Re: ios_in_flight of CONFIG_BLK_STATS (still) negative in 2.4

[Zlatko Calusic <zlatko.calusic@iskon.hr>]

Steffen Maier <smaier@users.sourceforge.net> writes:

> Hello,
>
> in this posting I would like to analyze and discuss a bug(?), that
> causes the number of I/O requests in flight (ios_in_flight) of the
> extended block device statistics (CONFIG_BLK_STATS) to become and stay
> negative on various kernels of version 2.4. In turn, this leads to
> erroneous display of 100% utilization for certain classes of harddisk
> devices with 'iostat -x' and probably other statistic tools relying on
> the 13th (last but two) field (ios_in_flight) of extended block device
> statistics listed in /proc/partitions.
>
> First I encountered the behavior with 2.4.24 but could reproduce it
> with 2.4.33-rc3 (which is now the latest 2.4 kernel version 2.4.33).
> There have been various postings concerning this problem in recent
> years but to the best of my knowledge no ultimate fix has been merged
> into the mainstream kernel so far.

Have you tried this patch: http://linux.inet.hr/iostat_patch_2.4.html

It was made against 2.4.24 so you'll probably need to tweak it a
little bit before applying it cleanly to a newer 2.4, but I think it
deserves it's chance.
-- 
Zlatko
-
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/