Hello,
[email protected], le dim. 29 mars 2026 21:29:35 +0100, a ecrit:
> So, I propose this RFC patch for discussion. Currently, I'm not
> very fond of it as it may pose unforeseeable consequences for
> callers of diskfs_lookup which rely on its trimming side effect.
Yes, application calling e.g. mkdir("foo/", 0777); should succeed, and
currently glibc translates that into dir_mkdir ("foo/"), so it's not so
simple. And the details are complex :/
> Maybe an assert would be better than just returning EINVAL.
We'd have to filter out in all places before putting an assert,
otherwise it's an easy DoS.
Samuel
