AOSP Automotive 9


Huh. I was all set to test today, and got hit with the bluetooth scan bug:

Edit: Got that straightened out and managed to pair it with my phone, I have the hikey960 tethering via bluetooth, and my phone has network via wifi.

So far, I’ve made it through more than a dozen runs of speedtest and have NOT managed to crash the bluetooth. I’ll keep hammering it, but for now it is very promising. If it stands up to this today, I’ll try raising the baud rate tomorrow, see if it can handle 3M.

… now past 30 runs and still going. It never would have made it this far before. It would usually crash out in 2-3 runs, 5 tops.


@doitright Great to hear! Those changes got merged yesterday. I also posted a workaround to bug #808 to gerrit to hopefully prod the developers to look a bit more into the issue.

Be aware, with the dma enablement, we do occasionally trip over a k3dma issue:

But I suspect we can get a solution for that soon.


Yeah, I noticed that dma bug, and was going to ask you about it. So far, it seems that even with that dma bug, its a hell of a lot more stable than before. I’m pushing close to a hundred runs of speedtest and its still holding.

Is it possible that that bug doesn’t affect the Android 4.9 kernel? Up at the top you mention mainline WIP kernel.


Potentially! I’ve not looked closely at the k3dma 4.9 delta from upstream.


Hmm, its actually looking like I might not be tripping it just because I’m not running it as fast. Your kernel has uart4 set for 3000000 baud. AOSP 4.9 is at 921600. I won’t be able to test out higher speeds today, but I should be able to by tomorrow.

I might have hit that one.
Unfortunately, I don’t have any uart board with me, so I didn’t actually see what it said, but the board appeared to shut down, for lack of a better way to describe it, and I couldn’t get any action (including via USB-C) except by pulling the plug. This board also has a Toshiba UFS and running UEFI, so its hitting that bug too, which obviously isn’t doing me any favors.