)]}'
{
  "log": [
    {
      "commit": "228da42898c4f7bd72d9c1ee4135108e8d40d860",
      "tree": "a780ed018bfeb97c174958f188c770c74a48bad9",
      "parents": [
        "54a15182e05ca757db3bb90a4135e9f8fd3c84f2"
      ],
      "author": {
        "name": "Chris Caputo",
        "email": "ccaputo@alt.net",
        "time": "Sat Jul 18 05:44:03 2009 +0000"
      },
      "committer": {
        "name": "Paul Jakma",
        "email": "paul@quagga.net",
        "time": "Sun Jul 19 18:28:08 2009 +0100"
      },
      "message": "[bgpd] Stability fixes including bugs 397, 492\n\nI\u0027ve spent the last several weeks working on stability fixes to bgpd.\nThese patches fix all of the numerous crashes, assertion failures, memory\nleaks and memory stomping I could find.  Valgrind was used extensively.\n\nAdded new function bgp_exit() to help catch problems.  If \"debug bgp\" is\nconfigured and bgpd exits with status of 0, statistics on remaining\nlib/memory.c allocations are printed to stderr.  It is my hope that other\ndevelopers will use this to stay on top of memory issues.\n\nExample questionable exit:\n\n  bgpd: memstats: Current memory utilization in module LIB:\n  bgpd: memstats:  Link List                     :          6\n  bgpd: memstats:  Link Node                     :          5\n  bgpd: memstats:  Hash                          :          8\n  bgpd: memstats:  Hash Bucket                   :          2\n  bgpd: memstats:  Hash Index                    :          8\n  bgpd: memstats:  Work queue                    :          3\n  bgpd: memstats:  Work queue item               :          2\n  bgpd: memstats:  Work queue name string        :          3\n  bgpd: memstats: Current memory utilization in module BGP:\n  bgpd: memstats:  BGP instance                  :          1\n  bgpd: memstats:  BGP peer                      :          1\n  bgpd: memstats:  BGP peer hostname             :          1\n  bgpd: memstats:  BGP attribute                 :          1\n  bgpd: memstats:  BGP extra attributes          :          1\n  bgpd: memstats:  BGP aspath                    :          1\n  bgpd: memstats:  BGP aspath str                :          1\n  bgpd: memstats:  BGP table                     :         24\n  bgpd: memstats:  BGP node                      :          1\n  bgpd: memstats:  BGP route                     :          1\n  bgpd: memstats:  BGP synchronise               :          8\n  bgpd: memstats:  BGP Process queue             :          1\n  bgpd: memstats:  BGP node clear queue          :          1\n  bgpd: memstats: NOTE: If configuration exists, utilization may be expected.\n\nExample clean exit:\n\n  bgpd: memstats: No remaining tracked memory utilization.\n\nThis patch fixes bug #397: \"Invalid free in bgp_announce_check()\".\n\nThis patch fixes bug #492: \"SIGBUS in bgpd/bgp_route.c:\nbgp_clear_route_node()\".\n\nMy apologies for not separating out these changes into individual patches.\nThe complexity of doing so boggled what is left of my brain.  I hope this\nis all still useful to the community.\n\nThis code has been production tested, in non-route-server-client mode, on\na linux 32-bit box and a 64-bit box.\n\nRelease/reset functions, used by bgp_exit(), added to:\n\n  bgpd/bgp_attr.c,h\n  bgpd/bgp_community.c,h\n  bgpd/bgp_dump.c,h\n  bgpd/bgp_ecommunity.c,h\n  bgpd/bgp_filter.c,h\n  bgpd/bgp_nexthop.c,h\n  bgpd/bgp_route.c,h\n  lib/routemap.c,h\n\nFile by file analysis:\n\n* bgpd/bgp_aspath.c: Prevent re-use of ashash after it is released.\n\n* bgpd/bgp_attr.c: #if removed uncalled cluster_dup().\n\n* bgpd/bgp_clist.c,h: Allow community_list_terminate() to be called from\n  bgp_exit().\n\n* bgpd/bgp_filter.c: Fix aslist-\u003ename use without allocation check, and\n  also fix memory leak.\n\n* bgpd/bgp_main.c: Created bgp_exit() exit routine.  This function frees\n  allocations made as part of bgpd initialization and, to some extent,\n  configuration.  If \"debug bgp\" is configured, memory stats are printed\n  as described above.\n\n* bgpd/bgp_nexthop.c: zclient_new() already allocates stream for\n  ibuf/obuf, so bgp_scan_init() shouldn\u0027t do it too.  Also, made it so\n  zlookup is global so bgp_exit() can use it.\n\n* bgpd/bgp_packet.c: bgp_capability_msg_parse() call to bgp_clear_route()\n  adjusted to use new BGP_CLEAR_ROUTE_NORMAL flag.\n\n* bgpd/bgp_route.h: Correct reference counter \"lock\" to be signed.\n  bgp_clear_route() now accepts a bgp_clear_route_type of either\n  BGP_CLEAR_ROUTE_NORMAL or BGP_CLEAR_ROUTE_MY_RSCLIENT.\n\n* bgpd/bgp_route.c:\n  - bgp_process_rsclient(): attr was being zero\u0027ed and then\n    bgp_attr_extra_free() was being called with it, even though it was\n    never filled with valid data.\n\n  - bgp_process_rsclient(): Make sure rsclient-\u003egroup is not NULL before\n    use.\n\n  - bgp_processq_del(): Add call to bgp_table_unlock().\n\n  - bgp_process(): Add call to bgp_table_lock().\n\n  - bgp_update_rsclient(): memset clearing of new_attr not needed since\n    declarationw with \"\u003d { 0 }\" does it.  memset was already commented\n    out.\n\n  - bgp_update_rsclient(): Fix screwed up misleading indentation.\n\n  - bgp_withdraw_rsclient(): Fix screwed up misleading indentation.\n\n  - bgp_clear_route_node(): Support BGP_CLEAR_ROUTE_MY_RSCLIENT.\n\n  - bgp_clear_node_queue_del(): Add call to bgp_table_unlock() and also\n    free struct bgp_clear_node_queue used for work item.\n\n  - bgp_clear_node_complete(): Do peer_unlock() after BGP_EVENT_ADD() in\n    case peer is released by peer_unlock() call.\n\n  - bgp_clear_route_table(): Support BGP_CLEAR_ROUTE_MY_RSCLIENT.  Use\n    struct bgp_clear_node_queue to supply data to worker.  Add call to\n    bgp_table_lock().\n\n  - bgp_clear_route(): Add support for BGP_CLEAR_ROUTE_NORMAL or\n    BGP_CLEAR_ROUTE_MY_RSCLIENT.\n\n  - bgp_clear_route_all(): Use BGP_CLEAR_ROUTE_NORMAL.\n\n  Bug 397 fixes:\n\n    - bgp_default_originate()\n    - bgp_announce_table()\n\n* bgpd/bgp_table.h:\n  - struct bgp_table: Added reference count.  Changed type of owner to be\n    \"struct peer *\" rather than \"void *\".\n\n  - struct bgp_node: Correct reference counter \"lock\" to be signed.\n\n* bgpd/bgp_table.c:\n  - Added bgp_table reference counting.\n\n  - bgp_table_free(): Fixed cleanup code.  Call peer_unlock() on owner if\n    set.\n\n  - bgp_unlock_node(): Added assertion.\n\n  - bgp_node_get(): Added call to bgp_lock_node() to code path that it was\n    missing from.\n\n* bgpd/bgp_vty.c:\n  - peer_rsclient_set_vty(): Call peer_lock() as part of peer assignment\n    to owner.  Handle failure gracefully.\n\n  - peer_rsclient_unset_vty(): Add call to bgp_clear_route() with\n    BGP_CLEAR_ROUTE_MY_RSCLIENT purpose.\n\n* bgpd/bgp_zebra.c: Made it so zclient is global so bgp_exit() can use it.\n\n* bgpd/bgpd.c:\n  - peer_lock(): Allow to be called when status is \"Deleted\".\n\n  - peer_deactivate(): Supply BGP_CLEAR_ROUTE_NORMAL purpose to\n    bgp_clear_route() call.\n\n  - peer_delete(): Common variable listnode pn.  Fix bug in which rsclient\n    was only dealt with if not part of a peer group.  Call\n    bgp_clear_route() for rsclient, if appropriate, and do so with\n    BGP_CLEAR_ROUTE_MY_RSCLIENT purpose.\n\n  - peer_group_get(): Use XSTRDUP() instead of strdup() for conf-\u003ehost.\n\n  - peer_group_bind(): Call bgp_clear_route() for rsclient, and do so with\n    BGP_CLEAR_ROUTE_MY_RSCLIENT purpose.\n\n  - bgp_create(): Use XSTRDUP() instead of strdup() for peer_self-\u003ehost.\n\n  - bgp_delete(): Delete peers before groups, rather than after.  And then\n    rather than deleting rsclients, verify that there are none at this\n    point.\n\n  - bgp_unlock(): Add assertion.\n\n  - bgp_free(): Call bgp_table_finish() rather than doing XFREE() itself.\n\n* lib/command.c,h: Compiler warning fixes.  Add cmd_terminate().  Fixed\n  massive leak in install_element() in which cmd_make_descvec() was being\n  called more than once for the same cmd-\u003estrvec/string/doc.\n\n* lib/log.c: Make closezlog() check fp before calling fclose().\n\n* lib/memory.c: Catch when alloc count goes negative by using signed\n  counts.  Correct #endif comment.  Add log_memstats_stderr().\n\n* lib/memory.h: Add log_memstats_stderr().\n\n* lib/thread.c: thread-\u003efuncname was being accessed in thread_call() after\n  it had been freed.  Rearranged things so that thread_call() frees\n  funcname.  Also made it so thread_master_free() cleans up cpu_record.\n\n* lib/vty.c,h: Use global command_cr.  Add vty_terminate().\n\n* lib/zclient.c,h: Re-enable zclient_free().\n"
    },
    {
      "commit": "0b2aa3a0a8b095bdef1eddda117d173af75dede2",
      "tree": "29201cb2beec187409e842b4078ed8a8047d3f11",
      "parents": [
        "7593fddfa1558d086e3142ad96cf2790007f5d84"
      ],
      "author": {
        "name": "Paul Jakma",
        "email": "paul.jakma@sun.com",
        "time": "Sun Oct 14 22:32:21 2007 +0000"
      },
      "committer": {
        "name": "Paul Jakma",
        "email": "paul.jakma@sun.com",
        "time": "Sun Oct 14 22:32:21 2007 +0000"
      },
      "message": "[bgpd] Merge AS4 support\n\n2007-10-14 Paul Jakma \u003cpaul.jakma@sun.com\u003e\n\n\t* NEWS: Note that MRT dumps are now version 2\n\t* (general) Merge in Juergen Kammer\u0027s AS4 patch.\n\n2007-09-27 Paul Jakma \u003cpaul.jakma@sun.com\u003e\n\n\t* bgp_aspath.c: (assegment_normalise) remove duplicates from\n\t  from sets.\n\t  (aspath_reconcile_as4) disregard a broken part of the RFC around\n\t  error handling in path reconciliation.\n\t* aspath_test.c: Test dupe-weeding from sets.\n\t  Test that reconciliation merges AS_PATH and AS4_PATH where\n\t  former is shorter than latter.\n\n2007-09-26 Paul Jakma \u003cpaul.jakma@sun.com\u003e\n\n\t* aspath_test.c: Test AS4_PATH reconcilation where length\n\t  of AS_PATH and AS4_PATH is same.\n\n2007-09-25 Paul Jakma \u003cpaul.jakma@sun.com\u003e\n\n\t* bgp_open.c: (peek_for_as4_capability) Fix to work.\n\t* bgp_packet.c: (bgp_open_receive) Fix sanity check of as4.\n\t* tests/bgp_capability_test.c: (general) Extend tests to validate\n\t  peek_for_as4_capability.\n\t  Add test of full OPEN Option block, with multiple capabilities,\n\t  both as a series of Option, and a single option.\n\t  Add some crap to beginning of stream, to prevent code depending\n\t  on getp \u003d\u003d 0.\n\n2007-09-18 Paul Jakma \u003cpaul.jakma@sun.com\u003e\n\n\t* bgp_open.c: (bgp_capability_as4) debug printf inline with others.\n\t  (peek_for_as4_capability) There\u0027s no need to signal failure, as\n\t  failure is better dealt with through full capability parser -\n\t  just return the AS4, simpler.\n\t* bgp_packet.c: (bgp_open_receive) Update to match\n\t  peek_for_as4_capability change.\n\t  Allow use of BGP_AS_TRANS by 2b speakers.\n\t  Use NOTIFY_OPEN_ERR rather than CEASE for OPEN parsing errors.\n\t  (bgp_capability_msg_parse) missing argument to debug print\n\t  (bgp_capability_receive) missing return values.\n\t* tests/bgp_capability_test.c: (parse_test) update for changes to\n\t  peek_for_as4_capability\n\n2007-07-25 Paul Jakma \u003cpaul.jakma@sun.com\u003e\n\n\t* Remove 2-byte size macros, just make existing macros take\n\t  argument to indicate which size to use.\n\t  Adjust all users - typically they want \u00271\u0027.\n\t* bgp_aspath.c: (aspath_has_as4) New, return 1 if there are any\n\t  as4\u0027s in a path.\n\t  (aspath_put) Return the number of bytes actually written, to\n\t  fix the bug Juergen noted: Splitting of segments will change\n\t  the number of bytes written from that already written to the\n\t  AS_PATH header.\n\t  (aspath_snmp_pathseg) Pass 2-byte flag to aspath_put. SNMP\n\t  is still defined as 2b.\n\t  (aspath_aggregate) fix latent bug.\n\t  (aspath_reconcile_as4) AS_PATH+NEW_AS_PATH reconciliation\n\t  function.\n\t  (aspath_key_make) Hash the AS_PATH string, rather than\n\t  just taking the addition of assegment ASes as the hash value,\n\t  hopefully sligthly more collision resistant.\n\t  (bgp_attr_munge_as4_attrs) Collide the NEW_ attributes\n\t  together with the OLD 2-byte forms, code Juergen\n\t  had in bgp_attr_parse but re-organised a bit.\n\t  (bgp_attr_parse) Bunch of code from Juergen moves\n\t  to previous function.\n\t  (bgp_packet_attribute) Compact significantly by\n\t  just /always/ using extended-length attr header.\n\t  Fix bug Juergen noted, by using aspath_put\u0027s\n\t  (new) returned size value for the attr header rather\n\t  than the (guesstimate) of aspath_size() - the two could\n\t  differ when aspath_put had to split large segments, unlikely\n\t  this bug was ever hit in the \u0027wild\u0027.\n\t  (bgp_dump_routes_attr) Always use extended-len and\n\t  use aspath_put return for header length. Output 4b ASN\n\t  for AS_PATH and AGGREGATOR.\n\t* bgp_ecommunity.c: (ecommunity_{hash_make,cmp}) fix\n\t  hash callback declarations to match prototypes.\n\t  (ecommunity_gettoken) Updated for ECOMMUNITY_ENCODE_AS4,\n\t  complete rewrite of Juergen\u0027s changes (no asdot support)\n\t* bgp_open.c: (bgp_capability_as4) New, does what it says\n\t  on the tin.\n\t  (peek_for_as4_capability) Rewritten to use streams and\n\t  bgp_capability_as4.\n\t* bgp_packet.c: (bgp_open_send) minor edit\n\t  checked (in the abstract at least) with Juergen.\n\t  Changes are to be more accepting, e.g, allow AS_TRANS on\n\t  a 2-byte session.\n\t* (general) Update all commands to use CMD_AS_RANGE.\n\t* bgp_vty.c: (bgp_clear) Fix return vals to use CMD_..\n\t  Remove stuff replicated by VTY_GET_LONG\n\t  (bgp_clear_vty) Return bgp_clear directly to vty.\n\t* tests/aspath_test.c: Exercise 32bit parsing. Test reconcile\n\t  function.\n\t* tests/ecommunity_test.c: New, test AS4 ecommunity changes,\n\t  positive test only at this time, error cases not tested yet.\n\n2007-07-25 Juergen Kammer \u003cj.kammer@eurodata.de\u003e\n\n\t* (general) AS4 support.\n\t* bgpd.h: as_t changes to 4-bytes.\n\t* bgp_aspath.h: Add BGP_AS4_MAX and BGP_AS_TRANS defines.\n\t* bgp_aspath.c: AS_VALUE_SIZE becomes 4-byte, AS16_VALUE_SIZE\n\t  added for 2-byte.\n\t  Add AS16 versions of length calc macros.\n\t  (aspath_count_numas) New, count number of ASes.\n\t  (aspath_has_as4) New, return 1 if there are any as4\u0027s in a\n\t  path.\n\t  (assegments_parse) Interpret assegment as 4 or 2 byte,\n\t  according to how the caller instructs us, with a new\n\t  argument.\n\t  (aspath_parse) Add use32bit argument to pass to\n\t  assegments_parse. Adjust all its callers to pass 1, unless\n\t  otherwise noted.\n\t  (assegment_data_put) Adjust to be able to write 2 or 4 byte\n\t  AS, according to new use32bit argument.\n\t  (aspath_put) Adjust to write 2 or 4.\n\t  (aspath_gettoken) Use a long for passed in asno.\n\t* bgp_attr.c: (attr_str) Add BGP_ATTR_AS4_PATH and\n\t  BGP_ATTR_AS4_AGGREGATOR.\n\t  (bgp_attr_aspath) Call aspath_parse with right 2/4 arg, as\n\t  determined by received-capability flag.\n\t  (bgp_attr_aspath_check) New, code previously in attr_aspath\n\t  but moved to new func so it can be run after NEW_AS_PATH\n\t  reconciliation.\n\t  (bgp_attr_as4_path) New, handle NEW_AS_PATH.\n\t  (bgp_attr_aggregator) Adjust to cope with 2/4 byte ASes.\n\t  (bgp_attr_as4_aggregator) New, read NEW_AGGREGATOR.\n\t  (bgp_attr_parse) Add handoffs to previous parsers for the two\n\t  new AS4 NEW_ attributes.\n\t  Various checks added for NEW/OLD reconciliation.\n\t  (bgp_packet_attribute) Support 2/4 for AS_PATH and\n\t  AGGREGATOR, detect when NEW_ attrs need to be sent.\n\t* bgp_debug.{c,h}: Add \u0027debug bgp as4\u0027.\n\t* bgp_dump.c: MRTv2 support, unconditionally enabled, which\n\t  supports AS4. Based on patches from Erik (RIPE?).\n\t* bgp_ecommunity.c: (ecommunity_ecom2str) ECOMMUNITY_ENCODE_AS4\n\t  support.\n\t* bgp_open.c: (peek_for_as4_capability) New, peek for AS4\n\t  capability prior to full capability parsing, so we know which\n\t  ASN to use for struct peer lookup.\n\t  (bgp_open_capability) Always send AS4 capability.\n\t* bgp_packet.c: (bgp_open_send) AS4 handling for AS field\n\t  (bgp_open_receive) Peek for AS4 capability first, and figure\n\t  out which AS to believe.\n\t* bgp_vty.c: (bgp_show_peer) Print AS4 cap\n\t* tests/aspath_test.c: Support asn32 changes, call aspath_parse\n\t  with 16 bit.\n\t* vtysh/extract.pl: AS4 compatibility for router bgp ASNUMBER\n\t* vtysh/extract.pl.in: AS4 compatibility for router bgp ASNUMBER\n\t* vtysh/vtysh.c: AS4 compatibility for router bgp ASNUMBER\n"
