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

Re: [Xen-devel][PATCH] patch to support super page (2M) with EPT



Exactly. The interface for populate_physmap() is clear — if you ask for order-9 allocations then that is what you must get. Otherwise the allocation fails. It is up to the caller to retry with order-0 allocations _if_ that is a suitable fallback.

 -- Keir

On 13/5/08 14:39, "Li, Xin B" <xin.b.li@xxxxxxxxx> wrote:

if 2M page allocation fails, the domain builer will try to use 4K allocation instead.
-Xin


 

From: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx  [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of Huang2,  Wei
Sent: 2008å5æ13æ 21:37
To: Keir Fraser; Xin, Xiaohui;  xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: RE: [Xen-devel][PATCH] patch  to support super page (2M) with EPT

 
 
 

Memory.c  looks more invasive because it takes care of failure cases using 4KB pages.  Xiaohui’s patch tries to allocate pages using extend_order. But if this  request fails for any reason, the guest cannot be started anymore.  



-Wei



 
 

From: Keir Fraser  [mailto:keir.fraser@xxxxxxxxxxxxx]
Sent: Tuesday, May 13, 2008 3:47  AM
To: Huang2, Wei; Xin, Xiaohui;  xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: Re: [Xen-devel][PATCH] patch  to support super page (2M) with EPT

 
 
Overall your  changes to common code seem a bit more invasive than in the Intel patch. In  particular I don’t understand why you made such changes to common/memory.c.  The other patch makes far fewer changes (and even some of those would go away  since they have erroneously changed the populate_physmap interface). So my  feeling is that the Intel patch is a slightly more elegant base to start with:  extra changes that your patch makes really need to be accounted  for.

 -- Keir

On 12/5/08 18:28, "Huang2, Wei"  <Wei.Huang2@xxxxxxx> wrote:

Here is  a revised version. I get rid of 4MB support, as suggested. I did not see the  issue mentioned by Xiaohui related to splitting last 2M into 4KB pages. But  anyway, I attached two versions for your reference. Keir, please let me know  if you have comments.

 
-Wei





From: Keir Fraser [mailto:keir.fraser@xxxxxxxxxxxxx]  
Sent: Monday, May 12, 2008 2:03 AM
To: Xin, Xiaohui;  Huang2, Wei; xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: Re:  [Xen-devel][PATCH] patch to support super page (2M) with EPT

Yes,  absolutely no need for 4M page support. We do not support the 32-bit non-PAE  build target any more.

 -- Keir

On 12/5/08 06:04, "Xin,  Xiaohui" <xiaohui.xin@xxxxxxxxx> wrote:

Some  comments here:
1) Basically 4M pages  allocations is not hardware  naturally for EPT, we only use 2M super pages now.  
I remembered that  Keir said that 2M pages allocation is sufficient, and he  removed all the  pure 32bit support already.
2)  If we don’t allocate  the last 2M  area with 4kb pages, the EPT will meet some problem. Xen will set  one of  the 4k page
     there to be invalid,   logically that means we should invalid the all the 2M page if we  allocate it  with 2M, and then the  
     special pages Xen used  in the high end  of the guest memory can not be used then. May we know how you  cope with   that?
 
Thanks
Xiaohui
 





From:   xen-devel-bounces@xxxxxxxxxxxxxxxxxxx [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxx]   On Behalf Of Huang2, Wei
Sent: Monday, May 12, 2008  12:36  PM
To: Xin, Xiaohui;   xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: RE: [Xen-devel][PATCH]  patch  to support super page (2M) with EPT

This  is the latest one I created. Please review it and  I will  re-submit.
 
1.
       It  includes the  patch for p2m-ept.c, directly from your previous patch.   

2.        Xc_hvm_create.c   is based on my original approach. It includes support for both 2MB and  4MB  pages. Also it considers the case of odd page size (such as 255MB).  But I did  not allocate the last 2MB area using 4KB  pages. Let me  know if it is a  big issue.

3.        The  rest are  pretty similar.


Thanks,
 
-Wei
 

From:  xen-devel-bounces@xxxxxxxxxxxxxxxxxxx [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxx]   On Behalf Of Huang2, Wei
Sent: Sunday, May 11, 2008  3:34  PM
To: Xin, Xiaohui;   xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: RE: [Xen-devel][PATCH]  patch  to support super page (2M) with EPT

Could  we work together for a common solution? As far  as I can see, it largely  overlaps with my super page patch. The major  difference is between p2m.c  and  p2m-ept.c.
 
-Wei
 

From:  xen-devel-bounces@xxxxxxxxxxxxxxxxxxx [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxx]   On Behalf Of Xin, Xiaohui
Sent: Friday, May 09, 2008  4:11  AM
To: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject:   [Xen-devel][PATCH] patch to support super page (2M) with   EPT

Attached are the  patches to support  super page with EPT. We only support 2M size. And  shadow may still work fine  with 4K pages.
The patches can be split  into 3 parts. Apply order is as  attached.
 
tool.diff
To  allocate 2M physical contiguous memory  in guest except the first 2M and  the last 2M.
The first 2M covers special  memory, and Xen use the last  few pages in guest memory to do special  things.
We let them to be 4K  pages as normal.
super_page_common.patch  
To modify the p2m  interfaces by adding an order parameter, such as   guest_physmap_add_page(), p2m_set_entry(),   etc.
p2m-ept-file.patch
           To   handle the EPT tables to support super page.              
 
 
Signed-off-by:   Xin Xiaohui <xiaohui.xin@xxxxxxxxx>
Signed-off-by: Li Xin, B   <xin.b.li@xxxxxxxxx>
 

 

 





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






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

 



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

 


Rackspace

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