Re: [pyusb-users] USB topology

[Nicolas P. <nic...@aa...>]

Le 13/05/2016 à 17:36, Jeffrey Nichols a écrit :
> On 5/13/2016 10:39 AM, Nicolas Pinault wrote:
>> Le 13/05/2016 à 15:10, Jeffrey Nichols a écrit :
>> I already tried to use "port_numbers" but this field was always blank.
>> After updating libusb-1.0, this field is not blank anymore.
>>
>> What  I don't understand is that bus/port_number numbers are not unique.
>> On my system :
>>   IdVendor8086 IdProduct=1E26 bus=1 Address=1 port=0
>>   IdVendor8086 IdProduct=1E2D bus=2 Address=1 port=0
>> * IdVendor0204 IdProduct=6025 bus=1 Address=6 port=1 ports = (1, 6, 1)
>>   IdVendor0424 IdProduct=2513 bus=1 Address=2 port=6 ports = (1, 6)
>> + IdVendor046D IdProduct=C061 bus=2 Address=5 port=1 ports = (1, 1)
>>   IdVendor0529 IdProduct=0001 bus=1 Address=4 port=5 ports = (1, 5)
>>   IdVendor2CFF IdProduct=0002 bus=1 Address=5 port=2 ports = (1, 6, 2)
>>   IdVendor413C IdProduct=2107 bus=2 Address=2 port=2 ports = (1, 2)
>> + IdVendor8087 IdProduct=0024 bus=2 Address=4 port=1 ports = (1,)
>> * IdVendor8087 IdProduct=0024 bus=1 Address=3 port=1 ports = (1,)
>>   IdVendor8564 IdProduct=4000 bus=3 Address=20 port=8 ports = (8,)
>>   IdVendor8086 IdProduct=1E31 bus=3 Address=1 port=0
>>
>> The lines marked with a + or a * contains same bus/port_number numbers.
>> Only port_numbers field is different.
>> What does this mean ?
> When the bus is different, they're on totally different USB buses 
> (presumably this corresponds internally to USB root hubs, I'm not sure 
> of the technical details). That's always the first level of 
> differentiation.
>
Sure. It's so obvious I didn't mentioned it.
> The port always corresponds to the final number in the ports tuple. 
> It's the port number on the last hub in the hierarchy. It's not unique 
> enough to use to distinguish positions in the hierarchy (even when 
> combined with bus), and not terribly useful in my opinion.
That's what I discovered. Only port_numbers is useful.
>
> I use the tuple (bus, port_numbers) in my application to determine a 
> device's physical location, so when I tell a device to reboot into 
> bootloader mode I can find the new device in the same location (as a 
> device anywhere else could not be the one I'm looking for).
My use case is different but the idea is the same.

Nicolas