From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262198AbTJ3G2I (ORCPT ); Thu, 30 Oct 2003 01:28:08 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262202AbTJ3G2I (ORCPT ); Thu, 30 Oct 2003 01:28:08 -0500 Received: from mail.mediaways.net ([193.189.224.113]:44005 "HELO mail.mediaways.net") by vger.kernel.org with SMTP id S262198AbTJ3G2E (ORCPT ); Thu, 30 Oct 2003 01:28:04 -0500 Subject: Re: usb devices get redetected all the time in 2.6.0-test{7,9} From: Soeren Sonnenburg To: Greg KH Cc: Linux Kernel In-Reply-To: <20031029200112.GB4434@kroah.com> References: <1067411725.1577.19.camel@localhost> <20031029200112.GB4434@kroah.com> Content-Type: text/plain Message-Id: <1067495225.1529.14.camel@localhost> Mime-Version: 1.0 Date: Thu, 30 Oct 2003 07:27:06 +0100 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, 2003-10-29 at 21:01, Greg KH wrote: > On Wed, Oct 29, 2003 at 08:15:26AM +0100, Soeren Sonnenburg wrote: > > Hi. > > > > When I attach a camera to a 2.4. based machine hotplug starts up some > > script to download images of that camera and fine thats it. > > > > In 2.6.0-testX I see kernel messages like: > > > > hub 2-0:1.0: new USB device on port 1, assigned address 2 > > bus usb: add device 2-1 > > bound device '2-1' to driver 'usb' > > bus usb: add device 2-1:1.0 > > bus usb: remove device 2-1:1.0 > > bus usb: add device 2-1:1.0 > > bus usb: remove device 2-1:1.0 > > bus usb: add device 2-1:1.0 > > bus usb: remove device 2-1:1.0 > > bus usb: add device 2-1:1.0 > > [...] > > > > and several of these hotplug scripts get started... Is this a wanted > > behaviour and something has to be fixed in userspace or is it a kernel > > bug ? > > Does this settle down after a bit, or does the add/remove stuff keep > going on for a while? I realized that when I disable the hotplug script the camera is detected only once (as it should). The script fetches the directory listing off the camera... thus is probably opening+closing the device once. So I wonder whether this could somehow trigger an usb disconnect on close() making the usb driver to redetect the device ?! > Can you enable CONFIG_USB_DEBUG and see if you get any more information > in the kernel debug log as to why this is happening? If still needed I will do. Soeren