Wireshark-dev: Re: [Wireshark-dev] Wireshark-dev Digest, Vol 23, Issue 11
From: Philip Frey1 <PHF@xxxxxxxxxxxxxx>
Date: Tue, 8 Apr 2008 09:10:43 +0200
Hi Jaap,
thanks for the pointer. I have created a new bug report and added the diff there.
I hope it worked. Can you see my request?
Many thanks,
Philip
"Jaap Keuter"
<jaap.keuter@xxxxxxxxx>
Sent by: wireshark-dev-bounces@xxxxxxxxxxxxx 04.04.2008 20:28
|
|
Hi,
Please adhere to the patch submission policy: http://wiki.wireshark.org/SendingFilesToWireshark
That ensures things are picked up and tracked.
Thanx,
Jaap
From: wireshark-dev-bounces@xxxxxxxxxxxxx [mailto:wireshark-dev-bounces@xxxxxxxxxxxxx] On Behalf Of Philip Frey1
Sent: 2008-Apr-04 18:23
To: wireshark-dev@xxxxxxxxxxxxx
Cc: geyves@xxxxxxxxxxxxxxx
Subject: Re: [Wireshark-dev] Wireshark-dev Digest, Vol 23, Issue 11
Hi Guy,
> Message: 2
> Date: Fri, 04 Apr 2008 02:17:34 -0700
> From: Guy Harris <guy@xxxxxxxxxxxx>
> Subject: Re: [Wireshark-dev] Share CRC32C code between MPA and iSCSI
> To: Developer support list for Wireshark <wireshark-dev@xxxxxxxxxxxxx>
> Cc: markb@xxxxxxxxxx, geyves@xxxxxxxxxxxxxxx
> Message-ID: <47F5F22E.2040102@xxxxxxxxxxxx>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> Philip Frey1 wrote:
>
> > we have developed a dissector for the MPA (Marker PDU Aligned Framing)
> > protocol which is part of the iWARP stack.
> > Since we also need to do some CRC32C calculations (same as iSCSI),
>
> ...and, it appears, SCTP.
Good catch.
> The SCTP dissector has the same CRC table; its CRC routine starts with
> 0xFFFFFFFF - which, when byte-swapped, is unchanged - and then adds in
> to the CRC a bunch of bytes and 0's, flips the bits in the result, and
> then swaps the bytes. Your CRC32 routine doesn't flip the bits, so the
> right way to combine them might be to:
>
> export the CRC32 table from crc32c.c, declaring it in crc32.h;
>
> moving the CRC32C macro from packet-sctp.c to crc32.h;
>
> moving the CRC32C_SWAP macro from crc32c.c to crc32.h (or see whether
> GLib 1.2.x and 2.x define something equivalent, and use that);
>
> have calculate_crc32() use the CRC32C macro:
>
> guint32 calculate_crc32(const void *buf, int len, guint32 crc)
> {
> const guint8 *p = (const guint8 *)buf;
> crc = CRC32C_SWAP(crc);
> while (len-- > 0) {
> CRC32C(crc, *p++);
> }
> return CRC32C_SWAP(crc);
> }
>
> have sctp_crc32c() use CRC32C_PRELOAD instead of ~0L.
>
> There might be some other merging that could be done.
I have implemented exactly as you suggested (assuming I understood your suggestion
correctly). :-D
Please find attached the modified files:
(I merged everything into the crc32-files and got rid of the crc32c files)
If that is what you had in mind, then I will contribute it by sending my diff
to the bug-list of wireshark (as suggested in the README.developer).
Cheers,
Phil_______________________________________________
Wireshark-dev mailing list
Wireshark-dev@xxxxxxxxxxxxx
http://www.wireshark.org/mailman/listinfo/wireshark-dev
- References:
- Re: [Wireshark-dev] Wireshark-dev Digest, Vol 23, Issue 11
- From: Jaap Keuter
- Re: [Wireshark-dev] Wireshark-dev Digest, Vol 23, Issue 11
- Prev by Date: Re: [Wireshark-dev] Wireshark Portable
- Next by Date: [Wireshark-dev] multiple PDU's
- Previous by thread: Re: [Wireshark-dev] Wireshark-dev Digest, Vol 23, Issue 11
- Next by thread: Re: [Wireshark-dev] Is there anyway to control what will be display in the "protocol column"?
- Index(es):