| erle | josch it could enable a reproducible offline build | 00:09 |
|---|---|---|
| erle | first step to reproducible obv | 00:09 |
| erle | the provenance of “this thing that the CI spit out” is not too questionable, but it is a bit weird | 00:09 |
| josch | erle: how do you want to achieve an offline build for something that needs to access the debian archive to download packages? | 00:11 |
| josch | erle: when you run the image build scripts multiple time, the output will be bit-by-bit identical | 00:11 |
| erle | let future me worry about the debian archive | 00:12 |
| josch | there are many definitions of what people consider being "reproducible" -- what kind is it that you find the current script lacking? | 00:12 |
| erle | well it starts with the bootloader binary being downloaded from the CI | 00:13 |
| josch | if you like, we can add a sha256sum to the script that makes sure that the u-boot binary is exactly what we expect | 00:14 |
| erle | this command here: /usr/lib/apt/apt-helper download-file https://source.mnt.re/reform/reform-boundary-uboot/-/jobs/artifacts/v3/raw/flash.bin\?job\=build flash.bin | 00:14 |
| josch | yes? | 00:15 |
| josch | erle: can you please elaborate where you want to go and why -- what is your motivation? | 00:19 |
| erle | i want to rebuild a disk image while offline and also introduce caching into the build process | 00:19 |
| erle | i started this once, then stopped, today remembered about it | 00:20 |
| josch | should be trivial with apt-cacher | 00:20 |
| erle | well, for everything | 00:21 |
| josch | what else is getting downloaded other than u-boot and Debian packages? | 00:21 |
| erle | the question is rather, what is being cached. the answer is: next to nothing. the build steps are not separated. | 00:22 |
| josch | correct, but if you want to cache stuff, why don't you just put apt-cacher in the middle and call it a day? | 00:22 |
| erle | consider the following https://mister-muffin.de/p/UmOD.png | 00:22 |
| erle | that is a partial dependency tree from my first attempt | 00:23 |
| erle | which i shamefully abandoned | 00:23 |
| josch | erle: you are not working on sysimage-v3, are you? | 00:23 |
| josch | erle: these days, kernel, dtb and configs are all Debian packages | 00:24 |
| erle | well, i said this was an old attempt | 00:25 |
| - chomwitt (QUIT: Ping timeout: 244 seconds) (~chomwitt@2a02:587:dc1f:5600:28f0:c157:4d87:4f4e) | 00:42 | |
| + littlebo1eep (~alMalsamo@gateway/tor-sasl/almalsamo) | 00:58 | |
| - littlebo1eep (QUIT: Ping timeout: 240 seconds) (~alMalsamo@gateway/tor-sasl/almalsamo) | 01:08 | |
| + littlebo1eep (~alMalsamo@gateway/tor-sasl/almalsamo) | 01:22 | |
| - mjw (QUIT: Quit: Leaving) (~mark@gnu.wildebeest.org) | 01:29 | |
| + vkoskiv (~vkoskiv@89-166-62-97.bb.dnainternet.fi) | 01:32 | |
| - littlebo1eep (QUIT: Ping timeout: 240 seconds) (~alMalsamo@gateway/tor-sasl/almalsamo) | 01:37 | |
| minute | erle: maybe you want to review the whole new structure first before drawing conclusions. | 01:54 |
| minute | erle: full recognition before etc etc | 01:55 |
| erle | minute maybe i should! but i want caching on the level of artifacts | 01:56 |
| erle | because with redo i can do “redo-ifchange $target“ and have it only regenerate if the prerequisites changed | 01:56 |
| erle | i have not seen that in the current (?) script, which requires manual deletion of the written image | 01:57 |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20) | 02:01 | |
| - mtm (QUIT: Ping timeout: 255 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 02:02 | |
| bluerise | cinap_lenwerk: congrats | 02:19 |
| bluerise | cinap_lenwerk: how? | 02:19 |
| - Christoph_ (QUIT: Remote host closed the connection) (~Christoph@p54bf605d.dip0.t-ipconnect.de) | 02:25 | |
| - bkeys (QUIT: Remote host closed the connection) (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 02:52 | |
| + bkeys (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 02:56 | |
| bkeys | I accidentally dd'ed over my Reform on the SD card, I figure now is then the time to boot from nvme | 03:14 |
| bkeys | And try to get mainline Fedora booted on here | 03:18 |
| bkeys | I don't think anyone is awake. But I got debian booting from emmc, I figured it would be uboot only on emmc | 03:49 |
| bkeys | I have Fedora on my nvme drive, basically unmodified. | 03:49 |
| - bkeys (QUIT: Ping timeout: 255 seconds) (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 03:56 | |
| + mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 04:08 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:20) | 05:48 | |
| + chomwitt (~chomwitt@2a02:587:dc1f:5600:b0ac:64b:5946:8d06) | 06:15 | |
| - mlarkin (QUIT: Quit: Lost terminal) (~mlarkin@047-048-086-214.biz.spectrum.com) | 07:08 | |
| vkoskiv | Has anyone tried to get OBS working? Wondering if it would work at least somewhat with hardware accelerated encoding, if supported. | 08:50 |
| vkoskiv | Looks like it wants a newer version of opengl than is supported. | 08:50 |
| vkoskiv | It initializes EGL 1.4 and then fails to create a context. | 08:51 |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20) | 08:55 | |
| kfx | obs barely works right on my 40-core desktop with 128gb of ram running a commercially-supported linux distro | 09:02 |
| - thum (QUIT: Ping timeout: 248 seconds) (~thum@user/thum) | 09:12 | |
| + MajorBiscuit (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 10:03 | |
| vkoskiv | Yeah I'm not expecting too much, just wondering if someone got it running yet | 10:08 |
| + mjw (~mjw_@2001:1c06:2488:1400:9e5c:8eff:fe8f:a440) | 11:16 | |
| + thum (~thum@user/thum) | 12:01 | |
| - buckket (QUIT: Quit: buckket) (~buckket@pdp8.buckket.org) | 12:01 | |
| minute | we don't have hw encode yet, only decode | 12:15 |
| + Christoph_ (~Christoph@p54bf61d7.dip0.t-ipconnect.de) | 13:02 | |
| - MajorBiscuit (QUIT: Quit: WeeChat 3.5) (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 13:04 | |
| + MajorBiscuit (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 13:08 | |
| - MajorBiscuit (QUIT: Quit: WeeChat 3.5) (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 13:14 | |
| + MajorBiscuit (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 13:27 | |
| + mark_ (~mark@gnu.wildebeest.org) | 13:38 | |
| - MajorBiscuit (QUIT: Ping timeout: 248 seconds) (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 14:00 | |
| - mtm (QUIT: Ping timeout: 246 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 14:02 | |
| + MajorBiscuit (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 14:03 | |
| - MajorBiscuit (QUIT: Client Quit) (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 14:04 | |
| + MajorBiscuit (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 14:06 | |
| + bkeys (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 14:11 | |
| erle | minute funny battery board bug: when enough cells are dead and the reform is on for long enough, it will show all cells at 5.3 – but of course, when i remove the charging power to call the bluff the reform turns off. | 14:14 |
| vkoskiv | Looks like OBS would like OpenGL 3.3 but the theoretical max for vivante is 3.1 | 14:21 |
| vkoskiv | But that's fine, I can wait for a newer SoM down the line :D | 14:22 |
| vkoskiv | First time I'm seeing 100% signal strength here on the train. I think my parents' house just has poor wifi. | 14:25 |
| bkeys | Okay, so as I mentioned last night I am on the war path to get Fedora on my Reform. | 14:26 |
| bkeys | I have my serial adapter up to the Reform like it says in the manual, but I am not seeing any output on my serial monitor | 14:27 |
| vkoskiv | Weird, ping doesn't work on this system. | 14:27 |
| vkoskiv | strace shows some EACCES and EPERM errors and it just returns right away with code 2 | 14:27 |
| josch | vkoskiv: that's a filesystem problem -- try running these tools as root. I'm in the process of fixing this with the maintainer of e2fsprogs | 14:31 |
| josch | bkeys: is this with fedora or the debian sysimage-v3? | 14:32 |
| bkeys | Debian image at the moment, I think I just have to configure the uboot on the Reform to run bootefi instead of what it is you guys are doing and Fedora should boot | 14:32 |
| vkoskiv | josch: works, thanks! | 14:32 |
| bkeys | But right now I am not getting any output whatsoever on the serial, so I have no way to manually mess with uboot | 14:33 |
| bkeys | I have the correct baud rate, I set it up for software flow control (which has to do with my input, not the serial output) | 14:33 |
| bkeys | If I cannot get a serial console, my next guess is to build a uboot image for the Reform that just runs bootefi out of the box | 14:34 |
| bkeys | Even once it boots to Linux I get no output, which is strange I should be able to get a full TTY and everything from Linux via serial console | 14:35 |
| josch | bkeys: i am told that you should at least see serial output of u-boot -- do you also not see that? | 14:35 |
| josch | (i cannot verify that myself because my S1 is broken) | 14:35 |
| bkeys | I wish I could see the serial output of uboot | 14:36 |
| josch | bkeys: but you are using the default sysimage-v3 without having changed anything with u-boot on there? | 14:36 |
| bkeys | I spammed keypresses right after booting the Reform and it still went to the OS uninterrupted so it's not seeing my input either | 14:36 |
| bkeys | Yes I haven't changed anything on uboot | 14:37 |
| bkeys | This is uboot/debian on the emmc at this point | 14:37 |
| josch | bkeys: can i build an image for you that you flash to your sd-card and try out a hunch I have? :) | 14:37 |
| josch | oh, so you have the dip-switch changed so that your u-boot comes from emmc and not from the sd-card? | 14:38 |
| bkeys | Yes I did that last night | 14:38 |
| josch | ah okay | 14:38 |
| bkeys | And sure, send me whatever image you wanna test out | 14:38 |
| bkeys | With you image do I have to flip the switch again cause it's on SD card? | 14:38 |
| josch | hrm... that doesn't make it as easy to try out experiments from sd-card -- didn't you say you only want to switch to emmc booting later? | 14:38 |
| josch | yes, unfortunately :/ | 14:39 |
| bkeys | Yeah, but I am feeling determined so I figure if I burn a bridge perhaps it will motivate me to my end goal | 14:39 |
| josch | bkeys: so i'm building an image for you but before you test that, I guess you made sure that your TX is connected to RX and RX to TX and GND to GND, right? I'm successfully connecting to my reform over serial by running `sudo screen /dev/ttyUSB0 115200`. I'm using an FTDI232 set to 3.3 V and here is my setup: https://mister-muffin.de/p/t7No.jpg https://mister-muffin.de/p/M8dC.jpg -- if you tried all | 14:53 |
| josch | that and that doesn't work, stay tuned, image is building. :) | 14:53 |
| bkeys | Alright I am back | 15:13 |
| bkeys | Yeah I double checked the pins | 15:13 |
| bkeys | I am using minicom rather than screen | 15:13 |
| josch | yeah, shouldn't matter -- i was just posting it just in case you are interested in a known-good configuration | 15:13 |
| bkeys | Yeah screen isn't showing anything either | 15:14 |
| bkeys | I just have some USB to TTL adapter | 15:14 |
| josch | in that case, in about 2 minutes, download this and flash it to your sd-card: https://mister-muffin.de/reform/reform-rescue-system.img | 15:14 |
| josch | this is a rescue image that is nearly exactly the same as the stock rescue image but it will output to S2 instead of S1 | 15:14 |
| josch | S2 is connector J20 just to the left of S1 on this photo: https://mister-muffin.de/p/t7No.jpg | 15:15 |
| bkeys | https://alexnld.com/wp-content/uploads/2018/04/58df7051-4a29-4afb-9da1-e60cd7c55a5d.jpg | 15:15 |
| bkeys | What I have looks similar to this | 15:15 |
| josch | i can't read what chip that is or what signal voltage you've set it to | 15:16 |
| bkeys | Those are the 3 pins I am using now | 15:16 |
| bkeys | Is there a different set of pins I am supposed to use by default? | 15:16 |
| bkeys | Oh there are the ones right next to it | 15:17 |
| bkeys | So I should probably put them up on the pins that go straight up, rather than out the side of the board I am guessing | 15:17 |
| josch | the ones you see connected in my photo are connected to S1 | 15:17 |
| josch | my S1 is broken though, so I have to use S2 which is the connector to the left of S1 in the photo | 15:17 |
| josch | yes, S2 has pins perpendicular to the board and the pins of S1 go out to the side | 15:18 |
| josch | the image I've prepared and which you can download now from https://mister-muffin.de/reform/reform-rescue-system.img will output kernel messages to S2 instead of S1 | 15:18 |
| josch | maybe see if that works -- if yes, then your S1 is broken just as mine was ;) | 15:18 |
| bkeys | So is my S1 broken? | 15:19 |
| josch | not necessarily | 15:19 |
| josch | but this is why we try it out | 15:19 |
| josch | i'll be afk for 15 minutes | 15:20 |
| bkeys | Okay | 15:21 |
| - bkeys (QUIT: Quit: With every step we take, danger will follow closely) (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 15:38 | |
| + bkeys (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 15:38 | |
| josch | bkeys: did you have any success? | 15:48 |
| bkeys | I got the switch flipped back to sd card | 16:00 |
| bkeys | Trying to burn it now, this pinebook pro isn't that good | 16:00 |
| bkeys | It's being a pain with my samba share which is where your image is mounted | 16:00 |
| bkeys | I mean stored | 16:03 |
| bkeys | It should be noted that the system image on SD card doesn't have serial output either | 16:03 |
| josch | bkeys: okay, failed experiment then | 16:04 |
| josch | bkeys: you tested your serial adapter with another device to make sure that your adapter is not the problem? | 16:04 |
| josch | i tested mine with a raspi | 16:05 |
| bkeys | Well I haven't tried your image yet | 16:05 |
| bkeys | And yes I used it a few weeks ago on a different SBC and it worked fine | 16:05 |
| josch | ah i thought you meant my image when you were talking about "the system image on SD card" :) | 16:06 |
| bkeys | Gosh this pinebook pro is just awful | 16:06 |
| + mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 16:09 | |
| bkeys | I will say once I switched to SD card for some reason neither S1 or S2 even worked | 16:09 |
| bkeys | Like it didn't even pick it up as a serial device and the adapter didn't light up | 16:10 |
| minute | bkeys: try swapping rx/tx, make sure you have 115200 baud, try another uart usb adapter | 16:10 |
| bkeys | I know I got 115200 baud | 16:11 |
| bkeys | If serial won't work then I would think the next thing is to build a uboot image that is configured to simply run bootefi | 16:11 |
| + bkeys1 (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 16:15 | |
| bkeys1 | Now I am on USB wireless, hopefully it won't be so unstable | 16:15 |
| - bkeys (QUIT: Ping timeout: 250 seconds) (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 16:16 | |
| * bkeys1 -> bkeys | 16:16 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:20) | 16:17 | |
| + bkeys1 (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 16:30 | |
| - bkeys (QUIT: Ping timeout: 244 seconds) (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 16:31 | |
| * bkeys1 -> bkeys | 16:31 | |
| bkeys | Okay the pinebook pro is so bad with my network I have downloaded josch's image to my power9 server, and am copying it to an external hard drive and delivering it to the pinebook pro that way | 16:36 |
| bkeys | Gosh what a pain; makes me realize how easy samba shares and network makes things | 16:36 |
| bkeys | josch: Okay I got your image burned on the SD card | 16:39 |
| josch | \o/ | 16:40 |
| bkeys | Yeah the adapter isn't lighting up at all | 16:42 |
| bkeys | I got another one it came with new in packaging; let me open it. | 16:42 |
| bkeys | Okay I remove the extension cable and it solved it | 16:46 |
| bkeys | But no serial output on S2 | 16:47 |
| bkeys | The LED on the adapter is blinking like it sees something, but nothing coming out of minicom at all | 16:47 |
| bkeys | Nothing on S1 either | 16:49 |
| bkeys | Maybe I have the wrong kind of adapter. | 16:49 |
| bkeys | Mine is USB to TTL | 16:49 |
| bkeys | I am not certain if that is any different than USB to UART | 16:49 |
| bkeys | josch: And I take it there is no other way to access uboot? | 16:50 |
| josch | bkeys: u-boot can also output on hdmi -- but to input something you need serial on S1 (u-boot cannot do S2) | 16:50 |
| bkeys | Is USB to UART different than USB to TTL? | 16:51 |
| minute | not sure. usb to rs232 is certainly wrong :D | 16:53 |
| bkeys | Well what adapter do you recommend? | 16:54 |
| bkeys | The next option I see is that I try to build uboot so that it just uses bootefi instead of y'alls boot scripts | 16:55 |
| vagrantc | not sure of the exact u-boot config, but if there are no boot scripts or extlinux, u-boot should fall back to EFI ... | 17:01 |
| vagrantc | (unless it's not using distro_bootcmd) | 17:02 |
| josch | vagrantc: should distro_cmd not be doing that? | 17:02 |
| vagrantc | josch: in general, yes :) | 17:03 |
| bkeys | Well I have hooked up an external HDMI screen to the reform before and I got no graphical output | 17:03 |
| bkeys | From uboot or from linux even | 17:04 |
| vagrantc | you'll really want access to the serial output... | 17:04 |
| bkeys | I'd agree but I don't know if I have a way of doing that atm | 17:04 |
| bkeys | It seems all my immediate options are exhausted | 17:04 |
| vagrantc | what adapter do you have? | 17:05 |
| bkeys | A USB to TTL one | 17:06 |
| vagrantc | and you've tried swapping the tx/rx both ways | 17:06 |
| bkeys | My guess is that it's the wrong adapter | 17:06 |
| vagrantc | should be ok | 17:06 |
| bkeys | I haven't tried swapping them no | 17:06 |
| vagrantc | and you have ground? | 17:06 |
| bkeys | Yes | 17:06 |
| vagrantc | try swapping them | 17:06 |
| josch | with my image you should get something independent on whether you connect your serial to S1 or S2 because u-boot outputs on S1 and linux to S2 | 17:07 |
| vagrantc | it's kind of arbitrary what is rx and what is tx ... depends which perspective you're looking at it from | 17:07 |
| bkeys | Swapped it and I get nothing on S2 | 17:08 |
| bkeys | Oh wait, I am getting Linux on S2! | 17:08 |
| josch | hah! | 17:08 |
| bkeys | Okay lets move around and see if S1 works | 17:08 |
| josch | nice :) | 17:08 |
| vagrantc | when getting nothing, always try swapping rx/tx :) | 17:08 |
| bkeys | My guess is that it's a per adapter thing | 17:09 |
| bkeys | Okay I got a uboot prompt | 17:09 |
| bkeys | Yay | 17:09 |
| vagrantc | it depends on if the adapter is labeled "plug this wire into the rx of the device" or "plug the output of the device into this rx on your adapter" | 17:10 |
| bkeys | I see | 17:10 |
| bkeys | Well I got a uboot prompt | 17:10 |
| vagrantc | hope abounds! | 17:10 |
| bkeys | I have mainline Fedora on the nvme, I need to do bootefi but it bootefi and boot from nvme | 17:10 |
| vagrantc | does the u-boot build support loading off NVMe ? | 17:11 |
| bkeys | So I can run bootefi but it needs a image, my guess is grub on the nvme | 17:11 |
| josch | at least your S1 is working -- when you do uboot stuff it's hard to do without it XD | 17:11 |
| bkeys | If I got mainline Fedora working on here and I got the board repaired this thing would be such a cool laptop | 17:12 |
| bkeys | I mean it's already cool; but mainline Fedora support would be great | 17:12 |
| josch | i don't think u-boot can access nvme | 17:13 |
| bkeys | So how does the reform system image boot from nvme? | 17:13 |
| vkoskiv | I'd like to flash the rescue system on my emmc to update it, but I forget where that was documented? | 17:14 |
| vkoskiv | I think i twas just setting some flag to enable writing and then dd? | 17:14 |
| vagrantc | bkeys: put an EFI boot partition on microSD or eMMC or USB | 17:14 |
| josch | vkoskiv: reform-flash-rescue | 17:14 |
| vkoskiv | Should have guessed it :D | 17:14 |
| bkeys | So Fedora has an EFI boot partition built into the image; essentially I copy that partition over to emmc? | 17:14 |
| vagrantc | bkeys: might work ... not super familiar | 17:15 |
| vkoskiv | I'll do the flash when I'm not on a flaky train hotspot. | 17:16 |
| bkeys | Okay, I am serial console'd into the reform | 17:16 |
| josch | awesome :) | 17:17 |
| bkeys | So gparted doesn't show mmc does it? | 17:17 |
| vkoskiv | Speaking of data, I was using my reform last night to clone some hard drives and it was behaving really weird. Slow transfers that often just stopped after a gigabyte or so. | 17:17 |
| vkoskiv | I did the same on another laptop and it worked fine. | 17:18 |
| vkoskiv | I was in a rush so I didn't investigate further. Any known issues with the usb ports or anything like that? | 17:19 |
| minute | vkoskiv: which usb port were you using? there are 2 on a hub and 1 direct | 17:25 |
| minute | the one close to the HDMI port is direct | 17:25 |
| josch | is that documented in the handbook? | 17:26 |
| - MajorBiscuit (QUIT: Quit: WeeChat 3.5) (~MajorBisc@wlan-145-94-166-231.wlan.tudelft.nl) | 17:26 | |
| vkoskiv | I tried them all | 17:26 |
| vkoskiv | I can do more testing this weekend. | 17:26 |
| vkoskiv | I do seem to recall the one next to the HDMI port was a bit faster | 17:26 |
| minute | maybe power issue? | 17:26 |
| vkoskiv | Could be placebo since I recall reading about it a while back | 17:26 |
| vkoskiv | Very possible! What kind of current are these ports rated for? | 17:27 |
| minute | i have made huge transfers with usb sticks and didn't have this issue | 17:27 |
| vkoskiv | I plugged in this SATA usb 3 adapter that powered a laptop sized SSD drive | 17:28 |
| vkoskiv | I did get an external power supply for the bigger mechanical hard drive I was cloning. | 17:28 |
| vkoskiv | That one seemed to work fine. | 17:28 |
| vkoskiv | I cloned the old drive onto the nvme and then from the file to the ssd | 17:29 |
| vkoskiv | And that second step which would use more current from the port didn | 17:29 |
| vkoskiv | didn't work* | 17:29 |
| minute | ah yeah i had some issues with a usb sata ssd adapter before... could be power issue indeed. | 17:30 |
| minute | these things are kinda flaky sometimes | 17:30 |
| vkoskiv | It's called HARDware for a reason. | 17:30 |
| bkeys | It'd be really cool to have edk2 on this thing | 17:32 |
| minute | with the soquartz+rcm4 you have edk2 ^^^^ | 17:32 |
| bkeys | When do you think that will be a thing? | 17:33 |
| vagrantc | are those adapter boards available yet? | 17:33 |
| minute | oh wait actually there is edk2 for imx8m, from microsoft/nxp | 17:33 |
| bkeys | See I know I.MX8 has decent 3D performance so I can live with less CPU performance | 17:33 |
| minute | vagrantc: they exist in my lab, but we don't sell 'em yet because i didn't have time for full validation... jacqueline also made one and is working on it | 17:34 |
| bkeys | 4GB of RAM can be a bottleneck, but I can do a swapfile on nvme | 17:34 |
| minute | bkeys: https://github.com/ms-iot/imx-iotcore | 17:34 |
| vkoskiv | 19.7% packet loss on this train jouney. Not bad. (using the train public wifi) | 17:34 |
| vkoskiv | s/jouney/journey | 17:34 |
| bkeys | minute: I will take a look once I get Fedora on here | 17:34 |
| vagrantc | minute: that's what i had thought ... you taunt us with your early access :) heh. | 17:34 |
| bkeys | Right now I am reinstalling KDE so I can run gparted and make a UEFI partition for the emmc | 17:34 |
| minute | vagrantc: if you have soldering skills you can download rcm4 ^^ | 17:35 |
| bkeys | What is the 3D performance of soquarts versus the I.MX8? | 17:36 |
| - chomwitt (QUIT: Ping timeout: 258 seconds) (~chomwitt@2a02:587:dc1f:5600:b0ac:64b:5946:8d06) | 17:36 | |
| minute | no idea | 17:36 |
| minute | i haven't got soquartz to run | 17:36 |
| minute | https://wiki.pine64.org/wiki/SOQuartz | 17:36 |
| vagrantc | minute: not myself, but i know a few people ... although would kind of like to get the original module working reasonably well first ... not that i've put much time into it | 17:36 |
| minute | > Mali-G52 2EE Bifrost GPU@800MHz | 17:36 |
| minute | Supports 1600 Mpix/s fill rate when at 800MHz clock frequency | 17:37 |
| minute | Supports 38.4 GLOP/s when at 800MHz clock frequency | 17:37 |
| bkeys | You get opengl 3.2 which aint bad | 17:37 |
| bkeys | more than I.mx8 | 17:37 |
| vagrantc | 8GB of ram is pretty exciting | 17:37 |
| minute | > According to the i.MX8M reference manual, the GC7000lite at 800Mhz is capable of 1.6 Giga pixel/sec fill rate | 17:37 |
| bkeys | but is that with etnaviv? | 17:38 |
| minute | bkeys: do you get gl 3.2 with mesa? | 17:38 |
| bkeys | nope | 17:38 |
| bkeys | 2.0 I believe | 17:38 |
| minute | gles probably? | 17:38 |
| minute | i mean on bifrost | 17:38 |
| minute | with panfrost | 17:38 |
| minute | don't they have vulkan now? | 17:39 |
| bkeys | this pinebook pro has panfrost on it | 17:39 |
| bkeys | they are working on it last I heard | 17:39 |
| minute | https://www.collabora.com/news-and-blog/blog/2021/06/11/open-source-opengl-es-3.1-on-mali-gpus-with-panfrost/ | 17:39 |
| bkeys | I got 3.1 on here | 17:39 |
| minute | reads like gles 3.1 to me, and that was 1 year ago | 17:39 |
| minute | bkeys: aha! so that's better than etnaviv which is at 2.0 | 17:40 |
| bkeys | Yeah, I plan on getting an adapter board if it means 3.1 and possible edk2 | 17:40 |
| minute | https://www.phoronix.com/scan.php?page=news_item&px=PanVK-Lands-In-Mesa | 17:40 |
| bkeys | josch: So gparted says the entire emmc is filled with MNTRESCUE; but not all 16 gigs are being used, I take it I am safe to shrink the partition | 17:55 |
| minute | why would you want to shrink it? | 17:57 |
| bkeys | To make room for EFI partition | 17:57 |
| minute | ah | 18:03 |
| bkeys | So I have the EFI partition on the mmcblk0p2 partition, I am trying to figure out how to mount it on uboot | 18:12 |
| bkeys | I found the grub.efi file that I believe I will have to load | 18:12 |
| bkeys | I do ls mmc 0:2 and I can see inside the partition | 18:13 |
| bkeys | I want to look inside of the EFI folder within it though, how do I do that? | 18:13 |
| bkeys | ls mmc 0:2 EFI | 18:14 |
| - cinap_lenwerk (PART: !!unknown attribute: msg!!) (~cinap_len@232.red-83-48-121.staticip.rima-tde.net) | 18:19 | |
| - bkeys (QUIT: Ping timeout: 256 seconds) (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 18:29 | |
| - mjw (QUIT: Killed (NickServ (GHOST command used by mark_!~mark@gnu.wildebeest.org))) (~mjw_@2001:1c06:2488:1400:9e5c:8eff:fe8f:a440) | 18:36 | |
| * mark_ -> mjw | 18:36 | |
| + wielaard (~mjw_@2001:1c06:2488:1400:9e5c:8eff:fe8f:a440) | 18:36 | |
| + bkeys (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 18:38 | |
| bkeys | minute, josch: So there is a BOOTAA64.EFI file in the EFI partition, how do I tell uboot that is the file I want to boot efi from? | 18:39 |
| + mlarkin (~mlarkin@047-048-086-214.biz.spectrum.com) | 18:52 | |
| bkeys | Okay guys, I got a grub command line! | 18:55 |
| - bkeys (QUIT: Remote host closed the connection) (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 19:02 | |
| + bkeys (~Thunderbi@static-198-54-135-69.cust.tzulo.com) | 19:03 | |
| minute | bkeys: great! | 19:04 |
| minute | bkeys: how did you do it? | 19:04 |
| bkeys | So I wrote the mainline Fedora image to my nvme drive, then I found this | 19:04 |
| bkeys | https://u-boot.readthedocs.io/en/latest/develop/uefi/uefi.html | 19:04 |
| bkeys | I loaded the grubaa64.efi just like it told me, I had to adjust the variable names, like we don't use kernel_addr_r on the reform, it's just loadaddr | 19:05 |
| bkeys | So once I specified that as well as the device tree included with your system image, I simply ran bootefi and I got a grub prompt | 19:05 |
| minute | neat, and grub already shows fedora? | 19:05 |
| bkeys | Once I figure this out end to end i will document this and post a guide on the community forums | 19:05 |
| bkeys | No, I need help with this | 19:06 |
| minute | ah, grub prompt, not menu, right | 19:06 |
| bkeys | I got a grub command prompt | 19:06 |
| bkeys | No menu on the screen or on serial console no | 19:06 |
| minute | unfortunately i don't know much about grub | 19:06 |
| bkeys | vmlinuz wasn't showing up, so my guess is I had the EFI partition mounted, but not the boot partition | 19:06 |
| bkeys | Me neither | 19:06 |
| bkeys | But it should "just work" | 19:06 |
| bkeys | It's not getting a kernel, which makes sense | 19:06 |
| minute | IIRC grub needs to load a bunch of modules to have any functionality, and a cfg to show a menu | 19:07 |
| bkeys | I have learned a lot about uboot, if I can get mainline Fedora on here like this I can do it on anything | 19:07 |
| bkeys | Yes and the Fedora image shows those files, I just have to present those files to grub from uboot somehow | 19:07 |
| minute | i see | 19:07 |
| bkeys | But if I can get this working it should be the same on other mainline distros; and hopefully the community can branch out from the system image | 19:08 |
| bkeys | But first things first I gotta figure this out | 19:08 |
| bkeys | I believe the grub.cfg file will be presented to grub via the bootargs | 19:09 |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20) | 19:14 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:40) | 19:20 | |
| bkeys | minute: I am gonna go into the fedora-arm-installer source code and see what it does on ARM boards | 19:26 |
| josch | bkeys: this is awesome news! Please keep us posted. :) | 19:34 |
| bkeys | Yeah, how I see it is if we can get more mainline distro images booting on here (given they work well) then it might take some load off of y'all | 19:34 |
| bkeys | I just wish I knew what the bootargs variable was for other systems | 19:35 |
| bkeys | I actually have an SBC that already runs mainline Fedora and uses uboot; I could bring it over here and see what it does | 19:35 |
| josch | if I'm reading include/config_distro_bootcmd.h correctly, then https://source.mnt.re/reform/reform-boundary-uboot/-/merge_requests/6 it also runs bootefi, so maybe this MR will make efi boot work already automatically | 19:41 |
| bkeys | I changed the bootcmd to automatically go to the grub command line, but it doesn't persist despite me running saveenv, does anyone know why that might be? | 19:42 |
| bkeys | Maybe you guys can answer that question, I'm gonna go out to my lab and get the SBC that already runs grub so I can see what it does | 19:45 |
| bkeys | Well this thing has a lot more in it's env than the reform that's for sure | 19:53 |
| bkeys | josch, minute: Do y'all wannna take a look at the printenv for this other SBC? | 19:54 |
| bkeys | I am making some sense of it, I can tell a lot of it has nothing to do with EFI | 19:54 |
| josch | bkeys: sure, if you copypaste it somewhere | 19:55 |
| josch | bkeys: but vagrantc can probably make more sense of it than I can :D | 19:55 |
| bkeys | Booting /efi\boot\bootaa64.efi | 19:59 |
| bkeys | I saw that line, that's relevant | 19:59 |
| bkeys | That is when it was booting | 19:59 |
| bkeys | That lets me know what EFI file I need to boot | 19:59 |
| bkeys | josch, minute, vagrantc: https://paste.centos.org/view/2e66b624 | 20:03 |
| bkeys | Here is the printenv of my Khadas VIM3 it runs mainline Fedora with grub EFI using uboot | 20:03 |
| josch | bkeys: yes, that's distro_bootcmd stuff, so that's exactly what https://source.mnt.re/reform/reform-boundary-uboot/-/merge_requests/6 is implementing. | 20:03 |
| bkeys | Okay so I probably wanna use this branch | 20:04 |
| bkeys | Is there a build artifact I can flash to my emmc? | 20:05 |
| minute | josch: is that safe to merge? | 20:05 |
| bkeys | I mean I am definitely willing to test it if someone has an image to share with me | 20:05 |
| josch | bkeys: you can just switch to that branch and rebuild to try it out | 20:06 |
| bkeys | Yeah I just figure I got my reform kinda taken apart and such; but I can build it | 20:06 |
| bkeys | I'll just SSH into it | 20:06 |
| josch | minute: it is, because the reform-system-image script uses a fixed tag on that repo, so even after merging it, it will not automatically end up in the images we create | 20:06 |
| josch | minute: i think the other consumer of that repo is sigrid with void linux but i think sigrid also uses a fixed tag? | 20:07 |
| sigrid | yep, I use a tag | 20:09 |
| bkeys | sigrid: So on Void Linux are you setting up Grub EFI? | 20:11 |
| sigrid | nope. I use u-boot | 20:17 |
| sigrid | (without efi) | 20:17 |
| bkeys | josch: Okay I cloned your repo and am trying to build off the distro_bootcmd branch | 20:19 |
| bkeys | Okay I had to do make menuconfig (I never built uboot from source before), josch, do I have to change anything? | 20:20 |
| josch | you don't need to do menuconfig | 20:20 |
| josch | bkeys: https://community.mnt.re/t/uboot-build-fails-due-to-build-flags/1002/2 | 20:20 |
| bkeys | Now do I need to build this, or is it available as an artifact from your pull request? | 20:21 |
| josch | bkeys: it will only be available as an artifact after lukas merged my MR | 20:23 |
| josch | but building is simple, just apt install the packages and run those two commands from that post | 20:23 |
| bkeys | Where is flash.bin located? | 20:25 |
| bkeys | The make command finished | 20:25 |
| bkeys | Ah, had a few missing packages | 20:30 |
| bkeys | Okay josch, I dd this flash.bin file to /dev/mmcblk0boot0 or something like that, right? | 20:30 |
| josch | flash.bin is in your current directory | 20:31 |
| josch | you flash it to your sd-card like this: | 20:31 |
| josch | dd if=flash.bin of=/dev/mmcblk0 conv=notrunc bs=1k seek=33 | 20:32 |
| josch | flashing to emmc is a bit more tricky, you first need to run: | 20:32 |
| josch | echo 0 > /sys/class/block/mmcblk0boot0/force_ro | 20:32 |
| josch | and then: | 20:33 |
| josch | dd if=/boot/flash.bin of=/dev/mmcblk0boot0 bs=1024 seek=33 | 20:33 |
| josch | and then lock it again: | 20:33 |
| josch | echo 1 > /sys/class/block/mmcblk0boot0/force_ro | 20:33 |
| josch | during development, I'd do everything from the sd-card first and migrate to emmc or nvme once stuff starts working | 20:33 |
| josch | operating the dip switch or writing to emmc is just more painful than swapping the sd-card | 20:33 |
| bkeys | Yeah I borked the install on the emmc, time to revert to SD card | 20:35 |
| vkoskiv | reform-boot-config says it can't find /dev/nvme0n1p1 | 20:40 |
| vkoskiv | I did set ROOTPART | 20:40 |
| minute | i just merged | 20:41 |
| bkeys | How long will it take flash.bin to show up as a build artifact minute? It'd be nice to not have to build it again | 20:44 |
| josch | bkeys: https://source.mnt.re/reform/reform-boundary-uboot/-/jobs/809/artifacts/browse | 20:46 |
| erle | josch minute so artifacts are only generated for merged stuff and not for branches? | 20:49 |
| josch | erle: yes, the ci is set up such that only the reform projects are run with it | 20:50 |
| bkeys | Ugh, now I am failing to boot, it doesn't see mmcblk0p1 | 20:54 |
| kfx | I wish u-boot would boot off the sd card if present and fall back to the emmc :( | 20:55 |
| vkoskiv | How do I configure the ROOTPART to point to my encrypted + LVM volume? | 21:01 |
| vkoskiv | I tried it, flipped the switch but I ended up in the emmc rescue system | 21:02 |
| vkoskiv | It didn't even ask me to unlock the drive | 21:02 |
| vkoskiv | I set up the thing with the reform-setup-encrypted-nvme script | 21:03 |
| bkeys | josch: So I flashed flash.bin to the front of the SD card, booted from it and I still don't have any distroboot stuffo n here | 21:04 |
| bkeys | Then again ithas my changes on here I was trying to save earlier??? | 21:05 |
| josch | kfx: that's exactly what our u-boot does | 21:09 |
| josch | vkoskiv: can i come back to you later? i have my baby sleeping on my right now XD | 21:10 |
| vkoskiv | Yeah absolutely no hurry | 21:10 |
| vkoskiv | I'm going to be messing with this thing all weekend :D | 21:10 |
| josch | bkeys: that sounds like it's not using u-boot from your sd-card but from emmc | 21:10 |
| bkeys | Let me take off the heatsink and check... | 21:11 |
| bkeys | The switch is turned towards ON josch | 21:14 |
| kfx | josch: the flash.bin from job 809 is configured for emmc use, and will do sdcard -> emmc boot order? | 21:16 |
| kfx | worth a try | 21:17 |
| minute | ON means SD card. | 21:18 |
| josch | kfx: this is what this line is supposed to do: https://source.mnt.re/reform/reform-boundary-uboot/-/commit/2dd6c0e7e2ee54784169d68933ab5b2e44881585#fee690d794fc086e909bc80b182c8942ac1fbe9d_179_179 | 21:21 |
| josch | if it doesn't try sdcard first, then that's a bug i will fix | 21:21 |
| kfx | roger | 21:22 |
| kfx | will report results | 21:22 |
| bkeys | Indeed it is reading the uboot from the emmc | 21:25 |
| bkeys | Let me take the heatsink back off | 21:25 |
| - Christoph_ (QUIT: Remote host closed the connection) (~Christoph@p54bf61d7.dip0.t-ipconnect.de) | 21:27 | |
| + Christoph_ (~Christoph@p54bf61d7.dip0.t-ipconnect.de) | 21:30 | |
| vkoskiv | I ran ROOTPART=/dev/reformvg/root reform-boot-config --emmc emmc | 21:52 |
| vkoskiv | To try and get this to boot from my nvme | 21:52 |
| vkoskiv | Flipped the switch to OFF and now the emmc doesn't even boot anymore | 21:52 |
| vkoskiv | I must have messed it up somehow | 21:52 |
| vkoskiv | Apparently the switch is on, it's trying the sd card... | 21:54 |
| vkoskiv | Huh, I did flip it. It's OFF but it's still trying the sd | 21:55 |
| minute | what do you mean by "it is trying" | 21:56 |
| minute | the switch tells the imx to load uboot from emmc or sd | 21:56 |
| vkoskiv | I flipped the switch to OFF, it wouldn't boot, I put an SD in and it booted from there | 22:00 |
| vkoskiv | I now messed with the switch, trying again | 22:00 |
| vkoskiv | Yeah something is weird, it's still expecting an SD card even though the switch is off | 22:02 |
| kfx | josch: I'm guessing this is not desired output from that flash.bin http://sciops.net/paste/1654891315.cb36eecb | 22:02 |
| vkoskiv | I *think* it's supposed to use the emmc when the switch is off. | 22:04 |
| vkoskiv | That's the way it was before. | 22:04 |
| vkoskiv | I might have mixed it up though. The docs don't say which setting is which | 22:05 |
| kfx | vkoskiv: minute said "on means sd card" earlier | 22:06 |
| vkoskiv | Ah. Well, on mine it tries the SD card also when it's set to OFF | 22:07 |
| vkoskiv | Yeah looks like both on or off it is booting from the SD card | 22:08 |
| vkoskiv | I'm suspecting a hardware fault in the switch | 22:09 |
| minute | maybe the emmc bootloader is invalid and it falls back to sd? | 22:12 |
| vkoskiv | Oh, that would be more likely if that's possible | 22:12 |
| minute | did you change the bootconfig dip switches perhaps? | 22:12 |
| vkoskiv | Depends on where those are located | 22:12 |
| vkoskiv | I've only touched the one on the SoM | 22:12 |
| minute | on the motherboard | 22:12 |
| minute | ok | 22:12 |
| minute | might be a fallback then | 22:12 |
| minute | I haven't tried to nuke the emmc bootloader myself to see what happens | 22:12 |
| vkoskiv | What I want is for this to boot from my luks+lvm nvme ssd without the sd card installed | 22:12 |
| vkoskiv | I didn't know it would do a fallback, so a software problem seems more likely | 22:13 |
| minute | i think for that to work you need to flash v3 rescue image to emmc | 22:13 |
| vkoskiv | Not sure how to go about fixing this now | 22:13 |
| vkoskiv | I did that earlier and it booted fine | 22:13 |
| vkoskiv | Then I did something and it stopped booting | 22:13 |
| vkoskiv | I'll try reflashing it | 22:13 |
| minute | so... flash it back again | 22:13 |
| minute | yeah | 22:13 |
| + chomwitt (~chomwitt@2a02:587:dc1f:5600:518c:c264:c8fd:1f6b) | 22:15 | |
| vkoskiv | the flash script spits a warnin: target 'reform_crypt' not found in /etc/crypttab | 22:15 |
| vkoskiv | Not sure if that's a problem | 22:15 |
| minute | which flash script? | 22:16 |
| vkoskiv | reform-flash-rescue, I think? | 22:17 |
| vkoskiv | emmc boots again! | 22:17 |
| vkoskiv | But yeah I now want it to use my luks+lvm nvme instead of kicking me to the recovery img | 22:18 |
| minute | ok, i'm not sure how that works with v3, as i haven't tried it myself | 22:18 |
| minute | i personally boot from sd->encrypted nvme | 22:19 |
| vkoskiv | Yeah that's what I've been using and that works nicely | 22:19 |
| vkoskiv | Maybe it's preferable | 22:19 |
| minute | not sure what's different @ emmc boot process | 22:20 |
| vkoskiv | I just thought I don't want the SD card sitting there for every boot | 22:20 |
| minute | yeah, i mean what is technically different that would cause it to not work | 22:22 |
| minute | anyway i'm pretty exhausted from the week, gonna chill out a bit | 22:22 |
| vkoskiv | Yeah, have a fun weekend! | 22:24 |
| vkoskiv | I know I will, this thing is very cool :D | 22:24 |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:40) | 22:27 | |
| minute | i have a prototype trackball (rp2040 based) with white switches, the clicking is frankly addictive | 22:45 |
| vkoskiv | I like the CLACK these brown ones make too :D | 22:47 |
| vkoskiv | I get this sometimes when booting and in dmesg: | 22:47 |
| vkoskiv | nvme nvme0: I/O 54 QID 1 timeout, completion polled | 22:47 |
| vkoskiv | Not sure I understand what that means | 22:48 |
| vkoskiv | I'm using an Intel 600p in here. | 22:48 |
| vkoskiv | Seems to work fine. | 22:48 |
| vkoskiv | Also - Anyone here set up TRIM? is that even a thing anymore? | 22:50 |
| kfx | trim works normally for me, just set the discard option on your luks device and also in the filesystem mount options | 22:52 |
| kfx | I don't know the systemd magic for it; I use it with alpine linux | 22:52 |
| minute | i read once that discard shouldn't be used nowadays? | 22:53 |
| vkoskiv | minute: *hint hint* reform merch pls | 22:54 |
| vkoskiv | I'd wear the heck out of a black shirt with a white MNT logo :D | 22:54 |
| vkoskiv | s/reform merch/MNT merch | 22:54 |
| minute | yes, we're gonna launch some merch :3 | 22:56 |
| kfx | minute: that's urban legend junk from the arch wiki. zomg speed haxors don't like continuous trim; they prefer to let a systemd service schedule it during lower-load moments | 22:56 |
| kfx | there's no meaningful difference, really, it's just another thing that systemd has declared it should be in charge of because we peons aren't smart enough | 22:56 |
| minute | ok | 22:56 |
| kfx | to be honest, because debian is systemd-based it's probably much simpler to just `systemctl enable fstrim.timer` than going through the trouble of setting up continuous trim | 22:58 |
| josch | kfx: yes, this is not the u-boot output that i got when i tried it | 23:00 |
| josch | vkoskiv: so, after flashing a recent sysimage-v3 onto an sd-card you want to run the following commands to set up booting from encrypted nvme: | 23:04 |
| josch | reform-setup-encrypted-nvme | 23:04 |
| josch | if you answer y at the second prompt, then this will also call reform-migrate for you which in turn will also call reform-boot-config with the right argument | 23:05 |
| vkoskiv | I did this part, I'm running off the nvme + sd at the moment | 23:05 |
| josch | after this is done, reboot and confirm that this setup is working and successfully boots your encrypted system on nvme | 23:05 |
| josch | good | 23:05 |
| josch | second step is setting up booting from emmc instead of sd | 23:05 |
| josch | for that, you first run reform-flash-rescue so that your emmc contains an up-to-date sysimage-v3 | 23:06 |
| vkoskiv | I did that and verified that it's now v3 | 23:07 |
| josch | good | 23:07 |
| josch | then, lastly you run: | 23:07 |
| josch | ROOTPART="/dev/reformvg/root" reform-boot-config --emmc sd | 23:08 |
| vkoskiv | Huh, I've put the last arg as emmc this whole time :D | 23:08 |
| josch | this will configure the boot partion on emmc such that it will boot from nvme | 23:08 |
| josch | the last arg is ignored | 23:08 |
| vkoskiv | I'll try this | 23:08 |
| josch | so putting emmc there doesn't change its behviour | 23:08 |
| josch | the last arg is a shortcut to setting ROOTPART to some default values | 23:09 |
| josch | if that doesn't work, i need more information about the things that do happen for you after running all of these commands | 23:09 |
| vkoskiv | running now | 23:09 |
| vkoskiv | Then flip the switch right? | 23:10 |
| vkoskiv | Is 'systemctl poweroff' the recommended way to shut down? It doesn't seem to shut down programs gracefully | 23:10 |
| vkoskiv | Weird. The circle key stopped working | 23:13 |
| josch | vkoskiv: you have to press it for longer than a second for the menu to show up | 23:14 |
| vkoskiv | I know. | 23:14 |
| vkoskiv | Holding it for 30+ sec, nothing. | 23:14 |
| _E | bkeys: glad someone's taking the time to do it The Right Way™ rather than using my workarounds | 23:15 |
| josch | vkoskiv: maybe you are hit by this: https://community.mnt.re/t/power-issues-and-lpc-wakeup/596 | 23:15 |
| vkoskiv | disconnect+reconnect of batteries fixed it | 23:16 |
| vkoskiv | I'm fairly certain the lpc+kbd firmware need flashing | 23:16 |
| vkoskiv | Haven't goetten to that yet | 23:16 |
| vkoskiv | Looks like it won't boot from emmc now. | 23:16 |
| vkoskiv | So something got messed up there again | 23:17 |
| vkoskiv | Same as last time when I tried to get this working | 23:17 |
| vkoskiv | SoM switch is OFF, and it does nothing when I boot | 23:18 |
| vkoskiv | If I stick in an SD card, it boots fine from there. | 23:18 |
| josch | vkoskiv: can you connect to your reform via serial and see if at least u-boot loads? | 23:19 |
| vkoskiv | I don't think I have the right hardware | 23:20 |
| vkoskiv | I have a random DB9 usb serial thing | 23:20 |
| vkoskiv | I'll need an adapter I guess? | 23:20 |
| josch | db9 sounds like rs232 and that's the wrong thing | 23:21 |
| vkoskiv | Yeah it's rs232 | 23:21 |
| josch | more precisely you need some usb->ttl adapter at 3.3 V signal level | 23:22 |
| vkoskiv | I also have this random guruplug jtag board with rs232 and uart :D | 23:23 |
| vkoskiv | And jtag, of course. | 23:23 |
| vkoskiv | I'm happy booting with the SD for now. I think I'll go about flashing the LPC, keyboard and maybe trackball firmware | 23:24 |
| vkoskiv | Since I'd like to work on improving those anyway | 23:24 |
| josch | you can try another thing | 23:28 |
| josch | you can take an empty sd card (just zero the partition table) and just flash u-boot onto it | 23:28 |
| josch | u-boot should first try booting from sd-card (and fail) and then try emmc (and succeeds because you ran reform-boot-config --emmc earlier) | 23:29 |
| josch | since you only have u-boot on that sd-card and not kernel nor initrd, you can remove the sd-card immediately after u-boot loaded the kernel | 23:29 |
| josch | this means you can "boot from sd" but still have a usable sd-card slot during normal operation | 23:29 |
| josch | you'd then have u-boot on an sd-card, kernel+initrd on emmc and rootfs on nvme | 23:30 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
