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 --- net/openvswitch/flow_netlink.h | 71 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 71 insertions(+) create mode 100644 net/openvswitch/flow_netlink.h (limited to 'net/openvswitch/flow_netlink.h') diff --git a/net/openvswitch/flow_netlink.h b/net/openvswitch/flow_netlink.h new file mode 100644 index 000000000..ff8cdecbe --- /dev/null +++ b/net/openvswitch/flow_netlink.h @@ -0,0 +1,71 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* + * Copyright (c) 2007-2013 Nicira, Inc. + */ + + +#ifndef FLOW_NETLINK_H +#define FLOW_NETLINK_H 1 + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include "flow.h" + +size_t ovs_tun_key_attr_size(void); +size_t ovs_key_attr_size(void); + +void ovs_match_init(struct sw_flow_match *match, + struct sw_flow_key *key, bool reset_key, + struct sw_flow_mask *mask); + +int ovs_nla_put_key(const struct sw_flow_key *, const struct sw_flow_key *, + int attr, bool is_mask, struct sk_buff *); +int parse_flow_nlattrs(const struct nlattr *attr, const struct nlattr *a[], + u64 *attrsp, bool log); +int ovs_nla_get_flow_metadata(struct net *net, + const struct nlattr *a[OVS_KEY_ATTR_MAX + 1], + u64 attrs, struct sw_flow_key *key, bool log); + +int ovs_nla_put_identifier(const struct sw_flow *flow, struct sk_buff *skb); +int ovs_nla_put_masked_key(const struct sw_flow *flow, struct sk_buff *skb); +int ovs_nla_put_mask(const struct sw_flow *flow, struct sk_buff *skb); + +int ovs_nla_get_match(struct net *, struct sw_flow_match *, + const struct nlattr *key, const struct nlattr *mask, + bool log); + +int ovs_nla_put_tunnel_info(struct sk_buff *skb, + struct ip_tunnel_info *tun_info); + +bool ovs_nla_get_ufid(struct sw_flow_id *, const struct nlattr *, bool log); +int ovs_nla_get_identifier(struct sw_flow_id *sfid, const struct nlattr *ufid, + const struct sw_flow_key *key, bool log); +u32 ovs_nla_get_ufid_flags(const struct nlattr *attr); + +int ovs_nla_copy_actions(struct net *net, const struct nlattr *attr, + const struct sw_flow_key *key, + struct sw_flow_actions **sfa, bool log); +int ovs_nla_add_action(struct sw_flow_actions **sfa, int attrtype, + void *data, int len, bool log); +int ovs_nla_put_actions(const struct nlattr *attr, + int len, struct sk_buff *skb); + +void ovs_nla_free_flow_actions(struct sw_flow_actions *); +void ovs_nla_free_flow_actions_rcu(struct sw_flow_actions *); + +int nsh_hdr_from_nlattr(const struct nlattr *attr, struct nshhdr *nh, + size_t size); + +#endif /* flow_netlink.h */ -- cgit v1.3.1