Re: [PATCH v4 1/2] Provide in-kernel headers for making it easy to extend the kernel

From: Alexei Starovoitov
Date: Mon Mar 11 2019 - 21:45:33 EST

On Mon, Mar 11, 2019 at 09:28:23PM -0400, Steven Rostedt wrote:
> On Mon, 11 Mar 2019 20:39:12 -0400
> Joel Fernandes <joel@xxxxxxxxxxxxxxxxx> wrote:
> > I think even though the kernel-headers can't have information about all data
> > structures, they do already contain a lot of data structure definitions we
> > need already. And anything needed can/should arguably be moved to include/ if
> > they are really needed for kernel extension by something "external" to the
> > kernel such as kernel modules or eBPF, right?
> That's not my worry. I would like to be able to easily walk data
> structures from within the kernel, without having to do a lot of work
> in userspace to get that information. The kprobe_events could then be
> passed type casts or such to access data fields of arguments to
> functions and such.
> >
> > In any case, such a solution such as what Steve suggested, still cannot do
> > what we can with headers - such as build kernel modules on the fly using the
> > C-compiler without any auto-generation of C code from any debug artifiacts.
> > Think systemtap working with the module-backend without any need for
> > linux-headers package on the file system. So such a solution would still be a
> > bit orthogonal in scope to what this proposed solution can solve IMO.
> >
> With the information I would like to have, it would be trivial to read
> the data to create the header files needed for modules.

what you're asking for we already have. It's called BTF.
pahole takes vmlinux dwarf and convert it into ~1Mbyte of BTF that includes
all kernel types.
With gzip it can be compressed further if necessary.
We also have a prototype to generate all_vmlinux_types.h from BTF.
But it's not a substitute for kernel headers.
We've had a long discussion during last LPC regarding this:
tldr: many tracing use cases will be solved with BTF, but kernel headers
are here to stay.