waugh@dg-rtp.dg.com (Matthew Waugh) (09/27/89)
We have a number of Sun workstations that have started to report the following errors: le0: Received packet with ENP bit in rmd cleared le0: Received packet with STP bit in rmd cleared This means nothing to me - does anybody know what particular protocol problem it's trying to tell me about. This has just started to happen, so that leads me to believe something else changed on the wire and upset these machines, but since I don't know what the Sun thinks has been violated, it's hard to know where to start looking. Please e-mail me any responses, and thanks for any help, Mat ------------------------------------------------------------------------ Matthew Waugh waugh@dg-rtp.dg.com RTP Network Services {world}!mcnc!rti!dg-rtp!waugh Data General Corp. RTP, NC. (919)-248-6344
hwajin@wrs.wrs.com (Hwajin Bae) (09/29/89)
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