Date: Fri, 24 Mar 95 10:20:18 MST From: terry@cs.weber.edu (Terry Lambert) To: bde@zeta.org.au (Bruce Evans) Cc: FreeBSD-hackers@FreeBSD.org, wilko@yedi.iaf.nl Subject: Re: obscure NMI Message-ID: <9503241720.AA09650@cs.weber.edu> In-Reply-To: <199503240935.TAA12296@godzilla.zeta.org.au> from "Bruce Evans" at Mar 24, 95 07:35:31 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> >Any suggestions on: > > NMI port 61 a0, port 70 7f, port 461 10 > >? > > >This happens just after the npx probe message is displayed on _every_ > >boot (1.1.5). The system board is a Philips P3464 EISA, with 20Mb mem. > > NMI was used on XTs for reporting FPU exceptions. It is even less > suitable for this purpose than the IRQ13 used on AT's, but XTs have > even less spare interrupts than AT's and 8086's don't even generate > exceptions for illegal opcodes. > > The npx probe generates an FPU exception. Perhaps the P3464 generates > both IRQ13 and NMI for reporting FPU exceptions. This braindamage is > required for compatibility but it is usually implemented in software > (by jumping from the IRQ13 handler to the NMI handler after dismissing > the IRQ13). There is some very interesting code in the Linux kernel to tell which way the FPU behaves and deal with it correctly thereafter. This is right before they check if you have a buggy Pentium or not (to deal with that) and right after they check if the WP bit is honored in protected mode or not and deal with that. I would suggest something similar should be done. Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9503241720.AA09650>