Re: [PATCH v7 3/3] meta: Deploy image build artifacts into distro- and image-specific subdirs

["'MOESSBAUER, Felix' via isar-users" <isar-users@googlegroups.com>]

On Fri, 2026-04-10 at 12:46 +0200, Zhihang Wei wrote:
> On 4/10/26 11:44, MOESSBAUER, Felix wrote:
> > On Thu, 2026-02-26 at 17:26 +0100, Zhihang Wei wrote:
> > > From: Ilia Skochilov <iskochilov@ilbers.de>
> > > 
> > > When building different distros or images for the same machine (e.g.
> > > phyboard-mira-bullseye and phyboard-mira-bookworm, or isar-image-ci
> > > and isar-image-base), artifacts with identical names (e.g. DTB files)
> > > may be deployed to the same location, resulting file conflicts and
> > > build failures.
> > > 
> > > Deploy image build artifacts into a subdirectory named
> > > ${DISTRO}-${IMAGE_PN} to avoid such collisions.
> > > 
> > > Since the directory now encodes the distro and image name, remove the
> > > ${DISTRO} and ${IMAGE_PN} prefixes from artifact filenames to
> > > keep the names short and clear.
> > > 
> > [shortening a bit]
> > 
> > > +IMAGE_PN ?= ""
> > > +DEPLOY_DIR_IMAGE = "${DEPLOY_DIR}/images/${MACHINE}${@('/%s-%s' % (d.getVar('DISTRO'), d.getVar('IMAGE_PN'))) if d.getVar('IMAGE_PN') != '' else ''}"
> > This cannot work, as the value of IMAGE_PN is late-expanded, hence
> > different in the initrd recipe and the image recipe. The
> > DEPLOY_DIR_IMAGE variable must only use data that is machine specific,
> > not image specific (distro might be fine, though as this is identical
> > for both initrd and image).
> Understand the merge_wic_sbom problem here.
> 
> "merge_wic_sbom" cares only about the customized initrd, right? 
> 

Yes.

It also needs the imager sbom and the rootfs sbom, but the problematic
initrd line [1] is anyways only active if a custom initrd is used.

[1]
https://github.com/ilbers/isar/blob/master/meta/classes-recipe/imagetypes_wic.bbclass#L223

> Then we
> could probably change image artifacts (kernel, image, default initrd)
> deploy path only in the image.bbclass, and not touching
> DEPLOY_DIR_IMAGE.
> > Voting for a revert of this patchset.
> > 
> > PS: Regarding the DTB issue? Can't we simply prefix the DTBs with the
> > IMAGE_PN? In do_copy_boot_files we know the IMAGE_PN and during wic-
> > imaging we can remove the prefix again.
> Prefixing DTBs with IMAGE_PN only may not be enough. We have two test 
> cases: - Same machine, different distros, same image - Same machine, 
> same distro, different images Both failed before this patch.

Let's see if my proposal passes the new CI test. I'm still not
convinced that the TMPDIR split is the right approach either. Upstream
I only find a TMPDIR split proposal in meta-ti [2], but not in the poky
/ OE core layer.

[2]
https://lists.yoctoproject.org/g/meta-ti/topic/patch_multiconfig_use/91431962

Felix

> Zhihang

-- 
You received this message because you are subscribed to the Google Groups "isar-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isar-users+unsubscribe@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/isar-users/0cbdfdc26b691518b17f0d229c4b8163f81d9682.camel%40siemens.com.