Thanks for your suggestion.
In my opinion the null point is a incorrect state for the acl, then I
return -EINVAL.
I will check this state transition in XFS.
2009/12/21 Christoph Hellwig <[email protected]>:
> I like taking these checks into posix_acl_valid, but I think the patch
> submission needs a bit more work.
>
> All the patches are extremly whitespace mangled. ?And I don't think
> splitting them up makes a whole lot of sense, when we do API changes
> like this we usually fix up all callers. ?So please try to fix your
> mailer, merged them all into one, and maybe also chose a more
> descriptive subject line, e.g.
>
> Subject: take checks for NULL and error pointers into posix_acl_valid()
>
>
> And btw, at least the XFS change seems incorrect - previously we
> returned NULL acl pointer and this patch changes it to -EINVAL.
>
>