| - bkeys (QUIT: Ping timeout: 264 seconds) (~Thunderbi@45.134.140.153) | 00:29 | |
| - mjw (QUIT: Ping timeout: 248 seconds) (~mjw@gnu.wildebeest.org) | 00:31 | |
| + bkeys (~Thunderbi@45.134.140.153) | 00:58 | |
| - bkeys (QUIT: Read error: Connection reset by peer) (~Thunderbi@45.134.140.153) | 01:00 | |
| + bkeys (~Thunderbi@45.134.140.153) | 01:01 | |
| - bkeys (QUIT: Client Quit) (~Thunderbi@45.134.140.153) | 01:04 | |
| + bkeys (~Thunderbi@45.134.140.153) | 01:05 | |
| - bkeys (QUIT: Client Quit) (~Thunderbi@45.134.140.153) | 01:05 | |
| + bkeys (~Thunderbi@45.134.140.153) | 01:05 | |
| + piroko (~piroko@104.225.216.16) | 01:28 | |
| + jacobk (~quassel@47-186-105-237.dlls.tx.frontiernet.net) | 03:03 | |
| - bluerise (QUIT: Ping timeout: 255 seconds) (~bluerise@p5b0ac2ab.dip0.t-ipconnect.de) | 03:13 | |
| - kop316 (QUIT: Remote host closed the connection) (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 04:02 | |
| + kop316 (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 04:03 | |
| - kop316 (QUIT: Remote host closed the connection) (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 04:04 | |
| + kop316 (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 04:04 | |
| + bluerise (~bluerise@p5b2116b7.dip0.t-ipconnect.de) | 04:24 | |
| + mrdaught (~mrdaught@bras-base-toroon01zb3-grc-78-184-147-249-58.dsl.bell.ca) | 05:35 | |
| - mrdaught (QUIT: Read error: Connection reset by peer) (~mrdaught@bras-base-toroon01zb3-grc-78-184-147-249-58.dsl.bell.ca) | 05:38 | |
| + mrdaught (~mrdaught@bras-base-toroon01zb3-grc-78-184-147-249-58.dsl.bell.ca) | 05:39 | |
| - jacobk (QUIT: Quit: No Ping reply in 180 seconds.) (~quassel@47-186-105-237.dlls.tx.frontiernet.net) | 06:04 | |
| - aloo_shu (QUIT: Ping timeout: 272 seconds) (~aloo_shu@85.51.17.53) | 06:04 | |
| + jacobk (~quassel@47-186-105-237.dlls.tx.frontiernet.net) | 06:10 | |
| + aloo_shu (~aloo_shu@85.51.17.53) | 06:13 | |
| - mrdaught (QUIT: Remote host closed the connection) (~mrdaught@bras-base-toroon01zb3-grc-78-184-147-249-58.dsl.bell.ca) | 06:16 | |
| + mrdaught (~mrdaught@bras-base-toroon01zb3-grc-78-184-147-249-58.dsl.bell.ca) | 06:17 | |
| + mundane-1emantic (~user@static-198-54-134-176.cust.tzulo.com) | 06:20 | |
| mundane-1emantic | Is there any other special steps to flash the pocket sdcard images? When I flash them, they don't seem to show up as having partitions or filesystems (in the past the regular reform images do show this) | 06:21 |
|---|---|---|
| mundane-1emantic | I'm just following the "README.flashing" instructions for now | 06:22 |
| voltagex | the battery switch is really hard to hit :( | 06:31 |
| - murphnj (QUIT: Ping timeout: 248 seconds) (~murph@user/murphnj) | 07:08 | |
| + murphnj (~murph@user/murphnj) | 07:22 | |
| - aloo_shu (QUIT: Ping timeout: 252 seconds) (~aloo_shu@85.51.17.53) | 07:34 | |
| + aloo_shu (~aloo_shu@90.166.193.235) | 07:37 | |
| - mundane-1emantic (QUIT: Ping timeout: 252 seconds) (~user@static-198-54-134-176.cust.tzulo.com) | 08:25 | |
| + mjw (~mjw@gnu.wildebeest.org) | 09:16 | |
| josch | whoops, we didn't re-generate the system images since the kernel fix and thus the pocket images were still without partition table due to u-boot on them | 09:22 |
| josch | this is corrected now | 09:22 |
| josch | thanks to the reduced reform-handbook size, images are now 1.15 GB down from 1.3 GB | 09:22 |
| josch | on reform.d.n our images are down to 813 MB from 915 MB | 09:23 |
| minute | josch: great, thank you! | 09:49 |
| - jacobk (QUIT: Ping timeout: 260 seconds) (~quassel@47-186-105-237.dlls.tx.frontiernet.net) | 11:04 | |
| + jacobk (~quassel@47-186-105-237.dlls.tx.frontiernet.net) | 11:04 | |
| - _alice (QUIT: Read error: Connection reset by peer) (sid544964@ilkley.irccloud.com) | 11:10 | |
| + _alice (sid544964@id-544964.ilkley.irccloud.com) | 11:10 | |
| amospalla | josch: I included the u-boot messages with the printenv on the forums. Forgot that part hihihi. | 12:22 |
| grimmware | Does anyone have a link for the ugreen PD charger which is known to charge the pocket *after* it's powered on? | 12:32 |
| grimmware | None of my chargers (steam deck, apple, Anker) will actually charge the thing after it's up | 12:33 |
| josch | amospalla: thank you! | 12:43 |
| josch | minute: according to the u-boot output by amospalla, u-boot on the pocket reform does not set the ${fdtfile} environment variable. But having that variable set is crucial for images that include support for multiple boards as that is the variable which communicates which device tree to load | 12:47 |
| josch | minute: and for the pocket, it should be fdtfile=freescale/imx8mp-mnt-pocket-reform.dtb | 12:47 |
| amospalla | I do not know what you are talking about, but sounds very good ^^ | 12:48 |
| josch | minute: the recent reform-tools release made u-boot on emmc non-updatable but these kind of issues are reasons for why we *do* want to support users updating their u-boot | 12:48 |
| josch | amospalla: nah, it's rather negative news, unfortunately :/ | 12:48 |
| amospalla | s | 12:48 |
| amospalla | oh :( | 12:49 |
| josch | it's a bit of a pain because it means users have to manually update their u-boot and then you cannot rely on users having done that and if they do that, it involes a bit of a risk of them accidentally soft-bricking their device if something should go wrong | 12:49 |
| amospalla | thank you for the explanation | 12:51 |
| josch | minute: this is documented in u-boot sources as doc/develop/distro.rst where ${fdtfile} is marked as "Mandatory" in the section "Required Environment Variables" | 12:56 |
| - mjw (QUIT: Ping timeout: 252 seconds) (~mjw@gnu.wildebeest.org) | 13:12 | |
| * Guest3131 -> mjw | 13:20 | |
| - iank (QUIT: Quit: ZNC 1.8.2+deb2+deb11u1 - https://znc.in) (~iank@fsf/staff/iank) | 13:25 | |
| + iank (~iank@fsf/staff/iank) | 13:27 | |
| - mrdaught (QUIT: Remote host closed the connection) (~mrdaught@bras-base-toroon01zb3-grc-78-184-147-249-58.dsl.bell.ca) | 13:31 | |
| + mrdaught (~mrdaught@184.147.249.58) | 13:31 | |
| josch | bluerise: as somebody more familiar with u-boot than i, can you confirm my findings related to ${fdtfile} above? Should that variable get set by the board file? | 14:01 |
| minute | josch: ah, i see. | 14:12 |
| bluerise | depends on the U-Boot version I think. with mainline I'd assume that the Kconfig or defconfig for the board selects it | 14:13 |
| bluerise | https://github.com/u-boot/u-boot/blob/master/configs/imx8mq_reform2_defconfig#L14 | 14:13 |
| josch | bluerise: right but you agree that it should be set, and that not setting it is not optimal? | 14:15 |
| josch | thank you for your input :) | 14:15 |
| bluerise | Honestly I don't know anymore because mainline has changed to some "standard boot" thingy instead of distroboot | 14:16 |
| bluerise | with distroboot, I think fdtfile is set automagically | 14:16 |
| josch | minute: this also answers the question: "can there be a single u-boot for all rk3588 platforms" -- the answer is "no" if different device trees need to be passed to linux | 14:16 |
| bluerise | some u-boot boards have code that selects different FDT files | 14:16 |
| josch | bluerise: using ${soc}-${board}.dtb, right? | 14:17 |
| bluerise | last time I saw that it was C code that updated the... I guess ${fdtfile} | 14:17 |
| bluerise | variable | 14:17 |
| bluerise | ok let me regroup | 14:18 |
| bluerise | the easiest case is you define a board that has a single DTB | 14:18 |
| bluerise | and then it depends on what you need to achieve. enable/disable overlays/devices? | 14:18 |
| bluerise | https://github.com/u-boot/u-boot/blob/master/board/solidrun/clearfog/clearfog.c#L273 | 14:19 |
| bluerise | this board for instance is basically the "same" with minor changes here and there, hence opting for updating the fdtfile | 14:20 |
| josch | thank you, seems we should find a way to do something similar for pocket and rk3588 u-boot | 14:22 |
| minute | hunch: maybe the cheapest way is just to create a big amount of images, the extra storage cost is still less than the engineering time and handling the complexity of "flexible images" | 14:29 |
| minute | but maybe in the future the size of the individual images can be shrunk still | 14:29 |
| minute | i.e. making more stuff optional/net installed | 14:30 |
| minute | i'm now downloading the latest image build for rk3588 and trying it on next :3 | 14:30 |
| [tj] | freebsd has a generic image for armv7 which you have to add your own u-boot to | 14:32 |
| [tj] | it is a mixed bag, good for the project but a terrible user experience | 14:32 |
| Zaba | well, in the hypothetical bright bootstd future, a sufficiently modern and fully-featured u-boot would be able to boot Linux from extlinux.conf or any spiritual equivalent to that (i.e., firmware-agnostic configuration), and so you’d just have u-boot programmed into an SPI flash or something and boot from images that would only have standard file systems and no boot firmware on them | 14:42 |
| Zaba | I’ve got one rk3568 SBC set up like that, it’s pretty nice | 14:43 |
| - mrdaught (QUIT: Read error: Connection reset by peer) (~mrdaught@184.147.249.58) | 14:45 | |
| minute | i'm using custom extlinux.confs on most of my reform devices for debugging etc | 14:52 |
| minute | maybe we could migrate imx8mq users to it at some point, if well planned | 14:53 |
| minute | now working on reform next battery firmware :0 | 14:54 |
| minute | btw, rk3588 image freezes on boot of course, because of the removed hdmi1 support in the kernel... still sth for my todo list to forward port that | 14:55 |
| minute | josch: my system didn't boot the first time, and the second time, it didn't do the rootfs resizing. can i ask the system to do that now? | 15:05 |
| minute | also, hostname choice was somehow not applied | 15:06 |
| minute | so now everything is meeeega delayed because it always wants to look up "reform" hostname that doesn't exist | 15:07 |
| minute | like, sudo is super slow | 15:07 |
| minute | this used to work though. does the debian-packaged reform-setup-wizard set the hostname at exit, josch? | 15:08 |
| josch | minute: yes, but where are we using the debian-packaged reform-setup-wizard? Moving to the packaged one is this MR https://source.mnt.re/reform/reform-system-image/-/merge_requests/97 | 15:35 |
| josch | minute: it has been 3 months since i opened that MR, so i have to test things again | 15:35 |
| josch | minute: but unless you are trying out that MR, the images should not contain the debian-packaged reform-setup-wizard | 15:36 |
| minute | josch: ah ok! sorry, i thought maybe the change in behavior came from that :3 | 15:36 |
| minute | josch: ok, thanks for clarifying, then it's a new mystery | 15:36 |
| josch | minute: extlinux.conf on imx8mq is not only tricky because of potentially old u-boot but also because of: how do you switch dtbs without using boot.scr? | 15:36 |
| minute | josch: i specify the dtb in extlinux.conf | 15:37 |
| josch | manually? | 15:37 |
| josch | then that will be overridden the next time u-boot-menu runs | 15:37 |
| josch | and u-boot-menu runs on every kernel upgrade to write the kernel versions into the extlinux.conf | 15:37 |
| minute | josch: so it can't write the FDT= parameter? then it needs to be fixed maybe | 15:38 |
| minute | sorry, not FDT=, just FDT | 15:38 |
| minute | also, there's a symlink "dtb" in /boot but i guess noone uses that | 15:39 |
| minute | what i mean is, one could just let uboot load /dtb and replace that symlink if needed | 15:40 |
| josch | minute: but that symlink is kernel version specific and thus also changed on every kernel update | 15:40 |
| minute | ok, not sure what dtb switching is needed then | 15:41 |
| josch | because the dtb path is kernel version specific | 15:41 |
| minute | is it about using uboot menu with multiple kernels and a dtb for each kernel? | 15:41 |
| minute | then uboot menu must be enhanced to couple FDT entries to kernel entries | 15:42 |
| josch | i mean the /boot/dtb link currently points to (for me) dtbs/6.9.10-mnt-reform-arm64/amlogic/meson-g12b-bananapi-cm4-mnt-reform2.dtb | 15:43 |
| josch | in there is the kernel version, so when a new kernel is installed something has to know for which version to put what | 15:44 |
| + mrdaught (~mrdaught@172.58.139.90) | 15:44 | |
| josch | that something is currently flash-kernel which can read /etc/flash-kernel/machine | 15:44 |
| josch | to my knowledge, that mechanism does not exist for u-boot-menu | 15:44 |
| - mrdaught (QUIT: Read error: Connection reset by peer) (~mrdaught@172.58.139.90) | 15:45 | |
| - sir-photch (QUIT: Ping timeout: 252 seconds) (~m-hy5poy@static.93.70.235.167.clients.your-server.de) | 15:46 | |
| - bkeys (QUIT: Quit: bkeys) (~Thunderbi@45.134.140.153) | 15:46 | |
| + bkeys (~Thunderbi@45.134.140.153) | 15:46 | |
| + sir-photch (~m-hy5poy@static.93.70.235.167.clients.your-server.de) | 15:50 | |
| minute | josch: can this mechanism be added to u-boot-menu? | 15:55 |
| minute | josch: so it would write the FDT entries into extlinux.conf | 15:55 |
| minute | oh interesting, i just noted that rk3588 is actually using our uboot from microsd card | 15:56 |
| minute | the imx8mq feeling... | 15:56 |
| minute | a small thing to fix > Model: Radxa ROCK 5 Model B | 15:56 |
| josch | minute: i (probably) stand corrected. The fdt switching could be implemented by writing the desired dtbname as variable U_BOOT_FDT into /etc/u-boot-menu/conf.d/reform.conf | 15:58 |
| minute | aha aha | 16:07 |
| josch | now we need somebody with an imx8mq to test :) | 16:58 |
| josch | kinda regretting not having one at home anymore... XD | 16:59 |
| minute | we have a tray of weird ones, can check it one day :D | 17:01 |
| josch | yeah, like my very first one which had broken serial :) | 17:02 |
| minute | yeah that's my go-to one for testing!! | 17:02 |
| josch | it was otherwise working fine -- broken serial was just very bad in the super early days when we were still mucking around with kernel, initramfs etc | 17:03 |
| + bkeys1 (~Thunderbi@45.134.140.153) | 17:03 | |
| minute | yeah | 17:03 |
| - bkeys (QUIT: Ping timeout: 252 seconds) (~Thunderbi@45.134.140.153) | 17:05 | |
| * bkeys1 -> bkeys | 17:05 | |
| - bkeys (QUIT: Remote host closed the connection) (~Thunderbi@45.134.140.153) | 17:14 | |
| - hairu (QUIT: Remote host closed the connection) (m-uotkmd@user/hairu) | 17:15 | |
| + hairu (m-uotkmd@user/hairu) | 17:16 | |
| + bkeys (~Thunderbi@45.134.140.153) | 17:16 | |
| josch | aha, i think i figured it out by staring at u-boot-menu code. The magic content for /etc/u-boot-menu/conf.d/reform.conf in case of a311d is: | 17:23 |
| josch | U_BOOT_FDT_DIR=/dtbs/ | 17:23 |
| josch | U_BOOT_FDT=/amlogic/meson-g12b-bananapi-cm4-mnt-reform2.dtb | 17:23 |
| josch | slashes *are* important | 17:23 |
| josch | i'll read some irc backlog to figure out whether i'm missing something... | 17:24 |
| josch | hrm.... | 17:31 |
| josch | flash-kernel is still required to copy the correct dtb into /boot | 17:31 |
| josch | we could also copy a snippet into /etc/kernel/postinst.d/zz-sync-dtb which always copies *all* dtbs from /usr/lib/linux-image-* into /boot | 17:32 |
| josch | that would be around 50 MB per kernel version though... | 17:32 |
| josch | we could also only select those dtbs that are reform related, either by using a curated list or by matching against the dtb name | 17:33 |
| josch | that could then have the advantage that /boot could become device agnostic as well | 17:33 |
| josch | but that is not something that we need extlinux.conf for, this also works with flash-kernel | 17:33 |
| josch | the prerequisite is, that /boot has all the relevant dtbs and that u-boot sets ${fdtfile} to the correct value | 17:34 |
| josch | which is not the case for the pocket (yet) | 17:34 |
| josch | vagrantc proposed to throw out flash-kernel in 2022 | 17:34 |
| josch | aha, past me was clever and left a lot of notes in /usr/share/u-boot-menu/conf.d/reform.conf | 17:41 |
| josch | that was in the beginning of May... | 17:42 |
| josch | gosh... sometimes my memory really fails me... i have no recollection of having written this... | 17:43 |
| + mrdaught (~mrdaught@172.58.139.90) | 17:44 | |
| - mrdaught (QUIT: Read error: Connection reset by peer) (~mrdaught@172.58.139.90) | 17:45 | |
| - bkeys (QUIT: Ping timeout: 258 seconds) (~Thunderbi@45.134.140.153) | 17:56 | |
| ^alex | awoo | 17:58 |
| minute | josch: haha i have the same issue... i guess normal that one can't keep the details of all these hundreds of subprojects in the brain | 18:08 |
| josch | minute: this is very releaving to hear, honestly :) | 18:09 |
| bluerise | I'd recommend not trying to fit in 2 boards into a single u-boot image, instead providing a better installer/upgrader | 18:14 |
| bluerise | but I mean, you do you | 18:14 |
| josch | bluerise: we already have reform-flash-uboot but upgrading u-boot always carries some danger with it... | 18:15 |
| bluerise | yeah, same risk if you have two images or a single image. when you upgrade u-boot, it might break | 18:15 |
| josch | bluerise: if it were up to you, how would you design the system? | 18:45 |
| + mark_ (~mjw@gnu.wildebeest.org) | 18:53 | |
| - mjw (QUIT: Killed (tungsten.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 18:54 | |
| * mark_ -> mjw | 18:54 | |
| + Guest7334 (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 18:54 | |
| * cwebber- -> cwebber | 19:08 | |
| bluerise | I can speak about what I did in the past, which seemed to work reasonably well. | 19:11 |
| bluerise | a) installs/upgrades transparent to the user -- they shouldn't need to see what's going on under the hood | 19:12 |
| bluerise | b) different products have different DTBs if necessary. often caused by major HW design differences (i.MX8MQ vs i.MX8MP) | 19:12 |
| bluerise | c) don't diverge DTBs if you don't have to. e.g. a pmic change, or DDR RAM setting change... have some HW major/minor versions e.g. through GPIOs/resistors around so that U-Boot can enable/disable features | 19:13 |
| bluerise | -- c) might not always great for upstreamability, but it's easier to manage as a product, and better than having 1000 different dtbs around solely because you needed to update some HW due to supply chain constraints | 19:14 |
| josch | bluerise: would you like to see u-boot upgraded transparently during normal system upgrades, for example? or would you prefer to be in charge when you want to flash a new u-boot? | 19:16 |
| josch | and thank you for your perspective :) | 19:16 |
| bluerise | From a usability point of view, doing it transparently is better. How would a user know whether it's safe or not to do the upgrade? Shouldn't be up for the user to decide, but whoever is packaging and shipping the new version needs to figure out: Is this safe? Will it run whatever the user boots? Is it backwards compatible? What could break? | 19:21 |
| ch | transparent is bad when there is no fallback | 19:22 |
| bluerise | Yeah but even then when someone opts to do the upgrade and it breaks, they are fucked the same way. Difference might be how many setups you're breaking | 19:23 |
| bluerise | Alternative is to stay on the oldest version as long as possible, which nowadays might work well enough since there's not going to be a lot of breaking ABI changes between U-Boot and whatever you're booting next. | 19:24 |
| bluerise | Or maybe introduce some dependency tracking of: can't upgrade to this version of Reform Software if you don't also update u-boot | 19:25 |
| bluerise | does RK3588 on Reform support going maskrom mode? | 19:26 |
| + gustav28 (~gustav@c-df53205c.019-141-67626730.bbcust.telenor.se) | 20:02 | |
| - cobra (QUIT: Ping timeout: 260 seconds) (~cobra@user/Cobra) | 20:13 | |
| + cobra (~cobra@user/Cobra) | 20:15 | |
| - cobra (QUIT: Quit: ZNC 1.8.2 - https://znc.in) (~cobra@user/Cobra) | 20:21 | |
| + cobra (~cobra@user/Cobra) | 20:25 | |
| josch | bluerise: disappointingly, the debian dependency system does not support depending on "this kernel is running" or "this bootloader is used". Those parts are unfortunately outside the control of the dependency system... :/ | 20:26 |
| josch | i'm going to work on a patch to the pocket reform u-boot because for example the debian-installer images ship /extlinux/extlinux.conf which just contains "fdtdir /dtbs/" for each entry and distro_boot is supposed to figure out the correct dtb for the platform without being told by the extlinux.conf | 20:28 |
| - natalie- (QUIT: Ping timeout: 252 seconds) (~natalie@user/natalie) | 20:47 | |
| + natalie (~natalie@user/natalie) | 20:50 | |
| Twodisbetter | josch: thanks | 21:23 |
| - colinsane (QUIT: Ping timeout: 260 seconds) (~colinunin@97-113-70-61.tukw.qwest.net) | 21:37 | |
| + mrdaught (~mrdaught@172.58.139.90) | 21:45 | |
| - mrdaught (QUIT: Read error: Connection reset by peer) (~mrdaught@172.58.139.90) | 21:45 | |
| + colinsane (~colinunin@97-113-136-73.tukw.qwest.net) | 22:13 | |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-df53205c.019-141-67626730.bbcust.telenor.se) | 22:15 | |
| - jacobk (QUIT: Ping timeout: 248 seconds) (~quassel@47-186-105-237.dlls.tx.frontiernet.net) | 22:26 | |
| + bkeys (~Thunderbi@50.96.250.19) | 22:32 | |
| - bkeys (QUIT: Ping timeout: 248 seconds) (~Thunderbi@50.96.250.19) | 22:47 | |
| - mjw (QUIT: Ping timeout: 258 seconds) (~mjw@gnu.wildebeest.org) | 23:50 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!