[prev in list] [next in list] [prev in thread] [next in thread] 

List:       xen-users
Subject:    Re: [Xen-users] perf hardware event on ARM64
From:       Jintack Lim <jintack () cs ! columbia ! edu>
Date:       2015-10-25 22:51:45
Message-ID: CAHyh4xhVFueDazEkQrpOUdmXo=VjWWm+zWU4p+68S74M=++C2w () mail ! gmail ! com
[Download RAW message or body]

[Attachment #2 (multipart/alternative)]


On Fri, Oct 23, 2015 at 9:44 AM, Ian Campbell <ian.campbell@citrix.com>
wrote:

> On Fri, 2015-10-23 at 09:22 -0400, Jintack Lim wrote:
> > Hi,
> >
> > I'm working on ARM64 (AppliedMicro=E2=84=A2 X-Gene).
> > I wanted to run perf to collect some hardware events, but I got some
> > trouble.
> > Basically I have two questions.
> >
> > 1.
> > I found that pmu is not registered in the device tree,
> > while it is registered on native linux (installed on the same type of
> > machine) .
> > Can I just modify device tree at booting time to make it look like nati=
ve
> > linux?
> > Will it work?
>
> > Additional information is at the bottom.
> >
> > 2.
> > I found that running perf on Xen 4.5.0 resulted in segmentation fault (=
or
> > system crash),
> > however it runs well on the latest Xen on git server.
> > Can somebody point the patch that makes this change?
>
> 4.5.0 was vulnerable to XSA-93:
>
> http://xenbits.xen.org/xsa/advisory-93.html
>
> The core issue with PMU is that Xen currently does not know how to contex=
t
> switch the relevant state when changing guests. Fixing that would be more
> work than just fiddling with device tree.
>
> You might be able to bodge something useful (but unsupported) together by
> just allowing only dom0 to access these registers and to see the values
> collected from the whole system (i.e. don't worry about context switching=
).
> How useful such numbers would be depends somewhat on what you are hoping =
to
> find out from this exercise.
>

My machine has 8 cores,
and I'm running dom0 on core 0~3, and domU on 4~7.
I wonder if running perf on dom0 can profile hardware events on cores, on
which dom0 is not running.
If not, then I should run perf on domU I guess.


>
> Ian.
>
>
>

[Attachment #5 (text/html)]

<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Oct \
23, 2015 at 9:44 AM, Ian Campbell <span dir="ltr">&lt;<a \
href="mailto:ian.campbell@citrix.com" target="_blank" \
onclick="window.open(&#39;https://mail.google.com/mail/?view=cm&amp;tf=1&amp;to=ian.ca \
mpbell@citrix.com&amp;cc=&amp;bcc=&amp;su=&amp;body=&#39;,&#39;_blank&#39;);return \
false;">ian.campbell@citrix.com</a>&gt;</span> wrote:<br><blockquote \
class="gmail_quote" style="margin:0px 0px 0px \
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><span \
class="">On Fri, 2015-10-23 at 09:22 -0400, Jintack Lim wrote:<br> &gt; Hi,<br>
&gt;<br>
&gt; I&#39;m working on ARM64 (AppliedMicro™ X-Gene).<br>
&gt; I wanted to run perf to collect some hardware events, but I got some<br>
&gt; trouble.<br>
&gt; Basically I have two questions.<br>
&gt;<br>
&gt; 1.<br>
&gt; I found that pmu is not registered in the device tree,<br>
&gt; while it is registered on native linux (installed on the same type of<br>
&gt; machine) .<br>
&gt; Can I just modify device tree at booting time to make it look like native<br>
&gt; linux?<br>
&gt; Will it work?<br>
<br>
&gt; Additional information is at the bottom.<br>
&gt;<br>
&gt; 2.<br>
&gt; I found that running perf on Xen 4.5.0 resulted in segmentation fault (or<br>
&gt; system crash),<br>
&gt; however it runs well on the latest Xen on git server.<br>
&gt; Can somebody point the patch that makes this change?<br>
<br>
</span>4.5.0 was vulnerable to XSA-93:<br>
<br>
<a href="http://xenbits.xen.org/xsa/advisory-93.html" rel="noreferrer" \
target="_blank">http://xenbits.xen.org/xsa/advisory-93.html</a><br> <br>
The core issue with PMU is that Xen currently does not know how to context<br>
switch the relevant state when changing guests. Fixing that would be more<br>
work than just fiddling with device tree.<br>
<br>
You might be able to bodge something useful (but unsupported) together by<br>
just allowing only dom0 to access these registers and to see the values<br>
collected from the whole system (i.e. don&#39;t worry about context switching).<br>
How useful such numbers would be depends somewhat on what you are hoping to<br>
find out from this exercise.<br></blockquote><div><br></div><div>My machine has 8 \
cores,</div><div>and I&#39;m running dom0 on core 0~3, and domU on 4~7.</div><div>I \
wonder if running perf on dom0 can profile hardware events on cores, on which dom0 is \
not running.</div><div>If not, then I should run perf on domU I guess.</div><div>  \
</div><blockquote class="gmail_quote" style="margin:0px 0px 0px \
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
 <span class=""><font color="#888888"><br>
Ian.<br>
<br>
<br>
</font></span></blockquote></div><br></div></div>



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

[prev in list] [next in list] [prev in thread] [next in thread]