When BIRD has no free memory mapped, it allocates several pages in
advance just to be sure that there is some memory available if needed.
This hysteresis tactics works quite well to reduce memory ping-ping with
kernel.
Yet it had a subtle bug: this pre-allocation didn't take a memory
coldlist into account, therefore requesting new pages from kernel even
in cases when there were other pages available. This led to slow memory
bloating.
To demonstrate this behavior fast enough to be seen well, you may:
* temporarily set the values in sysdep/unix/alloc.c as follows to
exacerbate the issue:
#define KEEP_PAGES_MAIN_MAX 4096
#define KEEP_PAGES_MAIN_MIN 1000
#define CLEANUP_PAGES_BULK 4096
* create a config file with several millions of static routes
* periodically disable all static protocols and then reload config
* log memory consumption
This should give you a steady growth rate of about 16kB per cycle. If
you don't set the values this high, the issue happens much more slowly,
yet after 14 days of running, you are going to see an OOM kill.
After this fix, pre-allocation uses the memory coldlist to get some hot
pages and the same test as described here gets you a perfectly stable
constant memory consumption (after some initial wobbling).
Thanks to NIX-CZ for reporting and helping to investigate this issue.
Thanks to Santiago for finding the cause in the code.
The usage pattern implemented in allocator seems to be incompatible with
transparent huge pages, as memory released using madvise(MADV_DONTNEED)
with regular page size and alignment does not seem to trigger demotion
of huge pages back to regular pages, even when significant number of
pages is released. Even if demotion is triggered when system memory
is low, it still breaks memory accounting.
Log message before aborting due to watchdog timeout. We have to use
async-safe write to debug log, as it is done in signal handler.
Minor changes from committer.
When there is a continuos stream of CLI commands, cli_get_command()
always returns 1 (there is a new command). Anyway, the socket receive
buffer was reset only when there was no command at all, leading to a
strange behavior: after a while, the CLI receive buffer came to its end,
then read() was called with zero size buffer, it returned 0 which was
interpreted as EOF.
The patch fixes that by resetting the buffer position after each command
and moving remaining data at the beginning of buffer.
Thanks to Maria Matejka for examining the bug and for the original bugfix.
BIRD keeps a previous (old) configuration for the purpose of undo. The
existing code frees it after a new configuration is successfully parsed
during reconfiguration. That causes memory usage spikes as there are
temporarily three configurations (old, current, and new). The patch
changes it to free the old one before parsing the new one (as user
already requested a new config). The disadvantage is that undo is
not available after failed reconfiguration.
Memory unmapping causes slow address space fragmentation, leading in
extreme cases to failing to allocate pages at all. Removing this problem
by keeping all the pages allocated to us, yet calling madvise() to let
kernel dispose of them.
This adds a little complexity and overhead as we have to keep the
pointers to the free pages, therefore to hold e.g. 1 GB of 4K pages with
8B pointers, we have to store 2 MB of data.
This is a reimplementation of commit 0f2be469f8
by Alexander Zubkov. In the master branch, changes in commit eb937358
broke setting of channel preference for alien routes learned during
scan. The preference was set only for async routes.
The original solution is extended here to accomodate for v3 specifics.
Changes in commit eb937358 broke setting of channel preference for alien
routes learned during scan. The preference was set only for async routes.
Move common attribute processing part of functions krt_learn_async() and
krt_learn_async() to a separate function to have only one place for such
changes.
Remove compile-time sysdep option CONFIG_ALL_TABLES_AT_ONCE, replace it
with runtime ability to run either separate table scans or shared scan.
On Linux, use separate table scans by default when the netlink socket
option NETLINK_GET_STRICT_CHK is available, but retreat to shared scan
when it fails.
Running separate table scans has advantages where some routing tables are
managed independently, e.g. when multiple routing daemons are running on
the same machine, as kernel routing table modification performance is
significantly reduced when the table is modified while it is being
scanned.
Thanks Daniel Gröber for the original patch and Toke Høiland-Jørgensen
for suggestions.
The learnt routes are now pushed all into the connected table, not only
the best one. This shouldn't do any damage in well managed setups, yet
it should be noted that it is a change of behavior.
If anybody misses a feature which they implemented by misusing this
internal learn table, let us know, we'll consider implementing it in a
better way.
Passing protocol to preexport was in fact a historical relic from the
old times when channels weren't a thing. Refactoring that to match
current extensibility needs.
There were quite a lot of conflicts in flowspec validation code which
ultimately led to some code being a bit rewritten, not only adapted from
this or that branch, yet it is still in a limit of a merge.
For now, all route attributes are stored as eattrs in ea_list. This
should make route manipulation easier and it also allows for a layered
approach of route attributes where updates from filters will be stored
as an overlay over the previous version.
As there is either a nexthop or another destination specification
(or othing in case of ROAs and Flowspec), it may be merged together.
This code is somehow quirky and should be replaced in future by better
implementation of nexthop.
Also flowspec validation result has its own attribute now as it doesn't
have anything to do with route nexthop.
This doesn't do anything more than to put the whole structure inside
adata. The overall performance is certainly going downhill; we'll
optimize this later.
Anyway, this is one of the latest items inside rta and in several
commits we may drop rta completely and move to eattrs-only routes.
Some tokens are both keywords and symbols. For now, we allow only
specific keywords to be redefined; in future, more of the keywords may
be added to this category.
The redefinable keywords must be specified in any .Y file as follows:
toksym: THE_KEYWORD ;
See proto/bgp/config.Y for an example.
Also dropped a lot of unused terminals.
Changes in internal API:
* Every route attribute must be defined as struct ea_class somewhere.
* Registration of route attributes known at startup must be done by
ea_register_init() from protocol build functions.
* Every attribute has now its symbol registered in a global symbol table
defined as SYM_ATTRIBUTE
* All attribute ID's are dynamically allocated.
* Attribute value custom formatting hook is defined in the ea_class.
* Attribute names are the same for display and filters, always prefixed
by protocol name.
Also added some unit testing code for filters with route attributes.
This commit removes the EAF_TYPE_* namespace completely and also for
route attributes, filter-based types T_* are used. This simplifies
fetching and setting route attributes from filters.
Also, there is now union bval which serves as an universal value holder
instead of private unions held separately by eattr and filter code.
When BIRD was munmapping too many pages, it sometimes aborted, saying
that munmap failed with "Not enough memory" as the address space was
getting more and more fragmented.
There is a workaround in place, simply keeping that page for future use,
yet it has never been compiled in because I somehow forgot to include
errno.h. And because I also thought that somebody may have ENOMEM not
defined (why?!), there was a check which quietly omitted that
workaround.
Anyway, ENOMEM is POSIX. It's an utter nonsense to check for its
existence. If it doesn't exist, something is broken.
Add option to socket interface for nonlocal binding, i.e. binding to an
IP address that is not present on interfaces. This behaviour is enabled
when SKF_FREEBIND socket flag is set. For Linux systems, it is
implemented by IP_FREEBIND socket flag.
Minor changes done by commiter.
This commit prevents use-after-free of routes belonging to protocols
which have been already destroyed, delaying also all the protocols'
shutdown until all of their routes have been finally propagated through
all the pipes down to the appropriate exports.
The use-after-free was somehow hypothetic yet theoretically possible in
rare conditions, when one BGP protocol authors a lot of routes and the
user deletes that protocol by reconfiguring in the same time as next hop
update is requested, causing rte_better() to be called on a
not-yet-pruned network prefix while the owner protocol has been already
freed.
In parallel execution environments, this would happen an inter-thread
use-after-free, causing possible heisenbugs or other nasty problems.
There is a simple universal IO loop, taking care of events, timers and
sockets. Primarily, one instance of a protocol should use exactly one IO
loop to do all its work, as is now done in BFD.
Contrary to previous versions, the loop is now launched and cleaned by
the nest/proto.c code, allowing for a protocol to just request its own
loop by setting the loop's lock order in config higher than the_bird.
It is not supported nor checked if any protocol changed the requested
lock order in reconfigure. No protocol should do it at all.
In previous versions, every thread used its own time structures,
effectively leading to different time in every thread and strange
logging messages.
The time processing code now uses global atomic variables to keep
current time available for fast concurrent reading and safe updates.
* internal tables are now more standalone, having their own import and
export hooks
* route refresh/reload uses stale counter instead of stale flag,
allowing to drop walking the table at the beginning
* route modify (by BGP LLGR) is now done by a special refeed hook,
reimporting the modified routes directly without filters
Channels have now included rt_import_req and rt_export_req to hook into
the table instead of just one list node. This will (in future) allow for:
* channel import and export bound to different tables
* more efficient pipe code (dropping most of the channel code)
* conversion of 'show route' to a special kind of export
* temporary static routes from CLI
The import / export states are also updated to the new algorithms.
Routes are now allocated only when they are just to be inserted to the
table. Updating a route needs a locally allocated route structure.
Ownership of the attributes is also now not transfered from protocols to
tables and vice versa but just borrowed which should be easier to handle
in a multithreaded environment.
We can also quite simply allocate bigger blocks. Anyway, we need these
blocks to be aligned to their size which needs one mmap() two times
bigger and then two munmap()s returning the unaligned parts.
The user can specify -B <N> on startup when <N> is the exponent of 2,
setting the block size to 2^N. On most systems, N is 12, anyway if you
know that your configuration is going to eat gigabytes of RAM, you are
almost forced to raise your block size as you may easily get into memory
fragmentation issues or you have to raise your maximum mapping count,
e.g. "sysctl vm.max_map_count=(number)".
Add a wrapper function in sysdep to get random bytes, and required checks
in configure.ac to select how to do it. The configure script tries, in
order, getrandom(), getentropy() and reading from /dev/urandom.
From now, there are no auxiliary pointers stored in the free slab nodes.
This led to strange debugging problems if use-after-free happened in
slab-allocated structures, especially if the structure's first member is
a next pointer.
This also reduces the memory needed by 1 pointer per allocated object.
OTOH, we now rely on pages being aligned to their size's multiple, which
is quite common anyway.
In general, events are code handling some some condition, which is
scheduled when such condition happened and executed independently from
I/O loop. Work-events are a subgroup of events that are scheduled
repeatedly until some (often significant) work is done (e.g. feeding
routes to protocol). All scheduled events are executed during each
I/O loop iteration.
Separate work-events from regular events to a separate queue and
rate limit their execution to a fixed number per I/O loop iteration.
That should prevent excess latency when many work-events are
scheduled at one time (e.g. simultaneous reload of many BGP sessions).
This is an implementation of draft-walton-bgp-hostname-capability-02.
It is implemented since quite some time for FRR and in datacenter, this
gives a nice output to avoid using IP addresses.
It is disabled by default. The hostname is retrieved from uname(2) and
can be overriden with "hostname" option. The domain name is never set
nor displayed.
Minor changes by committer.
So one can define kernel protocol template without channels.
For other protocols, it is either irrelevant or already done.
Thanks to Clemens Schrimpe for the bugreport.
The log subsystem should be locked earlier, as default_log_list() may
internally manipulate with the current_log_list (if it is also a default
log list).
The static logging structures are reused, we need to reinitialize them
otherwise add_tail() would fail in debug build. Reinitializing these
structures should be fine as the list they belong to is being
reinitialized on entry to the very same function.
Thanks to Andreas Rammhold and Mikael Magnusson for patches.
This is a quick workaround for an issue where configured logfiles are
opened/created during parsing of a config file even when parse-and-exit
option is active. We should later refactor the logging code to avoid
opening log during parsing altogether.
When dynamic BGP with remote range is configured, MD5SIG needs to use
newer socket option (TCP_MD5SIG_EXT) to specify remote addres range for
listening socket.
Thanks to Adam Kułagowski for the suggestion.
The old code stored route verdicts and temporary routes directly in
rtable. The new code do not store received routes (it immediately
compares them with exported routes and resolves conflicts) and uses
internal bitmap to keep track of which routes were received and which
needs to be reinstalled.
By not putting 'invalid' temporary routes to rtable, we keep rtable
in consistent state, therefore scan no longer needs to be atomic
operation and could be splitted to multiple events.
Use a hierarchical bitmap in a routing table to assign ids to routes, and
then use bitmaps (indexed by route id) in channels to keep track whether
routes were exported. This avoids unreliable and inefficient re-evaluation
of filters for old routes in order to determine whether they were exported.
Names read from texfiles in /etc/iproute2/* are normalized by replacing
non-alphanumeric chars with underscore. The patch fixes handling of
uppercase letters, which were handled as non-alphanumberic.
Thanks to Igor Gavrilov for the bugreport.
The C11 specification allows only sig_atomic_t and _Atomic variable
access. All other accesses to global variables are undefined behavior.
Using int was probably OK on x86 and x86_64; yet there were some reports
from other architectures (especially some MIPS) that in rare cases,
after issuing SIGHUP, BIRD did strange things.
The command initiating planned graceful restart including bird shutdown
should be called 'graceful restart' instead of 'graceful down', as the
later should be reserved for graceful shutdown in style of RFC 8326.
When 'graceful down' command is entered, protocols are shut down
with regard to graceful restart. Namely Kernel protocol does
not remove routes and BGP protocol does not send notification,
just closes the connection.
Support for dynamically spawning BGP protocols for incoming connections.
Use 'neighbor range' to specify range of valid neighbor addresses, then
incoming connections from these addresses spawn new BGP instances.
The temporary atttributes are no longer removed by ea_do_prune(), but
they are undefined by store_tmp_attrs() protocol hooks. This fixes
several bugs where temporary attributes were removed when they should
not or not removed when they should be. The flag EAF_TEMP is no longer
needed and was removed.
Update all protocol make_tmp_attrs() / store_tmp_attrs() hooks to use
helper functions and to handle unset attributes properly.
Also fix some related bugs like improper handling of empty eattr list.
This is a major change of how the filters are interpreted. If everything
works how it should, it should not affect you unless you are hacking the
filters themselves.
Anyway, this change should make a huge improvement in the filter performance
as previous benchmarks showed that our major problem lies in the
recursion itself.
There are also some changes in nest and protocols, related mostly to
spreading const declarations throughout the whole BIRD and also to
refactored dynamic attribute definitions. The need of these came up
during the whole work and it is too difficult to split out these
not-so-related changes.
Since v2 we have multiple listening BGP sockets, and each BGP protocol
has associated one of them. Use listening socket that accepted the
incoming connection as a key in the dispatch process so only BGP
protocols assocaited with that listening socket can be selected.
This is necesary for proper dispatch when VRFs are used.
FreeBSD silently changes TTL to 1 when MSG_DONTROUTE is used, even when
it is explicitly set to another value. That breaks TTL security sockets,
including BFD which always uses TTL 255. Bad FreeBSD!
The old behavior was that enabling debugging did many nontrivial changes
in BIRD behavior. The patch changes it that these changes are generally
independent. Compiling with --enable-debug now just enables compile-time
debug macros, but do not automatically activate debug mode (-d) nor local
mode (-l). Debug mode with output to file (-D) do not force foreground
mode (-f), therefore there is no need for backgroud option (-b), which is
removed. Also fixes a bug when the default log target in -D mode was
stderr instead of given debug file.
Once upon a time, far far away, there were the old Bird developers
discussing what direction of route flow shall be called import and
export. They decided to say "import to protocol" and "export to table"
when speaking about a protocol. When speaking about a table, they
spoke about "importing to table" and "exporting to protocol".
The latter terminology was adopted in configuration, then also the
bird CLI in commit ea2ae6dd0 started to use it (in year 2009). Now
it's 2018 and the terminology is the latter. Import is from protocol to
table, export is from table to protocol. Anyway, there was still an
import_control hook which executed right before route export.
One thing is funny. There are two commits in April 1999 with just two
minutes between them. The older announces the final settlement
on config terminology, the newer uses the other definition. Let's see
their commit messages as the git-log tool shows them (the newer first):
commit 9e0e485e50
Author: Martin Mares <mj@ucw.cz>
Date: Mon Apr 5 20:17:59 1999 +0000
Added some new protocol hooks (look at the comments for better explanation):
make_tmp_attrs Convert inline attributes to ea_list
store_tmp_attrs Convert ea_list to inline attributes
import_control Pre-import decisions
commit 5056c559c4
Author: Martin Mares <mj@ucw.cz>
Date: Mon Apr 5 20:15:31 1999 +0000
Changed syntax of attaching filters to protocols to hopefully the final
version:
EXPORT <filter-spec> for outbound routes (i.e., those announced
by BIRD to the rest of the world).
IMPORT <filter-spec> for inbound routes (i.e., those imported
by BIRD from the rest of the world).
where <filter-spec> is one of:
ALL pass all routes
NONE drop all routes
FILTER <name> use named filter
FILTER { <filter> } use explicitly defined filter
For all protocols, the default is IMPORT ALL, EXPORT NONE. This includes
the kernel protocol, so that you need to add EXPORT ALL to get the previous
configuration of kernel syncer (as usually, see doc/bird.conf.example for
a bird.conf example :)).
Let's say RIP to this almost 19-years-old inconsistency. For now, if you
import a route, it is always from protocol to table. If you export a
route, it is always from table to protocol.
And they lived happily ever after.
The new MRT protocol is responsible for periodic RIB table dumps in the
MRT format (RFC 6396). Also the existing code for BGP4MP MRT dumps is
refactored and splitted between BGP to MRT protocols, will be more
integrated into MRT in the future.
Example:
protocol mrt {
table "*";
filename "%N_%F_%T.mrt";
period 60;
}
It is partially based on the old MRT code from Pavel Tvrdik.
Allow to specify log file size limit and ensure that log file is rotated
to secondary name to avoid exceeding of log size limit.
The patch also fixes a bug related to keeping old fds open after
reconfiguration and using old fds after 'configure undo'.
no more warnings
No more warnings over me
And while it is being compiled all the log is black and white
Release BIRD now and then let it flee
(use the melody of well-known Oh Freedom!)
BSD systems cannot use SO_DONTROUTE, because it does not work properly
with multicast packets (perhaps it tries to find iface based on multicast
group address). But we can use MSG_DONTROUTE sendmsg() flag for unicast
packets. Works on FreeBSD, is ignored on OpenBSD and is broken on NetBSD
(i guess due to integrated routing table and ARP table).
This is a fundamental change of an original (1999) concept of route
processing inside BIRD. During import/export, there was a temporary
ea_list created which was to be used instead of the another one inside
the route itself.
This led to some confusion, quirks, and strange filter code that handled
extended route attributes. Dropping it now.
The protocol interface has changed in an uniform way -- the
`struct ea_list *attrs` argument has been removed from store_tmp_attrs(),
import_control(), rt_notify() and get_route_info().
Use full time precision to initialize random generator. The old
code was prone to initialize it to the same values in specific
circumstances (boot without RTC, multiple VMs starting at once).
This patch adds support for source-specific IPv6 routes to BIRD core.
This is based on Dean Luga's original patch, with the review comments
addressed. SADR support is added to network address parsing in confbase.Y
and to the kernel protocol on Linux.
Currently there is no way to mix source-specific and non-source-specific
routes (i.e., SADR tables cannot be connected to non-SADR tables).
Thanks to Toke Hoiland-Jorgensen for the original patch.
Minor changes by Ondrej Santiago Zajicek.
On Linux, setting the ToS will also set the priority and the range of
accepted values is quite limited (masked by 0x1e). Therefore, 0xc0 is
translated to a priority of 0, not something we want, overriding the
"7" priority which was set previously explicitely. To avoid that, just
move setting priority later in the code.
Thanks to Vincent Bernat for the patch.
A filter should log messages only if executed explicitly (e.g., during
route export or route import). When a filter is executed for technical
reasons (e.g., to establish whether a route was exported before), it
should run silently.
