question about xl migrate

question about xl migrate

[Zhang, Chunyu]

hi all

i have two question about xl migrate

write_batch
120     for ( i = 0; i < nr_pfns; ++i )
121     {
122         types[i] = mfns[i] = ctx->save.ops.pfn_to_gfn(ctx,
123                                                       ctx->save.batch_pfns[i]);
124 
125         /* Likely a ballooned page. */
126         if ( mfns[i] == INVALID_MFN )
127         {
128             set_bit(ctx->save.batch_pfns[i], ctx->save.deferred_pages);
129             ++ctx->save.nr_deferred_pages;
130         }
131     }

1.  line 122
in hvm mode, is not mfns, but gfns.
right?

2. line 125
in hvm mode,would not be a balloon page.
gfn would not be INVALID_MFN.
mfn would be INVALID_MFN.
right?

if what i think is right, maybe we can change those code ?


--------------
--------------
A new email address of FJWAN is launched from Apr.1 2007.
The updated address is: zhangcy@cn.fujitsu.com
--------------------------------------------------
Zhang Chunyu
Nanjing Fujitsu Nanda Software Tech. Co., Ltd.(FNST)
No. 6 Wenzhu Road, Nanjing, 210012, China
TEL：+86+25-86630566-9566
FUJITSU INTERNAL：7998-9566
FAX：+86+25-83317685
EMail：zhangcy@cn.fujitsu.com
--------------------------------------------------
This communication is for use by the intended recipient(s) only and may contain information that is privileged, confidential and exempt from disclosure under applicable law. If you are not an intended recipient of this communication, you are hereby notified that any dissemination, distribution or copying hereof is strictly prohibited.  If you have received this communication in error, please notify me by reply e-mail, permanently delete this communication from your system, and destroy any hard copies you may have printed.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: question about xl migrate

[Andrew Cooper]

On 17/05/16 10:01, Zhang, Chunyu wrote:
> hi all
>
> i have two question about xl migrate
>
> write_batch
> 120     for ( i = 0; i < nr_pfns; ++i )
> 121     {
> 122         types[i] = mfns[i] = ctx->save.ops.pfn_to_gfn(ctx,
> 123                                                       ctx->save.batch_pfns[i]);
> 124 
> 125         /* Likely a ballooned page. */
> 126         if ( mfns[i] == INVALID_MFN )
> 127         {
> 128             set_bit(ctx->save.batch_pfns[i], ctx->save.deferred_pages);
> 129             ++ctx->save.nr_deferred_pages;
> 130         }
> 131     }
>
> 1.  line 122
> in hvm mode, is not mfns, but gfns.
> right?

Correct.  The array should be named "gfns" rather than "mfns", which is
the appropriate term for both PV and HVM guests.

>
> 2. line 125
> in hvm mode,would not be a balloon page.
> gfn would not be INVALID_MFN.
> mfn would be INVALID_MFN.
> right?

I don't understand what you asking here.

~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: question about xl migrate

[Zhang, Chunyu]

hi Andrew
>On 17/05/16 10:01, Zhang, Chunyu wrote:
>> hi all
>>
>> i have two question about xl migrate
>>
>> write_batch
>> 120     for ( i = 0; i < nr_pfns; ++i )
>> 121     {
>> 122         types[i] = mfns[i] = ctx->save.ops.pfn_to_gfn(ctx,
>> 123                                                       ctx->save.batch_pfns[i]);
>> 124 
>> 125         /* Likely a ballooned page. */
>> 126         if ( mfns[i] == INVALID_MFN )
>> 127         {
>> 128             set_bit(ctx->save.batch_pfns[i], ctx->save.deferred_pages);
>> 129             ++ctx->save.nr_deferred_pages;
>> 130         }
>> 131     }
>>
>> 1.  line 122
>> in hvm mode, is not mfns, but gfns.
>> right?
>
>Correct.  The array should be named "gfns" rather than "mfns", which is
>the appropriate term for both PV and HVM guests.
>
>>
>> 2. line 125
>> in hvm mode,would not be a balloon page.
>> gfn would not be INVALID_MFN.
>> mfn would be INVALID_MFN.
>> right?
>
>I don't understand what you asking here.

i think those code should delete:
>> 125         /* Likely a ballooned page. */
if page is ballooed, gfns is not  INVALID_MFN, but mormal.
>> 126         if ( mfns[i] == INVALID_MFN )
>> 127         {
>> 128             set_bit(ctx->save.batch_pfns[i], ctx->save.deferred_pages);
>> 129             ++ctx->save.nr_deferred_pages;
>> 130         }
those code is not for balloon.

when xc_get_pfn_type_batch is called,
ballooned page type is XEN_DOMCTL_PFINFO_XTAB
XEN_DOMCTL_PFINFO_XTAB is for ballooned pages.

>
>~Andrew
>
>

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: question about xl migrate

[Andrew Cooper]

>>> 2. line 125
>>> in hvm mode,would not be a balloon page.
>>> gfn would not be INVALID_MFN.
>>> mfn would be INVALID_MFN.
>>> right?
>> I don't understand what you asking here.
> i think those code should delete:
>>> 125         /* Likely a ballooned page. */
> if page is ballooed, gfns is not  INVALID_MFN, but mormal.

Perhaps for HVM guests.  Definitely not for PV guests however.

>>> 126         if ( mfns[i] == INVALID_MFN )
>>> 127         {
>>> 128             set_bit(ctx->save.batch_pfns[i], ctx->save.deferred_pages);
>>> 129             ++ctx->save.nr_deferred_pages;
>>> 130         }
> those code is not for balloon.
>
> when xc_get_pfn_type_batch is called,
> ballooned page type is XEN_DOMCTL_PFINFO_XTAB
> XEN_DOMCTL_PFINFO_XTAB is for ballooned pages.

That is covered by the later check.  This code needs to work for PV
guests as well as HVM guests.  You absolutely can't go deleting this
clause, or you will break migration for PV guests.

~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: question about xl migrate

[Zhang, Chunyu]

hi Andrew
>
>>>> 2. line 125
>>>> in hvm mode,would not be a balloon page.
>>>> gfn would not be INVALID_MFN.
>>>> mfn would be INVALID_MFN.
>>>> right?
>>> I don't understand what you asking here.
>> i think those code should delete:
>>>> 125         /* Likely a ballooned page. */
>> if page is ballooed, gfns is not  INVALID_MFN, but mormal.
>
>Perhaps for HVM guests.  Definitely not for PV guests however.
get it.
>
>>>> 126         if ( mfns[i] == INVALID_MFN )
>>>> 127         {
>>>> 128             set_bit(ctx->save.batch_pfns[i], ctx->save.deferred_pages);
>>>> 129             ++ctx->save.nr_deferred_pages;
>>>> 130         }
>> those code is not for balloon.
>>
>> when xc_get_pfn_type_batch is called,
>> ballooned page type is XEN_DOMCTL_PFINFO_XTAB
>> XEN_DOMCTL_PFINFO_XTAB is for ballooned pages.
>
>That is covered by the later check.  This code needs to work for PV
>guests as well as HVM guests.  You absolutely can't go deleting this
>clause, or you will break migration for PV guests.
get it.
thanks..
>
>~Andrew
>
>

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel