2025-10-11.log

joschhow should the image be named?00:00
joschreform-system-efi.img.xz?00:00
minutejosch: oh, would that be multiarch somehow?00:00
minutejosch: i thought there had to be arm64 and x64 ones00:00
joschyou want to create these for x86_64 as well?00:01
minutejosch: so, reform-system-arm64.img.xz and reform-system-x86_64.img.xz etc :D00:01
minuteah wait, efi should be in there00:01
joschdo you want to distribute artifacts for amd64 on source.mnt.re as well or is it enough to run ./mkimage.sh on a amd64 machine whenever an image for x86_64 is needed?00:02
minutereform-system-efi-arm64.img.xz00:02
+ Ar|stote|is (~linx@149.210.0.217)00:02
minutejosch: if not too hard, i'd like to distribute x64 as well sure00:02
minutejosch: unless that's somehow a maintenance burden for you in terms of the scripts00:02
joschi think it's more a problem of gitlab-ci.yml fiddling00:03
joschi'd create a second job for amd6400:03
joschand then the arm64 job would run on one runner and the amd64 job on the other00:03
joschboth create artifacts which then get distributed together00:03
minutejosch: ah yeah that sounds great 00:03
minutejosch: it would be cool if people could try out our image on a pc or in a vm, that's why i thought that was interesting (and for testing install stuff things on pcs in vms)00:04
minuteor if a tiny AMD som shows up one day...00:05
joschanother issue is the device tree00:09
+ colinsane (~colinunin@97-113-89-123.tukw.qwest.net)00:10
joschwith the systemd efi stub it can be embedded into the stub00:10
joschand it probabl needs to be but that would make the image platform specific00:10
joschthere is a project which builds the stub together witha  mechanism that auto-chooses the dtb on boot00:10
joschhere it is (with my comments): https://bugs.debian.org/111206300:11
minutejosch: funny, chris is the other engineer @ linaro who works with us on the quasar integration00:13
joschnice :)00:14
minuteinteresting > Stubble will by default override the firmware provided dtb with an embedded one if they share the same compatible00:14
minutethat sounds pretty good?00:14
minuteso, edk2 or uboot-efi in any case have a built-in dtb when deployed on a module, and it could be replaced based on the "compatible" matching, sure00:16
minute(when loading linux i mean)00:16
joschthe ITP is still open, if they need a DD to upload stuff, things can be arranged in case we need stubble for the reform00:24
minutejosch: i don't know these acronyms but i can forward a link to this chat log later if you think that's useful :D00:28
joschi don't know if stubble is needed for the reform00:29
joschmaybe edk2 or uboot-efi can ship a dtb that is sufficient for a device agnostic efi stub to boot linux? 00:29
joschlets start with rebasing the commits from that MR00:30
minutejosch: yeah, ok00:42
- jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu)00:59
wickedshelljosch: thanks for doing all the detailed testing of battery lifes. I meant to say something earlier, but I was out for the last week and a bit! It's a nicely detailed comparison, and I really like the assement of which module is better for what.01:01
wickedshellI'm still trying to sit and be happy with the imx8mp :P As long as I'm not asking it to do a large compile of something new it's more then fast enough for what I do on it. (Notably that doesn't involve a lot of web usage, as that seems to be flakey with both the built in or the extra module). But the wifi flakiness may just mean I need to look at rerouting my antenna cables01:02
+ LainIwakura (~LainIwaku@user/LainIwakura)01:04
+ TechnoWizard (~wiz@user/TechnoWizard)01:39
- TechnoWizard (QUIT: Ping timeout: 256 seconds) (~wiz@user/TechnoWizard)02:58
+ jogu (~jogu@user/jogu)03:02
+ swish (~swish@2600:1700:510:e610:86da:309c:ef17:962)03:12
- swish (QUIT: Client Quit) (~swish@2600:1700:510:e610:86da:309c:ef17:962)03:12
+ farrisswisher (~farrisswi@2600:1700:510:e610:86da:309c:ef17:962)03:12
farrisswisherhey folks, I'm looking into the possibility of getting a reform (or pocket) to replace my framework 16 preorder, was wondering if anyone has tried their hand at doing any audio production on the platform? I know using VST plugins etc would be a non-starter since those are typically built for x86 only, but I'm hopeful that open source plugins would03:15
farrisswisherat least work fine if built from source?03:15
- jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu)03:22
- farrisswisher (QUIT: Quit: Client closed) (~farrisswi@2600:1700:510:e610:86da:309c:ef17:962)03:25
- nsc (QUIT: Ping timeout: 264 seconds) (~nicolas@i5C74DEA3.versanet.de)03:34
+ nsc (~nicolas@i5C74DF4C.versanet.de)03:36
- paperManu (QUIT: Ping timeout: 240 seconds) (~paperManu@64.58.44.160)03:42
amatechawould a 2230 m.2 fit in the reform next (2242)?03:43
amatechaI'm guessing not but I know very little about m.2 form factor lol03:44
- jacqueline (QUIT: Ping timeout: 246 seconds) (~jacquelin@user/jacqueline)04:14
- mjw (QUIT: Ping timeout: 240 seconds) (~mjw@gnu.wildebeest.org)04:18
+ jacqueline (~jacquelin@user/jacqueline)04:23
- GNUmoon2 (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon)04:38
+ GNUmoon2 (~GNUmoon@gateway/tor-sasl/gnumoon)04:39
- GNUmoon2 (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon)04:46
- Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149.210.0.217)05:12
+ Ar|stote|is (~linx@149.210.0.217)05:12
+ emdevt (~emdevt@2001:7e8:fa02:9801:87d0:3a5f:5000:3a4d)06:12
joschamatecha: there are extension adapters which allow you to install shorter cards into slots meant for longer ones07:36
joschwickedshell: i'm still typing this on my a311d -- doing this assessment also had a very selfish interest that i want to convince myself that switching to rk3588 is actually okay :D07:37
amatechaoh interesting josch that's great to know, thanks07:45
+ GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon)07:57
joschminute: i found out what the dtb symlink is good for: it's used by old u-boot versions on imx8mq and imx8m+ which did not set ${fdtfile} in their env -- if u-boot doesn't know the dtb it falls back to /boot/dtb08:42
+ rwa_ (~rene@2001:9e8:3387:2700:af2e:608b:aa59:bfb1)08:59
- emdevt (QUIT: Remote host closed the connection) (~emdevt@2001:7e8:fa02:9801:87d0:3a5f:5000:3a4d)09:04
rwa_Warzone 2100 running fine on the RK3588 via flatpak :)09:29
+ jogu (~jogu@user/jogu)09:50
+ gustav2 (~gustav@c-78-82-52-97.bbcust.telenor.se)10:02
- jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu)10:09
+ BoostisBetter (4a410829d7@irc.cheogram.com)10:28
- LainIwakura (QUIT: Quit: Client closed) (~LainIwaku@user/LainIwakura)10:42
Ar|stote|isThe reform next uses 2242 because of space constraints? Are there extenders to mount a 2280 elsewhere?11:49
- rwa_ (QUIT: Quit: Leaving) (~rene@2001:9e8:3387:2700:af2e:608b:aa59:bfb1)12:06
minuteactually i've changed to 2230 as the default + 2242 option, because 2230 is more popular/affordable now12:14
+ emdevt (~emdevt@2001:7e8:fa02:9801:87d0:3a5f:5000:3a4d)12:46
Ar|stote|isLots of gaming handhelds bigger supply!13:27
Ar|stote|isAI has increased the price of nvme ssds though. The new gpu era is upon us.13:28
minutefor all those datacenters?13:31
minutebut yeah, steam deck etc have helped with making small ssds popular13:31
minutei'm gonna do another reform next mb spin and am torn if i keep microsd express slot (internal, on motherboard top side) or try to change it to another 2230 m.2 slot. not sure how much of a squeeze it will be with trackpad on top though.13:32
Ar|stote|ishttps://www.tweaktown.com/news/108074/phison-ceo-says-severe-nand-flash-shortages-and-unprecedented-ssd-demand-could-last-10-years/index.html13:33
Ar|stote|isMicrosd express? The new one the switch 2 has?13:34
+ paperManu (~paperManu@64.58.44.160)13:45
minuteAr|stote|is: yeah kind of a coincidence that they decided to go with that, too13:59
minutethe cards are still quite expensive though14:00
Ar|stote|isYeah but it will be the future most probably14:05
Ar|stote|isEgwn though internal doesn't seem that useful imho14:05
Ar|stote|isEven*14:05
Ar|stote|isDo we have any news regarding that Qualcomm som? When will it be available and more specifics?14:18
minuteAr|stote|is: https://social.treehouse.systems/@cas/11535180962440884514:21
minutevideo https://mastodon.social/@mntmn/11535161342558721614:22
minuteand a bit more ramblings from today https://mastodon.social/@mntmn/11535532713986290614:23
- cow321 (QUIT: Ping timeout: 244 seconds) (~deflated8@user/meow/deflated8837)15:25
cwebberhello hello15:27
cwebberoh cool15:27
+ cow321 (~deflated8@user/meow/deflated8837)15:29
cwebbermy pocket has stopped charging with even the more powerful chargers. Gonna order https://www.adafruit.com/product/1905 and see if it helps15:35
cwebberI did leave it off for a few months15:35
cwebberbut it's not powering at all now15:35
cwebberI did try flashing the firmware earlier but didn't succeed, the script had aborted, and I didn't continue trying to figure out what the deal was (it said the firmware hash mismatched what it expected iirc) so I simply haven't dealt with it15:36
cwebberI've had to replace 3 spicy pillows (!!!) on my System76 laptop though15:37
cwebberso I'm getting sick of that thing15:37
cwebberwant to use the pocket more15:37
cwebberI guess my whole "Well I want a device that can suspend"... given that my laptop literally can't run on batteries at all with its third spicy pillow removed, I'm a little more willing to accept the current state of power management on the RK358815:40
- emdevt (QUIT: Ping timeout: 252 seconds) (~emdevt@2001:7e8:fa02:9801:87d0:3a5f:5000:3a4d)17:00
+ thelounge8040 (~thelounge@gidzit.org)17:01
- thelounge8040 (QUIT: Client Quit) (~thelounge@gidzit.org)17:04
+ paperManu_ (~paperManu@142.169.16.202)17:07
joschcwebber: do you need assistance?17:22
joschyou could try getting the battery board out of the loop by bridging pins 2&3 of the motherboard battery connector17:22
- paperManu_ (QUIT: Ping timeout: 240 seconds) (~paperManu@142.169.16.202)17:52
+ mjw (~mjw@gnu.wildebeest.org)17:56
minutecwebber: here is the troubleshooting process https://support.mnt.re/#devices/pocket_reform/charger_module_test18:20
minutejosch: about fat32 /boot, and /dtb, why does that need to be a symlink and cannot be simply a copy? i really don't understand the benefits of symlinks there. if anything, i find them confusing in practice18:20
+ johl (~johl@dynamic-093-128-030-172.93.128.pool.telefonica.de)18:54
- johl (QUIT: Quit: leaving) (~johl@dynamic-093-128-030-172.93.128.pool.telefonica.de)18:56
+ rwa_ (~rene@2001:9e8:3387:2700:af2e:608b:aa59:bfb1)18:59
joschminute: oh it absolutely can be a copy, would work the same way19:03
joschmaybe flash-kernel would have to be adjusted to support this but it can probably be done19:03
joschthis remark was not meant as an argument against converting /boot from ext4 to fat3219:04
joschmy reasons against converting /boot from ext4 to fat32 are much less tangible and rather are like: fat32 only has short filenames, short labels, no journal, is not 2038-safe and so on19:05
joschin essence all arguments that boil down to: fat32 is an outdated filesystem from the last century lacking all the nice things that we learned to like from semi-modern filesystems19:05
joschmy other concern would be: i have never seen a setup which added the efi partiton into /boot in practice. All efi setups that i've come across put the efi partition on its own small partition. But i don't know if that has a technical reason or not.19:06
joschfat32: it's case-insensitive... it's just all so jucky, no? do i make any sense? :D19:09
rwa_sure, fat32 is for sd-cards and usb-sticks, but even there nowadays exFAT is the way to go19:11
rwa_fat32 is not only century, is even from the last millenium... :D19:11
rwa_+last19:11
joschLike I wrote here https://floss.social/@josch/115356137255845924 I think I'd really like the input of somebody who knows enough about EFI. "abusing" /boot as the efi partition is technically possible, but is it a good idea? I'd like to avoid creating a franken-setup :)19:17
joschin any case, the answer to this question is not blocking my work right now19:18
joschthere is a lot in above MR which I'm moving over to main bit by bit19:18
rwa_franken-boot :D19:19
minutejosch: i see! it's a pity that EFI has this design flaw19:34
minutethey should really update to a modern fs19:34
f_hiya, does anybody know if the MNT reform could support 2 ext. displays at the same time?19:38
minutef_: yes, rk3588-dsi reform (or pocket) can, but needs a flex cable and hdmi socket somewhere :D19:44
f_\o/19:45
minutewe have one extra 20pin fpc for hdmi on rcore.19:45
rwa_:o19:45
minutewhich could also be displayport with a different dts19:45
f_I'm considering mnt reform as my next laptop to replace my current hp laptop from 2011 haha19:46
f_Displayport would be perfect19:46
minutei need to validate this displayport stuff next week, already soldered the breakout19:46
f_I use one displayport-first monitor along with a VGA-only monitor19:46
f_minute: that's really cool :D19:47
f_Reform definitely goes on my list of next laptop to use then19:48
minuteon sm8550 i think we can also have 3 displays, but that module will still take a while :D19:48
f_heh, I'm not looking for lots of performance :P19:48
minutehaha ok :D rk should be more to your taste as it has only 2 blobs and no hypervisor etc19:49
f_but if sm8550 is available when my current laptop dies then that'd be even cooler :D19:49
f_yeah RK seems like my taste19:49
minutei hope they won't make rk3688 more blobby...19:50
f_I originally thought a311d would be cool and even cooler assuming I finish reverse-engineering A311D bl2 by then19:50
f_but apparently can't do 2 ext. displays19:50
minuteno, and sadly also we never got dsi+hdmi to work at the same time too, even if it was said to be technically possible19:51
minutef_: did something ever happen with a311d2 upstreaming btw?19:51
f_dsi+hdmi wouldn't matter all that much, my current HP laptop already has trouble handling my 2 external displays + the internal eDP one19:51
minute(just curious)19:51
f_minute: Haven't checked, sorry19:51
minuteok19:52
f_all I know is amlogic is trying to upstream some drivers for some increasingly new SoC and being gaslighted on mailing lists for doing it wrong19:52
f_Whatever S4/T7 means...19:52
minuteaw19:52
minutewhat are the specs?19:53
f_ /A519:53
f_I don't know what these codenames refer to, I'll have to check19:53
minuteah ok19:53
f_their codenames make less and less sense with every new product line19:53
f_A5 is apparently Amlogic S905X5M19:54
minutei found the low energy use of a311d quite impressive but 4gb ram limit was a hard sell for many19:54
minuteand the fact that internal+external disp. couldn't be used together 19:54
f_^ wouldn't be a downgrade for me, I already can't use the internal disp with 2 ext. monitors lol .. nvidia shenanigans19:55
f_not only that but it barely handles my 1440p ultrawide monitor19:55
f_just barely19:55
minuteaw... i'm using 1440p external monitor with my rk3588 pocket reform at home, it's very nice19:56
f_It's a hp laptop from 2011 with nvidia fermi graphics, what did you expect :P19:56
f_Fermi support in nouveau seems to mostly be in a state similar to that slice of pizza I'd leave in the fridge for 2 months19:57
minuteheh. but you get gl 4.x?19:57
f_barely19:57
f_anything that dares try to use the GPU shows up garbled19:58
f_amlogic-s4 apparently is Amlogic S905Y419:58
f_All I remember about these newer SoCs is they tripled the number of firmware blobs required to boot19:59
f_AFAIk they're also not really supported by U-Boot yet19:59
rwa_i don't know much about it, but all this boot stuff seems to be quite a mess on all those ARM-based SBC20:01
f_depends which ;)20:01
f_The reason they're a huge mess on the newer aml SoCs for me is probably because I haven't looked much at them20:01
rwa_well...all? :D20:01
f_and you haven't seen x86_64 :D20:02
rwa_yeah, sure, i barely get in touch with the low level stuff20:02
f_ACTION wants to get off that boat20:02
- SadOp (QUIT: Quit: Konversation terminated!) (~jan@81.209.189.43)20:05
f_At least now I know what to get next when my laptop finally dies20:07
f_just need to know when it's planning to die now20:08
joschthe framework desaster is letting a bunch of really nice list of laptop alternatives surface, for example this one: https://exquisite.social/@h3artbl33d/11535641140001327020:14
joschsome of theme were news to me and buying used is great advice in general :)20:15
rwa_i guess most of them are "designed in Europe, made in China" ... i was quite interested in StarLabs back then, but besides their Linux focus i have no clue about "repairability"20:18
rwa_to me it pretty much looks like "pay a premium for a generic laptop because its labeled by a small company"20:20
rwa_is my brain right that the trackpad for the classic Reform has been discontinueed?20:22
minutejosch: this sounds like there's no uboot on emmc https://community.mnt.re/t/reform-setup-encrypted-disk-fails/3911/2120:32
minutejosch: so uboot on sd card boots the old system20:32
minutejosch: but that's weird indeed, as it should prioritize /boot on sdcard, right?20:33
minutejosch: or, the sd card was already modified by the setup/migrate script, thus now preferring to boot the old system on emmc?20:33
minutejosch: i'm wondering if they shouldn't just use reform-flash-rescue emmc and call it a day...20:34
chorcrwa_: yeah, most of those are clevo/tongfang chassis, nothing wrong with that, but they are hardly designed in Europe; for me nowadays it's either MNT or some refurb corpo laptop, but I'm priviledged to do most of my computing on a workstation anyway20:42
f_josch: what is the framework disaster?20:43
rwa_f_: https://crimier.github.io/posts/Framework-Omarchy/20:46
rwa_chorc: so not even "designed in Europe"...thats worse than i expected :(20:47
f_oh the omarchy junk20:47
f_I was kinda considering Framework too but not really too much given it's x86_6420:47
minutethere's the powerpc laptop from france but it is progressing slowly afaik20:47
f_and also windows-first20:47
minuteyeah20:47
rwa_chorc: i have a "gaming rig", but my "work" can easily be done on the Reform...thats wy i'd like to get my hands on a "Desktop Reform" :D20:47
minutei also have a gaming rig... ps5 lol20:48
f_my gaming rig is my nintendo wii20:48
rwa_i managed to keep my hands of those consoles (XBox/PS), but i'm in the steam hell...20:48
f_I love playing games for children as an adult20:48
rwa_atleast i can install any linux distribution on Legion Go20:49
