Remove NFSv4 inheritance flags in setfacl recursive mode

[mhorne063_gmail.com]

Summary

When modifying ACLs recursively, strip the inheritance flags from each of the ACL entries.

This code was extracted from the patch originally submitted to PR 155163

[mhorne063_gmail.com]

No code changes, just updated from changes made to D15060

[emaste]

bin/setfacl/setfacl.c

129

This should probably have a comment to give more information about what is meant by "sanitize".

153–155

style(9) puts the operator at the end of the line when wrapping is necessary

[lattera-gmail.com]

It has been quite a few years since I originally wrote this code. This might be able to carve out some time these next two weeks to refactor it. There are parts of this patch that feel a bit awkward and could likely be improved upon.

Part of me wonders if the inheritance sanitization code should live in libc with the rest of the ACL code. Doing so would allow other projects that deal with ACLs to take advantage of it.

[mhorne063_gmail.com]

Cleaned this up a bit. The logic is the same but should be a bit more obvious now.

Unless I missed something there was no reason to be calling acl_dup() like we did in the last iteration of the patch.

Also made it more obvious that we were ignoring return values by casting them to void

[markj]

bin/setfacl/setfacl.c

143

The callers are already checking R_flag, this is redundant.

154

Don't we also need to call acl_set_flagset_np()? This call is just modifying a stack-local variable.

[markj]

bin/setfacl/setfacl.c

131

This should probably be documented in the man page.

149

acl_flagset will be left uninitialized if an error occurs. This may not be possible in practice, but we should handle it anyway.

[markj]

bin/setfacl/setfacl.c

154

Ignore this, I misunderstood the API: acl_flagset_t is a uint16_t *, not a uint16_t.

[mhorne063_gmail.com]

Ahh I see, I misunderstood the purpose of duplicating the ACL. Much better now.

Thanks Mark for being so quick to fix this up and commit.