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/ceph/auth_none.c | 146 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 146 insertions(+) create mode 100644 net/ceph/auth_none.c (limited to 'net/ceph/auth_none.c') diff --git a/net/ceph/auth_none.c b/net/ceph/auth_none.c new file mode 100644 index 000000000..99e1f3e10 --- /dev/null +++ b/net/ceph/auth_none.c @@ -0,0 +1,146 @@ +// SPDX-License-Identifier: GPL-2.0 + +#include + +#include +#include +#include +#include + +#include +#include + +#include "auth_none.h" + +static void reset(struct ceph_auth_client *ac) +{ + struct ceph_auth_none_info *xi = ac->private; + + xi->starting = true; +} + +static void destroy(struct ceph_auth_client *ac) +{ + kfree(ac->private); + ac->private = NULL; +} + +static int is_authenticated(struct ceph_auth_client *ac) +{ + struct ceph_auth_none_info *xi = ac->private; + + return !xi->starting; +} + +static int should_authenticate(struct ceph_auth_client *ac) +{ + struct ceph_auth_none_info *xi = ac->private; + + return xi->starting; +} + +static int ceph_auth_none_build_authorizer(struct ceph_auth_client *ac, + struct ceph_none_authorizer *au) +{ + void *p = au->buf; + void *const end = p + sizeof(au->buf); + int ret; + + ceph_encode_8_safe(&p, end, 1, e_range); + ret = ceph_auth_entity_name_encode(ac->name, &p, end); + if (ret < 0) + return ret; + + ceph_encode_64_safe(&p, end, ac->global_id, e_range); + au->buf_len = p - (void *)au->buf; + dout("%s built authorizer len %d\n", __func__, au->buf_len); + return 0; + +e_range: + return -ERANGE; +} + +static int build_request(struct ceph_auth_client *ac, void *buf, void *end) +{ + return 0; +} + +/* + * the generic auth code decode the global_id, and we carry no actual + * authenticate state, so nothing happens here. + */ +static int handle_reply(struct ceph_auth_client *ac, u64 global_id, + void *buf, void *end, u8 *session_key, + int *session_key_len, u8 *con_secret, + int *con_secret_len) +{ + struct ceph_auth_none_info *xi = ac->private; + + xi->starting = false; + ceph_auth_set_global_id(ac, global_id); + return 0; +} + +static void ceph_auth_none_destroy_authorizer(struct ceph_authorizer *a) +{ + kfree(a); +} + +/* + * build an 'authorizer' with our entity_name and global_id. it is + * identical for all services we connect to. + */ +static int ceph_auth_none_create_authorizer( + struct ceph_auth_client *ac, int peer_type, + struct ceph_auth_handshake *auth) +{ + struct ceph_none_authorizer *au; + int ret; + + au = kmalloc_obj(*au, GFP_NOFS); + if (!au) + return -ENOMEM; + + au->base.destroy = ceph_auth_none_destroy_authorizer; + + ret = ceph_auth_none_build_authorizer(ac, au); + if (ret) { + kfree(au); + return ret; + } + + auth->authorizer = (struct ceph_authorizer *) au; + auth->authorizer_buf = au->buf; + auth->authorizer_buf_len = au->buf_len; + auth->authorizer_reply_buf = NULL; + auth->authorizer_reply_buf_len = 0; + + return 0; +} + +static const struct ceph_auth_client_ops ceph_auth_none_ops = { + .reset = reset, + .destroy = destroy, + .is_authenticated = is_authenticated, + .should_authenticate = should_authenticate, + .build_request = build_request, + .handle_reply = handle_reply, + .create_authorizer = ceph_auth_none_create_authorizer, +}; + +int ceph_auth_none_init(struct ceph_auth_client *ac) +{ + struct ceph_auth_none_info *xi; + + dout("ceph_auth_none_init %p\n", ac); + xi = kzalloc_obj(*xi, GFP_NOFS); + if (!xi) + return -ENOMEM; + + xi->starting = true; + + ac->protocol = CEPH_AUTH_NONE; + ac->private = xi; + ac->ops = &ceph_auth_none_ops; + return 0; +} -- cgit v1.3.1