soekris net5501 and netbsd-5 /boot
Greg Troxel
I have a net5501, which was running netbsd-5 from april or so fine,
except for the 2-minute boot delay problem. I updated /netbsd and
userland in plafe, and in addition updated /boot and bootxx_ffsv1. The
system will now get to the point where I can tell it to boot /netbsd or
/netbsd.ok (my old kernel), and for either will print the sizes line and
then nothing. So bootxx_ffsv1 loads /boot, and /boot seems to load the
kernel but immediately have trouble.
My next step is to boot it diskless and see what's going on. It seems
like the new /boot or bootxx_ffsv1 is somehow broken, which seems rather
unlikely. I'll report back, but any clues are welcome.
David Laight
>
> I have a net5501, which was running netbsd-5 from april or so fine,
> except for the 2-minute boot delay problem. I updated /netbsd and
> userland in plafe, and in addition updated /boot and bootxx_ffsv1. The
> system will now get to the point where I can tell it to boot /netbsd or
> /netbsd.ok (my old kernel), and for either will print the sizes line and
> then nothing. So bootxx_ffsv1 loads /boot, and /boot seems to load the
> kernel but immediately have trouble.
What are the console settings?
It might be that you've lost the serial console, everything up to the
point you've mentioned is likely to be using bios calls - which work.
The serial console parameters are in a patchable header in the bootcode
(the copy in bootxx_ffsv1 wiill be the one that is used).
The system might even be coming up far enough to let you login remotely!
David
--
David Laight: da...@l8s.co.uk
--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-...@muc.de
Greg Troxel
David Laight <da...@l8s.co.uk> writes:
> On Mon, Aug 01, 2011 at 07:35:54PM -0400, Greg Troxel wrote:
>>
>> I have a net5501, which was running netbsd-5 from april or so fine,
>> except for the 2-minute boot delay problem. I updated /netbsd and
>> userland in plafe, and in addition updated /boot and bootxx_ffsv1. The
>> system will now get to the point where I can tell it to boot /netbsd or
>> /netbsd.ok (my old kernel), and for either will print the sizes line and
>> then nothing. So bootxx_ffsv1 loads /boot, and /boot seems to load the
>> kernel but immediately have trouble.
>
> What are the console settings?
> It might be that you've lost the serial console, everything up to the
> point you've mentioned is likely to be using bios calls - which work.
Thanks very much for the hint; that was indeed the issue.
Interrupting the boot and telling /boot "console com0" made the system
boot ok.
> The serial console parameters are in a patchable header in the bootcode
> (the copy in bootxx_ffsv1 wiill be the one that is used).
(For the archives:)
It's not obvious from the man page, but installboot can show the
existing options, via verbose editing and not writing anything.
root 18 /usr/mdec #> installboot -v -e /dev/rwd0a
File system: /dev/rwd0a
Boot options: timeout 5, flags 0, speed 9600, ioaddr 0, console pc
and then it's just "-v -e -o console=com0" to set it.
It seems like a bug that installing a new bootstrap doesn't preserve
options. Do others feel that way?
> The system might even be coming up far enough to let you login remotely!
For reasons I don't understand, while after setting "console com0" to
/boot, the system came up and was on the net, without that it never did.
David Laight
>
> It seems like a bug that installing a new bootstrap doesn't preserve
> options. Do others feel that way?
Not as a default action.
Otherwise it is difficult to get a system 'clean'.
> > The system might even be coming up far enough to let you login remotely!
>
> For reasons I don't understand, while after setting "console com0" to
> /boot, the system came up and was on the net, without that it never did.
Possibly the kernel starts talking directly to the io/mem ports of the
assumed vga output screen. I suspect that doesn't work well.
The code might even try to trap into the bios - which will work
even less well.
Greg Troxel
David Laight <da...@l8s.co.uk> writes:
> On Wed, Aug 03, 2011 at 10:14:05AM -0400, Greg Troxel wrote:
>>
>> It seems like a bug that installing a new bootstrap doesn't preserve
>> options. Do others feel that way?
>
> Not as a default action.
> Otherwise it is difficult to get a system 'clean'.
Fair enough. I have documented the existing behavior.