Date: Sat, 9 Aug 2008 13:33:38 +0300 From: Kostik Belousov <kostikbel@gmail.com> To: Robert Watson <rwatson@freebsd.org> Cc: src-committers@freebsd.org, John Baldwin <jhb@freebsd.org>, Peter Jeremy <peterjeremy@optushome.com.au>, Ed Schouten <ed@freebsd.org>, cvs-all@freebsd.org, cvs-src@freebsd.org Subject: Re: cvs commit: src/sys/dev/io iodev.c Message-ID: <20080809103338.GN97161@deviant.kiev.zoral.com.ua> In-Reply-To: <alpine.BSF.1.10.0808091127170.36489@fledge.watson.org> References: <200808081343.m78DhwYE068477@repoman.freebsd.org> <200808081226.32089.jhb@freebsd.org> <20080809001256.GL64458@server.vk2pj.dyndns.org> <alpine.BSF.1.10.0808091127170.36489@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--02C+GlUcNvfRYA2K Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Sat, Aug 09, 2008 at 11:27:58AM +0100, Robert Watson wrote: >=20 > On Sat, 9 Aug 2008, Peter Jeremy wrote: >=20 > >On 2008-Aug-08 12:26:31 -0400, John Baldwin <jhb@freebsd.org> wrote: > >>It should be setting D_TRACKCLOSE though so that close() reliably clear= s=20 > >>the flag even in single-threaded processes. You can still get odd=20 > >>behavior if you explicitly open it twice in an app and then close one o= f=20 > >>the two fd's. You will no longer have IO permission even though you sti= ll=20 > >>have one fd open. However, if you do that I think you deserve what you= =20 > >>asked for. :) > > > >That behaviour may be legitimate: Your code links with libraries foo an= d=20 > >bar that each independently open /dev/io so they can frob different thin= gs=20 > >in IO space. libfoo needs ongoing access to device foo and so keeps its= =20 > >descriptor open. libbar only needs once-off access to device bar and so= =20 > >closes /dev/io once it's finished its initialisation. Libraries foo and= =20 > >bar are completely independent and shouldn't need to know anything about= =20 > >each other and your app shouldn't need to know that libraries it's using= =20 > >frob around in IO space. >=20 > If that's the view, there should probably be a per-process counter,=20 > although this is all a bit tricky anyway since file descriptors and=20 > processes have a tenuous relationship. Another interesting issue is the close on exec, esp. for /dev/io. It seems that Linux recently grown full new API to handle FD_CLOEXEC races, see http://udrepper.livejournal.com/20407.html --02C+GlUcNvfRYA2K Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (FreeBSD) iEYEARECAAYFAkidcoEACgkQC3+MBN1Mb4iHpwCgmHo9VoMHSNUazRS/18mmkgTs g3gAoIcIXqc8YmW5mqfW/j8aAY6A//Q3 =PpBH -----END PGP SIGNATURE----- --02C+GlUcNvfRYA2K--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20080809103338.GN97161>