Re: [HOWTO] SGI tftp boot with ARCLoad
Florian Lohoff <flo@rfc822.org> wrote:
Hi,
> Try extending the config - i mean - the config should not load either
> if the prom does - just append more than a couple of tftp blocks
> of comments - if it fails the prom peeks at it - Maybe the prom
> just looks at files which are >2 tftp blocks or something.
So, here's the result:
- bootp():/arc.cf directly from the PROM: throws an error after the
second block (can be expected in this case, it's not a bootable
file after all)
- loading from arcload: works flawlessly
Interestingly enough, I saw that the transfer of the kernel image
ended prematurely with the same error being thrown back to the
server. Though arcload carried on and booted the machine with what it
got. The initrd failed to load as usual.
Given that the error thrown back to the server is "Disk full or
allocation exceeded" according to wireshark (though it complains that
anything that's TFTP from the PROM and not an ACK is malformed ...),
it's possible that in the initrd case the PROM is really out of space
to store it, especially given it threw an error for the kernel too.
But that'd be funny because the kernel is larger than the initrd and
if the PROM uses any space for buffering, it should be released once
the "file" is closed.
I'm 90% sure the PROM is doing something behind my back; when opening
a file on the network, the PROM very clearly loads the file or part of
it when the open call is issued. (obviously the open call needs a
return value so it has to check the file is available, and it returns
a handle containing the file size among other things)
When loading the kernel image, it reports the size of the file
properly after the open call, when loading the initrd it reports 0 as
the file size.
I guess I've just been screwed over by an SGI PROM once more. Hrmpf :(
JB.
--
Julien BLACHE - Debian & GNU/Linux Developer - <jblache@debian.org>
Public key available on <http://www.jblache.org> - KeyID: F5D6 5169
GPG Fingerprint : 935A 79F1 C8B3 3521 FD62 7CC7 CD61 4FD7 F5D6 5169
Reply to: