On Tue, Jun 05, 2018 at 10:29:48PM +0200, Stefan Sperling wrote:
> On Tue, Jun 05, 2018 at 10:24:09PM +0200, Remi Locherer wrote:
> > Hi,
> >
> > the ospf6ctl manual misses two database filters.
> >
> > OK?
> >
> > Remi
>
> ok
me too
> >
> >
> > Index: ospf6ctl.8
> >
On Wed, Jun 06, 2018 at 08:06:30AM +0200, Remi Locherer wrote:
> Hi,
>
> RfC 5340 says that for intra area prefix LSAs metric should be set to 0
> in case of point-to-multipoint or loopback interfaces. Otherwise metric
> should be set to the value of the interfaces output cost.
>
> ospf6d
On Wed, Jun 06, 2018 at 11:04:56PM +0200, Claudio Jeker wrote:
> So the announce keyword in bgpd is massivly overloaded.
> It is one of the most common things new bgpd users are unsure about.
> These are all possible announce options:
>
> announce (IPv4|IPv6) (none|unicast|
On Thu, Jun 07, 2018 at 10:20:17AM +0100, Stuart Henderson wrote:
> On 2018/06/07 10:26, Claudio Jeker wrote:
> > On Wed, Jun 06, 2018 at 11:04:56PM +0200, Claudio Jeker wrote:
> > > So the announce keyword in bgpd is massivly overloaded.
> > > It is one of the most c
On Wed, Jun 06, 2018 at 08:17:54PM +0200, Remi Locherer wrote:
> On Wed, Jun 06, 2018 at 09:01:49AM +0200, Claudio Jeker wrote:
> > On Wed, Jun 06, 2018 at 08:06:30AM +0200, Remi Locherer wrote:
> > > Hi,
> > >
> > > RfC 5340 says that for intra area p
So the announce keyword in bgpd is massivly overloaded.
It is one of the most common things new bgpd users are unsure about.
These are all possible announce options:
announce (IPv4|IPv6) (none|unicast|vpn)
announce as-4byte (yes|no)
announce capabilities (yes|no)
announce refresh (yes|no)
On Thu, Jun 07, 2018 at 12:14:07PM +0200, Claudio Jeker wrote:
> On Thu, Jun 07, 2018 at 10:20:17AM +0100, Stuart Henderson wrote:
> > On 2018/06/07 10:26, Claudio Jeker wrote:
> > > On Wed, Jun 06, 2018 at 11:04:56PM +0200, Claudio Jeker wrote:
> > > > So the announ
Hi,
This adds a protection to handle aspaths overly large attributes in bgpd.
The main idea is to protect other bgp routes downstream for hitting the
limit with is often not well catched.
The limit is currently a bit arbitarily set to 4096 - 1024 which is afaik
the same limit that bird uses.
On Thu, May 31, 2018 at 11:12:38AM +, Job Snijders wrote:
> On Wed, May 30, 2018 at 03:18:45PM +0200, Claudio Jeker wrote:
> > This adds a protection to handle aspaths overly large attributes in
> > bgpd. The main idea is to protect other bgp routes downstream for
> &
On Sun, Jun 24, 2018 at 05:58:09PM -0400, Rob Pierce wrote:
> It looks like a BER problem found while testing the new ldap client (with an
> empty password) was already addressed in snmpd back in 2010 by martinh.
>
> In LDAP under a CONTEXT class, 0 corresponds to LDAP_AUTH_SIMPLE. This is
>
My plan is to make struct prefix the main object that holds references to
aspath, nexthop and rib. This diff is the next little step towards that.
Pass a struct prefix to rde_filter and remove some of the arguments which
are now fetched from struct prefix. Currently some trickery is needed
for
On Wed, Jun 27, 2018 at 10:05:52AM -0400, Rob Pierce wrote:
> This ber.c change has been in ldapd since rev 1.1 and was applied to snmpd
> back
> in 2012. The following diff applies the change to the ldap client and ypldap.
