From: utzoo!decvax!harpo!npoiv!npois!ucbvax!C70:info-cpm Newsgroups: fa.info-cpm Title: Solution to the 2651/2661 Problem Article-I.D.: ucb.1515 Posted: Wed Jul 14 23:20:43 1982 Received: Fri Jul 16 01:09:44 1982 >From FJW@Mit-Mc Wed Jul 14 23:20:29 1982 It turned out that I simply didn't apply the originally suggested fix in the "right" place. (I put it in the REMSTAT routine.) The problem with MODEM and friends is the garbage collection that is done after a block is sent: it does a direct call to REMIN (or more likely in your implementations, just an IN DPORT), twice. That is where the check of the xmit buffer was not done. The last character sent, which should have been the checksum byte in my case, ended up being replaced by the last character received - the initial NAK! Thus, the real and normally superfluous fix is to always check the status of the xmit buffer before reading the receive buffer... To correct the previous rumor: it is a second-source supplier of the 2661 who is holding orders pending a fix to a "masking problem." --Frank