Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 28 Nov 2011 17:20:43 -0500
From:      John Baldwin <jhb@freebsd.org>
To:        John Nielsen <lists@jnielsen.net>
Cc:        "freebsd-current@freebsd.org" <freebsd-current@freebsd.org>
Subject:   Re: loader crash / BTX halted on 9.0-RC2 DVD with AMD pseudo-RAID
Message-ID:  <201111281720.44003.jhb@freebsd.org>
In-Reply-To: <929879BF-521C-43A1-8D63-DF2B04B6D013@jnielsen.net>
References:  <848DEEC1-570F-43F8-B432-A34F81014CD0@jnielsen.net> <201111221026.23015.jhb@freebsd.org> <929879BF-521C-43A1-8D63-DF2B04B6D013@jnielsen.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tuesday, November 22, 2011 10:07:52 pm John Nielsen wrote:
> On Nov 22, 2011, at 10:26 AM, John Baldwin <jhb@freebsd.org> wrote:
> 
> > On Monday, November 21, 2011 1:45:36 pm John Nielsen wrote:
> >> This weekend I downloaded the Freebsd 9.0 RC2 amd64 ISO image and burned it 
> > to a DVD. I have a computer that currently runs Windows 7 but I plan to 
> > install FreeBSD on it in the near future so I booted it up from the DVD to 
> > check the hardware/driver status. Much to my dismay, the boot loader crashed 
> > right away (register dump followed by "BTX halted") and the computer 
> > immediately rebooted. I took a video with my phone so I could capture the 
> > crash message, screenshot here:
> >> 
> >> http://picpaste.com/pics/BTXcrash.1321899682.jpg
> >> 
> >> I then tried tweaking a few BIOS settings and found that turning off the 
> > built-in pseudo-RAID allowed the DVD to boot normally. I changed the SATA type 
> > from "RAID" to "AHCI". Fortunately I plan to use the controller in AHCI mode 
> > for the FreeBSD installation so this won't end up being a problem for me, but 
> > I still thought it was worth reporting.
> > 
> > Hmmm, so this is odd.  It died with an Invalid TSS exception on the iret 
> > instruction at the end of the return-from-real-mode trampoline in BTX.  
> > Looking at the dump I noticed that PSL_NT is set in %eflags, so for some 
> > reason the iret was trying to do a nested task return.  We shouldn't let
> > that flag leak out of any real mode code.  Try this patch perhaps:
> 
> Thanks for looking!
> 
> I put gptboot on a USB stick and tried it with and without the patch.
> Identical behavior in both cases to booting from the DVD (only faster)--BTX
> dump and an instant reboot. I didn't do a screen capture yet but will be
> happy to tomorrow if it will help.

A screen capture would be useful.  It may be that I did not fix the right
copy of the flags.

-- 
John Baldwin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201111281720.44003.jhb>