[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Updated installer images



Hi Frank!

On 09/21/2017 09:59 AM, Frank Scheiner wrote:
>> No, not at all. If you want to install on 32-bit hardware, you still
>> need to
>> use the powerpc NETINST image which can be found here:
>>
>>> http://jenkins.kfreebsd.eu/jenkins/view/cd/job/debian-cd_sid_powerpc/ws/
> 
> Now that's interesting, I assumed the powerpc architecture was dropped
> with Stretch. But apparently it's still in Sid. I appreciate this for
> all my 32 bit PowerPC machines.

powerpc was dropped as a *release* architecture for Stretch. However,
that does not mean it cannot continue to exist in sid. In fact, powerpc
is not even maintained in Debian Ports but the packages are still built
on the regular buildds maintained by DSA (Debian's official sysadmin
team) and the packages are uploaded to the main FTP archive instead
of the Debian Ports FTP archive.

Thus, making powerpc a release architecture again is currently a matter
of flipping a switch, at least technically.

>> This must a bug specific to Apple (New World?) machines. I don't see this
>> problem when installing on my POWER7 machine.
> 
> Christoph mentioned [#825840] as possible cause. From what I read it
> also looks like this does not affect people that use serial consoles,
> but I haven't yet tried that on my Xserve G5.
> 
> [#825840]: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=825840

It apparently affects ATI devices only when using the Open Firmware
framebuffer driver. Not so much of a big deal, I think. Most likely
a regression that one should be able by bi-secting the kernel.

>>> https://anonscm.debian.org/cgit/d-i/partman-partitioning.git/tree/lib/disk-label.sh
>>>
>>
>> I will do that tomorrow.
> 
> Great, I'll follow this to see how the process works.

However, I'm still unsure what partitioning scheme and bootloader we are
going to use for ppc64. So, please let's continue discuss this decision
first.

>> However, I'm still not sure whether we should actually use GRUB on
>> ppc64 by default.
> 
> Maybe both should be used for a time frame. Yaboot works pretty well for
> network booting - although not related to on disk installations -
> without much adaptation between Apple G5s and IBM POWER5s (and possibly
> newer machines) and on-disk file systems or its default configurations
> are not an issue in this case. But maybe Grub can work pretty well
> there, too, I just haven't tested it yet.

I'm solely talking about the default bootloader which gets installed
by Debian Installer. Maintaining two bootloaders in Debian Installer
is a bit difficult and requires lots of testing, it should be possible,
however.

But the problem with yaboot is also that it is no longer maintained
upstream [1] which means it can potentially cause more trouble in
the future.

Ideally, we would be using GRUB on as many platforms as possible
as this results in fewer moving parts in the whole software
stack and consequently less maintenance burden and bugs, see [2].

There are, of course, some systems left which cannot use GRUB. But
we can either try to migrate them to u-boot, if possible, or just
continue to carry their specific bootloaders.

Either way, I would be very interested to hear about test results
of GRUB on ppc64. There are instructions on how to set up GRUB on
ppc64 and it should generally work. However, I don't know what
possible the limitations are.

Adrian

> [1] http://yaboot.ozlabs.org/
> [2]
https://www.redhat.com/archives/fedora-devel-list/2008-January/msg00861.html

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913


Reply to: