lkml.org 
[lkml]   [2002]   [Jul]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: 64 bit netdev stats counter
Alan Cox wrote:
>
> On Fri, 2002-07-12 at 22:58, David S. Miller wrote:
> > 32-bit values aren't atomic either, what is the issue?
> > We don't use atomic_t ops on these counters so they aren't
> > guarenteed in any way right now even. GCC is going to
> > output "incl MEM" or similar for net_stats->counter++, since
> > it lacks the 'lock;' prefix it is not atomic.
>
> The behaviour is quite different though. On a 32bit counter the worst we
> do is lose a few counts. On a 64bit one on 32bit cpus its quite likely
> gcc will output
>
> increment low 32bit
> if zero
> increment high
>
> Which means we can rapidly get 2^32 out of sync

Isn't this the same as 32-bit counters on a machine that doesn't do atomic
32-bit ops? Although in that case you could only be 2^16 off...

Chris

--
Chris Friesen | MailStop: 043/33/F10
Nortel Networks | work: (613) 765-0557
3500 Carling Avenue | fax: (613) 765-2986
Nepean, ON K2H 8E9 Canada | email: cfriesen@nortelnetworks.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:27    [W:0.113 / U:0.648 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site