RE: [RFC PATCH 0/3] kernel: add support for 256-bit IO access

[David Laight]

From: Thomas Gleixner
> Sent: 20 March 2018 09:41
> On Tue, 20 Mar 2018, Ingo Molnar wrote:
> > * Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
...
> > > And if we go down that road then we want a AVX based memcpy()
> > > implementation which is runtime conditional on the feature bit(s) and
> > > length dependent. Just slapping a readqq() at it and use it in a loop does
> > > not make any sense.
> >
> > Yeah, so generic memcpy() replacement is only feasible I think if the most
> > optimistic implementation is actually correct:
> >
> >  - if no preempt disable()/enable() is required
> >
> >  - if direct access to the AVX[2] registers does not disturb legacy FPU state in
> >    any fashion
> >
> >  - if direct access to the AVX[2] registers cannot raise weird exceptions or have
> >    weird behavior if the FPU control word is modified to non-standard values by
> >    untrusted user-space
> >
> > If we have to touch the FPU tag or control words then it's probably only good for
> > a specialized API.
> 
> I did not mean to have a general memcpy replacement. Rather something like
> magic_memcpy() which falls back to memcpy when AVX is not usable or the
> length does not justify the AVX stuff at all.

There is probably no point for memcpy().

Where it would make a big difference is memcpy_fromio() for PCIe devices
(where longer TLP make a big difference).
But any code belongs in its implementation not in every driver.
The implementation of memcpy_toio() is nothing like as critical.

If might be the code would need to fallback to 64bit accesses
if the AVX(2) registers can't currently be accessed - maybe some
obscure state....

However memcpy_to/fromio() are both horrid at the moment because
they result in byte copies!

	David