Re: gcc compiler version / kernel problems

[Greg Julius <fromRedHat outtacyte com>]

At 03:08 PM 1/14/03, you wrote:
> > I did a lsmod to see what modules were loaded and in active use.  Is this 
> > a good way to get a hint?
>
> Yes, look for any SCSI drivers or the ext3 filesystem.  You can also
> look in /etc/modules.conf and look for any "alias scsi_hostadapter"
> lines.

Nothing there for scsi


> Also, do a "mount" command.  If your root filesystem is on /dev/hdxy
> (x and y can be any digit), then you don't need a SCSI driver since
> your root filesystem is on IDE.  Also, if you don't see "ext3" as the
> filesystem type for "/", then you don't need the ext3 module to boot.

My root filesystem is on hda5  so IDE (knew that)
and it is ext3 so I built that into my kernel via the option (not module)


> If none of that stuff exists, you can really delete the initrd image
> and remove the reference to it in /etc/lilo.conf or
> /boot/grub/grub.conf.

That's just it, if it ain't there, it fails with a kernel panic VFS.  I was 
sure I didn't need an initrd, so I didn't build one.  When it panicked, I 
decided, what the hey, I'll put one in as the RH ones all have it.  Presto 
it booted.  I can't see what it thinks it needs.  Unless its because some 
of the other stuff in the initrd (clearly, but what?).


> > Regarding file sizes, those were the stderr output sizes listed.  The 
> > stdout file sizes were MUCH larger and full of the housekeeping and 
> > tracking stuff.  I wasn't worried about that stuff, just the stderr files 
> > seemed to me a bit large (except the one that ended up with zero bytes ;)   ).
>
> Well, I'd need to see the actual messages to make a determination.
> Generally speaking, if the make doesn't abort with an error, you most
> likely have a viable kernel and modules.

Kool, that's good to know.


Thanks for your help - you and others have been invaluable to me.  At least 
I've got the kernel running.
-g