| Age | Commit message (Collapse) | Author |
|---|
|
* sigevent.c: (quagga_signal_handler) add a global caught flag, set
the flags to a constant rather increment to be kinder.
(quagga_sigevent_process) new function, to do core of what
quagga_signal_timer did. dont block signals at all as sig->caught
is volatile sig_atomic_t and should be safe to access from signal
and normal contexts. The signal blocking is unneeded paranoia, but
is left intact under an ifdef, should some platform require it.
Check global caught flag before iterating through array.
(quagga_signal_timer) nearly everything moved to
quagga_sigevent_process. Left in under ifdef, in case some
platform could use a regular timer check for signals.
* sigevent.h: quagga_sigevent_process declaration.
* thread.c: (thread_fetch) check for signals at beginning of
scheduler loop, check for signals if select returns EINTR.
|
|
* configure.ac: modify default CFLAGS to be compiler agnostic
build Makefile for tests/ subdir.
|
|
* redhat/quagga.spec.in: update for shared lib install, *.la
should be in -devel
|
|
* if.c: (if_cmp_func) Add comment suggested by Greg Troxel.
|
|
* ospf_packet.c: (ospf_ls_upd_send_queue_event) Partial fix for
problem reported by Peter Frost amongst others, where function
will spin indefinitely if update list contains LSAs greater than
MTU-headers or other condition leading to update list never being
cleared. Problem of what to do with these LSAs remains.
(ospf_make_ls_upd) add comment about large LSA problem,
indentation cleanup.
|
|
* Makefile.am (EXTRA_DIST): Add missing \, so tools stuff is
really in distfile.
|
|
* kernel_socket.c (rtm_flag_dump): terminate buffer with '\0', not '0'.
This is arguably not a security problem, since strlcat is used to copy
to the buffer -gdt.
|
|
* sigevent.c: Don't block SIGTRAP and SIGKILL. Blocking SIGTRAP
confuses gdb, at least on NetBSD 2.0_BETA, where the block
succeeds.
|
|
|
|
2) Following common style is always better anyway.
|
|
|
|
|
|
Merge of GNU Zebra cvs2svn changesets r799, r800 and r807.
* bgp_dump.c: (bgp_dump_attr) cleanup. return status code. check
attributes present before printing.
* bgp_dump.c: update bgp_dump_attr prototype.
* bgp_packet.c: (bgp_update_receive) init attrstr. check status
of bgp_dump_attr. Log end-of-rib UPDATEs.
|
|
* jhash.{c,h}: New files. Bob Jenkins' public domain hashing
function, as implemented in linux kernel by David Miller.
|
|
* Merge Kunihiro's 'show route-map' change and add
compatibility aliases for route-map continue
|
|
|
|
|
|
* if.c: (if_cmp_func) fix for interface names where name is same,
but one has no number, eg "devtyp" and "devtyp0".
|
|
* bgp_packet.c: (bgp_collision_detect) Send NOTIFY on new socket
if that is connection we're closing.
(bgp_read) invalid marker check applies to KEEPALIVE too.
* bgp_route.c: Ignore multicast NRLI, dont send NOTIFY.
|
|
* ripng_interface.c (ripng_multicast_join): Use privs to do join,
to work around bug on gif(4) on NetBSD 1.6.2.
|
|
|
|
NEWS file.
|
|
|
|
libzebra and libospfapiclient are now built shared, and linked shared
with the daemons. This reduces the memory needed when running
multiple daemons; each daemon is at least 150k smaller. Static
libraries are still built, and libtool should use them on platforms
which don't have shared libaries. As with autoconf, the user of a
distribution does not need libtool; one just needs that to build from
CVS.
libospf.a is still a non-shared library, and still installed, not
because that makese sense, but because I don't understand why it is
the way it is now.
Note that the tree was tagged 'libtool-before' just before this commit.
|
|
Remove disthook that removed the crud that was copied in due to this.
|
|
if --enable-vtysh, look for perl, and substitute into vtysh/extract.pl,
rather than assuming perl is in /usr/bin
Look for tputs in libtermcap and libcurses, in addition to
previously-searched locations. Follow GNU readlines search order.
