**** BEGIN LOGGING AT Mon Jul 24 03:00:02 2017 Jul 24 07:08:56 hi jkridner: any updates on h/w?? Jul 24 07:09:37 nerdboy:,ds2: where are you, I am feeling alone .. :( Jul 24 07:14:45 * nerdboy about to go to sleep... Jul 24 07:15:17 i'll have to look at your branch in the morning Jul 24 07:15:30 just ping me whenever you are free Jul 24 13:16:18 ravikp7: where is _device defined? Jul 24 13:17:18 jkridner: used it by mistake, corrected now Jul 24 13:17:42 commit an hour ago Jul 24 13:17:50 k. with the "new" spl, I have uboot running. Jul 24 13:18:08 that is, I have the transfer compelete.... never see UMS start. Jul 24 13:18:19 ravikp7: likely due to no card inserted. Jul 24 13:18:27 or no eMMC on my board. Jul 24 13:18:49 jkridner: I'm also not using an sd card Jul 24 13:18:57 we'll want to revisit the uboot code to handle both eMMC and uSD and deal with when they don't exist and when they are inserted. :-/ Jul 24 13:19:15 ravikp7: I'm using a PocketBone, which doesn't have an on-board eMMC. Jul 24 13:19:56 I'm thinking PocketBone users would exclusively use this tool for booting/flashing. Jul 24 13:20:19 though I suppose etcher.io and another adapter would work fine for them. Jul 24 13:20:30 jkridner: k, code for server seems good to go now? any further changes? Jul 24 13:21:05 btw, nice to see less statefulness in the code! Jul 24 13:21:26 I'm able to run the server and reset the board and see the SPL transfer happen again. that's exactly what I expected. Jul 24 13:21:56 please keep this sort of clean-up coming. I think the code is getting better in that regard. Jul 24 13:22:25 what more could be done for clean up, any suggestions? Jul 24 13:25:36 ravikp7: big stateful thing to be cleaned up right now is the fact you don't parse the IN packets to decide what to do. Jul 24 13:27:18 jkridner: I parse the packets against all possible, then decide what do do? current checking out one byte to identify the request isn't a good solution? Jul 24 13:27:27 instead of being stateful such as "1) getBOOTP & sendBOOTP, 2) getARP and sendARP, 3) getTFTP and sendTFTP", you should just respond to whatever request you get. Jul 24 13:27:58 well, you only have to handle those three I think. if you get something other than those three, generate some warnings. Jul 24 13:28:15 jkridner: It's stating that only after identifying the request Jul 24 13:29:23 jkridner: I redirect inTransfer again on receiving request other than those 3, shouldn't that work? Jul 24 13:29:42 I'm saying this line shouldn't exist: https://github.com/ravikp7/node-beagle-boot/blob/master/main.js#L163 Jul 24 13:30:03 instead, you should have a handler installed for the inEndpoint that should handle whatever type of packet you get. Jul 24 13:30:53 jkridner: it's not what I wanted to show you, see the other branch Jul 24 13:31:00 true-server Jul 24 13:31:14 ah! Jul 24 13:31:19 yes! Jul 24 13:33:22 jkridner: now, things seem good? stating request handling is done only after identifying request Jul 24 13:34:15 FYI, I'll abandon https://github.com/jadonk/node-beagle-boot/tree/stringify now, but wanted for you to see this stringify package in case we wanted more debug statements. Jul 24 13:35:15 jkridner: yeah, was thinking of using it for more debug info Jul 24 13:35:48 * jkridner wishes we could get some console output/input from u-boot. Jul 24 13:37:15 jkridner: why would someone want to mount sd card through this server? Jul 24 13:37:31 inserted in board Jul 24 13:37:32 ravikp7: PocketBone only has uSD. Jul 24 13:38:22 I wonder if it is easy to patch the uboot bootcmd script on the binary. Jul 24 13:38:34 jkridner: yeah, should be Jul 24 13:39:30 not sure if there is a CRC, but the string shows up clearly enough. ;-) Jul 24 13:39:44 "ums 0 mmc 1" Jul 24 13:40:24 ravikp7: did you rebuild 'uboot' and 'spl' from source? do you have the build instructions? if we can do so, we can bisect to find the issue. Jul 24 13:41:38 jkridner: yeah, I tried bisect, same finding as what m_w found Jul 24 13:43:11 ravikp7: which was? Jul 24 13:43:38 jkridner: finding the logs.. Jul 24 13:44:11 jkridner: see the thread https://groups.google.com/forum/#!topic/beagleboard-gsoc/8osyK8ttvVk Jul 24 13:45:22 jkridner: using this config am335x_evm_usbspl_config Jul 24 13:48:55 jkridner: merging true-server branch with master Jul 24 13:48:59 fyi, did "LANG=C sed "s/ums 0 mmc 1/ums 0 mmc 0/" bin/uboot > bin/uboot-mmc-0" and got this working with PocketBone! :-) Jul 24 13:49:38 unfortunately, I had to insert the card while the code was loading because this particular SD card tries (and fails) to boot. hope it is clean with a clean card. Jul 24 13:50:51 that e-mail thread had fallen out of my mind. thanks. Jul 24 13:52:59 seems the u-boot list wasn't in copy. :( Jul 24 13:53:54 jkridner: I sent emails to uboot list, m_w also replied to thread, still no helpful replies :( Jul 24 13:55:16 jkridner: copy that uboot list thread to you? Jul 24 13:56:28 hmmm... my gmail address seems to be devoid of recent u-boot list traffic. Jul 24 13:59:49 jkridner: copied to you Jul 24 13:59:51 seems I've been off the list for a long time and didn't notice. :( Jul 24 14:13:28 * jkridner wonders why rx_submit would be called without a request. m_w did you think through that any? Jul 24 14:16:04 * jkridner looks at https://pastebin.com/t2WNLQsQ which is a very helpful bisect log Jul 24 14:16:39 * jkridner is confused why patch level 238 is run. 177 looks good and 178 looks bad, showing the bad patch. Jul 24 14:17:09 oh, wait, 178 is good. Jul 24 14:17:13 now I'm extra confused. Jul 24 14:19:03 ravikp7: do you understand the bisect log and how it points out the issue? Jul 24 14:19:33 200 is good Jul 24 14:19:56 238 is bad. Jul 24 14:21:02 I don't see any runs between 200 and 238. is the log truncated? Jul 24 14:24:16 jkridner: we'd have to consider this also, that the uboot version that we started bisect with, worked on linux and not on OSX Jul 24 14:27:46 jkridner: I didn't complete the bisect, got a low spec pc, was taking much time to compile uboot, just got few compiles which showed similar results. that log is from m_w Jul 24 14:29:19 yeah, but we should be able to utilize m_w's result. I'm looking for the type of failure that was generated by e02ee2548afe22f7d197b89df2ae7bd72f1df630 Jul 24 15:07:22 BOOT_COMMAND patch out of tree? Jul 24 15:08:34 duh: https://pastebin.com/rZ3R3zRs Jul 24 15:11:42 jkridner: something also should be there before ums command? Jul 24 15:12:25 should it? Jul 24 15:12:58 no idea, just asking Jul 24 15:16:53 jkridner: should I run ums command for both emmc and sd, and mount both if available? Jul 24 15:17:30 yeah.... not sure of the failure modes. looking at getting mainline working first. Jul 24 15:18:17 jkridner: using uboot binary build from mainline Jul 24 15:35:05 but that one isn't working to load uboot, correct? Jul 24 15:35:19 jkridner: yes Jul 24 17:40:26 jkridner: if we want to mount sd card only in pocketbone, I can use this bootcmd " if mmc dev 1; then; ums 0 mmc 1; else ums 0 mmc 0; fi; " Jul 24 17:47:21 ravikp7: pocketbone? Jul 24 17:47:34 ravikp7: you are adding support for it as well? Jul 24 17:49:47 m_w: same processor, so I guess would work on that too without additional effort. jkridner tested on it, it works. we'll just mount sd card instead of emmc there Jul 24 17:50:41 all Bone family covered for direct flashing :) Jul 24 17:52:26 cool Jul 24 18:04:20 m_w: yeah, seems really important for PocketBone users IMO to be able to use this app. hopefully "ums 0 mmc 0,1" does 90% of the work. Jul 24 18:04:35 It worked fine with "ums 0 mmc 0" Jul 24 18:05:22 jkridner: so that program in circuit instead of on their PC? Jul 24 18:05:36 jkridner: how about just mounting sd card only when emmc is not present? Jul 24 18:07:15 m_w: yes, no additional adapter required.... really, I'd want u-boot-ums to be just one option of what can be loaded. booting to a ramdisk should also be an option so no uSD would even be required to use the board. Jul 24 18:07:49 that should work when emmc is disabled by uboot_overlays right? Jul 24 18:07:55 jkridner: very nice, have you been able to get any further with the u-boot usb issue upstream? Jul 24 18:07:56 ravikp7's framework should make it easy to plug in different responses to the TFTP requests. Jul 24 18:08:12 m_w: I rebuilt u-boot from mainline to test. Jul 24 18:08:28 m_w: just went upstairs to fetch a Black (instead of PocketBone) and a serial adapter. Jul 24 18:08:35 going to try now. Jul 24 18:08:52 need to find a couple USB cables now. Jul 24 18:09:32 jkridner: okay, I think that it is related to the memory but am not sure Jul 24 18:09:35 grabbing mini cables from the closet for Black. Jul 24 18:09:51 m_w: because of the new single-config? Jul 24 18:09:57 yeah Jul 24 18:10:15 jkridner: you have the bisect right? Jul 24 18:10:40 I would have spent a bit more time debugging but got "busy" Jul 24 18:10:57 m_w: the boot bisect log doesn't reach the end. Jul 24 18:11:23 I didn't translate the -00XXX indexes to position in the commit log.... Jul 24 18:11:33 but, I didn't see any runs between 200 and 238. Jul 24 18:11:49 so, I didn't see the final working or initial failing boot log. Jul 24 18:13:40 hmmmm... something must be wrong with this serial adapter... no output. Jul 24 18:25:08 ugh.... for the love of... can't find a freakin' serial cable. Jul 24 18:30:04 been there Jul 24 18:44:56 either that or you have too many of the wrong ones... Jul 24 18:53:41 thetransformerr[: can you add hookup description to your readmes? does it use prudaq quick start jumpers? does it use a scope? what is the output pin? Jul 24 18:54:50 * jkridner won't get back to debugging u-boot today. :( Jul 24 18:58:48 and what is the clock source? the jumper for that also needs to be correct Jul 24 19:02:06 jkridner: got etcher sdk spec/documentation here https://github.com/resin-io/etcher/blob/sdk-spec/docs/sdk-spec.markdown , will be released in a week. It includes all the modules, they've also got option to directly fetch images from http/https . Jul 24 19:04:47 jkridner: without sdk, I can can just get flash working for uncompressed images as the decompression module isn't available standalone. Others drivelist and etcher-image-write are available as standalone modules for use Jul 24 19:08:08 thetransformerr[: also, the libzmq3 package already in debian is version 4.0.5 - do you need version 4.1 or is 4.0.x good enough? Jul 24 23:57:18 libzmq3 is good enough, In my kernel it wasnt there, so I just menyioned it Jul 25 01:02:17 zmq has a kernel driver? Jul 25 01:03:40 you want me to push a little extra bashiness? Jul 25 01:04:34 * nerdboy 's desktop is *still* building stupid webkit since this morning... Jul 25 01:07:20 since i have 4.9 kernel installed Jul 25 01:07:43 so, hookup documentation... Jul 25 01:08:00 pwm out is on P9.22 right? Jul 25 01:08:28 how did you hook up the rest? using scope/siggen? Jul 25 01:18:03 hmm, seems like zeromq does about the same thing redis-ipc does Jul 25 01:18:13 just more complicated Jul 25 01:26:31 I didnt knew about redis-ipc, and I came to know that zeromq has feature of transfering message reliably, Jul 25 01:27:01 I used arduino for wave gen and a scope Jul 25 01:28:24 today I would do a full test for python and c files, and I received LM-35 as temp sensor, that can be used with multi channel sampling but Jul 25 01:28:37 for single axis I am waiting for dht Jul 25 01:30:23 not actually kernel, image I meant to say :), Jul 25 01:34:25 and also in setup.sh, many times export pwm work but many times it displays device busy, I think it occurs when pwm had already been exported Jul 25 01:35:08 but I cant figure out how to determine that in script "||true" should have handled that Jul 25 01:35:50 or there is another reason which I cant determine for showing "device or resource busy" message Jul 25 01:58:44 the actual pwm0 thing isn't created until you export so you could check for that Jul 25 01:59:07 you can also read what's in export Jul 25 02:01:10 thetransformerr[: there are some (broken) examples in the exploring beaglebone stuff on github Jul 25 02:01:36 i saw the book pdf on google too Jul 25 02:02:21 * nerdboy bought the physical book so didn't feel bad about grabbing the pdf Jul 25 02:03:22 can you post your arduino test setup? maybe add a test directory and source code? Jul 25 02:03:45 i got a few of those around here somewhere **** ENDING LOGGING AT Tue Jul 25 03:00:01 2017