Huh? That's only one pointer comparison, in a path where we find
other similar comparisons (in permission(), in cached_lookup() for
revalidating the inode, etc...). And not to mention the loop in
lookup_dentry(), which is executed every time before we call lookup().
And if we opted for the name_translate_char solution, we would do a
"costly" pointer comparison for each _character_ of the name, instead
of once per path component.
> And what I'm more
>worried about is actually the fact that the call can change the dentry
>outside the VFS layer - something I don't like from a conceptual
>standpoint.
Well, no dentry is changed outside the VFS layer, only a qstr which
is not yet attached to a dentry at that point. But maybe that is what
you meant?
Regards,
Alain