Re: CVS commit: src/sys

[Ryota Ozaki <ozaki-r%netbsd.org@localhost>]

On Sat, Apr 20, 2019 at 6:45 PM Ryota Ozaki <ozaki-r%netbsd.org@localhost> wrote:
>
> On Fri, Apr 19, 2019 at 6:49 PM Kamil Rytarowski <n54%gmx.com@localhost> wrote:
> >
> > On 19.04.2019 11:41, J. Hannken-Illjes wrote:
> > >> On 19. Apr 2019, at 03:52, Ryota Ozaki <ozaki-r%netbsd.org@localhost> wrote:
> > >>
> > >> Module Name: src
> > >> Committed By:        ozaki-r
> > >> Date:                Fri Apr 19 01:52:56 UTC 2019
> > >>
> > >> Modified Files:
> > >>      src/sys/kern: kern_lwp.c kern_softint.c subr_psref.c
> > >>      src/sys/rump/kern/lib/libsysproxy: sysproxy.c
> > >>      src/sys/sys: lwp.h userret.h
> > >>
> > >> Log Message:
> > >> Implement a simple psref leak detector
> > >>
> > >> It detects leaks by counting up the number of held psref by an LWP and checking
> > >> its zeroness at the end of syscalls and softint handlers.  For the counter, a
> > >> unused field of struct lwp is reused.
> > >
> > > For DIAGNOSTIC-only operations LWP specific data
> > > (see kern/subr_lwp_specificdata.c) is a better choice.
> > >
> >
> > I wanted to propose the same. An exampe of this is in KCOV.
>
> Thanks.  I'll try it.  (I'll be AFK for the next few days...)

I'm sorry for delaying this task.  Finally I have benchmarked a revised patch
(our benchmarking setups have been out of service for a couple of weeks...).

Performance degradation of IP forwarding is 3%.  Is it acceptable as
DIAGNOSTIC?

Thanks,
  ozaki-r