21/12/13 18:02, intrigeri wrote:
> Hi,
>
> shipping a 64-bit kernel is a pre-requisite for UEFI support, as many
> (if not all) 64-bit EFI firmwares don't support booting a 32-bit
> kernel, and we won't be supporting any (rare, and mostly obsolete)
> 32-bit UEFI hardware.
>
> The feature/amd64-kernel branch implements this: it replaces our good
> old 686-pae kernel with a 64-bit one, and adjusts various pieces of
> documentation and the test suite (I've not tested this part yet, still
> lacking a suitable setup) accordingly.
>
> Note that the code bits of this branch have been in experimental for
> many weeks now.
>
> Drawback: possibly some UX degradation inside VirtualBox, see ticket.
> But the VirtualBox guest modules have been broken since March, and
> anyway it seems obvious to me that if we had to choose between
> supporting running Tails as a guest inside VirtualBox, or running it
> on UEFI hardware, we would choose the latter.
Agreed. I suppose it's a bit sad, but frankly the only thing I miss from
the guest additions is the ability to set a higher resolution than
1024x768. Perhaps we can at least get that to work at some point.
> Candidate for 0.23, please review'n'merge into devel.
>
> Assigned to anonym, who's going to be the RM for 0.23, but it would be
> really great if other people did test builds from experimental on
> various hardware to ensure it works fine
Review passed.
I've tested it successfully on the following systems:
* An old Pentium M (i686 with PAE)
* Two semi-modern x86_64 laptops
* VirtualBox (64-bit only)
* libvirt/kvm (both 32-bit and x86_64)
No issues at all, so I'm pleased and ready to merge, except I'd like an
answer to my question below first.
> (modulo the broken memory
> erasure on shutdown, but that's another story, thanks in advance for
> not reporting this again and again).
So the plan is that I should merge the branch despite this regression?
Cheers!
