David Lamparter [Mon, 13 Apr 2015 07:50:00 +0000 (09:50 +0200)]
bgpd: speed up "no-hit" withdraws for routeservers
This accelerates handling of incoming Withdraw messages for routes that
don't exist in the table to begin with. Cisco IOS 12.4(24)T4 has a bug
in this regard - it sends withdraws instead of doing nothing for
prefixes that are filtered.
Pulling up the adj_in removal in Quagga should have no ill effect, but
we can avoid the costly iteration over all rsclients if there was no
adj_in entry.
Performance impact of this change on routeserver with 3 buggy peers,
startup/sync time:
before patch: 143.12 seconds (user cpu)
after patch: 7.01 seconds (user cpu)
Many thanks to Nick Hilliard & INEX for providing real-world test data!
Signed-off-by: David Lamparter <equinox@opensourcerouting.org> Acked-by: Paul Jakma <paul@jakma.org>
Paul Jakma [Tue, 20 Jan 2015 15:45:36 +0000 (15:45 +0000)]
ospfd: Remove the blocking of opaque LSAs origination & flooding 'optimisation'
* Opaque support contains some kind of hack/optimisation to
origination/flooding to suppress some origins/floods until an opaque LS
Acks are received. Previous versions of the code have already been shown
to have bugs in them (see e16fd8a5, e.g.). It seems over-complex and fragile,
plus its conceptually the wrong place to try implement flooding hacks that,
AFAICT, do not depend particularly on the semantics of opaque LSA.
David Lamparter [Tue, 5 May 2015 09:10:20 +0000 (11:10 +0200)]
lib/cli: reduce strcmp in CLI hot paths
Er, no idea how anyone could ever have thought that it would be a good
idea to have a zillion of strcmp() calls in the CLI's active paths, just
to compare against things like "A.B.C.D".
Reduces 40k prefix list load time from 1.65s to 1.23s (1.34:1).
Acked-by: Paul Jakma <paul@jakma.org>
[v2: killed CMDS_* macros] Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
(cherry picked from commit 10bac80195cf5a781da6e4415e6580fd7080f734)
Timo Teräs [Wed, 29 Apr 2015 06:43:01 +0000 (09:43 +0300)]
lib: constify sockunion api
Add const to read-only api calls.
Signed-off-by: Timo Teräs <timo.teras@iki.fi> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
(cherry picked from commit 81b139bdd530adda045d22a4daf0054b89703dab)
David Lamparter [Tue, 3 Mar 2015 09:00:43 +0000 (10:00 +0100)]
build: add --enable-werror
This allows enabling -Werror in a consistent way. Note that this is
different from just specifiying it in CFLAGS, since that would either
break configure tests (if done on ./configure), or would override
configure's CFLAGS (if done on make).
Using --enable-werror instead provides a new WERROR variable that is
additionally used during make with a consistent set of warning flags.
The tests/ directory is exempt. (Rationale being, better to have more
tests than pedantically complain about them.)
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
David Lamparter [Wed, 4 Mar 2015 06:18:24 +0000 (07:18 +0100)]
vtysh: drop unused variables & RETSIGTYPE
Drop unused return values in vtysh. Also gets rid of the rather funny
prototyping of signal setup in vtysh - which as a side effect makes it
not need AC_TYPE_SIGNAL in configure.ac anymore. It wasn't used
sensibly to begin with...
David Lamparter [Tue, 3 Mar 2015 09:30:27 +0000 (10:30 +0100)]
ospf6d: use existing union, avoid strict-aliasing
There are preexisting fields u.lp.id and u.lp.adv_router in struct
prefix that do the same thing as these type-punning pointer derefs.
Use these and shut up the strict-aliasing warnings.
Donald Sharp [Wed, 8 Jun 2016 15:47:33 +0000 (11:47 -0400)]
pimd: Fix register message checksum
The register message checksum was being calculated over
the first 4 bytes of the packet, instead of the first
8 bytes. From the RFC:
PIM Version, Type, Reserved, Checksum
Described in Section 4.9. Note that in order to reduce
encapsulation overhead, the checksum for Registers is done only
on the first 8 bytes of the packet, including the PIM header and
the next 4 bytes, excluding the data packet portion. For
interoperability reasons, a message carrying a checksum
calculated over the entire PIM Register message should also be
accepted. When calculating the checksum, the IPv6 pseudoheader
"Upper-Layer Packet Length" is set to 8.
Ticket: CM-11265 Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
kernel_delete_ipv6_old(), removed in 51bdeba a little while ago, was the
last user of netlink_route() and kernel_rtm_ipv6(). Everything else
uses the _multipath variants of these functions.
David Lamparter [Tue, 3 Mar 2015 07:54:54 +0000 (08:54 +0100)]
bgpd, zebra: fix struct/pointer sizeof mixups
Two places were taking sizeof(pointer) instead of the sizeof(struct),
while performing operations on the struct. Both are initialisation
functions; I guess we haven't seen fallout since they weren't critical.
Fix anyway.
[v2: fix mistake that actually broke bgpd RS workqueue init] Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
(cherry picked from commit d43f8b39b075fe60e0c8fdb33b07b284d3fae503)
David Lamparter [Tue, 3 Mar 2015 08:07:25 +0000 (09:07 +0100)]
*: add/cleanup initialisers
There were some (inconsequential) warnings about uninitialised use of
variables. Also, in one case, sub-structs were mixed in initialisation,
which doesn't quite work as intended.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
Lou Berger [Tue, 12 Jan 2016 18:41:55 +0000 (13:41 -0500)]
bgpd: wire up VPNv6 protocol processing
There wasn't much missing for VPNv6 to begin with; just a few bits of
de- & encoding and a few lists to be updated.
Signed-off-by: Lou Berger <lberger@labn.net> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
[Editorial note: Signed-off-by may imply an authorship claim, but need not]
Lou Berger [Tue, 12 Jan 2016 18:41:46 +0000 (13:41 -0500)]
lib: fix bookkeeping for libreadline malloc()s
When libreadline is used, we mistakenly mix in strdup() done in
libreadline with Quagga's lib/memory bookkeeping/counting, leading to
counter underflows on MTYPE_TMP.
Signed-off-by: Lou Berger <lberger@labn.net> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
(cherry picked from commit 672900382d47137638086bd8351b2678f589a546)
Donald Sharp [Thu, 7 Jan 2016 14:33:28 +0000 (09:33 -0500)]
lib, bgpd: Fixup afi_t to be an enum and cleanup zebra.h
This code change does two things:
1) Removes ZEBRA_AFI_XXX #defines since they were redundant information
2) Switches afi_t to an enumerated type so that the compiler
can do a bit more compile time checking.
Paul Jakma [Tue, 1 Dec 2015 14:32:11 +0000 (14:32 +0000)]
bgpd: bgp_scan shouldn't queue up route_nodes with no routes for processing
* bgp_nexthop.c: (bgp_scan) There is little point queueing an rn with no routing
information for processing.
* bgp_route.c: (bgp_process) Do nothing on rn's with no routes. Add an assert
for now, to try catch any other cases, but prob should be removed.
(bgp_best_selection) rn with no routes == finish early.
Paul Jakma [Wed, 25 Nov 2015 17:14:35 +0000 (17:14 +0000)]
bgpd: Check capability falls on right multiple of size, where possible.
* bgp_open.c: (cap_modsizes) Table of multiple a capability's data size
should fall on, if applicable.
(bgp_capability_parse) Check the header lengthcap_modsizes should fall on.
Inspiration from Cumulus bgpd-capability-cleanup.patch patch, with a
slightly different approach.
Timo Teräs [Thu, 22 Oct 2015 08:35:18 +0000 (11:35 +0300)]
bgpd: update rtt on soft clear
rtt is calculated dynamically by the kernel. Refresh it on
soft clear.
Fixes: ef757700d0 "bgpd: allow using rtt in route-map's set metric" Signed-off-by: Timo Teräs <timo.teras@iki.fi>
(cherry picked from commit 5a2a1ec18c89daec5de6690a9b0f47c0d11a0f2d)
Timo Teräs [Thu, 22 Oct 2015 08:35:17 +0000 (11:35 +0300)]
bgpd: check rtt later after the real peer is known
OPEN message handler moves the connection from the temporary
"struct peer" (used to accept it) to the real "struct peer" based
on the configuration. RTT needs to be updated only to the real
struct peer, and this patch moves the RTT query to point where
realpeer is known.
Fixes: ef757700d0 "bgpd: allow using rtt in route-map's set metric" Signed-off-by: Timo Teräs <timo.teras@iki.fi>
(cherry picked from commit 0edba8b6ad9c83fa0a3cc58765fe9f123f4109ac)
Timo Teräs [Wed, 24 Jun 2015 12:27:21 +0000 (15:27 +0300)]
bgpd: Make bgp_info_cmp robust to paths that do not have su_remote info
My original su_remote == NULL check is not correct. It seems that
* bgp_route.c: (bgp_info_cmp) Some bgp_info is compared with su_remote=NULL
and it's supposed to be perfectly legal. E.g. configured subnet announces
("network a.b.c.d/n"). Ensure bgp_info_cmp is robust if such a path gets
as far as the neighbour address comparison step.
Timo Teräs [Fri, 22 May 2015 10:41:00 +0000 (13:41 +0300)]
zebra: simplify redistribution code
Merge the conditionals as one to avoid code duplication.
Signed-off-by: Timo Teräs <timo.teras@iki.fi> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
(cherry picked from commit f85592e05ae6463727433893e61afd1081fcf7e0)
Timo Teräs [Wed, 29 Apr 2015 06:43:04 +0000 (09:43 +0300)]
bgpd: allow using rtt in route-map's set metric
Useful when the BGP neighbors are over tunnels that have large
differences in geographic distances and RTTs. Especially useful
for DMVPN setups to allow preferring closes hub.
The parameter is added as new alias command as otherwise it seems
the command parser is not able to match it properly (it seems
merging is done for the various 'set metric' route-map objects in
different routing engines). For same reason also they are listed
as three separate options: optional +/- seems not possibly easily.
Related research papers:
http://www.pps.univ-paris-diderot.fr/~jch/research/delay-based.pdf
http://arxiv.org/pdf/1309.0632.pdf
Paper on similar extension to Babel:
http://www.pps.univ-paris-diderot.fr/~jch/research/rapport-jonglez-2013.pdf
Signed-off-by: Timo Teräs <timo.teras@iki.fi> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
(cherry picked from commit ef757700d0fd51dc0b46df9d3631208919f9b779)
Use common code to parse, validate and adjust the route-map
objects that contain a simple integer value. This also allows
compiling the add/sub format metric object.
Signed-off-by: Timo Teräs <timo.teras@iki.fi> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
(cherry picked from commit 38f22ab07c8e1f0f12e23c2e2d0a3e1f71bef695)
David Lamparter [Tue, 3 Mar 2015 07:51:53 +0000 (08:51 +0100)]
*: use void * for printing pointers
On higher warning levels, compilers expect %p printf arguments to be
void *. Since format string / argument warnings can be useful
otherwise, let's get rid of this noise by sprinkling casts to void *
over printf calls.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
David Lamparter [Tue, 3 Mar 2015 07:48:11 +0000 (08:48 +0100)]
*: use long long to print time_t
Since we can't assume time_t to be long, int, or even long long, this
consistently uses %lld/long long (or %llu/unsigned long long in a few
cases) to print time_t/susecond_t values. This should fix a bunch of
warnings, on NetBSD in particular.
(Unfortunately, there seems to be no "PRId64" style printing macro for
time_t...)
* linklist.{c,h}: (listnode_move_to_tail) new unction to move a
listnode to tail of list.
* ospf_packet.c: (ospf_write) remove debug that seemed to be mostly covered
by existing debug.
Use listnode_move_to_tail to just move the list node to the end of the
tail, rather than freeing the one to hand and allocing a new one.
Donald Sharp [Wed, 10 Jun 2015 00:22:42 +0000 (20:22 -0400)]
Add code to extract.pl.in to prevent further cli function overwrites
Currently extract.pl.in is used to build the vtysh cli. When two
different cli's collide with the same command name, the original
cli is never called, because it is dropped. This code notes the
silent drop and tracks the number of drops. If they change then
the code will fail the build. The current number of drops was
figured out by running extract.pl and counting up the drops
then adding code to compare the numbers returned.
If you have added to the problem, the solution is to fix your cli
command to not stomp on someone else's command. If you have removed
a stomp, safely modify extract.pl.in as part of your commit.
Signed-off-by: Donald Sharp <sharpd at cumulusnetworks.com> Acked-by: Vincent Jardin <vincent.jardin@6wind.com>
David Lamparter [Mon, 18 Nov 2013 22:04:27 +0000 (23:04 +0100)]
lib: unstupidify thread debug information
the library's thread scheduling functions keep track of the thread
function's name, so far so good. However, copying the compiler-provided
constant into a buffer inside the thread structure is plain useless.
Also, strip_funcname() was trying to support something that never
happens.
Instead, let's use some bytes here to track where threads are scheduled
from. Another commit will print that information on crashes.
Ripping out useless stuff: -64 bytes in the thread structure
Re-add as const ptr: +8 bytes
Extra debug info: +12 bytes
David Lamparter [Tue, 3 Mar 2015 07:52:22 +0000 (08:52 +0100)]
bgpd: don't use #ifdef inside macro args
Using #ifdef inside preprocessor macro argument lists is not guaranteed
to work. In reality it mostly does, but we don't need these ifdefs for
HAVE_IPV6 anymore, so let's get rid of the warning nonetheless.
Milan Kocian [Mon, 1 Dec 2014 12:48:25 +0000 (12:48 +0000)]
bgpd: fix negative values in output
Negative value in output of ecommunities (and as numbers)
seems odd :-). This patch fixes it. And add minor formating
modification, better for big as numbers.
Signed-off-by: Milan Kocian <milon@wq.cz> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
(cherry picked from commit cb4fc59c8a0f9df81109d38acbeaab5627e361f5)
David Lamparter [Thu, 14 May 2015 12:47:05 +0000 (14:47 +0200)]
tests: add testcli reference in/out & do DejaGNU
This adds reference in & output for the previously added testcli tool,
to check basic CLI parsing/help functions. Unlike "testcommands", this
one doesn't depend on compile-time system details.
David Lamparter [Tue, 5 May 2015 09:04:59 +0000 (11:04 +0200)]
tests: add CLI dummy command-exec tool
This adds some common CLI testtool code as well as a tool that has a
bunch of commands to be poked for their correct processing.
The tool doesn't work correctly from a script at stdin at this point
because the vty code will throw away all buffered when it sees EOF, so
the tail end of the input file is lost.
David Lamparter [Sat, 22 Nov 2014 18:31:33 +0000 (10:31 -0800)]
build: remove --disable-ipv6
Building with IPv6 disabled tends to break rather often and sprinkles
ugly #ifdefs around the code. All that only to support systems where
the C library doesn't have IPv6 capability.
The year now being 2015, if this is a problem the thing to fix is the C
library.
The implication of this patch is that future patches need not care about
HAVE_IPV6 = 0 and may remove ifdefs gratuitously. This patch doesn't
remove these ifdefs to not create unneccessary churn.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org> Acked-by: Paul Jakma <paul@jakma.org>
David Lamparter [Sat, 28 Jun 2014 20:07:41 +0000 (22:07 +0200)]
build: remove --enable-solaris parameter
This switch controlled descending into the solaris/ subdirectory, which
contains package descriptions and init scripts. If they're not
appropriate, they'd better be removed outright.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org> Acked-by: Greg Troxel <gdt@ir.bbn.com> Acked-by: Feng Lu <lu.feng@6wind.com> Acked-by: Paul Jakma <paul@jakma.org>
David Lamparter [Sat, 28 Jun 2014 19:26:36 +0000 (21:26 +0200)]
zebra, ripngd: remove ::/64 special-casing
In the 90ies, IPv4 was believed to exist within IPv6, with some kernels
implementing this belief in code... Our code here is keyed to "#ifdef
LINUX", yet no Linux from the past 10 years had this, making the code
completely useless.
FreeBSD 10.0 does in fact have a "::/96 via ::1 dev lo0 reject" route.
IMHO we shouldn't mess with that, the admin can filter as neccessary
anyway.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org> Acked-by: Greg Troxel <gdt@ir.bbn.com> Acked-by: Feng Lu <lu.feng@6wind.com>
[DL: slightly adjusted commit message to remove misunderstanding] Acked-by: Paul Jakma <paul@jakma.org>
Donald Sharp [Wed, 1 Jun 2016 19:31:02 +0000 (15:31 -0400)]
pimd: Add support for displaying ip mroute
When you enter a static mroute under an interface
the 'show run' is not displaying this information.
Add code to allow this.
Ticket: CM-11257 Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com> Reviewed-by: Daniel Walton <dwalton@cumulusnetworks.com> Reviewed-by: Don Slice <dslice@cumulusnetworks.com>
Donald Sharp [Wed, 1 Jun 2016 01:27:48 +0000 (21:27 -0400)]
pimd: Allow command zclient to find it's data
pim has two zclient sockets to zebra. One
is used exclusively to do mrib lookups. The
other is to do the normal day to day communication
between pim and zebra. With the change
to the zebra api to send up all data to all
sockets this caused the mrib lookup socket
to accumulate data in between mrib lookups.
So if at some point in time we get upcoming
data but no mrib lookups modify the code
to find the mrib lookup it is looking for.
Long term we need to figure something else out
but this change will get us moving forward again.
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com> Reviewed-by: Daniel Walton <dwalton@cumulusnetworks.com>
projects / mirror_frr.git / log