Enables IPS functionality on Windows using the open-source
(LGPLv3/GPLv2) WinDivert driver and API.
From https://www.reqrypt.org/windivert-doc.html : "WinDivert is a
user-mode capture/sniffing/modification/blocking/re-injection package
for Windows Vista, Windows Server 2008, Windows 7, and Windows 8.
WinDivert can be used to implement user-mode packet filters, packet
sniffers, firewalls, NAT, VPNs, tunneling applications, etc., without
the need to write kernel-mode code."
- adds `--windivert [filter string]` and `--windivert-forward [filter
string]` command-line options to enable WinDivert IPS mode.
`--windivert[-forward] true` will open a filter for all traffic. See
https://www.reqrypt.org/windivert-doc.html#filter_language for more
information.
Limitation: currently limited to `autofp` runmode.
Additionally:
- `tmm_modules` now zeroed during `RegisterAllModules`
- fixed Windows Vista+ `inet_ntop` call in `PrintInet`
- fixed `GetRandom` bug (nonexistent keys) on fresh Windows installs
- fixed `RandomGetClock` building on Windows builds
- Added WMI queries for MTU
This is a DHCP decoder and logger written in Rust. Unlike most
parsers, this one is stateless so responses are not matched
up to requests by Suricata. However, the output does contain
enough fields to match them up in post-processing.
Rules are included to alert of malformed or truncated options.
Add a new parser for Internet Key Exchange version (IKEv2), defined in
RFC 7296.
The IKEv2 parser itself is external. The embedded code includes the
parser state and associated variables, the state machine, and the
detection code.
The parser looks the first two messages of a connection, and analyzes
the client and server proposals to check the cryptographic parameters.
Add Ja3GetHash() to return the content of the JA3 hash buffer from the
TLS session.
Example:
function init (args)
local needs = {}
needs["protocol"] = "tls"
return needs
end
function setup (args)
filename = SCLogPath() .. "/ja3_hash.log"
file = assert(io.open(filename, "a"))
end
function log (args)
ja3_hash = Ja3GetHash()
if ja3_hash == nil then
return
end
file:write(ja3_hash .. "\n")
file:flush()
end
function deinit (args)
file:close()
end
In the (useless) example above, each JA3 hash is logged to a log file.
Match on JA3 hash using ja3_hash keyword, e.g:
alert tls any any -> any any (msg:"JA3 hash test";
ja3_hash;
content:"e7eca2baf4458d095b7f45da28c16c34";
sid:1;)
Implement SMB app-layer parser for SMB1/2/3. Features:
- file extraction
- eve logging
- existing dce keyword support
- smb_share/smb_named_pipe keyword support (stickybuffers)
- auth meta data extraction (ntlmssp, kerberos5)
Allows matching on stickybuffers. Like dsize, it allows matching on
exact values, greater than and less than, and ranges.
For streaming buffers, such as HTTP bodies, the final size of the
body is only known at the end of the transaction.
This patch implements bypass capability for af-packet.
The filter only bypass TCP and UDP in IPv4 and IPv6. It don't
don't bypass IPv6 with extended headers.
This patch also introduces a bypassed flow manager that takes
care of timeouting the bypassed flows. It uses a 60 sec
timeout on flow. As they are supposed to be active we can
try that. If they are not active then we don't care to get them
back in Suricata.
This patch introduces the ebpf cluster mode. This mode is using
an extended BPF function that is loaded into the kernel and
provide the load balancing.
An example of cluster function is provided in the ebpf
subdirectory and provide ippair load balancing function.
This is a function which uses the same method as
the one used in autofp ippair to provide a symetrical
load balancing based on IP addresses.
A simple filter example allowing to drop IPv6 is added to the
source.
This patch also prepares the infrastructure to be able to load
and use map inside eBPF files. This will be used later for flow
bypass.
This adds a new module that permits to decompress
swf file compressed with zlib or lzma algorithms.
The API that performs decompression will take a compressed
buffer and build a new decompressed buffer following the
FWS format which represents an uncompressed file.
The maximum buffer that can be created is up to 50mb.
TFTP parsing and logging written in Rust.
Log on eve.json the type of request (read or write), the name of the file and
the mode.
Example of output:
"tftp":{"packet":"read","file":"rfc1350.txt","mode":"octet"}
Filestore v2 is starts as a copy of log-filestore with the
following changes.
- NSS is required as file names as based on the SHA256.
- Work/tmp files are stored in a temp. directory, then
moved into a directory tree where the directory names
are the first 2 characters of the hex SHA256.
- Removes the need for a waldo file or pid in the filenames.
This keyword mathes on ftp operation STOR and RETR. It will allow
rules writer to select if the alert has to be on a put or a fetch
operation.
It is now possible to write a signature like:
alert ftp-data any any -> any any (msg:"FTP data get firwmare"; ftdata_command:retr; sid:2; rev:1;)
to alert when a file is retrieved from a FTP server.
This patch provides a working expectation system. This will allow
suricata to have a way to identify parallel connections opened by
a protocol such as FTP.
Expectation are a chained list and there is a cleaning by timeout
of the entries.
This patch also defined a counter of expectations that is also
used to check if we need to query IPPairs. This way we only query
the IPPairs store if we have an expectation.
Add a common structure definition to contain all required elements
required for registering a parser.
This also reduces the risk of incorrectly registering a parser: the
compiler will type-check functions.
The registration function allows factorization of the code. It can be
used to register parsers, but is not mandatory.
If extra registration code (for functions not in the structure)
it is still possible by calling the C functions after the registration.
Reimplement keyword to match on SHA-1 fingerprint of TLS
certificate as a mpm keyword.
alert tls any any -> any (msg:"TLS cert fingerprint test";
tls_cert_fingerprint;
content:"4a:a3:66:76:82:cb:6b:23:bb:c3:58:47:23:a4:63:a7:78:a4:a1:18";
sid:12345;)
added util-napatech module which contains implementation threads
for processing statistics. And modified source-napatech and
runmode-napatech to instantiate the threads.
napatech: Implementation of packet counters
napatech: implementation of statistics counters
napatech: Implementation of packet counters.
napatech: added util-napatech module
napatech: added utils-napatech module.
added include declaration and napatech specific structure when HAVE_NAPATECH
is defined.
Added util-napatech module to project.
Since the parser now also does nfs2, the name nfs3 became confusing.
As it's still in beta, we can rename so this patch renames all 'nfs3'
logic to simply 'nfs'.
The target keyword allows rules writer to specify information about
target of the attack. Using this keyword in a signature causes
some fields to be added in the EVE output. It also fixes ambiguity
in the Prelude output.
Rust is currently optional, use the --enable-rust configure
argument to enable Rust.
By default Rust will be built in release mode. If debug is enabled
then it will be built in debug mode.
On make dist, "cargo vendor" will be run to make a local copy
of Rust dependencies for the distribution archive file.
Add autoconf checks to test for the vendored source, and if it
exists setup the build to use the vendored code instead of
fetching it from the network.
Also, as Cargo requires semantic versioning, the Suricata version
had to change from 4.0dev to 4.0.0-dev.
Set flags by default:
-Wmissing-prototypes
-Wmissing-declarations
-Wstrict-prototypes
-Wwrite-strings
-Wcast-align
-Wbad-function-cast
-Wformat-security
-Wno-format-nonliteral
-Wmissing-format-attribute
-funsigned-char
Fix minor compiler warnings for these new flags on gcc and clang.
Make stream engine use the streaming buffer API for it's data storage.
This means that the data is stored in a single reassembled sliding
buffer. The subleties of the reassembly, e.g. overlap handling, are
taken care of at segment insertion.
The TcpSegments now have a StreamingBufferSegment that contains an
offset and a length. Using this the segment data can be retrieved
per segment.
Redo segment insertion. The insertion code is moved to it's own file
and is simplified a lot.
A major difference with the previous implementation is that the segment
list now contains overlapping segments if the traffic is that way.
Previously there could be more and smaller segments in the memory list
than what was seen on the wire.
Due to the matching of in memory segments and on the wire segments,
the overlap with different data detection (potential mots attacks)
is much more accurate.
Raw and App reassembly progress is no longer tracked per segment using
flags, but there is now a progress tracker in the TcpStream for each.
When pruning we make sure we don't slide beyond in-use segments. When
both app-layer and raw inspection are beyond the start of the segment
list, the segments might not be freed even though the data in the
streaming buffer is already gone. This is caused by the 'in-use' status
that the segments can implicitly have. This patch accounts for that
when calculating the 'left_edge' of the streaming window.
Raw reassembly still sets up 'StreamMsg' objects for content
inspection. They are set up based on either the full StreamingBuffer,
or based on the StreamingBufferBlocks if there are gaps in the data.
Reworked 'stream needs work' logic. When a flow times out the flow
engine checks whether a TCP flow still needs work. The
StreamNeedsReassembly function is used to test if a stream still has
unreassembled segments or uninspected stream chunks.
This patch updates the function to consider the app and/or raw
progress. It also cleans the function up and adds more meaningful
debug messages. Finally it makes it non-inline.
Unittests have been overhauled, and partly moved into their own files.
Remove lots of dead code.
Implement common code to easily add more per HTTP header detection
keywords.
Implement http_accept sticky buffer. It operates on the HTTP Accept
header.
eve: detects libevent for async redis at configure
eve: moves redis output code to new file - util-log-redis.{c,h}
eve: redis ECHO and QUIT commands for async mode
eve: redis output defaults if conf is missing
Match on TLS certificate serial number using tls_cert_serial
keyword, e.g.:
alert tls any any -> any any (msg:"TLS cert serial test";
tls_cert_serial; content:"5C:19:B7:B1:32:3B:1C:A1";
sid:12345;)
Matches on the start of a HTTP request or response.
Uses a buffer constructed from the request line and normalized request
headers, including the Cookie header.
Or for the response side, it uses the response line plus the
normalized response headers, including the Set-Cookie header.
Both buffers are terminated by an extra \r\n.
A sticky buffer that allows content inspection on a contructed buffer
of HTTP header names. The buffer starts with \r\n, the names are
separated by \r\n and the end of the buffer contains an extra \r\n.
E.g. \r\nHost\r\nUser-Agent\r\n\r\n
The leading \r\n is to make sure one can match on a full name in all
cases.
Due to the use of AFL_LOOP and initialization/deinit outside of it,
part of the fuzzing relied on the global 'state' in flow and defrag.
Because of this crashes that were found could not be reproduced. The
saved crash input was only the last in the series.
This patch addresses that. It requires a new output directory 'dump'
where the packet fuzzers will store all their input. If the AFL_LOOP
fails the files will not be removed and this 'serie' can be read
again for reproducing the issue.
e.g.: AFL would work with:
--afl-decoder-ppp=@@
and after a crash is found the produced serie can be read with:
--afl-decoder-ppp-serie=1486656919-514163
The series have a timestamp as name and a suffix that controls the
order in which the files will be 'replayed' in Suricata.
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.
Some code won't work well when the OS doesn't allow RWX pages. This
page introduces a check for runtime evaluation of the OS' policy on
this.
Thanks to Shawn Webb from HardenedBSD for suggesting this solution.
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
Move engine and registration into the keyword file.
Register as 'ALPROTO_UNKNOWN' instead of per alproto. The
registration will only apply it to those rules that have
events set.
Introduce prefilter keyword to force a keyword to be used as prefilter.
e.g.
alert tcp any any -> any any (content:"A"; flags:R; prefilter; sid:1;)
alert tcp any any -> any any (content:"A"; flags:R; sid:2;)
alert tcp any any -> any any (content:"A"; dsize:1; prefilter; sid:3;)
alert tcp any any -> any any (content:"A"; dsize:1; sid:4;)
In sid 2 and 4 the content keyword is used in the MPM engine.
In sid 1 and 3 the flags and dsize keywords will be used.
In preparation of the introduction of more general purpose prefilter
engines, rename PatternMatcherQueue to PrefilterRuleStore. The new
engines will fill this structure a similar way to the current mpm
prefilters.
This adds a new keyword which permits to call the
bypass callback when a sig is matched.
The callback must be called when the match of the sig
is complete.
Detection plugin for TLS certificate fields notBefore and notAfter.
Supports equal to, less than, greater than, and range operations
for both keywords. Dates can be represented as either ISO 8601 or
epoch (Unix time).
Examples:
alert tls [...] tls_cert_notafter:1445852105; [...]
alert tls [...] tls_cert_notbefore:<2015-10-22T23:59:59; [...]
alert tls [...] tls_cert_notbefore:>2015-10-22; [...]
alert tls [...] tls_cert_notafter:2000-10-22<>2020-05-15; [...]
Moved and adapted code from detect-filemd5 to util-detect-file-hash,
generalised code to work with SHA-1 and SHA-256 and added necessary
flags and other constants.
Add a new API to store data from streaming sources, like HTTP body
processing or TCP data.
Currently most of the code uses a pattern of list of data chunks
(e.g. TcpSegment) that is reassembled into a large buffer on-demand.
The Streaming Buffer API changes the logic to store the data in
reassembled form from the start, with the segments/chunks pointing
to the reassembled data.
The main buffer storing the data slides forward, automatically or
manually. The *NoTrack calls allows for a segmentless mode of
operation.
This approach has two main advantages:
1. accessing the reassembled data is virtually cost-free
2. reduction of allocations and memory management
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.
Match on server name indication (SNI) extension in TLS using tls_sni
keyword, e.g:
alert tls any any -> any any (msg:"SNI test"; tls_sni;
content:"example.com"; sid:12345;)
Dump a json record containing all sigs that need to be inspected after
prefilter. Part of profiling. Only dump if threshold is met, which is
currently set by:
--set detect.profiling.inspect-logging-threshold=200
A file called packet_inspected_rules.json is created in the default
log dir.
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.
Support TLS in Lua detection scripts.
function init (args)
local needs = {}
needs["tls"] = tostring(true)
return needs
end
function match(args)
version, subject, issuer, fingerprint = TlsGetCertInfo();
if version == nil then
return 0
end
str = string.format("Version %s\nIssuer %s\nSubject %s\nFingerprint %s",
version, issuer, subject, fingerprint)
SCLogInfo(str);
return 1
end
Remove most of the CFLAGS updates from configure. Flags are now (mostly)
set in AM_CLFLAGS.
Update all -DBLAH additions to CFLAGS to use AC_DEFINE([BLAH], ...)
Improve Lua vs LuaJIT checking.
Improve the configure output a bit.
Lots of smaller cleanups.
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.
Implement LINKTYPE_NULL for pcap live and pcap file.
From: http://www.tcpdump.org/linktypes.html
"BSD loopback encapsulation; the link layer header is a 4-byte field,
in host byte order, containing a PF_ value from socket.h for the
network-layer protocol of the packet.
Note that ``host byte order'' is the byte order of the machine on
which the packets are captured, and the PF_ values are for the OS
of the machine on which the packets are captured; if a live capture
is being done, ``host byte order'' is the byte order of the machine
capturing the packets, and the PF_ values are those of the OS of
the machine capturing the packets, but if a ``savefile'' is being
read, the byte order and PF_ values are not necessarily those of
the machine reading the capture file."
Feature ticket #1445
- Created app-layer-htp-xff.c and app-layer-htp-xff.h
- Added entries in the Makefile.am
- Added the necessary configuration options to EVE alert section
- Updated Unified2 XFF configuration comments and removed unnecessary whitespace
- Created a generic function to parse the configuration
- Release the flow locks sooner and remove debug logging
- Added XFF support to EVE alert output
Convert regular 'stats.log' output to this new API.
In addition to the current stats value, also give the last value. This
makes it easy to display the difference.
Add the modbus.function and subfunction) keywords for public function match in rules (Modbus layer).
Matching based on code function, and if necessary, sub-function code
or based on category (assigned, unassigned, public, user or reserved)
and negation is permitted.
Add the modbus.access keyword for read/write Modbus function match in rules (Modbus layer).
Matching based on access type (read or write),
and/or function type (discretes, coils, input or holding)
and, if necessary, read or write address access,
and, if necessary, value to write.
For address and value matching, "<", ">" and "<>" is permitted.
Based on TLS source code and file size source code (address and value matching).
Signed-off-by: David DIALLO <diallo@et.esia.fr>
Decode Modbus request and response messages, and extracts
MODBUS Application Protocol header and the code function.
In case of read/write function, extracts message contents
(read/write address, quantity, count, data to write).
Links request and response messages in a transaction according to
Transaction Identifier (transaction management based on DNS source code).
MODBUS Messaging on TCP/IP Implementation Guide V1.0b
(http://www.modbus.org/docs/Modbus_Messaging_Implementation_Guide_V1_0b.pdf)
MODBUS Application Protocol Specification V1.1b3
(http://www.modbus.org/docs/Modbus_Application_Protocol_V1_1b3.pdf)
Based on DNS source code.
Signed-off-by: David DIALLO <diallo@et.esia.fr>
This patch adds a new Log API for streaming data such as TCP reassembled
data and HTTP body data. It could also replace Filedata API.
Each time a new chunk of data is available, the callback will be called.
EVE logging is a really good substitute for pcapinfo. Suriwire is
now supporting EVE output so it is not anymore necessary to have
pcapinfo in Suricata.
Remove local copies from each MPM file and use include file instead.
Might be better to also add util-memcpy.c rather than inlining it each time,
to get smaller code, since only seems to be used at initialization.
When running on a TILEncore-Gx PCIe card, setting the filetype of fast.log
to pcie, will open a connection over PCIe to a host application caleld
tile-pcie-logd, that receives the alert strings and writes them to a file
on the host. The file name to open is also passed over the PCIe link.
This allows running Suricata on the TILEncore-Gx PCIe card, but have the
alerts logged to the host system's file system efficiently. The PCIe API that
is used is the Tilera Packet Queue (PQ) API which can access PCIe from User
Space, thus avoiding system calls.
Created util-logopenfile-tile.c and util-logopen-tile.h for the TILE
specific PCIe logging functionality.
Using Write() and Close() function pointers in LogFileCtx, which
default to standard write and close for files and sockets, but are
changed to PCIe write and close functions when a PCIe channel is
openned for logging.
Moved Logging contex out of tm-modules.h into util-logopenfile.h,
where it makes more sense. This required including util-logopenfile.h
into a couple of alert-*.c files, which previously were getting the
definitions from tm-modules.h.
The source and Makefile for tile-pcie-logd are added in contrib/tile-pcie-logd.
By default, the file name for fast.log specified in suricata.yaml is used as
the filename on the host. An optional argument to tile-pcie-logd, --prefix=,
can be added to prepend the supplied file path. For example, is the file
in suricata.yaml is specified as "/var/log/fast.log" and --prefix="/tmp",
then the file will be written to "/tmp/var/log/fast.log".
Check for TILERA_ROOT environment variable before building tile_pcie_logd
Building tile_pcie_logd on x86 requires the Tilera MDE for its PCIe libraries
and API header files. Configure now checs for TILERA_ROOT before enabling
builing tile_pcie_logd in contrib/tile_pcie_logd
- restore json output-file.[ch] as output-json-file.[ch] after rebase conflict
- fix Makefile.am after merge conflict
- some dev-log-api-v4.0 rebase json fallout cleanup
A new logger API for registering file storage handlers. Where the
FileLog handler is called once per file, this handler will be called
for each data chunk so that storing the entire file is possible.
The logger call in the API is as follows:
typedef int (*FiledataLogger)(ThreadVars *, void *thread_data,
const Packet *, const File *, const FileData *, uint8_t flags);
All data is const, thus should be read only. The final flags field
is used to indicate to the caller that the file is new, or if it's
being closed.
Files use an internal unique id 'file_id' which can be used by the
loggers to create unique file names. This id can use the 'waldo'
feature of the log-filestore module. This patch moves that waldo
loading and storing logic to this API's implementation. A new
configuration directive 'file-store-waldo: <filename>' is added,
but the existing waldo settings will also continue to work.
This patch introduces a new logging API for logging extracted file info.
It allows for registration of a callback that is called once per file:
when it's considered 'closed'.
Users of this API register their Log Function through:
OutputRegisterFileModule()
The API uses a magic settings globally. This might be changed later.
This patch introduces a new API for logging transactions from
tx-aware app layer protocols. It runs all the registered loggers
from a single thread module. This thread module takes care of the
transaction handling and flow locking. The logger just gets a
transaction to log out.
All loggers for a protocol will be run at the same time, so there
will not be any timing differences.
Loggers will no longer act as Thread Modules in the strictest sense.
The Func is NULL, and SetupOuputs no longer attaches them to the
thread module chain individually. Instead, after registering through
OutputRegisterTxModule, the setup data is used in the single logging
module.
The logger (LogFunc) is called for each transaction once, at the end
of the transaction.
This patch introduces a new API for outputs that log based on the
packet, such as alert outputs. In converts fast-log to the new API.
The API gets rid of the concept of each logger being a thread module,
but instead there is one thread module that runs all packet loggers.
Through the registration function OutputRegisterPacketModule a log
module can register itself to be considered for each packet.
Each logger registers itself to this new API with 2 functions and the
OutputCtx object that was already used in the old implementation.
The function pointers are:
LogFunc: the log function
ConditionFunc: this function is called before the LogFunc and only
if this returns TRUE the LogFunc is called.
For a simple alert logger like fast-log, the condition function will
simply return TRUE if p->alerts.cnt > 0.
Move app layer event handling into app-layer-event.[ch].
Convert 'Set' macro's to functions.
Get rid of duplication in Set and SetRaw. Set now calls SetRaw.
Fix potentential int overflow condition in the event storage.
Update callers.
This patch introduces wrapper functions around allocation functions
to be able to have a global HTP memcap. A simple subsitution of
function was not enough because allocated size needed to be known
during freeing and reallocation.
The value of the memcap can be set in the YAML and is left by default
to unlimited (0) to avoid any surprise to users.
Split threads.h into several files, where each of these files defines
all lock types and macro's.
threads.h defines the normal case
threads-debug.h defines the debug variants
threads-profile.h defines the lock profiling variants
Finally, threads-arch-tile.h moves the Tilera specifics out
Move SIMD the implementations of SigMatchSignaturesBuildMatchArray()
for SSE3 and Tile out of detect.c to reduce the size of the file.
Also moved SIMD unit tests to detect-simd.c
Aho-Corasick mpm optimized for Tilera Tile-Gx architecture. Based on the
util-mpm-ac.c code base. The primary optimizations are:
1) Matching function used Tilera specific instructions.
2) Alphabet compression to reduce delta table size to increase cache
utilization and performance.
The basic observation is that not all 256 ASCII characters are used by
the set of multiple patterns in a group for which a DFA is
created. The first reason is that Suricata's pattern matching is
case-insensitive, so all uppercase characters are converted to
lowercase, leaving a hole of 26 characters in the
alphabet. Previously, this hole was simply left in the middle of the
alphabet and thus in the generated Next State (delta) tables.
A new, smaller, alphabet is created using a translation table of 256
bytes per mpm group. Previously, there was one global translation
table for converting upper case to lowercase.
Additional, unused characters are found by creating a histogram of all
the characters in all the patterns. Then all the characters with zero
counts are mapped to one character (0) in the new alphabet. Since
These characters appear in no pattern, they can all be mapped to a
single character and still result in the same matches being
found. Zero was chosen for the value in the new alphabet since this
"character" is more likely to appear in the input. The unused
character always results in the next state being state zero, but that
fact is not currently used by the code, since special casing takes
additional instructions.
The characters that do appear in some pattern are mapped to
consecutive characters in the new alphabet, starting at 1. This
results in a dense packing of next state values in the delta tables
and additionally can allow for a smaller number of columns in that
table, thus using less memory and better packing into the cache. The
size of the new alphabet is the number of used characters plus 1 for
the unused catch-all character.
The alphabet size is rounded up to the next larger power-of-2 so that
multiplication by the alphabet size can be done with a shift. It
might be possible to use a multiply instruction, so that the exact
alphabet size could be used, which would further reduce the size of
the delta tables, increase cache density and not require the
specialized search functions. The multiply would likely add 1 cycle to
the inner search loop.
Since the multiply by alphabet-size is cleverly merged with a mask
instruction (in the SINDEX macro), specialized versions of the
SCACSearch function are generated for alphabet sizes 256, 128, 64, 32
and 16. This is done by including the file util-mpm-ac-small.c
multiple times with a redefined SINDEX macro. A function pointer is
then stored in the mpm context for the search function. For alpha bit
sizes of 8 or smaller, the number of states usually small, so the DFA
is already very small, so there is little difference using the 16
state search function.
The SCACSearch function is also specialized by the size of the value
stored in the next state (delta) tables, either 16-bits or 32-bits.
This removes a conditional inside the Search function. That
conditional is only called once, but doesn't hurt to remove
it. 16-bits are used for up to 32K states, with the sign bit set for
states with matches.
Future optimization:
The state-has-match values is only needed per state, not per next
state, so checking the next-state sign bit could be replaced with
reading a different value, at the cost of an additional load, but
increasing the 16-bit next state span to 64K.
Since the order of the characters in the new alphabet doesn't matter,
the new alphabet could be sorted by the frequency of the characters in
the expected input stream for that multi-pattern matcher. This would
group more frequent characters into the same cache lines, thus
increasing the probability of reusing a cache-line.
All the next state values for each state live in their own set of
cache-lines. With power-of-two sizes alphabets, these don't overlap.
So either 32 or 16 character's next states are loaded in each cache
line load. If the alphabet size is not an exact power-of-2, then the
last cache-line is not completely full and up to 31*2 bytes of that
line could be wasted per state.
The next state table could be transposed, so that all the next states
for a specific character are stored sequentially, this could be better
if some characters, for example the unused character, are much more
frequent.
This patch is using the qa/log directory to store the output
of the check. In case of success, the directory is deleted.
In case of failure, the directory remains in place.
This should fixes#910.
Victor Julien
Eric Leblond
Jason Ish
Jacob Masen-Smith
Pierre Chifflier
Mats Klepsland
Giuseppe Longo
Clement Galland
Dana Helwig
Phil Young
fooinha
kwong
Duarte Silva
tobiass1
Justin Viiret
Tom DeCanio
Giuseppe Longo
gureedo
Ken Steele
DIALLO David
David Abarbanel
Jason Ish
Nelson Escobar
Anoop Saldanha
Victor Julien