Date: Sat, 23 Mar 96 21:54 WET From: uhclem@nemesis.lonestar.org (Frank Durda IV) To: hackers@freebsd.org Subject: Re: Crash advice needed APPENDIX B Message-ID: <m0u0gtH-000CPpC@nemesis.lonestar.org>
next in thread | raw e-mail | index | archive | help
[0]Granted, FreeBSD 2.0.5 did not like the cache when booting [0]a compressed kernel. It would always fail during uncompression. [0]This was fixed in 2.1.0 and I have seen no other solid [0]errors of that type. [1]"Rodney W. Grimes" <rgrimes@GndRsh.aac.dev.com> writes: [1]Big indication your board does infact have a cache coherency problem. Not exactly, since once the system was installed on the hard disk, everything works, and serious loads (make worlds plus other stuff) all run without incident. All I would do was pull the cache during the floppy boot, install the system then stick the cache back in once the system was on the hard disk. (Didn't have to do that in 1.1.5.1 either.) Note also that three versions of the cache module from two completely different vendors exhibited this same problem on 2.0.5, and I had all the Intel erratas on the earlier parts (which I did not use), so I don't know precisely what the problem was. I do know that the cache module could be fooled by changing the state of the A20 line, or any external remapping or banking of RAM (obviously), but I assume we would turn on A20 and leave it on forever. I believe Bruce identified some flaw in the way the uncompress was being done that affected some systems with particular cache systems and a change was made on purpose for 2.1.0 that made things happy. Perhaps he will expound on what the change was. Frank Durda IV <uhclem@nemesis.lonestar.org>|"Now with brighter colors! or uhclem%nemesis@rwsystr.nkn.net | Whiter whites! Internet support | in all applets! Coming soon or ...letni!rwsys!nemesis!uhclem | to the commodity computer!"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?m0u0gtH-000CPpC>