Date: Sat, 27 Sep 1997 09:16:35 -0700 (PDT) From: "Rodney W. Grimes" <rgrimes@GndRsh.aac.dev.com> To: Don.Lewis@tsc.tdk.com (Don Lewis) Cc: dg@root.com, tlambert@primenet.com, tarkhil@mgt.msk.ru, hackers@FreeBSD.ORG, stable@FreeBSD.ORG Subject: Re: 'fxp' driver/hardware lossage (was Re: Alexander B. Povol's mail) Message-ID: <199709271616.JAA20171@GndRsh.aac.dev.com> In-Reply-To: <199709270811.BAA00487@salsa.gv.tsc.tdk.com> from Don Lewis at "Sep 27, 97 01:11:51 am"
next in thread | previous in thread | raw e-mail | index | archive | help
> On Sep 26, 11:10pm, "Rodney W. Grimes" wrote: > } Subject: Re: 'fxp' driver/hardware lossage (was Re: Alexander B. Povol's m > } > You can't send a packet to yourself under normal circumstances (ethernet > } > is a simplex device). > } > } Only broken NIC chips are simplex, even the 82586 is actually capable > } of hering itself talk on the wire, though most drives do not set > } the chip into this mode. Either way ``ethernet is _not_ a simplex > } device''. > > That's only true of coaxial media. Over twisted pair, if you see > data start arriving on the receive pair while your transmitting then > you have to assume that the data is being sent by another station > and you've got a collision situation. Add ``in half duplex mode'' and the above holds water. But not for full duplex mode. My statement still holds true ``ethernet is _NOT_ a simplex device''. Why do you the the driver flag ``SIMPLEX'' exists, to deal with the fact that some ethernet interfaces on some systems do infact actuall here them selves talk, infact that use to be pretty much the norm back in the days of VAXEN. Here is another one to think about.... ``If you see data arriving on the receive pair _AND_ the mac source address is not yours you have a collision, or if you are talking to a full duplex port in full duplex mode you are using it.'' These types of designs preclude the use of the cheap TP<->AUI converter chips that just trash a 10Mhz signal onto CD when they see RD go active with XMTD active. Does Intel buy chance use something as stupid as the old 82503 or 82506 for the 10Mb TP interface? > However, you may have the option > to loop back your transmitted data to your receiver after the receiver > stage where collision sensing is done. I believe this is known as > natural loopback. I don't think it's commonly used. More common that looping it after the TP interface chip, and often implemented right in the NIC chip itself. -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation, Inc. Reliable computers for FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199709271616.JAA20171>