Skip to content

Keep delegate-consumed mutable buffers above the delegate - #21507

Open
Conarnar wants to merge 1 commit into
pytorch:mainfrom
Conarnar:export-D114288921
Open

Keep delegate-consumed mutable buffers above the delegate#21507
Conarnar wants to merge 1 commit into
pytorch:mainfrom
Conarnar:export-D114288921

Conversation

@Conarnar

Copy link
Copy Markdown
Contributor

Summary:
tag_constant_data freezes a const/param/buffer into a delegate when all of its
users are partitioned. It recognized a mutated buffer only when a direct user's
node name is a key in buffers_to_mutate. That misses the case where a buffer
both feeds a delegate and is mutated inside it: the mutation is produced by a
getitem off the call_delegate, so the buffer's only direct user is the
delegate node (whose name is not a buffers_to_mutate key), and the buffer is
wrongly frozen as constant data instead of staying a method-owned mutable buffer.

Additionally treat a placeholder as a mutated buffer when its FQN is a
buffers_to_mutate target, so a buffer a delegate updates in place stays owned
above the delegate. This is the case anticipated by
test_not_delegate_mutable_buffers ("consider when the delegate can consume the
mutable buffer") — e.g. a backend that embeds a pre-compiled engine which updates
a KV cache in place.

Authored with AI assistance (Claude Code).

Differential Revision: D114288921

Summary:
`tag_constant_data` freezes a const/param/buffer into a delegate when all of its
users are partitioned. It recognized a *mutated* buffer only when a direct user's
node name is a key in `buffers_to_mutate`. That misses the case where a buffer
both feeds a delegate and is mutated *inside* it: the mutation is produced by a
`getitem` off the `call_delegate`, so the buffer's only direct user is the
delegate node (whose name is not a `buffers_to_mutate` key), and the buffer is
wrongly frozen as constant data instead of staying a method-owned mutable buffer.

Additionally treat a placeholder as a mutated buffer when its FQN is a
`buffers_to_mutate` target, so a buffer a delegate updates in place stays owned
above the delegate. This is the case anticipated by
`test_not_delegate_mutable_buffers` ("consider when the delegate can consume the
mutable buffer") — e.g. a backend that embeds a pre-compiled engine which updates
a KV cache in place.

Authored with AI assistance (Claude Code).

Differential Revision: D114288921
Copilot AI review requested due to automatic review settings July 30, 2026 23:54
@pytorch-bot

pytorch-bot Bot commented Jul 30, 2026

Copy link
Copy Markdown

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21507

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

❌ 2 New Failures

As of commit a8bd74e with merge base 43cb2b2 (image):

NEW FAILURES - The following jobs have failed:

  • Cadence Build & Test / hifi-build / hifi4 (gh)
    ##[error]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)
    ##[error]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-cla meta-cla Bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jul 30, 2026

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

@meta-codesync

meta-codesync Bot commented Jul 30, 2026

Copy link
Copy Markdown
Contributor

@Conarnar has exported this pull request. If you are a Meta employee, you can view the originating Diff in D114288921.

@github-actions

Copy link
Copy Markdown

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.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. meta-exported

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants