Date: Wed, 7 Jun 95 19:45 CDT From: uhclem%nemesis@fw.ast.com (Frank Durda IV) To: phk@ref.tfs.com, jkh@freebsd.org, davidg@freebsd.org, hackers@freebsd.org, bugs@freebsd.org Subject: Re: Pouls info request on cache issue Message-ID: <m0sJVjN-0004vyC@nemesis.lonestar.org>
next in thread | raw e-mail | index | archive | help
[2]phk says: [2]Please report this: [2] size kernel.MFS [2] kzip kernel.MFS [2] size kernel.MFS.kz # size kernel.MFS text data bss dec hex 913408 1159168 76080 2148656 20c930 # kzip kernel.MFS # size kernel.MFS.kz text data bss dec hex 1040384 0 2560 1042944 fea00 I then halted the system and attempted to boot from kernel.MFS.kz. If the cache (Intel or IDT modules) was present, it hung with: Boot: kernel.MFS.kz Booting wd(0,a)kernel.MFS.kz @ 0x300000 text=0xfe000 data=0x0 bss=0xa00 symbols=[+0x600+0x4+0x250+0x4+0x1c4] total=0x3ff 424 entry point=0x300000 Uncompressing kernel...done Booting the kernelel...done [note overlapping message] BIOS basemem (639K) != RTC basemem (640K) [in bold] and it hangs. If you boot kernel.MFS.kz with the cache disconnected, it runs fine. If you boot using kernel.MFS, it works fine with the cache installed or removed. FYI, this system has a Phoenix BIOS. [1]However, Bruce indicated that the boot procedure saves (where?) and then [1]reloads a block of low memory while decompressing, and then boots. Note [1]that an artifact of that reload is the scrambled messages I see on the screen [1]when the system hangs. Bruce felt the reload was unnecessary. [1]If you agree, you might comment it out and roll a kernel for me to try [1]that has that step missing. [1] [2]That would be in /sys/i386/boot/kzipboot/something... I will see about putting a debug step after unzip that forces a cache flush and see what that does. I'll also try pulling the save/restore since its only remaining purpose apparently is to allow the last status message to be displayed before the kernel completely takes over. Umm, you guys don't use INT 10 to display those messages do you? INT 10 tinkers with a lot of things, including A20 line. If there is anything you want tried here, let me know. If this drags on, I guess I could arrange to ship one of these systems somewhere, provided it comes back and someone has the time to look at the problem. :-) I got mail from someone with Micron MLBs that are also acting up in a similar way... but then PC Week says Windows '95 hates some Micron MLBs. Gee, I might have to go out and buy one... :-) Frank Durda IV (Win '95.9 beta beater by day) uhclem%nemesis@fw.ast.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?m0sJVjN-0004vyC>