Skip to content

Conversation

@IgWod
Copy link
Contributor

@IgWod IgWod commented Oct 18, 2025

When serializing SPIR-V MLIR containing externally linked function with debug enabled, the serialization crashes as printValueIDMap tries to print a key value that has been already destroyed. This happen as for externally linked function the body of the function is erased, that causes arguments to be destroyed as well, but the valueIDMap was never updated.

When serializing SPIR-V MLIR containing externally linked function
with debug enabled, the serialization crashes as `printValueIDMap`
tries to print a key value that has been already destroyed. This
happen as for externally linked function the body of the function
is erased, that causes arguments to be destroyed as well, but the
valueIDMap was never updated.
@github-actions
Copy link

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

@llvmbot
Copy link
Member

llvmbot commented Oct 18, 2025

@llvm/pr-subscribers-mlir-spirv

@llvm/pr-subscribers-mlir

Author: Igor Wodiany (IgWod)

Changes

When serializing SPIR-V MLIR containing externally linked function with debug enabled, the serialization crashes as printValueIDMap tries to print a key value that has been already destroyed. This happen as for externally linked function the body of the function is erased, that causes arguments to be destroyed as well, but the valueIDMap was never updated.


Full diff: https://github.com/llvm/llvm-project/pull/164098.diff

2 Files Affected:

  • (modified) mlir/lib/Target/SPIRV/Serialization/SerializeOps.cpp (+9)
  • (modified) mlir/test/Target/SPIRV/function-decorations.mlir (+2-1)
diff --git a/mlir/lib/Target/SPIRV/Serialization/SerializeOps.cpp b/mlir/lib/Target/SPIRV/Serialization/SerializeOps.cpp
index e9b180a70bb23..3c2c65bf64ed6 100644
--- a/mlir/lib/Target/SPIRV/Serialization/SerializeOps.cpp
+++ b/mlir/lib/Target/SPIRV/Serialization/SerializeOps.cpp
@@ -311,6 +311,15 @@ LogicalResult Serializer::processFuncOp(spirv::FuncOp op) {
     op.addEntryBlock();
     if (failed(processFuncParameter(op)))
       return failure();
+
+    // Erasing the body of the function destroys arguments, so we need to remove
+    // them from the map to avoid problems when processing invalid values used
+    // as keys. We have already serialized function arguments so we probably can
+    // remove them from the map as external function will not have any uses.
+    for (Value arg : op.getArguments())
+      if (valueIDMap.count(arg))
+        valueIDMap.erase(arg);
+
     // Don't need to process the added block, there is nothing to process,
     // the fake body was added just to get the arguments, remove the body,
     // since it's use is done.
diff --git a/mlir/test/Target/SPIRV/function-decorations.mlir b/mlir/test/Target/SPIRV/function-decorations.mlir
index ef627145aac3e..6098e42f063a2 100644
--- a/mlir/test/Target/SPIRV/function-decorations.mlir
+++ b/mlir/test/Target/SPIRV/function-decorations.mlir
@@ -1,4 +1,5 @@
-// RUN: mlir-translate -no-implicit-module -test-spirv-roundtrip -split-input-file -verify-diagnostics %s | FileCheck %s
+// RUN: mlir-translate --no-implicit-module --test-spirv-roundtrip --split-input-file %s | FileCheck %s
+// RUN: mlir-translate --no-implicit-module --test-spirv-roundtrip --split-input-file --debug %s | FileCheck %s
 
 spirv.module Logical GLSL450 requires #spirv.vce<v1.0, [Shader, Linkage], []> {
     spirv.func @linkage_attr_test_kernel()  "DontInline"  attributes {}  {

@IgWod
Copy link
Contributor Author

IgWod commented Oct 18, 2025

This is from my personal account as I felt like fixing it in my spare time :) (cc @IgWod-IMG)

@github-actions
Copy link

⚠️ We detected that you are using a GitHub private e-mail address to contribute to the repo.
Please turn off Keep my email addresses private setting in your account.
See LLVM Developer Policy and LLVM Discourse for more information.

@IgWod
Copy link
Contributor Author

IgWod commented Oct 18, 2025

I've addressed the feedback. Feel free to merge once it passes CI (I don't have permission on this account), otherwise I'll merge Monday morning from my work machine.

@IgWod-IMG IgWod-IMG merged commit 94647ee into llvm:main Oct 20, 2025
10 of 11 checks passed
@github-actions
Copy link

@IgWod Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

@IgWod IgWod deleted the serialization-crash branch October 20, 2025 12:49
googlewalt added a commit that referenced this pull request Oct 20, 2025
#164319)

This way, testing with --debug flag can correctly specify that it
requires assertions.

This is a fix for #164098
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants