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/admin-guide/index.rst | 192 ++++++++++++++++++++++++++++++++++++ 1 file changed, 192 insertions(+) create mode 100644 Documentation/admin-guide/index.rst (limited to 'Documentation/admin-guide/index.rst') diff --git a/Documentation/admin-guide/index.rst b/Documentation/admin-guide/index.rst new file mode 100644 index 000000000..cd28dfe91 --- /dev/null +++ b/Documentation/admin-guide/index.rst @@ -0,0 +1,192 @@ +================================================= +The Linux kernel user's and administrator's guide +================================================= + +The following is a collection of user-oriented documents that have been +added to the kernel over time. There is, as yet, little overall order or +organization here — this material was not written to be a single, coherent +document! With luck things will improve quickly over time. + +General guides to kernel administration +--------------------------------------- + +This initial section contains overall information, including the README +file describing the kernel as a whole, documentation on kernel parameters, +etc. + +.. toctree:: + :maxdepth: 1 + + README + devices + + features + +A big part of the kernel's administrative interface is the /proc and sysfs +virtual filesystems; these documents describe how to interact with tem + +.. toctree:: + :maxdepth: 1 + + sysfs-rules + sysctl/index + cputopology + abi + +Security-related documentation: + +.. toctree:: + :maxdepth: 1 + + hw-vuln/index + LSM/index + perf-security + +Booting the kernel +------------------ + +.. toctree:: + :maxdepth: 1 + + bootconfig + kernel-parameters + efi-stub + initrd + + +Tracking down and identifying problems +-------------------------------------- + +Here is a set of documents aimed at users who are trying to track down +problems and bugs in particular. + +.. toctree:: + :maxdepth: 1 + + reporting-issues + reporting-regressions + quickly-build-trimmed-linux + verify-bugs-and-bisect-regressions + bug-hunting + bug-bisect + tainted-kernels + ramoops + dynamic-debug-howto + init + kdump/index + perf/index + pstore-blk + clearing-warn-once + kernel-per-CPU-kthreads + lockup-watchdogs + RAS/index + sysrq + + +Core-kernel subsystems +---------------------- + +These documents describe core-kernel administration interfaces that are +likely to be of interest on almost any system. + +.. toctree:: + :maxdepth: 1 + + cgroup-v2 + cgroup-v1/index + cpu-isolation + cpu-load + mm/index + module-signing + namespaces/index + numastat + pm/index + syscall-user-dispatch + +Support for non-native binary formats. Note that some of these +documents are ... old ... + +.. toctree:: + :maxdepth: 1 + + binfmt-misc + java + mono + + +Block-layer and filesystem administration +----------------------------------------- + +.. toctree:: + :maxdepth: 1 + + bcache + binderfs + blockdev/index + cifs/index + device-mapper/index + ext4 + filesystem-monitoring + nfs/index + iostats + jfs + md + ufs + xfs + +Device-specific guides +---------------------- + +How to configure your hardware within your Linux system. + +.. toctree:: + :maxdepth: 1 + + acpi/index + aoe/index + auxdisplay/index + braille-console + btmrvl + dell_rbu + edid + gpio/index + hw_random + laptops/index + lcd-panel-cgram + media/index + nvme-multipath + parport + pnp + rapidio + rtc + serial-console + svga + thermal/index + thunderbolt + vga-softcursor + video-output + +Workload analysis +----------------- + +This is the beginning of a section with information of interest to +application developers and system integrators doing analysis of the +Linux kernel for safety critical applications. Documents supporting +analysis of kernel interactions with applications, and key kernel +subsystems expectations will be found here. + +.. toctree:: + :maxdepth: 1 + + workload-tracing + +Everything else +--------------- + +A few hard-to-categorize and generally obsolete documents. + +.. toctree:: + :maxdepth: 1 + + ldm + unicode -- cgit v1.3.1