From 0d6526f82c3cdefcca47f73f5fc08dc6f335eac6 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Mon, 21 Sep 2026 17:37:22 -0700 Subject: sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling bug alb_break_llc() decides whether to break LLC preference during active load balance. It does so by testing that every runnable fair task on the source rq prefers its LLC: env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable But the two counters cover different sets. nr_pref_llc_running is updated in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued, so it follows queued tasks. h_nr_runnable is updated in set_delayed()/ clear_delayed() and drops delay-dequeued tasks. So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks, alb_break_llc() returns false, and active balance is free to pull a task off its preferred LLC. Active balance only moves runnable tasks, and this is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB skips the per-task test in can_migrate_task(). The runnable set is the one we want. Fix it on the counter side. A task should be counted in nr_pref_llc_running exactly while it is both queued on its preferred LLC (pref_llc_queued) and runnable (!sched_delayed). Define that membership once in task_pref_llc_runnable(), and adjust the counter only through pref_llc_running_inc()/pref_llc_running_dec() from the four sites that change either input: account_llc_enqueue(), account_llc_dequeue(), set_delayed() and clear_delayed(). Gating every update on the same predicate keeps the delay, wake and dequeue paths from double-counting or underflowing; see the comments at those sites for the ordering. nr_llc_running and sd->llc_counts are not touched and stay on queued semantics. Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC") Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/ Reported-by: Zhan Xusheng Suggested-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Kayra Cizmeci Cc: # v7.2.x Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com --- Documentation/userspace-api/liveupdate.rst | 51 ++++++++++++++++++++++++++++++ 1 file changed, 51 insertions(+) create mode 100644 Documentation/userspace-api/liveupdate.rst (limited to 'Documentation/userspace-api/liveupdate.rst') diff --git a/Documentation/userspace-api/liveupdate.rst b/Documentation/userspace-api/liveupdate.rst new file mode 100644 index 000000000..cdd74627a --- /dev/null +++ b/Documentation/userspace-api/liveupdate.rst @@ -0,0 +1,51 @@ +.. SPDX-License-Identifier: GPL-2.0 + +================ +Live Update uAPI +================ +:Author: Pasha Tatashin + +ioctl interface +=============== +.. kernel-doc:: kernel/liveupdate/luo_core.c + :doc: LUO ioctl Interface + +ioctl uAPI +=========== +.. kernel-doc:: include/uapi/linux/liveupdate.h + +Note on Compatibility +===================== + +Note that the Live Update feature is still under development and subject to +change, so compatibility is not guaranteed when kexec rebooting between +different kernel versions. This is expected to change and stabilize in a future +version. + +Userspace Integration +===================== + +systemd (since version v261) uses LUO to preserve its per-service file +descriptor store across a kexec-based live update. Services opt in by setting +``FileDescriptorStoreMax=`` and ``FileDescriptorStorePreserve=`` in their unit, +and push file descriptors with a name into the store via +``sd_pid_notify_with_fds(... "FDSTORE=1\nFDNAME=foo")``. + +Services may also create their own LUO sessions (via ``/dev/liveupdate``) and +push the resulting session fds into their file descriptor store like any other +fd. systemd detects such session fds and handles them accordingly, and +hands the re-retrieved session fd back to the service after kexec, using the +existing file descriptor store service interface. + +For details, see: + +- `File Descriptor Store `_ +- `systemd.service(5) FileDescriptorStorePreserve= + `_ +- `sd_pid_notify_with_fds(3) + `_ + +See Also +======== + +- :doc:`Live Update Orchestrator ` -- cgit v1.3.1