dingens
Member
- Joined
- Jun 1, 2009
- Messages
- 409
Good news
I remember DJWillis posting in the Beagleboard IRC about using G1's driver long ago. I should check the log, there could be new info.
Edit:
DJWillis explains the driver situation:
—
Anyway, we actually have 2 open source options so I’ll give people a little background.
One is the TI WiLink 4 based ‘partial’ driver that was opened as part of the Google Android G1 release.
This driver is unfortunately a mess and really has no place in a modern system (ancient design architecture, lots of semi-bespoke shims, some really nasty bugs etc.).
We can use this driver if you don’t mind setting it up from one clunky user space app that crashes a lot and lacks solid support for things like WEP/WPA etc. – Really this driver is only of use for hardware validation not real use from users. Even after I hacked in Wireless Extensions support (yep, the driver is THAT old) it’s still not a driver I would want to use (sorry but I hate lots of nasty random mem leaks and errors).
—
The next open option is the WL12xx driver written by Kalle Valo of Nokia. This is a nice modern driver (MAC80211) and is already in Wireless-Testing (as of a few days ago) and on its way to mainline.
This driver was SPI only but a gent by the name of Bob Copeland added the start of SDIO support and we are working off that base to bring the driver up on the Pandora so we have a nice solid mainline driver to use that is actively developed, supports the Pandora out of the gate, and is easy to maintain.
As of now I have the WL2151 firmware loading on the Pandora with the new driver and lots of nice signs of life but the work is a little way away from passing traffic (the Linux SDIO stack is causing me pain).
Not sure if that counts as an update .
If anyone wants patches (or is familiar with the kernels SDIO stack) drop me a bell.
John
I remember DJWillis posting in the Beagleboard IRC about using G1's driver long ago. I should check the log, there could be new info.
Edit:
DJWillis explains the driver situation:
I’ll elaborate sometime on that but lets just say that is not a viable option for many reasons.I don’t see why they couldn’t have just included the closed source WIFI drier in the firmware until the open source one was ready; add it the first update.
—
Anyway, we actually have 2 open source options so I’ll give people a little background.
One is the TI WiLink 4 based ‘partial’ driver that was opened as part of the Google Android G1 release.
This driver is unfortunately a mess and really has no place in a modern system (ancient design architecture, lots of semi-bespoke shims, some really nasty bugs etc.).
We can use this driver if you don’t mind setting it up from one clunky user space app that crashes a lot and lacks solid support for things like WEP/WPA etc. – Really this driver is only of use for hardware validation not real use from users. Even after I hacked in Wireless Extensions support (yep, the driver is THAT old) it’s still not a driver I would want to use (sorry but I hate lots of nasty random mem leaks and errors).
—
The next open option is the WL12xx driver written by Kalle Valo of Nokia. This is a nice modern driver (MAC80211) and is already in Wireless-Testing (as of a few days ago) and on its way to mainline.
This driver was SPI only but a gent by the name of Bob Copeland added the start of SDIO support and we are working off that base to bring the driver up on the Pandora so we have a nice solid mainline driver to use that is actively developed, supports the Pandora out of the gate, and is easy to maintain.
As of now I have the WL2151 firmware loading on the Pandora with the new driver and lots of nice signs of life but the work is a little way away from passing traffic (the Linux SDIO stack is causing me pain).
Not sure if that counts as an update .
Once I get things settled (hopefully) I want to have a play with trying to get hostapd going as I have some funky uses for a mobile access point in mind .but I have been getting some helpful advice on that front so the plan is to sink a few hours into that over the weekend. Started to push a few simple patches upstream for things like chip ID's so with any luck by the time the driver hits mainline the Pandora should 'just work' if I can nail all the SDIO bits.
If anyone wants patches (or is familiar with the kernels SDIO stack) drop me a bell.
John
Last edited by a moderator: