Re: [PATCH] ath10k: improve power save performance for sdio

[Wen Gong <wgong@codeaurora.org>]

On 2020-04-17 15:17, Kalle Valo wrote:
> Wen Gong <wgong@codeaurora.org> writes:
> 
>> On 2020-04-16 20:38, Kalle Valo wrote:
>>> Wen Gong <wgong@codeaurora.org> writes:
>>> 
>> 
>>> 
>>>> +static inline int ath10k_hif_set_mbox_sleep(struct ath10k *ar,
>>>> bool enable_sleep)
>>>> +{
>>>> +	if (ar->hif.ops->set_mbox_sleep)
>>>> +		return ar->hif.ops->set_mbox_sleep(ar, enable_sleep);
>>>> +	return 0;
>>>> +}
>>> 
>>> I don't think we need to add another hif op for this. I sent v2 which
>>> uses existing op.
>> 
>> Yes, I see it. but I see it removed ath10k_hif_set_mbox_sleep in
>> ath10k_core_start, expected it will not effect the patch's power save.
> 
> So the idea is that the same functionality is still there as with your
> version, the callpath is just a bit different:
> 
> ath10k_core_start()
>     -> ath10k_hif_start_post()
>         -> ath10k_sdio_hif_start_post()
>             -> ath10k_sdio_set_mbox_sleep()
> 
> But please do double check that I didn't miss anything.
yes, I see this now, so it not miss in v2.
@@ -1749,6 +1819,8 @@  static int ath10k_sdio_hif_start_post(struct 
ath10k *ar)
  		ar_sdio->swap_mbox = false;
  	}

+	ath10k_sdio_hif_set_mbox_sleep(ar, true);
+
  	return 0;
  }