reopen 426147
severity 426147 normal
thanks

(Oops, I didn't mean to close the bug, so I'm reopening it for now.)

On Mon, May 28, 2007 at 12:54:35PM -0400, Andrew Paprocki wrote:
> The config is attached to this msg. This is the config for the kernel
> supplied on the Debian installer disc, which is what I was using.
> 
> This issue is even more strange.. after I finally got the system
> booted by doing the partitioning/formatting on another box the mkfs.*
> binaries work fine on the installed system. They only seem to generate
> a floating point error when booted in the installation kernel off the
> DVD. (I got the error by running the binaries by hand in busybox).

OK, so there are a couple of things going on here.  First of all, the
kernel which is suppled with the Debian cd/dvd for installation
purposes isn't necessarily the same as the kernel which is booted by
the install CD.

Do you have the dmesg command available in the boot environment?  If
so, check to see if the kernel is emitting the following in the dmesg
log:

math-emulation not enabled and no coprocessor found.

Whe the math emualation codeis not enabled, it will print the above
message and then send a floating point exception to the program.  

> The device I am using is a Geode LX 800, detailed here:
> http://www.amd.com/us-en/ConnectivitySolutions/ProductInformation/0,,50_2330_9863_13022%5E13058,00.html
> 
> " Integrated FPU that supports the Intel MMX(r) and AMD 3DNow!™
> Technology instruction sets"

OK, earlier you said GX3, and from some google searches it seems that
a number of GX3's don't have FPU's.  And apparently some GX3's have
broken error reporting, so the use of the integrated FPU is sometimes
disabled, at least on NetBSD.  So I am a bit paranoid about the FPU.
Hence, please humor me in checking the dmesg log, would you?

> If the busybox binaries are a different build, then it looks like that
> is a place to look, as the normal package installed binaries work
> fine.

what do you mean by "busybox binaries" ?  Do you mean the busybox pre
version 1.40 that had parts of e2fsprogs built in?  Or did you mean
running e2fsck from a busybox shell?  In any case, if the issue is
that it doesn't work in the installer enviroment, it's almost
certainly an interaller bug of some kind; either in the kernel, or C
libraries provided in the installation environment.

And where did was this part of the Debian installation DVD, or
somewhere other bootable rescue CD/DVD?

                                        - Ted


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to