Index | Thread | Search

From:
Tom Smyth <tom.smyth@wirelessconnect.eu>
Subject:
Re: nsh needs to adapt to mbuf.h changes
To:
Theo de Raadt <deraadt@openbsd.org>
Cc:
Stefan Sperling <stsp@stsp.name>, ports <ports@openbsd.org>
Date:
Mon, 21 Jul 2025 18:15:50 +0100

Download raw body.

Thread
sorry for my lack of clarity,

I was proposing 1) to release  a new version of nsh  as it has been
stable , and get the updated release into ports for users of release /
stable  OpenBSD
and then 2)  apply  fix to  master branch of nsh for users of current

so users of OpenBSD7.7 can have the benefits of the small incremental
improvements in nsh prior to this change for OpenBSD Current,



On Mon, 21 Jul 2025 at 18:09, Theo de Raadt <deraadt@openbsd.org> wrote:
>
> I don't understand.
>
> This should not go to -stable.
>
> Tom Smyth <tom.smyth@wirelessconnect.eu> wrote:
>
> > Stuart, all,
> > Sorry for missing this, and not replying sooner,
> >
> >
> > It might be worth doing a release for those running stable, and then
> > adding in the fix with an ifdef
> > Ill have a chat with Stefan, offline,
> >
> > On Wed, 25 Jun 2025 at 01:14, Stuart Henderson <stu@spacehopper.org> wrote:
> > >
> > > quick heads-up, this change in base needs nsh to catch up. I didn't
> > > check but I assume the existing code is going to be similar enough to
> > > what's in usr.bin/netstat/mbuf.c that it should be easy to adapt.
> > >
> > >
> > > ----- Forwarded message from Alexander Bluhm <bluhm@openbsd.org> -----
> > >
> > > From: Alexander Bluhm <bluhm@openbsd.org>
> > > Date: Sun, 22 Jun 2025 05:34:40 -0600 (MDT)
> > > To: source-changes@cvs.openbsd.org
> > > Subject: CVS: cvs.openbsd.org: src
> > >
> > > CVSROOT:        /cvs
> > > Module name:    src
> > > Changes by:     bluhm@cvs.openbsd.org   2025/06/22 05:34:40
> > >
> > > Modified files:
> > >         sys/kern       : kern_sysctl.c uipc_mbuf.c uipc_mbuf2.c
> > >         sys/sys        : mbuf.h
> > >         usr.bin/netstat: mbuf.c
> > >
> > > Log message:
> > > Protect mbuf statistics counter with splnet.
> > >
> > > MP safe conters with per CPU memory still need spl protection when
> > > called from interrupt context.  Some increments of mbuf counters
> > > had that, others did not.  Introduce inline function mbstat_inc()
> > > and enum counter types for mbufs, like we have elsewhere.  With
> > > that mbufs counter increment takes splnet() everywhere.
> > >
> > > OK mvs@ dlg@
> > >
> > >
> > > ----- End forwarded message -----
> >
> >
> >
> > --
> > Kindest regards,
> > Tom Smyth.
> >



-- 
Kindest regards,
Tom Smyth.