[Mlir-commits] [mlir] [mlir][bufferization] Fix to_buffer being incorrectly hoisted by LICM (PR #188997)
llvmlistbot at llvm.org
llvmlistbot at llvm.org
Fri Mar 27 06:55:09 PDT 2026
github-actions[bot] wrote:
<!--PREMERGE ADVISOR COMMENT: Linux-->
# :penguin: Linux x64 Test Results
* 7731 tests passed
* 604 tests skipped
* 1 test failed
## Failed Tests
(click on a test name to see its output)
### MLIR
<details>
<summary>MLIR.Dialect/Tensor/bufferize.mlir</summary>
```
Exit Code: 1
Command Output (stdout):
--
# RUN: at line 1
/home/gha/actions-runner/_work/llvm-project/llvm-project/build/bin/mlir-opt /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir --one-shot-bufferize="dialect-filter=tensor,bufferization copy-before-write unknown-type-conversion=identity-layout-map" -cse -split-input-file | /home/gha/actions-runner/_work/llvm-project/llvm-project/build/bin/FileCheck /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir
# executed command: /home/gha/actions-runner/_work/llvm-project/llvm-project/build/bin/mlir-opt /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir '--one-shot-bufferize=dialect-filter=tensor,bufferization copy-before-write unknown-type-conversion=identity-layout-map' -cse -split-input-file
# note: command had no output on stdout or stderr
# executed command: /home/gha/actions-runner/_work/llvm-project/llvm-project/build/bin/FileCheck /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir
# .---command stderr------------
# | /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir:572:16: error: CHECK-DAG: expected string not found in input
# | // CHECK-DAG: %[[dim0:.*]] = memref.dim %[[m1]], %[[c0]]
# | ^
# | <stdin>:428:91: note: scanning from here
# | func.func @tensor.pad(%arg0: tensor<?x10xindex>, %arg1: index, %arg2: index, %arg3: index) -> tensor<?x?xindex> {
# | ^
# | <stdin>:428:91: note: with "m1" equal to "0"
# | func.func @tensor.pad(%arg0: tensor<?x10xindex>, %arg1: index, %arg2: index, %arg3: index) -> tensor<?x?xindex> {
# | ^
# | <stdin>:428:91: note: with "c0" equal to "c0"
# | func.func @tensor.pad(%arg0: tensor<?x10xindex>, %arg1: index, %arg2: index, %arg3: index) -> tensor<?x?xindex> {
# | ^
# | <stdin>:435:2: note: possible intended match here
# | %dim = memref.dim %4, %c0 : memref<?x10xindex>
# | ^
# | /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir:639:11: error: CHECK: expected string not found in input
# | // CHECK: %[[SUBVIEW2:.*]] = memref.subview %[[ALLOC]][8] [8] [1]
# | ^
# | <stdin>:495:26: note: scanning from here
# | memref.copy %0, %subview_0 : memref<8xf32> to memref<8xf32, strided<[1], offset: 8>>
# | ^
# | <stdin>:495:26: note: with "ALLOC" equal to "alloc"
# | memref.copy %0, %subview_0 : memref<8xf32> to memref<8xf32, strided<[1], offset: 8>>
# | ^
# | <stdin>:495:38: note: possible intended match here
# | memref.copy %0, %subview_0 : memref<8xf32> to memref<8xf32, strided<[1], offset: 8>>
# | ^
# | /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir:681:11: error: CHECK: expected string not found in input
# | // CHECK: %[[SUBVIEW1:.*]] = memref.subview %[[ALLOC]][0, 0] [8, %[[F_DIM]]] [1, 1]
# | ^
# | <stdin>:528:68: note: scanning from here
# | %alloc = memref.alloc(%4) {alignment = 64 : i64} : memref<8x?xf32>
# | ^
# | <stdin>:528:68: note: with "ALLOC" equal to "alloc"
# | %alloc = memref.alloc(%4) {alignment = 64 : i64} : memref<8x?xf32>
# | ^
# | <stdin>:528:68: note: with "F_DIM" equal to "dim"
# | %alloc = memref.alloc(%4) {alignment = 64 : i64} : memref<8x?xf32>
# | ^
# | <stdin>:530:8: note: possible intended match here
# | %subview = memref.subview %alloc[0, 0] [8, %dim_1] [1, 1] : memref<8x?xf32> to memref<8x?xf32, strided<[?, 1]>>
# | ^
# | /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir:707:11: error: CHECK: expected string not found in input
# | // CHECK: %[[SUBVIEW1:.*]] = memref.subview %[[ALLOC]][0, 0] [%[[NON_CONCAT_DIM]], %[[F_DIM]]] [1, 1]
# | ^
# | <stdin>:556:33: note: scanning from here
# | %dim_2 = memref.dim %alloc, %c0 : memref<?x?xf32>
# | ^
# | <stdin>:556:33: note: with "ALLOC" equal to "alloc"
# | %dim_2 = memref.dim %alloc, %c0 : memref<?x?xf32>
# | ^
# | <stdin>:556:33: note: with "NON_CONCAT_DIM" equal to "dim_2"
# | %dim_2 = memref.dim %alloc, %c0 : memref<?x?xf32>
# | ^
# | <stdin>:556:33: note: with "F_DIM" equal to "dim_0"
# | %dim_2 = memref.dim %alloc, %c0 : memref<?x?xf32>
# | ^
# | <stdin>:558:6: note: possible intended match here
# | %subview = memref.subview %alloc[0, 0] [%dim_2, %dim_3] [1, 1] : memref<?x?xf32> to memref<?x?xf32, strided<[?, 1]>>
# | ^
# | /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir:737:44: error: undefined variable: $sum_map
# | // CHECK: %[[OFFSET:.*]] = affine.apply #[[$sum_map]]()[%[[F_DIM]], %[[G_DIM]]]
# | ^
# | <stdin>:582:28: note: with "F_DIM" equal to "dim"
# | memref.copy %1, %subview_1 : memref<8x?xf32> to memref<8x?xf32, strided<[10, 1], offset: ?>>
# | ^
# | <stdin>:582:28: note: with "G_DIM" equal to "dim_0"
# | memref.copy %1, %subview_1 : memref<8x?xf32> to memref<8x?xf32, strided<[10, 1], offset: ?>>
# | ^
# | <stdin>:583:2: note: possible intended match here
# | %3 = affine.apply #map()[%dim, %dim_0]
# | ^
# |
# | Input file: <stdin>
# | Check file: /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/Tensor/bufferize.mlir
# |
# | -dump-input=help explains the following input dump.
# |
# | Input was:
# | <<<<<<
# | .
# | .
# | .
# | 423:
# | 424: // -----
# | 425: #map = affine_map<()[s0, s1] -> (s0 + s1 + 5)>
# | 426: #map1 = affine_map<()[s0, s1] -> (s0 + s1 + 10)>
# | 427: module {
# | 428: func.func @tensor.pad(%arg0: tensor<?x10xindex>, %arg1: index, %arg2: index, %arg3: index) -> tensor<?x?xindex> {
# | dag:572'0 X~~~~~~~~~~~~~~~~~~~~~~~~ error: no match found
# | dag:572'1 with "m1" equal to "0"
# | dag:572'2 with "c0" equal to "c0"
# | 429: %0 = bufferization.to_buffer %arg0 : tensor<?x10xindex> to memref<?x10xindex>
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 430: %1 = bufferization.to_buffer %arg0 : tensor<?x10xindex> to memref<?x10xindex>
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 431: %2 = bufferization.to_buffer %arg0 : tensor<?x10xindex> to memref<?x10xindex>
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 432: %3 = bufferization.to_buffer %arg0 : tensor<?x10xindex> to memref<?x10xindex>
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 433: %4 = bufferization.to_buffer %arg0 : tensor<?x10xindex> to memref<?x10xindex>
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 434: %c0 = arith.constant 0 : index
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 435: %dim = memref.dim %4, %c0 : memref<?x10xindex>
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | dag:572'3 ? possible intended match
# | 436: %c5 = arith.constant 5 : index
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 437: %c1 = arith.constant 1 : index
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 438: %dim_0 = memref.dim %3, %c1 : memref<?x10xindex>
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 439: %dim_1 = memref.dim %2, %c0 : memref<?x10xindex>
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 440: %5 = affine.apply #map()[%dim_1, %arg2]
# | dag:572'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | .
# | .
# | .
# | 490: %1 = bufferization.to_buffer %arg0 : tensor<8xf32> to memref<8xf32>
# | 491: %alloc = memref.alloc() {alignment = 64 : i64} : memref<16xf32>
# | 492: %subview = memref.subview %alloc[0] [8] [1] : memref<16xf32> to memref<8xf32, strided<[1]>>
# | 493: memref.copy %1, %subview : memref<8xf32> to memref<8xf32, strided<[1]>>
# | 494: %subview_0 = memref.subview %alloc[8] [8] [1] : memref<16xf32> to memref<8xf32, strided<[1], offset: 8>>
# | 495: memref.copy %0, %subview_0 : memref<8xf32> to memref<8xf32, strided<[1], offset: 8>>
# | check:639'0 X~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ error: no match found
# | check:639'1 with "ALLOC" equal to "alloc"
# | check:639'2 ? possible intended match
# | 496: %2 = bufferization.to_tensor %alloc : memref<16xf32> to tensor<16xf32>
# | check:639'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 497: return %2 : tensor<16xf32>
# | check:639'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 498: }
# | check:639'0 ~~~
# | 499: }
# | check:639'0 ~~
# | 500:
# | check:639'0 ~
# | .
# | .
# | .
# | 523: %3 = bufferization.to_buffer %arg0 : tensor<8x?xf32> to memref<8x?xf32>
# | 524: %c1 = arith.constant 1 : index
# | 525: %dim = memref.dim %1, %c1 : memref<8x?xf32>
# | 526: %dim_0 = memref.dim %0, %c1 : memref<8x?xf32>
# | 527: %4 = affine.apply #map()[%dim, %dim_0]
# | 528: %alloc = memref.alloc(%4) {alignment = 64 : i64} : memref<8x?xf32>
# | check:681'0 X error: no match found
# | check:681'1 with "ALLOC" equal to "alloc"
# | check:681'2 with "F_DIM" equal to "dim"
# | 529: %dim_1 = memref.dim %3, %c1 : memref<8x?xf32>
# | check:681'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 530: %subview = memref.subview %alloc[0, 0] [8, %dim_1] [1, 1] : memref<8x?xf32> to memref<8x?xf32, strided<[?, 1]>>
# | check:681'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | check:681'3 ? possible intended match
# | 531: memref.copy %3, %subview : memref<8x?xf32> to memref<8x?xf32, strided<[?, 1]>>
# | check:681'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 532: %dim_2 = memref.dim %2, %c1 : memref<8x?xf32>
# | check:681'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 533: %subview_3 = memref.subview %alloc[0, %dim_1] [8, %dim_2] [1, 1] : memref<8x?xf32> to memref<8x?xf32, strided<[?, 1], offset: ?>>
# | check:681'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 534: memref.copy %2, %subview_3 : memref<8x?xf32> to memref<8x?xf32, strided<[?, 1], offset: ?>>
# | check:681'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 535: %5 = bufferization.to_tensor %alloc : memref<8x?xf32> to tensor<8x?xf32>
# | check:681'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | .
# | .
# | .
# | 551: %c1 = arith.constant 1 : index
# | 552: %dim_0 = memref.dim %1, %c1 : memref<?x?xf32>
# | 553: %dim_1 = memref.dim %0, %c1 : memref<?x?xf32>
# | 554: %5 = affine.apply #map()[%dim_0, %dim_1]
# | 555: %alloc = memref.alloc(%dim, %5) {alignment = 64 : i64} : memref<?x?xf32>
# | 556: %dim_2 = memref.dim %alloc, %c0 : memref<?x?xf32>
# | check:707'0 X~~~~~~~~~~~~~~~~~~ error: no match found
# | check:707'1 with "ALLOC" equal to "alloc"
# | check:707'2 with "NON_CONCAT_DIM" equal to "dim_2"
# | check:707'3 with "F_DIM" equal to "dim_0"
# | 557: %dim_3 = memref.dim %4, %c1 : memref<?x?xf32>
# | check:707'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 558: %subview = memref.subview %alloc[0, 0] [%dim_2, %dim_3] [1, 1] : memref<?x?xf32> to memref<?x?xf32, strided<[?, 1]>>
# | check:707'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | check:707'4 ? possible intended match
# | 559: memref.copy %4, %subview : memref<?x?xf32> to memref<?x?xf32, strided<[?, 1]>>
# | check:707'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 560: %dim_4 = memref.dim %3, %c1 : memref<?x?xf32>
# | check:707'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 561: %subview_5 = memref.subview %alloc[0, %dim_3] [%dim_2, %dim_4] [1, 1] : memref<?x?xf32> to memref<?x?xf32, strided<[?, 1], offset: ?>>
# | check:707'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 562: memref.copy %3, %subview_5 : memref<?x?xf32> to memref<?x?xf32, strided<[?, 1], offset: ?>>
# | check:707'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 563: %6 = bufferization.to_tensor %alloc : memref<?x?xf32> to tensor<?x?xf32>
# | check:707'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | .
# | .
# | .
# | 577: %dim = memref.dim %2, %c1 : memref<8x?xf32>
# | 578: %subview = memref.subview %alloc[0, 0] [8, %dim] [1, 1] : memref<8x10xf32> to memref<8x?xf32, strided<[10, 1]>>
# | 579: memref.copy %2, %subview : memref<8x?xf32> to memref<8x?xf32, strided<[10, 1]>>
# | 580: %dim_0 = memref.dim %1, %c1 : memref<8x?xf32>
# | 581: %subview_1 = memref.subview %alloc[0, %dim] [8, %dim_0] [1, 1] : memref<8x10xf32> to memref<8x?xf32, strided<[10, 1], offset: ?>>
# | 582: memref.copy %1, %subview_1 : memref<8x?xf32> to memref<8x?xf32, strided<[10, 1], offset: ?>>
# | check:737'0 X~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ error: match failed for invalid pattern
# | check:737'1 undefined variable: $sum_map
# | check:737'2 with "F_DIM" equal to "dim"
# | check:737'3 with "G_DIM" equal to "dim_0"
# | 583: %3 = affine.apply #map()[%dim, %dim_0]
# | check:737'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | check:737'4 ? possible intended match
# | 584: %subview_2 = memref.subview %alloc[0, %3] [8, 2] [1, 1] : memref<8x10xf32> to memref<8x2xf32, strided<[10, 1], offset: ?>>
# | check:737'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 585: memref.copy %0, %subview_2 : memref<8x2xf32> to memref<8x2xf32, strided<[10, 1], offset: ?>>
# | check:737'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 586: %4 = bufferization.to_tensor %alloc : memref<8x10xf32> to tensor<8x10xf32>
# | check:737'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 587: return %4 : tensor<8x10xf32>
# | check:737'0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | 588: }
# | check:737'0 ~~~
# | .
# | .
# | .
# | >>>>>>
# `-----------------------------
# error: command failed with exit status: 1
--
```
</details>
If these failures are unrelated to your changes (for example tests are broken or flaky at HEAD), please open an issue at https://github.com/llvm/llvm-project/issues and add the `infrastructure` label.
https://github.com/llvm/llvm-project/pull/188997
More information about the Mlir-commits
mailing list