Performance Counter Addresses

Downing, Evan P edowning3 at gatech.edu
Mon Oct 5 02:16:30 UTC 2015


> What's PANDA's recording performance overhead?


Recording overhead is about 1.8x

Replaying overhead is about 3.6x

This is according to their paper back in 2014. It may have changed since then.

> For that to work, your best bet is probably to use rr to record a Firefox session, running the whole thing under QEMU-KVM with PMU enabled. Then switch the QEMU mode to PANDA, and
> run rr replay in it. But you would need to extend QEMU/TCG to support the retired-conditional-branches counter.

Very interesting idea. I am interested, however, in seeing how these downloaded files and installed plugins possibly affect the overall operating system. Would the rr replay reveal this activity on the guest OS when replayed on top of PANDA?

~Evan

________________________________
From: rocallahan at gmail.com <rocallahan at gmail.com> on behalf of Robert O'Callahan <robert at ocallahan.org>
Sent: Sunday, October 4, 2015 10:05 PM
To: Downing, Evan P
Cc: Kyle Huey; rr-dev at mozilla.org
Subject: Re: Performance Counter Addresses

On Mon, Oct 5, 2015 at 2:52 PM, Downing, Evan P <edowning3 at gatech.edu<mailto:edowning3 at gatech.edu>> wrote:

I'm gathering data such as network traffic, plugins installed, files downloaded using Firefox, etc.


I could just use PANDA, but extracting the information could take a long time.

The performance overhead is large.

What's PANDA's recording performance overhead?


It would be faster if I used rr to extract preliminary information to determine if the data is worth studying more, and then perform an in-depth analysis using PANDA.

Sort of like zooming in on the problem (using PANDA) once I've discovered it (using rr).

For that to work, your best bet is probably to use rr to record a Firefox session, running the whole thing under QEMU-KVM with PMU enabled. Then switch the QEMU mode to PANDA, and run rr replay in it. But you would need to extend QEMU/TCG to support the retired-conditional-branches counter.

Rob
--
lbir ye,ea yer.tnietoehr  rdn rdsme,anea lurpr  edna e hnysnenh hhe uresyf toD
selthor  stor  edna  siewaoeodm  or v sstvr  esBa  kbvted,t rdsme,aoreseoouoto
o l euetiuruewFa  kbn e hnystoivateweh uresyf tulsa rehr  rdm  or rnea lurpr
.a war hsrer holsa rodvted,t  nenh hneireseoouot.tniesiewaoeivatewt sstvr  esn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/rr-dev/attachments/20151005/d83d9a03/attachment.html>


More information about the rr-dev mailing list