**** BEGIN LOGGING AT Mon Mar 04 02:59:59 2013 Mar 04 04:23:23 http://www.thingiverse.com/thing:54771 <-- beagle case Mar 04 04:45:39 mrpackethead: have a 3D printer? Mar 04 04:45:58 yes in deed i do. Mar 04 04:48:29 mrpackethead: what type? Mar 04 04:48:42 makerbot replicator 2 Mar 04 04:48:48 sorry replicator dual Mar 04 04:48:56 replicator 2 is the new model. Mar 04 04:49:13 dual head extrusion Mar 04 05:25:26 * mranostay remembers why he doesn't do development work on a MBA Mar 04 05:30:19 ok, is it just me, or is this backtrace completely useless and insane https://lkml.org/lkml/2013/3/2/170 Mar 04 05:31:10 hi folks, anyone has a good recommendations on git kernel repo for beaglebone? I'd like something I can just type make after configuring manually. There are so many places to get it from not to mention the ones which are over a year old... Mar 04 05:31:37 aquired siglock in vblank_timer_fn? say what? Mar 04 05:31:45 er, acquired Mar 04 05:32:23 does this mean that someone is touching freed memory? Mar 04 05:33:30 are you running run an encrypted fs? Mar 04 05:33:36 yes Mar 04 05:33:40 ahhhh Mar 04 05:33:43 not root, but home Mar 04 05:34:04 what in the backtrace indicates that to you? Mar 04 05:34:12 the binary appears to be located on an encrypted fs? Mar 04 05:34:22 ah, yes it is Mar 04 05:34:28 well... the calls to execv tells me you are starting something Mar 04 05:34:36 then I see stuff about cryptomgr Mar 04 05:35:09 Russ: that is certainly not easily parsed Mar 04 05:35:59 so either the crypto manager is acquiring a lock when it shouldn't (prehaps for a crypto acc but the platform suggests that isn't likely) or prehaps the crypto code wanted entropy and ran into an issue thus blowing up Mar 04 05:36:57 the actual dead lock doesn't happen until much later Mar 04 05:37:30 you seem to have stack or some kind of randomization enabled too Mar 04 05:37:44 but the deadlock is in a do_execve Mar 04 05:38:10 who knows, whatever the ubuntu kernel configuration is Mar 04 05:38:18 fuse is involved? Mar 04 05:38:36 which flavor of encryption is used? Mar 04 05:38:43 fuse/loopback/other? Mar 04 05:38:46 actually wait... Mar 04 05:38:56 the binary isn't encrypted, I just have encrypted ~/Private Mar 04 05:39:08 ecryptfs Mar 04 05:39:52 it isn't the binary Mar 04 05:40:00 it is somethign to do with it starting the script interperter Mar 04 05:41:13 I'm confused why it thinks (&sighand->siglock)->rlock has anything to do with (&new->fa_lock)->rlock Mar 04 05:41:36 or how vblank_disable_fn takes (&sighand->siglock)->rlock Mar 04 05:42:26 prehaps it is related to the context it is running in Mar 04 05:42:59 wild ass guess - the way things are acquring entropy could be doing bad things Mar 04 05:43:25 this only happens when I have uml running Mar 04 05:43:42 which UML are you referring to? Mar 04 05:43:54 user mode linux Mar 04 05:44:00 ahhh Mar 04 05:44:18 I wonder if the randomization stuff is interacting Mar 04 05:44:34 you have an elf binary (from the host kernel side) running another elf binary (UML instance) Mar 04 05:44:53 uml is just a user process, it doesn't do anything a user process can't do Mar 04 05:45:12 yes, I am familiar with UML Mar 04 05:45:31 wait are we on topic here? Mar 04 05:45:37 and just to be clear, this issue is happening on the host, not the guest Mar 04 05:45:50 * mranostay hides Mar 04 05:45:52 but normal processes rarely fork() and not immediately exec Mar 04 05:49:12 a capture from an actual deadlock doesn't help things much, as it deadlocks in a slightly different place on a different lock Mar 04 05:49:22 https://lkml.org/lkml/2013/3/2/127 Mar 04 05:50:09 I wonder if sigIO is the culprit Mar 04 05:50:55 send_sigio is called by kill_fasync Mar 04 05:51:40 maybe its actually between tasklist_lock and nonblocking_pool.lock? Mar 04 05:52:18 possible Mar 04 05:54:34 getting somewhere I think Mar 04 05:54:55 release_task takes tasklist_lock and then calls __exit_signal/posix_cpu_timers_exit Mar 04 05:55:24 posix_cpu_timers_exit calls add_device_randomness, which calls mix_pool_bytes, which takes nonblocking_pool.lock Mar 04 05:55:41 that doesn't seem good Mar 04 05:55:51 well that's one direction, now I need to find the other direction Mar 04 05:57:49 ok, account takes nonblocking_pool.lock, which calls kill_fasync, which calls send_sigio, which takes tasklist_lock Mar 04 06:00:27 prpplague: so looking at the minnowboard.. any reason they didn't make J19 a little more breadboard friendly? :) Mar 04 06:01:02 that implies something useful Mar 04 06:01:05 we can't have that Mar 04 06:01:26 it is like asking...why not slap a clovertrail or clovertrail II on there instead Mar 04 06:01:58 heh Mar 04 06:02:51 a low cost clovertrail II board would actually be interesting but can't have that... it'd screw up the economy and send thousands of arm related engineers to the unemployment line Mar 04 06:03:31 heh Mar 04 06:03:37 I'm sure that's the rationale. Mar 04 06:11:02 ok, so I found the commit that causes that problem...but I've been having the hardlock since before that commit, crap Mar 04 06:11:20 just hadn't bothered to hookup netconsole and lockdep till now Mar 04 06:12:47 maybe lockdep is not enabled by default on ubuntu? Mar 04 06:12:55 so all you get is mystery locks :D Mar 04 06:16:13 mranostay: we're discussing things in #minnowboard Mar 04 06:16:30 ds2, I recompiled the kernel with lockdep Mar 04 06:20:03 was suggesting maybe it is a latent ubuntu problem Mar 04 06:23:27 Anybody knows about a beaglebone advanced guide (specifically looking for which kernel source repo to use, a bit confused with all the option's I'm running into). I want to build my own kernel, not install somebody else's distro. Already made my own rootfs with buildroot but the kernel in their makefile looks a bit outdated. Mar 04 06:25:12 just pick one and run with it Mar 04 06:25:33 had another hang while I was rebuilding a kernel with a fix Mar 04 06:25:39 this hang looks different Mar 04 06:26:59 http://hastebin.com/dijutejomi.xml Mar 04 06:28:20 grrr #@$!#@$ Ubuntu Mar 04 06:29:22 wtf is that? Mar 04 06:29:34 all I get is some javascript spyware and no data Mar 04 06:30:08 http://pastebin.com/i0WNcvyQ Mar 04 06:30:16 hastebin is cool, but it does use javascript Mar 04 06:30:29 it has nothing but javascript spyware Mar 04 06:30:37 no data. I looked at the source after seeing a blank page Mar 04 06:33:00 hmmm looks the same Mar 04 06:33:09 only complication appears to be the console is F'ed here Mar 04 06:33:13 this one comes through credit_entropy_bits Mar 04 06:33:19 its a netconsole capture Mar 04 06:33:34 I know... but the net console appears to be F'ed Mar 04 06:34:11 this one is a hard lock Mar 04 06:34:24 but the underlying stuff seems to be very similiar Mar 04 06:34:44 yes, and sadly while lockdep did let me find a circular locking dependancy, it doesn't appear to be the one that is causing me problems Mar 04 06:36:48 ok, I've installed a kernel that fixes the circular dep I found, lets see if that lets lockdep get any additional info Mar 04 06:36:52 rebooting Mar 04 06:52:49 need tutorials on beagleboard xm bringup Mar 04 07:22:58 * LetoThe2nd need caffeine Mar 04 07:29:44 LetoThe2nd: trolls! Mar 04 07:30:06 troll troll troll Mar 04 07:30:56 mru: trolling in the eastern world? Mar 04 07:30:57 mranostay: trölls? Mar 04 07:31:23 oh you Mar 04 07:36:49 JIHAD! Mar 04 07:39:12 arrest that man --> KotH Mar 04 07:39:26 why? what have i done wrong? Mar 04 07:39:36 i just called people for a crusade against the infidels Mar 04 07:39:43 you assume that is a requirement Mar 04 07:39:51 something any believer should do, regularly Mar 04 07:41:09 KotH, define "infidels" Mar 04 07:41:31 that's highly dependent on the point of view one takes ;) Mar 04 07:41:52 != your group Mar 04 07:41:53 mranostay: Error: "=" is not a valid command. Mar 04 07:42:12 can someone please kill that bot Mar 04 07:42:24 !self-destruct Mar 04 07:42:25 mru: Error: "self-destruct" is not a valid command. Mar 04 07:42:37 who's bot is it? Mar 04 07:44:23 mru: or kill the person that put it here Mar 04 07:44:32 will that get rid of th bot? Mar 04 07:46:21 ok right have them log off then kill them Mar 04 07:56:22 ka6sox: as you are here... you said a couple of days ago something about striplines for producing short pulses Mar 04 07:56:33 ka6sox: would you be so kind and elaborate that a little bit? Mar 04 08:01:27 KotH, sure.... Mar 04 08:02:01 the reason for that is to match the line for short pulses and reduce the capacitance to let them have nice rising edges. Mar 04 08:02:23 ah.. Mar 04 08:02:25 ok Mar 04 08:02:48 * KotH thought that matching was an obvious thing to do, at those frequencies :) Mar 04 08:07:15 yawn Mar 04 08:08:16 moaning Mar 04 08:10:57 KotH, what is the frequency and PRF? Mar 04 08:12:50 mru: maybe somebody could sponsor Amber a camera Mar 04 08:13:24 ? Mar 04 08:15:31 ka6sox: PRF? Mar 04 08:15:54 ka6sox: i want to produce pulses, shorter than 1ns, with a repetition rate higher than 100MHz, if possible 700MHz Mar 04 08:16:37 ka6sox: the ultimate goal would be 100ps pulses with 700MHz Mar 04 08:18:49 two shiney new beagle-cases just coming out of the printer Mar 04 08:19:02 so 100pS wide with a PRF of 700mhz??? Mar 04 08:19:27 yes, if possible Mar 04 08:19:41 what voltage are the pulses? Mar 04 08:19:56 <10V, <30mA Mar 04 08:20:08 i'd like to drive a laserdiode with that Mar 04 08:21:03 mru: https://plus.google.com/u/0/100254004947968890609/posts/BytbE91S8gz Mar 04 08:21:17 mru: also please note, your food has been sponsored by ARM Mar 04 08:21:28 did I mention it was sponsored by ARM? Mar 04 08:22:17 what a blur Mar 04 08:22:33 SF to HK in how many days? Mar 04 08:22:59 got home monday, left saturday Mar 04 08:23:20 plenty enough time to do laundry and repack bags Mar 04 08:24:02 er G+ fail Mar 04 08:24:04 "Tomorrow, March 5, 9:00 AM HKT" Mar 04 08:24:12 :) Mar 04 08:24:29 we're living in the future over here Mar 04 08:24:29 wtf is that in my local time? Mar 04 08:24:41 hehe Mar 04 08:25:14 oh, it was an invite Mar 04 08:25:26 future time makes perfect sense then Mar 04 08:25:38 Hello. Assuming I want to change u-boot to include Simon Glass' patches, do I also have to recompile the MLO ? Mar 04 08:27:27 mru: so an hour from now? Mar 04 08:27:49 ka6sox: i guess, you dont have any good idea how to get such short pulses? Mar 04 08:27:54 KotH, so effectively 20Ghz Mar 04 08:28:03 juup Mar 04 08:28:10 mranostay: west coast has different arithmetic? Mar 04 08:28:19 mru, yes Mar 04 08:28:42 KotH, not without getting very fancy. Mar 04 08:28:47 ^^' Mar 04 08:28:49 the pulses will be ugly Mar 04 08:28:49 thought so Mar 04 08:28:51 KotH: use bit banging and Moore's law Mar 04 08:28:53 juup Mar 04 08:28:53 and wiat Mar 04 08:29:02 wait* Mar 04 08:29:09 mranostay: it's 4.29pm here now Mar 04 08:29:14 av500: unfortunately, moores law is limited by physics Mar 04 08:29:19 i see that now Mar 04 08:29:22 the 4th Mar 04 08:29:30 and physics is limited by? Mar 04 08:29:36 av500: getting 1ns pulses is "easy" with logic, shorter is kind of difficult Mar 04 08:29:54 ka6sox: thanks for your help anyways Mar 04 08:30:18 I routinely work up to 17ghz...but its not expecting what you are doing. Mar 04 08:30:55 hehe Mar 04 08:31:06 i routinely work between 120hz and 8kHz Mar 04 08:31:15 yes, getting a continous sinus is easier at those frequencies, than pulses Mar 04 08:31:16 so probalby not much help to you Mar 04 08:31:25 I work 9 to 5 Mar 04 08:31:36 Pulses are a PITA Mar 04 08:31:46 ka6sox: P is for pulse? Mar 04 08:31:52 av500, ypu Mar 04 08:32:00 s/ypu/yup/ Mar 04 08:32:15 its the stupid risetime BS. Mar 04 08:33:52 ka6sox: i can probably get away with 300ps-500ps pulses... but that highly depends on the laser diode... and i currently dont have the equipment to test that Mar 04 08:34:14 ka6sox: so i'd like to have something with which i can produce shorter pulses if necessary Mar 04 08:34:59 KotH: btw, what is the use case? Mar 04 08:35:10 av500: jihad Mar 04 08:35:23 av500: building a phaser out of off-the-shelf components Mar 04 08:35:34 KotH, make sure your wires are SHORT...and both capacitance and Inductance free. Mar 04 08:36:09 no such thing as inductance-free wires Mar 04 08:36:20 av500: or alternatively, if that is not possible an optical pll for offset frequencies in the GHz range, where normal photodiodes dont work anymore Mar 04 08:36:35 ka6sox: how do you keep a wire capacitance and inductance free? Mar 04 08:36:42 use wireless Mar 04 08:37:11 KotH, build the entire thing on the SAME substrate. Mar 04 08:37:52 ka6sox: ah.. and everything impedance matched? Mar 04 08:37:58 yes Mar 04 08:39:48 mru, you have to eliminate the wire. Mar 04 08:40:48 you need a retriggerable monostable oscillator with a nasty AMP driving this UGLY load called your diode :P Mar 04 08:41:10 juup Mar 04 08:41:23 current idea is to use an avalanche transistor Mar 04 08:41:35 but i am not sure whether it can short pulses like that Mar 04 08:41:50 * KotH adds a "produce" somewhere in that sentence Mar 04 08:42:12 just bitbang it Mar 04 08:42:35 and what makes the diode more ugly than it has to be: they come in TO-52 and TO-48 only.. no SMD cases ^^' Mar 04 08:43:11 KotH, then you are pissing up a rope...they wont' got that fast. Mar 04 08:43:24 ^^; Mar 04 08:43:25 -t Mar 04 08:44:09 anyways.. gtg.. lecture is calling Mar 04 08:44:16 thanks again! Mar 04 08:44:21 neither of those packages are very LOW inductance or capacitance... Mar 04 08:44:27 GLHF.... Mar 04 08:45:02 (he needs a Krause-Ogle Box Structure...) Mar 04 09:21:46 LetoThe2nd: it's the guy who just needs an ADC... Mar 04 09:21:55 dm8tbr: i know. Mar 04 09:22:02 ah, good :) Mar 04 09:33:00 * dm8tbr gets more popcorn Mar 04 09:40:48 dm8tbr: not necesary. Mar 04 10:01:28 * av500 makes sure that pfefferz knows his breakfast was sponsored by ARM Mar 04 10:02:28 by adding silicium to it? Mar 04 10:02:49 no, just big signs stating the same Mar 04 10:03:36 or by a 'less-heat-producing' coffee maker... Mar 04 10:32:52 on the bone is it possible to tftp a new u-boot to auto-update it afterwards ? Mar 04 10:33:14 ? Mar 04 10:33:23 but since its SW, anything is possible Mar 04 10:34:45 when I do tftp addr u-boot.bin it does find the file provided by my tftp server Mar 04 10:34:51 but it hangs there Mar 04 10:34:59 maybe I wrongly specified the address Mar 04 10:43:19 sniff the wire Mar 04 10:44:49 can anyone running the CCO 3.8 beaglebone kernel successfully run perf, without it hanging the system on exit? Mar 04 10:57:03 where do I find the memory layout for the bone ? Mar 04 11:03:15 okay I tftp'd correctly using a TI document, RAM starts at 82000000 apparently, but how to know that if there's no tutorial readily available ? Mar 04 11:03:42 RAM starts at 0x80000000 