**** BEGIN LOGGING AT Mon May 14 03:00:02 2018 May 14 05:30:39 hi May 14 10:56:50 Hi Everyone May 14 10:56:57 I am facing a problem May 14 10:57:20 while using the overlay in u-boot May 14 10:57:40 Can anybody help May 14 10:58:38 mrigendra: not if you do not ask a meaningful question. May 14 10:59:24 I am not sure if I am doing anything stupid. May 14 10:59:46 https://groups.google.com/forum/#!category-topic/beagleboard/newbies/4oo4oL20BEs May 14 11:00:33 I am trying to load an overlay at boot time and use corresponding i2c driver (adxl34x.c) May 14 11:00:37 to load with it. May 14 11:01:21 I have posted the question link that should be better to look at. May 14 11:02:20 Kindly check if I have done things correctly, or if not, what did I missed. May 14 11:06:42 anybody there? May 14 11:07:55 certainly somebody is here, but if nobody answers, probably nobody knows. (at least i do not) May 14 11:08:09 just wait, and maybe ask again in a couple of hours May 14 11:09:21 Ahh ok. I'll wait May 14 11:13:01 hmm May 14 11:32:58 here with experience in uboot overlays? May 14 11:33:23 Anyone here with experience in uboot overlays? May 14 11:45:32 https://groups.google.com/forum/#!category-topic/beagleboard/newbies/4oo4oL20BEs May 14 11:51:56 Hi, May 14 11:52:56 Why my LCD7 clone (00A2) works on a beaglebone white but not on a black ?? May 14 11:53:23 (same identical debian image) May 14 11:55:08 fred__tv: does it conflict with the eMMC? May 14 11:56:30 hmmmm....both are using sdcard boot , does any pin conflicts with eMMC on black ? May 14 11:58:02 yes, even if you boot from SD-card, you MUST disable the eMMC - if there are conflicting pin uses May 14 11:59:10 even with simple console image, the white displays tux logo then cli, the black completely dead... May 14 11:59:27 what is the debug UART output? May 14 11:59:42 (but it starts as I can connect via ssh) May 14 12:00:03 not tried yet.. May 14 12:00:44 'compleatly dead' and 'can ssh in' are fundamentally conflicting statements May 14 12:03:39 I mean the LCD is completely dead, BBB works May 14 12:04:46 probably I need to recall different dtb May 14 12:06:40 the kernel messages might provide some insight May 14 12:17:16 I have few doubts regarding overlay. Can anyone clarify? May 14 12:21:39 nikhil: no, noone can clarify them since nobody can read your mind about what doubts you have or clarifications you need May 14 12:22:26 fred__tv: did you disable eMMC in /boot/uEnv.txt ? May 14 12:22:39 fred__tv: LCD7 A2 conflicts with eMMC May 14 12:23:36 No I am booting from sdcard. I don't use emmc for booting. May 14 12:23:41 that's not what I asked May 14 12:23:56 did you explicitly *disable* eMMC ? May 14 12:24:24 ##This will work with: Angstrom's 2013.06.20 u-boot. loadaddr=0x82000000 fdtaddr=0x88000000 rdaddr=0x88080000 initrd_high=0xffffffff fdt_high=0xffffffff enable_uboot_overlays=1 disable_uboot_overlay_video=1 disable_uboot_overlay_audio=1 disable_uboot_overlay_wireless=1 uboot_overlay_addr0=/lib/firmware/BB-I2C2-ADXL-00A0.dtbo #for single partitions: mmcroot=/dev/mmcblk0p1 loadximage=load mmc 0:1 ${loadaddr} /boot/vmlinuz-${un May 14 12:24:50 fred__tv: oh sorry, I thought you replied to me but I see now it was nikhil May 14 12:25:19 No I didn't disable emmc May 14 12:25:20 nikhil: please pay attention to whether people are talking to you or to someone else May 14 12:25:41 nikhil: and do not repeat yourself every 10-15 minutes. May 14 12:26:14 nikhil: and don't paste multiline output in irc, use a paste service like pastebin.com May 14 12:26:55 Is it better if I show the google group link where I have asked the question? May 14 12:27:09 That contains most of whatever I am going to ask. May 14 12:27:28 I see the link. I didn't know it was you since you changed your name May 14 12:28:25 I was in office, closed that window. May 14 12:32:10 your uEnv.txt is doing all sorts of weird shit that largely overrides u-boot's own logic, hence overlays aren't being applied at all May 14 12:33:22 This uEnv.txt was given here http://eewiki.net/display/linuxonarm/BeagleBone+Black#BeagleBoneBlack-U-BootOverlays May 14 12:33:43 you're also using a customized u-boot or something? May 14 12:33:52 2018.03-dirty May 14 12:34:26 I am not doing any customisations, everything has been pulled from that same eewiki May 14 12:34:47 that link doesn't contain any of the weird stuff you're showing in your groups post May 14 12:35:03 this is what a normal /boot/uEnv.txt looks like => https://pastebin.com/JVr005Cf May 14 12:35:12 (/uEnv.txt is normally not used, only /boot/uEnv.txt ) May 14 12:35:58 but you patched and built u-boot yourself instead of getting a prebuilt one? May 14 12:36:26 I'm pretty sure the prebuilt ones do not have a "-dirty" version May 14 12:37:22 Ok, if you search on that eewiki page with this "Dealing with old Bootloader in eMMC" May 14 12:37:49 you will see the uEnv.txt there. This uEnv.txt does not clear the emmc bootloader. May 14 12:38:01 just get rid of an old bootloader in eMMC May 14 12:38:10 wipe it or replace it with a new one May 14 12:39:12 but I still don't understand why your u-boot even cares about it May 14 12:39:31 can you please get one from https://rcn-ee.com/repos/bootloader/am335x_boneblack/ ? May 14 12:40:17 If I take prebuilt one, I cannot see the source nor I can learn anything. May 14 12:40:36 and when all the instructions are already there on eewiki why would i use prebuilt ones. May 14 12:41:07 but it will let you figure out what's causing your problems May 14 12:41:24 start with something that works, then if you want replace things by custom built stuff May 14 12:41:54 then at least when it breaks you know which part is the problem May 14 12:43:06 first confirm it works with the prebuilt bootloader, no /uEnv.txt, and your overlay enabled via /boot/uEnv.txt (i.e. ignoring all the comments it would look something like https://pastebin.com/H5t2nQUs ) May 14 12:48:02 Yes the prpblem was uEnv.txt . May 14 12:48:17 I changed to yours and added and my overlay is added at boot. May 14 12:52:04 is there an HARDWARE conflict between eMMC and LCD7 A2 ? May 14 12:53:26 nikhil: as a general rule, avoid anything dated many years ago :P merely mentioning that 2013 angstrom u-boot probably scared people away May 14 12:53:43 fred__tv: yes May 14 12:54:00 fred__tv: that's why you need to explicitly disable eMMC via /boot/uEnv.txt May 14 12:54:53 but, If I need LCD7 with eMMC ??? is this why rev A3 was made ? May 14 12:58:31 yes i guess a slight change will make things break in BBB. May 14 12:58:49 fred__tv: yes May 14 12:59:09 One last question, if I compiled a driver as a module, will it be loaded once the device overlay is loaded? May 14 12:59:54 yes, kernel modules containing drivers get automatically loaded whenever a device is encountered that needs that driver May 14 12:59:54 I think it should be loaded by kernel as kernel keep a table with the driver supporting devices. May 14 13:01:09 ok. thats it for me. as of now my driver is not getting loaded but I will check it myself. May 14 13:01:18 many thanks for the help. May 14 13:03:01 nikhil: the 'compatible' property in DT needs to match one of the strings in the driver's of_match_table May 14 13:05:31 zmatt: yes that is the only thing to look. May 14 13:05:42 hmm, that part looks fine actually May 14 13:06:03 try running "sudo depmod" May 14 13:06:13 ok May 14 13:07:02 normally that would be run automatically after installing a new kernel, but I've seen that sometimes fail to happen for some reason May 14 13:09:40 btw i2c2 on P9.19-20 is enabled by default, it's quite pointless to manually enable it May 14 13:10:56 adding the node for your i2c device to &i2c2 should be all that's required May 14 13:10:56 umm I checked with depmod. The driver is not loaded. May 14 13:11:31 it's still not loaded automatically at boot? that's really odd May 14 13:13:03 oh! if you're using an initramfs, try updating that with "sudo update-initramfs -u" ... or better yet, just remove initramfs since it's not useful and just slows down boot May 14 13:14:15 (initramfs is only required if the kernel is not able to mount the root filesystem on its own with its built-in drivers) May 14 13:14:23 I don't think I ever flashed initramfs. Its all filesystem, zimage, uboot.img only May 14 13:15:17 initramfs is a file in /boot May 14 13:16:30 a small name mismatch can cause my driver not to load, correct. May 14 13:17:20 of course May 14 13:17:34 like compatible = "adi,adxl345" in overlay file and compatible = "adi,adxl34x" in driver. May 14 13:18:07 the driver source code says adi,adxl34x is deprecated and adi,adxl345 should be used instead May 14 14:28:26 Corrected name in the driver and yes, it works. May 14 14:30:56 uhh what? May 14 14:31:21 if you need to patch the driver you're probably doing something wrong May 14 14:33:05 if you put compatible = "adi,adxl345", "adi,adxl34x"; in your DT then it will comply with the recommendation to use adi,adxl345 while still being compatible with old kernel that only support "adi,adxl34x" (if it is necessary to support such an old kernel) May 14 14:45:59 correct, I changed compatible as you said, previously it was mismatching with my device id. May 14 14:46:15 you're using an old kernel? May 14 14:46:46 VERSION = 4 PATCHLEVEL = 4 SUBLEVEL = 127 EXTRAVERSION = NAME = Blurry Fish Butt May 14 14:46:46 I didn't say anything about changing compatible, I said it looked fine to me May 14 14:46:52 Is this old? May 14 14:47:03 4.4 is pretty old yes May 14 14:48:27 I said about this May 14 14:48:32 in the driver static const struct i2c_device_id adxl34x_id[] = { { "adxl345", 0 }, { } }; May 14 14:48:55 it is using "adxl345" but inside DT it is using adi,adxl34x May 14 14:49:05 then how can it match and do the probing. May 14 14:49:43 so I think it was looking fine to me too but clearly because of this mismatch my driver didn't loaded. May 14 14:49:47 this doesn't match what you linked to your in google groups post, https://pastebin.com/ef8Uh06D (lines 142-153) May 14 14:49:53 which is what I was looking at May 14 14:50:44 Ohh I am so sorry, I have changed just this thing today and not updated their May 14 14:50:49 Hi, what to do with certificate problems? May 14 14:51:09 On a fresh install I can't pull from Git, nor can I use the update-kernel script because of not trusted certificates May 14 14:51:16 just changed adxl34x to adxl345 in place of it while experimenting. May 14 14:51:21 I already tried reinstalling ca-certificates May 14 14:52:56 nikhil: anyway, like I said, just use compatible = "adi,adxl345", "adi,adxl34x"; to retain compatibility with old kernel versions May 14 14:54:12 bou4: you will need to be more specific. what is giving what error about what, and what happened when you tried to update ca-certificates? May 14 14:54:13 zmatt :Yes, thanks for help. I will keep this in my head and have documented this procedure in a blog too, so that it can be used in future. May 14 14:56:08 https://pastebin.com/sY0Qq8Ws May 14 14:56:23 this after a fresh reflash and apt-get update / upgrade May 14 14:56:38 brb May 14 14:57:54 bou4: hmm that's not very specific. maybe try wget https://github.com/RobertCNelson/boot-scripts/ to see if you get a more informative error May 14 15:03:38 Okay, so I'm pretty confused by u-boot. I'm trying to netboot a kernel image (this time, instead of compiling my own I took the uImage from a beaglebone install media old enough to have one (angstrom) so I know I have a good kernel), and I keep getting "ERROR: can't get kernel image!". I'm a little confused by how there are three seperate addresses in the boot process, one that I copy the image wo with tftpboot, one called "Load Address" and one called "En May 14 15:05:31 n8vi: your message was too long and got truncated. also, why aren't you following my example of using a FIT? May 14 15:06:56 I wasn't able to get that to work either May 14 15:07:32 what problem were you having? May 14 15:07:52 Let me retrace my steps on that and I'll give you a full report May 14 15:08:04 since I've used it successfully both for usb boot and ethernet boot May 14 15:10:00 what format kernel do you use? I just see "kernel" in your tree. Is that a uImage or a zImage or what, or does it not matter? May 14 15:10:47 oh, is it just the "Image" file May 14 15:10:57 zImage May 14 15:11:13 okay, thanks. May 14 15:12:28 I've never used an uImage, nor really know what it is even. my vague impression was