Official Blog- Keep It Going

[nikkopt]

About the WiFi, and quoting:

Also, notaz and DJWillis are slowly stepping forward in getting WiFi working.
While it's not perfect, the packet loss is decreasing and things are getting better.
It might still be a long way (or it might be finished soon, you never know with such things), but it's improving.

Is the packet loss issue really a driver problem and not hardware? What i mean is, ED said that a closed source driver DOES exist but they don't use it because the company name is Open Pandora and not Closed Pandora. Does the wifi work ok with this driver? And why not release the Pandora with it until the open source one is finished? I think people prefer having that than wifi not working..

Source: here

 

[MDave]

I wouldn't mind using a closed source wifi driver as a temporary solution until the open source one is fixed

 

[EvilDragon]

About the WiFi, and quoting:

Also, notaz and DJWillis are slowly stepping forward in getting WiFi working.
While it's not perfect, the packet loss is decreasing and things are getting better.
It might still be a long way (or it might be finished soon, you never know with such things), but it's improving.

Is the packet loss issue really a driver problem and not hardware? What i mean is, ED said that a closed source driver DOES exist but they don't use it because the company name is Open Pandora and not Closed Pandora. Does the wifi work ok with this driver? And why not release the Pandora with it until the open source one is finished? I think people prefer having that than wifi not working..

Source: here

Well, AFAIK there are no packet losses anymore, but WPA2 is not working yet and speed is still a bit slow.
That seems to be connected to SDIO power management, as far as I did understand DJWillis and notaz.
Disabling the power management of the WiFi driver did stop the packet losses, now the speed issue comes from power management in SDIO.

There are lots of patches and fixes that need to be done to the driver, but they are getting there

 

[notaz]

Is the packet loss issue really a driver problem and not hardware? What i mean is, ED said that a closed source driver DOES exist but they don't use it because the company name is Open Pandora and not Closed Pandora. Does the wifi work ok with this driver? And why not release the Pandora with it until the open source one is finished? I think people prefer having that than wifi not working..

The closed source TI driver became open after G1/HTC Dream was released thanks to Google, but it has a problem - it doesn't support Linux wireless stack properly and needs custom tools to operate. Some time later Nokia developed another driver which does fully support Linux and is part of mainline kernel, which we are using now. The old TI driver doesn't perform any better than Nokia one (after my adjustments) so there is no point using that.

 

[nikkopt]

Thanks for the replies

 

[LuvClub]

Well, AFAIK there are no packet losses anymore, but WPA2 is not working yet and speed is still a bit slow.
That seems to be connected to SDIO power management, as far as I did understand DJWillis and notaz.
Disabling the power management of the WiFi driver did stop the packet losses, now the speed issue comes from power management in SDIO.

There are lots of patches and fixes that need to be done to the driver, but they are getting there

I am slightly concerned by the problems with the WiFi and I am not afraid to say that I do not understand any of the issues at all, but hopefully it would be sorted out soon.

Could you confirm that the all known issues are due to software problems which should be fixable and that the WiFi should work as expected at a future date without using workarounds or at a slowed rate.

If horror of horrors it is found to be an unfixable hardware fault would OP guarantee a no quibble refund?

 

[WhiteBat]

I am slightly concerned by the problems with the WiFi and I am not afraid to say that I do not understand any of the issues at all, but hopefully it would be sorted out soon.

Could you confirm that the all known issues are due to software problems which should be fixable and that the WiFi should work as expected at a future date without using workarounds or at a slowed rate.

If horror of horrors it is found to be an unfixable hardware fault would OP guarantee a no quibble refund?

But Please tell me that at some point, there is an understanding that the WIFI will be 100% at some point once its sorted... right?

Yes, it will be
First we were fearing it might be hardware as well, that was when the drivers weren't working. The drivers ARE now working. It's the SDIO bus that's making trouble at the moment, and as notaz pointed out, that's on the other OMAP systems as well, so it's not a driver issue.

For a better understanding:
To achieve WiFi, you need multiple things:
1) A driver that knows how to talk to the WiFi module. That's what did cost DJWillis a lot of his time and he got it to work.
2) Some way that the SoC can communicate to the WiFi module. This can be USB, it can be SDIO. USB is working nice, but SDIO is pretty new and therefore still buggy in the Linux world. We're using SDIO, and that needs to be fixed.

http://www.gp32x.de/board/index.php?/topic/52211-thoughts-on-why-isnt-software-donei-payed-i-expect-etc/page__view__findpost__p__820235