Mar 04 11:05:01 ah alright. is it the am35xx datasheet I have to look at ? same goes for u-boot.img Mar 04 11:05:18 (i.e. I don't know where it starts/ends) Mar 04 11:18:18 guillaume2: the AM335x Technical Reference Manual is the place to look for anything... technical Mar 04 11:19:10 jackmitchell, would it also be the place to look for the starting address of u-boot ? i'm not sure Mar 04 11:19:12 and you can just start reading in it if you can't sleep as well Mar 04 11:21:38 guillaume2: have you read this: http://processors.wiki.ti.com/index.php/AM335x_U-Boot_User%27s_Guide Mar 04 11:22:11 I was reading exactly that Mar 04 11:22:31 there's a graph that maps the NAND Mar 04 11:22:44 does it strictly apply to the bone ? Mar 04 11:23:04 no, it strictly applies to the AM335x processor I would imagine Mar 04 11:24:14 well, I could always erase 0x0008000 - 1, test if it boots, then erase 0x0008000 and if it doesn't then u-boot maybe starts here Mar 04 11:25:27 jackmitchell: I haven't tried using perf yet Mar 04 11:25:54 koenkooi: it's been broken for a while; I can't remember if it has worked in anything but the 3.2.x series for me yet Mar 04 11:25:59 hello Mar 04 11:26:07 world Mar 04 11:26:15 it would be interesting to know if I have a perf problem or if it is a kernel problem Mar 04 11:26:32 I struggle to get any debug out put as the kernel hangs hard without a sniff of a bt Mar 04 11:35:33 crap, I killed my memory card Mar 04 11:36:01 with an assault reader? Mar 04 11:36:39 with an erase bomb Mar 04 11:38:41 i'll have my gparted necromancer go grind some XP because atm it's not powerful enough to revive it Mar 04 11:50:45 oooh that's interesting. It seems as if it's the SD Card port on my computer that has died Mar 04 12:25:01 ka6sox-away: why would i need a krause-ogle box? Mar 04 13:17:12 did anyone get the LCD7 cape working with the 3.8 patch stack? it doesn't seem to configure the correct display mode... Mar 04 13:49:05 av500: perfect start of monday quote, "but since its SW, anything is possible" Mar 04 13:49:14 * mdp feels energized Mar 04 13:49:31 too many energy drinks? Mar 04 13:49:42 av500 is my motivational speaker Mar 04 13:50:05 mdp: https://plus.google.com/112266164281670850856/posts/2ywuCxV19pn Mar 04 13:50:09 you could say he's a motivational poster Mar 04 13:50:31 av500, that's the first one that got me going ;) Mar 04 13:50:56 how much long until friday? Mar 04 13:51:09 av500: where is that pic from? Mar 04 13:51:19 the internet Mar 04 13:51:25 google image search will tell you Mar 04 13:51:28 ok Mar 04 13:51:32 mdp: too long Mar 04 13:51:41 normally I add src links if I have them Mar 04 13:51:54 av500: so no special relation to that pic? Mar 04 13:52:53 other than my eyes beholding it, no Mar 04 13:53:40 http://yx.fi/5jo Mar 04 13:53:46 av500: you're spending too much time with mru... your language is getting more and more arcane Mar 04 13:54:40 yes, as usual it has been tumblr-dried for a while Mar 04 13:54:55 seems the original was a Nescafe ad Mar 04 13:59:01 KotH: he's embracing the Olde English now Mar 04 14:00:19 mdp: you mean, in a couple of months he'll be as stiff necked as an old victorian? Mar 04 14:00:23 * mdp constructs the "what I did last week and what I plan to get done this week" mail :( Mar 04 14:00:27 Le Horror! Mar 04 14:00:41 KotH: yes Mar 04 14:00:46 http://www.marvell.com/company/news/pressDetail.do?releaseID=3576 Mar 04 14:02:01 it's the year of the ARM-linux servers... Mar 04 14:02:10 or will they run windows-rt ;) Mar 04 14:03:05 Jacmet: I got seriously delayed on the repost due to an intermittent PLL lock issue in that ti814x support..thought it was toolchain and finally resolved it for real in the v3 patch ;) Mar 04 14:03:49 mdp: yes, I saw ;) Mar 04 14:04:15 mdp: I've also gotten caught up in all kinds of other things, so I haven't been able to progress much on the 816x stuff Mar 04 14:04:39 Jacmet: had to dump the backstory..when I switch to UART SPL for my pcie enablement, it failed 100% of the time Mar 04 14:04:42 great! Mar 04 14:05:08 that cost me at least a couple days last week chasing ghosts Mar 04 14:05:32 mdp :/ Mar 04 14:05:59 Jacmet: I've got to switch to some kernel tasks for a few days before -rc1 comes out though...then back to pcie Mar 04 14:06:46 mdp: ehh, -rc1 is out Mar 04 14:07:26 -rc2 ;) Mar 04 14:07:53 mdp: ;) Mar 04 14:10:45 Jacmet: it's a bit of a PITA to rebase stuff during the merge window ;) Mar 04 14:11:13 mdp: yeah, I can imagine Mar 04 15:10:42 anyone have an idea how to set the controller data for platform devices when using device tree? it was easy before, just in the board file, but i'm completely stuck on the new DT way Mar 04 15:12:15 cmicali, err, which controller? Mar 04 15:12:35 panto: mcspi - i want to use this patch https://patchwork.kernel.org/patch/2026411/ (which I applied) Mar 04 15:12:54 but I am feeling like an idiot that i can't figure out how to get the new cs_per_word to = 1 Mar 04 15:15:36 data->cs_per_word = of_property_read_bool(node, "ti,cs-per-word"); Mar 04 15:15:57 that patch you're using isn't DT enabled Mar 04 15:31:26 panto: thanks - i was thinking of adding that, but wasn't sure if there was another way that was intended - i'll make that edit, thanks! Mar 04 15:56:31 panto, heh, I just read what you linked..not remembering that I replied ;) Mar 04 15:56:48 panto, I feel like the Memento guy Mar 04 15:57:31 mdp: write "buy milk" on your firehead? Mar 04 15:57:35 forehead Mar 04 15:57:37 mdp, I can't remember either :) Mar 04 15:57:38 firey too Mar 04 15:58:21 hi guys, when building the kernel "bitbake virtual/kernel -c compile -f" is it possible not to build using linux-mainline-3.2 but instead linux-3.0 or linux-omap-2.6.39? the problem is that the compiler is not taking the patches of 3.0/omap-2.6.39 which include the hack for bbxm for 1ghz. Mar 04 15:58:47 is that something to change in local.conf? Mar 04 16:00:16 or is it safe to copy the patches from 3.0/omap-2.6.39 to the mainline-3.2 folder and change the .bb file? Mar 04 16:14:27 hi Mar 04 16:14:48 i've managed to install debian armhf on beagleboard-xm Mar 04 16:15:31 but i can't made it to use more than 1024x768 dvi-d output Mar 04 16:16:05 i know it has this capability since the default angstrom distribution correctly shows up Mar 04 16:16:31 i've tried the video mode hd720 over the kernel cmdline Mar 04 16:16:34 to no avail Mar 04 16:19:41 now i'm trying to compile a new kernel from github.com/RobertCNelson/stable-kernel Mar 04 16:20:08 don't know which version is best for me Mar 04 17:34:25 troll troll troll Mar 04 17:34:34 your boat Mar 04 17:36:08 that doesnt fit ... it continues with "gently" Mar 04 17:39:28 furiously down the interwebz? Mar 04 17:40:38 better :) Mar 04 17:47:32 tubes Mar 04 17:47:56 transistors Mar 04 17:54:21 avalanche diodes Mar 04 17:56:30 i hate snow Mar 04 17:56:39 or we talking about something different? Mar 04 17:57:06 * dm8tbr likes snow Mar 04 17:57:21 +20cm in the last 2 days Mar 04 17:57:42 snow trolls ? Mar 04 17:57:55 yeti-trolls Mar 04 17:58:05 you heard of them, but you've never seen them Mar 04 17:59:59 heh Mar 04 18:00:24 if you do see them, it's the *last* thing you see... Mar 04 18:00:38 * ogra_ is actually waiting for some intresting trolling after canonicals recent announcement :) Mar 04 18:01:15 ogra_, what *this* time? Mar 04 18:01:52 https://lists.ubuntu.com/archives/ubuntu-devel/2013-March/036776.html Mar 04 18:02:44 If all the entries in "Device Drivers" Kconfig were sorted alphabetically, would the world end? Mar 04 18:03:18 mdp, us OLD tymers would probably be lost. Mar 04 18:03:31 I was thinking I might be lost for a while too Mar 04 18:03:52 it only today occurred to me that it must suck for newbies Mar 04 18:03:56 oh well Mar 04 18:06:53 the problem began when you had to scroll Mar 04 18:07:41 and everyone knows kernel people have only limited screen size with no more than 80 characters wide ;) Mar 04 18:08:07 and 24 tall Mar 04 18:08:32 +1 Mar 04 18:08:39 is there something else available? Mar 04 18:08:51 no Mar 04 18:15:08 but google will help, at least with the scroll thing. maybe the next standard will be 1080x1920 Mar 04 18:15:28 easy to realize ;) Mar 04 18:15:56 2k, that's a lot of letters... Mar 04 18:16:09 you want to put all Kconfig in one line? Mar 04 18:18:54 <_av500_> ogra_: awesome troll fodder Mar 04 18:19:38 <_av500_> ogra_: but its not as bad as switching to .RPM Mar 04 18:19:39 NIH trolls wakeup Mar 04 18:20:55 4k Mar 04 18:21:04 Mir (Russian: Мир, IPA: [ˈmʲir]; lit. Peace or World) Mar 04 18:21:44 hippies? Mar 04 18:22:17 <_av500_> ynezz: in my name, I prefer mir to be world Mar 04 18:22:34 <_av500_> ogra_: and it mentions android, awesome Mar 04 18:22:42 <_av500_> and going full GL/EGL Mar 04 18:22:49 <_av500_> that will appease the unity 2d crowd Mar 04 18:22:52 and Linux 2 times Mar 04 18:22:56 <_av500_> and test driven Mar 04 18:23:03 <_av500_> all in all a wonderful plan Mar 04 18:23:33 and qt Mar 04 18:23:43 <_av500_> ynezz: linux twice is awesome Mar 04 18:23:59 <_av500_> even spaceman said linux yestereday Mar 04 18:24:11 I wonder how it could've passed their filters Mar 04 18:24:14 * _av500_ made a screenshot Mar 04 18:24:20 good point Mar 04 18:24:59 <_av500_> though I learned today why they do not mention linux, they are secretly planning to use a BSD kernel Mar 04 18:25:13 maybe just that sed script s/Linux/Ubuntu/ is running in the morning... Mar 04 18:26:49 <_av500_> "...Adaptable to future requirements..." Mar 04 18:26:54 <_av500_> omg, the have a time machine Mar 04 18:27:02 <_av500_> they* Mar 04 18:27:33 <_av500_> "...we decided to name it Mir (as in the space station)..." Mar 04 18:27:41 ouch Mar 04 18:27:44 <_av500_> so, not peace Mar 04 18:27:59 <_av500_> but rather, something flying around held together by chicken wire Mar 04 18:28:13 hm, Mir didn't exploded? Mar 04 18:28:22 <_av500_> I think it was deorbited Mar 04 18:28:36 <_av500_> due to lack of maintenance Mar 04 18:29:00 well, romantic end Mar 04 18:30:17 <_av500_> for those on board, yes Mar 04 18:30:21 it was deorbited Mar 04 18:31:36 It did manage to damage australia, but not nearly as much as was hoped for Mar 04 18:31:54 or was that a different satellite Mar 04 18:32:00 Developers plan to use Linux in half of their upcoming embedded projects, according to preliminary data from an annual EE Times embedded market survey. And Android leads the Linux pack. Mar 04 18:32:01 The data from EE Times’s “2013 Embedded Market Study” were disclosed to embedded market executives by UBM Tech vice president David Blaza and EE Times editor-in-chief Alex Wolfe at Embedded World in Nuremburg last week. Mar 04 18:32:04 Russ: that was SpaceLab iirc Mar 04 18:32:18 and it was meant to hit DC Mar 04 18:32:21 skylab Mar 04 18:32:25 er skylab Mar 04 18:32:37 space.. sky.. same thing Mar 04 18:32:38 android leads the linux pack Mar 04 18:32:56 android != linux as mru would say Mar 04 18:33:09 i woud'nt ahve thought so either. Mar 04 18:33:27 Anstrom is 4% Mar 04 18:33:37 debian is 8% Mar 04 18:33:41 but you can install linux.apk on it Mar 04 18:33:52 mranostay, skynet? Mar 04 18:34:19 koen: you say it Angst ROM right? Mar 04 18:34:38 KoenogenMOD Mar 04 18:35:19 <_av500_> lol Mar 04 18:35:25 <_av500_> waffleROM Mar 04 18:36:45 i'm trying to stream data from an SPI device, and despite sending 2+ transfers per message using spi_async w/ DMA enabled, i always get at least a 0.1ms delay after each api transfer - is this just the price of linux not being realtime, or is there some magic bit I need to twiddle to make this better? scope output: http://uploads.hipchat.com/29857/194444/97eb6y2wmzs763o/upload.png Mar 04 18:37:35 did you check the kernel driver for a msleep(100)? Mar 04 18:37:45 :) Mar 04 18:38:00 driver is mine, and removed all of the msleeps by hand Mar 04 18:38:12 maybe i should msleep(-100) Mar 04 18:38:36 what happened when you used the existing spi driver? Mar 04 18:38:56 ohhhh...can't stop the fortan numbering :-/ Mar 04 18:38:59 * mdp cries Mar 04 18:39:17 I think you mean, human numbering Mar 04 18:39:24 no sorry, the driver is written from scratch, but it's really simple - it's just using spi_async with transfers that are 1k big, and it fires another spi_async in the completion handler of the last transfer Mar 04 18:39:30 Russ, silly humans Mar 04 18:39:32 just start with O and proceed with P Mar 04 18:39:59 why didn't you just try the host driver that already exists first? Mar 04 18:40:03 <_av500_> mdp: use roman numbering Mar 04 18:40:17 <_av500_> ttyOIV Mar 04 18:40:41 just let me grieve Mar 04 18:40:56 <_av500_> mdp: no way, its only monday Mar 04 18:41:01 <_av500_> just fucking do it! Mar 04 18:41:46 russ: host driver? i'm using the stock mcspi spi-omap2 driver for the master, but for the actual device driver there isn't one generally available for the chip i'm using Mar 04 18:42:04 ok, I thought you meant you wrote a host driver from scratch Mar 04 18:44:53 troll troll troll Mar 04 18:45:20 I'm doubtful that a 100ms delay has anything to do with realtime Mar 04 18:45:46 ah sorry no, and it's not 100ms sorry it's 0.1-0.2ms Mar 04 18:46:13 so 100 to 200us Mar 04 18:46:18 yeah exactly Mar 04 18:46:54 still seems pretty high, especially at high spi bus speeds, so i suspect i'm not doing something right Mar 04 18:47:25 how long did you calculate that it would take? Mar 04 18:48:19 i have the SPI bus clocked at 3mhz, so the clock period is 0.33us, so ideally the delay between spi messages would be zero (so 0.33us) Mar 04 18:48:45 the spi spec doesn't give a mininum time between transfers? Mar 04 18:49:09 i don't think so.. i think the notion of a 'transfer' is really just baked into the kernel spi interface Mar 04 18:49:37 what is the delay_usec value is your spi_ioc_transfer? Mar 04 18:49:44 manually set to 0 Mar 04 18:50:52 and i'm adding two transfers to the message with spi_message_add_tail, and i see that 100-200ms delay between those two transfers Mar 04 18:50:53 and have you looked at the mcspi controller documentation? Mar 04 18:51:23 yeah just started to about 30 minutes ago, to see if there is something i'm missing, nothing obvious jumping out so far though Mar 04 18:51:51 I'm not aware of anything Mar 04 18:54:06 thanks.. i think given some of the timing requirements i have i may try to use the PRU instead Mar 04 19:06:52 Russ, hehe, spi spec ;) Mar 04 19:07:59 haven't read it yet, is it as detailed and specific as the i2c spec? Mar 04 19:08:39 mranostay: android is linux Mar 04 19:08:56 mranostay, fix your cape yet? Mar 04 19:09:06 I even have the t-shirt Mar 04 19:11:57 not yet Mar 04 19:12:15 Russ, as it doesn't exist..it definitely less detailed than the i2c spec Mar 04 19:13:25 and the i2c spec doesn't really define much in the first place Mar 04 19:13:43 smbus! Mar 04 19:14:12 koen, you are clearly stuck in x86 land ;) Mar 04 19:14:14 what really sucks is intel actually decided to implement smbus controllers rather than i2c controllers Mar 04 19:14:22 can't talk to some i2c devices Mar 04 19:14:34 koen, next you'll be asking Tartarus about fixes to the AM335x BIOS Mar 04 19:15:58 at least I had the source for that Mar 04 19:16:02 * koen coughs Mar 04 19:16:19 TI G+ feed is great, "New EFuse News!!!" Mar 04 19:16:25 efuse!!! Mar 04 19:16:31 * Tartarus runs Mar 04 19:16:34 TI rEFuses to comment? Mar 04 19:17:11 I always thought those police things looked more like chariots Mar 04 19:17:56 sucks when you blow the wrong efuses Mar 04 19:46:19 Russ: FYI i've found a couple reports that a 80-150us delay in the SPI stuff is unavoidable due to the wait being queued in the linux work queue, sound like it's actually not possible to work around it (http://en.usenet.digipedia.org/thread/18417/2926/, http://permalink.gmane.org/gmane.linux.distributions.gumstix.general/52852( Mar 04 19:47:47 cmicali live is hard in kernelland Mar 04 19:47:51 ups life Mar 04 19:48:01 haha indeed Mar 04 19:48:06 hard and undeterministic Mar 04 19:48:30 cmicali: yeah, it's not made for determinism between transfers Mar 04 19:49:07 i knew that, and yet, i pretended and hoped it would be just enough for my needs Mar 04 19:49:08 depending on your cs mode and your driver using bit-banged CS assertion/deassertion, that adds more Mar 04 19:49:22 i am a huge fan of the head in the sand approach Mar 04 19:49:29 <_av500_> ti g+ added Mar 04 19:49:51 trollbang again Mar 04 19:49:57 gm av500 Mar 04 19:50:03 cmicali: to your credit, you did http://26-26-54.hardwarebug.org/30 Mar 04 19:50:23 _av500_: welcome to the TI G+ community Mar 04 19:50:33 <_av500_> cmicali: what do you need in terms of timing? Mar 04 19:50:34 mdp: amazing Mar 04 19:50:37 _av500_: I replied once to tell them the website sucked Mar 04 19:51:21 <_av500_> cmicali: a transfer every X ms? or just a burst of tranfers all back to back? Mar 04 19:51:22 av500: i want to sample an adc regularly at about 2msps (works out to 40mhz spi transfer) - trick is no gaps, or else you have to post-process the data to deal with them Mar 04 19:51:44 <_av500_> no gaps, so constant tranfers? Mar 04 19:51:55 yeah, constsant transfers Mar 04 19:52:05 <_av500_> /CS never to go high? Mar 04 19:52:11 <_av500_> looked at DMA? Mar 04 19:52:13 unfortunately CS needs to go high every single word Mar 04 19:52:29 <_av500_> so your transfer is one word only? Mar 04 19:52:31 but thanks to a patch someone made recently the McSPI controller has an option to toggle CS after every word Mar 04 19:52:39 <_av500_> ah Mar 04 19:52:45 so i did that, and i'm using DMA Mar 04 19:52:51 and it works great Mar 04 19:53:08 it's just the timing gaps between each transfer - no other way i can see using the linux kernel spi API Mar 04 19:53:19 <_av500_> how long is one? Mar 04 19:53:24 <_av500_> transfer Mar 04 19:53:48 doesn't really matter, i just want to sample at a regular rate, so the larger i make my transfers (spi_transfer.len) the better the data is, because there's no gaps inside the transfer Mar 04 19:53:55 but as soon as it's over, i get a bit of a delay before the next one's over Mar 04 19:54:11 and there doesn't seem to be a notion of continuous transfer / ring buffer in the spi.c api Mar 04 19:54:13 <_av500_> ic Mar 04 19:54:18 <_av500_> not in the API Mar 04 19:54:20 right Mar 04 19:54:32 i have a feeling if i programmed directly against the mcspi controller it might be possible Mar 04 19:54:34 <_av500_> but I guess you can have the DMA complete IRQ trigger the next DMA Mar 04 19:55:01 so that's what I'm doing with the SPI API right now (calling another spi_async in the completion handler (IRQ handler) of the previous one) Mar 04 19:55:15 but inside the mcspi driver it's queuing that transfer which causes the delay Mar 04 19:55:16 <_av500_> yeah, but that takes to long Mar 04 19:55:41 cmicali: yes, what you need is something akin to the dmaengine cyclic transfer...for the spi subsystem Mar 04 19:55:50 <_av500_> yes Mar 04 19:56:03 it seems like someone must have run into this kind of thing before Mar 04 19:56:07 <_av500_> hmm, can the DMA chain on its own? Mar 04 19:56:12 i didn't think trying to do this would be unique Mar 04 19:56:31 <_av500_> like using DMA to even write the next set of SPI controller registers Mar 04 19:56:46 <_av500_> so it autoloops Mar 04 19:56:48 yeah Mar 04 19:56:54 that is exactly what i was thinking, so i'm reading the mcspi controller docs now Mar 04 19:56:59 <_av500_> and all you need is to empty the ring buffer fast enough Mar 04 19:57:00 trying to see if it can do that on it's own Mar 04 19:57:05 <_av500_> yep Mar 04 19:57:14 instead of trying to simulate that in software Mar 04 19:57:29 the other option i was thinking is just bit bang everything using the PRU Mar 04 19:57:30 <_av500_> because even if it was instant, you would still have IRQ masked from time to time Mar 04 19:57:31 _av500_, you can do that with bare metal edma Mar 04 19:57:39 <_av500_> mdp: yep Mar 04 19:59:00 does using the PRU for this seem like it's asking for trouble, or generally an OK idea? Mar 04 19:59:09 <_av500_> PRU should be fine Mar 04 19:59:28 <_av500_> i'd find the DMA chaining more elegant Mar 04 19:59:38 yeah PRU seems like the brute force approach Mar 04 19:59:42 <_av500_> but then, I just watch you write the code :) Mar 04 19:59:47 lol Mar 04 19:59:49 haha Mar 04 19:59:56 great seats! Mar 04 20:00:01 <_av500_> first row Mar 04 20:00:10 and then...we begin the code mocking! Mar 04 20:00:19 <_av500_> gentle at first Mar 04 20:00:21 hihi Mar 04 20:00:27 <_av500_> to lure him in Mar 04 20:00:34 Lure(tm) Mar 04 20:00:57 luke Mar 04 20:01:51 a solid mocking would probably help a lot at this point Mar 04 20:02:24 uhm how to mock kernelland Mar 04 20:02:30 "A 2.5GHz 2.2mW/25μW On/Off-State Power 2psrms-Long-Term-Jitter Digital Clock Multiplier with 3-Reference-Cycles Power-On Time" is needed Mar 04 20:02:40 you're just hitting a "corner case" that the spi subsystem wasn't designed for Mar 04 20:02:59 yeah that is what i guessed Mar 04 20:03:08 the kernel is full of stuff like that Mar 04 20:03:30 it can be an epic journey that would make Homer proud to fix such things Mar 04 20:03:36 a bit new new to kernel, so also have the problem that i don't know what i don't know Mar 04 20:03:37 we will watch Mar 04 20:04:30 we'll kick it off with a little light reading of the McSPI Register Manual Mar 04 20:04:35 to relax and sooth Mar 04 20:05:00 s/sooth/soothe Mar 04 20:05:31 cmicali but you are doing better than 90% joining here Mar 04 20:05:43 *encourage* Mar 04 20:06:14 <_av500_> +1 Mar 04 20:06:49 i have promised myself i will document/post some of the hurdles of getting started Mar 04 20:07:08 cmicali: also take a look at the edma chapter and how that integrates with the mcspi dma reqs Mar 04 20:07:09 for other people getting started Mar 04 20:07:38 "the assault blog" Mar 04 20:07:41 ;) Mar 04 20:08:15 cmicali: you can chain one param set to another..causing the second to start transferring when the other completes Mar 04 20:08:52 mdp: thx - looking for that now Mar 04 20:08:59 that's the am33xx way to keep a set of dma buffers continuously filling for the spi fifo Mar 04 20:09:26 the important thing would be a pair of link param sets, one to write the registers to kick Mar 04 20:09:35 the hard thing is ordering Mar 04 20:09:59 can can't set the dma req in the registers until that transfer param set is set Mar 04 20:10:17 if you try to avoid a s/w control path for managing this Mar 04 20:10:30 er, direct control path Mar 04 20:10:34 anyway, have fun Mar 04 20:11:33 yes, having fun Mar 04 20:13:32 ubuntu goes NIH oh my god MIR Mar 04 20:13:54 <_av500_> woglinde: shhh Mar 04 20:14:03 we still good 'ol linux though..for those not wanting to run ubuntu or android Mar 04 20:14:05 yes fun so far, thanks a lot for the direction Mar 04 20:14:32 <_av500_> woglinde: they should have called it VLADI-MIR of course Mar 04 20:14:37 hehe Mar 04 20:14:42 cmicali: if you find that to be fun..you'll probably get along fine here ;) Mar 04 20:14:46 or ISS-MIR Mar 04 20:14:51 <_av500_> lol Mar 04 20:14:57 <_av500_> EZMIR Mar 04 20:15:20 woglinde: NIH syndrome 100% Mar 04 20:15:24 mdp :) Mar 04 20:16:06 hm or is it the replecement for phone than Mar 04 20:16:11 hmhm Mar 04 20:16:13 we will see Mar 04 20:16:50 <_av500_> woglinde: no, for all Mar 04 20:16:54 <_av500_> phone and tablet first Mar 04 20:16:57 <_av500_> desktop later Mar 04 20:17:11 desktop already has qt Mar 04 20:17:31 <_av500_> that is not a WM Mar 04 20:17:38 <_av500_> unless you use QWS Mar 04 20:17:38 av500 yes thats what I meant Mar 04 20:17:44 sorry Mar 04 20:18:00 the wm isn't of interest, the api to use it is. Mar 04 20:18:00 aholler unity goes QML Mar 04 20:18:32 and g* is a major failure Mar 04 20:18:52 <_av500_> indeed Mar 04 20:19:48 hm ah reading the whole article now Mar 04 20:19:59 * ogra_ grins Mar 04 20:20:02 what is a major failure? Mar 04 20:20:10 Fire. Mar 04 20:20:14 * woglinde sees orga whine Mar 04 20:20:19 <_av500_> no, g+, I do not like Will Wheaton Mar 04 20:20:21 me ? Mar 04 20:20:35 in the future Mar 04 20:20:42 woglinde, DUDE ! the last two weeks were the most entertaining ones in my career !!! Mar 04 20:20:42 when MIR will not be ready Mar 04 20:20:47 _av500_, you might like... Mar 04 20:20:53 ogra LIAR Mar 04 20:20:57 *g* Mar 04 20:20:58 :) Mar 04 20:21:12 MIR ist uebel... Mar 04 20:21:13 woglinde: glib, gnome and whatelse reinvents oo in c Mar 04 20:21:26 but back to the major failure Mar 04 20:21:41 aholler I see nothing wrong in qt/qml Mar 04 20:21:56 glib is not oo Mar 04 20:22:24 qt isn't the failure, g* is Mar 04 20:22:47 g+? Mar 04 20:22:52 <_av500_> gtk Mar 04 20:22:52 g- Mar 04 20:22:54 *g* Mar 04 20:22:59 gg Mar 04 20:23:08 hm glib is okay too Mar 04 20:23:14 makes some c life easier Mar 04 20:24:38 but as I said some time ago we will have a year with high expectations Mar 04 20:24:47 bb10 firefoxos ubuntuphone Mar 04 20:25:09 and sailfishos Mar 04 20:25:09 it all will end up with a browser as os Mar 04 20:25:13 I forgot Mar 04 20:25:23 aholler, i highly doubt that Mar 04 20:25:41 * woglinde wonders why cannonical didnot bought jolla Mar 04 20:25:59 because we need to build a space station :) Mar 04 20:26:04 <_av500_> woglinde: one word: RPM Mar 04 20:26:08 all this nice qt dev Mar 04 20:26:09 s Mar 04 20:26:12 and there is that Mar 04 20:26:49 but there is at least close collaboration going on so porting will be easy Mar 04 20:26:51 av500 you rip out packaging Mar 04 20:27:01 thats not hard Mar 04 20:27:03 <_av500_> woglinde: you define yourself by packaging Mar 04 20:27:12 <_av500_> no? Mar 04 20:27:18 eh? Mar 04 20:27:32 you put RPM into play Mar 04 20:27:35 <_av500_> yes Mar 04 20:27:39 well, to run in the ubuntu infrastructure it would have tio become deb based indeed Mar 04 20:27:42 <_av500_> because to some people THAT matters Mar 04 20:27:43 I talked about the real coding Mar 04 20:27:49 but re-packing isnt hard Mar 04 20:28:21 or did you meant redhat should buy jolla to make feodora phone? Mar 04 20:28:37 <_av500_> fedphone Mar 04 20:29:00 with built in wire-tap? Mar 04 20:43:39 bed time Mar 04 20:43:44 cya tomorrow trolls Mar 04 20:52:12 trolls don't sleep Mar 04 20:54:02 <_av500_> thus he is no troll Mar 04 20:54:33 quite Mar 04 20:55:06 is he lighter than a duck? Mar 04 20:55:20 that's witches Mar 04 20:55:29 if he's not a troll... Mar 04 20:58:19 gm mru Mar 04 20:58:32 how is hongkong? Mar 04 20:58:55 it got late at the bar... Mar 04 21:01:15 <_av500_> did I mention I hate the tiny reset on the BBB? Mar 04 21:01:32 _av500_: no, but we all do Mar 04 21:01:50 <_av500_> I might make a reset-cape Mar 04 21:01:54 is it worse than the on the regular bone? Mar 04 21:02:02 better Mar 04 21:02:06 why would anyone ever need reset? Mar 04 21:02:09 same Mar 04 21:02:15 better or worse Mar 04 21:02:19 ds2: some people write buggy code Mar 04 21:02:21 ups is here with my lian li Mar 04 21:02:27 stop writing buggy code! Mar 04 21:02:36 the reset switch is the punishment for buggy code Mar 04 21:02:46 thought about making a debug cape with serial, ina219's, etc Mar 04 21:02:50 ds2, I can't, I'm creating a bunch of blog entries to guide those new to bone Mar 04 21:02:54 need buggy code Mar 04 21:02:55 software access to reset Mar 04 21:03:00 * _av500_ loads a shotgun shell with tiny buttons Mar 04 21:03:03 then you need to be punished ;) Mar 04 21:03:16 btw, great quote from the bar tonight: "my kid only top-posted once" Mar 04 21:05:52 ouch Mar 04 21:07:27 better they learn early Mar 04 21:08:06 but should we teach them tolerance of top posters? Mar 04 21:08:15 certainly not! Mar 04 21:08:48 nice swiss people, where is koth Mar 04 21:09:03 they said no to fucking expensive olympic winter games Mar 04 21:09:17 * KotH hides Mar 04 21:09:18 smart people Mar 04 21:09:40 so at the moment there is nothing for 2018 Mar 04 21:09:42 we had our experiences with euro08.. and people didnt like it Mar 04 21:10:18 koth hm winter games are bit diffrent Mar 04 21:10:21 the logistics of bringing all the hookers in alone is a lot of work Mar 04 21:10:29 and more crowded usally Mar 04 21:10:37 mdp *g* Mar 04 21:10:47 mdp the could use parachutes Mar 04 21:10:52 heh Mar 04 21:11:09 woglinde: yeah.. but the image sticks... Mar 04 21:11:18 <_av500_> mdp: it also gets real cold outside in the evening Mar 04 21:11:41 woglinde: people dont like to get their wallets turned over by some capitalistic pick.. even if he is swiss himselv Mar 04 21:11:46 himself* Mar 04 21:12:18 koth yes yes Mar 04 21:12:33 KotH: you mean forced investment with no ROI is out of style? Mar 04 21:12:43 thats why I read this blog/side Mar 04 21:12:52 http://www.jensweinreich.de Mar 04 21:13:48 maybe they are just realistic forecasted that there will be no snow in 2018 Mar 04 21:13:58 -are Mar 04 21:14:31 hm ah no it was 2022 not 2018 is in south korea Mar 04 21:14:34 <_av500_> hmm, mainline uboot has no fastboot Mar 04 21:14:39 av500 yes Mar 04 21:14:41 woglinde: what really pissed people off was that bars, pubs and restaurants that didnt drink the fifa koolaid were forced to be walled off, so people so people from inside couldnt watch the live screenings of the matches shown on the street Mar 04 21:15:05 koth its the same with olympic comitee Mar 04 21:15:19 woglinde: or that the military and police had to provide security, but wasnt paid for Mar 04 21:15:35 woglinde: while fifa made billions...literaly Mar 04 21:15:56 koth yes yes same with olympic Mar 04 21:16:04 juup Mar 04 21:16:12 thats why people dont like it :) Mar 04 21:16:34 oh no it is jkridner Mar 04 