Merge branch 'rcu/misc' into next

- In order to prepare the layout for nohz_full work deferral to
  user exit, the context tracking state must shrink the counter
  of transitions to/from RCU not watching. The only possible hazard
  is to trigger wrap-around more easily, delaying a bit grace periods
  when that happens. This should be a rare event though. Yet add
  debugging and torture code to test that assumption.

- Fix memory leak on locktorture module

- Annotate accesses in rculist_nulls.h to prevent from KCSAN warnings.
  On recent discussions, we also concluded that all those WRITE_ONCE()
  and READ_ONCE() on list APIs deserve appropriate comments. Something
  to be expected for the next cycle.

- Provide a script to apply several configs to several commits with torture.

- Allow torture to reuse a build directory in order to save needless
  rebuild time.

- Various cleanups.

include/linux/context_tracking_state.h

@@ -18,12 +18,6 @@ enum ctx_state {
 	CT_STATE_MAX		= 4,
 };
 
-/* Odd value for watching, else even. */
-#define CT_RCU_WATCHING CT_STATE_MAX
-
-#define CT_STATE_MASK (CT_STATE_MAX - 1)
-#define CT_RCU_WATCHING_MASK (~CT_STATE_MASK)
-
 struct context_tracking {
 #ifdef CONFIG_CONTEXT_TRACKING_USER
 	/*
@@ -44,9 +38,45 @@ struct context_tracking {
 #endif
 };
 
+/*
+ * We cram two different things within the same atomic variable:
+ *
+ *                     CT_RCU_WATCHING_START  CT_STATE_START
+ *                                |                |
+ *                                v                v
+ *     MSB [ RCU watching counter ][ context_state ] LSB
+ *         ^                       ^
+ *         |                       |
+ * CT_RCU_WATCHING_END        CT_STATE_END
+ *
+ * Bits are used from the LSB upwards, so unused bits (if any) will always be in
+ * upper bits of the variable.
+ */
 #ifdef CONFIG_CONTEXT_TRACKING
+#define CT_SIZE (sizeof(((struct context_tracking *)0)->state) * BITS_PER_BYTE)
+
+#define CT_STATE_WIDTH bits_per(CT_STATE_MAX - 1)
+#define CT_STATE_START 0
+#define CT_STATE_END   (CT_STATE_START + CT_STATE_WIDTH - 1)
+
+#define CT_RCU_WATCHING_MAX_WIDTH (CT_SIZE - CT_STATE_WIDTH)
+#define CT_RCU_WATCHING_WIDTH     (IS_ENABLED(CONFIG_RCU_DYNTICKS_TORTURE) ? 2 : CT_RCU_WATCHING_MAX_WIDTH)
+#define CT_RCU_WATCHING_START     (CT_STATE_END + 1)
+#define CT_RCU_WATCHING_END       (CT_RCU_WATCHING_START + CT_RCU_WATCHING_WIDTH - 1)
+#define CT_RCU_WATCHING           BIT(CT_RCU_WATCHING_START)
+
+#define CT_STATE_MASK        GENMASK(CT_STATE_END,        CT_STATE_START)
+#define CT_RCU_WATCHING_MASK GENMASK(CT_RCU_WATCHING_END, CT_RCU_WATCHING_START)
+
+#define CT_UNUSED_WIDTH (CT_RCU_WATCHING_MAX_WIDTH - CT_RCU_WATCHING_WIDTH)
+
+static_assert(CT_STATE_WIDTH        +
+	      CT_RCU_WATCHING_WIDTH +
+	      CT_UNUSED_WIDTH       ==
+	      CT_SIZE);
+
 DECLARE_PER_CPU(struct context_tracking, context_tracking);
-#endif
+#endif	/* CONFIG_CONTEXT_TRACKING */
 
 #ifdef CONFIG_CONTEXT_TRACKING_USER
 static __always_inline int __ct_state(void)

include/linux/rculist_nulls.h

@@ -138,7 +138,7 @@ static inline void hlist_nulls_add_tail_rcu(struct hlist_nulls_node *n,
 
 	if (last) {
 		WRITE_ONCE(n->next, last->next);
-		n->pprev = &last->next;
+		WRITE_ONCE(n->pprev, &last->next);
 		rcu_assign_pointer(hlist_nulls_next_rcu(last), n);
 	} else {
 		hlist_nulls_add_head_rcu(n, h);
@@ -148,8 +148,8 @@ static inline void hlist_nulls_add_tail_rcu(struct hlist_nulls_node *n,
 /* after that hlist_nulls_del will work */
 static inline void hlist_nulls_add_fake(struct hlist_nulls_node *n)
 {
-	n->pprev = &n->next;
-	n->next = (struct hlist_nulls_node *)NULLS_MARKER(NULL);
+	WRITE_ONCE(n->pprev, &n->next);
+	WRITE_ONCE(n->next, (struct hlist_nulls_node *)NULLS_MARKER(NULL));
 }
 
 /**

kernel/locking/locktorture.c

@@ -103,8 +103,8 @@ static const struct kernel_param_ops lt_bind_ops = {
 	.get = param_get_cpumask,
 };
 
-module_param_cb(bind_readers, &lt_bind_ops, &bind_readers, 0644);
-module_param_cb(bind_writers, &lt_bind_ops, &bind_writers, 0644);
+module_param_cb(bind_readers, &lt_bind_ops, &bind_readers, 0444);
+module_param_cb(bind_writers, &lt_bind_ops, &bind_writers, 0444);
 
 long torture_sched_setaffinity(pid_t pid, const struct cpumask *in_mask, bool dowarn);
 
@@ -1211,6 +1211,10 @@ static void lock_torture_cleanup(void)
 			cxt.cur_ops->exit();
 		cxt.init_called = false;
 	}
+
+	free_cpumask_var(bind_readers);
+	free_cpumask_var(bind_writers);
+
 	torture_cleanup_end();
 }
 

kernel/rcu/Kconfig.debug

@@ -213,4 +213,19 @@ config RCU_STRICT_GRACE_PERIOD
 	  when looking for certain types of RCU usage bugs, for example,
 	  too-short RCU read-side critical sections.
 
+
+config RCU_DYNTICKS_TORTURE
+	bool "Minimize RCU dynticks counter size"
+	depends on RCU_EXPERT && !COMPILE_TEST
+	default n
+	help
+	  This option sets the width of the dynticks counter to its
+	  minimum usable value.  This minimum width greatly increases
+	  the probability of flushing out bugs involving counter wrap,
+	  but it also increases the probability of extending grace period
+	  durations.  This Kconfig option should therefore be avoided in
+	  production due to the consequent increased probability of OOMs.
+
+	  This has no value for production and is only for testing.
+
 endmenu # "RCU Debugging"

kernel/rcu/rcutorture.c

@@ -2438,10 +2438,8 @@ static bool rcu_torture_one_read(struct torture_random_state *trsp, long myid)
 	newstate = rcutorture_extend_mask(rtors.readstate, trsp);
 	WARN_ON_ONCE(newstate & RCUTORTURE_RDR_UPDOWN);
 	rcutorture_one_extend(&rtors.readstate, newstate, trsp, rtors.rtrsp++);
-	if (!rcu_torture_one_read_start(&rtors, trsp, myid)) {
-		rcutorture_one_extend(&rtors.readstate, 0, trsp, rtors.rtrsp);
+	if (!rcu_torture_one_read_start(&rtors, trsp, myid))
 		return false;
-	}
 	rtors.rtrsp = rcutorture_loop_extend(&rtors.readstate, trsp, rtors.rtrsp);
 	rcu_torture_one_read_end(&rtors, trsp);
 	return true;

tools/testing/selftests/rcutorture/bin/kvm-again.sh

@@ -31,7 +31,7 @@ fi
 if ! cp "$oldrun/scenarios" $T/scenarios.oldrun
 then
 	# Later on, can reconstitute this from console.log files.
-	echo Prior run batches file does not exist: $oldrun/batches
+	echo Prior run scenarios file does not exist: $oldrun/scenarios
 	exit 1
 fi
 
@@ -68,7 +68,7 @@ usage () {
 	echo "       --datestamp string"
 	echo "       --dryrun"
 	echo "       --duration minutes | <seconds>s | <hours>h | <days>d"
-	echo "       --link hard|soft|copy"
+	echo "       --link hard|soft|copy|inplace|inplace-force"
 	echo "       --remote"
 	echo "       --rundir /new/res/path"
 	echo "Command line: $scriptname $args"
@@ -121,7 +121,7 @@ do
 		shift
 		;;
 	--link)
-		checkarg --link "hard|soft|copy" "$#" "$2" 'hard\|soft\|copy' '^--'
+		checkarg --link "hard|soft|copy|inplace|inplace-force" "$#" "$2" 'hard\|soft\|copy\|inplace\|inplace-force' '^--'
 		case "$2" in
 		copy)
 			arg_link="cp -R"
@@ -132,6 +132,14 @@ do
 		soft)
 			arg_link="cp -Rs"
 			;;
+		inplace)
+			arg_link="inplace"
+			rundir="$oldrun"
+			;;
+		inplace-force)
+			arg_link="inplace-force"
+			rundir="$oldrun"
+			;;
 		esac
 		shift
 		;;
@@ -172,21 +180,37 @@ fi
 
 echo ---- Re-run results directory: $rundir
 
-# Copy old run directory tree over and adjust.
-mkdir -p "`dirname "$rundir"`"
-if ! $arg_link "$oldrun" "$rundir"
+if test "$oldrun" != "$rundir"
 then
-	echo "Cannot copy from $oldrun to $rundir."
-	usage
-fi
-rm -f "$rundir"/*/{console.log,console.log.diags,qemu_pid,qemu-pid,qemu-retval,Warnings,kvm-test-1-run.sh.out,kvm-test-1-run-qemu.sh.out,vmlinux} "$rundir"/log
-touch "$rundir/log"
-echo $scriptname $args | tee -a "$rundir/log"
-echo $oldrun > "$rundir/re-run"
-if ! test -d "$rundir/../../bin"
-then
-	$arg_link "$oldrun/../../bin" "$rundir/../.."
+	# Copy old run directory tree over and adjust.
+	mkdir -p "`dirname "$rundir"`"
+	if ! $arg_link "$oldrun" "$rundir"
+	then
+		echo "Cannot copy from $oldrun to $rundir."
+		usage
+	fi
+	rm -f "$rundir"/*/{console.log,console.log.diags,qemu_pid,qemu-pid,qemu-retval,Warnings,kvm-test-1-run.sh.out,kvm-test-1-run-qemu.sh.out,vmlinux} "$rundir"/log
+	touch "$rundir/log"
+	echo $scriptname $args | tee -a "$rundir/log"
+	echo $oldrun > "$rundir/re-run"
+	if ! test -d "$rundir/../../bin"
+	then
+		$arg_link "$oldrun/../../bin" "$rundir/../.."
+	fi
+else
+	# Check for a run having already happened.
+	find "$rundir" -name console.log -print > $T/oldrun-console.log
+	if test -s $T/oldrun-console.log
+	then
+		echo Run already took place in $rundir
+		if test "$arg_link" = inplace
+		then
+			usage
+		fi
+	fi
 fi
+
+# Find runs to be done based on their qemu-cmd files.
 for i in $rundir/*/qemu-cmd
 do
 	cp "$i" $T

