ccing Johannes Berg since upstream change (mac80211-next) breaks build:
In the commit log its written: remove SUPPORTS_80211_ENCAP HW flag
Any sane reasons for doing that? mac80211 fails to build because of
removed flags, this is on backports-5.3-rc4
Other than that the feature delivers the promised:
QCA9880 iperf between wired and wireless machine
native mode 449Mbps
ethernet mode 522Mbps
On 18/12/2019 23:45, Tom Psyborg wrote:
> ccing Johannes Berg since upstream change (mac80211-next) breaks build:
>
> In the commit log its written: remove SUPPORTS_80211_ENCAP HW flag
>
> Any sane reasons for doing that? mac80211 fails to build because of
> removed flags, this is on backports-5.3-rc4
>
> Other than that the feature delivers the promised:
> QCA9880 iperf between wired and wireless machine
> native mode 449Mbps
> ethernet mode 522Mbps
>
superseded due to johill's fixup of the patch, new version coming up
John
Am 18.12.2019 um 23:45 schrieb Tom Psyborg:
> ccing Johannes Berg since upstream change (mac80211-next) breaks build:
>
> In the commit log its written: remove SUPPORTS_80211_ENCAP HW flag
>
> Any sane reasons for doing that? mac80211 fails to build because of
> removed flags, this is on backports-5.3-rc4
>
> Other than that the feature delivers the promised:
> QCA9880 iperf between wired and wireless machine
> native mode 449Mbps
> ethernet mode 522Mbps
johns patch does not work for qca9880 wave 1 chipsets. it works only for
10.4 firmares like 9984, ipq40xx etc. the 9880 has no benefit from it
and has no effect.
so your test seem to have false results
Sebastian
>
On 19/12/2019, Sebastian Gottschall <[email protected]> wrote:
>
> Am 18.12.2019 um 23:45 schrieb Tom Psyborg:
>> ccing Johannes Berg since upstream change (mac80211-next) breaks build:
>>
>> In the commit log its written: remove SUPPORTS_80211_ENCAP HW flag
>>
>> Any sane reasons for doing that? mac80211 fails to build because of
>> removed flags, this is on backports-5.3-rc4
>>
>> Other than that the feature delivers the promised:
>> QCA9880 iperf between wired and wireless machine
>> native mode 449Mbps
>> ethernet mode 522Mbps
> johns patch does not work for qca9880 wave 1 chipsets. it works only for
> 10.4 firmares like 9984, ipq40xx etc. the 9880 has no benefit from it
> and has no effect.
>
> so your test seem to have false results
>
> Sebastian
>
>>
>
no way. check that you applied code correctly and enabled ethernetmode
parameter. iperf or ping also wouldn't start from wired machine in
ethernetmode unless pinged from wireless machine first
> no way. check that you applied code correctly and enabled ethernetmode
> parameter. iperf or ping also wouldn't start from wired machine in
> ethernetmode unless pinged from wireless machine first
let me verify this again. maybe i'm wrong.
>
in core.c there is also needed ethernetmode check in
ath10k_core_init_firmware_features
ar->wmi.rx_decap_mode = ATH10K_HW_TXRX_NATIVE_WIFI
replace with
if (ethernetmode)
ar->wmi.rx_decap_mode = ATH10K_HW_TXRX_ETHERNET
else
ar->wmi.rx_decap_mode = ATH10K_HW_TXRX_NATIVE_WIFI
QSDK driver sets it like that when using eth mode
On 19/12/2019, John Crispin <[email protected]> wrote:
> On 18/12/2019 23:45, Tom Psyborg wrote:
>> ccing Johannes Berg since upstream change (mac80211-next) breaks build:
>>
>> In the commit log its written: remove SUPPORTS_80211_ENCAP HW flag
>>
>> Any sane reasons for doing that? mac80211 fails to build because of
>> removed flags, this is on backports-5.3-rc4
>>
>> Other than that the feature delivers the promised:
>> QCA9880 iperf between wired and wireless machine
>> native mode 449Mbps
>> ethernet mode 522Mbps
>>
>
> superseded due to johill's fixup of the patch, new version coming up
> John
>
in case it doesn't fix up scan crash i can provide more details.
triggering scan (from luci at least) would always crash the driver
unless vdev was already started, right after the message: mac create
vdev 0 map ffff