Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!julius.cs.uiuc.edu!apple!vsi1!daver!bungi.com!news From: gs@vw25.chips.com (George Scolaro) Newsgroups: comp.sys.nsc.32k Subject: Re: Running without an FPU Message-ID: Date: 20 Sep 90 00:10:56 GMT Sender: news@daver.bungi.com Lines: 26 Approved: news@daver.bungi.com [In the message entitled "Re: Running without an FPU" on Sep 19, 22:30, Eyal Lebedinsky writes:] > > Set a watchdog timer. > try accessing the FPU. > If the next instruction is reached - I have an FPU. > If the timer interrupt arrives - I have no FPU. > > Now, do we have a timer? will the interrupt be acknowledged if the FPU bus > cycle is stuck? The FPU access is treated as a slave access by the 32000 series. The slave access is a non-interruptible transfer, hence no FPU -> system hang. What we really need is an FPU emulation library for the 32000 which is linked into the kernel. This does mean we need 2 versions of the kernel, or at least one byte that is set/cleared to configure it. And yes we have lots of timers, one in each DUART and a couple in the ICU. regards, -- George Scolaro (gs@vw25.chips.com) Chips & Technologies (408) 434-0600 3050 Zanker Road San Jose, CA 95134