Re: [PATCH v10 04/28] x86/fpu/xstate: Modify address finders to handle both static and dynamic buffers

[Thomas Gleixner]

On Sun, Oct 03 2021 at 22:35, Chang Seok Bae wrote:
> On Oct 1, 2021, at 06:15, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
>
> Okay, a NULL pointer is odd and it as an argument should be avoided. Defining
> a separate struct fpu for the initial state can make every function expect a
> valid struct fpu pointer.
>
> I think that the patch set will have such order (once [1] is dropped out) of,
>     - patch1 (new): a cleanup patch for fpstate_init_xstate() in patch1
>     - patch2 (new): the above init_fpu goes into this, and 
>     - patch3-5: changes arguments to fpu,

So actually I sat down over the weekend and looked at that again. Adding this
to struct fpu is wrong. The size and features information belongs into
something like this:

struct fpstate {
	unsigned int		size;
        u64			xfeatures;

        union fpregs_state	regs;
};

Why? Simply because fpstate is the container for the dynamically sized
regs and that's where it semantically belongs.

While staring at that I just started to cleanup stuff all over the place
to make the integration of this simpler.

The patches are completely untested and have no changelogs yet, but if
you want a preview, I've uploaded a patch series to:

    https://tglx.de/~tglx/patches.tar

I'm still staring at some of the dynamic feature integrations, but this
is roughly where this should be heading.

Thanks,

        tglx