Re: proposed fix for some race conditions in mkdir and install

[Jim Meyering]

Paul Eggert <address@hidden> wrote:
> Jim Meyering <address@hidden> writes:
>
>> By the way, I've just realized that it *is* possible to implement these
>> mkdir-like functions in a thread-safe manner (i.e. traversing via openat,
>> operating on files via mkdirat, chmodat, etc.), while remaining efficient
>> (O(n)).  The only assumption is that the target directory name does not
>> contain too long a sequence of existing unreadable directories (sum
>> of length of names + slashes > PATH_MAX).
>
> I initially tried something like that, but got too confused by it,
> perhaps partly because I wanted to fall back on something like the
> current approach when mkdirat didn't work (which is a common case
> these days, even on GNU/Linux systems, alas).

What do you mean?
Of course, for at-function-syscall support, you need Linux 2.6.16 or
newer, but gnulib's mkdirat emulation works well also with /proc/self/fd/N
support, which is available back at least to linux-2.4.x.
Admittedly, /proc isn't always available, but I wouldn't say that's "common".

> The current code is still too complicated for my taste --

I agree.  But we can save the rewrite for later :)

> I'd be quite
> surprised if it were bug-free.  I considered ripping out the
> open(".")/fchdir part of the code, since it's simpler if you assume
> that you _always_ create a subprocess, but decided that in this case
> simplicity had to yield to performance.

I'm glad you did.