RE: [Xen-ia64-devel][PATCH] fixed an issue about ptc.g

["Xu, Anthony" <anthony.xu@xxxxxxxxx>]

Hi Kouya,

VHPT is walked by hardware, if we don't insert entry into head of VHPT,
Hardware can't walk VHPT collision chain, and then another tlb miss happens, 
this
Tlb miss is not necessary and impact performance.

While VTLB is walked by software, we don't need to insert the entry into head 
of VTLB.

Anthony

>-----Original Message-----
>From: Kouya SHIMURA [mailto:kouya@xxxxxxxxxxxxxx]
>Sent: 2007年4月24日 13:25
>To: Xu, Anthony
>Cc: xen-ia64-devel
>Subject: Re: [Xen-ia64-devel][PATCH] fixed an issue about ptc.g
>
>Hi Anthony,
>
>I'm wondering why you don't use the same logic in vtlb_insert() as
>vmx_vhpt_insert().
>To enclose it with local_irq_disable/enable is not enough?
>
>Thanks,
>Kouya
>
>Xu, Anthony writes:
> > If one vcpu is execute ptc.g, while the other vcpu is execute itc.
> >
> > the VHPT(VTLB) entry which should be purged may survive unexpected.
> >
> > Then issue happens.

_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel