Re: [PATCH] xen: correctly check for pending events when restoring irq flags

[Ian Campbell <Ian.Campbell@citrix.com>]

On Thu, 2012-04-26 at 19:44 +0100, David Vrabel wrote:
> From: David Vrabel <david.vrabel@citrix.com>
> 
> In xen_restore_fl_direct(), xen_force_evtchn_callback() was being
> called even if no events were pending.

In actual fact it seems that the callback was actually being called if
and only if no events were pending? Which makes me wonder how it used to
work at all!

> diff --git a/arch/x86/xen/xen-asm.S b/arch/x86/xen/xen-asm.S
> index 79d7362..3e45aa0 100644
> --- a/arch/x86/xen/xen-asm.S
> +++ b/arch/x86/xen/xen-asm.S
> @@ -96,7 +96,7 @@ ENTRY(xen_restore_fl_direct)
>  
>  	/* check for unmasked and pending */
>  	cmpw $0x0001, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_pending
> -	jz 1f
> +	jnz 1f
>  2:	call check_events
>  1:

Took me a while, this is a bit tricksy (and it may well be too early for
me to be decoding it) since the check here is trying to check both
pending and masked in a single cmpw, but I think this is correct. It
will call check_events now only when the combined mask+pending word is
0x0001 (aka unmasked, pending).

Acked-by: Ian Campbell <ian.campbell@citrix.com>

Does xen_irq_enable_direct have the same sort of issue? No, in that case
we are doing a straight forward test of pending without involving masked
(since it has just unmasked) and so jz is correct.

Ian.