sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling buggrafted

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 <zhanxusheng@xiaomi.com>
Suggested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Kayra Cizmeci <kayracizmeci@gmail.com>
Cc: <stable@kernel.org> # v7.2.x
Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com

1 files changed, 102 insertions, 0 deletions

diff --git a/net/mac80211/eht.c b/net/mac80211/eht.c
new file mode 100644
index 000000000..23fe82df6
--- /dev/null
+++ b/net/mac80211/eht.c
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * EHT handling
+ *
+ * Copyright(c) 2021-2026 Intel Corporation
+ */
+
+#include "driver-ops.h"
+#include "ieee80211_i.h"
+
+void
+ieee80211_eht_cap_ie_to_sta_eht_cap(struct ieee80211_sub_if_data *sdata,
+				    struct ieee80211_supported_band *sband,
+				    const u8 *he_cap_ie, u8 he_cap_len,
+				    const struct ieee80211_eht_cap_elem *eht_cap_ie_elem,
+				    u8 eht_cap_len,
+				    struct link_sta_info *link_sta)
+{
+	struct ieee80211_sta_eht_cap *eht_cap = &link_sta->pub->eht_cap;
+	struct ieee80211_he_cap_elem *he_cap_ie_elem = (void *)he_cap_ie;
+	u8 eht_ppe_size = 0;
+	u8 mcs_nss_size;
+	u8 eht_total_size = sizeof(eht_cap->eht_cap_elem);
+	u8 *pos = (u8 *)eht_cap_ie_elem;
+
+	memset(eht_cap, 0, sizeof(*eht_cap));
+
+	if (!eht_cap_ie_elem ||
+	    !ieee80211_get_eht_iftype_cap_vif(sband, &sdata->vif))
+		return;
+
+	mcs_nss_size = ieee80211_eht_mcs_nss_size(he_cap_ie_elem,
+						  &eht_cap_ie_elem->fixed,
+						  sdata->vif.type ==
+							NL80211_IFTYPE_STATION);
+
+	eht_total_size += mcs_nss_size;
+
+	/* Calculate the PPE thresholds length only if the header is present */
+	if (eht_cap_ie_elem->fixed.phy_cap_info[5] &
+			IEEE80211_EHT_PHY_CAP5_PPE_THRESHOLD_PRESENT) {
+		u16 eht_ppe_hdr;
+
+		if (eht_cap_len < eht_total_size + sizeof(u16))
+			return;
+
+		eht_ppe_hdr = get_unaligned_le16(eht_cap_ie_elem->optional + mcs_nss_size);
+		eht_ppe_size =
+			ieee80211_eht_ppe_size(eht_ppe_hdr,
+					       eht_cap_ie_elem->fixed.phy_cap_info);
+		eht_total_size += eht_ppe_size;
+
+		/* we calculate as if NSS > 8 are valid, but don't handle that */
+		if (eht_ppe_size > sizeof(eht_cap->eht_ppe_thres))
+			return;
+	}
+
+	if (eht_cap_len < eht_total_size)
+		return;
+
+	/* Copy the static portion of the EHT capabilities */
+	memcpy(&eht_cap->eht_cap_elem, pos, sizeof(eht_cap->eht_cap_elem));
+	pos += sizeof(eht_cap->eht_cap_elem);
+
+	/* Copy MCS/NSS which depends on the peer capabilities */
+	memset(&eht_cap->eht_mcs_nss_supp, 0,
+	       sizeof(eht_cap->eht_mcs_nss_supp));
+	memcpy(&eht_cap->eht_mcs_nss_supp, pos, mcs_nss_size);
+
+	if (eht_ppe_size)
+		memcpy(eht_cap->eht_ppe_thres,
+		       &eht_cap_ie_elem->optional[mcs_nss_size],
+		       eht_ppe_size);
+
+	eht_cap->has_eht = true;
+
+	/*
+	 * The MPDU length bits are reserved on all but 2.4 GHz and get set via
+	 * VHT (5 GHz) or HE (6 GHz) capabilities.
+	 */
+	if (sband->band != NL80211_BAND_2GHZ)
+		return;
+
+	switch (u8_get_bits(eht_cap->eht_cap_elem.mac_cap_info[0],
+			    IEEE80211_EHT_MAC_CAP0_MAX_MPDU_LEN_MASK)) {
+	case IEEE80211_EHT_MAC_CAP0_MAX_MPDU_LEN_11454:
+		link_sta->pub->agg.max_amsdu_len =
+			IEEE80211_MAX_MPDU_LEN_VHT_11454;
+		break;
+	case IEEE80211_EHT_MAC_CAP0_MAX_MPDU_LEN_7991:
+		link_sta->pub->agg.max_amsdu_len =
+			IEEE80211_MAX_MPDU_LEN_VHT_7991;
+		break;
+	case IEEE80211_EHT_MAC_CAP0_MAX_MPDU_LEN_3895:
+	default:
+		link_sta->pub->agg.max_amsdu_len =
+			IEEE80211_MAX_MPDU_LEN_VHT_3895;
+		break;
+	}
+
+	ieee80211_sta_recalc_aggregates(&link_sta->sta->sta);
+}