LWN: Comments on "2009 Linux Plumbers Conference Call For Topics"

2009 Linux Plumbers Conference Call For Topics

Jitesh_Iyer — Fri, 26 Mar 2010 06:24:06 +0000

Yes, That was a great event on Linux. Thanks for sharing information about the conference. By the way I use to participate in the conference often. Recently I have participated in the Cloudslam 2010 which is the world's largest and 2nd virtual conference. That was a great conference through which I have gathered more information on Cloud computing and its innovations and trends. I have registered and got the details about the conference through http://cloudslam10.com

Control Group Microconference

balbir_singh — Thu, 30 Apr 2009 11:22:49 +0000

Thanks, yes, he should definitely be included. I don't see control groups being accepted as a mini-conference topic at plumbers so far. Just FYI

Balbir

Control Group Microconference

legoater — Thu, 30 Apr 2009 09:50:24 +0000

you might consider inviting the liblxc maintainer: Daniel Lezcano.

Constraint-Based Power Management Microconference

pwsan — Sat, 18 Apr 2009 01:11:17 +0000

Would suggest Magnus Damm (SH-Mobile), Ben Dooks (S3Cxxx), Kevin Hilman and Paul Walmsley (OMAP).

- Paul

Security Microconf

pragmatine — Thu, 16 Apr 2009 12:47:11 +0000

Its interesting to see the focus on 'Securing the desktop', but none of the current access control LSMs in the kernel are really targeted to this - they focus on static, server oriented type policies, whereas the desktop is a much more dynamic environment, and providing a least authority implementation for desktop apps using these existing MAC frameworks seems far too cumbersome - and they do not provide the kind of flexibility that is really needed - perhaps a focus on more desktop oriented frameworks such as PULSE[1, 2] (yes I am pushing my own previous research work) would be good.

[1] http://crpit.com/abstracts/CRPITV81Murray.html
[2] http://sourceforge.net/projects/pulse-lsm/

topic suggestion: improving debugging tools

bauermann — Fri, 27 Mar 2009 21:22:36 +0000

Forgot to mention the expected deliverables:

- a roadmap for 2009 which is more aligned to developer needs
- Python API improvements based on ideas and needs of developers
- general usability improvements from feedback in the miniconf

Sorry if I don't have anything more specific. Like I said, the way I see it, the miniconf would be a "developers meets users" kind of thing.

Also it would be nice to have some experts who work on other debug-related tools, or even GCC debuginfo, like Roland McGrath, Alexandre Oliva and Michael Matz.

topic suggestion: improving debugging tools

bauermann — Fri, 27 Mar 2009 19:28:07 +0000

GDB development was slow for some time, but has been catching up in the last few years (Tom Tromey says GDB is going through a Renaissance). The GDB 7.0 release will ship some exciting features (like Python scripting support, pretty-printing, non-stop multithread debugging and others). I believe we now have the manpower to modernize GDB even more, in order to help Linux developers.

Hence, we could have a miniconf to bring GDB developers and users together to brainstorm on what we should work next: what are GDB's biggest weaknesses, and what would be killer features to have, for instance.

The people to call are a subset of (or even the whole of) the current active GDB developers and maintainers: Daniel Jacobowitz, Joel Brobecker, Pedro Alves, Tom Tromey, Hui Zhu, Thiago Bauermann (yours truly), Doug Evans, Pierre Muller (sorry if I forgot someone).

2009 Linux Plumbers Conference Call For Topics

mbr — Mon, 02 Mar 2009 10:00:11 +0000

Embedded/Mobile Linux.

There are more and more Linux powered devices nowadays. However, many device vendors use Linux like a proprietary OS. They choose a kernel, heavily patch it to work on their device and then add even more heavily patched user space. Probably having a miniconf at Linux plumbers will draw more attention to embedded folks problems and issues.

Control Group Microconference

balbir_singh — Mon, 23 Feb 2009 20:02:54 +0000

Thanks! The people to bring together would be Dhaval Giani, Balbir Singh, Paul Menage, Kamezawa Hiroyuki, Li Zefan, Dan Smith, Kosaki Motohiro, Vivek Goyal, Bharata B Rao, Peter Zijlstra, Serge Hallyn, Srivatsa Vaddagiri, Matt Helsley, Jiri Slaby, Pavel Emelianov. There are other people actively contributing to aspects of cgroups as well, but the names mentioned would be a good starting place for the discussions on this topic.

Others, please feel free to add your name if you felt I left your name out of this topic, especially the IO controller folks.

Control Group Microconference

PaulMcKenney — Mon, 23 Feb 2009 19:36:48 +0000

Very interesting topic!

Who would be the right people to bring together to make forward progress on this topic?

Audio Microconference

PaulMcKenney — Mon, 23 Feb 2009 19:18:46 +0000

The only reasonable response is "we hear you!". ;-)

That said, what patches would you be expecting from such a microconf, and what people should be brought together in oder to generate them?

2009 Linux Plumbers Conference Call For Topics

balbir_singh — Thu, 19 Feb 2009 13:28:37 +0000

CGroups (Control Groups) is a new feature that is being enabled in several distributions, including but not limited to openSUSE, Ubuntu and Fedora. Resource management is being considered as a Fedora 11 Feature (see https://fedoraproject.org/wiki/Features/ResourceMgt). A lot of work has been done in the kernel in the form of the common infrastructure and addition of controllers (memory, CPU, network). libcgroup is being developed to deal with the challenges in user space of controlling and managing control groups. We propose a conference topic for cgroups to discuss the future of control groups, the development of newer controllers like the IO controller and the impact of control groups on various subsystems and how to address the user space challenge of configuration, unification of diverse APIs presented by the various kernel resource controllers. CGroups will being to become an important aspect of system management tools, since they provide both measurement and control of kernel resources. At the conference we would like to explore the user level and kernel design aspects to help better integration of the new feature for common desktop users and large data center operators. The topic also provides an opportunity for libvirt and liblxc to present their user space implementation of cgroups and to discuss the potential to merge all the three approaches.

Security Microconf

jamesmrh — Thu, 19 Feb 2009 00:58:31 +0000

I've also had some interest expressed via private email -- it seems we'd easily fill a day for a security microconf. I'd be happy to volunteer as the runner.

2009 Linux Plumbers Conference Call For Topics

mezcalero — Wed, 18 Feb 2009 23:43:45 +0000

We want an audio microconf again!

boot/init revisited

PaulMcKenney — Wed, 18 Feb 2009 22:35:41 +0000

I bet you are correct, and besides, greedy people like myself would ask "5 seconds is nice, but why not faster?". At least I am no longer a teenager, so I won't be asking for anything faster than a few hundred milliseconds. ;-)

Arjan van de Ven, Aoke Kok, and Keith Packard appear in the LWN article reporting on last year's session. Any thoughts on who else would help push this forward?

Configuration and Storage of Configuration Data

PaulMcKenney — Wed, 18 Feb 2009 01:32:56 +0000

You could go almost anywhere with this one! :-)

Making configuration more flexible and usable is certainly hard to argue with, and a HIG even for Linux kernel configuration parameters would be useful to me personally (and perhaps even more useful to the people who must live with the kernel configuration parameters that I create).

What scope were you thinking in terms of? Who would be the interested parties?

boot/init revisited

davej — Wed, 18 Feb 2009 00:14:04 +0000

I'll bet by the time plumbers comes around, every distro will still not come anywhere near the famous 5 second boot-up. Though everyone will surely be doing better than they were a year ago.

What stumbling blocks will remain? how can we work better together to make these changes happen?

Last years event saw a lot of cross-distro collaboration during the mini-hackfest that followed the boot/init miniconf. I'm sure there are more wins to share here.

Security Microconf

pcmoore — Tue, 17 Feb 2009 23:18:16 +0000

Let's not forget virtualization security, there are a few things going on in the libvirt space (ACLs, sVirt) that look interesting. I also think there is work to be done to help make the labeled networking bits in the kernel more appealing for non-hardcore users (some friendly tools?), but that may not be a very popular topic (although it remains dear to me <g>).

Regardless, I agree that 2.5 hours would likely be nowhere near long enough for all the topics that would likely fall into a "Security Microconf".

Making User-Space I/O Not Suck

aliguori — Tue, 17 Feb 2009 21:07:25 +0000

I must confess that my first reaction to the UNIX interfaces back in the early 1980s was not all that positive, so I can relate! Of course, these interfaces aren't all that much younger than I am, and much has changed during the intervening years, virtualization being but one such change! My guess is that you are primarily interested in the interaction of userspace I/O and virtualization, though I suspect that non-virtualized environments must be taken into account as well.

There are two important issues for virtualization. The first is performance. Since virtualizations acts as a layer between hardware and traditional OSes, there is a great deal of pressure in the virtualization layer to perform as well as possible.

For block devices, the ideal situation would be to be able to submit multiple asynchronous IO requests from the single context, and receive a notification through a file descriptor upon completion. Since we lack a generic mechanism to submit asynchronous requests, that's the first stumbling block. Zach's acall interface and Ingo's syslets both try to address this.

You also have an issue of functionality. The interface that we have to a virtual guest is a bit richer than the existing IO interfaces. When emulating a SCSI adapter, for instance, we receive requests with Simple, Ordered, or Head tags which control the completion order of each request when submitted asynchronously. There is no way, today, to submit barrier requests from userspace (let alone this richer barriers).

Networking is much the same. Today, the interface we have is tun/tap for sending raw network IO. We very much want to improve the performance of network IO by having zero-copy TX/RX. For TX, we can do this reasonably with some of the patches Rusty has posted previously. RX is much harder because in general 1) we don't have an interface for publishing RX buffers to the kernel 2) you need an interface for the network drivers to have them receive to a preallocated RX buffer. There is some discussion around making use of VMDq or SR-IOV to enable this.

You mention Zach and Rusty. Who else would need to be involved to make good forward progress? What forward progress (patches!) would it be reasonable to expect from sessions at the Linux Plumbers Conference, and what prework would be required in order to make such progress possible?

This is an ongoing activity involving a lot of people. I think the progress that could made be depends on who would attend to be honest. So far, all of the usual suspects have been involved in these discussions but a lot of things have been stalled by lack of consensus of a common direction (acall vs. syslets vs. linux-aio is a good example of this).

Making User-Space I/O Not Suck

PaulMcKenney — Tue, 17 Feb 2009 18:12:47 +0000

Constraint-Based Power Management Microconference

PaulMcKenney — Tue, 17 Feb 2009 17:17:26 +0000

Power management certainly continues to be of great interest to many people, so great suggestion! And constraint-based power management APIs have received recent attention as well.

Who would be the right people to bring together to make real implementation progress in this area?

Of course, I would not want to discourage any progress towards a constraint-based power-management implementation before LPC!

2009 Linux Plumbers Conference Call For Topics

Burgundavia — Mon, 16 Feb 2009 08:04:31 +0000

Talk to the crew that organizes the Ubuntu conferences. They have plenty of experience of what works and what doesn't when you need to capture audio in a noisy environment.

2009 Linux Plumbers Conference Call For Topics

russell — Sun, 15 Feb 2009 06:02:22 +0000

How about configuration, and the storage of configuration data. Also, what should be in configuration data. Some people want lots of options and flexibility. Others want usability and GUI interfaces.

Can we have both? A usability layer on top of easy to find, parse and edit text files?

What about documentation?

Could plumbers develop something like the GNOME HIG but for configuration?

Security Microconf

niv — Sat, 14 Feb 2009 17:48:37 +0000

If 2.5 hours are not enough - we can probably accommodate somehow (longer sessions? multiple?). Don't let that stop anyone from proposing ideas - our format isn't exactly carved in stone. If you have an suggestion or an estimate for how much time the session might take, feel free to provide that info too.

Security Microconf

jamesmrh — Fri, 13 Feb 2009 23:55:19 +0000

I think it'd be useful to hold a miniconf on the topic of Linux security integration.

We have a rapidly growing set of security features in Linux, including: Access control (SELinux, SMACK, Tomoyo etc.), Integrity, Crypto, FS Capabilities, Desktop (XACE), Storage (LNFS) and likely soon Anti-malware.

Developers in these areas, along with general developers (e.g. distro integration folk) could get together to discuss issues relating to how these components can be integrated and leveraged. Possible topics might include:

- Utilizing TPM with disk encryption
- Encouraging ISV adoption of security features (v. hard problem)
- Improving the end-user experience
- Best practices for integrating FS caps (etc.) into distros
- Securing the desktop
- WIP updates for major projects in development (e.g. anti-malware).

I wonder if 2.5 hours would be enough, though.

2009 Linux Plumbers Conference Call For Topics

aliguori — Fri, 13 Feb 2009 20:17:35 +0000

How about "Making userspace I/O not suck" :-)

Zach Brown has proposed acalls as a replacement for the problematic linux-aio subsystem. KVM would be an obvious consumer of such an interface although it's still not sufficient for our needs.

Rusty Russell has proposed vringfd as a mechanism for improving userspace raw networking performance. Unfortunately, this does not seem to have much of a future so brainstorming is required for a better interface.

In general, the interfaces we've inherited from Unix leaves a great deal to be desired.

2009 Linux Plumbers Conference Call For Topics

niv — Fri, 13 Feb 2009 20:01:57 +0000

hverkuil, thanks for the update - that bit about "As a result of last year's plumbers conference a new API was designed for DVB tuners..." was good to know!

We'd like to track how effective we were in helping to solve issues in Linux and moving code fwd (so which topics were useful to talk about and which possibly were less so.. :)). One of the key goals of the LPC is to get developers together to address thorny problems, resolve issues, design solutions and get new stuff into Linux.

2009 Linux Plumbers Conference Call For Topics

hverkuil — Fri, 13 Feb 2009 19:36:44 +0000

In 2008 I discussed as part of the V4L-DVB microconf extending V4L2 to
support complex media devices. But most of the work until today has been
an effort to lay the foundation of that in the form of an improved V4L2
framework. Hopefully by the time of the conference we have managed to
port all V4L2 drivers over to that framework and we can by then start to
build on it.

This topic would involve explaining the new framework, the problems
encountered while converting drivers to it, and future ideas on how to
improve the framework in order to simplify the V4L2 drivers. In addition,
there is a lot of work being done on merging the Texas Instruments omap
drivers, and hopefully in the near future the davinci drivers, also from
TI. It would be an interesting sub-topic in its own right.

As a result of last year's plumbers conference a new API was designed for
DVB tuners. It would be good to see the end result of that as well.

The complexity of implementing support for hybrid tuners (analog and
digital) is another topic I would like to see.

2009 Linux Plumbers Conference Call For Topics

niv — Fri, 13 Feb 2009 19:29:36 +0000

More specifically, I should have said - where there were author slides, we'll put them up.

For a lot of the discussion that did not involve a prepared presentation, we had hoped to provide audio/video recording, but that wasn't successful, unfortunately.

Speaking as someone organizing the thing this year, we hope to have better recordings and better capture the output of those sessions, including slides.

2009 Linux Plumbers Conference Call For Topics

niv — Fri, 13 Feb 2009 19:26:44 +0000

We will definitely be improving the website, IT support, and that kind of stuff. Thanks for the input! If you have any other suggestions, let us know...

2009 Linux Plumbers Conference Call For Topics

biehl — Fri, 13 Feb 2009 18:58:07 +0000

Not a specific topic, but maybe the focus on uploading presentations could be improved? The effort seemed lackluster last time:

http://wiki.linuxplumbersconf.org/presentation_slides

2009 Linux Plumbers Conference Call For Topics

mgross — Fri, 13 Feb 2009 17:30:41 +0000

Constraint based power management micro conf topic.

I would like a micro conference to focus on ideas, issues, methods and problems around implementing power management based on constraining the power states the system is allowed to enter. Examples of constraint based PM include PM_QOS, clock framework, and Android's wake_lock PM design. The goal is to make progress at getting more useful constraint based PM solutions into upstream.