Profile cover photo
Profile photo
Duncan Laurie
About
Communities and Collections
View all
Posts

Post has shared content
LCA2013 was a lot of fun and I highly recommend attending!

I had hoped to be able to bring and talk about Pixel at this event but unfortunately the release schedule did not line up.

I made a slight modification to the online slides at http://bit.ly/chromefw to indicate that Pixel is the second device to feature our open source Embedded Controller firmware.

Linux Weekly News did a nice write-up of my talk if people are interested in reading about it rather than watching it: http://lwn.net/Articles/537791/
Google Chrome OS is not a regular Operating System. You already knew that. 

But did you know Chrome OS Team was putting a lot of effort into its custom firmware and that it was open-sourced?
Yup, if you want to build a secure device and a polished user experience, you need to control that. And what is better than open-sourcing it?

I'd recommend you watch the first part of this talk at linux.conf.au 2013 event to understand that the firmware's world is not the Care Bears' world and that it's not easy to deal with big vendors when you're building from scratch a device.

Slides are available at http://bit.ly/chromefw
Add a comment...

Post has shared content
+Linus Torvalds I have been attempting to add this comment to the original post since last night but it does not show up for anyone but me.  Perhaps it will show up at some point but until then maybe this will be visible.

The resume is failing in this case because the firmware is seeing the TPM in an unexpected state and deciding not to trust it.

If you load the TPM driver (I used modprobe tpm_tis force=1 interrupts=0) so that the TPM is initialized and suspended properly then it will be happier.  This looks to be compiled into our kernel directly so we did not see the dependency.

There is an issue with the tpm_tis driver suspending the first time (sometimes the first few times) after it is loaded.  I am not sure why this happens yet, but once it starts working it seems to continue to work.

We do have a few TPM driver patches in our 3.4 tree that may need to be reconciled and/or sent upstream.  There might also be something in the normal/verified ChromeOS boot path that is putting the TPM into a different state which could be altering the driver behavior.  I will investigate this further.
Pixel update: current git now gets you a pretty complete experience (ie the touchpad works).

And I pushed out the Pixel changes from the Pixel.

That said, there's a few oddities still, related to it not being a normal PC. +David Miller is working on getting the keyboard mapping DTRT automatically, and there's something wrong with suspend.
Add a comment...
Wait while more posts are being loaded