Implement a special sticky buffer to select frames for inspection.
This keyword takes an argument to specify the per protocol frame type:
alert <app proto name> ... frame:<specific frame name>
Or it can specify both in the keyword:
alert tcp ... frame:<app proto name>.<specific frame name>
The latter is useful in some cases like http, where "http" applies to
both HTTP and HTTP/2.
alert http ... frame:http1.request;
alert http1 ... frame:request;
Examples:
tls.pdu
smb.smb2.hdr
smb.smb3.data
Consider a rule like:
alert tcp ... flow:to_server; content:"|ff|SMB"; content:"some smb 1 issue";
this will scan all toserver TCP traffic, where it will only be limited by a port,
depending on how rules are grouped.
With this work we'll be able to do:
alert smb ... flow:to_server; frame:smb1.data; content:"some smb 1 issue";
This rule will only inspect the data portion of SMB1 frames. It will not affect
any other protocol, and it won't need special patterns to "search" for the
SMB1 frame in the raw stream.
The idea of stream frames is that the applayer parsers can tag PDUs and
other arbitrary frames in the stream while parsing. These frames can then
be inspected from the rule language. This will allow rules that are more
precise and less costly.
The frames are stored per direction in the `AppLayerParserState` and will only
be initialized when actual frames are in use. The per direction storage has a
fixed size static portion and dynamic support for a larger number. This is done
for effeciency.
When the Stream Buffer slides, frames are updated as they use offsets relative
to the stream. A negative offset is used for frames that started before the
current window.
Frames have events to inspect/log parser errors that don't fit the TX model.
Frame id starts at 1. So implementations can keep track of frame ids where 0
is not set.
Frames affect TCP window sliding. The frames keep a "left edge" which
signifies how much data to keep for frames that are still in progress.
Set RSS hash function according to Intel ICE PMD available hash functions
Set hash functions according to the support by the ICE PMD, so that no warning
regarding RSS setting is issued.
Set RSS hash function according to Intel IXGBE PMD available hash functions.
During configuration, a warning appeared stating that RSS hash function
has been changed from one value to the other. This has meant that
the supported hash functions did not cover all required hash functions
by the configuration. This commit solves the warning.
Due to peculiar behavior of i40e PMD driver, the RSS is required to be set
via rte_flow rules or a hash filter as compared to other NICs where RSS is
configured through port configuration structure.
RTE_FLOW rules are created on 5-tuples (as opposed to 3-tuple configured
on the other NICs). Fragmented traffic have been tested with this setup
and it has been proven that fragmented packets of the same flow are
received on the same queue. At the same time, setting 3-tuple on rte_flow
rules have not yield in the expected results.
Notes from the experiments:
- Configuration of 5-tuple (as is in the commit):
fragmented and nonfragmented packets are received by the same workers
even when I applied seed to alter them via tcpreplay-edit (option --seed)
- Setting only ETH_RSS_FRAG_IPV4 and ETH_RSS_IPV4 (i.e. setting 3-tuple):
when setting ETH_RSS_IPV4, the PMD driver says that pctype is not
supported (generally this means that the "type" of traffic is not
a valid configuration for the i40e)
- Setting only ETH_RSS_FRAG_IPV4 and ETH_RSS_NONFRAG_IPV4_OTHER:
this doesn't work well, packets of the same flow are received on
the different workers (my explanation is that the fragmented packets are
matched with ETH_RSS_FRAG_IPV4 but the other UDP packets are not matched
with ETH_RSS_NONFRAG_IPV4_OTHER rte_flow rule (they would be matched with
ETH_RSS_NONFRAG_IPV4_UDP).
Register a new runmode - DPDK. This enables a new flag on Suricata start
(--dpdk).
With the flag given, DPDK runmode is enabled.
Runmode loads the configuration and then initializes EAL.
If successful, it configures the physical NICs according to the configuration
file. After that, worker threads are initialized and then are in continuous
receive loop.
adds a container, ie a thread safe hash table whose
key is the filename
keep a tree of unordered ranges, up to a memcap limit
adds HTPFileOpenWithRange to handle like HTPFileOpen
if there is a range : open 2 files, one for the whole reassembled,
and one only for the current range
Including the whole directory results in .deps files ending up
in the distribution archive which shouldn't be there. Instead
we have to list all the test sources individually.
Move tests in a seperate commit so that we can use the previous one for
regression testing. This also gets rid of the temporary glue that made
the C tests work with the rust implementation.
Renaming was done with shell commands, git mv for moving the files and content like
find -iname '*.c' | xargs sed -i 's/ikev1/ike/g' respecting the different mixes of upper/lower case.
Split the headers and source into 2 variables. Headers are
marked noinst so they don't get automatically installed on
"make install". Instead they will be installed by a custom
Makefile target, "make install-headers".
Fix another issue with library ordering when breaking apart
LDFLAGS from LIBS for outputting usable command lines for
users of a Suricata library.
RUST_LDADD should just contain the extra libs required by
Rust, not the actual Suricata Rust library.
As we don't install the libraries by default, provide a make target,
"install-library" to install the libsuricata library files.
If shared library support exists, both the static and shared
libraries will be installed, otherwise only the static libraries
will be installed.
Building the shared library on Linux is not something by default.
Instead a user must opt-in to building by running the
"make libsuricata.so" target in the src/ directory.
Currently shared library support is only available on Linux. More
OSs will be supported as we can test them.
With the circular reference gone, we can now make use
of a convenience library for the Suricata program
as well as any other programs that depend on the same
source such as the fuzzer.
While its not a libtool convenience library, it serves
the same purpose and is a common idiom in Make and CMake
projects whereas the COMMON_SOURCES approach was more
of a hack we had to resort to until the circular
reference was resolved.
Expose the "SuricataContext" required by Rust as a function. During
normal startup we register this context with the Rust code, but
plugins written in Rust will need to get the same registration
done, but to do this in a plugin, the plugin code must
call and set the context within its address space.
A filetype plugin is a plugin that implements an eve filetype. Most
of the current filetypes could likely be implemented as such a plugin.
Such a plugin must implement Open, Close and Write, where Write
is provided the formatted JSON to be logged.
This commit also includes the plumbing for plugin loading. Example
plugin to come.
Plugins are loaded by the "plugin" section in the configuration
file:
plugins:
- /path/to/directory/plugins
- /path/to/plugin_file.so
This can also be done on the command line with:
--set plugins.0=/path/plugin_file.so
Goals:
- reduce locking
- take advantage of 'hot' caches
- better locality
Locking reduction
New flow spare pool. The global pool is implmented as a list of blocks,
where each block has a 100 spare flows. Worker threads fetch a block at
a time, storing the block in the local thread storage.
Flow Recycler now returns flows to the pool is blocks as well.
Flow Recycler fetches all flows to be processed in one step instead of
one at a time.
Cache 'hot'ness
Worker threads now check the timeout of flows they evaluate during lookup.
The worker will have to read the flow into cache anyway, so the added
overhead of checking the timeout value is minimal. When a flow is considered
timed out, one of 2 things happens:
- if the flow is 'owned' by the thread it is handled locally. Handling means
checking if the flow needs 'timeout' work.
- otherwise, the flow is added to a special 'evicted' list in the flow
bucket where it will be picked up by the flow manager.
Flow Manager timing
By default the flow manager now tries to do passes of the flow hash in
smaller steps, where the goal is to do full pass in 8 x the lowest timeout
value it has to enforce. So if the lowest timeout value is 30s, a full pass
will take 4 minutes. The goal here is to reduce locking overhead and not
get in the way of the workers.
In emergency mode each pass is full, and lower timeouts are used.
Timing of the flow manager is also no longer relying on pthread condition
variables, as these generally cause waking up much quicker than the desired
timout. Instead a simple (u)sleep loop is used.
Both changes reduce the number of hash passes a lot.
Emergency behavior
In emergency mode there a number of changes to the workers. In this scenario
the flow memcap is fully used up and it is unavoidable that some flows won't
be tracked.
1. flow spare pool fetches are reduced to once a second. This avoids locking
overhead, while the chance of success was very low.
2. getting an active flow directly from the hash skips flows that had very
recent activity to avoid the scenario where all flows get only into the
NEW state before getting reused. Rather allow some to have a chance of
completing.
3. TCP packets that are not SYN packets will not get a used flow, unless
stream.midstream is enabled. The goal here is again to avoid evicting
active flows unnecessarily.
Better Localily
Flow Manager injects flows into the worker threads now, instead of one or
two packets. Advantage of this is that the worker threads can get packets
from their local packet pools, avoiding constant overhead of packets returning
to 'foreign' pools.
Counters
A lot of flow counters have been added and some have been renamed.