21:16:44 run! Mar 04 21:16:49 web interface? is jkridner a noob today? Mar 04 21:16:50 koth problem is that most people only see it afterwards or have to pay for Mar 04 21:16:52 looks like panto already did. Mar 04 21:17:01 firewall :( Mar 04 21:17:04 jkridner gm Mar 04 21:17:38 jkridner: ssh on a non-standard port? Mar 04 21:17:38 woglinde: as i said, we had euro08 to teach people Mar 04 21:17:45 koth intressting davos and st. moritz vote for it Mar 04 21:17:47 jkridner: if firewall block your path, get a bigger dragon Mar 04 21:17:51 jkridner: Hi Mar 04 21:18:10 woglinde: towns that consist mostly of hotels.. they would make a lot of money Mar 04 21:18:51 *sigh* multiarch earlyprintk only for "uart1" Mar 04 21:18:54 more patches Mar 04 21:18:58 patch patch patch Mar 04 21:19:07 hack hack hack Mar 04 21:19:52 anybody here boot 3.9-rc1 on bone classic? Mar 04 21:20:07 koth yes right and expensive shopping malls at least in st. moritz Mar 04 21:20:21 mdp exact steps please Mar 04 21:20:23 *nod* Mar 04 21:20:37 woglinde: looking for someone to PM me stepz Mar 04 21:20:38 woglinde: the alps are already horribly expensive Mar 04 21:20:41 woglinde: even for swiss Mar 04 21:21:16 so enough offtopic Mar 04 21:21:25 off is on Mar 04 21:21:33 woglinde: no such thing Mar 04 21:21:36 and it's nearly friday already Mar 04 21:21:46 it's sooo friday Mar 04 21:23:20 bah, magic defconfig issue Mar 04 21:25:05 mdp haha Mar 04 21:31:27 -- Mar 04 21:31:29 When I heard they had elected a new Pope I was kind of hoping he'd be all "hey guys! Being gay is totally okay! Oh, also we are sorry for all the persecution and crusades and child molestation and stuff." I should have known that Emperor-Palpatine-lookin' motherfucker wouldn't do it, though. Mar 04 21:31:35 -- Mar 04 21:31:41 * KotH is reading old stuff...if anyone is asking Mar 04 21:35:22 whats wrong with sith? Mar 04 21:36:38 <_av500_> Tartarus: ping Mar 04 21:37:26 woglinde: the sparks and flashes Mar 04 21:40:30 <_av500_> hrm Mar 04 21:40:39 <_av500_> Error: unrecognized/unsupported machine ID (r1 = 0x00000e05). Mar 04 21:40:47 <_av500_> stock uboot on BBB Mar 04 21:42:00 voldemort, right Mar 04 21:42:01 ? Mar 04 21:42:34 <_av500_> yes Mar 04 21:42:38 Tartarus: ^^^^ Mar 04 21:42:43 A1 or A4? Mar 04 21:42:44 <_av500_> yes Mar 04 21:42:46 <_av500_> 4 Mar 04 21:43:03 <_av500_> I used the demo img from koen and just uboot-img from mainline Mar 04 21:43:20 Uh Mar 04 21:43:25 That's trying to boot a kernel? Mar 04 21:43:27 Or U-Boot? Mar 04 21:44:22 this PSU is still good, right? http://i.imgur.com/4EbuYO2.jpeg Mar 04 21:44:43 <_av500_> yes Mar 04 21:44:58 <_av500_> -> craigslist Mar 04 21:44:59 eh, I have such one too Mar 04 21:45:23 I think its getting ready to bloom Mar 04 21:45:32 just with closed capacitors Mar 04 21:45:46 <_av500_> Tartarus: trying to boot a kernel Mar 04 21:45:47 hehe Mar 04 21:46:00 <_av500_> Tartarus: from mainline uboot Mar 04 21:46:10 OK, did you pass an fdt? Mar 04 21:46:15 Or trying some crazy vendor tree? :) Mar 04 21:46:21 just fixed a friend's crappy LG TV with nice bloomin' caps like that Mar 04 21:46:44 <_av500_> Tartarus: I did make CROSS_COMPILE=arm-linux- am335x_evm Mar 04 21:47:05 _av500_: Right, and you get that error when trying to bootm a kernel right? Mar 04 21:47:09 <_av500_> yes Mar 04 21:47:10 *evil* vendor tree Mar 04 21:47:23 <_av500_> it was a booting system, the only thing I changed was uboot.img Mar 04 21:47:33 And not MLO? Mar 04 21:47:36 <_av500_> nope Mar 04 21:47:38 <_av500_> same MLO Mar 04 21:47:43 Try current? Mar 04 21:47:49 <_av500_> current? Mar 04 21:47:56 From the same build as u-boot Mar 04 21:48:12 <_av500_> I can Mar 04 21:48:19 <_av500_> but why? Mar 04 21:48:21 But, uh, same exact steps work with the older u-boot? Mar 04 21:48:25 <_av500_> yes Mar 04 21:48:29 <_av500_> from the bone demo image Mar 04 21:49:10 And you aren't relying on findfdt returning the right tree for that board are you? koen's patches are still outstanding Mar 04 21:49:38 <_av500_> I do not know Mar 04 21:49:56 OK, how do you boot normally? Mar 04 21:50:09 <_av500_> as I said, it was a demo image Mar 04 21:50:21 <_av500_> that had MLO+uboot+kenrel Mar 04 21:50:22 <_av500_> etc Mar 04 21:50:27 <_av500_> and I want to hack on uboot Mar 04 21:50:30 <_av500_> so I compiled one Mar 04 21:50:41 <_av500_> thinking mainline would be fine Mar 04 21:50:47 <_av500_> but it seems not Mar 04 21:50:54 lesson leaned, don't hack ;) Mar 04 21:51:32 It is fine Mar 04 21:51:53 bootcmd hasn't been updated for fdt Mar 04 21:52:45 <_av500_> I see findfdt in env Mar 04 21:52:48 Yes Mar 04 21:52:52 But it doesn't have.. Mar 04 21:52:58 bah, marked it as deferred Mar 04 21:53:02 but a check for bonelt Mar 04 21:53:05 <_av500_> ah yes Mar 04 21:53:14 <_av500_> no BOBONELT Mar 04 21:53:17 <_av500_> no BONELT Mar 04 21:54:29 Yeah, Koen patched it, but also did some related changes that need further updates Mar 04 21:54:33 <_av500_> Tartarus: where would I find said patches? Mar 04 21:55:10 http://patchwork.ozlabs.org/project/uboot/list/?submitter=10456&state=* Mar 04 21:55:19 v2 2/3, 3/3 Mar 04 21:55:51 <_av500_> so I dont need ext4? Mar 04 21:56:24 Well, if you're using koen's image I guess you need 1/3 too, yeah, sorry Mar 04 21:56:29 <_av500_> np Mar 04 21:56:53 <_av500_> Tartarus: another Q: do you now any mainline close uboot with fastboot? Mar 04 21:57:24 _av500_: http://patchwork.ozlabs.org/user/bundle/3913/?state=* Mar 04 21:57:35 Technical, not legal hurdles never addressed Mar 04 21:57:47 Pick it up, v2 it and lots of people will be happy Mar 04 21:58:09 <_av500_> that url asks me for a login Mar 04 21:58:18 ha Mar 04 21:58:19 sec Mar 04 21:58:25 http://patchwork.ozlabs.org/patch/126797/ Mar 04 21:58:30 http://patchwork.ozlabs.org/patch/126798/ Mar 04 22:00:07 <_av500_> nice Mar 04 22:00:14 <_av500_> I found rowboat with fastboot as well Mar 04 22:00:21 <_av500_> but the git log made me step back :) Mar 04 22:00:27 row row Mar 04 22:00:49 <_av500_> too many "cleaning" and "housekeeping" commits Mar 04 22:00:55 The "good" news is the fastboot stuff is contained to boards/ Mar 04 22:01:09 Or is, in one of the vendor trees I have seen exists Mar 04 22:01:25 why I want fastboot protocol on uboot? Mar 04 22:01:45 <_av500_> woglinde: because I dont want to connect a 3rd cable Mar 04 22:01:54 use dfu Mar 04 22:01:58 <_av500_> I have one usb for power and one for serial Mar 04 22:02:03 <_av500_> dfu is an option Mar 04 22:02:08 _av500_: Are you getting that error after it successfully loads a kernel image? And are you running 3.2? Mar 04 22:02:35 <_av500_> dwalsh_: I think it is resolved Mar 04 22:02:38 I found that after updating u-boot once I needed to specify the machine id through: env set machid 0x Mar 04 22:02:42 <_av500_> missing the BONELT fdt Mar 04 22:02:48 DFU isn't quite there for am335x Mar 04 22:02:57 <_av500_> ah Mar 04 22:03:22 <_av500_> well, anything that lets me download a kenrel without sd card dj action Mar 04 22:03:23 I've got nand patches on top of panto's stuff, but that needs addressing since it breaks mmc Mar 04 22:03:29 and for bonelt we need mmc to work Mar 04 22:03:39 <_av500_> I would be happy with RAM Mar 04 22:03:40 _av500_: usb networking is in there, out of the box Mar 04 22:03:55 <_av500_> hmm Mar 04 22:03:56 setenv ethact usb_ether Mar 04 22:04:03 <_av500_> oh Mar 04 22:04:07 I'll assume you've taken network manager outback and shot it Mar 04 22:04:16 (Well, if you don't setenv ethact it tries cpsw then usb) Mar 04 22:04:54 haha Mar 04 22:04:57 yes network Mar 04 22:06:19 _av500_: http://processors.wiki.ti.com/index.php/AMSDK_u-boot_User%27s_Guide#Booting_U-Boot_from_the_network has the rest of the "magic" for dhcp and /etc/network/interfaces for USB booting Mar 04 22:06:45 With the voldemort board you should be able to boot completely from USB, if no SD card is in and eMMC lacks valid MLO/U-Boot Mar 04 22:06:48 ti wiki usefull at least? Mar 04 22:07:00 woglinde: I can only vouch for the pages I edit Mar 04 22:07:38 <_av500_> Tartarus: thx Mar 04 22:07:54 uh danzig in the radio Mar 04 22:14:33 Tartarus, +1 on your pages ;) Mar 04 22:14:44 "Tom's Wiki" Mar 04 22:15:08 yes an not outdated uboot wiki side Mar 04 22:15:44 hm arago toolchain Mar 04 22:20:34 <_av500_> Tartarus: so am335x_evm_usbspl? Mar 04 22:20:51 yes, which might require a few more patches, sigh Mar 04 22:20:59 <_av500_> sigh Mar 04 22:21:16 <_av500_> actually only kernel from usb would be enough Mar 04 22:22:01 And that's fine mailine, really Mar 04 22:22:47 <_av500_> ok Mar 04 22:29:39 <_av500_> Tartarus: ok, boots with koen patches Mar 04 22:30:14 k Mar 04 22:33:10 why am I the only one that gets build errors when I don't have CONFIG_FS_FAT and CONFIG_FS_EXT4 enabled? Mar 04 22:33:29 apparently the overrides work for the rest of the world Mar 04 22:38:30 my longtime sensortag tester is ready: http://dpaste.de/LCehV/ ;) Mar 04 22:40:01 <_av500_> \o/ Mar 04 22:40:04 <_av500_> will it rain? Mar 04 22:40:27 I first have to put the tag outside ;) Mar 04 22:40:46 would be to nice if that temp aleady would have been outside Mar 04 22:43:00 ready for gnuplot: http://dpaste.de/EOT5o/ ;) Mar 04 22:43:27 gn Mar 04 22:43:45 mdp lol I read EXFAT Mar 04 22:44:49 aholler: congratz ;) Mar 04 22:45:06 * djlewis is still working on his Mar 04 22:45:18 easier than fiddling with 800 or 433mhz stuff Mar 04 22:46:09 i'm using 2.4G Mar 04 23:08:07 mru: trolls! Mar 04 23:12:14 <_av500_> romans, Mar 04 23:13:15 romantic trolls ? Mar 04 23:16:32 <_av500_> hmm, upstart hasa GUI now Mar 04 23:16:41 <_av500_> will systemd get one? Mar 04 23:17:16 <_av500_> hmmmm, mir draws the usual NIH rants fromthe usual crowd Mar 04 23:17:43 mir is not Linux! Mar 04 23:18:38 hehe Mar 04 23:19:30 eh? Mar 04 23:21:01 <_av500_> Crofton|work: ood one Mar 04 23:21:05 <_av500_> good even Mar 04 23:25:40 <_av500_> damn you blue leds Mar 04 23:27:12 _av500_: the board that doesn't exist? Mar 04 23:27:20 those are bright as hell Mar 04 23:28:20 http://www.thinkgeek.com/product/eeb6/ Mar 04 23:29:22 <_av500_> LOL: ...has someone forked emir yet? its been like a half a day.... Mar 04 23:29:42 <_av500_> mranostay: yes Mar 04 23:29:48 <_av500_> sharpie or solder... Mar 04 23:31:42 _av500_: black electrical tape Mar 04 23:36:40 <_av500_> I want them dimmed Mar 04 23:36:43 <_av500_> not off Mar 04 23:37:15 <_av500_> damn, the sharpie ink just evaporated off them Mar 04 23:38:19 <_av500_> ah, better now Mar 04 23:38:41 <_av500_> dimmed Mar 04 23:38:45 I hate blue LEDs Mar 04 23:38:52 <_av500_> alan_o: they hate you back Mar 04 23:39:08 They hated me first :) Mar 04 23:39:22 <_av500_> I guess I will swap one for orange, to have #tealandorange Mar 04 23:47:47 _av500_: no loving for blue LEDs? Mar 04 23:47:48 :D Mar 04 23:49:06 <_av500_> nope Mar 04 23:49:57 it is a conspiracy to drive people like us away Mar 04 23:50:00 those light up my closet Mar 04 23:50:09 ds2: along with no-NAND Mar 04 23:50:17 mranostay: there is NAND on the BBB Mar 04 23:50:46 those blue LEDs is another thing that gets F'ed if you try to run it on batteries Mar 04 23:50:59 <_av500_> ds2: it is managed NAND Mar 04 23:51:07 <_av500_> you do not like that! Mar 04 23:51:10 _av500_: iNAND :P Mar 04 23:51:15 <_av500_> hate it please! Mar 04 23:51:20 ds2: desolder them? Mar 04 23:51:26 or increase the current limit Mar 04 23:51:31 resistors Mar 04 23:51:42 <_av500_> in fact, remove the LED triggers and be done Mar 04 23:51:45 mranostay: that is an option Mar 04 23:51:47 <_av500_> nobody forces you to blink them Mar 04 23:51:56 change them back to green Mar 04 23:51:59 or red or amber Mar 04 23:52:07 damn 3.2V Vf Mar 04 23:52:09 <_av500_> or apricot Mar 04 23:52:19 <_av500_> mauve Mar 04 23:52:23 but I got better things to do Mar 04 23:52:34 like work around the damn erratas on that chip Mar 04 23:52:42 _av500_: done? Mar 04 23:52:47 and getting snapshotboot to work on it ;) Mar 04 23:52:48 make them rgb so people can choose Mar 04 23:53:19 <_av500_> +1 Mar 04 23:53:55 <_av500_> opensource is all about choice Mar 04 23:53:56 +1 to what? Mar 04 23:54:07 i like led-show of the lm4-launchpad ;) Mar 04 23:55:13 why have LEDs when you can have a LCD :D Mar 04 23:55:50 <_av500_> meh, have HDMI out to display a 3d rendering of a blinking blue led on a 50" TV Mar 04 23:56:15 problem is with this chip, that will turn out red ;) Mar 04 23:56:28 <_av500_> RLOD? Mar 04 23:56:33 errata Mar 04 23:56:45 so you better write code that blinks a red LED to get a blue one ;) Mar 04 23:56:52 or bit bang the 3d Mar 04 23:57:08 efuse it Mar 04 23:58:11 <_av500_> ds2: lets makeit a black LED Mar 04 23:58:14 <_av500_> for everybody Mar 04 23:58:21 <_av500_> less choice is better Mar 04 23:58:43 ir-led Mar 04 23:59:15 bbbbgone Mar 05 00:01:23 yes, I like black LEDs Mar 05 00:02:05 bitbang vga? Mar 05 01:03:56 well, thats sad. Mar 05 01:08:31 troll troll troll Mar 05 01:09:05 i just have sent back my $35,000 PCB milling system Mar 05 01:09:15 and they are refunding my money Mar 05 01:09:27 very annoying. Mar 05 01:12:19 ah, that was woglinde, never mind... Mar 05 01:13:41 I thought Mir was son of something related to Tizen? Mar 05 01:26:01 what was wrong with it? Mar 05 01:27:51 mru: XorA|gone owes you a beer Mar 05 01:39:36 rofl, if you google for "corporate takeover of open source linaro" the first hit is the linaro career page Mar 05 01:40:34 http://www.mail-archive.com/linux-omap@vger.kernel.org/msg85510.html Mar 05 01:42:28 heh Mar 05 01:46:01 Crofton|work: hehe Mar 05 01:50:24 corporate overlords pay well i'm told Mar 05 01:52:37 mranostay: corporate overlords feed well ;-) Mar 05 01:53:16 we are all !@#@!#$s for money :) Mar 05 02:18:33 Crofton, rmk seems to be a foul mood from time to time Mar 05 02:18:39 sure Mar 05 02:19:00 but the google query is amusing Mar 05 02:19:26 I just listened to Corbets Linaro Connect keynote. It is interesting Mar 05 02:36:04 koen: ping Mar 05 02:40:32 the little "sleds" for the beagl are quite good Mar 05 02:40:40 keeps everything off the desk Mar 05 02:43:28 hmm...wonder how hard it would be to make the abs slightly conductive Mar 05 02:43:51 Crofton|work: the LED needs a heatsink? Mar 05 02:44:13 because really, plastic can really build up a nice static charge Mar 05 02:45:26 you feed it 30 volts at up to 1A Mar 05 02:46:23 has anyone had an SD-Card fail, in a way that pulls the Power Supply of a beaglebone done? Mar 05 02:46:29 down Mar 05 02:49:11 mranostay, once I get it hooled up on the bench, I'll have a better idea of the heat situation Mar 05 02:49:46 Crofton|work: what are you using it for exactly? Mar 05 02:57:25 amusing myself Mar 05 02:57:36 https://fbcdn-sphotos-b-a.akamaihd.net/hphotos-ak-prn1/r270/421785_10151480179392661_748751452_n.jpg **** ENDING LOGGING AT Tue Mar 05 02:59:58 2013