Re: [PATCH v2] conf: Add support for new debian distro "buster"

[Henning Schild <henning.schild@siemens.com>]

Am Thu, 16 Aug 2018 17:14:00 +0200
schrieb Jan Kiszka <jan.kiszka@siemens.com>:

> On 2018-08-16 17:10, Henning Schild wrote:
> > Am Thu, 16 Aug 2018 17:04:53 +0200
> > schrieb Jan Kiszka <jan.kiszka@siemens.com>:
> >   
> >> On 2018-08-16 17:03, Henning Schild wrote:  
> >>> Am Thu, 16 Aug 2018 16:51:29 +0200
> >>> schrieb Jan Kiszka <jan.kiszka@siemens.com>:
> >>>      
> >>>> On 2018-08-16 16:49, [ext] Henning Schild wrote:  
> >>>>> Am Thu, 16 Aug 2018 14:57:44 +0300
> >>>>> schrieb "Maxim Yu. Osipov" <mosipov@ilbers.de>:
> >>>>>         
> >>>>>> Hi Henning,
> >>>>>>
> >>>>>> It seems that under jessie we have similar problem with
> >>>>>> unsupported syscall with qemuarm as we had with qemuarm64 on
> >>>>>> stretch:
> >>>>>>
> >>>>>> cat
> >>>>>> build/tmp/work/debian-buster-armhf/buildchroot-target/temp/log.do_build:
> >>>>>> === Setting up libperl5.26:armhf (5.26.2-7) ...
> >>>>>> Setting up man-db (2.8.4-2) ...
> >>>>>> Building database of manual pages ...
> >>>>>> qemu: Unsupported syscall: 383
> >>>>>> ===
> >>>>>>
> >>>>>> May be it makes sense to exclude for now qemuarm from buster
> >>>>>> patch (at least while we support jessie)?  
> >>>>>
> >>>>> No. For arm64 i agree, we have that problem and can not build
> >>>>> buster images from any distro < buster. But for arm we just have
> >>>>> that problem for debian8 and not 9, so arm should be part of the
> >>>>> patch.  
> >>>>
> >>>> So, qemu-user-static from buster works fine? Is there possibly
> >>>> some backport of that available for stretch?  
> >>>
> >>> I did not try that yet, and there do not seem to be backports.
> >>>      
> >>
> >> As it's "static", we would likely be fine with the buster package -
> >> provided that helps.  
> > 
> > I am just testing that ... will let you guys know.
> >   
> 
> Perfect.

Buster can build itself ... so it does make sense to include the full
code into Isar, in case someone is already on buster. I did send a v3.

> > Even if in Isar we will either have to document that or try and
> > issue warnings. Putting that qemu in "our" container would only
> > work for people using that.  
> 
> It would be a workaround. I parallel, the issue could be brought up 
> upstream. I'm sure Debian is interested as well to support 
> qemu-debootstrap on stretch for buster...

Sure, but debian bug reporting is painfull since you need a debian
machine to do so.

Henning

> Jan
>