Anstrom boot hang with MACHINE=i686-generic and DISTRO=anstrom-2010.x

Anstrom boot hang with MACHINE=i686-generic and DISTRO=anstrom-2010.x

[Greg Kogut]

I've tried both console-image and x11-image. 

During the boot process they both hang.  All of the rcS.d scripts appear to run,
but I never get to a getty screen.   It appears to hang either just before getty
runs, or getty itself is hanging.  

I've tried booting at run-level one, and toying with a very minimal set of
startup scripts.

I thought maybe that getty was configured to display to ttyS0, etc, but it
should be going to tty1 - my DVI output.   I added some lines to inittab to
spawn gettys to multiple tty's, but no luck.

The hardware is just a modern mini-mobo.  Core Duo, DVI out, ext2 disk.  

Any ideas?

Re: Anstrom boot hang with MACHINE=i686-generic and DISTRO=anstrom-2010.x

[Greg Kogut]

Greg Kogut <gkogut <at> yahoo.com> writes:

> 
> I've tried both console-image and x11-image. 


Edit:

I was able to to boot fully after appending "-o rw" to the "root"
kernel parameter.

I suspect the issue was that the read-only mount was not allowing
udev to correctly populate /dev with the tty devices passed to 
getty, and getty was flummoxed.

So I'm amending my question:  Help a n00b -what's the standard
practice with respect to maintaining an embedded-safe read-only
rootfs while allowing udev to do its thang?   Separate partition
for /dev? 

Re: Anstrom boot hang with MACHINE=i686-generic and?DISTRO=anstrom-2010.x

[Denys Dmytriyenko]

On Tue, Nov 08, 2011 at 11:19:45PM +0000, Greg Kogut wrote:
> Greg Kogut <gkogut <at> yahoo.com> writes:
> 
> > 
> > I've tried both console-image and x11-image. 
> 
> 
> Edit:
> 
> I was able to to boot fully after appending "-o rw" to the "root"
> kernel parameter.
> 
> I suspect the issue was that the read-only mount was not allowing
> udev to correctly populate /dev with the tty devices passed to 
> getty, and getty was flummoxed.
> 
> So I'm amending my question:  Help a n00b -what's the standard
> practice with respect to maintaining an embedded-safe read-only
> rootfs while allowing udev to do its thang?   Separate partition
> for /dev? 

Actually, /dev is usually a separate partition already, when managed by devfs, 
udev or mdev. Unless you manage /dev manually with static devnodes, which is 
not the case for you, as you mentioned udev... Check your /etc/fstab or 
mount/mtab to see how /dev is mounted - in most cases should be RAM-based 
tmpfs handled by udev. Also, you may want to disable devcache by editing 
/etc/default/udev and commenting the DEVCACHE variable.

-- 
Denys

Re: Anstrom boot hang with MACHINE=i686-generic and DISTRO=anstrom-2010.x

[Koen Kooi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Op 09-11-11 00:19, Greg Kogut schreef:
> Greg Kogut <gkogut <at> yahoo.com> writes:
> 
>> 
>> I've tried both console-image and x11-image.
> 
> 
> Edit:
> 
> I was able to to boot fully after appending "-o rw" to the "root" kernel
> parameter.
> 
> I suspect the issue was that the read-only mount was not allowing udev to
> correctly populate /dev with the tty devices passed to getty, and getty
> was flummoxed.
> 
> So I'm amending my question:  Help a n00b -what's the standard practice
> with respect to maintaining an embedded-safe read-only rootfs while
> allowing udev to do its thang?   Separate partition for /dev?

You're still following the deprecated OE-classic workflow. If you follow the
instructions at http://www.angstrom-distribution.org/building-angstrom you
will get the updated angstrom configs that work with RO /

regards,

Koen

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)
Comment: GPGTools - http://gpgtools.org

iD8DBQFOulN9MkyGM64RGpERAgkLAJsHQ6t7GR+JUBu5BeaTVz/YYK/KpQCguTWU
XMTgjD846DcyqCIr9B7JUhc=
=3H0D
-----END PGP SIGNATURE-----