Greg KH <greg@...> writes: > > On Wed, Feb 25, 2009 at 06:30:00AM +0000, Martin Willms wrote: > > > > Hello, > > > > I just saw something "unexpected" using our USB driver for our ADU-07 > > measurement systems. We use a measurement hardware with Cypress EZ USB > > controller connected to an CPU board running an Linux system. As the EZ USB is > > soft booted, it is probed as Cypress device after power on. After software > > beeing downloaded to it it renumerates and probes as ADU-07. > > > > So far so good. Now I had problems, if doing a warm start of the Linux system, > > not switching power off of the Cypress controller. In this case the Cypress > > controller would not renumerate as it does not change its VID/PID. As a result > > my USB driver did not work properly anymore. > > > > After investigating this issue I found out, that the Linux kernel calls the > > probe function for the device before calling the init function of the module. > > That's impossible, how would the kernel know to call the probe function > if it's not even registered yet? > > Now the probe function can be called before the init function returns, > right after usb_register_driver() is called, which is probably what > happened here, right? That's expected for devices that are already > plugged in, and your driver should be able to handle it. > > Do you have a pointer to your driver somewhere so that we can look at it > to see if there are any other problems? > > thanks, > > greg k-h > -- > To unsubscribe from this list: send the line "unsubscribe linux-usb" in > the body of a message to majordomo@... > More majordomo info at http://vger.kernel.org/majordomo-info.html > > Hello, Yes, you are right. usb_register is called right at the beginning of the init function. Afterwards I am doing my initialisations (allocating buffers, starting kernel thread, ...). Is it OK to call the usb_register right at the end of the init function? In this case I would have to free up all allocated resources, if registering will fail, but I think that you are right that this will be the correct way to do this. Up to now our driver is not part of the kernel. I am not shure, if it would help anyone as it only works with our devices. But maybe it would be a good idea to make it part of the new "scrap" tree of kernel modules. I will think about it. thanks, Martin -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
