Re: mkdir -p and EROFS

[Jim Meyering]

Paul Eggert <address@hidden> wrote:
> address@hidden (Eric Blake) writes:
>> The algorithm change between 5.3.0 and 5.90 in lib/mkdir-p.c to
>> try mkdir() first instead of stat(), and key off of EEXIST, breaks
>> when mkdir() fails with EROFS on an intermediate directory when
>> the writable directory has been mounted inside a read-only tree.
>
> Thanks for reporting this.  It is indeed a bug in
> coreutils/lib/mkdir-p.c.  It can be triggered by other errors too.
> This is hard to write a test case for, but I'd like to fix things.
> Does the following patch work for you?
>
> Jim, if this works for Eric, is it OK to install this patch at this late
> date?

Nice patch.
This seems safe enough, so yes, please apply.
Thanks!

> 2005-10-12  Paul Eggert  <address@hidden>
>
>       * mkdir-p.c (make_dir_parents): Don't fail if an intervening mkdir
>       fails due to EROFS, or due to EEXIST or other reasons for that matter.
>       Problem reported by Eric Blake.
>       (ENOSYS): Remove; no longer needed.