The Linux Kernel Mailing List
 help / color / mirror / Atom feed
* [PATCH] time: x86: Fix race switching from vsyscall to non-vsyscall clock
@ 2012-03-09 22:04 John Stultz
  2012-03-12 16:08 ` Thomas Gleixner
  0 siblings, 1 reply; 2+ messages in thread
From: John Stultz @ 2012-03-09 22:04 UTC (permalink / raw)
  To: LKML; +Cc: John Stultz, Andy Lutomirski, Thomas Gleixner, John Stultz

When switching from a vsyscall capable to a non-vsyscall capable
clocksource, there was a small race, where the last vsyscall
gettimeofday before the switch might return a invalid time value
using the new non-vsyscall enabled clocksource values after the
switch is complete.

This is due to the vsyscall code checking the vclock_mode once
outside of the seqcount protected section. After it reads the
vclock mode, it doesn't re-check that the sampled clock data
that is obtained in the seqcount critical section still matches.

The fix is to sample vclock_mode inside the protected section,
and as long as it isn't VCLOCK_NONE, return the calculated
value. If it has changed and is now VCLOCK_NONE, fall back
to the syscall gettime calculation.

CC: Andy Lutomirski <luto@amacapital.net>
CC: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <john.stultz@lianro.org>
---
 arch/x86/vdso/vclock_gettime.c |   25 ++++++++++++++++++++-----
 1 files changed, 20 insertions(+), 5 deletions(-)

diff --git a/arch/x86/vdso/vclock_gettime.c b/arch/x86/vdso/vclock_gettime.c
index 6bc0e72..e7cf1dd 100644
--- a/arch/x86/vdso/vclock_gettime.c
+++ b/arch/x86/vdso/vclock_gettime.c
@@ -85,21 +85,28 @@ notrace static inline long vgetns(void)
 notrace static noinline int do_realtime(struct timespec *ts)
 {
 	unsigned long seq, ns;
+	int mode;
 	do {
 		seq = read_seqbegin(&gtod->lock);
+		mode = gtod->clock.vclock_mode;
 		ts->tv_sec = gtod->wall_time_sec;
 		ts->tv_nsec = gtod->wall_time_nsec;
 		ns = vgetns();
 	} while (unlikely(read_seqretry(&gtod->lock, seq)));
+
 	timespec_add_ns(ts, ns);
+	if (mode == VCLOCK_NONE)
+		return -1;
 	return 0;
 }
 
 notrace static noinline int do_monotonic(struct timespec *ts)
 {
 	unsigned long seq, ns, secs;
+	int mode;
 	do {
 		seq = read_seqbegin(&gtod->lock);
+		mode = gtod->clock.vclock_mode;
 		secs = gtod->wall_time_sec;
 		ns = gtod->wall_time_nsec + vgetns();
 		secs += gtod->wall_to_monotonic.tv_sec;
@@ -116,6 +123,8 @@ notrace static noinline int do_monotonic(struct timespec *ts)
 	ts->tv_sec = secs;
 	ts->tv_nsec = ns;
 
+	if (mode == VCLOCK_NONE)
+		return -1;
 	return 0;
 }
 
@@ -158,19 +167,25 @@ notrace int __vdso_clock_gettime(clockid_t clock, struct timespec *ts)
 {
 	switch (clock) {
 	case CLOCK_REALTIME:
-		if (likely(gtod->clock.vclock_mode != VCLOCK_NONE))
-			return do_realtime(ts);
+		if (likely(gtod->clock.vclock_mode != VCLOCK_NONE)) {
+			if (do_realtime(ts))
+				goto fallback;
+			return 0;
+		}
 		break;
 	case CLOCK_MONOTONIC:
-		if (likely(gtod->clock.vclock_mode != VCLOCK_NONE))
-			return do_monotonic(ts);
+		if (likely(gtod->clock.vclock_mode != VCLOCK_NONE)) {
+			if (do_monotonic(ts))
+				goto fallback;
+			return 0;
+		}
 		break;
 	case CLOCK_REALTIME_COARSE:
 		return do_realtime_coarse(ts);
 	case CLOCK_MONOTONIC_COARSE:
 		return do_monotonic_coarse(ts);
 	}
-
+fallback:
 	return vdso_fallback_gettime(clock, ts);
 }
 int clock_gettime(clockid_t, struct timespec *)
-- 
1.7.3.2.146.gca209


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] time: x86: Fix race switching from vsyscall to non-vsyscall clock
  2012-03-09 22:04 [PATCH] time: x86: Fix race switching from vsyscall to non-vsyscall clock John Stultz
@ 2012-03-12 16:08 ` Thomas Gleixner
  0 siblings, 0 replies; 2+ messages in thread
From: Thomas Gleixner @ 2012-03-12 16:08 UTC (permalink / raw)
  To: John Stultz; +Cc: LKML, Andy Lutomirski, John Stultz

On Fri, 9 Mar 2012, John Stultz wrote:

> When switching from a vsyscall capable to a non-vsyscall capable
> clocksource, there was a small race, where the last vsyscall
> gettimeofday before the switch might return a invalid time value
> using the new non-vsyscall enabled clocksource values after the
> switch is complete.
> 
> This is due to the vsyscall code checking the vclock_mode once
> outside of the seqcount protected section. After it reads the
> vclock mode, it doesn't re-check that the sampled clock data
> that is obtained in the seqcount critical section still matches.
> 
> The fix is to sample vclock_mode inside the protected section,
> and as long as it isn't VCLOCK_NONE, return the calculated
> value. If it has changed and is now VCLOCK_NONE, fall back
> to the syscall gettime calculation.
> 
> CC: Andy Lutomirski <luto@amacapital.net>
> CC: Thomas Gleixner <tglx@linutronix.de>
> Signed-off-by: John Stultz <john.stultz@lianro.org>
> ---
>  arch/x86/vdso/vclock_gettime.c |   25 ++++++++++++++++++++-----
>  1 files changed, 20 insertions(+), 5 deletions(-)
> 
> diff --git a/arch/x86/vdso/vclock_gettime.c b/arch/x86/vdso/vclock_gettime.c
> index 6bc0e72..e7cf1dd 100644
> --- a/arch/x86/vdso/vclock_gettime.c
> +++ b/arch/x86/vdso/vclock_gettime.c
> @@ -85,21 +85,28 @@ notrace static inline long vgetns(void)
>  notrace static noinline int do_realtime(struct timespec *ts)
>  {
>  	unsigned long seq, ns;
> +	int mode;
>  	do {
>  		seq = read_seqbegin(&gtod->lock);
> +		mode = gtod->clock.vclock_mode;
>  		ts->tv_sec = gtod->wall_time_sec;
>  		ts->tv_nsec = gtod->wall_time_nsec;
>  		ns = vgetns();
>  	} while (unlikely(read_seqretry(&gtod->lock, seq)));
> +
>  	timespec_add_ns(ts, ns);
> +	if (mode == VCLOCK_NONE)
> +		return -1;

Can't we just return mode and avoid repeated conditionals all over the
place ?

>  {
	res = VCLOCK_NONE;

>  	switch (clock) {
>  	case CLOCK_REALTIME:
> -		if (likely(gtod->clock.vclock_mode != VCLOCK_NONE))
> -			return do_realtime(ts);

  			res = do_realtime(ts);
		break;			


>  	case CLOCK_MONOTONIC_COARSE:
>  		return do_monotonic_coarse(ts);
>  	}

	return res != VCLOCK_NONE ? 0 : vdso_fallback_gettime(clock, ts);

Thanks,

	tglx

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2012-03-12 16:08 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-03-09 22:04 [PATCH] time: x86: Fix race switching from vsyscall to non-vsyscall clock John Stultz
2012-03-12 16:08 ` Thomas Gleixner

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox