Rtl8812au: v5.1.5 Alfa AC1900 not changing to USB3 mode

[lhernanz]

Hi,

as the title says, the Alfa stays in USB2 mode. I have tracked the problem to the code below. For whatever the reason (I have the impression that the EEPROMUsbSwitch variable is not properly initialized in the 8814 driver but I have not tested this thoroughly), EEPROMUsbSwitch is false and this stops the mode change. Commenting the two lines solves the problem.

Regards.

--- a/os_dep/linux/usb_intf.c
+++ b/os_dep/linux/usb_intf.c
@@ -649,15 +649,17 @@ static int usb_reprobe_switch_usb_mode(PADAPTER Adapter)
        u8 ret = _FALSE;

        /* efuse not allow driver to switch usb mode */
-       if (pHalData->EEPROMUsbSwitch == _FALSE)
-               goto exit;
+       //if (pHalData->EEPROMUsbSwitch == _FALSE)
+       //      goto exit;

[kimocoder]

Thanks for the report, will take a look shortly :)

[lhernanz]

Thanks,
I will keep testing the changes, although AC is not working with this version of the driver (see me comment on the AC bug) and, therefore, I am thinking about testing a previous version of the driver.
Regards

[kimocoder]

Please be specific and do some test before the commit/changes are made.

[lhernanz]

Sorry, I have not explained myself properly.

The bug report and fix is for v5.1.5. I have tested the change and I can confirm that the card changes to USB3.0 as expected after applying the patch.

My point about the need to move to a different version is that, in v5.1.5, AC is not working (at least for this card), so having USB3.0 is a moth point as you are not going to get wifi speeds above what USB 2.0 can support. Of course, the ideal thing would be to be able to fix the AC problem too :)

Regards

[lhernanz]

Hi,

Uhm, I am not entirely sure. In the v5.1.5 you can already see some similar code doing some similar (there are important differences between the drivers, though) things. See the function rtw_cfg80211_create_vht_cap in 5.1.5 vs the rtw_cfg80211_init_vht_capab in 4.3.21.

But clearly the code in 5.1.5 is not working for AC, so it could be that the code is incomplete in that version. It is kind of amazing that Realtek of Alfa are not providing updated drivers for their products that are fully functional.

[lhernanz]

Is there any real advantage of using v5.1.5 vs v4.3.21 as of today? As you say, code quality in v5.1.5 seems far from ideal.

[kimocoder]

@lhernanz this issue has been resolved :)

[lhernanz]

Hi,
the new code is not working for me. I am now connecting at USB2.0 speeds again. The problem still is the if (pHalData->EEPROMUsbSwitch == _FALSE) comparison that you have uncommented in the last commit. If that comparison succeed, you get out of the method, therefore, you never get to execute the new code that you have added (that is working for me if I comment the mentioned statement as I was suggesting at the beginning of this thread).

My impression is that the variable pHalData->EEPROMUsbSwitch is not initialized in the 8814 HAL code as the only references that I can find to that variable are in the 8812 HAL code:

hal/rtl8812a/usb/usb_halinit.c
2258:       pHalData->EEPROMUsbSwitch = _FALSE;
2261:       pHalData->EEPROMUsbSwitch = (PROMContent[EEPROM_USB_MODE_8812] & BIT1) >> 1;
2263:   RTW_INFO("Usb Switch: %d\n", pHalData->EEPROMUsbSwitch);

Regards

[kimocoder]

Yes, personally haven't got time for testing at the moment.
@evilphish is working on a solution for it, opened the issue report again so the discussion may be seen by others too. Thanks

[evilphish]

@lhernanz Hey there, my 8814 unit does not arrive until in 2 days or so, so I can not test this fix yet. If you do not want to wait, feel free to grab the sources from https://github.com/evilphish/rtl8812au (just committed a fix) and check whether it works now. If it works we can merge it back.

[kimocoder]

Got a friend to test it, and the device (on 8814au) is now working inn SuperSpeed on bcdUSB 3.00.
The patch seem to work great! Will have those logs sent first thing in the morning but it finally seems to be resolved.

[lhernanz]

Hi,
The @evilphish patch is working for me too!

I guess that we can close this one. We "only" need to make the AC mode to work and it will be almost perfect (have not checked if you have made any progress on that one lately). It is kind of crazy the mess that Realtek is creating and the low quality of the code.

Thanks for the help guys!

[kimocoder]

Confirmed working for 8814, how about 8812? If that works, we'll closed it..

[evilphish]

That's the thing. It is changing to USB3 mode but this results in a drop in reception.
You can actually see more networks with the 8812 if you plug it into an old USB2 rather than a USB3 port. This leads me to believe that something is not working right with the mode switching. Even though the device does do the switch and is then listed as a USB3 device.

But I think we can close this and continue the discussion regarding the 8812 in the thread where this issue first manifested. https://github.com/aircrack-ng/rtl8812au/issues/29#issuecomment-338819141

[kimocoder]

Closed. Continue om @evilphish issue report. Thanks