The SMB::State$recent_files field is meant to have expiring entries.
However, due to usage of &default=string_set(), the &read_expire
attribute is not respected causing unbounded state growth. Replace
&default=string_set() with &default=set().
Thanks to ya-sato on Slack for reporting!
Related: zeek/zeek-docs#179, #3513.
This is similar to what the external corelight/zeek-smb-clear-state script
does, but leverages the smb2_discarded_messages_state() event instead of
regularly checking on the state of SMB connections.
The pcap was created using the dperson/samba container image and mounting
a share with Linux's CIFS filesystem, then copying the content of a
directory with 100 files. The test uses a BPF filter to imitate mostly
"half-duplex" traffic.
When a negotiate request offers no dialects, but the response contains
an ntlm record which selects a dialect, a script error is triggered.
$ zeek -C -r ./f2b0e.pcap 'DPD::ignore_violations+={ Analyzer::ANALYZER_SMB }'
1668615340.837882 expression error in /home/awelzel/corelight-oss/zeek/scripts/base/protocols/smb/./smb1-main.zeek, line 96: no such index (SMB1::c$smb_state$current_cmd$smb1_offered_dialects[SMB1::response$ntlm$dialect_index])
Script error triggered by fuzzing when testing Tim's all-the-fuzzing branch.
When a CREATE request contains the FILE_DELETE_ON_CLOSE option and
the subsequent CREATE response indicates success, we now raise the
smb2_file_delete event to log a delete action in smb_files.log and
also give users a way to handle this scenario.
The provided pcap was generated locally by recording a smbtorture run
of the smb2.delete-on-close-perms test case.
Placed the create_options into the CmdInfo record for potential
exposure in smb_cmd.log (wasn't sure how that would look so left it
for the future).
Fixes#2276.
Due to different double precision on M1, file IDs for SMB could end up
changing on M1 because the access time of a file goes into their
computation. The real solution for this would be changing Zeek's
internal "time" representation to uint64; that's planned, but requires
major surgery. For now, this PR changes the SMB code to also pass SMB's
original time representation (which is a uint64) into script-land, and
then use that for computing the file ID.
Closes#1406
This avoids a problem identified by amanbansal2709 in pull
request #1288. I fixed it in a different way than that pull request
by making sure the ts field is always set so that this isssue doesn't
return in the future.
This adds a "policy" hook into the logging framework's streams and
filters to replace the existing log filter predicates. The hook
signature is as follows:
hook(rec: any, id: Log::ID, filter: Log::Filter);
The logging manager invokes hooks on each log record. Hooks can veto
log records via a break, and modify them if necessary. Log filters
inherit the stream-level hook, but can override or remove the hook as
needed.
The distribution's existing log streams now come with pre-defined
hooks that users can add handlers to. Their name is standardized as
"log_policy" by convention, with additional suffixes when a module
provides multiple streams. The following adds a handler to the Conn
module's default log policy hook:
hook Conn::log_policy(rec: Conn::Info, id: Log::ID, filter: Log::Filter)
{
if ( some_veto_reason(rec) )
break;
}
By default, this handler will get invoked for any log filter
associated with the Conn::LOG stream.
The existing predicates are deprecated for removal in 4.1 but continue
to work.
* origin/topic/jsiwek/empty-lines:
Add 'smtp_excessive_pending_cmds' weird
Fix SMTP command string comparisons
Improve handling of empty lines in several text protocol analyzers
Add rate-limiting sampling mechanism for weird events
Teach timestamp canonifier about timestamps before ~2001
SMB error handling improved. The analyzer isn't destroyed when a problem
is encoutered anymore. The flowbuffer in the parser is now flushed and
the analyzer is set to resync against an SMB command. This was needed
because there is some state about open files that is kept within the
parser itself which was being destroyed and that was causing analysis
after content gaps or parse errors to be faulty. The new mechanism
doesn't detroy the parser so parsing after gaps is improved.
DCE_RPC handling in SMB is improved in the edge case where a drive
mapping isn't seen. There is a new const named SMB::pipe_filenames
which is used as a heuristic for identifying "files" opened on named
pipe shares. If the share mapping type isn't known and a filename
in this set is found, the share type will change to "PIPE" by
generating an event named "smb_pipe_connect_heuristic". Reads and
writes to that file will be sent to the DCE_RPC analyzer instead of
to the files framework.
The concept of "unknown" share types has been removed due to the new
heuristic detection of share types.
Some general clean up of how the SMB cmd log is written and when.
- Looser coupling between these analyzers.
- New ntlm.log (still pretty early)
- Improved string handling for NTLM (convert UTF16 to UTF8)
- SMB2 analyzer now supports GSSAPI.
- Improved abstraction of DCE_RPC operations (still not finished)
- Lots of whitespace cleanup.
There were some cases where the log would be missing a field
or data wouldn't get sent to file analysis. At least some of
this is fixed now and I get confused a bit less when I look
at the logs now.
Also, I made the default handling "FILE" so that things like
FILE_UNKNOWN wouldn't show up in the logs so regularly. It's
technically correct that way, but it doesn't look good and it's
correct as FILE often enough that it make sense to make it the
default I think.
- Actually get the path into the smb_files.log now.
- When a share root is having the "create" message used on it,
instead of giving a null file name, now give a special
indicator of "<share_root>".
- Update test baselines.
