LWN: Comments on "Android kernel notes from LPC 2020"

lol

cbushey — Sat, 26 Sep 2020 23:04:36 +0000

I needed a good chuckle. Thank you.

Android kernel notes from LPC 2020

tbird20d — Mon, 21 Sep 2020 01:19:06 +0000

What phone vendor does not release their source? I'm not aware of any.

Android kernel notes from LPC 2020

darwi — Mon, 14 Sep 2020 06:53:06 +0000

> We might even be able to sandbox them.

How do you plan to sandbox Linux drivers? They access core kernel memory from everywhere, and thus normal hardware-based memory separation / address-spaces, afaict, cannot work.

Android kernel notes from LPC 2020

smurf — Sat, 12 Sep 2020 13:29:32 +0000

Unfortunately yes.

On the other hand, it's an ABI, so at least it can be instrumented to discover what the module in question is doing with the rest of the kernel. We might even be able to sandbox them. That alone is a huge improvement WRT the current state of affairs.

Android kernel notes from LPC 2020

Sesse — Fri, 11 Sep 2020 08:56:34 +0000

Where “as before” for many vendors means not at all? :-)

Android kernel notes from LPC 2020

atnot — Fri, 11 Sep 2020 07:57:32 +0000

This was asked in the session and there's no difference, vendors will release their sources as before

Android kernel notes from LPC 2020

mathstuf — Thu, 10 Sep 2020 21:30:49 +0000

I think they'd have a hard time arguing (with a straight face) that their Android-specific kernel modifications weren't somehow derivative of the kernel itself… How often do these components have existing support in other OS deployments that can be claimed to have been "just a port" (akin to nvidia's driver argument)?

Android kernel notes from LPC 2020

meyert — Thu, 10 Sep 2020 21:21:22 +0000

So vendor KMI modules will be binary only available? Or also source code?
If binary only wouldn't the generic kernel image approach make the open sourcing situation worse than currently?
There may are thousands vendor kernels but at least you can get the source code.

Android kernel notes from LPC 2020

Wol — Thu, 10 Sep 2020 18:49:08 +0000

No they're pseudonymous, like me :-)

Cheers,
Wol

Android kernel notes from LPC 2020

nijhof — Thu, 10 Sep 2020 18:46:37 +0000

Are anonymous VMAs still anonymous if they've got a name?