Date: Thu, 12 Feb 2009 10:45:52 +0300 From: Ivan Shcheklein <shcheklein@gmail.com> To: freebsd-bugs@freebsd.org, rwatson@freebsd.org Subject: Re: kern/130348: [socket] accept() prematurely allocates an inheritable descriptor [regression] Message-ID: <3034886f0902112345nfae93e6i55489dd3fb6cd7c7@mail.gmail.com> In-Reply-To: <200902111540.n1BFeCJ5068985@freefall.freebsd.org> References: <200902111540.n1BFeCJ5068985@freefall.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Feb 11, 2009 at 6:40 PM, <rwatson@freebsd.org> wrote: > Synopsis: [socket] accept() prematurely allocates an inheritable descriptor > [regression] > > State-Changed-From-To: analyzed->patched > State-Changed-By: rwatson > State-Changed-When: Wed Feb 11 15:38:28 UTC 2009 > State-Changed-Why: > Fix committed to 8.x; transition to patched until MFC. > > Hi Ivan: > > Thanks for this bug report; per commentary in the commit and the PR, there > is an unavoidable race here due to the nature of the API, but I have made > a change to our fork(2) code so that it is quite a narrow race (consistent > with that found previously on FreeBSD and on other platforms) rather than > a wide one. I will merge this fix to 7.x in a week or so once it has > settled. If you're able to apply the patch manually to your local tree > and confirm it fixes the problem you were seeing, that would be helpful. > > Thanks, > > > http://www.freebsd.org/cgi/query-pr.cgi?pr=130348 > Works fine on 7.1. Thank you, Robert. Also, I think we will use select() to avoid this race condition at all.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3034886f0902112345nfae93e6i55489dd3fb6cd7c7>