This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug bootstrap/11932] [3.4/3.5 regression] Parallel bootstrap failure with sparc libiberty multilib
- From: "neroden at twcny dot rr dot com" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 5 Feb 2004 00:27:09 -0000
- Subject: [Bug bootstrap/11932] [3.4/3.5 regression] Parallel bootstrap failure with sparc libiberty multilib
- References: <20030815185146.11932.lucier@math.purdue.edu>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Additional Comments From neroden at twcny dot rr dot com 2004-02-05 00:27 -------
Subject: Re: [3.4/3.5 regression] Parallel bootstrap
failure with sparc libiberty multilib
ghazi at gcc dot gnu dot org wrote:
> ------- Additional Comments From ghazi at gcc dot gnu dot org 2004-02-04 17:04 -------
> I saw this bug again today on the 3.4 branch on sparc-sun-solaris2.7.
>
> FYI, for me it only happens about once per month using -j10 or -j20
> bootstraps. So it's hard to say whether the mkinstalldirs patch fixes it for
> sure. I'll have to put the patch in my tree and let a couple of months go by
> without seeing the failure to form an opinion. :-)
>
> Another option would be to somehow serialize the creation of these directories
> so we won't encounter the race condition.
That's harder to do correctly than it looks. But worse, it amounts to
pretty much the same thing as the mkinstalldirs patch, only done in lots
of different places ad-hoc. :-( Which is why I prefer the
mkinstalldirs change.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11932