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, 208 insertions, 0 deletions

diff --git a/include/math-emu/quad.h b/include/math-emu/quad.h
new file mode 100644
index 000000000..616113626
--- /dev/null
+++ b/include/math-emu/quad.h
@@ -0,0 +1,208 @@
+/* Software floating-point emulation.
+   Definitions for IEEE Quad Precision.
+   Copyright (C) 1997,1998,1999 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Richard Henderson (rth@cygnus.com),
+		  Jakub Jelinek (jj@ultra.linux.cz),
+		  David S. Miller (davem@redhat.com) and
+		  Peter Maydell (pmaydell@chiark.greenend.org.uk).
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If
+   not, write to the Free Software Foundation, Inc.,
+   59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
+
+#ifndef  __MATH_EMU_QUAD_H__
+#define  __MATH_EMU_QUAD_H__
+
+#if _FP_W_TYPE_SIZE < 32
+#error "Here's a nickel, kid. Go buy yourself a real computer."
+#endif
+
+#if _FP_W_TYPE_SIZE < 64
+#define _FP_FRACTBITS_Q         (4*_FP_W_TYPE_SIZE)
+#else
+#define _FP_FRACTBITS_Q		(2*_FP_W_TYPE_SIZE)
+#endif
+
+#define _FP_FRACBITS_Q		113
+#define _FP_FRACXBITS_Q		(_FP_FRACTBITS_Q - _FP_FRACBITS_Q)
+#define _FP_WFRACBITS_Q		(_FP_WORKBITS + _FP_FRACBITS_Q)
+#define _FP_WFRACXBITS_Q	(_FP_FRACTBITS_Q - _FP_WFRACBITS_Q)
+#define _FP_EXPBITS_Q		15
+#define _FP_EXPBIAS_Q		16383
+#define _FP_EXPMAX_Q		32767
+
+#define _FP_QNANBIT_Q		\
+	((_FP_W_TYPE)1 << (_FP_FRACBITS_Q-2) % _FP_W_TYPE_SIZE)
+#define _FP_IMPLBIT_Q		\
+	((_FP_W_TYPE)1 << (_FP_FRACBITS_Q-1) % _FP_W_TYPE_SIZE)
+#define _FP_OVERFLOW_Q		\
+	((_FP_W_TYPE)1 << (_FP_WFRACBITS_Q % _FP_W_TYPE_SIZE))
+
+#if _FP_W_TYPE_SIZE < 64
+
+union _FP_UNION_Q
+{
+   long double flt;
+   struct 
+   {
+#if __BYTE_ORDER == __BIG_ENDIAN
+      unsigned sign : 1;
+      unsigned exp : _FP_EXPBITS_Q;
+      unsigned long frac3 : _FP_FRACBITS_Q - (_FP_IMPLBIT_Q != 0)-(_FP_W_TYPE_SIZE * 3);
+      unsigned long frac2 : _FP_W_TYPE_SIZE;
+      unsigned long frac1 : _FP_W_TYPE_SIZE;
+      unsigned long frac0 : _FP_W_TYPE_SIZE;
+#else
+      unsigned long frac0 : _FP_W_TYPE_SIZE;
+      unsigned long frac1 : _FP_W_TYPE_SIZE;
+      unsigned long frac2 : _FP_W_TYPE_SIZE;
+      unsigned long frac3 : _FP_FRACBITS_Q - (_FP_IMPLBIT_Q != 0)-(_FP_W_TYPE_SIZE * 3);
+      unsigned exp : _FP_EXPBITS_Q;
+      unsigned sign : 1;
+#endif /* not bigendian */
+   } bits __attribute__((packed));
+};
+
+
+#define FP_DECL_Q(X)		_FP_DECL(4,X)
+#define FP_UNPACK_RAW_Q(X,val)	_FP_UNPACK_RAW_4(Q,X,val)
+#define FP_UNPACK_RAW_QP(X,val)	_FP_UNPACK_RAW_4_P(Q,X,val)
+#define FP_PACK_RAW_Q(val,X)	_FP_PACK_RAW_4(Q,val,X)
+#define FP_PACK_RAW_QP(val,X)		\
+  do {					\
+    if (!FP_INHIBIT_RESULTS)		\
+      _FP_PACK_RAW_4_P(Q,val,X);	\
+  } while (0)
+
+#define FP_UNPACK_Q(X,val)		\
+  do {					\
+    _FP_UNPACK_RAW_4(Q,X,val);		\
+    _FP_UNPACK_CANONICAL(Q,4,X);	\
+  } while (0)
+
+#define FP_UNPACK_QP(X,val)		\
+  do {					\
+    _FP_UNPACK_RAW_4_P(Q,X,val);	\
+    _FP_UNPACK_CANONICAL(Q,4,X);	\
+  } while (0)
+
+#define FP_PACK_Q(val,X)		\
+  do {					\
+    _FP_PACK_CANONICAL(Q,4,X);		\
+    _FP_PACK_RAW_4(Q,val,X);		\
+  } while (0)
+
+#define FP_PACK_QP(val,X)		\
+  do {					\
+    _FP_PACK_CANONICAL(Q,4,X);		\
+    if (!FP_INHIBIT_RESULTS)		\
+      _FP_PACK_RAW_4_P(Q,val,X);	\
+  } while (0)
+
+#define FP_ISSIGNAN_Q(X)		_FP_ISSIGNAN(Q,4,X)
+#define FP_NEG_Q(R,X)			_FP_NEG(Q,4,R,X)
+#define FP_ADD_Q(R,X,Y)			_FP_ADD(Q,4,R,X,Y)
+#define FP_SUB_Q(R,X,Y)			_FP_SUB(Q,4,R,X,Y)
+#define FP_MUL_Q(R,X,Y)			_FP_MUL(Q,4,R,X,Y)
+#define FP_DIV_Q(R,X,Y)			_FP_DIV(Q,4,R,X,Y)
+#define FP_SQRT_Q(R,X)			_FP_SQRT(Q,4,R,X)
+#define _FP_SQRT_MEAT_Q(R,S,T,X,Q)	_FP_SQRT_MEAT_4(R,S,T,X,Q)
+
+#define FP_CMP_Q(r,X,Y,un)	_FP_CMP(Q,4,r,X,Y,un)
+#define FP_CMP_EQ_Q(r,X,Y)	_FP_CMP_EQ(Q,4,r,X,Y)
+
+#define FP_TO_INT_Q(r,X,rsz,rsg)	_FP_TO_INT(Q,4,r,X,rsz,rsg)
+#define FP_TO_INT_ROUND_Q(r,X,rsz,rsg)	_FP_TO_INT_ROUND(Q,4,r,X,rsz,rsg)
+#define FP_FROM_INT_Q(X,r,rs,rt)	_FP_FROM_INT(Q,4,X,r,rs,rt)
+
+#define _FP_FRAC_HIGH_Q(X)	_FP_FRAC_HIGH_4(X)
+#define _FP_FRAC_HIGH_RAW_Q(X)	_FP_FRAC_HIGH_4(X)
+
+#else   /* not _FP_W_TYPE_SIZE < 64 */
+union _FP_UNION_Q
+{
+  long double flt /* __attribute__((mode(TF))) */ ;
+  struct {
+#if __BYTE_ORDER == __BIG_ENDIAN
+    unsigned sign  : 1;
