* Thomas Maier-Komor <[EMAIL PROTECTED]> [2006-04-27 08:19]:
> But what can one do for disks/metadisks? There I have the same problem
> that they aren't enumerated in a sequence, but instead their
> enumeration can have holes. I tried to analyze what kstat does
> executing "apptrace -T libkstat kstat", but there seems to be a bug in
> apptrace which makes apptrace terminate at once.
 
  Use 
  
  truss -t \!all -u libkstat:* kstat

  instead.

> Is it possible to walk the datastructures of libkstat? As I'd also
> like to support Solaris 8 and 9, I am not really asking for an
> enhancements of libkstat, but for an idea or pointer, how one can
> access the relevant data. I've looked a little bit into libdevinfo,
> but I am not sure, if it will give me everything I need. Additionally,
> libdevinfo is out of sync with libkstat concerning dynamic
> reconfiguration making everything even more difficult.
> 
> Any ideas or pointers to docs (commented headers are fine).

  (You can ask for enhancements of libkstat, once you're ready.  There
  is also the Perl module, and I would like, at some point, to get Peter
  Tribble and Gary to figure out what a public org.opensolaris.os.kstat
  Java class should look like.)

  In any case, the easiest example is to look at iostat/mpstat/vmstat,
  which were the initial kstat consumers.  Jonh Levon recently
  commonized the code these tools use; for disks, look at

  http://cvs.opensolaris.org/source/xref/on/usr/src/cmd/stat/iostat/iostat.c

  and the functions it calls.

  - Stephen

-- 
Stephen Hahn, PhD  Solaris Kernel Development, Sun Microsystems
[EMAIL PROTECTED]  http://blogs.sun.com/sch/
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to