From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?ISO-8859-1?Q?S=E9bastien_Dugu=E9?= Subject: Re: Kernel patches enabling better POSIX AIO (Was Re: [3/4] kevent: AIO, aio_sendfile) Date: Mon, 04 Sep 2006 16:36:40 +0200 Message-ID: <1157380601.3895.53.camel@frecb000686> References: <44C77C23.7000803@redhat.com> <44C796C3.9030404@us.ibm.com> <1153982954.3887.9.camel@frecb000686> <44C8DB80.6030007@us.ibm.com> <44C9029A.4090705@oracle.com> <1154024943.29920.3.camel@dyn9047017100.beaverton.ibm.com> <44C90987.1040200@redhat.com> <1154034164.29920.22.camel@dyn9047017100.beaverton.ibm.com> <1154091500.13577.14.camel@frecb000686> <44DCDE73.9030901@redhat.com> <20060812182928.GA1989@in.ibm.com> <44DE27AB.7040507@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: suparna@in.ibm.com, Badari Pulavarty , Zach Brown , Christoph Hellwig , Evgeniy Polyakov , lkml , David Miller , netdev , linux-aio@kvack.org Return-path: Received: from ecfrec.frec.bull.fr ([129.183.4.8]:18155 "EHLO ecfrec.frec.bull.fr") by vger.kernel.org with ESMTP id S1751400AbWIDOgs convert rfc822-to-8bit (ORCPT ); Mon, 4 Sep 2006 10:36:48 -0400 To: Ulrich Drepper In-Reply-To: <44DE27AB.7040507@redhat.com> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Sat, 2006-08-12 at 12:10 -0700, Ulrich Drepper wrote: > Suparna Bhattacharya wrote: > > I am wondering about that too. IIRC, the IO_NOTIFY_* constants are = not > > part of the ABI, but only internal to the kernel implementation. I = think > > Zach had suggested inferring THREAD_ID notification if the pid spec= ified > > is not zero. But, I don't see why ->sigev_notify couldn't used dire= ctly > > (just like the POSIX timers code does) thus doing away with the=20 > > new constants altogether. Sebestian/Laurent, do you recall? >=20 > I suggest to model the implementation after the timer code which does > exactly what we need. >=20 Will do. >=20 > > I'm guessing they are being used for validation of permissions at t= he time > > of sending the signal, but maybe saving the task pointer in the ioc= b instead > > of the pid would suffice ? >=20 > Why should any verification be necessary? The requests are generated= in > the same process which will receive the notification. Even if the PO= SIX > process (aka, kernel process group) changes the IDs the notifications > should be set. The key is that notifications cannot be sent to anoth= er > POSIX process. >=20 > Adding this as a feature just makes things so much more complicated. >=20 Agreed. S=E9bastien. --=20 ----------------------------------------------------- S=E9bastien Dugu=E9 BULL/FREC:B1-247 phone: (+33) 476 29 77 70 Bullcom: 229-7770 mailto:sebastien.dugue@bull.net Linux POSIX AIO: http://www.bullopensource.org/posix http://sourceforge.net/projects/paiol -----------------------------------------------------