Date: Fri, 14 Apr 1995 18:06:32 -0700 (PDT) From: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com> To: phk@ref.tfs.com (Poul-Henning Kamp) Cc: cmf@ins.infonet.net, current@FreeBSD.org Subject: Re: Interesting (and odd) effect in -current Message-ID: <199504150106.SAA02129@gndrsh.aac.dev.com> In-Reply-To: <199504150051.RAA18822@ref.tfs.com> from "Poul-Henning Kamp" at Apr 14, 95 05:51:13 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > Rod, > > Wouldn't it be a good idea in general if we zero all of the RAM in locore.s > (except the messagebuffer that is) ? > > That way we will have flushed all the caches, and reset all parity bits... IMHO, no, this should have already been done by the BIOS at power up time. Humm... I just came up with a fast and dirty way to find out if we are ever reading memory we have not written into (should never ever occur, right, as that would be using unitialized data). If it wasn't so chip set specific, you could actually turn parity off, use the special ports to write bad parity in all of memory. Then let things fly. But I have digressed, if the BIOS didn't manage to get this write at power on, you would get NMI interrupts no matter what OS you ran. I don't see a reason to add code to FreeBSD that really belongs in the BIOS in the off chance that some really rare broken motherboard could then work. If some one can prove to me that a production motherboard failed to initialize the parity bits correctly I will replace the motherboard for them, and be good riddens of one more piece of junk to have to try and support with redundant code. -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Custom computers for FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199504150106.SAA02129>