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, 227 insertions, 0 deletions

diff --git a/net/9p/error.c b/net/9p/error.c
new file mode 100644
index 000000000..8ba8afc91
--- /dev/null
+++ b/net/9p/error.c
@@ -0,0 +1,227 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Error string handling
+ *
+ * Plan 9 uses error strings, Unix uses error numbers.  These functions
+ * try to help manage that and provide for dynamically adding error
+ * mappings.
+ *
+ *  Copyright (C) 2004 by Eric Van Hensbergen <ericvh@gmail.com>
+ *  Copyright (C) 2002 by Ron Minnich <rminnich@lanl.gov>
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/module.h>
+#include <linux/list.h>
+#include <linux/jhash.h>
+#include <linux/errno.h>
+#include <linux/hashtable.h>
+#include <net/9p/9p.h>
+
+/**
+ * struct errormap - map string errors from Plan 9 to Linux numeric ids
+ * @name: string sent over 9P
+ * @val: numeric id most closely representing @name
+ * @namelen: length of string
+ * @list: hash-table list for string lookup
+ */
+struct errormap {
+	char *name;
+	int val;
+
+	int namelen;
+	struct hlist_node list;
+};
+
+#define ERRHASH_BITS 5
+static DEFINE_HASHTABLE(hash_errmap, ERRHASH_BITS);
+
+/* FixMe - reduce to a reasonable size */
+static struct errormap errmap[] = {
+	{"Operation not permitted", EPERM},
+	{"wstat prohibited", EPERM},
+	{"No such file or directory", ENOENT},
+	{"directory entry not found", ENOENT},
+	{"file not found", ENOENT},
+	{"Interrupted system call", EINTR},
+	{"Input/output error", EIO},
+	{"No such device or address", ENXIO},
+	{"Argument list too long", E2BIG},
+	{"Bad file descriptor", EBADF},
+	{"Resource temporarily unavailable", EAGAIN},
+	{"Cannot allocate memory", ENOMEM},
+	{"Permission denied", EACCES},
+	{"Bad address", EFAULT},
+	{"Block device required", ENOTBLK},
+	{"Device or resource busy", EBUSY},
+	{"File exists", EEXIST},
+	{"Invalid cross-device link", EXDEV},
+	{"No such device", ENODEV},
+	{"Not a directory", ENOTDIR},
+	{"Is a directory", EISDIR},
+	{"Invalid argument", EINVAL},
+	{"Too many open files in system", ENFILE},
+	{"Too many open files", EMFILE},
+	{"Text file busy", ETXTBSY},
+	{"File too large", EFBIG},
+	{"No space left on device", ENOSPC},
+	{"Illegal seek", ESPIPE},
+	{"Read-only file system", EROFS},
+	{"Too many links", EMLINK},
+	{"Broken pipe", EPIPE},
+	{"Numerical argument out of domain", EDOM},
+	{"Numerical result out of range", ERANGE},
+	{"Resource deadlock avoided", EDEADLK},
+	{"File name too long", ENAMETOOLONG},
+	{"No locks available", ENOLCK},
+	{"Function not implemented", ENOSYS},
+	{"Directory not empty", ENOTEMPTY},
+	{"Too many levels of symbolic links", ELOOP},
+	{"No message of desired type", ENOMSG},
+	{"Identifier removed", EIDRM},
+	{"No data available", ENODATA},
+	{"Machine is not on the network", ENONET},
+	{"Package not installed", ENOPKG},
+	{"Object is remote", EREMOTE},
+	{"Link has been severed", ENOLINK},
+	{"Communication error on send", ECOMM},
+	{"Protocol error", EPROTO},
+	{"Bad message", EBADMSG},
+	{"File descriptor in bad state", EBADFD},
+	{"Streams pipe error", ESTRPIPE},
+	{"Too many users", EUSERS},
+	{"Socket operation on non-socket", ENOTSOCK},
+	{"Message too long", EMSGSIZE},
+	{"Protocol not available", ENOPROTOOPT},
+	{"Protocol not supported", EPROTONOSUPPORT},
+	{"Socket type not supported", ESOCKTNOSUPPORT},
+	{"Operation not supported", EOPNOTSUPP},
+	{"Protocol family not supported", EPFNOSUPPORT},
+	{"Network is down", ENETDOWN},
+	{"Network is unreachable", ENETUNREACH},
+	{"Network dropped connection on reset", ENETRESET},
