Commit Graph
196 Commits
Author SHA1 Message Date
Ivan Nardi 91bb77a880 A final(?) effort to reduce memory usage per flow (#1389)
Remove some unused fields and re-organize other ones.
In particular:
* Update the parameters of `ndpi_ssl_version2str()` function
* Zattoo, Thunder: these timestamps aren't really used.
* Ftp/mail: these protocols are dissected only over TCP.
* Attention must be paid to TLS.Bittorrent flows to avoid invalid
read/write to `flow->protos.bittorrent.hash` field.

This is the last(?) commit of a long series (see 22241a1d, 227e586e,
730c2360, a8ffcd8b) aiming to reduce library memory consumption.

Before, at nDPI 4.0 (more precisly, at a6b10cf7, because memory stats
were wrong until that commit):
```
nDPI Memory statistics:
	nDPI Memory (once):      221.15 KB
	Flow Memory (per flow):  2.94 KB
```
Now:
```
nDPI Memory statistics:
	nDPI Memory (once):      231.71 KB
	Flow Memory (per flow):  1008 B       <---------
```
i.e. memory usage per flow has been reduced by 66%, dropping below the
psychological threshold of 1 KB.

To further reduce this value, we probably need to look into #1279:
let's fight this battle another day.
2021-12-22 19:54:06 +01:00
Ivan Nardi 7153b8933c Improve/add several protocols (#1383)
Improve Microsoft, GMail, Likee, Whatsapp, DisneyPlus and Tiktok
detection.
Add Vimeo, Fuze, Alibaba and Firebase Crashlytics detection.

Try to differentiate between Messenger/Signal standard flows (i.e chat)
and their VOIP (video)calls (like we already do for Whatsapp and
Snapchat).

Add a partial list of some ADS/Tracking stuff.

Fix Cassandra, Radius and GTP false positives.
Fix DNS, Syslog and SIP false negatives.

Improve GTP (sub)classification: differentiate among GTP-U, GTP_C and
GTP_PRIME.

Fix 3 LGTM warnings.
2021-12-18 13:24:51 +01:00
Ivan Nardi 58b33dcb2d TLS: add support for IPV6 in Subject Alt Names field (#1385) 2021-12-18 12:03:46 +01:00
Ivan Nardi ec6eea0c23 TLS: fix usage of certificate cache (#1384) 2021-12-18 12:03:32 +01:00
Luca Deri 85b396ca43 Fixed issue that prevented alt certificate names to be fully detected when ipAddress and rfc822Name were specified in certificates 2021-12-07 14:45:42 +01:00
Ivan Nardi a8ffcd8bb0 Rework how hostname/SNI info is saved (#1330)
Looking at `struct ndpi_flow_struct` the two bigger fields are
`host_server_name[240]` (mainly for HTTP hostnames and DNS domains) and
`protos.tls_quic.client_requested_server_name[256]`
(for TLS/QUIC SNIs).

This commit aims to reduce `struct ndpi_flow_struct` size, according to
two simple observations:
 1) maximum one of these two fields is used for each flow. So it seems safe
to merge them;
 2) even if hostnames/SNIs might be very long, in practice they are rarely
longer than a fews tens of bytes. So, using a (single) large buffer is a
waste of memory for all kinds of flows. If we need to truncate the name,
we keep the *last* characters, easing domain matching.

Analyzing some real traffic, it seems safe to assume that the vast
majority of hostnames/SNIs is shorter than 80 bytes.

Hostnames/SNIs are always converted to lowercase.

Attention was given so as to be sure that unit-tests outputs are not
affected by this change.

Because of a bug, TLS/QUIC SNI were always truncated to 64 bytes (the
*first* 64 ones): as a consequence, there were some "Suspicious DGA
domain name" and "TLS Certificate Mismatch" false positives.
2021-11-24 10:46:48 +01:00
Ivan Nardi afc2b641eb Fix writes to flow->protos union fields (#1354)
We can write to `flow->protos` only after a proper classification.

This issue has been found in Kerberos, DHCP, HTTP, STUN, IMO, FTP,
SMTP, IMAP and POP code.
There are two kinds of fixes:
 * write to `flow->protos` only if a final protocol has been detected
 * move protocol state out of `flow->protos`
The hard part is to find, for each protocol, the right tradeoff between
memory usage and code complexity.

Handle Kerberos like DNS: if we find a request, we set the protocol
and an extra callback to further parsing the reply.

For all the other protocols, move the state out of `flow->protos`. This
is an issue only for the FTP/MAIL stuff.

Add DHCP Class Identification value to the output of ndpiReader and to
the Jason serialization.

Extend code coverage of fuzz tests.

Close #1343
Close #1342
2021-11-15 16:20:57 +01:00
Ivan Nardi 55880e4ae4 TLS: fix two warnings (#1365)
Disable unit tests on CI for big-endian target. We know we have multiple
issues on big-endian architectures (see #1312) and so the unit tests
always fail there. Ignore this error for the time being and let the CI
pass if we don't have other issues.

Remove an unused automa definition
2021-11-02 22:08:15 +01:00
Luca Deri 1f6c94e663 Updated test results 2021-11-02 12:08:20 +01:00
Luca Deri b11065b2e3 Fixed certificate mismatch for TLS flows with no client hello observed 2021-11-02 12:06:18 +01:00
Luca Deri a9aa46287d Little change in TLS debug traces 2021-10-28 10:16:33 +02:00
Luca Deri 5283755afa Fixed TLS certificate dissection of subjectAltName 2021-10-27 22:51:59 +02:00
Luca Deri aaa474a37d Minor code cleanup 2021-10-27 09:32:11 +02:00
Ivan Nardi 6edb7bedd7 Avoid overwriting valid protocol in ndpi_detection_giveup (#1360)
We should avoid updating any valid protocol in `ndpi_detection_giveup`; we
should try to find a proper classification only if the flow is still
completely unclassified.

For example in the attached pcap there is a valid TLS session, recognized
as such by TLS dissector. However, the `ndpi_detection_giveup`function
updates it to "HTTP/TLS" (!?) simply because the server port is 80.

Note that the real issue is not the wrong classification, but the
wrong access to `flow->protos` union. If we already set some fields of
`flow->protos` and we change the protocol in `ndpi_detection_giveup`, we
might end up freeing some invalid pointers in `ndpi_free_flow_data`
(no wonder this issue has been found while fuzzing #1354)

Fix GIT and TLS dissectors (issues found by CI fuzzer)
2021-10-27 09:23:07 +02:00
Toni 41765efcf8 Detect invalid characters in text and set a risk. Fixes #1347. (#1363)
Signed-off-by: Toni Uhlig <matzeton@googlemail.com>
2021-10-26 21:34:01 +02:00
Ivan Nardi 333a6d60e8 TLS: fix a heap-buffer-overflow (#1356)
Revert of c3d1c697
Error reproducible with the attached pcap and valgrind
2021-10-22 14:57:49 +02:00
Luca Deri c3d1c69722 Fixed heap-buffer-overflow in TLS dissector 2021-10-20 22:07:47 +02:00
Toni ed51987e3a Fix broken fuzz_process_packet fuzzer by adding a call to ndpi_finalize_initialization(). (#1334)
* fixed several memory errors (heap-overflow, unitialized memory, etc)
 * ability to build fuzz_process_packet with a main()
   allowing to replay crash data generated with fuzz_process_packet
   by LLVMs libfuzzer
 * temporarily disable fuzzing if `tests/do.sh`
   executed with env FUZZY_TESTING_ENABLED=1

Signed-off-by: Toni Uhlig <matzeton@googlemail.com>
2021-10-18 23:16:32 +02:00
Ivan Nardi 2dfc478ad4 Fix compilation with clang-13 or if some debug macros are enabled (#1326) 2021-10-06 18:32:15 +02:00
Luca Deri 10576eab0a TLS obsolete protocol is set when TLS < 1.2 (used to be 1.1) 2021-10-05 16:44:56 +02:00
Ivan Nardi 730c2360bd Remove struct ndpi_packet_struct from struct ndpi_flow_struct (#1319)
There are no real reasons to embed `struct ndpi_packet_struct` (i.e. "packet")
in `struct ndpi_flow_struct` (i.e. "flow"). In other words, we can avoid
saving dissection information of "current packet" into the "flow" state,
i.e. in the flow management table.

The nDPI detection module processes only one packet at the time, so it is
safe to save packet dissection information in `struct ndpi_detection_module_struct`,
reusing always the same "packet" instance and saving a huge amount of memory.
Bottom line: we need only one copy of "packet" (for detection module),
not one for each "flow".

It is not clear how/why "packet" ended up in "flow" in the first place.
It has been there since the beginning of the GIT history, but in the original
OpenDPI code `struct ipoque_packet_struct` was embedded in
`struct ipoque_detection_module_struct`, i.e. there was the same exact
situation this commit wants to achieve.

Most of the changes in this PR are some boilerplate to update something
like "flow->packet" into something like "module->packet" throughout the code.
Some attention has been paid to update `ndpi_init_packet()` since we need
to reset some "packet" fields before starting to process another packet.

There has been one important change, though, in ndpi_detection_giveup().
Nothing changed for the applications/users, but this function can't access
"packet" anymore.
The reason is that this function can be called "asynchronously" with respect
to the data processing, i.e in context where there is no valid notion of
"current packet"; for example ndpiReader calls it after having processed all
the traffic, iterating the entire session table.

Mining LRU stuff seems a bit odd (even before this patch): probably we need
to rethink it, as a follow-up.
2021-10-05 15:49:52 +02:00
Ivan Nardi 6325aebda6 TLS: avoid zeroing large structures (#1300)
Zeroing large structures (i.e. size > KB) is quite costly (from a CPU point
of view): we can safely avoid doing that for a couple of big structures.

Standard and Valgrind tests have been diverging quite a lot: it is time
to re-sync them. Use the same script and enable Valgrind via an
enviroment variable:

  NDPI_TESTS_VALGRIND=1 ./tests/do.sh
2021-09-16 14:34:59 +02:00
Ivan Nardi 0eb7a0388c Hangout: fix usage of STUN function (#1299) 2021-09-13 18:25:43 +02:00
Ivan Nardi 9e0f0ce3df Fix access to some TLS fields in flow structure (#1277)
Fields 'tls.hello_processed` and `tls.subprotocol_detected` are used by
QUIC (i.e UDP...), too.
2021-08-20 18:11:37 +02:00
Ivan Nardi 8fdffbf3a1 Compile everything with "-W -Wall -Wno-unused-parameter" flags (#1276)
Fix all the warnings.

Getting rid of "-Wno-unused-parameter" is quite complex because some
parameters usage depends on compilation variable (i.e.
`--enable-debug-messages`).

The "-Werror" flag has been added only in Travis builds to avoid
breaking the builds to users using uncommon/untested
OS/compiler/enviroment.

Tested on:
* x86_64; Ubuntu 20.04; gcc 7,8,9,10,11; clang 7,8,9,10,11,12
* x86_64; CentOS 7.7; gcc 4.8.5 (with "--disable-gcrypt" flag)
* Raspberry 4; Debian 10.10; gcc 8.3.0
2021-08-20 18:11:13 +02:00
Luca Deri 821f4c9249 Fixed some invalid TLS guesses 2021-08-17 19:04:07 +02:00
Luca Deri 4183718952 Added TLS fatal alert flow risk 2021-08-07 19:40:44 +02:00
Luca Deri 51e4888442 Implemented ALPN automa for checking uncommon ALPNs 2021-07-24 17:50:32 +02:00
Luca Deri 61fc5be202 Reworked flow risk implementation 2021-07-23 17:27:15 +02:00
Toni 6ad0d6666c Implemented function to retrieve flow information. #1253 (#1254)
* fixed [h]euristic typo

Signed-off-by: Toni Uhlig <matzeton@googlemail.com>
2021-07-23 10:37:20 +02:00
Toni 32275543c4 Added risk: TLS_EXTENSION_SUSPICIOUS (#1252)
* validates client/server hello TLS extensions
 * inspects content for some extensions

Signed-off-by: Toni Uhlig <matzeton@googlemail.com>
2021-07-19 16:23:24 +02:00
pacantandpacant 488166833e Fixed TLS certificate threshold (#1248)
Co-authored-by: pacant <a.pace97@outlook.com>
2021-07-16 13:52:55 +02:00
Luca Deri b95bd0358f Code cleanup 2021-07-14 11:19:24 +02:00
pacantandpacant 19a29e1e22 TLS Risks - Certificate Validity Too Long (#1239)
* Added flow risk: TLS certificate too long

* Added flow risk: TLS certificate too long

* Date for TLS limit added

* TLS certificate check fixed

Co-authored-by: pacant <a.pace97@outlook.com>
2021-07-14 11:13:22 +02:00
Ivan Nardi 46c32d2433 TLS: fix a memory error in JA3 code (#1227)
protocols/tls.c:1856:5: runtime error: index 256 out of bounds for type 'char [256]'
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior protocols/tls.c:1856:5
2021-07-05 09:42:52 +02:00
Toni ced6fca184 Fixed off-by-one memory error for TLS-JA3. (#1222)
Signed-off-by: Toni Uhlig <matzeton@googlemail.com>
2021-06-29 15:33:12 +02:00
Luca Deri 0cbb4b2ffa Restored 32 bit value in ndpi_match_string_value() 2021-06-21 17:06:22 +02:00
Vitaly Lavrov be808c30f3 Code review. (#1205)
The common actions required to call the ac_automata_search() function
have been moved to the ndpi_match_string_common function. This made it
possible to simplify the ndpi_match_string, ndpi_match_string_protocol_id,
ndpi_match_string_value, ndpi_match_custom_category, ndpi_match_string_subprotocol,
ndpi_match_bigram, ndpi_match_trigram functions.

Using u_int16_t type for protocol identifiers when working with the
ahocorasick library (changes src/include/ndpi_api.h.in and src/include/ndpi_typedefs.h).

Reworked "finalization" of all AC_AUTOMATA_t structures.

Changing the order of fields in the ndpi_call_function_struct structure
reduces the size of the ndpi_detection_module_struct structure by 10 kB (for x86_64).
2021-06-15 11:28:09 +02:00
Ivan Nardi f4dbeecba9 TLS: improve check for common ALPNs (#1191)
Facebook is still using its own ALPN for HTTP2 as well
2021-06-01 16:50:28 +02:00
Luca eb0115d4e0 Added TLS check to avoid crash with UDP-based traffic 2021-05-25 09:03:24 +02:00
Luca Deri d485c1cecb Improved TLS browser detection heuristics 2021-05-19 21:28:33 +02:00
Luca Deri ca15e3295e Added risk/score dump (ndpiReader -h)
Added ndpi_dump_risks_score() API score
2021-05-18 19:34:17 +02:00
Luca Deri 1ec621c85b Added further checks 2021-05-15 19:53:46 +02:00
Luca Deri abd6bce6f9 Added TLS certifiacate caching
Added Fortigate protocol
2021-05-15 10:52:16 +02:00
Luca Deri ac1eaca8a6 Added browser TLS heuristic 2021-05-13 20:00:27 +02:00
Luca Deri a62be9b8ec Implemented heuristic to detect Safari and Firefox TLS browsing 2021-05-13 12:37:07 +02:00
Toni 5918a6542d Improved SSL certificate name wildcard handling and risk. #1182 (#1183)
Signed-off-by: Toni Uhlig <matzeton@googlemail.com>
2021-05-11 21:38:26 +02:00
Ivan Nardi 2ca37ff8be TLS: fix extraction for TLS signature algorithms (#1180)
```
==69562==ERROR: AddressSanitizer: heap-buffer-overflow on address 0x6100009000fb at pc 0x7f41882003a7 bp 0x7f4183cfbfc0 sp 0x7f4183cfb768
READ of size 32 at 0x6100009000fb thread T1
    #0 0x7f41882003a6 in __interceptor_memcpy ../../../../src/libsanitizer/sanitizer_common/sanitizer_common_interceptors.inc:827
    #1 0x560b2d7462a1 in processClientServerHello protocols/tls.c:1647
    #2 0x560b2d73be6a in processTLSBlock protocols/tls.c:712
    #3 0x560b2d73e61f in ndpi_search_tls_udp protocols/tls.c:968
```
2021-05-09 15:10:41 +02:00
Ivan Nardi 12c6ab768d TLS: fix another use-of-uninitialized-value error in ClientHello parsing (#1179)
Error detected with valgrind.

==13127== Conditional jump or move depends on uninitialised value(s)
==13127==    at 0x483EF58: strlen (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==13127==    by 0x1A93B6: ndpi_strdup (ndpi_main.c:159)
==13127==    by 0x1C07CC: processClientServerHello (tls.c:1678)
==13127==    by 0x1C0C4C: processTLSBlock (tls.c:712)
==13127==    by 0x1C0C4C: ndpi_search_tls_tcp.part.0 (tls.c:849)

See also 8c3674e9
2021-05-09 15:10:14 +02:00
Luca ae2470fad4 Initial work towards detection via TLS of browser types 2021-05-06 21:42:06 +02:00