f_I can boot debian on my wii20:50
f_and play intensive AAA games such as supertux220:50
chorcrwa_: I'm on the other side, I need quite a bit of umpf for what I do, and I've got it solved either by employee or just generic PC made from gaming parts :-)20:51
rwa_fair enough :)20:51
chorcminute: powerpc, really?20:52
rwa_what a hell of a FrankenPad :D https://www.tpart.net/product/x210ai-completed-laptop-ultra-7-165h/20:53
chorcback in early 2000s I had access to SparcBook, it was something, but it wasn't made for fun20:54
rwa_oh, trixie backports ship mesa 25.2.4? maybe it's time to move my "gaming rig" to Debian as well *thinking*20:56
- rwa_ (QUIT: Quit: Leaving) (~rene@2001:9e8:3387:2700:af2e:608b:aa59:bfb1)21:01
minutesebastian reichel figured out that suspend issue caused by emmc on rk3588 can be fixed by removing "supports-cqe" flag from device tree https://gitlab.collabora.com/hardware-enablement/rockchip-3588/linux/-/issues/43#note_24823921:20
+ TechnoWizard (~TechnoWiz@user/TechnoWizard)21:23
minutehm, the following works for me with mmc drivers unloaded: cd /sys/power; echo test_resume > disk; echo devices > pm_test; echo disk > state21:27
+ paperManu_ (~paperManu@64.58.44.160)21:54
- TechnoWizard (QUIT: Ping timeout: 250 seconds) (~TechnoWiz@user/TechnoWizard)22:00
minutebut "processors" test failed22:10
- RandyK (QUIT: Ping timeout: 272 seconds) (~RandyK@user/randyk)22:10
+ RandyK (~RandyK@user/randyk)22:10
- gustav2 (QUIT: Quit: Quit) (~gustav@c-78-82-52-97.bbcust.telenor.se)22:15
+ TechnoWizard (~TechnoWiz@user/TechnoWizard)22:24
- TechnoWizard (QUIT: Remote host closed the connection) (~TechnoWiz@user/TechnoWizard)22:24
- RandyK (QUIT: Remote host closed the connection) (~RandyK@user/randyk)22:27
+ RandyK (~RandyK@user/randyk)22:27
+ TechnoWizard (~TechnoWiz@user/TechnoWizard)22:48
- ezequielg (QUIT: Server closed connection) (sid363064@id-363064.uxbridge.irccloud.com)23:00
+ ezequielg (sid363064@id-363064.uxbridge.irccloud.com)23:00
joschminute: yes, i too suspected that there was no uboot on emmc which is why i asked for a dump of the first 16MBit of emmc to confirm this theory. I thought you are shipping the units with u-boot flashed to emmc on rk3588 do you not?23:16
joschi didn't just suggest to try "reform-flash-uboot emmc" because then we would never find out what the original issue was and that means we cannot fix it for the next person23:17
joschfor example, reform-setup-encrypted-disk probably should warn if it cannot find latest u-boot on emmc to prevent situations such as these, no?23:19
- TechnoWizard (QUIT: Ping timeout: 256 seconds) (~TechnoWiz@user/TechnoWizard)23:29
manisminute: the PowerPC notebook is designed in Italy, if you meant that.23:39
minutejosch: yes of course we ship uboot on emmc. they probably overwrote it23:47
joschyes, likely23:47
joschi want to put in some checks to guard against these situations23:47
minutejosch: and yes, good point @ checking for bootability, that would be quite useful23:47
joschminute: i must admit now that it's really cool to have serial via usb-c integrated into motherboard 3.0. So often can we not get any early-boot output because the user doesn't happen to have a uart adapter at home. The odd thing here was just that SW4 had all switches off which had to be fixed first.23:52
minutejosch: right. i forgot to put a step in the shipping/qa procedure to switch these switches on. OTOH it's more secure like that.23:53
minutei.e. an adversarial power supply can't just mess with your serial port then ^^23:54
joschindeed23:56
joschespecially tricky once serial login works on rk3588 again :)23:56
+ TechnoWizard (~TechnoWiz@user/TechnoWizard)23:57

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!