Re: [PATCH 3/6] forkables: Create forkable hardlinks, yet unused.

[Corinna Vinschen <corinna-cygwin at cygwin dot com>]

On Nov 16 13:34, Michael Haubenwallner wrote:
> (sorry about previous empty mail)
> 
> Hi Corinna,
> 
> This is a fixup for the race condition where multiple processes failed
> to concurrently create identical hardlinks.
> 
> So I'm quite successful with the forkable hardlinks now...

I'm still pretty unhappy with this patch.  It adds *lots* of code
to handle a seldom border case.

Assuming you perform some action which starts lots of processes.
Like, say, a bigger build.  Let's say, you install the coreutils
source package and run `time cygport coreutils.cygport prep build'.

If you do this thrice, once without your patch, once with your patch
but without utilizing it, and once with your patch and utilizing it,
how do they compare?  Do you have numbers?


Thanks,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

Attachment: signature.asc
Description: PGP signature