Fix regression in get_partition_node_name

[uecasm]

Fixes part of this issue

Changes proposed in this pull request:

• Fixes regression in get_partition_node_name for targets that do not support the START column in lsblk.
  • This restores prior behaviour, which may reintroduce the bug that was intended to fix, but that's better than breaking existing functionality.

[schaefi]

@uecasm Uhh, good catch thanks.

Can you give me a short background explanation for which devices we hit the regression ? The places in code where we call it targets the storage disk which is expected to hold a partition table and thus proper start addresses. Thanks

[schaefi]

Ah forget my last question, you referenced an issue link.. just reading

[schaefi]

A reproducer to the issue would be great. I believe if you can send

sfdisk -d /dev/nvme0n1 > table

From an affected system, I should be able to reproduce this also on a loop device

Thoughts ?

[uecasm]

This isn't device-specific, it's an issue with the version of lsblk. As noted, I'm testing on a Leap 15.5 target and the version in there does not support this column. The START column was not added until util-linux-systemd 2.38, and 15.5 has 2.37.4.

[schaefi]

This isn't device-specific, it's an issue with the version of lsblk. As noted, I'm testing on a Leap 15.5 target and the version in there does not support this column. The START column was not added until util-linux-systemd 2.38, and 15.5 has 2.37.4.

I see, thanks and the comment makes that part clear 👍