Bug#842617: Please also ship systemd-boot on arm64

2018-11-23 Thread Alexander Kurtz
On Sun, 2018-11-18 at 15:56 +0100, Michael Biebl wrote:
> Alexander, I've re-read your reply in
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=842617#32
> 
> If in the feature you actually plan to use the armhf bits on say the
> Raspberry Pi 2, I'm happy to recondider.

I'm sorry, but my only ARM machine is a RPI3 which supports arm64, so
I'll probably never use armhf myself. However, in addition to u-boot's
"fake" EFI implementation [0], there is now also a build of EDK2 for
32-bit ARM virtual machines available in Debian [1], so having systemd-
boot available on arm *might* have some benefit after all.

Best regards

Alexander Kurtz

[0] https://packages.debian.org/sid/u-boot-rpi
[1] https://packages.debian.org/sid/qemu-efi-arm


signature.asc
Description: This is a digitally signed message part


Bug#842617: Please also ship systemd-boot on arm64

2018-11-18 Thread Michael Biebl
Am 18.11.18 um 12:54 schrieb Martin Pitt:
> Hello Michael,
> 
> Michael Biebl [2018-11-17 15:47 +0100]:
>>> I know that arm64 EFI boot works (Canonical uses it in their internal
>>> OpenStack deployment), but as far as I know there is no existing
>>> armhf EFI implementation; so this would mean to ship dead bits. Or are
>>> you aware of any platform where this could actually be tested?
>>
>> Martin, do you still have concerns regarding enabling gnu-efi support
>> for armhf?
>> Personally I don't have any experience in that regard, all I can say is
>> that we apparently ship
>> https://packages.debian.org/sid/armhf/grub-efi-arm
>> (with a popcon count of 0 :-o )
> 
> Not a strong veto, if that somehow makes the packaging simpler and more
> consistent across architctures. I still don't quite like shipping dead bits
> that can't actually run anywhere, but they are relatively harmless, too.

Just wanted to get your input on this matter.
On second thought, let's just wait until we have a user request for this
where the user is able to actually use and test the functionality.

Alexander, I've re-read your reply in
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=842617#32

If in the feature you actually plan to use the armhf bits on say the
Raspberry Pi 2, I'm happy to recondider.

Regards,
Michael

-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature


Bug#842617: Please also ship systemd-boot on arm64

2018-11-18 Thread Martin Pitt
Hello Michael,

Michael Biebl [2018-11-17 15:47 +0100]:
> > I know that arm64 EFI boot works (Canonical uses it in their internal
> > OpenStack deployment), but as far as I know there is no existing
> > armhf EFI implementation; so this would mean to ship dead bits. Or are
> > you aware of any platform where this could actually be tested?
> 
> Martin, do you still have concerns regarding enabling gnu-efi support
> for armhf?
> Personally I don't have any experience in that regard, all I can say is
> that we apparently ship
> https://packages.debian.org/sid/armhf/grub-efi-arm
> (with a popcon count of 0 :-o )

Not a strong veto, if that somehow makes the packaging simpler and more
consistent across architctures. I still don't quite like shipping dead bits
that can't actually run anywhere, but they are relatively harmless, too.

Martin


signature.asc
Description: PGP signature


Bug#842617: Please also ship systemd-boot on arm64

2018-11-17 Thread Michael Biebl
On Mon, 31 Oct 2016 07:46:13 +0200 Martin Pitt  wrote:
> Alexander Kurtz [2016-10-30 22:05 +0100]:
> > Thanks a lot for your quick reply! You might want to consider also
> > adding "armhf" to that list, since both gnu-efi [0] and u-boot-rpi [1]
> > are also available there (in contrast to "armel" where gnu-efi is
> > missing).
> 
> I know that arm64 EFI boot works (Canonical uses it in their internal
> OpenStack deployment), but as far as I know there is no existing
> armhf EFI implementation; so this would mean to ship dead bits. Or are
> you aware of any platform where this could actually be tested?

Martin, do you still have concerns regarding enabling gnu-efi support
for armhf?
Personally I don't have any experience in that regard, all I can say is
that we apparently ship
https://packages.debian.org/sid/armhf/grub-efi-arm
(with a popcon count of 0 :-o )

Regards,
Michael

-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature


Bug#842617: Please also ship systemd-boot on arm64

2016-10-31 Thread Alexander Kurtz
On Mon, 2016-10-31 at 07:46 +0200, Martin Pitt wrote:
> I know that arm64 EFI boot works (Canonical uses it in their internal
> OpenStack deployment), but as far as I know there is no existing
> armhf EFI implementation; so this would mean to ship dead bits. Or
> are you aware of any platform where this could actually be tested?

Well, as I mentioned there's u-boot which apparently provides an UEFI
implementation nowadays [0], so the armhf variant of [1] should
theoretically be usable for booting an unmodified Debian on a Raspberry
Pi 2 [2]. Additionally, OVMF is also available for ARM (without the 64)
[3], though this is currently not packaged in Debian, see [4].

Please note that I've not yet tested any of this, but I think it would
be great if at some point you could boot Debian using systemd-boot no
matter if you're on armhf, arm64, i386 or amd64 on both virtual and
physical hardware. That's certainly not going to be easy, but fixing
the Debian packaging of systemd seemed like the lowest hanging fruit!
;-)

Best regards

Alexander Kurtz

[0] https://media.ccc.de/v/946-uefi-grub2-on-raspberry-pi
[1] https://packages.debian.org/sid/u-boot-rpi
[2] https://wiki.debian.org/RaspberryPi
[3] https://www.kraxel.org/repos/jenkins/edk2/
[4] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=842683

signature.asc
Description: This is a digitally signed message part


Bug#842617: Please also ship systemd-boot on arm64

2016-10-30 Thread Martin Pitt
Alexander Kurtz [2016-10-30 22:05 +0100]:
> Thanks a lot for your quick reply! You might want to consider also
> adding "armhf" to that list, since both gnu-efi [0] and u-boot-rpi [1]
> are also available there (in contrast to "armel" where gnu-efi is
> missing).

I know that arm64 EFI boot works (Canonical uses it in their internal
OpenStack deployment), but as far as I know there is no existing
armhf EFI implementation; so this would mean to ship dead bits. Or are
you aware of any platform where this could actually be tested?

Thanks,

Martin

-- 
Martin Pitt| http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)


signature.asc
Description: PGP signature


Bug#842617: Please also ship systemd-boot on arm64

2016-10-30 Thread Julian Andres Klode
On 30 October 2016 at 22:05, Alexander Kurtz  wrote:
> On Sun, 2016-10-30 at 22:45 +0200, Martin Pitt wrote:
>> But indeed gnu-efi does exist on arm64 in testing/unstable now, so I
>> added it
>
> Thanks a lot for your quick reply! You might want to consider also
> adding "armhf" to that list, since both gnu-efi [0] and u-boot-rpi [1]
> are also available there (in contrast to "armel" where gnu-efi is
> missing).
>

Yes, (JFTR) armel is missing some assembly stuff used by gnu-efi, see:

ttps://buildd.debian.org/status/fetch.php?pkg=gnu-efi&arch=armel&ver=3.0.2-1&stamp=1431473135

-- 
Julian Andres Klode  - Debian Developer, Ubuntu Member

See http://wiki.debian.org/JulianAndresKlode and http://jak-linux.org/.



Bug#842617: Please also ship systemd-boot on arm64

2016-10-30 Thread Alexander Kurtz
On Sun, 2016-10-30 at 22:45 +0200, Martin Pitt wrote:
> But indeed gnu-efi does exist on arm64 in testing/unstable now, so I
> added it

Thanks a lot for your quick reply! You might want to consider also
adding "armhf" to that list, since both gnu-efi [0] and u-boot-rpi [1]
are also available there (in contrast to "armel" where gnu-efi is
missing).

Thanks again!

Alexander Kurtz

[0] https://packages.debian.org/sid/gnu-efi
[1] https://packages.debian.org/sid/u-boot-rpi

signature.asc
Description: This is a digitally signed message part


Bug#842617: Please also ship systemd-boot on arm64

2016-10-30 Thread Martin Pitt
Control: tag -1 pending

Hello Alexander,

Alexander Kurtz [2016-10-30 21:32 +0100]:
> While the i386 and amd64 variants of the systemd package do ship the
> systemd-boot binary [0,1], it seems the arm64 variant does not [2] (at
> least there is nothing under /usr/lib/systemd/boot/). As Fedora manages
> to ship the related files just fine [3], I guess this is just a simple
> oversight in the Debian packaging.

Somewhere between "no gnu-efi available when we first looked at this"
and "nobody tested it on arm64". But indeed gnu-efi does exist on
arm64 in testing/unstable now, so I added it:

  https://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=b1a3bccd7

This will have to be reverted for the jessie backport, though.

Martin
-- 
Martin Pitt| http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)


signature.asc
Description: PGP signature


Bug#842617: Please also ship systemd-boot on arm64

2016-10-30 Thread Alexander Kurtz
Package: systemd
Version: 231-10

Hi!

While the i386 and amd64 variants of the systemd package do ship the
systemd-boot binary [0,1], it seems the arm64 variant does not [2] (at
least there is nothing under /usr/lib/systemd/boot/). As Fedora manages
to ship the related files just fine [3], I guess this is just a simple
oversight in the Debian packaging. Given that u-boot can apparently
nowadays provide an UEFI emulation [4], this might make things like
"bootctl install" on your Raspberry Pi 3 possible!

Best regards

Alexander Kurtz

[0] https://packages.debian.org/sid/i386/systemd/filelist
[1] https://packages.debian.org/sid/amd64/systemd/filelist
[2] https://packages.debian.org/sid/arm64/systemd/filelist
[3] http://koji.fedoraproject.org/koji/rpminfo?rpmID=8320101
[4] https://media.ccc.de/v/946-uefi-grub2-on-raspberry-pi

signature.asc
Description: This is a digitally signed message part