- Removed unnecessary assignment of the data field
- Removed else condition (same function called for IPv4 and IPV6)
- Fixed constants to be a power of two (used in bitwise operations)
The field ext_pkt was cleaned before calling the release function.
The result was that IPS mode such as the one of AF_PACKET were not
working anymore because they were not able to send the data which
were initially pointed by ext_pkt.
This patch moves the ext_pkt cleaning to the cleaning macro. This
ensures that the cleaning is done for allocated and pool packets.
Call PACKET_RELEASE_REFS from PacketPoolGetPacket() so that
we only access the large packet structure just before actually
using it. Should give better cache behaviour.
The Source Routing Header had routing defined as a char* for a field
of variable size. Since that field was not being used in the code, I
removed the pointer and added a comment.
Structures that are used to cast packet data into fields need to be packed
so that the compiler doesn't add any padding to these fields. This also helps
Tile-Gx to avoid unaligned loads because the compiler will insert code to
handle the possible unaligned load.
Reported in bug 1238 is an issue where stream reassembly can be
disrupted.
A packet that was in-window, but otherwise unexpected set the
window to a really low value, causing the next *expected* packet
to be considered out of window. This lead to missing data in the
stream reassembly.
The packet was unexpected in various ways:
- it would ack unseen traffic
- it's sequence number would not match the expected next_seq
- set a really low window, while not being a proper window update
Detection however, it greatly hampered by the fact that in case of
packet loss, quite similar packets come in. Alerting in this case
is unwanted. Ignoring/skipping packets in this case as well.
The logic used in this patch is as follows. If:
- the packet is not a window update AND
- packet seq > next_seq AND
- packet acq > next_seq (packet acks unseen data) AND
- packet shrinks window more than it's own data size
THEN set event and skip the packet in the stream engine.
So in case of a segment with no data, any window shrinking is rejected.
Bug #1238.
Until now the time out handling in defrag was done using a single
uint32_t that tracked seconds. This lead to corner cases, where
defrag trackers could be timed out a little too early.
If a next header / protocol is encountered that we can't handle (yet)
set an event. Disabled the rule by default.
decode-event:ipv6.unknown_next_header;
Fix or rather implement handling of unfragmentable exthdrs in ipv6.
The exthdr(s) appearing before the frag header were copied into the
reassembled packet correctly, however the stripping of the frag header
did not work correctly.
Example:
The common case is a frag header directly after the ipv6 header:
[ipv6 header]->[frag header]->[icmpv6 (part1)]
[ipv6 header]->[frag header]->[icmpv6 (part2)]
This would result in:
[ipv6 header]->[icmpv6]
The ipv6 headers 'next header' setting would be updated to point to
whatever the frag header was pointing to.
This would also happen when is this case:
[ipv6 header]->[hop header]->[frag header]->[icmpv6 (part1)]
[ipv6 header]->[hop header]->[frag header]->[icmpv6 (part2)]
The result would be:
[ipv6 header]->[hop header]->[icmpv6]
However, here too the ipv6 header would have been updated to point
to what the frag header pointed at. So it would consider the hop header
as if it was an ICMPv6 header, or whatever the frag header pointed at.
The result is that packets would not be correctly parsed, and thus this
issue can lead to evasion.
This patch implements handling of the unfragmentable part. In the first
segment that is stored in the list for reassembly, this patch detects
unfragmentable headers and updates it to have the last unfragmentable
header point to the layer after the frag header.
Also, the ipv6 headers 'next hdr' is only updated if no unfragmentable
headers are used. If they are used, the original value is correct.
Reported-By: Rafael Schaefer <rschaefer@ernw.de>
Bug #1244.
Some tests are already registered via the function
AppLayerParserRegisterProtocolUnittests. So we don't need to
egister them during runmode initialization.
Exponentially increase the memory allocated for new states when adding new
states, then at the end resize down to the actually final size so that no space is wasted.
On non-TLS systems, check each time the Thread Local Storage
is requested and if it has not been initialized for this thread, initialize it.
The prevents not initializing the worker threads in autofp run mode.
Use new management API to run the flow recycler.
Make number of threads configurable:
flow:
memcap: 64mb
hash-size: 65536
prealloc: 10000
emergency-recovery: 30
managers: 2
recyclers: 2
This sets up 2 flow recyclers.
Use new management API to run the flow manager.
Support multiple flow managers, where each of them works with it's
own part of the flow hash.
Make number of threads configurable:
flow:
memcap: 64mb
hash-size: 65536
prealloc: 10000
emergency-recovery: 30
managers: 2
This sets up 2 flow managers.
Handle misc tasks only in instance 1: Handle defrag hash timeout
handing, host hash timeout handling and flow spare queue updating
only from the first instance.
Currently management threads do their own thread setup and handling. This
patch introduces a new way of handling management threads.
Functionality that needs to run as a management thread can now register
itself as a regular 'thread module' (TmModule), where the 'Management'
callback is registered.
The flow end flags field is filled by the flow manager or the flow
hash (in case of forced timeout of a flow) to record the timeout
conditions in the flow:
- emergency mode
- state
- reason (timed out or forced)
Add logging to the flow logger.
Thread was killed by the generic TmThreadKillThreads instead of
the FlowKillFlowRecyclerThread. The latter wakes the thread up, so
that shutdown is quite a bit faster.
For netflow logging track TCP flags per stream direction. As the struct
had no more space left without expanding it, the flags and wscale
fields are now compressed.
Most flows are marked for clean up by the flow manager, which then
passes them to the recycler. The recycler logs and cleans up. However,
under resource stress conditions, the packet threads can recycle
existing flow directly. So here the recycler has no role to play, as
the flow is immediately used.
For this reason, the packet threads need to be able to invoke the
flow logger directly.
The flow logging thread ctx will stored in the DecodeThreadVars
stucture. Therefore, this patch makes the DecodeThreadVars an argument
to FlowHandlePacket.
Now that we use 'filetype' instead of 'type', we should also
use 'regular' instead of 'file'.
Added fallback to make sure we stay compatible to old configs.
At -O1+ in both Gcc and Clang, PacketPoolWait would optimize the
wait loop in the wrong way. Adding a compiler barrier to prevent
this optimization issue.
Fix pcap packet acquisition methods passing 0 to pcap_dispatch.
Previously they passed the packet pool size, but the packet_q_len
variable was now hardcoded at 0.
This patch sets packet_q_len to 64. If packet pool is empty, we fall
back to direct alloc. As the pcap_dispatch function is only called
when packet pool is not empty, we alloc at most 63 packets.
Better handle the autofp case where one thread allocates the majority
of the packets and other threads free those packets.
Add a list of locally pending packets. The first packet freed goes on the
pending list, then subsequent freed packets for the same Packet Pool are
added to this list until it hits a fixed number of packets, then the
entire list of packets is pushed onto the pool's return stack. If a freed
packet is not for the pending pool, it is freed immediately to its pool's
return stack, as before.
For the autofp case, since there is only one Packet Pool doing all the
allocation, every other thread will keep a list of pending packets for
that pool.
For the worker run mode, most packets are allocated and freed locally. For
the case where packets are being returned to a remote pool, a pending list
will be kept for one of those other threads, all others are returned as before.
Which remote pool for which to keep a pending list is changed each time the
pending list is returned. Since the return pending pool is cleared when it is
freed, then next packet to be freed chooses the new pending pool.
Using a stack for free Packet storage causes recently freed Packets to be
reused quickly, while there is more likelihood of the data still being in
cache.
The new structure has a per-thread private stack for allocating Packets
which does not need any locking. Since Packets can be freed by any thread,
there is a second stack (return stack) for freeing packets by other threads.
The return stack is protected by a mutex. Packets are moved from the return
stack to the private stack when the private stack is empty.
Returning packets back to their "home" stack keeps the stacks from getting out
of balance.
The PacketPoolInit() function is now called by each thread that will be
allocating packets. Each thread allocates max_pending_packets, which is a
change from before, where that was the total number of packets across all
threads.
Whenever DETECT_CONTENT_NOCASE is set for a BoyerMoore matcher, the
function BoyerMooreCtxToNocase() must be called. This call was missing
in AppLayerProtoDetectPMRegisterPattern().
Also created BoyerMooreNocaseCtxInit() that calls BoyerMooreCtxToNocase()
to make some code cleaner and safer.
Rather than converting the search string to lower case while searching,
convert it to lowercase during initialization.
Changes the Boyer Moore search API for take BmCtx
Change the API for BoyerMoore to take a BmCtx rather than the two parts that
are stored in the context. Which is how it is mostly used. This enforces
always calling BoyerMooreCtxToNocase() to convert to no-case.
Use CtxInit and CtxDeinit functions to create and destroy the context,
even in unit tests.
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.
When using multi mode, the filename can use a few variables:
%n -- thread number, where the 1st thread has 1, and it increments
%i -- thread id (system thread id, similar to pid)
%t -- timestamp, where seconds or seconds+usecs depends on
the ts-format option.
Example:
filename: filename: pcaps/%n/pcap.%t
This will translate to: pcaps/3/pcap.1256792217 for the 3rd thread.
Note that while it's possible to use directories, they won't be
created. So make sure they exist.
This patch adds a field 'is_private' to PcapLogData, so that the
using thread knows if it needs to lock access to it or not.
Reshuffle PcapLogData to roughly match order of access.
This patch implements a new mode in pcap-logging: 'multi'. It stores
a pcap file per logger thread, instead of just one file globally.
This removes lock contention, so it brings a lot more performance.
The trade off is that there are now mulitple files where there would
be one before.
Files have a thread id added to their name: base_name.tid.ts, so by
we have something like: "log.pcap.20057.1254500095".
PcapLog uses the global data structure PcapLogData as thread data
as well. This is possible because all operations on it are locked.
This patch introduces PcapLogThreadData. It contains a pointer to
the PcapLogData. Currently to the global instance, but in the future
it may hold a thread-local instance of PcapLogData.
Add profiling to a logfile. Default is $log_dir/pcaplog_stats.log
The counters for open, close, rotate, write and handles are written
to it, as well as:
- total bytes written
- cost per MiB
- cost per GiB
Option is disabled by default.
Tracks: file open, file close, file rotate (which includes open and
close), file write and open handles.
Open handles measures the cost of open the libpcap handles.
When the config is missing, DefragPolicyGetHostTimeout will default
to returning -1. This will effectively set no timeout at all, leading
to defrag trackers being freed too early.
This patch is fixing an issue in defragmentation code. The
insertion of a fragment in the list of fragments is done with
respect to the offset of the fragment. But the code was using
the original offset of the fragment and not the one of the
new reconstructed fragment (which can be different in the
case of overlapping segment where the left part is trimmed).
This case could lead to some evasion techniques by causing
Suricata to analyse a different payload.
This patch fixes the following issue reported by valgrind:
31 errors in context 1 of 1:
Conditional jump or move depends on uninitialised value(s)
at 0x8AB2F8: UnixSocketPcapFilesCheck (runmode-unix-socket.c:279)
by 0x97725D: UnixCommandBackgroundTasks (unix-manager.c:368)
by 0x97BC52: UnixManagerThread (unix-manager.c:884)
by 0x6155F6D: start_thread (pthread_create.c:311)
by 0x6E3A9CC: clone (clone.S:113)
The running field in PcapCommand was not initialized.
This patch fixes an issue in unix socket handling. It is possible
that a socket did disconnect when analysing a command and because
the data treatment is done in a loop on clients this was leading
to a update of the list of clients during the loop. So we need
in fact to use TAILQ_FOREACH_SAFE instead of TAILQ_FOREACH.
Reported-by: Luigi Sandon <luigi.sandon@gmail.com>
Fix-suggested-by: Luigi Sandon <luigi.sandon@gmail.com>
It is possible to have a non-contiguous CPU set, which was not being
handled correctly on the TILE architecture.
Added a "rank" field in the ThreadVar to store the worker's rank separately
from the cpu for this case.
When applying wildcard thresholds (with sid = 0 and/or gid = 0) it's wrong
to exit on the first signature already having an event filter. Indeed,
doing so results in the theshold not being applied to all subsequent
signatures. Change the code in order to skip signatures with event
filters instead of breaking out of the loop.
If a live reload signal was given before the engine was fully started
up (e.g. pcap file thread waiting for a disk to spin up), a segv could
occur.
This patch only enables live reloads after the threads have been
started up completely.
*** CID 1211009: Bad bit shift operation (BAD_SHIFT)
/src/output-json-http.c: 265 in JsonHttpLogJSON()
259 /* log custom fields if configured */
260 if (http_ctx->fields != 0)
261 {
262 HttpField f;
263 for (f = HTTP_FIELD_ACCEPT; f < HTTP_FIELD_SIZE; f++)
264 {
>>> CID 1211009: Bad bit shift operation (BAD_SHIFT)
>>> In expression "1 << f", left shifting by more than 31 bits has undefined behavior. The shift amount, "f", is as much as 46.
265 if ((http_ctx->fields & (1<<f)) != 0)
266 {
267 /* prevent logging a field twice if extended logging is
268 enabled */
269 if (((http_ctx->flags & LOG_HTTP_EXTENDED) == 0) ||
270 ((http_ctx->flags & LOG_HTTP_EXTENDED) !=
________________________________________________________________________________________________________
*** CID 1211010: Bad bit shift operation (BAD_SHIFT)
/src/output-json-http.c: 492 in OutputHttpLogInitSub()
486 {
487 if ((strcmp(http_fields[f].config_field,
488 field->val) == 0) ||
489 (strcasecmp(http_fields[f].htp_field,
490 field->val) == 0))
491 {
>>> CID 1211010: Bad bit shift operation (BAD_SHIFT)
>>> In expression "1 << f", left shifting by more than 31 bits has undefined behavior. The shift amount, "f", is as much as 46.
492 http_ctx->fields |= (1<<f);
493 break;
494 }
495 }
496 }
497 }
StreamTcpSetDisableRawReassemblyFlag() has the same effect as
AppLayerParserTriggerRawStreamReassembly in that it will force the
raw reassembly to flush out asap. So it is redundant to call both.
Implement StreamTcpSetDisableRawReassemblyFlag() which stops raw
reassembly for _NEW_ segments in a stream direction.
It is used only by TLS/SSL now, to flag the streams as encrypted.
Existing segments will still be reassembled and inspected, while
new segments won't be. This allows for pattern based inspection
of the TLS handshake.
Like is the case with completely disabled 'raw' reassembly, the
logic is that the segments are flagged as completed for 'raw' right
away. So they are not considered in raw reassembly anymore.
As no new segments will be considered, the chunk limit check will
return true on the next call.
Have a single function StreamTcpReturnSegmentCheck determine if a
segment is ready to be removed from the stream.
Handle FLOW_NOPAYLOAD_INSPECT in raw reassembly.
httplog_ctx->fields would not be initialized before setting flags in
it:
Scanbuild:
output-json-http.c:491:46: warning: The left expression of the compound assignment is an uninitialized value. The computed value will also be garbage
http_ctx->fields |= (1<<f);
~~~~~~~~~~~~~~~~ ^
1 warning generated.
Drmemory:
~~27874~~ Error #1: UNINITIALIZED READ: reading register eax
~~27874~~ # 0 JsonHttpLogJSON [/home/buildbot/qa/buildbot/donkey/drmemory/Suricata/src/output-json-http.c:260]
~~27874~~ # 1 JsonHttpLogger [/home/buildbot/qa/buildbot/donkey/drmemory/Suricata/src/output-json-http.c:375]
Just memset the whole structure right after initialition.
Fix issue detected byCoverity:
*** CID 1197756: Bad bit shift operation (BAD_SHIFT)
/src/util-rohash.c: 74 in ROHashInit()
68 }
69 if (hash_bits < 4 || hash_bits > 32) {
70 SCLogError(SC_ERR_HASH_TABLE_INIT, "invalid hash_bits setting, valid range is 4-32");
71 return NULL;
72 }
73
>>> CID 1197756: Bad bit shift operation (BAD_SHIFT)
>>> In expression "1U << hash_bits", left shifting by more than 31 bits has undefined behavior. The shift amount, "hash_bits", is as much as 32.
74 uint32_t size = hashsize(hash_bits) * sizeof(ROHashTableOffsets);
75
76 ROHashTable *table = SCMalloc(sizeof(ROHashTable) + size);
77 if (unlikely(table == NULL)) {
78 SCLogError(SC_ERR_HASH_TABLE_INIT, "failed to alloc memory");
79 return NULL;
This was only a potential issue as ROHashInit was only called with
hash_bits 16 in the code.
Bug #1170.
During socket creation all error cases were leading to suricata to
retry the opening of capture. This patch updates this behavior to
have fatal and recoverable error case. In case of a fatal error,
suricata is leaving cleanly.
Change GAP detection logic. If we encounter missing data before
last_ack, we know we have missed data. The receiving host has ack'd
it already, so a retransmission of the missing data is highly
unlikely.
AppLayer reassembly correctly only flags a segment as done when it's
completely used in reassembly. Raw reassembly could flag a partially
used segment as complete as well. In this case the segment could be
discarded early. Further reassembly would miss data, leading to a
gap. Due to this, up to 'window size' bytes worth of segments could
remain in the session for a long time, leading to memory resource
pressure.
This patch sets the flag correctly.
Some traffic is very unbalanced. For example a 4 bytes request
followed by 12mb of response data. If the 4 bytes don't lead to
the protocol being detected, then app layer processing won't
start, but it will not give up either. It will just wait for more
data. This leads to piling up data on the opposite side.
Observed:
TS: 4 bytes. PP failed (bit set), PM has not given up (bit not set).
This makes sense as max_depth is not yet reached.
TC: 12mb. PP and PM failed (bits set).
As ts-PM never gives up, we never consider proto detect complete,
so all segments in either direction are still kept in the session.
This patch adds a cutoff for this case:
- IF for TS we have PP bit set, PM not set, AND
- we have TC both bits set, AND
- proto is unknown, AND
- TC data is 100k already, THEN
- give up on protocol detection.
The same for the opposite direction.
The reassembly gap detection makes use of the window. However, in
midstream mode the window size we use is unreliable, as we have to
assume window scaling is in place. This patch improves midstream
handling of those cases.
In midstream mode we may encounter a case where the data we is beyond
the isn, but below last_ack. This means we're missing some data, that
is already acked so it won't be retransmitted. Therefore, we can
conclude it's a data gap.
If we're getting a lot of data in one direction and the proto for this
direction is unknown, proto detect will hold up segments in the segment
list in the stream. They are held so that if we detect the protocol on
the opposing stream, we can still parse this side of the stream as well.
However, some sessions are very unbalanced. FTP data channels, large
PUT/POST request and many others, can lead to cases where we would have
to store many megabytes worth of segments before we see the opposing
stream. This leads to risks of resource starvation.
In this patch, a cutoff point is enforced. If we've stored 100k in one
direction and we've seen no data in the other direction, we give up.
If we've given up, the applayer_proto_detection_skipped event is set.
app-layer-event: applayer_proto_detection_skipped;
If a TCP session is midstream, we may end up with a case where the
start of an HTTP request is missing. We won't detect HTTP based on
the request. However, the reply is fine, so we detect HTTP anyway.
This leads to passing the incomplete request to the htp parser.
This has been observed, where the http parser then saw many bogus
requests in the incomplete data. This is not limited to HTTP.
To counter this case, a midstream session MUST find it's protocol
in the toserver direction. If not, we assume the start of the
request/toserver is incomplete and no reliable detection and parsing
is possible. So we give up.
If midstream is enabled and the first packet is the syn/ack packet from
the 3whs, initialized server.last_ack to the packets seq.
This fixes tracking the session.
Extended data were freed before the release function was called.
The result was that, in AF_PACKET IPS mode, the release function
was only sending void data because it the content of the extended
data is the content of the packet.
This patch updates the code to have the freeing of extended data
done in the cleaning function for a packet which is called by the
release function. This improves consistency of the code and fixes
the bug.
The direction specific masks were not used correctly. The toserver ones
were only used for 'dp' registrations, the toclient ones only for 'sp'.
The patch merges them.
If a flow matches both an 'sp' based PP registration and a 'dp' based,
until now we would only check the 'dp' one. This patch changes that. It
will inspect both.
Instead of the notion of toserver and toclient protocol detection, use
destination port and source port.
Independent of the data direction, the flow's port settings will be used
to find the correct probing parser, where we first try the dest port,
and if that fails the source port.
Update the configuration file format, where toserver is replaced by 'dp'
and toclient by 'sp'. Toserver is intrepreted as 'dp' and toclient as
'sp' for backwards compatibility.
Example for dns:
dns:
# memcaps. Globally and per flow/state.
#global-memcap: 16mb
#state-memcap: 512kb
# How many unreplied DNS requests are considered a flood.
# If the limit is reached, app-layer-event:dns.flooded; will match.
#request-flood: 500
tcp:
enabled: yes
detection-ports:
dp: 53
udp:
enabled: yes
detection-ports:
dp: 53
Like before, progress of protocol detection is tracked per flow direction.
Bug #1142.
Instead of error phrone externs with macro's, use functions with a local
static enum var instead.
- EngineModeIsIPS(): in IPS mode
- EngineModeIsIDS(): in IDS mode
To set the modes:
- EngineModeSetIDS(): IDS mode (default)
- EngineModeSetIPS(): IPS mode
Bug #1177.
The SLOAD define using __insn_ld2s_L2 is used to provide a compiler
hint that the load will come from the L2 cache instead of the L1. It
also specifies that it is a 2 byte signed load. For the Tiny MPM, that
needs to be a 1-byte load, which is what is specified in util-ac-mpm-tile.c,
but the #undef was removing that definition.
Previously, the alstate use in the main detect loop was unsafe. The
alstate pointer would be set duing a lock, but it would again be used
after one or more lock/unlock cycles. If the data pointed to would
disappear, a dangling pointer would be the result.
Due to they way flows are cleaned up using reference counting and
such, changes of this happening were very small. However, at least
one path can lead to this situation. So it had to be fixed.
Make DeStateDetectContinueDetection get it's own alstate pointer instead
of using the one that was passed to it. We now get and use it only
inside a flow lock.
Make DeStateDetectStartDetection get it's own alstate pointer instead
of using the one that was passed to it. We now get and use it only
inside a flow lock.
This is an intrusive change. This patch modifies the way AMATCH
inspection uses locking.
So far, each keyword did it's own locking. This lead to a situation
where a 'alstate' pointer was passed around that was not always
protected by a lock.
This patch moves the locking to the Stateful detection functions.
This patch fixes a warning message when suricata is started without
giving an interface name on the command line. The code was trying
to get the MTU even if pcap_dev was not set.
FreeBSD 10 32-bit with clang 3.3:
log-tlslog.c:172:14: error: format specifies type 'long' but the argument has type 'time_t' (aka 'int') [-Werror,-Wformat]
p->ts.tv_sec,
^~~~~~~~~~~~
1 error generated.
detect-engine-payload.c:508:27: warning: format specifies type 'long' but the argument has type 'time_t' (aka 'int') [-Wformat]
printf("%ld.%06ld\n", tv_diff.tv_sec, (long int)tv_diff.tv_usec);
~~~ ^~~~~~~~~~~~~~
%d
1 warning generated.
Add output 'free list' that contains all the output ctx' that need
cleanup at shutdown. It differs from the runmode output list in that
it will also contain a 'parent' for the submodules that share the
context of it's parent.
For the loggers that we allow only one instance for: tls, ssh, drop, we
track active loggers through Output*Enable functions. Add Disable
functions to mirror this. They are to be called from the shutdown funcs
those loggers use.
This is a beginning of implementation for bug #1660:
https://redmine.openinfosecfoundation.org/issues/1160
This patch adds a cleaning function for each logger of new type
(packet, tx and file). These functions are called in RunModeShutDown().
The state of this patch is that it is crashing suricata when sending
pcap to analyse:
- At first pcap if tx and file cleaning function are called
- At second pcap if only packet cleaning function is called
The cause in first case is unknown. In second case this is due to
the necessity of cleaning the list of logger registered to a logging
type.
The OpenSSL implementation of RFC 6520 (Heartbeat extension) does not
check the payload length correctly, resulting in a copy of at most 64k
of memory from the server (ref: CVE-2014-0160).
This patch adds support for decoding heartbeat messages (if not
encrypted), and checking several parts (type, length and padding).
When an anomaly is detected, a TLS event is raised.
In case of multiple transactions, the stored AMATCH list would not have
been reset, but it would still be reconsidered. Even though none would
match, the engine would still conclude that the rule matched.
Pierre Chifflier pointed out that a rule like:
alert tls any any -> any any (msg:"TLS store"; tls.issuerdn:!"C=FR"; tls.store;)
was alerting but not storing the certificate. If the filter was
removed:
alert tls any any -> any any (msg:"TLS store"; tls.store;)
then tls.store is working as expected.
This was linked with fact that logging is only done once for a SSL
state. So without filter, once we have the info we can log and we
run the storage. But when there is a filter, we log and then there
is a filter analysis and alerting. And as logging as already be done
we don't enter in the logging function and there is no storage.
This patch forces the entrance in the log function when there is a
request for TLS storage. And it adds an exit in the logging function
to only do the storage part if the TLS state has already being logged.
Previously the sync code would depend on traffic to complete. This
patch adds poll support and can complete the setup if the poll timeout
is reached as well.
Part of bug #1130.
This patch is updating af-packet to discard packets that have been
sent to a socket before all socket in a fanout group have been setup.
Without this, there is no way to assure that all packets for a single
flow will be treated by the same thread.
Tests have been done on a system with an ixgbe network card. When using
'cluster_flow' load balancing and disactivating receive hash on the iface:
ethtool -K IFACE rxhash off
then suricata is behaving as expected and all packets for a single flow
are treated by the same thread.
For some unknown reason, this is not the case when using cluster_cpu. It
seems that in that case the load balancing is not perfect on the card side.
The rxhash offloading has a direct impact on the cluster_flow load balancing
because load balancing is done by using a generic hash key attached to
each skb. This hash can be computed by the network card or can be
computed by the kernel. In the xase of a ixgbe network card, it seems there
is some issue with the hash key for TCP. This explains why it is necessary to
remove the rxhash offloading to have a correct behavior. This could also
explain why cluster_cpu is currently failing because the card is using the
same hash key computation to do the RSS queues load balancing.
This allows for registering a keyword under another name while keeping
the old name active and supported.
Do this for 'luajit', which can now also be used as just 'lua'.
The HTTP module of Eve didn't register itself with the app-layer
for HTTP. This meant that if no other HTTP logger was active, the
HTTP logging in Eve wouldn't work.
This patch makes the HTTP Eve module register itself correctly.
Bug #1133.
To remain constistent with the other logs, set the event type to
the same name as the structure containing the defails. In this
case fileinfo.
Part of bug #1127.
Due to what appears to be an issue in logstash, the 'file' part of
the file event types was masked by a field that logstash-forwarder
added itself.
Since logstash-forwarder is an important part of the logstash stack,
this patch works around the issue by renaming our 'file' structure
to 'fileinfo', thus resolving the naming conflict.
Bug #1127
Move pfring_enable_ring to the start of ReceivePfringLoop() so that
it's guaranteed to be called after all threads have called
pfring_set_cluster first.
This is necessary because pfring will already make packets available
to thread N, while thread N+1 is still registering itself. This leads
to cases where the first packet(s) of a flow are processed by a
different thread in Suricata than the later ones.
This is a race condition only at start up. New flows after the pfring
initialization is complete will not be influenced by this.
Bug #1129.
This patch updates the timestamp format used in eve loggin.
It uses a ISO 8601 comptatible string. This allow tools parsing
the output to easily detect adn/or use the timestamp.
In the EVE JSON output, the value of the timestamp key has been
changed to 'timestamp' (instead of 'time'). This allows tools
like Splunk to detect the timestamp and use it without configuration.
Logstash configuration is simple:
input {
file {
path => [ "/usr/local/var/log/suricata/eve.json" ]
codec => json
type => "suricata-log"
}
}
filter {
if [type] == "suricata-log" {
date {
match => [ "timestamp", "ISO8601" ]
}
}
}
In splunk, auto detection of the fle format is failling and it seems
you need to define a type to parse JSON in
$SPLUNK_DIR/etc/system/local/props.conf:
[suricata]
KV_MODE = json
NO_BINARY_CHECK = 1
TRUNCATE = 0
Then you can simply declare the log file in
$SPLUNK_DIR/etc/system/local/inputs.conf:
[monitor:///usr/local/var/log/suricata/eve.json]
sourcetype = suricata
In both cases the timestamp are correctly imported by
the tools.
PF_RING is delivering the packet with VLAN header stripped. This
patch updates the code to get the information from PF_RING extended
header information.
This patch uses the new function SCKernelVersionIsAtLeast to know
that we've got a old kernel that do not strip the VLAN header from
the message before sending it to userspace.
In case of 'alert ip' rules that have ports, the port checks would
be bypassed for non-port protocols, such as ICMP. This would lead to
a rule matching: a false positive.
This patch adds a check. If the rule has a port setting other than
'any' and the protocol is not TCP, UDP or SCTP, then we rule won't
match.
Rules with 'alert ip' and ports are rare, so the impact should be
minimal.
Bug #611.
Eve-log would call GET_VLAN_ID on the packets vlan header if p->vlan_idx
was bigger than 0. GET_VLAN_ID would then unconditionally dereference
p->vlanh[0] or [1]. However, there are a number of cases in which these
pointers are not set. Defrag pseudo packets, AF_PACKET and in the future
PF_RING, do set the id's, but not the header pointers.
This patch adds 2 new macro's which are wrappers around a function:
VLAN_GET_ID1 and VLAN_GET_ID2 get the id's by calling DecodeVLANGetId.
This function will return the correct id.
Bug #1120.
app-layer-ssh.c:165:5: warning: Value stored to 'input_len' is never read
input_len -= 1;
^ ~
1 warning generated.
app-layer-ssh.c:160:5: warning: Value stored to 'input_len' is never read
input_len -= 4;
^ ~
1 warning generated.
Previously the software version would only contain up to the first
space.
E.g. in SSH-2.0-OpenSSH_4.7p1 Debian-8ubuntu3
It would contain "OpenSSH_4.7p1".
This patch changes the behavior to:
"OpenSSH_4.7p1 Debian-8ubuntu3"
This patch adds a check that was missing when specifying BPF filter
from a file. Suricata behavior should have been the same as when
BPF filter is specified on command line.
Fix warning spotted by clang on FreeBSD:
source-ipfw.c:241:49: warning: use of logical '||' with constant operand [-Wconstant-logical-operand]
if (suricata_ctl_flags & (SURICATA_STOP || SURICATA_KILL)) {
^ ~~~~~~~~~~~~~
source-ipfw.c:241:49: note: use '|' for a bitwise operation
if (suricata_ctl_flags & (SURICATA_STOP || SURICATA_KILL)) {
^~
|
Use same logic as the one used in other capture mode.
In the case of running mode like NFQ there is no need possibility
to compute the statistics as it is done in LiveDevice (drop and
checksum count are meaningless).
This patch adds a function that allow running mode to disable the
display of the counters at exit.
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.
This is required because SCMemcmpLowercase() expects it first argument
to be already lowercase for the comparison. This is done by using
memcpy_tolower() for NO_CASE patterns.
This addresses code review comments from Victor.
The case_state in MPMs was just to track when a pid could have no-case and
case-sensitive matches for the same PID. Now that can't happen after fixing
bug 1110, so remove the code and storage for case_state.
This fixes bug 1110. When assigning PIDs, use the NO_CASE flag when comparing
for duplicates. The state of the flag must be the same, but also use the same
type of comparisons when checking for duplicates.
Previously, "foo":CS would match with "foo":CI when it should not.
and "foo":CI would not match "FoO":CI when it should. Both of those
cases are fixed with this change.
This then allows simplifying the use of pid in MPMs because now if they
pids match, then so do the flags, so checking the flags is not required.
In the stream engine, Init() can fail if the memcap is reached. In this
case the segment was not freed by PoolGet:
==8600== Thread 1:
==8600== 70,480 bytes in 1,762 blocks are definitely lost in loss record 611 of 612
==8600== at 0x4C2A2DB: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==8600== by 0x914CC8: TcpSegmentPoolAlloc (stream-tcp-reassemble.c:166)
==8600== by 0xA0D315: PoolGet (util-pool.c:297)
==8600== by 0x9302CD: StreamTcpGetSegment (stream-tcp-reassemble.c:3768)
==8600== by 0x921FE8: StreamTcpReassembleHandleSegmentHandleData (stream-tcp-reassemble.c:1873)
==8600== by 0x92EEDA: StreamTcpReassembleHandleSegment (stream-tcp-reassemble.c:3584)
==8600== by 0x8D3BB1: HandleEstablishedPacketToServer (stream-tcp.c:1969)
==8600== by 0x8D7F98: StreamTcpPacketStateEstablished (stream-tcp.c:2323)
==8600== by 0x8F13B8: StreamTcpPacket (stream-tcp.c:4243)
==8600== by 0x8F2537: StreamTcp (stream-tcp.c:4485)
==8600== by 0x95DFBB: TmThreadsSlotVarRun (tm-threads.c:559)
==8600== by 0x8BE60D: TmThreadsSlotProcessPkt (tm-threads.h:142)
tcp.segment_memcap_drop | PcapFile | 1762
This patch fixes PoolGet to both Cleanup and Free the Alloc'd data in
case Init fails.
==15745== 3 bytes in 1 blocks are definitely lost in loss record 5 of 615
==15745== at 0x4C2A2DB: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==15745== by 0x71858C1: strdup (strdup.c:42)
==15745== by 0xA20814: SCProtoNameInit (util-proto-name.c:75)
==15745== by 0x952D1B: PostConfLoadedSetup (suricata.c:1983)
==15745== by 0x9537CD: main (suricata.c:2112)
Also, clean up and add a check to make sure it's initialized only once.
The full path of the script names is stored in a buffer that wasn't
freed at exit.
==24195== 41 bytes in 1 blocks are definitely lost in loss record 300 of 613
==24195== at 0x4C2A2DB: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==24195== by 0x565D06: DetectLoadCompleteSigPath (detect.c:251)
==24195== by 0x7CABE8: DetectLuajitParse (detect-luajit.c:595)
==24195== by 0x7CD2AE: DetectLuajitSetup (detect-luajit.c:827)
==24195== by 0x7DC273: SigParseOptions (detect-parse.c:547)
==24195== by 0x7DDC75: SigParse (detect-parse.c:856)
==24195== by 0x7E1C2B: SigInitHelper (detect-parse.c:1336)
==24195== by 0x7E2968: SigInit (detect-parse.c:1559)
==24195== by 0x7E37B1: DetectEngineAppendSig (detect-parse.c:1831)
==24195== by 0x566D17: DetectLoadSigFile (detect.c:335)
==24195== by 0x567636: SigLoadSignatures (detect.c:423)
==24195== by 0x951A97: LoadSignatures (suricata.c:1816)
This patch frees the buffer.
For packets that were freed, not recycled, profiling memory wasn't
freed:
==15745== 13,312 bytes in 8 blocks are definitely lost in loss record 611 of 615
==15745== at 0x4C2C494: calloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==15745== by 0xA190D5: SCProfilePacketStart (util-profiling.c:963)
==15745== by 0x4E4345: PacketGetFromAlloc (decode.c:134)
==15745== by 0x83FE75: FlowForceReassemblyPseudoPacketGet (flow-timeout.c:276)
==15745== by 0x8413BF: FlowForceReassemblyForHash (flow-timeout.c:588)
==15745== by 0x841897: FlowForceReassembly (flow-timeout.c:716)
==15745== by 0x9540F6: main (suricata.c:2296)
==15745==
==15745== 14,976 bytes in 9 blocks are definitely lost in loss record 612 of 615
==15745== at 0x4C2C494: calloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==15745== by 0xA190D5: SCProfilePacketStart (util-profiling.c:963)
==15745== by 0x4E4345: PacketGetFromAlloc (decode.c:134)
==15745== by 0x83FE75: FlowForceReassemblyPseudoPacketGet (flow-timeout.c:276)
==15745== by 0x841508: FlowForceReassemblyForHash (flow-timeout.c:620)
==15745== by 0x841897: FlowForceReassembly (flow-timeout.c:716)
==15745== by 0x9540F6: main (suricata.c:2296)
This patch addresses that.
If lock profiling was compiled in, but disabled in the config a
serious memory leak condition was triggered.
Valgrind output:
==11169== 9,091,248 bytes in 189,401 blocks are definitely lost in loss record 564 of 564
==11169== at 0x4C2A2DB: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==11169== by 0xABC44C: LockRecordAdd (util-profiling-locks.c:112)
==11169== by 0xABC950: SCProfilingAddPacketLocks (util-profiling-locks.c:141)
==11169== by 0xA04CD5: TmThreadsSlotVarRun (tm-threads.c:562)
==11169== by 0x958793: TmThreadsSlotProcessPkt (tm-threads.h:142)
==11169== by 0x9599C3: PcapFileCallbackLoop (source-pcap-file.c:172)
==11169== by 0x56FC130: ??? (in /usr/lib/x86_64-linux-gnu/libpcap.so.1.4.0)
==11169== by 0x959D24: ReceivePcapFileLoop (source-pcap-file.c:210)
==11169== by 0xA05B9E: TmThreadsSlotPktAcqLoop (tm-threads.c:703)
==11169== by 0x6155F6D: start_thread (pthread_create.c:311)
==11169== by 0x6E399CC: clone (clone.S:113)
Some of the packets counters were using a 32bit integer. Given the
bandwidth that is often seen, this is not a good idea. This patch
switches to 64bit counter.
Packets counter is incremented in AFPDumpCounters and it was
also incremented during packet reading. The result was a value
that is twice the expected result.
Spotted-by: Victor Julien <victor@inliniac.net>
Until now a PoolInit failure for the segment pools would result in
an abort() through BUG_ON(). This patch adds a proper error message,
then exits.
Bug #1108.
When TcpSegmentPoolInit fails (e.g. because of a too low memcap),
it would free the segment. However, the segment memory is managed
by the Pool API, which would also free the same memory location.
This patch fixes that.
Also, memset the structure before any checks are done, as the segment
memory is passed to TcpSegmentPoolCleanup in case of error as well.
Bug #1108
Fixes:
** CID 1075221: Dereference after null check (FORWARD_NULL)
/src/suricata.c: 1344 in ParseCommandLine()
The reason it gave this warning is that in other paths using optarg
there was a check, so the checker assumed optarg can be NULL.
This patch fixes:
** CID 1187544: Missing break in switch (MISSING_BREAK)
/src/decode-icmpv6.c: 268 in DecodeICMPV6()
** CID 1187545: Missing break in switch (MISSING_BREAK)
/src/decode-icmpv6.c: 270 in DecodeICMPV6()
** CID 1187546: Missing break in switch (MISSING_BREAK)
/src/decode-icmpv6.c: 272 in DecodeICMPV6()
** CID 1187547: Missing break in switch (MISSING_BREAK)
/src/decode-icmpv6.c: 274 in DecodeICMPV6()
It duplicates the logic instead of adding 'fall through' statements
as the debug statements were wrong and confusing. For ND_REDIRECT
all 5 ND_* types would have been printed.
By assuming that HTPCallbackRequestLine would always be run first,
an memory leak was introduced. It would not check if user data already
existed in the tx, causing it to overwrite the user data pointer is
it already existed.
Bug #1092.
In really short Suricata runtimes the capture counters would not
be updated. This patch does a force update at the end of the
capture loops in pcap and af-packet.
The probing parser registration function
AppLayerProtoDetectPPParseConfPorts was a void, meaning it would
give no feedback to the registering protocol implementation. If a
config was missing, it would just give up.
This patch changes it to return a bool. 0 if no config was found,
1 if a config was found.
This allows the caller to setup a default case.
The probing parser detection ports yaml settings of the TCP part
of the DNS parser accidentally used udp as protocol string, causing
the wrong part of the YAML to be evaluated.
If the protocol is disabled, app-layer-event would print a cryptic
error message. This patch makes sure we inform the user the protocol
is in fact disabled.
This patch introduces a new counter "decoder.vlan_qinq". It counts
packets that have more than two stacked vlan layers.
Packets with 2 vlan layers will both increment "decoder.vlan" and
"decoder.vlan_qinq".
A node isn't known to be a sequence node until the YAML is parsed.
If a node sequence node was set on the command line, promote
it to a sequence node when it is discovered by YAML to be
a sequence node.
Fixes comment #18 in issue 921.
When creating a pseudo packet with the reassembled IP packet, the
parent's vlan id or id's are also needed. The defrag packet is run
through decode and the flow engine, where the vlan id is necessary
for connecting the packet to the correct flow.
Some old distribution don't ship recent enough linux header. This
result in TP_STATUS_VLAN_VALID being undefined. This patch defines
the constant and use it as it is used in backward compatible method
in the code: the flag is not set by kernel and a test on vci value
will be made.
This should fix https://redmine.openinfosecfoundation.org/issues/1106
Flow-timeout code injects pseudo packets into the decoders, leading
to various issues. For a full explanation, see:
https://redmine.openinfosecfoundation.org/issues/1107
This patch works around the issues with a hack. It adds a check to
each of the decoder entry points to bail out as soon as a pseudo
packet from the flow timeout is encountered.
Ticket #1107.
FlowReference stores the flow in the destination pointer and increases
the flow reference counter (use_cnt). This should only be called once
per destination pointer. The reference counter is decremented when
FlowDereference is called. Multiple FlowReference calls would lead to
multiple use_cnt bumps, while there would be only one FlowRereference.
This lead to a use_cnt that would never become 0, meaning the flow
would stay in the hash for the entire lifetime of the process.
The fix here is to check if the destination pointer is already set to
the flow. If so, we don't increase the reference counter.
As this is really a bug, this condition will lead to a BUG_ON if the
DEBUG_VALIDATION checking is enabled.
The HTP config tree is a radix. The lookups are updated to the new API.
The return of user_data is treated as a succesful lookup, instead of
the node itself.
This patch updates all the radix tests to the new API. In most cases
it just passes a NULL user data return pointer.
It also removes the tests related to SC_RADIX_NODE_USERDATA, as this
macro is removed.
Bug #1073
The radix tree stores user data. However, it had no function to return
this data to the consumers of the API. Instead, on lookup, it would
set a field "user_data_result" in the nodes prefix structure which
could then be read by the caller.
Apart for this not being a very nice design as it exposes API internals
to the caller, it is not thread safe. By updating the global data
structure without any form (or suggestion) of locking, threads could
overwrite the same field unexpectedly.
This patch modifies the lookup logic to get rid of this stored
user_data_result. Instead, all the lookup functions how take an
addition argument: void **user_data_result.
Through this pointer the user data is returned. It's allowed to be
NULL, in this case the user data is ignored.
This is a significant API change, that affects a lot of tests and
callers. These will be updated in follow up patches.
Bug #1073.
Handles ND_ROUTER_SOLICIT, ND_ROUTER_ADVERT, ND_NEIGHBOUR_ADMIN,
ND_NEIGHBOUR_SOLICIT and ND_REDIRECT. Don't set ICMPV6_UNKONWN_CODE
if code is the expected value of 0.
This patch uses the new function SCKernelVersionIsAtLeast to know
that we've got a old kernel that do not strip the VLAN header from
the message before sending it to userspace.
Since commit in kernel
commit a3bcc23e890a6d49d6763d9eb073d711de2e0469
Author: Ben Greear <greearb@candelatech.com>
Date: Wed Jun 1 06:49:10 2011 +0000
af-packet: Add flag to distinguish VID 0 from no-vlan.
a flag is set to indicate VLAN has been set in packet header.
As suggested in commit message, using a test of the flag followed
by a check on vci value ensure backward compatibility of the test.
AppLayerParserProtocolIsTxEventAware would check if a proto is tx
event aware by checking if it had registered a StateHasEvents function.
However, this is an optimization function. This patch changes it to
use the StateGetEvents function instead, which is a better indicator.
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
Generate the alert string into a temporary buffer before aquiring the
file lock. Only hold the file lock while writing the alert string to the
file.
In the case of multiple alerts, it would be better to generate all the
alerts, then aquire the lock once and write them all and then flush.
Changed PrintRawLineHexFp, which printed to a file, to PrintBufferRawLineHex,
that puts the same output into a string buffer. It was only used by fast.log.
Fix issue where negating a range containing a negation would fail.
E.g. HOME_NET: [192.168.0.0/16,!192.168.10.0], can be used in a rule
as !$HOME_NET.
Also, fix another parsing issue:
If the negation range would be bigger than the 'positive' range, parsing
wouldn't be correct. Now this case is rejected.
E.g. [192.168.1.3,!192.168.0.0/16] is now explicitly rejected
Ticket 1079.
End profiling inside the lock for a tunnel packet as otherwise another
thread may already free the packet while the profiling code runs.
SEGV's observed and now gone.
The TM_FLAG_LOGAPI_TM flag indicates that a module is run by the log
api, not by the 'regular' thread module call functions.
Set flag in all all Log API users' registration code.
Purpose of this flag is in profiling. In profiling output it will be
used to list log api thread modules separately.
The log API calls thread modules directly, so the TMM profiling logic
can be applied to it. This patch does so.
The "Thread Module" out now again lists the individual loggers. As the
module are normally called much less frequently the numbers are hard to
compare to pre-log-api numbers.
Add option "profiling.sample-rate":
# Run profiling for every xth packet. The default is 1, which means we
# profile every packet. If set to 1000, one packet is profiled for every
# 1000 received.
#sample-rate: 1000
This allows for configuration of the sample rate.
Instead of a large (6k+) structure in the Packet, make the profiling
storage dynamic. To do this the Packet->profile is now a pointer.
Initial support for selective sampling, e.g. only profile every
1000th packet.
A negated match is matching if the tested field is NULL. But as it
is not set, nor negated nor normal test must match.
Without this patch, a rule like:
alert tls any any -> any any (msg:"negated match"; tls.subject:!"CN=home.regit.org"; sid:1; rev:1;)
is alerting for all connections. Event if they are done on a certificate
with matching subject. This was due to the fact that tls protocol
is discovered before the handshake is complete. Thus the condition
on tls is true with a NULL tls.subject. And code was returning a
positive match in the case of a NULL subject and a signature with
a negated match.
Introduce a utility function to convert an array of bytes into a
null-terminated string:
char *BytesToString(const uint8_t *bytes, size_t nbytes);
All non-printables are copied over, except for '\0', which is
turned into literal '\' '0' in the string. So the resulting string
may be bigger than the input.
In various places SCStrndup was used to 'dup' a bstr string, however
libhtp provides bstr_util_strdup_to_c for this. As this is a cleaner
interface, it's preferred.
This patch separates http keys from file to have a different value
list:
{
"time":"01\/31\/2014-12:04:52.837245","event_type":"file","src_ip":"5.3.1.1","src_port":80,"dest_ip":"1.8.1.9","dest_port":9539,"proto":"TCP",
"http":{"url":"/foo/","hostname":"bar.com","http_refer":"http:\/\/bar.org","http_user_agent":"Mozilla\/5.0"},
"file":{"filename":"bar","magic":"unknown","state":"CLOSED","stored":false,"size":21}
}
One interest of this modification is that it is possible to use the
same key as the one used in http events. Thus correlating both type
of events is trivial. On code side, this will permit to factorize
the code by simply asking the underlying protocol to output its
info in a json object.
Second interest is that adding file extraction for a new protocol
will result in only changing the protocol specific json list.
This patch adds an event_type key to the generated events. Current
value is one of "dns", "alert, "file", "tls", "http", "drop". It is
then easy to differentiate in log analysis tools the events based on
source inside Suricata.
Without this patch DNS answers for a single query are stored in a
single json event. The result is an array in the object like this one:
{"type":"answer","id":45084,"rrname":"s-static.ak.facebook.com","rrtype":"CNAME","ttl":734},
{"type":"answer","id":45084,"rrname":"s-static.ak.facebook.com.edgekey.net","rrtype":"CNAME","ttl":1710},
This type of output is not well supported in logstash. It is
displayed as it is written above and it is not possible to
query the fields.
I think the reason is that this is not logical if we consider search
query. For example if we search for "rrname" equal "s-static.ak.facebook.com"
we got one entry with two values in it. That's against the logic
of event. Furthermore, if we want to get a complete query, we can
used the id.
This patch splits the answer part in mulitple message. The result
is then accepted by logstash and fields can be queried easily.
This patch updates DNS field name to be in sync with RFC 2629:
https://github.com/adulau/pdns-qof
This will allow to easily use Suricata with other passive DNS tools.
This patch is synchronizing key name with Common Information Model.
It updates key name following what is proposed in:
http://docs.splunk.com/Documentation/PCI/2.0/DataSource/CommonInformationModelFieldReference
The interest of these modifications is that using the same key name
as other software will provide an easy to correlate and improve
data. For example, geoip setting in logstash can be applied on
all src_ip fields allowing geoip tagging of data.
Both the drop and tls logs are currently not designed to have multiple
instances running. So until that is changed, error out if more than one
instance is started.
To support the 'eve-log' idea, we need to be able to force all log
modules to be enabled by the master eve-log module, and need to be
able to make all logs go into a single file. This didn't fit the
API so far, so added the sub-module concept.
A sub-module is a regular module, that registers itself as a sub-
module of another module:
OutputRegisterTxSubModule("eve-log", "JsonHttpLog", "http",
OutputHttpLogInitSub, ALPROTO_HTTP, JsonHttpLogger);
The first argument is the name of the parent. The 4th argument is
the OutputCtx init function. It differs slightly from the non-sub
one. The different is that in addition to it's ConfNode, it gets
the OutputCtx from the parent. This way it can set the parents
LogFileCtx in it's own OutputCtx.
The runmode setup code will take care of all the extra setup. It's
possible to register a module both as a normal module and as a sub-
module, which can operate at the same time.
Only the TxLogger API is handled in this patch, the rest will be
updated later.
- 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
The stream chunk pool contains preallocating stream chunks (StreamMsg).
These are used for raw reassembly, used in raw content inspection by
the detection engine. The default setting so far has been 250, which
was hardcoded. This meant that in setups that needed more, allocs and
frees would be happen constantly.
This patch introduces a yaml option to set the 'prealloc' value in the
pool. The default is still 250.
stream.reassembly.chunk-prealloc
Related to feature #1093.
The stream reassembly engine uses a set of pools in which preallocated
segments are stored. There are various pools each with different packet
sizes. The goal is to lower memory presure. Until now, these pools were
hardcoded.
This patch introduces the ability to configure them fully from the yaml.
There can be at max 256 of these pools.
Yaml layout is as follows:
stream:
reassemble:
segments:
- size: 2048
prealloc: 3000
- size: 4
prealloc: 1000
- size: 1024
prealloc: 2000
The size is the packet size. The prealloc value indicates how many
segments are set up at startup.
The pools have no limit wrt how many segments can be used of a certain
size. If the engine needs more than the prealloc size, segments are
malloc'd and free'd. The only limit here is the stream.reassemble.memcap.
If the yaml part if omitted, the default values are the same as before.
Feature #1093
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.
The new API call:
int AppLayerParserProtocolHasLogger(uint8_t ipproto,
AppProto alproto)
Returns TRUE if a logger is registered on the ip/alproto pair, and
FALSE otherwise.
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.
Fixes configure enabling of prelude. CFLAGS is reset, so the previous
adding of -DPRELUDE was nixed. Using AC_DEFINE now.
Cleanups:
- make functions static
- simplify handling of no prelude support
- move registration to the bottom
Remove separate ipv4 and ipv6 registration functions.
Make all functions static.
Move registration function to the bottom.
Simplify OS_WIN32 wrappers usage.
This patch converts log-tls to use the packet logger API. The packet
logger API was choosen as the TLS parser is not transaction aware.
To make sure the state is only logged once, the flag
SSL_AL_FLAG_STATE_LOGGED was added to the parser. This flag is checked
by the condition function, and set at the end of the Logger function.
Make all functions static. Remove separate ipv4 and ipv6 registration
functions. Move register function to the bottom so that we no longer
need function prototype declarations.
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.
If a protocol parser is active without a logger when detection is
disabled, the transaction handling logic would fail. Now it will
return the proper tx id so we can clean up the complete transactions.
When running w/o detect, TX cleanup handling needs to ignore the
inspect_id as it's only updated by detect.
This patch introduces a new ActiveTx handler for logging only:
AppLayerTransactionGetActiveLogOnly
If --disable-detection is passed on the commandline, this function
is registered.
Flow timeout code keeps track of thread module running detect, and
fails (hard) if it doesn't find it.
This changeset retrieves the global g_detect_disabled and passes
it to the timeout handling code during setup.
This global will be set to TRUE if detect is disabled. The reason for
adding a global is that there currently is no clean way to pass
configuration options to management threads.
If FlowForceReassemblyForFlowV2 can't get packets to inject into the
engine, until now it would bail and retry later. In case of resource
starvation issues, this would cause a lot of lock contention, as the
flow manager would try over and over again.
This patch limits FlowForceReassemblyForFlowV2 to one try per flow,
if it fails... bad luck. It will only fail in serious conditions,
which means we must prefer the health of the engine over the proper
inspection of the flow in question.
AppLayer Proto detection code uses a mix of pattern search and
"probing parsers". The pattern search validates potential matches
using a single pattern search algo. The code was using SpmSearch
for this, but this made it inefficient as it builds a BoyerMoore
context for each search. This lead to significant memory pressure,
especially on high speed/bandwidth boxes.
This patch switches the search calls to BoyerMoore and BoyerMoore-
Nocase directly. This can be done as the ctx' were available already.
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.
Don't use FlowGetProtoMapping at runtime, use f->protomap instead.
Add safety check to make sure its value is within range, as it's
used to index an array.
Update unittests to initialized flows (somewhat).
Remove unnecessay inlining.
Rename functions with wrong naming scheme. E.g. AllocAppLayer.. instead
of AppLayer..Alloc.
Use AppProto instead of uint16_t.
Convert u16 ipproto cases to u8.
This patch updates the DNS counters from the main AppLayer entry
functions. Due to the limited scope of AppLayerThreadCtx some of
the logic had to be implemented in app-layer.c, where it doesn't
belong.
To be able to register counters from AppLayerGetCtxThread, the
ThreadVars pointer needs to be available in it and thus in it's
callers:
- AppLayerGetCtxThread
- DecodeThreadVarsAlloc
- StreamTcpReassembleInitThreadCtx
Add memuse tracking and memcap checking to the DNS parsers. Memuse
is tracked globally and per flow (state).
Memcaps are also checked per flow and globally before memory allocs
are done.
In some cases the TLS state pointers to subject and issuerdn could
be overwritten by a new memory allocation, causing us to loose
track of the old.
This has been observed in the case of improper VLAN handling, where
it was suspected that multiple unrelated TLS streams were mangled
together.
The loop is freeing elements so we need to use the safe version
of TIALQ_FOREACH.
This fixes a valgrind error:
Thread 1 Suricata-Main:
Invalid read of size 8
at 0x8E129C: LiveDeviceListClean (util-device.c:167)
by 0x89B742: main (suricata.c:2284)
Address 0x8382988 is 24 bytes inside a block of size 40 free'd
at 0x4C2A70C: free (vg_replace_malloc.c:468)
by 0x8E1297: LiveDeviceListClean (util-device.c:179)
by 0x89B742: main (suricata.c:2284)
Buffers in per thread HTTP header, client body and server body storage
would be freed based on the usage indicator instead of the size
indicator.
As the usage indicator (e.g. hsbd_buffers_list_len) could be reset
while leaving the memory untouched for later reuse, the free function
would not iterate over all memory blocks.
Removed DrMemory suppressions as well.
Bug #980.
Logic of patch 98e4a14f6d was correct
but implementation is wrong because TP_STATUS_KERNEL is equal to
zero and thus can not be evaluated in a binary operation. This patch
updates the logic by doing two tests.
Reported-by: Alessandro Guido
In preparation of a patchset that will allow for disabling the detect
module, this patch introduces a way to register a function for getting
the lowest active tx id. This is used by the app layer for cleaning up
transactions that already fully inspected, and by the flow timeout code
to determine if a flow is fully inspected and logged at timeout.
The registration function RegisterAppLayerGetActiveTxIdFunc allows for
registration of a custom function of type:
uint64_t (*GetActiveTxIdFunc)(Flow *f, uint8_t flags);
If no function is called, AppLayerTransactionGetActiveDetectLog is used,
which implements the existing behaviour of considering both the
inspect_id's and the log_id.
In AppLayerTransactionsCleanup instead of figuring out 'done' tx id's
itself, now call AppLayerTransactionGetActive for both directions to
figure out the completed TX id's.
StreamMsgs would be stored in a per thread queue before being
attached to the tcp ssn. This is unnecessary, so this patch
removes this queue and puts the smsgs into the ssn directly.
Large patch as it affects a lot of tests.
StreamMsg' flow reference was used mostly to make sure a flow would
not get removed from the hash before inspection. For this it needed
to reference the flow use_cnt reference counter. Nowadays we have
more advanced flow timeout handling. This will make sure that if
there still are pending smsgs' in a flow, these will still be
processed.
Preparation for removing flow pointer from StreamMsg. Instead of
getting the ssn indirectly through StreamMsg->flow, we pass it
directly as all callers have it already.
StreamSmgs are used for raw stream reassembly only. They could also
be used to tell the rest of the engine about sequence gaps. This was
a left over from the older implementation, where the app layer used
the smsgs as well.
This patch updates the logic of the packet acquisition loop. When
the reader loop function is called and when the data to read
at offset is a without data (kernel) or still used by suricata. We
try to iter for a loop on the ring to try to find kernel put by
data.
As we are entering the function because the poll said there was some
data. This allow us to jump to the data added to the ring by the
kernel.
When using suricata in autofp mode, with multiple detect threads and
packet acquisition threads attached to a dedicated CPU, the reader
loop function was looping really fast because poll call was returning
immediatly because we did read the data available.
This patch adds a memory counter for HTP memory usage. As
there is no thread variables available in application layer
the counter has been added to the TCP reassembly thread.
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.
When running with sgh-mpm-context: full, many more MPMs are created
(16K) and many are small. If they have less than 128 states, they only
need 1 byte for the next state instead of 2 bytes, cutting the size of
the next-state table in half. This reduces total memory usage.
Since that makes 3 different state sizes (1, 2 and 4 bytes), rather
than going from 2 copies of the code to create the MPM to 3, I
factored out the code that fills the next-state table into three
functions so that all the other code could be the same.
The search function is now parameterize for 8-bit and 16-bit state
sizes and alphabet sizes 8, 16, 32, 64, 128 and 256.
Live device counter was in fact the number of packets seen by suricata
and not the total number of packet reported by pfring. This patch fixes
this by using counter provided by kernel instead.
Pfring kernel counter is per socket and is not cleared after read.
So to get the number of packet on the interface we can add the new
value for this thread and add it to the interface counter.
Live device counter was in fact the number of packets seen by suricata
and not the total number of packet reported by kernel. This patch fixes
this by using counter provided by kernel instead.
The counter is Clear On Read, so by adding the value fetch at each call
and earch sockets we get the number of packets and drops for the
interface.
- Seems to be a regression introduced in the commit
796bfab231 (fix was already done in commit
ee0b21652b)
- Doesn't happen with htplib v0.5.6, but it does in the latest, v0.5.9
For AppLayerThreadCtx, AppLayerParserState, AppLayerParserThreadCtx
and AppLayerProtoDetectThreadCtx, use opaque pointers instead of
void pointers.
AppLayerParserState is declared in flow.h as it's part of the Flow
structure.
AppLayerThreadCtx is declared in decode.h, as it's part of the
DecodeThreadVars structure.
A number of unittests would lead to clang build errors because
of unsafe det_ctx ptr usage. This patch fixes these and inits
det_ctx to NULL in the other detect tests.
app-layer.[ch], app-layer-detect-proto.[ch] and app-layer-parser.[ch].
Things addressed in this commit:
- Brings out a proper separation between protocol detection phase and the
parser phase.
- The dns app layer now is registered such that we don't use "dnstcp" and
"dnsudp" in the rules. A user who previously wrote a rule like this -
"alert dnstcp....." or
"alert dnsudp....."
would now have to use,
alert dns (ipproto:tcp;) or
alert udp (app-layer-protocol:dns;) or
alert ip (ipproto:udp; app-layer-protocol:dns;)
The same rules extend to other another such protocol, dcerpc.
- The app layer parser api now takes in the ipproto while registering
callbacks.
- The app inspection/detection engine also takes an ipproto.
- All app layer parser functions now take direction as STREAM_TOSERVER or
STREAM_TOCLIENT, as opposed to 0 or 1, which was taken by some of the
functions.
- FlowInitialize() and FlowRecycle() now resets proto to 0. This is
needed by unittests, which would try to clean the flow, and that would
call the api, AppLayerParserCleanupParserState(), which would try to
clean the app state, but the app layer now needs an ipproto to figure
out which api to internally call to clean the state, and if the ipproto
is 0, it would return without trying to clean the state.
- A lot of unittests are now updated where if they are using a flow and
they need to use the app layer, we would set a flow ipproto.
- The "app-layer" section in the yaml conf has also been updated as well.
of the archaic features we use in the app layer. We will reintroduce this
parser shortly. Also do note that keywords that rely on the ssh parser
would now be disabled.
Coverity 1139544
If strdup would fail, 'node' was freed but it wasn't set to NULL. The
code then returned node. The caller would not detect there was an error
and use the freed pointer.
By moving FlowReference() out of FlowGetFlowFromHash() and into the one
function that calls it, all the flow functions take const Packet * instead
of Packet *.
A ptr to local var is stored in the radix tree currently,
this patch permits to alloc space to store host timeout
and thus also free it when data is removed.
defrag-config.c: In function 'DefragParseParameters':
defrag-config.c:105: warning: passing argument 2 of 'DefragPolicyAddHostInfo' from incompatible pointer type
make[3]: *** [defrag-config.o] Error 1
This would only happen in memory failure conditions.
util-decode-der.c:634:27: warning: Potential leak of memory pointed to by 'child'
return (Asn1Generic *)node;
stream-tcp-reassemble.c:2569:17: warning: Value stored to 'seg' is never read
seg = seg->next;
^ ~~~~~~~~~
stream-tcp-reassemble.c:2587:17: warning: Value stored to 'seg' is never read
seg = seg->next;
These were only used if debug is enabled.
app-layer-dns-tcp.c:407:13: warning: Value stored to 'length' is never read
length = *data;
app-layer-dns-udp.c:236:13: warning: Value stored to 'length' is never read
length = *data;
Detect when default_packet_size is zero, which enables zero-copy mode for
pfring and in that case, do what AF Packet does and set pkt_ext pointer to
the data and set PKT_ZERO_COPY flag.
The uint8_t *pkt in the Packet structure always points to the memory
immediately following the Packet structure. It is better to simply
calculate that value every time than store the 8 byte pointer.
Some of the fields in the SCACTileCtx struct are only used to create the MPM,
but are not needed to search the MPM. Create a new structure to contain just
the data needed by AC Search. After creating the MPM, copy the data into the
new structure and then free the memory only needed during initialization.
This reduces the size of the AC-Tile MPM context from 1360 bytes down to 296
bytes.
Add two new mPIPE load-balancing configuration options in suricata.yaml.
1) "sticky" which keep sending flows to one CPU, but if that queue is full,
don't drop the packet, move the flow to the least loaded queue.
2) Round-robin, which always picks the least full input queue for each
packet.
Allow configuring the number of packets in the input queue (iqueue) in
suricata.yaml.
For the mPipe.buckets configuration, which must be a power of 2, round
up to the next power of two, rather than report an error.
Added mpipe.min-buckets, which defaults to 256, so if the requested number
of buckets can't be allocated, Suricata will keep dividing by 2 until either
it succeeds in allocating buckets, or reaches the minimum number of buckets
and fails.
In SigMatchSignatures, the value p->flow doens't change, but GCC can't
figure that out, so it reloads p->flow many times during the function.
When p->flow is loaded into the variable pflow once at the start of the
function, the compile then doesn't need to reload it.
[src/app-layer-htp.c:1967] -> [src/app-layer-htp.c:1978]: (warning) \
Possible null pointer dereference: tx - otherwise it is redundant \
to check it against null.
pcap has a callback function that is called for each packet. Once a
second, it's meant to 'dump stats'. However, the timing logic was
broken, so it would actually dump stats for each packet.
By moving the stats second timer into the thread vars, next calls of
the callback will be able to use the stored time.
Flow timeout code worked by luck when checking if a flow still needed
reassembly for app layer inspection or logging. It would check for a
part of raw reassembly (smsg list) to determine if detection was
needed. In this case it would also process app layer cleanup,
including logging.
Introduced AppLayerTransactionGetActive which returns the lowest tx_id
in a direction that still needs some work.
FlowForceReassemblyNeedReassmbly now uses it to determine if the
applayer still needs work.
Converted FlowForceReassemblyForHash to use the checking function
FlowForceReassemblyNeedReassmbly as well, so that checking if a flow
needs work is now unified.
Raw reassembly is used only by the detection engine. For users only
caring about logging it's a significant overhead, both in cpu and
memory usage.
The option is called 'raw' and lives under the stream.reassembly
options.
stream:
memcap: 32mb
checksum-validation: yes # reject wrong csums
inline: auto # auto will use inline mode in IPS mode, yes or no set it statically
reassembly:
memcap: 64mb
depth: 1mb # reassemble 1mb into a stream
toserver-chunk-size: 2560
toclient-chunk-size: 2560
randomize-chunk-size: yes
#randomize-chunk-range: 10
raw: false # <- new option
Spotted out by clang:
source-erf-dag.h|25 col 9| warning: '__SOURCE_ERR_DAG_H__'
is used as a header guard here, followed by #define of a different macro
[-Wheader-guard]
Prevents benign log message of parent nodes of final values being
redefined (which ends up having no affect as the final nodes
are protected from being removed).
If we have multiple layer of tunnel, the decoding of initial
Packet will recurse in DecodeTunnel function called in
PacketTunnelPktSetup. If we are not setting the pseudo
packet root before calling DecodeTunnel (as done in previous
code), then the tunnel root will no be correct for the lower
layer packets. This result in an counter problem and a suricata
failure after some time.
If a session only has data in one direction, like ftp data sessions,
protocol detection will only run in one direction. This led to a
situation where reassembly would hold all the segments as proto
detection was never flagged as complete.
This patch introduces a limit for protocol detection in this case.
If the limit is reached, detection will give up.
This patch adds a '-k' option to suricata to be able to specify
the checksum validation to use. If '-k all' is used, checksum
validation is forced. If '-k none' is used, no checksum validation
is made.
Message output in case of detection of a pcap file with a probable
cheksum issue has been updated to indicate that '-k' is a solution.
This patch adds support for checksum-checks in the pcap-file running
mode. This is the same functionnality as the one already existing for
live interface.
It can be setup in the YAML:
pcap-file:
checksum-checks: auto
A message is displayed for small pcap to warn that invalid checksum
rate is big on the pcap file and that checksum-check could
be set to no.
This patch set a new value in pkt->flag to signal that a packet is
invalid during decoding. The patch has been obtained via a coccinelle
transformation.
This patch adds and increments a invalid packet counter. It
does this by introducing PacketDecodeFinalize function
This function is incrementing the invalid counter and is also
signalling the packet to CUDA.
Remove all flags indicating the buffer type. They were only used
at parse time.
Because of this the DetectPcreData_ structure could shrink to 32
bytes.
This patch replaces PacketPseudoPktSetup by a better named
PacketTunnelPktSetup function which is also in charge of doing
the decoding of the tunneled packet.
This allow to clean the code. But it also fixes an issue.
Previously, if the DecodeTunnel function was failling (cause of
an invalid packet mainly), the result was that the original packet
to be considered as a tunnel packet (and not inspected by payload
detection).
In some cases, the decoding is not possible and some really invalid
packet can be created. This is in particular the case of tunnel. In
that case, it is more interesting to forget about the tunneled
packet and only consider the original packet.
DecodeTunnel function is maked as warn_unused_result because it is
meaningful for the decoder to know if the underlying data were not
correct. And in this case, only focus detection on the content.
Fix sigatures not supporting [10.0.0.0/24, !10.1.1.1] notation when
used directly in a rule instead of through a variable.
Add tests for Bugs #815 and #920.
When inspecting HTTP bodies there are several limits involved.
In this patch the reaching of the body limit will trigger body
inspection.
Without this, the body would only be inspected when inspection
limits "request-body-minimal-inspect-size" or
"response-body-minimal-inspect-size" were reached. If the body
limit was smaller than this value, the body would only be
inspected at the end of the tx or stream.
Duarte Silva
Victor Julien
Eric Leblond
Ken Steele
Giuseppe Longo
sxhlinux
Alexander Gozman
Alexander Gozman
Matt Carothers
sxhlinux
Mats Klepsland
Alessandro Guido
Jason Ish
jeka
Tom DeCanio
Will Metcalf
Pierre Chifflier
Anoop Saldanha