| Age | Commit message (Collapse) | Author |
|---|
|
the OSPF API-server (exporting the LSDB and allowing announcement of
Opaque-LSAs) writes past the end of fixed on-stack buffers. This leads
to an exploitable stack overflow.
For this condition to occur, the following two conditions must be true:
- Quagga is configured with --enable-opaque-lsa
- ospfd is started with the "-a" command line option
If either of these does not hold, the relevant code is not executed and
the issue does not get triggered.
Since the issue occurs on receiving large LSAs (larger than 1488 bytes),
it is possible for this to happen during normal operation of a network.
In particular, if there is an OSPF router with a large number of
interfaces, the Router-LSA of that router may exceed 1488 bytes and
trigger this, leading to an ospfd crash.
For an attacker to exploit this, s/he must be able to inject valid LSAs
into the OSPF domain. Any best-practice protection measure (using
crypto authentication, restricting OSPF to internal interfaces, packet
filtering protocol 89, etc.) will prevent exploitation. On top of that,
remote (not on an OSPF-speaking network segment) attackers will have
difficulties bringing up the adjacency needed to inject a LSA.
This patch only performs minimal changes to remove the possibility of a
stack overrun. The OSPF API in general is quite ugly and needs a
rewrite.
Reported-by: Ricky Charlet <ricky.charlet@hp.com>
Cc: Florian Weimer <fweimer@redhat.com>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
Note that list of prereq versions is out of date.
Add DejaGnu for testing.
Change references to CVS to git.
Signed-off-by: Greg Troxel <gdt@ir.bbn.com>
|
|
Use check_PROGRAMS instead of noinst_PROGRAMS in tests/Makefile.am
to build the tests only when make check is actually run.
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
|
|
The export statement is specific to GNU make and breaks the build
with BSD make. I couldn't observe any difference in behaviour
between having the export present and absent, therefore, just remove it.
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
|
|
An ospf router should accept a new maxage LSA into its lsdb if it has any
neighbors in state Exchange or Loading. ospfd would however only account
for neighbors on the same interface which does not seem to be a valid
optimization.
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Signed-off-by: David Lamparter <equinox@diac24.net>
|
|
ospf_maxage_lsa_remover whould check whether to yield,
but run on anyway.
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
Signed-off-by: Joachim Nilsson <troglobit@gmail.com>
Signed-off-by: David Lamparter <equinox@diac24.net>
|
|
This is work in progress at Cumulus Networks.
|
|
Wrap the few libzebra test programs we have up for DejaGNU.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
this just wraps the existing test programs in expect wrappers that make
their results usable to DejaGNU.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
DejaGNU seems to be the 'standard' GNU test framework (which by itself
doesn't say much), but it seems relatively usable and the "remote
system" capabilities might come in handy for virtualisation-based tests
for kernel interactions or something.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
commit c81ee5c... "ospfd: Optimize and improve SPF nexthop calculation"
subtly changed semantics of routes calculated over pointopoint links by
removing the nexthop IP address and instead using an ifindex route.
This breaks calculation of AS-Ext routes with a forwarding address since
in ospf_ase_complete_direct_routes() this will be hit:
if (op->nexthop.s_addr == 0)
op->nexthop.s_addr = nexthop.s_addr;
thus turning the route unusable by having an invalid nexthop.
Fix by restoring the nexthop IP on routes over PtP links. This also
allows running multi-access (Ethernet) interfaces in PtP mode again.
This bug is a regression against 0.99.21 and only present in 0.99.22.
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
[patch description and code comments rewritten]
Acked-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Acked-by: James Li <jli@cumulusnetworks.com>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
Register the vtysh socket in Vvty_serv_thread so it will be
correctly closed on vty_reset instead of being leaked.
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
ripd had a check to restrict metric 0 to only directly connected routes.
This check was implemented by checking against Connected as route type.
This is, however, incorrect -- all routes that directly use an interface
without a nexthop should be treated as directly connected and passed off
with metric 0.
ripngd does not posess such a check and was not touched.
Reported-by: Sean Fulton <sean@gcnpublishing.com>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
While the actual build failures have been fixed independently by
d1d3ac9 "build: reorder libraries to address linker error", libzebra
still does not reference libcap. This will lead to more build failures
if someone else tries to use libzebra and doesn't add libcap.
Let's just add libcap here and be done with it.
I've not added libcap to the _DEPENDENCIES variable above since libcap
is a system library. Actually, the whole _DEPENDENCIES thing is rather
fishy; automake automatically sets _DEPENDENCIES from _LIBADD. For the
sake of not breaking stuff that works (especially since most autotools
stuff is arcane magic), I'm leaving it alone...
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
If configured without opaque LSA support, the old code would incorrectly
associate type 5 LSAs with an area.
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
bgpd tests don't compile or run with --disable-bgpd, let's catch this in
the Makefile.
Reported-by: Joachim Nilsson <troglobit@gmail.com>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
the original version of this had issues with tagless repositories; to
fix that I removed the "-g" part from one of the regexes. I then failed
to add those 2 characters back, leading to version numbers like
"0.99.220123456" instead of "0.99.22-ga123456". Let's put the "-g"
back...
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
commit 4afa50b added few lines that are syntactically incorrect
with leading plus sign.
Cc: Denis Ovsienko <infrastation@yandex.ru>
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
The homenet OSPFv3 extensions are not only relevant TODO items, but also
suitable for GSoC students.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
the TODO was last touched in 2006. This is a first pass at cleaning it
up, motivated primarily by the need for an up-to-date idea list for the
Google Summer of Code 2013.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
INSTALL.quagga.tex:
Given the statement that it's viewed as a bug if quagga doesn't
build on OS versions on the list, prune the list to the set for
which there would be near-universal agreement that it's a bug.
Clarify that the response to a system on the list not building might
be dropping it from the list. (Time marches on, and these lists are
not necessarily maintained. As an example, the comment saying
FreeBSD4 support was iffy is now 6 years old.)
Delete old discussion of ancient texinfo.
Delete discussion of NetBSD versions before 4 (as no longer relevant).
|
|
Dynamically grow the hash table index if the chains get too long.
If expansion doesn't help keep chain length short, then stop expanding,
to avoid bad behavior if there is a poor hash function.
Not a new idea, based on concepts in uthash.
Depends on my previous patch to restrict hash to power of 2.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
[profiling results: sum of cycles spent in hash_get/jhash with RIPE RIS
test data (single simple BGP peer) improved to 69% of previously spent]
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
By forcing the hash table size to be a power of 2, a potentially
expensive divide can be replaced by a mask operation. Almost all
usage of the hash table was using default size of 1024. Only places
with different size was thread library (1011) and bgp aspath.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
|
|
both Quagga and RPM have moved a bit since this was last touched.
Should now work again on CentOS 5 and 6.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
The if ($file =~ /lib/) path matching logic is supposed to
match Quagga's lib directory only but will match all path
having lib in it such as /var/lib/jenkins/quagga/...
Fix by matching both lib and file: lib/keychain.c etc.
Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
This patch resolves the significance of order of group and password
statements.
It prevents passwords from being lost in cases where all
three conditions apply:
1. the peer is member of a group with or without group password
2. the peer has an individual password set
3. the peer is added to a group within an address-family ipv6
section
In addition this patch prevents the same issue in cases, where an IPv4
peer's password is set first and the peer is added to a group
afterwards.
Adding a peer to a group cancels his individual password. Without ipv6
this is not a problem, because choosing the right order of config
statements will do (set password only after adding peer to group).
When adding the peer to a group within the address-family
section, his password is definitely lost. The same workaround (ie.
setting the password after the address-family section) can not be used,
because "show run" will print the configuration statements in the wrong
order.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
so net.core.rmem_max must not be adjusted. Requires
linux kernel >= 2.6.14, falls back to SO_RCVBUF on error
Signed-off-by: Ulrich Weber <ulrich.weber@sophos.com>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
|
|
Not only was there a minor typo in the "pragma weak" preprocessor
checks, but also were the tests not behaving as needed - they only
indicated support for the /first/ method of implementing weak aliases,
which on Linux is __attribute__ and not #pragma.
* m4/ax_sys_weak_alias.m4: set defines for _all_ weak alias methods
* zebra/kernel_null.c: fix typo
Cc: Doug VanLeuven <roamdad@sonic.net>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
in particular,
- add IS-IS to some listings
- list Solaris & OSX as "some work required"
- remove OS version numbers. We have no base to specify any of them.
- list supported C compilers (gcc, clang, icc)
- cut the Quagga 2.0 stuff that promises QoS and firewall functionality
|
|
|
|
Commit 8692c50652 introduced a bug where bgpd would crash on
soft-reconfiguration.
This happens e.g. when there are filtered unicast routes because
rn->info is NULL in that case, which the code did not account for.
Reported-by: Paweł Staszewski <pstaszewski@itcare.pl>
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
turns out, bgp_mp_reach_parse really doesn't like getting garbage
attribute input. In particular, attr->extra better be NULL or we
merrily go trample random places (like our stack).
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
NB: these tests test for current implementation state, not for RFC
conformance. In particular, behaviour with confederations in AS4_PATH
as well as reconcilation of short AS_PATH + AS4_PATH is currently NOT
conforming to RFC 4893/6793.
* tests/aspath_test.c: add capability to put both AS4_PATH & AS_PATH,
add test for AS4_PATH w/o AS_PATH, update confederation test
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
commit fe9bb64... "bgpd: CVE-2012-1820, DoS in bgp_capability_orf()"
made the length test in bgp_capability_orf_entry() stricter and is now
causing us to refuse (with CEASE) ORF capabilites carrying any excess
data. This does not conform to the robustness principle as laid out by
RFC1122 ("be liberal in what you accept").
Even worse, RFC5291 is quite unclear on how to use the ORF capability
with multiple AFI/SAFIs. It can be interpreted as either "use one
instance, stuff everything in" but also as "use multiple instances".
So, if not for applying robustness, we end up clearing sessions from
implementations going by the former interpretation. (or if anyone dares
add a byte of padding...)
Cc: Denis Ovsienko <infrastation@yandex.ru>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
bgp_attr_munge_as4_attrs would previously try to reintegrate an AS4_PATH
with a NULL AS_PATH, leading to a rather nasty SEGV. Let's go by
RFC6793 and treat missing AS_PATH as 0-length AS_PATH, which in turn
means discarding the AS4_PATH.
[NB: we don't actually stick to the actual rule, which is discarding
AS4_PATH if it's longer than AS_PATH; indeed we should probably fix that
too]
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
this is not a full release version, so neither release notes nor
documentation are updated yet. Also, signing the tag with my private
GPG key instead of the Quagga one.
|
|
Incorporate a patch by Svetozar Mihailov which implements
default-originate route-maps to behave as expected, i.e. allowing
the default route to be advertised conditionally, depending on a
criterion given by the route-map.
I am aware that the performance attributes of the following implementation
are far from optimal. However, this affects only code paths belonging to
a feature that is broken without this patch, therefore, it seems reasonable
to me to have this in the mainline for now.
Cc: Svetozar Mihailov <quagga@j.zarhi.com>
Reported-by: Sébastien Cramatte <scramatte@gmail.com>
Signed-off-by: Christian Franke <chris@opensourcerouting.org>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
This pushes out the NOTIFY message before closing a connection.
Previously, the TCP_CORK bandwidth optimization code caused NOTIFY
messages to disappear prior to when the connection is closed.
* bgpd/bgp_packet.c: unset CORK, set NODELAY, and replace
writen() by more correct write()
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
This reverts commit b07458a055493dd37cb955ae90f11ae8bc334d3a.
On second thought, the right way to do this is with rename(), not by
introducing a lock that can potentially even stall bgpd.
Reported-by: Christian Franke <chris@opensourcerouting.org>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
Quagga makes bad assumptions about sockaddr_dl (on NetBSD, but possibly
on other systems as well). Particularly, sizeof(struct sockaddr_dl)
returns a size that does not include the full sdl_data field, leading to
not enough data being copied. This breaks IPv6 RAs in particular, as
a broken mac address from sockaddr_dl will be included in the packets.
From: Matthias-Christian Ott <ott@mirix.org>
Tested-by: Uwe Toenjes <6bone@6bone.informatik.uni-leipzig.de>
[further simplified + more comments]
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
|
|
This script compiles Quagga in a variety of configurations and
optionally with LLVM and ICC (if those are installed).
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
Newer MacOSX versions have support for both IPv6 advanced socket API
RFCs (2292 and 3542) switchable in compile time, but neither of these
is default for some strange reason. RFC3542 will be default in future,
but for now we have to declare that we want to use the RFC3542 API
before including <netinet/in.h>.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
AM_CONFIG_HEADER has been deprecated for many years and is removed
completely from automake 1.13.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
Keep data flowing, uncork after each BGP_WRITE_PACKET_MAX.
This makes TCP send data sooner, since thread may not be scheduled
again for a a longish time because of new UPDATE's coming in.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.org>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
The readtime value is for diagnostic, and doesn't have to be highly
accurate. This also fixes a problem where the readtime was being measured
with system clock, but the peer_uptime() was comparing with bgp_clock.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
Measuring the resource usage of threads is moderately expensive
since it requires doing an additional system call everytime a
thread context switches. Make it possible to disable this with
a configuration option.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
|
The flag bit BGP_NODE_PROCESS_SCHEDULED is checked but never set.
This causes route node to be scheduled multiple times under load.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
|
