Rick Lindsley wrote:
>
> Here's a patch to put sard changes similar to those in 2.4 into 2.5. I
> say "similar" because the I/O subsystem has changed sufficiently in 2.5
> that making them exactly the same might be more effort that it's
> worth. Still, we do record per-partition reads/writes/sectors, and
> per-disk stats on the same plus queue time and number of I/O merges.
> Once applied, "cat /proc/diskstats" outputs this information.
It would be very nice to get better disk accounting into the kernel.
> Because in 2.5.34, gendisk->part[0] no longer is an hd_struct that
> refers to the whole disk, there wasn't a convenient place to record
> this information. I gratuitously added an hd_struct to gendisk to have
> a place to store the information, below, but that's distasteful and
> ugly. I'd like to move it to a different place.
>
> Also, with this patch, we are collecting stats twice, once for these
> stats and once for /proc/stat (kstat). That's stupid and I'd like to
> get the stats only once and use them, perhaps, in two places.
kstat should be a lighter-weight per-cpu thing. But the current
disk accounting in there would make it 12 kilobytes per CPU.
My vote: remove the disk accounting from kernel_stat and use this.
> What follows works, but needs refinements. Comments welcome.
What are those refinements?
What userspace tools are available for interpreting this information?
-
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/
This archive was generated by hypermail 2b29 : Sun Sep 15 2002 - 22:00:27 EST