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, 87 insertions, 0 deletions

diff --git a/net/wireless/michael-mic.c b/net/wireless/michael-mic.c
new file mode 100644
index 000000000..ec5164756
--- /dev/null
+++ b/net/wireless/michael-mic.c
@@ -0,0 +1,87 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Michael MIC implementation - optimized for TKIP MIC operations
+ * Copyright 2002-2003, Instant802 Networks, Inc.
+ */
+#include <linux/types.h>
+#include <linux/bitops.h>
+#include <linux/export.h>
+#include <linux/ieee80211.h>
+#include <linux/unaligned.h>
+
+struct michael_mic_ctx {
+	u32 l, r;
+};
+
+static void michael_block(struct michael_mic_ctx *mctx, u32 val)
+{
+	mctx->l ^= val;
+	mctx->r ^= rol32(mctx->l, 17);
+	mctx->l += mctx->r;
+	mctx->r ^= ((mctx->l & 0xff00ff00) >> 8) |
+		   ((mctx->l & 0x00ff00ff) << 8);
+	mctx->l += mctx->r;
+	mctx->r ^= rol32(mctx->l, 3);
+	mctx->l += mctx->r;
+	mctx->r ^= ror32(mctx->l, 2);
+	mctx->l += mctx->r;
+}
+
+static void michael_mic_hdr(struct michael_mic_ctx *mctx, const u8 *key,
+			    struct ieee80211_hdr *hdr)
+{
+	u8 *da, *sa, tid;
+
+	da = ieee80211_get_DA(hdr);
+	sa = ieee80211_get_SA(hdr);
+	if (ieee80211_is_data_qos(hdr->frame_control))
+		tid = ieee80211_get_tid(hdr);
+	else
+		tid = 0;
+
+	mctx->l = get_unaligned_le32(key);
+	mctx->r = get_unaligned_le32(key + 4);
+
+	/*
+	 * A pseudo header (DA, SA, Priority, 0, 0, 0) is used in Michael MIC
+	 * calculation, but it is _not_ transmitted
+	 */
+	michael_block(mctx, get_unaligned_le32(da));
+	michael_block(mctx, get_unaligned_le16(&da[4]) |
+			    (get_unaligned_le16(sa) << 16));
+	michael_block(mctx, get_unaligned_le32(&sa[2]));
+	michael_block(mctx, tid);
+}
+
+void michael_mic(const u8 *key, struct ieee80211_hdr *hdr,
+		 const u8 *data, size_t data_len, u8 *mic)
+{
+	u32 val;
+	size_t block, blocks, left;
+	struct michael_mic_ctx mctx;
+
+	michael_mic_hdr(&mctx, key, hdr);
+
+	/* Real data */
+	blocks = data_len / 4;
+	left = data_len % 4;
+
+	for (block = 0; block < blocks; block++)
+		michael_block(&mctx, get_unaligned_le32(&data[block * 4]));
+
+	/* Partial block of 0..3 bytes and padding: 0x5a + 4..7 zeros to make
+	 * total length a multiple of 4. */
+	val = 0x5a;
+	while (left > 0) {
+		val <<= 8;
+		left--;
+		val |= data[blocks * 4 + left];
+	}
+
+	michael_block(&mctx, val);
+	michael_block(&mctx, 0);
+
+	put_unaligned_le32(mctx.l, mic);
+	put_unaligned_le32(mctx.r, mic + 4);
+}
+EXPORT_SYMBOL_GPL(michael_mic);