From: Marcus Folkesson <marcus.folkesson@gmail.com>
To: "Martin Hundebøll" <martin@geanix.com>
Cc: Ross Burton <Ross.Burton@arm.com>,
openembedded-core <openembedded-core@lists.openembedded.org>
Subject: Re: [OE-core] [PATCH v4 0/2] image-bootfiles: new class
Date: Tue, 11 Jun 2024 17:04:08 +0200 [thread overview]
Message-ID: <ZmhnaAv3KziwhCKF@gmail.com> (raw)
In-Reply-To: <e21366a98f2769cb7460ad6ebfe86ad3a51c24cd.camel@geanix.com>
[-- Attachment #1: Type: text/plain, Size: 3570 bytes --]
Hello Martin,
On Tue, Jun 11, 2024 at 04:44:04PM +0200, Martin Hundebøll wrote:
> On Tue, 2024-06-11 at 15:22 +0200, Marcus Folkesson wrote:
> > Hi Ross,
> >
> > On Tue, Jun 11, 2024 at 10:37:06AM +0000, Ross Burton wrote:
> > > On 30 May 2024, at 10:53, Marcus Folkesson via
> > > lists.openembedded.org
> > > <marcus.folkesson=gmail.com@lists.openembedded.org> wrote:
> > > > The image-bootfiles class is used to put all files listed in
> > > > IMAGE_BOOT_FILES into the root filesystem.
> > > >
> > > > IMAGE_BOOT_FILES is used by the bootimg-partition wic plugin to
> > > > put the
> > > > files into a boot partition.
> > > > Be able to list files as "boot files" in e.g. your .conf or image
> > > > files
> > > > instead of install those in every recipe is a good thing.
> > > >
> > > > It is not always desired to have a separate boot partition for
> > > > boot
> > > > files. Sometimes it could be good to have them as a part of the
> > > > root
> > > > filesystem.
> > > >
> > > > For example, if a double copy strategy is used for update the
> > > > system,
> > > > then you probably want to update both the boot files and root
> > > > filesystem
> > > > at the same time as there may be dependencies.
> > >
> > > In my mind, IMAGE_BOOT_FILES is a workaround for the fact that some
> > > /boot partitions (such as ones generated by wic) are not managed by
> > > bitbake directly. If you have a setup where you just have a / that
> > > contains /boot isn’t adding eg grub to IMAGE_INSTALL sufficient to
> > > get it in the right place in the rootfs?
> >
> > I don't know about the workaround, but it would'nt surprise me as it
> > is
> > not handled by bitbake as it is now.
> >
> > For some packages yes, but not for all. If you, for instance, have an
> > embedded
> > system where you depend on other files that are critical for the boot
> > process, there is no uniform way to specify that for those files.
> >
> > IMAGE_BOOT_FILES is good as it let you include e.g. ramdisks and such
> > that does not have installation scripts to the boot partition.
> >
> > The use case that I had was that I was first using a separate boot
> > partition using the bootimg-partition wic plugin. Everything was
> > good.
> > Then I wanted to include those files into the root filesystem instead
> > to
> > be able to do an atomic update on everything, but there is no good
> > way
> > to achive that.
>
> I'd suggest updating the recipe providing those boot files, so it
> installs them into /boot (maybe in addition to deploying them to
> $DEPLOYDIR).
Yes, that is the option.
>
> > This image-class make the swap from bootimg-partition to rootfs
> > seamless as it uses the same mechanics for both implementations.
>
> I have always disliked the concept of recipes pulling files from
> ${DEPLOY_DIR_IMAGE}. That folder is an output folder, and should not be
> used as input for other packages.
>
> Instead, a proper do_populate_deploy task similar to
> do_populate_sysroot would make it possible to DEPEND += on deployed
> files from other recipes.
I would probably have implemented it differently if we did not already
had the bootimg-partition plugin that works this way.
I think there is a point in having them work in a similar a way.
>
> But this is probably a discussion for another mail thread. Sorry for
> raising my opinion so late in the process.
No problem, all feedback is good. Thank you
>
> // Martin
Best regards,
Marcus Folkesson
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
next prev parent reply other threads:[~2024-06-11 14:57 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-05-30 9:53 [PATCH v4 0/2] image-bootfiles: new class Marcus Folkesson
2024-05-30 9:53 ` [PATCH v4 1/2] bootimg-partition: break out code to a common library Marcus Folkesson
2024-05-30 9:53 ` [PATCH v4 2/2] image-bootfiles.bbclass: new class, copy boot files to root filesystem Marcus Folkesson
2024-05-31 12:09 ` Quentin Schulz
2024-06-16 4:26 ` Konrad Weihmann
2024-06-17 6:23 ` [OE-core] " Marcus Folkesson
2024-06-11 10:37 ` [OE-core] [PATCH v4 0/2] image-bootfiles: new class Ross Burton
2024-06-11 13:22 ` Marcus Folkesson
2024-06-11 14:44 ` Martin Hundebøll
2024-06-11 15:04 ` Marcus Folkesson [this message]
2024-07-16 11:28 ` Ross Burton
2024-07-22 7:36 ` Marcus Folkesson
2024-06-14 20:25 ` Marcus Folkesson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ZmhnaAv3KziwhCKF@gmail.com \
--to=marcus.folkesson@gmail.com \
--cc=Ross.Burton@arm.com \
--cc=martin@geanix.com \
--cc=openembedded-core@lists.openembedded.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox