Commit Graph

11739 Commits (f2ab5803fbd88ae613429a382f28e81841832d0d)
 

Author SHA1 Message Date
Victor Julien ee41c0e293 flow/spare: implement pool shrinking
Remove at most one block per run, so it shrinks slowly.
5 years ago
Victor Julien b3599507f4 flow: redesign of flow timeout handling
Goals:
- reduce locking
- take advantage of 'hot' caches
- better locality

Locking reduction

New flow spare pool. The global pool is implmented as a list of blocks,
where each block has a 100 spare flows. Worker threads fetch a block at
a time, storing the block in the local thread storage.

Flow Recycler now returns flows to the pool is blocks as well.

Flow Recycler fetches all flows to be processed in one step instead of
one at a time.

Cache 'hot'ness

Worker threads now check the timeout of flows they evaluate during lookup.
The worker will have to read the flow into cache anyway, so the added
overhead of checking the timeout value is minimal. When a flow is considered
timed out, one of 2 things happens:

- if the flow is 'owned' by the thread it is handled locally. Handling means
  checking if the flow needs 'timeout' work.

- otherwise, the flow is added to a special 'evicted' list in the flow
  bucket where it will be picked up by the flow manager.

Flow Manager timing

By default the flow manager now tries to do passes of the flow hash in
smaller steps, where the goal is to do full pass in 8 x the lowest timeout
value it has to enforce. So if the lowest timeout value is 30s, a full pass
will take 4 minutes. The goal here is to reduce locking overhead and not
get in the way of the workers.

In emergency mode each pass is full, and lower timeouts are used.

Timing of the flow manager is also no longer relying on pthread condition
variables, as these generally cause waking up much quicker than the desired
timout. Instead a simple (u)sleep loop is used.

Both changes reduce the number of hash passes a lot.

Emergency behavior

In emergency mode there a number of changes to the workers. In this scenario
the flow memcap is fully used up and it is unavoidable that some flows won't
be tracked.

1. flow spare pool fetches are reduced to once a second. This avoids locking
   overhead, while the chance of success was very low.

2. getting an active flow directly from the hash skips flows that had very
   recent activity to avoid the scenario where all flows get only into the
   NEW state before getting reused. Rather allow some to have a chance of
   completing.

3. TCP packets that are not SYN packets will not get a used flow, unless
   stream.midstream is enabled. The goal here is again to avoid evicting
   active flows unnecessarily.

Better Localily

Flow Manager injects flows into the worker threads now, instead of one or
two packets. Advantage of this is that the worker threads can get packets
from their local packet pools, avoiding constant overhead of packets returning
to 'foreign' pools.

Counters

A lot of flow counters have been added and some have been renamed.

Overall the worker threads increment 'flow.wrk.*' counters, while the flow
manager increments 'flow.mgr.*'.

Additionally, none of the counters are snapshots anymore, they all increment
over time. The flow.memuse and flow.spare counters are exceptions.

Misc

FlowQueue has been split into a FlowQueuePrivate (unlocked) and FlowQueue.
Flow no longer has 'prev' pointers and used a unified 'next' pointer for
both hash and queue use.
5 years ago
Victor Julien e0aa7c1dbc unittests: check for flow memuse 5 years ago
Victor Julien f50c7b6d11 flow-manager: call other timeouts max once a second
Call Defrag and others only once per second. Flow Manager may wake
up (much) more often when flow engine is under resource pressure.
As this does not affect Defrag and others, it only unnecessarily
adds load.
5 years ago
Victor Julien 6814f08e93 flow-manager: only update FlowBucket::next_ts if it changed 5 years ago
Victor Julien 1d6d7f0858 flow: unref flow at end of flow worker 5 years ago
Victor Julien 2fd7c87f22 flow: do timeout checks before tuple compare 5 years ago
Victor Julien 48605f4867 flow: don't reorder list on lookup
Reduces cache misses.
5 years ago
Victor Julien 2a872ccb86 flow: timeout check on flow lookup 5 years ago
Victor Julien 7583a6c37c flow: simplify hash lookup logic
Remove double compare paths in favor of a single unified path.
5 years ago
Victor Julien 8b016cff4b flow: only move lastts forward
Pcaps with timestamps jumping around could confuse flow timeout
handling otherwise.
5 years ago
Victor Julien afd4a8012e flow/worker: check pkt src using DEBUG_VALIDATE_BUG_ON 5 years ago
Jeff Lucovsky be57f6d3cc rust/rdp: Fix use of incorrect buffer
This commit updates the connection confirmation parsing function to use
the correct buffer when determining the packet type.
5 years ago
Shivani Bhardwaj 2529eb35e1 smb: use uuid crate 5 years ago
Jason Ish e6668560e0 rust: only run cbindgen if needed
Only run cbindgen when necessary. This is a bit tricky. When
building a dist we want to unconditionally build the headers.

When going through a "make; sudo make install" type process,
cbindgen should not be run as the headers already exist, are
valid, and the environment under sudo is more often than
not suitable to pick up the Rust toolchains when installed
with rustup.

For the normal "make" case we have the gen/rust-bindings.h file
depend on library file, this will cause it to only be rebuilt
if the code was modified.

For "make dist" we unconditionally create "dist/rust-bindings.h".
This means the generated file could be in 2 locations, so update
configure.ac, and the library search find to find it.

The "gen/rust-bindings.h" should be picked up first if it exists,
for those who develop from a dist archive where "dist/rust-bindings.h"
also exists.

Not completely happy having the same file in 2 locations, but not
sure how else to get the dependency tracking correct.
5 years ago
Jason Ish fd5d8b78d0 alert/eve: add snmp metadata for rdp alerts
Redmine issue:
https://redmine.openinfosecfoundation.org/issues/3441
5 years ago
Jason Ish ef0ebc9550 alert/eve: add snmp metadata for snmp alerts
Redmine issue:
https://redmine.openinfosecfoundation.org/issues/3441
5 years ago
Shivani Bhardwaj 4c7f55e636 dcerpc: handle gap for TCP streams 5 years ago
Jason Ish 6cff558663 dns: add tailing data to split tcp unit test
Add trailing data to the complete payload to test the case
where data is consumed, but still incomplete.
5 years ago
Jason Ish cc154ce152 dns: fix subtraction overflow in debug message
The math for the return value wasn't updated in the debug log
to match the logic and could lead to a subtraction overflow
5 years ago
Philippe Antoine 80ccbff28a dns: return status from parsing tcp request 5 years ago
Philippe Antoine 1296740471 dns: fix incomplete results
and messages parsing whose length is >= 32768
5 years ago
James Dutrisac 8d5e54c046 pcap: recusively reading pcaps / documentation
Changes to doc/userguide/partials/options.rst for feature 2363
   (reading pcaps recursively)
5 years ago
James Dutrisac db5b73d9d6 pcap: read directories recursively
Describe Changes
- Added ability to recursively read pcap directories
- src/suricata.c: addition of new command line parameter
    --pcap-file-recursive
- src/source-pcap-file.c: parsing of the command line argument
- src/source-pcap-file-directory-helper.h: two thread vars tracking
    directory depth and should recurse
- src/util-error.c / src/util-error.h:
   Added new warning code "SC_WARN_PATH_READ_ERROR"
- Redmine ticket: https://redmine.openinfosecfoundation.org/issues/2363

Ticket: #2363
5 years ago
James Dutrisac 5a92d0a704 path: introduce path handling util funcs
This commit provides changes to util-path.c and util-path.h
to support the recursive reading of directories. It adds
4 functions.
- SCIsRegularFile to provide OS independent file info.
- SCIsRegularDirectory to provide OS independent directory info.
- SCRealPath is an OS independent wrapper for realpath.
- PathJoin to manage path resolution logic.
5 years ago
Philippe Antoine 3cfc1fcb07 stream/app-layer: break loop on proto change 5 years ago
Shivani Bhardwaj c9a637d854 datasets: fix null pointer deref 5 years ago
Shivani Bhardwaj a8f147d17a datasets: Init even in socket mode
Closes redmine ticket 3476.
5 years ago
Sascha Steinbiss 93eef1da84 detect-mqtt: unify error handling in rule parsing
This is meant to provide a single path to the error case.
This might help make things more clear for static
checkers.
5 years ago
Sascha Steinbiss 5dc21b0e09 detect-engine: initialize struct fields 5 years ago
Philippe Antoine 9a954e944d http: merge duplicated code
HtpRequestBodyHandlePUT and HtpRequestBodyHandlePOST
5 years ago
Philippe Antoine fe6950de08 http: use more precise parameter in HtpRequestBodySetupMultipart
So that we can see that it does not have to handle gaps
5 years ago
Philippe Antoine e13b319beb http: remove unused code
HtpRequestBodySetupPUT function
So that we can see that we do not use data=NULL in there
5 years ago
Philippe Antoine 28050c967f file: handles gaps natively
ie data=NULL and len>0 parameters
5 years ago
Philippe Antoine f0159b2fd2 util: PrintRawDataFp handles null
for gaps which are data=NULL and len>0
5 years ago
Philippe Antoine 83567e8732 files: rust closes files even on 0 length chunk 5 years ago
Philippe Antoine 6343920dfc applayer: allow rust parsers to have only one probe 5 years ago
Sascha Steinbiss 4e1a41a17d output-json: add MAC address output
This commit adds MAC address output to the EVE-JSON format. We follow the
remarks made in Redmine ticket #962: for packets, log MAC src/dst as a
scalar field in EVE; for flows, log MAC src/dst as lists in EVE. Field names
are different between flow and packet context to avoid type confusion
(src_mac vs. src_macs). Configuration approach and JSON representation is
taken from previous GitHub PR #2700.
5 years ago
Jeff Lucovsky c42574169e output/anomaly: Restrict anomaly logger count
This commit restricts the anomaly logger count. The restriction is
necessary due to state maintenance in the logger that doesn't scale
beyond a single logger.

Until that issue's solved, when multiple anomaly loggers are configured,
an error message will be emitted to highlight the restriction.
5 years ago
Philippe Antoine 61c327dd80 signature: checks for integer overflow in limits propagation 5 years ago
Victor Julien 5db1d9b841 eve/yaml: move mqtt down 5 years ago
Victor Julien 1c748f394b fuzz/pcap: enable MQTT parser 5 years ago
Shivani Bhardwaj e9fe5ada7f datasets: reload static sets 5 years ago
Sascha Steinbiss c31360070b rust/mqtt: add MQTT parser 5 years ago
Philippe Antoine a5572890a9 detect: adds engine for u8 keywords 5 years ago
Jeff Lucovsky a5d30a3220 doc/output: Document multithreaded eve option 5 years ago
Jeff Lucovsky 30ae98f658 output/json: Multi-threaded EVE logging support
This commit modifies the JSON loggers with changes necessary to support
multi-threaded EVE output.

Each "thread-init" function sets up the per-thread log file context for
subsequent calls to the JSON output to buffer function.
5 years ago
Jeff Lucovsky aa20770277 log: Support multi-threaded eve output. 5 years ago
Jeff Lucovsky 15b4554ab3 output: Check for fwrite_unlocked
This commit creates a macro for fwrite_unlocked which is probed during
configuration time.
5 years ago
Jeff Lucovsky 3d0f353ee5 output: Correct typos 5 years ago