| minute | josch: ahh why does it fail? | 00:19 |
|---|---|---|
| minute | tada https://mnt.re/system-image | 00:20 |
| minute | josch: i made a bash script with some jq magic and a cronjob | 00:20 |
| minute | it redirects to the latest successful job with ref=main | 00:21 |
| minute | in project 3 (reform-system-image) | 00:21 |
| josch | minute: i'm not quite sure why it fails -- there are some a311d warnings for the dtb though | 00:22 |
| minute | josch: ok, will probably look into it tomorrow | 00:24 |
| josch | minute: thank you for fixing https://mnt.re/system-image -- long term we might want more tags and releases i think. Sometimes it would be nice to have a few weeks or month old images available | 00:24 |
| josch | minute: nooo! you have holidays! | 00:24 |
| minute | ok :D | 00:30 |
| josch | oh whoops, found the problem | 00:32 |
| minute | ohh? | 01:00 |
| noam | it me | 01:00 |
| noam | i the problem | 01:00 |
| josch | i just threw out the cadence patch stack and replaced it with the new v15 version | 01:19 |
| josch | at least for this kernel version, that was much less problematic | 01:19 |
| - bluerise (QUIT: Ping timeout: 256 seconds) (~bluerise@pc19f8b89.dip0.t-ipconnect.de) | 01:22 | |
| + bluerise (~bluerise@pc19f8d63.dip0.t-ipconnect.de) | 01:22 | |
| minute | josch: great to hear | 01:23 |
| josch | indeed -- but who knows whether this works out-of-the-box or not :D | 01:24 |
| josch | and the mipi dsi patch was already in 20240506_heiko_mipi_dsi_phy_for_rk3588.patch | 01:31 |
| josch | and CONFIG_PHY_ROCKCHIP_SAMSUNG_DCPHY=m is already in the config | 01:32 |
| minute | nice | 01:36 |
| - mjw (QUIT: Ping timeout: 256 seconds) (~mjw@gnu.wildebeest.org) | 01:56 | |
| + wakest (m-a7d6fe@45.77.48.108) | 02:46 | |
| - chomwitt (QUIT: Ping timeout: 264 seconds) (~chomwitt@2a02:587:7a02:1a00:1ac0:4dff:fedb:a3f1) | 03:16 | |
| + chomwitt (~chomwitt@2a02:587:7a09:e500:1ac0:4dff:fedb:a3f1) | 03:18 | |
| - nsc (QUIT: Ping timeout: 255 seconds) (~nicolas@i5C74DC29.versanet.de) | 03:24 | |
| + nsc (~nicolas@i5C74DF86.versanet.de) | 03:26 | |
| - chomwitt (QUIT: Ping timeout: 264 seconds) (~chomwitt@2a02:587:7a09:e500:1ac0:4dff:fedb:a3f1) | 03:29 | |
| josch | so... the good news is, that kernel 6.9 built successfully and it even includes cdns_mhdp_imx.ko | 07:19 |
| josch | the bad news is, that then the job fails when it uploads the artifacts: Uploading artifacts as "archive" to coordinator... 201 Created id=4654 responseStatus=201 Created token=glcbt-64 | 07:19 |
| + pr8dan (~pr8dan@user/pr8dan) | 08:02 | |
| pr8dan | So, newbie here trying to get nonbooting pocket back up (did an apt upgrade yesterday). I am just download an img to boot from sd. The general idea would then to copy vmlinuz and modules to emmc? be to copy vmlinuz | 08:10 |
| josch | pr8dan: yes, that worked for pandora yesterday | 08:24 |
| pr8dan | ok thx - will try and feedback thereafter | 08:25 |
| josch | pr8dan if you or pandora or anybody else wants to help out, we just built kernel 6.9 which may or may not fix this problem. If you want to help, feel free to start from a vanilla system image, download linux-image-6.9.2-mnt-reform-arm64*.deb from here: https://source.mnt.re/reform/reform-debian-packages/-/jobs/4654/artifacts/browse/changes/ install and boot and see what happens | 08:26 |
| josch | Lukas is on holiday, so we can help out by doing our own testing and maybe have things fixed when they come back from thair vacation. :) | 08:27 |
| - pr8dan (QUIT: Ping timeout: 240 seconds) (~pr8dan@user/pr8dan) | 08:46 | |
| pandora | I can do that after breakfast | 09:21 |
| josch | that would be awesome! | 09:25 |
| + pr8dan (~pr8dan@user/pr8dan) | 09:32 | |
| pr8dan | ok can confirm that booting from sd card works again (probably not too surprising) | 10:10 |
| + megasquid (~megasquid@user/megasquid) | 10:15 | |
| josch | pr8dan: maybe not surprising but it's always re-assuring to have more people confirm results :) | 10:25 |
| pr8dan | I just copied vmlinuz from the sd to the emmc boot and rebooted - i did not yet copy any modules. rebooting works. where would I find problems with modules? dmesg? | 10:25 |
| josch | pr8dan: if rebooting works after just copying vmlinuz and if your system seems otherwise normal, then this is probably all there is to it! | 10:28 |
| josch | in that case, thank you for trying out just copying over vmlinuz -- this then means that whatever happened, happend in kernel image itself... mysterious... | 10:29 |
| pr8dan | in dmesg i do get some errors on direct firmware load for ark6_wlan and imx-sdma (but as you can guess i am pretty clueless) | 10:32 |
| josch | pr8dan: those should be fine | 10:33 |
| pr8dan | i will install your deb above on sd and feedback | 10:33 |
| josch | thank you! :) | 10:33 |
| ch | josch: the artifact upload is not the reason for an 'exit code 1' failure | 10:41 |
| josch | ch: do you know what is? | 10:41 |
| ch | josch: AFAIK 'exit code N' stuff is propagated from the last thing that happened before 'Running after_script' | 10:42 |
| josch | interesting... then i have to investigate further, thank you! | 10:42 |
| ch | (it also says 'Uploading artifacts for failed job', so at that point it was already *failed*) | 10:42 |
| josch | aha i didn't spot that, nice catch! :) | 10:43 |
| ch | i would imagine the grep didn't match or something | 10:43 |
| josch | yes, i have to improve the reporting for that check so that it's easy to spot the issue | 10:44 |
| - nybble (QUIT: Ping timeout: 260 seconds) (~sprang@user/csprng) | 10:44 | |
| josch | that's it! the driver is now stored in: | 10:45 |
| josch | /usr/lib/modules/6.9.2-mnt-reform-arm64/kernel/drivers/gpu/drm/bridge/cadence/cdns-mhdp8546.ko.xz | 10:45 |
| josch | ch: thank you! :) | 10:46 |
| pr8dan | pocket: install linux-image-6.9.2-mnt-reform-arm64_6.9.2-1~exp1+reform20240621T224758Z_arm64.deb over sd with clean img and it fails to boot again | 10:46 |
| josch | drats :( | 10:46 |
| josch | pr8dan: thank you for trying this out | 10:47 |
| josch | this will require more debugging... | 10:47 |
| pr8dan | happy to help (in a limited way) | 10:47 |
| + reform1462 (~user@business-90-187-186-49.pool2.vodafone-ip.de) | 10:50 | |
| reform1462 | so u want us to test this file: https://source.mnt.re/reform/reform-debian-packages/-/jobs/4654/artifacts/file/changes/linux-image-6.9.2-mnt-reform-arm64_6.9.2-1~exp1+reform20240621T224758Z_arm64.deb | 10:51 |
| josch | reform1462: yes, pr8dan already tried it without success | 10:52 |
| pandora | Oh yeah missed the last messages | 10:53 |
| - reform1462 (QUIT: Remote host closed the connection) (~user@business-90-187-186-49.pool2.vodafone-ip.de) | 11:10 | |
| + mjw (~mjw@gnu.wildebeest.org) | 11:30 | |
| pandora | @josch where are the build scripts for the kernel located? I would like to have a look how it is done | 11:32 |
| + andreas-e (~Andreas@2001:861:c4:f2f0::c64) | 12:10 | |
| - megasquid (QUIT: Ping timeout: 256 seconds) (~megasquid@user/megasquid) | 12:13 | |
| josch | pandora: this is the script: https://source.mnt.re/reform/reform-debian-packages/-/blob/main/linux/build.sh?ref_type=heads | 12:14 |
| josch | pandora: how you call it is documented in the README | 12:14 |
| josch | it's complicated because the Debian linux kernel packaging is not intended to be used by downstreams like us | 12:14 |
| josch | i sat down together with Debian kernel maintainer Ben Hutchings a few weeks ago and we improved the situation a bit thanks to their input | 12:15 |
| josch | but it's still quite a bit fiddle unfortunately | 12:15 |
| pandora | i will have a look at it and see how much i understand :D | 12:32 |
| - andreas-e (QUIT: Quit: Leaving) (~Andreas@2001:861:c4:f2f0::c64) | 12:41 | |
| - pr8dan (QUIT: Ping timeout: 240 seconds) (~pr8dan@user/pr8dan) | 12:51 | |
| josch | pandora: feel free to ask me anything :) | 12:53 |
| + pr8dan (~pr8dan@pub151248192164.dh-hfc.datazug.ch) | 13:00 | |
| - pr8dan (QUIT: Changing host) (~pr8dan@pub151248192164.dh-hfc.datazug.ch) | 13:00 | |
| + pr8dan (~pr8dan@user/pr8dan) | 13:00 | |
| - pr8dan (QUIT: Remote host closed the connection) (~pr8dan@user/pr8dan) | 13:05 | |
| noam | josch: mother's maiden name? /s | 13:25 |
| josch | noam: can you explain the intended sarcasm please? i'm not familiar with the expression i'm afraid :) | 13:38 |
| noam | josch: it's often used as a "security" question | 13:41 |
| noam | if people answer it honestly, which AFAIK many do, then if you know their mother's maiden name, you don't need their password on a _lot_ of sites | 13:41 |
| noam | I used a randomly generated one for my bank account IIRC | 13:42 |
| noam | had to read out a buncha symbols to the exasperated person on the other end of the phone because apparently they had a human look it over lol | 13:42 |
| noam | "right square bracket" "_That_'s what that is!" | 13:43 |
| noam | fun times! | 13:43 |
| josch | oooh yes, i know the problem :) | 13:55 |
| pandora | Thats an interesting construct in the shell script :D | 14:13 |
| pandora | && exit 1 || : | 14:13 |
| josch | timeless classic :) | 15:24 |
| - cobra (QUIT: Ping timeout: 264 seconds) (~cobra@user/Cobra) | 15:35 | |
| + cobra (~cobra@user/Cobra) | 15:41 | |
| minute | CSS can make a big impact on gtk4 performance on imx8mplus https://mastodon.social/@mntmn/112660672927882029 | 16:17 |
| josch | huh? the scrolling, sure -- but even the mouse pointer becomes less responsive? i do not remember this happening from back when i had the imx8mq | 16:27 |
| josch | this is really bad XD | 16:28 |
| minute | yeah it happens when the gpu's fillrate is exhausted | 16:30 |
| minute | and some gui toolkits are really good at managing that | 16:30 |
| minute | the imx8mp's gpu is slower than the imx8mq's (less shader cores) | 16:30 |
| minute | but for example chromium and qt squeeze usable performance from it | 16:31 |
| minute | i think (without looking deeply) gtk does a lot of repaints when something small changes, and in a nested hierarchy of tons of elements this then all gets uploaded as textures or so | 16:31 |
| minute | so the gpu is kind of swamped for a moment | 16:32 |
| minute | also the cursor stutters because there are no overlay planes like even the amiga had in the 1980s (sprites) | 16:34 |
| minute | we certainly have regressed in the way we are using the resources of computers | 16:34 |
| josch | yeah, why optimize if our processors are so fast that we don't notice how inefficient our code is... | 16:35 |
| minute | a fun excercise is to set a custom css of * { border: 1px solid red; } in gtk inspector :D | 16:37 |
| josch | this discovery is so wild... | 16:38 |
| - ggoes (QUIT: Quit: ZNC - https://znc.in) (~gregf@fsf/staff/ggoes) | 17:36 | |
| + ggoes (~gregf@fsf/staff/ggoes) | 17:36 | |
| + jacobk (~quassel@2600:1700:9e1e:7800:c501:186:b0fc:52ef) | 17:42 | |
| + nybble (~sprang@user/csprng) | 18:33 | |
| svp | been reading up on the rcm4 docs, are the raspi and banana cm4s slightly different in terms of pinout? reading the online manual of both bundle options, the raspi manual mentions nvme yes mpcie no but the banana manual mentions nvme no mpcie yes. plus, is the sd/emmc quirk specific to the raspi as well? i was curious about trying a little more exotic modules too so i want to know what options i have available. | 18:58 |
| minute | svp: the pinout / adapter is the same | 19:25 |
| minute | svp: the banana adapter docs are more up to date, rpi version had very low sales. but will bring them up to date soon | 19:25 |
| minute | svp: the ssd situation is identical, both work with m.2 adapter in mpcie slot | 19:25 |
| minute | svp: bpi is more flexible in terms of emmc/sd, can do both in parallel. for rpi it's more tricky | 19:26 |
| minute | svp: rpi can do dual display (2x hdmi, one of them converted to edp), for bpi we use mipi to edp, and you can use hdmi externally but not both at the same time | 19:27 |
| minute | (the latter is a limitation of drivers) | 19:27 |
| svp | so the sd shares the emmc bus with both configurations but one is more lenient than the other? | 19:28 |
| minute | svp: no, on bpi emmc and sd are separate | 19:29 |
| minute | svp: on rpi they are shared | 19:29 |
| svp | ah! noted. | 19:29 |
| minute | i think rpi has a secondary sd card option but IIRC you need to move some resistors on our adapter for that | 19:29 |
| minute | but haven't looked into that in a while | 19:29 |
| svp | yeah i did find that passage written somewhere | 19:30 |
| svp | i think itd be pretty fun to try something exotic like those risc-v CM4-drop-in-replacements, but juuust in case i should probably grab one or two CM4-slotting "backplane" boards in case something has to be repurposed... | 19:40 |
| josch | svp: oh somebody is trying that already! :D | 19:44 |
| svp | oh! how're the reports looking so far? | 19:44 |
| josch | there hasn't been a status update for a while... let me check the log to see who that was... | 19:44 |
| josch | ah it was the milkv.io mars-cm, right? | 19:45 |
| svp | yep, that very one | 19:45 |
| josch | so far, the only one who was actually running riscv on the reform was Lukas via fpga :) | 19:48 |
| josch | svp: nope, sorry i cannot find it. maybe i misremember | 19:50 |
| - ggoes (QUIT: Quit: ZNC - https://znc.in) (~gregf@fsf/staff/ggoes) | 19:51 | |
| + ggoes (~gregf@fsf/staff/ggoes) | 19:54 | |
| svp | rip, i'll report back if i can get my hands on one | 19:57 |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 20:33 | |
| mtm | is the version 2 Reform motherboard available for purchase? Couldn't find it on the store | 20:34 |
| josch | mtm: it's the version 2.5 right now -- do you need version 2? | 20:47 |
| josch | and the 2.5 is out of stock right now: https://shop.mntre.com/products/mnt-reform-motherboard-25 | 20:47 |
| mtm | ah, yes, 2.5. I've got an RK3588 on order and was thinking about getting the newer motherboard with it. Then taking the old bits (original SoC, motherboard) and building a sort of wooden cyberdeck from them | 21:01 |
| - jacobk (QUIT: Ping timeout: 268 seconds) (~quassel@2600:1700:9e1e:7800:c501:186:b0fc:52ef) | 21:02 | |
| mtm | any word on a re-vamped camera? | 21:03 |
| minute | mtm: we def wanna do one but no resources at the moment | 21:10 |
| mtm | gotcha | 21:16 |
| mtm | will the rk3588 ship around the same time the new 2.5 motherboards get in? If it's reasonably close maybe I can add the motherboard to the order to save a little shipping | 21:18 |
| + megasquid (~megasquid@210.205.160.108.in-addr.arpa) | 21:48 | |
| - megasquid (QUIT: Changing host) (~megasquid@210.205.160.108.in-addr.arpa) | 21:55 | |
| + megasquid (~megasquid@user/megasquid) | 21:55 | |
| + ZylonMaster (~hjcs@syn-098-015-248-249.res.spectrum.com) | 22:25 | |
| sevan | hello :) | 23:29 |
| sevan | just a heads of in case it wasn't spotted but the apt repo advertises a kernel update which there's to package for, so when it comes to fetch during upgrade it returns 404 & fails | 23:30 |
| sevan | Err:1 https://mntre.com/reform-debian-repo reform/main arm64 linux-image-6.8.12-mnt-reform-arm64 arm64 6.8.12-1+reform20240605T104210Z 404 Not Found [IP: 91.250.115.15 443] | 23:32 |
| josch | sevan: that is intentional | 23:32 |
| sevan | Error: Failed to fetch https://mntre.com/reform-debian-repo/pool/main/l/linux/linux-image-6.8.12-mnt-reform-arm64_6.8.12-1%2breform20240605T104210Z_arm64.deb 404 Not Found [IP: 91.250.115.15 443] | 23:32 |
| sevan | josch: ah, no worries. | 23:32 |
| josch | sevan: https://mastodon.social/@mntmn/112653820899266095 | 23:33 |
| sevan | wait, I'm not one of those fancy Pocket Reform owners | 23:34 |
| sevan | wholesome Reform 2 here. | 23:34 |
| josch | sevan: if so, would you mind upgrading to the latest kernel and reporting whether this makes your reform fail to boot or not? | 23:34 |
| sevan | well, since its Saturday night, why not :) | 23:35 |
| josch | sevan: cool! do you want to try out the old 6.8 kernel or the new 6.9? | 23:35 |
| sevan | currently on 6.7.12 | 23:36 |
| sevan | so 6.8? | 23:36 |
| josch | sevan: okay, please download and install this: https://source.mnt.re/reform/reform-debian-packages/-/jobs/4584/artifacts/file/repo/pool/main/l/linux/linux-image-6.8.12-mnt-reform-arm64_6.8.12-1+reform20240605T104210Z_arm64.deb | 23:39 |
| sevan | ok | 23:40 |
| sevan | installing... probably going to be a while since it'll trigger a rebuild of the zfs modules via dkms :/ | 23:45 |
| josch | i'm going to be here for a while longer :) | 23:46 |
| sevan | ok :) | 23:46 |
| megasquid | About to try and desolder L10 and L11 on the motherboard. I tried once without flux and couldn't get the solder to melt. I've since picked up a desoldering braid and flux. I'll be setting my iron to 380 C. I realize a hot air reflow station would be best, but don't have one. Any advice before I give it a go? Planning on using the pointiest tip I have for the iron. | 23:50 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!