home *** CD-ROM | disk | FTP | other *** search
- Path: news.ultranet.com!usenet
- From: "Albert P. Belle Isle" <belleisl@cerberus-sys.com>
- Newsgroups: alt.winsock.trumpet,alt.winsock,comp.dcom.modems
- Subject: Re: Frame check error
- Date: Fri, 15 Mar 1996 12:07:10 -0500
- Organization: Cerberus Systems, Inc.
- Message-ID: <3149A3BE.7CD2@cerberus-sys.com>
- References: <4hhg08$c2t@mcmail.CIS.McMaster.CA> <4iaa9h$ntb@server.cntfl.com>
- NNTP-Posting-Host: apb-p5-90.cerberus-sys.com
- Mime-Version: 1.0
- Content-Type: text/plain; charset=us-ascii
- Content-Transfer-Encoding: 7bit
- X-Mailer: Mozilla 2.01 (Win16; I)
-
- John Farrell wrote:
- >
- > I have the same problem, and I'm using a 28.8 USR on a 486/66 clone.
- > I know, in a general way, what causes the error, what I'm wondering is
- > how to fix it. Or if it is unfixable. I have tried adding some stuff
- > to my systeml.ini file, and to my config.sys; but it didn't seem to do
- > much good. My settings are MTU 1500, TCP RWIN 4096, TCP MSS 1460,
- > and I have the baud rate on trumpet setup at 38400. Would messing
- > around with the com port settings in control panel help any? I tried
- > increasing the baud rate there, but it didn't seem to make any
- > difference, and I even messed around with the timeslice and that other
- > stuff in there, but being a civil engineer all I know how to do is
- > design a bridge, not fix my telcom stuff.
- >
- > I've read that the proper UART will help, but I thought that came on
- > the card with the USR 28.8, also I have heard that one needs to be
- > sure that the modem is set up for error correction. How?
- >
- > I've read that this is a problem caused by the way windows works. Can
- > it be cured on this machine, or on any machine running 3.1?
- >
- > What about windows 95?
-
- John:
-
- Many of the settings in your TCP/IP stack deal with things like the
- maximum number of data bytes that should be sent to you in a single
- packet (a TCP/IP transaction).
-
- In a dial-in stack (like Trumpet WinSock) there is also a mechanism for
- fooling the TCP/IP part of the stack into thinking that the packets full
- of bytes it finds in the area of your RAM that it uses for buffers got there
- from a direct TCP/IP connection to the Internet - even though they really
- came through your RS232 ("com") port, after being transmitted through a pair
- of modems communicating over an analog telephone connection. The lower-level
- protocol for doing this is either PPP or one of the versions of SLIP.
-
- PPP wraps 1500-byte IP transmission units, along with a check-sum, into
- data "frames" which are sent over your access provider's RS232 serial
- port, through his modem to your modem, then through your serial port into
- the portion of your machine's memory containing the buffers TCPMan(ager)
- has allocated as the logical "socket" for a particular TCP/IP-using
- application program.
-
- This "data link layer" underneath the TCP/IP (packet) transactions must
- flawlessly send every byte, without errors, if the packet is to be
- successfully reassembled in your buffers. If it isn't, TCP will wait for
- a re-transmission, wasting a lot of time. UDP, which doesn't account for
- each byte and is used for things like RealAudio, will just loose the packet.
-
- If the protocol you're using to simulate the direct TCP/IP connection is
- the Point-to-Point Protocol, you'll see "PPP Frame Check Error" in PPP
- Trace, because the check-sum your machine computed didn't match the one
- included with the data bytes in the PPP frame.
-
- One source of data link layer errors is individual byte-garbling by noise
- on the phone line that's not corrected by your modem. Be sure that your
- modem's init string has not turned-off its V42 error correction.
-
- A more common cause of data link layer errors is when your modem, having
- been promised by your com port settings that your CPU will empty bytes
- from the UART at a particular rate, accepts more bytes from your ISP's
- modem and writes them on top of the ones your CPU still hasn't gotten
- around to emptying from the UART. This is called a com overrun error.
-
- In any case, the TCP layer is a user of what comes over the data link layer,
- and its parameter settings (MSS/RWIN/RTOmax) have no more influence on
- this problem than changing the size of the viewing window in the browser
- software that uses the results of the TCP transactions. You have to get
- rid of the byte-dropping (overruns) or byte-garbling (modem symbol errors)
- to make a solid data link layer, before trying to run higher-layer software
- on top of it.
-
- I've tried to elaborate on this in the sections of the below-listed FAQ
- entitled "MTU/MSS/RWIN vs Com Overrun Errors" and "SLIP, PPP and Frame
- Check Errors," if you're interested in more details. The section entitled
- "Com Port Overrun Errors" deals with some solutions you might try.
-
- Hope this helps.
-
- Regards,
-
- Al
-
-
-
- --
- ==================================================================
- Albert P. Belle Isle
- Cerberus Systems, Inc.
-
- Al's Winsock Tuning FAQ -
- http://www.cerberus-sys.com/~belleisl/mtu_mss_rwin.html
- ==================================================================
-