chown: use Link instead of Chown for dup inodes

[giuseppe]

If the inode was already encountered and chowned, use link(2) instead
of chown(2).

This is needed when the underlying storage (as it could be overlay
with index=off) breaks the hard link on copy up.

#1144 added the initial
check.

Closes: #1257

Signed-off-by: Giuseppe Scrivano gscrivan@redhat.com

[flouthoc]

LGTM

[nalind]

This comment really confused me - it didn't describe the checks on the next line, and after suggesting that if the checks were true, we still needed to do something, it just returned.

[nalind]

Do we want to take the hit of checking inode numbers after the system.Lchown(), and re-linking things that we know were linked before we started walking the tree?

[nalind]

I suppose if we only checked for "that multiply-linked thing got a new inode number" when st.Nlinks > 1, it might not hurt too bad. If we kept a mapping from old inode numbers to new link targets, and relinked it instead of chowning the no-longer-linked lower item, would that work?

[giuseppe]

I suppose if we only checked for "that multiply-linked thing got a new inode number" when st.Nlinks > 1, it might not hurt too bad. If we kept a mapping from old inode numbers to new link targets, and relinked it instead of chowning the no-longer-linked lower item, would that work?

that is a much better solution. I've played with it and it seems to work fine, I'll test it more and update the PR

[rhatdan]

LGTM
@nalind PTAL

[nalind]

Are there any tests we can add for this?

[nalind]

LGTM