Allow uint64-backed FreeableBuffers larger than size_t (#23364)

[cmt0]

Summary:

Follow-on to D120705447, which made DataLoader source offsets 64-bit but kept load sizes size_t, capping a named-data resource at 4 GiB on 32-bit targets. Addresses the review feedback there that weights over 4 GiB should not have to be split into multiple segments.

The size is now 64-bit along the named-data path. DataLoader::load_at_offset takes a uint64_t size, FlatTensorDataMap::get_data no longer rejects segments over SIZE_MAX, and FreeableBuffer stores a uint64_t size exposed via size_uint64(). size() aborts if the value doesn't fit, as data() already does for uint64-backed buffers. DualAddressDataLoader and TCEBackend are updated to match.

On a 32-bit core:

• 4 GiB weight via NamedDataMap::get_data: supported when the loader returns a uint64-backed buffer. The CPU only forwards the 64-bit address and size to a device, e.g. the TCE weight blob. CPU-pointer loaders still return NotSupported.

Review follow-up: separate LegacyUInt64Data callback storage and preserve implicit conversion of legacy callbacks, including captureless lambdas. The legacy callback alias and constructor remain deprecated. Added move/free and 32-bit bounds regression coverage. Embedded consumers retain C++17 support.

Review follow-up implemented with Codex.

Reviewed By: rascani

Differential Revision: D122398972

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/23364

• 📄 Preview Python docs built from this PR

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

⏳ 11 Pending, 2 Unrelated Failures

As of commit 7bb8e5a with merge base 8e4ec35 ():

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• pull / unittest-editable / linux / linux-job (gh) (similar failure)
  RuntimeError: Command bash /__w/_temp/exec_script failed with exit code 1

BROKEN TRUNK - The following job failed but was present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / unittest / linux / linux-job (gh) (trunk failure)
  RuntimeError: Command bash /__w/_temp/exec_script failed with exit code 1

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

[meta-codesync]

@cmt0 has exported this pull request. If you are a Meta employee, you can view the originating Diff in D122398972.

[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.

[rascani]

Review automatically exported from Phabricator review in Meta.