cvs commit: src/usr.bin/pkill pkill.c

[Robert Watson]

On Fri, 26 Mar 2004, Garance A Drosehn wrote:

> At 1:11 PM -0500 3/26/04, Robert Watson wrote:
> >On Fri, 26 Mar 2004, Garance A Drosehn wrote:
> >
> >>    Modified files:
> >>      usr.bin/pkill        pkill.c
> >>    Log:
> >  >   Switch to checking P_KTHREAD instead of P_SYSTEM when
> >  >   deciding what "system processes" to always ignore.
> >
> >Either our definition or my understanding of P_KTHREAD vs
> >P_SYSTEM is weak.  P_SYSTEM seems to include init, and
> >P_KTHREAD seems to mean things that really are kprocs, but
> >there was some recent discussion of the details of this on
> >arch@ with regard to whether signals should be delivered to
> >things that are arguably kernel threads or processes.
> 
> It would not surprise me if P_KTHREAD is not the most correct check for
> me to make here, but in my debugging it sure seems like using P_SYSTEM
> results in some processes getting skipped that I do not expect to be
> skipped. 

I'm actually not complaining about your change (if anything, it seems
right to me), but rather pointing out that our implementation may be
confused and that has resulted in what you bumped into.

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert at fledge.watson.org      Senior Research Scientist, McAfee Research