Subject: Re: tape drive instrumentation, again
To: None <blymn@baesystems.com.au>
From: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
List: tech-kern
Date: 03/12/2006 21:53:37
> Folks,
>
> This is the next iteration of the tape/nfs statistics. I have made
> the disk_* functions wrappers around the iostat_* equivalents. This
> minimises the code changes. I have also done the same with
> disk_find(), this is a wrapper around the iostat_find() function but
> it returns the struct disk *. I also change the sysctl name from
> drivenames to iostatnames, I have left the nfs names in the list
> because I do think they could be useful.
>
> As before, a tar file containing the extra files plus the patch are on
> ftp.netbsd.org in /pub/NetBSD/misc/blymn/iostat_patch.tar.gz. Below
> is the patch from the tar ball so people can review it.
looks good enough to me.
minor comments:
- use struct member prefixes ("sc_", "nm_"), at least where other
existing members have it.
- how about having iostat_seek() instead of "seek++"?
- use malloc/free rather than MALLOC/FREE
unless it's in a critical path.
- is there any point to update CTL_HW_NAMES?
i think you can just remove it.
- is anyone use _resetstat()? how about just removing it?
- please make sysctl_hw_disknames etc static.
- is "#include <sys/lock.h>" in iostat.h necessary?
- can you do whitespace cleanups as another commit?
YAMAMOTO Takashi