[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH] x86/watchdog: Use real timestamps for watchdog timeout



>>> On 24.05.13 at 14:00, Andrew Cooper <andrew.cooper3@xxxxxxxxxx> wrote:
> The TSC scale is recalculated as part of local_time_calibration().  I
> couldn't spot any other sensible way of converting a TSC delta to
> something approximating seconds.  Have I missed something?

Since we're not after a precise amount of time, using cpu_khz as
basis and accepting that the period until the watchdog may fire
could be, say, twice the nominal amount doesn't seem that bad
to me.

Or, if we're going the double checking route, retaining the current
counting and simply using way too low a TSC delta to alter the
printed message, largely like what Tim suggested, could be an
option. In that case, even a factor 10 mis-scaling of the TSC count
wouldn't hurt. The only real problem would be when the TSC stops
counting while in deep C states, but that again is a non-issue here
because you can't at the same time get NMI storms and be in deep
C states permanently.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.