[Bug 1811216] Re: Support both armhf and arm64 raspi3 in flash-kernel

[Adam Smith]

This changes the fdt_addr_r to 0x03000000.  Although the previous number
did not work, I believe the fdt_addr_r will have been changed by many
users.  For example, Ryan Finnie's pi3 images have this set to
0x02008000.  So this updated flash-kernel package will stop those
images/installations booting (because the pi's config.txt will still be
using the old address).  This is potentially going to upset a lot of
people.

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to flash-kernel in Ubuntu.
https://bugs.launchpad.net/bugs/1811216

Title:
  Support both armhf and arm64 raspi3 in flash-kernel

Status in flash-kernel package in Ubuntu:
  Fix Released
Status in flash-kernel source package in Bionic:
  Fix Committed
Status in flash-kernel source package in Cosmic:
  Fix Committed

Bug description:
  [Impact]

  Currently the boot.scr provided by flash-kernel for rpi3 uses bootz for booting the kernel, an uboot command that only works for armhf kernels. This means that in case the boot.scr on a pi3 arm64 gets updated on upgrade, the system will stop booting. Easiest way is to support separate bootscripts for arm64 and armhf.
  Also, to fully support all the strange cases for arm64, the boot address parameters need to be adjusted for a bigger kernel. The arm64 kernel, in contrast to the armhf one, is not compressed and takes up much more space in memory.

  This is basically only needed for bionic as we don't intend to build
  raspi3 images on cosmic, but SRUs will be provided for all series.

  [Test Case]

   * Boot up an arm64 based image on a raspi3 and log into it
   * Run flash-kernel
   * Check for errors
   * Check if /boot/firmware/boot.scr still has booti as its boot command
   * Reboot
   * Check if system still boots correctly

  [Regression Potential]

  The only regression potential is in actually making the image not
  booting, so running the SRU test case should validate that. There's
  basically no other regression potential I can think of.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/flash-kernel/+bug/1811216/+subscriptions