From: compudj@krystal.dyndns.org (Mathieu Desnoyers)
Subject: [ltt-dev] TODO list before releasing LTTng-buffering to LKML
Date: Wed, 8 Oct 2008 22:21:01 -0400 [thread overview]
Message-ID: <20081009022101.GA16931@Krystal> (raw)
In-Reply-To: <20081008010211.GA26361@Krystal>
* Mathieu Desnoyers (compudj at krystal.dyndns.org) wrote:
> Hi!
>
> Just to keep you posted on the current development, this is what is left
> on my roadmap before I post a minimal LTTng (only the timestamping,
> event ID management and buffering) to LKML :
>
> Todo list before posting LTTng-buffering to LKML
>
> - Create a simplified ltt-relay (use interrupt disable instead of
> lockless algo)
I implemented this one yesterday night. Basically, it's the original
lockless scheme, except that it uses only standard variables (no atomic
or local atomic ops) and it also uses a per-buffer spinlock (because
readers can be on different cpus) and interrupt disable. Here are the
performance results : they are not that far from the lockless scheme. It
adds about 1% performance hit on tbench on the x86_64 over the lockless
scheme. See lttng-transport-locked.patch in LTTng 0.36 for details.
Performance tests, tbench, flight recorder trace, default instrumentation
Dual quad-core x86_64, 2.0GHz
Instrumentation dynamically disabled : 1883.09 MB/s
Markers connected : 1812.14 MB/s
Lockless scheme, flight recorder : 925.68 MB/s
Per-cpu-buffer spinlock and interrupt disable, flight recorder : 871.78 MB/s
Global spinlock and interrupt disable, flight recorder : 153.74 MB/s
Single Core Pentium 4, noreplace-smp, 3.0GHz
Instrumentation dynamically disabled : 146.85 MB/s
Markers connected : 144.32 MB/s
Lockless scheme, flight recorder : 89.59 MB/s
Per-cpu-buffer spinlock and interrupt disable, flight recorder : 86.05 MB/s
Mathieu
> - merge a simplified lttv in lttng
>
>
> Todo list after posting first LTTng-buffering release to LKML :
>
> - assign marker IDs per channel
> - switch ltt-control.ko (currently over netlink) to debugfs
> - switch marker list (and marker activation (currently in /proc/ltt) to debugfs
> Marker directory. One file per marker.
> - Create a ltt-ascii.ko kernel module which merge-sorts the buffers and exports
> them to userspace through a debugfs file.
>
> If some of you are interested to work on any of these, you are welcome.
> I'll be glad to post more details about these items upon request.
>
> Mathieu
>
> --
> Mathieu Desnoyers
> OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68
>
> _______________________________________________
> ltt-dev mailing list
> ltt-dev at lists.casi.polymtl.ca
> http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
>
--
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68
next prev parent reply other threads:[~2008-10-09 2:21 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-10-08 1:02 Mathieu Desnoyers
2008-10-09 2:21 ` Mathieu Desnoyers [this message]
2008-10-24 20:38 ` Mathieu Desnoyers
2008-10-27 3:10 ` Zhaolei
2008-10-14 8:50 ` Lai Jiangshan
2008-10-14 9:17 ` Andrew McDermott
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20081009022101.GA16931@Krystal \
--to=compudj@krystal.dyndns.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox