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

From: Avi Kivity
Date: Sun Jun 29 2008 - 08:32:49 EST


Andi Kleen wrote:
Avi Kivity wrote:

This is fixable.

Sure nearly everything is fixable, but why would you want to do that?

For me the best fix currently seems to be to just not do that
when it hurts.


We already use sse in the kernel (raid), presumably disabling preemption there hurts some workloads.

With sse/avx gaining more features, we may see more requirements for kernel fpu.

We could change kernel_fpu_begin() not to disable
preemption, but instead set a task flag. When we get the "no device"
fault, if the flag is set, save the fpu state into the kernel fpu save
area

What kernel fpu save area do you mean?

A new one, of course.

--
error compiling committee.c: too many arguments to function

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/