On Thu, 8 Jul 2010, Mulyadi Santosa wrote:
> On Thu, Jul 8, 2010 at 02:13, Robert P. J. Day <rpjday@xxxxxxxxxxxxxx> wrote:
> > in fact, that's exactly the right value based on this snippet from
> > init/main.c:
> >
> > unsigned long loops_per_jiffy = (1<<12);
> >
> > which is, of course, 4096.
>
> Thanks for sharing....so it's fixed in 64 bit environment...I
> assumen it's x64...not IA 64. So pls CMIIW about my assumption
you are correct, and yesterday was the first time i'd noticed that.
when i first gdb dumped that variable and saw 4096, i immediately
assumed i'd done something wrong because that value couldn't possibly
be right. i stand corrected, so gdb appears to be working
*reasonably* correctly.
> > but i still don't know why i can't print the changing value of
> > jiffies if i clear the cache with:
> >
> > (gdb) core-file /proc/kcore
>
> But, is it changing if you quit gdb and run it again? Just a stupid
> confirmation..
yup, tried that, no change. and i tried it again several hours
later just now -- no difference. so i'm thinking it's a
64-bit-related issue. beyond that, i'm baffled.
rday
p.s. the way i try to figure out what the symbol name is is to grep
through /proc/kallsyms, as in
$ grep jiffies /proc/kallsyms
so if anyone else has a 64-bit system, and wants to try that, feel
free to suggest a symbol name to dump that might make sense. for now,
in gdb, i'm running:
(gdb) p __jiffies
--
========================================================================
Robert P. J. Day Waterloo, Ontario, CANADA
Top-notch, inexpensive online Linux/OSS/kernel courses
http://crashcourse.ca
Twitter: http://twitter.com/rpjday
LinkedIn: http://ca.linkedin.com/in/rpjday
========================================================================
