Re: top(1) stopped working on -current

Giorgos Keramidas wrote this message on Thu, Apr 27, 2006 at 17:13 +0300:
> On 2006-04-27 14:42, Jiawei Ye <leafy7382_at_gmail.com> wrote:
> > leafy_at_chihiro:~$ top
> > top: sysctl(kern.ccpu...) failed: No such file or directory
> >
> > -current as of 4 hrs ago.
> 
> The following change renames `kern.ccpu' to `kern.sched.ccpu', but it is
> kind of incomplete, IMHO.

I'll get to this in a few hours....  sorry for the breakage...  :(

/me just woke up.

> # Index: src/sys/kern/sched_4bsd.c
> # ===================================================================
> # --- src/sys/kern/sched_4bsd.c   (revision 19)
> # +++ src/sys/kern/sched_4bsd.c   (working copy)
> # _at__at_ -31,11 +31,11 _at__at_
> #   * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
> #   * SUCH DAMAGE.
> #   */
> #
> #  #include <sys/cdefs.h>
> # -__FBSDID("$FreeBSD: src/sys/kern/sched_4bsd.c,v 1.78 2006/04/17 18:20:37 jhb Exp $");
> # +__FBSDID("$FreeBSD: src/sys/kern/sched_4bsd.c,v 1.79 2006/04/26 19:42:38 jmg Exp $");
> #
> #  #include "opt_hwpmc_hooks.h"
> #
> #  #define kse td_sched
> #
> # _at__at_ -404,11 +404,11 _at__at_
> #  #define        loadfactor(loadav)      (2 * (loadav))
> #  #define        decay_cpu(loadfac, cpu) (((loadfac) * (cpu)) / ((loadfac) + FSCALE))
> #
> #  /* decay 95% of `ke_pctcpu' in 60 seconds; see CCPU_SHIFT before changing */
> #  static fixpt_t ccpu = 0.95122942450071400909 * FSCALE; /* exp(-1/20) */
> # -SYSCTL_INT(_kern, OID_AUTO, ccpu, CTLFLAG_RD, &ccpu, 0, "");
> # +SYSCTL_INT(_kern_sched, OID_AUTO, ccpu, CTLFLAG_RD, &ccpu, 0, "");
> #
> #  /*
> #   * If `ccpu' is not equal to `exp(-1/20)' and you still want to use the
> #   * faster/more-accurate formula, you'll have to estimate CCPU_SHIFT below
> #   * and possibly adjust FSHIFT in "param.h" so that (FSHIFT >= CCPU_SHIFT).
> 
> The main problems with this change are:
> 
>     1) It doesn't update consumers of the kern.ccpu sysctl that we
>        already have in the tree (i.e. top(1) which you saw breaking).
> 
>     2) It doesn't update the `kern.ccpu' sysctl in the sched_ule.c
>        scheduler, so it creates an inconsistency between different
>        FreeBSD scheduling systems.
> 
> I believe John-Mark can fix this easily, so I've copied him in this
> message.  In the mean time, you can revert sched_4bsd.c to revision 1.78
> and try to see if this fixes the problems top(1) is having.

-- 
  John-Mark Gurney				Voice: +1 415 225 5579

     "All that I will do, has been done, All that I have, has not."