| Commit message (Collapse) | Author | Age |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixed errors for bigendian platforms in ndpiReader.
All address and port comparisons and hash calculations are done with
endian in mind.
The get_ndpi_flow_info() function searched for an existing flow for the
forward and reverse direction of the packet.
The ndpi_workflow_node_cmp() function looked for a flow regardless of
the packet's direction. This is what led to an error in determining the
direction of transmission of the packet.
Fixed error in "synscan" test: the number of packets in the forward and
reverse direction is incorrectly defined (verified via tcpdump).
Fixed bug with icmp protocol checksum check for big endian platforms.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Add few scripts to easily update some IPs lists
Some IPs lists should be updated frequently: try to easy the process.
The basic idea is taken from d59fefd0 and a8fe74e5 (for Azure
addresses): one specific .c.inc file and one script for each protocol.
Add the possibility to don't load a specific list.
Rename the old NDPI_PROTOCOL_HOTMAIL id to NDPI_PROTOCOL_MS_OUTLOOK,
to identify Hotmail/Outlook/Exchange flows.
TODO: ipv6
Remove the 9 addresses associated to BitTorrent: they have been added in
e2f21116 but it is not clear why all the traffic to/from these ips
should be classified as BitTorrent.
* Added quotes
* Added quotes
Co-authored-by: Luca Deri <lucaderi@users.noreply.github.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We should have two protocols in classification results only when the
"master" protocol allows some sub-protocols.
Classifications like `AmazonAWS`, `TLS/AmazonAWS`, `DNS/AmazonAWS` are
fine. However classifications like `NTP/Apple`, `BitTorrent/Azure`,
`DNScrypt.AmazonAWS` or `NestLogSink.Google` are misleading.
For example, `ndpiReader`shows `BitTorrent/Azure` flows under `Azure`
statistics; that seems to be wrong or, at least, very misleading.
This is quite important since we have lots of addresses from CDN
operators.
The only drawback of this solution is that right now ICMP traffic is
classified simply as `ICMP`; if we are really interested in ICMP stuff
we can restore the old behaviour later.
|
| |
|
| |
Fix: 7a3aa41a
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
As a general rule, the higher the confidence value, the higher the
"reliability/precision" of the classification.
In other words, this new field provides an hint about "how" the flow
classification has been obtained.
For example, the application may want to ignore classification "by-port"
(they are not real DPI classifications, after all) or give a second
glance at flows classified via LRU caches (because of false positives).
Setting only one value for the confidence field is a bit tricky: more
work is probably needed in the next future to tweak/fix/improve the logic.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Signed-off-by: Toni Uhlig <matzeton@googlemail.com>
Co-authored-by: Luca Deri <lucaderi@users.noreply.github.com>
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
The goal is to have a (roughly) idea about how many packets nDPI needs
to properly classify a flow.
Log this information (and guessed flows number too) during unit tests,
to keep track of improvements/regressions across commits.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
Before
Accuracy 66%, Precision 86%, Recall 38%
After
Accuracy 71%, Precision 89%, Recall 49%
|
| |
|
|
|
| |
Reworked Tor dissector embedded in TLS (fixes #1141)
Removed false positive on HTTP User-Agent
|
| | |
|
| | |
|
|
|
* QUIC: fix return value on error path on quic_cipher_init()
* QUIC: allow dissection of sessions forcing version negotiation
Enhance heuristic to avoid false positives.
|