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/net/tun_proto.h | 50 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 50 insertions(+) create mode 100644 include/net/tun_proto.h (limited to 'include/net/tun_proto.h') diff --git a/include/net/tun_proto.h b/include/net/tun_proto.h new file mode 100644 index 000000000..7b0de7852 --- /dev/null +++ b/include/net/tun_proto.h @@ -0,0 +1,50 @@ +#ifndef __NET_TUN_PROTO_H +#define __NET_TUN_PROTO_H + +#include +#include + +/* One byte protocol values as defined by VXLAN-GPE and NSH. These will + * hopefully get a shared IANA registry. + */ +#define TUN_P_IPV4 0x01 +#define TUN_P_IPV6 0x02 +#define TUN_P_ETHERNET 0x03 +#define TUN_P_NSH 0x04 +#define TUN_P_MPLS_UC 0x05 + +static inline __be16 tun_p_to_eth_p(u8 proto) +{ + switch (proto) { + case TUN_P_IPV4: + return htons(ETH_P_IP); + case TUN_P_IPV6: + return htons(ETH_P_IPV6); + case TUN_P_ETHERNET: + return htons(ETH_P_TEB); + case TUN_P_NSH: + return htons(ETH_P_NSH); + case TUN_P_MPLS_UC: + return htons(ETH_P_MPLS_UC); + } + return 0; +} + +static inline u8 tun_p_from_eth_p(__be16 proto) +{ + switch (proto) { + case htons(ETH_P_IP): + return TUN_P_IPV4; + case htons(ETH_P_IPV6): + return TUN_P_IPV6; + case htons(ETH_P_TEB): + return TUN_P_ETHERNET; + case htons(ETH_P_NSH): + return TUN_P_NSH; + case htons(ETH_P_MPLS_UC): + return TUN_P_MPLS_UC; + } + return 0; +} + +#endif -- cgit v1.3.1