| Age | Commit message (Collapse) | Author |
|---|
|
|
|
catching this.
|
|
* kernel_socket.c (kernel_read): Look up interfaces by index
first, so that state changes which do not include a sockaddr_dl
now work. Add many sanity checks. In
particular, do not assume that a sockaddr_dl follows a message
without checking the ifm_addrs flags, and do not trust the length
in a sockaddr_dl. Add/clarify many comments.
|
|
- add privs support
- use misc quagga's definitions
- make it compile"able"
- fix segfault cases related to hostname()
- add debug isis xxx command
This patch has been approved by Paul Jakma.
|
|
fixes rest of bug #62
|
|
files, defaulting to sysconfdir (matching previous behavior).
This is needed to support (cleanly) NetBSD pkgsrc, which requires that
example config files go in $(prefix)/share/examples/pkgname, rather
than in $(prefix)/etc/pkgname.
|
|
* zebra/rt_netlink.c: netlink_parse_info() ignore messages which are
not from kernel. Reported to RH by Herbert Xu. See
http://rhn.redhat.com/errata/RHSA-2003-307.html and CAN-2003-0858.
|
|
* zebra/zebra_rib.c: Revert patch (dating from zebra.org) which
caused zebra to read all routes in all tables, rather than just
the main table. See [quagga-dev 280].
|
|
* zebra/zebra_rib.c: (nexthop_active_update) Check for multipath
limit when setting changed flag to avoid spurious changes.
(static_install_ipv{4,6}) dont uninstall by default, might not be
required - avoid spurious uninstalls.
(static_uninstall_ipv{4,6}) only uninstall the route if its
actually FIB route.
|
|
* configure.ac: Check for fcntl()
* {bgpd,ospf,ospf6d,ripd,ripngd}/Makefile.am: Install conf file via
regular automake means, not magic install incantations, see
bug #38.
* lib/Makefile.am: install the headers, needed to link libzebra.a
(and hence libospf.a, OSPF-API, etc.)
|
|
* zebra/kernel_socket.c: HAVE_IPV6 conditional for WRAPUP when
HAVE_SA_LEN is not defined. bcopy -> memcpy, bcopy is not
portable.
|
|
* lib/regex.c: bzero -> memset
* zebra/ioctl.c: ditto. bzero is not portable.
|
|
* zebra/connected.c: PtP revert fixup. Zebra was not creating
connected route for PtP peer.
|
|
* zebra/ipforward_proc.c: (ipforward) Close the fd for
/proc/net/snmp. See [quagga-dev 284]
|
|
* zebra/zebra_rib.c: Fix possible dangling reference to rib
route_nodes - unlock it the appropriate number of times. (twice,
because of the implicit lock). see [quagga-dev 251].
|
|
* zebra/connected.c: revert the 'generic PtP' patch as it causes
far too many problems. People who use FreeSWAN should investigate
native linux ipsec.
* zebra/rt_netlink.c: ditto
* lib/if.c: ditto
* ripd/ripd.h: ditto
* ripd/ripd.c: ditto
* ripd/rip_interface.c: ditto
* ospfd/ospfd.c: ditto
* ospfd/ospf_snmp.c: ditto
* bgpd/bgp_nexthop.c: ditto
|
|
* zebra/kernel_socket.c: Fix up WRAPUP macro to deal with multiple
address families in the absence of sa_len element in struct
sockaddr.
(ifm_read): Handle solaris 9 if_msghdr_t.
Deal with interfaces which are incomplete, lookup on name rather
than the placeholder interface index of -1.
|
|
* bgpd/bgp_routemap.c: attr->med is type u_in32_t, should be
compared with UINT32_MAX
* ospfd/ospfd.c: remove redundant assert
* zebra/rtadv.c: add missing include for zebra/rib.h
|
|
* lib/Makefile.am: Do not use a lib (libcap) as a dependency
* zebra/Makefile.am: Link in libcap
|
|
* zebra/main.c: Fixup ZEBRA_{GROUP,USER} defines ->
QUAGGA_{USER,GROUP}
* bgpd/bgp_main.c: ditto
* ospf6d/ospf6_main.c: ditto
* ospfd/ospf_main.c: ditto
* ripd/rip_main.c: ditto
* ripngd/ripng_main.c: ditto
* zebra/main.c: ditto
|
|
* Add missing 'i' to getopts, short form of --pid_file.
see http://bugzilla.quagga.net/show_bug.cgi?id=25
|
|
* lib/if.c: (if_cmp_func) new function, compare interface names in
alphabetical order.
(if_create) Take name as argument and add interface in sorted order.
(if_get_by_name),(interface_cmd) fixup calls to if_create - see
above.
(if_init) register list comparison function.
* lib/if.h: Add comparison function, modify if_create prototype.
* lib/zclient.c: Modify call to if_create.
* ospfd/ospf_interface.c: (ospf_vl_new) modify call to if_create.
change sprintf to snprintf.
* zebra/kernel_socket.c: (ifm_read) modify call to if_create.
|
|
* lib/version.h: add ZEBRA_URL (unused for now)
* lib/vty.c: CMD_ERR_NOTHING_TODO when reading conf file should not
be fatal. slight reformating.
* ospfd/ospf_zebra.c: ignore reject/blackhole routes if zebra sends
these type of routes. probably should be a new type of route to
allow daemons to more easily choose whether to redistribute them
- rathen than just a flag (eg for reject/blackhole).
reorder the is_prefix_default test for ZEBRA_IPV4_ROUTE_DELETE to
avoid the inverted test - slightly more readable.
* redhat/zebra.spec.in: Add ospfapi port to services file, if
with_ospfapi.
* zebra/rib.h: Change nexthop types to an enum.
* zebra/rt_netlink.c: run it through indent -nut.
Add nexthop_types_desc[] descriptive array for nexthop types.
(netlink_route_multipath) debug statements indicate which branch
they are in and print out nexthop type.
* zebra/zebra_rib.c: slight reformatting.
* zebra/zebra_vty.c: Pass ZEBRA_FLAG_BLACKHOLE flag to
static_add_ipv4() if Null0 route is configured. print out Null0 if
STATIC_IPV4_BLACKHOLE route, and ignore flags (shouldnt be
possible to set flags from vty) for config and show route.
|
|
* zebra/if_proc.c: zebra does not close the file after reading
/proc/net/dev for retrieving interface names and statistics.
call fclose(fp) before returning from ifstat_update_proc()
or interface_list_proc()
|
|
* zebra/if_ioctl.c: Fix missing declaration for if_fake_index.
See bug id #24: http://bugzilla.dishone.st/show_bug.cgi?id=24
reported by Vsevolod Sipakov.
|
|
* zebra/kernel_socket.c: Update for new zebrad global struct
* zebra/ipforward_solaris.c: Add missing privs.h include. Uncomment
/dev/ip define.
* lib/zebra.h: Add typedefs for SunOS.
|
|
* zebra/rt_netlink.c: Debug statements added to
netlink_route_multipath()
* zebra/zebra_rib.c: If route has a gateway, delete only existing
route with that specified gateway.
|
|
* Fix lib/smux.c's reliance on daemons exporting struct
thread_master *master.
|
|
* lib/vty.{c,h}: Remove vty layer depending on a 'master' global,
pass the thread master in explicitly to vty_init. Sort out some
header dependency problems with lib/command.h
* zebra/: Move globals to struct zebrad. Update vty_init().
* (.*)/\1_main.c: update call to vty_init().
|
|
* zebra/ioctl.c: Properly zero-out struct ifreq before passing it
to the SIOCSIFFLAGS ioctl(2), so that it doesn't
occasionally trigger promisc mode on interface.
