maxwelld | yeah i broke d4 battery once, its cable got disconnected from the battery, and it was my best battery. | 00:06 |
---|---|---|
maxwelld | and i dont like when u cannot have a spare battery with u and u need a screwdriver to change the battery. | 00:10 |
maxwelld | another reason to want to use d3. | 00:10 |
maxwelld | it has normal battery. | 00:10 |
maxwelld | and n900's battery was hotswappable. if u did it fast enough u could change the battery without shutting down the phone. | 00:10 |
dsc_ | lol | 00:11 |
Juest | maxwelld: what's the timing for hotswapping without power off? | 00:12 |
sicelo | with mainline kernel, there's no timing ... as long as your charger can supply enough current to keep the device powered. you can even run it off the charger the entire time | 00:14 |
sicelo | with fremantle's bme, it was a very few seconds. 1 to 2, i'd say | 00:17 |
freemangordon | tmlind: is it ok to call hrtimer_start() with zero timeout? | 08:04 |
tmlind | freemangordon: hmm don't know what happens then | 08:11 |
freemangordon | well, my idea was to avoid initialization delays. but maybe it does not make sense | 08:18 |
tmlind | it might be ok, i don't know | 08:33 |
uvos__ | i think the "non removeable" battery in d4 is a blessing in desguise | 10:53 |
uvos__ | since it allowed motorola used screw terminals for power. which now makes it easy for us to wire in any battery | 10:54 |
uvos__ | which is a good thing given the last eb41 rolled off the assembly line in 2014 or so | 10:54 |
Wizzup | arno11: so you wanted me to build these for extras right? | 11:29 |
Wizzup | arno11: can you make some wiki entries for it? | 11:29 |
arno11 | Wizzup: yeah, i tried to build deb packages locally but failed because of makefile/rules errors i'm not familiar with. Otherwise binaries works perfectly, no particular errors. | 11:54 |
arno11 | and for wiki, no probs i already posted things on TMO so just copy paste needed | 11:55 |
Wizzup | ah, ok | 12:02 |
arno11 | i mean i need your help to fix makefile/rules stuff and to be sure it doesn't try to install old maemo gui libs we can't find/replace anymore. the generic menu (picodrive or pcsx) are ok and usable. | 12:04 |
arno11 | sorry i'm doing too much things at the same time: for picodrive there is just a makefile error, no troubles with maemo stuff (not based on old maemo picodrive) | 12:21 |
freemangordon | uvos__: this commit https://github.com/maemo-leste/droid4-linux/commit/4af467160d52cc169bb5c27cab2a062bb4f14bea is broken | 12:26 |
freemangordon | why was it pushed at all? I just provided it for testing | 12:26 |
freemangordon | at the end it turned out the only missing thing was a call to init_waitqueue_head | 12:27 |
uvos__ | sicelo requested it | 12:27 |
freemangordon | ok, but now the kernel doe not even build :) | 12:27 |
uvos__ | i had no idea it was your patch | 12:27 |
uvos__ | i know it failed | 12:27 |
freemangordon | ok | 12:28 |
freemangordon | shall I revert it? | 12:28 |
freemangordon | or could you do it? | 12:28 |
uvos__ | sure, but i have to add some mdm fixes | 12:28 |
uvos__ | so dont build it yet | 12:28 |
freemangordon | ok, then you do it | 12:28 |
freemangordon | (revert 4af467160d52cc169bb5c27cab2a062bb4f14bea) | 12:28 |
uvos__ | right | 12:28 |
freemangordon | thanks | 12:29 |
uvos__ | so call to init_waitqueue_head | 12:29 |
uvos__ | where is this? | 12:29 |
uvos__ | (patch) | 12:29 |
freemangordon | https://github.com/maemo-leste/droid4-linux/commit/4af467160d52cc169bb5c27cab2a062bb4f14bea#diff-7ea64515dff442a6a9d926f9cd9d151420588a6fbfbe8d0cd51ebab05ef83b12R244 | 12:29 |
uvos__ | ok | 12:29 |
freemangordon | but it should be init_waitqueue_head(&ir_rx51.wqueue); as original code uses static global struct | 12:29 |
freemangordon | anyway, I don;t think we shall waste time on that | 12:30 |
uvos__ | why not (i really have no idea what this is about) | 12:30 |
freemangordon | I think I have a patch for ir_pwm that will make it work properly on n900 | 12:30 |
uvos__ | different driver, ok | 12:30 |
freemangordon | this is IR TX | 12:30 |
uvos__ | ok | 12:31 |
freemangordon | yes, and the driver we're trying to fix is marked for removal anyways | 12:31 |
uvos__ | i presume ir_pwm can use any pwm device and this one is omap3/n900 specific | 12:31 |
uvos__ | ok | 12:31 |
freemangordon | yes | 12:31 |
uvos__ | ill just revert then | 12:31 |
freemangordon | see "media: rc: remove ir-rx51 in favour of generic pwm-ir-tx" | 12:32 |
freemangordon | right | 12:32 |
Wizzup | arno11: check | 12:32 |
freemangordon | uvos__: actually there is nothing omap/n900 specific in ir-rx51 | 12:32 |
Wizzup | just trying to have a canonical point for this stuff in the wiki | 12:33 |
Wizzup | freemangordon: other than the name :D | 12:33 |
freemangordon | both drivers are very similar, besides ir-rx51 works on n900 while pwm-ir-tx does not | 12:33 |
freemangordon | Wizzup: not really | 12:33 |
freemangordon | ir-rx51 uses hrtimer expicitly, while pwm-ir-tx uses usleep_range() | 12:33 |
freemangordon | and that's what I think breaks it on n900 | 12:34 |
freemangordon | I have a patch that fixes that, hopefully | 12:34 |
freemangordon | will ask sicelo and/or arno11 ti test it later on | 12:34 |
Wizzup | okay cool | 12:41 |
Wizzup | freemangordon: the name was mostly a joke since it is called rx51 | 12:41 |
freemangordon | sicelo: arno11: please test https://pastebin.com/bS7Q43Mg | 12:41 |
freemangordon | once we have it working, I'll send a proper patch to lkml | 12:42 |
freemangordon | Wizzup: right :) | 12:42 |
uvos__ | presumably pwm-ir-tx would not exist if the author of ir-rx51 had more foresight with the name | 12:47 |
freemangordon | heh | 12:48 |
sicelo | freemangordon: thanks. arno, i don't have my leste sd card handy again. let me know which kernel you're on, and i can build you the module so you test with Pierogi | 13:14 |
Daanct12 | in case if anyone wasnt aware of it already, u-boot just dropped the n900 upstream | 13:21 |
Daanct12 | end of an era i guess :\ | 13:22 |
sicelo | Daanct12: https://libera.irclog.whitequark.org/maemo-leste/2023-09-05#34872479; | 13:24 |
sicelo | anyway, the existing uboot will still work, and someone might pick it up again in future (I would like to, but seems that's a tad too low-level for me) | 13:25 |
Daanct12 | yeah, the 2013 u-boot still works | 13:25 |
sicelo | or the unfortunate fallout gets resolved too (because it's not really dropped for technical reasons ...) | 13:26 |
Daanct12 | the commit log says it was mostly because of the usb code being legacy | 13:26 |
arno11 | sicelo: ok cool, i'm under 6.1.48 :) | 16:59 |
sicelo | arno11: https://phoseka.com/pwm-ir-tx.ko ... | 20:33 |
sicelo | sha512sum: b0c6de105eceb228859e5e1162e55ada6c4e3924c54d679f275a284d56f4c0dd49f1fb0753cb47ac668b5042048b967f4e696c9f2451143d793624fd68d96f04 | 20:33 |
arno11 | thx i'll be ready to test in 1 hour | 20:39 |
freemangordon | sicelo: arno11: if it works, I want you to test another (much simpler) patch, ok? | 20:42 |
arno11 | ok | 20:42 |
arno11 | freemangordon: sicelo: i tried the patch: works well when you press one key then when you try again it causes a kernel oops. | 21:40 |
freemangordon | :( | 21:42 |
arno11 | indeed :( | 21:42 |
freemangordon | ok, seems I'll have to bring my dev n900 back to life | 21:43 |
freemangordon | not now though | 21:43 |
arno11 | maybe the problem is not the patch but dts config ? | 21:44 |
freemangordon | should not cause oops | 21:44 |
arno11 | ok | 21:45 |
freemangordon | can you somehow get the oops? | 21:46 |
arno11 | if it's the same kind of oops i think only Wizzup and is serial could help | 21:48 |
sicelo | arno11: you can use mtdoops, but it's a bit unreliable sometimes. works well when it wants to :-) | 21:52 |
freemangordon | yeah, mtdoops should work | 21:52 |
sicelo | reboot the device, then select to use the uboot console. there you can type the following ... it's doable even on our tiny keyboard: | 21:53 |
arno11 | ok i can't this night but i can definitely try tomorrow | 21:53 |
sicelo | `setenv bootargs console=tty0 debug mtdoops.mtddev=log mtdoops.record_size=8192 rootwait root=/dev/mmcblk0p2` | 21:53 |
arno11 | ok | 21:53 |
freemangordon | sicelo: how do you build the module to be compatible with zImage from the .deb? | 21:55 |
sicelo | then you can boot. you can confirm this cmdline worked by `cat /proc/cmdline` in the running system. if it worked, then you can do the test, and on next boot, you can do `cat /dev/mtd2ro | less` to hopefully get the oops | 21:55 |
arno11 | ok | 21:55 |
sicelo | freemangordon: i built the whole kernel, but didn't make deb :-) | 21:55 |
freemangordon | ok, but you provided only pwm-ir-tx.ko to arno11 | 21:56 |
freemangordon | so, what is your workflow? | 21:56 |
freemangordon | ofc you build the whole kernel, the question is how to replace only that .ko | 21:56 |
sicelo | i check out leste's kernel, build the whole thing, and send him the ko, because i think it doesn't depend on others | 21:57 |
freemangordon | ok, but what about module symbo versions? | 21:57 |
freemangordon | also, how do you build it? | 21:57 |
sicelo | i'm guessing all that matches, since it's still technically same kernel? | 21:57 |
freemangordon | make ARCH=arm -j4 INSTALL_MOD_STRIP=1 modules | 21:57 |
freemangordon | not really | 21:57 |
freemangordon | ? | 21:57 |
sicelo | make ARCH=arm CROSS_COMPILE=arm-linux-gnueabihf- INSTALL_MOD_PATH=${mod_dir} INSTALL_MOD_STRIP=1 modules_install | 21:58 |
freemangordon | hmm | 21:58 |
arno11 | sicelo's process works with your ir-rx51 patch | 21:58 |
freemangordon | ok, lemme try | 21:58 |
freemangordon | sicelo: so, you first build zImage and then apply th epatch? | 21:59 |
sicelo | no. | 22:00 |
sicelo | build the kernel first, then zimage | 22:00 |
freemangordon | what is "kernel first"? | 22:00 |
freemangordon | not uImage ;) | 22:00 |
sicelo | um, :p | 22:00 |
sicelo | ok my script has `systemd-inhibit make -j $(($(nproc) + 1)) O=${output_dir} zImage modules omap3-n900.dtb` | 22:00 |
sicelo | with ARCH/CROSS_COMPILE already exported | 22:01 |
freemangordon | same here | 22:01 |
sicelo | so it's fine. | 22:01 |
freemangordon | so, where are your modules installed? | 22:01 |
sicelo | in .output/modules_dir ... i customized that with the 0=${outpt_dir} | 22:02 |
freemangordon | full path please | 22:02 |
freemangordon | I mean: | 22:02 |
freemangordon | modules are installed in kernel version directory | 22:03 |
freemangordon | what is the kernel version directory your modules are installed in? | 22:03 |
sicelo | here's the script https://paste.debian.net/1291768/ | 22:04 |
sicelo | .output/modules_dir/lib/modules/6.1.48/ | 22:04 |
freemangordon | no way :) | 22:06 |
sicelo | :p | 22:06 |
sicelo | why not, heh? | 22:06 |
freemangordon | if you have modified code, it shoudl change the kernel version | 22:06 |
freemangordon | ../modules/lib/modules/6.1.48-41200-g5319954d333d-dirty/ | 22:06 |
freemangordon | for example | 22:06 |
sicelo | that happens if you didn't commit the code, iirc | 22:06 |
freemangordon | right, but even if you commit, it still changes the kernel version | 22:07 |
sicelo | no idea. it doesn't happen in my case, and i build lots of N900 kernels (for my pmOS maintenance work) | 22:07 |
freemangordon | sicelo: right, but there must be something else | 22:08 |
freemangordon | some env var or something | 22:08 |
sicelo | maybe kernel config option | 22:08 |
freemangordon | hmm | 22:08 |
sicelo | this is the only script i use, nothing else :-) | 22:08 |
freemangordon | ok, lemme try | 22:08 |
sicelo | i just (un)comment the appropriate parts at a given moment | 22:09 |
sicelo | i should (someday) fix it up properly | 22:09 |
freemangordon | sicelo: what does scripts/setlocalversion prints? | 22:12 |
freemangordon | 22:12 | |
sicelo | ~/Git/linux-kernel$ scripts/setlocalversion | 22:13 |
sicelo | Error: kernelrelease not valid - run 'make prepare' to update it | 22:13 |
freemangordon | ugh | 22:13 |
freemangordon | how's that? | 22:13 |
freemangordon | there is no include/config/auto.conf it seems | 22:14 |
freemangordon | hmm, maybe because of your output dir, lemme try | 22:14 |
freemangordon | sicelo: still installing to ../modules/lib/modules/6.1.48-41200-g5319954d333d-dirty | 22:37 |
sicelo | interesting | 22:37 |
freemangordon | mhm | 22:37 |
sicelo | maybe give arno11 the ko file generated in yours. may actually just work fine ootb :-) | 22:39 |
freemangordon | no way | 22:39 |
sicelo | ... | 22:40 |
freemangordon | BTW, how is that new .ko loaded without proper dtb? | 22:40 |
sicelo | i think he did adjust dtb to use pwm-ir-tx | 22:43 |
freemangordon | ok | 22:43 |
freemangordon | sicelo: so, CONFIG_LOCALVERSION_AUTO=y in your config? | 22:44 |
arno11 | yep, using a custom uImage | 22:44 |
freemangordon | and modules dir is still 6.1.48? | 22:45 |
sicelo | grep CONFIG_LOCALVERSION_AUTO Git/linux-kernel/.output/.config | 22:46 |
sicelo | CONFIG_LOCALVERSION_AUTO=y | 22:46 |
freemangordon | do you have .scmversion file by chance? | 22:47 |
sicelo | yes, it's empty though | 22:49 |
freemangordon | well... | 22:50 |
freemangordon | that's why you don't get version changed | 22:50 |
freemangordon | ok, I'll do the same | 22:50 |
sicelo | ah. i can't even recall when i set it, but now that you mention, i do vaguely remember something of that sort | 22:51 |
freemangordon | heh :) | 22:51 |
freemangordon | arno11: may I have your dts changes in regards to pwm-ir-tx? | 22:53 |
sicelo | you just replace the part that says ir-rx51 with pwm-ir-tx :-) | 22:53 |
freemangordon | ok | 22:54 |
freemangordon | also, where to get pierogi from? | 22:54 |
arno11 | yep i just replaced "nokia,n900-ir" by "pwm-ir-tx" | 22:54 |
freemangordon | maybe just provide the binary | 22:55 |
sicelo | freemangordon: may i just pick your brain ... it's really small, i promise. | 22:55 |
sicelo | https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/blob/main/src/core/devices/wwan/nm-modem-ofono.c#L1278-L1280 | 22:55 |
arno11 | https://github.com/arno1103/pierogi | 22:55 |
sicelo | i want to replace that `goto out` with s = "0.0.0.0"; | 22:56 |
sicelo | any obvious gotchas? e.g. pointer problems? | 22:56 |
sicelo | arno11: maybe you can save the deb for him somewhere if you can? | 22:57 |
freemangordon | arno11: yeah, if you have deb around it will be easier for me | 22:58 |
freemangordon | sicelo: not at a glance (obvious issues) | 22:58 |
sicelo | cool. thanks | 22:59 |
arno11 | ok but not easy from my n900 (and again the baby...) | 22:59 |
freemangordon | ok, I'll build it | 22:59 |
sicelo | arno11: let me see if i have the one you sent me | 22:59 |
arno11 | sicelo: i think i sent you the deb file iirc | 22:59 |
arno11 | cool thx | 22:59 |
sicelo | freemangordon: https://phoseka.com/pierogi_1.1.41_armhf.deb | 23:01 |
sicelo | this is my friend's domain and vps ... i'm abusing ... /me hides | 23:01 |
arno11 | :) | 23:02 |
freemangordon | HTTP request sent, awaiting response... 403 Forbidden | 23:02 |
freemangordon | :) | 23:02 |
freemangordon | check file permissions | 23:02 |
sicelo | try now | 23:03 |
freemangordon | now it is ok | 23:04 |
uvos | https://uvos.xyz/llama270b.txt <--- lol 70b llama2 knows about leste, is a bit out of date however | 23:31 |
uvos | pretty interesting how mutch information on niche topics its able to absorb | 23:31 |
uvos | in just some 40 gb of wights (quantized) | 23:32 |
freemangordon | arno11: does not crash here | 23:37 |
freemangordon | what profile do you use? | 23:37 |
freemangordon | and I am able to control my toshiba TV | 23:40 |
freemangordon | hmm, ok, it oopsed :) | 23:41 |
arno11 | :) | 23:42 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!