Re: [PATCH] ath5k: Does not connect reliably on 2.6.27

Re: [PATCH] ath5k: Does not connect reliably on 2.6.27

[Nick Kossifidis]

2008/10/14 Elias Oltmanns <eo@nebensachen.de>:
> Hi all,
>
> after a cold boot into 2.6.27, ath5k doesn't work on my machine. The
> sequence
>
> # ifconfig ath0 up
> # iwlist ath0 ap
>
> results in a message informing me that no list of APs is available for
> that device. I have identified the following commit as the guilty one:
>
> commit 2203d6be7ed17af81a1dc35a0af9806086743b02
> Author: Nick Kossifidis <mick@madwifi.org>
> Date:   Sun Jul 20 06:36:52 2008 +0300
>
>    ath5k: Misc hw_reset updates
>
>     * Update hw_reset to calculate some of the values we were using as
>       static
>     * Increase activation to rx delay
>
>    Changes-licensed-under: ISC
>    Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
>    Signed-off-by: John W. Linville <linville@tuxdriver.com>
>
> The minimal patch to fix the issue for me is attached. However, I am
> wondering whether the preceding writes to register 0xa228 should be
> dropped or at least adjusted somehow. Perhaps those of you who know the
> recently released HAL from Atheros can throw some light on the matter.
>
> Regards,
>
> Elias
> ---
> diff --git a/drivers/net/wireless/ath5k/hw.c b/drivers/net/wireless/ath5k/hw.c
> index dc51b84..b1cec2a 100644
> --- a/drivers/net/wireless/ath5k/hw.c
> +++ b/drivers/net/wireless/ath5k/hw.c
> @@ -862,7 +862,7 @@ int ath5k_hw_reset(struct ath5k_hw *ah, enum ieee80211_if_types op_mode,
>                        data = 0;
>
>                        /* Just write 0x9b5 ? */
> -                       /* ath5k_hw_reg_write(ah, 0x000009b5, 0xa228); */
> +                       ath5k_hw_reg_write(ah, 0x000009b5, 0xa228);
>                        ath5k_hw_reg_write(ah, 0x0000000f, AR5K_SEQ_MASK);
>                        ath5k_hw_reg_write(ah, 0x00000000, 0xa254);
>                        ath5k_hw_reg_write(ah, 0x0000000e, AR5K_PHY_SCAL);
>

I think i've fixed that (it's a buggy mask some lines above) on an
earlier patch (the one that introduced AR2425 support) but i'll look
into it again ;-)

-- 
GPG ID: 0xD21DB2DB
As you read this post global entropy rises. Have Fun ;-)
Nick

Re: [PATCH] ath5k: Does not connect reliably on 2.6.27

[Elias Oltmanns]

"Nick Kossifidis" <mickflemm@gmail.com> wrote:
> 2008/10/14 Elias Oltmanns <eo@nebensachen.de>:
>> Hi all,
>
>>
>> after a cold boot into 2.6.27, ath5k doesn't work on my machine. The
>> sequence
>>
>> # ifconfig ath0 up
>> # iwlist ath0 ap
>>
>> results in a message informing me that no list of APs is available for
>> that device.
[...]
>> The minimal patch to fix the issue for me is attached. However, I am
>> wondering whether the preceding writes to register 0xa228 should be
>> dropped or at least adjusted somehow. Perhaps those of you who know the
>> recently released HAL from Atheros can throw some light on the matter.
>>
>> Regards,
>>
>> Elias
>> ---
>> diff --git a/drivers/net/wireless/ath5k/hw.c b/drivers/net/wireless/ath5k/hw.c
>> index dc51b84..b1cec2a 100644
>> --- a/drivers/net/wireless/ath5k/hw.c
>> +++ b/drivers/net/wireless/ath5k/hw.c
>> @@ -862,7 +862,7 @@ int ath5k_hw_reset(struct ath5k_hw *ah, enum ieee80211_if_types op_mode,
>>                        data = 0;
>>
>>                        /* Just write 0x9b5 ? */
>> -                       /* ath5k_hw_reg_write(ah, 0x000009b5, 0xa228); */
>> +                       ath5k_hw_reg_write(ah, 0x000009b5, 0xa228);
>>                        ath5k_hw_reg_write(ah, 0x0000000f, AR5K_SEQ_MASK);
>>                        ath5k_hw_reg_write(ah, 0x00000000, 0xa254);
>>                        ath5k_hw_reg_write(ah, 0x0000000e, AR5K_PHY_SCAL);
>>
>
> I think i've fixed that (it's a buggy mask some lines above) on an
> earlier patch (the one that introduced AR2425 support) but i'll look
> into it again ;-)

Can you drop me a link or a commit + tree so I can test it?

Regards,

Elias