[Merge] ~codyshepherd/livecd-rootfs/+git/livecd-rootfs:fix/riscv64-manifest-gen into livecd-rootfs:ubuntu/jammy

[Cody Shepherd]

> The arch specific things are in the case...esac clause which is where you're adding the call to create_manifest - but how is it different from the generic one?

Without my change, inside the subarch if-branch of the riscv64-specific switch case, the install_grub() function returns early without creating manifests, etc. This causes hook failures later (specifically for CPC, in the qcow2-image.binary), when those manifests are expected to exist. i.e. I'm fixing an edge case where manifests never get generated.

> Can you confirm that the manifests produces after running your tests do not have duplicate snap lists?

I don't follow your logic for what condition might occur to produce duplicate snap listings. Under normal circumstances (i.e. the $SUBARCH if-branch of the riscv64 switch case not being taken), create_manifest() gets called at the end of install_grub(). My change adds a call to create_manifest() directly before a return from install_grub(). In both of these cases, the create_manifest() function is called directly before leaving install_grub(); I don't see an opportunity for repeated calls to snap-seed-parse. Can you elaborate on the scenario you are considering?
-- 
https://code.launchpad.net/~codyshepherd/livecd-rootfs/+git/livecd-rootfs/+merge/475962
Your team Ubuntu Core Development Team is subscribed to branch livecd-rootfs:ubuntu/jammy.