**** BEGIN LOGGING AT Mon Sep 24 02:59:59 2018 Sep 24 09:58:23 LetoThe2nd: feel free to invite Sigyn ( freenode antispam bot ) Sep 24 10:11:30 how to find bord image Sep 24 10:12:28 look at the channel topic Sep 24 10:12:37 "neeext!" Sep 24 10:44:51 lol Sep 24 10:58:58 and I was even helpful Sep 24 10:59:53 I could have answered "get camera, take picture of board, board image is on camera" Sep 24 11:25:51 alright, so mem sleep is also broken on bbb, not just bbbw. Sep 24 11:26:00 sigh Sep 24 11:41:11 wasn't it a kernel config issue? Sep 24 11:41:46 I don't think there's any hw restriction preventing the use of suspend or deepsleep on the bbb Sep 24 11:48:50 zmatt: there was a kernel config issue, yes. And it went unnoticed for quite a long time which means nobody is actually testing PM. And so it's not wonder it got broken. Sep 24 11:49:21 zmatt: and it's also no wonder that other PM issues crept in because nobody gives a shit. Sep 24 11:50:26 fortunately I have a working image and a broken image and I can probably make a differential analysis Sep 24 12:00:45 hm, 50 mA in DS0 is also not good Sep 24 12:03:21 oh that is interesting Sep 24 12:03:28 pm33xx pm33xx: PM: Successfully put all powerdomains to target state Sep 24 12:03:36 so it thinks it reached the target state Sep 24 14:25:40 thinkfat: be sure to notify rcn about the kernel config issues Sep 24 14:25:55 zmatt: already done, he fixed it in his tree Sep 24 14:26:15 zmatt: but current images still seem to be output without the fix Sep 24 14:29:36 i want access control for my own company Sep 24 14:29:44 what i need to buy from beagleboard? Sep 24 14:30:40 i want door access control Sep 24 14:30:47 i can use beagleboard ? Sep 24 14:32:47 dokuzi_: your question is too vague to be answered Sep 24 14:34:37 already i have 5 device for access control installed in my office's dorrs Sep 24 14:34:39 doors* Sep 24 14:34:53 to control check in/out of my worker Sep 24 14:35:36 something like this https://www.plexa.com/it/soluzioni/controllo-accessi Sep 24 14:40:56 dokuzi_: it's as if you're asking "I want a house, can I make it with these stones?". Perhaps technically yes, but considering how many other components are required, and the expertise required to make the end result from those components, you're probably not capable of doing so if you're asking a question like this Sep 24 14:42:36 i want to know if beagleboard has some device and software for "door access control" Sep 24 14:42:38 this is Sep 24 14:42:45 beagleboard has or not something like this? Sep 24 14:45:03 dokuzi_: no, just get an off the shelf sulotion. Sep 24 14:45:40 (while its technically possible, its pointless if what you need is already perfectly available) Sep 24 15:06:41 hi Sep 24 15:07:48 we have required San cloud Beagle Bone Enhanced (Sancloud BBE WiFi) Sep 24 15:07:59 Upgraded Ethernet speed (gigabit) 1GB DDR3 RAM 4GB 8-bit eMMC on-board flash storage 3D graphics accelerator NEON floating-point accelerator 2x PRU 32-bit microcontrollers Sep 24 15:08:08 this is our end user requirement Sep 24 15:09:03 please email us prices on below email Sep 24 15:09:13 sales@alphaint.com.pk Sep 24 15:09:22 waiting for your urgent response Sep 24 15:09:23 tahir: contact sancloud, not us Sep 24 15:09:28 they have a webshop Sep 24 15:09:51 this is a support forum, not a sales department Sep 24 15:10:12 :) Sep 24 15:10:15 we do not directly represent any particular manufacturer Sep 24 15:10:41 can u please send me sales department email id Sep 24 15:11:39 is google broken? Sep 24 15:12:00 http://www.sancloud.co.uk Sep 24 15:12:02 this is their website Sep 24 15:12:08 it has a webshop and a contact form Sep 24 15:13:10 ok thanks Sep 24 16:48:27 zmatt: you're just too kind ;) Sep 24 16:50:06 sometimes Sep 24 19:58:55 zmatt: this hwmod patch seems to be forever ignored http://lists.busybox.net/pipermail/buildroot/2016-July/167543.html Sep 24 19:59:21 zmatt: but there is nothing in the devicetree either that prevents shutdown of debugss when linux boots Sep 24 19:59:41 zmatt: I remember I had the same problem when I started looking into ds0 the first time Sep 24 20:00:00 I think you can just stick a property on debugss in DT if you want to force it to remain active Sep 24 20:00:08 ti,no-idle; or something like that Sep 24 20:00:23 zmatt: yes, only that there isn't even a node for it in the device tree Sep 24 20:00:52 zmatt: can I just put one, without a compatible, with just a ti,hwmods = "debugss" ? Sep 24 20:00:59 lemme check how I declared it Sep 24 20:02:32 I have this: https://pastebin.com/raw/kakfWC1y although I don't really remember if I had a specific reasoning for it Sep 24 20:02:53 seems like a sane enough declaration though Sep 24 20:04:41 I'm adding ti,no-reset-on-init; Sep 24 20:04:46 just for good measure Sep 24 20:04:59 iirc, you cannot idle it anyway Sep 24 20:05:55 hmmm I could swear uboot used to load some overlays on the bbb... Sep 24 20:05:57 kinda sucks that they fucked up the icepick <-> prcm integration. jtag debuggability is not supposed to depend on the OS leaving clocks enabled Sep 24 20:06:01 it doesn't now... Sep 24 20:06:11 default images use u-boot overlays yes Sep 24 20:06:30 yeah, it should load some, but I don't see it happening any more Sep 24 20:06:36 I wonder what I messed up... Sep 24 20:06:52 if you're booting from sd card, beware of the possibility of an old u-boot on eMMC fucking things up Sep 24 20:07:04 that might very well be. Sep 24 20:07:11 I'm indeed booting from SD card Sep 24 20:07:18 wipe eMMC to force u-boot to be loaded from sd card Sep 24 20:07:31 (or at least wipe sector 256) Sep 24 20:07:37 humm Sep 24 20:07:50 sudo blkdiscard /dev/mmcblk1 if you don't care about eMMC contents Sep 24 20:07:54 my working setup is on eMMC, I still need that for comparison... Sep 24 20:07:57 ah ok Sep 24 20:08:11 then power up with S2 button held down Sep 24 20:08:17 better. Sep 24 20:08:20 to temporarily remove eMMC from the boot order Sep 24 20:08:56 yes, that was it! Sep 24 20:08:58 (the setting remains in effect even if you reboot. it's only sampled at power up) Sep 24 20:13:45 now that's interesting Sep 24 20:13:57 hmm? Sep 24 20:14:01 when I boot with the old bootloader from emmc, DS0 works Sep 24 20:14:13 when I boot with the new loader from SD card, DS0 doesn't work Sep 24 20:14:45 probably because cape-universal is enabling a billion devices and one of their drivers fucks up? Sep 24 20:16:20 that's at least a hypothesis... Sep 24 20:17:43 try without the eqep one Sep 24 20:18:13 I vaguely remember looking at it and feeling my heart fill with darkness Sep 24 20:18:27 hm Sep 24 20:18:39 I'm disabling u-boot capemgr alltogether Sep 24 20:19:04 u-boot overlays doesn't directly affect anything though, it's just a way to compose the DT passed to the kernel Sep 24 20:20:33 without it it uses the old monolithic DT instead of assembling one from a barebone base DT and a bunch of overlays Sep 24 20:25:43 also if u-boot overlays are disabled then the capemgr kernel module (runtime overlays) will be active again by default, which you may not want Sep 24 20:26:13 Hey, guys. If I have custom am335x boards with specialized ethernet PHYs connected to the MII, and if I transmit a frame from one board to another and I can see the frame on the MII on both sides of the link but the OS on the receiving side doesn't increment its received packet counter, what could I be missing? mii-tool on both sides correctly reports the values of the PHY's registers and shows link up, I can set the registers with https://github.com/wkz/p Sep 24 20:26:30 ok, first check with u-boot overlays disabled and cape universal disabled, ds0 works again Sep 24 20:26:40 I feel like it must be something simple and stupid Sep 24 20:27:06 thinkfat: note that if u-boot overlays are disabled, most settings in /boot/uEnv.txt (e.g. the option for cape-universal) will do nothing anymore Sep 24 20:27:52 testing now with u-boot overlays but without cape-universal Sep 24 20:28:20 n8vi: your text was cut off (halfway in the url) Sep 24 20:28:42 I can set the registers with https://github.com/wkz/phytool, and all my pinmuxing appears correct Sep 24 20:28:53 ok, already broken without cape-universal Sep 24 20:28:59 so, it must be one of the overlays Sep 24 20:29:06 thinkfat: interesting Sep 24 20:29:09 or the base DT Sep 24 20:29:14 I'm seeing the frame on both sides of the MII with a logic analyzer Sep 24 20:30:00 that is to say on the TX side of the MII on the sending board and on the RX side of the MII on the receiving board Sep 24 20:30:41 so your link is up and you can transmit packets, but nothing is received? not even any error? hmz Sep 24 20:31:02 yup Sep 24 20:31:28 from "ifconfig eth0": Sep 24 20:31:29 RX packets 0 bytes 0 (0.0 B) Sep 24 20:31:29 RX errors 0 dropped 0 overruns 0 frame 0 Sep 24 20:31:55 from mii-tool: eth0: 100 Mbit, full duplex, link ok Sep 24 20:33:01 Looks the same on both sides and I've tried transmitting packets both ways Sep 24 20:33:23 not sure, sorry. I'd probably try to see if there's anything interesting to see in the statistics registers of the ethernet subsystem Sep 24 20:33:36 I've looked all through them Sep 24 20:33:42 they look totally happy Sep 24 20:33:58 I feel like there's a "receiver enable" that I need to set on the am335x or something Sep 24 20:34:20 how have you checked those registers? Sep 24 20:34:25 because, evidently, transmit is working fine Sep 24 20:35:00 I've looked at them with "mii-tool -v -v" and with the "phytool" program I mentioned earlier Sep 24 20:35:16 I wrote a python wrapper to decode all the bits for this specific PHY Sep 24 20:35:43 there's all sorts of setup that needs to be done to receive packets, but obviously the kernel will normally do that. I'm assuming your hardware situation is same as on the bbb (apart from maybe different phy) and ditto the DT Sep 24 20:35:55 no I don't mean the phy, I mean the ethernet subsystem of the AM335x Sep 24 20:36:06 hmmm Sep 24 20:36:36 How do I find those registers? Sep 24 20:37:46 zmatt: hm, when I use the u-boot capemgr it always loads a boneblack-uboot-univ.dtb Sep 24 20:37:47 * n8vi finds some interesting google results and starts reading Sep 24 20:38:08 zmatt: but that already contains the cape-universal stuff Sep 24 20:38:27 n8vi: https://pastebin.com/raw/3XkcYpAq Sep 24 20:38:40 thinkfat: only the pinmux groups, but they're not referenced by default Sep 24 20:38:42 iirc Sep 24 20:38:53 awesome. thanks, zmatt, I'll take a look Sep 24 20:39:39 zmatt: hm Sep 24 20:39:58 zmatt: all the x_y_pinmux nodes have status = "okay" Sep 24 20:40:16 and compatible = "bone-pinmux-helper" Sep 24 20:40:21 so I guess they are being used Sep 24 20:40:55 eh, there are pinmux helpers in the base DT? (apart from the two pins normally used for cape i2c, iirc P9.19 and P9.20 ?) Sep 24 20:41:35 depends on what base dt is for you Sep 24 20:41:50 am335x-boneblack-uboot.dts or something like that Sep 24 20:42:04 when I enable u-boot overlays, it switches to a boneblack-uboot-univ.dtb Sep 24 20:42:27 that sounds like it loads the cape-universal stuff Sep 24 20:43:07 anyway, not solving that riddle right now Sep 24 20:43:11 time to end the day Sep 24 20:43:17 ttyl Sep 24 20:46:07 yep that means you have cape-universal enabled Sep 24 20:48:47 based on a quick look at the boot script, if enable_uboot_cape_universal is set (non-empty) then it'll load am335x-boneblack-uboot-univ.dtb, otherwise am335x-boneblack-uboot.dtb Sep 24 21:42:12 HELP Sep 24 21:43:19 I am trying to figure out how to write a CSV. file for beaglebone. I apologize in advance as a I am a novice Sep 24 21:43:44 the same way you would on any other linux system Sep 24 22:14:03 zmatt: did you read my last question? or my message was blocked? Sep 24 22:14:23 didn't see it? Sep 24 22:16:09 okay here it is: I hope this doesn't sound vage: How is processed the "compatible" line in the DT?. i.e. who reads that line, and what acctions are taken after parsing it? (e.g. In C, the preprocessor replaces #include, yadda yadda...) Sep 24 22:17:16 it's not clear if you mean how device tree sources are parsed or what is being done with the value of that property Sep 24 22:17:55 both (maybe I'm asking too much information) Sep 24 22:18:38 but the "compatible" property is a string-array (i.e. one or more NUL-terminated strings concatenated together) which the kernel uses to locate a matching driver Sep 24 22:20:19 So the kernel is processing that line itself? (the binary version of that line I guess) Sep 24 22:20:56 the binary version yes Sep 24 22:21:05 see e.g. hexdump -C /proc/device-tree/compatible Sep 24 22:24:54 although dts has various ways of providing the value of a property (e.g. as strings or 32-bit integers), ultimately the contents of a property is just a sequence of bytes. DT is not typed Sep 24 22:30:01 then, it loads the associated module/driver? if it is a loadable module? Sep 24 22:35:53 yeah, loading uses the modalias system, similar to how it finds a kernel module for a pci device or usb device based on their identification Sep 24 22:38:36 and some lines of that DT node, are passed to such module by the kernel? Sep 24 22:40:41 the driver has access to the DT node and can read its properties, iterate over its child nodes, etc Sep 24 22:42:29 I love those details Sep 24 22:45:13 how things works, and not just expecting them to work Sep 24 22:45:20 e.g. in drivers you may find of_property_* calls to read DT properties, or the newer device_property_*/fwnode_property_* calls (meant to unify ACPI properties and DT properties) Sep 24 22:57:35 thanks zmatt :D **** ENDING LOGGING AT Tue Sep 25 02:59:59 2018