urtwn(4): remove duplicate initialization of some registers from urtwn_mac_init()

[avos]

Test Plan

Tested with RTL8188EU, STA/MONITOR modes.

[kevlo]

The vendor driver doesn't set R92C_BCN_MAX_ERR for for RTL8188EUS/RTL8188ETV
and I don't see duplicate initialization of some registers.
Could you point me out? Thanks.

[avos]

sys/dev/usb/wlan/if_urtwnreg.h

1101

R92C_BCN_MAX_ERR (0x55d)

1126

R92C_BCN_MAX_ERR (0x55d)

[avos]

sys/dev/usb/wlan/if_urtwnreg.h

1092

and here (R92C_AGGLEN_LMT (0x458))

1115

Also, R92C_AGGLEN_LMT (0x458) is overriden here; however I don't know to do with it (remove it from here? or from urtwn_init()?)

[kevlo]

Not sure I like this. Part of the reason is we don't know much about
the internal workings of the chip. As for initializations of MAC/BB/RF,
you can see the vendor driver does set those registers.
The device may not work properly unless they are set up correctly.

[avos]

In D3845#79972, @kevlo wrote:

Not sure I like this. Part of the reason is we don't know much about
the internal workings of the chip. As for initializations of MAC/BB/RF,
you can see the vendor driver does set those registers.

Can you specify this place? (the vendor driver is rtl8188C_8192C_usb_linux_v4.0.2_9000.20130911.tar.gz, am I right?)

The device may not work properly unless they are set up correctly.

I have not seen any regressions with this patch + current behavior corrupts RX filter.

[kevlo]

In D3845#80121, @s3erios_gmail.com wrote:

In D3845#79972, @kevlo wrote:

Not sure I like this. Part of the reason is we don't know much about
the internal workings of the chip. As for initializations of MAC/BB/RF,
you can see the vendor driver does set those registers.

Can you specify this place? (the vendor driver is rtl8188C_8192C_usb_linux_v4.0.2_9000.20130911.tar.gz, am I right?)

Yes, it's for RTL8188CU/RTL8192CU. For RTL8188EUS/RTL8188ETV, the driver
is RTL8188EUS_RTL8189ES_linux_v4.1.8_9499.20131104.zip.

The device may not work properly unless they are set up correctly.

I have not seen any regressions with this patch + current behavior corrupts RX filter.

[avos]

Remove EDCA_VI_PARAM modification; vendor driver sets this value after common initialization procedure.

[adrian]

I'm okay with this as long as kevin is.

I'll go and test it all next week (after you commit it) and report back issues.

just note - duplicate registers in configuration writing doesn't mean it's duplicate work - sometimes these chips are buggy or have .. interesting behaviour that requires multiple register reads/writes in order to advance some internal state engine.

[kevlo]

That's exactly my concern.

[avos]

Obsoleted by https://github.com/s3erios/rtwn.