[ET-VK][ez] Fix use-after-free bug in Vulkan queue creation

[SS-JIA]

Summary:

A use-after-free bug in find_compute_queues() was discovered using Valgrind. The function created a local queue_priorities vector inside a loop and stored its data pointer in VkDeviceQueueCreateInfo. When the vector went out of scope, its memory was freed, but vkCreateDevice() later accessed this freed memory. Fixed by adding a queue_priorities parameter to persist the data until after vkCreateDevice() completes.

Problem

A use-after-free bug was discovered in find_compute_queues() using Valgrind.
The function created a local std::vector queue_priorities inside a
loop and stored its data pointer in VkDeviceQueueCreateInfo. When the vector
went out of scope at the end of each iteration, its memory was freed. Later,
when vkCreateDevice() accessed these queue priorities, it read from freed
memory.

Investigation

Valgrind reported:

• Invalid read of size 4 at 0xb3bdd60 (freed memory)
• Block was freed by operator delete in find_compute_queues()
• Block was allocated by operator new in find_compute_queues()
• Error occurred during vkCreateDevice() call

Fix

Modified find_compute_queues() to accept an additional parameter
std::vector<std::vector>& queue_priorities that persists the
queue priority data until after vkCreateDevice() completes. This ensures
the memory remains valid when Vulkan needs to access it.

Updated all call sites:

• create_logical_device()
• Adapter constructor (external device variant)

Verification

Valgrind results before fix: 296 errors from 13 contexts, 1 Invalid read
Valgrind results after fix: 295 errors from 12 contexts, 0 Invalid reads ✓

Remaining errors are in NVIDIA drivers and third-party libraries.

cc @manuelcandales @digantdesai @cbilgin

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16367

• 📄 Preview Python docs built from this PR

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

❌ 1 Cancelled Job, 1 Unrelated Failure

As of commit 128b9e8 with merge base 0ee2f49 ():

CANCELLED JOB - The following job was cancelled. Please retry:

• pull / test-samsung-models-linux / linux-job (gh)
  ##[error]The operation was canceled.

UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:

• pull / android / run-emulator (gh) (#16137)
  Timeout waiting for emulator to boot.

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

[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.

[meta-codesync]

@SS-JIA has imported this pull request. If you are a Meta employee, you can view this in D89687612.