disk: Default to direct-io=off in losetup wrapper

[ckyrouac]

losetup defaults to on when the arg is not present

I mistakenly tested the original wrapper changes with my patched image that included the bootc fix. Verified this works on my machine.

[cgwalters]

losetup defaults to on when the arg is not present

Wait sorry...are you sure? We went over this in containers/bootc#375 (comment)

[ckyrouac]

that's the behavior I see on my machine, and this is in the man page:

       --direct-io[=on|off]
           Enable or disable direct I/O for the backing file. The optional argument can be either on or off. If the optional argument is omitted, it defaults to on.

❯ losetup --version
losetup from util-linux 2.39.4

[ckyrouac]

also I meant to say "direct-io" defaults to on

[ckyrouac]

https://github.com/util-linux/util-linux/blob/926b6077333554924756ba648c9df38c803c48bc/sys-utils/losetup.c#L818

not familiar with getopt_long but maybe this clause is always executed which is why it defaults to on?

[cgwalters]

It's confusingly phrased...I took a stab at clarifying the upstream docs util-linux/util-linux#3002

not familiar with getopt_long but maybe this clause is always executed which is why it defaults to on?

That bit should only be executed if the argument was provided.

[ckyrouac]

That bit should only be executed if the argument was provided.
makes sense

ah, there was a separate issue preventing the partitions being created when I tried losetup without the direct-io argument earlier. Just did another test and it defaults to off for me when the arg is omitted. There's also a bug in the in-line bash script that threw me off, it passes the original arguments instead of the modified args. I guess adding --direct-io=off at the end of the command overrides the previous arg that set it to on. I'll just leave it as-is since it's working.

[cgwalters]

There's also a bug in the in-line bash script that threw me off, it passes the original arguments instead of the modified args.

haha oh man, yes

[germag]

I guess adding --direct-io=off at the end of the command overrides the previous arg that set it to on. I'll just leave it as-is since it's working.

Yes, it's an unintended behavior of getopt, but now it's a "de facto" standard