Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!purdue!bu-cs!oliveb!amdahl!rtech!wrs!hwajin From: hwajin@wrs.wrs.com (Hwajin Bae) Newsgroups: comp.dcom.lans Subject: Re: Sun interfaces - strange error report Message-ID: <766@wrs.wrs.com> Date: 28 Sep 89 20:51:55 GMT References: <1366@xyzzy.UUCP> Reply-To: hwajin@wrs.wrs.com (Hwajin Bae) Distribution: usa Organization: Wind River Systems, Emeryville, CA Lines: 21 In article <1366@xyzzy.UUCP> waugh@dg-rtp.dg.com (Matthew Waugh) writes: > le0: Received packet with ENP bit in rmd cleared > le0: Received packet with STP bit in rmd cleared When an ethernet frame received cannot fit into a single buffer in the receive ring, LANCE will try to use data-chaining to use multiple buffers for a packet. The first one will have the STP (start-of-packet) bit set to 1 and the last one will have the ENP (end-of-packet) bit set. Only the last one will have the valid ethernet frame length in the RMD. One common problem with most LANCE drivers is that they assume that there won't be any data-chaining initiated by LANCE. They usually drop packets that doesn't have both STP and ENP bits set and not handle the input data-chaing initiated by LANCE at all. This works well only if each of your RMD buffer is big enough to hold maximum ethernet frame size (1516). If your RMD buffer unit size is smaller than that -- say 1024, LANCE will occasionally try to use data-chaining. If your driver doesn't handle data-chaining on the input side you will get these errors. -- Hwajin Bae Wind River Systems, Emeryville, CA