[patch 1/1] image tools extention not install suggested deb

[patch 1/1] image tools extention not install suggested deb

[Roberto A. Foglietta]

[-- Attachment #1.1: Type: text/plain, Size: 159 bytes --]

Hi all,

 the option --no-install-suggests is proposed to be used into inage tools
extension but regressions should be considered. In attachment.

 Cheers, R-

[-- Attachment #1.2: Type: text/html, Size: 289 bytes --]

[-- Attachment #2: 0001-image-tools-extension-uses-no-install-suggests.patch --]
[-- Type: application/x-patch, Size: 1455 bytes --]

[patch 1/1] linux module postinst generalisation

[Roberto A. Foglietta]

[-- Attachment #1.1: Type: text/plain, Size: 239 bytes --]

Hi all,

 linux module postinst generalisation is needed because sometimes (corner
cases) ls /lib/modules includes not valid items that make depmod fails in
total. In this way, it fails specifically only and no damage is done.

Cheers, R-

[-- Attachment #1.2: Type: text/html, Size: 489 bytes --]

[-- Attachment #2: 0001-linux-module-postinst-generalisation.patch --]
[-- Type: application/x-patch, Size: 839 bytes --]

Re: [patch 1/1] linux module postinst generalisation

[Henning Schild]

Hey Roberto,

that patch should be inline so it can be reviewed and later applied. You
did send it as attachment. So now people have a harder time reviewing
it and replying to certain lines of it, and anyone who wanted to apply
it would maybe not be able to use "git am" as they are used to.

Am Fri, 16 Sep 2022 14:44:47 +0200
schrieb "Roberto A. Foglietta" <roberto.foglietta@gmail.com>:

> Hi all,
> 
>  linux module postinst generalisation is needed because sometimes
> (corner cases) ls /lib/modules includes not valid items that make
> depmod fails in total. In this way, it fails specifically only and no
> damage is done.

I think that patch kind of makes sense. But maybe you can go into more
detail how such a "broken" entry can be there in the first place.
Normally one would expect an isar-built image to have exactly one
kernel and one "working" module directory for that.
Maybe you should rather find the package that drops invalid directories
and repair that. Or maybe it is not "invalid" after all and our "ls"
matching is too simple.
And an interator over multiple directories would maybe also fail once
it finds the one "broken" directory.

regards,
Henning

> 
> Cheers, R-
> 

Re: [patch 1/1] linux module postinst generalisation

[Roberto A. Foglietta]

[-- Attachment #1: Type: text/plain, Size: 664 bytes --]

Il Lun 19 Set 2022, 09:31 Henning Schild <henning.schild@siemens.com> ha
scritto:

> Hey Roberto,
>
> [...]
>

And an interator over multiple directories would maybe also fail once
> it finds the one "broken" directory.
>

One iteration might fail but nicely because others succeed. Yes, in theory
should not be found any invalid entry in that folder but integraring 3rd
party proprietary packages, it could happen and it is much easier to
generalise the approach rather than fix that packages.

Moreover, why not something like

depmod -a ${KERNEL_VERSION}

After all, the opposite of a generalisation is a specific call that cannot
fails.

I hope this helps, R-

[-- Attachment #2: Type: text/html, Size: 1604 bytes --]