Re: [PATCH v3 0/5] Rebuild mount logic

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

Am Thu, 10 Jun 2021 10:12:24 +0300
schrieb Anton Mikanovich <amikan@ilbers.de>:

> 08.06.2021 11:54, Henning Schild wrote:
> > I think the core of the problem is that "umount" can fail and that
> > failure is not handled. Before we maybe ignored it ...
> >
> > In the logs i sent it failed with "target is busy"
> >
> > Henning  
> 
> I've found the reason and it was quite strange:
> The construction `$(($(< filename) + 1))` with `1` inside the file
> was returning 1, while equivalent `$(($(cat filename) + 1))` was
> working correctly.
> So I've changed it to little bit slower `cat` version and now there
> is no issues when building Isar inside the container.
> Please look at the patch: "[PATCH] buildchroot: Remove bashism when 
> reading file content"

Cool, reason is that bitbake is not using bash. It might be using
whatever the default shell is. At least i think so.

Thanks!
Henning