Un-skip sum dim=None U55/U85 tests; fix stale addmm xfail reason

[psiddh]

Summary:
Two test-hygiene fixes in the Arm backend test suite, both reducing the
ai_infra_mobile_platform SKIPPING oncall bucket for executorch/backends/arm.

• test_sum.py: test_sum_u55_INT_1_0 / test_sum_u85_INT_1_0 were skipping
  the dim_None and dim_None_4d_tensor cases because the FVP bundled program
  cannot serialize None as a model input. dim/keepdim were passed as
  forward arguments, so they became model inputs. A new SumDimNone module
  bakes dim=None into the module, leaving only the tensor as an input, so the
  cases now run and pass on both U55 and U85 instead of being skipped. The
  non-None cases are unchanged. (The TOSA path already ran these un-skipped.)

• test_addmm.py: the test_addmm_16a8w_u55_INT xfail reason claimed "Vela
  compilation fails with 'Invalid arguments'", but the test actually fails at
  runtime on the Corstone-300 U55 FVP (method load status 0x14); Vela
  compilation succeeds. Corrected the reason string to describe the observed
  failure so the next investigator is not misled. The test still xfails.

Authored with Claude Code.

Differential Revision: D113676338

cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21407

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures

As of commit 5be00a1 with merge base 55d693b ():

NEW FAILURES - The following jobs have failed:

• Cadence Build & Test / hifi-build / hifi4 (gh)
  Refusing to check out fork pull request code from a 'pull_request_target' workflow. This workflow runs with the base repository's GITHUB_TOKEN, secrets, default-branch cache scope, and runner access. Fetching and executing a fork's code in that trusted context commonly leads to "pwn request" vulnerabilities. To opt in, review the risks at https://gh.io/securely-using-pull_request_target and set 'allow-unsafe-pr-checkout: true' on the actions/checkout step.
• Cadence Build & Test / vision-build / vision (gh)
  Refusing to check out fork pull request code from a 'pull_request_target' workflow. This workflow runs with the base repository's GITHUB_TOKEN, secrets, default-branch cache scope, and runner access. Fetching and executing a fork's code in that trusted context commonly leads to "pwn request" vulnerabilities. To opt in, review the risks at https://gh.io/securely-using-pull_request_target and set 'allow-unsafe-pr-checkout: true' on the actions/checkout step.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[meta-codesync]

@psiddh has exported this pull request. If you are a Meta employee, you can view the originating Diff in D113676338.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[Copilot]

Pull request overview

This PR improves Arm backend test hygiene by (1) enabling previously skipped Ethos-U55/U85 sum(dim=None) test cases in the FVP path by baking dim=None into the module, and (2) correcting a stale/incorrect xfail reason for an addmm int16 test so failures are attributed to the right stage (runtime load vs. Vela compile).

Changes:

• Add a SumDimNone module and route dim=None test vectors through it so the FVP bundled program doesn’t need to serialize None as an input.
• Remove skip logic for dim=None cases in U55/U85 sum INT tests by switching to module+inputs selection.
• Update test_addmm_16a8w_u55_INT xfail reason to reflect the observed U55 FVP load failure (status 0x14).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
backends/arm/test/ops/test_sum.py	Introduces SumDimNone + helper to un-skip dim=None U55/U85 cases in the FVP pipeline.
backends/arm/test/ops/test_addmm.py	Updates the xfail reason string to match the actual failing stage (FVP load).

Comments suppressed due to low confidence (2)

backends/arm/test/ops/test_sum.py:124

• test_data is a callable parameter (provided by common.parametrize) and is used as such via _module_and_inputs(test_data). The current Tuple annotation is inaccurate and inconsistent with other tests in this file.

def test_sum_u55_INT_1_0(test_data: Tuple):

backends/arm/test/ops/test_sum.py:137

• Same as above: test_data is a callable from common.parametrize, but is annotated as Tuple. Align the signature with the rest of the file for clarity and static checking.

def test_sum_u85_INT_1_0(test_data: Tuple):

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.