Add support for aboot bootc images #1033
Conversation
There was a problem hiding this comment.
Thank you, this looks fine. I think we will need some sort of test though because this feature will move into the generic "images" library in the near future and without a test there is much more regression risk.
(unrelated rambling) I would actually really like to see this being declarative and tied into our imagestypes.yaml we have for non-bootc distros. I.e. the bootc image would contain a partition table YAML and we would have the ability to put payload path into that. This would allow people to build bootc images that contain arbitrary firmware files that then would be written to their (special) partitions (and tranditional images would support it in the same way as part of their normal imagetypes.yaml if we ever build imags for boards like the dragonboard that require this kind of partition payload on the written image) Doing aboot/ukiboot would just be a special case of this more general feature (I hope this extremly short summary makes sense, happy to expand if anything in unclear). But we are not there yet so having this to unblock you is fine.
alexlarsson
force-pushed
the
aboot-support
branch
from
5565830 to
820fffd
Compare
|
I agree that long term we want to allow the image itself to be much more descriptive and flexible. That is however a long term plan, and I would really like to have this in for now as some kind of basic support so we can start moving towards bootc. I will look into figuring out some test for this. |
I fully agree with this. It will be great to have for a lot of special SBC (and similar) scenarios that we don't cover well now. |
alexlarsson
force-pushed
the
aboot-support
branch
from
820fffd to
b6704ce
Compare
|
Added a test in test_manifest.py |
alexlarsson
force-pushed
the
aboot-support
branch
2 times, most recently
from
d8e44bc to
ffa413e
Compare
achilleas-k
changed the title
bib/cmd/bootc-image-builder/image.go
Outdated
| @@ -251,6 +252,23 @@ func genPartitionTable(c *ManifestConfig, customizations *blueprint.Customizatio | |||
| } | |||
| } | |||
|
|
|||
| if c.SourceInfo != nil && c.SourceInfo.KernelInfo != nil && c.SourceInfo.KernelInfo.HasAbootImg { | |||
| idx := slices.IndexFunc(partitionTable.Partitions, func(part disk.Partition) bool { | |||
| return part.Label == "aboot_a" || part.Label == "ukiboot_a" || part.Type == "46" | |||
There was a problem hiding this comment.
This all feels like it could use some comments with links to more information
There was a problem hiding this comment.
Added some comments with links
| testdata_path = tmp_path / "testdata" | ||
| testdata_path.write_text("some test data", encoding="utf-8") | ||
|
|
||
| # Create derived container with the custom partitioning with an aboot |
If the target image contains an aboot.img, then we automatically write it to the A slot boot partition.
This adds a test that a write-device stage is correctly generated if the partition table contains the right partition and an aboot.img in the modules dir.
alexlarsson
force-pushed
the
aboot-support
branch
from
ffa413e to
b2683b6
Compare
|
Reminder for myself - this will need to get ported to https://github.com/osbuild/images/tree/main/pkg/distro/bootc so that we can merge #1014 |
This detects if a aboot.img is in the kernel module directory, and if so, adds an osbuild stage that writes it to the a slot boot partition. This currently supports traditional aboot (boot_a partition), and ukiboot with EFI (ukiboot_a partition).