**** BEGIN LOGGING AT Tue Mar 30 02:59:56 2021 Mar 30 07:03:19 Thank you @kveremitz and @zmatt for the response. Mar 30 07:04:33 I checked the link https://pastebin.com/z7a7Tn4C and removed the Capacitor C24 and C30 in the BBB. Mar 30 07:05:37 note that all of my testing was done with a reset extender present... technically when the caps are removed, the reset pulse is shorter than required by the phy's datasheet. I do not know whether or not that can cause problems Mar 30 07:06:21 (that's the purpose of the big cap in the first place: to extend the reset time to meet the phy's requirements) Mar 30 07:07:17 Now after the reboot, power on the Ethernet port led is Working Okay. But Ethernet port assigned with invaild IP address. Mar 30 07:07:39 IP address assignment has nothing to do with hardware, that's purely software settings Mar 30 07:08:07 Ok got it. Mar 30 07:08:21 with the default settings it will try to acquire an IP address using DHCP Mar 30 07:08:35 (like every OS does by default) Mar 30 07:08:49 Okay. Mar 30 07:08:57 In the same link https://pastebin.com/z7a7Tn4C, the last two points:- Our current workaround consists of: Mar 30 07:08:57 1. a voltage supervisor (TPS3839K) with a push-pull output connected to the BBB reset (P9.10) via 330Ω series resistor (to limit current), both to ensure reset is asserted for longer and to make it rise faster after deassertion. This should greatly reduce (but not eliminate) phy issues. Mar 30 07:08:58 2. a Schottky diode from the pmic power button input (P9.09) to the VDD_3V3 to ensure that if the beaglebone powers off, it automatically powers back on. This allows software to power-cycle the board if the phy issue is detected. Mar 30 07:08:58 Is there any cricuit is available. Please provide schematic if you have any. Mar 30 07:10:35 I mean, that's a pretty concrete description of the components we used and how they were connected to the beaglebone, but I guess I could doodle a schematic Mar 30 07:11:19 Yes please. Mar 30 07:11:32 Let me try that. Mar 30 07:23:46 Subash: https://photos.app.goo.gl/2g4GoVLgzXucAxYK6 .. note that the schottky diode on the left needs to be combined with a software solution that detects when the phy is in a bad state and powers cycles the system (by powering off and letting the diode turn the system back on), it also precludes being able to do a clean shutdown of the system since the diode will always immediately turn the system ... Mar 30 07:23:51 ...back on, hence it may not be suitable for everyone Mar 30 07:24:55 while the reset extender on the right does not fully fix the problem, it merely reduces its occurrence, unless you also remove the caps from the beaglebone Mar 30 07:25:54 (a true proper fix would be to use a gpio to reset the ethernet phy instead of using the processor reset signal, or using a different phy altogether, but neither of these fixes is practical to patch onto a beaglebone) Mar 30 07:30:34 jkridner[m]1: has it never been considered to swap out the ethernet phy for a different one? given how this problem has plagued people since the BBB was introduced Mar 30 07:31:06 it's the BBB's only hardware erratum that really has no good workaround Mar 30 07:33:38 Hello I am not able to connect to bluetooth on Beaglebone Ai. I have Debian 10 with LXDE Mar 30 07:50:17 Okay. Thank you for the schematic @zmatt. Let me try that and come back to you. Mar 30 07:51:26 Subash: it may also be worth following https://github.com/beagleboard/beaglebone-black/issues/4 where this issue is discussed, in case there are any updates Mar 30 07:54:27 @zmatt: Okay i will check the above github link. Mar 30 10:52:36 which header is that zmatt ? Mar 30 10:53:00 and isn't gerald still in the pcb-design 'camp'? thought he was pretty OK myself Mar 30 10:53:16 ooh nice hack taggin the reset to gpio :D Mar 30 10:53:25 (reading issue on GH) Mar 30 10:53:44 which header? you mean the one with "P9" written above it? Mar 30 10:54:16 not sure what you mean by the second question Mar 30 10:54:35 oof P9 lol, my bad, not used to handwriting :D Mar 30 10:55:07 and I wouldn't call using a gpio for the phy reset a "hack", unless you manage to actually patch a beaglebone with it then it definitely sounds like a hack Mar 30 10:55:52 but for a design revision it would be the best solution if swapping out the phy for one that doesn't suck isn't an option Mar 30 10:56:16 they could reuse the emmc reset gpio since that doesn't work anymore anyway :P Mar 30 12:38:26 m Mar 30 13:32:13 so I note from your comment. But yes the 'hack' was physically altering an existing piece of hardware to perform conceptual/feasibility testing ... Mar 30 13:32:29 :P Mar 30 13:32:34 aka "hacking" :) Mar 30 13:36:45 zmatt: can you suggest a phy that doesn't suck? Mar 30 13:39:14 in a package of similar size, obviously Mar 30 13:52:03 mru: nope, no idea Mar 30 13:52:45 are those lan87xx ones really so bad, though, if connected properly? Mar 30 13:54:45 mru: if they randomly manage to mis-sample strapping options that are held at a constant level throughout the power-up sequence, with a reset that fully complies with the requirements and recommendations in the datasheet? Mar 30 13:55:05 yea, it is really that bad :P Mar 30 13:55:14 oh, I didn't realise it was that bad Mar 30 13:55:56 can't say I've ever noticed that on other boards with that phy Mar 30 13:56:21 I'm sure there's some kind of explanation, but I haven't been able to identify it Mar 30 13:57:09 I like to at least double the documented reset times, just in case Mar 30 13:57:25 unless that's totally unreasonable Mar 30 13:57:28 the only factor I've found that had any impact was the rise time of the reset signal, which was directly and strongly correlated with failure rate Mar 30 13:57:51 extending the reset time without reducing the rise time of the reset signal had no effect iirc Mar 30 13:59:05 the datasheet says nothing about rise time Mar 30 13:59:10 only that it has to be monotonic Mar 30 13:59:13 correct Mar 30 14:11:16 mru: chart of test results: https://docs.google.com/spreadsheets/d/e/2PACX-1vTpgNnrxhtiKuQkZ_KeqjzT4uEemMZHWxhm6AYgNIZ0mu8L0brOyVA7JqsYZhF8YcmnTlATSwLMt1WY/pubchart?oid=1115484504&format=interactive Mar 30 14:11:49 this is for one particular beaglebone that had a high failure rate Mar 30 14:12:30 and with external reset-extender that extended the reset time to many times what was required Mar 30 14:13:16 so the conclusion is don't use an rc delay for the reset? Mar 30 14:14:12 yeah, but the purpose of the RC was to extend the PMIC's reset time to meet the requirements of the phy, so if you're going to remove the caps you still need to replace them with something else, like a separate reset circuit Mar 30 14:14:23 or use a gpio to create the ethernet phy reset Mar 30 14:14:55 always connect resets to gpios if any are available Mar 30 14:15:07 sooner or later you'll need it Mar 30 14:16:12 like I said in the issue, the eMMC reset line could be reused since it's useless (its purpose was to be able to hold the eMMC in reset if you reuse the pins, but while this worked for the old micron eMMC, it does not work for kingston eMMC whose reset input is edge-triggered) Mar 30 14:16:24 *eMMC reset gpio Mar 30 18:14:45 What is the difference between the beagle bone black rev c from ghi and the beaglebone black rev c from seeed studio? Mar 30 18:32:33 (none) Mar 30 20:02:35 yeah, should be 0 difference. Mar 30 20:02:48 I had someone complain they saw a difference related to eMMC.... Mar 30 20:03:07 which is particularly odd, since we have a direct relationship with Kingston to supply the same part to both. Mar 30 20:03:33 well, technically they said a difference between Embest and Seeed, but same story there. Mar 30 20:05:31 thanks all who joined forum.beagleboard.org. The main group will probably start moving next week where we'll encourage people to start posting new threads to forum.beagleboard.org/c/general. This week, business as normal on the big list, but small ones like x15, beaglev, gsoc, should all move over. Mar 30 20:07:13 the idea would be to send out a big notice and give people a couple of weeks to finish old threads on the google group while starting new threads on forum.beagleboard.org. Then, we'd mark the GoogleGroup as read-only and merge General and the GoogleGroup archive on forum.beagleboard.org. Mar 30 20:07:40 right now is the trial on the smaller, less-active groups to see where the pain points are. Mar 30 20:08:00 the e-mail integration might be the biggest pain point right now, so feedback on that is especially important. Mar 30 20:09:13 glad to say goodbye to google in that sense ... though never been a big participant :/ Mar 30 20:10:04 Nice! Mar 30 20:12:38 Well, I tried to sign in via Github.com's account under myself. Github.com replied to me but the forum email never showed (at least, not yet). Mar 30 20:14:46 Oh! Mar 30 20:15:05 I forgot to mention. I will have access to power now at the Maker Faire. Tricksters! Mar 30 20:16:20 Are they doing a f2f Maker Faire? Mar 30 20:17:22 set_: does the e-mail address on github match the e-mail address used to post to the forum? Mar 30 20:17:57 Yes but w/ Masks. Mar 30 20:18:03 Yes. Mar 30 20:18:28 Will the Beagle-V be fast enough for desktop usage? Mar 30 20:18:35 @jkridner: I signed into the forum via github credentials like w/ a given option. Mar 30 20:18:38 I can merge the accounts if you completed a login so you can have permission to post with your other e-mail address. Mar 30 20:18:41 demostanis[m]: define desktop Mar 30 20:18:57 demostanis[m]: think Pi3-ish. Mar 30 20:19:11 you won't be playing the latest games for sure Mar 30 20:19:46 They had four or five options given to sign in, i.e. Twitter, Github, and some others. Mar 30 20:19:59 I used the Github option. Mar 30 20:20:36 Web browsing, code editing, office, simple video editing, COMPILING RUST CODE Mar 30 20:21:03 Github replied and let me know that an application was used to sign in w/ github credentials but there was no email verification from the forum.beagleboard.org site (not yet). Mar 30 20:21:13 web browsing ... that's pretty much run arbitrary code of arbitrary complexity Mar 30 20:21:28 I doubt you'll have much fun compiling rust Mar 30 20:21:44 and web browsers like to have at least 8GB of ram Mar 30 20:21:44 set_: I don't know that we send out e-mail verification if you use a social sign-in. Mar 30 20:22:06 * jkridner wishes everyone hadn't killed OpenID. Social logins suck. Mar 30 20:22:33 Oh. I think that may be needed. I can try again to sign in. Cross-site scripting used to be frowned up too. Mar 30 20:23:38 BeagleV will have an 8GB variant. Mar 30 20:24:25 Is it better than a bad Dell Optilex? Mar 30 20:24:32 Your account is awaiting activation, use the forgot password link to issue another activation email. <<< This is what is listed after the Github.com usage for sign in. Mar 30 20:24:34 we are going to start calling this BeagleV as BeagleV Starlight in the somewhat near future to better match the Chinese branding and help give more visibility to StarFive on this one. Mar 30 20:25:09 set_: can you send me the e-mail used on github.com so I can check the logs? Mar 30 20:25:15 Sure. Mar 30 20:25:25 Please wait while I get the email. Mar 30 20:27:28 https://pastebin.com/k9PCkEDQ Mar 30 20:27:40 It shows the time and date on the bottom of the email. Mar 30 20:29:04 @jkridner: There is a new way to sign into git and github.com w/ git now. Mar 30 20:29:36 This might be the issue. They have another way, i.e. some long alpha-numeric code instead of passwords. Mar 30 20:30:31 yeah, application thingies. let me check the logs. Mar 30 20:33:31 Okay... Mar 30 20:33:46 set_: it shows the github login associated with your account... but says it hasn't verified your primary e-mail. Mar 30 20:34:11 I guess because they don't match. That shouldn't be a big deal, but I'm not sure how to disable that verification step for known-good users. Mar 30 20:34:20 Oh. Mar 30 20:34:24 I can try another way. Mar 30 20:34:24 wouldn't it be simpler to just register normally? Mar 30 20:34:33 Sure. I can try that idea... Mar 30 20:34:52 set_: I'll look if it actually sent a verification e-mail. Mar 30 20:35:24 Right. The pop-up from the forum stated I should expect a verification email. Mar 30 20:37:49 I tried w/ the email used from the google groups now. This also stated to wait for a verification email. I will reply in a bit. Mar 30 20:38:19 set_: it says you updated your profile. Mar 30 20:38:57 Yep. I just used the @gmail.com address and it worked like a charm. Mar 30 20:39:12 I got the verification email in seconds. Mar 30 20:40:19 I used a different email address this time. Maybe my provider is slow. Who knows? Mar 30 20:40:29 if beagle forums are migrating OFF google I might just check them out .. that was a blocker before.. Mar 30 20:42:26 blocker how? Mar 30 20:45:47 $google. Their interface is Terrible for groups imho. Mar 30 20:46:23 but then forums for me have always been phpBB or SMF :) Mar 30 20:46:29 I like Discourse though Mar 30 20:47:37 Speakin' of Discourse, how do they stay in business. They are providing a bunch of servers free, right? Mar 30 20:49:56 Anyway...up, up, and otay! Mar 30 20:51:21 set_: we're just using their code, not their servers. Mar 30 20:52:43 There's no support from Google and they change the interface at whim. The big question for me with Discourse is if we can do good e-mail and chat integration. Most users don't want to visit our website. I just want to make sure search engines get answers and we get visibility to update them with more recent/accurate answers. Mar 30 21:20:03 if people don't want to visit your website, you need a better website Mar 30 21:20:08 more like facebook Mar 30 21:20:13 people seem to love facebook Mar 30 21:29:41 hmmm... set_ took off. If back, I must notify that e-mail did go out according to the logs to verify the primary e-mail address. Mar 30 21:30:22 mru: I'll tell the web guy to make it more like facebook just for you. :-D Mar 30 21:30:47 I'm not people Mar 30 21:31:31 and I don't want people visiting our website---only crawlers. Mar 30 21:32:56 is there some sort of reverse robots.txt that tells everything except crawlers to get lost? Mar 30 21:33:17 you could do it by matching the user agent, obviously Mar 30 21:34:13 jkridner: "I don't want people visiting our website---only crawlers" ? ehh what? Mar 30 21:34:39 I actually visited the website yesterday Mar 30 21:34:48 didn't find what I needed though Mar 30 21:35:23 which was an image old enough to boot on a beaglebone white rev A2 Mar 30 21:36:34 odd... would think you could. Mar 30 21:36:46 that's so old it was never sold Mar 30 21:36:53 and some things are slightly different Mar 30 21:37:25 Debian 7.5 (BeagleBone, BeagleBone Black - 2GB SD) 2014-05-14 - more info - md5: 35877ce21e8ed0eb1bdc6819ad71c317 too old for you? Mar 30 21:37:35 oh. Mar 30 21:37:41 jkridner: relevant eMMC differences over time have been the transition from micron to kingston (from which point on it become impossible to hold eMMC in reset) and the transition from Kingston EMMC04G-S100 (eMMC v4.5) to Kingston EMMC04G-M627 (eMMC v5.1) which isn't supported on ancient 3.8 kernels without a trivial patch (which rcn has backported to the wheezy/3.8 series so people who are determined ... Mar 30 21:37:44 I can't even remember that long ago. :-D Mar 30 21:37:47 ...to be stuck on that ancient kernel still only need to update to the latest patch version) Mar 30 21:37:53 jkridner: that one screamed something about a missing eeprom Mar 30 21:38:02 mru: maybe ask rcn? Mar 30 21:38:07 rcn-ee did a patch to 3.8 kernels, no? Mar 30 21:38:13 jkridner: yeah that's what I just said Mar 30 21:38:16 I only wanted to check something with the eth phy Mar 30 21:38:27 I didn't make it that far in your message. :-D Mar 30 21:38:28 it was simpler to remove it and solder it to another board Mar 30 21:38:35 * jkridner has a 12 word limit. Mar 30 21:39:23 jkridner: ... Mar 30 21:39:57 mru: lol, BBW A2 is so old that it predates the revision log in the BBW schematic Mar 30 21:41:13 mru: but if it used to be supported it's strange that it no longer is, that probably means it's something minor overlooked in board detection, hence my suggestion to ping rcn if you care about using that ancient board Mar 30 21:41:27 I don't think it ever was really supported Mar 30 21:41:33 they only made a few of that revision Mar 30 21:42:17 if it's a prototype then it's possible no image that has ever been on beagleboard.org will work, it could have a custom u-boot build Mar 30 21:42:37 I would have had something built at the time Mar 30 21:42:57 hmm, since it's the first am335x-based board it probably initially had no board detection in u-boot whatsoever Mar 30 21:43:35 and, judging by the "screaming about a missing eeprom", no ability to do board detection either :P Mar 30 21:44:04 could be a blank eeprom, don't recall the exact message Mar 30 21:44:30 my inclination would be to patch u-boot to ignore board detection and hardcode the dtb for the BBW Mar 30 21:44:38 or, did you get into a u-boot prompt? Mar 30 21:44:40 it's quite possible that whatever we used on those boards never made it into mainline u-boot Mar 30 21:45:05 I no longer care about making it work Mar 30 21:45:39 since I got what I actually wanted with a soldering iron instead Mar 30 21:55:16 jkridner: kernel 3.14 and uboot 2021 is the target.. >,< LUL [bad vendor, bad] Mar 30 21:57:24 kveremitz: ew what? Mar 30 21:57:50 zmatt: the chap I Was talking about an am3352? over in the uboot channel Mar 30 21:58:01 I was asking about serial debug here before.. Mar 30 21:58:08 why would you want a bleading-edge u-boot + prehistoric kernel Mar 30 21:58:19 oops I got that wrong then .. Mar 30 21:58:24 2014? Mar 30 21:58:49 200??!? Mar 30 21:59:32 [Mon 7:44 pm] I'll happily join the beagle channel as that's an interesting board, however the manufacturer SD runs on 3.14 and uboot 2014.07 :) I've been told in #yocto that it also means that the dtb used for a 3.14 kernel very likely won't work in modern kernels Mar 30 21:59:34 history ftw. Mar 30 22:01:00 kveremitz: serial console hasn't really changed significantly over time, other than the ttyO* -> ttyS* transition, though ttyO* also still works anyway (in both u-boot and kernel) Mar 30 22:01:27 kveremitz: and why are you proxying for him instead of letting him "happily" join this channel? ;P Mar 30 22:01:46 * kveremitz looks around .. Mar 30 22:01:49 ah the question isn't beagle-related? Mar 30 22:01:55 correct Mar 30 22:01:58 originally Mar 30 22:02:10 but I figured the am3352 is close enough .. Mar 30 22:02:41 normally if you build u-boot for some device it will come with correct console settings baked in Mar 30 22:02:41 uh, dtbs need to align with kernel versions by kernel design. Mar 30 22:02:52 alex88> Hi everyone, I'm trying to get u-boot to work on a TI AM3552 board, we have the manufacturer SD image that works and our own yocto based image that doesn't output anything to the serial console, I've been told it might be a dtb problem, however the dtb files are in the ext4 /boot dir, not the boot partition Mar 30 22:03:16 there's so many different ways uboot could be configured/booted .. -sigh- Mar 30 22:03:36 if u-boot is not outputting anything then u-boot is misbuilt Mar 30 22:03:40 but yes, I pointed out there would probably be more people here could help with a TI-based boostrap than over there.. Mar 30 22:03:49 (ironically) Mar 30 22:04:05 this isn't particularly TI-specific either afaik Mar 30 22:04:57 also: note GMT+1 timestamps ;P Mar 30 22:07:53 https://github.com/u-boot/u-boot/blob/v2014.07/include/configs/am335x_evm.h#L85 Mar 30 22:08:06 pretty sure that defines the serial console that will be used by u-boot Mar 30 22:09:09 and it's also passed to the kernel's "console" parameter Mar 30 22:10:04 fair, but the dtbs will need to be present and accessible for that to work, as jk points out.. if they are in the wrong place or missing... Mar 30 22:10:05 so if they're using UART0 for the serial console it should Just Work unless their board init fails to setup pinmux or something Mar 30 22:10:11 yup Mar 30 22:10:24 dtb doesn't matter until the kernel boots Mar 30 22:10:34 oh because uboot uses internal, true Mar 30 22:10:50 doesn't it use a FDT? Mar 30 22:10:54 not in 2014 Mar 30 22:11:02 gotcha good point Mar 30 22:11:19 nor even in 2017 Mar 30 22:11:41 and even now it uses dtb(s) that are compiled into u-boot, not ones loaded from the boot disk Mar 30 22:12:05 since otherwise you have a chicken-and-egg problem :P Mar 30 22:12:25 since that dtb would contain the definitions needed to locate the boot disk in the first place Mar 30 22:15:29 jkridner: well, that wasn't the original intention... DT in its original incarnation was a serialization of firmware data structures hence was not kernel-dependent, and even with DTBs written by humans it was (is?) the intention that it could be forward-compatible... that intention definitely failed though :P Mar 30 22:24:58 Hi, i'm trying to boot a beagleboard c4 revision, but don't find an actual img to copy to sd. What is the latest img to this board? Mar 30 22:25:32 https://beagleboard.org/latest-images has a "Stretch for BeagleBoard via microSD card" image Mar 30 22:26:20 "OMAP3/DM3730 Debian 9.5 2018-10-07 4GB SD LXQT" Mar 30 22:26:53 I tried this image but I'm getting black screen in hdmi port. Is necessary some change in configuration files to enable hdmi output? Mar 30 22:27:09 or a log file in sd to see the errors Mar 30 22:27:50 eh, I don't know anything about the old omap3-based beagleboards, sorry... those are from before my time :) Mar 30 22:28:00 maybe someone else knows Mar 30 22:28:25 thanks Mar 31 02:16:40 I managed to get a output from my beagleboard c4 but is showing this error at boot: *** Warning - bad CRC or NAND, using default environment. Anyone have a procedure to upgrade the nand? Current is U-Boot 2009.06-rc2 (Sep 09 2009 - 13:12:40) Mar 31 02:35:28 hmm Mar 31 02:39:20 FilipeF: does it boot from sd card if you hold down the user/boot button at power-on Mar 31 02:40:37 (this will prioritize sd card over nand) Mar 31 02:45:28 if that works then wiping nand from u-boot should allow it to boot from sd card without having to hold down the user/boot button Mar 31 02:47:31 zmatt: thanks, trying here, Uncompressing Linux... done, booting the kernel. Mar 31 02:48:13 FilipeF: I found some comment saying the command to wipe nand is "nand erase.chip" Mar 31 02:48:24 (from the u-boot commandline) Mar 31 02:50:59 see also https://github.com/beagleboard/Latest-Images/issues/48 ... someone else's attempts to get a recent image working on the classic beagleboard **** ENDING LOGGING AT Wed Mar 31 02:59:56 2021