Re: [Xen-ia64-devel] [PATCH][3/3] Steal time accounting forPVdomain/IA64 TAKE2

[Atsushi SAKAI <sakaia@xxxxxxxxxxxxxx>]

Hi, Alex

 Thank you for various comments. It helps many
I will post a patch tomorrow, now testing.

Anyway I reply it in inline.

> Why do we have to do these one at a time, instead of in bulk.  Seems
> very inefficient.  Also, please explain the local_irq_enable/disable at
> the end of each.

I re-check and confirmed that only do_timer needs multiple call.
But this do_timer loop is not needed for later 2.6.19 
since do_timer spec is changed from register value to tick counts.

For local_irq_enable/disable Issue,
I saw 2.6.20 kernel and found timer_interrupt has local_irq_enable/disable.
But this issue is effective for > 1024CPU case,
So I will omit it since current CPU# is 64.

Reference.
http://www.gelato.unsw.edu.au/archives/linux-ia64/0610/19314.html

> Shouldn't this be in an #ifdef CONFIG_XEN?  HYPERVISOR_vcpu_op isn't
> going to compile otherwise.  It should probably have a stub for the
> non-CONFIG_XEN case so you don't have to #ifdef out the caller.

I forgot to add #ifdef CONFIG_XEN. 
I will fix it.

> Is there any reason to filter out the other defined vcpu_ops?  Maybe we
> should add support for them now. 

I filter vcpu_op, just because I check only one hypercall.
I consider it as my backlog.

Thanks
Atsushi SAKAI




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