>
> Ok?
OK claudio@
> Index: usr.bin/ldap/ber.c
>
On Tue, May 01, 2018 at 11:53:16AM +0200, Sebastian Benoit wrote:
> Florian Obser(flor...@openbsd.org) on 2018.04.30 18:27:52 +0200:
> > The -d flag should be a no-op in monitor mode since it does not modify
> > the routing table.
> >
> > However, if -d is provided route monitor lists all
On Sat, Oct 28, 2017 at 11:06:16AM +0200, Mike Belopuhov wrote:
> On Thu, Oct 26, 2017 at 08:58 +0200, Claudio Jeker wrote:
> > On Wed, Oct 25, 2017 at 11:46:05PM +0200, Mike Belopuhov wrote:
> > > On Wed, Oct 25, 2017 at 21:56 +0200, Claudio Jeker wrote:
> > > > W
On Tue, Jul 03, 2018 at 03:45:49PM -0400, Rob Pierce wrote:
> On Sat, Jun 30, 2018 at 02:04:16PM -0400, Rob Pierce wrote:
> > I recently committed a piece of BER code synchronizing in the wrong
> > direction
> > (i.e. from the ldap instances to the snmpd instance). sthen@ noticed a break
> > in
On Tue, Jul 03, 2018 at 05:38:18PM -0400, Rob Pierce wrote:
> On Tue, Jul 03, 2018 at 09:25:06PM +0100, Stuart Henderson wrote:
> > On 2018/07/03 22:17, Claudio Jeker wrote:
> > > I have a hard time to understand why this is needed in snmpd.
> > > For single cha
There is no need to run against both input filters.
path_update() and prefix_remove() are both smart enough to handle all
cases (similar to a regular update). Should make reloads a bit faster.
More precise
- prefix_remove() of a non existing prefix is a NOP
- path_update() does itself a
On Thu, Jan 11, 2018 at 11:51:25AM +0100, Peter Hessler wrote:
> On 2018 Jan 11 (Thu) at 20:41:19 +1000 (+1000), David Gwynne wrote:
> :vlan and trunk are definitely ethernet interfaces.
> :
>
> Is there anything in OpenBSD that isn't an ethernet inteface?
>
How about gif(4), gre(4), tun(4),
On Mon, Jan 08, 2018 at 11:50:54AM +0100, Alexander Bluhm wrote:
> On Mon, Jan 08, 2018 at 11:49:05AM +1000, David Gwynne wrote:
> > i want this so it makes mpls_input have the same function signature
> > as all the other protocol input functions we care about.
> >
> > it also helps mpls_input
On Wed, Feb 07, 2018 at 07:28:42PM +0100, Ingo Schwarze wrote:
> Hi,
>
> 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 confusing and the already
> excessively
On Thu, Feb 08, 2018 at 04:27:57PM +1000, David Gwynne wrote:
> some interfaces could support copying the ttl from an inner packet to an
> outer one, but there's no obvious way to configure that at the moment.
>
> this borrows the semantic of using -1 for a special thing from
> vnetid. this
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 see
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
On Sun, Feb 04, 2018 at 12:42:22AM +0100, Remi Locherer wrote:
> Hi
>
> This adds a new feature to ospfd: depend on interface.
>
> A ospfd.conf using it looks like this:
>
> --%<--
> redistribute default depend on carp0
> area 0.0.0.0 {
> interface em2 { depend on carp0 }
> [...]
>
This diff changes the way bgpd does updates. Instead of having its own
special update queue/tree it uses a regular RIB (Adj-RIB-Out) to store all
updates to be sent. Stuff that has been sent is linked to the prefixes
queue. On the peer there are also queues for updates and withdraws.
The whole
On Sun, Feb 11, 2018 at 02:17:38AM +0100, Sebastian Benoit wrote:
>
> Hi,
>
> here is the ospfd part again, now the socket option is set on reload as
> well.
>
> ok?
>
> (benno_route_priofilter_2_ospfd.diff)
>
> diff --git usr.sbin/ospfd/kroute.c usr.sbin/ospfd/kroute.c
> index
On Sat, Feb 10, 2018 at 09:03:10PM +1000, David Gwynne wrote:
> ipv6 packets dont have a DF bit you can set in the header, you have
> to pass it as an option to ip6_output if you want that behaviour
> on output. however, ip6_send makes it hard to pass ip6 options to
> ip6_output because you give
Similar to IPv4 this will inject all the magic reject routes for IPv6.
This includes the bad networks in the 6to4 range (2002::/16) and some
other magic.
Here it what you get:
Internet6:
DestinationGateway Flags Refs Use Mtu Prio Iface
::/96
Since today lo(4) will create 127.0.0.1 and the 127/8 reject route
automatically. This means /etc/netstart no longer needs to do that.
I will commit this in a few days so that people can make sure to have
-current kernels.
--
:wq Claudio
Index: etc/netstart
On Mon, Feb 05, 2018 at 06:52:39PM +0100, Jeremie Courreges-Anglas wrote:
>
> Hi,
>
> while reviewing an snmpd diff, I noticed that the fd in struct ber was
> always set to -1; and indeed snmpd, ldapd and ypldap only pass buffers
> to the ber API. So this diff removes support for direct
Forwarding to tech, since people may have missed this.
Hope to commit this soon.
- Forwarded message from Claudio Jeker <cje...@diehard.n-r-g.com> -
Date: Sun, 24 Dec 2017 19:07:11 +0100
From: Claudio Jeker
To: Mischa Peters
Subject: Re: relayd stops processing traffic intermit
On Wed, Jan 03, 2018 at 03:46:13PM +0100, Martin Pieuchot wrote:
> These defines do not help, they'll make further changes more complicated
> to understand, so I'd like to remove them. ok?
Yes, this is just being lazy in the wrong place.
OK claudio@
> Index: kern/uipc_usrreq.c
>
On Thu, Jan 04, 2018 at 03:58:07PM +0100, Martin Pieuchot wrote:
> 'struct socket' is exposed to userland via . Most of
> our programs in base do not need it. Since I'm going to change the
> content of this struct for MP works, I'd like to reduce the impact on
> userland.
>
> Here's a diff to
On Thu, Jun 21, 2018 at 10:33:37AM +0200, Reyk Floeter wrote:
> Hi,
>
> the _url code was broken and disabled in ypldap's aldap - I fixed it
> for ldap(1). The other chunk is a DEBUG message fix, not compiled by
> default.
>
> OK?
Looks good to me. OK claudio@
> Index:
Nothing is using the per aspath counts in the RDE.
Remove them, makes the code simpler.
OK?
--
:wq Claudio
Index: rde.h
===
RCS file: /cvs/src/usr.sbin/bgpd/rde.h,v
retrieving revision 1.168
diff -u -p -r1.168 rde.h
--- rde.h
This is the first step of some larger reshuffling of how the RDE is
working. One of the things needed is proper reference counting for
nexthops since I want to kill nexthop_link and nexthop_unlink in the long
run.
Even though an intermediat step the result is IMO a lot cleaner than
before.
On Thu, Aug 02, 2018 at 02:56:59PM +0200, Claudio Jeker wrote:
> On Tue, Jul 31, 2018 at 05:39:31PM +0200, Claudio Jeker wrote:
> > Instead of walking the Adj-RIB-In per RIB walk it once and check per
> > prefix if the RIB needs an update or not. This will make it easier to ma
On Fri, Aug 03, 2018 at 11:01:39AM +0200, Claudio Jeker wrote:
> On Thu, Aug 02, 2018 at 02:56:59PM +0200, Claudio Jeker wrote:
> > On Tue, Jul 31, 2018 at 05:39:31PM +0200, Claudio Jeker wrote:
> > > Instead of walking the Adj-RIB-In per RIB walk it once and check per
>
So the reshuffling made it possible to simply run the reconfigure
of the RIBs in the background and no longer lock up the RDE for
extended times. In short this means that during 'bgpctl reload'
other bgpctl calls no longer block and that also UPDATE from neighbors
are fully processed.
This does
Since the RIBs have been fully split there is no way that an update of a
Loc-RIB is modifying the Adj-RIB-In. Because of that there is no need
for the comment and also for the more complex loop construct.
--
:wq Claudio
Index: rde.c
On Mon, Jul 30, 2018 at 04:39:39PM -0400, Rob Pierce wrote:
> Some public ber functions sneaked in below the internal functions comment.
> Move
> them up so the comment regains its former truthiness.
>
> Ok?
OK claudio@
> Index: usr.bin/ldap/ber.c
>
This brings the network code more in line with what regular update
processing does. It adds the prefix to Adj-RIB-In and if "log update" is
set it will also log the addition and removal of a prefix.
To make the output of log update nicer I pimped peerself and also
initialize the remote address to
Per rfc6472 AS_SET should no longer be used but some AS still do.
Until now source-as would take the rightmost AS number of an AS_PATH no
matter if it was an AS_SEQUENCE or an AS_SET. Thit is not correct. Also
because AS_SET are used in aggregation source-as should match against the
aggregator AS
On Mon, Aug 06, 2018 at 03:37:38PM +0200, Sebastian Benoit wrote:
> Ayaka Koshibe(akosh...@openbsd.org) on 2018.08.05 17:43:34 -0700:
> > Hi,
> >
> > I noticed that if you try to move an immovable lo(4) to a nonexistent
> > rdomain,
> > a new routing table will be created despite the operation
On Fri, Aug 03, 2018 at 04:31:58PM +0200, Claudio Jeker wrote:
> So the reshuffling made it possible to simply run the reconfigure
> of the RIBs in the background and no longer lock up the RDE for
> extended times. In short this means that during 'bgpctl reload'
> other bgpctl cal
On Sat, Aug 11, 2018 at 09:54:44PM -0400, Rob Pierce wrote:
> In aldap.c, tls_write(2) and write(2) also return ssize_t, so both error and
> wrote have been changed accordingly.
>
> ok?
OK claudio
> Index: usr.bin/ldap/aldap.c
>
On Thu, Aug 09, 2018 at 10:37:50PM +0200, Sebastian Benoit wrote:
> Claudio Jeker(cje...@diehard.n-r-g.com) on 2018.08.09 21:59:16 +0200:
> > On Thu, Aug 09, 2018 at 03:10:11PM +0200, Claudio Jeker wrote:
> > > Per rfc6472 AS_SET should no longer be used but some AS still d
On Thu, Aug 09, 2018 at 03:10:11PM +0200, Claudio Jeker wrote:
> Per rfc6472 AS_SET should no longer be used but some AS still do.
> Until now source-as would take the rightmost AS number of an AS_PATH no
> matter if it was an AS_SEQUENCE or an AS_SET. Thit is not correct. Also
> be
On Wed, Aug 22, 2018 at 12:12:10AM +0200, Remi Locherer wrote:
> On Tue, Aug 21, 2018 at 05:54:18PM +0100, Stuart Henderson wrote:
> > On 2018/08/21 17:16, Remi Locherer wrote:
> > > Hi tech,
> > >
> > > recently we had a short outage in our network. A script started an
> > > additional
> > >
On Fri, Aug 24, 2018 at 11:57:54AM +0100, Ricardo Mestre wrote:
> Hi,
>
> Similar to the patch I just sent for smsc(4) this one also has a misplaced
> parenthesis in an if-clause. NetBSD fixed this on rev 1.13.
