Linux 2.6.39 Kernel Merge Window Closes With -rc1

[phoronix]

Phoronix: Linux 2.6.39 Kernel Merge Window Closes With -rc1

While we have already been benchmarking code for the Linux 2.6.39 kernel a fair amount at Phoronix with the Nouveau page-flipping and z-compression merge plus Nouveau Fermi acceleration, only this afternoon did Linus Torvalds tag the first release candidate for this next major kernel update...

http://www.phoronix.com/vr.php?view=OTI3MA

[allquixotic]

Anything interesting in here for either i965 or r800? Bugfixes are always fantastic, but it seems that all the major performance work on r600g has died down markedly since the December-January timeframe.

[not.sure]

Anything interesting in here for either i965 or r800? Bugfixes are always fantastic, but it seems that all the major performance work on r600g has died down markedly since the December-January timeframe.

That supposedly happens mostly in mesa, not the kernel.

[Nevertime]

The one thing that I really would like to see added would be support for my broadcom wifi. Don't know if there's any plans to make their older wifi chipsets work though

[elanthis]

Somewhat game related non-news for this release:

I made a half-hearted attempt at cleaning up the xpad driver, but no luck getting in this release. Probably not the next one, either.

The driver right now is utterly incapable of emulating the essential features of XInput (the Windows XInput, not the X11 XInput). It's basically just a half-assed joystick driver, which is completely uninteresting. There's a strong reason developers love XInput and avoid DirectInput on most Windows titles now. Nobody but a handful of weirdos even want generic joystick support, except in very specialized games like flight/driving sims.

After I add the extra API for the missing core features (player index support mostly), I also wanted to add low battery notification and take a crack at adding an ALSA/audio feature, too.

Unfortunately, aside from a lack of time, the complete half-broken state of the driver is making it hard for me as a first-time kernel input developer. The damn thing didn't do hotplug at all for a few months, and now force feedback seems to be broken at the USB/input layer (the xpad ff handlers aren't even getting invoked in my tests, and I'm getting an oh so useful "invalid value" error when I run any tests, including the standard fftest utility; sigh), and Fedora as usual can't even get udev to set the proper device permissions by default (so even if force feedback worked, you wouldn't be able to use it without chmod'ing the js device as root).

Go go Linux.

[AnonymousCoward]

Somewhat game related non-news for this release:

I made a half-hearted attempt at cleaning up the xpad driver, but no luck getting in this release. Probably not the next one, either.

Have you heard of the userspace xboxdrv driver? I'm not sure if it will be usable for your purposes, but as an end user it's a great improvement over the kernel based driver.

http://pingus.seul.org/~grumbel/xboxdrv/