+    unsigned exp   : _FP_EXPBITS_Q;
+    unsigned long frac1 : _FP_FRACBITS_Q-(_FP_IMPLBIT_Q != 0)-_FP_W_TYPE_SIZE;
+    unsigned long frac0 : _FP_W_TYPE_SIZE;
+#else
+    unsigned long frac0 : _FP_W_TYPE_SIZE;
+    unsigned long frac1 : _FP_FRACBITS_Q-(_FP_IMPLBIT_Q != 0)-_FP_W_TYPE_SIZE;
+    unsigned exp   : _FP_EXPBITS_Q;
+    unsigned sign  : 1;
+#endif
+  } bits;
+};
+
+#define FP_DECL_Q(X)		_FP_DECL(2,X)
+#define FP_UNPACK_RAW_Q(X,val)	_FP_UNPACK_RAW_2(Q,X,val)
+#define FP_UNPACK_RAW_QP(X,val)	_FP_UNPACK_RAW_2_P(Q,X,val)
+#define FP_PACK_RAW_Q(val,X)	_FP_PACK_RAW_2(Q,val,X)
+#define FP_PACK_RAW_QP(val,X)		\
+  do {					\
+    if (!FP_INHIBIT_RESULTS)		\
+      _FP_PACK_RAW_2_P(Q,val,X);	\
+  } while (0)
+
+#define FP_UNPACK_Q(X,val)		\
+  do {					\
+    _FP_UNPACK_RAW_2(Q,X,val);		\
+    _FP_UNPACK_CANONICAL(Q,2,X);	\
+  } while (0)
+
+#define FP_UNPACK_QP(X,val)		\
+  do {					\
+    _FP_UNPACK_RAW_2_P(Q,X,val);	\
+    _FP_UNPACK_CANONICAL(Q,2,X);	\
+  } while (0)
+
+#define FP_PACK_Q(val,X)		\
+  do {					\
+    _FP_PACK_CANONICAL(Q,2,X);		\
+    _FP_PACK_RAW_2(Q,val,X);		\
+  } while (0)
+
+#define FP_PACK_QP(val,X)		\
+  do {					\
+    _FP_PACK_CANONICAL(Q,2,X);		\
+    if (!FP_INHIBIT_RESULTS)		\
+      _FP_PACK_RAW_2_P(Q,val,X);	\
+  } while (0)
+
+#define FP_ISSIGNAN_Q(X)		_FP_ISSIGNAN(Q,2,X)
+#define FP_NEG_Q(R,X)			_FP_NEG(Q,2,R,X)
+#define FP_ADD_Q(R,X,Y)			_FP_ADD(Q,2,R,X,Y)
+#define FP_SUB_Q(R,X,Y)			_FP_SUB(Q,2,R,X,Y)
+#define FP_MUL_Q(R,X,Y)			_FP_MUL(Q,2,R,X,Y)
+#define FP_DIV_Q(R,X,Y)			_FP_DIV(Q,2,R,X,Y)
+#define FP_SQRT_Q(R,X)			_FP_SQRT(Q,2,R,X)
+#define _FP_SQRT_MEAT_Q(R,S,T,X,Q)	_FP_SQRT_MEAT_2(R,S,T,X,Q)
+
+#define FP_CMP_Q(r,X,Y,un)	_FP_CMP(Q,2,r,X,Y,un)
+#define FP_CMP_EQ_Q(r,X,Y)	_FP_CMP_EQ(Q,2,r,X,Y)
+
+#define FP_TO_INT_Q(r,X,rsz,rsg)	_FP_TO_INT(Q,2,r,X,rsz,rsg)
+#define FP_TO_INT_ROUND_Q(r,X,rsz,rsg)	_FP_TO_INT_ROUND(Q,2,r,X,rsz,rsg)
+#define FP_FROM_INT_Q(X,r,rs,rt)	_FP_FROM_INT(Q,2,X,r,rs,rt)
+
+#define _FP_FRAC_HIGH_Q(X)	_FP_FRAC_HIGH_2(X)
+#define _FP_FRAC_HIGH_RAW_Q(X)	_FP_FRAC_HIGH_2(X)
+
+#endif /* not _FP_W_TYPE_SIZE < 64 */
+
+#endif /* __MATH_EMU_QUAD_H__ */