| Age | Commit message (Collapse) | Author |
|---|
|
I've spent the last several weeks working on stability fixes to bgpd.
These patches fix all of the numerous crashes, assertion failures, memory
leaks and memory stomping I could find. Valgrind was used extensively.
Added new function bgp_exit() to help catch problems. If "debug bgp" is
configured and bgpd exits with status of 0, statistics on remaining
lib/memory.c allocations are printed to stderr. It is my hope that other
developers will use this to stay on top of memory issues.
Example questionable exit:
bgpd: memstats: Current memory utilization in module LIB:
bgpd: memstats: Link List : 6
bgpd: memstats: Link Node : 5
bgpd: memstats: Hash : 8
bgpd: memstats: Hash Bucket : 2
bgpd: memstats: Hash Index : 8
bgpd: memstats: Work queue : 3
bgpd: memstats: Work queue item : 2
bgpd: memstats: Work queue name string : 3
bgpd: memstats: Current memory utilization in module BGP:
bgpd: memstats: BGP instance : 1
bgpd: memstats: BGP peer : 1
bgpd: memstats: BGP peer hostname : 1
bgpd: memstats: BGP attribute : 1
bgpd: memstats: BGP extra attributes : 1
bgpd: memstats: BGP aspath : 1
bgpd: memstats: BGP aspath str : 1
bgpd: memstats: BGP table : 24
bgpd: memstats: BGP node : 1
bgpd: memstats: BGP route : 1
bgpd: memstats: BGP synchronise : 8
bgpd: memstats: BGP Process queue : 1
bgpd: memstats: BGP node clear queue : 1
bgpd: memstats: NOTE: If configuration exists, utilization may be expected.
Example clean exit:
bgpd: memstats: No remaining tracked memory utilization.
This patch fixes bug #397: "Invalid free in bgp_announce_check()".
This patch fixes bug #492: "SIGBUS in bgpd/bgp_route.c:
bgp_clear_route_node()".
My apologies for not separating out these changes into individual patches.
The complexity of doing so boggled what is left of my brain. I hope this
is all still useful to the community.
This code has been production tested, in non-route-server-client mode, on
a linux 32-bit box and a 64-bit box.
Release/reset functions, used by bgp_exit(), added to:
bgpd/bgp_attr.c,h
bgpd/bgp_community.c,h
bgpd/bgp_dump.c,h
bgpd/bgp_ecommunity.c,h
bgpd/bgp_filter.c,h
bgpd/bgp_nexthop.c,h
bgpd/bgp_route.c,h
lib/routemap.c,h
File by file analysis:
* bgpd/bgp_aspath.c: Prevent re-use of ashash after it is released.
* bgpd/bgp_attr.c: #if removed uncalled cluster_dup().
* bgpd/bgp_clist.c,h: Allow community_list_terminate() to be called from
bgp_exit().
* bgpd/bgp_filter.c: Fix aslist->name use without allocation check, and
also fix memory leak.
* bgpd/bgp_main.c: Created bgp_exit() exit routine. This function frees
allocations made as part of bgpd initialization and, to some extent,
configuration. If "debug bgp" is configured, memory stats are printed
as described above.
* bgpd/bgp_nexthop.c: zclient_new() already allocates stream for
ibuf/obuf, so bgp_scan_init() shouldn't do it too. Also, made it so
zlookup is global so bgp_exit() can use it.
* bgpd/bgp_packet.c: bgp_capability_msg_parse() call to bgp_clear_route()
adjusted to use new BGP_CLEAR_ROUTE_NORMAL flag.
* bgpd/bgp_route.h: Correct reference counter "lock" to be signed.
bgp_clear_route() now accepts a bgp_clear_route_type of either
BGP_CLEAR_ROUTE_NORMAL or BGP_CLEAR_ROUTE_MY_RSCLIENT.
* bgpd/bgp_route.c:
- bgp_process_rsclient(): attr was being zero'ed and then
bgp_attr_extra_free() was being called with it, even though it was
never filled with valid data.
- bgp_process_rsclient(): Make sure rsclient->group is not NULL before
use.
