Linux wireless GSoC 2010 project ideas
in [Kernel]
|
Prev: perf,hw_breakpoint,kgdb: No mutex taken forkerneldebugger
Next: ARM: Add sh7372 and AP4EVB support
From: Pat Erley on 1 Feb 2010 23:40 On 02/01/10 17:12, Nick Bowler wrote: > On 22:14 Mon 01 Feb , Maxim Levitsky wrote: >> 2 - Allow a wireless card to associate to several APs at once (if they >> share the frequency). >> Implement this at least for ath5/9k. Madwifi had that feature. Maybe >> ath9k has. If it has that feature, test it. > > I haven't actually tried this particular case, but this is ostensibly > already supported by mac80211: iw allows you to create several wireless > interfaces that share a phy, which can then be configured into different > modes (I've used station + monitor successfully in the past). > With ath9k I have used: AP + MP AP + STA AP + AP (Open and WPA2) using both the debug interface to ath9k's internal multi-vif support (never tested MP with ath9k multi-vif) and mac80211. I know that if you use the ath9k internal method, you can operate on multiple channels at a sacrifice to throughput. I have not tested if mac80211 supports multiple channels. It's as easy as: AP+MP and AP+STA: Start Hostapd Create interface with iw use second interface. AP+AP Configure hostapd for multiple bss start hostapd Pat Erley -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Holger Schurig on 3 Feb 2010 04:30 > Also, like I said, this should be a analysis, and not > wardriving/hacking tool. > And with GUI. Please define "analysis". What exactly should be analyzed? Kismet can do: * construct list of all beaconing APs and IBSS devices * associate SSID to beacons because of assoc/auth req/resp * construct a list of all stations around * show used protection schema * get simple packet statistics * get simple signal level figures * ... But "analysis" can obviously be much more. Some of the analysis should also be done on driver level, e.g. extend the nl80211-site-survey into mac80211 and add noise reporting to some drivers (I have something preliminary for ath5k). But also extend it to give channel occupation statistics. And for a GUI, you could run kismet-server in the background and write a GUI (e.g. in Qt, so that it runs on X11/Mac/Win). If the required analysis isn't easily put into kismet, you can of course write a new application if you please so :-) I for example wrote a simple Qt application that has 25% of kismet's capability, but with a GUI. Unfortunately still Qt3. -- http://www.holgerschurig.de -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Hin-Tak Leung on 4 Feb 2010 00:00 On 1/28/10, Till Kamppeter <till.kamppeter(a)gmail.com> wrote: > Great to hear from you. > > I will apply again for the LF entering as a mentoring organization and I > will add your projects to the list of proposed projects for the LF > application. Perhaps at this point, it might be a good idea to share some insights for reasons of failures/non-competition. I think the first year (2008) of LF's participation, it was 6/8 competion over all? How did 2009 do? Luis mentioned that 2/3 failed out of the 3 wireless-related project, but AFAIK there were about 10 accepted in 2009 under the LF umbrella - what's the completion rate for the other 7? It is probably a good idea to write some of that up (non-competition and the possibility of re-do) in one of Linux Foundation's wiki pages. AFAIK the overall non-competition rate across all organization is about 10% so the 2/8 in 2008 wasn't too bad. > P. S.: Do you know someone with knowledge in data compression? I will > run a project at OpenPrinting which is about data compression. What are you thinking of? :-) Hin-Tak -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Luis R. Rodriguez on 22 Feb 2010 16:30 On Thu, Jan 28, 2010 at 1:38 PM, Luis R. Rodriguez <mcgrof(a)gmail.com> wrote: > Google has confirmed it will have a Google Summer of Code for 2010 > [1]. Last year we had a few projects suggested (4) and accepted (3) > under the Linux Foundation sponsoring organization umbrella [2]. > Unfortunately out of the three projects that were approved only one > completed successfully, that of the adding AP support to Network > Manger. I haven't seen specific updates to the progress of that but I > do know some patches were indeed submitted to help with this effort. > Perhaps the student can elaborate more. > > The other projects that did not pass are up as suggestion for this > year again, but am hoping there are more. If you do have a project > idea please just go ahead and add your idea to the list of possible > projects [3]; you don't have to fill out a full page for it for now > but the more details you can add the better. If the Linux Foundation > does give us a few slots I recommend we be a little more strict about > acceptance criteria since our failure rate was pretty high (2/3) and > it would be better to see other projects get accepted if we do not > have the confidence our projects will be completed. One possibility to > help with the success rate of our projects might be to narrow the > scope down a little more. I think the testing and GeoClue project > might have been a little too ambitious and although we did have pretty > excited students we saw no progress at all. > > If you have ideas for projects just feel free to add to the wiki. We > should strive to get all project ideas finalized by the middle of > February, latest the end of February. Hopefully towards the end of > February we can see who would be willing to mentor each project. > Google plans on starting to accept would-be-mentor organization > applications on March 8th so we'll need our ideas finalized well > before that so we can send them as suggestions to the Linux Foundation > to see if we can get a few good project candidates accepted. > > [1] http://groups.google.com/group/google-summer-of-code-discuss/browse_thread/thread/d839c0b02ac15b3f > [2] http://wireless.kernel.org/en/developers/GSoC/2009 > [3] http://wireless.kernel.org/en/developers/GSoC/2010 Its the end of February, so we need to finalize the GSoC projects. I'm only up to mentor the automation of testing, with a reduced scope of using just having something register to nl80211 events and have a set of tests cases run. Orbit testing and integration is completely optional. If anyone cares to mentor any of the other projects please let me know as otherwise we just have one project on the 802.11 front. Luis -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Frederic Weisbecker on 9 Mar 2010 17:50
On Fri, Jan 29, 2010 at 12:51:59AM +0100, Till Kamppeter wrote: > No problem for students to suggest topics. We are not restricted to > wireless in terms of kernel. We support kernel projects in general. > > Please give your suggestion as answer to this e-mail, doing "Reply to > all", as I am not a kernel export, the other participants of this thread > should have a look at the student's project ideas and help finding > mentors. > > Till Ok, I hope it's not too late. Here is one subject that I would like to apply for: = Improving tracing in perf events / ftrace = Using trace events under perf events is still a young feature and needs various improvements. - Syscall events can carry only raw values and addresses. We need to make them "aware" of strings and structures contents from userspace. - Make function / function graph tracers usable by perf - Optimize tracing fast-path - Improve perf tools to better handle trace events scalability - Provide new perf tools that exploit trace events (scheduler migration analysis, etc...) - Implement a per context excluding (eg: exclude irqs, exclude functions, etc...) - There are always things to do there. Of course, each of these elements require a lot of work, so not all of them can be done in two months, and some of them can be started already before the summer. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ |