**** BEGIN LOGGING AT Tue Jun 20 03:00:03 2017 Jun 20 10:17:31 Hello. I vaguely remember that u-boot was interruptible in 'normal boot' mode with CTRL+C but I've got a board (not beagle) here I can't interrupt. Is this a new secure boot behaviour? Jun 20 11:20:45 IIRC it's not CTRL-C but something else like any key, space or enter and I'd expect that you can turn that behavior off Jun 20 11:33:04 Hi All, Had anyone experienced working with UniFlash for BBB flashing? Jun 20 11:57:17 cyberguy: I don't see it listed as supported, and seeing that it's a TI tool, you probably will have more luck asking on E2E Jun 20 11:57:58 das: it is possible to build an u-boot that isn't interruptible, or requires a specific key or sequence of keys (and not necessarily tell you) Jun 20 11:59:47 tbr: I already suggested E2E several times, but he seems happier by asking here daily Jun 20 12:03:26 I am having trouble with enabling ADC when bootin up, anyone here that might point me in the right direction ? Jun 20 12:07:38 Do I must use the EEPROM with BBB? Can I not install it? Jun 20 12:09:10 EPARSE Jun 20 12:09:51 ? Jun 20 12:10:31 ? Jun 20 12:10:36 zmatt : Do I must install the BBB EEPROM on my custom board? Jun 20 12:11:09 u-boot uses the EEPROM to detect which board it is Jun 20 12:11:24 you could build a custom u-boot which doesn't require it Jun 20 12:11:48 you might want to consider including the EEPROM anyway however, it's a handy place to store your board's revision number and serial number Jun 20 12:12:25 makes it easier to support multiple boards in the future Jun 20 12:12:28 I Have another EEPROM on board (I have RTC chip with EEPROM ) Jun 20 12:12:35 ok Jun 20 12:12:53 then you could use that to detect your board in u-boot I guess Jun 20 12:13:50 see board/ti/am335x/ and board/ti/common/ for inspiration Jun 20 12:14:23 So I need to write to the EEPROM this data with my custom board...? Jun 20 12:15:17 well if the EEPROM isn't at the same i2c address as the usual board detect EEPROM then there's also no urgent reason to use exactly the same format, although it may be convenient Jun 20 12:15:40 that's ultimately up to you as designer Jun 20 12:16:52 I suggest you examine the code in the two directories I just mentioned to understand how the board detect logic currently works, and use that as inspiration Jun 20 12:18:30 this assumes you're okay with customizing u-boot... if your board is closely compatible with the beaglebone and you want to be able to use the stock u-boot, you will need an eeprom at the address that u-boot expects Jun 20 12:20:29 the cautious approach would be to just include it in the schematic... it's always an option to decide later to omit it during production Jun 20 12:21:47 great zmatt thanks Jun 20 12:22:24 Ok, I am seeking some guidance here, I'm running debian 8.8 on a beaglebone white, I do get ADC to work when adding BB-ADC to slots manually, I then did this in rc.local and after reboot the ADC worked but my gpio's wasn't there anymore ? Jun 20 12:22:27 another question , there is a driver i need which is now approved at net-next , I understand it takes time until it'll be in the mainline linux. Jun 20 12:22:44 can I just take the patches and patch it to the current linux? Jun 20 12:24:39 the only way to discover the answer to that is "try it and see" Jun 20 12:24:59 with some luck you can just merge that branch using git Jun 20 12:25:08 or cherry-pick the specific patches you need Jun 20 12:25:57 Alrightythen2: ah, that's a much more specific question Jun 20 12:26:16 Alrightythen2: sounds like cape-universal being auto-disabled Jun 20 12:27:21 I read about that in the uEnv.txt ,, so i tried using rc.local instead but it seems like it still auto-disables Jun 20 12:27:51 Alrightythen2: "cape-universal" conflicts with nearly every overlay, which is why it's automatically disabled if any other overlay has been loaded during boot... the ADC overlay is a rare example of an overlay that *doesn't* conflict but I guess they don't make that exception Jun 20 12:28:09 you can manually load the cape-universal overlay of course Jun 20 12:28:37 I was trying to find out about how to do that Jun 20 12:28:52 boot normally, check the slots file to find its name Jun 20 12:29:09 then load it the same way you load the adc overlay Jun 20 12:29:36 there are a few variants of cape-universal, hence I suggest checking which one it loads by default to be sure you pick the right one Jun 20 12:29:50 Alrighty then,, look like its called univ-all Jun 20 12:31:15 tiny bit off topick but, is there a reason i cant sudo echo BB-ADC > slots ? I need to enter su , then it works ? Jun 20 12:32:11 He only sudos the echo the BB-ADC is 'a new comment' for him Jun 20 12:32:27 you can sudo sh -c 'echo BB-ADC > slots' Jun 20 12:33:03 Alrightythen2: yeah, the >slots is redirecting the output of "sudo echo BB-ADC" Jun 20 12:33:26 another solution is echo BB-ADC | sudo tee slots Jun 20 12:33:37 Aaahh,, that make so much sense, I'm relearning linux again after 30yrs so what i remember is still kinda random lol ,, 10+Karma to you buddy ;) Jun 20 12:36:05 tbr, zmatt : thx Jun 20 12:36:12 prob locked then Jun 20 12:37:03 das: there are of course always ways around it, e.g. by replacing the u-boot by one with different configuration Jun 20 12:38:06 it's a BGA NAND I can't reprogram it i'm afraid Jun 20 12:38:37 I never said those ways would be *convenient* Jun 20 12:38:40 :) Jun 20 12:39:45 if the system boots linux though you could probably overwrite whatever portion of NAND u-boot resides in, but it would be a bit risky Jun 20 12:40:02 presumably there are also different ways to boot the board, e.g. via jtag Jun 20 12:40:11 nah JTAG's not routed Jun 20 12:40:12 for once Jun 20 12:40:27 usually JTAG works out of the box so I don't have to bother :) Jun 20 12:41:15 and since it's all BGA packaging I can't identifify and solder the JTAG/SWD pings Jun 20 12:41:18 pins* Jun 20 12:41:41 it has no alternative boot methods? they flash the NAND before soldering the board with no way to recover a hosed board? Jun 20 12:42:22 not sure. I assume JTAG's disabled in factory after the chip's been programmed ? Jun 20 12:42:48 which would mean there's a header somewhere though Jun 20 12:43:03 all I found is a UART to the processor Jun 20 12:43:15 well if it's mass-produced than flashing the NAND before soldering *is* an option Jun 20 12:43:22 have you checked if it sends CCC? Jun 20 12:43:35 ah alright, I'm not well versed in production options Jun 20 12:43:42 also if you manage to hold NAND in reset, the soc might try UART Jun 20 12:43:54 what SoC is it anyway? Jun 20 12:44:08 oh, right, might not be TI :) Jun 20 12:44:21 freescale Jun 20 12:44:35 that's a company, not a SoC Jun 20 12:44:40 :P Jun 20 12:44:44 yeah I'm looking at my notes Jun 20 12:44:59 markings have been removed Jun 20 12:45:04 oh fun Jun 20 12:45:24 MX6 something says uboot Jun 20 12:45:35 so there's also good chance they really locked up boot Jun 20 12:45:43 (by programming a public key into eFUSE) Jun 20 12:45:52 pubkey hash ? Jun 20 12:45:59 usually Jun 20 12:46:37 tbr: what's that CCC thing you mentioned? to me it means smth else :) Jun 20 12:46:59 that's the TI soc signal for "give me something to boot!" Jun 20 12:47:02 last time I checked there wasn't public docs on imx6 security, but there are on imx53 security (including the efuse chains and such) Jun 20 12:47:10 yeah, attempt to boot via XMODEM Jun 20 12:47:14 and it's not just 3 Cs it's arbitrary length Jun 20 12:47:15 probably not applicable here Jun 20 12:47:32 ack, as it's not TI Jun 20 12:47:48 ah damn ^^ Jun 20 12:47:49 Has anyone been trying stuff with Yocto on the BB(B)? I'm currently trying to activate the pwm and adc but can't seem to get it to work. The topics that I found on stack and other sites were often not completed. So if anyone has a tutorial or slightest tip that would potentially make my week. Jun 20 12:48:26 das: so, whatever you're trying to do may require finding a vulnerability rather than replacing u-boot :) Jun 20 12:48:59 das: fault injection may work, i.e. if u-boot can't find a bootable kernel it may drop into the commandline Jun 20 12:49:08 Duffy: if you are trying to build an device image using the OE environment and Yocto provided recipes, then you should probably just make it use a custom devicetree Jun 20 12:49:39 zmatt: this needs lab furnitures and knowledge I do not have Jun 20 12:50:05 also might be interesting to see what happens if the FSL SoC can't load U-boot (holding nand in reset or such) Jun 20 12:50:08 das: in some cases it only needs a paperclip or screwdriver, but the BGAs are making things more complicated Jun 20 12:50:30 NAND and SoC are BGA Jun 20 12:50:31 tbr: unlikely in this case Jun 20 12:50:37 ok Jun 20 12:50:49 I don't know a simple way around this Jun 20 12:51:15 but it was a good idea though, probably would've tried it on TSOP :p Jun 20 12:51:32 Thats basically what I'm trying right now but he (bitbake) always exits with different errors that I can't seem to find in the same or smiliar context as of what I'm doing there. Jun 20 12:51:33 randomly shorting nand pins/vias once the board enters U-Boot and before it loads the kernel? Jun 20 12:51:58 is this really how fault injections works? ^^ Jun 20 12:51:59 Duffy: might want to ask on #oe or #yocto or such Jun 20 12:52:02 undervolting may work as substitute for that Jun 20 12:52:25 flash generally requires higher voltage to work correctly than the cpu Jun 20 12:52:26 that's ... interesting Jun 20 12:52:49 will require a lot of tweaking but mmmh Jun 20 12:53:49 also the voltage on the flash supply rail needs to be low, not just the feed voltage to the board, I'd expect Jun 20 12:53:57 alternatives I can think of would have a high risk of bricking Jun 20 12:54:00 yes Jun 20 12:55:23 I'm in the yocto channel aswell but it seems that noone is home or my question seems to easy. Jun 20 12:55:26 like breaking into the running (linux?) system and zeroing the part of flash where the kernel stored... but that'd be a yolo leap of faith that u-boot will give you a prompt in that situation Jun 20 12:56:10 (Seems to easy so that I could find the solution myself, didnt want to sound cocky) Jun 20 12:57:05 tbr: I didn't get that, I think the NAND's Vcc is the same as the CPU Jun 20 12:57:34 that's not very likely actually Jun 20 12:57:43 the NAND's array supply is typically 3.3V Jun 20 12:57:49 das: yes, but it's probably not the same as the supply voltage you hook up to the board. so just turning supply down won't help Jun 20 12:57:50 I think Jun 20 12:59:37 if you say so, I'm not experienced enough to be sure Jun 20 12:59:56 thanks a lot though, will think about fault injections :) Jun 20 13:00:38 another keyword in this context is "glitching" Jun 20 13:00:54 with overvolts and the likes? Jun 20 13:01:13 usually I've seen it mentioned when they uncapped the IC Jun 20 13:01:55 it's essentially synonymous to fault injection in this context Jun 20 13:02:20 uncapping is rarely needed for fault injection Jun 20 13:02:32 just use your high-energy ion beam ;-) Jun 20 13:02:56 I've got the poor man's lab ^^ Jun 20 13:03:37 imx53 docs are also crazy helpful on security stuff... maybe freescale eventually gets equally complacent about imx6 security info Jun 20 13:04:05 prob using similar mechanisms, wouldn't hurt if I read those Jun 20 13:04:08 like, it almost seems they just dumped internal info on the web for imx53 Jun 20 13:04:23 including which metal layer to modify for certain overrides Jun 20 13:04:31 oh Jun 20 13:09:18 of course none of that info should let you bypass security if they did stuff right Jun 20 13:09:30 good security doesn't require obscurity Jun 20 13:15:31 yeah I'm probably not going to bypass the founder's security Jun 20 13:15:46 if anything it's the manufacturer's implementation Jun 20 13:18:49 hw bugs are possible, sw bugs are close to guaranteed though :) you just gotto find them Jun 20 19:00:30 i'm using BBB : will uboot start with empty eeprom? Jun 20 20:35:12 What version of Eagle is used for the Beagle Bone Black layout? I have used version 6.6, 7.7 8.2 to load the file all of which indicate an error that it is a not an EAGLE design Jun 20 20:36:19 https://github.com/beagleboard/beaglebone-black-wireless Jun 20 21:27:34 I just upgraded to debian 8.7 I cannot login as root anymore, is there a password, it used to be blank Jun 20 21:51:32 cyberguy: a BBB doesn't have an empty EEPROM, you already explained you had a custom board so please don't lie. i already explained to you that u-boot uses it for board detection, and I told you where to look in the u-boot source code to see how that works Jun 20 21:51:44 Mbrown: BBB isn't an eagle design Jun 20 21:52:45 Bernard_: use sudo Jun 20 21:57:48 Bernard_: allowing root login without password was a rather serious security hole. if you set a password on the root account (sudo passwd) then you should be able to login as root again Jun 20 21:58:28 that's generally not necessary though, you can just stick to sudo (or use sudo -i to get a root shell) Jun 20 22:02:45 I used sudo passwd, changed the password successfully but I still cannot login as root Jun 20 22:03:58 yes, you need to override things. That ensures that you understand that you are doing something potentially very stupid. Jun 20 22:04:22 /etc/ssh/sshd_config can be tuned Jun 20 22:11:34 the problem is that I use SSH and I have use ssh debian user credentials and I cannot overwrite some files Jun 20 23:14:23 tbr: ah right, they probably set PermitRootLogin to prohibit-password (its default value) ? Jun 20 23:14:39 oh well, bernard left Jun 21 00:38:13 zmatt: yes thats the default for openssh now .. Jun 21 01:10:22 hello all - i have a strange need. i need to get a beagleboard derivative to boot. said board lacks a eeprom. ive rewritten a few lines of uboot to populate the header with what i need. i am able to boot into a minimal kernel image just fine. now im trying to boot a more sophisticated kernel build but i get kernel panics. i suspect this is because the kernel is trying to read from an eeprom that doesnt exist. long story short - Jun 21 01:10:23 where can i find capemanager in the kernel? i want to experiment by editing the source and building a new kernel Jun 21 01:10:36 btw, thanks in advance for any help you can give Jun 21 01:11:34 i have a functioning build system and a software background - i just dont know where to look for the capemanager source files Jun 21 01:14:26 heres my boot log Jun 21 01:14:28 https://pastebin.com/ZEBASbX8 Jun 21 01:35:12 veremit: I know that, but not on bb.org images formerly :) Jun 21 01:36:25 t_brah: reading from a nonexistent eeprom shouldn't be a reason for panic... Jun 21 01:36:52 but why would you care about capemanager on a custom board? Jun 21 01:37:42 it crashes in ethernet code, nothing to do with capemanager Jun 21 01:38:14 though there are also various errors earlier Jun 21 01:39:05 oh wow ancient kernel too Jun 21 01:39:18 ancient u-boot too for good measure Jun 21 01:45:57 t_brah: line 128 sounds bad, line 152 looks weird as hell, line 219 isn't good, line 249 is already very bad and my feeling says probably related to the subsequent crash Jun 21 01:48:44 but seriously, consider first ditching the ancient crap, especially the kernel (although the u-boot is pretty old too) Jun 21 01:50:05 thanks for the direction! on it Jun 21 01:51:38 and of course customize the DT to reflect any differences between the bbb and your board, and you'll probably want to disable cape manager while you're there Jun 21 01:53:36 I've already customized the dtb to manually add the hdmi device and to remove the eeprom - does that sound sane? the board is exactly the same as a bbb Jun 21 01:54:03 well, no, it's not exactly the same since you removed the eeprom :) Jun 21 01:54:15 yeah, tell me about it Jun 21 01:54:31 which is probably costing you way more trouble than removing it was worth Jun 21 01:54:32 "oh itll be easy, just get android to run on it" hyuk hyyuk Jun 21 01:55:00 dont worry ive already given the ee hell Jun 21 01:55:06 they removed it for "cost reasons" Jun 21 01:55:11 lol Jun 21 01:55:30 yes, because everyone knows the smallest i2c eeprom you can find on the market is surely an expensive piece of silicon Jun 21 01:56:27 never mind that you might want some place to store your board identification and revision so you can in the future support multiple products/revisions with the same software base Jun 21 01:57:19 android... I don't even think TI supports android anymore since they exited the mobile market? Jun 21 01:57:41 lol yeah, the only android support i could find was bbbandroid Jun 21 01:57:49 might need to make a frankenimage for these guys Jun 21 01:57:54 *shudder* Jun 21 01:58:55 the last android sdk from TI for am335x is "jelly bean 4.2.2", released in 2013 Jun 21 01:59:43 someone released a repo that built android 4.4.4 aosp on top of a 3.8 kernel Jun 21 02:00:11 thats what im trying to hack together Jun 21 02:00:33 sounds like fun fun fun :) Jun 21 02:00:38 i hate it - but i need to pay the irs sooooooooo Jun 21 02:00:42 hehe Jun 21 02:01:35 ill see if i can replace the ancient kernel with a modern beagleboneblack/kernel repo Jun 21 02:01:47 as for uboot ehhhhhhh Jun 21 02:01:49 it works Jun 21 02:02:47 yeah u-boot's age is less urgent Jun 21 02:04:31 we use 4.9-bone (+ patches + custom config) on all beaglebones currently Jun 21 02:05:30 why android though? especially since with TI support gone you'll be stuck on old sgx drivers forever Jun 21 02:06:00 because the customer developed some app that links to a phone Jun 21 02:06:04 and the sunk cost problem Jun 21 02:07:58 oh well :) Jun 21 02:08:04 sounds like a joy Jun 21 02:22:17 4.9 bone is a good stable kernel? Jun 21 02:23:08 https://github.com/beagleboard/linux/tree/4.9 Jun 21 02:23:23 thinking of pulling this kernel and merging it with bbbandroid Jun 21 02:26:43 works for me anyway Jun 21 02:27:00 merging... ? with a 3.8-derivative? Jun 21 02:27:16 oh hell no Jun 21 02:27:20 phew Jun 21 02:27:43 im just blowing away the whole 3.8 kernel folder and replacing it with this one Jun 21 02:27:48 I don't think there's a deeper merge-conflict hell I could possibly conceive Jun 21 02:28:12 im far, far too lazy to attempt such nonsense Jun 21 02:29:14 especially since 3.8-bone was a really hacked-up 3.8 with tons of patches Jun 21 02:29:24 thanks so much man for the help so far Jun 21 02:29:34 i was wandering a tunnel of insanity for a good month Jun 21 02:30:06 np, I don't mind distraction from writing code that interfaces with dbus... in nodejs... Jun 21 02:30:27 ... are you sure you dont mind? Jun 21 02:30:55 it sounds to me like you truly love this amazing abortion Jun 21 02:31:31 oh javascript is truly a wonderful language Jun 21 02:31:53 except for its strings Jun 21 02:31:57 and bitwise ops Jun 21 02:32:06 and grammar Jun 21 02:32:09 and everything else really Jun 21 02:32:30 [] + [] = "" Jun 21 02:32:40 such magnificent code Jun 21 02:32:45 === even Jun 21 02:32:52 don't forget it has three different flavors of "equal" Jun 21 02:32:59 [] + {} = [object Object] Jun 21 02:33:05 well at least it isnt PHP Jun 21 02:33:12 {} + [] = 0 Jun 21 02:33:17 such beauty Jun 21 02:33:26 {} + {} = NaN Jun 21 02:33:31 https://www.destroyallsoftware.com/talks/wat\ Jun 21 02:33:32 one of those doesn't actually work in practice anymore Jun 21 02:33:33 https://www.destroyallsoftware.com/talks/wat Jun 21 02:33:40 oh! good! Jun 21 02:33:41 I think {} + [] Jun 21 02:33:43 LOL Jun 21 02:34:00 oh it does Jun 21 02:34:01 hmm Jun 21 02:34:02 glad to hear it....?????? Jun 21 02:34:07 oh lol Jun 21 02:34:11 ah, {} + {} Jun 21 02:34:15 yields '[object Object][object Object]' Jun 21 02:34:34 I have no clue why automatic semicolon insertion doesn't kick in there like it does for {} + [] Jun 21 02:34:50 anymore at least, since evidently it used to Jun 21 02:36:07 it used to be {} + {} => {}; +{} => +'[object Object]' => NaN Jun 21 02:37:10 while apparently it's still the case that {} + [] => {}; +[] => +'' => 0 Jun 21 02:38:30 although I'm kinda curious why these cases aren't treated the same, I'm not even going to try to dig in the grammar again.... it's so horrible that the C++ grammar is simple and clean by comparison Jun 21 02:39:13 at least C++ doesn't fucking parameterize their non-terminals Jun 21 02:42:43 kinda creepy that the decision on whether to interpret {} as an empty block or an empty object literal depends on such far lookahead Jun 21 02:47:06 kernels should work fine w/o the eeprom Jun 21 02:47:20 yeah Jun 21 02:47:41 done a few boards w/o them and no issues Jun 21 02:47:53 u-boot cares, kernel doesn't **** ENDING LOGGING AT Wed Jun 21 03:00:03 2017