- bgp_processq_del(): Add call to bgp_table_unlock().
- bgp_process(): Add call to bgp_table_lock().
- bgp_update_rsclient(): memset clearing of new_attr not needed since
declarationw with "= { 0 }" does it. memset was already commented
out.
- bgp_update_rsclient(): Fix screwed up misleading indentation.
- bgp_withdraw_rsclient(): Fix screwed up misleading indentation.
- bgp_clear_route_node(): Support BGP_CLEAR_ROUTE_MY_RSCLIENT.
- bgp_clear_node_queue_del(): Add call to bgp_table_unlock() and also
free struct bgp_clear_node_queue used for work item.
- bgp_clear_node_complete(): Do peer_unlock() after BGP_EVENT_ADD() in
case peer is released by peer_unlock() call.
- bgp_clear_route_table(): Support BGP_CLEAR_ROUTE_MY_RSCLIENT. Use
struct bgp_clear_node_queue to supply data to worker. Add call to
bgp_table_lock().
- bgp_clear_route(): Add support for BGP_CLEAR_ROUTE_NORMAL or
BGP_CLEAR_ROUTE_MY_RSCLIENT.
- bgp_clear_route_all(): Use BGP_CLEAR_ROUTE_NORMAL.
Bug 397 fixes:
- bgp_default_originate()
- bgp_announce_table()
* bgpd/bgp_table.h:
- struct bgp_table: Added reference count. Changed type of owner to be
"struct peer *" rather than "void *".
- struct bgp_node: Correct reference counter "lock" to be signed.
* bgpd/bgp_table.c:
- Added bgp_table reference counting.
- bgp_table_free(): Fixed cleanup code. Call peer_unlock() on owner if
set.
- bgp_unlock_node(): Added assertion.
- bgp_node_get(): Added call to bgp_lock_node() to code path that it was
missing from.
* bgpd/bgp_vty.c:
- peer_rsclient_set_vty(): Call peer_lock() as part of peer assignment
to owner. Handle failure gracefully.
- peer_rsclient_unset_vty(): Add call to bgp_clear_route() with
BGP_CLEAR_ROUTE_MY_RSCLIENT purpose.
* bgpd/bgp_zebra.c: Made it so zclient is global so bgp_exit() can use it.
* bgpd/bgpd.c:
- peer_lock(): Allow to be called when status is "Deleted".
- peer_deactivate(): Supply BGP_CLEAR_ROUTE_NORMAL purpose to
bgp_clear_route() call.
- peer_delete(): Common variable listnode pn. Fix bug in which rsclient
was only dealt with if not part of a peer group. Call
bgp_clear_route() for rsclient, if appropriate, and do so with
BGP_CLEAR_ROUTE_MY_RSCLIENT purpose.
- peer_group_get(): Use XSTRDUP() instead of strdup() for conf->host.
- peer_group_bind(): Call bgp_clear_route() for rsclient, and do so with
BGP_CLEAR_ROUTE_MY_RSCLIENT purpose.
- bgp_create(): Use XSTRDUP() instead of strdup() for peer_self->host.
- bgp_delete(): Delete peers before groups, rather than after. And then
rather than deleting rsclients, verify that there are none at this
point.
- bgp_unlock(): Add assertion.
- bgp_free(): Call bgp_table_finish() rather than doing XFREE() itself.
* lib/command.c,h: Compiler warning fixes. Add cmd_terminate(). Fixed
massive leak in install_element() in which cmd_make_descvec() was being
called more than once for the same cmd->strvec/string/doc.
* lib/log.c: Make closezlog() check fp before calling fclose().
* lib/memory.c: Catch when alloc count goes negative by using signed
counts. Correct #endif comment. Add log_memstats_stderr().
* lib/memory.h: Add log_memstats_stderr().
* lib/thread.c: thread->funcname was being accessed in thread_call() after
it had been freed. Rearranged things so that thread_call() frees
funcname. Also made it so thread_master_free() cleans up cpu_record.
* lib/vty.c,h: Use global command_cr. Add vty_terminate().
* lib/zclient.c,h: Re-enable zclient_free().
|
|
2008-07-22 Paul Jakma <paul.jakma@sun.com>
* bgp_community.c: (community_str2com) assigns defaults to local
vars
* bgp_attr.c: (bgp_attr_parse) match format specifier to arg
* bgp_table.{c,h}: (bgp_table_top) can take a * to a const, quelling
warning in bgp_route.c
|
|
2008-07-02 Stephen Hemminger <stephen.hemminger@vyatta.com>
* bgp_table.{c,h}: (bgp_table_finish) Take a double pointer and
scrub pointer in caller, so fixing double-free crashes seen
with route-server-client configuration.
* *.c: Adjust all callers of bgp_table_finish
2008-07-02 Paul Jakma <paul.jakma@sun.com>
* bgp_table.{c,h}: (bgp_node_delete, bgp_table_free) shouldn't be
exported.
|
|
2006-03-12 Paul Jakma <paul.jakma@sun.com>
* bgp_attr.h: (struct attr) rearrange fields to avoid
wasted padding between them as much as possible.
(attr_count,attr_unknown_count) export new functions to
return number of counts of cached attributes.
* bgp_attr.c: (attr_count,attr_unknown_count) new functions to
return number of counts of cached attributes.
* bgp_route.h: (struct bgp_info) rearrange fields to avoid
wasted padding.
* bgp_table.h: (struct bgp_table) Add a count field, of number
of nodes in the table.
(struct bgp_node) rearrange fields to avoid
wasted padding between them, though I don't think there
was any in this case.
* bgp_table.c: (bgp_node_{delete,get}) Maintain the table node count.
(bgp_table_count) new function to access the table count.
|
|
2006-02-21 Paul Jakma <paul.jakma@sun.com>
* bgpd.h: move the clear_node_queue to be peer specific.
Add a new peer status flag, PEER_STATUS_CLEARING.
* bgp_table.h: (struct bgp_table) Add fields to record afi,
safi of the table.
(bgp_table_init) Take afi and safi to create table for.
* bgp_table.c: (bgp_table_init) record the afi and safi.
* bgp_nexthop.c: Update all calls to bgp_table_init.
* bgp_vty.c: ditto.
* bgpd.c: ditto.
* bgp_fsm.c: (bgp_timer_set) dont bring up a session which is
clearing.
* bgp_route.c: (general) Update all bgp_table_init calls.
(bgp_process_{rsclient,main}) clear_node is serialised
via PEER_STATUS_CLEARING and fsm now.
(struct bgp_clear_node_queue) can be removed. struct bgp_node
can be the queue item data directly, as struct peer can be
kept in the new wq global user data and afi/safi can be
retrieved via bgp_node -> bgp_table.
(bgp_clear_route_node) fix to get peer via wq->spec.data,
afi/safi via bgp_node->bgp_table.
(bgp_clear_node_queue_del) no more item data to delete, only
unlock the bgp_node.
(bgp_clear_node_complete) only need to unset CLEARING flag
and unlock struct peer.
(bgp_clear_node_queue_init) queue attaches to struct peer
now. record peer name as queue name.
(bgp_clear_route_table) If queue transitions to active,
serialise clearing by setting PEER_STATUS_CLEARING rather
than plugging process queue, and lock peer while queue
active.
Update to pass only bgp_node as per-queue-item specific data.
|
|
* (global) The great bgpd extern and static'ification.
* bgp_routemap.c: remove unused ROUTE_MATCH_ASPATH_OLD code
(route_set_metric_compile) fix u_int32_t to ULONG_MAX comparison
warnings.
* bgp_route.h: (bgp_process, bgp_withdraw, bgp_update) export these
used by various files which had their own private declarations,
in the case of mplsvpn - incorrect.
|
|
* bgpd/(general) refcount struct peer and bgp_info, hence allowing us
add work_queues for bgp_process.
* bgpd/bgp_route.h: (struct bgp_info) Add 'lock' field for refcount.
Add bgp_info_{lock,unlock} helper functions.
Add bgp_info_{add,delete} helpers, to remove need for
users managing locking/freeing of bgp_info and bgp_node's.
* bgpd/bgp_table.h: (struct bgp_node) Add a flags field, and
BGP_NODE_PROCESS_SCHEDULED to merge redundant processing of
nodes.
* bgpd/bgp_fsm.h: Make the ON/OFF/ADD/REMOVE macros lock and unlock
peer reference as appropriate.
* bgpd/bgp_damp.c: Remove its internal prototypes for
bgp_info_delete/free. Just use bgp_info_delete.
* bgpd/bgpd.h: (struct bgp_master) Add work_queue pointers.
(struct peer) Add reference count 'lock'
(peer_lock,peer_unlock) New helpers to take/release reference
on struct peer.
* bgpd/bgp_advertise.c: (general) Add peer and bgp_info refcounting
and balance how references are taken and released.
(bgp_advertise_free) release bgp_info reference, if appropriate
(bgp_adj_out_free) unlock peer
(bgp_advertise_clean) leave the adv references alone, or else
call bgp_advertise_free cant unlock them.
(bgp_adj_out_set) lock the peer on new adj's, leave the reference
alone otherwise. lock the new bgp_info reference.
(bgp_adj_in_set) lock the peer reference
(bgp_adj_in_remove) and unlock it here
(bgp_sync_delete) make hash_free on peer conditional, just in
case.
* bgpd/bgp_fsm.c: (general) document that the timers depend on
bgp_event to release a peer reference.
(bgp_fsm_change_status) moved up the file, unchanged.
(bgp_stop) Decrement peer lock as many times as cancel_event
canceled - shouldnt be needed but just in case.
stream_fifo_clean of obuf made conditional, just in case.
(bgp_event) always unlock the peer, regardless of return value
of bgp_fsm_change_status.
* bgpd/bgp_packet.c: (general) change several bgp_stop's to BGP_EVENT's.
(bgp_read) Add a mysterious extra peer_unlock for ACCEPT_PEERs
along with a comment on it.
* bgpd/bgp_route.c: (general) Add refcounting of bgp_info, cleanup
some of the resource management around bgp_info. Refcount peer.
Add workqueues for bgp_process and clear_table.
(bgp_info_new) make static
(bgp_info_free) Ditto, and unlock the peer reference.
(bgp_info_lock,bgp_info_unlock) new exported functions
(bgp_info_add) Add a bgp_info to a bgp_node in correct fashion,
taking care of reference counts.
(bgp_info_delete) do the opposite of bgp_info_add.
(bgp_process_rsclient) Converted into a work_queue work function.
(bgp_process_main) ditto.
(bgp_processq_del) process work queue item deconstructor
(bgp_process_queue_init) process work queue init
(bgp_process) call init function if required, set up queue item
and add to queue, rather than calling process functions directly.
(bgp_rib_remove) let bgp_info_delete manage bgp_info refcounts
(bgp_rib_withdraw) ditto
(bgp_update_rsclient) let bgp_info_add manage refcounts
(bgp_update_main) ditto
(bgp_clear_route_node) clear_node_queue work function, does
per-node aspects of what bgp_clear_route_table did previously
(bgp_clear_node_queue_del) clear_node_queue item delete function
(bgp_clear_node_complete) clear_node_queue completion function,
it unplugs the process queues, which have to be blocked while
clear_node_queue is being processed to prevent a race.
(bgp_clear_node_queue_init) init function for clear_node_queue
work queues
(bgp_clear_route_table) Sets up items onto a workqueue now, rather
than clearing each node directly. Plugs both process queues to
avoid potential race.
(bgp_static_withdraw_rsclient) let bgp_info_{add,delete} manage
bgp_info refcounts.
(bgp_static_update_rsclient) ditto
(bgp_static_update_main) ditto
(bgp_static_update_vpnv4) ditto, remove unneeded cast.
(bgp_static_withdraw) see bgp_static_withdraw_rsclient
(bgp_static_withdraw_vpnv4) ditto
(bgp_aggregate_{route,add,delete}) ditto
(bgp_redistribute_{add,delete,withdraw}) ditto
* bgpd/bgp_vty.c: (peer_rsclient_set_vty) lock rsclient list peer
reference
(peer_rsclient_unset_vty) ditto, but unlock same reference
* bgpd/bgpd.c: (peer_free) handle frees of info to be kept for lifetime
of struct peer.
(peer_lock,peer_unlock) peer refcount helpers
(peer_new) add initial refcounts
(peer_create,peer_create_accept) lock peer as appropriate
(peer_delete) unlock as appropriate, move out some free's to
peer_free.
(peer_group_bind,peer_group_unbind) peer refcounting as
appropriate.
(bgp_create) check CALLOC return value.
(bgp_terminate) free workqueues too.
* lib/memtypes.c: Add MTYPE_BGP_PROCESS_QUEUE and
MTYPE_BGP_CLEAR_NODE_QUEUE
|
|
* bgp_fsm.h: Add extern qualifier to exported functions
* bgp_nexthop.c: add static to nexthop specific globals
* *.h: Add guard defines
|
|
(at Technical University of Madrid as part of Euro6ix Project)
Enhanced Route Server functionality and Route-Maps:
* bgpd/bgpd.h: Modified 'struct peer' and 'struct bgp_filter' to
support rs-clients. A 'struct bgp_table *rib' has been added to the
first (to mantain a separated RIB for each rs-client) and two new
route-maps have been added to the last (for import/export policies).
Added the following #defines: RMAP_{IN|OUT|IMPORT|EXPORT|MAX},
PEER_RMAP_TYPE_{IMPORT|EXPORT} and BGP_CLEAR_SOFT_RSCLIENT.
* bgpd/bgpd.c: Modified the functions that create/delete/etc peers in
order to consider the new fields included in 'struct peer' for
supporting rs-clients, i.e. the import/export route-maps and the
'struct bgp_table'.
* bgpd/bgp_route.{ch}: Modified several functions related with
receiving/sending announces in order to support the new Route Server
capabilities.
Function 'bgp_process' has been reorganized, creating an auxiliar
function for best path selection ('bgp_best_selection').
Modified 'bgp_show' and 'bgp_show_route' for displaying information
about any RIB (and not only the main bgp RIB).
Added commands for displaying information about RS-clients RIBs:
'show bgp rsclient (A.B.C.D|X:X::X:X)', 'show bgp rsclient
(A.B.C.D|X:X::X:X) X:X::X:X/M', etc
* bgpd/bgp_table.{ch}: The structure 'struct bgp_table' now has two
new fields: type (which can take the values BGP_TABLE_{MAIN|RSCLIENT})
and 'void *owner' which points to 'struct bgp' or 'struct peer' which
owns the table.
When creating a new bgp_table by default 'type=BGP_TABLE_MAIN' is set.
* bgpd/bgp_vty.c: The commands 'neighbor ... route-server-client' and
'no neighbor ... route-server-client' now not only set/unset the flag
PEER_FLAG_RSERVER_CLIENT, but they create/destroy the 'struct
bgp_table' of the peer. Special actions are taken for peer_groups.
Command 'neighbor ... route-map WORD (in|out)' now also supports two
new kinds of route-map: 'import' and 'export'.
Added commands 'clear bgp * rsclient', etc. These commands allow a new
kind of soft_reconfig which affects only the RIB of the specified
RS-client.
Added commands 'show bgp rsclient summary', etc which display a
summary of the rs-clients configured for the corresponding address
family.
* bgpd/bgp_routemap.c: A new match statement is available,
'match peer (A.B.C.D|X:X::X:X)'. This statement can only be used in
import/export route-maps, and it matches when the peer who announces
(when used in an import route-map) or is going to receive (when used
in an export route-map) the route is the same than the one specified
in the statement.
For peer-groups the statement matches if the specified peer is member
of the peer-group.
A special version of the command, 'match peer local', matches with
routes originated by the Route Server (defined with 'network ...',
redistributed routes and default-originate).
* lib/routemap.{ch}: Added a new clause 'call NAME' for use in
route-maps. It jumps into the specified route-map and when it returns
the first route-map ends if the called RM returns DENY_MATCH, or
continues in other case.
|
|
|