>
> OK?
OK claudio
> Index: if_upgt.c
>
On Tue, Aug 28, 2018 at 01:19:39PM +0200, Remi Locherer wrote:
> On Tue, Aug 28, 2018 at 07:56:43AM +0200, Claudio Jeker wrote:
> > On Mon, Aug 27, 2018 at 11:33:19PM +0200, Remi Locherer wrote:
> > > On Fri, Aug 24, 2018 at 12:21:31PM +0200, Remi Locherer wrote:
> > > &
Currently if a path is rejected because of parse errors there is only
a generic error message logged but it is not clear which prefix caused it.
Lets make this at least more obvious when 'log updates' is used.
OK?
--
:wq Claudio
Index: rde.c
This is in preparation for introducing as-sets (a fast lookup table for
when you want to make sure that your peering partner is realy only passing
you traffic he should).
To make as-set possible lets do some cleanup beforehands. This mainly
removes one element from the filter_as struct, uses
On Mon, Aug 27, 2018 at 12:11:43PM +0200, Claudio Jeker wrote:
> This is in preparation for introducing as-sets (a fast lookup table for
> when you want to make sure that your peering partner is realy only passing
> you traffic he should).
>
> To make as-set possible lets
On Mon, Aug 27, 2018 at 11:33:19PM +0200, Remi Locherer wrote:
> On Fri, Aug 24, 2018 at 12:21:31PM +0200, Remi Locherer wrote:
> > On Fri, Aug 24, 2018 at 08:58:12AM +0200, Claudio Jeker wrote:
> > > On Wed, Aug 22, 2018 at 12:12:10AM +0200, Remi Locherer wrote:
> > > &
On Wed, Aug 29, 2018 at 01:07:33PM +0200, Claudio Jeker wrote:
> This is the bgpd diff that allows bgpctl to show invalid / error paths
> which act as an implicit withdraw.
>
> While there also fix 'bgpctl show rib in nei foo' since until now that
> code actually printed the same
This is the bgpd diff that allows bgpctl to show invalid / error paths
which act as an implicit withdraw.
While there also fix 'bgpctl show rib in nei foo' since until now that
code actually printed the same as 'bgpctl show rib nei foo'.
The code is a bit shuffled to make the if statement in
There is no reason why bgpd must be running to do a bgpctl show mrt call.
This diff moves the SHOW_MRT case up to the IRRFILTER one and also makes
the pledge() calls stricter. I see no reason why rpath or wpath are
needed and neither did benno@ remember why.
OK?
--
:wq Claudio
Index: bgpctl.c
This diffs adds timestamps similar to kdump to bgpctl show mrt output
when messages or states are shown.
Currently it looks like this:
1531425344.00 153.5.146.105[23456] -> 62.48.0.253[0]: size 63
0.00 153.5.146.105[23456] -> 62.48.0.253[0]: Connect -> OpenSent
0.00
For "dump (all|updates) (in|out)" it totally makes sense to use the
extended timestamp format that includes microseconds as well. This
way it is better visible when updates got processed.
Table dumps are not affected since there the timestamp is not that
useful.
bgpdump shows now the time with
On Mon, Jul 23, 2018 at 04:17:07PM -0400, Rob Pierce wrote:
> It looks like some code was shuffled around in revision 1.34 in which
> snmpe_application was renamed to smi_application and this prototype was
> missed.
>
> ok?
OK claudio@
> Index: snmpe.c
>
This diff simply changes all the nexthop accesses from asp->nexthop to
calling prefix_nexthop() which currently does the same. This will ease
moving the nexthop into struct prefix and is part of a larger diff that
does exactly that.
OK?
--
:wq Claudio
Index: mrt.c
This adds the missing parser for messages aka updates.
It is reusing some of bgpd's functions which I already moved to util.c for
that and reimplements a few other bits. It also extends the current
attribute printing code to support all attributes even those that are not
seen in the ususal show
Fix a printf and change the way data is increased so that at the end of
the function it is not overflowing. In the VPN4 case data is increased by
sizeof(u_int64_t) but later on the data is increased by the full length
which moves the pointer beyond the end. Now this is not really a problem
since
This is the next small bit of rde rebuilding.
Move nexthop into struct filterstate (including nexthop flags) and use
them whereever possible. In some places (path_update, softreconfig_out
handler) the nexthop state are folded back into the rde_aspath.
This also fixes a possible crash when it
On Wed, Jul 25, 2018 at 12:27:29PM +0200, Sebastien Marie wrote:
> On Mon, Jul 16, 2018 at 11:37:41AM +0200, Sebastien Marie wrote:
> > Hi,
> >
> > I am looking at a way to auto-logout some workstation on inactivity.
>
> ping.
>
> no problem if there is no interest for such thing in xidle, I
On Tue, Jul 24, 2018 at 05:46:15PM +0100, Jason McIntyre wrote:
> On Tue, Jul 24, 2018 at 04:18:38PM +0200, Sebastian Benoit wrote:
> > phessler@ noticed that we dont document which redistribute settings
> > cause the priority filter to be enabled/disabled.
> >
> > Add this text:
> >
> >
On Tue, Jul 24, 2018 at 01:00:03PM +0200, Claudio Jeker wrote:
> After a lot of preparation diffs this does the actuall change of moving
> the nexthop form struct rde_aspath to struct prefix. Also the nexthop
> flags are moved into their own flag variable (reducing the flag masking
> t
After a lot of preparation diffs this does the actuall change of moving
the nexthop form struct rde_aspath to struct prefix. Also the nexthop
flags are moved into their own flag variable (reducing the flag masking
that happened before).
I tested this with most nexthop combos I can think of and
On Tue, Sep 04, 2018 at 08:15:23PM +0200, Denis Fondras wrote:
> Hi,
>
> While generating big prefixlists macros with bgpq3 (big like
> as2914:as-europe-v6
> for example), OpenBGPd cannot load the config file (error "string too long").
> This diff implements a dynamic buffer to overcome this
Currently a lot of different operations are implemented for prefix
matching. This is not really needed because most can be expressed with the
range expression (prefixlen A - B). The only exceptions are the != and ><
operators. Both of them are a bit excotic and may be removed at a later time.
On Fri, Sep 07, 2018 at 12:23:05PM +0200, Klemens Nanni wrote:
> henning@ removed this functionality years ago, see the lo.4 diff:
>
> revision 1.27
> date: 2011/04/02 14:38:09; author: henning; state: Exp; lines: +2
> -37;
> rmeove the link1 hack, it is in the way, it is
On Sun, Sep 09, 2018 at 03:17:19AM -0700, Ayaka Koshibe wrote:
> Hi,
>
> This is a new version of a previous diff that I had for making switchd(8)
> ignore PACKET_IN messages generated from looped traffic. Currently, it will
> respond to the PACKET_IN with an invalid PACKET_OUT onto
On Sun, Sep 09, 2018 at 08:41:07AM +0200, Alexandr Nedvedicky wrote:
> Hello,
>
> while poking around 'XXXSMP' comments in net/if.c, I've noticed
> the 'if_cloners_lock' can be removed. The thing is the list of
> cloners gets populated/modified at system boot only, while kernel
> attaches device
On Tue, Sep 04, 2018 at 05:49:29PM +0200, Denis Fondras wrote:
> On Tue, Sep 04, 2018 at 01:56:12PM +0200, Claudio Jeker wrote:
> > On Mon, Aug 27, 2018 at 12:11:43PM +0200, Claudio Jeker wrote:
> > > This is in preparation for introducing as-sets (a fast lookup table for
&
t.ps != NULL) &&
@@ -615,6 +621,13 @@ rde_filter_equal(struct filter_head *a,
& PREFIXSET_FLAG_DIRTY) != 0)) {
log_debug("%s: prefixset %s has changed",
__func__, fa->match.prefixset.name);
+
.ps->dirty) {
log_debug("%s: prefixset %s has changed",
__func__, fa->match.prefixset.name);
return (0);
Index: rde_trie.c
===
RCS file: rde_trie.
On Mon, Aug 27, 2018 at 12:11:43PM +0200, Claudio Jeker wrote:
> This is in preparation for introducing as-sets (a fast lookup table for
> when you want to make sure that your peering partner is realy only passing
> you traffic he should).
>
> To make as-set possible lets
Yet another minor bgpd diff from a much bigger diff. This adds
inet4applymask() similar to inet6applymask() and starts using it in a few
places. This makes some of the INET vs INET6 cases more similar.
OK?
--
:wq Claudio
Index: bgpd.h
wrote:
> > > On Wed, Jul 25, 2018 at 12:55:48PM +0200, Claudio Jeker wrote:
> > > > On Wed, Jul 25, 2018 at 12:27:29PM +0200, Sebastien Marie wrote:
> > > > > On Mon, Jul 16, 2018 at 11:37:41AM +0200, Sebastien Marie wrote:
> > > > >
On Tue, Sep 11, 2018 at 12:01:28AM +0200, Alexander Bluhm wrote:
> Hi,
>
> I don't see any use for the INPCBHASH() macros. Just expand them.
>
> ok?
OK claudio@
> bluhm
>
> Index: netinet/in_pcb.c
> ===
> RCS file:
On Wed, Jul 04, 2018 at 11:51:46PM +0200, Sebastian Benoit wrote:
> Remi noticed that ospfd does no longer see new interfaces.
>
> The priority filter should only work on routing messages that
> have a rtm_priority.
>
> So these are out:
>
> RTM_DELADDR
> RTM_NEWADDR (struct ifa_msghdr)
>
Next step on the bigger RIB refactor. Introduce a filterstate instead of
passing multiple things around. As a benefit this reduces the amount of
malloc() / free() calls because the rde_aspath is now on the stack.
After that more changes will become possible like moving nexthop into the
state, etc.
On Thu, Jul 05, 2018 at 10:51:24PM +0100, Stuart Henderson wrote:
> I noticed relatively high cpu use from slaacd on a BGP router
> that was undergoing some route churn earlier (no interfaces were
> actually configured to use slaac).
>
> routesock is currently unfiltered so will see a lot of
t;);
>
> - if ((frontend_routesock = socket(PF_ROUTE, SOCK_RAW | SOCK_CLOEXEC, 0))
> - < 0)
> + if ((frontend_routesock = socket(PF_ROUTE, SOCK_RAW | SOCK_CLOEXEC,
> + AF_INET6)) < 0)
> fatal("route socket");
>
> rtfilter = ROUTE_
On Tue, Jul 10, 2018 at 01:56:37PM +0200, Florian Riehm wrote:
> Hi,
>
> If intra area prefixes move from one router to another router, cloning routes
> may become gateway routes and contrary. The kernel does not allow to change
> the
> flags RTF_GATEWAY / RTF_CLONING in RTM_CHANGE messages, but
On Tue, Jul 10, 2018 at 12:18:15PM +0200, Florian Riehm wrote:
> Hi,
>
> since we use multiple cloning routes (mpath) if more than one ip address
> exists in the same network, the routes are distinguished by their gateway
> address, which is the associated interface address.
> The ospfd has to
On Tue, Jul 10, 2018 at 11:15:51AM +0200, Sebastian Benoit wrote:
> we only support up to RT_TABLEID_MAX rtables.
>
> ok?