Reported/tested by: Vladimir B. Grebenschikov <vova@sw.ru>
(patch snarfed taken from FreeBSD ports)
|
|
* (zebra/rtadvc.): can not set checksum for ICMPV6, see:
From: Jun-ichiro itojun Hagino <itojun@iijlab.net>
Subject: [zebra 11052] setsockopt(IPV6_CHECKSUM) does not work for ICMPv6
socket
setsockopt(IPV6_CHECKSUM) does not work for ICMPv6 socket.
RFC2292 may not be clear enough about it (it talks about raw socket
and ICMPv6 socket mixed up), but the next revision
(draft-ietf-ipngwg-rfc2292bis-02.txt) is very clear about it.
> An attempt to set IPV6_CHECKSUM for an ICMPv6 socket will fail.
|
|
* Revert Cougar's sort interface names patch, causes problems with
enabling of interfaces for OSPF in ospfd.
|
|
|
|
|
|
* Merge of zebra privileges
|
|
|
|
to always be considered to be a blackhole route. (if gate is passed in, it
cant be).
|
|
|
|
|
|
|
|
|
|
working ipforward_solaris.c
|
|
ensure only interfaces that are inactive or do not exist can be deleted
|
|
Add the 'no interface' command to all the daemons and vtysh. now it's
possible to delete interface from routeing daemons as well only if it
doesn't exist in os.
http://hasso.linux.ee/zebra/ht-no_interface_fix.patch
|
|
|
|
http://hasso.linux.ee/zebra/ht-no_interface-23052003.patch
Add the 'no interface' command to all the daemons and vtysh.
|
|
|
|
Subject: [zebra 19084] Re: suspected memory leakage upon static route
I believe this is a more complete patch, it also addresses
allocating/freeing of nexthop->ifname string buffer (1) using
XSTRDUP/XFREE macros (thus correctly updating memory statistics), and
(2) for all nexthop->type cases that carry an ifname string.
|
|
Subject: [zebra 19080] suspected memory leakage upon static route deletion
upon deletion of a static route entry, at the end of
zebra/zebra_rib.c/static_delete_ipv4(), there's a call to XFREE to
deallocate the 'struct static_ipv4'. However, in the case of a static with
ifname as nexthop (type==STATIC_IPV4_IFNAME), this struct holds a pointer to
a dynamically allocated ifname string buffer (si->gate.ifname, see its
allocation within static_add_ipv4() using XSTRDUP). IMO, the attached patch
is required.
Notes: the same applies to IPv6 statics as well (fix included in patch);
and, I admit I haven't yet tested that, but it seems simple enough to be
working... ;->
|
|
Subject: [zebra 18677] zebra initialisation bug and patch
Hi All,
I have found a bug in zebra that prevents its routing table and
interface database from being initialised properly. The problem occurs
when a request is made via the netlink socket but the kernel produces a
EWOULDBLOCK/EAGAIN when the result is trying to be retrieved via a
recvmsg(). Zebra does not do anything about this and continues to
function (with an empty routing table and interface list) as if nothing
has happened. With no such information the routing protocol dosn't work!
Two functions are called during the initialisation of Zebra:
interface_lookup_netlink() and netlink_route_read() - obtaining the
interfaces and routing table from the kernel respectively. These are the
only time these functions are called.
These functions, interface_lookup_netlink() and netlink_route_read(),
use netlink_parse_info() to recieve the data from the netlink socket.
The problem is, netlink_parse_info() returns (without error) when its
call to recvmsg() results in an errno EWOULDBLOCK/EAGAIN. This behaviour
is expected by other funtions calling netlink_parse_info() -
netlink_parse_info is simply recalled at a later stage. However, on
initialisation it is never recalled.
Since zebra is expected to nothing else during initialisation it was
easiest to temporarily change the netlink socket to BLOCK and wait
indefinently until the kernel responds with the required information.
Attached is a patch with these changes.
Comments and questions are welcome.
Please inform me if this patch is added to the Zebra source.
--israel
|
