[3825] in linux-net channel archive
Re: bsd_comp errors
daemon@ATHENA.MIT.EDU (Lex Spoon)
Wed Jul 24 01:07:41 1996
To: aonline@access.tnet.com.au (Michael Slater)
Date: Mon, 22 Jul 1996 17:39:07 -0400 (EDT)
Cc: john@nlc.net.au, linux-net@vger.rutgers.edu
In-Reply-To: <Pine.LNX.3.91.960720123847.13288A-100000@access.tnet.com.au> from "Michael Slater" at Jul 20, 96 12:42:19 pm
Reply-To: sspoon@clemson.edu
From: "Lex Spoon" <sspoon@clemson.edu>
Michael Slater wrote:
> On 17 Jul 1996, John Saunders wrote:
> > Graham Mitchell (gmitch@woodlea.wintermute.co.uk) wrote:
> > >
> > > Jul 5 03:53:20 woodlea linux: bsd_decomp0: bad sequence # 4435, expected 4430
> > > Jul 5 19:02:50 woodlea linux: bsd_decomp0: peer should have cleared dictionary on 5905
> > > Jul 5 19:02:50 woodlea linux: bsd_decomp0: bad code 0x180 oldcode=0x4e max_ent=0x100 explen=4 seqno=5907
> > > Jul 5 19:02:51 woodlea pppd[3822]: Lost compression sync: disabling compression
> > >
(not sure I got the attributions right, but everyone agreed
they are getting bsd_comp errors when using PPP between to
Linux boxes).
For the record, it occasionally happens between Linux and slirp-1.0c, also.
Most frustrating. Looking in the source one time (yikes -- that stuff
was well over my head), I found this lovely comment:
/*
* Decompress "BSD Compress".
*
* Because of patent problems, we return DECOMP_ERROR for errors
* found by inspecting the input data and for system problems, but
* DECOMP_FATALERROR for any errors which could possibly be said to
* be being detected "after" decompression. For DECOMP_ERROR,
* we can issue a CCP reset-request; for DECOMP_FATALERROR, we may be
* infringing a patent of Motorola's if we do, so we take CCP down
* instead.
*
In the following function, there are many returns of DECOMP_FATALERROR,
one of which is accompanied by a printk() matchting the "bad code" message
above.
By the way, how can such an idea be patented???? is it that hard
a concept to reset a system when it crashes? Anyone who has ever
used Microsoft Windows knows this! Hmm, maybe I'll go for a patent
on opening the door back up after I close it on my finger.
Final note:
For slirp users, "just let the modem do the compression" is nothing
like BSD compression is awesome; the bottom line is that
it allows reasonable performance of multiple sessions when one of the
sessions is hogging bandwidth. Even though it means turning the modem
off and redialing every once in a while, I still choose to use
BSD compression.
lex
