On Mon, 26 Feb 2024 at 09:17, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> Are tiny reads (handwaving: 100 bytes or less) really worth optimizing
> for to that degree?

So this ended up just bothering me, and dammit, we already *have* that
stack buffer, except it's that 'folio batch' buffer.

Which is already 128 bytes of stack space (on 64-bit).

So, just hypothetically, let's say that *before* we start using that
folio batch buffer for folios, we use it as a dummy buffer for small
reads.

So we'd make that 'fbatch' thing be a union with a temporary byte buffer.

That hypothetical patch might look something like this TOTALLY UNTESTED CRAP.

Anybody interested in seeing if something like this might actually
work? I do want to emphasize the "something like this".

This pile of random thoughts ends up compiling for me, and I _tried_
to think of all the cases, but there might be obvious thinkos, and
there might be things I just didn't think about at all.

I really haven't tested this AT ALL. I'm much too scared. But I don't
actually hate how the code looks nearly as much as I *thought* I'd
hate it.

              Linus