Re: RTL8192CU continually reconnecting

[Larry Finger <Larry.Finger@lwfinger.net>]

On 09/15/2013 12:48 PM, Mark Cave-Ayland wrote:
> On 15/09/13 14:50, Larry Finger wrote:
>
>> There is a known problem with the dynamic power management code for this
>> chip. Unfortunately, the problem does not affect my system. At the
>> moment, my device that uses rtl8192cu has been connected for 12 hours
>> with no disconnects. The only evidence of a bug here is poor performance.
>
> Did my email to the list yesterday provide any further clues at all? I know that
> you're looking at re-working the power code, but if there is any chance of a
> stop-gap fix for 3.12 then I know it would be greatly appreciated :)

Yes, you provided some clues, but as I have no knowledge of the internal 
workings of the chips, I dare not set parameters willy-nilly as that might cause 
chips to burn out and fail.

The vendor driver provides better performance for my device than does rtl8192cu. 
I would provide the patches needed to compile it on a modern kernel; however, it 
causes a kernel panic whenever you try to unload it. I have no time nor 
inclination to debug that. Thus no one should use it unless you want to fix that 
problem.

I am trying to change the in-kernel dm code to match the vendor driver. That is 
difficult as nearly every variable is renamed.

Any patches will be pushed when they are ready, and not before. I will, however, 
provide them for testing. It remains to be seen if any such patches will be 
suitable for 3.12 and backporting to 3.11.

Larry