Overall the worker threads increment 'flow.wrk.*' counters, while the flow
manager increments 'flow.mgr.*'.
Additionally, none of the counters are snapshots anymore, they all increment
over time. The flow.memuse and flow.spare counters are exceptions.
Misc
FlowQueue has been split into a FlowQueuePrivate (unlocked) and FlowQueue.
Flow no longer has 'prev' pointers and used a unified 'next' pointer for
both hash and queue use.
This commit adds MAC address output to the EVE-JSON format. We follow the
remarks made in Redmine ticket #962: for packets, log MAC src/dst as a
scalar field in EVE; for flows, log MAC src/dst as lists in EVE. Field names
are different between flow and packet context to avoid type confusion
(src_mac vs. src_macs). Configuration approach and JSON representation is
taken from previous GitHub PR #2700.
Fixes https://redmine.openinfosecfoundation.org/issues/2689
Adds a new source file to handle this keyword.
And modifies documentation, Makefile, and registration accordingly.
url_decode decodes url-encoded data, ie replacing '+' with space
and '%HH' with its value.
Match on Hassh using ssh.hassh, ssh.hassh.server, ssh.hassh.string, ssh.hassh.server.string keywords, e.g:
alert ssh any any -> any any (msg:"match SSH hash"; ssh.hassh; content:"xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"; sid:1000010;)
alert ssh any any -> any any (msg:"match SSH hash-server"; ssh.hassh.server; content:"xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"; sid:1000020;)
alert ssh any any -> any any (msg:"match SSH hash-string"; ssh.hassh.string; content:"none,zlib@openssh.com,zlib"; sid:1000030;)
alert ssh any any -> any any (msg:"match SSH hash-server-string"; ssh.hassh.server.string; content:"umac-64-etm@openssh.com,umac-128-etm@openssh.com,"; sid:1000040;)
File store v1 has been deprecated and was scheduled for removal
by June 2020.
Log an error if a file-store configuration is loaded without
version set to 2.
This commit adds support for the Remote Framebuffer Protocol (RFB) as
used, for example, by various VNC implementations. It targets the
official versions 3.3, 3.7 and 3.8 of the protocol and provides logging
for the RFB handshake communication for now. Logged events include
endpoint versions, details of the security (i.e. authentication)
exchange as well as metadata about the image transfer parameters.
Detection is enabled using keywords for:
- rfb.name: Session name as sticky buffer
- rfb.sectype: Security type, e.g. VNC-style challenge-response
- rfb.secresult: Result of the security exchange, e.g. OK, FAIL, ...
The latter could be used, for example, to detect brute-force attempts
on open VNC servers, while the name could be used to map unwanted VNC
sessions to the desktop owners or machines.
We also ship example EVE-JSON output and keyword docs as part of the
Sphinx source for Suricata's RTD documentation.
Add a rule keyword, dns.opcode to match on the opcode flag
found in the DNS request and response headers.
Only exact matches are allowed with negation.
Examples:
- dns.opcode:4;
- dns.opcode:!1;
Datasets are sets/lists of data that can be accessed or added from
the rule language.
This patch implements 3 data types:
1. string (or buffer)
2. md5
3. sha256
The patch also implements 2 new rule keywords:
1. dataset
2. datarep
The dataset keyword allows matching against a list of values to see if
it exists or not. It can also add the value to the set. The set can
optionally be stored to disk on exit.
The datarep support matching/lookups only. With each item in the set a
reputation value is stored and this value can be matched against. The
reputation value is unsigned 16 bit, so values can be between 0 and 65535.
Datasets can be registered in 2 ways:
1. through the yaml
2. through the rules
The goal of this rules based approach is that rule writers can start using
this without the need for config changes.
A dataset is implemented using a thash hash table. Each dataset is its own
separate thash.
Thread safe hash table implementation based on the Flow hash, IP Pair
hash and others.
Hash is array of buckets with per bucket locking. Each bucket has a
list of elements which also individually use locking.
Victor Julien
Jason Ish
Lukas Sismis
Philippe Antoine
Jeff Lucovsky
Shivani Bhardwaj
Simon Dugas
Sascha Steinbiss
frank honza
Emmanuel Thompson
Carl Smith
Ali Jad Khalil
Victor Julien
Vadym Malakhatko
Pierre Chifflier
Danny Browning
Giuseppe Longo
Zach Kelly
Henrik Lund Kramshoej