Set default timeout for the flow manager to wake up to 1 second. The 0.4 sec
performed best on a Xeon, but in kvm vm's it was horrible:
32 bit vm: 60% cpu for flowmgr when idle.
64 bit vm: 30% cpu for flowmgr when idle.
With the 1 second timeout both are at 0.3% cpu.
Short sleep can lead to some really annoying performance issue in
some environnement like virtual systems. This technic was used in
the flow manager. This patch uses an alternate approach based on
a timed condition which is triggered each time a new flow has to
be created. This avoid to run out of flow. A counter is also done
to be able not to run the cleaning code at each new flow.
Filestore keyword by default (... filestore; ... ) marks only the file in the
same direction as the rule match for storing. This makes sense when inspecting
individual files (filemagic, filename, etc) but not so much when looking at
suspicious file requests, where the actual file is in the response.
The filestore keyword now takes 2 optional options:
filestore:<direction>,<scope>;
By default the direction is "same as rule match", and scope is "currently
inspected file".
For direction the following values are possible: "request" and "to_server",
"response" and "to_client", "both".
For scope the following values are possible: "tx" for all files in the current
HTTP/1.1 transation, "ssn" and "flow" for all files in the session/flow.
For the above case, where a suspious request should lead to a response file
download, this would work:
alert http ... content:"/suspicious/"; http_uri; filestore:response; ...
Add counter to the stream reassembly engine to count stream gaps. Stream gaps
are the result of missing packets (usually due to packet loss). This missing
data stops the reassembly for the app layer.
In stateful detection only inspect the file portion of the rule after all
other conditions matched. This to prevent "filestore" from tagging files
for storage during a partial match.
Add a couple of unittests to test the behaviour change.
Add pruning of files in memory so we keep only memory what we really need.
Fix magic logic.
Reset file part of the de_state on receiving another file in the same tx.
- Enable response body tracking
- Enable file extraction for responses
- File store meta file includes magic, close reason.
- Option to force magic lookup for all stored files.
- Fix libmagic calls thead safety.
Introduce a separate FlowAddress structure for holding the ipv4 or ipv6 address
that doesn't have the family in it like the Address structure. Instead, the
family is stored in the flow as a flag: FLOW_IPV4 and FLOW_IPV6.
Add macro's to check the family, copy the address, etc.
Update many unittests to reflect these changes. Introduce unittest helper
functions for creating and initializing a flow and freeing it again.
On 64 bit this shrinks the flow with 8 bytes.
Removal of per flow 'aldata' array. It contained a ptr for each ALPROTO. Instead now we have 2 ptrs in the flow: alparser and alstate.
Various cleanups and dead code removal from the app layer API.
Should safe 100+ bytes memory per flow on 64 bit.
Updated lots of unittests to reflect these changes.
commit eff08f93d8
Author: Anoop Saldanha <poonaatsoc@gmail.com>
Date: Thu Nov 3 14:31:24 2011 +0530
update failing unittest to reflect the mpm design update
Fixed a bug in the mpm code that would make all the changes in the commit just undone wrong.
This patch introduces a function called SigGetThresholdTypeIter
which iterate on all Threshold for a given signature returning
the next DetectThresholdData.
Added stream counters:
- tcp.reassembly_memuse -- current memory use by reassembly in bytes
- tcp.memuse -- current memory use by stream tracking in bytes
- tcp.reused_ssn -- ssn reused by new session with identical tuple
- tcp.no_flow -- TCP packets with no flow - indicating flow engine memory at its limits
With the introduction of the PrintInet function there was almost
no difference between IPv4 and IPv6 HTTP logging functions. This
patch adds a wrapper that factorizes the code.
This patch fixes the unittest for IPV4 defrag. The direct usage
of the pkt pointer in the Packet structure is not allowed. This
is fixed by using PacketCopyData function.
This modification was requiring some other fixes, like using
memcmp to compare data instead of an iteration on pkt pointer.
This patch modifies output module loading to only trigger alert
message for non existing modules when they are loaded. It also
warn about unified1 removal.
Attach multiple packets to an event instead of using one
event data per packet. This is currrently unsupported by
reporting frontend but at least we don't have multiple
alerts.
Some old version of barnyard2 were needing a workaround in the
packet header building. THis patch introduces a enable-old-barnyard2
configure flag which can be used to restore this behaviour.
This patch adds the logging of stream segments. Among other
modifications, it uses a wrapper to fwrite to permit to update
file statistics in an automated manner. Some memcpy have also
been avoided by using pointer to the data.
This patch introduces a function called StreamMsgForEach which
can be used to run a callback on all segments of a stream. This
is currently only supported for TCP as this is the only streaming
aware protocol.
When no interface was specified on command line, the workers and
single mode where not able to start due to the fact there was no
registered interfaces.
This patch adds a new alert format called pcap-info. It aims at
providing an easy to parse one-line per-alert format containing
the packet id in the parsed pcap for each alert. This permit to
add information inside the pcap parser.
This format is made to be used with suriwire which is a plugin for
wireshark. Its target is to enable the display of suricata results
inside wireshark.
This format doesn't use append mode per default because a clean file
is needed to operate with wireshark.
The format is a list of values separated by ':':
Packet number:GID of matching signature:SID of signature:REV of signature:Flow:To Server:To Client:0:0:Message of signature
The two zero are not yet used values. Candidate for usage is the
part of the packet that matched the signature.
Previous commits have considerabily empowered the "single" mode which
could contain multiple threads. This behaviour was not a target for
this runmode and the following patch remedies to the situation by
introducing the "workers" mode where each thread do all the tasks
from acquisition to logging. This runmode is currently implemented
for af-packet and pf-ring.
A devide configuration can be used by multiple threads. It is thus
necessary to wait that all threads stop using the configuration before
freeing it. This patch introduces an atomic counter and a free function
which has to be called by each thread when it will not use anymore
the structure. If the configuration is not used anymore, it is freed
by the free function.
This patch adds a --pcap option which can be used to select or
an interface if an argument is provided or the interfaces defined
in the configuration file.
This patch convert pfring to pktacqloop and use the new factorisation
function. This also fixes commmand line parsing of pfring which is now
able to work like af-packet:
- 'suricata -c s.yaml --pfring' start suricata with all interfaces in
conf
- 'suricata -c s.yaml --pfring=eth2' start suricata on eth2
Major fixes for the tag subsystem:
- Removed TimeGet call from tag packet runtime to safe a gettimeofday
- Removed unused lock from data type
- Fixed broken first packet skip logic
- Fix broken reference counter logic
- Fix memory leak on tag expiration
- Cleaned up code
This patch adds support for the replace keyword. It is used with
content to change selected part of the payload. The major point
with this patch is that having a replace keyword made necessary
to avoid all stream level check because we need to access to the
could-be-modified packet payload.
One of the main difficulty is to handle complex signature. If there is
other content check, we must do the substitution when we're sure all
match are valid. The patch adds an attribute to the thread context
variable to be able to deal with recursivity of the match function.
Replace is only activated in IPS mode and apply only to raw match.
This patch make use of bit shift to rewrite some of the mask constants.
It also delete an unused flag value and suppress the associated dead code.
The numeric value of the flag is now used by the flag needed for replace
code.
Many, many pcre: signatures specify (...) when the more efficient
(?:...) is all that is needed. This change attempts to force
PCRE_NO_AUTO_CAPTURE on all unnamed capture groups, reverting to
capturing when necessary, e.g., when \1 is referenced.
A race condition was observed when leaving NFQ. This was caused by
the queue handle being accessed after been nullified. This patch
uses the handle mutex to protect the destruction and adds tests
on nullity to avoid crashed.
This patch renames DECODER_SET_EVENT, DECODER_ISSET_EVENT and some
other structures to ENGINE equivalent to take into account the fact
the event list is now related to all engines and not only to decoder.
Decode signature are using the fact that no proto is set on packet
to increase the matching speed. This is not the case of stream and
other engine events. Thus a difference needs to be made.
This patch adds an alias to the 'engine-event' keyword. It is now
possible to access to the stream events via the 'stream-event'
keyword. A simple transformation is done:
stream-event:reassembly_segment_before_base_seq
is a shortcut for:
engine-event:stream.reassembly_segment_before_base_seq
This patch renames the 'decode-event' keyword to 'engine-event' and
keep it for backword compatibility of rulesets. All *DecodeEvent*
references in the code are replaced by EngineEvent version.
This patch changes the option name. af-packet long option is
now used instead of -a to mimic pfring behaviour.
This patch improves the standard parsing of the command line.
Running
suricata -c suricata.yaml --af-packet
will start a suricata running in AF_PACKET mode listening on all
interfaces defined in the suricata.yaml configuration file. The
traditionnal syntax:
suricata -c suricata.yaml --af-packet=ppp0
will start a suricata listening on ppp0 only.
This patch adds a new function which build the list of interfaces to
use by parsing the configuration file. This is using the new format
and thus only af-packet can benefit of this feature.
This patch adds multi interface support to AF_PACKET. A structure
is used at thread creation to give all needed information to the
input module. Parsing of the options is done in runmode preparation
through a dedicated function which return the configuration in a
structure usable by thread creation.
The input modules are needing a per interface configuration. This
implies some new operations to be able to parse easily te configuration.
The syntax of the configuration file is for example:
af-packet:
- interface: eth0
threads: 2
- interface: eth1
threads: 3
We need a way to express get a configuration variable for interface[eth0].
This is by using ConfNodeLookupKeyValue() to get the matching node. And
after that value can be fetch by using ConfGetChildValue*() functions.
This patch handles the end of AF_PACKET socket support work. It
provides conditional compilation, autofp and single runmode.
It also adds a 'defrag' option which is used to activate defrag
support in kernel to avoid rx_hash computation in flow mode to fail
due to fragmentation.
This patch contains some fixes by Anoop Saldanha, and incorporate
change following review by Anoop Saldanha and Victor Julien.
AF_PACKET support is only build if the --enable-af-packet flag is
given to the configure command line. Detection of code availability
is also done: a check of the existence of AF_PACKET in standard
header is done. It seems this variable is Linux specific and it
should be enough to avoid compilation of AF_PACKET support on other
OSes.
Compilation does not depend on up-to-date headers on the system. If
none are present, wemake our own declaration of FANOUT variables. This
will permit compilation of the feature for system where only the kernel
has been updated to a version superior to 3.1.
Per packet per app layer parser profiling. Example summary output:
Per App layer parser stats:
App Layer IP ver Proto cnt min max avg
-------------------- ------ ----- ------ ------ ---------- -------
ALPROTO_HTTP IPv4 6 163394 126 38560320 42814
ALPROTO_FTP IPv4 6 644 117 26100 2566
ALPROTO_TLS IPv4 6 670 117 7137 799
ALPROTO_SMB IPv4 6 114794 126 225270 957
ALPROTO_DCERPC IPv4 6 5207 126 25596 1266
Also added to the csv out.
In the csv out there is a new column "stream (no app)" that removes the
app layer parsers from the stream tracking. So raw stream engine performance
becomes visible.
Per packet profiling uses tick based accounting. It has 2 outputs, a summary
and a csv file that contains per packet stats.
Stats per packet include:
1) total ticks spent
2) ticks spent per individual thread module
3) "threading overhead" which is simply calculated by subtracting (2) of (1).
A number of changes were made to integrate the new code in a clean way:
a number of generic enums are now placed in tm-threads-common.h so we can
include them from any part of the engine.
Code depends on --enable-profiling just like the rule profiling code.
New yaml parameters:
profiling:
# packet profiling
packets:
# Profiling can be disabled here, but it will still have a
# performance impact if compiled in.
enabled: yes
filename: packet_stats.log
append: yes
# per packet csv output
csv:
# Output can be disabled here, but it will still have a
# performance impact if compiled in.
enabled: no
filename: packet_stats.csv
Example output of summary stats:
IP ver Proto cnt min max avg
------ ----- ------ ------ ---------- -------
IPv4 6 19436 11448 5404365 32993
IPv4 256 4 11511 49968 30575
Per Thread module stats:
Thread Module IP ver Proto cnt min max avg
------------------------ ------ ----- ------ ------ ---------- -------
TMM_DECODEPCAPFILE IPv4 6 19434 1242 47889 1770
TMM_DETECT IPv4 6 19436 1107 137241 1504
TMM_ALERTFASTLOG IPv4 6 19436 90 1323 155
TMM_ALERTUNIFIED2ALERT IPv4 6 19436 108 1359 138
TMM_ALERTDEBUGLOG IPv4 6 19436 90 1134 154
TMM_LOGHTTPLOG IPv4 6 19436 414 5392089 7944
TMM_STREAMTCP IPv4 6 19434 828 1299159 19438
The proto 256 is a counter for handling of pseudo/tunnel packets.
Example output of csv:
pcap_cnt,ipver,ipproto,total,TMM_DECODENFQ,TMM_VERDICTNFQ,TMM_RECEIVENFQ,TMM_RECEIVEPCAP,TMM_RECEIVEPCAPFILE,TMM_DECODEPCAP,TMM_DECODEPCAPFILE,TMM_RECEIVEPFRING,TMM_DECODEPFRING,TMM_DETECT,TMM_ALERTFASTLOG,TMM_ALERTFASTLOG4,TMM_ALERTFASTLOG6,TMM_ALERTUNIFIEDLOG,TMM_ALERTUNIFIEDALERT,TMM_ALERTUNIFIED2ALERT,TMM_ALERTPRELUDE,TMM_ALERTDEBUGLOG,TMM_ALERTSYSLOG,TMM_LOGDROPLOG,TMM_ALERTSYSLOG4,TMM_ALERTSYSLOG6,TMM_RESPONDREJECT,TMM_LOGHTTPLOG,TMM_LOGHTTPLOG4,TMM_LOGHTTPLOG6,TMM_PCAPLOG,TMM_STREAMTCP,TMM_DECODEIPFW,TMM_VERDICTIPFW,TMM_RECEIVEIPFW,TMM_RECEIVEERFFILE,TMM_DECODEERFFILE,TMM_RECEIVEERFDAG,TMM_DECODEERFDAG,threading
1,4,6,172008,0,0,0,0,0,0,47889,0,0,48582,1323,0,0,0,0,1359,0,1134,0,0,0,0,0,8028,0,0,0,49356,0,0,0,0,0,0,0,14337
First line of the file contains labels.
2 example gnuplot scripts added to plot the data.