[fedora-arm] Issues to solve before F19 ARM GA

[Peter Robinson]

On Sun, May 26, 2013 at 6:15 AM, Brendan Conoboy <blc at redhat.com> wrote:
> Even though F19 ARM Beta bits are making their way to the mirrors now there
> is still quite a bit yet to be done before GA.  In addition to beta release
> notes still needing to be written there are still bugs and regressions to
> track down.  Let's squash as many of these as possible in the next couple
> weeks, and preferably spin test images early&often to confirm the fixes.
>
> General image issues:
>  o LVM console spam on first boot
>  o The initial-setup does not appear to run

Not sure when the first rcX is out to test Dennis's fix.

>  o Restorecon runs on first boot- can we avoid this?
>
> Versatile express:
>  o No u-boot - would be nice to have for virt integration

It would be nice, not a regression though, by "virt integration" do
you mean a single .img for use with virt-manager or something else?

>  o No boot-vexpress script or pre-extracted kernels - was in F18.
>  o No graphics support - regression since F18 (Kernel 3.6)
>
> TI Boards in general:
>  o Images, as distributed, need manual steps to be made bootable.
>  o An easy script to handle the manual steps is not yet available.
>  o Supplied vfat partition possibly badly formed.
>  o No graphics - regression since F18 (Kernel 3.6)
>
> OMAP3 (Beagle boards):
>  o The supplied uboot does not load boot.scr.
>  o The supplied kernel fails to boot.

This is a nice to have. 3.8.x works minus USB, in theory this should
now be fixed (might be a 3.10.x fix though).

> AM335x (Beagle bone boards):
>  o Missing kernel support - this is probably an F19 remix or F20 feature.

I'm working on this as I get time and hoping to land something for
Final but agreed it's a nice to have.

> Tegra2 (Trimslice, specifically)
>  o SSD support is unreliable due to evident hardware bug.
>  o X, though functional, is not stable.

Again I think both of those are nice to have. It's solid as a rock on
mine when used as a /home directory even if I thump it with builds.

> Anaconda installs:
>  o a-b-c needs to be run manually in %post

Is that a major blocker and how else would it run? Presumably as part
of the kernel install / anaconda boot loader setup?

> Notwithstanding the above, OMAP4 (Panda), Trimslice, Highbank and Versatile
> Express are in decent shape, which is why the Beta is going out now.  Plus
> it gives a readily available set of bits for advancing these defects.  If
> you are aware of additional issues please followup here- let's get them all
> identified then work out solutions.  Thanks!

Agreed HW side is in reasonable shape. I would also like to bring to
attention the ARMTracker blocker bugs and if people have some spare
cycles to go through the list and see what they can help in fixing.
I'm reviewing package failure lists and (re)fixing or filing some bugs
as I get the time and it would be good if some people with more
programming skills can help fix some.

Peter