Re: usb: dwc2: Problem with remote wakeup implementation

[Maxim Devaev <mdevaev@gmail.com>]

В Mon, 4 Apr 2022 07:30:25 +0000
Minas Harutyunyan <Minas.Harutyunyan@synopsys.com> пишет:

> Hi Maxim,
> 
> According databook RmtWkUpSig bit description below. You should also 
> consider LPM state.
> 
> Remote Wakeup Signaling (RmtWkUpSig)
> When the application sets this bit, the core initiates remote
> signaling to wake up the USB host. The application must Set this bit
> to instruct the core to exit the Suspend state. As specified in the
> USB 2.0 specification, the application must clear this bit 1-15 ms
> after setting it.
> If LPM is enabled and the core is in the L1 (Sleep) state, when the
> application sets this bit, the core initiates L1 remote signaling to
> wake up the USB host. The application must set this bit to instruct
> the core to exit the Sleep state. As specified in the LPM
> specification, the hardware automatically clears this bit 50
> microseconds (TL1DevDrvResume) after being set by the
> application. The application must not set this bit when GLPMCFG
> bRemoteWake from the previous LPM transaction is zero.
> 
> If it will not help, please check power status of core while resuming: 
> hibernated, in partial power down, etc.
> 
> Thanks,
> Minas
> 

Thank you so much! This is very useful.
I fixed the problem, it was related with the new clock gating code.

I'll try to make a clear patch and submit it to the kernel.