Hi marchueu, these FBDEV removal CLs unfortunately breaks all ToT factory install shims.
I previously recommended to not removing fbdev support so fast ( https://code.google.com/p/chrome-os-partner/issues/detail?id=57809#c22 ) . Although we (factory team) has finished the preliminary implementation of using frecon on factory shims, but that was not widely validated yet.
For example, chunsten just found today that on Gru, same code will hang when doing switch_root. Also, all x86 factory shims are now all displaying nothing.
Can you revert and hold the migration until the factory team has validated the remaining images (factory_shim, netboot x x64, arm) all work?
@10: we gave enough advance warning (2 years), so we're not going to revert, sorry. You have factory branches where you can reenable fbdev if you want, but we certainly don't support it.
@11: > we gave enough advance warning (2 years), so we're not going to revert, sorry.
frecon support in initramfs environment (recovery, factory shim, netboot) was not available until very recently(https://code.google.com/p/chrome-os-partner/issues/detail?id=55250 , 3 months ago); and it caused problems that was only fixed two weeks ago ( https://code.google.com/p/chrome-os-partner/issues/detail?id=57809 )
> You have factory branches where you can reenable fbdev if you want, but we certainly don't support it.
The problem was that gru factory branch had an early cut of kernel that did not support fbdev correctly; to re-base to a working version we need to find a good one from ToT, and then realized it's all broken.
And there are projects which do not have factory branches yet.
Anyway if removing fbdev is already decided to go, we will try to fix factory shim as soon as possible; but we'll definitely need help in issue 636671 to make changes in frecon once we've found root cause (from what I've heard now, it's stuck in switch_root if frecon was running - I'll investigate more).
We should also remove the code (from libchromeos-ui, I think) for launching the X server. I can do this.
I think that powerd may also have some code around active-VT detection that I delete too.
Oh, and debugd should stop trying to include Xorg.0.log in feedback reports and chrome://system. Right now we always get "cat: /var/log/Xorg.0.log: No such file or directory".
Comment 1 by ihf@chromium.org
, Oct 13 2016