Re: [RFC PATCH 06/11] x86,fpu: lazily skip fpu restore with eager fpu mode, too

[Oleg Nesterov]

On 01/11, riel@xxxxxxxxxx wrote:
>
> If the next task still has its FPU state present in the FPU registers,
> there is no need to restore it from memory.

Another patch I can't understand...

> --- a/arch/x86/include/asm/fpu-internal.h
> +++ b/arch/x86/include/asm/fpu-internal.h
> @@ -435,13 +435,9 @@ static inline void switch_fpu_prepare(struct task_struct *old, struct task_struc
>  		old->thread.fpu.last_cpu = ~0;
>  		if (preload) {
>  			new->thread.fpu_counter++;
> -			if (!use_eager_fpu() && fpu_lazy_restore(new, cpu))
> -				/* XXX: is this safe against ptrace??? */
> -				__thread_fpu_begin(new);
> -			else {
> +			set_thread_flag(TIF_LOAD_FPU);
> +			if (!fpu_lazy_restore(new, cpu))
>  				prefetch(new->thread.fpu.state);
> -				set_thread_flag(TIF_LOAD_FPU);
> -			}

It is not clear to me why do we set TIF_LOAD_FPU if fpu_lazy_restore()
succeeds. __thread_fpu_begin() is cheap.

At the same time, if switch_fpu_finish() does fpu_lazy_restore() anyway,
why this patch doesn't remove it from switch_fpu_prepare() ?

However,

> @@ -466,6 +462,10 @@ static inline void switch_fpu_finish(void)
>
>  	__thread_fpu_begin(tsk);
>
> +	/* The FPU registers already have this task's FPU state. */
> +	if (fpu_lazy_restore(tsk, raw_smp_processor_id()))
> +		return;
> +

Now that this is called before return to user-mode, I am not sure this is
correct. Note that __kernel_fpu_begin() doesn't clear fpu_owner_task if
use_eager_fpu().

OK, interrupted_kernel_fpu_idle() should fail in this case... but as we
already discussed this means the perfomance regression, so this should
be changed.

Oleg.

--
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/