[v2] wifi: rtl8xxxu: Enable AP mode for RTL8192CU (RTL8188CUS)

From: Ezra Buehler <ezra.buehler@husqvarnagroup.com>

This allows the driver to be used in wireless access point mode on the
AT91SAM9G25-based GARDENA smart Gateway.

Unfortunately, the data throughput in AP mode appears to be lower than
with the vendor driver (or in STA mode). Especially when sending, the
data rate is significantly lower. My measurements performed with iperf3
and an Edimax EW-7811Un (VID: 7392, PID: 7811) showed a maximum TX rate
of about 4 Mbits/sec compared to the ~24 Mbits/sec measured with the
rtl8192cu driver.

Although the performance might be good enough for our use case, this is
something that should be further looked into.

Signed-off-by: Ezra Buehler <ezra.buehler@husqvarnagroup.com>
---
When comparing the register values dumped for the two drivers in AP
mode, there are quite some differences. So, no idea where to start
digging deeper. Any pointers would be highly appreciated.

Cheers,
Ezra.
---
Changes in v2:
- Set max_macid_num to correct value
- Make commit message more elaborate, mention bad TX performance
- Link to v1: https://lore.kernel.org/20241108195511.1338147-1-ezra@easyb.ch
---
 drivers/net/wireless/realtek/rtl8xxxu/8192c.c | 2 ++
 1 file changed, 2 insertions(+)


base-commit: 361cb056e2468be534f47c1a6745f96581a721e3

Ezra Buehler <ezra@easyb.ch> wrote:

> From: Ezra Buehler <ezra.buehler@husqvarnagroup.com>
> 
> This allows the driver to be used in wireless access point mode on the
> AT91SAM9G25-based GARDENA smart Gateway.
> 
> Unfortunately, the data throughput in AP mode appears to be lower than
> with the vendor driver (or in STA mode). Especially when sending, the
> data rate is significantly lower. My measurements performed with iperf3
> and an Edimax EW-7811Un (VID: 7392, PID: 7811) showed a maximum TX rate
> of about 4 Mbits/sec compared to the ~24 Mbits/sec measured with the
> rtl8192cu driver.
> 
> Although the performance might be good enough for our use case, this is
> something that should be further looked into.
> 
> Signed-off-by: Ezra Buehler <ezra.buehler@husqvarnagroup.com>

1 patch(es) applied to rtw-next branch of rtw.git, thanks.

4828f572b20b wifi: rtl8xxxu: Enable AP mode for RTL8192CU (RTL8188CUS)

---
https://github.com/pkshih/rtw.git

Am 03.02.25 um 03:40 schrieb Ping-Ke Shih:
> Ezra Buehler <ezra@easyb.ch> wrote:
> 
>> From: Ezra Buehler <ezra.buehler@husqvarnagroup.com>
>>
>> This allows the driver to be used in wireless access point mode on the
>> AT91SAM9G25-based GARDENA smart Gateway.
>>
>> Unfortunately, the data throughput in AP mode appears to be lower than
>> with the vendor driver (or in STA mode). Especially when sending, the
>> data rate is significantly lower. My measurements performed with iperf3
>> and an Edimax EW-7811Un (VID: 7392, PID: 7811) showed a maximum TX rate
>> of about 4 Mbits/sec compared to the ~24 Mbits/sec measured with the
>> rtl8192cu driver.
>>
>> Although the performance might be good enough for our use case, this is
>> something that should be further looked into.
>>
>> Signed-off-by: Ezra Buehler <ezra.buehler@husqvarnagroup.com>
> 
> 1 patch(es) applied to rtw-next branch of rtw.git, thanks.

Is there a reason for not setting macid in rtl8xxxu_fill_txdesc_v1() and 
rtl8xxxu_update_rate_mask() like it was done for the gen2 case?

> 
> 4828f572b20b wifi: rtl8xxxu: Enable AP mode for RTL8192CU (RTL8188CUS)
> 
> ---
> https://github.com/pkshih/rtw.git
>

Martin Kaistra <martin.kaistra@linutronix.de> wrote:
> Am 03.02.25 um 03:40 schrieb Ping-Ke Shih:
> > Ezra Buehler <ezra@easyb.ch> wrote:
> >
> >> From: Ezra Buehler <ezra.buehler@husqvarnagroup.com>
> >>
> >> This allows the driver to be used in wireless access point mode on the
> >> AT91SAM9G25-based GARDENA smart Gateway.
> >>
> >> Unfortunately, the data throughput in AP mode appears to be lower than
> >> with the vendor driver (or in STA mode). Especially when sending, the
> >> data rate is significantly lower. My measurements performed with iperf3
> >> and an Edimax EW-7811Un (VID: 7392, PID: 7811) showed a maximum TX rate
> >> of about 4 Mbits/sec compared to the ~24 Mbits/sec measured with the
> >> rtl8192cu driver.
> >>
> >> Although the performance might be good enough for our use case, this is
> >> something that should be further looked into.
> >>
> >> Signed-off-by: Ezra Buehler <ezra.buehler@husqvarnagroup.com>
> >
> > 1 patch(es) applied to rtw-next branch of rtw.git, thanks.
> 
> Is there a reason for not setting macid in rtl8xxxu_fill_txdesc_v1() and
> rtl8xxxu_update_rate_mask() like it was done for the gen2 case?

rtl8xxxu_fill_txdesc_v3() doesn't fill macid neither. Quickly search for
the code filling tx desc [1]. I think we can have a
   #define TXDESC32_MACID_MASK GENMASK(4, 0)

The h2c command format of rtl8xxxu_update_rate_mask() seemingly should be
equal to rtl8192c_set_raid_cmd() [2], but not.

Can anyone help to spin patches and test if it works?

[1] https://github.com/lwfinger/rtl8192cu/blob/master/hal/rtl8192cu_xmit.c#L257
[2] https://github.com/lwfinger/rtl8192cu/blob/master/hal/rtl8192c_cmd.c#L201

On Mon, Feb 3, 2025 at 9:51 AM Ping-Ke Shih <pkshih@realtek.com> wrote:
>
> Martin Kaistra <martin.kaistra@linutronix.de> wrote:
> >
> > Is there a reason for not setting macid in rtl8xxxu_fill_txdesc_v1() and
> > rtl8xxxu_update_rate_mask() like it was done for the gen2 case?
>
> rtl8xxxu_fill_txdesc_v3() doesn't fill macid neither. Quickly search for
> the code filling tx desc [1]. I think we can have a
>    #define TXDESC32_MACID_MASK GENMASK(4, 0)
>
> The h2c command format of rtl8xxxu_update_rate_mask() seemingly should be
> equal to rtl8192c_set_raid_cmd() [2], but not.
>
> Can anyone help to spin patches and test if it works?

The suggested changes seem to help. A quick test shows TX rates of
around 20 Mbits/sec. I will prepare the patches soon.

>
> [1] https://github.com/lwfinger/rtl8192cu/blob/master/hal/rtl8192cu_xmit.c#L257
> [2] https://github.com/lwfinger/rtl8192cu/blob/master/hal/rtl8192c_cmd.c#L201
>

I've also noticed that the TX rate is not reported correctly. "iw
station dump" shows "tx bitrate: (unknown)" and "hostapd_cli all_sta"
displays "tx_rate_info=0".

Any ideas why that could be?

Cheers,
Ezra.

On 04/02/2025 21:59, Ezra Buehler wrote:
> On Mon, Feb 3, 2025 at 9:51 AM Ping-Ke Shih <pkshih@realtek.com> wrote:
>>
>> Martin Kaistra <martin.kaistra@linutronix.de> wrote:
>>>
>>> Is there a reason for not setting macid in rtl8xxxu_fill_txdesc_v1() and
>>> rtl8xxxu_update_rate_mask() like it was done for the gen2 case?
>>
>> rtl8xxxu_fill_txdesc_v3() doesn't fill macid neither. Quickly search for
>> the code filling tx desc [1]. I think we can have a
>>    #define TXDESC32_MACID_MASK GENMASK(4, 0)
>>
>> The h2c command format of rtl8xxxu_update_rate_mask() seemingly should be
>> equal to rtl8192c_set_raid_cmd() [2], but not.
>>
>> Can anyone help to spin patches and test if it works?
> 
> The suggested changes seem to help. A quick test shows TX rates of
> around 20 Mbits/sec. I will prepare the patches soon.
> 
>>
>> [1] https://github.com/lwfinger/rtl8192cu/blob/master/hal/rtl8192cu_xmit.c#L257
>> [2] https://github.com/lwfinger/rtl8192cu/blob/master/hal/rtl8192c_cmd.c#L201
>>
> 
> I've also noticed that the TX rate is not reported correctly. "iw
> station dump" shows "tx bitrate: (unknown)" and "hostapd_cli all_sta"
> displays "tx_rate_info=0".
> 
> Any ideas why that could be?
> 
> Cheers,
> Ezra.

Looks like rtl8xxxu_sta_statistics() was not updated for AP mode.
See rtw_ops_sta_statistics() and rtw_fw_ra_report_iter() in rtw88
for inspiration.

But this chip doesn't report the TX rate, so rtl8xxxu can only
report the theoretical maximum. It does this already for station
mode in rtl8xxxu_bss_info_changed(). I guess you can do the same
for AP mode in rtl8xxxu_sta_add().