**** BEGIN LOGGING AT Tue Oct 04 02:59:58 2016 Oct 04 07:12:12 amitof: Disable the emmc, you then need to boot off an mmc card. Those are the last pins I would use, better to disable hdmi first Oct 04 10:01:00 Hi all, i am about to build a debian kernel for my BBB. In the manual there are the kernels from v4.1 to v4.7 Oct 04 10:01:05 any thoughts about them? Oct 04 10:01:45 depends on what you want to achieve Oct 04 10:04:17 is there more information available about the specific kernel and their advantages? Oct 04 10:30:13 tlab: Nice, I will try that. I also does not get pinmuxes to show for my interrupt pins Oct 04 10:35:51 If i want to achieve a short startup time on my BBB, which kernel is the best? Oct 04 10:38:53 scp: all of them, given a streamlined config. You'll spend most time in init anyway Oct 04 10:40:10 ok thanks tbr Oct 04 10:48:56 tlab: Il have to try it out later. Had some luck enabling debugging on the kernel last night. I have some freetime to night after my lectures. Will update you on how it went then :) Oct 04 10:51:19 does beagleboard need server? Oct 04 11:41:00 vickie: Probably not, but please explain your question as in what you are expaecting the server to do Oct 04 12:28:35 tlab: the irq probably still hasn't been requested by the mcp driver, given that (for whatever reason) it still doesn't seem to be working Oct 04 13:05:26 Anybody encrypted their rootfs on a beaglebone black or green ? Oct 04 13:08:18 LaMs: i haven't even been able to mount luks encrypted volumes on my BBB :( Oct 04 13:10:16 zykotick9: What was the issues for you ? I'm trying to figure out how it's done, to be more specific, i'm trying to understand if I need to modify U-Boot in some way. Right now I'm in a catch-22 Oct 04 13:11:46 LaMs: sorry - i have NO idea. I'll be keeping an eye on any _real_ replies you get. Good luck! Oct 04 13:32:04 Is there actually a setting to disable all joind/quit/timeout status messages in this chat? Oct 04 13:33:45 That's a function of your client. Oct 04 13:34:58 scp: you might want to see http://wiki.xkcd.com/irc/Hide_join_part_messages Oct 04 13:45:58 Hi everyone, I'm experience problems with a new BeagleBoard X15 that freezes/hangs during normal operation. I'm unsure on what forum I can use to address my problems. Should I consult the Beagleboard forum (https://groups.google.com/forum/#!forum/beagleboard-x15) or Texas Instruments e2e forum in this matter? // Max Oct 04 15:05:43 tlab: So, here is my dts file. I compiled it and moved it to boot so that I can boot using the dtb. It still doesn't show any pinmuxes after updating kernel to 4.4.23-ti-r51 Oct 04 15:05:51 http://dpaste.com/1M1F8M3 Oct 04 15:06:42 spidev shows two devices and spi_loobpack.c works Oct 04 15:08:33 And here is some debugging output Oct 04 15:08:34 http://dpaste.com/0G5WZK8 Oct 04 15:09:53 And here is the output from showpin: http://dpaste.com/3HAKNTV Oct 04 15:17:42 hi. i'm experimenting with device tree overlays, and i'm trying to debug my first one. i'm able to comile it, load and unload, but when i change anything, copy to /lib/firmware and load again, it loads previous version. even if i remove the dtbo file from /lib/firmware it loads successfully. i suspect it is cached somewhere. how do i clean it and make it load the newes version? Oct 04 15:28:22 Hello Oct 04 15:30:06 First of all: the "Asking Rules" doesn't exist Oct 04 15:31:31 How is it possible to start a selfmade *.js Script on the Beagle board? I find only the "Run" Windows in the Demo Section. I like to use the Terminal Oct 04 15:37:15 Unable to connect. Interesting. Running CLI is a generic Linux question, but "start a .js" isn't really a lot of info. Perhaps search including what you want to do, like "interactive" or "on boot" or "on interval". Oct 04 15:41:50 First start the Script once for testing and debuging. After Completing the Script i have read the "autorun" tutorial. But at the moment i dont like to restart the BB after each change Oct 04 15:42:14 but i have found now the right command "node .js" Oct 04 15:43:25 \o/ Oct 04 15:44:43 NadeoWolf: "how to ask smart questions" you mean? yeah someone needs to fix that link: catb.org for some reason doesn't support https anymore, replace it by http Oct 04 15:45:24 not only. also the "this IRC survival guide" Oct 04 15:46:09 ah, I have no idea what was in that Oct 04 15:46:19 shouldn't be too hard to find general resources on using IRC though Oct 04 15:48:42 (and freenode-specific info can be found here -> https://freenode.net/kb/all ) Oct 04 15:52:07 any device trees hackes? i’m having an issue with reloading my DT overlay. when i try lo load a newer version of my overlay, it still loads the old one (even when nothing is is /lib/firmware, so the should be an error). i suspect some caching. does anyone know anything about it? Oct 04 15:52:44 which kernel version? does rebooting fix the problem? Oct 04 15:53:45 yes. rebooting helps, but it is quite annoying to reboot every time whiile i’m experimenting. version is 4.4.22-ti-r48 Oct 04 15:54:56 NadeoWolf: archive.org to the rescue! https://web.archive.org/web/20150531122326/http://www.mahdiyusuf.com/post/6176918750/survival-guide-irc-1 Oct 04 15:55:30 hi guys. iam new to beagle board i want to contribute to the community where do i start from? Oct 04 15:56:33 NadeoWolf: and actually it definitely looks like important advice Oct 04 15:56:48 NadeoWolf: it also looks like the link has been broken already a very long time and you're the first to have mentioned it Oct 04 15:58:28 xD i'm a long IRC user, so i just know the "Default" rules.. only my english is not so good.. xD Oct 04 15:59:26 yeah I didn't say you needed that advice specifically, but a _lot_ of people that come in here clearly could have used it Oct 04 15:59:57 e.g. asking "is anyone here?" and then leaving one minute later Oct 04 16:00:05 hahaha Oct 04 16:00:35 i have found another problem: the Cloud 9 IDE is maybe not the best solution , because the require a Credit Card for registration Oct 04 16:01:03 *the = it Oct 04 16:01:35 huh, registration? what? I never use cloud9 but I do know it's (by default) installed on the BBB itself and Just Works, it doesn't require registration of any sort Oct 04 16:01:48 what???? how??? Oct 04 16:02:04 the links brings me each time on the cloud9 Homepage Oct 04 16:02:22 what is the link on the BB? Oct 04 16:02:51 eh it's just a web-based IDE running on locally a webserver on the BBB itself Oct 04 16:03:12 I think port 9000, but there should be a link to it on the page served on port 80 Oct 04 16:03:15 192.168.7.2 Oct 04 16:03:19 i know. but is default on or do i neet to start it first? Oct 04 16:03:24 default on Oct 04 16:04:03 nothing on port 9000 Oct 04 16:04:57 I might have remembered the port wrong, it's been a while since I last looked at it and I don't have any of that stuff installed on our beaglebones Oct 04 16:05:45 i log in with mirc.. one moment.. Oct 04 16:06:23 back xD Oct 04 16:08:08 i found the port... "3000" Oct 04 16:08:16 ah, I was close Oct 04 16:08:26 O:) Oct 04 16:09:20 is close if we remember that they are nearly 65000 ports xD Oct 04 16:09:30 uff is so hard to write in english xD Oct 04 16:09:31 well I only got one digit wrong Oct 04 16:09:39 true xDDD Oct 04 16:11:20 you can also get a list of listening tcp ports with: ss -4lnt Oct 04 16:16:23 zmatt: or netstat Oct 04 16:18:40 yeah netstat is an older alternative Oct 04 16:19:34 xD Oct 04 16:20:25 ss is included with iproute2 (like the ip utility) Oct 04 16:20:38 How does one go about enabling and using GPIOs on 4.4.x kernels? I am pouring some code from 3.8.13 and it looks like I got SPI working, but GPIO isn't doing anything. Oct 04 16:21:22 puzz-mobile: if the pin is muxed to gpio (default typically) then you can just export the gpio same as before Oct 04 16:22:32 you can check pinmux state e.g. with my show-pins util ( https://github.com/mvduin/bbb-pin-utils ) Oct 04 16:30:38 how many Wh can i attach on the 5V pin? :s i think 72Wh is to match.... ?? Oct 04 16:31:25 what do you mean? Wh is a unit of energy (1 Wh = 3.6 kJ) Oct 04 16:33:11 ok. i debugged my DT overlay and it appears to load correctly. now i want leds-gpio driver (which is built-in in this kernel, not built as a module), to reinitialize and reread new configuration from overlaid DT. i can unbind it by “echo leds > /sys/bus/platform/drivers/leds-gpio/unbind”. but how do i re-bind back, so that it rereads the DT? is there a way to do it? Oct 04 16:33:20 “echo leds > /sys/bus/platform/drivers/leds-gpio/bind” wouldn’t work. Oct 04 16:34:34 that would be the way to rebind it Oct 04 16:34:51 i'm planing to control my Digital-LED-Stripe WS2812B .. but i need another powersupply..... Oct 04 16:35:29 olddbg: if rebinding doesn't work I'd suggest customizing the main device tree instead of using overlays Oct 04 16:36:11 zmatt: Super useful tools. Thank you. Looks like absolutely zero of the pins defaulted to GPIOs. Oct 04 16:36:34 huh Oct 04 16:36:35 zmatt: bash: echo: write error: No such device Oct 04 16:37:02 zmatt: this is what i was trying to avoid :) Oct 04 16:37:13 olddbg: must be some weird problem specific to leds-gpio Oct 04 16:37:19 funny, I actually avoid overlays :) Oct 04 16:38:14 zmatt: I think I just failed at disabling HDMI. Oct 04 16:38:28 the only benefit I see for overlays is being able to experiment with different DT settings without reboot, but unloading and reloading an overlay often results in a mess anyway (up to and including kernel panics) Oct 04 16:38:33 well, i though it was new and fashionable way to tackle with hw descriptions. sigh. well, i think i’ll just resort to a static DT Oct 04 16:39:03 no they were originally designed to allow autodetection of capes Oct 04 16:40:48 but overlays have all sorts of limitations that customizing the main dt doesn't, their syntax is horrible (although https://github.com/mvduin/overlay-utils helps), they're a source of unreliability since not all of kernel appreciates dynamic changes at runtime, and they have terrible diagnostics Oct 04 16:41:18 oh yes, i noticed that about diags :) Oct 04 16:41:55 dtc can do almost no error checking, e.g. it won't catch typos in references like it would in the main dts Oct 04 16:42:19 the kernel ought to catch this and give a useful error message, but well... Oct 04 16:43:00 yeah. so i was decompiing the resulting dtbos and verified references manually Oct 04 16:43:47 I once considered making some util that would verify an overlay against the main dt, but then I remembered I don't really care about overlays anyway Oct 04 16:44:00 puzz-mobile: http://elinux.org/Beagleboard:BeagleBone_Debian_Image_Migration#Disabling_eMMC_or_HDMI Oct 04 16:44:01 hehe Oct 04 16:48:11 zmatt: I was using the emmc overlay... This is weird. Oct 04 16:50:05 olddbg: I still want better tools for generating device trees though. device trees were designed to be easy to generate and parse by computers, not by people Oct 04 16:51:52 very true. i don’t have much experience with DTs, but so far it was quite annoying. Oct 04 16:52:58 yeah it's a pretty stupid simple format, which is why it requires a fair amount of redundancy Oct 04 16:54:00 DTs in general have a bigger problem though, namely that on modern SoCs you need a directed graph instead of a tree to accurately model the hardware Oct 04 16:55:27 (or rather more than one; at the very least a memory/mmio interconnect graph, an event/interrupt signal graph, a power delivery graph) Oct 04 16:55:36 apple's IOKit does some of that stuff better Oct 04 17:06:11 oh and a clock graph of course Oct 04 17:09:16 zmatt: Any explanation of the columns of show-pins? Oct 04 17:09:16 I have http://7e02ea1a021d16a7.paste.se/ and I'm particularly interested in knowing that P8.27-P8.29 are acting as GPIO outputs. Oct 04 17:10:57 lo >> means they're outputs driven low Oct 04 17:11:06 << hi means an input measuring high Oct 04 17:11:39 (I'm not 100% sure whether these would be inverted for gpios marked active-low btw, I just realized I haven't tested that yet) Oct 04 17:14:17 first column is pin description, then pin number (index in pinmux array), slew, receiver-enabled, pull, mux, function description (incl gpio value and direction if known), usage (as reported by kernel), and in parens the pinmux node if known Oct 04 17:14:44 sysfs is the usage reported for gpios manually exported via sysfs Oct 04 17:15:27 (gpio-of-helper can be used to define pins in DT and have them exported automatically, there a custom label would be shown in the usage column) Oct 04 17:17:07 Thank you so much. This is very helpful. Oct 04 17:18:34 your P9.15 shows pretty bizarre muxing Oct 04 17:29:07 Yeah that one is screwy. I don't know what is wrong. That should be an input like P9.12. Oct 04 17:29:41 I've been terrified of getting into this device tree stuff and I've been avoiding it like the plague, but it's looking more and more like that's what this is coming to. Oct 04 17:34:54 it looks more intimidating from a distance Oct 04 17:35:33 once you get used to it it's not really scary anymore, mostly just annoying Oct 04 17:46:58 just-let-me-use-my-pins.dtb Oct 04 17:54:41 well if pins was all that it's about then cape-universal might actually be a good idea, but it isn't Oct 04 17:55:50 Haha. No it is not. I think it's time I just learned how to set up a proper .dts for this instead of using overlays. Oct 04 18:01:40 tlab: laurittr: ah good, I see rcn has now disabled the "edma workaround" dt fragment that was causing spi problems Oct 04 18:01:50 https://groups.google.com/d/msg/beagleboard/j4JSF6rKW4g/dmJFVG1dAwAJ Oct 04 18:04:43 lol, and some other eh.. minor bugfixes Oct 04 18:04:44 r47: Oct 04 18:04:45 tilcdc fixes (lcd panels work again) Oct 04 19:58:35 zmatt: But it still doesnt work.. Il try some more tomorrow, but I really think my dts is right Oct 04 20:58:00 Finally got this code working. The SPI interface in 4.4.21 wasn't releasing the CS line like it did in 3.8.13. Oct 04 20:58:09 laurittr, zmatt mine doesn't work either only thing I have noticed so far is the irq don't say they are being used in show-pins, but maybe they never show up being used? I don't know Oct 04 20:58:27 show-pins http://pastebin.com/JVY1gHiQ Oct 04 21:05:21 if anyone is interested in purchasing 20 beaglebones tell me! Oct 04 21:05:55 Hello Oct 04 21:13:56 ogres: I hope you're not still talking about those clones and trying defraud people by claiming to be selling beaglebones Oct 04 21:14:10 no I am telling the buyer that they are clones Oct 04 21:14:42 and letting them go for $250 Oct 04 21:15:04 you just called them beaglebones. they are not Oct 04 21:15:59 seriously, stop doing this or I'm going to see if I can get you banned here Oct 04 21:16:18 http://www.ebay.com/itm/Antminer-S4-S4-S5-S5-S7-BB-version-BeagleBone-Black-BB-Black-BB-board-/262641562578?hash=item3d26a833d2:g:4tMAAOSwNRdX5wbp Oct 04 21:17:05 so yea they are clones and I planned on disclosing that Oct 04 21:18:14 tlab: Yeah, i get the same pinmux. irq pins not showing in pinmux Oct 04 21:18:35 yes, you made a nice comparison after all (http://i.imgur.com/96N4U6b.jpg) showing many ways in which they are failing to be beaglebones; apart from an am3352 being used hence no PRUSS and no SGX Oct 04 21:19:09 but don't call them beaglebones in the first place, they are not members of the beaglebone family Oct 04 21:19:22 Hello, they know if there is any tutorial to install the PowerVR graphics module to have 3D acceleration? Oct 04 21:20:15 ogres: that ebay page btw seems to be a different clone compared to the one in your image Oct 04 21:22:01 tlab: what's P9.42 ? it's showing muxed to gpio but listed as being part of spi1 Oct 04 21:22:09 (pin 104) Oct 04 21:23:36 and the new kernel (or rather its device tree files) fixes the problem of not being able to enable spi through overlay Oct 04 21:24:01 puzz-mobile: it does for me (unless you explicitly request cs to remain asserted after the transfer) Oct 04 21:24:58 tlab: laurittr: i.e. if you're using a custom dt then the new kernel probably won't matter Oct 04 21:27:59 zmatt: cs_change set to 1? Oct 04 21:28:38 puzz-mobile: ah, I'll let you in on a secret: when that flag is set on the last transfer of a single request, it means the opposite Oct 04 21:28:59 basically it's cs_not_the_default_behaviour Oct 04 21:29:27 the default being to keep cs asserted between the transfers of one request and then deassert it Oct 04 21:29:40 if 3.8 did it otherwise, that's a bug Oct 04 21:30:02 Hahaha. Shame on me for not reading spidev.c. Oct 04 21:30:15 what did you expect? documentation? Oct 04 21:30:19 (I got bitten by this one too btw) Oct 04 21:31:58 zmatt: In that pastebin from tlab there are 2 entries for p9.42.. What does that mean? Oct 04 21:32:36 laurittr: there really are two cpu pins that connect to that expansion header pin Oct 04 21:32:53 that's why I quickly added I meant pin 104 Oct 04 21:34:18 i have checked the kernal build and apparmor is enabled in the config yet i am getting this error AppArmor not available as kernel LSM this is on ubunut 16.04 for the beaglebone. what am i missing Oct 04 21:35:30 this is in reference to http://elinux.org/BeagleBoardUbuntu Oct 04 21:36:05 Ok, that makes sense, thats why the other one should be disabled. Oct 04 21:36:49 ahh right I did that, silly me for asking Oct 04 21:36:51 duh Oct 04 21:36:58 sorry, multitasking takes its toll Oct 04 21:36:59 :P Oct 04 21:37:30 yeah it's not required but I do it as good form, and to at least make sure there are no pull conflicts Oct 04 21:38:26 zmatt: I noticed you also got bitten by the same pin as me, P9_15. I think your bug report about conflicting pulls on that pin might be why they're muxed so strangely now. Oct 04 21:38:58 that seems very unlikely Oct 04 21:39:14 the only thing required to fix the pull conflict is fixing the pull direction of one of them Oct 04 21:39:48 Fair enough. I don't know what happened there. I just made an overlay for the one pin and called it a day. Oct 04 21:40:08 tlab: when a pin is used as irq by some driver it will most definitely show up as being in use Oct 04 21:40:20 And I couldn't find a commit that fixed the conflicting pulls, so I pushed it in the back of my mind. Oct 04 21:40:45 it should be fixed in u-boot Oct 04 21:40:53 if it isn't already, I haven't checked recently Oct 04 21:41:02 dmesg says: “AppArmor: AppArmor disabled by boot time parameter” but i cant seem to find where that option is set, if anyone can help it would be greatly appreciated Oct 04 21:41:46 device trees can fix it too, but only by claiming the pin, which prevents it from being used for a real purpose Oct 04 21:42:27 johnorjias: http://lmgtfy.com/?q=AppArmor+disabled+by+boot+time+parameter Oct 04 21:42:40 zmatt it is the same clone that I have Oct 04 21:43:24 ogres: the photo doesn't match http://i.imgur.com/96N4U6b.jpg Oct 04 21:43:29 crap gotto run to catch train, bbl Oct 04 21:43:32 i didnt take that picture Oct 04 21:43:35 i found it on the net Oct 04 21:43:39 but its literally the same thing Oct 04 21:43:44 the one on ebay is the same I have Oct 04 21:48:20 yes i know how to google. i am not finding anything benifical to my question, good to know you don’t know the answer either Oct 04 22:08:17 yay, made it Oct 04 22:08:44 ogres: in that case you may want to investigate what that big ic on the bottom side is Oct 04 22:17:59 zmatt, are you saying the new kernel will not fix our spi issue because we are using dt? Oct 04 22:44:15 * tlab is lost Oct 04 23:15:35 tlab: it fixes a problem you've already worked around Oct 04 23:16:17 I have no idea what your current problem is Oct 04 23:16:42 yeah I don't either lol Oct 04 23:18:05 I know enough to barely tinker I think Oct 05 00:40:26 laurittr, http://dev.iachieved.it/iachievedit/gpio-chip-selects-with-the-beaglebone/ (not sure if it's relevant) Oct 05 01:06:36 tlab: you're using regular chip selects rather than gpio chip selects Oct 05 01:06:52 you can try using gpio chip selects though (or a mix) Oct 05 01:07:41 I loaded an old image with 3.8 kernel just to see what show-pins looked like Oct 05 01:07:43 http://pastebin.com/Wj8ihmNX Oct 05 01:08:32 looks as I'd expect Oct 05 01:11:39 I recently browsed the current version of the mcspi driver and it left me a bit puzzled how it still works with regular chip selects other than cs0... maybe it doesn't, although it should then still work with cs0 (and higher chip selects if defined as gpio chip selects), also I think those changes were only made very recently but I'm not sure Oct 05 01:12:25 neither mcp is working for you however so I don't think the chip selects are the problem Oct 05 01:13:02 the major diff I see is the irq isn't listed in the show-pins Oct 05 01:13:20 which implies the driver hasn't requested the irq yet Oct 05 01:13:34 but I think so far you haven't seen any sign of life from the mcp driver yet on 4.x ? Oct 05 01:13:54 no Oct 05 01:14:27 I think maybe at one time we had all can show in ifconfig, but I think maybe I wasn't using the right can port on the board Oct 05 01:14:46 I have all 3 wired to the same bus now Oct 05 01:15:43 the can bus they're wired to (if any) obviously can't have any effect on whether they show up in ifconfig or not Oct 05 01:16:28 yeah but I think I tried to bring them up and I thought they did nothing, but may have been because they were not wired Oct 05 01:17:26 but maybe I recall incorrectly Oct 05 01:17:27 if you got the can interfaces for the mcps to show up on 4.x that would be pretty important though, what did you do? Oct 05 01:17:32 ok Oct 05 01:17:48 I thought it was something you did with the TT3201 file Oct 05 01:18:29 I haven't done anything with the file since the initial version Oct 05 01:18:32 https://github.com/mvduin/overlay-utils/commits/master/TT3201-001-01.dtsi Oct 05 01:18:43 I thought you changed it once Oct 05 01:18:50 git log shows I didn't Oct 05 01:19:03 yeah Oct 05 01:19:20 maybe it was the can0 can1 files Oct 05 01:22:57 I probably just remember incorrectly Oct 05 01:23:13 the mcp driver sure doesn't seem to bother logging anything on error Oct 05 01:23:38 I'd put a boatload of debug messages in that probe function and recompile Oct 05 01:23:42 I tried that dynamic debug, but I got same results as laurittr Oct 05 01:24:00 dynamic debug won't help if there are no debug messages to enable Oct 05 01:24:45 http://lxr.free-electrons.com/source/drivers/net/can/spi/mcp251x.c#L1027 Oct 05 01:24:46 I'm not even sure if I used it correctly Oct 05 01:24:53 that whole function contains zero debug messages Oct 05 01:25:06 zero messages of any kind actually, also no errors if anything goes wrong Oct 05 01:25:55 so that's not terribly informative Oct 05 01:26:15 can compile it on bbb? Oct 05 01:27:35 I wouldn't feel an urge to try that, it takes long enough on a 2.5GHz quad-core xeon Oct 05 01:28:03 can just do the module ? Oct 05 01:31:06 maybe but I'm not sure how to do that. plus even merely cloning the linux repo on a bbb will already be painful. you don't have any decently beefy x86 machine running linux available? Oct 05 01:31:43 or even a mediocre x86 machine running linux actually, if the alternative is a bbb :P Oct 05 01:32:04 yeah I do, i7 Oct 05 01:32:22 I have fedora running in virtualbox with 4 of the cores Oct 05 01:32:46 off a diff hd Oct 05 01:32:52 why then would you even contemplate using the bbb :P Oct 05 01:33:11 portability Oct 05 01:33:18 ? Oct 05 01:33:31 can take it places Oct 05 01:33:33 is small Oct 05 01:33:42 how does that matter for recompiling the kernel? Oct 05 01:34:02 oh I thought you ment contemplate using the bb at all Oct 05 01:34:05 no Oct 05 01:34:16 I see, for compiling Oct 05 01:34:29 eh I'm in no rush Oct 05 01:34:37 but I'll try cross compile Oct 05 01:34:49 the bb-kernel repo scripts are already designed for it Oct 05 01:35:43 at least on debian it will even tell you which packages you need to install if anything is missing, not sure whether it has that feature on other distros Oct 05 01:36:09 which repo scripts? Oct 05 01:36:28 https://github.com/RobertCNelson/ti-linux-kernel-dev/tree/ti-linux-4.4.y <-- to build a 4.4-ti series kernel Oct 05 01:37:09 weird why does this not show the dirs, https://github.com/RobertCNelson/ti-linux-kernel-dev Oct 05 01:37:14 https://github.com/RobertCNelson/bb-kernel is same thing for -bone kernels Oct 05 01:37:21 you need to pick a branch Oct 05 01:37:42 ahh damn ok Oct 05 01:38:26 first one has more than just bone Oct 05 01:38:31 second one is bone specific? Oct 05 01:39:02 well 4.4-ti series is currently the default even for beaglebones Oct 05 01:39:44 though the -bone kernels are more beaglebone-specific and I think closer to mainline than the -ti kernels Oct 05 01:40:03 I don't really have good overview of what the differences are currently Oct 05 01:40:30 but the -ti kernels probably also have a lot of beagleboard-x15 specific focus currently Oct 05 01:40:57 guy at work is using x15 Oct 05 01:42:52 I select a branch but doesn't give me a good git url Oct 05 01:43:48 branch isn't part of a git url, you use e.g. git clone -b BRANCH URL Oct 05 01:44:03 (or switch branch after clone) Oct 05 01:45:31 looks like there is at least one debug message during probe -> http://lxr.free-electrons.com/source/drivers/net/can/spi/mcp251x.c#L662 Oct 05 01:45:54 it gets called very late though, so if anything went wrong earlier it will never get there Oct 05 01:46:44 cat /sys/kernel/debug/dynamic_debug/control | grep spi Oct 05 01:46:51 am I using that correctly? Oct 05 01:47:48 useless use of cat ;P (grep spi FILE), other than that it depends on what you want to do Oct 05 01:48:58 I didn't know if that was spi debug stuff or not Oct 05 01:49:21 the control file lists all dynamically togglable debug messages currently known to the kernel Oct 05 01:49:28 build_kernel.sh works with fedora, had me install redhat-lsb-core Oct 05 01:50:50 to enable debug messages in a module immediately when it loads you can use e.g. modprobe mcp251x dyndbg=+p Oct 05 01:51:04 ahh Oct 05 01:51:08 or put options mcp251x dyndbg=+p in a .conf file in /etc/modprobe.d/ Oct 05 01:51:25 or use a kernel commandline parameter Oct 05 01:53:56 thanks for helping Oct 05 01:54:23 it's downloading now, I'm going to bed; tomorrow I'll compile and try after I put in some debug statements Oct 05 01:54:53 yeah, just scatter some dev_dbg() calls around Oct 05 01:54:55 but I can add dev_dbg() or something else Oct 05 01:54:58 ok Oct 05 01:55:26 dev_err is more appropriate for real errors Oct 05 01:56:04 there are two dev_dbg Oct 05 01:56:19 later **** ENDING LOGGING AT Wed Oct 05 02:59:58 2016