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

[Daniel Colascione]

On Mon, Mar 18, 2019 at 11:58 AM Alan Cox <gnomes@xxxxxxxxxxxxxxxxxxx> wrote:
>
> > I think the compressed tarball is much simpler/easier overall.  If
> > someone really wants the filesystem, they just uncompress it into a
> > tmpfs mount.  It's much less moving kernel code to worry about.
>
> There is an even simpler approach. The people who want this for whatever
> strange reason are Android folks. Android lives on flash, so all they have
> to do is put the headers in a flash file system that is updated with the
> kernel and mount it wherever they like. Simple matter of a bit of
> devicetree no ?

Did you read the rest of the thread? We talked a lot about the
problems with filesystem-based approaches. It's not just "Android
folks" who want BPF-based tools to Just Work, and it's not "strange"
to want that --- what's strange is this reflexive opposition to a
pragmatic feature (one that nobody has to use) that will make a lot of
system analysis cases Just Work. There's nothing wrong with having the
kernel describe itself, and there's no reason to push tons of
error-prone metadata tracking to userspace when the kernel can just
talk about itself in a guaranteed-correct manner. Every argument
against this work is also an argument against /proc/config.gz.
Completely unsurprisingly, /proc/config.gz is in practice super
useful.