On Thu, Feb 28, 2019 at 05:34:44PM +0900, Masami Hiramatsu wrote:
Hi Joel,
On Wed, 27 Feb 2019 14:37:47 -0500 "Joel Fernandes (Google)" joel@joelfernandes.org wrote:
Introduce in-kernel headers and other artifacts which are made available as an archive through proc (/proc/kheaders.tar.xz file). This archive makes it possible to build kernel modules, run eBPF programs, and other tracing programs that need to extend the kernel for tracing purposes without any dependency on the file system having headers and build artifacts.
On Android and embedded systems, it is common to switch kernels but not have kernel headers available on the file system. Raw kernel headers also cannot be copied into the filesystem like they can be on other distros, due to licensing and other issues. There's no linux-headers package on Android. Further once a different kernel is booted, any headers stored on the file system will no longer be useful. By storing the headers as a compressed archive within the kernel, we can avoid these issues that have been a hindrance for a long time.
Hmm, isn't it easier to add kernel-headers package on Android?
I have already been down that road. In the Android ecosystem, the Android teams only provide a "userspace system image" which goes on the system partition of the flash (and a couple other images are also provided but system is the main one). The system image cannot contain GPL source code. It is also not possible to put kernel headers for every kernel version on the system images that ship and is not practical. Android boots on 1000s of forked kernels. It does not make sense to provide headers on the system image for every kernel version and I already had many discussions on the subject with the teams, it is something that is just not done. Now for kernel modules, there's another image called the "vendor image" which is flashed onto the vendor parition, this is where kernel modules go. This vendor image is not provided by Google for non-Pixel devices. So we have no control over what goes there BUT we do know that kernel modules that are enabled will go there, and we do have control over enforcing that certain kernel modules should be built and available as they are mandatory for Android to function properly. We would also possibly make it a built-in option as well. Anyway my point is keeping it in the kernel is really the easiest and the smartest choice IMO.
The feature is also buildable as a module just in case the user desires it not being part of the kernel image. This makes it possible to load and unload the headers on demand. A tracing program, or a kernel module builder can load the module, do its operations, and then unload the module to save kernel memory. The total memory needed is 3.8MB.
But it also requires to install build environment (tools etc.) on the target system...
Yes, that's true. Check the other thread with Masahiro that we are discussing this point on and let us continue discussing there: https://lore.kernel.org/patchwork/patch/1046307/#1238223 https://lore.kernel.org/patchwork/patch/1046307/#1238491
The code to read the headers is based on /proc/config.gz code and uses the same technique to embed the headers.
To build a module, the below steps have been tested on an x86 machine: modprobe kheaders rm -rf $HOME/headers mkdir -p $HOME/headers tar -xvf /proc/kheaders.tar.xz -C $HOME/headers >/dev/null cd my-kernel-module make -C $HOME/headers M=$(pwd) modules rmmod kheaders
It seems a bit complex, but no difference from compared with carrying kheaders.tar.gz. I think we would better have a psudo filesystem which can mount this compressed header file directly :) Then it becomes simpler, like
modprobe headerfs mkdir $HOME/headers mount -t headerfs $HOME/headers
And this doesn't consume any disk-space.
I felt using a compressed tar is really the easiest way because of all the tools are already available. There isn't a compressed in-ram filesystem right now that I'm aware off that can achieve the kind of high compression ratio this patchset does.
thanks,
- Joel