Re: ABI change for device drivers using future AVX instruction set

[Arjan van de Ven <arjan@infradead.org>]

On Wed, 25 Jun 2008 17:32:36 +0200
Agner Fog <agner@agner.org> wrote:

> 
> One problem that has not been resolved yet, AFAIK, is how to handle
> the possible use of YMM registers in device drivers. Should these
> registers be saved before calling a device driver from an interrupt
> or should it be the responsibility of the device driver?

the answer is: you don't use AVX in drivers just as you don't use SSE
in drivers.

Let me repeat this loud and clear:

It is not allowed to use floating point, SSE of AVX in device drivers.

(there are few places that very carefully do this anyway, the raid5
code being the most obvious one, but we should not add more)


> 2. There is a performance cost to using XSAVE / XRESTOR.

a patch introducing xsave/xrestor is being discussed already here...
it's not really more expensive.

> 
> 3. When compiling a device driver, the compiler may insert implicit 
> calls to library functions such as memcpy and memset. These functions 
> typically have a CPU dispatcher that chooses the largest register
> size available. The device driver may therefore use YMM registers
> without the knowledge of the programmer and without compiling with
> the AVX switch on.

this is not correct; the library functions are provided by the kernel
and do not use AVX etc.


-- 
If you want to reach me at my work email, use arjan@linux.intel.com
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org