[Snort-devel] [snort-devel] Chainning pre-processors

Emiliano Fausto emiliano.fausto at ...2499...
Thu Dec 5 13:40:30 EST 2013


Hui,

I was really trying to get Stream5 working and I couldn't, I'm pretty sure
that the problem is not related to Stream5, but I just can't get the
preprocessor to rebuilt TCP segments.

In wireshark I can see there are lot of TCP segmented packets and if I
select to reassemble the HTTP ones, it does it well. Backing to SRNOT, for
some reason the statistics that the Stream5 are showing are that there
where some TCP sessions, there were discarded TCP packets, etc. But in the
line: "TCP Rebuilt packets" shows a 0.

I tried with HTTP so that I can have a known protocol to start with as you
suggested me, and tried to download a wallpaper, which (due to its size)
was delivered to me in several packets.

I changed the snort.conf configuration file, to look like this:

preprocessor stream5_global: track_tcp yes, \
    track_udp no, \
    track_icmp no, \
    max_tcp 262144, \
    max_active_responses 2, \
    min_response_seconds 5, \
    show_rebuilt_packets
preprocessor stream5_tcp: policy linux, \
    overlap_limit 10, small_segments 3 bytes 150, timeout 180, \
    ports both all

I also played a little bit with my own preprocessor to print the Flags that
the packet brings, and they change based on the priority I assign to the
AddPreproc(). But I tried different ones, from FIRST to LAST and none of
them matched the & FLAG_REBUILT_STREAM flag.

I was looking at the stream5 documentation (the README file), also searched
different threads in the snort users and devel community, but none of them
helped me.

Do you know where can I get more information or examples on how to use
properly the Stream5 preprocess, or if I'm missing something else to make
it works? maybe create a generic rule, or something like that?

I really thank you in advance for all your help on this,
Emiliano.



2013/12/5 Hui Cao <hcao at ...402...>

>  Make sure you have frag3 enabled. You can try the application
> preprocessors (such as http, smtp, ssl, sip etc) to see how it works.  If
> the packets are reassembled, you can see that from snort exit stats for
> stream etc.
>
> Stream will reassemble the payloads of TCP.
>
> Best,
> Hui.
>
>
> On 12/05/2013 08:26 AM, Emiliano Fausto wrote:
>
> Hello Hui,
>
>  you are totally right. I was confused, I didn't have to assemble
> fragmented IP packets.
>
>  So, I configured the Stream5 preprocessor in order it to take into
> account the port of my application (it's 9090).
>
>  I let my AddPreproc() with priority set to PRIORITY_APPLICATION so that
> it's registered after the Stream5.
>
>  Then, I tried to get any packet with (p->flags & FLAG_REBUILT_STREAM),
> but none of them matched, it seems that the Stream5 preprocessor isn't
> reassembling my TCP segments.
>
>  The SNORT configuration file is set like this:
>
>  preprocessor stream5_global: track_tcp yes, \
>     track_udp yes, \
>     track_icmp no, \
>     max_tcp 262144, \
>     max_udp 131072, \
>     max_active_responses 2, \
>     min_response_seconds 5
> preprocessor stream5_tcp: policy windows, detect_anomalies, require_3whs
> 180, \
>     overlap_limit 10, small_segments 3 bytes 150, timeout 180, \
>     ports client 21 22, \
>     ports both 9090
> preprocessor stream5_udp: timeout 180
>
>  My app works on port 9090.
>
>  Maybe, the problem is not related to stream5, maybe the problem is
> originated because the one in charge of doing the segments reassembly is my
> preprocessor, at the application level, and not the stream5 preprocessor at
> the TCP level.
>
>  More on this, into wireshark I can see all the packets of the protocol
> TCP segmented, and if I go to preferences-->protocols-->TCP it has an
> option that says: "allow dissectors to reassemble TCP". It means that the
> reassemble isn't being done at the TCP layer, but into the proper
> application.
>
>  Do you know if SNORT has something of this? Or the Stream5 should
> reassemble the payloads of the TCP packets?
>
>  Thanks a lot in advance,
> Emiliano.
>
>
> 2013/12/4 Hui Cao <hcao at ...402...>
>
>>  Frag3 will deal with IP fragmentation. You might want reassembled data
>> from TCP. This means your preprocessor should be after stream and you
>> should check
>> (SFSnortPacket*)ipacketp)->flags & FLAG_REBUILT_STREAM. Snort will only
>> reassemble up to the flush point defined (typically 16K). You can't expect
>> snort will give all the data from one reassembled packet.
>>
>> Fragmentation from wireshark might mean TCP segmentation. You might get
>> full IP packets.
>>
>> Best,
>> Hui.
>>  On 12/04/2013 03:17 PM, Emiliano Fausto wrote:
>>
>> Hui,
>>
>>  yes, I understood what you told me about filtering just the
>> de-fragmented packets, that's why I added these 3 lines:
>>
>>  SFSnortPacket *p = (SFSnortPacket *)pkt;
>> int fla = (p->flags & FLAG_REBUILT_FRAG);
>> _dpd.logMsg("Flags set? %d",fla);
>>
>>  But I generate fragmented packets (which for instance Wireshark detects
>> and show as fragmented and the possibility of reassembly them) and I'm
>> always getting with each one the output: "Flags set? 0".
>>
>>  I receive all the fragmented packets, but not the last one reassembled.
>>
>>  Regards,
>> Emiliano.
>>
>>
>> 2013/12/4 Hui Cao <hcao at ...402...>
>>
>>> It will still get all packets including the raw packets. You need to
>>> use packet flags to filter them.
>>>
>>> Best,
>>> Hui.
>>>
>>> On Wed, Dec 4, 2013 at 3:00 PM, Emiliano Fausto
>>>  <emiliano.fausto at ...2499...> wrote:
>>> > Yes,
>>> >
>>> > that's what I thought, but for some reason the TCP packets keep on
>>> coming
>>> > fragmented to my preprocessor.
>>> >
>>> > No I took off all the preprocessors and just let the frag3 and mines,
>>> I'll
>>> > try to figure out if they are called in the correct order, but they
>>> are not
>>> > being chained.
>>> >
>>> > Thanks,
>>> > Emiliano.
>>> >
>>> >
>>> > 2013/12/4 Hui Cao <hcao at ...402...>
>>> >>
>>> >> It looks good to me.
>>> >>
>>> >> Best,
>>> >> Hui.
>>> >>
>>> >> On Wed, Dec 4, 2013 at 2:44 PM, Emiliano Fausto
>>> >> <emiliano.fausto at ...2499...> wrote:
>>> >> > Hi Hui,
>>> >> >
>>> >> > I've seen that I was using PRIORITY_TRANSPORT, which is lower than
>>> the
>>> >> > PRIORITY_NETWORK that uses frag3.
>>> >> >
>>> >> > Anyway, I put the priority: PRIORITY_LAST to my own preprocessor,
>>> but
>>> >> > when
>>> >> > the TCP packets keep arriving fragmented to my preprocessor.
>>> >> >
>>> >> > Is there anything else I should take into account?
>>> >> >
>>> >> > Thanks in advance,
>>> >> > Emiliano.
>>> >> >
>>> >> >
>>> >> > 2013/12/4 Hui Cao <hcao at ...402...>
>>> >> >>
>>> >> >> In sr/preprocids.h
>>> >> >>
>>> >> >> Best,
>>> >> >> Hui.
>>> >> >>
>>> >> >> On 12/04/2013 02:36 PM, Emiliano Fausto wrote:
>>> >> >>
>>> >> >> Great,
>>> >> >>
>>> >> >> so, the pre-processors are "chained" by default, and the order that
>>> >> >> SNORT
>>> >> >> follows to call them is set by the PRIORITY variable.
>>> >> >>
>>> >> >> Do you know where's defined this PRIORITY variable? Because I saw
>>> that
>>> >> >> the
>>> >> >> frag3 is being registered with PRIORITY_NETWORK, so I'd like to
>>> set the
>>> >> >> priority of my own preprocessor as (PRIORITY_NETWORK -1).
>>> >> >>
>>> >> >> Thanks in advance,
>>> >> >> Emiliano
>>> >> >>
>>> >> >>
>>> >> >> 2013/12/4 Hui Cao <hcao at ...402...>
>>> >> >>>
>>> >> >>> sc means snort configuration. We use PRIORITY to sort the
>>> processing.
>>> >> >>> All
>>> >> >>> processors enabled will be called and processed based on
>>> priority. You
>>> >> >>> have
>>> >> >>> to rely on the code to figure out what exactly snort does.
>>> >> >>>
>>> >> >>> The checking is correct. You will only process rebuilt packets.
>>> >> >>>
>>> >> >>> Best,
>>> >> >>> Hui.
>>> >> >>> On 12/04/2013 02:19 PM, Emiliano Fausto wrote:
>>> >> >>>
>>> >> >>> Hello Hui,
>>> >> >>>
>>> >> >>> thanks a lot for your answer.
>>> >> >>>
>>> >> >>> Right now I have registered my preprocessor (let's call it
>>> >> >>> examplePreprocess as you said, because right now I'm using the one
>>> >> >>> provided
>>> >> >>> with the DPX) with this line:
>>> >> >>>
>>> >> >>> _dpd.addPreproc(ExampleProcess, PRIORITY_TRANSPORT, 10000,
>>> >> >>> PROTO_BIT__TCP);
>>> >> >>>
>>> >> >>> So, the only change is to add previous to the parameter
>>> >> >>> ExampleProcess,
>>> >> >>> the "sc". What does it mean? Do you know if there's any
>>> documentation
>>> >> >>> about
>>> >> >>> this chaining preprocesses?
>>> >> >>>
>>> >> >>> So, checking the flags, should be:
>>> >> >>>
>>> >> >>> (SFSnortPacket*)tcppacket)->flags & FLAG_REBUILT_FRAG
>>> >> >>>
>>> >> >>> right?
>>> >> >>>
>>> >> >>> Thanks again!
>>> >> >>> Emiliano.
>>> >> >>>
>>> >> >>>
>>> >> >>>
>>> >> >>> Then, I'll have to register my own preprocessor where?
>>> >> >>>
>>> >> >>>
>>> >> >>> 2013/12/4 Hui Cao <hcao at ...402...>
>>> >> >>>>
>>> >> >>>> Yes, it is possible. You can register you preprocessor like this:
>>> >> >>>>
>>> >> >>>> _dpd.addPreproc( sc, ExampleProcess, PRIORITY_TRANSPORT,
>>> You_PP_ID,
>>> >> >>>> PROTO_BIT__IP );
>>> >> >>>>
>>> >> >>>> Remember check the following flag in your ExampleProcess:
>>> >> >>>>
>>> >> >>>> (SFSnortPacket*)ipacketp)->flags & FLAG_REBUILT_FRAG
>>> >> >>>>
>>> >> >>>> Best,
>>> >> >>>> Hui.
>>> >> >>>>
>>> >> >>>>
>>> >> >>>> On 12/04/2013 12:52 PM, Emiliano Fausto wrote:
>>> >> >>>>
>>> >> >>>> Hi everybody,
>>> >> >>>>
>>> >> >>>> I'm creating a new preprocessor which needs to have the whole
>>> content
>>> >> >>>> in
>>> >> >>>> a packet which was fragmented.
>>> >> >>>>
>>> >> >>>> So I thought of using the frag3 preprocessor to re-assembly the
>>> >> >>>> packets,
>>> >> >>>> and then, when this reassembly is done, sent it to my own
>>> >> >>>> preprocessor.
>>> >> >>>>
>>> >> >>>> Do you know if this is possible? May I have the output of frag3
>>> being
>>> >> >>>> the input of my own preprocessor?
>>> >> >>>>
>>> >> >>>> Regards,
>>> >> >>>> Emiliano.
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> ------------------------------------------------------------------------------
>>> >> >>>> Sponsored by Intel(R) XDK
>>> >> >>>> Develop, test and display web and hybrid apps with a single code
>>> >> >>>> base.
>>> >> >>>> Download it for free now!
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> >> >>>> _______________________________________________
>>> >> >>>> Snort-devel mailing list
>>> >> >>>> Snort-devel at lists.sourceforge.net
>>> >> >>>> https://lists.sourceforge.net/lists/listinfo/snort-devel
>>> >> >>>> Archive:
>>> >> >>>>
>>> http://sourceforge.net/mailarchive/forum.php?forum_name=snort-devel
>>> >> >>>>
>>> >> >>>> Please visit http://blog.snort.org for the latest news about
>>> Snort!
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> ------------------------------------------------------------------------------
>>> >> >>>> Sponsored by Intel(R) XDK
>>> >> >>>> Develop, test and display web and hybrid apps with a single code
>>> >> >>>> base.
>>> >> >>>> Download it for free now!
>>> >> >>>>
>>> >> >>>>
>>> >> >>>>
>>> http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
>>> >> >>>> _______________________________________________
>>> >> >>>> Snort-devel mailing list
>>> >> >>>> Snort-devel at lists.sourceforge.net
>>> >> >>>> https://lists.sourceforge.net/lists/listinfo/snort-devel
>>> >> >>>> Archive:
>>> >> >>>>
>>> http://sourceforge.net/mailarchive/forum.php?forum_name=snort-devel
>>> >> >>>>
>>> >> >>>> Please visit http://blog.snort.org for the latest news about
>>> Snort!
>>> >> >>>
>>> >> >>>
>>> >> >>>
>>> >> >>
>>> >> >>
>>> >> >
>>> >
>>> >
>>>
>>
>>
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.snort.org/pipermail/snort-devel/attachments/20131205/e0b6cc22/attachment.html>


More information about the Snort-devel mailing list