Date: Thu, 1 Oct 1998 01:01:49 +0200 (SAT) From: Robert Nordier <rnordier@nordier.com> To: shmit@kublai.com Cc: rnordier@nordier.com, freebsd-bugs@FreeBSD.ORG Subject: BTX/firmware incompatibility problem (was i386/8105) Message-ID: <199809302301.BAA00418@ceia.nordier.com> In-Reply-To: <19980930174254.E20854@kublai.com> from Brian Cully at "Sep 30, 98 05:42:54 pm"
index | next in thread | previous in thread | raw e-mail
Brian Cully wrote: > On Wed, Sep 30, 1998 at 11:13:09PM +0200, Robert Nordier wrote: > > However, I'd be interested in getting details of the register dump. > > Not on account of btxboot (which is unimportant) but because it is > > test code for BTX, which the new 3-stage bootstrap also uses. > > (Even just the "eip=" value would be a help in deciding whether this > > is a btxboot, BTX, or kernel problem.) > > Sure. It'll probably help if I tell you what I did to install it, as well, > no? :-) > > About an hour ago I CVSupped the system (I made world about four or five > hours ago, with freshly CVSupped sources). > > natasya:~/src/btxboot-0.8.1% cd /sys/boot/i386/ > natasya:/sys/boot/i386% make clean && make > [snip] > natasya:/sys/boot/i386% cd ~/src/btxboot-0.8.1/ > natasya:~/src/btxboot-0.8.1% make clean && make > natasya:~/src/btxboot-0.8.1% su -m > Password: > natasya:~/src/btxboot-0.8.1# disklabel -B -b ./btxboot1 -s ./btxboot2 /dev/rda0s1 > > Then I rebooted and got this: > > F1 . . . BSD > > Default: F1 > int=0000000e err=00000005 efl=00030202 eip=00001af0 > eax=00000800 ebx=000003a2 ecx=00000000 edx=00000080 > esi=00001264 edi=00000000 ebp=000003aa esp=0000039e > cs=c800 ds=0040 es=9f77 fs=0000 gs=0000 ss=9f77 > cs:eip=3a 54 1b 75 08 80 7c 10-55 75 02 eb 05 5e 1f e9 > ss:esp=64 12 40 00 00 00 00 00-00 00 4b 15 00 00 00 00 Useful, thanks. This is the sort of firmware-related problem I was hoping to isolate by encouraging people to try the btxboot code. For some reason, your hard drive BIOS is checking for specific values at low memory locations 0x0040:0x127f and 0x0040:0x1274, something it has no particular business doing. (Trying to protect itself against some particularly virulent strain of DOS or Windows, perhaps?) The new boot code is running in fully-protected mode, and isn't expecting that some foreign code is going to be doing (inherently unreliable) heuristic checking based on the contents of its own private memory space. > > And there you go. Should we stop hijacking Jordan's threads and move > this somewhere else? :-) OK, done. (Sorry, Jordan. :-) You could really file a separate pr for this (since the forthcoming /sys/boot/i386/* code isn't going to work either for this firmware), though I'll be taking a look at it anyway. -- Robert Nordier To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-bugs" in the body of the messagehome | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199809302301.BAA00418>