    },
    {
      "commit": "94f2b3923e9663d0355a829f22e4e31cf68ee7b8",
      "tree": "59c0291a76b1c97fbb089abb479307f19cdfd4c1",
      "parents": [
        "7cad40007b9714dba65065916ea68577f9d3d6dd"
      ],
      "author": {
        "name": "paul",
        "email": "paul",
        "time": "Tue Jun 28 12:44:16 2005 +0000"
      },
      "committer": {
        "name": "paul",
        "email": "paul",
        "time": "Tue Jun 28 12:44:16 2005 +0000"
      },
      "message": "2005-06-28 Paul Jakma \u003cpaul.jakma@sun.com\u003e\n\n\t* (global) The great bgpd extern and static\u0027ification.\n\t* bgp_routemap.c: remove unused ROUTE_MATCH_ASPATH_OLD code\n\t  (route_set_metric_compile) fix u_int32_t to ULONG_MAX comparison\n\t  warnings.\n\t* bgp_route.h: (bgp_process, bgp_withdraw, bgp_update) export these\n\t  used by various files which had their own private declarations,\n\t  in the case of mplsvpn - incorrect.\n"
    },
    {
      "commit": "00d252cb5ff21ecc537ab40385316b41622a95e2",
      "tree": "c0132f4be53d8ae251ecb1dd8c39816f8734d08f",
      "parents": [
        "269d74fdc39a612da8f627bf97628c68d25a16ab"
      ],
      "author": {
        "name": "paul",
        "email": "paul",
        "time": "Mon May 23 14:19:54 2005 +0000"
      },
      "committer": {
        "name": "paul",
        "email": "paul",
        "time": "Mon May 23 14:19:54 2005 +0000"
      },
      "message": "2005-05-23 Paul Jakma \u003cpaul@dishone.st\u003e\n\n\t* bgp_fsm.h: Add extern qualifier to exported functions\n\t* bgp_nexthop.c: add static to nexthop specific globals\n\t* *.h: Add guard defines\n"
    },
    {
      "commit": "9834cd0f43bc45f04303b0f742faacf3bc5eac45",
      "tree": "b0198301ead5b6d8051a95689a5ce59c228c7116",
      "parents": [
        "cc1131ab011afc4f06c1a967e9c0cbd77cc88d58"
      ],
      "author": {
        "name": "paul",
        "email": "paul",
        "time": "Sat Oct 18 01:01:19 2003 +0000"
      },
      "committer": {
        "name": "paul",
        "email": "paul",
        "time": "Sat Oct 18 01:01:19 2003 +0000"
      },
      "message": "2003-10-18 Lorenzo Colitti \u003clorenzo@ripe.net\u003e\n\n        * bgpd/bgp_dump.{c,h}: (bgp_dump_interval_add) Dump at discrete\n          fixed intervals rather than fixed intervals from startup time.\n          (bgp_dump_interval_func) Dont return immediately if file cant be\n          openeded, but reschedule interval dumps, even - admin might\n          fix problem in meantime. Close the dump file in between intervals.\n          (bgp_dump_init) account for MSG header when initialising stream\n\t  size.\n"
    },
    {
      "commit": "718e3744195351130f4ce7dbe0613f4b3e23df93",
      "tree": "bac2ad39971cd43f31241ef123bd4e470f695ac9",
      "parents": [],
      "author": {
        "name": "paul",
        "email": "paul",
        "time": "Fri Dec 13 20:15:29 2002 +0000"
      },
      "committer": {
        "name": "paul",
        "email": "paul",
        "time": "Fri Dec 13 20:15:29 2002 +0000"
      },
      "message": "Initial revision\n"
    }
  ]
}