anti/DECNET
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

merge: testing → main (reconcile 2-week divergence)

Browse Source
This commit is contained in:
anti
2026-04-28 18:36:00 -04:00
parent 499836c9e4
commit 862e4dbb31
1235 changed files with 160255 additions and 7996 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
decnet/prober/osfp/__init__.py Normal file
View File

@@ -0,0 +1,27 @@
"""Passive + active OS fingerprinting providers.
Consumed by the profiler's `sniffer_rollup` (and, longer-term, by a
dedicated prober pass). Each provider implements `base.Provider`: given
a dict of observed TCP/IP quirks (window, wscale, mss, options
signature, TTL, etc.), return a best-match OS label with confidence.
Layout mirrors `decnet/geoip/` and `decnet/bus/`: `base.py` defines the
protocol, `factory.py` is the only sanctioned accessor, and each
concrete source (p0f today, nmap-osdb / DECNET-observed later) lives in
its own subpackage. Don't import concrete provider classes directly —
use :func:`factory.get_provider` or :func:`factory.get_all_providers`.
"""
from decnet.prober.osfp.base import OsMatch, Provider
from decnet.prober.osfp.factory import (
get_all_providers,
get_provider,
reset_cache,
)
__all__ = [
"OsMatch",
"Provider",
"get_all_providers",
"get_provider",
"reset_cache",
]

59
decnet/prober/osfp/base.py Normal file
View File

@@ -0,0 +1,59 @@
"""OS-fingerprint provider protocol + OsMatch result shape.
Each concrete provider (p0f v2 today; nmap-osdb / DECNET-observed DB
later) implements `Provider`. Callers go through
:func:`decnet.prober.osfp.factory.get_provider` or
:func:`decnet.prober.osfp.factory.get_all_providers` — direct imports
of a concrete class are forbidden, mirroring the convention in
``decnet/geoip`` and ``decnet/bus``.
"""
from __future__ import annotations
from abc import ABC, abstractmethod
from dataclasses import dataclass
from typing import Any, Optional
@dataclass(frozen=True)
class OsMatch:
"""The result of matching an observation against a provider's DB.
Consumers should prefer higher ``confidence``. Providers compute
confidence as the fraction of signature fields that matched exactly
(vs. wildcard / modulo / "any" predicates) — a signature with every
field constrained scoring 1.0, one with every field wildcarded
approaching 0.0. This is explicit so the profiler can pick the
most-specific match when multiple providers fire.
"""
os: str
flavor: str
confidence: float
provider: str
is_userland: bool = False
def __str__(self) -> str:
tag = "userland" if self.is_userland else self.os
return f"{tag} {self.flavor} ({self.confidence:.2f} via {self.provider})"
class Provider(ABC):
"""Abstract OS-fingerprint source.
Providers consume a dict of observed TCP/IP quirks (``window``,
``wscale``, ``mss``, ``options_sig``, ``ttl``, ``df``,
``total_len``, ``quirks`` — not all fields required) and return a
best-match :class:`OsMatch` or ``None`` when nothing matches.
Providers MUST NOT raise on malformed or partial input — the
upstream caller (`profiler/fingerprint.py::sniffer_rollup`) runs
on data that may be missing any or all fields depending on the
event mix, and a raising provider would wedge every attacker
profile rebuild. Return ``None`` instead.
"""
name: str
@abstractmethod
def match(self, obs: dict[str, Any]) -> Optional[OsMatch]:
"""Return best-match OsMatch for *obs*, or None."""

87
decnet/prober/osfp/factory.py Normal file
View File

@@ -0,0 +1,87 @@
"""OS-fingerprint provider factory.
Dispatch is env-driven (``DECNET_OSFP_PROVIDERS``, comma-separated),
with ``p0f-v2`` as the current default. Structure mirrors
:mod:`decnet.geoip.factory` exactly: lazy singletons, a ``reset_cache``
for tests, no dialect-specific globals past this module.
Callers have two entry points:
- :func:`get_provider` — fetch one provider by name (or the default).
Used by anything that wants a single authoritative answer.
- :func:`get_all_providers` — fetch the full priority chain as a list.
Used by the profiler's :func:`~decnet.profiler.fingerprint.sniffer_rollup`
to try each provider in turn and take the highest-confidence match
across all of them.
Reserved names ``dbip`` / ``maxmind`` don't apply here — we use
``nmap-osdb`` (pending Fyodor's grant) and ``decnet-observed`` (our
own DB of honeypot-captured signatures) as the reserved slots that
raise :class:`NotImplementedError` until their subpackages ship.
"""
from __future__ import annotations
import os
from typing import Optional
from decnet.prober.osfp.base import Provider
_DEFAULT_PROVIDERS = "p0f-v2"
# Lazy singletons, one per name, keyed by the env-selected order so
# resetting the env (via reset_cache in tests) rebuilds cleanly.
_cached: dict[str, Provider] = {}
def _configured_names() -> list[str]:
raw = os.environ.get("DECNET_OSFP_PROVIDERS", _DEFAULT_PROVIDERS)
return [n.strip() for n in raw.split(",") if n.strip()]
def _build(name: str) -> Provider:
if name == "p0f-v2":
from decnet.prober.osfp.p0f.provider import P0fV2Provider
return P0fV2Provider()
if name in ("nmap-osdb", "decnet-observed"):
raise NotImplementedError(
f"OS-fingerprint provider {name!r} is reserved but not yet wired."
)
raise ValueError(f"Unsupported OS-fingerprint provider: {name!r}")
def get_provider(name: Optional[str] = None) -> Provider:
"""Return a single provider — *name* if given, otherwise the first
entry of ``DECNET_OSFP_PROVIDERS`` (default ``p0f-v2``).
Lazily built, memoised. Callers MUST go through this or
:func:`get_all_providers` — direct imports of the concrete
provider class are forbidden per the provider-subpackage convention.
"""
if name is None:
names = _configured_names()
name = names[0] if names else _DEFAULT_PROVIDERS
cached = _cached.get(name)
if cached is not None:
return cached
provider = _build(name)
_cached[name] = provider
return provider
def get_all_providers() -> list[Provider]:
"""Return every configured provider, in priority order.
Declared order in ``DECNET_OSFP_PROVIDERS`` IS priority order. The
consumer (``sniffer_rollup``) iterates and picks the best-scoring
match across all of them; a later provider CAN beat an earlier one
if its signature is more specific, so the "priority" is a tiebreaker,
not a short-circuit.
"""
return [get_provider(n) for n in _configured_names()]
def reset_cache() -> None:
"""Forget memoised providers — tests use this when monkeypatching
``DECNET_OSFP_PROVIDERS`` or ``decnet/prober/osfp/p0f/data/``."""
_cached.clear()

6
decnet/prober/osfp/p0f/__init__.py Normal file
View File

@@ -0,0 +1,6 @@
"""p0f v2 fingerprint database provider.
Upstream: https://lcamtuf.coredump.cx/p0f.shtml (p0f v2.0.8, 2006)
License: LGPL-2.1 (preserved in data/LICENSE.p0f-upstream)
Used in DECNET under GPL-3.0 via LGPL-2.1 §3 — see data/README.md.
"""

498
decnet/prober/osfp/p0f/data/LICENSE.p0f-upstream Normal file
View File

@@ -0,0 +1,498 @@
GNU LESSER GENERAL PUBLIC LICENSE
Version 2.1, February 1999
Copyright (C) 1991, 1999 Free Software Foundation, Inc.
59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
Everyone is permitted to copy and distribute verbatim copies
of this license document, but changing it is not allowed.
Preamble
The licenses for most software are designed to take away your
freedom to share and change it. By contrast, the GNU General Public
Licenses are intended to guarantee your freedom to share and change
free software--to make sure the software is free for all its users.
This license, the Lesser General Public License, applies to some
specially designated software packages--typically libraries--of the
Free Software Foundation and other authors who decide to use it. You
can use it too, but we suggest you first think carefully about whether
this license or the ordinary General Public License is the better
strategy to use in any particular case, based on the explanations below.
When we speak of free software, we are referring to freedom of use,
not price. Our General Public Licenses are designed to make sure that
you have the freedom to distribute copies of free software (and charge
for this service if you wish); that you receive source code or can get
it if you want it; that you can change the software and use pieces of
it in new free programs; and that you are informed that you can do
these things.
To protect your rights, we need to make restrictions that forbid
distributors to deny you these rights or to ask you to surrender these
rights. These restrictions translate to certain responsibilities for
you if you distribute copies of the library or if you modify it.
For example, if you distribute copies of the library, whether gratis
or for a fee, you must give the recipients all the rights that we gave
you. You must make sure that they, too, receive or can get the source
code. If you link other code with the library, you must provide
complete object files to the recipients, so that they can relink them
with the library after making changes to the library and recompiling
it. And you must show them these terms so they know their rights.
We protect your rights with a two-step method: (1) we copyright the
library, and (2) we offer you this license, which gives you legal
permission to copy, distribute and/or modify the library.
To protect each distributor, we want to make it very clear that
there is no warranty for the free library. Also, if the library is
modified by someone else and passed on, the recipients should know
that what they have is not the original version, so that the original
author's reputation will not be affected by problems that might be
introduced by others.
Finally, software patents pose a constant threat to the existence of
any free program. We wish to make sure that a company cannot
effectively restrict the users of a free program by obtaining a
restrictive license from a patent holder. Therefore, we insist that
any patent license obtained for a version of the library must be
consistent with the full freedom of use specified in this license.
Most GNU software, including some libraries, is covered by the
ordinary GNU General Public License. This license, the GNU Lesser
General Public License, applies to certain designated libraries, and
is quite different from the ordinary General Public License. We use
this license for certain libraries in order to permit linking those
libraries into non-free programs.
When a program is linked with a library, whether statically or using
a shared library, the combination of the two is legally speaking a
combined work, a derivative of the original library. The ordinary
General Public License therefore permits such linking only if the
entire combination fits its criteria of freedom. The Lesser General
Public License permits more lax criteria for linking other code with
the library.
We call this license the "Lesser" General Public License because it
does Less to protect the user's freedom than the ordinary General
Public License. It also provides other free software developers Less
of an advantage over competing non-free programs. These disadvantages
are the reason we use the ordinary General Public License for many
libraries. However, the Lesser license provides advantages in certain
special circumstances.
For example, on rare occasions, there may be a special need to
encourage the widest possible use of a certain library, so that it becomes
a de-facto standard. To achieve this, non-free programs must be
allowed to use the library. A more frequent case is that a free
library does the same job as widely used non-free libraries. In this
case, there is little to gain by limiting the free library to free
software only, so we use the Lesser General Public License.
In other cases, permission to use a particular library in non-free
programs enables a greater number of people to use a large body of
free software. For example, permission to use the GNU C Library in
non-free programs enables many more people to use the whole GNU
operating system, as well as its variant, the GNU/Linux operating
system.
Although the Lesser General Public License is Less protective of the
users' freedom, it does ensure that the user of a program that is
linked with the Library has the freedom and the wherewithal to run
that program using a modified version of the Library.
The precise terms and conditions for copying, distribution and
modification follow. Pay close attention to the difference between a
"work based on the library" and a "work that uses the library". The
former contains code derived from the library, whereas the latter must
be combined with the library in order to run.
GNU LESSER GENERAL PUBLIC LICENSE
TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
0. This License Agreement applies to any software library or other
program which contains a notice placed by the copyright holder or
other authorized party saying it may be distributed under the terms of
this Lesser General Public License (also called "this License").
Each licensee is addressed as "you".
A "library" means a collection of software functions and/or data
prepared so as to be conveniently linked with application programs
(which use some of those functions and data) to form executables.
The "Library", below, refers to any such software library or work
which has been distributed under these terms. A "work based on the
Library" means either the Library or any derivative work under
copyright law: that is to say, a work containing the Library or a
portion of it, either verbatim or with modifications and/or translated
straightforwardly into another language. (Hereinafter, translation is
included without limitation in the term "modification".)
"Source code" for a work means the preferred form of the work for
making modifications to it. For a library, complete source code means
all the source code for all modules it contains, plus any associated
interface definition files, plus the scripts used to control compilation
and installation of the library.
Activities other than copying, distribution and modification are not
covered by this License; they are outside its scope. The act of
running a program using the Library is not restricted, and output from
such a program is covered only if its contents constitute a work based
on the Library (independent of the use of the Library in a tool for
writing it). Whether that is true depends on what the Library does
and what the program that uses the Library does.
1. You may copy and distribute verbatim copies of the Library's
complete source code as you receive it, in any medium, provided that
you conspicuously and appropriately publish on each copy an
appropriate copyright notice and disclaimer of warranty; keep intact
all the notices that refer to this License and to the absence of any
warranty; and distribute a copy of this License along with the
Library.
You may charge a fee for the physical act of transferring a copy,
and you may at your option offer warranty protection in exchange for a
fee.
2. You may modify your copy or copies of the Library or any portion
of it, thus forming a work based on the Library, and copy and
distribute such modifications or work under the terms of Section 1
above, provided that you also meet all of these conditions:
a) The modified work must itself be a software library.
b) You must cause the files modified to carry prominent notices
stating that you changed the files and the date of any change.
c) You must cause the whole of the work to be licensed at no
charge to all third parties under the terms of this License.
d) If a facility in the modified Library refers to a function or a
table of data to be supplied by an application program that uses
the facility, other than as an argument passed when the facility
is invoked, then you must make a good faith effort to ensure that,
in the event an application does not supply such function or
table, the facility still operates, and performs whatever part of
its purpose remains meaningful.
(For example, a function in a library to compute square roots has
a purpose that is entirely well-defined independent of the
application. Therefore, Subsection 2d requires that any
application-supplied function or table used by this function must
be optional: if the application does not supply it, the square
root function must still compute square roots.)
These requirements apply to the modified work as a whole. If
identifiable sections of that work are not derived from the Library,
and can be reasonably considered independent and separate works in
themselves, then this License, and its terms, do not apply to those
sections when you distribute them as separate works. But when you
distribute the same sections as part of a whole which is a work based
on the Library, the distribution of the whole must be on the terms of
this License, whose permissions for other licensees extend to the
entire whole, and thus to each and every part regardless of who wrote
it.
Thus, it is not the intent of this section to claim rights or contest
your rights to work written entirely by you; rather, the intent is to
exercise the right to control the distribution of derivative or
collective works based on the Library.
In addition, mere aggregation of another work not based on the Library
with the Library (or with a work based on the Library) on a volume of
a storage or distribution medium does not bring the other work under
the scope of this License.
3. You may opt to apply the terms of the ordinary GNU General Public
License instead of this License to a given copy of the Library. To do
this, you must alter all the notices that refer to this License, so
that they refer to the ordinary GNU General Public License, version 2,
instead of to this License. (If a newer version than version 2 of the
ordinary GNU General Public License has appeared, then you can specify
that version instead if you wish.) Do not make any other change in
these notices.
Once this change is made in a given copy, it is irreversible for
that copy, so the ordinary GNU General Public License applies to all
subsequent copies and derivative works made from that copy.
This option is useful when you wish to copy part of the code of
the Library into a program that is not a library.
4. You may copy and distribute the Library (or a portion or
derivative of it, under Section 2) in object code or executable form
under the terms of Sections 1 and 2 above provided that you accompany
it with the complete corresponding machine-readable source code, which
must be distributed under the terms of Sections 1 and 2 above on a
medium customarily used for software interchange.
If distribution of object code is made by offering access to copy
from a designated place, then offering equivalent access to copy the
source code from the same place satisfies the requirement to
distribute the source code, even though third parties are not
compelled to copy the source along with the object code.
5. A program that contains no derivative of any portion of the
Library, but is designed to work with the Library by being compiled or
linked with it, is called a "work that uses the Library". Such a
work, in isolation, is not a derivative work of the Library, and
therefore falls outside the scope of this License.
However, linking a "work that uses the Library" with the Library
creates an executable that is a derivative of the Library (because it
contains portions of the Library), rather than a "work that uses the
library". The executable is therefore covered by this License.
Section 6 states terms for distribution of such executables.
When a "work that uses the Library" uses material from a header file
that is part of the Library, the object code for the work may be a
derivative work of the Library even though the source code is not.
Whether this is true is especially significant if the work can be
linked without the Library, or if the work is itself a library. The
threshold for this to be true is not precisely defined by law.
If such an object file uses only numerical parameters, data
structure layouts and accessors, and small macros and small inline
functions (ten lines or less in length), then the use of the object
file is unrestricted, regardless of whether it is legally a derivative
work. (Executables containing this object code plus portions of the
Library will still fall under Section 6.)
Otherwise, if the work is a derivative of the Library, you may
distribute the object code for the work under the terms of Section 6.
Any executables containing that work also fall under Section 6,
whether or not they are linked directly with the Library itself.
6. As an exception to the Sections above, you may also combine or
link a "work that uses the Library" with the Library to produce a
work containing portions of the Library, and distribute that work
under terms of your choice, provided that the terms permit
modification of the work for the customer's own use and reverse
engineering for debugging such modifications.
You must give prominent notice with each copy of the work that the
Library is used in it and that the Library and its use are covered by
this License. You must supply a copy of this License. If the work
during execution displays copyright notices, you must include the
copyright notice for the Library among them, as well as a reference
directing the user to the copy of this License. Also, you must do one
of these things:
a) Accompany the work with the complete corresponding
machine-readable source code for the Library including whatever
changes were used in the work (which must be distributed under
Sections 1 and 2 above); and, if the work is an executable linked
with the Library, with the complete machine-readable "work that
uses the Library", as object code and/or source code, so that the
user can modify the Library and then relink to produce a modified
executable containing the modified Library. (It is understood
that the user who changes the contents of definitions files in the
Library will not necessarily be able to recompile the application
to use the modified definitions.)
b) Use a suitable shared library mechanism for linking with the
Library. A suitable mechanism is one that (1) uses at run time a
copy of the library already present on the user's computer system,
rather than copying library functions into the executable, and (2)
will operate properly with a modified version of the library, if
the user installs one, as long as the modified version is
interface-compatible with the version that the work was made with.
c) Accompany the work with a written offer, valid for at
least three years, to give the same user the materials
specified in Subsection 6a, above, for a charge no more
than the cost of performing this distribution.
d) If distribution of the work is made by offering access to copy
from a designated place, offer equivalent access to copy the above
specified materials from the same place.
e) Verify that the user has already received a copy of these
materials or that you have already sent this user a copy.
For an executable, the required form of the "work that uses the
Library" must include any data and utility programs needed for
reproducing the executable from it. However, as a special exception,
the materials to be distributed need not include anything that is
normally distributed (in either source or binary form) with the major
components (compiler, kernel, and so on) of the operating system on
which the executable runs, unless that component itself accompanies
the executable.
It may happen that this requirement contradicts the license
restrictions of other proprietary libraries that do not normally
accompany the operating system. Such a contradiction means you cannot
use both them and the Library together in an executable that you
distribute.
7. You may place library facilities that are a work based on the
Library side-by-side in a single library together with other library
facilities not covered by this License, and distribute such a combined
library, provided that the separate distribution of the work based on
the Library and of the other library facilities is otherwise
permitted, and provided that you do these two things:
a) Accompany the combined library with a copy of the same work
based on the Library, uncombined with any other library
facilities. This must be distributed under the terms of the
Sections above.
b) Give prominent notice with the combined library of the fact
that part of it is a work based on the Library, and explaining
where to find the accompanying uncombined form of the same work.
8. You may not copy, modify, sublicense, link with, or distribute
the Library except as expressly provided under this License. Any
attempt otherwise to copy, modify, sublicense, link with, or
distribute the Library is void, and will automatically terminate your
rights under this License. However, parties who have received copies,
or rights, from you under this License will not have their licenses
terminated so long as such parties remain in full compliance.
9. You are not required to accept this License, since you have not
signed it. However, nothing else grants you permission to modify or
distribute the Library or its derivative works. These actions are
prohibited by law if you do not accept this License. Therefore, by
modifying or distributing the Library (or any work based on the
Library), you indicate your acceptance of this License to do so, and
all its terms and conditions for copying, distributing or modifying
the Library or works based on it.
10. Each time you redistribute the Library (or any work based on the
Library), the recipient automatically receives a license from the
original licensor to copy, distribute, link with or modify the Library
subject to these terms and conditions. You may not impose any further
restrictions on the recipients' exercise of the rights granted herein.
You are not responsible for enforcing compliance by third parties with
this License.
11. If, as a consequence of a court judgment or allegation of patent
infringement or for any other reason (not limited to patent issues),
conditions are imposed on you (whether by court order, agreement or
otherwise) that contradict the conditions of this License, they do not
excuse you from the conditions of this License. If you cannot
distribute so as to satisfy simultaneously your obligations under this
License and any other pertinent obligations, then as a consequence you
may not distribute the Library at all. For example, if a patent
license would not permit royalty-free redistribution of the Library by
all those who receive copies directly or indirectly through you, then
the only way you could satisfy both it and this License would be to
refrain entirely from distribution of the Library.
If any portion of this section is held invalid or unenforceable under any
particular circumstance, the balance of the section is intended to apply,
and the section as a whole is intended to apply in other circumstances.
It is not the purpose of this section to induce you to infringe any
patents or other property right claims or to contest validity of any
such claims; this section has the sole purpose of protecting the
integrity of the free software distribution system which is
implemented by public license practices. Many people have made
generous contributions to the wide range of software distributed
through that system in reliance on consistent application of that
system; it is up to the author/donor to decide if he or she is willing
to distribute software through any other system and a licensee cannot
impose that choice.
This section is intended to make thoroughly clear what is believed to
be a consequence of the rest of this License.
12. If the distribution and/or use of the Library is restricted in
certain countries either by patents or by copyrighted interfaces, the
original copyright holder who places the Library under this License may add
an explicit geographical distribution limitation excluding those countries,
so that distribution is permitted only in or among countries not thus
excluded. In such case, this License incorporates the limitation as if
written in the body of this License.
13. The Free Software Foundation may publish revised and/or new
versions of the Lesser General Public License from time to time.
Such new versions will be similar in spirit to the present version,
but may differ in detail to address new problems or concerns.
Each version is given a distinguishing version number. If the Library
specifies a version number of this License which applies to it and
"any later version", you have the option of following the terms and
conditions either of that version or of any later version published by
the Free Software Foundation. If the Library does not specify a
license version number, you may choose any version ever published by
the Free Software Foundation.
14. If you wish to incorporate parts of the Library into other free
programs whose distribution conditions are incompatible with these,
write to the author to ask for permission. For software which is
copyrighted by the Free Software Foundation, write to the Free
Software Foundation; we sometimes make exceptions for this. Our
decision will be guided by the two goals of preserving the free status
of all derivatives of our free software and of promoting the sharing
and reuse of software generally.
NO WARRANTY
15. BECAUSE THE LIBRARY IS LICENSED FREE OF CHARGE, THERE IS NO
WARRANTY FOR THE LIBRARY, TO THE EXTENT PERMITTED BY APPLICABLE LAW.
EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR
OTHER PARTIES PROVIDE THE LIBRARY "AS IS" WITHOUT WARRANTY OF ANY
KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE
LIBRARY IS WITH YOU. SHOULD THE LIBRARY PROVE DEFECTIVE, YOU ASSUME
THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
16. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN
WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY
AND/OR REDISTRIBUTE THE LIBRARY AS PERMITTED ABOVE, BE LIABLE TO YOU
FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR
CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE
LIBRARY (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING
RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A
FAILURE OF THE LIBRARY TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF
SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
DAMAGES.
END OF TERMS AND CONDITIONS
How to Apply These Terms to Your New Libraries
If you develop a new library, and you want it to be of the greatest
possible use to the public, we recommend making it free software that
everyone can redistribute and change. You can do so by permitting
redistribution under these terms (or, alternatively, under the terms of the
ordinary General Public License).
To apply these terms, attach the following notices to the library. It is
safest to attach them to the start of each source file to most effectively
convey the exclusion of warranty; and each file should have at least the
"copyright" line and a pointer to where the full notice is found.
<one line to give the library's name and a brief idea of what it does.>
Copyright (C) <year> <name of author>
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
Also add information on how to contact you by electronic and paper mail.
You should also get your employer (if you work as a programmer) or your
school, if any, to sign a "copyright disclaimer" for the library, if
necessary. Here is a sample; alter the names:
Yoyodyne, Inc., hereby disclaims all copyright interest in the
library `Frob' (a library for tweaking knobs) written by James Random Hacker.
<signature of Ty Coon>, 1 April 1990
Ty Coon, President of Vice

72
decnet/prober/osfp/p0f/data/README.md Normal file
View File

@@ -0,0 +1,72 @@
# p0f v2 fingerprint database (vendored)
This directory contains the p0f v2.0.8 TCP/IP fingerprint database as
published by Michal Zalewski in 2006, vendored here so DECNET's prober
and profiler can do passive / active OS fingerprinting without a runtime
network fetch.
## What's in here
| File | Purpose | Sigs |
|-----------------------|-----------------------------------------------|------|
| `p0f.fp` | SYN fingerprints (passive, incoming) | 262 |
| `p0fa.fp` | SYN-ACK fingerprints (active probe responses) | 61 |
| `p0fr.fp` | RST+ fingerprints (reset-response quirks) | 46 |
| `p0fo.fp` | "stray" fingerprints | 6 |
| `LICENSE.p0f-upstream`| Verbatim LGPL-2.1 text from upstream | — |
## Provenance
**Authoritative source:** Debian snapshot archive, `p0f_2.0.8.orig.tar.gz`.
- Archive URL: `https://snapshot.debian.org/archive/debian-archive/20120328T092752Z/debian/pool/main/p/p0f/p0f_2.0.8.orig.tar.gz`
- SHA-1 (upstream-recorded by Debian): `7b4d5b2f24af4b5a299979134bc7f6d7b1eaf875`
Files in this directory are byte-identical copies of the corresponding
files inside `p0f_2.0.8.orig.tar.gz::p0f/{doc/COPYING, *.fp}`.
## License + DECNET-side licensing stance
Upstream files are licensed under the **GNU Lesser General Public
License, version 2.1** (see `LICENSE.p0f-upstream` — verbatim copy of
upstream's `doc/COPYING`). Attribution belongs to Michal Zalewski and
the named contributors in the original upstream `CREDITS` file.
DECNET is licensed under **GPL-3.0-or-later**. LGPL-2.1 §3 explicitly
permits converting an LGPL-2.1 work to any version of the GPL at the
recipient's choice. DECNET exercises that conversion for the vendored
files: when consumed as part of DECNET they are effectively under
GPL-3.0. The upstream LGPL-2.1 notice is preserved so:
- Recipients of DECNET see the full chain (original LGPL-2.1 → §3
conversion → GPL-3.0), and
- Anyone who wants to use these signatures under LGPL-2.1 terms
(e.g. in an unrelated library) can still do so by pulling the files
directly from upstream.
## Modifications to upstream
**None.** The four `.fp` files in this directory are verbatim copies.
Any DECNET-authored additions go into a sibling file (`p0f-decnet.fp`,
currently absent) under GPL-3.0, loaded by the same parser. Keeping
upstream untouched means:
1. Syncing future upstream changes is a one-step file replacement.
2. Attribution is unambiguous: entries in `p0f*.fp` here are Michal's,
entries in `p0f-decnet.fp` are DECNET's.
3. If we ever want to contribute signatures back to upstream, it's a
one-file diff.
## Refreshing upstream
```
curl -O https://snapshot.debian.org/archive/debian-archive/20120328T092752Z/debian/pool/main/p/p0f/p0f_2.0.8.orig.tar.gz
echo "7b4d5b2f24af4b5a299979134bc7f6d7b1eaf875 p0f_2.0.8.orig.tar.gz" | sha1sum -c
tar xzf p0f_2.0.8.orig.tar.gz
cp p0f/p0f.fp p0f/p0fa.fp p0f/p0fr.fp p0f/p0fo.fp decnet/prober/osfp/p0f/data/
cp p0f/doc/COPYING decnet/prober/osfp/p0f/data/LICENSE.p0f-upstream
```
p0f v2 is no longer actively maintained upstream (last release 2006),
so refreshes are effectively N/A — but the procedure is recorded for
the case where a mirror we trust publishes a signed rebuild.

834
decnet/prober/osfp/p0f/data/p0f.fp Normal file
View File

@@ -0,0 +1,834 @@
#
# p0f - SYN fingerprints
# ----------------------
#
# .-------------------------------------------------------------------------.
# | The purpose of this file is to cover signatures for incoming TCP/IP |
# | connections (SYN packets). This is the default mode of operation for |
# | p0f. This is also the biggest and most up-to-date set of signatures |
# | shipped with this project. The file also contains a detailed discussion |
# | of all metrics examined by p0f, and some practical notes on how to |
# | add new signatures. |
# `-------------------------------------------------------------------------'
#
# (C) Copyright 2000-2006 by Michal Zalewski <lcamtuf@coredump.cx>
#
# Each line in this file specifies a single fingerprint. Please read the
# information below carefully before attempting to append any signatures
# reported by p0f as UNKNOWN to this file to avoid mistakes. Note that
# this file is compatible only with the default operation mode, and not
# with -R or -A options (SYN+ACK and RST+ modes).
#
# We use the following set metrics for fingerprinting:
#
# - Window size (WSS) - a highly OS dependent setting used for TCP/IP
# performance control (max. amount of data to be sent without ACK).
# Some systems use a fixed value for initial packets. On other
# systems, it is a multiple of MSS or MTU (MSS+40). In some rare
# cases, the value is just arbitrary.
#
# NEW SIGNATURE: if p0f reported a special value of 'Snn', the number
# appears to be a multiple of MSS (MSS*nn); a special value of 'Tnn'
# means it is a multiple of MTU ((MSS+40)*nn). Unless you notice the
# value of nn is not fixed (unlikely), just copy the Snn or Tnn token
# literally. If you know this device has a simple stack and a fixed
# MTU, you can however multiply S value by MSS, or T value by MSS+40,
# and put it instead of Snn or Tnn. One system may exhibit several T
# or S values. In some situations, this might be a source of some
# additional information about the setup if you have some time to dig
# thru the kernel sources; in some other cases, like Windows, there seem
# to be a multitude of variants and WSS selection algorithms, but it's
# rather difficult to find a pattern without having the source.
#
# If WSS looks like a regular fixed value (for example is a power of two),
# or if you can confirm the value is fixed by looking at several
# fingerprints, please quote it literaly. If there's no apparent pattern
# in WSS chosen, you should consider wildcarding this value - but this
# should be the last option.
#
# NOTE: Some NAT devices, such as Linux iptables with --set-mss, will
# modify MSS, but not WSS. As a result, MSS is changed to reflect
# the MTU of the NAT device, but WSS remains a multiple of the original
# MSS. Fortunately for us, the source device would almost always be
# hooked up to Ethernet. P0f handles it automatically for the original
# MSS of 1460, by adding "NAT!" tag to the result.
#
# In certain configurations, Linux erratically (?) uses MTU from another
# interface on the default gw interface. This only happens on systems with
# two network interfaces. Thus, some Linux systems that do not go thru NAT,
# but have multiple interfaces instead, will be also tagged this way.
#
# P0f recognizes and automatically wildcards WSS of 12345, as generated
# by sendack and sendsyn utilities shipped with the program, when
# reporting a new signature. See test/sendack.c and test/sendsyn.c for more
# information about this.
#
# - Overall packet size - a function of all IP and TCP options and bugs.
# While this is partly redundant in the real world, we record this value
# to capture rare cases when there are IP options (which we do not currently
# examine) or packet data past the headers. Both situations are rare.
#
# Packet size MAY be wildcarded, but the meaning of the wildcard is
# very special, and means the packet must be larger than PACKET_BIG
# (defined in config.h as 100). This is usually not necessary, except
# for some really broken implementations in RST+ mode. For more information,
# see p0fr.fp. P0f automatically wildcards big packets when reporting
# new signatures.
#
# NEW SIGNATURE: Copy this value literally.
#
# - Initial TTL - We check the actual TTL of a received packet. It can't
# be higher than the initial TTL, and also shouldn't be dramatically
# lower (maximum distance is defined in config.h as 40 hops).
#
# NEW SIGNATURE: *Never* copy TTL from a p0f-reported signature literally.
# You need to determine the initial TTL. The best way to do it is to
# check the documentation for a remote system, or check its settings.
# A fairly good method is to simply round the observed TTL up to
# 32, 64, 128, or 255, but it should be noted that some obscure devices
# might not use round TTLs (in particular, some shoddy appliances and
# IRIX and Tru64 are known to use "original" initial TTL settings). If not
# sure, use traceroute or mtr to see how far you are from the host.
#
# Note that -F option overrides this check if no signature can be found.
#
# - Don't fragment flag (DF) - some modern OSes set this to implement PMTU
# discovery. Others do not bother.
#
# NEW SIGNATURE: Copy this value literally. Note: this setting is
# sometimes cleared by firewalls and/or certain connectivity clients.
# Try to find out what's the actual state for a given OS if you see both,
# and add the right one. P0f will automatically detect a case when a
# firewall removed the DF flag and will append "(firewall!)" suffix to
# the signature, so if the DF version is the right one, don't add no-DF
# variant, unless it has a different meaning.
#
# - Maximum segment size (MSS) - this setting is usually link-dependent. P0f
# uses it to determine link type of the remote host.
#
# NEW SIGNATURE: Always wildcard this value, except for rare cases when
# you have an appliance with a fixed value, know the system supports only
# a very limited number of network interface types, or know the system
# is using a value it pulled out of nowhere. I use specific unique MSS
# to tell Google crawlbots from the rest of Linux population, for example.
#
# If a specific MSS/MTU is unique to a certain link type, be sure to
# add it to mtu.h instead of creating several variants of each signature.
#
# - Window scaling (WSCALE) - this feature is used to scale WSS.
# It extends the size of a TCP/IP window to 32 bits, of sorts. Some modern
# systems implement this feature.
#
# NEW SIGNATURE: Observe several signatures. Initial WSCALE is often set
# to zero or other low value. There's usually no need to wildcard this
# parameter.
#
# - Timestamp - some systems that implement timestamps set them to
# zero in the initial SYN. This case is detected and handled appropriately.
#
# NEW SIGNATURE: Copy T or T0 option literally.
#
# - Selective ACK permitted - a flag set by systems that implement
# selective ACK functionality,
#
# NEW SIGNATURE: copy S option literally.
#
# - NOP option - its presence, count and sequence is a useful OS-dependent
# characteristic,
#
# NEW SIGNATURE: copy N options literally.
#
# - Other and unrecognized options (TTCP-related and such) - implemented by
# some eccentric or very buggy TCP/IP stacks ;-),
#
# NEW SIGNATURE: copy ? options literally.
#
# - EOL option. Contrary to the popular belief, the presence of EOL
# option is actually quite rare, most systems just NOP-pad to the
# packet boundary.
#
# NEW SIGNATURE: copy E option literally.
#
# - The sequence of TCP all options mentioned above - this is very
# specific to the implementation,
#
# NEW SIGNATURE: Copy the sequence literally.
#
# - Quirks. Some buggy stacks set certain values that should be zeroed in a
# TCP packet to non-zero values. This has no effect as of today, but is
# a valuable source of information. Some systems actually seem to leak
# memory there. Other systems just exhibit harmful but very specific
# behavior. This section captures all unusual yes-no properties not
# related to the main and expected header layout. We detect the following:
#
# - Data past the headers. Neither SYN nor SYN+ACK packets are supposed
# to carry any payload. If they do, we should take notice. The actual
# payload is not examined, but will be displayed if use the -X option.
# Note that payload is not unusual in RST+ mode (see p0fr.fp), very
# rare otherwise.
#
# - Options past EOL. Some systems have some trailing data past EOL
# in the options section of TCP/IP headers. P0f does not examine this
# data as of today, simply detects its presence. If there is a
# confirmed sizable population of systems that have data past EOL, it
# might be a good idea to look at it. Until then, you have to recompile
# p0f with DEBUG_EXTRAS set or use -x to display this data,
#
# - Zero IP ID. This again is a (mostly) harmless setting to use a fixed
# IP ID for packets with DF set. Some systems reportedly use zero ID,
# most OSes do not. There is a very slight probability of a false
# positive when IP ID is "naturally" chosen to be zero on a system
# that otherwise does set proper values, but the probability is
# neglible (if it becomes a problem, recompile p0f with IGNORE_ZEROID
# set in the sources).
#
# - IP options specified. Usually, packets do not have any IP options
# set, but there can be some. Until there is a confirmed sizable
# population of systems that do have IP options in a packet, p0f
# does not examine those in detail, but it might change (use
# DEBUG_EXTRAS or -x to display IP options if any found),
#
# - URG pointer value. SYN packets do not have URG flag set, so the
# value in URG pointer in TCP header is ignored. Most systems set it
# to zero, but some OSes (some versions of Windows, for example) do
# not zero this field or even simply leak memory; the actual value is
# not examined, because most cases seem to be just random garbage
# (you can use DEBUG_EXTRAS or -x to report this information though);
# see doc/win-memleak.txt for more information,
#
# - "Unused" field value. This should be always zero, but some systems
# forget to clear it. This might result in some funny issues in the
# future. P0f checks for non-zero value (and will display it if
# DEBUG_EXTRAS is set, or you can use -x),
#
# - ACK number non-zero. ACK value in SYN packets with no ACK flag
# is disregarded and is usually set to zero (just like with URG
# pointer), but some systems forget to do it. The exact value is
# not examined (but will be displayed with DEBUG_EXTRAS, or you can
# use -x). Note that this is not an anomaly in SYN+ACK and RST+ modes,
#
# - Non-zero second timestamp. The initial SYN packet should have the
# second timestamp always zeroed. SYN+ACK and RST+ may "legally" have
# this quirk though,
#
# - Unusual flags. If, in addition to SYN (or SYN+ACK), there are some
# auxilinary flags that do not modify the very meaning of a packet,
# p0f records this (this can be URG, PUSH, or something else).
#
# Note: ECN flags (ECE and CWR) are ignored and denoted in a separate
# way. ECN is never by default, because some systems can't handle it,
# and it probably does not make much sense to include it in signatures
# right now.
#
# - TCP option segment parsing problems. If p0f fails to decode options
# because of a badly broken packet, it records this fact.
#
# There are several other quirks valid only in RST+ mode, see p0fr.fp for
# more information. Those quirks are unheard of in SYN and SYN+ACK
# modes.
#
# NEW SIGNATURE: Copy "quirks" section literally.
#
# We DO NOT use ToS for fingerprinting. While the original TCP/IP
# fingerprinting research believed this value would be useful for this
# purpose, it is not. The setting is way too often tweaked by network
# devices.
#
# To wildcard MSS, WSS or WSCALE, replace it with '*'. You can also use a
# modulo operator to match any values that divide by nnn - '%nnn' (and,
# as stated above, WSS also supports special values Snn and Tnn).
#
# Fingerprint entry format:
#
# wwww:ttt:D:ss:OOO...:QQ:OS:Details
#
# wwww - window size (can be * or %nnn or Sxx or Txx)
# "Snn" (multiple of MSS) and "Tnn" (multiple of MTU) are allowed.
# ttt - initial TTL
# D - don't fragment bit (0 - not set, 1 - set)
# ss - overall SYN packet size (* has a special meaning)
# OOO - option value and order specification (see below)
# QQ - quirks list (see below)
# OS - OS genre (Linux, Solaris, Windows)
# details - OS description (2.0.27 on x86, etc)
#
# If OS genre starts with '*', p0f will not show distance, link type
# and timestamp data. It is useful for userland TCP/IP stacks of
# network scanners and so on, where many settings are randomized or
# bogus.
#
# If OS genre starts with @, it denotes an approximate hit for a group
# of operating systems (signature reporting still enabled in this case).
# Use this feature at the end of this file to catch cases for which
# you don't have a precise match, but can tell it's Windows or FreeBSD
# or whatnot by looking at, say, flag layout alone.
#
# If OS genre starts with - (which can prefix @ or *), the entry is
# not considered to be a real operating system (but userland stack
# instead). It is important to mark all scanners and so on with -,
# so that they are not used for masquerade detection (also add this
# prefix for signatures of application-induced behavior, such as
# increased window size with Opera browser).
#
# Option block description is a list of comma or space separated
# options in the order they appear in the packet:
#
# N - NOP option
# E - EOL option
# Wnnn - window scaling option, value nnn (or * or %nnn)
# Mnnn - maximum segment size option, value nnn (or * or %nnn)
# S - selective ACK OK
# T - timestamp
# T0 - timestamp with zero value
# ?n - unrecognized option number n.
#
# P0f can sometimes report ?nn among the options. This means it couldn't
# recognize this option (option number nn). It's either a bug in p0f, or
# a faulty TCP/IP stack, or, if the number is listed here:
#
# http://www.iana.org/assignments/tcp-parameters
#
# ...the stack might be simply quite exotic.
#
# To denote no TCP options, use a single '.'.
#
# Quirks section is usually an empty list ('.') of oddities or bugs of this
# particular stack. List items are not separated in any way. Possible values:
#
# P - options past EOL,
# Z - zero IP ID,
# I - IP options specified,
# U - urg pointer non-zero,
# X - unused (x2) field non-zero,
# A - ACK number non-zero,
# T - non-zero second timestamp,
# F - unusual flags (PUSH, URG, etc),
# D - data payload,
# ! - broken options segment.
#
# WARNING WARNING WARNING
# -----------------------
#
# Do not add a system X as OS Y just because NMAP says so. It is often
# the case that X is a NAT firewall. While nmap is talking to the
# device itself, p0f is fingerprinting the guy behind the firewall
# instead.
#
# When in doubt, use common sense, don't add something that looks like
# a completely different system as Linux or FreeBSD or LinkSys router.
# Check DNS name, establish a connection to the remote host and look
# at SYN+ACK (p0f -A -S should do) - does it look similar?
#
# Some users tweak their TCP/IP settings - enable or disable RFC1323,
# RFC1644 or RFC2018 support, disable PMTU discovery, change MTU, initial
# TTL and so on. Always compare a new rule to other fingerprints for
# this system, and verify the system isn't "customized". It is OK to
# add signature variants caused by commonly used software (PFs, security
# packages, etc), but it makes no sense to try to add every single
# possible /proc/sys/net/ipv4/* tweak on Linux or so.
#
# KEEP IN MIND: Some packet firewalls configured to normalize outgoing
# traffic (OpenBSD pf with "scrub" enabled, for example) will, well,
# normalize packets. Signatures will not correspond to the originating
# system (and probably not quite to the firewall either).
#
# NOTE: Try to keep this file in some reasonable order, from most to
# least likely systems. This will speed up operation. Also keep most
# generic and broad rules near ehe end.
#
# Still decided to add signature? Let us know - mail a copy of your discovery
# to lcamtuf@coredump.cx. You can help make p0f better, and I can help you
# make your signature more accurate.
#
##########################
# Standard OS signatures #
##########################
# ----------------- AIX ---------------------
# AIX is first because its signatures are close to NetBSD, MacOS X and
# Linux 2.0, but it uses a fairly rare MSSes, at least sometimes...
# This is a shoddy hack, though.
45046:64:0:44:M*:.:AIX:4.3
16384:64:0:44:M512:.:AIX:4.3.2 and earlier
16384:64:0:60:M512,N,W%2,N,N,T:.:AIX:4.3.3-5.2 (1)
32768:64:0:60:M512,N,W%2,N,N,T:.:AIX:4.3.3-5.2 (2)
65535:64:0:60:M512,N,W%2,N,N,T:.:AIX:4.3.3-5.2 (3)
65535:64:0:64:M*,N,W1,N,N,T,N,N,S:.:AIX:5.3 ML1
# ----------------- Linux -------------------
S1:64:0:44:M*:A:Linux:1.2.x
512:64:0:44:M*:.:Linux:2.0.3x (1)
16384:64:0:44:M*:.:Linux:2.0.3x (2)
# Endian snafu! Nelson says "ha-ha":
2:64:0:44:M*:.:Linux:2.0.3x (MkLinux) on Mac (1)
64:64:0:44:M*:.:Linux:2.0.3x (MkLinux) on Mac (2)
S4:64:1:60:M1360,S,T,N,W0:.:Linux:2.4 (Google crawlbot)
S4:64:1:60:M1430,S,T,N,W0:.:Linux:2.4-2.6 (Google crawlbot)
S2:64:1:60:M*,S,T,N,W0:.:Linux:2.4 (large MTU?)
S3:64:1:60:M*,S,T,N,W0:.:Linux:2.4 (newer)
S4:64:1:60:M*,S,T,N,W0:.:Linux:2.4-2.6
S3:64:1:60:M*,S,T,N,W1:.:Linux:2.6, seldom 2.4 (older, 1)
S4:64:1:60:M*,S,T,N,W1:.:Linux:2.6, seldom 2.4 (older, 2)
S3:64:1:60:M*,S,T,N,W2:.:Linux:2.6, seldom 2.4 (older, 3)
S4:64:1:60:M*,S,T,N,W2:.:Linux:2.6, seldom 2.4 (older, 4)
T4:64:1:60:M*,S,T,N,W2:.:Linux:2.6 (older, 5)
S4:64:1:60:M*,S,T,N,W5:.:Linux:2.6 (newer, 1)
S4:64:1:60:M*,S,T,N,W6:.:Linux:2.6 (newer, 2)
S4:64:1:60:M*,S,T,N,W7:.:Linux:2.6 (newer, 3)
T4:64:1:60:M*,S,T,N,W7:.:Linux:2.6 (newer, 4)
S20:64:1:60:M*,S,T,N,W0:.:Linux:2.2 (1)
S22:64:1:60:M*,S,T,N,W0:.:Linux:2.2 (2)
S11:64:1:60:M*,S,T,N,W0:.:Linux:2.2 (3)
# Popular cluster config scripts disable timestamps and
# selective ACK:
S4:64:1:48:M1460,N,W0:.:Linux:2.4 in cluster
# This happens only over loopback, but let's make folks happy:
32767:64:1:60:M16396,S,T,N,W0:.:Linux:2.4 (loopback)
32767:64:1:60:M16396,S,T,N,W2:.:Linux:2.6 (newer, loopback)
S8:64:1:60:M3884,S,T,N,W0:.:Linux:2.2 (loopback)
# Opera visitors:
16384:64:1:60:M*,S,T,N,W0:.:-Linux:2.2 (Opera?)
32767:64:1:60:M*,S,T,N,W0:.:-Linux:2.4 (Opera?)
# Some fairly common mods & oddities:
S22:64:1:52:M*,N,N,S,N,W0:.:Linux:2.2 (tstamp-)
S4:64:1:52:M*,N,N,S,N,W0:.:Linux:2.4 (tstamp-)
S4:64:1:52:M*,N,N,S,N,W2:.:Linux:2.6 (tstamp-)
S4:64:1:44:M*:.:Linux:2.6? (barebone, rare!)
T4:64:1:60:M1412,S,T,N,W0:.:Linux:2.4 (rare!)
# ----------------- FreeBSD -----------------
16384:64:1:44:M*:.:FreeBSD:2.0-4.2
16384:64:1:60:M*,N,W0,N,N,T:.:FreeBSD:4.4 (1)
1024:64:1:60:M*,N,W0,N,N,T:.:FreeBSD:4.4 (2)
57344:64:1:44:M*:.:FreeBSD:4.6-4.8 (RFC1323-)
57344:64:1:60:M*,N,W0,N,N,T:.:FreeBSD:4.6-4.9
32768:64:1:60:M*,N,W0,N,N,T:.:FreeBSD:4.8-5.1 (or MacOS X 10.2-10.3)
65535:64:1:60:M*,N,W0,N,N,T:.:FreeBSD:4.7-5.2 (or MacOS X 10.2-10.4) (1)
65535:64:1:60:M*,N,W1,N,N,T:.:FreeBSD:4.7-5.2 (or MacOS X 10.2-10.4) (2)
65535:64:1:60:M*,N,W0,N,N,T:Z:FreeBSD:5.1 (1)
65535:64:1:60:M*,N,W1,N,N,T:Z:FreeBSD:5.1 (2)
65535:64:1:60:M*,N,W2,N,N,T:Z:FreeBSD:5.1 (3)
65535:64:1:64:M*,N,N,S,N,W1,N,N,T:.:FreeBSD:5.3-5.4
65535:64:1:64:M*,N,W1,N,N,T,S,E:P:FreeBSD:6.x (1)
65535:64:1:64:M*,N,W0,N,N,T,S,E:P:FreeBSD:6.x (2)
65535:64:1:44:M*:Z:FreeBSD:5.2 (RFC1323-)
# 16384:64:1:60:M*,N,N,N,N,N,N,T:.:FreeBSD:4.4 (tstamp-)
# ----------------- NetBSD ------------------
16384:64:0:60:M*,N,W0,N,N,T:.:NetBSD:1.3
65535:64:0:60:M*,N,W0,N,N,T0:.:-NetBSD:1.6 (Opera)
16384:64:1:60:M*,N,W0,N,N,T0:.:NetBSD:1.6
65535:64:1:60:M*,N,W1,N,N,T0:.:NetBSD:1.6W-current (DF)
65535:64:1:60:M*,N,W0,N,N,T0:.:NetBSD:1.6X (DF)
32768:64:1:60:M*,N,W0,N,N,T0:.:NetBSD:1.6Z or 2.0 (DF)
32768:64:1:64:M1416,N,W0,S,N,N,N,N,T0:.:NetBSD:2.0G (DF)
32768:64:1:64:M*,N,W0,S,N,N,N,N,T0:.:NetBSD:3.0 (DF)
# ----------------- OpenBSD -----------------
16384:64:1:64:M*,N,N,S,N,W0,N,N,T:.:OpenBSD:3.0-3.9
57344:64:1:64:M*,N,N,S,N,W0,N,N,T:.:OpenBSD:3.3-3.4
16384:64:0:64:M*,N,N,S,N,W0,N,N,T:.:OpenBSD:3.0-3.4 (scrub)
65535:64:1:64:M*,N,N,S,N,W0,N,N,T:.:-OpenBSD:3.0-3.4 (Opera?)
32768:64:1:64:M*,N,N,S,N,W0,N,N,T:.:OpenBSD:3.7
# ----------------- Solaris -----------------
S17:64:1:64:N,W3,N,N,T0,N,N,S,M*:.:Solaris:8 (RFC1323 on)
S17:64:1:48:N,N,S,M*:.:Solaris:8 (1)
S17:255:1:44:M*:.:Solaris:2.5-7 (1)
# Sometimes, just sometimes, Solaris feels like coming up with
# rather arbitrary MSS values ;-)
S6:255:1:44:M*:.:Solaris:2.5-7 (2)
S23:64:1:48:N,N,S,M*:.:Solaris:8 (2)
S34:64:1:48:M*,N,N,S:.:Solaris:9
S34:64:1:48:M*,N,N,N,N:.:Solaris:9 (no sack)
S44:255:1:44:M*:.:Solaris:7
4096:64:0:44:M1460:.:SunOS:4.1.x
S34:64:1:52:M*,N,W0,N,N,S:.:Solaris:10 (beta)
32850:64:1:64:M*,N,N,T,N,W1,N,N,S:.:Solaris:10 (1203?)
32850:64:1:64:M*,N,W1,N,N,T,N,N,S:.:Solaris:9.1
# ----------------- IRIX --------------------
49152:60:0:44:M*:.:IRIX:6.2-6.4
61440:60:0:44:M*:.:IRIX:6.2-6.5
49152:60:0:52:M*,N,W2,N,N,S:.:IRIX:6.5 (RFC1323+) (1)
49152:60:0:52:M*,N,W3,N,N,S:.:IRIX:6.5 (RFC1323+) (2)
61440:60:0:48:M*,N,N,S:.:IRIX:6.5.12-6.5.21 (1)
49152:60:0:48:M*,N,N,S:.:IRIX:6.5.12-6.5.21 (2)
49152:60:0:64:M*,N,W2,N,N,T,N,N,S:.:IRIX:6.5 IP27
# ----------------- Tru64 -------------------
# Tru64 and OpenVMS share the same stack on occassions.
# Relax.
32768:60:1:48:M*,N,W0:.:Tru64:4.0 (or OS/2 Warp 4)
32768:60:0:48:M*,N,W0:.:Tru64:5.0 (or OpenVMS 7.x on Compaq 5.0 stack)
8192:60:0:44:M1460:.:Tru64:5.1 (no RFC1323) (or QNX 6)
61440:60:0:48:M*,N,W0:.:Tru64:v5.1a JP4 (or OpenVMS 7.x on Compaq 5.x stack)
# ----------------- OpenVMS -----------------
6144:64:1:60:M*,N,W0,N,N,T:.:OpenVMS:7.2 (Multinet 4.3-4.4 stack)
# ----------------- MacOS -------------------
S2:255:1:48:M*,W0,E:.:MacOS:8.6 classic
16616:255:1:48:M*,W0,E:.:MacOS:7.3-8.6 (OTTCP)
16616:255:1:48:M*,N,N,N,E:.:MacOS:8.1-8.6 (OTTCP)
32768:255:1:48:M*,W0,N:.:MacOS:9.0-9.2
32768:255:1:48:M1380,N,N,N,N:.:MacOS:9.1 (OT 2.7.4) (1)
65535:255:1:48:M*,N,N,N,N:.:MacOS:9.1 (OT 2.7.4) (2)
# ----------------- Windows -----------------
# Windows TCP/IP stack is a mess. For most recent XP, 2000 and
# even 98, the pathlevel, not the actual OS version, is more
# relevant to the signature. They share the same code, so it would
# seem. Luckily for us, almost all Windows 9x boxes have an
# awkward MSS of 536, which I use to tell one from another
# in most difficult cases.
8192:32:1:44:M*:.:Windows:3.11 (Tucows)
S44:64:1:64:M*,N,W0,N,N,T0,N,N,S:.:Windows:95
8192:128:1:64:M*,N,W0,N,N,T0,N,N,S:.:Windows:95b
# There were so many tweaking tools and so many stack versions for
# Windows 98 it is no longer possible to tell them from each other
# without some very serious research. Until then, there's an insane
# number of signatures, for your amusement:
S44:32:1:48:M*,N,N,S:.:Windows:98 (low TTL) (1)
8192:32:1:48:M*,N,N,S:.:Windows:98 (low TTL) (2)
%8192:64:1:48:M536,N,N,S:.:Windows:98 (13)
%8192:128:1:48:M536,N,N,S:.:Windows:98 (15)
S4:64:1:48:M*,N,N,S:.:Windows:98 (1)
S6:64:1:48:M*,N,N,S:.:Windows:98 (2)
S12:64:1:48:M*,N,N,S:.:Windows:98 (3
T30:64:1:64:M1460,N,W0,N,N,T0,N,N,S:.:Windows:98 (16)
32767:64:1:48:M*,N,N,S:.:Windows:98 (4)
37300:64:1:48:M*,N,N,S:.:Windows:98 (5)
46080:64:1:52:M*,N,W3,N,N,S:.:Windows:98 (RFC1323+)
65535:64:1:44:M*:.:Windows:98 (no sack)
S16:128:1:48:M*,N,N,S:.:Windows:98 (6)
S16:128:1:64:M*,N,W0,N,N,T0,N,N,S:.:Windows:98 (7)
S26:128:1:48:M*,N,N,S:.:Windows:98 (8)
T30:128:1:48:M*,N,N,S:.:Windows:98 (9)
32767:128:1:52:M*,N,W0,N,N,S:.:Windows:98 (10)
60352:128:1:48:M*,N,N,S:.:Windows:98 (11)
60352:128:1:64:M*,N,W2,N,N,T0,N,N,S:.:Windows:98 (12)
# What's with 1414 on NT?
T31:128:1:44:M1414:.:Windows:NT 4.0 SP6a (1)
64512:128:1:44:M1414:.:Windows:NT 4.0 SP6a (2)
8192:128:1:44:M*:.:Windows:NT 4.0 (older)
# Windows XP and 2000. Most of the signatures that were
# either dubious or non-specific (no service pack data)
# were deleted and replaced with generics at the end.
65535:128:1:48:M*,N,N,S:.:Windows:2000 SP4, XP SP1+
%8192:128:1:48:M*,N,N,S:.:Windows:2000 SP2+, XP SP1+ (seldom 98)
S20:128:1:48:M*,N,N,S:.:Windows:SP3
S45:128:1:48:M*,N,N,S:.:Windows:2000 SP4, XP SP1+ (2)
40320:128:1:48:M*,N,N,S:.:Windows:2000 SP4
S6:128:1:48:M*,N,N,S:.:Windows:XP, 2000 SP2+
S12:128:1:48:M*,N,N,S:.:Windows:XP SP1+ (1)
S44:128:1:48:M*,N,N,S:.:Windows:XP SP1+, 2000 SP3
64512:128:1:48:M*,N,N,S:.:Windows:XP SP1+, 2000 SP3 (2)
32767:128:1:48:M*,N,N,S:.:Windows:XP SP1+, 2000 SP4 (3)
# Windows 2003 & Vista
8192:128:1:52:M*,W8,N,N,N,S:.:Windows:Vista (beta)
32768:32:1:52:M1460,N,W0,N,N,S:.:Windows:2003 AS
65535:64:1:52:M1460,N,W2,N,N,S:.:Windows:2003 (1)
65535:64:1:48:M1460,N,N,S:.:Windows:2003 (2)
# Odds, ends, mods:
S52:128:1:48:M1260,N,N,S:.:Windows:XP/2000 via Cisco
65520:128:1:48:M*,N,N,S:.:Windows:XP bare-bone
16384:128:1:52:M536,N,W0,N,N,S:.:Windows:2000 w/ZoneAlarm?
2048:255:0:40:.:.:Windows:.NET Enterprise Server
44620:64:0:48:M*,N,N,S:.:Windows:ME no SP (?)
S6:255:1:48:M536,N,N,S:.:Windows:95 winsock 2
32000:128:0:48:M*,N,N,S:.:Windows:XP w/Winroute?
16384:64:1:48:M1452,N,N,S:.:Windows:XP w/Sygate? (1)
17256:64:1:48:M1460,N,N,S:.:Windows:XP w/Sygate? (2)
# No need to be more specific, it passes:
*:128:1:48:M*,N,N,S:U:-Windows:XP/2000 while downloading (leak!)
# ----------------- HP/UX -------------------
32768:64:1:44:M*:.:HP-UX:B.10.20
32768:64:1:48:M*,W0,N:.:HP-UX:11.00-11.11
# Whoa. Hardcore WSS.
0:64:0:48:M*,W0,N:.:HP-UX:B.11.00 A (RFC1323+)
# ----------------- RiscOS ------------------
16384:64:1:68:M1460,N,W0,N,N,T,N,N,?12:.:RISC OS:3.70-4.36 (inet 5.04)
12288:32:0:44:M536:.:RISC OS:3.70 inet 4.10
4096:64:1:56:M1460,N,N,T:T:RISC OS:3.70 freenet 2.00
# ----------------- BSD/OS ------------------
8192:64:1:60:M1460,N,W0,N,N,T:.:BSD/OS:3.1-4.3 (or MacOS X 10.2)
# ---------------- NetwonOS -----------------
4096:64:0:44:M1420:.:NewtonOS:2.1
# ---------------- NeXTSTEP -----------------
S8:64:0:44:M512:.:NeXTSTEP:3.3 (1)
S4:64:0:44:M1024:.:NeXTSTEP:3.3 (2)
# ------------------ BeOS -------------------
1024:255:0:48:M*,N,W0:.:BeOS:5.0-5.1
12288:255:0:44:M*:.:BeOS:5.0.x
# ------------------ OS/400 -----------------
8192:64:1:60:M1440,N,W0,N,N,T:.:OS/400:V4R4/R5
8192:64:0:44:M536:.:OS/400:V4R3/M0
4096:64:1:60:M1440,N,W0,N,N,T:.:OS/400:V4R5 + CF67032
28672:64:0:44:M1460:A:OS/390:?
# ------------------ ULTRIX -----------------
16384:64:0:40:.:.:ULTRIX:4.5
# ------------------- QNX -------------------
S16:64:0:44:M512:.:QNX:demodisk
16384:64:0:60:M1460,N,W0,N,N,T0:.:QNX:6.x
# ------------------ Novell -----------------
16384:128:1:44:M1460:.:Novell:NetWare 5.0
6144:128:1:44:M1460:.:Novell:IntranetWare 4.11
6144:128:1:44:M1368:.:Novell:BorderManager ?
# According to rfp:
6144:128:1:52:M*,W0,N,S,N,N:.:Novell:Netware 6 SP3
# -------------- SCO UnixWare ---------------
S3:64:1:60:M1460,N,W0,N,N,T:.:SCO:UnixWare 7.1
S17:64:1:60:M*,N,W0,N,N,T:.:SCO:UnixWare 7.1.x
S23:64:1:44:M1380:.:SCO:OpenServer 5.0
# ------------------- DOS -------------------
2048:255:0:44:M536:.:DOS:Arachne via WATTCP/1.05
T2:255:0:44:M984:.:DOS:Arachne via WATTCP/1.05 (eepro)
16383:64:0:44:M536:.:DOS:Unknown via WATTCP (epppd)
# ------------------ OS/2 -------------------
S56:64:0:44:M512:.:OS/2:4
28672:64:0:44:M1460:.:OS/2:Warp 4.0
# ----------------- TOPS-20 -----------------
# Another hardcore MSS, one of the ACK leakers hunted down.
0:64:0:44:M1460:A:TOPS-20:version 7
# ------------------ AMIGA ------------------
S32:64:1:56:M*,N,N,S,N,N,?12:.:AMIGA:3.9 BB2 with Miami stack
# ------------------ Minix ------------------
# Not quite sure.
# 8192:210:0:44:M1460:X:@Minix:?
# ------------------ Plan9 ------------------
65535:255:0:48:M1460,W0,N:.:Plan9:edition 4
# ----------------- AMIGAOS -----------------
16384:64:1:48:M1560,N,N,S:.:AMIGAOS:3.9 BB2 MiamiDX
# ----------------- FreeMiNT ----------------
S44:255:0:44:M536:.:FreeMiNT:1 patch 16A (Atari)
###########################################
# Appliance / embedded / other signatures #
###########################################
# ---------- Firewalls / routers ------------
S12:64:1:44:M1460:.:@Checkpoint:(unknown 1)
S12:64:1:48:N,N,S,M1460:.:@Checkpoint:(unknown 2)
4096:32:0:44:M1460:.:ExtremeWare:4.x
S32:64:0:68:M512,N,W0,N,N,T,N,N,?12:.:Nokia:IPSO w/Checkpoint NG FP3
S16:64:0:68:M1024,N,W0,N,N,T,N,N,?12:.:Nokia:IPSO 3.7 build 026
S4:64:1:60:W0,N,S,T,M1460:.:FortiNet:FortiGate 50
8192:64:1:44:M1460:.:@Eagle:Secure Gateway
# ------- Switches and other stuff ----------
4128:255:0:44:M*:Z:Cisco:7200, Catalyst 3500, etc
S8:255:0:44:M*:.:Cisco:12008
S4:255:0:44:M536:Z:Cisco:IOS 11.0
60352:128:1:64:M1460,N,W2,N,N,T,N,N,S:.:Alteon:ACEswitch
64512:128:1:44:M1370:.:Nortel:Contivity Client
# ---------- Caches and whatnots ------------
8190:255:0:44:M1428:.:Google:Wireless Transcoder (1)
8190:255:0:44:M1460:.:Google:Wireless Transcoder (2)
8192:64:1:64:M1460,N,N,S,N,W0,N,N,T:.:NetCache:5.2
16384:64:1:64:M1460,N,N,S,N,W0,N:.:NetCache:5.3
65535:64:1:64:M1460,N,N,S,N,W*,N,N,T:.:NetCache:5.3-5.5 (or FreeBSD 5.4)
20480:64:1:64:M1460,N,N,S,N,W0,N,N,T:.:NetCache:4.1
S44:64:1:64:M1460,N,N,S,N,W0,N,N,T:.:NetCache:5.5
32850:64:1:64:N,W1,N,N,T,N,N,S,M*:.:NetCache:Data OnTap 5.x
65535:64:0:60:M1460,N,W0,N,N,T:.:CacheFlow:CacheOS 4.1
8192:64:0:60:M1380,N,N,N,N,N,N,T:.:CacheFlow:CacheOS 1.1
S4:64:0:48:M1460,N,N,S:.:Cisco:Content Engine
27085:128:0:40:.:.:Dell:PowerApp cache (Linux-based)
65535:255:1:48:N,W1,M1460:.:Inktomi:crawler
S1:255:1:60:M1460,S,T,N,W0:.:LookSmart:ZyBorg
16384:255:0:40:.:.:Proxyblocker:(what's this?)
65535:255:0:48:M*,N,N,S:.:Redline: T|X 2200
# ----------- Embedded systems --------------
S9:255:0:44:M536:.:PalmOS:Tungsten T3/C
S5:255:0:44:M536:.:PalmOS:3/4
S4:255:0:44:M536:.:PalmOS:3.5
2948:255:0:44:M536:.:PalmOS:3.5.3 (Handera)
S29:255:0:44:M536:.:PalmOS:5.0
16384:255:0:44:M1398:.:PalmOS:5.2 (Clie)
S14:255:0:44:M1350:.:PalmOS:5.2.1 (Treo)
16384:255:0:44:M1400:.:PalmOS:5.2 (Sony)
S23:64:1:64:N,W1,N,N,T,N,N,S,M1460:.:SymbianOS:7
8192:255:0:44:M1460:.:SymbianOS:6048 (Nokia 7650?)
8192:255:0:44:M536:.:SymbianOS:(Nokia 9210?)
S22:64:1:56:M1460,T,S:.:SymbianOS:? (SE P800?)
S36:64:1:56:M1360,T,S:.:SymbianOS:60xx (Nokia 6600?)
S36:64:1:60:M1360,T,S,W0,E:.:SymbianOS:60xx
32768:32:1:44:M1460:.:Windows:CE 3
# Perhaps S4?
5840:64:1:60:M1452,S,T,N,W1:.:Zaurus:3.10
32768:128:1:64:M1460,N,W0,N,N,T0,N,N,S:.:PocketPC:2002
S1:255:0:44:M346:.:Contiki:1.1-rc0
4096:128:0:44:M1460:.:Sega:Dreamcast Dreamkey 3.0
T5:64:0:44:M536:.:Sega:Dreamcast HKT-3020 (browser disc 51027)
S22:64:1:44:M1460:.:Sony:Playstation 2 (SOCOM?)
S12:64:0:44:M1452:.:AXIS:Printer Server 5600 v5.64
3100:32:1:44:M1460:.:Windows:CE 2.0
####################
# Fancy signatures #
####################
1024:64:0:40:.:.:-*NMAP:syn scan (1)
2048:64:0:40:.:.:-*NMAP:syn scan (2)
3072:64:0:40:.:.:-*NMAP:syn scan (3)
4096:64:0:40:.:.:-*NMAP:syn scan (4)
1024:64:0:40:.:A:-*NMAP:TCP sweep probe (1)
2048:64:0:40:.:A:-*NMAP:TCP sweep probe (2)
3072:64:0:40:.:A:-*NMAP:TCP sweep probe (3)
4096:64:0:40:.:A:-*NMAP:TCP sweep probe (4)
1024:64:0:60:W10,N,M265,T,E:P:-*NMAP:OS detection probe (1)
2048:64:0:60:W10,N,M265,T,E:P:-*NMAP:OS detection probe (2)
3072:64:0:60:W10,N,M265,T,E:P:-*NMAP:OS detection probe (3)
4096:64:0:60:W10,N,M265,T,E:P:-*NMAP:OS detection probe (4)
1024:64:0:60:W10,N,M265,T,E:PF:-*NMAP:OS detection probe w/flags (1)
2048:64:0:60:W10,N,M265,T,E:PF:-*NMAP:OS detection probe w/flags (2)
3072:64:0:60:W10,N,M265,T,E:PF:-*NMAP:OS detection probe w/flags (3)
4096:64:0:60:W10,N,M265,T,E:PF:-*NMAP:OS detection probe w/flags (4)
32767:64:0:40:.:.:-*NAST:syn scan
12345:255:0:40:.:A:-p0f:sendsyn utility
# UFO - see tmp/*:
56922:128:0:40:.:A:-@Mysterious:port scanner (?)
5792:64:1:60:M1460,S,T,N,W0:T:-@Mysterious:NAT device (2nd tstamp)
S12:128:1:48:M1460,E:P:@Mysterious:Chello proxy (?)
S23:64:1:64:N,W1,N,N,T,N,N,S,M1380:.:@Mysterious:GPRS gateway (?)
#####################################
# Generic signatures - just in case #
#####################################
*:128:1:52:M*,N,W0,N,N,S:.:@Windows:XP/2000 (RFC1323+, w, tstamp-)
*:128:1:52:M*,N,W*,N,N,S:.:@Windows:XP/2000 (RFC1323+, w+, tstamp-)
*:128:1:52:M*,N,N,T0,N,N,S:.:@Windows:XP/2000 (RFC1323+, w-, tstamp+)
*:128:1:64:M*,N,W0,N,N,T0,N,N,S:.:@Windows:XP/2000 (RFC1323+, w, tstamp+)
*:128:1:64:M*,N,W*,N,N,T0,N,N,S:.:@Windows:XP/2000 (RFC1323+, w+, tstamp+)
*:128:1:48:M536,N,N,S:.:@Windows:98
*:128:1:48:M*,N,N,S:.:@Windows:XP/2000

208
decnet/prober/osfp/p0f/data/p0fa.fp Normal file
View File

@@ -0,0 +1,208 @@
#
# p0f - SYN+ACK fingerprints
# --------------------------
#
# .-------------------------------------------------------------------------.
# | The purpose of this file is to cover signatures for outgoing TCP/IP |
# | connections (SYN+ACK packets). This mode of operation can be enabled |
# | with -A option. Please refer to p0f.fp for information on the metrics |
# | used to create a signature, and for a guide on adding new entries to |
# | those files. This database is somewhat neglected, and is looking for a |
# | caring maintainer. |
# `-------------------------------------------------------------------------'
#
# (C) Copyright 2000-2006 by Michal Zalewski <lcamtuf@coredump.cx>
#
# Plenty of signatures contributed in bulk by rain forest puppy, Paul Woo and
# Michael Bauer.
#
# Submit all additions to the authors. Read p0f.fp before adding any
# signatures. Run p0f -A -C after making any modifications. This file is
# NOT compatible with SYN, RST+, or stray ACK modes. Use only with -A option.
#
# Feel like contributing? You can run p0f -A -K, then test/tryid -iR nnn...
#
# IMPORTANT INFORMATION ABOUT THE INTERDEPENDENCY OF SYNs AND SYN+ACKs
# --------------------------------------------------------------------
#
# Some systems would have different SYN+ACK fingerprints depending on
# the system that sent SYN. More specifically, RFC1323, RFC2018 and
# RFC1644 extensions sometimes show up only if SYN had them enabled.
#
# Also, some silly systems may copy WSS from the SYN packet you've sent,
# in which case, you need to wildcard the value. Use test/sendsyn.c, which
# uses a distinct WSS of 12345, to test for this condition if unsure.
#
# IMPORTANT INFORMATION ABOUT DIFFERENCES IN COMPARISON TO p0f.fp:
# ----------------------------------------------------------------
#
# - 'A' quirk would be present on almost every signature here. ACK number
# is unusual for SYN packets, but is a commonplace in SYN+ACK packets,
# of course. It is still possible to have a signature without 'A', when
# the ACK flag is present but the value is zero - this, however, is
# very uncommon.
#
# - 'T' quirk would show up on almost all signatures for systems implementing
# RFC1323. The second timestamp is only unusual for SYN packets. SYN+ACK
# are expected to have it set.
#
##########################
# Standard OS signatures #
##########################
# ---------------- Linux -------------------
32736:64:0:44:M*:A:Linux:2.0
S22:64:1:60:M*,S,T,N,W0:AT:Linux:2.2
S22:64:1:52:M*,N,N,S,N,W0:A:Linux:2.2 w/o timestamps
5792:64:1:60:M*,S,T,N,W0:AT:Linux:older 2.4
5792:64:1:60:M*,S,T,N,W0:ZAT:Linux:recent 2.4 (1)
S4:64:1:44:M*:ZA:Linux:recent 2.4 (2)
5792:64:1:44:M*:ZA:Linux:recent 2.4 (3)
S4:64:1:52:M*,N,N,S,N,W0:ZA:Linux:2.4 w/o timestamps
# --------------- Windows ------------------
65535:128:1:64:M*,N,W0,N,N,T0,N,N,S:A:Windows:2000 SP4
S44:128:1:64:M*,N,W0,N,N,T0,N,N,S:A:Windows:XP SP1
S12:128:1:64:M*,N,W0,N,N,T0,N,N,S:A:Windows:2000 (SP1+)
S6:128:1:44:M*:A:Windows:NT 4.0 SP1+
65535:128:1:48:M*,N,N,S:A:Windows:98 (SE)
65535:128:1:44:M*:A:Windows:2000 (1)
16616:128:1:44:M*:A:Windows:2003
16384:128:1:44:M*:A:Windows:2000 (2)
S16:128:1:44:M*:A:Windows:2000 (3)
# ------------------- OpenBSD --------------
17376:64:1:64:M*,N,N,S,N,W0,N,N,T:AT:OpenBSD:3.3
# ------------------- NetBSD ----------------
16384:64:0:60:M*,N,W0,N,N,T0:AT:NetBSD:1.6
# ----------------- HP/UX ------------------
32768:64:1:44:M*:A:HPUX:10.20
# ----------------- Tru64 ------------------
S23:60:0:48:M*,N,W0:A:Tru64:5.0 (1)
65535:64:0:44:M*:A:Tru64:5.0 (2)
# ----------------- Novell -----------------
6144:128:1:52:M*,W0,N,S,N,N:A:Novell:Netware 6.0 (SP3)
32768:128:1:44:M*:A:Novell:Netware 5.1
# ------------------ IRIX ------------------
60816:60:1:60:M*,N,W0,N,N,T:AT:IRIX:6.5.0
# ----------------- Solaris ----------------
49232:64:1:64:N,N,T,M*,N,W0,N,N,S:AT:Solaris:9 (1)
S1:255:1:60:N,N,T,N,W0,M*:AT:Solaris:7
24656:64:1:44:M*:A:Solaris:8
33304:64:1:60:N,N,T,M*,N,W1:AT:Solaris:9 (2)
# ----------------- FreeBSD ----------------
65535:64:1:60:M*,N,W1,N,N,T:AT:FreeBSD:5.0
57344:64:1:44:M*:A:FreeBSD:4.6-4.8
65535:64:1:44:M*:A:FreeBSD:4.4
57344:64:1:48:M1460,N,W0:A:FreeBSD:4.6-4.8 (wscale)
57344:64:1:60:M1460,N,W0,N,N,T:AT:FreeBSD:4.6-4.8 (RFC1323)
# ------------------- AIX ------------------
S17:255:1:44:M536:A:AIX:4.2
S12:64:0:44:M1460:A:AIX:5.2 ML04 (1)
S42:64:0:44:M1460:A:AIX:5.2 ML04 (2)
# ------------------ BSD/OS ----------------
S6:64:1:60:M1460,N,W0,N,N,T:AT:BSD/OS:4.0.x
# ------------------ OS/390 ----------------
2048:64:0:44:M1460:A:OS/390:?
# ------------------ Novell ----------------
6144:128:1:44:M1400:A:Novell:iChain 2.2
# ------------------ MacOS -----------------
33304:64:1:60:M*,N,W0,N,N,T:AT:MacOS:X 10.2.6
#################################################################
# Contributed by Ryan Kruse <rkruse@alterpoint.com> - trial run #
#################################################################
# S4:255:0:44:M1024:A:Cisco:LocalDirector
# 1024:255:0:44:M536:A:Cisco,3COM,Nortel:CatIOS,SuperStack,BayStack
# S16:64:0:44:M512:A:Nortel:Contivity
# 8192:64:0:44:M1460:A:Cisco,Nortel,SonicWall,Tasman:Aironet,BayStack Switch,Soho,1200
# 4096:255:0:44:M1460:A:Cisco:PIX,CatOS
# 8192:128:0:44:M1460:A:Cisco:VPN Concentrator
# 8192:128:0:60:M1460,N,W0,N,N,T:AT:Cisco:VPN Concentrator
# 4096:32:0:44:M1460:A:Cisco,3COM,Extreme,Nortel:Catalyst Switch CatOS,CoreBuilder,Summit,Passport
# S4:255:0:44:M536:ZA:Cisco:IOS
# 1024:32:0:44:M1480:UA:Nortel:BayStack Switch
# 4096:60:0:44:M1460:A:Adtran:NetVanta
# 4096:64:0:44:M1008:A:Adtran:TSU
# S4:32:0:44:M1024:A:Alcatel:Switch
# S8:255:0:44:M536:ZA:Cisco:IOS
# 50:255:0:44:M536:ZA:Cisco:CatIOS
# 512:64:0:40:.:A:Dell:Switch
# 4096:64:0:40:.:A:Enterasys:Vertical Horizon Switch
# 17640:64:1:44:M1460:A:F5,Juniper,RiverStone:BigIP,Juniper OS,Router 7.0+
# 16384:64:0:44:M1460:A:Foundry,SonicWall:BigIron,TZ
# 4096:64:0:44:M1452:A:HP:ProCurve Switch
# 1024:64:0:44:M1260:A:Marconi:ES
# 10240:30:0:44:M1460:A:Milan:Switch
# 4096:64:0:44:M1380:A:NetScreen:Firewall
# S32:64:0:44:M512:A:Nokia:CheckPoint
# 1024:64:0:44:M536:A:Nortel:BayStack Switch
# 4128:255:0:44:M*:ZA:Cisco:IOS
# 1024:16:0:44:M536:A:Nortel:BayStack Switch
# 1024:30:0:44:M1480:A:Nortel:BayStack Switch
# S4:64:0:44:M1460:A:Symbol:Spectrum Access Point
# S2:255:0:44:M512:A:ZyXEL:Prestige
# S16:255:0:44:M1024:A:ZyXEL:ZyAI
###########################################
# Appliance / embedded / other signatures #
###########################################
16384:64:1:44:M1460:A:F5:BigIP LB 4.1.x (sometimes FreeBSD)
4128:255:0:44:M*:ZA:Cisco:Catalyst 2900 12.0(5)
4096:60:0:44:M*:A:Brother:HL-1270N
S1:30:0:44:M1730:A:Cyclades:PR3000
8192:64:1:44:M1460:A:NetApp:Data OnTap 6.x
5792:64:1:60:W0,N,N,N,T,M1460:ZAT:FortiNet:FortiGate 50
S1:64:1:44:M1460:A:NetCache:5.3.1
S1:64:0:44:M512:A:Printer:controller (?)
4096:128:0:40:.:A:Sequent:DYNIX 4.2.x
S16:64:0:44:M512:A:3Com:NBX PBX (BSD/OS 2.1)
16000:64:0:44:M1442:A:CastleNet:DSL router
S2:64:0:44:M32728:A:D-Link:DSL-500
S4:60:0:44:M1460:A:HP:JetDirect A.05.32
8576:64:1:44:M*:A:Raptor:firewall
S12:64:1:44:M1400:A:Cequrux Firewall:4.x
2048:255:0:44:M1400:A:Netgear:MR814
16384:128:0:64:M1460,N,W0,N,N,T0,N,N,S:A:Akamai:??? (1)
16384:128:0:60:M1460,N,W0,N,N,T0:A:Akamai:??? (2)
8190:255:0:44:M1452:A:Citrix:Netscaler 6.1
# Whatever they run. EOL boys...
S6:128:1:48:M1460,E:PA:@Slashdot:or BusinessWeek (???)

48
decnet/prober/osfp/p0f/data/p0fo.fp Normal file
View File

@@ -0,0 +1,48 @@
#
# p0f - stray ACK signatures
# --------------------------
#
# .-------------------------------------------------------------------------.
# | The purpose of this file is to cover signatures for stray ACK packets |
# | (established session data). This mode of operation is enabled with -O |
# | option and is HIGHLY EXPERIMENTAL. Please refer to p0f.fp for more |
# | information on the metrics used and for a guide on adding new entries |
# | to this file. This database is looking for a caring maintainer. |
# `-------------------------------------------------------------------------'
#
# (C) Copyright 2000-2006 by Michal Zalewski <lcamtuf@coredump.cx>
#
# Submit all additions to the authors. Read p0f.fp before adding any
# signatures. Run p0f -O -C after making any modifications. This file is
# NOT compatible with SYN, SYN+ACK or RST+ modes. Use only with -O option.
#
# IMPORTANT INFORMATION ABOUT THE INTERDEPENDENCY OF SYNs AND ACKs
# ----------------------------------------------------------------
#
# Some systems would have different ACK fingerprints depending on the initial
# SYN or SYN+ACK received from the other party. More specifically, RFC1323,
# RFC2018 and RFC1644 extensions sometimes show up only if the other party had
# them enabled. Hence, the reliability of ACK fingerprints may be affected.
#
# IMPORTANT INFORMATION ABOUT DIFFERENCES IN COMPARISON TO p0f.fp:
# ----------------------------------------------------------------
#
# - Packet size MUST be wildcarded. ACK packets, by their nature, have
# variable sizes, depending on the amount of data carried as a payload.
#
# - Similarly, 'D' quirk is not checked for, and is not allowed in signatures
# in this file. A good number of ACK packets have payloads.
#
# - PUSH flag is excluded from 'F' quirk checks in this mode.
#
# - 'A' quirk is not a bug; all AC packets should have it set; also,
# 'T' quirk is not an anomaly; its absence on systems with T option is.
#
32767:64:1:*:N,N,T:AT:Linux:2.4.2x (local?)
*:64:1:*:.:A:Linux:2.4.2x
32736:64:0:*:.:A:Linux:2.0.3x
57600:64:1:*:N,N,T:AT:FreeBSD:4.8
%12:128:1:*:.:A:Windows:XP
S44:128:1:*:.:A:Windows:XP

193
decnet/prober/osfp/p0f/data/p0fr.fp Normal file
View File

@@ -0,0 +1,193 @@
#
# p0f - RST+ signatures
# ---------------------
#
# .-------------------------------------------------------------------------.
# | The purpose of this file is to cover signatures for reset packets |
# | (RST and RST+ACK). This mode of operation can be enabled with -A option |
# | and is considered to be least accurate. Please refer to p0f.fp for more |
# | information on the metrics used and for a guide on adding new entries |
# | to this file. This database is looking for a caring maintainer. |
# `-------------------------------------------------------------------------'
#
# (C) Copyright 2000-2006 by Michal Zalewski <lcamtuf@coredump.cx>
#
# Submit all additions to the authors. Read p0f.fp before adding any
# signatures. Run p0f -R -C after making any modifications. This file is
# NOT compatible with SYN, SYN+ACK, or stray ACK modes. Use only with -R
# option.
#
# IMPORTANT INFORMATION ABOUT THE INTERDEPENDENCY OF SYNs AND RST+ACKs
# --------------------------------------------------------------------
#
# Some silly systems may copy WSS from the SYN packet you've sent,
# in which case, you need to wildcard the value. Use test/sendsyn.c for
# "connection refused" and test/sendack.c for "connection dropped" signatures
# - both tools use a distinct WSS of 12345, which is an easy way to tell
# if WSS should be wildcarded.
#
# IMPORTANT INFORMATION ABOUT COMMON IMPLEMENTATION FLAWS
# -------------------------------------------------------
#
# There are several types of RST packets you will surely encounter.
# Some systems, including most reputable ones, are severily brain-damaged
# and generate some illegal combinations from time to time. This is WAY
# more common than with other packet types, because a broken RST does not
# have any immediately noticable consequences; besides, the RFC793 is fairly
# difficult to comprehend when it comes to this type of responses.
#
# P0f will give you a hint on new RST signatures, but it is your duty to
# diagnose the problem and append the proper description when adding the
# signature. Below is a list of valid and invalid states:
#
# - "Connection refused" message: this is a RST+ACK packet, SEQ number
# set to zero, ACK number non-zero. This is a valid response and
# is denoted by p0f as "refused" (quirk combination: K, 0, A).
#
# There are some very cases when this is incorrectly sent in response
# to an unexpected ACK packet.
#
# - Illegal combination: RST+ACK packet, SEQ number set to zero, ACK
# number zero. This is denoted by p0f as "invalid-K0" (quirk combination:
# K and 0, no A).
#
# - Illegal combination: RST+ACK, SEQ number non-zero, ACK number zero
# or non-zero. This is denoted by p0f as "invalid-K" and
# "invalid-KA", respectively (quirk combinations, K, sometimes A, no 0).
#
# This combination is frequently generated by Cisco routers in certain
# configurations in response to ACK (!). Brain dead, by all means, and
# usually a result of (incorrectly) setting ACK flag on a valid RST packet.
#
# - "Connection dropped": RST, sequence number non-zero, ACK zero or
# non-zero. This is denoted as "dropped" and "dropped 2" respectively
# (quirk combinations: no K, sometimes A, no 0). While the ACK value should
# be zeroed, it is not strictly against the RFC, and some systems either
# leak memory there or set it to the value of SEQ.
#
# The latter variant, with non-zero ACK, is particularly common on
# Windows.
#
# - Ilegal combination: RST, SEQ number zero, ACK zero or non-zero.
# Denoted as "invalid-0" and "invalid-0A". Obviously incorrect, and
# will not have the desired effect.
#
# Ok. That's it. RFC793 does not get much respect nowadays.
#
# IMPORTANT INFORMATION ABOUT DIFFERENCES IN COMPARISON TO p0f.fp:
# ----------------------------------------------------------------
#
# - Packet size may be wildcarded. The meaning of wildcard is, however,
# hardcoded as 'size > PACKET_BIG' (defined as 100 in config.h). This is
# because some stupid devices (including Ciscos) tend to send back RST
# packets quoting anything you have sent them in ACK packet previously.
# Use sparingly, only if -X confirms the device actually bounces back
# whatever you send.
#
# - A new quirk, 'K', is introduced to denote RST+ACK packets (as opposed
# to plain RST). This quirk is only compatible with this mode.
#
# - A new quirk, 'Q', is used to denote SEQ number equal to ACK number.
# This happens from time to time in RST and RST+ACK packets, but
# is practically unheard of in other modes.
#
# - A new quirk, '0', is used to denote packets with SEQ number set to 0.
# This happens on some RSTs, and is once again unheard of in other modes.
#
# - 'D' quirk is not a bug; some devices send verbose text messages
# describing why a connection got dropped; it's actually suggested
# by RFC1122. Of course, some systems have their own standards, and
# put all kinds of crap in their RST responses (including FreeBSD and
# Cisco). Use -X to examine those values.
#
# - 'A' and 'T' quirks are not an anomaly in certain cases for the reasons
# described in p0fa.fp.
#
################################
# Connection refused - RST+ACK #
################################
0:255:0:40:.:K0A:Linux:2.0/2.2 (refused)
0:64:1:40:.:K0A:FreeBSD:4.8 (refused)
0:64:1:40:.:K0ZA:Linux:recent 2.4 (refused)
0:128:0:40:.:K0A:Windows:XP/2000 (refused)
0:128:0:40:.:K0UA:-Windows:XP/2000 while browsing (refused)
######################################
# Connection dropped / timeout - RST #
######################################
0:64:1:40:.:.:FreeBSD:4.8 (dropped)
0:255:0:40:.:.:Linux:2.0/2.2 or IOS 12.x (dropped)
0:64:1:40:.:Z:Linux:recent 2.4 (dropped)
0:255:1:40:.:Z:Linux:early 2.4 (dropped)
0:32:0:40:.:.:Xylan:OmniSwitch / Linksys WAP11 AP (dropped)
0:64:1:40:.:U:NetIron:load balancer (dropped)
0:128:1:40:.:QA:Windows:XP/2000 (dropped 2)
0:128:1:40:.:A:-Windows:XP/2000 while browsing (1) (dropped 2)
0:128:1:40:.:QUA:-Windows:XP/2000 while browsing (2) (dropped 2)
0:128:1:40:.:UA:-Windows:XP/2000 while browsing a lot (dropped 2)
0:128:1:40:.:.:@Windows:98 (?) (dropped)
0:64:0:40:.:A:Ascend:TAOS or BayTech (dropped 2)
*:255:0:40:.:QA:Cisco:LocalDirector (dropped 2)
0:64:1:40:.:A:Hasbani:WindWeb (dropped 2)
S23:255:1:40:.:.:Solaris:2.5 (dropped)
#######################################################
# Connection dropped / timeout - RST with description #
#######################################################
0:255:1:58:.:D:MacOS:9.x "No TCP/No listener" (seldom SunOS 5.x) (dropped)
0:255:1:53:.:D:MacOS:8.5 "no tcp, reset" (dropped)
0:255:1:65:.:D:MacOS:X "tcp_close, during connect" (dropped)
0:255:1:54:.:D:MacOS:X "tcp_disconnect" (dropped)
0:255:1:62:.:D:HP/UX:? "tcp_fin_wait_2_timeout" (dropped)
32768:255:1:54:.:D:MacOS:8.5 "tcp_disconnect" (dropped)
0:255:1:63:.:D:@Unknown: "Go away" device (dropped)
0:255:0:62:.:D:SunOS:5.x "new data when detached" (1) (dropped)
32768:255:1:62:.:D:SunOS:5.x "new data when detached" (2) (dropped)
0:255:1:67:.:D:SunOS:5.x "tcp_lift_anchor, can't wait" (dropped)
0:255:0:46:.:D:HP/UX:11.00 "No TCP" (dropped)
# More obscure ones:
# 648:255:1:54:.:D:MacOS:??? "tcp_disconnect" (dropped)
# 0:45:1:53:.:D:MacOS:7.x "no tcp, reset" (dropped)
##############################################
# Connection dropped / timeout - broken RSTs #
##############################################
S12:255:1:58:.:KAD:Solaris:2.x "tcp_disconnect" (dropped, lame)
S43:64:1:40:.:KA:AOL:proxy (dropped, lame)
*:64:1:40:.:KA:FreeBSD:4.8 (dropped, lame)
*:64:1:52:N,N,T:KAT:Linux:2.4 (?) (dropped, lame)
0:255:0:40:.:KAF:3Com:SuperStack II (dropped, lame)
*:255:0:40:.:KA:Intel:Netport print server (dropped, lame)
*:150:0:40:.:KA:Linksys:BEF router (dropped, lame)
*:32:0:44:.:KZD:@NetWare:??? "ehnc" (dropped, lame)
0:64:0:40:.:KQ0:BayTech:RPC-3 telnet host (dropped, lame)
#############################################
# Connection dropped / timeout - extra data #
#############################################
*:255:0:*:.:KAD:Cisco:IOS/PIX NAT + data (1) (dropped, lame)
0:255:0:*:.:D:Windows:NT 4.0 SP6a + data (dropped)
0:255:0:*:.:K0AD:Isolation:Infocrypt accelerator + data (dropped, lame)
*:255:0:*:.:AD:Cisco:IOS/PIX NAT + data (2) (dropped)
*:64:1:*:N,N,T:KATD:Linux:2.4 (?) + data (dropped, lame)
*:64:1:*:.:KAD:FreeBSD:4.8 + data (dropped, lame)

243
decnet/prober/osfp/p0f/format.py Normal file
View File

@@ -0,0 +1,243 @@
"""p0f v2 ``.fp`` file parser.
Format (from the DSL spec at the top of every shipped ``.fp`` file):
wwww:ttt:D:ss:OOO:QQ:OS:Details
Where:
wwww — window size: literal int | '*' | '%nnn' | 'Snn' | 'Tnn'
ttt — initial TTL (literal int: 32/64/128/255 typically)
D — DF bit: '0' or '1'
ss — total IP packet length: literal int | '*' | '%nnn'
OOO — option order: comma/space-separated tokens, or '.' for none.
Tokens: N, E, S, T, T0, P, Wnnn/W*/W%nnn, Mnnn/M*/M%nnn, ?n
QQ — quirks: concatenated single-letter flags, or '.' for none.
Flags: P, Z, I, U, X, A, T, F, D, !, K, Q, 0, R
OS — genre, optionally prefixed '-' (userland), '@' (group),
'*' (random/bogus), or combinations (e.g. '-@Windows').
Details — free-text flavor/version.
Lines starting with '#' and blank lines are skipped.
"""
from __future__ import annotations
import logging
import re
from pathlib import Path
from typing import Optional
from decnet.prober.osfp.p0f.signature import (
IntSpec,
OptionToken,
Signature,
WindowSpec,
precompute_specificity,
)
logger = logging.getLogger("decnet.prober.osfp.p0f.format")
_OPTION_TOKEN_RE = re.compile(r"^([NESTPE]|T0|[MW\?])(\*|%\d+|\d+)?$")
class P0fParseError(ValueError):
"""Raised on genuinely malformed signature lines. The loader
catches these and skips the offending line with a logger warning —
one bad row doesn't disable the whole DB."""
def parse_p0f_v2(path: Path) -> list[Signature]:
"""Parse a p0f v2 ``.fp`` file and return a list of Signatures.
Malformed lines are logged at WARNING and skipped rather than
aborting the whole load — the vendored DB has ~375 entries and one
corrupt row shouldn't prevent the other 374 from being usable.
"""
out: list[Signature] = []
with path.open("r", encoding="utf-8", errors="replace") as fh:
for lineno, raw in enumerate(fh, 1):
line = raw.strip()
if not line or line.startswith("#"):
continue
try:
sig = _parse_line(line)
except P0fParseError as exc:
logger.warning(
"p0f parse: skipping %s:%d — %s", path.name, lineno, exc,
)
continue
out.append(sig)
logger.debug("p0f parse: loaded %d signatures from %s", len(out), path.name)
return out
def _parse_line(line: str) -> Signature:
parts = line.split(":", 7)
if len(parts) < 7:
raise P0fParseError(f"expected 7+ colon-delimited fields, got {len(parts)}")
if len(parts) == 7:
parts = [*parts, ""] # empty details
wss_s, ttl_s, df_s, tot_s, opts_s, quirks_s, os_s, details = parts
wss = _parse_wss(wss_s)
ttl = _parse_int_field(ttl_s, field="ttl")
df = _parse_df(df_s)
total_len = _parse_int_spec(tot_s)
options = _parse_options(opts_s)
quirks = _parse_quirks(quirks_s)
os_name, is_userland, is_approx, is_random = _parse_os_genre(os_s)
sig = Signature(
wss=wss,
ttl=ttl,
df=df,
total_len=total_len,
options=options,
quirks=quirks,
os=os_name,
flavor=details.strip(),
notes="",
is_userland=is_userland,
is_approximate=is_approx,
is_random=is_random,
)
# Replace specificity (frozen dataclass field default) with the
# computed value via dataclasses.replace.
from dataclasses import replace
return replace(sig, specificity=precompute_specificity(sig))
def _parse_wss(s: str) -> WindowSpec:
s = s.strip()
if s == "*":
return WindowSpec("any")
if s.startswith("%"):
try:
return WindowSpec("mod", int(s[1:]))
except ValueError as exc:
raise P0fParseError(f"bad mod window {s!r}") from exc
if s.startswith("S"):
try:
return WindowSpec("mss_mul", int(s[1:]))
except ValueError as exc:
raise P0fParseError(f"bad Snn window {s!r}") from exc
if s.startswith("T"):
try:
return WindowSpec("mtu_mul", int(s[1:]))
except ValueError as exc:
raise P0fParseError(f"bad Tnn window {s!r}") from exc
try:
return WindowSpec("literal", int(s))
except ValueError as exc:
raise P0fParseError(f"bad literal window {s!r}") from exc
def _parse_int_field(s: str, *, field: str) -> int:
"""Parse a bare int field (used for TTL). No wildcards allowed."""
try:
return int(s.strip())
except ValueError as exc:
raise P0fParseError(f"bad {field}: {s!r}") from exc
def _parse_df(s: str) -> Optional[bool]:
s = s.strip()
if s == "*":
return None
if s == "0":
return False
if s == "1":
return True
raise P0fParseError(f"bad DF {s!r}; expected 0/1/*")
def _parse_int_spec(s: str) -> IntSpec:
s = s.strip()
if s == "*":
return IntSpec("any")
if s.startswith("%"):
try:
return IntSpec("mod", int(s[1:]))
except ValueError as exc:
raise P0fParseError(f"bad mod int {s!r}") from exc
try:
return IntSpec("literal", int(s))
except ValueError as exc:
raise P0fParseError(f"bad literal int {s!r}") from exc
def _parse_options(s: str) -> tuple[OptionToken, ...]:
s = s.strip()
if s in (".", ""):
return (OptionToken("."),)
normalized = s.replace(",", " ")
tokens: list[OptionToken] = []
for raw in normalized.split():
tok = raw.strip()
if not tok:
continue
tokens.append(_parse_option_token(tok))
if not tokens:
return (OptionToken("."),)
return tuple(tokens)
def _parse_option_token(raw: str) -> OptionToken:
# T0 — timestamp zero (not the TCP option '?0').
if raw == "T0":
return OptionToken("T0")
m = _OPTION_TOKEN_RE.match(raw)
if not m:
raise P0fParseError(f"bad option token {raw!r}")
kind, val_raw = m.group(1), m.group(2)
if kind in ("N", "E", "S", "T", "P"):
return OptionToken(kind)
# M / W / ? expect a numeric predicate (or wildcard).
if val_raw is None:
raise P0fParseError(f"option {kind!r} missing required value")
if val_raw == "*":
spec = IntSpec("any")
elif val_raw.startswith("%"):
try:
spec = IntSpec("mod", int(val_raw[1:]))
except ValueError as exc:
raise P0fParseError(f"bad {kind} mod value {val_raw!r}") from exc
else:
try:
spec = IntSpec("literal", int(val_raw))
except ValueError as exc:
raise P0fParseError(f"bad {kind} literal value {val_raw!r}") from exc
return OptionToken(kind, spec)
def _parse_quirks(s: str) -> frozenset[str]:
s = s.strip()
if s == "." or not s:
return frozenset()
# Quirks are a concatenated string of single-letter flags. '!' is a
# valid quirk too.
return frozenset(c for c in s if not c.isspace())
def _parse_os_genre(s: str) -> tuple[str, bool, bool, bool]:
"""Strip p0f's genre-prefix modifiers and return (os_name, is_userland, is_approx, is_random)."""
is_userland = False
is_approx = False
is_random = False
s = s.strip()
# Prefixes can stack in any order — strip them all.
changed = True
while changed and s:
changed = False
if s.startswith("-"):
is_userland = True
s = s[1:]
changed = True
elif s.startswith("@"):
is_approx = True
s = s[1:]
changed = True
elif s.startswith("*"):
is_random = True
s = s[1:]
changed = True
return s, is_userland, is_approx, is_random

109
decnet/prober/osfp/p0f/provider.py Normal file
View File

@@ -0,0 +1,109 @@
"""p0f v2 Provider — loads the vendored .fp databases and matches
against observed TCP quirks.
Four databases ship under ``data/``:
p0f.fp — SYN fingerprints (passive / sniffer-captured inbound).
p0fa.fp — SYN-ACK fingerprints (prober active-probe responses).
p0fr.fp — RST+ fingerprints (reset-response quirks).
p0fo.fp — "stray" packet fingerprints.
The provider routes incoming observations to the right sig list based
on ``obs["context"]`` — see :meth:`P0fV2Provider.match` — and returns
the highest-specificity matching :class:`OsMatch` or ``None``.
DECNET-authored additions can land in ``p0f-decnet.fp`` (same
directory, loaded if present) under GPL-3.0. None exist today — the
plan deferred writing any to a later commit — but the provider
already picks it up when it appears.
"""
from __future__ import annotations
import logging
from pathlib import Path
from typing import Any, Optional
from decnet.prober.osfp.base import OsMatch, Provider
from decnet.prober.osfp.p0f.format import parse_p0f_v2
from decnet.prober.osfp.p0f.signature import Signature
logger = logging.getLogger("decnet.prober.osfp.p0f.provider")
# Directory containing the vendored .fp files.
_DATA_DIR: Path = Path(__file__).resolve().parent / "data"
# Which .fp files feed each observation context.
_CONTEXT_DBS: dict[str, tuple[str, ...]] = {
"syn": ("p0f.fp", "p0f-decnet.fp"),
"synack": ("p0fa.fp",),
"rst": ("p0fr.fp",),
"stray": ("p0fo.fp",),
}
class P0fV2Provider(Provider):
"""Match observations against the p0f v2 database."""
name = "p0f-v2"
def __init__(self, data_dir: Optional[Path] = None) -> None:
self._data_dir = (data_dir or _DATA_DIR).resolve()
self._sigs_by_context: dict[str, list[Signature]] = {}
self._load()
def _load(self) -> None:
for context, filenames in _CONTEXT_DBS.items():
merged: list[Signature] = []
for name in filenames:
path = self._data_dir / name
if not path.is_file():
# p0f-decnet.fp is optional; all others are required.
if name.startswith("p0f-decnet"):
continue
logger.warning("p0f-v2: missing required DB file %s", path)
continue
try:
merged.extend(parse_p0f_v2(path))
except OSError as exc:
logger.warning("p0f-v2: could not load %s: %s", path, exc)
self._sigs_by_context[context] = merged
logger.debug("p0f-v2: %s context loaded %d signatures", context, len(merged))
def match(self, obs: dict[str, Any]) -> Optional[OsMatch]:
"""Return the highest-specificity matching signature, or None.
``obs["context"]`` selects the DB slice; default is "syn"
(passive observation, which is 80%+ of the event stream).
Invalid contexts return None rather than raising.
"""
context = obs.get("context", "syn")
sigs = self._sigs_by_context.get(context)
if not sigs:
return None
best: tuple[float, Signature] | None = None
for sig in sigs:
score = sig.score(obs)
if score is None:
continue
if best is None or score > best[0]:
best = (score, sig)
# Short-circuit on a perfect match — can't beat 1.0.
if best[0] >= 1.0:
break
if best is None:
return None
score, sig = best
return OsMatch(
os=sig.os,
flavor=sig.flavor,
confidence=score,
provider=self.name,
is_userland=sig.is_userland,
)
def signature_counts(self) -> dict[str, int]:
"""For diagnostics / tests — how many sigs loaded per context."""
return {ctx: len(sigs) for ctx, sigs in self._sigs_by_context.items()}

287
decnet/prober/osfp/p0f/signature.py Normal file
View File

@@ -0,0 +1,287 @@
"""p0f v2 signature + observation matching/scoring.
A :class:`Signature` is one parsed row from a ``.fp`` file. A match
against an observation dict (the kind ``sniffer_rollup`` hands us)
returns a confidence score in [0, 1], with higher scores indicating
more-specific matches. Wildcards and modulo predicates match but
contribute less to the confidence than an exact literal match, so
when multiple signatures fire against one observation we can pick the
most-specific one.
Observation dict shape (all keys optional — a provider returns None
if too few match-relevant fields are present):
{
"window": int | None, # TCP window size
"mss": int | None, # TCP MSS option value
"wscale": int | None, # TCP window-scale option value
"ttl": int | None, # initial-TTL bucket (32/64/128/255)
"df": bool | None, # IP Don't-Fragment flag
"total_len": int | None, # IP total length (SYN)
"options_sig": str | None, # e.g. "M,N,W,T" or "M1460,N,W7,S"
"quirks": frozenset[str] | None, # e.g. {"Z", "P"}
}
The scoring is our extension — upstream p0f is "first match wins"
using the order of entries in ``.fp``. We score so the factory can
compare across multiple DB files (p0f.fp + p0fa.fp) and return the
winner objectively.
"""
from __future__ import annotations
import re
from dataclasses import dataclass, field
from typing import Any, Optional
# ─── Field predicates ──────────────────────────────────────────────────────
@dataclass(frozen=True)
class WindowSpec:
"""Parsed 'wss' field. Encodes p0f v2's window-size predicate DSL:
- 'literal' → observed window == value
- 'mss_mul' → observed window == MSS * value (p0f "Snn")
- 'mtu_mul' → observed window == (MSS+40) * value (p0f "Tnn")
- 'mod' → observed window % value == 0 (p0f "%nnn")
- 'any' → wildcard (p0f "*")
"""
kind: str
value: Optional[int] = None
def matches(self, window: Optional[int], mss: Optional[int]) -> bool:
if self.kind == "any":
return True
if window is None:
return False
if self.kind == "literal":
return window == self.value
if self.kind == "mod":
return self.value is not None and self.value > 0 and (window % self.value == 0)
if self.kind == "mss_mul":
return mss is not None and self.value is not None and window == mss * self.value
if self.kind == "mtu_mul":
return mss is not None and self.value is not None and window == (mss + 40) * self.value
return False
@dataclass(frozen=True)
class IntSpec:
"""Wildcard-or-modulo int predicate, used for MSS / wscale / total_len."""
kind: str # 'literal' | 'mod' | 'any'
value: Optional[int] = None
def matches(self, observed: Optional[int]) -> bool:
if self.kind == "any":
return True
if observed is None:
return False
if self.kind == "literal":
return observed == self.value
if self.kind == "mod":
return self.value is not None and self.value > 0 and (observed % self.value == 0)
return False
@dataclass(frozen=True)
class OptionToken:
"""One TCP option as it appears in a signature's options list.
- kind='N' EOL 'E' SACK-permitted 'S' timestamp 'T' zero-timestamp 'T0'
- kind='M' MSS option, value = IntSpec
- kind='W' window-scale option, value = IntSpec
- kind='?' unknown option number, value = IntSpec (literal = option number)
- kind='.' no-options sentinel (singleton — matches only empty option list)
"""
kind: str
value: Optional[IntSpec] = None
def matches_literal(self, token: "OptionToken") -> bool:
"""True when *this* signature token matches an observed *token*.
Signature-side carries the wildcard/modulo predicate; observed
side is always a literal (or kind-only for flag options).
"""
if self.kind != token.kind:
return False
if self.value is None:
return True
if token.value is None:
return False
# Both have IntSpecs — match via predicate.
return self.value.matches(token.value.value)
# ─── Signature ─────────────────────────────────────────────────────────────
@dataclass(frozen=True)
class Signature:
"""One parsed row from a p0f v2 .fp file.
``label_prefix`` captures p0f's os-genre modifiers:
- ``-`` userland stack (not a real OS; flagged scanner/browser)
- ``@`` approximate / group match
- ``*`` random or bogus userland
These prefixes are stripped from ``os``; the flags survive here
for the profiler to decide e.g. "do I promote nmap to tool_guesses?"
"""
wss: WindowSpec
ttl: int
df: Optional[bool]
total_len: IntSpec
options: tuple[OptionToken, ...] # in order; use (OptionToken('.'),) for none
quirks: frozenset[str]
os: str
flavor: str
notes: str
is_userland: bool = False # '-' prefix
is_approximate: bool = False # '@' prefix
is_random: bool = False # '*' prefix (distinct from wildcard)
# Cache: a crude "specificity budget" precomputed at parse time.
# Higher = more constrained fields, used as a tie-breaker when two
# signatures match the same observation.
specificity: float = field(default=0.0)
def score(self, obs: dict[str, Any]) -> Optional[float]:
"""Return a confidence in [0, 1] on match, or None if any field
rejects the observation.
Soft-field semantics: ``df`` and ``total_len`` are treated as
"skip check when observation is missing" — the sniffer doesn't
currently emit either, and a literal-constraint sig shouldn't
reject a match solely because the observation is upstream-
incomplete. Hard fields (``window``, ``ttl``, ``options_sig``,
``quirks``) still hard-reject on absent or mismatched input —
those are the real discriminators."""
mss = obs.get("mss")
# Window (hard)
if not self.wss.matches(obs.get("window"), mss):
return None
# TTL — initial-TTL bucket must match exactly. The profiler is
# expected to have rounded the observed TTL up to the nearest
# bucket already via decnet.sniffer.p0f.initial_ttl. (hard)
obs_ttl = obs.get("ttl")
if obs_ttl is None or obs_ttl != self.ttl:
return None
# DF (soft — skip when unknown)
if self.df is not None:
obs_df = obs.get("df")
if obs_df is not None and bool(obs_df) != self.df:
return None
# Total length (soft — skip when unknown)
obs_total = obs.get("total_len")
if obs_total is not None and not self.total_len.matches(obs_total):
return None
# Options (hard)
if not _options_match(self.options, obs.get("options_sig")):
return None
# Quirks — must match as a set. (hard)
obs_quirks = obs.get("quirks") or frozenset()
if not isinstance(obs_quirks, frozenset):
obs_quirks = frozenset(obs_quirks)
if self.quirks != obs_quirks:
return None
# All fields matched — return the precomputed specificity.
return self.specificity
def _options_match(sig_opts: tuple[OptionToken, ...], obs_sig: Optional[str]) -> bool:
"""Match signature option sequence against observation's comma/space-
separated option string."""
obs_tokens = _parse_observation_options(obs_sig)
# Special case: signature is '.' (no-options sentinel).
if len(sig_opts) == 1 and sig_opts[0].kind == ".":
return len(obs_tokens) == 0
if len(sig_opts) != len(obs_tokens):
return False
return all(s.matches_literal(o) for s, o in zip(sig_opts, obs_tokens))
_OBS_TOKEN_RE = re.compile(r"^([A-Z\?])(\d+)?$")
def _parse_observation_options(opts_sig: Optional[str]) -> list[OptionToken]:
"""Convert the observation-side options string (from
tcp_syn_fingerprint / tcpfp_fingerprint SD fields) into a list of
literal OptionTokens. Accepts comma or space delimiters and tokens
like 'M1460', 'W7', 'T', 'T0', 'N', 'E', '?47'.
"""
if not opts_sig:
return []
normalized = opts_sig.replace(",", " ")
out: list[OptionToken] = []
for raw in normalized.split():
token = raw.strip()
if not token:
continue
if token == "T0": # nosec B105 — TCP option name ("Timestamp zero"), not a credential
out.append(OptionToken("T0"))
continue
m = _OBS_TOKEN_RE.match(token)
if not m:
# Unknown token — represent as opaque "?" with no value so
# nothing matches it. Better than raising.
out.append(OptionToken("?", IntSpec("literal", -1)))
continue
kind, num = m.group(1), m.group(2)
if num is None:
out.append(OptionToken(kind))
else:
out.append(OptionToken(kind, IntSpec("literal", int(num))))
return out
def precompute_specificity(sig: Signature) -> float:
"""Crude specificity score used when comparing matching signatures.
Each field contributes a weight; wildcards and modulo predicates
contribute less. Tuned so a fully-literal signature scores ~1.0 and
a near-wildcard signature scores ~0.1.
"""
w = 0.0
total = 0.0
# Window (weight 3 — very discriminating)
total += 3
if sig.wss.kind == "literal":
w += 3.0
elif sig.wss.kind in ("mss_mul", "mtu_mul"):
w += 2.5
elif sig.wss.kind == "mod":
w += 1.5
# TTL — always literal, contributes a flat 1
total += 1
w += 1.0
# DF (weight 1)
total += 1
if sig.df is not None:
w += 1.0
# Total length (weight 1)
total += 1
if sig.total_len.kind == "literal":
w += 1.0
elif sig.total_len.kind == "mod":
w += 0.5
# Options (weight 3 — highly discriminating when literal)
total += 3
if not (len(sig.options) == 1 and sig.options[0].kind == "."):
literal_opts = sum(
1 for o in sig.options
if o.value is None or o.value.kind == "literal"
)
if sig.options:
w += 3.0 * (literal_opts / len(sig.options))
else:
# "no options" is itself a signal.
w += 2.0
# Quirks (weight 1 — most sigs have no quirks so this is a small edge)
total += 1
if sig.quirks:
w += 1.0
return round(w / total, 4)