Hi Phil, Brian and others,

I'd like to point to systemd errors caused by same names for GPT
partitions. Ours all have the name "primary" at the moment causing
"journalctl -p err -b0" to show error messages like:

"Dev dev-disk-by\x2dpartlabel-primary.device appeared twice with
different sysfs paths"

Also gdisk sets default names like e.g. "Linux filesystem".

We should be able to provide GPT partitions without a name. The UEFI
spec doesn't define if that name may be empty. It just has to be NULL
terminated. Or can you point to a document stating explicitly that the
name may not be empty?

Should we change the mkpart syntax from
mkpart name [fs-type] start end
mkpart [name [fs-type]] start end

That doesn't provide us a chance to provide the fs-type with empty name.

Or should we use a special keyword like this:
mkpart \<empty\> [fs-type] start end

I prefer the special key word. That just causes
to be created in case this feature is not implemented.

What are your thoughts?


Reply via email to