Re: [RFC] x86: save_args out of line

[Alexander van Heukelum]

On Mon, 17 Nov 2008 22:22:16 +0300, "Cyrill Gorcunov"
<gorcunov@xxxxxxxxx> said:
> [Andi Kleen - Mon, Nov 17, 2008 at 07:23:40PM +0100]
> ...
> | 
> | Ok. Hopefully it's worth the effort. The branch misprediction bubble
> | should not be too bad, perhaps it'll make up for the other cycles
> | you're adding. But even if it doesn't decreasing cache line foot print
> is 
> | always a good thing.
> ...
> 
> May I turn in? :)

Sure!

> Original .macro interrupt func has
> 
> 	testl $3, CS(%rdi)
> 	je 1f
> 	SWAPGS
> 1:	incl %gs:pda_irqcount

inc changes flags...

> 	jne 2f
> 	pop %rax
> 	mov %gs:pda_irqstackptr,%rsp
> 	push %rax
> 2:	TRACE_IRQS_OFF
> 
> Wouldn't we help the branch predictor a bit by
> 
> 	testl $3, CS(%rdi)
> 	je 1f
> 	SWAPGS
> 	jne 2f

... so this is not the same.

> 1:	pop %rax
> 	mov %gs:pda_irqstackptr,%rsp
> 	push %rax	
> 2:	incl %gs:pda_irqcount
> 	TRACE_IRQS_OFF
> 
> I hope I didn't miss anything but maybe it's just a churning
> or plain wrong. Don't shoot me :)

I'm sure it helps the branch predictor, though ;).

Alexander

> 		- Cyrill -
-- 
  Alexander van Heukelum
  heukelum@xxxxxxxxxxx

-- 
http://www.fastmail.fm - A fast, anti-spam email service.

--
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/