All of lore.kernel.org
 help / color / mirror / Atom feed
* About the oprofile tools and related profiling tools
@ 2011-05-30  3:25 Lv, Hui
  2011-05-31  9:44 ` George Dunlap
  2011-05-31 14:35 ` Konrad Rzeszutek Wilk
  0 siblings, 2 replies; 7+ messages in thread
From: Lv, Hui @ 2011-05-30  3:25 UTC (permalink / raw)
  To: xen-devel


[-- Attachment #1.1: Type: text/plain, Size: 699 bytes --]

Hi, list

I found the oprofile tools were not supported in PVOPS domain0(from jeremy's tree). Someone has made the patch to enable it. When I applied such patch, xenoprofile can work. But the results were not trusted. (I think, part of this reason was due to xen, which changed the NMI handling way, but did not update the xenoprofile part)
So my question is that, does anyone still maintain this tools? Or, do xen guys have a plan to enable more powerful profiling tools like 'perf' in native Linux or something else.
One of my basic feeling about xen's performance profiling tools is that the functional level profiling tools like oprofile were neglected recently.

Best regards!
Lv,hui


[-- Attachment #1.2: Type: text/html, Size: 5647 bytes --]

[-- Attachment #2: Type: text/plain, Size: 138 bytes --]

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

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: About the oprofile tools and related profiling tools
  2011-05-30  3:25 About the oprofile tools and related profiling tools Lv, Hui
@ 2011-05-31  9:44 ` George Dunlap
  2011-05-31 14:35 ` Konrad Rzeszutek Wilk
  1 sibling, 0 replies; 7+ messages in thread
From: George Dunlap @ 2011-05-31  9:44 UTC (permalink / raw)
  To: Lv, Hui; +Cc: xen-devel

On Mon, May 30, 2011 at 4:25 AM, Lv, Hui <hui.lv@intel.com> wrote:
> I found the oprofile tools were not supported in PVOPS domain0(from jeremy’s
> tree). Someone has made the patch to enable it. When I applied such patch,
> xenoprofile can work. But the results were not trusted. (I think, part of
> this reason was due to xen, which changed the NMI handling way, but did not
> update the xenoprofile part)
>
> So my question is that, does anyone still maintain this tools? Or, do xen
> guys have a plan to enable more powerful profiling tools like ‘perf’ in
> native Linux or something else.
>
> One of my basic feeling about xen’s performance profiling tools is that the
> functional level profiling tools like oprofile were neglected recently.

Indeed, they have rather been.  I'd love xenoprofile to be in a better
working state, but unfortunately I simply don't have the time to work
on it.  But I'd be willing to do what I could to help someone who did
step up to make things work.

 -George

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: About the oprofile tools and related profiling tools
  2011-05-30  3:25 About the oprofile tools and related profiling tools Lv, Hui
  2011-05-31  9:44 ` George Dunlap
@ 2011-05-31 14:35 ` Konrad Rzeszutek Wilk
  2011-05-31 15:42   ` Lv, Hui
  1 sibling, 1 reply; 7+ messages in thread
From: Konrad Rzeszutek Wilk @ 2011-05-31 14:35 UTC (permalink / raw)
  To: Lv, Hui; +Cc: xen-devel

On Mon, May 30, 2011 at 11:25:10AM +0800, Lv, Hui wrote:
> Hi, list
> 
> I found the oprofile tools were not supported in PVOPS domain0(from jeremy's tree). Someone has made the patch to enable it. When I applied such patch, xenoprofile can work. But the results were not trusted. (I think, part of this reason was due to xen, which changed the NMI handling way, but did not update the xenoprofile part)

Hm, I don't think the Jeremy's tree had support for it.

Mike Petullo wrote a patch for the oprofile support and Xen for 2.6.39 kernels.

If you want to try it out, I would suggest you play with devel/next-2.6.39 in my
git tree git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git.

> So my question is that, does anyone still maintain this tools? Or, do xen guys have a plan to enable more powerful profiling tools like 'perf' in native Linux or something else.

Would love to, except that I don't think anybody has signed up for that yet.

> One of my basic feeling about xen's performance profiling tools is that the functional level profiling tools like oprofile were neglected recently.
> 
> Best regards!
> Lv,hui
> 

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

^ permalink raw reply	[flat|nested] 7+ messages in thread

* RE: About the oprofile tools and related profiling tools
  2011-05-31 14:35 ` Konrad Rzeszutek Wilk
@ 2011-05-31 15:42   ` Lv, Hui
  2011-05-31 16:02     ` Konrad Rzeszutek Wilk
  0 siblings, 1 reply; 7+ messages in thread
From: Lv, Hui @ 2011-05-31 15:42 UTC (permalink / raw)
  To: Konrad Rzeszutek Wilk, dunlapg, George Dunlap; +Cc: xen-devel

Thanks Konrad and George!

I have tried Mike Petullo's patched pvops domain0 and xenoprofile on it can work. 

But I did not get the right sample results of HVM guest (No samples obtained). So I guess, something should be wrong from hypervisor part in profiling HVM guest.

So did anyone successfully profile the HVM based guest using this patch?

Anyway, this patch is very useful. Thanks Mike Petullo's work. :) 

Best regards,

Lv, Hui

-----Original Message-----
From: Konrad Rzeszutek Wilk [mailto:konrad.wilk@oracle.com] 
Sent: Tuesday, May 31, 2011 10:36 PM
To: Lv, Hui
Cc: xen-devel@lists.xensource.com
Subject: Re: [Xen-devel] About the oprofile tools and related profiling tools

On Mon, May 30, 2011 at 11:25:10AM +0800, Lv, Hui wrote:
> Hi, list
> 
> I found the oprofile tools were not supported in PVOPS domain0(from jeremy's tree). Someone has made the patch to enable it. When I applied such patch, xenoprofile can work. But the results were not trusted. (I think, part of this reason was due to xen, which changed the NMI handling way, but did not update the xenoprofile part)

Hm, I don't think the Jeremy's tree had support for it.

Mike Petullo wrote a patch for the oprofile support and Xen for 2.6.39 kernels.

If you want to try it out, I would suggest you play with devel/next-2.6.39 in my
git tree git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git.

> So my question is that, does anyone still maintain this tools? Or, do xen guys have a plan to enable more powerful profiling tools like 'perf' in native Linux or something else.

Would love to, except that I don't think anybody has signed up for that yet.

> One of my basic feeling about xen's performance profiling tools is that the functional level profiling tools like oprofile were neglected recently.
> 
> Best regards!
> Lv,hui
> 

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

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: About the oprofile tools and related profiling tools
  2011-05-31 15:42   ` Lv, Hui
@ 2011-05-31 16:02     ` Konrad Rzeszutek Wilk
  2011-05-31 16:21       ` W. Michael Petullo
  0 siblings, 1 reply; 7+ messages in thread
From: Konrad Rzeszutek Wilk @ 2011-05-31 16:02 UTC (permalink / raw)
  To: Lv, Hui; +Cc: George Dunlap, dunlapg, xen-devel

On Tue, May 31, 2011 at 11:42:38PM +0800, Lv, Hui wrote:
> Thanks Konrad and George!
> 
> I have tried Mike Petullo's patched pvops domain0 and xenoprofile on it can work. 
> 
> But I did not get the right sample results of HVM guest (No samples obtained). So I guess, something should be wrong from hypervisor part in profiling HVM guest.
> 
> So did anyone successfully profile the HVM based guest using this patch?

You might want to email Mike to ask that.
> 
> Anyway, this patch is very useful. Thanks Mike Petullo's work. :) 

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: About the oprofile tools and related profiling tools
  2011-05-31 16:02     ` Konrad Rzeszutek Wilk
@ 2011-05-31 16:21       ` W. Michael Petullo
  2011-06-01  0:16         ` Lv, Hui
  0 siblings, 1 reply; 7+ messages in thread
From: W. Michael Petullo @ 2011-05-31 16:21 UTC (permalink / raw)
  To: Konrad Rzeszutek Wilk; +Cc: George Dunlap, dunlapg, xen-devel, Lv, Hui

>> Thanks Konrad and George!
>> 
>> I have tried Mike Petullo's patched pvops domain0 and xenoprofile on it can work. 
>> 
>> But I did not get the right sample results of HVM guest (No samples
>> obtained). So I guess, something should be wrong from hypervisor part
>> in profiling HVM guest.
>> 
>> So did anyone successfully profile the HVM based guest using this patch?
 
> You might want to email Mike to ask that.

I have only used my variation of the 2.6.39 OProfile patch to passively
profile a PV kernel. Note that most of the work was done previously;
all I did was port it to Linux 2.6.39. I have not tried it on a HVM
kernel. My latest 2.6.39 patch is at:

http://www.flyn.org/patches/linux-xen-passive-oprofile/linux-2.6.39.rc5.git5-xen-passive-oprofile.patch.gz

-- 
Mike

:wq

^ permalink raw reply	[flat|nested] 7+ messages in thread

* RE: About the oprofile tools and related profiling tools
  2011-05-31 16:21       ` W. Michael Petullo
@ 2011-06-01  0:16         ` Lv, Hui
  0 siblings, 0 replies; 7+ messages in thread
From: Lv, Hui @ 2011-06-01  0:16 UTC (permalink / raw)
  To: W. Michael Petullo, Konrad Rzeszutek Wilk
  Cc: George Dunlap, dunlapg, xen-devel

Thanks Mike. I can try you latest patch and see what happened to HVM guest:)

-----Original Message-----
From: W. Michael Petullo [mailto:mike@flyn.org] 
Sent: Wednesday, June 01, 2011 12:21 AM
To: Konrad Rzeszutek Wilk
Cc: Lv, Hui; George Dunlap; dunlapg@gmail.com; xen-devel@lists.xensource.com
Subject: Re: [Xen-devel] About the oprofile tools and related profiling tools

>> Thanks Konrad and George!
>> 
>> I have tried Mike Petullo's patched pvops domain0 and xenoprofile on it can work. 
>> 
>> But I did not get the right sample results of HVM guest (No samples
>> obtained). So I guess, something should be wrong from hypervisor part
>> in profiling HVM guest.
>> 
>> So did anyone successfully profile the HVM based guest using this patch?
 
> You might want to email Mike to ask that.

I have only used my variation of the 2.6.39 OProfile patch to passively
profile a PV kernel. Note that most of the work was done previously;
all I did was port it to Linux 2.6.39. I have not tried it on a HVM
kernel. My latest 2.6.39 patch is at:

http://www.flyn.org/patches/linux-xen-passive-oprofile/linux-2.6.39.rc5.git5-xen-passive-oprofile.patch.gz

-- 
Mike

:wq

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2011-06-01  0:16 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-05-30  3:25 About the oprofile tools and related profiling tools Lv, Hui
2011-05-31  9:44 ` George Dunlap
2011-05-31 14:35 ` Konrad Rzeszutek Wilk
2011-05-31 15:42   ` Lv, Hui
2011-05-31 16:02     ` Konrad Rzeszutek Wilk
2011-05-31 16:21       ` W. Michael Petullo
2011-06-01  0:16         ` Lv, Hui

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.