Paul Jakma [Fri, 8 Apr 2011 11:44:43 +0000 (12:44 +0100)]
general: remove inline qualifiers and move in-header functions to objects
* (general) Move functions in headers into files, to be compiled into
shared object files. Remove inline qualifier from functions. Let the
compiler do the work.
When the same ip address is used on several interfaces,
and one of them gets deleted (or equivalent: set to down and
then address removed), rib_delete_ipv[46] will also remove
the connected route from other interfaces.
rib_delete_ipv[46] is called twice when an interface is
deleted:
- for the "ifdown" event
- for the address removal
(note: this may be specific to the netlink interface of linux)
The second call does not find the connected route to that same
ifindex anymore, but deletes similar connected routes to any
other ifindex instead.
Reason: the ifindex check is on the same level as the check
for ZEBRA_ROUTE_CONNECT/NEXTHOP_TYPE_IFINDEX. If everything
matches except for the ifindex, the "else" part (intended for
different route types) is executed, thus removing the route
from the wrong interface.
fix: move ifindex check inside the "then" part of the check
for ZEBRA_ROUTE_CONNECT/NEXTHOP_TYPE_IFINDEX. Now connected
routes to other ifindexes will not spill over to the "else"
part for different route types anymore.
* fix disagreement with C99 in zlog_debug calls
Format specifier in some zlog_debug calls for size_t values
was changed in order to C99 ('%u' -> '%zu').
* fix -Wsign-compare warnings
Type of return value of ospf6_packet_max() was changed.
Denis Ovsienko [Sun, 18 Dec 2011 12:27:02 +0000 (16:27 +0400)]
ospfd: fix bug in NSSA ABR status check
* ospf_abr.c
* ospf_abr_nssa_am_elected(): feed "best" instead of "address of best"
into IPV4_ADDR_CMP(), because "best" is a pointer; also, mean s_addr
field of the structures to get better typed pointers
Denis Ovsienko [Sun, 18 Dec 2011 11:40:17 +0000 (15:40 +0400)]
lib: address type-punned pointers in prefix_same()
IPV4_ADDR_SAME() wasn't passed the right union member, this could cause
a bug due to strict-aliasing. IPV6_ADDR_SAME() case got its fix before
the error could be created by macro upgrade.
Denis Ovsienko [Sat, 17 Dec 2011 15:39:30 +0000 (19:39 +0400)]
bgpd: justify checks for IPv4 class D/E
* lib/prefix.h
* IPV4_CLASS_DE(): make consistent with counterpart macros
* bgp_packet.c
* bgp_open_receive(): test using macro instead of ">="
* bgp_route.c
* bgp_update_rsclient(): idem
* bgp_update_main(): idem
ZEBRA_HELLO message is used by routing daemons to inform zebra
what type of routes daemon will be announcing to zebra. Also
zebra uses route_type_oaths array to track which daemon announces
which protocol. Zebra mops up routes if daemon didn't for some
reason.
Denis Ovsienko [Sun, 4 Dec 2011 19:21:55 +0000 (23:21 +0400)]
lib: optimize ip_masklen()
The new implementation makes use of a 64KB mapping table, which makes it
possible to compute masklen faster and with constant execution time. The
map also allows for additional version of the function, which can detect
errors in input argument.
The previous implementation had a variable cost of execution, which
depended on masklen in a non-linear manner, and at its worst (/31) was
4 times slower, than the new implementation. The only case of old
function just slightly outperforming the new one is /0, which is of
little practical interest.
(This patch was modified to leave calls to stream_getl() in place, they
are necessary for the stream's internal pointer to advance to the
correct position. -- Denis)
Signed-off-by: Denis Ovsienko <infrastation@yandex.ru>
Fix gcc warnings about varables that are set but never used.
Denis Ovsienko [Fri, 14 Oct 2011 17:59:58 +0000 (21:59 +0400)]
ospf6d: decode message type with LOOKUP()
* ospf6_message.h
* OSPF6_MESSAGE_TYPE_CANONICAL(): dismiss
* OSPF6_MESSAGE_TYPE_NAME(): dismiss
* ospf6_message.c
* ospf6_message_type_str: rewrite as a message list, add max value
* ospf6_packet_examin(): update to use LOOKUP()
* ospf6_receive(): idem
* ospf6_send(): idem
Denis Ovsienko [Sat, 8 Oct 2011 13:22:45 +0000 (17:22 +0400)]
ospf6d: make some old sizing checks assertions
As long as ospf6_packet_examin() is now the single checkpoint for
received packets, most of the old checks performed elsewhere can
be converted into assert() constructs. Malformed input data at
respective points can be attributed solely to a programming error,
not a malformed packet.
David Lamparter [Sun, 4 Oct 2009 14:21:49 +0000 (16:21 +0200)]
lib: add sockopt helper for setting IPV6_V6ONLY and use it
getaddrinfo returns a list of socket parameters for listening. it
will contain both IPv4 and IPv6 listening sockets. unless we use
IPV6_V6ONLY on the IPv6 ones, only the socket listed first will
work. if the IPv4 one came first, the IPv6 one would get an
"Address in use" error.
this functionality was already present for bgpd and its listening
sockets. as it is needed for vtys as well, make it a common helper.
David Lamparter [Tue, 15 Sep 2009 23:52:42 +0000 (01:52 +0200)]
lib: put route_types.txt to real use
this replaces most occurences of routing protocol lists by preprocessor
defines from route_types.h. the latter is autogenerated from
route_types.txt by a perl script (previously awk). adding a routing
protocol now is mostly a matter of changing route_types.txt and log.c.
David Lamparter [Wed, 26 Aug 2009 22:28:28 +0000 (00:28 +0200)]
rib: default distance value for new protocols (v2)
adding protocols and forgetting to update zebra_rib.c currently causes
a beyond end of array access for the default distance value. fix by
making 150 the default distance for unknown routing protocols.
150 is higher than all other IGPs but lower than iBGP (200)
Tom Goff [Wed, 25 Nov 2009 20:36:06 +0000 (20:36 +0000)]
zebra: set the interface link-layer socket address
* kernel_socket.c: (ifm_read) The reorganization from commit 6fe70d1b35c189cb1e488b2c26551ba7baac6148 removed setting the
interface link-layer socket address structure on routing socket
platforms. This restores saving the link-layer information because
it might be used elsewhere: rtm_write() when no gateway is given,
if_dump_vty() and rtadv_send_packet().
Ulrich Weber [Wed, 16 Nov 2011 15:32:12 +0000 (19:32 +0400)]
bgp: use VTY_GET_INTEGER_RANGE() in bgp_clear()
Second patch replaces "VTY_GET_LONG ("AS", as_ul, arg);"
by "VTY_GET_INTEGER_RANGE ("AS", as, arg, 1, BGP_AS4_MAX);"
as done in all other code, which parses AS numbers.
bgpd: store nexthop info for redistributed IPV6 routes
BGP was ignoring nexthop info for static and other redistributed
routes for IPv6. Build extra attribute info to store the nexthop.
See also:
https://bugzilla.vyatta.com/show_bug.cgi?id=6073
ospfd: avoid redundant lookup in ospf_redistribute_withdraw
The old algorithim looked up every node twice as it withdrew
the entry. It looks up entry once in redistribute_withdraw, then
looks it up again info_delete. Use result of first lookup
to do the free directly.
This may explain the slow performance observed in
https://bugzilla.vyatta.com/show_bug.cgi?id=4421
lib: call filter delete hook before freeing access list
The delete_hook was being run after calling access list delete function.
This would cause ospf to dereference a NULL, in ospf_filter_update
because 'access->name' was already freed.
See also:
https://bugzilla.vyatta.com/show_bug.cgi?id=7654
ospf6d: check MTU with message header size in mind
* ospf6_message.c: (ospf6_packet_max): new function, return maximum IPv6
payload on an interface; (ospf6_hello_send, ospf6_dbdesc_send,
ospf6_dbdesc_send_newone, ospf6_lsreq_send, ospf6_lsupdate_send_neighbor,
ospf6_lsupdate_send_interface, ospf6_lsack_send_neighbor,
ospf6_lsack_send_interface): compare message size with the maximum
payload instead of the MTU.
Chris Luke [Tue, 18 Oct 2011 13:26:51 +0000 (17:26 +0400)]
bgpd: justify capabilities for TOS setting
To set the TOS bits on TCP connections, platforms that restrict
capabilities need the priv level to be raised before the sockopt
is set, and this requires the ZCAP_NET_ADMIN priv.
* bgp_main.c: update _caps_p to include ZCAP_NET_ADMIN
* bgp_network.c
* bgp_connect(): request ZPRIVS_RAISE/ZPRIVS_LOWER
* bgp_listener(): request ZPRIVS_RAISE earlier
Denis Ovsienko [Tue, 18 Oct 2011 09:55:07 +0000 (13:55 +0400)]
bgpd: use bgp_attr_malformed()
Some of the recent attribute flags/length checks copied from QRE use
bgp_notify_send_with_data() directly, but master branch assumes
using bgp_attr_malformed().
Denis Ovsienko [Mon, 10 Oct 2011 17:08:33 +0000 (21:08 +0400)]
bgpd: add flag checks for MP_(UN)REACH_NLRI
* bgp_attr.[ch]
* bgp_mp_reach_parse(): add extra arguments and a uniform flag
check block
* bgp_mp_unreach_parse(): idem
* bgp_attr_parse(): provide extra arguments
* bgp_mp_attr_test.c
* parse_test(): justify respective calls
* bgp_attr.c
* bgp_attr_cluster_list(): accept extra argument, add checks for
"optional", "transitive" and "partial" bits, log each error
condition independently
* bgp_attr_parse(): provide extra arguments
* bgp_attr.c
* bgp_attr_originator_id(): accept extra argument, add checks for
"optional", "transitive" and "partial" bits, log each error
condition independently
* bgp_attr_parse(): provide extra arguments
IPv6 supports the same concept of differentiated service for routing
protocols as IPv4, but like too many things, the standards committee
decided that having two names for the same thing wasn't good enough and
introduced a third more generic term transport class.
The socket option to set transport class works the same as IPv4, but the
arguments are different.
* lib/sockopt.[ch]
* setsockopt_ipv6_tclass(): new function
* bgpd/bgp_network.c
* bgp_connect(): set socket option
* bgp_listener(): set socket option
* ospf6d/ospf6_network.c
* ospf6_set_transport_class(): new function
* ospf6_serv_sock(): set socket option
* ripngd/ripngd.c
* ripng_make_socket(): set socket option
Denis Ovsienko [Wed, 12 Oct 2011 09:54:21 +0000 (13:54 +0400)]
bgpd: fix regression in improved attr flag checks
Commit 2febf323411c1aed9d7694898f852ce2ef36a7e5 assumed every flag
bit except optional/transitive/partial unset, which at times could
not be true for "extended length" bit.
* bgp_attr.c
* bgp_attr_origin(): exclude BGP_ATTR_FLAG_EXTLEN from comparison
* bgp_attr_nexthop(): idem
* bgp_attr_med(): idem
* bgp_attr_local_pref(): idem
* bgp_attr_atomic(): idem
Peter Pentchev [Mon, 12 Sep 2011 12:30:31 +0000 (16:30 +0400)]
bgpd: fix parsing of graceful restart cap. (#663)
"While setting up a testbed, I ran across a little problem in the
parsing of the "graceful restart" BGP capability that resulted in
Quagga not actually activating it for the peer in question - when
the peer sent a single AFI/SAFI block."
* bgp_open.c
* bgp_capability_restart(): actually process the last AFI/SAFI block
Denis Ovsienko [Sat, 8 Oct 2011 09:54:48 +0000 (13:54 +0400)]
bgpd: improve attr length error handling (BZ#679)
* bgp_attr.c
* bgp_attr_parse(): provide extra argument to bgp_attr_aggregator()
* bgp_attr_local_pref(): use bgp_notify_send_with_data()
* bgp_attr_atomic(): idem
* bgp_attr_aggregator(): idem
Do not check each of the Optional/Transitive/Partial attribute
flag bits, when their only valid combination is known in advance,
but still perform bit-deep error message logging. This change
assumes unused (low-order) 4 bits of the flag octet cleared.
* bgp_attr.c
* bgp_attr_origin(): rewrite check
* bgp_attr_nexthop(): idem
* bgp_attr_med(): idem
* bgp_attr_local_pref(): idem
* bgp_attr_atomic(): idem
ORIGIN handling function used to have "partial" bit check and recent
commits added it for NEXT_HOP, MULTI_EXIT_DISC and ATOMIC_AGGREGATE
cases. This commit adds "partial" check for AS_PATH and LOCAL_PREF
cases, which should leave attributes 1 through 6 inclusive completely
covered with attribute flags checks.
* bgp_attr.c
* bgp_attr_origin(): use bit-by-bit checks for better diagnostics
* bgp_attr_aspath(): add flag check
* bgp_attr_local_pref(): idem
* lib/prefix.h
* IPV4_CLASS_DE(): new helper macro
* bgp_attr.c
* bgp_attr_nexthop(): add check for "partial" bit, refresh flag error
reporting, explain meaning of RFC4271 section 6.3 and implement it
lib: provide more information in case of failed LOOKUP.
* log.[ch]
* mes_lookup: add a parameter with the name of the message list, print
the name in case of failure.
* LOOKUP macro: pass the name of the message list.