**** BEGIN LOGGING AT Mon Jul 14 02:59:59 2014
Jul 14 03:43:10 <Abhishek__> testing Socket.IO links
Jul 14 03:59:16 <jkridner|m> hi all
Jul 14 03:59:38 <Abhishek__> good evening jkridner
Jul 14 03:59:59 <jkridner|m> 1:30 is just too freakin late
Jul 14 04:00:12 <jkridner|m> I will try to hang around
Jul 14 04:16:35 <ds2> you need a better timezone ;)
Jul 14 05:21:32 <tridge> jkridner: hi. My network link is a bit dodgy, so may not be here for whole meeting
Jul 14 05:21:46 <tridge> (I'm on a bus between canberra and sydney at the moment)
Jul 14 05:23:01 <tridge> jkridner: the big news is the BBB/PXF controlled its first vehicle last week, a rover. It didn't drive very well due to the MPU6k issues, but at least its on a vehicle now :)
Jul 14 05:27:33 <jkridner|m> hi tridge
Jul 14 05:28:00 <jkridner|m> thats a start
Jul 14 05:29:54 <vvu> morning!
Jul 14 05:29:54 * vvu is a bit laggy from last night match
Jul 14 05:30:23 <Abhishek_> hi all
Jul 14 05:30:36 <alexanderhiam> ahoy
Jul 14 05:35:46 <vmayoral> Good morning
Jul 14 05:38:00 <tridge> hi victor
Jul 14 05:38:13 <tridge> vmayoral: see my msg about the force push to BeaglePilot/master
Jul 14 05:38:47 <tridge> vmayoral: sorry that was needed - I wanted to get the BBB port into ardupilot/master and needed to rework the commits to break some of them up
Jul 14 05:39:42 <vmayoral> tridge:np. Great having the code merged
Jul 14 05:40:13 <tridge> vmayoral: it also changes SPIDriver a bit, to make it easier to port to new boards
Jul 14 05:40:51 <tridge> vmayoral: have you had a chance to look at the LSM9DSO driver any more?
Jul 14 05:42:09 <vmayoral> tridge: no i did not yet. I was away for some days but i'm now on my way back. As soon as I arrive home i'll test the PWM, wrap up the ROS work and look at the LSM.
Jul 14 05:42:22 <vmayoral> Is there anything elae u want me to do?
Jul 14 05:43:09 <tridge> i think we should re-work the 9250 driver to use the time based wait_for_sample(), add the dual-pole filtering and try sampling at 1KHz
Jul 14 05:43:44 <tridge> that will raise the SPI load a lot (by factor of 5x), but should reduce the time spent in wait_for_sample()
Jul 14 05:45:24 <tridge> to make that work we'll need to change LinuxScheduler::_timer_thread() to not drift in time
Jul 14 05:45:36 <tridge> right now it just does _microsleep(1000) between calls
Jul 14 05:46:32 <tridge> it will need to instead have a "next_timer_tick" and sleep the right number of microseconds for that deadline
Jul 14 05:47:34 <tridge> that will be needed to ensure we don't lose any ticks, and that we average 1kHz
Jul 14 05:48:09 <tridge> later I think we'll need a thread per SPI bus, with an API to ask for regular transfers
Jul 14 05:52:20 <av500> gm
Jul 14 05:53:18 <jkridner|m> quiet. no other updates?
Jul 14 05:53:55 <alexanderhiam> jkridner|m: did you see karki's command mapping?
Jul 14 05:54:43 <jkridner|m> nope. link?
Jul 14 05:54:52 <alexanderhiam> https://gist.github.com/deepakkarki/c2cccf889b91da660bc9
Jul 14 05:58:56 <jkridner|m> looks like a good start
Jul 14 05:59:35 * jkridner|m falls asleep
Jul 14 06:01:39 <Abhishek_> biot tested and uncovered a bug in the sigrok bindings, captures less than 512K bytes in size were affected. Has been fixed now.
Jul 14 06:02:22 <Abhishek_> he tested with a small python script to transmit out SPI data, and then captured and viewed with sigrok-cli on the BBB
Jul 14 06:10:08 <ds2> 6
Jul 14 06:10:41 <Abhishek_> 7
Jul 14 06:53:28 <av500> 8
Jul 14 08:43:51 <vvu> praveendath92: had any luck with the soft lockup ?
Jul 14 08:44:30 <praveendath92> vvu: I tried with the prefetch added but no luck :(
Jul 14 08:44:32 <praveendath92> Same error.
Jul 14 08:45:08 <praveendath92> https://github.com/praveendath92/udlfb/blob/master/udlfb.c#L623
Jul 14 08:45:53 <vvu> do the conversion maybe there is some trick there :)
Jul 14 08:46:06 <vvu> they cast it to some const unsigned long
Jul 14 08:48:15 <praveendath92> I tried w/o compress_line and leaving trim_hline
Jul 14 08:48:19 <praveendath92> same error.
Jul 14 08:48:54 <vvu> is it hard to reverse the compresion ?
Jul 14 08:49:06 <vvu> leaving this as is and doing work on the android side ?
Jul 14 08:49:27 <praveendath92> May be not.
Jul 14 08:49:49 <praveendath92> It's just that having raw data will help.
Jul 14 08:50:01 <praveendath92> Debugging could be difficult.
Jul 14 08:50:26 <vvu> so find where you have the buffer ready and call usb_bulk_transfer from there
Jul 14 08:50:35 <praveendath92> Since I got this error, I should restart :/
Jul 14 08:50:51 <vvu> so everytime you get this error u need restart ?
Jul 14 08:51:25 <praveendath92> Yep.
Jul 14 08:51:35 <vvu> install a VM
Jul 14 08:51:38 <vvu> and use that one
Jul 14 08:51:40 <praveendath92> When I try to unload the module my pc freezes.
Jul 14 08:52:21 <praveendath92> A Linux on VM running on Linux.
Jul 14 08:52:40 <vvu> why not, at least you do not restart physical machine always
Jul 14 08:53:24 <praveendath92> Hmm
Jul 14 08:53:27 <vvu> also praveendath92 there are some more tricks in the compress_hline
Jul 14 08:53:34 <vvu> more prefetch* things
Jul 14 08:55:09 <praveendath92> Hmm.
Jul 14 08:56:06 <praveendath92> He is using rlx compression not rl.
Jul 14 08:56:48 <praveendath92> Never heard of that one, rlx.
Jul 14 08:59:43 <praveendath92> vvu: VM suggestion?
Jul 14 08:59:54 <vvu> virtualbox ?
Jul 14 09:00:01 <vvu> vmware player, pick one
Jul 14 09:00:23 <praveendath92> virtualbox.
Jul 14 09:07:07 <vvu> praveendath92: can you enable the compression functions and comment out the prefetching instructions
Jul 14 09:07:21 <vvu> praveendath92: http://lwn.net/Articles/444336/
Jul 14 09:07:43 <vvu> prefetch just brings data into cache while you are processing something, used for optimizations nothing more
Jul 14 09:07:54 * praveendath92 is wondering he didn't try that.
Jul 14 09:08:34 <praveendath92> And prefetchw to fetch 1st cache line.
Jul 14 09:09:28 <praveendath92> Give me 10 mins. Waiting for the download to finish and then I will reboot to try that.
Jul 14 09:09:33 <vvu> oki
Jul 14 09:09:54 <vvu> and use a vm not to go offline irc
Jul 14 09:11:49 <praveendath92> Yes. yes.
Jul 14 09:12:03 <praveendath92> I should also install headers now.
Jul 14 09:12:16 <praveendath92> In VM.
Jul 14 11:37:43 <karki_> alexanderhiam, panto, jkridner : do we have a hangout today with chris?
Jul 14 11:53:24 <alexanderhiam> karki_: I believe Chris said he could make it this week
Jul 14 11:53:39 <karki_> yeah, I recall now!
Jul 14 11:54:00 <alexanderhiam> though it may be in an hour
Jul 14 12:50:39 * jkridner yawns
Jul 14 12:51:09 <jkridner> I will put a call into everyone and see if it happens.
Jul 14 12:57:35 <jkridner> panto, karki_: hangout started.
Jul 14 12:59:23 <karki_> jkridner : k
Jul 14 13:01:23 <karki_> jkridner : in a few min, my browser is giving problems :/
Jul 14 13:06:20 * karki_ aargh....
Jul 14 13:20:34 <karki_> jkridner : bad internet :/
Jul 14 13:52:09 <karki_> panto : ping
Jul 14 13:52:29 <jkridner> hi karki_
Jul 14 13:52:34 <jkridner> hi panto, alexanderhiam as well.
Jul 14 13:52:36 <karki_> hi
Jul 14 13:53:31 <alexanderhiam> karki_: another way to think of the soft PWM in pruspeak is that it would let you have more than 8 PWMs in any userspace program
Jul 14 13:54:06 <jkridner> karki_: I'm concerned you might not finish the basic tasks.... I like that you are looking forward to make the architecture right for JIT and various hardware peripherals, but I think we need to agree on what are the most basic components to make sure you push this across some sort of finish line that can be built upon by others moving forward (and yourself).
Jul 14 13:54:28 <karki_> I'm not working on the JIT now
Jul 14 13:54:52 <karki_> just want to make sure I'm not working in a way that will make adding a JIT tricky later on
Jul 14 13:55:12 <jkridner> I see a lot of value in the soft PWMs.... but mostly I see it as a way to have a secondary hard real-time loop doing useful things and having one of the PRUs acting as a traffic cop.
Jul 14 13:55:23 <jkridner> it is overall a very useful and extensible architecture...
Jul 14 13:55:40 <jkridner> new messages can be added easily, etc.
Jul 14 13:56:31 * alexanderhiam wonders how well soft PWM would work implemented in the main pruspeak event loop
Jul 14 13:56:49 <alexanderhiam> karki_: have you done any timing tests?
Jul 14 13:57:12 <karki_> not yet
Jul 14 13:57:28 <karki_> but then we'd have a lot of stuff running on one PRU
Jul 14 13:57:32 <jkridner> alexanderhiam: sounds like a difficult architecture to maintain if put into the main event loop...
Jul 14 13:57:33 <karki_> thats 12K
Jul 14 13:57:43 <jkridner> It was my original thought to put such things there....
Jul 14 13:57:56 <jkridner> but, offloading to a dedicated loop on the other PRU seems more general to me.
Jul 14 13:57:56 <karki_> oops 8K of inst space only
Jul 14 13:58:13 <karki_> jkridner : like how panto does in his code?
Jul 14 13:58:23 <jkridner> yes.
Jul 14 13:59:24 <alexanderhiam> jkridner: are you thinking the second PRU would have a more general event loop you could use for other things as well, or just a dedicated PWM driver?
Jul 14 13:59:26 <karki_> I feel soft pwm is easier, it's just that I don't want to redo stuff when we move to JIT which we will (after GSoC)
Jul 14 13:59:38 <jkridner> it allows for optimizations to done to the interpreter such that no instruction takes more than a few cycles to execute as longer tasks that require periodic attention are handed off to the other PRU.
Jul 14 13:59:58 <jkridner> alexanderhiam: I'm thinking it would be more general, but start with a PWM example.
Jul 14 14:00:29 <alexanderhiam> ok, so a custom driver could potential run in the same event loop as the PWM driver
Jul 14 14:00:35 <alexanderhiam> potentially*
Jul 14 14:02:30 * jkridner would like to understand this JIT architecture.
Jul 14 14:03:24 <karki_> wait... how the event loop works is it keeps which pin to toggle next, right? different pins need to be toggled at different times, so the pru keeps track of that in some sort of a structure, and when it is time  it toggles the pin output, right?
Jul 14 14:03:41 <alexanderhiam> right
Jul 14 14:04:09 <karki_> so how can we be doing many other stuff other than PWM here, wont it mess up the time accuracy?
Jul 14 14:04:12 <karki_> panto ?
Jul 14 14:04:58 * jkridner wonders if JIT targets the interpreter or the secondary operations.
Jul 14 14:05:29 <alexanderhiam> potentially, you'd want to figure out a reasonable maximum frequency for PWM considering that there may be other tasks
Jul 14 14:05:51 <vvu> jkridner: an offtopic question, is there any chance that the debian images for emmc BBB will get smaller ?
Jul 14 14:06:07 <vvu> i tried the usb flasher with the latest image and it is bigger than my available RAM
Jul 14 14:06:17 <vvu> now i am trying to find a workaround this thing
Jul 14 14:06:30 <karki_> alexanderhiam : yeah, and try to figure how many other stuff you can do before you start messing up the frequency!
Jul 14 14:06:59 <alexanderhiam> but don't worry too much about that now
Jul 14 14:08:50 <karki_> k, but still want panto to jump in before I start coding. don't want to write code which I will throw away later.
Jul 14 14:09:17 <alexanderhiam> karki_: what about the ADC, do you know how you'll implement that yet?
Jul 14 14:10:49 <karki_> alexanderhiam : I have the StarterWare and texane (https://github.com/texane/pru_sdk/tree/master/example/pruss_adc) resources for the actual coding part
Jul 14 14:11:38 <karki_> but if you are asking about how I will handle ADC architecturally in PRU Speak - no, haven't yet decided how it will be done.
Jul 14 14:12:09 * karki_ goes for dinner, will be back in half an hour!
Jul 14 14:34:29 <jkridner> vvu: I don't think so, though there is a more minimal *alternative* image that is getting some attention.
Jul 14 14:35:11 <vvu> mmh i am having some issues with u-boot, the work i stopped last year seems to be not working anymore
Jul 14 14:35:17 <vvu> and i forgot some hacks i did
Jul 14 14:35:23 <vvu> need to debug again
Jul 14 14:35:30 <vvu> maybe in a week or so i can up a new system :)
Jul 14 14:38:00 <jkridner> vvu: I'd prefer your work be built on buildroot for the flashing system.
Jul 14 14:38:27 <vvu> can you give me more details
Jul 14 14:39:23 <jkridner> https://github.com/jadonk/buildroot/commit/367dc8d1be24095614b5d35d76133513d02578b1
Jul 14 14:42:56 <vvu> jkridner: and on this buildroot you can make custom commands like flashing via usb ?
Jul 14 15:12:27 <jkridner> I added some stuff to run scripts at boot.
Jul 14 15:12:39 <jkridner> Not that tag though... you have to look for the patch that does that.
Jul 14 15:12:57 <jkridner> I grabbed a script from the root FAT directory, but you could do it lots of ways if you just look how I did it.
Jul 14 15:13:13 <jkridner> nice thing is it builds quickly all from source.
Jul 14 15:13:26 <jkridner> like Angstrom, but simpler. much simpler.
Jul 14 15:16:10 <vvu> mhmh i will see, 1st wanna make the tool working again
Jul 14 15:44:00 <karki_> praveendath92 : hey!
Jul 14 15:44:17 <praveendath92> karki_: Hi
Jul 14 15:44:27 <karki_> hows it going ?
Jul 14 15:45:12 <praveendath92> Not very good.
Jul 14 15:45:39 <karki_> I was wondering if the android app part which you are writing in java could be written in JS as an HTML5 app.
Jul 14 15:45:44 <praveendath92> I have to understand some code which I think I'm really bad at.
Jul 14 15:45:56 <karki_> I hate reading code too :'(
Jul 14 15:46:10 <praveendath92> Using phonegap?
Jul 14 15:46:26 * Abhishek_ reads ~10 lines of code for every 1 line of code written
Jul 14 15:46:51 <karki_> yeah
Jul 14 15:46:51 <praveendath92> I'm personally don't like doing that and besides I wonder if Phonegap supports Usb related stuff yet.
Jul 14 15:47:20 <praveendath92> Even Google hasn't given a proper documentation Accessories.
Jul 14 15:47:26 <karki_> you prefer native android(java) over JS frameworks?
Jul 14 15:47:31 <praveendath92> Why karki_ ?
Jul 14 15:47:40 <praveendath92> For phone apps, yes.
Jul 14 15:47:48 <Abhishek_> karki_: I presume it is for the web interface of BS?
Jul 14 15:47:57 <karki_> yeah
Jul 14 15:47:58 <praveendath92> Oh.
Jul 14 15:48:08 <karki_> but I don't need  the usb stuff for that
Jul 14 15:48:18 <karki_> though I need HTML5
Jul 14 15:48:23 <Abhishek_> I got some basic Socket.IO communication working today
Jul 14 15:48:38 <karki_> I was wondering about the USB stuff for another project.....
Jul 14 15:48:43 <praveendath92> Abhishek_: With JS?
Jul 14 15:48:47 <Abhishek_> yeah
Jul 14 15:49:07 <Abhishek_> Node.JS + Express + (Jade)
Jul 14 15:49:10 <praveendath92> Nice. That can be helpful if we want mix things.
Jul 14 15:49:38 * praveendath92 wondering if it possible to write native and html5 code in phonegap.
Jul 14 15:51:57 <karki_> Abhishek_ : could you give me some resources from where you started off?
Jul 14 15:52:30 <Abhishek_> karki_: http://socket.io/get-started/chat/
Jul 14 15:52:31 <vvu> who wants to share a faster internet connection when downloading kernel sources ? :)
Jul 14 15:52:52 * Abhishek_ wishes he could share his LTE connection
Jul 14 15:53:38 <praveendath92> vvu: What for?
Jul 14 15:54:08 <vvu> i started fixing my project from last year and want to recompile the latest bbb kernel, need a custom config not default one
Jul 14 15:55:19 <praveendath92> Oh.
Jul 14 16:01:06 <karki_> guys, I need some help. This semester we have a course on mobile system engineering. Our lab work is completely based on a project we choose to do. I *need* ideas ;)
Jul 14 16:02:30 <av500> scope?
Jul 14 16:03:01 <karki_> well 4 people across 3 months, part  time :)
Jul 14 16:03:06 <av500> no
Jul 14 16:03:11 <karki_> oh
Jul 14 16:03:12 <av500> scope of that  course
Jul 14 16:03:17 <av500> what is it about?
Jul 14 16:04:05 <karki_> they start off with internals of mobile systems (hardware, GSM/CDMA, etc.)
Jul 14 16:04:19 <karki_> followed by some simple apps in java
Jul 14 16:04:29 <karki_> then some kernel stuff and internals
Jul 14 16:04:48 <karki_> then HTML5, cloud, IoT etc
Jul 14 16:04:52 <av500> ah
Jul 14 16:05:40 <karki_> Though the project is not limited to what is taught. :)
Jul 14 16:05:56 <av500> yes, this I know
Jul 14 16:06:11 <av500> 5 years in #beagle taught me that
Jul 14 16:07:56 <jkridner> vvu: To build with buildroot, I ran patch.sh from the latest beagleboard kernel, then pushed it to a git tree. Then building with buildroot was super easy, just using the commit id.
Jul 15 02:05:35 <Abhishek_> Morning all
Jul 15 02:05:56 <Abhishek_> Travelling back to India today
Jul 15 02:29:36 <karki_> college tomorrow?
**** ENDING LOGGING AT Tue Jul 15 02:59:59 2014