+	{"Software caused connection abort", ECONNABORTED},
+	{"Connection reset by peer", ECONNRESET},
+	{"No buffer space available", ENOBUFS},
+	{"Transport endpoint is already connected", EISCONN},
+	{"Transport endpoint is not connected", ENOTCONN},
+	{"Cannot send after transport endpoint shutdown", ESHUTDOWN},
+	{"Connection timed out", ETIMEDOUT},
+	{"Connection refused", ECONNREFUSED},
+	{"Host is down", EHOSTDOWN},
+	{"No route to host", EHOSTUNREACH},
+	{"Operation already in progress", EALREADY},
+	{"Operation now in progress", EINPROGRESS},
+	{"Is a named type file", EISNAM},
+	{"Remote I/O error", EREMOTEIO},
+	{"Disk quota exceeded", EDQUOT},
+/* errors from fossil, vacfs, and u9fs */
+	{"fid unknown or out of range", EBADF},
+	{"permission denied", EACCES},
+	{"file does not exist", ENOENT},
+	{"authentication failed", ECONNREFUSED},
+	{"bad offset in directory read", ESPIPE},
+	{"bad use of fid", EBADF},
+	{"wstat can't convert between files and directories", EPERM},
+	{"directory is not empty", ENOTEMPTY},
+	{"file exists", EEXIST},
+	{"file already exists", EEXIST},
+	{"file or directory already exists", EEXIST},
+	{"fid already in use", EBADF},
+	{"file in use", ETXTBSY},
+	{"i/o error", EIO},
+	{"file already open for I/O", ETXTBSY},
+	{"illegal mode", EINVAL},
+	{"illegal name", ENAMETOOLONG},
+	{"not a directory", ENOTDIR},
+	{"not a member of proposed group", EPERM},
+	{"not owner", EACCES},
+	{"only owner can change group in wstat", EACCES},
+	{"read only file system", EROFS},
+	{"no access to special file", EPERM},
+	{"i/o count too large", EIO},
+	{"unknown group", EINVAL},
+	{"unknown user", EINVAL},
+	{"bogus wstat buffer", EPROTO},
+	{"exclusive use file already open", EAGAIN},
+	{"corrupted directory entry", EIO},
+	{"corrupted file entry", EIO},
+	{"corrupted block label", EIO},
+	{"corrupted meta data", EIO},
+	{"illegal offset", EINVAL},
+	{"illegal path element", ENOENT},
+	{"root of file system is corrupted", EIO},
+	{"corrupted super block", EIO},
+	{"protocol botch", EPROTO},
+	{"file system is full", ENOSPC},
+	{"file is in use", EAGAIN},
+	{"directory entry is not allocated", ENOENT},
+	{"file is read only", EROFS},
+	{"file has been removed", EIDRM},
+	{"only support truncation to zero length", EPERM},
+	{"cannot remove root", EPERM},
+	{"file too big", EFBIG},
+	{"venti i/o error", EIO},
+	/* these are not errors */
+	{"u9fs rhostsauth: no authentication required", 0},
+	{"u9fs authnone: no authentication required", 0},
+	{NULL, -1}
+};
+
+/**
+ * p9_error_init - preload mappings into hash list
+ *
+ */
+
+int p9_error_init(void)
+{
+	struct errormap *c;
+	u32 hash;
+
+	/* load initial error map into hash table */
+	for (c = errmap; c->name; c++) {
+		c->namelen = strlen(c->name);
+		hash = jhash(c->name, c->namelen, 0);
+		INIT_HLIST_NODE(&c->list);
+		hash_add(hash_errmap, &c->list, hash);
+	}
+
+	return 1;
+}
+EXPORT_SYMBOL(p9_error_init);
+
+/**
+ * p9_errstr2errno - convert error string to error number
+ * @errstr: error string
+ * @len: length of error string
+ *
+ */
+
+int p9_errstr2errno(char *errstr, int len)
+{
+	int errno;
+	struct errormap *c;
+	u32 hash;
+
+	errno = 0;
+	c = NULL;
+	hash = jhash(errstr, len, 0);
+	hash_for_each_possible(hash_errmap, c, list, hash) {
+		if (c->namelen == len && !memcmp(c->name, errstr, len)) {
+			errno = c->val;
+			break;
+		}
+	}
+
+	if (errno == 0) {
+		/* TODO: if error isn't found, add it dynamically */
+		errstr[len] = 0;
+		pr_err("%s: server reported unknown error %s\n",
+		       __func__, errstr);
+		errno = ESERVERFAULT;
+	}
+
+	return -errno;
+}
+EXPORT_SYMBOL(p9_errstr2errno);