linux-kernel - [PATCH] sched_clock: prevent scd->clock from moving backwards, take #2

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Wed, 31 Dec 2008 09:54:17 +0100
From:	Ingo Molnar <mingo@...e.hu>
To:	Jayson King <dev@...sonking.com>
Cc:	linux-kernel@...r.kernel.org, torvalds@...ux-foundation.org,
	a.p.zijlstra@...llo.nl, shaggy@...ux.vnet.ibm.com, rjw@...k.pl,
	fabio.comolli@...il.com
Subject: [PATCH] sched_clock: prevent scd->clock from moving backwards,
	take #2


* Jayson King <dev@...sonking.com> wrote:

> Greetings,
>
> I noticed large latency spikes in 2.6.28 (and 2.6.27.10) which would 
> make the system unresponsive for minutes at a time (alt+Fn switching 
> wouldn't even work) when the system is under load. I could trigger the 
> latency by running a large make -j 3. Sometimes it would trigger in a 
> few seconds other times it would take a few minutes.
>
> I bisected it in 2.6.27.10 to d9a888b061f55534016d2d86a21639948312a117 
> -> Revert "sched_clock: prevent scd->clock from moving backwards" 
> (ca7e716c7833aeaeb8fedd6d004c5f5d5e14d325 in 2.6.28) and reverting it in 
> 2.6.28 make the latency spike go away. At least, so far no spike in over 
> 24 hours.

Could you try the patch below - does it solve the problem? (and if it 
solves the problem, and if you have working suspend-to-ram or hibernation 
on your system then please also double-check that those still work.)

	Ingo

------------------------>
>From 1c5745aa380efb6417b5681104b007c8612fb496 Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@...utronix.de>
Date: Mon, 22 Dec 2008 23:05:28 +0100
Subject: [PATCH] sched_clock: prevent scd->clock from moving backwards, take #2

Redo:

  5b7dba4: sched_clock: prevent scd->clock from moving backwards

which had to be reverted due to s2ram hangs:

  ca7e716: Revert "sched_clock: prevent scd->clock from moving backwards"

... this time with resume restoring GTOD later in the sequence
taken into account as well.

The "timekeeping_suspended" flag is not very nice but we cannot call into
GTOD before it has been properly resumed and the scheduler will run very
early in the resume sequence.

Cc: <stable@...nel.org>
Signed-off-by: Ingo Molnar <mingo@...e.hu>
---
 include/linux/time.h      |    1 +
 kernel/sched_clock.c      |    5 ++++-
 kernel/time/timekeeping.c |    7 +++++--
 3 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/include/linux/time.h b/include/linux/time.h
index ce321ac..fbbd2a1 100644
--- a/include/linux/time.h
+++ b/include/linux/time.h
@@ -105,6 +105,7 @@ extern unsigned long read_persistent_clock(void);
 extern int update_persistent_clock(struct timespec now);
 extern int no_sync_cmos_clock __read_mostly;
 void timekeeping_init(void);
+extern int timekeeping_suspended;
 
 unsigned long get_seconds(void);
 struct timespec current_kernel_time(void);
diff --git a/kernel/sched_clock.c b/kernel/sched_clock.c
index e8ab096..a0b0852 100644
--- a/kernel/sched_clock.c
+++ b/kernel/sched_clock.c
@@ -124,7 +124,7 @@ static u64 __update_sched_clock(struct sched_clock_data *scd, u64 now)
 
 	clock = scd->tick_gtod + delta;
 	min_clock = wrap_max(scd->tick_gtod, scd->clock);
-	max_clock = scd->tick_gtod + TICK_NSEC;
+	max_clock = wrap_max(scd->clock, scd->tick_gtod + TICK_NSEC);
 
 	clock = wrap_max(clock, min_clock);
 	clock = wrap_min(clock, max_clock);
@@ -227,6 +227,9 @@ EXPORT_SYMBOL_GPL(sched_clock_idle_sleep_event);
  */
 void sched_clock_idle_wakeup_event(u64 delta_ns)
 {
+	if (timekeeping_suspended)
+		return;
+
 	sched_clock_tick();
 	touch_softlockup_watchdog();
 }
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index fa05e88..900f1b6 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -46,6 +46,9 @@ struct timespec xtime __attribute__ ((aligned (16)));
 struct timespec wall_to_monotonic __attribute__ ((aligned (16)));
 static unsigned long total_sleep_time;		/* seconds */
 
+/* flag for if timekeeping is suspended */
+int __read_mostly timekeeping_suspended;
+
 static struct timespec xtime_cache __attribute__ ((aligned (16)));
 void update_xtime_cache(u64 nsec)
 {
@@ -92,6 +95,8 @@ void getnstimeofday(struct timespec *ts)
 	unsigned long seq;
 	s64 nsecs;
 
+	WARN_ON(timekeeping_suspended);
+
 	do {
 		seq = read_seqbegin(&xtime_lock);
 
@@ -299,8 +304,6 @@ void __init timekeeping_init(void)
 	write_sequnlock_irqrestore(&xtime_lock, flags);
 }
 
-/* flag for if timekeeping is suspended */
-static int timekeeping_suspended;
 /* time in seconds when suspend began */
 static unsigned long timekeeping_suspend_time;
 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/