On Mon, Sep 10, 2018 at 07:08:21PM -0600, Jason A. Donenfeld wrote:

Sometimes it's useful to amortize calls to XSAVE/XRSTOR and the related
FPU/SIMD functions over a number of calls, because FPU restoration is
quite expensive. This adds a simple header for carrying out this pattern:

    simd_context_t simd_context = simd_get();
    while ((item = get_item_from_queue()) != NULL) {
        encrypt_item(item, simd_context);
        simd_context = simd_relax(simd_context);
    }
    simd_put(simd_context);

The relaxation step ensures that we don't trample over preemption, and
the get/put API should be a familiar paradigm in the kernel.

Given that it's always supposed to be used like that, mightn't it be
better if simd_relax() took a pointer to the context, so the call is
just

    simd_relax(&simd_context);

?

The inlining means that there won't actually be a pointer dereference in
the emitted code.

If simd_put() also took a pointer then it could set the context back to
HAVE_NO_SIMD as well?

    - Kevin