**** BEGIN LOGGING AT Wed Dec 09 02:59:56 2020 Dec 09 03:03:52 i did read somewhere that you can solve this... Dec 09 03:04:12 i know you made a suggestion Dec 09 03:04:18 i'll have to try and find it again Dec 09 03:04:29 i was thinking about looking into libvirt's networking features Dec 09 03:04:32 NAT or proxy arp Dec 09 03:04:36 those are the options Dec 09 03:05:05 i think the solution i saw was a NAT solution but i think there was a specific not-to-crazy way to do it Dec 09 03:05:10 thats what i was probably going to do anyway Dec 09 03:05:29 i mean the DHCP server has created its own subnet so Dec 09 03:05:56 NAT is the simpler solution of the two, but doesn't allow devices on wifi to access the bbai ethernet side, which may be a problem or may be desired depending on the situation Dec 09 03:06:12 right that's sort of a concern Dec 09 03:06:25 ehh no, DHCP doesn't "create" a subnet... you created a subnet and deployed a DHCP server on it Dec 09 03:06:26 it's not because i'm writing and trying out different solutions software to traverse nats Dec 09 03:06:49 that's technically true but i mean, practically the only reason there is a subnet is because i did deploy a dhcp server on it Dec 09 03:07:03 no, absolutely not Dec 09 03:07:06 although i guess technically the first thing i did was give my BBB an ip address on eth0 Dec 09 03:07:14 exactly Dec 09 03:08:37 whats the server i'll use to deploy the NAT? Dec 09 03:08:56 service* Dec 09 03:09:05 proxy arp isn't too bad Dec 09 03:09:07 none, NAT is handled in the kernel Dec 09 03:09:13 thats what i just realized Dec 09 03:09:13 NAT can get messy Dec 09 03:09:15 you set it up using iptables or nftables Dec 09 03:09:17 it's probably iptables Dec 09 03:09:24 yeah i think i'm running iptables Dec 09 03:09:31 i mean i just assume everything is Dec 09 03:09:36 how many things are behind it? Dec 09 03:10:12 behind it? i mean tbh just one Dec 09 03:10:19 at any given time Dec 09 03:10:30 i can do it with ufw it seems.. Dec 09 03:10:35 will every possible one have the same IP address? Dec 09 03:10:44 if so proxy arp is brain dead simple Dec 09 03:11:21 i mean since there's only one, i can reduce DNSMASQs range to just one IP Dec 09 03:11:25 for DHCP Dec 09 03:12:03 dhcp-range=eth0,192.168.2.50,192.168.2.50,255.255.255.0,12h Dec 09 03:12:55 but i mean the device running the ARP + DHCP etc (the BB AI) is also on the network Dec 09 03:13:23 although i'm assuming it's not using the 192.168.2. since the default gateway is on a different subnet Dec 09 03:13:31 also i'm using words i don't really understand lol Dec 09 03:13:45 proxy arp used to be a simple way to turn a modem equipped workstation into a PPP server Dec 09 03:14:56 i'm getting the sense this is ancient tech Dec 09 03:15:11 since the tutorials i'm reading are like "if ur fortunate it enough to be using kernel 2.4..." Dec 09 03:16:01 iptables has change Dec 09 03:16:20 found the stack overflow Dec 09 03:17:11 rofl i dont believe the solution i'm reading Dec 09 03:17:21 its too easy Dec 09 03:17:26 https://superuser.com/questions/1024511/how-can-i-bridge-my-router-to-my-wired-lan-over-wifi Dec 09 03:18:10 i just feel like its missing iptables entries Dec 09 03:37:29 proxy arp doesn't involve iptables Dec 09 03:59:32 Just getting the Cortex-M4 to work on a TI board is hard enough w/ energia and CCS. Sheesh. I cannot wait to try the BBAI. Blah? Dec 09 04:00:01 10:00! Get your loafers on! Dec 09 04:43:13 wowwwww zmatt i got it to work Dec 09 04:43:27 very surprised, i patched together what little i knew, some snippets here and there Dec 09 04:43:54 although tbf only ipv4 works Dec 09 04:44:46 also thanks ds2 for your advice Dec 09 04:44:47 pretty sure linux also supports proxy arp for ipv6, but it'll have to be configured separately Dec 09 04:44:51 i have no idea what i did Dec 09 04:45:04 well you might hate but apparently... i'm using ip forwarding... and proxy arp? Dec 09 04:45:13 (i have no idea what i did) Dec 09 04:45:28 i'm writing out my configuration Dec 09 04:46:04 i wish i had two device on the ethernet so i could see what happens in that case Dec 09 04:46:05 ("proxy arp" for ipv6 is actually called proxy ndp) Dec 09 04:55:55 https://gist.github.com/ayjayt/a70134d914c4b6e650562dd96994b32e Dec 09 04:58:47 im so proud of this Dec 09 04:59:10 wait wtf Dec 09 04:59:30 why does this enable proxy arp but also configure NAT ... that makes absolutely no sense Dec 09 04:59:49 proxy ARP and NAT are normally mutually exclusive Dec 09 05:00:05 its doing its best Dec 09 05:00:10 pretty sure the proxy arp doesn't get used here Dec 09 05:00:12 and i love it Dec 09 05:00:19 yeah it's just enabled without configured? Dec 09 05:00:50 the explanation in the first source said u need arp w/ forwarding or something but i dont get it Dec 09 05:00:57 I mean, it gets used but defeated by the remaining configuration Dec 09 05:01:31 like, it's kinda wrong and broken but in a way that won't easily manifest itself as tangible problems Dec 09 05:01:47 i posted that in ##networking just to troll them Dec 09 05:01:58 because they had been so mean to me when i was trying to understand that issue we debugged a few days ago Dec 09 05:02:36 I suggest turning proxy arp off again (echo 0) on both interfaces, pretty sure things will still work Dec 09 05:02:37 if i really wanted to troll them i could write "wait why doesn't ipv6 work?" Dec 09 05:02:43 alright ill give it a shot Dec 09 05:03:08 that last line of his post is also weird... manually adding a route on the client should never be needed unless the dnsmasq configuration is broken Dec 09 05:03:23 maybe I'm wrong about proxy arp not causing problems :P Dec 09 05:03:43 well that gist is MINE Dec 09 05:03:44 anyway, ignoring badness, this is the NAT solution Dec 09 05:03:54 not the proxy arp solution Dec 09 05:04:03 like I said, proxy arp does not involve iptables Dec 09 05:04:21 but i had added that route because i had connected the ethernet prior to setting up ip routing Dec 09 05:04:26 i mean, ip forwarding Dec 09 05:04:50 it should be added as part of getting a dhcp lease Dec 09 05:05:01 i wonder if i set up forwarding before i connected the device to the ethernet if it would have automatically configured it as the default route Dec 09 05:05:06 i'm not aware if i need to add a command Dec 09 05:05:07 no Dec 09 05:05:10 and you were right about Dec 09 05:05:14 about arp* Dec 09 05:05:19 yes so i don't know the config Dec 09 05:05:40 without ip_forwarding the client just wouldn't be able to reach anything beyond the bbai Dec 09 05:06:20 so is dhcp (dnsmasq) supposed to have some config enabled that says to tell its clients that is a gateway? Dec 09 05:08:48 alright its actually bed time for me but ill be back around to check this out tomorrow since i'm interested in that question and also the arp solution. Dec 09 05:09:29 i got 11 likes on my "what is a memory map post" on the bbb facebook group so i guess tomorrow i'll be like "how does the chip boot?" and then maybe "what is priveledged mode on a processor" or something gnight Dec 09 05:10:10 by default dnsmasq will provide its own ip in the "router" option in dhcp responses Dec 09 05:11:13 (which can be overridden in case the router is a different host than the dhcp server) Dec 09 05:11:13 i feel like it didnt work out that way Dec 09 05:11:38 is the client configured as dhcp client? (not manual ip config) Dec 09 05:12:05 yea Dec 09 05:12:41 then the client should set up a default route automatically Dec 09 05:12:48 when it obtains its dhcp lease Dec 09 05:13:02 even if its a bad route? Dec 09 05:13:09 "a bad route" ? Dec 09 05:13:14 wtf is a "bad route" ? Dec 09 05:13:29 yeah like i mean, at the time it got the dhcp lease, there was no DNS or route to the internet through that gateway Dec 09 05:13:40 i didn't set up ipforwarding until afterwards Dec 09 05:13:42 the client has no way of knowing that Dec 09 05:14:08 so it must be a configuration some where in dnsmasq Dec 09 05:14:16 and "no internet" doesn't mean it's a bad route... lack of internet access might be completely intentional (e.g. some office networks) Dec 09 05:14:25 dnsmasq config looks fine Dec 09 05:14:45 i mean okay but if there is DNS wouldn't you consider that a bad route Dec 09 05:15:10 journalctl does say "DNS disabled" for dnsmasq Dec 09 05:15:12 if you unplug the client and plug it back in, what does the "ip route" show then? Dec 09 05:15:42 DNS or lack thereof is a completely separate issue from routing Dec 09 05:15:43 i kinda wanna remove the default i set first Dec 09 05:15:56 that will happen automatically when you unplug etherner Dec 09 05:16:47 ugh i mean it didn't even break the ssh pipe when unplugged the ethernet Dec 09 05:17:05 and it's configure is now exactly as it was before i disconnected it Dec 09 05:17:10 correct, but that's separate Dec 09 05:17:11 configuration* in ip route i mean Dec 09 05:17:20 I mean, if you're uncertain, check "ip route" while unplugged Dec 09 05:17:34 or delete the route and then unplug and plug back in Dec 09 05:17:39 lol theres a problem with that- let me start tmux and put it on a timer to do that Dec 09 05:18:12 uhh, we're talking about the route on the device connected via ethernet to the bbai.. which I assumed would be your pc Dec 09 05:18:39 otherwise how exactly are you reaching it? Dec 09 05:19:33 PC + BB AI on wireless, NVIDIA on BB AI ethernet, getting to NVIDIA through BB AI Dec 09 05:19:56 ah you're ssh'ing into the bbai and from where into the nvidia? Dec 09 05:20:05 *from there Dec 09 05:20:15 mhm Dec 09 05:21:01 btw, in case you're not familiar with it, ssh can make connections via an intermediate hop using the -J option, e.g. ssh -J debian@bbai username@nvidia Dec 09 05:21:40 anyway, then just remove the default route and then unplug ethernet and plug it back in Dec 09 05:22:17 oh i didn't know that, very nice Dec 09 05:22:28 yeah so i mean i did sleep 3; ip route Dec 09 05:22:30 disconnected, reconnected Dec 09 05:22:40 the line for default does disappear and return Dec 09 05:22:53 it does sound like proxy arp would be better in your case, it would allow you to ssh directly into the nvidia Dec 09 05:23:25 I wonder if parprouted and dhcrelay are a hassle to setup like that superuser.com seems to imply (but I question his competence) Dec 09 05:23:45 superuser.com guy seemed to think ARP was needed along w/ port forwarding? Dec 09 05:23:50 his solution wasn't a copletely solution Dec 09 05:23:54 complete* Dec 09 05:24:35 okay wow yeah now if i delete default route, unplug it, and plug it back in, its fine Dec 09 05:24:38 what makes you think that? Dec 09 05:24:38 it comes back as default Dec 09 05:25:11 I actually withdraw my comment about his competence.. I thought you got the iptables lines from that post, but I see you added those yourself Dec 09 05:25:37 right i followed his instructions and got no results Dec 09 05:26:06 although tbf i did skipped line #2 Dec 09 05:26:09 what ip range did you use on eth0 and which on wifi ? Dec 09 05:26:40 ugh i mean maybe it is line 2 that is essential? Dec 09 05:26:54 but he enables arp + ip_forward ! Dec 09 05:27:01 step 2 is extremely essential, as is the need to have eth0 use a subrange of wlan0 Dec 09 05:27:23 big oof Dec 09 05:27:33 my understanding of the concepts is lacking Dec 09 05:28:31 in your case the "which your router does not use for assigning IP addresses" can probably not be accomlished, though hopefully the dhcp server on your lan checks whether an IP is free (using an ARP probe) before handing it out via dhcp Dec 09 05:28:49 the better solution (parprouted + dhcrelay) avoids that pitfall Dec 09 05:29:24 since it would relay dhcp requests on ethernet to your wifi router, so it would be the one handing your nvidia its ip address Dec 09 05:29:37 (instead of using dnsmasq on the bbai) Dec 09 05:29:54 so the bbai would essentially become transparent for dhcp and for ipv4 Dec 09 05:30:12 (and with additional configuration it could also be for ipv6) Dec 09 05:30:13 my router is wicked insecure actually Dec 09 05:30:20 i dont want to mess up to much but i can see its lease range Dec 09 05:31:02 yeah that does sound better Dec 09 05:31:11 so now i think we're talking about three solutions Dec 09 05:31:29 1) NAT, 2) ARP w/ the superuser.com's setup, and 3) parprouted + dhcrelay Dec 09 05:31:43 the word "proxy" in "proxy ARP" is not optional :P Dec 09 05:31:52 sure Dec 09 05:31:56 but yeah Dec 09 05:32:39 alright bed time Dec 09 05:32:49 ugh i have to be up in six Dec 09 05:32:50 GOODNIGHT Dec 09 05:35:33 ayjay_t: good night Dec 09 05:40:15 ayjay_t: debian has a dhcp relay daemon called "dhcp-helper" that looks extremely trivial to configure Dec 09 09:58:25 LED from the M4. I got it w/ a lot of help and a more luck than architecture. Dec 09 09:59:12 Something about a long script in .c and another short .c script to run the confusion of it all. Phew. Dec 09 10:06:37 so, it seems one would have to "flash" the M4 from the am5729 to get output from the M4 or to use it as an input. Long dues. Dec 09 10:08:32 I think the source is going to be the issue, not the source for the actual input or output, but the source that tells the M4 from the am5729 that it needs to listen and listen now. Dec 09 10:11:34 As usual, I am excited and as you all know - while this usually does not account for anything - I will be testing and learning while testing. I think I am getting smarter! Ha. Dec 09 13:22:21 so my mind is blown that if i put eth0 on the same subnet as wlan0, albeit in a non-overlapping range, i just have to _enable_ _proxy_arp and ipv4_forwarding and the ethernet device will join the network correctly Dec 09 14:02:24 what is nmi_intc? Dec 09 14:02:45 non maskable interrupt C? Dec 09 14:11:54 konsgnxx: context? Dec 09 14:12:17 dts file eg:https://github.com/beagleboard/linux/blob/5.4/Documentation/devicetree/bindings/usb/fcs,fusb302.txt Dec 09 14:12:52 is that for allwinner devices specific? Dec 09 14:12:56 impossible to say, it's just an example Dec 09 14:13:11 it's just the interrupt controller to which this thing is connected Dec 09 14:13:47 gotcha, i found allwinner devices with that defined as a peripheral. assuming that is the case Dec 09 14:14:01 then the example was probably taken from a board with an allwinner device Dec 09 15:29:38 hmmm, why does the bb.org-overlays/include/dt-bindings/ have such a limited selection of include files? is it to not be redundant with the linux/include/dt-bindings folder? Dec 09 16:26:53 which one are you missing? Dec 09 16:27:10 maybe they're just added as needed Dec 09 17:43:12 hello, I'm having some trouble with my PMIC status driver, it runs but gives me this: "genirq: Flags mismatch irq 59. 00000000 (tps65217-status) vs. 00002000 (vbus)" per my internet searches it means the IRQ is already being used. Does that sound right? Dec 09 17:46:21 I came across an original beaglebone white, but I cannot find anyone using them on for projects. Is it too outdated to find projects online? Dec 09 17:57:14 TonyTheLion: hmm, vbus is your driver? what's tps65217-status though... Dec 09 18:02:24 like, the only tps65217 component drivers defined in the kernel (mainline and TI) are tps65217-pmic, tps65217-bl, tps65217-charger, and tps65217-pwrbutton Dec 09 18:04:24 the irq you need is only claimed by tps65217-charger, which you're not using anyway Dec 09 18:07:53 (it could probably share the irq with you if both you and it pass IRQF_SHARED to devm_request_threaded_irq() ) Dec 09 18:15:12 @zmatt: I see, but this device, (not strictly a beagle) is using the charger Dec 09 18:15:23 so I guess the double use of the irq is the issue here Dec 09 18:15:38 I thought you weren't? Dec 09 18:15:50 tps65217-status is the driver Dec 09 18:16:01 it has to charge a connected battery, so it must be using it Dec 09 18:16:19 idk what vbus here is Dec 09 18:16:46 it is listed as an interrupt in /proc/interrupts for tps65217, that I do know Dec 09 18:16:56 as in, you're using a battery connected to the tps65217's BAT terminal? Dec 09 18:17:31 yes Dec 09 18:20:43 so.. isn't the status of the charger driver already what you want? or do you for some reason need to know when vbus specifically is available (and not "vbus or ac") ? Dec 09 18:21:29 I named what you helped me create last week "tps65217-status" Dec 09 18:21:48 and that is giving me the error I gave above Dec 09 18:22:10 is vbus the USB? Dec 09 18:23:02 what I want is an interrupt that will tell me in userspace whether the USB is connected (ie the device is not running on battery) Dec 09 18:23:15 those are not synonyms Dec 09 18:23:31 excuse my lack of hardware knowledge Dec 09 18:23:32 "device is not running on battery" is exactly the status of the charger Dec 09 18:23:50 i.e. it's already what the tps65217-charger driver provides Dec 09 18:23:57 oh Dec 09 18:24:16 so somehwere in sysfs I ought to be able to query that information Dec 09 18:24:23 or via udev Dec 09 18:25:20 you can read it via udev or directly from sysfs (which is where udev also gets it anyway), and you can get notified of status changes via udev Dec 09 18:27:19 I don't know how I overlooked last time that you mentioned you're using a battery Dec 09 18:28:23 which would have enabled me to conclude that what you were asking wasn't actually what you want (and what you want is easier than what you asked) Dec 09 18:29:12 the error still doesn't make sense to me, but at least it no longer matters Dec 09 18:29:22 since you don't need any custom driver Dec 09 18:31:27 ah okay Dec 09 18:46:01 zmatt: did you mention me? I got a notice, but the element.io client is junk. Dec 09 18:58:26 Baaaaahhhhh damnit, seems pocketbeagle is using mmc1 as the sd card port, shouldn't try using if for wifi too.... bahhhhhhh Dec 09 18:58:46 sorry prototype, can't fix you..... :( Dec 09 18:59:32 ohhh, oops nvm, it uses mmc0 Dec 09 20:00:00 where would one find info on how to use beagleboard/image-builder? I gotta make some more kernel tweaks Dec 09 20:01:17 I always just did a build, the a rebuild ever after. Dunno what "image-builder" is. Dec 09 20:02:58 build n a script? Dec 09 20:04:27 build and rebuild are scripts that come down with the kernel source, iirc. Dec 09 20:19:32 konsgnxx: image-builder isn't for building a kernel, it's for building an image :P if you just want to build a new kernel check out rcn's kernel build repo Dec 09 20:20:07 konsgnxx: see https://pastebin.com/eLhrp1Hg Dec 09 20:20:35 jkridner[m]: lemme check what I've said to you recently Dec 09 20:24:16 jkridner[m]: oh yeah, that instead of embedding freenode's webchat into an (annoyingly small) iframe, it would probably be better to just have a big button that links to it. when freenode's webchat is at the top-level (instead of in an iframe) it does a much better job at notifying the user when someone says something to them (while they're on a different browser tab) Dec 09 20:25:07 k. Dec 09 20:29:12 better? https://beagleboard.org/chat/ Dec 09 20:30:36 jkridner[m]: also, in case it's of interest, someone recently received a BBBW with this EEPROM contents: https://pastebin.com/raw/PLGcxZvS (bogus board id, no serial, manufacturing date week 58 of 2008) ... not sure if this is still the same batch as the earlier BBBWs with bogus EEPROM, I thought those looked different but I might just remember wrong Dec 09 20:31:03 looks good to me Dec 09 20:32:47 hmmm.... 2008?!? yeah, pretty bogus. I'll need to look up what the bad GHI batch looked like. hope it is part of the same batch. I think a few hundred slipped through about 1-1/2 years ago or so. Dec 09 20:33:23 not just 2008, week 58 of 2008 ;-) Dec 09 20:33:38 generally not a very busy week of the year Dec 09 20:34:26 Only slightly less busy than week 57 I expect. Dec 09 22:02:31 AWOL? All w/out loans? Dec 09 22:02:52 Hey guys... Dec 09 22:02:54 Guys? Dec 09 22:03:09 Do not tell anyone else this idea but... Dec 09 22:03:25 I have mastered faking the dark side. Dec 09 22:03:26 Ha. Dec 09 22:04:31 Besides my poor taste in jokes, I am on my way to configuring these darn steppers to work w/in cahoots of one another. Dec 09 22:04:45 Pencil action is on the way and all b/c of the BBGG! Dec 09 22:05:23 Now, when I say take big leaps, I am microstepping. Ooh! Dec 09 22:06:51 I did one poor post that someone in life should correct but I am not sure who this person will be as of now. Dec 09 22:07:05 I can only stay patient. Dec 09 22:08:41 Fingers crossed and hands covered! Dec 09 22:10:05 I meant, "fingers crossed and behind my back." Dec 09 22:10:07 Dang it! Dec 09 22:18:12 ugh i need to write educational material today Dec 09 22:18:18 it's one paragraph how badly can i procrastinate Dec 09 22:18:19 hold me. Dec 09 22:35:38 Okay. Dec 09 22:35:47 HOlding and not letting google go. Dec 09 22:35:49 bbl! Dec 09 22:57:20 Phew. I am back. The world? Sheesh. Dec 09 23:02:56 I know...do not get up. "It is just him." I know, I know. At least some applause? **** ENDING LOGGING AT Thu Dec 10 02:59:56 2020