Luajit has a strange memory requirement, it's 'states' need to be in the
first 2G of the process' memory.
This patch improves the pool approach by moving it to the front of the
start up.
A new config option 'luajit.states' is added to control how many states
are preallocated. It defaults to 128.
Add a warning when more states are used then preallocated. This may fail
if flow/stream/detect engines use a lot of memory. Add hint at exit that
gives the max states in use if it's higher than the default.
Add support for the ENIP/CIP Industrial protocol
This is an app layer implementation which uses the "enip" protocol
and "cip_service" and "enip_command" keywords
Implements AFL entry points
Register keywords globally at start up.
Create a map of the registery per detection engine. This we need because
the sgh_mpm_context value is set per detect engine.
Remove APP_MPMS_MAX.
Add a generic 'capture' section to the YAML:
# general settings affecting packet capture
capture:
# disable NIC offloading. It's restored when Suricata exists.
# Enabled by default
#disable-offloading: false
#
# disable checksum validation. Same as setting '-k none' on the
# commandline
#checksum-validation: none
When running in live mode, the new default 'auto' value of
unix-command.enabled causes unix-command to be activated. This
will allow users of live capture to benefit from the feature and
result in no side effect for user running in offline capture.
We did ignore additional USR2 signals while a rule-reload was running.
This changes the counter to be incremented with every additional USR2
signal so we don't ignore them anymore but it's still limited to prevent
huge overload or even overflow.
As the logging modules are no longer threading modules, rename
them so they don't look like they are being registered as
threading modules.
Also, move the registration to the output.c which will handle
registration of the loggers.
Introduces a new thread module, TMM_LOGGER, which is the
root most logger.
Only handles loggers in the packet path, stats and flow
logging are not included.
The loggers are made up of a hierarchy of loggers. At the top we
have the root logger which is the main entry point to
logging. Under the root there exists parent loggers that are the
entry point for specific types of loggers such as packet logger,
transaction loggers, etc. Each parent logger may have 0 or more
loggers that actual handle the job of producing output to something
like a file.
Fix improper fread string handling. Improve error handling.
Skip trailing spaces for slightly more pretty printing.
Coverity CID 400763.
Thanks to Steve Grubb for helping address this issue.
The structure for create the alert preceding each tagged packet
was not being initialized, preventing tagged packets from being
logged.
Note: Snort unified2 does not precede tagged packets with an
alert like is done here, so this just fixes what the code
intended to do, it does not make it Snort unified2
compatible.
Address issue:
https://redmine.openinfosecfoundation.org/issues/1854
With the additional ParseInterfacesList the packet processing threads
were doubled since the Interface was included twice unless the device
was passed via the commandline with af-packet=IF.
The additonal ParseInterfacesList isn't necessary so remove it again
Now that the FlowWorker handles the TCP Stream directly, having
the TCP engine as a thread module is no longer needed.
This patch removes the registration.
Initial version of the 'FlowWorker' thread module. This module
combines Flow handling, TCP handling, App layer handling and
Detection in a single module. It does all flow related processing
under a single flow lock.
This new API allows for different SPM implementations, using a function
pointer table like that used for MPM.
This change also switches over the paths that make use of
DetectContentData (which previously used BoyerMoore directly) to the new
API.
NSS library was not deinit at exit resulting in memory leak. As
it is useless for a config test, the patch updates the code so it
is not initialized.
Patch also calls MagicDeinit to free memory used by libmagic.
This patch introduces a new set of commandline options meant for
assisting in fuzz testing the app layer implementations.
Per protocol, 2 commandline options are added:
--afl-http-request=<filename>
--afl-http=<filename>
In the former case, the contents of the file are passed directly to
the HTTP parser as request data.
In the latter case, the data is devided between request and responses.
First 64 bytes are request, then next 64 are response, next 64 are
request, etc, etc.
When fuzzing, AFL will create lots of malformed rules. We don't want
to error out on those. As we're fuzzing the parser any non-crash
should return 0. Crashes (ASAN or not) will return a non-0 code.
Add regex setup and free util functions. Keywords often use a regex
to parse rule input. Introduce a common function to do this setup.
Also create a list of registered regexes to free at engine shutdown.
Unit testing support macros for failing on expressions,
as well as passing tests on expressions.
If fatal unittests are enabled BUG_ON will be triggered for
an assertion providing the line number of the failure, otherwise
the test will simply fail.
Moved the fatal flag to a global var instead of a configuration
parameter for ease of access from a macro.
Instead of detect-engine which used a list for no good reason, use a
simple map now.
detect:
profile: medium
custom-values:
toclient-groups: 3
toserver-groups: 25
sgh-mpm-context: auto
inspection-recursion-limit: 3000
# If set to yes, the loading of signatures will be made after the capture
# is started. This will limit the downtime in IPS mode.
#delayed-detect: yes
Per rule group tracking of checks, use of lists, mpm matches,
post filter counts.
Logs SGH id so it can be compared with the rule_group.json output.
Implemented both in a human readable text format and a JSON format.
Simplify handling of USR2 signal. The SCLogInfo usage could lead to
dead locks as the SCLog API can do many complicated things including
memory allocations, syslog calls, libjansson message construction.
If an existing malloc call was interupted, it could lead to the
following dead lock:
0 __lll_lock_wait_private () at ../nptl/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:97
1 0x0000003140c7d2df in _L_lock_10176 () from /lib64/libc.so.6
2 0x0000003140c7ab83 in __libc_malloc (bytes=211543457408) at malloc.c:3655
3 0x0000003140c80ec2 in __strdup (s=0x259ca40 "[%i] %t - (%f:%l) <%d> (%n) -- ") at strdup.c:43
4 0x000000000059dd4a in SCLogMessageGetBuffer (tval=0x7fff52b47360, color=1, type=SC_LOG_OP_TYPE_REGULAR, buffer=0x7fff52b47370 "", buffer_size=2048,
log_format=0x259ca40 "[%i] %t - (%f:%l) <%d> (%n) -- ", log_level=SC_LOG_INFO, file=0x63dd00 "suricata.c", line=287, function=0x640f50 "SignalHandlerSigusr2StartingUp", error_code=SC_OK,
message=0x7fff52b47bb0 "Live rule reload only possible after engine completely started.") at util-debug.c:307
5 0x000000000059e940 in SCLogMessage (log_level=SC_LOG_INFO, file=0x63dd00 "suricata.c", line=287, function=0x640f50 "SignalHandlerSigusr2StartingUp", error_code=SC_OK,
message=0x7fff52b47bb0 "Live rule reload only possible after engine completely started.") at util-debug.c:549
6 0x000000000057e374 in SignalHandlerSigusr2StartingUp (sig=12) at suricata.c:287
7 <signal handler called>
8 _int_malloc (av=0x3140f8fe80, bytes=<value optimized out>) at malloc.c:4751
9 0x0000003140c7ab1c in __libc_malloc (bytes=296) at malloc.c:3657
10 0x0000000000504d55 in FlowAlloc () at flow-util.c:60
11 0x00000000004fd909 in FlowInitConfig (quiet=0 '\000') at flow.c:454
12 0x0000000000584c8e in main (argc=6, argv=0x7fff52b4a3b8) at suricata.c:2300
This patch simply sets a variable and lets the main loop act on that.
On systems like Gentoo where _FORTIFY_SOURCE is already defined like
FORTIFY_SOURCE=((defined __OPTIMIZE && OPTIMIZE > 0) ? 2 : 0) the use
within the printf function (%d) won't result in the correct value and
we end up with 'defined' undeclared compile error. This workaround makes
sure that just the resolved value is checked and then printed.
It is not possible to use simultaneously -l and unix socket
runmode because setting the log directory make it final so
not modifable by other call.
It is a implementation limitation but it does not make sense
to set logging directory to have it overwritten by the first
directory specified 'by pcap-file'. So it seems correct to
only trigger an error if this both options are used at the same
time.
When user asks for a configuration dump, it is useless to display
the version and CPU info. Also initializing the log system conduct
to overwrite the some log files and in particular suricata.log and
this is annoying as a command should not interfere with a running
daemon.
When there are many threads and/or the packet pool (max-pending-packets) is
small, a potential dead lock exists between the packet pool return pool
logic and the capture threads. The autofp workers together can have all the
packets in their return pools, while the capture thread(s) are waiting at an
empty pool. A race between the worker threads and the capture thread, where
the latter signals the former, is lost by the capture thread. Now everyone
is waiting.
To avoid this scenario, this patch makes the previously hardcoded 'return
pool' threshold dynamic based on the number of threads and the packet pool
size.
It sets the threshold to the max pending packets value, divided by the number
of lister threads. The max value hasn't changed. Normally, in the autofp
runmode these are the stream/detect/log worker threads.
The max_pending_return_packets value needs to stay below the packet pool size
of the 'producers' (normally pkt capture threads but also flow timeout
injection) to avoid the deadlock.
As it's quite impossible at this time to learn how many threads will be
created before starting the runmodes, and thus spawning the threads and
already initializing the packet pools, this code sets a global variable
after runmode setup, but before the threads are 'unpaused'.
An design error was made when doing the TLS storage module which
has been made dependant of the TLS logging. At the time there was
only one TLS logging module but there is now two different ones.
By putting the TLS store module in a separate module, we can now
use EVE output and TLS store at the same time.
Due to an error at initialization, the stream engine would not disable
'raw' reassembly automatically when --disable-detection was used.
This lead to segments not getting cleared from the segment lists.
Don't kill flow manager and recyclers before the rest of the threads. The
packet threads may still have packets from their pools. As the flow threads
would destroy their pools the packets would be lost.
This patch doesn't kill the threads, it just pulls them out of their run
loop and into a wait loop. The packet pools won't be cleared until all
threads are killed.
Wait for flow management threads to close before moving on to the
next steps in the shutdown process.
Don't destroy flow force reassembly packet pool too early. Worker
threads may still want to return packets to it.
Load the YAML into a prefix "detect-engine-reloads.N" where N is the
reload counter. This way we can load the updated config w/o overwriting
the current one.
Use new DetectEngineReload() function. It's called from the main loop
instead of it being spawned into it's own temporary thread. This greatly
simplifies the signal handling.
An added advantage is that this seems to improve the memory usage.
Related to bug #1358
The minimal detect engine has only the minimal memory use and setup
time. It's to be used for 'delayed' detect where the first detection
engine is essentially empty.
The threads setup are also minimal.
Instead of threading logic with dummy slots and all, use the regular
reload logic for delayed detect.
This means we pass a empty detect engine to the threads and then
reload (live swap) it as soon as the engine is running.
Update detect engine management to make it easier to reload the detect
engine.
Core of the new approach is a 'master' ctx, that keeps a list of one or
more detect engines. The detect engines will not be passed to any thread
directly, but instead will only be accessed through the detect engine
thread contexts. As we can replace those atomically, replacing a detect
engine becomes easier.
Each thread keeps a reference to its detect context. When a detect engine
is replaced or removed, it's added to a free list. Once its reference
count reaches 0, it is freed.
Victor Julien
Jason Ish
kwong
Duarte Silva
Eric Leblond
Andreas Herz
Arturo Borrero Gonzalez
Justin Viiret
Mats Klepsland
Alexander Gozman
Tom DeCanio