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 --- include/uapi/asm-generic/msgbuf.h | 49 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 49 insertions(+) create mode 100644 include/uapi/asm-generic/msgbuf.h (limited to 'include/uapi/asm-generic/msgbuf.h') diff --git a/include/uapi/asm-generic/msgbuf.h b/include/uapi/asm-generic/msgbuf.h new file mode 100644 index 000000000..6504d7b74 --- /dev/null +++ b/include/uapi/asm-generic/msgbuf.h @@ -0,0 +1,49 @@ +/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */ +#ifndef __ASM_GENERIC_MSGBUF_H +#define __ASM_GENERIC_MSGBUF_H + +#include +#include + +/* + * generic msqid64_ds structure. + * + * Note extra padding because this structure is passed back and forth + * between kernel and user space. + * + * msqid64_ds was originally meant to be architecture specific, but + * everyone just ended up making identical copies without specific + * optimizations, so we may just as well all use the same one. + * + * 64 bit architectures use a 64-bit long time field here, while + * 32 bit architectures have a pair of unsigned long values. + * On big-endian systems, the lower half is in the wrong place. + * + * Pad space is left for: + * - 2 miscellaneous 32-bit values + */ + +struct msqid64_ds { + struct ipc64_perm msg_perm; +#if __BITS_PER_LONG == 64 + long msg_stime; /* last msgsnd time */ + long msg_rtime; /* last msgrcv time */ + long msg_ctime; /* last change time */ +#else + unsigned long msg_stime; /* last msgsnd time */ + unsigned long msg_stime_high; + unsigned long msg_rtime; /* last msgrcv time */ + unsigned long msg_rtime_high; + unsigned long msg_ctime; /* last change time */ + unsigned long msg_ctime_high; +#endif + unsigned long msg_cbytes; /* current number of bytes on queue */ + unsigned long msg_qnum; /* number of messages in queue */ + unsigned long msg_qbytes; /* max number of bytes on queue */ + __kernel_pid_t msg_lspid; /* pid of last msgsnd */ + __kernel_pid_t msg_lrpid; /* last receive pid */ + unsigned long __unused4; + unsigned long __unused5; +}; + +#endif /* __ASM_GENERIC_MSGBUF_H */ -- cgit v1.3.1