Extend ExecuTorch reinplace pass (#19482)

[metascroy]

Summary:

This diff expands the ExecuTorch reinplace pass to be configurable for backends to use.

We can introduce more in-place portable ops to expand the default set in ExecuTorch over time as well.

Note: if two tensors have the same spec, they are now deduped and emitted as the same value. Previously (with copy_ writeback), the synthetic out got a new value, but with the same allocation info as the self value.

What's new

• reinplace_pass now takes ops_to_inplace= and inplace_overrides= so backends can extend the candidate set. The in-place form is auto-derived from the op schema, and registrations are validated up front.
• Mutated arg positions are derived from Tensor(a!) annotations, so ops with multiple mutated args are handled correctly. kwargs are also forwarded to the in-place op (e.g. accumulate= for index_put).
• Fixed a bug where, after a successful rewrite, the new node's input nodes were not added to seen_nodes, so an upstream candidate mutating one of those reads could incorrectly look safe to reinplace. Covered by a new index_put-only regression test.
• Memory planner now aliases an in-place op's result TensorSpec onto its mutated input's spec (schema-driven via output_to_aliased_input_map), so they share an allocation slot.
• Emitter dedupes by TensorSpec identity so two FX nodes sharing one spec share one Value in the lowered IR.

Reviewed By: JacobSzwejbka

Differential Revision: D104152427

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/19482

• 📄 Preview Python docs built from this PR

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

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• Run pull request jobs on OSDC runners in shadow mode

❌ 1 New Failure, 1 Unclassified Failure

As of commit 27995cb with merge base 65c7ee2 ():

NEW FAILURE - The following job has failed:

• pull / unittest-editable / macos / macos-job (gh)
  export/tests/test_target_recipes.py::TestTargetRecipes::test_mv2_model

UNCLASSIFIED FAILURE - DrCI could not classify the following job because the workflow did not run on the merge base. The failure may be pre-existing on trunk or introduced by this PR:

• Check Labels / Check labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

[meta-codesync]

@metascroy has exported this pull request. If you are a Meta employee, you can view the originating Diff in D104152427.

[JacobSzwejbka]

Review automatically exported from Phabricator review in Meta.

[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.