On "Tue, 24 Aug 2010 00:45:09 +0100",
David Brownlee <abs%netbsd.org@localhost> wrote:
>On 24 August 2010 00:06, Dan Oglesby <danno%serverunderground.com@localhost>
>wrote:
>
>>> As a final check, would you be willing to look at a -current boot
>>> floppy? - eg
>>> http://nyftp.netbsd.org/pub/NetBSD-daily/HEAD/201008100000Z/sparc/installat
>ion/floppy/
>>
>> The first disk boots up to the point where you get the little spinning stick
>/progress indicator, but after a couple of minutes it errors with:
>>
>> lseek section headers: invalid file offset
>> Cannot load netbsd: error=103
>>
>> device[/fd@1,f7200000:a] ("halt" to halt):
>>
>> Very similar to how the boot from floppy fails on 5.0.2 floppies.
>
>Hmm - thats coming from *_loadfile() in lib/libsa/loadfile_elf32.c, in
>a section which is (flags & (LOAD_SYM|COUNT_SYM)). I'm pretty sure the
>bootload on floppies is not supposed to load the symbol table to avoid
>just such an issue.
>
>I would assume that its failing in start_kernel() in
>arch/sparc/stand/ofwboot/boot.c
>
>I don't have a sparc box with floppy to hand, but I could point you at
>where to add some debugging (or put up a couple of debugging boot
>floppy images) if you were interested in looking at this?
>
>If someone actually familiar with the sparc and libsa code wants to
>chime in that would be not be a problem :)
I have a pair of IPCs with floppy drives, and I'd be happy to send one
to a good home. :-) The other, of course, I hope to find time to set
up for use/testing locally.
IIRC, both have ~2 GB drives and have relatively new battery-backed
nvram. One has OB 1.x and the other has 2.x. And, if I'm recalling which
is which correctly, the 2.x box is rather beat up while the 1.x box is
fairly pristine.
If anyone's interested I'll dig into the details.
Also, I have a Sparc 4 just sitting around, and a trio of Sun CRTs.
I'm sure there are other miscellaneous bits to go with all of this. :-)
Attachment:
pgp7rHVxzsI9p.pgp
Description: PGP signature