[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-ia64-devel] [PATCH 0/3] [RFC] clean up priv_handle_op() and paravirtualise vdso



Alex, could you apply this patch series?
priv_handle_op() should be fixed and There is no objection against
this patch series.

This patch focuses on performace, It should determined based on 
a some sort of benchmark that whether we should go for option B or not.
So I attached para-virtualized but non-trasnparent version of gate.S.
for those who want to evaluate transparent paravirtualization overhead.

Thanks

On Tue, May 30, 2006 at 12:09:40PM +0900, Isaku Yamahata wrote:
> 
> priv_handle_op() caused some troubles.
> the 1/3 patch fixes them. However it causes performance loss
> because vdso area always requires priv_handle_op() and itlb reflection.
> System call enterance is performance critical so that it is
> worthwhile for paravirtualization.
> The 3/3 patch paravirtualizes it.
> 
> There are choices to paravirtualize vdso area for transparent virtualization.
> A. add running_on_xen check to gate.S directly or
> B. prepare two version of gate.S, and use appropriate version at setup.
> However B. requires modification to additional several files.
> So My current choice is option A.
> 
> Should we go for option B?
> 
> _______________________________________________
> Xen-ia64-devel mailing list
> Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-ia64-devel
> 

-- 
yamahata

Attachment: gate.S-non-transparent
Description: Text document

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

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.