Date: Sat, 16 Dec 1995 06:48:22 +1100 From: Bruce Evans <bde@zeta.org.au> To: bde@zeta.org.au, nate@rocky.sri.MT.net Cc: FreeBSD-hackers@FreeBSD.org Subject: Re: sio overruns on 2.0.5R Message-ID: <199512151948.GAA28090@godzilla.zeta.org.au>
next in thread | raw e-mail | index | archive | help
>> >> After I switched to a Asus P55TP4XE 90Mc I see things like >> >> >> >> sio0: 5 more silo overflows (total 7) >> >> >I hate to sound like a broken record, but are you *sure* you're using >> >flow-control? Can you use stty on the connected end to make sure? >> >> Er, flow control has no effect on silo overflows. >Really? If a machine doesn't provide flow control can't you over-run >the buffers if the machine gets loaded down and the interrupt load gets >so bad that the sio interrupt handler isn't called enough. However, if If the interrupt load gets so bad that the sio interrupt handler isn't called enough, then the interrupt handler won't get called enough to control the flow. This is for 8250-16550's; other chips are better. Perhaps more importantly, the sender may not respond to the flow control until many characters later, so either the reciever's hardware must have enough buffering to hold that many characters, or the sio interrupt handler must be called before the receiver's buffers fill up. >you use flow control shouldn't it stop the buffer from getting >overflowed simply because no more data is being received. `silo' != `buffer'. FLow control stops the 2 layers of buffers in the driver from overflowing. There is a different message for each layer. Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199512151948.GAA28090>