-----Original Message----- From: Greg KH [mailto:greg@xxxxxxxxx] Sent: Friday, July 20, 2007 1:42 PM To: Rajendra Stalekar Cc: kernelnewbies@xxxxxxxxxxxx Subject: Re: doubts on device probing(auto-detection) On Fri, Jul 20, 2007 at 11:42:19AM +0530, Rajendra Stalekar wrote: > Hi, > > 1. When we probe a device, how does it find out that the device actually > exists and is functional? What type of device? ISA? PCI? USB? Firewire? Something else? They all work differently depending on the bus type. Rajendra:- PCI would be a good example I suppose. > 2. Do we probe for interrupts to check which interrupt line(s) can be > assigned to the device in question? USB devices don't have interrupts, so you are going to have to be more specific here :) > 3. When the driver is calling probe_irq_on, it returns a bit mask of > unassigned interrupts. In such case driver assigns an interrupt line for > that specific device, Am I right? No modern driver should ever call that function. Rajendra:- Is it because the devices are intelligent enough so that they announce the IRQ? > 4. If so, then how does the driver decide which interrupt line to assign to > the device? It uses the irq number that was assigned to it by the core already. > 5. Once the driver assigns the interrupt line, why does the programmer need > to explicity enable the interrupt for that device. Shouldn't that be done by > probe_irq_on as default? Again, don't use that old function. > 6. Why do we say that the programmer must be careful to enable interrupts on > the device after the call to the above function and disable them before the > probe_irq_off? Why are you calling it? :) thanks, greg k-h -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
