[9fans] Plan9/Bochs/kernel panic

[9fans] Plan9/Bochs/kernel panic

[Paul Swartz]

I tried installing Plan9 on Boches (CVS on Debian unstable, kernel
2.6.0-test9), but get a strange error when I boot from the floppy. 
The kernel loads and detects the emulated CPU, then dies immediatly
with a "panic: math emu".  Also interesting is before the "entry:
80100020" line, there's this strange line that appears to be some kind
of computation.  "1267078 => 745652+1116416+83748=1945816.".

Any ideas?

TIA, -p
--
      Paul Swartz
(o_   z3p at twistedmatrix dot com
//\   http://www.twistedmatrix.com/users/z3p.twistd/
V_/_  AIM: Z3Penguin

Re: [9fans] Plan9/Bochs/kernel panic

[Russ Cox]

> I tried installing Plan9 on Boches (CVS on Debian unstable, kernel
> 2.6.0-test9), but get a strange error when I boot from the floppy.
> The kernel loads and detects the emulated CPU, then dies immediatly
> with a "panic: math emu".  Also interesting is before the "entry:
> 80100020" line, there's this strange line that appears to be some kind
> of computation.  "1267078 => 745652+1116416+83748=1945816.".
>
> Any ideas?

The computation is the boot loader telling you its progress
in loading the kernel.  Disk size => text+data+bss = Mem size.

Beyond that I'm not much help.  Math emu means that the kernel
got a floating point exception, which should not happen.

Russ

Re: [9fans] Plan9/Bochs/kernel panic

[Russ Cox]

> Beyond that I'm not much help.  Math emu means that the kernel
> got a floating point exception, which should not happen.

I tracked this down.  It seems to be a bug in Bochs.
Real processors all let us turn off the FPU twice,
but if the FPU is already off, then Bochs causes an
exception when we "FWAIT" before turning it off again.
The Intel manual (I looked at the current online one
as well as an old 486 one) is very clear -- FWAIT causes
an exception if CR0.EM and CR0.TS are *both* set, whereas
Bochs causes an exception if *either* is set.
(To be fair, all the other floating point instructions
are supposed to cause the exception if *either* is set,
so it's not surprising that Bochs got this one exception
(no pun intended) wrong.)

Russ

Re: [9fans] Plan9/Bochs/kernel panic

[Russ Cox]

On 11/11/03, Russ Cox <rsc@swtch.com> wrote:
> I tracked this down.  It seems to be a bug in Bochs.
> Real processors all let us turn off the FPU twice,
> but if the FPU is already off, then Bochs causes an
> exception when we "FWAIT" before turning it off again.

I tracked this down again (I had quite a sense of deja vu
once I found the problem, but google failed me and I hadn't
thought to look in my gmail).  This time I fixed Plan 9 to
work around it.  There is a new /sys/src/9/pc/l.s on sources
that avoids the FWAIT when the FPU is already off.
With that change, Plan 9 boots in Bochs, though I
suggest qemu with the kqemu module unless you
are doing OS hacking and need Bochs's better debugging
features.

Russ