From owner-freebsd-gnome@FreeBSD.ORG Sun Feb 12 20:52:01 2006 Return-Path: X-Original-To: gnome@FreeBSD.org Delivered-To: freebsd-gnome@FreeBSD.ORG Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 194BB16A420; Sun, 12 Feb 2006 20:52:01 +0000 (GMT) (envelope-from marcus@marcuscom.com) Received: from creme-brulee.marcuscom.com (creme-brulee.marcuscom.com [24.172.16.118]) by mx1.FreeBSD.org (Postfix) with ESMTP id 1DEDE43D6E; Sun, 12 Feb 2006 20:51:50 +0000 (GMT) (envelope-from marcus@marcuscom.com) Received: from shumai.marcuscom.com (shumai.marcuscom.com [192.168.1.4]) by creme-brulee.marcuscom.com (8.13.4/8.13.4) with ESMTP id k1CKq7HM026046; Sun, 12 Feb 2006 15:52:07 -0500 (EST) (envelope-from marcus@marcuscom.com) From: Joe Marcus Clarke To: Jean-Yves Lefort In-Reply-To: <20060212214337.3e77f804.jylefort@FreeBSD.org> References: <1139185642.25725.167.camel@wolverine> <1139187760.730.10.camel@shumai.marcuscom.com> <1139647425.79253.4.camel@shumai.marcuscom.com> <1139674838.37533.78.camel@wolverine> <1139685927.59900.1.camel@shumai.marcuscom.com> <1139713316.37533.84.camel@wolverine> <1139713404.59900.36.camel@shumai.marcuscom.com> <1139726108.12426.18.camel@shumai.marcuscom.com> <1139730552.91280.4.camel@shumai.marcuscom.com> <20060212214337.3e77f804.jylefort@FreeBSD.org> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="=-42ELNYcW8AMgqyUfyLj9" Organization: MarcusCom, Inc. Date: Sun, 12 Feb 2006 15:51:47 -0500 Message-Id: <1139777507.20944.43.camel@shumai.marcuscom.com> Mime-Version: 1.0 X-Mailer: Evolution 2.4.2.1 FreeBSD GNOME Team Port Cc: gnome@FreeBSD.org, ssouhlal@FreeBSD.org Subject: Re: problems rhythmbox 0.9.3.1 X-BeenThere: freebsd-gnome@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: GNOME for FreeBSD -- porting and maintaining List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 12 Feb 2006 20:52:01 -0000 --=-42ELNYcW8AMgqyUfyLj9 Content-Type: text/plain Content-Transfer-Encoding: quoted-printable On Sun, 2006-02-12 at 21:43 +0100, Jean-Yves Lefort wrote: > On Sun, 12 Feb 2006 02:49:12 -0500 > Joe Marcus Clarke wrote: >=20 > > On Sun, 2006-02-12 at 01:35 -0500, Joe Marcus Clarke wrote: > > > On Sat, 2006-02-11 at 22:03 -0500, Joe Marcus Clarke wrote: > > > > On Sun, 2006-02-12 at 11:01 +0800, Khairil Yusof wrote: > > > > > On Sat, 2006-02-11 at 14:25 -0500, Joe Marcus Clarke wrote: > > > > >=20 > > > > > > And I am also loading songs over NFS....hmmm. What FAM system = are you > > > > > > using? > > > > >=20 > > > > > gamin with WANT_FAM_SYSTEM=3Dgamin in /etc/make.conf > > > >=20 > > > > You should be setting WITH_FAM_SYSTEM in make.conf now. In any eve= nt, > > > > I'm using gamin as well. > > >=20 > > > I found the problem. Since gamin is looking at NOTE_ATTRIB for chang= es, > > > even a file access change will trigger a FAM changed event. And sinc= e > > > such an event triggers a reload of the changed file in r-b, the > > > attributes change again, and then r-b loads the file again, ... and s= o > > > on. > > >=20 > > > It seems to me r-b is being too aggressive in what it considers a > > > change. I think a good solution might be to change > > > rhythmdb_check_changed_file() in rhythmdb.c to always return TRUE. T= his > > > will remove the "changed" file from the list of changed files, and it > > > should never trigger this cycle. > > >=20 > > > I'll give this a try, and let you know. > >=20 > > Sorry, this doesn't work. On closer inspection, all events would be > > ignored by r-b if this was done. I've copied Jean-Yves and Suleiman on > > this to get their opinion. I checked how Linux's inotify works, and > > they have separate events for st_atime changing vs. other attributes > > (IN_ACCESS vs. IN_ATTRIB). Perhaps we should do something similar. > >=20 > > In the meantime, the best solutions I can come up with are to disable > > NOTE_ATTRIB processing in gamin, or turn off library monitoring in > > rhythmbox. >=20 > The problem is specific to NFS: I don't receive NOTE_ATTRIB events > when I read a file on an UFS file system [1]. I've attached a patch > which forces polling for remote file systems. I'll commit it if > Khairil Yusof confirms that it fixes his problem. If you commit this, please also protect gam_poll_basic_init () behind USE_GAMIN_POLLER. I forgot that in my pass, and I've had it sitting on my drive waiting for this ATTRIB thing to be resolved. >=20 > [1] This is a kernel bug; if it is fixed, the problem described above > will resurface, and we'll probably have to disable NOTE_ATTRIB as Joe > suggests. That would suck, but it would be the best option for the time being. Getting a NOTE_ACCESS would be great moving forward. Joe --=20 PGP Key : http://www.marcuscom.com/pgp.asc --=-42ELNYcW8AMgqyUfyLj9 Content-Type: application/pgp-signature; name=signature.asc Content-Description: This is a digitally signed message part -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (FreeBSD) iD8DBQBD75/jb2iPiv4Uz4cRAo94AJ9XJ/65HXy+2kgF8FtU/5/ozS8nCACfZOkX zHc3uXhX9gpYSHYDKLhlxF4= =cXYy -----END PGP SIGNATURE----- --=-42ELNYcW8AMgqyUfyLj9--