OK
> (benno_bgpd_rtable_max.diff)
>
> Index: parse.y
> ===
> RCS file: /cvs/src/usr.sbin/bgpd/parse.y,v
>
On Tue, Jul 10, 2018 at 12:07:04AM +0200, Sebastian Benoit wrote:
> Sebastian Benoit(be...@openbsd.org) on 2018.07.10 00:06:06 +0200:
> > Claudio Jeker(cje...@diehard.n-r-g.com) on 2018.07.10 00:05:08 +0200:
> > > On Mon, Jul 09, 2018 at 11:58:05PM +0200, Seba
On Mon, Jul 09, 2018 at 11:45:41PM +0200, Sebastian Benoit wrote:
> Actually free the right thing in rib_free()
> Found by and with claudio.
>
> (benno_claudio_rde_rib_rib_free.diff)
OK claudio@
> Index: rde_rib.c
> ===
> RCS
On Mon, Jul 09, 2018 at 11:58:05PM +0200, Sebastian Benoit wrote:
>
> compare the right things here: we want to know if the flag has changed.
> Found with claudios help and patience.
>
> ok?
This is fucked up. :)
> (benno_claudio_rde_reconf_F_RIB_HASNOFIB.diff)
>
> Index: rde.c
>
Again lots of minuses that can be replaced with very few plus lines.
This moves the prefixlen check into rde_update_get_prefix*() functions
since those can already return an error that is actually the same.
OK?
--
:wq Claudio
Index: rde.c
This uses struct filterstate in rde_update_dispatch() and passes the
struct to a few functions down (rde_update_update, path_update and some
more friends). After that it should become easier to move nexthop and
later maybe the peer out of struct rde_aspath into struct prefix.
--
:wq Claudio
On Wed, Jul 11, 2018 at 12:43:41AM +0200, Sebastian Benoit wrote:
> hi,
>
> allows you to announce prefixes from the kernel routing table selected by
> priority.
>
> lightly tested, as in, the config part works.
>
> network inet priority 32
>
> ok?
This is fairly esotheric as a feature. It
On Wed, Jul 11, 2018 at 10:10:50AM +0200, Martin Pieuchot wrote:
> On 11/07/18(Wed) 09:55, Claudio Jeker wrote:
> > On busy servers I seen multiple RTM_LOSING message per second being
> > generated. This is not helpful (especially since nothing is doing
> > something with i
On busy servers I seen multiple RTM_LOSING message per second being
generated. This is not helpful (especially since nothing is doing
something with it). This diff removes the part where RTM_LOSING is
generated but at the same time adds some RTM_ADD / RTM_DELETE messages for
the dynamic routes
Especially the rde_aspath hash function is horrible.
Fix this by adding more bits to the SipHash which results in a better
spread. Also switch the stored hases to 64bit and save the one for
rde_aspath as well since it the compare a lot quicker.
--
:wq Claudio
Index: rde.h
On Wed, Jul 11, 2018 at 10:18:59PM +0200, Sebastian Benoit wrote:
>
> ok?
OK claudio@
> (benno_check_rdomain.diff)
>
> diff --git usr.sbin/bgpd/config.c usr.sbin/bgpd/config.c
> index 2a2e5dfa314..4f8134e6b68 100644
> --- usr.sbin/bgpd/config.c
> +++ usr.sbin/bgpd/config.c
> @@ -46,6 +46,7 @@
This diff ads enough extra code to parse and display mrt update messages.
Some code in bgpd needs to be moved to be reachable by bgpctl.
bgpctl code gets reshuffled so that mrt printing works without a running
bgpd and also to make the pledge more strict.
This is nice to see stuff like the full
On Mon, Jul 09, 2018 at 11:33:05AM +0200, Remi Locherer wrote:
> On Mon, Jul 09, 2018 at 10:42:16AM +0200, Claudio Jeker wrote:
> > On Mon, Jul 09, 2018 at 10:31:15AM +0200, Remi Locherer wrote:
> > > later on it is set with:
> > > log_setverbose(ospfd_con
501 - 600 of 2249 matches
Mail list logo