Repository navigation
sched: honor CPU binding of per-CPU kernel threads - #111
Merged
FengHeting merged 3 commits intoOct 22, 2026
Merged
Conversation
kthread_create_on_cpu() and kthread_run_on_cpu() promise a thread bound to one CPU, but create_kthread() only seeded task->cpu_id. Nothing in the scheduler remembered the request, so the first wakeup could place the thread elsewhere through select_wakeup_cpu_locked(), and newidle or periodic balancing could pull it to any CPU. Record the binding in task->cpu_pinned and honour it in the two places that pick a different CPU for an existing task: wakeup placement keeps a pinned task on its CPU, and the steal-candidate scan skips it, which covers both newidle and periodic balancing. Reproduce with two kbench same-CPU workers requested on CPU 1. Before this fix, wake placement and stealing can send either worker elsewhere; afterwards same-CPU rows consistently report [1>1,1>1]. The pin.log and heap A/B logs retain the CPU and migration evidence. This is a binding correctness fix, not a claim that cross-CPU ping-pong is faster. Validation: standalone origin/develop branch, GCC LTO build with virtio-gpu disabled and -Wl,-u,strcpy for the unrelated existing link defect; 5-vCPU runtime suites with the separately documented scheduler/kbench overlays. No source-runqueue minimum-size or idle-polling policy is introduced. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Validate the target CPU before queueing a pinned kthread creation request, so an invalid CPU is rejected instead of silently binding the thread to CPU 0. Honor the pin in wake placement even before sched_start(), where the BSP-only fallback would otherwise move a pinned thread to CPU 0 and rewrite its cpu_id permanently. Skip pinned entities in the steal scan without spending the movable candidate budget, and refuse explicit migration of a pinned task in task_set_cpu(), mirroring PF_NO_SETAFFINITY. Signed-off-by: MicroFish <2609948707@qq.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
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.
kthread_create_on_cpu()/kthread_run_on_cpu()只初始化 cpu_id，唤醒放置和负载均衡仍会迁移任务，实际不履行绑定 API 的承诺。新增 cpu_pinned，唤醒保持目标 CPU，newidle/periodic stealing 的候选扫描跳过它。复现：将两个 kbench worker 都请求放在 CPU 1，比较 same-cpu 行的实际 CPU 和任务迁移次数。现有
fix2-1.log中，wq same-cpu 一轮为[1>0m61214,1>4m55566]；修后pin.log及本轮 30 轮输出均为[1>1,1>1]。这改变基准实际执行位置，修前修后的耗时不能当成同一放置条件下的性能 A/B。因此本 PR 以正确性和迁移字段验证，不声称跨核延迟降低。
验证环境：5 vCPU、KVM、2 GiB、每次 QEMU 超时 60 秒。所有独立分支直接从
origin/develop@153578b5开出，目标均为develop；每个分支均独立完成 GCC 16.2 / LTO 构建，构建源只取git ls-files '*.c'，切分支后清除对应.o/.d。按已知基线问题临时关闭CONFIG_VIRTIO_GPU。除 strcpy PR 自身外，独立构建在链接命令临时加-Wl,-u,strcpy，绕过原始 develop 的已复现 LTO 缺陷；没有把其他 PR 的源码带进分支。运行基准时明确使用独立 sched 修复、strcpy 修复、kbench 和 kthread 绑定的共同测试覆盖层，以免把已知调度崩溃或实际 CPU 不符误算成待测修改的效果。覆盖层仅用于测试镜像，不构成提交依赖。
已否决方案：把偷取限制为源 rq 至少两个任务，曾使 ping-pong 约 3 µs 恶化到 7.8 µs；轮询 idle 无明显收益。这些改动均未采用。宿主负载高，仅把交替多次中位数中超过 20% 的差异视为性能结论。
🤖 Generated with ClaudeCode