For some unclear reasons, fuzzers using pl7m create huge corpus,
triggering OOM in oss-fuzz runs (where the memory RSS limit is set to
2560Mb). Example:
```
==25340== ERROR: libFuzzer: out-of-memory (used: 2564Mb; limit: 2560Mb)
To change the out-of-memory limit use -rss_limit_mb=<N>
Live Heap Allocations: 2364004039 bytes in 133791 chunks; quarantined: 60662293 bytes in 3664 chunks; 176432 other chunks; total chunks: 313887; showing top 95% (at most 8 unique contexts)
1285841683 byte(s) (54%) in 2956 allocation(s)
#0 0x56f814ef4bde in __interceptor_malloc /src/llvm-project/compiler-rt/lib/asan/asan_malloc_linux.cpp:69:3
#1 0x56f814e04416 in operator new(unsigned long) cxa_noexception.cpp:0
#2 0x56f814de6b2d in assign<unsigned char *, 0> /work/llvm-stage2/runtimes/runtimes-bins/compiler-rt/lib/fuzzer/libcxx_fuzzer_x86_64/include/c++/v1/vector:1443:3
#3 0x56f814de6b2d in operator= /work/llvm-stage2/runtimes/runtimes-bins/compiler-rt/lib/fuzzer/libcxx_fuzzer_x86_64/include/c++/v1/vector:1412:9
#4 0x56f814de6b2d in fuzzer::InputCorpus::AddToCorpus(std::__Fuzzer::vector<unsigned char, std::__Fuzzer::allocator<unsigned char>> const&, unsigned long, bool, bool, bool, std::__Fuzzer::chrono::duration<long long, std::__Fuzzer::ratio<1l, 1000000l>>, std::__Fuzzer::vector<unsigned int, std::__Fuzzer::allocator<unsigned int>> const&, fuzzer::DataFlowTrace const&, fuzzer::InputInfo const*) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerCorpus.h:221:10
#5 0x56f814de60e5 in fuzzer::Fuzzer::RunOne(unsigned char const*, unsigned long, bool, fuzzer::InputInfo*, bool, bool*) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:539:16
#6 0x56f814de7df2 in fuzzer::Fuzzer::ReadAndExecuteSeedCorpora(std::__Fuzzer::vector<fuzzer::SizedFile, std::__Fuzzer::allocator<fuzzer::SizedFile>>&) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:829:7
#7 0x56f814de8127 in fuzzer::Fuzzer::Loop(std::__Fuzzer::vector<fuzzer::SizedFile, std::__Fuzzer::allocator<fuzzer::SizedFile>>&) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:867:3
#8 0x56f814dd6736 in fuzzer::FuzzerDriver(int*, char***, int (*)(unsigned char const*, unsigned long)) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerDriver.cpp:914:6
#9 0x56f814e02c62 in main /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerMain.cpp:20:10
#10 0x7fa11e2c3082 in __libc_start_main /build/glibc-SzIz7B/glibc-2.31/csu/libc-start.c:308:16
1031350683 byte(s) (43%) in 2468 allocation(s)
#0 0x56f814ef4bde in __interceptor_malloc /src/llvm-project/compiler-rt/lib/asan/asan_malloc_linux.cpp:69:3
#1 0x56f814e04416 in operator new(unsigned long) cxa_noexception.cpp:0
#2 0x56f814de6b2d in assign<unsigned char *, 0> /work/llvm-stage2/runtimes/runtimes-bins/compiler-rt/lib/fuzzer/libcxx_fuzzer_x86_64/include/c++/v1/vector:1443:3
#3 0x56f814de6b2d in operator= /work/llvm-stage2/runtimes/runtimes-bins/compiler-rt/lib/fuzzer/libcxx_fuzzer_x86_64/include/c++/v1/vector:1412:9
#4 0x56f814de6b2d in fuzzer::InputCorpus::AddToCorpus(std::__Fuzzer::vector<unsigned char, std::__Fuzzer::allocator<unsigned char>> const&, unsigned long, bool, bool, bool, std::__Fuzzer::chrono::duration<long long, std::__Fuzzer::ratio<1l, 1000000l>>, std::__Fuzzer::vector<unsigned int, std::__Fuzzer::allocator<unsigned int>> const&, fuzzer::DataFlowTrace const&, fuzzer::InputInfo const*) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerCorpus.h:221:10
#5 0x56f814de60e5 in fuzzer::Fuzzer::RunOne(unsigned char const*, unsigned long, bool, fuzzer::InputInfo*, bool, bool*) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:539:16
#6 0x56f814de7635 in fuzzer::Fuzzer::MutateAndTestOne() /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:760:19
#7 0x56f814de8425 in fuzzer::Fuzzer::Loop(std::__Fuzzer::vector<fuzzer::SizedFile, std::__Fuzzer::allocator<fuzzer::SizedFile>>&) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:905:5
#8 0x56f814dd6736 in fuzzer::FuzzerDriver(int*, char***, int (*)(unsigned char const*, unsigned long)) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerDriver.cpp:914:6
#9 0x56f814e02c62 in main /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerMain.cpp:20:10
#10 0x7fa11e2c3082 in __libc_start_main /build/glibc-SzIz7B/glibc-2.31/csu/libc-start.c:308:16
```
See: https://oss-fuzz.com/testcase-detail/4717811415449600
See: https://oss-fuzz.com/testcase-detail/6164130982068224
Let's *try* the following workaround: set the parameter `-max-len` to
512K, to force the engine to not genereate inputs (i.e. pcap files...)
larger than 512K. Right now the value used is 1MB, i.e the default,
because we have file larger than 1MB in the initial seeds (i.e.
`/tests/pcaps/*`).
Let's hope than having smaller files lead to smaller corpus...
Update pl7m code (fix a Use-of-uninitialized-value error)
Pl7m is a custom mutator (used for structure aware fuzzing) for network
traffic packet captures (i.e. pcap files).
The output of the mutator is always a valid pcap file, containing the
same flows/sessions of the input file. That's it: the mutator only
changes the packet payload after the TCP/UDP header, keeping all the
original L2/L3 information (IP addresses and L4 ports).
See: https://github.com/IvanNardi/pl7m
```
AddressSanitizer:DEADLYSIGNAL
=================================================================
==29508==ERROR: AddressSanitizer: SEGV on unknown address 0x50710145d51d (pc 0x55cb788f25fe bp 0x7ffcfefa15f0 sp 0x7ffcfefa1240 T0)
==29508==The signal is caused by a READ memory access.
#0 0x55cb788f25fe in ndpi_search_zoom /home/ivan/svnrepos/nDPI/src/lib/protocols/zoom.c:210:24
#1 0x55cb787e9418 in check_ndpi_detection_func /home/ivan/svnrepos/nDPI/src/lib/ndpi_main.c:7174:6
#2 0x55cb7883f753 in check_ndpi_udp_flow_func /home/ivan/svnrepos/nDPI/src/lib/ndpi_main.c:7209:10
#3 0x55cb7883bc9d in ndpi_check_flow_func /home/ivan/svnrepos/nDPI/src/lib/ndpi_main.c:7240:12
```
Found by oss-fuzzer
See: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=69520
After a flow has been classified as RTP or RTCP, nDPI might analyse more
packets to look for STUN/DTLS packets, i.e. to try to tell if this flow
is a "pure" RTP/RTCP flow or if the RTP/RTCP packets are multiplexed with
STUN/DTLS.
Useful for proper (sub)classification when the beginning of the flows
are not captured or if there are lost packets in the the captured traffic.
Disabled by default
RFC4571 is not the only way to wrap RTP messages in TCP streams.
For example, when RTP is encapsulated over TURN flows (i.e. via DATA
attribute) there is no additional framing.
See also 6127e0490
TODO: enable parallel tests when using docker with no-x86_64 archs.
When I tried the obviuos solutions:
```
NDPI_FORCE_PARALLEL_UTESTS=1 NDPI_SKIP_PARALLEL_BAR=1 make check VERBOSE=1
```
I got:
```
Run configuration "caches_cfg" [--cfg=lru.ookla.size,0 --cfg=lru.msteams.ttl,1]
ookla.pcap /bin/sh: 1: run_single_pcap: not found
teams.pcap /bin/sh: 1: run_single_pcap: not found
Run configuration "caches_global" [--cfg=lru.ookla.scope,1 --cfg=lru.bittorrent.scope,1 --cfg=lru.stun.scope,1 --cfg=lru.tls_cert.scope,1 --cfg=lru.mining.scope,1 --cfg=lru.msteams.scope,1 --cfg=lru.stun_zoom.scope,1]
bittorrent.pcap /bin/sh: 1: run_single_pcap: not found
lru_ipv6_caches.pcapng /bin/sh: 1: run_single_pcap: not found
mining.pcapng /bin/sh: 1: run_single_pcap: not found
...
```
Running unit tests is quite a bottleneck while developing or while
waiting for GitHub CI results...
Try to run the tests in parallel, using the `parallel` tool.
By default, tests still run one after the other, as usual; to enable
parallel execution you need `NDPI_FORCE_PARALLEL_UTESTS=1 ./tests/do.sh`
Please note that the output is quite different in parallel mode!
A big part of the script has been rewritten to avoid code dupication
between "serial" and "parallel" path
On my notebook:
```
ivan@ivan-Latitude-E6540:~/svnrepos/nDPI(parallel)$ time ./tests/do.sh
[...]
real 3m12,684s
[...]
ivan@ivan-Latitude-E6540:~/svnrepos/nDPI(parallel)$ time NDPI_FORCE_PARALLEL_UTESTS=1 ./tests/do.sh
[...]
real 0m58,463s
```
IRC has its best times well behind, but there are still some servers
using it.
We should try to simplify the detection logic, still based on OpenDPI
logic.
Let's start with some easy changes:
* try to detect TLS connection via standard hostname/SNI matching,
removing an old heuristic (we have never had any trace matching it);
* add some basic server names;
* once we detect that the flow is IRC, we don't have to perform
anything else;
* remove HTTP stuff; real HTTP flows never trigger that data path
* use `ndpi_memmem()` when possible
There is some overlap between RTP and Raknet detection: give precedence
to RTP logic.
Consequences:
* Raknet might require a little bit more packets for some flows (not a
big issue)
* some very small (1-2 pkts) Raknet flows are not classified (not sure
what do do about that..)
```
SCARINESS: 12 (1-byte-read-heap-buffer-overflow)
#0 0x557f3a5b5100 in ndpi_get_host_domain /src/ndpi/src/lib/ndpi_domains.c:158:8
#1 0x557f3a59b561 in ndpi_check_dga_name /src/ndpi/src/lib/ndpi_main.c:10412:17
#2 0x557f3a51163a in process_chlo /src/ndpi/src/lib/protocols/quic.c:1467:7
#3 0x557f3a469f4b in LLVMFuzzerTestOneInput /src/ndpi/fuzz/fuzz_quic_get_crypto_data.c:44:7
#4 0x557f3a46abc8 in NaloFuzzerTestOneInput (/out/fuzz_quic_get_crypto_data+0x4cfbc8)
```
Some notes about the leak: if the insertion into the uthash fails (because of an
allocation failure), we need to free the just allocated entry. But the only
way to check if the `HASH_ADD_*` failed, is to perform a new lookup: a bit
costly, but we don't use that code in the fast-path.
See also efb261a95c
Credits for finding the issues to Philippe Antoine (@catenacyber) and his
`nallocfuzz` fuzzing engine
See: https://github.com/catenacyber/nallocfuzz
See: https://github.com/google/oss-fuzz/pull/9902
This cache was added in b6b4967aa, when there was no real Zoom support.
With 63f349319, a proper identification of multimedia stream has been
added, making this cache quite useless: any improvements on Zoom
classification should be properly done in Zoom dissector.
Tested for some months with a few 10Gbits links of residential traffic: the
cache pretty much never returned a valid hit.
Deciding when a session starts and ends is responsability of the
applicationi (via its flow manager)i, not of the library.
BTW, the removed code is incomplete at beast
Avoid code duplication between these two protocols.
We remove support for RTCP over TCP; it is quite rare to find this kind
of traffic and, more important, we have never had support for RTP
over TCP: we should try to add both detecion as follow-up.
Fix a message log in the LINE code
The new values has been checked against the ones reported by Wireshark.
Found while fixing a Use-of-uninitialized-value error reported by
oss-fuzz
```
==7582==WARNING: MemorySanitizer: use-of-uninitialized-value
#0 0x5a6549abc368 in ndpi_compute_ja4 ndpi/src/lib/protocols/tls.c:1762:10
#1 0x5a6549ab88a0 in processClientServerHello ndpi/src/lib/protocols/tls.c:2863:10
#2 0x5a6549ac1452 in processTLSBlock ndpi/src/lib/protocols/tls.c:909:5
#3 0x5a6549abf588 in ndpi_search_tls_tcp ndpi/src/lib/protocols/tls.c:1098:2
#4 0x5a65499c53ec in check_ndpi_detection_func ndpi/src/lib/ndpi_main.c:7215:6
```
See: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=68449&q=ndpi&can=1&sort=-id
eDonkey is definitely not as used as >10 years ago, but it seems it is
still active.
While having a basic TCP support seems easy, identification over UDP doesn't
work and it is hard to do it rightly (packets might be only 2 bytes long):
remove it.
Credits to V.G <v.gavrilov@securitycode.ru>
```
==22779==ERROR: AddressSanitizer: stack-buffer-overflow on address 0x7f0900701020 at pc 0x555bcd2a6f02 bp 0x7ffe3ba5e790 sp 0x7ffe3ba5e788
READ of size 1 at 0x7f0900701020 thread T0
#0 0x555bcd2a6f01 in shoco_decompress /home/ivan/svnrepos/nDPI/src/lib/third_party/src/shoco.c:184:26
#1 0x555bcd2a4018 in LLVMFuzzerTestOneInput /home/ivan/svnrepos/nDPI/fuzz/fuzz_alg_shoco.cpp:18:3
#2 0x555bcd1aa816 in fuzzer::Fuzzer::ExecuteCallback(unsigned char const*, unsigned long) (/home/ivan/svnrepos/nDPI/fuzz/fuzz_alg_shoco+0x4f816) (BuildId: c54d1c32163c9937e06f62127348ae6bd26d9309)
#3 0x555bcd193be8 in fuzzer::RunOneTest(fuzzer::Fuzzer*, char const*, unsigned long) (/home/ivan/svnrepos/nDPI/fuzz/fuzz_alg_shoco+0x38be8) (BuildId: c54d1c32163c9937e06f62127348ae6bd26d9309)
#4 0x555bcd1996fa in fuzzer::FuzzerDriver(int*, char***, int (*)(unsigned char const*, unsigned long)) (/home/ivan/svnrepos/nDPI/fuzz/fuzz_alg_shoco+0x3e6fa) (BuildId: c54d1c32163c9937e06f62127348ae6bd26d9309)
#5 0x555bcd1c3c92 in main (/home/ivan/svnrepos/nDPI/fuzz/fuzz_alg_shoco+0x68c92) (BuildId: c54d1c32163c9937e06f62127348ae6bd26d9309)
#6 0x7f090257a082 in __libc_start_main /build/glibc-e2p3jK/glibc-2.31/csu/../csu/libc-start.c:308:16
#7 0x555bcd18e96d in _start (/home/ivan/svnrepos/nDPI/fuzz/fuzz_alg_shoco+0x3396d) (BuildId: c54d1c32163c9937e06f62127348ae6bd26d9309)
Address 0x7f0900701020 is located in stack of thread T0 at offset 4128 in frame
#0 0x555bcd2a3d97 in LLVMFuzzerTestOneInput /home/ivan/svnrepos/nDPI/fuzz/fuzz_alg_shoco.cpp:5
This frame has 4 object(s):
[32, 4128) 'out' (line 9) <== Memory access at offset 4128 overflows this variable
[4256, 8352) 'orig' (line 9)
```
Found by oss-fuzzer.
See: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=68211
Microsoft UDP traffic over port ~3478 is voip traffic, using some kind
of proprietary STUN-like protocol: so use the most specific protocol id.
More important, we definitely want `Stun/Skype_TeamsCall` and not
`Stun/Skype_Teams`
Enable parsing of Mapped-Address attribute for all STUN flows: that
means that STUN classification might require more packets.
Add a configuration knob to enable/disable this feature.
Note that we can have (any) STUN metadata also for flows *not*
classified as STUN (because of DTLS).
Add support for ipv6.
Restore the correct extra dissection logic for Telegram flows.
The bug is triggered when `pe_offset == (u_int32_t)-1`
```
AddressSanitizer:DEADLYSIGNAL
=================================================================
==23719==ERROR: AddressSanitizer: SEGV on unknown address 0x5081000002b3 (pc 0x55c69274ac72 bp 0x7ffffffc8e70 sp 0x7ffffffc8cc0 T0)
==23719==The signal is caused by a READ memory access.
#0 0x55c69274ac72 in ndpi_search_portable_executable /home/ivan/svnrepos/nDPI/src/lib/ndpi_main.c:8191:7
#1 0x55c69271606b in ndpi_internal_detection_process_packet /home/ivan/svnrepos/nDPI/src/lib/ndpi_main.c:8596:5
#2 0x55c69270f58f in ndpi_detection_process_packet /home/ivan/svnrepos/nDPI/src/lib/ndpi_main.c:8629:22
#3 0x55c6926a07e7 in LLVMFuzzerTestOneInput /home/ivan/svnrepos/nDPI/fuzz/fuzz_process_packet.c:24:5
#4 0x55c6925a79b6 in fuzzer::Fuzzer::ExecuteCallback(unsigned char const*, unsigned long) (/home/ivan/svnrepos/nDPI/fuzz/fuzz_process_packet+0x64e9b6) (BuildId: ec46c60ec7e03ebfb3d825bd6308d0a8d6e9803b)
#5 0x55c692590d48 in fuzzer::RunOneTest(fuzzer::Fuzzer*, char const*, unsigned long) (/home/ivan/svnrepos/nDPI/fuzz/fuzz_process_packet+0x637d48) (BuildId: ec46c60ec7e03ebfb3d825bd6308d0a8d6e9803b)
#6 0x55c69259685a in fuzzer::FuzzerDriver(int*, char***, int (*)(unsigned char const*, unsigned long)) (/home/ivan/svnrepos/nDPI/fuzz/fuzz_process_packet+0x63d85a) (BuildId: ec46c60ec7e03ebfb3d825bd6308d0a8d6e9803b)
#7 0x55c6925c0e02 in main (/home/ivan/svnrepos/nDPI/fuzz/fuzz_process_packet+0x667e02) (BuildId: ec46c60ec7e03ebfb3d825bd6308d0a8d6e9803b)
#8 0x7f8e99793082 in __libc_start_main /build/glibc-wuryBv/glibc-2.31/csu/../csu/libc-start.c:308:16
#9 0x55c69258baed in _start (/home/ivan/svnrepos/nDPI/fuzz/fuzz_process_packet+0x632aed) (BuildId: ec46c60ec7e03ebfb3d825bd6308d0a8d6e9803b)
```
Found by oss-fuzzer
See: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=67881
Tradeoff between key comparison efficiency (i.e. no `memcmp`) and key
length.
At least in the ipv4 cases, we have no more different entries with the
same key.
```
AddressSanitizer:DEADLYSIGNAL
=================================================================
==29723==ERROR: AddressSanitizer: SEGV on unknown address 0x000000000000 (pc 0x562910b70ddb bp 0x7ffcb22c5b70 sp 0x7ffcb22c5a80 T0)
==29723==The signal is caused by a READ memory access.
==29723==Hint: address points to the zero page.
#0 0x562910b70ddb in binary_fuse16_contain /home/ivan/svnrepos/nDPI/src/lib/./third_party/include/binaryfusefilter.h:492:8
#1 0x562910b70bbe in ndpi_bitmap64_isset /home/ivan/svnrepos/nDPI/src/lib/ndpi_bitmap64.c:178:10
#2 0x562910788fd3 in ndpi_domain_classify_longest_prefix /home/ivan/svnrepos/nDPI/src/lib/ndpi_domain_classify.c:261:5
#3 0x56291078940e in ndpi_domain_classify_contains /home/ivan/svnrepos/nDPI/src/lib/ndpi_domain_classify.c:291:9
#4 0x56291069a392 in LLVMFuzzerTestOneInput /home/ivan/svnrepos/nDPI/fuzz/fuzz_ds_domain_classify.cpp:52:5
```
Found by oss-fuzz
See: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=67369
See: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=67372
```
Direct leak of 12 byte(s) in 1 object(s) allocated from:
#0 0x55779e1a46ff in malloc (/home/ivan/svnrepos/nDPI/example/ndpiReader+0x8706ff) (BuildId: 14c2fc626744710d49d652ea1c5bbb24a8cbab4f)
#1 0x55779e2120c7 in ndpi_malloc_wrapper /home/ivan/svnrepos/nDPI/example/ndpiReader.c:298:10
#2 0x55779e5fa215 in ndpi_malloc /home/ivan/svnrepos/nDPI/src/lib/ndpi_memory.c:60:25
#3 0x55779e5fa500 in ndpi_strdup /home/ivan/svnrepos/nDPI/src/lib/ndpi_memory.c:113:13
#4 0x55779e42153c in processClientServerHello /home/ivan/svnrepos/nDPI/src/lib/protocols/tls.c:2554:46
#5 0x55779e4359a1 in processTLSBlock /home/ivan/svnrepos/nDPI/src/lib/protocols/tls.c:908:5
#6 0x55779e432de7 in ndpi_search_tls_tcp /home/ivan/svnrepos/nDPI/src/lib/protocols/tls.c:1097:2
#7 0x55779e4133f9 in ndpi_search_tls_wrapper /home/ivan/svnrepos/nDPI/src/lib/protocols/tls.c:2913:5
```
Found by oss-fuzz
See: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=67250
The basic idea is to have the following logic:
* pattern "DOMAIN" matches the domain itself (i.e exact match) *and* any
subdomains (i.e. "ANYTHING.DOMAIN")
* pattern "DOMAIN." matches *also* any strings for which is a prefix
[please, note that this kind of match is handy but it is quite
dangerous...]
* pattern "-DOMAIN" matches *also* any strings for which is a postfix
Examples:
* pattern "wikipedia.it":
* "wikipiedia.it" -> OK
* "foo.wikipedia.it -> OK
* "foowikipedia.it -> NO MATCH
* "wikipedia.it.com -> NO MATCH
* pattern "wikipedia.":
* "wikipedia.it" -> OK
* "foo.wikipedia.it -> OK
* "foowikipedia.it -> NO MATCH
* "wikipedia.it.com -> OK
* pattern "-wikipedia.it":
* "wikipedia.it" -> NO MATCH
* "foo.wikipedia.it -> NO MATCH
* "0001-wikipedia.it -> OK
* "foo.0001-wikipedia.it -> OK
Bottom line:
* exact match
* prefix with "." (always, implicit)
* prefix with "-" (only if esplicitly set)
* postfix with "." (only if esplicitly set)
That means that the patterns cannot start with '.' anymore.
Close#2330
Skype has been using standard protocols (STUN/ICE or TLS) for a long,
long time, now. Long gone are the days of Skype as a distribuited
protocol.
See: #2166
Follow-up of 4543385d10
Remove trailing spaces for any HTTP header (we already remove leading
spaces)
We want:
* a "normalized" string in `flow->host_server_name`, but
* to parse the original string for flow risk checking
`ndpi_hostname_sni_set()` is a private function, so there is no need to
export its flags.
Add a simple job with macos-14 on M1.
https://github.blog/changelog/2024-01-30-github-actions-introducing-the-new-m1-macos-runner-available-to-open-source/)
There are some issues with external dependencies (they are installed but
autoconf script doens;t find them) so keep it simple.
On macos-13 it seems that:
* there is no `realpath` program (even if coreutils has been
installed...)
* most of the filesystem is read only (we can't write on /usr/lib).
So I change
```
make install DESTDIR=$(realpath _install)
ls -alhHR _install
```
to
```
DESTDIR=/tmp/ndpi make install
ls -alhHR /tmp/ndpi
```
for all the jobs
Fix a warning on GitHub logs:
```
Node.js 16 actions are deprecated. Please update the following actions
to use Node.js 20: actions/checkout@v3. For more information see:
https://github.blog/changelog/2023-09-22-github-actions-transitioning-from-node-16-to-node-20/.
```
Add the concept of "global context".
Right now every instance of `struct ndpi_detection_module_struct` (we
will call it "local context" in this description) is completely
independent from each other. This provide optimal performances in
multithreaded environment, where we pin each local context to a thread,
and each thread to a specific CPU core: we don't have any data shared
across the cores.
Each local context has, internally, also some information correlating
**different** flows; something like:
```
if flow1 (PeerA <-> Peer B) is PROTOCOL_X; then
flow2 (PeerC <-> PeerD) will be PROTOCOL_Y
```
To get optimal classification results, both flow1 and flow2 must be
processed by the same local context. This is not an issue at all in the far
most common scenario where there is only one local context, but it might
be impractical in some more complex scenarios.
Create the concept of "global context": multiple local contexts can use
the same global context and share some data (structures) using it.
This way the data correlating multiple flows can be read/write from
different local contexts.
This is an optional feature, disabled by default.
Obviously data structures shared in a global context must be thread safe.
This PR updates the code of the LRU implementation to be, optionally,
thread safe.
Right now, only the LRU caches can be shared; the other main structures
(trees and automas) are basically read-only: there is little sense in
sharing them. Furthermore, these structures don't have any information
correlating multiple flows.
Every LRU cache can be shared, independently from the others, via
`ndpi_set_config(ndpi_struct, NULL, "lru.$CACHE_NAME.scope", "1")`.
It's up to the user to find the right trade-off between performances
(i.e. without shared data) and classification results (i.e. with some
shared data among the local contexts), depending on the specific traffic
patterns and on the algorithms used to balance the flows across the
threads/cores/local contexts.
Add some basic examples of library initialization in
`doc/library_initialization.md`.
This code needs libpthread as external dependency. It shouldn't be a big
issue; however a configure flag has been added to disable global context
support. A new CI job has been added to test it.
TODO: we should need to find a proper way to add some tests on
multithreaded enviroment... not an easy task...
*** API changes ***
If you are not interested in this feature, simply add a NULL parameter to
any `ndpi_init_detection_module()` calls.
QUIC decryption fails when the Client Hello is split into multiple UDP
packets and these packets have different Destination Connection IDs
(because the server told the client to switch to a different CID; see
RFC 9000 7.2)
```
The Destination Connection ID field from the first Initial packet sent by
a client is used to determine packet protection keys for Initial packets.
[..]
Upon first receiving an Initial or Retry packet from the server, the
client uses the Source Connection ID supplied by the server as the
Destination Connection ID for subsequent packets
```
From a logical point of view, the ciphers used for decryption should be
initialized only once, with the first Initial pkt sent by the client and
kept for later usage with the following packets (if any).
However it seems that we can safely initialize them at each packet, if
we keep using the DCID of the **first** packet sent by the client.
Keep initializing the ciphers at each packet greatly simplifie this patch.
This issue has been undetected for so long because:
* in the vast majority of the cases we only decrypt one packet per flow;
* the available traces with the Client Hello split into multiple packets
(i.e. cases where we need to decrypt at least two packets per flow) were
created in a simple test environment to simulate Post-Quantum handshake,
and in that scenario the client sent all the packets (with the same
DCID) before any reply from the server.
However, in the last months all major browsers started supporting PQ
key, so it is now common to have split CH in real traffic.
Please note that in the attached example, the CH is split into 2
(in-order) fragments (in different UDP packets) and the second one in
turn is divided into 9 (out-of-order) CRYPTO frames; the reassembler
code works out-of-the-box even in this (new) scenario.
Avoid collisions between user-ids and internal-ids protocols in the
`example/protos.txt` file.
Add a new value for the classification confidence:
`NDPI_CONFIDENCE_CUSTOM_RULE`
With `./example/ndpiReader -p example/protos.txt -H` we now see also the
custom protocols and their internal/external ids:
```
nDPI supported protocols:
Id Userd-id Protocol Layer_4 Nw_Proto Breed Category
0 0 Unknown TCP X Unrated Unspecified
...
387 387 Mumble UDP X Fun VoIP
388 388 iSCSI TCP Acceptable Unspecified
389 389 Kibana TCP Acceptable Unspecified
390 390 TestProto TCP Acceptable Unspecified
391 391 HomeRouter TCP Acceptable Unspecified
392 392 CustomProtocol TCP Acceptable Unspecified
393 393 AmazonPrime TCP Acceptable Unspecified
394 394 CustomProtocolA TCP Acceptable Unspecified
395 395 CustomProtocolB TCP Acceptable Unspecified
396 800 CustomProtocolC TCP Acceptable Unspecified
397 1024 CustomProtocolD TCP Acceptable Unspecified
398 2048 CustomProtocolE TCP Acceptable Unspecified
399 2049 CustomProtocolF TCP Acceptable Unspecified
400 2050 CustomProtocolG TCP Acceptable Unspecified
401 65535 CustomProtocolH TCP Acceptable Unspecified
```
We likely need to take a better look in general at the iteration between
internal and external protocols ids...
This PR fixes the issue observed in
https://github.com/ntop/nDPI/pull/2274#discussion_r1460674874 and in
https://github.com/ntop/nDPI/pull/2275.
Some changes in the parameters names.
Add a fuzzer to fuzz the configuration file format.
Add the infrastructure to configuratin callbacks.
Add an helper to map LRU cache indexes to names.
When switching to (D)TLS dissector from the STUN one, we need to clear
any flow risks set from the latter (because we don't have anymore
`NDPI_PROTOCOL_STUN` in the classification results)
Allow to disable this check.
I don't know how much sense these limits have in the application
(especially with those default values...) since we have always had a
hard limit on the library itself (`max_packets_to_process` set to 32).
The only value might be that they provide different limits for TCP and
UDP traffic.
Keep them for the time being...
* fix header length check: we must take into account optional extension
header
* it seems the check on UTPv1 header is weak only in the DATA case: wait
for multiple packets only in that case
Remove some unreached/duplicated code.
Add error checking for `atoi()` calls.
About `isdigit()` and similar functions. The warning reported is:
```
Negative Character Value help
isdigit() is invoked here with an argument of signed type char, but only
has defined behavior for int arguments that are either representable
as unsigned char or equal to the value of macro EOF(-1).
Casting the argument to unsigned char will avoid the undefined behavior.
In a number of libc implementations, isdigit() is implemented using lookup
tables (arrays): passing in a negative value can result in a read underrun.
```
Switching to our macros fix that.
Add a check to `check_symbols.sh` to avoid using the original functions
from libc.
We shouldn't have the same default port associated with multiple
protocols. The key reason is that we use only one of them (for
classification by-port, for example) and which one we really use
depends on the internal order of the protocolsi's initialization (which
is unknown to the user).
We have 3 port collisions:
* 8009: AJP and CiscoVPN
* 445: SMBv1 and SMBv23
* 8080: HTTP_Connect and HTTP_Proxy
I think that the two last cases don't really matter...
About the first one: it seems that CiscoVPN is retired from 2011/2014
(see
https://www.cisco.com/c/en/us/obsolete/security/cisco-vpn-client.html)
and superseded by AnyConnect, which uses different ports (see
https://community.cisco.com/t5/network-security/what-protocols-are-used-on-anyconnect-mobility-client/td-p/4713525).
Therefore it should be safe to remove 8009 from the CiscoVPN default
ports.
Having smaller traces help fuzzing: we want the fuzzers to mutate
"interesting" packets analyzed by nDPI, i.e. the first packets of each
flows.
Try hard to keep the same classification and extraction capabilities
The main difference with the original function is that we allow to
specify the base.
Credit for the original idea and the first implementation to @0xA50C1A1