Fix async queue teardown#1083
Merged
Merged
Conversation
AnastaZIuk
changed the title
Erfan-Ahmadi
marked this pull request as ready for review
July 13, 2026 09:05
Erfan-Ahmadi
added a commit
that referenced
this pull request
Jul 17, 2026
* copyMemoryToImage بعدز هئحمعئثدفثی - Add hostImageCopy as MOVE_TO_LIMIT in device_features.json. - Query and enable VkPhysicalDeviceHostImageCopyFeaturesEXT through limits.hostImageCopy. - Add EUF_HOST_TRANSFER_BIT and map it to VK_IMAGE_USAGE_HOST_TRANSFER_BIT_EXT. - Wire VK_FORMAT_FEATURE_2_HOST_IMAGE_TRANSFER_BIT_EXT into hostImageTransfer format usage. - Add host image copy flags and SMemoryToImageCopy. - Add ILogicalDevice::copyMemoryToImage validation and CVulkanLogicalDevice backend. - Validate host-transfer image creation and EHICF_MEMCPY_BIT full-subresource rules. * Add `transitionImageLayout` function * one small dublicated bug fix * Add Vulkan image copy and layout transition support * - Added support in `CVulkanPhysicalDevice` to query and process `hostImageCopyProperties`, including `identicalMemoryTypeRequirements`, `copySrcLayouts`, and `copyDstLayouts`. - Updated `ILogicalDevice` to validate source and destination image layouts for host image copy operations in `copyMemoryToImage`, `copyImageToMemory`, `copyImageToImage`, and `transitionImageLayout`. - Extended `device_limits.json` to include new fields for `hostImageCopySrcLayouts`, `hostImageCopyDstLayouts`, and `identicalMemoryTypeRequirements`. * Added `optimalTilingLayoutUUID` to `IPhysicalDevice` * fix PR comments from Erfan * getImageSubresourceLayout implemented * Update examples_tests submodule * Revert "Update examples_tests submodule" This reverts commit 11801ba. * Restore validateImageCreationAgainstDevice lost in master merge * Update examples_tests to branch merged with examples master * Revert "Update examples_tests to branch merged with examples master" This reverts commit d0dd4fb. * update examples to master (before final merge) * update fixed examples * Fix async queue teardown (#1083) * Add async queue teardown diagnostics * Fix async queue teardown * update examples --------- Co-authored-by: Erfan Ahmadi <ahmadierfan99@gmail.com> Co-authored-by: Arkadiusz Lachowicz <34793522+AnastaZIuk@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Fix async queue teardown by draining accepted work before request storage is destroyed.
Root cause
The lifecycle issue comes from
9414a480ffe, whereIThreadHandlerstarted stopping its worker thread from the base destructor. For CRTP queue users this is too late because derived resources and request storage are destroyed first.Later async queue changes in
de28767cc1,418db0d4c1, and87eeacb250kept the same lifetime model while request slots started carrying associated futures and cancelled requests relied on the worker for cleanup. That made the old teardown ordering visible as an intermittent Debug assert whenrequest_poolwas destroyed before all slots were recycled.This matches the random failure seen around #1079. It is a queue lifecycle bug, not a DXC compile failure.
Fix
Queue owners now call
shutdown()from their own destructors. Shutdown closes submissions, waits for activerequest()calls to finish publishing, drains pending and cancelled slots, and only then stops the worker thread.