[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Installing Wheezy on DreamPlug: Unable to make the system bootable

Le samedi 31 août 2013 à 20:14 +0100, Ian Campbell a écrit :
> On Sat, 2013-08-31 at 21:08 +0200, Paul Kocialkowski wrote:
> > Le samedi 31 août 2013 à 20:01 +0100, Ian Campbell a écrit :
> > > > In any case, there is a bug there -- my guess is that
> > > > flash-kernel-installer hangs because of the partitioning layout for some
> > > > reason. Note that running flash-kernel from the installed working system
> > > > with my custom layout did work.
> > > 
> > > That's very odd. I'm afraid that without breaking in to get some logs
> > > I'm not sure how to take this any further.
> > 
> > Perhaps you could reproduce the bug by using my custom partitioning
> > layout and try to get logs from there?
> My spare DP is in the middle of being used to debug #719680.
> Please can you file a bug with the info from this thread, lets say
> against flash-kernel for the time being, so this doesn't get forgotten.

Sure, I'll do it soon enough.

>From the working installation's installer syslog, I can tell that
generating a new initramfs and the u-boot images comes right next after
the u-boot-tools's installation.
Now my guess is that the problem is in flash-kernel's
flash_kernel_set_root, which is the one dealing with the rootfs.
In a case, it is possible that flash_kernel_set_root waits for stdin and
hence would make the whole thing hang (read _ignored) in the installer
context. It is called as a hook when generating the initramfs.

It seems like a long shot since this case is not supposed to be reached,
but I also noticed that the images I generated on the working
installation wouldn't boot without the root=/dev/sda2 cmdline argument
from the bootloader (and I don't mean the ones I generated using my
computer and uploaded to the Dreamplug to make it boot the first time).

However I didn't get the error message that's supposed to come before
(read _ignored). So I might very well be wrong, but there is no doubt
the issue happens when running something from flash-kernel, and
flash_kernel_set_root is definitely related to the rootfs.

What do you think?

> I've reinstalled dozens of times on my spare one (testing the installer)
> and it's not showing any signs. If it was going to die after a single
> digit number of installations it probably wouldn't be suitable to use in
> general either.

That's good to know -- thanks for sharing.

Reply to: