Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!mips!sgi!shinobu!odin!sgi.com!scotth From: scotth@corp.sgi.com (Scott Henry) Newsgroups: comp.protocols.time.ntp Subject: Re: ntp for the ibm risc-6000 (AIX3.1)? Message-ID: Date: 21 Dec 90 12:45:30 GMT References: <9012171806.AA11783@merit.edu> <1990Dec19.010235.29690@acheron.uucp> Sender: news@odin.corp.sgi.com (Net News) Reply-To: scotth@sgi.com (Scott Henry) Distribution: comp Organization: Silicon Graphics Inc, Mountain View, CA Lines: 23 In-Reply-To: clarke@acheron.uucp's message of 19 Dec 90 01:02:35 GMT ec> From article <9012171806.AA11783@merit.edu>, by katz@MERIT.EDU (Dave Katz): > You can get a modified version of xntp from pub/xntp/xntp.rs6000.tar.Z > on merit.edu (35.1.1.42). Beware that you need a repaired kernel > from IBM to run NTP. > Request APAR A12990 for this fix. ec> Also be aware that this does not cure everything. The 6K seems to ec> have problems with SIGIO. You can run as a client or lock to your ec> local clock, but trying to run a Spectracom is doomed to failure. I ec> am working on this right now with Austin. I don't know if this applies to IBM RS-6000 machines or not, but on Silicon Graphics systems, SIGIO works only on sockets, and SIGPOLL works on streams devices (/dev/tty*). Once I discovered that, our TrueTime clock seems to be working fine. Since input_handler() doesn't care which signal invokes it, the changes were rather minor. -- Scott Henry / Traveller on Dragon Wings Information Services, / Help! My disclaimer is missing! Silicon Graphics, Inc / Politicians no baka!