Linux Headers May Soon Be Available In-Kernel Via /proc

[Vistaus]

Wouldn't this only ever be useful for Android where each device/variant needs a fork of the kernel.

But why would they include this when they're now seriously working on Fuchsia, the successor to Android, and even hired a hell of a veteran for it?
https://mobilesyrup.com/2019/01/23/g...teran-fuchsia/

[mifritscher]

Why not a squashfs? This way it would be mountable, enabling direct access. a tgz must be decompressed to access the files...

[pal666]

if android doesn't ship kernel headers, then correct solution is to ship kernel headers, not to bundle them into kernel

[pal666]

But why would they include this when they're now seriously working on Fuchsia, the successor to Android

because fuchsia is not successor to linux. no single company can compete with linux. google can't even maintain their fork of linux and are working on merging it back and switching to mainline kernels. fuchsia being successor is just fantasy of uninformed journalists

[pal666]

A typical distro won't need it - it already has a powerful packaging system and a kernel-header package that matches the kernel itself. But embedded devices are a very different beast

lolwut? android can't ship in one update both kernel and headers?

[[email protected]]

Why not a squashfs? This way it would be mountable, enabling direct access. a tgz must be decompressed to access the files...

I am quite agnostic to this proposal, but why not maintain a virtual file system from which userspace can read the headers also use the usual filesystem features (cache to ram, stream from disk). Heck, the .ko could even contain a full filesystem that supports compression.

Now, I might ask, why undergo all the trouble of including the kernel headers, and not the kernel sources themselves? *that* would be useful for android devices

[Grogan]

That will be more for compiling modules, the headers you want on your system are the sanitized headers that were used for your glibc build (on our PCs with Linux distros anyway). It's better than it used to be, these used to be cleaned up by hand but for some time now the kernel build system has a facility to generate them (currently "make headers_install")

[Vistaus]

because fuchsia is not successor to linux. no single company can compete with linux. google can't even maintain their fork of linux and are working on merging it back and switching to mainline kernels. fuchsia being successor is just fantasy of uninformed journalists

So then why have they hired that Apple veteran to work on Fuchsia and why are they working with Huawei on the first Fuchsia device? Fuchsia's gotta be *something* for them to hire that veteran and to work on actual consumer devices powered by it.

[smitty3268]

if android doesn't ship kernel headers, then correct solution is to ship kernel headers, not to bundle them into kernel

Yeah, why isn't that the obvious solution here? I'm not sure i get why this was even proposed.

[rene]

what a waste of memory resources, that is what a file system and proper files in /usr/include where invented for a couple of decades ago, also 99% of the users will never need them on the Android phone to start with :-/