Currently the rtl8723ae driver is broken (since v4.7).
Connection to AP is lost very often, especially when
the signal level is not very good.
The main issue is the power save mode is basically
not working, and seems to trigger a firmware bug.
So I had to take out the FW LPS mode handling.
While debugging the driver I found a couple related issues,
for instance that the signal level in dm.undec_sm_pwdb
is no longer accurate (may be even much too high) when no more
packets are received, and it increases the likelihood to receive
something if the input gain is set to maximum.
The patch was tested with the rtl8723ae PCI card in my laptop
against a FRITZ!Box 7590 AP -- the WiFi connection works now
very reliable for me.
V2 of the patch adjusts the defaults of swlps and fwlps module
parameters to match the firmware capabilities instead of removing
the whole code, so it can be easily re-activated once a firmware
update is available.
Bernd Edlinger (4):
rtlwifi: rtl8723ae: Take the FW LPS mode handling out
rtlwifi: rtl8723ae: Don't use dm.undec_sm_pwdb for input gain control
when no beacon was received in the connected state
rtlwifi: rtl8723ae: Re-introduce
rtl8723e_dm_refresh_rate_adaptive_mask
rtlwifi: Move the clearing of rtlpriv->link_info.num_rx_inperiod in
rtl_watchdog_wq_callback a few lines down
drivers/net/wireless/realtek/rtlwifi/base.c | 4 +-
drivers/net/wireless/realtek/rtlwifi/core.c | 2 +
.../net/wireless/realtek/rtlwifi/rtl8723ae/dm.c | 95 +++++++++++++++++++++-
.../net/wireless/realtek/rtlwifi/rtl8723ae/sw.c | 8 +-
4 files changed, 101 insertions(+), 8 deletions(-)
--
1.9.1
On 1/5/19 12:38 PM, Bernd Edlinger wrote:
> Currently the rtl8723ae driver is broken (since v4.7).
>
> Connection to AP is lost very often, especially when
> the signal level is not very good.
>
> The main issue is the power save mode is basically
> not working, and seems to trigger a firmware bug.
> So I had to take out the FW LPS mode handling.
>
> While debugging the driver I found a couple related issues,
> for instance that the signal level in dm.undec_sm_pwdb
> is no longer accurate (may be even much too high) when no more
> packets are received, and it increases the likelihood to receive
> something if the input gain is set to maximum.
>
> The patch was tested with the rtl8723ae PCI card in my laptop
> against a FRITZ!Box 7590 AP -- the WiFi connection works now
> very reliable for me.
>
> V2 of the patch adjusts the defaults of swlps and fwlps module
> parameters to match the firmware capabilities instead of removing
> the whole code, so it can be easily re-activated once a firmware
> update is available.
>
>
> Bernd Edlinger (4):
> rtlwifi: rtl8723ae: Take the FW LPS mode handling out
> rtlwifi: rtl8723ae: Don't use dm.undec_sm_pwdb for input gain control
> when no beacon was received in the connected state
> rtlwifi: rtl8723ae: Re-introduce
> rtl8723e_dm_refresh_rate_adaptive_mask
> rtlwifi: Move the clearing of rtlpriv->link_info.num_rx_inperiod in
> rtl_watchdog_wq_callback a few lines down
The subject on all 4 of these need to be modified. Only the first line will be
shown as the "subject", which will not make much sense.
In addition, you need to include the changes in the cover letter, and in the
actual patch after the --- line. In that section, you mention what changes were
made in each revision, then follow with another --- line. That info is there for
the maintainers - it will be stripped before the patch is merged.
Larry