refactor(irq-tuning): improve robustness and fix some corner cases#1010
Merged
xu282934741 merged 5 commits intokubewharf:mainfrom Nov 4, 2025
Merged
refactor(irq-tuning): improve robustness and fix some corner cases#1010xu282934741 merged 5 commits intokubewharf:mainfrom
xu282934741 merged 5 commits intokubewharf:mainfrom
Conversation
…ue and irq chore(irq-tuning): improve universality of getting mapping of nic queue and irq Signed-off-by: 张浩宇 <zhanghaoyu.zhy@bytedance.com>
…ent number of leading "0" fix(irq-tuning): fix ComparesHexBitmapStrings two strings with different number of leading "0" character. Signed-off-by: 张浩宇 <zhanghaoyu.zhy@bytedance.com>
…after irq affinity write After setting the kernel's IRQ affinity, the kernel cannot guarantee that an immediate subsequent read will return the latest value; in some cases, it may return a stale value. This is because if irq_do_set_affinity returns EBUSY, the kernel may invoke irqd_set_move_pending to set the IRQD_SETAFFINITY_PENDING flag. Subsequently, the next IRQ affinity write operation will store the newly affinitized CPU mask in irq_desc->pending_mask. At a later point, irq_move_masked_irq will copy irq_desc->pending_mask to irq_desc->desc->irq_common_data.affinity and then clear the IRQD_SETAFFINITY_PENDING flag within the interrupt context. The IRQ affinity read function show_irq_affinity first reads irq_desc->desc->irq_common_data.affinity and then checks whether the IRQD_SETAFFINITY_PENDING flag is set. If the flag is set, it retrieves the value from irq_desc->pending_mask. If the QRM's IRQ affinity read syscall has already read the stale irq_desc->desc->irq_common_data.affinity but not yet read irq_desc->pending_mask, and the interrupt context's irq_move_masked_irq clears the IRQD_SETAFFINITY_PENDING flag at this point, QRM will end up reading a stale value. Even if the SetIrqAffinity function can obtain the latest affinity value before returning, it still cannot ensure that this latest value originates from irq_desc->desc->irq_common_data.affinity. If the value is instead fetched from irq_desc->pending_mask, a subsequent nic.sync may retrieve a stale value when a race condition occurs between the interrupt context's irqd_set_move_pending and QRM's IRQ affinity read operations. Therefore, we choose to trust the kernel: if SetIrqAffinity returns no error, the IRQ affinity is considered to have been set correctly, and there is no need for a subsequent sync from the kernel to verify its correctness. Instead, IRQ affinity will be synced from the kernel during periodic syncNics operations, with the values read from the kernel compared against those managed by QRM. Signed-off-by: 张浩宇 <zhanghaoyu.zhy@bytedance.com>
chore(irq-tuning): ListNetNS ignores non-existed netns Signed-off-by: 张浩宇 <zhanghaoyu.zhy@bytedance.com>
syncNics only concerns NICs that can be shared among multiple containers. Therefore, when listing active uplink NICs (via ListActiveUplinkNics), container NICs (specifically SRIOV container NICs) should be filtered out. Signed-off-by: 张浩宇 <zhanghaoyu.zhy@bytedance.com>
zhanghaoyu1986
added
the
workflow/need-review
JulyWindK
approved these changes
Nov 3, 2025
Codecov Report❌ Patch coverage is ❌ Your patch status has failed because the patch coverage (26.34%) is below the target coverage (50.00%). You can increase the patch coverage or adjust the target coverage. Additional details and impacted files@@ Coverage Diff @@
## main #1010 +/- ##
==========================================
+ Coverage 58.68% 59.65% +0.97%
==========================================
Files 679 682 +3
Lines 77274 63929 -13345
==========================================
- Hits 45347 38137 -7210
+ Misses 27540 21385 -6155
- Partials 4387 4407 +20
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
xu282934741
reviewed
Nov 4, 2025
zhanghaoyu1986
changed the title
xu282934741
approved these changes
Nov 4, 2025
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
3 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.
What type of PR is this?
Bug fixes/Enhancements
What this PR does / why we need it:
fix some corner cases and improve the robustness and versatility