Am 09.07.2010 04:46, schrieb Gerardo Exequiel Pozzi:
>> What thing that may be improved here is:
>> * steps 4 and 6 instead of mounting on /tmpfs/mnt/imagename mount
>> under /mnt/imagename
>>    currently an ls /new_root/mnt will display empty dirs "overlay" and
>> "root-image".
>>    Now looks like a bit recursive.
>> * execute 3) then 2) so medium will me mounted directly to
>> /new_root/bootmnt
>>
> NO!, bind mount /bootmnt to /new_root/bootmnt should be keep...
> otherwise on "shutdown", resourse or device busy message.
> the same applies when copytoram
> 
> My hardcode archiso experiment: http://gist.github.com/468957 (I using
> using initscript.git + my pending patch for rc.shutdown from
> http://bugs.archlinux.org/task/19783)

It is still the right thing to do (if we can). We should rather have a
solution for the "device and ressource busy" problem, somehow telling
our initscripts to not try and umount the fs.

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to