On 8/21/24 17:03, Roderick Colenbrander wrote:

Ideally I would like some type of quirk, but I have no good way of
verifying (also across different 'clone' devices) how reliable this
current check would be. It feels a bit too much towards a magical
value. I don't see why the reports would differ besides some engineer
having made a typo.

The descriptors differ in several ways - why, I don't know.


  - Across a few DS4 2.0 that I have looked at, they seemed to all have 
the same descriptors.

  - Quite a few clones seemed to expose the exact same descriptor on 
USB, and those didn't need this quirk either.

  - I have seen two controllers that need this quirk, and they both have 
the *same* descriptor, which is close to, but still notably different 
from, the DS4 v2.0 descriptor.


So, while I don't have a large base to infer from, it seems like this 
particular "magic" descriptor and the quirk are correlated.


Do you have another idea for a specific enough difference that I could 
look for?

My gut feeling is that the previous fix is less fragile, so let's opt
for that one.

Let me know if you wish for me to resend v2 with a Signed-off-by. Or, I 
guess Jiri might just manually patch it in - please see this as 
permission to do so if you prefer this.



Max