**** BEGIN LOGGING AT Fri Oct 09 02:59:59 2015 Oct 09 04:08:43 Any clue on interfacing beagleboard-xm with WL18xx(WLAN)? Oct 09 04:09:33 Any clue on interfacing beagleboard-xm with WL18xx(WLAN)? Oct 09 04:17:46 hi Oct 09 04:18:45 hayb jon Oct 09 08:05:56 nyt: have you tried a 4.1 -ti kernel? Oct 09 08:07:10 nyt: the -bone branch is prone to fuck up the timing calculation (depending on video mode), that got rewritten in the -ti branch but hasn't been mainlined yet Oct 09 08:08:15 though that would result in consistent failure though, sounds like yours is stochastic? Oct 09 08:09:30 also, it's worth mentioning that a reboot never resets the hdmi framer, you need to poweroff and on again for that Oct 09 08:10:41 so if one kernel meddles with some particular setting of the hdmi framer while the other does not, then you'll get some interesting blend of settings by rebooting into the second kernel Oct 09 08:11:46 (unless the driver performs a soft-reset of the framer, haven't checked that, nor am I sure whether that actually sets the hdmi framer's registers or merely the functional logic) Oct 09 08:14:27 yeh i was using the ti kernel Oct 09 08:14:28 btw Oct 09 08:14:30 (never mind, it doesn't have soft-reset other than for audio and i2c-master) Oct 09 08:14:31 i figured it out Oct 09 08:14:31 ok Oct 09 08:14:36 if i boot with 3.8 Oct 09 08:14:39 then install 4.1 and reboot Oct 09 08:14:39 its good Oct 09 08:14:44 if i power off and power back on Oct 09 08:14:45 no good Oct 09 08:14:52 so it needs to boot with 3.8 first before 4.1 works Oct 09 08:14:55 ok, so 3.8 fixes some hdmi framer configuratio Oct 09 08:14:56 i dont know why but that's the dela Oct 09 08:14:59 yea Oct 09 08:15:06 that should be easy enough to locate Oct 09 08:15:08 its up on 3.8 right now at 720p50 Oct 09 08:15:11 :/ Oct 09 08:15:14 on 4.1 it was 720p60 Oct 09 08:15:23 when it came up Oct 09 08:15:31 I've heard someone also did work on the hdmi framer Oct 09 08:15:44 possibly not all work is good work :P Oct 09 08:16:15 gnna try 4.1.10-ti-r21 Oct 09 08:17:01 i also see linux-image-4.1.10-ti-git20151008-4b5cc48a Oct 09 08:17:50 that's... very fresh Oct 09 08:17:57 I wonder why rcn-ee made a git release Oct 09 08:18:52 I'm checking git log of the driver Oct 09 08:19:32 where is the ti gitt tree? Oct 09 08:21:31 you can view the source tree corresponding to each kernel version released by rcn-ee here -> https://github.com/RobertCNelson/linux-stable-rcn-ee/releases Oct 09 08:22:11 oh -r21 is fresher Oct 09 08:22:32 indeed Oct 09 08:25:42 ah, the git snapshot is missing all the usual patches Oct 09 08:25:48 (e.g. bone_capemgr) Oct 09 08:26:17 no joy on -r21 Oct 09 08:26:47 so yeah Oct 09 08:26:48 ok, so either 3.8-bone has a specific patch that fixes it, or 4.1 broke it Oct 09 08:26:52 theres some hdmi fuckery in 4.1 Oct 09 08:26:55 :( Oct 09 08:27:03 rolling back and rebuilding all my dtcs lol Oct 09 08:27:07 or missing hdmi-fuckery Oct 09 08:27:28 its weird 3.8 comes up w/ 720p at 50hz Oct 09 08:27:32 but when 4.1 comes up its proper at 60 Oct 09 08:27:42 edid handling changed Oct 09 08:28:07 yeah Oct 09 08:28:10 i noticed 4.1 was much better Oct 09 08:28:16 ill try forcing it Oct 09 08:28:27 https://github.com/RobertCNelson/linux-stable-rcn-ee/commit/07259f8b9fbd30fcef595e872deeea5ffab934ed Oct 09 08:30:17 lots Oct 09 08:31:08 or we need to diff the actual hdmi framer config somehow Oct 09 08:31:27 between non-working-4.1 and working-4.1 Oct 09 08:31:44 forcing 1920x1080@24 no joy Oct 09 08:31:53 which 4.1 is working? Oct 09 08:31:58 after rebooting from 3.8 Oct 09 08:32:04 that diff is for 4.3 Oct 09 08:32:17 oh Oct 09 08:32:18 no i see Oct 09 08:32:19 lots of patches Oct 09 08:32:24 it's in the history for 4.1-ti Oct 09 08:32:33 the diff between 3.8 and 4.1 is... big Oct 09 08:32:46 that patch is old Oct 09 08:32:48 might not even be that Oct 09 08:32:51 might be Oct 09 08:33:00 no I'm staring at the actual diff right now Oct 09 08:33:11 whtas weird is Oct 09 08:33:32 definitely lots of fixes and improvements Oct 09 08:33:41 so what o what did they break... Oct 09 08:34:20 root@backupbone:~# xrandr Oct 09 08:34:20 Screen 0: minimum 320 x 200, current 1280 x 1024, maximum 2048 x 2048 Oct 09 08:34:20 HDMI-0 connected 1280x1024+0+0 (normal left inverted right x axis y axis) 1209mm x 680mm Oct 09 08:34:20 1280x720 60.0 + 50.0 59.9 Oct 09 08:34:20 1920x1080 30.0 25.0 24.0 30.0 24.0 24.0 Oct 09 08:34:22 1280x1024 60.0* Oct 09 08:34:23 1440x900 59.9 Oct 09 08:34:24 1280x800 59.9 Oct 09 08:34:27 1152x864 75.0 Oct 09 08:34:28 1024x768 75.1 70.1 60.0 Oct 09 08:34:31 832x624 74.6 Oct 09 08:34:32 please use pastebin or such Oct 09 08:34:33 800x600 72.2 75.0 60.3 Oct 09 08:34:34 720x576 50.0 Oct 09 08:34:37 720x480 60.0 59.9 Oct 09 08:34:38 640x480 75.0 72.8 66.7 60.0 59.9 Oct 09 08:34:41 720x400 70.1 Oct 09 08:34:44 i can see the edid shit Oct 09 08:34:47 im justnot getting any joy Oct 09 08:34:51 sorry Oct 09 08:34:53 don't paste anything that's more than 1 or 2 lines into the chat Oct 09 08:34:55 forcing modes does nothing Oct 09 08:36:08 will try reseating hdmi while live Oct 09 08:36:10 you did try making sure it's not in display sleep? (e.g. press a key on a keyboard) Oct 09 08:36:53 yeah Oct 09 08:36:59 i also tried using xset Oct 09 08:37:03 and turning it on and off with dpms Oct 09 08:37:11 though still, if it works on reboot from 3.8 but not on cold boot in 4.1 or reboot from 4.1 then the only thing I can think of is a hdmi framer setting Oct 09 08:37:31 since it has no reset input Oct 09 08:37:58 but then I'd need to find a setting that's written by 3.8 and not overwritten by 4.1 Oct 09 08:38:55 can you try bisecting? i.e. try intermediate kernel versions to see when they broke it? Oct 09 08:39:06 like, does 3.12 or 3.14 work ? Oct 09 08:39:32 cant test atm Oct 09 08:39:41 3.14 -ti is still being maintained Oct 09 08:39:59 so that would be a good try Oct 09 08:41:31 k Oct 09 08:41:32 sec Oct 09 08:42:31 installing linux-image-3.14.54-ti-r77 Oct 09 08:45:05 also im pretty sure this worked on rev b ok Oct 09 08:45:13 but i cant be certain Oct 09 08:45:27 might have been an upgrade after booting 3.8 Oct 09 08:45:40 that would be my guess since there's no relevant hw change Oct 09 08:45:46 yea Oct 09 08:46:55 no joy there Oct 09 08:47:16 putting 3.8 back on for now and getting back to work Oct 09 08:50:25 what package is responsible for /lib/firmware? Oct 09 08:50:29 no idea Oct 09 08:50:38 will find it when this is done installing Oct 09 08:51:37 root@backupbone:/lib/firmware# dpkg -S BB-SPIDEV1-00A0.dtbo Oct 09 08:51:38 dpkg-query: no path found matching pattern *BB-SPIDEV1-00A0.dtbo* Oct 09 08:51:38 damn Oct 09 08:51:51 oh i can build em Oct 09 08:51:54 just gotta remove new dtc Oct 09 08:57:41 aaaaaaaaaand my bbb just roasted itself wtffffffffffff Oct 09 08:57:53 ?? Oct 09 08:58:00 has some io hooked up something might not be right Oct 09 08:58:06 maybe an external ps leaked to it Oct 09 08:58:16 its only hooked up to some npn transistors tho so Oct 09 08:58:17 i dunno Oct 09 08:58:24 really odd Oct 09 08:58:56 but roasted as in... it won't power on anymore? Oct 09 08:59:33 right Oct 09 08:59:38 power led blips when i hook up the ps Oct 09 08:59:40 thats it Oct 09 08:59:43 dead yes? Oct 09 08:59:55 the dreaded flashing power led of doom Oct 09 09:00:13 assuming there's no external stuff attached anymore, yeah that means it's officially time to RMA Oct 09 09:00:19 :/ Oct 09 09:00:23 wtff Oct 09 09:00:52 it means the PMIC sees an overcurrent/shortcircuit during power-on sequence Oct 09 09:01:10 hm Oct 09 09:01:19 not easily repairable i take it? Oct 09 09:01:45 if nothing is connected, that short-circuit is most likely internal to the CPU due to crispy-fried I/O Oct 09 09:01:53 ugh Oct 09 09:02:22 need one today in nyc Oct 09 09:02:25 dunno who will carry Oct 09 09:02:28 fuck Oct 09 09:08:10 language! Oct 09 09:08:34 I once talked with someone about the concept of an CAPE that basically just passes everything through, but with busswitches inserted that isolate all I/O when power is off or reset is asserted, and clip to power supply levels when not isolated Oct 09 09:08:43 and maybe some added ESD protection Oct 09 09:09:18 zmatt: a "nanny-cape" Oct 09 09:09:35 yep, plus it would make the I/O 5V-compatible Oct 09 09:09:58 it was actually on when it died so i dont even know why Oct 09 09:09:59 the only remaining worry would be overcurrent, other than that it would be reasonably abuse-proof Oct 09 09:10:02 it was running for hours Oct 09 09:10:21 nyt: then I'm thinking overcurrent rather than overvoltage Oct 09 09:10:29 dangit, my nanny-cape wouldn't have helped Oct 09 09:10:42 nyt: you did mention transistors... connected how exactly? Oct 09 09:10:52 bunch of 2n3904 Oct 09 09:11:04 bbb io pin -> transistor base Oct 09 09:11:16 without resistor? Oct 09 09:11:44 no resistor to npn Oct 09 09:11:58 oh wait no there are resistors Oct 09 09:12:02 my mistake Oct 09 09:12:14 15k ohm Oct 09 09:12:21 ok, so no worry there Oct 09 09:12:54 ground was shared between two power supplies though Oct 09 09:12:57 with no earth grond Oct 09 09:12:59 think that might be the issue Oct 09 09:13:10 it shouldn't be an issue Oct 09 09:13:57 so bbb ground was connected to ground of a dif 12v ps Oct 09 09:13:58 * av500 is happy that these days he works with $1.5 STM32F0 modules :) Oct 09 09:14:06 well a few of them actually Oct 09 09:14:06 cheap to burn Oct 09 09:14:07 if the power supplies are ungrounded, i.e. floating, then connecting their grounds together (properly) suffices to allow interaction between the two (subject to the usual power supply and I/O rules) Oct 09 09:14:14 since it's also connected to some h bridges Oct 09 09:14:23 but those are also via transistors with resistors Oct 09 09:15:02 and diodes Oct 09 09:15:39 ugh where to find one in asap at 9am :( Oct 09 09:30:40 nyt: ask on twitter Oct 09 09:31:46 what tag? Oct 09 09:33:00 hmm Oct 09 09:33:08 I dont use twitter :) Oct 09 09:33:12 but i heard its powerful Oct 09 09:36:02 nyt: https://plus.google.com/u/0/+VladimirPantelic/posts/BcvVZsQVjwn Oct 09 09:40:54 lol <3 Oct 09 09:41:29 trying to figure out what could have fried this Oct 09 09:42:15 vdd 3v3 to 4133a hall effect sensor input, with 15k ohm resistor to the sensor line as pullup, and ground back to dgnd Oct 09 09:42:38 could that have fried the chip? Oct 09 09:42:48 i assume that would just tax the vreg if anything Oct 09 09:43:03 hm but the ground to that board and voltage wasnt even plugged in Oct 09 09:43:29 though that board did have a ground hooked up to some other h bridge boards that shared the ground with the 4133a Oct 09 09:43:36 and the 4133a signal wires were still going back to the bbb Oct 09 09:44:23 everything else was hooked up to a board driving leds with 12v ps and some 2n3904s Oct 09 09:44:29 the SoC seems to be quite sensitive to seeing any voltage on pins while it's not powered Oct 09 09:44:38 it was powered on and running Oct 09 09:44:46 nyt: bookmark the link in case somebody responds :) Oct 09 09:44:57 i did Oct 09 09:45:04 k Oct 09 09:45:11 ty Oct 09 09:45:30 going to add some resistors to those hall effect sensor inputs I guess Oct 09 09:46:43 could it cause issues if i dont load capes for some of the pins? Oct 09 09:46:51 since i was messing with kernel stuff i wasnt loading any of the dtbos Oct 09 10:02:07 not really Oct 09 10:02:29 my only worry from your vague description is whether or not all PSUs were properly grounded together Oct 09 10:03:11 if not, exactly what their potentials were, especially with a 12V PSU in the mix Oct 09 10:03:21 pretty sure they were grounded Oct 09 10:03:24 i think the issue was some hall effect sensors Oct 09 10:03:30 that i had hooked up w/out resistors to the input pins Oct 09 10:03:51 unless those pins were configured as output that shouldn't be a problem Oct 09 10:07:03 they weren't configured as anything at the time Oct 09 10:07:07 didnt load a cape Oct 09 10:07:28 there was about 56-100mv of voltage on the outputs when i measured after Oct 09 10:13:42 that doesn't sound like the problem then Oct 09 10:14:04 I don't know, it happened to us too once, we never figured out the cause Oct 09 10:15:23 also, nice job ti... I wanted to look up the V/I curve for the output drivers in the IBIS data. After double-checking how to interpret the data, this is the result: Oct 09 10:15:28 http://gerbil.xs4all.nl/vi-curve.png Oct 09 10:16:19 hence we learn, that if you drive an output high, and its level is in fact high, then negligible current flows. On the other hand, if you forcibly raise the voltage more than a diode-drop above vdd, then much current will in fact flow inward! Oct 09 10:16:39 that's really the data I was looking for, thanks ti Oct 09 10:25:44 ah wait, nm Oct 09 10:26:50 got distracted by thinking pullup/pulldown meant pullup/pulldown Oct 09 10:27:06 while they meant drive high/low Oct 09 10:32:57 i just checked all my stuff, everything except the hall effect sensors was hooked up with 5k ohm resistors or hooked to switches that were open Oct 09 10:34:10 IIRC the wiki mentions one mystery situation that makes boards die like this. Don't remember details Oct 09 10:36:09 isn't a hall sensor a weak output anyway? Oct 09 10:36:44 looking closer Oct 09 10:36:49 the board it was connected to had a disconnected ground Oct 09 10:37:01 so they were getting ground from an external ps Oct 09 10:37:10 then the output went to the bbb Oct 09 10:37:28 dunno if enough current could have leaked through there to hurt it or not Oct 09 10:37:52 but like i said everything else is connected up to 5k ohm resistors or open switches with resistors Oct 09 10:37:56 hm Oct 09 10:38:07 if you set a pin to output Oct 09 10:38:11 and run it right to the bbb ground Oct 09 10:38:18 whats to prevent that from roasting the bbb Oct 09 10:39:31 "to the bbb ground" .. from what? you need a complete circuit Oct 09 10:40:35 if you mean drive current between different ground pins of the BBB, you'll mostly just make it warm Oct 09 10:41:36 if the driver is powered by a psu otherwise isolated from the BBB you'll drive its ground plane down compared to the BBB ground plane Oct 09 10:43:22 (seeing that being used to generate a negative supply rail can look really odd, with a switching power supply whose output is tied directly to ground... the ground of the power supply was actually the negative output rail) Oct 09 10:46:19 hm then i have no idea how this fried Oct 09 10:50:39 tbr: I just remembered we actually did find a likely candidate in our case... the adc powerup erratum combined with opamps (powered by vdd_adc) connected directly to analog inputs Oct 09 10:51:33 ah, not using any :/ Oct 09 10:51:41 all gpio Oct 09 11:01:25 on the bright side... the fractional quantum hall effect is fascinating Oct 09 11:01:29 (dangit wikipedia) Oct 09 11:15:56 Hey guys could you please try and help me with my "make" problem on the beaglebone black -> http://stackoverflow.com/questions/33036519/why-can-i-not-make-successfully-bbb-accesspoint-problems Oct 09 11:19:34 nyt: see your bookmakr Oct 09 11:28:30 word gonna call when they open Oct 09 11:28:33 also gonna call the local rat shacks Oct 09 11:28:46 aint they all closed? Oct 09 11:28:52 sold for parts Oct 09 11:28:55 boarded up Oct 09 11:29:35 nope Oct 09 11:29:43 full of vagrants Oct 09 11:58:10 Hey everyone the question has been upated -> http://stackoverflow.com/questions/33036519/why-can-i-not-make-successfully-bbb-accesspoint-problems/33036605#33036605 Oct 09 11:58:44 holy balls .. RS has bbb in stock! like .. Right Now ! Oct 09 12:05:19 where? Oct 09 12:05:50 RS Oct 09 12:05:52 online Oct 09 12:06:18 url? Oct 09 12:06:23 can't find when searching their site Oct 09 12:06:50 http://www.alliedelec.com/beagleboard-999-0001457-rev-c/70338169/ Oct 09 12:07:42 http://uk.rs-online.com/web/p/products/7753805/ Oct 09 12:07:46 oh rs-online Oct 09 12:07:48 thought radioshack Oct 09 12:07:48 lol Oct 09 12:08:21 I had a friend work next to RS warehouse Oct 09 12:08:25 so I had same day delivery Oct 09 12:08:31 order at noon, have it at 6 Oct 09 12:08:54 cool Oct 09 12:10:22 too bad he had to change jobs :( Oct 09 12:14:58 I can deal with next day .. even when they ship Sat :) Oct 09 12:15:25 just gotta separate work/home accounts atm .. :\ work doesn't like receiving personal invoices lol Oct 09 12:16:54 how many IOs are there available .. hmm .. elinux here we come .. Oct 09 12:17:14 stt_michael: there is always one IO less than you need :) Oct 09 12:19:25 +1 Oct 09 12:19:40 av500, this is true .. but the current configuration is already 2 short .. and the beagle would have many spare .. Oct 09 12:20:10 "every project evolves until it runs out of IOs" Oct 09 12:20:38 and then some .. Oct 09 12:20:43 yes Oct 09 12:20:43 or it never gets finished Oct 09 12:20:44 :D Oct 09 12:21:11 wanna see if I can get this rs485 thingy wrapped up Oct 09 12:21:29 just don't haev any 'play' hardware presently Oct 09 12:21:37 and some day I wanna play with the PRU Oct 09 12:22:12 before I start pushiin up the daisies hopefully .. Oct 09 12:33:05 right .. time for lunch Oct 09 12:33:08 bbiab! Oct 09 14:04:54 So, what prompted the use of uname boot rather than using symlinks? Oct 09 14:06:09 It seems to me it just made a very complicated solution to a simple problem. Oct 09 14:10:04 its the decade for that :) Oct 09 14:11:48 update-alternatives looks like an eligant solution. Oct 09 14:27:38 can anybody please try and help me with this ? -> http://stackoverflow.com/questions/33036519/why-can-i-not-make-successfully-bbb-accesspoint-problems/33036605#33036605 Oct 09 14:28:56 /usr/src/linux-headers-3.8.13-bone47/arch/armv7l/Makefile: No such file or directory Oct 09 14:29:00 do you have that file? Oct 09 14:29:21 no Oct 09 14:30:29 av500: I dont even have the arm7l file Oct 09 14:32:32 rcn-ee: Are you the guy who owns the website? Oct 09 14:32:35 rcn-ee? Oct 09 14:32:50 rcn-ee: I Could really use your help if that is the case -> http://stackoverflow.com/questions/33036519/why-can-i-not-make-successfully-bbb-accesspoint-problems/33036605#33036605 Oct 09 14:32:54 Cats_home, tis i i what's up.. Oct 09 14:33:10 ghez never expected you lol Oct 09 14:33:26 cart_man, answered both your emails' to the beagleboard group... Oct 09 14:33:57 rcn-ee: Ohh.. strange it did not send me a notice and I can not seem to find either...that is why Ive sent 2 : ( Oct 09 14:34:08 rcn-ee: I thought the forum was buggy...sorry about that Oct 09 14:34:19 rcn-ee: How do I get to my post...silly question I know Oct 09 14:34:46 use the button on the right "all topics" https://groups.google.com/forum/#!categories/beagleboard Oct 09 14:34:56 the "grouping" is a weird thing google does... Oct 09 14:35:12 most developers just read it as emails' and not as a forum... Oct 09 14:38:08 ohh shit ok so now the infamous timex.h problem occured.. which is a good sign but anybody know how I can fix it? Oct 09 14:38:25 rcn-ee: ^^ Oct 09 14:38:53 well it's 3.8.x.. linux-headers was majorly broken back then.. Oct 09 14:39:20 just grab the latest jessie iamge with "v4.1.x" cape manager is incluced... Oct 09 14:45:02 * zmatt growls at Gerard Oct 09 14:51:25 rcn-ee: Seems like its working.. will let you know!^^ Oct 09 15:00:56 zmatt: :) Oct 09 15:01:02 zmatt: duino? Oct 09 15:01:28 yeah, that remark was just flamebait Oct 09 15:02:19 :) Oct 09 15:02:37 (ask, and ye shall receive) Oct 09 15:03:32 dont make me recite the code of conduct! Oct 09 15:04:40 as a dramatic reading? Oct 09 15:05:09 amdrams tonight? Oct 09 15:08:53 I think I adequately extended the peace pipe again with my reply Oct 09 15:15:36 ja Oct 09 15:15:38 all is fine Oct 09 15:17:45 well not all is fine! there are so many wrongs in this world! ;) Oct 09 15:18:13 and I still need to do shopping, which involves high-effort stuff like getting dressed and leaving the house Oct 09 15:18:17 one for every day Oct 09 15:18:54 ooh, this one exploded pretty cool => https://en.wikipedia.org/wiki/File:Exploded_Electrolytic_Capacitor.jpg Oct 09 15:24:02 nyt: see your bookmark :) Oct 09 15:30:11 nyt: btw, reading back at what you said, the hall sensor was also powered via a resistor large enough to prevent any significant current flowing (unless the BBB pin was configured as output and driven high, which I think your hall sensor also would not have appreciated) Oct 09 16:10:29 rcn Oct 09 16:10:30 here/ Oct 09 16:10:37 found out what was up w/ 4.1 and 3.8 hdmi sort of Oct 09 16:14:18 nyt, what's up? Oct 09 16:28:59 so Oct 09 16:29:06 power off, boot with 3.8, get hdmi display Oct 09 16:29:10 upgrade to 4.1 Oct 09 16:29:14 still have hdmi working Oct 09 16:29:22 power off and back on Oct 09 16:29:24 hdmi no longer works Oct 09 16:36:43 its hdmi framer fun! yay! Oct 09 16:42:09 rcn-ee: in other words, 3.8 kernel is doing something with the (reset-insensitive) hdmi framer that 4.1 neglects to do Oct 09 16:42:37 zmatt, but only on rev c... nyt's got a rev b that has no issues. ;) Oct 09 16:43:17 he admitted he wasn't sure whether 4.1 was cold-booted there Oct 09 16:43:31 or warm-rebooted from a 3.8 kernel Oct 09 16:44:26 nyt: did you get a chance to try 3.14 btw? Oct 09 16:44:49 rcn-ee: given the amount of changes, bisecting seemed a good idea Oct 09 16:45:13 minus the 1k patch's on top of 3.8. ;) Oct 09 16:46:09 rcn-ee: is there a hdmi-unpatched 3.8 kernel in your debian repo for comparison? Oct 09 16:46:43 those patches weren't integrated with nor ported to 4.1 ? Oct 09 16:46:44 nyt's porting the "gem beagleboard" from 3.8-capemanager to 4.1-capemanager.. Oct 09 16:47:18 I have no idea what that means Oct 09 16:49:50 zmatt, oh i see... nope.. there is no un-patched 3.8... we didn't get mmc support on the am335x till like 3.12... Oct 09 16:50:12 ok Oct 09 16:50:16 most of the tilcd went in asis.. but rmk's been in charge of it ever sense... Oct 09 16:50:53 (and he's only working on it for armada, not sure he even has a bbb for verification..) Oct 09 16:52:56 I've been looking at the diff, most of the changes look like clear-cut improvements Oct 09 16:53:33 yeah he's been doing a nice job cleaning it up.. Oct 09 16:58:20 one thought I had was to dump the framer's settings both in non-working-4.1 and in working-4.1 (i.e. after reboot from 3.8) Oct 09 16:58:25 and diff them Oct 09 16:58:51 that may be easier than trying to locate the difference in code, due to the heavy restructuring Oct 09 16:59:44 probably Oct 09 16:59:52 why does nothing reset properly?! lol Oct 09 17:00:10 the hdmi framer has no reset input Oct 09 17:00:27 nor a way to fully soft-reset it Oct 09 17:00:42 you'd need to power-cycle it Oct 09 17:01:21 sounds like the same problem with the broadcom's on the wandboard .. although someone managed to leverage a gpio for that Oct 09 17:01:29 +1 to them Oct 09 17:01:32 well... the reset driver needs fixes... https://github.com/t-kristo/linux-pm/commits/4.3-rc3-prcm-reset-fwk-fixes Oct 09 17:01:45 job for zmatt .. ;P lol Oct 09 17:02:29 rcn-ee: that's about soc-internal resets though :P Oct 09 17:03:06 rcn-ee: the rev b might have been upgraded from 3.8 to 4.1 without ever power cycling, so can't say for certain its rev b related, i think its both and just a kernel issue Oct 09 17:04:37 nyt: I can see if I can figure out an easy way to dump the registers... though I guess I have some time since you fried your bbb anyway :P Oct 09 17:04:46 nyt, it's probally a kernel issue.. but can you double check: "sudo apt-get install linux-image-3.14.54-ti-r77" (note that v3.14.x doesn't have capemanager..) but then we can get an idea of where it broke.. Oct 09 17:05:21 did you source a new beagle yet, nyt? Oct 09 17:05:25 ti pretty much moved their tilcd/tda99x hacks from 3.14-ti to 4.1-ti Oct 09 17:05:47 at least someone was hacking it, though, rcn-ee ... Oct 09 17:06:14 stt_michael, that's why we are using it as base. ;) Oct 09 17:08:42 i tried 3.14.54 Oct 09 17:08:43 no joy Oct 09 17:08:50 power cycled with it, same issue Oct 09 17:08:59 yeah i went to crooklyn Oct 09 17:09:01 a rat shack had 2 Oct 09 17:09:04 the one by my house has 8 lol Oct 09 17:09:07 but thast 2 hours from where i am now Oct 09 17:29:35 :) Oct 09 17:32:30 nyt: do you have a serial cable btw? Oct 09 17:33:13 since the easiest way I can think of to dump the framer's registers is by doing so from u-boot Oct 09 17:33:44 since you can't really safely interact with the framer while the driver is loaded Oct 09 17:34:46 hmm, of course one can unload it and I think there are userland utilities for doing i2c Oct 09 17:34:54 ok, scrap the u-boot plan Oct 09 17:35:03 but first, I need to do shopping, so afk for a bit :) Oct 09 17:47:53 Who here is still using the original beagleboard? Oct 09 18:00:03 I am. Oct 09 18:00:08 best thing ever. Oct 09 18:01:08 zmatt: not here Oct 09 18:01:19 * tbr still has one in the drawer, and some other OMAP3 and OMAP4 devices Oct 09 18:01:34 all 3 generations of OMAP3 actually Oct 09 18:03:27 3 generations of omap3? Oct 09 18:04:00 I thought it was just omap34xx/35xx and omap36xx/dm37xx Oct 09 18:04:24 the one on the bb classic, the one on the bb xm and the one predating the bb classic Oct 09 18:04:47 that last one was something very early, but it made it into a product Oct 09 18:05:05 the 128Mb omap3/beagle's had a fun es omap3 core... Oct 09 18:07:06 would need to ask _av500_ to remind me what they used on gen6. I remember it had SGX disabled or rather broken. Oct 09 18:08:17 ohh wait omap34xx/35xx had a silicon revision major enough to warrent a new jtag id Oct 09 18:08:31 I guess that constitutes a separate generation of omap3 then indeed Oct 09 18:13:37 1.0 had a jtag id even lower than the dm64xx and dm33x/dm35x Oct 09 18:14:24 and it had a lot of revisions on the new id... 2.0 (jtag rev 1), 2.1 (2), 3.0 (3), 3.1 (4), 3.1.1 (5), 3.1.2 (7) Oct 09 18:14:32 very logical Oct 09 18:15:38 tbr: funny that the omap4 is probably more obsolete than the omap3 now Oct 09 18:15:47 yeah Oct 09 18:17:27 one nail in the coffin being that version 8 of the C6000 compiler dropped support for some "obsolete" DSP versions, and that included Tesla that's not only on the omap4 but also still on the omap5 Oct 09 18:19:50 omap4 has a c64+ which is supported Oct 09 18:19:51 eh Oct 09 18:19:51 omap3 Oct 09 18:21:27 tbr: there's also the am35xx which could be considered an omap3 generation... although I guess it's more an omap3-derivative Oct 09 18:21:39 omap3 was also pre-remoteproc... so it's dsp really only works in ancient kernels.. Oct 09 18:22:51 yeah O4 had half the size of DSP IIRC Oct 09 18:23:03 rcn-ee: the dsp works fine in recent kernels, getting it running is a simple matter of releasing the subsystem from reset, write code to its memory, write start address to a register (if the default doesn't suit you), release core from reset... and it's up Oct 09 18:23:13 no need to fancy kernel support to do that Oct 09 18:23:52 zmatt should be able to con remoteproc to do that. ;) Oct 09 18:24:36 tbr: yeah like a c64+ core hacked down them middle and glued onto another (but differently configured) instance of that same wierdass unicache/ammu thing that's in ducati/benelli Oct 09 18:24:51 tbr: there's not even an instruction set manual for it Oct 09 18:24:55 *public Oct 09 18:25:27 rcn-ee: dunno, I know absolutely nothing about remoteproc :) Oct 09 18:25:39 never used it, never looked at it Oct 09 18:26:09 me neither, i know it's used for the m4/dsp/pru's on the x15... Oct 09 18:28:47 Mmmmm remote proc Oct 09 18:30:13 rcn-ee: but then again I don't understand uio_pruss either.. afaict you can get equivalent functionality with uio_pdrv_genirq Oct 09 18:33:58 ok what's the latest kernel with SGX support on the original beagleboard? Oct 09 18:35:20 or, rather, is am335x (bbb) the only platform supported by TI's SGX and a recent-ish kernel at the moment? Oct 09 18:39:06 tuna_, well the next sgx release will be only, am33x/am4x/am5x.... Oct 09 18:39:27 and the am4x is very close to the omap4, so panda people might be happy... Oct 09 18:39:48 the am4x is closer to the am335x than it is to the omap4 Oct 09 18:40:08 afaik they only have the cortex-a9 in common (which isn't really relevant for friver support) Oct 09 18:40:17 *driver Oct 09 18:40:18 rcn-ee: I see. I guess it's time to retire omap3/4 for good if i need to get anything done with SGX Oct 09 18:40:38 last i looked, didn't the am4 also have dss2? so it has the same video hardware as the omap4.. Oct 09 18:40:55 rcn-ee: ahh, wait you may have a point there Oct 09 18:41:21 i see Oct 09 18:41:43 ok, time for phase 2/2 of shopping Oct 09 18:41:45 so at the moment, there is no love for omap3/4 in a recent version of SGX drivers Oct 09 18:41:54 that's why they are doing dmabuf... the am4/am5 have dss2 video, am3 has tilcd.. so dmabuf is nneed for sgx.. Oct 09 18:42:49 nyt: btw, dumping the framer's registers should be easy if anyone here knows how to do raw i2c from userspace... it behaves like a 256-byte eeprom (1-byte address), though it has multiple of such 256-byte pages: you select the page number by writing it to byte 0xff Oct 09 18:42:57 my notes on its register set: https://docs.google.com/spreadsheets/d/1Dxfjxc-T2hmnegCRblbsVK2kDqOs_0Yf4f5nbMCleuI/view Oct 09 18:43:14 tuna_, for omap3, our only real hope is the omap3-nokia.. those guys have been fixing up lots of omap3 issues on mainline.. ;) Oct 09 18:43:24 (n900) Oct 09 18:43:27 quite a few regs are said to be write-only... I guess you'll find out whether it truly is Oct 09 18:43:39 rcn-ee: good lead. I will check out omap3-nokia Oct 09 18:43:46 thanks Oct 09 18:44:11 but they also have the ancient nokia userspace... so they are still using the old sgx drivers with a new kernel.. Oct 09 18:44:22 nyt: probably a good idea to unload the tda998x driver first, but you may get away with dumping the regs while it's up Oct 09 18:44:30 anyway, brb Oct 09 18:45:42 "old sgx drivers" as in even older than the latest omap3-supported SGX driver? I'd expect the last omap3-supported SGX driver is pretty old by now Oct 09 18:45:45 ill have to revisit soonish Oct 09 18:45:50 cant do anything now Oct 09 18:45:56 but i mean its easy to test anyone can plug it in via hdmi Oct 09 18:46:38 nyt: hdmi works fine for me in 4.1, so it's presumably a specific interaction with your display Oct 09 18:47:43 (on second thought, phase 2/2 of shopping isn't really urgent, can defer till tomorrow) Oct 09 19:40:31 <_av500_> nyt: dont burn the new one too! Oct 09 19:50:31 hmm, c64+ and later DSPs have a single-cycle instruction for INTERCAL's mingle operator ($) Oct 09 19:51:37 I'm trying to install the drivers for beagle bone black on my computer, but every time I do, i get told the installation failed Oct 09 19:51:52 And I have no clue what's going on Oct 09 19:52:24 Caroline_: which windows version? Oct 09 19:53:12 windows 8 Oct 09 19:53:17 https://learn.sparkfun.com/tutorials/disabling-driver-signature-on-windows-8 Oct 09 19:53:17 64 bit Oct 09 20:07:22 Where can I find the layout of the EEPROM on the beaglebone itself? Oct 09 20:10:30 abferm, making a clone? Oct 09 20:25:29 isn't there a FSF project to reverse engineer the SGX drivers? Oct 09 20:27:31 ds2, that would be sweet Oct 09 20:27:50 ds2, it's the same one they annouced when the first beagle came out.. 6+ years ago... Oct 09 20:27:53 to get rid of the binary virus and be able to recompile against any kernel Oct 09 20:29:03 rob_w, the kernel modules actually haven't been bad.. https://github.com/RobertCNelson/ti-sdk-pvr Oct 09 20:29:03 i dont get why such companys keep this things binary secrets .. Oct 09 20:29:04 rcn-ee: no, I just want to grab serial number, etc. and use them for some operations. Oct 09 20:29:45 rob_w, it's imgtech... that keeps thing secret.. they dont' want anyone to know how crappy sgx is.. Oct 09 20:29:53 abferm, there are hints in the u-boot source Oct 09 20:29:59 mostly making sure my script is actually running on a beaglebone, not some other board. Oct 09 20:30:44 abferm, it varies for 3.8(eeprom)/4.1.x(early-nvmem)4.1.x(upstream-nvmem) Oct 09 20:30:47 crappy binary sgx < crappy freeed source sgx Oct 09 20:31:21 abferm, https://github.com/RobertCNelson/boot-scripts/blob/master/device/bone/tester/show-eeprom.sh Oct 09 20:32:51 rob_w, it's even worse than taht.. when we "had" x11 drivers, "fbdev" using neon was still faster... Oct 09 20:33:54 rcn-ee: looking for the data structure layout Oct 09 20:34:40 yeah .. and now x11 is dropped, right ? Oct 09 20:34:50 abferm, https://github.com/RobertCNelson/u-boot/blob/master/board/ti/am335x/board.h#L23 Oct 09 20:35:39 rob_w, we had it for omap3/dm37xx parts years ago... never got it for am33xx.. (omap4 -> omap5 had it), but they dropped it in the sgx repo.. Oct 09 20:36:31 rob_w, just watch http://git.ti.com/gitweb/?p=graphics/omap5-sgx-ddk-um-linux.git;a=summary Oct 09 20:36:33 which bugs me really.. we just finished a new am437x layout .. and now i dont have basic 3d in x11 Oct 09 20:37:11 use egl/wayland... Oct 09 20:37:17 or qt's eglfs... Oct 09 20:37:20 why oh why Oct 09 20:37:39 <_av500_> because Oct 09 20:37:50 do you know what ti means by a "null window" manager ? Oct 09 20:38:01 this is the term they use in the graphics sdk Oct 09 20:38:02 rob_w, no idea... we just pick up the scraps we get from ti and make things sorta work. ;) Oct 09 20:38:15 rob_w, i think it's dmabuf type apstraction.. Oct 09 20:38:17 <_av500_> rob_w: rememeber that rasterman of EFL was doing 2D faster in SW than SGX on omap3 :) Oct 09 20:38:41 hmm what a mess Oct 09 20:39:06 preching to the choir.. Oct 09 20:39:24 <_av500_> +1^111 Oct 09 20:39:37 <_av500_> rob_w: simply use android like normal people Oct 09 20:39:46 yeah right Oct 09 20:40:21 i will just rot in the corner and maintain my x11 apps till my end Oct 09 20:40:32 <_av500_> yes Oct 09 20:42:37 <_av500_> rcn-ee: why exaclty is the sgx driver linked to the type of window manager? Oct 09 20:42:42 <_av500_> I never understood that Oct 09 20:43:07 <_av500_> x11 vs android vs egl Oct 09 20:43:13 well, android uses egl... Oct 09 20:43:24 x11 is the odd one out... Oct 09 20:43:38 <_av500_> so its the flavour of GL? Oct 09 20:43:45 as long as you have egl... wayland/qt-eglfs/android will work on "it".. Oct 09 20:44:07 opengl is on top of egl: https://en.wikipedia.org/wiki/EGL_(API) Oct 09 20:45:27 <_av500_> eyes glaze over Oct 09 20:45:29 <_av500_> :) Oct 09 20:46:44 Basicly, as long as we get egl sgx drivers, we have a chance of doing something.. Oct 09 20:46:51 <_av500_> ok Oct 09 20:46:52 ti/imgtech needs working egl for android.. Oct 09 20:47:11 and we can use it for wayland.. (i haven't personally got it to 'really' work yet..) Oct 09 20:48:09 so sgx530 doent work with wayland ? Oct 09 20:49:00 rob_w, they say: wayland 1.3 right here: http://git.ti.com/gitweb/?p=graphics/omap5-sgx-ddk-um-linux.git;a=commit;h=994922a5655c63b05343a9238cb30e6ef61744be Oct 09 20:49:24 but that was for "omap5".. ti is updating things so that repo will "also" support am3/am4/am5... Oct 09 20:49:41 we've always had a second fiddle sgx-blob release... Oct 09 20:49:54 this is big change for us.. Oct 09 20:51:49 rcn-ee, so what is the way to go for a ti platform Oct 09 20:52:07 if i want basic 3d support , like clutter + gtk ? Oct 09 20:52:31 i'm going wayland, then run x11 apps on top of that.. Oct 09 20:52:38 hmm Oct 09 20:52:44 just read about that Oct 09 20:52:59 wonder if my clutter+gtk will work what that too Oct 09 20:52:59 <_av500_> rob_w: btw, where is my board? :)) Oct 09 20:53:03 :-) Oct 09 20:53:16 we just have the first 5 full setups on the table Oct 09 20:53:33 i bet u want one with lcd and cap touch , eh ? Oct 09 20:53:45 <_av500_> one of each :) Oct 09 20:53:53 ? Oct 09 20:53:57 <_av500_> will it run x11? Oct 09 20:54:00 <_av500_> :) Oct 09 20:54:12 well yes .. but no 3d yet Oct 09 20:54:14 ;-) Oct 09 20:54:18 only framebuffer 3d :-) Oct 09 20:54:36 <_av500_> our first omap3 design used no 3d either Oct 09 20:54:46 <_av500_> nor x11 Oct 09 20:54:55 <_av500_> it was all Kartoffeldruck Oct 09 20:55:19 <_av500_> heck, that was 2008 Oct 09 20:55:53 <_av500_> you were al in diapers! Oct 09 20:55:55 back then a blinky led was cool enough... Oct 09 20:55:59 but the platform feels good.. it hold 80 °C for 2 days now .. Oct 09 20:56:00 <_av500_> all* Oct 09 20:56:17 <_av500_> rob_w: I bet you are cheating at emmissions! Oct 09 20:56:32 nah, it's emissions cheating you... Oct 09 20:56:53 ... Oct 09 20:57:28 to cheat in a heating test, you need some sort of "software" cooling cheat Oct 09 20:57:49 if i get software to activly cool down a cpu , i will let you guys know Oct 09 20:58:28 <_av500_> sell it to VW Oct 09 20:58:36 <_av500_> I bet they have use for it Oct 09 20:59:03 it wouldnt surprise me if av500 has something todo with vw Oct 09 20:59:11 <_av500_> ... Oct 09 21:00:45 av500, did you also work the pru on any sitara yet ? Oct 09 21:01:35 Hey I managed to get my WiFi dongle to blink and hostapd to run. But I stil lcan not see my SSID when I try and log on with my cellphone Oct 09 21:02:29 cart_man, maybe then you only miss some hostap setting Oct 09 21:03:48 rob_w: Well is there a specific reason why they would rather use wlan1 rather than wlan0 IF wlan0 is the only Wireless interface I can see? Oct 09 21:04:48 rob_w: Today I tried everything. I downloaded and compiled drivers and created scripts and it just does not want to work...getting the dongle to flicker is soo far a milestone Oct 09 21:05:01 <_av500_> rob_w: nope Oct 09 21:05:01 wlan0 vs wlan1 are just some device names, you would want to use the one you actually have Oct 09 21:05:52 is your dongle working without hostap , so normal scanning ? Oct 09 21:06:43 Mmmmmm Oct 09 21:11:59 Ahm well if I say ifconfig I can see it is there set as static with an IP and all Oct 09 21:14:38 rob_w: any ideas? Oct 09 21:15:55 i meant , iwconfig scan , so without hostap as a normal client .. Oct 09 21:16:14 if the dongle works in this mode , then ur hardware driver is pretty far working Oct 09 21:19:27 can I temporarily set it up without editing files? Or is there some program that will help me do that? Oct 09 21:19:37 rob_w: ^^ Oct 09 21:20:10 u will need to start editing conf files i guess Oct 09 21:22:31 iwlist wlan0 scan does not seem to return any networks...there is a couple of them out thre actually Oct 09 21:24:42 rob_w: I blaclisted the Oct 09 21:25:08 rob_w: I blacklisted the drivers and installed my own that I had to do since the normal drivers did not really work to well with hostpad Oct 09 21:25:17 hostapd Oct 09 21:25:25 "not work well" ? Oct 09 21:25:51 rob_w: Yes apparently they had some well known issues on this particular chip Oct 09 21:25:51 i dont know any details, i am just generally speaking Oct 09 21:26:32 i would test if iwconfig also and if that breaks too then its rather a driver issue Oct 09 21:27:00 rob_w: I will quickly unblacklist the drivers and try again Oct 09 22:04:49 anyone have a good minicom alternative that supports utf-8? Oct 09 22:05:55 The utf-8 bit isn't super important, I'm just tired of minicom... Oct 09 22:10:49 gtkterm Oct 09 22:11:15 Any terminal alternatives? Oct 09 22:12:41 screen.. but i hate it as much as minicom... Oct 09 22:14:08 never used screen for serial... Oct 09 22:14:39 <_av500_> HYPERTRM.EXE? Oct 09 22:32:40 When is the driver for the eeprom loaded during boot? I have a firstboot service that I want to use to generate a unique machine-id based on the beaglebone's serial number, but when the script runs /sys/bus/i2c/devices/0-0050/eeprom does not exist. Oct 09 22:37:54 @_av500_: NO, JUST NO! Oct 09 22:38:59 Had to use hypertrm at a previous job to run some specific scripts... That was one of my worst computer related experiences ever. Oct 09 22:40:37 <_av500_> heh Oct 09 22:49:34 lol Oct 09 22:50:14 abferm: it probably appears asynchronously? use an udev rule or systemd service to trigger Oct 09 23:06:07 abferm: untested udev rule -> https://paste.debian.net/hidden/7c581356/ Oct 09 23:07:17 you can then use that as dependency for your firstboot service Oct 09 23:13:10 argh, I should really swap my dev bbb with one that has kingston instead of micron flash -.- Oct 09 23:34:39 abferm: more complete example, and tested -> https://paste.debian.net/hidden/3cf3b982/ Oct 09 23:40:40 the thing is a bit uglier than it ought to be, mostly as a result of the eeprom being a binattr instead of a proper device **** ENDING LOGGING AT Sat Oct 10 02:59:59 2015