**** BEGIN LOGGING AT Tue May 24 02:59:58 2016 May 24 09:48:29 hello May 24 09:48:54 anybody home? May 24 09:52:09 yes May 24 09:52:21 he waited a whole 2 minutes before leaving at least May 24 09:59:00 hi May 24 09:59:12 this is venkatesh May 24 09:59:36 i am working on beaglebone black May 24 09:59:42 i need a help May 24 10:01:09 PDogJr: haha ;) May 24 10:01:42 that guy didn't even last a minute May 24 10:02:28 I guess these people think we're all just sitting here staring at our IRC clients 24/7 May 24 10:03:10 elho May 24 10:03:43 any beagleboners out there? May 24 10:04:34 247 May 24 10:05:36 hehe May 24 10:06:07 * mtnman is a fresh beagleboner May 24 10:06:36 ...almost everything config'd.... May 24 10:09:02 have bluetooth keyboard running, but not yet able to get the bt speaker to connect. is there a way to figure out the hciXX for the speaker? May 24 10:35:01 hello May 24 10:53:03 is there an info bot on the channel? May 24 11:05:02 no May 24 11:05:05 mtnman: why? May 24 11:15:29 on the #debian the bot(S) are full of helpful info, whereas the channel seems pretty low traffic. May 24 11:16:36 *this* channel, that is May 24 11:18:42 well, this channel doesn't need a factoid bot that much May 24 11:19:26 is that so? May 24 13:34:40 ddrown: Just FYI, found a great solution here: https://eewiki.net/display/linuxonarm/NimbeLink+on+BeagleBone+Black+and+Green. Doing 'apt-get remove udhcpd --purge' will remove udhcpd -- which in turn will force g_serial instead of g_ether in the boot script you had referenced. So this solved my issue. May 24 13:35:39 I think you'll find g_ether still exists.. there's just nothing that uses it any more :D May 24 13:37:56 there should be at least 3x g_xxxx drivers loading .. mine always comes up as a composite device May 24 13:48:01 stt_michael: I'm using a console build -- so the /opt/scripts/boot/am335x_evm.sh picks between g_ether and g_serial in that case. And I wanted to force it to serial. May 24 13:49:29 Geof, oh ok May 24 13:49:46 you could always hack the script XD May 24 13:51:01 The logic appears to be: USB-image-file => multi, else if udhcpd or dnsmasq => ether, else serial. May 24 13:51:20 So removing udhcpd pushed it to the serial option on the console build. May 24 13:51:25 nice May 24 13:52:16 Yes, but I don't think I need udhcpd -- so that was an easy option, and recommended on eewiki -- so more expert than I. May 24 13:53:06 fair enough. Can't argue with RN .. thats f'sure May 24 13:53:58 stt_michael: This chatroom is worth its weight in gold. Thanks much. May 24 13:55:33 * _av500_ demands 5% May 24 14:57:00 morning May 24 14:59:35 Geof: I don't understand why it just doesn't always use a combined rndis+ecm+acm (+ mass storage if applicable) May 24 15:00:42 I setup a combined rndis+eem+ecm with microsoft "OS Descriptors" and that turned out to work out-of-the-box on linux (eem), mac (ecm), and windows (rndis) May 24 15:00:50 with no driver installation bullshit on windows May 24 15:10:33 zmatt: from what I can tell in the am335x_evm.sh script, it checks for the file: /var/local/usb_mass_storage.img. But that doesn't exist on the rcn console image. As a result, it goes on to check on the root drive being mmcblk0p1 or mmcblk1p1 which it is. Then it looks for udhcpd/dnsmasq and if they are there, does g_ether, and if not does g_serial. May 24 15:11:17 So per RCN's comments, I just removed udhcpd to make that test fail -- thereby getting g_serial. May 24 15:12:03 The only way it does g_multi is if the USB image is there. May 24 15:12:28 And in my situation, g_serial is all that I need. So I'm ok with the logic. May 24 15:14:34 From my experimentation, when it does g_ether, the serial port does not show up on Windows. Only the USB0 ethernet interface. May 24 15:15:13 I recently setup a combined rndis+eem+ecm with microsoft "OS Descriptors" and that turned out to work out-of-the-box on linux (eem), mac (ecm), and windows (rndis), without any driver setup bullshit on windows.... true triple-platform plug and play May 24 15:15:16 I'm turning off eth0 and usb0 in ifconfig -- to allow for using a cellular modem with ppp. May 24 15:15:45 it could easily be augmented with a serial console as well May 24 15:16:19 and/or a mass storage device if you'd fancy May 24 15:16:43 yay for the libcomposite driver :) May 24 15:17:09 That would be fine. I don't need mass storage, just serial. I was hesitant to modify the logic in am335x_evm.sh. Is that what you're suggesting? May 24 15:18:42 I don't use any of those scripts... "generic-boot-service" as the whole bunch of scripts together is called is horrifically slow and I disabled it entirely May 24 15:19:08 I think it was like 50% of my boot time May 24 15:20:02 Ok. I'm not an expert on bootup scenarios. But if you have guidance I can try it. I would like faster bootup if that's possible. May 24 15:20:45 this is what the bootchart of my test-bbb looks like at the moment, and I haven't really done any real optimization yet May 24 15:22:35 My ultimate goal is to boot up, start my cell modem ppp scripts, and then run two scripts for my application, in under 20 seconds. May 24 15:23:15 I have a hardware watchdog on my cape that needs to be fed in less than 20 seconds. May 24 15:23:33 I've had power on -> gui in less than 10 seconds May 24 15:24:02 That's awesome. Were you sending me a link? Your message above didn't have one. May 24 15:24:26 oh duh http://gerbil.xs4all.nl/boot.svg May 24 15:24:29 sorry May 24 15:24:36 this is a no-gui system though May 24 15:24:50 but since we were using a framebuffer gui the difference is minimal anyhow May 24 15:24:55 Yes, I'm no-gui. Just a console with NodeJS scripts running. May 24 15:25:37 u-boot also takes up way more time than it should, I've also been working on that part May 24 15:25:50 (not visible in the chart) May 24 15:27:37 ultimately it seems that reading data from eMMC (fourth line from the top) is the limiting factor May 24 15:27:47 so stripping stuff down helps a lot May 24 15:30:21 So how do I try that out? May 24 15:38:21 I just replace the boot chart by a more recent one that also included usb networking, although still the old g_ether May 24 15:38:27 try what out? May 24 15:41:58 I wonder why fsckd,service is showing weird like that May 24 15:45:11 oh yeah still need to file a bug on the default udev rules ( the /dev/disk/by-path rules can't cope with /dev/mmcblk*boot*) May 24 15:51:49 I guess I don't know how to replace the boot chart. May 24 15:54:01 the boot chart is a graph you can have systemd print to show you how your system startup progressed May 24 15:54:14 systemd-analyze plot >boot.svg May 24 15:54:55 I just showed mine for comparison May 24 16:03:53 zmatt: How does one make that pretty boot chart? May 24 16:04:04 17:54 < zmatt> systemd-analyze plot >boot.svg May 24 16:04:17 not really awake yet? :P May 24 16:04:18 Oh! That was the command. May 24 16:04:30 Apparently *not*. May 24 16:04:54 systemd-analyze has more useful commands, but plot is the best May 24 16:05:36 Thanks! May 24 16:06:04 the systemd-fsckd.service line I'm getting seems to be a fluke though... it's normally a socket-activated service May 24 16:06:43 This is what I'm getting right now -- no cell modem attached right now. http://imgh.us/boot_7.svg May 24 16:07:30 Argh. I hate when X decides to put the title bar off-window. May 24 16:08:13 lol, my system is up before yours has even just started systemd May 24 16:08:28 large kernel + initramfs are generally to blame for that May 24 16:09:14 I run with a much modestly configured kernel and no initramfs May 24 16:09:32 Same here. Takes over 5 sec just for systemd to be done. May 24 16:10:25 I see a LOT of cruft though. This good intel. Thanks again! May 24 16:10:36 to be started you mean May 24 16:10:51 Yeah. Sorry. May 24 16:13:42 a non-embedded is of course a different story May 24 16:13:54 there you typically don't tailor the kernel to the hardware May 24 16:14:21 zmatt: I manage ehrpwm with your help. But I'm doing wrong with ecap. Can't figure out why ? http://pastebin.com/ZwJiEDH7 May 24 16:15:04 Ok, I'll look into kernel tailoring and turning off initramfs. Thanks very much. May 24 16:15:05 dirkk: &ecapwm0 is an invalid reference May 24 16:15:07 I haven't learned enough to know if I want initramfs, but it is an embedded system. I'll toss learning that on the pile. May 24 16:15:26 Geof: those aren't your biggest time-consumers though May 24 16:16:57 Yes, networking service is huge. But I'll have to see what it is on my real system. This was just on a straight BBG without my cape. May 24 16:17:47 Ragnorok: you need initramfs to host the driver modules needed to boot your system, since on desktops/servers those can vary wildly (controller type? are you using lvm/raid? filesystem type?) May 24 16:18:39 Ragnorok: on embedded you know these in advance and compile the relevant block driver and filesystem into the kernel (this is already the case in rcn-ee's) May 24 16:18:40 No lvm, ext4 fs. May 24 16:18:57 yeah but the kernel you use on your desktop is generic May 24 16:19:02 Ok. I'm using a rcn kernel. May 24 16:19:21 Isn't 4.1.19-bone20 a rcn one? May 24 16:20:19 huh I thought due "This good intel" you were talking about your desktop/laptop rather than a bbb May 24 16:20:43 Geof: probably you're using ifupdown rather than a network manager May 24 16:20:56 Geof: I use systemd-networkd but that requires debian stretch May 24 16:21:10 Sorry. intel is military speak for "intelligence", ie info. May 24 16:21:31 ahhh, yeah I know the expression just didn't parse it as such May 24 16:22:15 another caution: initramfs understands filesystem uuids May 24 16:22:16 Roit. May 24 16:22:18 the kernel doesn't May 24 16:22:59 Roit toe. I don't think I'm using uuid right now anyway. May 24 16:24:04 so you probably have either /uEnv.txt or /boot/uEnv.txt setting the root device to mmcblk0p1 (or p2 if sufficiently ancient) May 24 16:25:18 It's whatever the default is at present. May 24 16:25:41 since even though eMMC is mmc1, all but the very latest kernels just order mmc devices in order of appearance, with no regard to preexisting numbering from DT May 24 16:26:00 "the default" means little since new images are made all the time May 24 16:26:29 zmatt : is it &ecap0 ? May 24 16:26:33 Sure. But 4.1.19-bone20 should have some default, and I haven't changed it. (grin) I don't see anything in uEnv.txt that seems to delcare the root. May 24 16:26:58 kernel version has nothing to do with it May 24 16:27:15 with image I meant filesystem image May 24 16:27:21 Ah. Roit. May 24 16:27:36 which also determines which u-boot you're using May 24 16:29:02 I should check out what that is. I think it's the one from 3.8.x, but that should be changing shortly. Another is getting a new console distro working. I'm mainly on the higher level stuff. May 24 16:29:35 Geof: if you're using ifupdown with eth0 configured for dhcp and don't have ethernet plugged in (or otherwise no response to dhcp) it delays boot hugely due to waiting for timeout May 24 16:30:35 hi, cannot find this anywhere - in bb.org overlays there is always some id suffix (eg. 00A1 or so) - what does it mean? id for each cape? version of overlay? May 24 16:30:44 maciejjo: version of the cape May 24 16:30:58 ok, thanks May 24 16:31:12 to make sure the correct overlay is loaded for the version of the cape you're using (it obtains that info from the cape's eeprom) May 24 16:31:39 so it is for capemgr yes? May 24 16:32:23 zmatt: I'm searching in am335x tech ref manual. Is there a doc where I could find those reference ? May 24 16:32:40 it is for what I just said. when capemgr detects a cape it gets its name and version from the cape EEPROM and tries to load /lib/firmware-CAPENAME-VERSION.dtbo May 24 16:32:48 dirkk: oh sorry I forgot to reply I think May 24 16:32:59 dirkk: no, TRM doesn't concern itself with linux, just the hardware May 24 16:33:16 dirkk: you have an invalid reference May 24 16:33:36 I'd expect the kernel to complain about that though if you try to load your overlay (see kernel log) May 24 16:35:20 zmatt: is &ecap0 the right one ? May 24 16:35:27 indeed May 24 16:36:11 you can find those labels (and the nodes they're referring to) in https://github.com/RobertCNelson/linux-stable-rcn-ee/blob/4.1.20-ti-r58/arch/arm/boot/dts/am33xx.dtsi May 24 16:36:32 replace the kernel version there by the kernel version you're using (although the file doesn't change much over time) May 24 16:38:54 you can also find all labels (and the device tree paths they refer to) in /proc/device-tree/__symbols__ May 24 16:39:24 (note that each file ends in a NUL byte rather than a newline) May 24 16:39:36 *the contents of each file May 24 16:42:09 On my real system, I don't need to bring up eth0 at all -- how do I disable that part of the bootup? May 24 16:42:34 I'm just going to use ppp over cell modem. May 24 16:43:44 on systems still using old ifupdown network configuration is done in /etc/network May 24 16:43:51 especially /etc/network/interfaces May 24 16:48:37 Thanks. I'll play around with things and see what I can do... May 24 16:50:51 be careful if you want to try omitting initramfs, it may require adjustments to one or more of { /uEnv.txt, /boot/uEnv.txt, /etc/fstab } May 24 16:52:25 zmatt: I made the change but same issue. Parse error ...:( May 24 16:52:41 dirkk: ohh, parse error, lemme see May 24 16:53:11 lol, "Warning: File 'Makefile' has modification time 20186 s in the future May 24 16:53:27 maybe consider configuring it with a valid NTP server address May 24 16:53:40 (it should normally get one from the dhcp server, but in your case apparently not) May 24 16:54:58 lol I know but for other pwm it works May 24 16:57:11 dirkk: btw, strongly consider renaming "pwmcap_pins: pwmcap" to "pwmcap0_pins: pwmcap0" or just "cap0_pins: cap0" ... pwmcap is ambiguous since there are multiple cap modules. maybe you don't need any other one now, if you later do and forget to avoid the name conflict you'd get strange problems May 24 16:57:21 still don't understand the parse errror though May 24 16:57:33 lemme try to compile it myself May 24 16:58:00 zmatt: ok thanks May 24 16:58:54 including the usage of the module ("pwmcap" instead of "cap") in the name is actually probably a good idea though May 24 17:00:55 it's probably something subtle and hard to spot... device tree sources are extremely fickle May 24 17:03:07 AHHH May 24 17:03:09 duhhh May 24 17:03:39 I believe you ;) May 24 17:04:29 check bone/pins.h . the constant P9_42 doesn't exist May 24 17:04:40 since there are *two* cpu pins connecting to that expansion header pin May 24 17:05:47 hahhh indeed May 24 17:08:36 so you need to use P9_42a May 24 17:09:56 and I recommend inculding PIN_GPOUT_NOPULL( P9_42b ) then to ensure the other pin is also reserved but disabled May 24 17:10:35 (configuring as gpio output (which is high-z by default) with no pull is the closest available to disabling a pin on the am335x) May 24 17:11:14 I miss the "safe mode" pinmux option of the dm814x :P it would detach the pin from *all* peripherals, including gpio May 24 17:14:44 zmatt: ok thanks for your help ;) dm814x ?? May 24 17:15:46 a "disallow output-enable" bit in pinconf would have been highly appreciated also, to make sure that accidently configuring a bidirectional pin of a peripheral module (e.g. eCAP) as output (pwm) while the board was designed to use it as input (capture) can't result in damage to the processor or external components May 24 17:16:49 dm814x (aka Centaurus) is the previous TI SoC I have experience with, and the am335x is architecturally strongly related to it May 24 17:18:53 TI doesn't seem to care anymore however about mass-marketing it... you can still buy it, but the's not much support and the "latest" (read: "last") official TI kernel is a heavily patched 2.6.37 May 24 17:19:10 not even bug fixes provided by users have been committed to git since May 24 17:19:34 it's really shameful how they treat the 81xx series since they're really nice processors May 24 17:20:13 especially the 811x (which was never even released to broadmarket) May 24 17:21:41 dm814x also seems to be the last processor where TI bothered to at least *try* to organize things in a clean and consistent fashion May 24 17:22:08 on the am335x it is clear they had absolutely no such aspirations May 24 17:22:15 what a mess -.- May 24 17:23:34 Haha .. is dm814x open-sourced ? May 24 17:24:28 what do you mean by that? May 24 17:26:00 support for dm81xx processors is btw still slowly trickling into mainline linux... I think it might already be bootable now (with a limited set of peripherals) May 24 17:27:30 maybe when I'm in a silly mood I'll dust off one of our dm814x-based prototype boards (written off as expensive mistake) and see if I can boot linux on it May 24 17:27:51 (I was previously doing baremetal coding on it only) May 24 17:29:05 Is it like am335x ? I mean open foc on ti wiki May 24 17:33:50 I still don't really understand the question. like the am335x it's a SoC you can buy, which is mostly documented, there's info on the wiki, an understaffed support forum, there is an sdk from TI (EZSDK 5.05.02, placed in "legacy" and marked "not recommended for new designs") May 24 17:35:28 there an evm board May 24 17:35:51 nothing like a small cheap board like the bbb though May 24 17:43:04 wouldn't ecap and ehrpwm be better? May 24 17:43:51 unfortunately the 811x however was withdrawn from the general public... no real docs have ever been published, there was just a briefly available evm ( http://support.spectrumdigital.com/boards/evm811x/revc/ ) but it quickly become automotive-only (dra62x) May 24 17:44:18 ds2: I have a habit of omitting prefixes and suffixes indicating how fancy/cool a peripheral is May 24 17:44:45 unless ambiguity might exist May 24 17:45:17 full names makes it easier to search in code May 24 17:45:50 true, although often you need regexp anyway May 24 17:46:32 it's kinda funny that they call it ehrpwm even though they declare a register range that doesn't include the HR register May 24 17:47:53 ds2: TI however does such things in practice too though (although they try avoid it in public docs) May 24 17:48:01 like calling IVA-HD just IVA May 24 17:48:46 and calling epwm ehrpwm is just annoying especially since probably virtually noone uses the HR functionality May 24 17:49:01 and they're otherwise identical May 24 17:49:57 but yeah I was in doubt about ecap vs cap May 24 17:50:43 but my reasoning was that if I was going to omit such prefixes/suffixes, I should do it consistently May 24 19:33:02 Anyone care to help with SDHC(mmcblk0)? I'm trying to write a correct uEnv.txt file on my SDHC card in order to force booting from SDHC. May 24 19:34:04 Seems I need to write mmcdev bootpart mmcroot optargs variables in uEnv.txt. May 24 19:40:30 I'm guessing that something like mmcroot=/dev/mmcblk0p2 or something needs to be there. May 24 21:31:08 hi, I want to declare some PWM node in devicetree, where can I find which PWM device and PWM number is mapped to specific BBB pin? May 24 21:31:47 I want to use P9_16 May 24 21:32:55 from what I see in srm this is ehrpwm1B_mux1 in mode 6 May 24 21:34:14 so I set the mode in pinmux, but now I have a node for my device under ocp, and I see I need to specify pwm property with pwms property May 24 21:34:56 and the second value in this should be "chip-relative PWM number" May 24 21:37:57 but where is it described? May 24 22:13:20 Hey thr... new to BBB setup .. but was wondering how to get the internet going using the ethernet cable ... while USB powered May 24 22:14:18 internet was working over USB ... but when I connect the ethernet cable ... the ip address has changed etc. and cloud9 cant connect over the 192.168.7.2 May 24 22:14:31 How do I query the new IP address without a monitor ? May 24 22:14:53 Your router should know that it just assigned a new DHCP client an address May 24 22:17:35 @rthille-ciena : That is probably true .. but I sit inside a university with connection to the internet all secured up May 24 22:17:49 and I do not have access to the router table May 24 22:18:32 unplug ethernet, get in via USB, look at logs to see what DHCP address was assigned. when you plug in the ethernet again, you'll probably get the same IP May 24 22:19:02 that sounds good, and could you tell me where I would find those logs .. May 24 22:24:13 hola May 24 22:24:20 can´t ceate socket: no space left om device May 24 22:24:31 que puedo hacer May 24 22:24:47 can´t ceate socket: no space left on device May 24 22:26:19 reset on power? May 24 22:30:00 Hello the black BEAGLEBONE presents the following fault I can do ceate can't socket: no space left on device May 24 22:30:21 Hello the black BEAGLEBONE presents the following fault I can do ceate can't socket: no space left on device May 24 23:10:08 Luis_: context? May 25 01:39:42 hello May 25 02:00:06 hello May 25 02:03:44 hello ayjay May 25 02:04:50 not much convo here May 25 02:07:58 yeah i mean people come and go May 25 02:10:56 besides, quality over quantity May 25 02:39:59 not sure about either here ayjay. nobody has responded to several questions over the past two days May 25 02:40:11 oh man that's rough May 25 02:40:40 here let me just give you zmatt's phone number.... May 25 02:40:56 well, day 3 as a beagleboner and its going okay aside from the wireless keyboard.... May 25 02:41:25 who is zmatt? May 25 02:41:44 ahha it's not important May 25 02:41:49 i'm glad your beaglebone experience is going wel May 25 02:41:58 and i'm sorry you're having trouble with a keyboard May 25 02:42:26 i am glad to back on a debian system. May 25 02:43:08 ayjay do you have much experience with bluetooth config? i have a device i cannot remove, and another that i cannot install! May 25 02:44:13 i don't unfortunately May 25 02:44:24 i don't really have much experience with anything, to be honest, i just do what i can as i go May 25 02:46:43 ok thanks anyhow **** ENDING LOGGING AT Wed May 25 02:59:58 2016