Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!utgpu!water!watnot!watmath!clyde!cbatt!ucbvax!LOGICON.ARPA!Perrine From: Perrine@LOGICON.ARPA.UUCP Newsgroups: mod.computers.vax Subject: Responses to VAX TBIA query Message-ID: <8702242023.AA11793@ucbvax.Berkeley.EDU> Date: Tue, 24-Feb-87 12:57:00 EST Article-I.D.: ucbvax.8702242023.AA11793 Posted: Tue Feb 24 12:57:00 1987 Date-Received: Fri, 27-Feb-87 00:12:55 EST Sender: daemon@ucbvax.BERKELEY.EDU Distribution: world Organization: The ARPA Internet Lines: 24 Approved: info-vax@sri-kl.arpa On Friday 24 Feb, I "shot a query into the net" about the VAX 780 TBIA register and 4.2BSD moving a 1 there when the Architecture doc called for a 0. Here are some of the responses I received: >>From: john@basser.oz.AU (john%basser.oz@SEISMO.CSS.GOV) >>Yep. TBIA is a write-only read-zero register. The _act of writing_ it >>invalidates the translation buffer; what bit pattern you write doesn't >>matter at all -- the hardware totally ignores it. >> >>From: "Lew Doll" >>It probably does not matter, since this register does not exist. It >>actually causes a branch in the microcode to a routine that invalidates >>each location, one at a time. >> >>From: Jason Venner >>This has been fixed in 4.3, with the most recent rev level, you must >>move a zero, previously a one worked correctly. Due to the response from the Berkeley person, and because we *are* running the latest HW/microcode rev, we will be rebuilding our 4.2 kernel to move the proper value to the TBIA register. I would advise others to consider doing the same. Tom Perrine ++ Logicon OSD ++ Perrine@LOGICON.ARPA