tools/testing/selftests/rcutorture/bin/kvm-series.sh

@@ -0,0 +1,116 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0+
+#
+# Usage: kvm-series.sh config-list commit-id-list [ kvm.sh parameters ]
+#
+# Tests the specified list of unadorned configs ("TREE01 SRCU-P" but not
+# "CFLIST" or "3*TRACE01") and an indication of a set of commits to test,
+# then runs each commit through the specified list of commits using kvm.sh.
+# The runs are grouped into a -series/config/commit directory tree.
+# Each run defaults to a duration of one minute.
+#
+# Run in top-level Linux source directory.  Please note that this is in
+# no way a replacement for "git bisect"!!!
+#
+# This script is intended to replace kvm-check-branches.sh by providing
+# ease of use and faster execution.
+
+T="`mktemp -d ${TMPDIR-/tmp}/kvm-series.sh.XXXXXX`"
+trap 'rm -rf $T' 0
+
+scriptname=$0
+args="$*"
+
+config_list="${1}"
+if test -z "${config_list}"
+then
+	echo "$0: Need a quoted list of --config arguments for first argument."
+	exit 1
+fi
+if test -z "${config_list}" || echo "${config_list}" | grep -q '\*'
+then
+	echo "$0: Repetition ('*') not allowed in config list."
+	exit 1
+fi
+
+commit_list="${2}"
+if test -z "${commit_list}"
+then
+	echo "$0: Need a list of commits (e.g., HEAD^^^..) for second argument."
+	exit 2
+fi
+git log --pretty=format:"%h" "${commit_list}" > $T/commits
+ret=$?
+if test "${ret}" -ne 0
+then
+	echo "$0: Invalid commit list ('${commit_list}')."
+	exit 2
+fi
+sha1_list=`cat $T/commits`
+
+shift
+shift
+
+RCUTORTURE="`pwd`/tools/testing/selftests/rcutorture"; export RCUTORTURE
+PATH=${RCUTORTURE}/bin:$PATH; export PATH
+. functions.sh
+
+ret=0
+nfail=0
+nsuccess=0
+faillist=
+successlist=
+cursha1="`git rev-parse --abbrev-ref HEAD`"
+ds="`date +%Y.%m.%d-%H.%M.%S`-series"
+startdate="`date`"
+starttime="`get_starttime`"
+
+echo " --- " $scriptname $args | tee -a $T/log
+echo " --- Results directory: " $ds | tee -a $T/log
+
+for config in ${config_list}
+do
+	sha_n=0
+	for sha in ${sha1_list}
+	do
+		sha1=${sha_n}.${sha} # Enable "sort -k1nr" to list commits in order.
+		echo Starting ${config}/${sha1} at `date` | tee -a $T/log
+		git checkout "${sha}"
+		time tools/testing/selftests/rcutorture/bin/kvm.sh --configs "$config" --datestamp "$ds/${config}/${sha1}" --duration 1 "$@"
+		curret=$?
+		if test "${curret}" -ne 0
+		then
+			nfail=$((nfail+1))
+			faillist="$faillist ${config}/${sha1}(${curret})"
+		else
+			nsuccess=$((nsuccess+1))
+			successlist="$successlist ${config}/${sha1}"
+			# Successful run, so remove large files.
+			rm -f ${RCUTORTURE}/$ds/${config}/${sha1}/{vmlinux,bzImage,System.map,Module.symvers}
+		fi
+		if test "${ret}" -eq 0
+		then
+			ret=${curret}
+		fi
+		sha_n=$((sha_n+1))
+	done
+done
+git checkout "${cursha1}"
+
+echo ${nsuccess} SUCCESSES: | tee -a $T/log
+echo ${successlist} | fmt | tee -a $T/log
+echo | tee -a $T/log
+echo ${nfail} FAILURES: | tee -a $T/log
+echo ${faillist} | fmt | tee -a $T/log
+if test -n "${faillist}"
+then
+	echo | tee -a $T/log
+	echo Failures across commits: | tee -a $T/log
+	echo ${faillist} | tr ' ' '\012' | sed -e 's,^[^/]*/,,' -e 's/([0-9]*)//' |
+		sort | uniq -c | sort -k2n | tee -a $T/log
+fi
+echo Started at $startdate, ended at `date`, duration `get_starttime_duration $starttime`. | tee -a $T/log
+echo Summary: Successes: ${nsuccess} Failures: ${nfail} | tee -a $T/log
+cp $T/log tools/testing/selftests/rcutorture/res/${ds}
+
+exit "${ret}"

tools/testing/selftests/rcutorture/configs/rcu/TREE04

@@ -16,3 +16,4 @@ CONFIG_DEBUG_OBJECTS_RCU_HEAD=n
 CONFIG_RCU_EXPERT=y
 CONFIG_RCU_EQS_DEBUG=y
 CONFIG_RCU_LAZY=y
+CONFIG_RCU_DYNTICKS_TORTURE=y