Xen devel - Oct 2006 - [PATCH] an obvious fix to PIC IO intercept

an obvious fix to PIC IO intercept.
In PIC IO, address from send_pio_req is physical address already.

Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
Signed-off-by: Xin Li <xin.b.li@intel.com>


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

Hi Keir, this fix is critical to stability of HVM.
BTW, we are fixing MOVS and INS/OUTS for HVM, hopefully sending the patch
tomorrow, acturally the code is totally messed up, althought they are nearly
never used, but we believe they do introduce stability issue.
thanks
-Xin 
>-----Original Message-----
>From: xen-devel-bounces@lists.xensource.com 
>[mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of Li, Xin B
>Sent: 2006年10月13日 3:10
>To: Xen Devel
>Subject: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>
>an obvious fix to PIC IO intercept.
>In PIC IO, address from send_pio_req is physical address already.
>
>Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
>Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
>Signed-off-by: Xin Li <xin.b.li@intel.com>
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On 13/10/06 9:43 pm, "Li, Xin B" <xin.b.li@intel.com> wrote:
> Hi Keir, this fix is critical to stability of HVM.
> BTW, we are fixing MOVS and INS/OUTS for HVM, hopefully sending the patch
> tomorrow, acturally the code is totally messed up, althought they are
nearly
> never used, but we believe they do introduce stability issue.
Hmm... So what OS is it that does INS/OUTS on the PIC?

 -- Keir



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

>-----Original Message-----
>From: Keir Fraser [mailto:Keir.Fraser@cl.cam.ac.uk] 
>Sent: 2006年10月14日 4:44
>To: Li, Xin B; Xen Devel
>Subject: Re: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>
>On 13/10/06 9:43 pm, "Li, Xin B" <xin.b.li@intel.com> wrote:
>
>> Hi Keir, this fix is critical to stability of HVM.
>> BTW, we are fixing MOVS and INS/OUTS for HVM, hopefully 
>sending the patch
>> tomorrow, acturally the code is totally messed up, althought 
>they are nearly
>> never used, but we believe they do introduce stability issue.
>
>Hmm... So what OS is it that does INS/OUTS on the PIC?
>
Vista.
-Xin

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

More detail on this PIC fix:
Before changeset 11684, we can boot 32bit Vista. But after 11684, we can not,
and Vista gets blocked in halt instruction virtualization, which is a known
issue. Then debug show:
1) only if removing a printk in send_pio_req will make it boot again.
2) Also we found if we compile xen with gcc 4.1.0, it can boot. Usually in our
team we compile xen with gcc 3.4.X.

Then we believe the right thing is Vista can *not* boot.

The root cause is in PIC code.  The reason why Vista can boot is that PIT is not
masked, but acturally Vista wants to mask it using OUTS, but because of the bug
in PIC code, the virt address is converted to phys address twice, we get a wrong
address of the mask value in guest momory :-(, so a random value is written into
PIC mask register.
So some cases it can boot, some cases it can not.

-Xin

>-----Original Message-----
>From: xen-devel-bounces@lists.xensource.com 
>[mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of Li, Xin B
>Sent: 2006年10月13日 3:10
>To: Xen Devel
>Subject: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>
>an obvious fix to PIC IO intercept.
>In PIC IO, address from send_pio_req is physical address already.
>
>Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
>Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
>Signed-off-by: Xin Li <xin.b.li@intel.com>
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

this patch is to fix INS/OUTS.
Thanks
-Xin
>-----Original Message-----
>From: Li, Xin B 
>Sent: 2006年10月14日 4:43
>To: Li, Xin B; Xen Devel
>Subject: RE: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>
>Hi Keir, this fix is critical to stability of HVM.
>BTW, we are fixing MOVS and INS/OUTS for HVM, hopefully 
>sending the patch tomorrow, acturally the code is totally 
>messed up, althought they are nearly never used, but we 
>believe they do introduce stability issue.
>thanks
>-Xin 
>
>>-----Original Message-----
>>From: xen-devel-bounces@lists.xensource.com 
>>[mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of Li, Xin B
>>Sent: 2006年10月13日 3:10
>>To: Xen Devel
>>Subject: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>>
>>an obvious fix to PIC IO intercept.
>>In PIC IO, address from send_pio_req is physical address already.
>>
>>Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
>>Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
>>Signed-off-by: Xin Li <xin.b.li@intel.com>
>>
>

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

No signed-off-by -- are you posting this just for info right now?

The other (PIC) patch is fine. I''ll take it for 3.0.3.

 -- Keir


On 13/10/06 10:44 pm, "Li, Xin B" <xin.b.li@intel.com> wrote:
>  this patch is to fix INS/OUTS.
> Thanks
> -Xin
> 
>> -----Original Message-----
>> From: Li, Xin B 
>> Sent: 2006年10月14日 4:43
>> To: Li, Xin B; Xen Devel
>> Subject: RE: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>> 
>> Hi Keir, this fix is critical to stability of HVM.
>> BTW, we are fixing MOVS and INS/OUTS for HVM, hopefully
>> sending the patch tomorrow, acturally the code is totally
>> messed up, althought they are nearly never used, but we
>> believe they do introduce stability issue.
>> thanks
>> -Xin 
>> 
>>> -----Original Message-----
>>> From: xen-devel-bounces@lists.xensource.com
>>> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of Li, Xin
B
>>> Sent: 2006年10月13日 3:10
>>> To: Xen Devel
>>> Subject: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>>> 
>>> an obvious fix to PIC IO intercept.
>>> In PIC IO, address from send_pio_req is physical address already.
>>> 
>>> Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
>>> Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
>>> Signed-off-by: Xin Li <xin.b.li@intel.com>
>>> 
>> 


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

Does it fix a reproducible issue? If not it can wait til after 3.0.3-0.

 -- Keir

On 13/10/06 11:01 pm, "Li, Xin B" <xin.b.li@intel.com> wrote:
> I want you to review first :-), if it''s ok, I will resend base on
new tip.
> 
> -Xin 
> 
>> -----Original Message-----
>> From: Keir Fraser [mailto:Keir.Fraser@cl.cam.ac.uk]
>> Sent: 2006年10月14日 5:46
>> To: Li, Xin B; Xen Devel
>> Subject: Re: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>> 
>> 
>> No signed-off-by -- are you posting this just for info right now?
>> 
>> The other (PIC) patch is fine. I''ll take it for 3.0.3.
>> 
>> -- Keir
>> 
>> 
>> On 13/10/06 10:44 pm, "Li, Xin B" <xin.b.li@intel.com>
wrote:
>> 
>>>  this patch is to fix INS/OUTS.
>>> Thanks
>>> -Xin
>>> 
>>>> -----Original Message-----
>>>> From: Li, Xin B
>>>> Sent: 2006年10月14日 4:43
>>>> To: Li, Xin B; Xen Devel
>>>> Subject: RE: [Xen-devel] [PATCH] an obvious fix to PIC IO
intercept
>>>> 
>>>> Hi Keir, this fix is critical to stability of HVM.
>>>> BTW, we are fixing MOVS and INS/OUTS for HVM, hopefully
>>>> sending the patch tomorrow, acturally the code is totally
>>>> messed up, althought they are nearly never used, but we
>>>> believe they do introduce stability issue.
>>>> thanks
>>>> -Xin 
>>>> 
>>>>> -----Original Message-----
>>>>> From: xen-devel-bounces@lists.xensource.com
>>>>> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf
>> Of Li, Xin B
>>>>> Sent: 2006年10月13日 3:10
>>>>> To: Xen Devel
>>>>> Subject: [Xen-devel] [PATCH] an obvious fix to PIC IO
intercept
>>>>> 
>>>>> an obvious fix to PIC IO intercept.
>>>>> In PIC IO, address from send_pio_req is physical address
already.
>>>>> 
>>>>> Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
>>>>> Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
>>>>> Signed-off-by: Xin Li <xin.b.li@intel.com>
>>>>> 
>>>> 
>> 


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

I want you to review first :-), if it''s ok, I will resend base on new
tip.

-Xin 
>-----Original Message-----
>From: Keir Fraser [mailto:Keir.Fraser@cl.cam.ac.uk] 
>Sent: 2006年10月14日 5:46
>To: Li, Xin B; Xen Devel
>Subject: Re: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>
>
>No signed-off-by -- are you posting this just for info right now?
>
>The other (PIC) patch is fine. I''ll take it for 3.0.3.
>
> -- Keir
>
>
>On 13/10/06 10:44 pm, "Li, Xin B" <xin.b.li@intel.com>
wrote:
>
>>  this patch is to fix INS/OUTS.
>> Thanks
>> -Xin
>> 
>>> -----Original Message-----
>>> From: Li, Xin B 
>>> Sent: 2006年10月14日 4:43
>>> To: Li, Xin B; Xen Devel
>>> Subject: RE: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>>> 
>>> Hi Keir, this fix is critical to stability of HVM.
>>> BTW, we are fixing MOVS and INS/OUTS for HVM, hopefully
>>> sending the patch tomorrow, acturally the code is totally
>>> messed up, althought they are nearly never used, but we
>>> believe they do introduce stability issue.
>>> thanks
>>> -Xin 
>>> 
>>>> -----Original Message-----
>>>> From: xen-devel-bounces@lists.xensource.com
>>>> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf 
>Of Li, Xin B
>>>> Sent: 2006年10月13日 3:10
>>>> To: Xen Devel
>>>> Subject: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>>>> 
>>>> an obvious fix to PIC IO intercept.
>>>> In PIC IO, address from send_pio_req is physical address
already.
>>>> 
>>>> Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
>>>> Signed-off-by: Xiaowei Yang <xiaowei.yang@intel.com>
>>>> Signed-off-by: Xin Li <xin.b.li@intel.com>
>>>> 
>>> 
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

No, I think we can put it after 3.0.3 is released :-)
-Xin 
>-----Original Message-----
>From: Keir Fraser [mailto:Keir.Fraser@cl.cam.ac.uk] 
>Sent: 2006年10月14日 5:54
>To: Li, Xin B; Xen Devel
>Subject: Re: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>
>Does it fix a reproducible issue? If not it can wait til after 3.0.3-0.
>
> -- Keir
>
>On 13/10/06 11:01 pm, "Li, Xin B" <xin.b.li@intel.com>
wrote:
>
>> I want you to review first :-), if it''s ok, I will resend 
>base on new tip.
>> 
>> -Xin 
>> 
>>> -----Original Message-----
>>> From: Keir Fraser [mailto:Keir.Fraser@cl.cam.ac.uk]
>>> Sent: 2006年10月14日 5:46
>>> To: Li, Xin B; Xen Devel
>>> Subject: Re: [Xen-devel] [PATCH] an obvious fix to PIC IO intercept
>>> 
>>> 
>>> No signed-off-by -- are you posting this just for info right now?
>>> 
>>> The other (PIC) patch is fine. I''ll take it for 3.0.3.
>>> 
>>> -- Keir
>>> 
>>> 
>>> On 13/10/06 10:44 pm, "Li, Xin B"
<xin.b.li@intel.com> wrote:
>>> 
>>>>  this patch is to fix INS/OUTS.
>>>> Thanks
>>>> -Xin
>>>> 
>>>>> -----Original Message-----
>>>>> From: Li, Xin B
>>>>> Sent: 2006年10月14日 4:43
>>>>> To: Li, Xin B; Xen Devel
>>>>> Subject: RE: [Xen-devel] [PATCH] an obvious fix to PIC IO 
>intercept
>>>>> 
>>>>> Hi Keir, this fix is critical to stability of HVM.
>>>>> BTW, we are fixing MOVS and INS/OUTS for HVM, hopefully
>>>>> sending the patch tomorrow, acturally the code is totally
>>>>> messed up, althought they are nearly never used, but we
>>>>> believe they do introduce stability issue.
>>>>> thanks
>>>>> -Xin 
>>>>> 
>>>>>> -----Original Message-----
>>>>>> From: xen-devel-bounces@lists.xensource.com
>>>>>> [mailto:xen-devel-bounces@lists.xensource.com] On
Behalf
>>> Of Li, Xin B
>>>>>> Sent: 2006年10月13日 3:10
>>>>>> To: Xen Devel
>>>>>> Subject: [Xen-devel] [PATCH] an obvious fix to PIC IO
intercept
>>>>>> 
>>>>>> an obvious fix to PIC IO intercept.
>>>>>> In PIC IO, address from send_pio_req is physical
address already.
>>>>>> 
>>>>>> Signed-off-by: Xiaohui Xin
<xiaohui.xin@intel.com>
>>>>>> Signed-off-by: Xiaowei Yang
<xiaowei.yang@intel.com>
>>>>>> Signed-off-by: Xin Li <xin.b.li@intel.com>
>>>>>> 
>>>>> 
>>> 
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel