Am 28.03.2016 um 13:17 schrieb Herbert Graeber
Am Sonntag, 27. März 2016, 22:46:17 CEST schrieb Alexander Graf:
On 26.03.16 13:01, Herbert Graeber wrote:
Am Freitag, 25. März 2016, 22:57:48 CET schrieb
Am Freitag, 25. März 2016, 21:13:17 CET schrieb
>> But the Problem is, that boot.script contains "rootflags=size=100%"
>> "size=100%" isn't an allowed mount option for ext4. This option
>> for tmpfs. I have found this option in JeOS-raspberrypi.kiwi and every
>> other of several *.kiwi files I looked into, too.
> right, but there is a kiwi-hooks/installUboot.sh script on firstboot
> in the image that should get executed
> and remove those flags:
> # On second boot, the rootfs is no longer tmpfs and dracut would
> the # command line argument, remove it again from the config
> for file in /etc/sysconfig/bootloader /etc/default/grub
> do ⇥ [ -e "$file" ] && sed -i -e
Those lines are missing in /kiwi-hooks/installUBoot.sh from openSUSE-
> can you provide the kiwi boot log ? perhaps there is a hint why the
> code isn't getting executed.
I can have a look tomorrow, but I think the missing lines are the cause
this an a log won't show more.
Today I tried the next image (2016.03.25-Build2.1). The problem remains.
I have attached /var/log/boot.kiwi, /boot/boot.script and /kiwk-hooks/
> What's strange is, that a raspberrypi2 image from the same date contains
> the right kiwi-hooks/installUBoot.sh
I think I've tracked it down. The problem was that the rpi2 image is
built from exactly the openSUSE:Factory:ARM JeOS package, while the rpi1
downstream image gets built from the upstream JeOS-raspberrypi
description plus a few modifications.
Because of the way OBS works, binary files don't get stored as diffs in
a linked repository but instead get overwritten. The installUboot.sh
script is part of a tgz file (uboot-setup-raspberrypi.tgz).
In between the last update to that tgz and today the rootflags line got
added to installUboot.sh, but wasn't updated in the downstream copy of
the tgz. I've done that manually now. I hope in the not too distant
future we can just throw away all of this downstream RPi stuff and have
great, working upstream Linux based RPi images for 1, 2 and 3 ;).
One thought: My first mail was meant as a status report, but lead to some
sort of bug triaging. Shall I make a bug report next time?