On Tuesday 30 August 2005 03.19, David Brownell wrote: > > From: Daniel Ritz <[EMAIL PROTECTED]> > > Date: Tue, 30 Aug 2005 00:39:36 +0200 > > > > ... > > > > if i didn't miss anything usb's check for knode_bus is also stupid > > workaround > > to avoid a double device_bind_driver()... > > USB is one of very few subsystems that really _needs_ to be using the > two phase init()/add() driver model calls instead of register(). > The state machine for a usb_interface is way too complicated to use > that single phase model! >
agreed. the device_bind_driver() API is just bad. (besides the comment says it's safe to call if from a drivers .probe which is not really true) > So those list checks just answer "has add(this) been called yet?", > using a cheap list test. It's not good to bind_driver() for a > device that's not yet known to the driver model, for one thing ... > i know, it made my rmmod hang :) but the check in USB should be an API in the driver model like alan stern mentions in another mail... > - Dave > rgds -daniel ------------------------------------------------------- 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-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
