search

RobertCNelson / stable-kernel

MIT License
74 stars 39 forks source link

Patches for Beagleboard xM at 1 GHz #15

Open teknoman117 opened 12 years ago

teknoman117 commented 12 years ago

I found some patches on the internet (credit is given) from an application specific buildroot setup for the Beagleboard xM that allowed it to run at 1 GHz on kernel 3.2.x. The other day you said it would be nice if I could get it working as well on 3.4.x, so I will try for that next ;)

RobertCNelson commented 12 years ago

Hi Nathaniel,

Couple things...

0002-Fix-sprz319-erratum-2.1.patch

"sprz319" is disabled as it causes classic Beagle Cx's to hard lock on bootup (checked this morning, it still locks up after "Uncompressing Linux... done, booting the kernel.")... I've debugged it previously, and it seems the omap36xx check is being ignored... I left it in the repo so xM users can easily enable it... So we can't enable it by default across all boards...

0021-Enable-1-GHz-clock-for-Beagleboard-xM.patch Yeap, this correctly enables 1Ghz operation, but it's still risky (xM life time wise) without the SmartReflex 1.5 patchset..

Angstrom used this against 3.0.x, some of these never made it upstream.. https://github.com/Angstrom-distribution/meta-ti/tree/master/recipes-kernel/linux/linux-3.0/pm-wip/voltdm

Still pretty cool it works now.. In the past i've enabled via: https://github.com/Angstrom-distribution/meta-ti/blob/master/recipes-kernel/linux/linux-3.0/beagle/0004-OMAP3-beagle-HACK-add-in-1GHz-OPP.patch

and it just hard locks on my xM... So really cool it boots now..

Nishanth Menon, was the last guy at TI working on this.. Haven't heard anything since: http://us.generation-nt.com/answer/patch-v3-00-10-pm-create-avs-adaptive-voltage-scaling-help-207248721.html?page=9

Regards,

teknoman117 commented 12 years ago

Okay, so I'll check those out when I get a chance. I guess that's why no one has ported the patches yet - no one wants to walk though over a hundred patches and inspect them for changes. As far as "forcing it," I hadn't realized that the SmartReflex driver didn't work properly, because I only enable the frequency increase if smart reflex was compiled into the kernel.

As far as the sprz319 erratum patch causing a lock up in the beagleboard Cx series, I'll just add a parameter to say that you are using a beagleboard xm so people don't have to walk through and un-comment it.

You mentioned that forcing it without the smart reflex driver being enabled was dangerous to its life span, unscientifically, for what its worth I hadn't noticed much of an increase in the heat output of the CPU on my Beagleboard xM rev. C. Also, the site that I initially found the patches on mentioned issues with the Beagleboard xM rev. A.