| minute | josch: what's the upside of cryptomgr tests? | 00:13 |
|---|---|---|
| minute | i thought only the tests were broken, not caam itself... or am i mistaken? | 00:13 |
| josch | minute: i have no clue about the detailts -- i just saw you including it in reform-system-image which means it stays on user's systems forever | 00:14 |
| minute | josch: afaik these are only a suite of self-tests | 00:15 |
| josch | yes, but you don't want to keep this forever because it's only for 6.5, right? | 00:16 |
| minute | hm https://lwn.net/Articles/400651/ | 00:16 |
| minute | > Though Xu posted a patch to default the self-tests to "off", it has not yet made its way into the mainline. Given Torvalds's statements, though, that will probably happen relatively soon | 00:17 |
| minute | > 2010 | 00:17 |
| josch | heh | 00:17 |
| josch | well, if the setting is supposed to stay in the kernel cmdline forever, things can as well remain as they are | 00:18 |
| minute | at the moment i would say so, in the long run it would be great if the kernel cmdline can be adjusted by updates | 00:19 |
| - S0rin (QUIT: Ping timeout: 240 seconds) (~S0rin@user/s0rin) | 00:19 | |
| + S0rin (~S0rin@user/s0rin) | 00:20 | |
| josch | we could use the method from my MR to do exactly that | 00:22 |
| josch | dependeng on the platform, different cmdline can be selected | 00:22 |
| minute | josch: ok cool | 00:22 |
| minute | btw for that fw blob, i'll have to do a few gymnastics: as i can only deploy to /boot, i have to pass another kernel cmdline to allow looking for firmware on /boot | 00:23 |
| josch | oh dear XD | 00:23 |
| + Gooberpatrol66 (~Gooberpat@user/gooberpatrol66) | 00:24 | |
| minute | maybe another option could be to mount the generated root partition and sneak it into /lib/firmware | 00:24 |
| josch | minute: why can you only deploy into /boot? i thought you were using the "--customize-hook="upload.... method? | 00:25 |
| minute | josch: yes, maybe i misunderstood something about mkimage.sh | 00:26 |
| minute | josch: does the second mmdebstrap see both partitions? | 00:26 |
| minute | i must say i was (and am) quite tired when i was working on this | 00:27 |
| josch | minute: the second mmdebstrap sees both but only /boot gets extracted from the result of it | 00:27 |
| josch | what deadline is hunting you right now that you are so short on time? | 00:28 |
| minute | josch: we have a big order backlog, and i want to ship things before i can concentrate on pocket reform finishing | 00:28 |
| josch | in "u-boot-config" in reform-ls1028a-uboot you could use CONFIG_USE_BOOTARGS=y and CONFIG_BOOTARGS to set ${bootargs} to ls1028a specific stuff like here: https://source.mnt.re/reform/reform-a311d-uboot/-/blob/main/bananapi-cm4-mnt-reform2_defconfig?ref_type=heads#L86 | 00:28 |
| minute | josch: yes, i did that already | 00:29 |
| josch | ah kk | 00:29 |
| minute | josch: but i'm not sure if that's very elegant :D | 00:29 |
| josch | vagrantc said another time that u-boot is supposed to set the platform-specific kernel cmdline | 00:30 |
| josch | because boot.scr as generated by flash-kernel is really bad at that | 00:30 |
| minute | josch: yeah, sorry, that's fine, i meant more like adjusting the cmdline to load firmware from /boot | 00:30 |
| minute | there's a parameter for that | 00:30 |
| josch | ah right | 00:30 |
| josch | i don't think i know the answer to that better than you do :) | 00:31 |
| minute | https://www.kernel.org/doc/html/v6.5/driver-api/firmware/fw_search_path.html | 00:31 |
| minute | 'firmware_class.path=$CUSTOMIZED_PATH' | 00:31 |
| minute | so i do firmware_class.path=/boot | 00:31 |
| minute | and in the driver i would request "ls1028a-mhdpfw.bin" | 00:31 |
| minute | which i place in /boot | 00:31 |
| minute | josch: but i'm thinking for example about: how can we load the fw in initramfs then? | 00:32 |
| minute | josch: will initramfs mount /boot? | 00:32 |
| minute | otherwise we won't have display | 00:33 |
| josch | i'm unsure | 00:33 |
| minute | in the end i would like u-boot to load the fw, but i don't know how | 00:34 |
| minute | in the vendor fw this is done with the special "hdp" command | 00:34 |
| josch | minute: but you are adding the "--customize-hook="upload.... to the first mmdebstrap invocation, right? | 00:34 |
| minute | vendor u-boot i mean | 00:34 |
| minute | josch: not yet | 00:34 |
| minute | josch: because then we would deploy the fw on all images. is that a good idea? | 00:34 |
| josch | no, you would write something like: | 00:35 |
| josch | ah | 00:35 |
| josch | no, you are right of course | 00:35 |
| josch | yes, the current method has the same rootfs for all platforms | 00:35 |
| josch | hrm... | 00:35 |
| minute | we could deploy it on /lib/firmware on all images, and if someone doesn't like it, they could delete the file ^^ | 00:35 |
| minute | but anyway, i realize there is no solution for initramfs | 00:36 |
| josch | not necessarily -- you can add any crazy script to the initramfs | 00:36 |
| josch | see my first boot merge request for reform-system-image | 00:37 |
| josch | i'm even repartitioning in the initramfs | 00:37 |
| josch | and modules are loaded in there as well | 00:37 |
| minute | ah oh, ok so i have to look at it tomorrow | 00:38 |
| josch | in a hook you can copy arbitrary files into the initramfs | 00:38 |
| josch | and then modprobe the module with the correct arguments for the firmware path, for example | 00:38 |
| minute | what is strange is that mainline u-boot actually has references to the hdp load command | 00:38 |
| minute | https://github.com/u-boot/u-boot/blob/b9b83a86f0e84e837191db120c279a9cc0e3434b/board/congatec/cgtqmx8/cgtqmx8.c#L436 | 00:38 |
| minute | https://github.com/u-boot/u-boot/blob/b9b83a86f0e84e837191db120c279a9cc0e3434b/include/configs/ls1028ardb.h#L124 | 00:39 |
| minute | but i don't see where this command is implemented (i don't think it is) | 00:39 |
| minute | josch: ok thanks, putting it in initramfs might be the best solution after all | 00:40 |
| minute | (until we have it in u-boot and u-boot also can already enable the display) | 00:40 |
| + jacobk (~quassel@utdpat241106.utdallas.edu) | 00:40 | |
| josch | minute: in drivers/video/imx/hdp.c there is U_BOOT_CMD(hdp, ..., do_hdp,... | 00:41 |
| josch | is it not that? | 00:41 |
| minute | lol wait | 00:42 |
| minute | josch: that file doesn't exist https://github.com/u-boot/u-boot/tree/master/drivers/video/imx | 00:43 |
| minute | josch: are you looking at vendor uboot? | 00:43 |
| josch | that was from boundary uboot | 00:43 |
| minute | josch: well, boundary uboot is imx uboot | 00:43 |
| josch | drivers/video/nxp/imx/hdmi/hdp_load.c | 00:43 |
| minute | we could of course try to use that instead of mainline | 00:43 |
| minute | just, sigh | 00:44 |
| josch | and drivers/video/nxp/layerscape/hdp_load.c | 00:44 |
| josch | that's not it? | 00:44 |
| minute | drivers/video/nxp does not exist in mainline uboot | 00:45 |
| josch | oh | 00:45 |
| josch | i'm still at vendor uboot | 00:45 |
| josch | my bad | 00:45 |
| minute | yeah, the problem is that that stuff requires all kinds of other code | 00:45 |
| minute | some weird API file etc | 00:45 |
| josch | bummer :( | 00:45 |
| minute | my first thought was to port it to mainline uboot but then i gave up because it looked like major spaghetti | 00:46 |
| josch | yeah, i guess going the initramfs route is the way forward | 00:46 |
| josch | one problem with that is, that this still needs the firmware somewhere on disk | 00:47 |
| josch | or otherwise, it's gone next time the initramfs is re-generated | 00:47 |
| minute | oh i see hm | 00:48 |
| josch | or... you could ship the firmware as part of flash.bin and the initramfs script extracts it from there instead of from a file on a filesystem :) | 00:48 |
| josch | that would avoid having the firmware on the rootfs on all platforms | 00:49 |
| josch | i now see why you want to put it into /boot | 00:49 |
| minute | is flash.bin in the initramfs? | 00:49 |
| minute | ah, you mean the initramfs generation script | 00:50 |
| minute | if it's on /boot, then it's right next to flash.bin :D | 00:50 |
| josch | yup | 00:50 |
| josch | the script generating the initramfs can copy it from /boot into the initramfs | 00:50 |
| minute | ok | 00:51 |
| josch | and a script that is run during boot can then load it | 00:51 |
| minute | josch: initramfs probably has a /boot as well? | 00:51 |
| minute | internally | 00:51 |
| josch | i don't know if it mounts it | 00:51 |
| minute | yeah, i thought as part of the cpio maybe | 00:51 |
| minute | well, in the initramfs it can just live in /lib/firmware | 00:52 |
| josch | reform-tools can install a script into /usr/share/initramfs-tools/hooks that copies the firmware into the initramfs *only* on ls1028a | 00:53 |
| minute | so, to summarize: 1. it's deployed on /boot 2. initramfs script puts it in /lib/firmware in the initramfs cpio 3. profit | 00:53 |
| minute | josch: ah! | 00:53 |
| josch | and reform-tools would also ship a script in /usr/share/initramfs-tools/scripts/ which is run during boot that then does something with this firmware | 00:53 |
| minute | josch: i'm not sure anything special needs to be done | 00:53 |
| josch | initramfs-tools calls "hooks" the scripts that are run on creation of the initramfs | 00:53 |
| minute | josch: initramfs boot will modprobe the module and the module will request it from /lib/firmware | 00:53 |
| josch | yes, hopefully | 00:54 |
| minute | it = the fw | 00:54 |
| josch | if not, then the second script can take care to modprobe with whatever arguments required | 00:54 |
| minute | ok | 00:54 |
| minute | all of these hoops just to load proprietary code... | 00:54 |
| josch | don't remind me of my nvidia story from last week XD | 00:55 |
| minute | haha which one is that | 00:55 |
| josch | nono | 00:55 |
| josch | it's just making everybody upset | 00:55 |
| minute | (at least i'm not putting it as a hex string in the driver) | 00:55 |
| josch | well, you could! :D | 00:55 |
| minute | yeah, but i just checked, even zipped it's like 36kB | 00:55 |
| josch | okay, sounds like a good plan for tomorrow | 00:56 |
| josch | i have to hit the pillow :) | 00:56 |
| minute | and it would make the kernel non-gpl | 00:56 |
| minute | n8n8 josch | 00:56 |
| josch | good night! | 00:56 |
| - XYZ (QUIT: Read error: Connection reset by peer) (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 01:45 | |
| + XYZ (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 01:48 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 01:49 | |
| - jacobk (QUIT: Ping timeout: 240 seconds) (~quassel@utdpat241106.utdallas.edu) | 01:58 | |
| - mtm (QUIT: Ping timeout: 260 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 02:03 | |
| - S0rin (QUIT: Ping timeout: 260 seconds) (~S0rin@user/s0rin) | 02:22 | |
| - XYZ (QUIT: Read error: Connection timed out) (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 02:23 | |
| + S0rin (~S0rin@user/s0rin) | 02:23 | |
| + XYZ (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 02:40 | |
| - XYZ (QUIT: Read error: Connection timed out) (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 03:15 | |
| - S0rin (QUIT: Ping timeout: 240 seconds) (~S0rin@user/s0rin) | 03:28 | |
| + S0rin (~S0rin@user/s0rin) | 03:33 | |
| + XYZ (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 03:33 | |
| - ajr (QUIT: Quit: Connection closed for inactivity) (uid609314@user/ajr) | 03:57 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 04:09 | |
| - XYZ (QUIT: Read error: Connection timed out) (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 04:21 | |
| + XYZ (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 04:38 | |
| - GNUmoon (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon) | 04:49 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 04:50 | |
| + jacobk (~quassel@utdpat241106.utdallas.edu) | 05:13 | |
| - XYZ (QUIT: Read error: Connection timed out) (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 05:13 | |
| - GNUmoon (QUIT: *.net *.split) (~GNUmoon@gateway/tor-sasl/gnumoon) | 05:39 | |
| - jacobk (QUIT: Ping timeout: 246 seconds) (~quassel@utdpat241106.utdallas.edu) | 06:03 | |
| + jacobk (~quassel@utdpat242064.utdallas.edu) | 07:16 | |
| + XYZ (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 07:46 | |
| - jacobk (QUIT: Ping timeout: 240 seconds) (~quassel@utdpat242064.utdallas.edu) | 07:54 | |
| + jacobk (~quassel@64.189.201.150) | 08:54 | |
| + pandora (uid585533@id-585533.ilkley.irccloud.com) | 10:12 | |
| + MajorBiscuit (~MajorBisc@2001:1c00:31c:8400:f184:4168:559b:d91b) | 10:37 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 10:39 | |
| minute | some cool pix https://mastodon.sdf.org/@SDF/111095624347351809 | 11:02 |
| - GNUmoon (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon) | 11:20 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 11:21 | |
| - klardotsh (QUIT: Ping timeout: 260 seconds) (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 11:31 | |
| grimmware | minute: I lol'd at games/catclock on Debian | 11:43 |
| + cliffya (~CliffyA@180-150-53-95.b49635.mel.static.aussiebb.net) | 13:23 | |
| - mjw (QUIT: Ping timeout: 260 seconds) (~mjw@gnu.wildebeest.org) | 13:26 | |
| * mark_ -> mjw | 13:28 | |
| abortretryfail | That is 9front my dude | 13:43 |
| abortretryfail | That red key backlight on the pocket tho.... >_> 🔥 | 13:45 |
| abortretryfail | I want that on my Reform, but do I want it enough to do a heck of a lot of SMD soldering? | 13:46 |
| josch | minute: i investigated how to best put the firmware into the initramfs | 13:49 |
| josch | minute: if you run `modinfo -F firmware yourmodule` does the firmware that "yourmodule" need show up? | 13:49 |
| josch | minute: because if yes, then mkinitramfs will automatically copy that modules into your initramfs if you put the module into /usr/share/initramfs-tools/modules.d/reform.conf in reform-tools | 13:50 |
| - mtm (QUIT: Ping timeout: 244 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 14:03 | |
| - cliffya (QUIT: Ping timeout: 260 seconds) (~CliffyA@180-150-53-95.b49635.mel.static.aussiebb.net) | 14:17 | |
| + cliffya (~CliffyA@180-150-53-95.b49635.mel.static.aussiebb.net) | 14:17 | |
| Asmadeus | m | 14:20 |
| - cliffya (QUIT: Ping timeout: 244 seconds) (~CliffyA@180-150-53-95.b49635.mel.static.aussiebb.net) | 14:23 | |
| + cliffya (~CliffyA@180-150-53-95.b49635.mel.static.aussiebb.net) | 14:23 | |
| josch | minute: you might be able to adjust details of this to do the firmware copying: https://source.mnt.re/reform/reform-tools/-/merge_requests/53 | 14:31 |
| - cliffya (QUIT: Ping timeout: 244 seconds) (~CliffyA@180-150-53-95.b49635.mel.static.aussiebb.net) | 14:47 | |
| + cliffya (~CliffyA@180-150-53-95.b49635.mel.static.aussiebb.net) | 14:47 | |
| - jacobk (QUIT: Ping timeout: 255 seconds) (~quassel@64.189.201.150) | 14:55 | |
| + jacobk (~quassel@129.110.242.224) | 15:07 | |
| - Boostisbetter (QUIT: Remote host closed the connection) (4a410829d7@irc.cheogram.com) | 15:19 | |
| - cliffya (QUIT: Quit: WeeChat 4.0.2) (~CliffyA@180-150-53-95.b49635.mel.static.aussiebb.net) | 15:20 | |
| + Boostisbetter (4a410829d7@irc.cheogram.com) | 15:22 | |
| - jacobk (QUIT: Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) (~quassel@129.110.242.224) | 15:51 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 16:09 | |
| minute | josch: thank you very much for the script! the last argument to copy_file is the destination, right? so shouldn't it go to /lib/firmware ? | 16:10 |
| josch | minute: ah whoops, yes, absolutely! | 16:57 |
| josch | copypaste error :) | 16:57 |
| + jacobk (~quassel@64.189.201.150) | 17:01 | |
| josch | minute: i just realized that this will not do the right thing when run as part of reform-system-image because in that setup, /proc/device-tree/model is not the source of truth but /etc/flash-kernel/machine is | 17:01 |
| josch | minute: i fixed both problems and amended my commit in that MR | 17:10 |
| josch | (still untested of course) | 17:10 |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 17:26 | |
| - eery (QUIT: Remote host closed the connection) (~eery@77.137.73.95) | 17:43 | |
| minute | josch: thank you! will test asap, after validating backlight pwm for ls1028a | 17:51 |
| minute | we had bad luck with standalone keyboard parts (cases) https://mastodon.social/@mntmn/111098258140121877 | 17:51 |
| josch | oh no, what a pain :( | 17:53 |
| minute | yeah, we're quite shocked tbh | 17:55 |
| minute | ah, debian is missing > CONFIG_PWM_FSL_FTM is not set | 17:57 |
| minute | "FTM" is their PWM engine | 17:57 |
| jfred | Oooooof | 17:58 |
| josch | it's so weird to think that there must've been a person who saw these parts and still decided "yep, looks good! lets ship those to the customer!" such a waste of time, money and resources :( | 18:14 |
| - jacobk (QUIT: Ping timeout: 248 seconds) (~quassel@64.189.201.150) | 18:25 | |
| minute | josch: yeah :( | 18:29 |
| + mark_ (~mjw@gnu.wildebeest.org) | 18:30 | |
| - MajorBiscuit (QUIT: Ping timeout: 258 seconds) (~MajorBisc@2001:1c00:31c:8400:f184:4168:559b:d91b) | 18:40 | |
| - mark_ (QUIT: Ping timeout: 258 seconds) (~mjw@gnu.wildebeest.org) | 19:11 | |
| + mark_ (~mjw@gnu.wildebeest.org) | 19:15 | |
| - mark_ (QUIT: Ping timeout: 258 seconds) (~mjw@gnu.wildebeest.org) | 19:20 | |
| + mark_ (~mjw@gnu.wildebeest.org) | 19:32 | |
| minute | ok, got pwm backlight working on the ls1028a | 19:33 |
| josch | \o/ | 19:38 |
| + jacobk (~quassel@129.110.242.224) | 19:42 | |
| minute | now to put everything together again... | 19:57 |
| Boostisbetter | Team MNT, I happy to announce that I am on the 52 succesful resume from suspend on my Reform with the 6.5 kernel | 20:02 |
| minute | debian is also missing CONFIG_FSL_RCPM=y | 20:07 |
| minute | Boostisbetter: cool! | 20:07 |
| - mjw (QUIT: Killed (NickServ (GHOST command used by mark_!~mjw@gnu.wildebeest.org))) (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:09 | |
| * mark_ -> mjw | 20:09 | |
| + mark_ (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:09 | |
| josch | minute: once you have a list of the missing kernel config options for ls1028a i will submit a merge request for the debian kernel so that they get included there as well. Then we don't have to deviate on that. | 20:10 |
| josch | those for imx8mq were approved quite quickly | 20:11 |
| minute | josch: cool! | 20:12 |
| minute | josch: ok so i merged https://source.mnt.re/reform/reform-tools/-/merge_requests/53 , and will deploy the fw to /boot in the second mmdebstrap invocation, right? | 20:13 |
| minute | as the second invocation runs flash-kernel, my guess is it will be picked up there | 20:14 |
| + klardotsh (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 20:22 | |
| - jacobk (QUIT: Ping timeout: 240 seconds) (~quassel@129.110.242.224) | 20:49 | |
| abortretryfail | weird, waybar's battery meter is wrong. | 21:20 |
| abortretryfail | circle+b says 62%, /sys/class/power_supply/8xlifepo4/capacity says 62, but waybar says 71% | 21:21 |
| + jacobk (~quassel@utdpat241106.utdallas.edu) | 21:22 | |
| + eery (~eery@77.137.73.95) | 21:28 | |
| - eery (QUIT: Ping timeout: 245 seconds) (~eery@77.137.73.95) | 21:35 | |
| minute | i pushed more ls1028a things. lets see if we have an almost-ready system image tomorrow | 21:39 |
| + eery (~eery@77.137.73.95) | 21:39 | |
| - mjw (QUIT: Ping timeout: 258 seconds) (~mjw@gnu.wildebeest.org) | 22:16 | |
| - GNUmoon (QUIT: Ping timeout: 252 seconds) (~GNUmoon@gateway/tor-sasl/gnumoon) | 22:17 | |
| - Boostisbetter (QUIT: Ping timeout: 258 seconds) (4a410829d7@irc.cheogram.com) | 22:27 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 22:31 | |
| - XYZ (QUIT: Ping timeout: 240 seconds) (~XYZ@37-48-2-164.nat.epc.tmcz.cz) | 22:39 | |
| + XYZ (~XYZ@78-80-18-84.customers.tmcz.cz) | 22:53 | |
| abortretryfail | Oh good, i'm not the only one who noticed that waybar is way off. https://community.mnt.re/t/lpc-spi-firmware-and-driver-prototype/1030/13 | 23:25 |
| + bkeys (~bkeys@c-67-191-106-248.hsd1.fl.comcast.net) | 23:34 | |
| bkeys | Hi all, I'm having an issue with my Reform. The batteries died (I have the protected battery boards) and now I get this when I plug it up to the charger | 23:35 |
| bkeys | https://postimg.cc/d7fswrDX | 23:37 |
| bkeys | Here is what my LPC screen is looking like | 23:38 |
| abortretryfail | It's all scrambled? | 23:38 |
| abortretryfail | Might be worth hitting the reset button on the keyboard and see if it comes back. You can get it with the corner of a credit card just above the F4 key without unscrewing the bezel. | 23:39 |
| - bkeys (QUIT: Quit: Client closed) (~bkeys@c-67-191-106-248.hsd1.fl.comcast.net) | 23:42 | |
| - vagrantc (QUIT: Ping timeout: 240 seconds) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 23:44 | |
| - GNUmoon (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon) | 23:53 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 23:54 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!