Clean up --enable-vtysh definition in configure.ac.
Add vtysh/vtysh_cmds.c to CLEANFILES. This is important if
extract.pl fails, so that 'make clean && make' will regenerate
vtysh_cmds.c rhather than using the old zero-length file.
|
|
|
|
* ChangeLog: fix my last update config.h -> zebra.h ;)
* zebra.h: Fix gcc check.
|
|
|
|
to read system one.
|
|
|
|
document this stuff. Nobody reads loooong changelog ;).
|
|
|
|
* filter.c: (access_list_remark_cmd) buffer_putstr doesnt need cast
to u_char. (ipv6_access_list_remark_cmd) ditto.
if.c: ditto
* network.c: (readn/writen) pointer arg should be type u_char.
* plist.c: needs to include stream.h, not declare stream functions
internally.
(various) Add static qualifier to internal functions.
(prefix_list_type_str) extraneous breaks in switch statement.
(ip_prefix_list_description_cmd) buffer_putstr doesnt need cast
* stream.h: depends on plist.h and export stream_put_prefix
* vty.c: (vty_<telnet option build functions>) should use
unsigned char, telnet options are 0 -> 255.
* zclient.c: various u_char<->char type cleanups.
* zebra.h: Having to define CMSG_* can apply to more than just
BSDI_NRL.
* ripd.c: (rip_distribute_update_all) distribute list hook
function pointer prototype requires struct prefix_list * arg.
(rip_distribute_update_all_wrapper) update to pass required arg,
NULL.
|
|
* config.h: __attribute__ is a gcc'ism
|
|
* ripd.h: Add define for the RIPv2 Authentication Data family
Move the auth type defines up to where other defines live.
Add RIP_AUTH_MD5_COMPAT_SIZE, for backwards compatible
md5->auth_len size. Add md5_auth_len field to struct
rip_interface: (rip_interface_new) Init md5_auth_len to compatible
size.
(ip_rip_authentication_mode_cmd) Extended to handle setting
md5 auth-length. Appropriate aliases added.
(no_ip_rip_authentication_mode_cmd) Reset md5_auth_len to
compatible size.
(rip_interface_config_write) Teach it about md5_auth_len.
_always_ write out the auth-length, so that everyone will get
the setting in their config file, and hence allow for a future
change of default for md5_auth_len to be less painful - every md5
user will have this setting in their config file.
ripd.c: (rip_packet_dump) Change nasty hard coded constants to
symbolic defines. Change various tests of 'ntoh.(variable) ==
constant' to test 'variable == ntoh.(constant)'. Clean up
indentation on some long lines.
(rip_auth_simple_password) ditto.
(rip_auth_md5) ditto, also add length argument and sanity check
md5 data offset field. Sanity check md5 auth length, accept RFC
or old-ripd/cisco lengths.
(rip_auth_md5_set) as per (rip_packet_dump), also write out
the configured md5 auth length for the interface (old-ripd or rfc)
(rip_read) as per (rip_packet_dump)
(rip_write_rte) ditto
(rip_response_process) ditto
(rip_write_rte) ditto
|
|
from starting.
|
|
|
|
|
|
* ospf_dump.c,ospf_lsa.c: Fix typos of merge of previous.
ospf_flood.c: (ospf_process_self_originated_lsa) fix zlog format.
|
|
|
|
|
|
* type mismatch fixes
|
|
* ripd.c: Interoperability fix. Correct value for MD5 auth length
is 16. Accept packets with this set to >= 16, and set to 16
ourselves.
|
|
* ospf_dump.c: (ospf_lsa_header_dump) LOOKUP can return null if
index is out of range.
ospf_flood.c: endianness fix
ospf_lsa.c: Missing ntohl's on (struct lsa *)->data->ls_seqnum
in various places.
|
|
|
|
* ripd.c: Fixup compile warnings
* rip_routemap.c: Ditto
|
|
* Correct RIP port, s/521/520/, reported by xantius@xantius.com
|
|
repository by Rivo Nurges.
|
|
|
