests, from the original larger PR that has been broken down: #92272

[llvmbot]

@llvm/pr-subscribers-mlir

Author: Jacob Yu (pingshiyu)

Changes

Trunc/ext operations regression tests, from the original larger PR that has been broken down: #92272

---

Full diff: https://github.com/llvm/llvm-project/pull/98182.diff

1 Files Affected:

• (added) mlir/test/Integration/Dialect/Arith/CPU/trunc-ext.mlir (+90)

diff --git a/mlir/test/Integration/Dialect/Arith/CPU/trunc-ext.mlir b/mlir/test/Integration/Dialect/Arith/CPU/trunc-ext.mlir
new file mode 100644
index 0000000000000..4c27f2bf7b318
--- /dev/null
+++ b/mlir/test/Integration/Dialect/Arith/CPU/trunc-ext.mlir
@@ -0,0 +1,90 @@
+// 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 @extsi_i1_i16(%v1 : i1) {
+    vector.print str "@extsi_i1_i16\n"
+    %0 = arith.extsi %v1 : i1 to i16
+    vector.print %0 : i16
+    return
+}
+
+func.func @extui_i1_i64(%v1 : i1) {
+    vector.print str "@extui_i1_i64\n"
+    %0 = arith.extui %v1 : i1 to i64
+    vector.print %0 : i64
+    return
+}
+
+func.func @trunci_i16_i8(%v1 : i16) {
+    vector.print str "@trunci_i16_i8\n"
+    %0 = arith.trunci %v1 : i16 to i8
+    vector.print %0 : i8
+    return
+}
+
+func.func @extsi() {
+    // ------------------------------------------------
+    // Test extending from i1
+    // ------------------------------------------------
+    %true = arith.constant -1 : i1
+
+    // extsi on 1 : i1
+    // extsi(1: i1) = -1 : i16
+    // CHECK-LABEL: @extsi_i1_i16
+    // CHECK-NEXT:  -1
+    func.call @extsi_i1_i16(%true) : (i1) -> ()
+
+    // ------------------------------------------------
+    // TODO: Test extension from i8, i16 etc..
+    // ------------------------------------------------
+
+    return
+}
+
+func.func @extui() {
+    // ------------------------------------------------
+    // Test extending from i1
+    // ------------------------------------------------
+    %true = arith.constant true
+
+    // extui should extend i1 with 0 bits not 1s
+    // extui(1 : i1) = 1 : i64
+    // CHECK-LABEL: @extui_i1_i64
+    // CHECK-NEXT:  1
+    func.call @extui_i1_i64(%true) : (i1) -> ()
+
+    // ------------------------------------------------
+    // TODO: Test extension from i8, i16 etc..
+    // ------------------------------------------------
+
+    return
+}
+
+func.func @trunci() {
+    // ------------------------------------------------
+    // Test truncating from i16
+    // ------------------------------------------------
+
+    // trunci on 20194 : i16
+    // trunci(20194 : i16) = -30 : i8
+    // CHECK-LABEL: @trunci_i16_i8
+    // CHECK-NEXT:  -30
+    %c20194 = arith.constant 20194 : i16
+    func.call @trunci_i16_i8(%c20194) : (i16) -> ()
+
+    // ------------------------------------------------
+    // TODO: Test truncation of i1, i8 etc..
+    // ------------------------------------------------
+
+    return
+}
+
+func.func @entry() {
+    func.call @extsi() : () -> ()
+    func.call @extui() : () -> ()
+    func.call @trunci() : () -> ()
+    return
+}

[kuhar]

LGTM % nit

[kuhar]

Why do we use -1 here but true in the test below? Should we make this consistent across the test cases?

[banach-space]

Given the discussion here:

• [mlir][arith] Add more canonicalization and integration tests coverage #92272,

and the resulting update to the MLIR guidelines on e2e tests:

• Add a paragraph on integration tests mlir-www#203,

I suggest that we park this for now.

@pingshiyu, thank you so much for your contribution and for bearing with us throughout this process. Apologies for not resolving the earlier discussion sooner—it would have saved you a lot of effort. I hope this doesn’t discourage you from contributing to MLIR in the future, and we truly appreciate your work!