Re: minimize routing socket desync

2018-02-08 Thread Sebastian Benoit
please skip over the #if 0, thats debugging to find the right values. As i said work in progress, feedback welcome. Sebastian Benoit(be...@openbsd.org) on 2018.02.09 04:49:43 +0100: > Hi, > > when bgpd changes a lot of routes (and runs with fib-updates yes), the > routing socket of ospfd (and os

minimize routing socket desync

2018-02-08 Thread Sebastian Benoit
Hi, when bgpd changes a lot of routes (and runs with fib-updates yes), the routing socket of ospfd (and ospf6d, ripd, eigrpd) will get a lot of routing messages. When the kernel cannot write all the messages intothe recipients route socket, it will set the desync flag, and the daemon has to read t

Re: ospfd getting confused about who is DR

2018-02-08 Thread Claudio Jeker
On Fri, Feb 09, 2018 at 03:39:43AM +0100, Claudio Jeker wrote: > On netsplits it can happen that on join multiple ospfd end up as DR. > In my case with 3 routers the one cut off stays DR even though the rest of > the network already has a DR and BDR. > > Looking into this it seems that in some cas

simplify kqueue timers

2018-02-08 Thread Jonathan Matthew
kqueue timers are specified in milliseconds, and timeout_add_msec can do the conversion to ticks for us, so we can trim a few lines here. Using timeout_add_msec preserves the 'at least one tick' behaviour that was added in r1.65. kqueue regress tests still pass. ok? Index: kern_event.c

ospfd getting confused about who is DR

2018-02-08 Thread Claudio Jeker
On netsplits it can happen that on join multiple ospfd end up as DR. In my case with 3 routers the one cut off stays DR even though the rest of the network already has a DR and BDR. Looking into this it seems that in some cases we don't issue an IF_EVT_NBR_CHNG and so the re-evaluation of DR/BDR d

Re: Additional AMD CPUID bits

2018-02-08 Thread Mike Larkin
On Fri, Feb 09, 2018 at 01:11:41PM +1100, Jonathan Gray wrote: > Additional AMD CPUID bits documented in > "Processor Programming Reference (PPR) for AMD Family 17h > Model 01h, Revision B1 Processors" > sure > Index: i386/include/specialreg.h > ==

Additional AMD CPUID bits

2018-02-08 Thread Jonathan Gray
Additional AMD CPUID bits documented in "Processor Programming Reference (PPR) for AMD Family 17h Model 01h, Revision B1 Processors" Index: i386/include/specialreg.h === RCS file: /cvs/src/sys/arch/i386/include/specialreg.h,v retrievi

Re: leave(1): schedule absolute alarm for start of minute

2018-02-08 Thread Theo Buehler
On Tue, Feb 06, 2018 at 11:22:52AM -0600, Scott Cheloha wrote: > Hi, > > When I schedule an alarm for an absolute time with minute granularity, > I expect the alarm to go off at the beginning of that minute. > > So, this: > > leave 1430 > > should go off at 14:30:00. > > The two-second s

pppoe timeout_add -> timeout_add_(m)sec

2018-02-08 Thread Jonathan Matthew
Except for one, these are all simple multiples of whole seconds, so we can convert to timeout_add_sec. The remaining one is a constant fraction of hz, which is easy to convert to timeout_add_msec. ok? Index: if_pppoe.c === RCS fil

[PATCH] dhclient.c

2018-02-08 Thread Holger Mikolon
Hi tech@, I'm following -current and recently observed that my laptop startup hangs at execution of netstart (when dhclient is called). Apparently, this is because I use the following dhclient.conf (since a couple of OpenBSD releases - so the issue is not in the config): interface "bge0" {

Re: Fix for vi(1) manpage Visual command

2018-02-08 Thread Christian Weisgerber
On 2018-02-07, Claudio Jeker wrote: > On Wed, Feb 07, 2018 at 07:28:42PM +0100, Ingo Schwarze wrote: >> while trying to fix the documentation of screen splitting in the vi(1) >> manual page, jmc@ suggested to stop documenting it at all. I like >> the idea because splitting is useless and confusi

regression in pflog output

2018-02-08 Thread Alexandr Nedvedicky
Hello, Matthias Pitzl discovered a regression introduced by my earlier commit [1]. Matthias has noticed the pflogd output changes for ruleset here: 8<---8<---8<--8< block out log quick from any to 1.1.1.1 block out log quick from any

Re: ospf6d only needs AF_INET6 route messages

2018-02-08 Thread Jeremie Courreges-Anglas
On Thu, Feb 08 2018, Sebastian Benoit wrote: > can someone confirm that ospf6d still works with this change? > > oks? ok > (benno_ospf6d_kroute.diff) > > diff --git usr.sbin/ospf6d/kroute.c usr.sbin/ospf6d/kroute.c > index 58e4f72f287..df7d56f09dd 100644 > --- usr.sbin/ospf6d/kroute.c > +++ usr.

Re: ospf6d only needs AF_INET6 route messages

2018-02-08 Thread Remi Locherer
On Thu, Feb 08, 2018 at 11:52:01AM +0100, Sebastian Benoit wrote: > can someone confirm that ospf6d still works with this change? > > oks? ospf6d works as before with this change. It also matches what ospfd does. OK remi@ > > (benno_ospf6d_kroute.diff) > > diff --git usr.sbin/ospf6d/kroute.c

ospf6d only needs AF_INET6 route messages

2018-02-08 Thread Sebastian Benoit
can someone confirm that ospf6d still works with this change? oks? (benno_ospf6d_kroute.diff) diff --git usr.sbin/ospf6d/kroute.c usr.sbin/ospf6d/kroute.c index 58e4f72f287..df7d56f09dd 100644 --- usr.sbin/ospf6d/kroute.c +++ usr.sbin/ospf6d/kroute.c @@ -102,7 +102,7 @@ kr_init(int fs) k

Re: fixing kvm lapic hangs

2018-02-08 Thread Jonathan Matthew
On Thu, Feb 08, 2018 at 04:32:29PM +1300, Jonathan Matthew wrote: > This diff (most of which has been around for a while) changes delay_func > when running in KVM to use pvclock (effectively the TSC) to determine when to > stop spinning. Since this is done in a KVM-specific driver, it won't have >

Re: sys_socket() plumbing

2018-02-08 Thread Martin Pieuchot
On 07/02/18(Wed) 16:11, Ted Unangst wrote: > Martin Pieuchot wrote: > > Diff below shuffles sys_socket() to look like sys_socketpair(). > > > > The goal is to do socket operations first in both functions. Since > > they don't need the KERNEL_LOCK(), we will be able to mark the syscalls > > NOLOCK

egre(4) - encapsulate ethernet over gre

2018-02-08 Thread David Gwynne
gre (the protocol) can easily encapsulate ethernet. this addes an egre interface that allows configuration of such a thing. the only caveat with it is that the protocol doesnt pad ethernet frames, so theyre misaligned. egre copes with this on input by copying the payload to new, aligned memory if