**** BEGIN LOGGING AT Sat Apr 14 02:59:59 2012
Apr 14 09:17:43 * mru reads felipe's "discussion"
Apr 14 09:44:52 <koen> mru: what, where?
Apr 14 09:45:03 <mru> lkml and g+
Apr 14 09:48:25 <aholler> hmm, that stable thingy, it's just annoying ;)
Apr 14 09:49:50 <ynezz> unstable stable, or stable unstable?
Apr 14 09:50:08 <aholler> actually both
Apr 14 09:50:32 <koen> mru: ah, that explains it, I blocked him on g+
Apr 14 09:50:32 <aholler> we should all stick to 2.6.32 ;)
Apr 14 09:51:23 <ynezz> bah, 2.6.26.5 ftw!
Apr 14 09:53:18 <ynezz> mru: link?
Apr 14 09:53:55 <koen> ynezz: https://lkml.org/lkml/2012/4/12/319 and up
Apr 14 09:54:15 <koen> felipe fails to grasp the concept of 'backport'
Apr 14 09:54:30 <aholler> do you all know him?
Apr 14 09:54:39 <ynezz> oh, thats too long for a weekend
Apr 14 09:54:54 * ynezz bookmarks to "read at work" folder
Apr 14 09:55:47 <aholler> he just is upset not to getting a bugfix fast into the stable tree.
Apr 14 09:57:06 <koen> aholler: felipe ususally forks your code and then accuses you of plagiarism if you do bugfixes on the original code, that happened to the pidgin folks
Apr 14 09:57:53 <koen> aholler: when I was at TI there was a discussing if we should involve in a project and if we did how we could stop him from pissing everyone off
Apr 14 09:57:59 <koen> discussion*
Apr 14 10:00:52 <aholler> fork the project ;)
Apr 14 10:07:07 <koen> I hate forks, but I adore the github type of forks
Apr 14 10:08:12 <koen> oooh
Apr 14 10:08:13 <koen> https://lkml.org/lkml/2012/4/12/434
Apr 14 10:08:18 <aholler> github is annoying as the forks there never disappear. Give it one or two years more and it's overcrowed with silly forks.
Apr 14 10:08:30 <koen> Linus: "You're a fucking moron."
Apr 14 10:08:36 <koen> I will print that out and frame it
Apr 14 10:09:24 <aholler> I found the last sentence much better
Apr 14 10:10:13 <mru> aholler: you don't suppose the ministry of silly forks will deal with them?
Apr 14 10:11:24 <aholler> the problem is that people are unable to delete the forks they done themself. that would help at least a bit.
Apr 14 10:11:58 <koen> you can delete a repo
Apr 14 10:12:04 <koen> I did that few times
Apr 14 10:12:17 <aholler> oh, than that got changed.
Apr 14 10:12:37 <koen> I delete/recreate to get rid of the fork label
Apr 14 10:12:45 <koen> gitorious on the other hand is a pain
Apr 14 10:13:15 <koen> you can't delete a repo if there are 'clones'
Apr 14 10:13:16 <mru> and it's down more often than it's up
Apr 14 10:13:20 <aholler> maybe it was gitorious, don't remember, but of them didn't allow to delete repos
Apr 14 10:13:39 <aholler> +one
Apr 14 10:13:40 <koen> I moved most (if not all) of my gitorious stuff to github
Apr 14 10:13:50 <koen> but I can't delete them
Apr 14 10:14:08 <koen> and deleting the content is just dangerous since I will screw it up myself :)
Apr 14 10:14:33 <koen> "only push this rm -rf commit to gitorious, whoops pushed it to every repo"
Apr 14 10:14:50 <mru> that can be undone
Apr 14 10:16:03 <koen> yes
Apr 14 10:16:11 <koen> but me being a moron can't be undone
Apr 14 10:16:41 <aholler> the net doesn't forget anything ;)
Apr 14 10:16:47 <mru> koen: you're confusing yourself with felipe
Apr 14 10:17:24 <koen> heh
Apr 14 10:18:02 <koen> I wish felipe would concentrate on coding instead of being silly
Apr 14 10:18:20 <mru> he is actually reasonably competent when he puts his mind to it
Apr 14 10:18:44 <koen> my point exactly
Apr 14 10:20:45 <koen> mru: what's teh libav plan for linuxtag?
Apr 14 10:21:10 <aholler> you all are visiting linuxtag?
Apr 14 10:21:39 <koen> I am, travel got approved :)
Apr 14 10:21:59 <mru> my evil boss made a fuss
Apr 14 10:23:02 <koen> wtf, linuxnacht tickets are €45?!?!
Apr 14 10:23:08 <mru> it sucks anyway
Apr 14 10:23:24 <aholler> yes, they should move it back to KA ;)
Apr 14 10:23:50 <mru> the first time in berlin was ok
Apr 14 10:23:56 <mru> since then it's only got worse
Apr 14 10:24:04 <koen> I liked the parkinglot next to the beach
Apr 14 10:24:30 <mru> but we can do that without paying 45 euros
Apr 14 10:24:41 <koen> exactly
Apr 14 10:25:06 <koen> €20 was what you would pay for dinner + beer in the city as well
Apr 14 10:29:06 <koen> and back to the stable situation: I had greg drop a patch from stable without any problems
Apr 14 10:29:34 <koen> he backported only a port of the twl audio changeset, so it broke audio on beagle
Apr 14 10:29:40 <koen> s/port/part/
Apr 14 10:30:00 <koen> report it, send revert, get new stable release a week later with working audio
Apr 14 10:36:27 <aholler> I think such depends on the phase of the moon, the weather, how kids are behaving, ...
Apr 14 10:37:49 <koen> I have learned to just give up at some point and go with the flow
Apr 14 10:38:01 <koen> in management speak "disagree and commit"
Apr 14 10:54:47 <_av500_> "don't shed where you bike"
Apr 14 16:34:59 <mranostay> morning friends and _av500_
Apr 14 16:36:11 <prpplague> mranostay: how dare you speak to us like that!
Apr 14 16:39:24 <mranostay> prpplague: just _av500_
Apr 14 16:54:40 <rlrosa> mornin
Apr 14 16:56:05 <rlrosa> av500, i got the logging thing solved. as you suggested, the bottleneck was the SD card. now i'm logging to a flash drive (connected to the USB port) and write() never takes over a second, and i have my steady-ish 10ms loop, and my quadcopter may not crash today :)
Apr 14 16:56:06 <rlrosa> thanks
Apr 14 16:57:43 <aholler> the problem isn't the sd-card, the problem is that your receiver gets interupted or blocked by the writing to the sd.
Apr 14 17:18:05 <rlrosa> aholler, that's true, but the SD card timing is bad...
Apr 14 17:18:23 <rlrosa> non blocking calls to write()  take several seconds sometimes...
Apr 14 17:22:09 <rlrosa> comm between logger and main is done using non blocking pipes, i don't understand why SD card halts my main. the only explanation i find is that the pipe get full (because logger stalls saving to SD), so main stalls writing to it
Apr 14 17:24:09 <_av500_> so make a longer pipe
Apr 14 17:24:17 <_av500_> a circular buffer
Apr 14 17:24:23 <_av500_> one thread writes, one reads
Apr 14 17:24:54 <_av500_> dead simple
Apr 14 17:25:01 <_av500_> you dont even need a mutex
Apr 14 17:26:16 <rlrosa> _av500_, can i specify in C what size i want to pipe to be or is it defined in the kernel ?
Apr 14 17:28:29 <_av500_> i dont know
Apr 14 17:28:34 <_av500_> but you dont need a pipe
Apr 14 17:28:43 <_av500_> use 2 threads and a circular buffer
Apr 14 17:29:00 <_av500_> your usb stick might as well block for 10s one day
Apr 14 17:29:06 <_av500_> you cannot be sure
Apr 14 17:29:11 <_av500_> so decouple it properly
Apr 14 17:29:26 <_av500_> and e.g. drop log data if the buffer does not empty
Apr 14 17:32:22 <rlrosa> the problem i see is that i have no way of telling when i have to drop log data, etc... i check if can read/write to pipes/SD using select, and it never complains about pipe being full or SD card being unavailable...
Apr 14 17:35:34 <aholler> you have to be sure the receiver doesn't get blocked
Apr 14 17:36:08 <_av500_> rlrosa: that does not matter
Apr 14 17:36:23 <_av500_> if your circular buffer reaches a certain state, you drop
Apr 14 17:36:38 <_av500_> regardless of whether the receiver thread is blocked or not
Apr 14 17:37:13 <aholler> if the receiver is blocked, the buffer doesn't fill up ;)
Apr 14 17:37:37 <rlrosa> right, that makes sense. thanks a lot guys :)
Apr 14 17:39:44 <aholler> rlrosa: I assume the process is blocked, not a part of it
Apr 14 17:40:18 <aholler> so if you write, the whole process waits when the card is busy
Apr 14 17:41:28 <aholler> I have such behaviour with the ubuntu on my dead-slow-netbook (which uses an slow 8gb-card internally).
Apr 14 17:41:52 <rlrosa> aholler, i didn't undertand..
Apr 14 17:44:01 <aholler> I want to say that you might not solve having the receiver and logger in the same process, oreven with having the receiver in userland. you still might try that, but having the receiver and a buffer inside a driver would be much more reliable
Apr 14 17:44:42 <mranostay> _av500_: reading your r-pi story link
Apr 14 17:44:50 <mranostay> i like the CE mark stuff :)
Apr 14 17:51:18 <_av500_> aholler: different threads should be fine
Apr 14 17:51:25 <_av500_> i am doing that since ages
Apr 14 17:51:50 <_av500_> 2 threads, circular buffer, done
Apr 14 17:52:00 <aholler> not sure. as said, my ubuntu stands still when the internal sd is busy
Apr 14 17:52:39 <aholler> that all depends on much stuff, e.g. if preempt is enabled and such
Apr 14 17:53:35 <rlrosa> aholler, i'm new to this level of code, so i think i'll follow _av500_ suggestion, since it seems easier then going to kernel level (i know nothing about that, yet :) )
Apr 14 17:53:59 <_av500_> aholler: swap?
Apr 14 17:54:38 <_av500_> aholler: whether the whole systems hangs due to sd should be easy to find out
Apr 14 17:54:49 <_av500_> printf and gettimeofday() is all you need
Apr 14 17:54:54 <aholler> yes, might be the reason. but userland is userland and he has 10ms as goal
Apr 14 17:55:03 <_av500_> 10 ms is fine
Apr 14 17:55:16 <_av500_> i do video playback and recording with 1ms in userland
Apr 14 17:55:33 <aholler> on a beagle?
Apr 14 17:55:37 <_av500_> on omap3
Apr 14 17:55:44 <_av500_> and arm9 before
Apr 14 17:55:46 <aholler> if you say so.
Apr 14 17:55:55 <_av500_> if all else fails, you put rootfs in ram
Apr 14 17:56:12 <_av500_> then sd can do whatever it wants
Apr 14 17:56:58 <_av500_> e.g. when you have a hdd based media player that has the hdd OFF 99% of the time :)
Apr 14 17:56:59 <aholler> just putting the log into ram would be enough ;)
Apr 14 17:57:13 <_av500_> aholler: and a big battery
Apr 14 17:58:23 <rlrosa> putting log into ram == giant buffer?
Apr 14 17:58:37 <aholler> rlrosa: just start with two threads and a mutex for the buffer
Apr 14 17:58:52 <aholler> if that doesn't work, you can still advance to a kernel-driver ;)
Apr 14 17:59:09 <_av500_> mutex not needed :)
Apr 14 17:59:27 <_av500_> writing a 32bit int is atomic
Apr 14 17:59:41 <aholler> in assembler
Apr 14 18:00:53 <_av500_> in c too
Apr 14 18:00:56 <aholler> there was the problem with usb and gcc 4.6 ;)
Apr 14 18:00:59 <aholler> so, no
Apr 14 18:01:06 <aholler> nothing guarantess that
Apr 14 18:01:35 <aholler> anyway, that are details ;)
Apr 14 18:01:38 <_av500_> unless you have a compiler that makes a 4x 8 bit store out of 32 bit
Apr 14 18:01:56 <aholler> which gcc 4.6 had done
Apr 14 18:02:26 <_av500_> ok, i assumed 4 byte alinged ints :)
Apr 14 18:02:42 <_av500_> but details indeed
Apr 14 18:03:43 <rlrosa> ok, i'll read about multithread and try that approach. thanks, again :)
Apr 14 18:04:53 <aholler> multithread and a ringbuffer inbetween them. and using a mutex shouldn't make that too slow and is save
Apr 14 18:06:03 <aholler> or use two process and network-sockets
Apr 14 18:06:25 <aholler> (non-blocking) if you don't want to dive into multithrading
Apr 14 18:06:32 <aholler> or just use syslog ;)
Apr 14 18:07:42 <rlrosa> aholler, ok, thanks!
Apr 14 21:33:16 <Guest30036> is it possible for the bootloader to become corrupted?
Apr 14 21:40:54 <me773> I have a beagle board xm c1
Apr 14 21:41:01 <me773> and it was running fine until I acidently poped out the SD card while it was running
Apr 14 21:41:08 <me773> now it does not boot at all, with the SD card in or out
Apr 14 21:41:15 <me773> just displays random text instead of a boot sequenc
Apr 14 21:41:20 <me773> is there a way to completely reset it?
Apr 14 21:46:10 <thurbad> prehaps you corrupted the SD?
Apr 14 21:46:37 <me773> shouldn;t it boot without the sd then?
Apr 14 21:46:51 <thurbad> from what?
Apr 14 21:46:53 <me773> into the bootloader?
Apr 14 21:47:29 <thurbad> the bootloader in rom just hands off to MLO on the SD VERY early
Apr 14 21:47:56 <me773> how can you tell if it is running or not?
Apr 14 21:48:10 <thurbad> do you have a serial port?
Apr 14 21:48:15 <me773> yes
Apr 14 21:48:26 <me773> I was using minicom
Apr 14 21:48:45 <thurbad> do you see anything when you press reset?
Apr 14 21:48:56 <me773> just a string of random text
Apr 14 21:49:40 <thurbad> the text is probably the string that says the CPU is still alive
Apr 14 21:50:17 <thurbad> it was like 40u on the rev c4 beagleboard, not sure what it is on the xM
Apr 14 21:50:40 <me773> so reformatting the SD would fix this?
Apr 14 21:51:09 <thurbad> perhaps
Apr 14 21:51:20 <thurbad> ver likely actually
Apr 14 21:51:36 <me773> how do you do that? Mine came formatted
Apr 14 21:51:50 <thurbad> you only had the one SD?
Apr 14 21:51:54 <me773> yes
Apr 14 21:59:04 <thurbad> http://code.google.com/p/beagleboard/wiki/LinuxBootDiskFormat should get you started asuming you have a linux box to do the partitioning with
Apr 14 22:00:50 <me773> thanx, need to get an SD card reader though...
Apr 14 22:01:44 <thurbad> ah,yeah that's helpful, they're only about $10 or so
Apr 14 22:02:00 <me773> k, thanks again!
Apr 15 00:38:52 <W1N9Zr0> can you use XDS100v2 to do source level debugging of kernel modules on beagleboard-xm?
Apr 15 01:16:48 <isaacbw> is the filesystem-based manipulation of GPIOs on the beaglebone the work of a kernel module?
**** ENDING LOGGING AT Sun Apr 15 02:59:59 2012