that it's something obscure or ancient May 14 15:12:46 okay. May 14 15:12:53 the fact that you had to fish one out of an ancient angstrom seems to confirm that May 14 15:12:58 What's fdt.dts? same as lunux.its? May 14 15:13:04 linux.fts May 14 15:13:06 no May 14 15:13:16 fdt.dts is the device tree May 14 15:14:17 actually ignore the part of my makefile where it builds the .dtb from .dts May 14 15:14:33 Can I just use a am335x-boneblack.dtb I have lying around here?? May 14 15:14:36 normally you'd use a dtb that was compiled when the kernel was compiled May 14 15:14:38 yes exactly May 14 15:14:41 okaty May 14 15:31:05 okay, so I've done "tftpboot :linux.itb", now it's just "boot"? May 14 15:44:29 n8vi: I just did tftpboot && bootm May 14 15:44:33 no sorry May 14 15:44:47 dhcp && bootm I think? I've mentioned it previously May 14 15:45:01 bootm. Okay, I'll try that one May 14 15:45:04 yep May 14 15:49:39 okay, we're back to "Starting kernel ..." a long pause and a reset, so I'm back to wondering if I'm compiling my kernel incorrectly May 14 15:50:02 which you should verify by booting it from sd card first May 14 15:51:29 The problem is I'm kinda not sure what I'm doing there either. The image I need for netbooting is a zImage, I guess, and the one on the sd card is vmlinuz ... May 14 15:51:42 which is also zImage May 14 15:51:53 is it? Are the two interchangable? May 14 15:52:09 they're the same thing yes May 14 15:52:14 I'm pretty sure May 14 15:52:31 If so, I'll just use a vmlinuz off a working sdcard and try netbooting that May 14 15:59:31 I've been using exfat on an USB to copy files from the beaglebone, but every now and again it crashes the unit. If I go w/ a smaller SD & fat32 it's fine. I'm hoping I'm missing something simple, as I just apt-get installed the exfat driver from whatever repo it lives in. May 14 15:59:45 Would be better to pull sources and build it? May 14 16:00:32 Oh. It crashes on unmount. Copy works fine. May 14 16:06:21 Ragnorok: why just use fat32 instead of exfat May 14 16:06:26 uh, sentence fail May 14 16:06:30 just use fat32 instead of exfat May 14 16:06:36 (or "why are you using exfat?") May 14 16:07:39 exfat is a proprietary and patented filesystem. it is unlikely to ever be supported well by linux May 14 16:11:33 lol Would like >32GB USB. All I've read indicates fat32 is limited to 32GB. Is that incorrect? May 14 16:12:50 uh, no? May 14 16:13:03 max size for a fat32 filesystem is 2 TB May 14 16:13:26 Weirdness. I searched all over for that. (shrug) There it is. May 14 16:13:53 Must've experienced vastly inferior google foo that day. It happens. May 14 16:14:30 your search apparently didn't include the wikipedia page for FAT May 14 16:14:57 which lists the max volume size on the right side near the top of the page May 14 16:15:40 Not *specifically*, but today I see that clearly. Guess I should have re-tried searching first. (smirk) May 14 16:16:02 Thanks! May 14 16:17:02 This is the sort of thing I found: https://www.techrepublic.com/blog/windows-and-office/format-fat32-drives-beyond-32gb-limit/ May 14 16:17:20 32 GB is not a limit for anything I'm aware of. the only context where I've seen that specific number show up is that the SD Card Association requires exfat to be used on sd cards larger than 32 GB May 14 16:17:54 It clearly says it's a windows issue as well, which I don't recall reading before. Annoying. May 14 16:17:54 that's really weird May 14 16:19:12 so that sounds like an intentional restriction in windows, presumably to nudge users towards using ntfs instead May 14 16:19:13 Guess I had to ask before I could read. I hate daze like that. May 14 16:19:31 or exfat May 14 16:19:39 Yeah. It do sound that way. May 14 16:22:57 lol, as long as it's smaller than 1 TB you can still use "format /FS:FAT32 X:" or whatever in the DOS command prompt May 14 16:23:14 so it's really just a restriction that exists in the user interface May 14 16:24:00 At least it was as I was hoping ... something simple ... that being "rtfm moron". I can live with that. May 14 16:24:12 yeah or "try it and see" May 14 16:24:46 mkfs.fat obviously would have worked without a problem May 14 16:25:09 I'll be finding out soon enough. (wink) May 14 16:25:13 any reason why you're using sd card rather than network? May 14 16:25:35 Unit is not in a networked enviroment. May 14 16:27:04 oh you're using it with keyboard and display connected to the beaglebone? May 14 16:27:18 I sometimes forget there are people who use the beaglebone like that May 14 16:28:09 I don't know if you recall a couple of years ago I was doing a recording system. One use case for that system is vehicular, where it has no network. The only way to get recordings off it is sneaker net. May 14 16:28:28 ok May 14 16:30:54 how would I wait using epoll on the change of state of a gpio pin? May 14 16:31:24 if I just wait on /sys/class/gpio/gpio46/value, it doesn't wait May 14 16:32:31 TonyTheLion: be sure to request the EPOLLPRI event, not EPOLLIN May 14 16:32:38 ah May 14 16:32:43 let me try that May 14 16:32:57 I know epoll not. Only used inotify so far. May 14 16:32:59 and configure /sys/class/gpio/gpio46/edge to 'rising', 'falling', or 'both' depending on which edges you want to detect May 14 16:33:09 epoll and inotify do totally different things May 14 16:33:22 I know. That's why I didn't answer. (grin) May 14 16:37:35 (also both kinda suck, even though they're the best thing available for their purpose on linux) May 14 16:40:22 (inotify sucks more) May 14 16:42:28 (have I mentioned yet that inotify sucks?) May 14 16:44:01 Let me check history. May 14 16:44:35 `open(gpio_pin_val, O_RDONLY | O_NONBLOCK))` are these the correct flags to open the `value` file with for the GPIO? May 14 16:44:59 the O_NONBLOCK does nothing, but other than that yesh May 14 16:46:14 note you can use pread to avoid having to use separate seek() and read() May 14 16:47:04 ohhh May 14 16:47:27 https://pastebin.com/z3ZFZAJB May 14 16:47:40 this is a C++ gpio class I wrote May 14 16:48:16 (the "die.h" it includes => https://pastebin.com/3rNiaWWa ) May 14 16:50:54 oh yeah and you should use O_CLOEXEC May 14 16:51:07 (not just for this but in general for any system call that produces a file descriptor) May 14 16:52:43 demo for my gpio class: https://pastebin.com/jn3K3A9J May 14 16:57:30 so you cannot just wait with a poll without adding an additionall while loop that checks the value? May 14 16:58:03 I did that to filter glitches May 14 16:58:46 so its possible it read a value that is incorrect? May 14 16:59:14 eh, no? May 14 16:59:30 I suspect it depends on what's driving the pin, what "incorrect" means. May 14 17:00:13 but because this demo is logging the "current value", I don't want it to log "high" multiple times in a row or "low" multiple times in a row May 14 17:00:27 which could happen if it transitions multiple times fast enough May 14 17:00:33 my program seems to just wait and changes in the pin value don't register at all May 14 17:00:48 did you configure the 'edge' attribute? May 14 17:01:07 I set it to both May 14 17:01:25 then there might be a bug May 14 17:01:30 you do read the correct value? May 14 17:01:55 does my example code work for you? (compile with -std=gnu++1z ) May 14 17:02:17 let me try your example May 14 17:02:21 gimme a few mins May 14 17:06:45 zmatt: your example works May 14 17:06:51 so my code is wrong May 14 17:06:54 weeeee May 14 17:07:02 zmatt ftw May 14 17:32:47 oh good, I've broken SPL/u-boot or my dhcp/tftp server in such a way that that won't even load. How the hell did I do that? May 14 17:33:25 * Ragnorok blames the electrons - if it weren't for them this problem wouldn't exist May 14 17:33:44 If it weren't for them, many problems wouldn't exist May 14 18:47:16 I want to open a file at startup in GUI May 14 18:47:27 anybody knows how to do it? May 14 18:50:25 I'm not familiar with lxqt May 14 19:49:39 anyone managed to run feeBSD on BBB May 14 19:50:17 I am booting up RaspBSD but there is no pck, git, gcc,wget curl.... How can I put software on it ? May 14 19:51:33 this sounds like a generic freebsd question rather than a beaglebone-related question, wouldn't asking in a freebsd channel make more sense? May 14 19:57:01 thank you May 14 20:20:14 Hi **** ENDING LOGGING AT Tue May 15 03:00:02 2018