Re: clock jump in 0.2.0.23-rc, too

[Scott Bennett <bennett@xxxxxxxxxx>]

     On Mon, 31 Mar 2008 22:49:05 -0700 Lucky Greeen <shamrock@xxxxxxxxxxxxxx>
wrote:
    [much deleted  --SB]
>
>There is a very simple way for you to test if the time jumps you have
>been seeing are caused by the TSCs between two CPU cores being out of
>sync: simply temporarily disable SMP in your kernel. If the problem goes
>away, the TSCs likely are out of sync. If so, you may want to find out
>of there is a way to disable FreeBSD's use of the TSC similar to the
>"notsc" option in Linux and see if that addresses the problem.
>
     Okay.  First off, here is the latest batch of messages.

Mar 31 22:30:27.110 [warn] Your system clock just jumped 101 seconds forward; assuming established circuits no longer work.
Apr 01 00:11:46.119 [notice] Tor has successfully opened a circuit. Looks like client functionality is working.
Apr 01 03:33:23.582 [warn] Your system clock just jumped 105 seconds forward; assuming established circuits no longer work.
Apr 01 03:35:44.206 [warn] Your system clock just jumped 141 seconds forward; assuming established circuits no longer work.
Apr 01 03:37:45.574 [warn] Your system clock just jumped 121 seconds forward; assuming established circuits no longer work.
Apr 01 03:38:43.885 [notice] Tor has successfully opened a circuit. Looks like client functionality is working.
Apr 01 04:15:30.446 [warn] Your system clock just jumped 110 seconds forward; assuming established circuits no longer work.
Apr 01 04:16:27.713 [notice] Tor has successfully opened a circuit. Looks like client functionality is working.
Apr 01 04:18:59.194 [warn] Your system clock just jumped 148 seconds forward; assuming established circuits no longer work.
Apr 01 04:21:19.749 [warn] Your system clock just jumped 140 seconds forward; assuming established circuits no longer work.
Apr 01 04:25:01.172 [warn] Your system clock just jumped 127 seconds forward; assuming established circuits no longer work.

     Right about this point--I don't know the exact time, but a date command
entered soon after showed Tue Apr  1 04:25:46 CDT 2008--I gave it a

	hellas # sysctl machdep.hyperthreading_allowed=0
	machdep.hyperthreading_allowed: 1 -> 0

However, another message soon arrived:

Apr 01 04:27:34.551 [warn] Your system clock just jumped 153 seconds forward; assuming established circuits no longer work.
Apr 01 04:29:33.772 [notice] Tried for 120 seconds to get a connection to [scrubbed]:80. Giving up.
Apr 01 04:29:56.323 [notice] Tor has successfully opened a circuit. Looks like client functionality is working.

So it would appear that the SMP issue is not operative here.
     Also, I'd like to reiterate that I never saw this behavior before running
0.2.0.21-rc, and now I've seen it with 0.2.0.22-rc and 0.2.0.23-rc.  I've been
using tor on my FreeBSD system since sometime in 2005 and as a server on the
same system since April 2007, well before 0.2.0.21-rc was made available.  I
don't know what is causing it, but it definitely is a problem.  I'll leave
hyperthreading disabled for a while to see what else happens, but eventually
I do want to turn it back on again.  Despite all the bashing by AMD fanatics,
hyperthreading really does get a lot more work done per core in a given time,
provided that work is broken into multiple threads or processes that are safe
to run in parallel.  It will be interesting indeed to see the difference in
throughput on Intel's forthcoming eight-cored chips for hyperthreading enabled
vs. disabled.
     Another tidbit of information is that I checked systat's vmstat display
once again after disabling hyperthreading and found that the total of ~4000
timer interrupts in each two-second interval are no longer split evenly between
cpu0 and cpu1.  Instead, all ~4000 interrupts occur on cpu0 and none at all
occur on cpu1.  Hmmm...I suppose I ought to comment out the "NumCPUs 2" line
in torrc until I reenable hyperthreading, but with the machine running at over
99% idle most of the day, it probably makes no real difference anyway.


                                  Scott Bennett, Comm. ASMELG, CFIAG
**********************************************************************
* Internet:       bennett at cs.niu.edu                              *
*--------------------------------------------------------------------*
* "A well regulated and disciplined militia, is at all times a good  *
* objection to the introduction of that bane of all free governments *
* -- a standing army."                                               *
*    -- Gov. John Hancock, New York Journal, 28 January 1790         *
**********************************************************************