[Bug 1899308] Re: failure to boot groovy daily (again)

[Steve Langasek]

On Thu, Oct 15, 2020 at 05:31:08PM -0000, Thomas Schmitt wrote:

> > sfdisk --no-reread -q $DEVICE -A 1 -X dos || true

> If this attributes the boot flag to the first partion of type EE, then
> this is a semi-violation of EFI specs and known to be punished by some
> firmwares. UEFI 2.8 says:
> 
>   "Table 20. Protective MBR Partition Record protecting the entire disk
>    BootIndicator   0   1   Set to 0x00 to indicate a non-bootable partition
>                            [...]
>                            Must be ignored by UEFI implementations."
> 
> It turned out back in 2015 that some EFIs interpret the latter statement
> as instruction to ignore the device if BootIndicator is present in the
> only existing partition. That's why --mbr-force-bootable was invented to
> mark a nearly non-existent partition as bootable.
> 
> The changelog description is wrong in these aspects:
> 
> > "If our partition table is GPT, we *may* have a protective MBR;"
> 
> We *must* have a protective MBR to have a valid GPT.
> 
> >  "we mark the iso9660 partition as bootable"
> 
> The ISO 9660 partition is number 1 in the GPT. Partition 1 in MBR is not
> mountable as ISO 9660 because it starts at LBA 1. 0 or 64 would be ok for
> mounting, but then it would not mark a valid GPT any more.
> 
> -----------------------------------------------------------------------
> 
> Casper needs to bring these bytes (decimal) into bytes 462 to 477 of
> the ISO image (i.e. the MBR):
> 
>   128   0  1   0   0   0   1   0   0   0   0   0   1   0   0   0
> 
> With bash i could do this by help of dd if=/dev/zero and echo -n $'\x80'
> $'\x01'. But with dash ...

This works under dash (and bash):

echo -n '\0200\00\01\00\00\00\01\00\00\00\00\00\01\00\00\00' | dd
of="$DEVICE" bs=1 seek=462 conv=notrunc

I'll switch to using that.

Thanks,
-- 
Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
Ubuntu Developer                                   https://www.debian.org/
slangasek at ubuntu.com                                     vorlon at debian.org

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to casper in Ubuntu.
https://bugs.launchpad.net/bugs/1899308

Title:
  failure to boot groovy daily  (again)

Status in Ubuntu CD Images:
  Fix Released
Status in casper package in Ubuntu:
  Triaged
Status in cd-boot-images-amd64 package in Ubuntu:
  Confirmed

Bug description:
  This is either a duplicate or return of
  - https://bugs.launchpad.net/ubuntu-cdimage/+bug/1883040
  - https://bugs.launchpad.net/ubuntu-cdimage/+bug/1886148

  Most probably "groovy daily won't boot anymore on some older BIOS
  boxes" or the 1883040 bug, as it's impacting only my older slow boxes
  (but it's also different).

  thumb-drive has been successfully used in QA live & install tests,
  however it won't boot on

  hp dc7700 (c2d-e6320, 5gb, nvidia quadro nvs 290)
  hp dc7900 (c2d-e8400, 4gb, intel 4 series integrated i915)

  The last timing issue also impacted d755-5 (older dell optiplex 755)
  when dc7700 was impacted, but that booted successfully this ISO.

  I see thumb-drive flash as it tries to boot, eventually it stops
  flashing, but no boot. At least 6 attempts made to boot, thumb-drive
  being tested on another 2x d755 boxes..

  (ISO written by mkusb/dus & gnome-disks [#11] is identical)

  ** Expected result

  It boots & is usable

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-cdimage/+bug/1899308/+subscriptions