From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?UTF-8?B?TWlrYSBQZW50dGlsw6Q=?= Date: Tue, 23 Aug 2005 20:09:17 +0000 Subject: Re: udevsynthesize as a udevstart+coldplug replacement Message-Id: <430B826D.4050308@kolumbus.fi> List-Id: References: <20050823191542.GA29091@vrfy.org> In-Reply-To: <20050823191542.GA29091@vrfy.org> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable To: linux-hotplug@vger.kernel.org Kay Sievers wrote: >On Tue, Aug 23, 2005 at 10:46:43PM +0300, Mika Penttil=C3=A4 wrote: > =20 > >>Kay Sievers wrote: >> >> =20 >> >>>Here is "udevsynthesize" a possible replacement for the udevstart/coldpl= ug >>>combination we currently have. Usually very early in the boot process, it >>>gets all available devices from sysfs and synthesizes the events these >>>devices would have generated at creation time. >>>The events are directly passed to the udevd daemom socket. udevd does al= l=20 >>>the >>>work and runs all the events asynchronously. udevsynthesize itself does >>>not read any udev rule. >>> >>>It scans block and class devices to create the the device nodes for alre= ady >>>available devices, which obsoletes the run of udevstart. >>>After that, it scans for bus devices, which events may load modules or >>>configure the device. If the bus type is known, the usual bus-specific >>>values are added to the event environment. >>> >>>With the "rules-in-daemon" patch, the whole udevstart/coldplug handling >>>(700 events) takes less than a second CPU time on a 1.7Ghz laptop. >>> >>>A typical boot script would do: >>>prepare tmpfs on /dev >>>create /dev/null >>>start /sbin/udevd >>>disable /proc/sys/kernel/hotplug (the input layer is still broken and=20 >>>needs a kernel patch) >>>run /sbin/udevsynthesize >>> =20 >>> > > =20 > >>What if udevd receives netlink messages and goes into "ignore unix=20 >>socket" mode before/while synthesize? >> =20 >> > >No such mode exists. :) Only events with a SEQNUM set will possibly be >ignored if received from userspace after netlink activity is detected. > >Kay > > =20 > Ok, there's the (seqnum !=3D 0 && netlink active) test which keeps these=20 messages passing through. Thanks, Mika ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net Linux-hotplug-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel