Re: [PATCH 2/2] ath9k: fix tx99 bus error

[Sven Eckelmann <sven@narfation.org>]

[-- Attachment #1: Type: text/plain, Size: 1656 bytes --]

On Dienstag, 20. Juni 2017 09:13:40 CEST miaoqing@codeaurora.org wrote:
> From: Miaoqing Pan <miaoqing@codeaurora.org>
> 
> The hard coded register 0x9864 and 0x9924 are invalid
> for ar9300 chips.
[...]
> -	REG_SET_BIT(ah, 0x9864, 0x7f000);
> -	REG_SET_BIT(ah, 0x9924, 0x7f00fe);

Sorry that this messages comes so later after the patch was accepted. But what 
were these things expected to do? My guess is that 0x9864 is AR_PHY_CCA and 
the other one is something else (AR_PHY_TIMING5?). But yes, these would be ar9002
and not AR9003.

What are the problems that we would expect when the CCA threshold and the CYCPWR 
threshold are no longer be set to the highest possible value? Are we now expecting 
that the device is not transmitting at 99% when it sees other signals?

Btw. why are you writing that ar9300 chips don't have this? It looks to me 
like the original code was taken from QCA's 9300 code [1]. Was it always broken
in the AR9300 hal and how was it now fixed with with the newer HALs?

Could it be that newer chips just have it mapped to a different location? AGC 
on the 9300 seems to be at 0x9e00 and maybe the cca register should have been 
0x9e1c (AR_PHY_CCA_0) and should be set to AR_PHY_CCA_THRESH62? And there is 
also a AR_PHY_TIMING5 (0x9810) which might have to be set to 
AR_PHY_TIMING5_CYCPWR_THR1 | AR_PHY_TIMING5_CYCPWR_THR1A. Of course, I have 
absolutely no idea whether these registers actually control the 
same thing and whether the settings are correct.

Kind regards,
	Sven

[1] https://github.com/freebsd/freebsd/blob/386ddae58459341ec567604707805814a2128a57/sys/contrib/dev/ath/ath_hal/ar9300/ar9300_tx99_tgt.c#L502

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]