Wizzup | ,3ot@(none):/# grep ^core_pwrdm /sys/kernel/debug/pm_debug/count | cut -d',' -f2, | 00:47 |
---|---|---|
Wizzup | OFF:351,RET:7 | 00:47 |
Wizzup | root@(none):/# uptime 05:35:42 up 3:35, 0 users, load average: 0.00, 0.00, 0.00 | 00:47 |
Wizzup | not bad, right? | 00:47 |
Wizzup | (that's with just busybox sh) | 00:47 |
lel | James-Hanson opened an issue: https://github.com/maemo-leste/bugtracker/issues/599 (Hildon fails to load theme on newer Droid 4 images) | 00:55 |
Wizzup | (replied) | 01:00 |
Wizzup | Pali: what do you think about extending include/configs/nokia_rx51.h for reading dtb/ftd files? | 01:17 |
Wizzup | currently it only works with initrd | 01:17 |
Pali | Do you mean to read DTB and use it for booting if exists? | 01:18 |
Wizzup | yes, like bootz kernaddr - dtbaddr | 01:18 |
Wizzup | currently it searches for any initrd if the env var is set, but it doesn't do that for dtb | 01:19 |
Pali | why not? seems like a good idea | 01:19 |
Wizzup | ok | 01:19 |
Pali | would you provide a patch for it? | 01:19 |
Wizzup | yeah, if there are no size concerns because it would increase the size of the script by a bit? | 01:19 |
Pali | you have to check | 01:20 |
Wizzup | I suppose they will also bug me for the musb dm conversion then ;) | 01:20 |
Pali | you can open a pull request to u-boot on github... opening pull request is just for running CI tests, nothing more | 01:20 |
Pali | and CI test checks also u-boot size for n900 | 01:20 |
Pali | so if it passes then there should not be any issue | 01:21 |
Wizzup | ok | 01:21 |
Pali | but IIRC kabel enabled LTO support for u-boot n900, so there should be more space | 01:22 |
Wizzup | ok | 01:22 |
freemangordon | Wizzup: I decided to revert changes I made for libhildonmime | 08:18 |
freemangordon | or at least will remove pre-depends I put there | 08:20 |
freemangordon | tmlind: what is the correct way to use clear_page()? | 09:36 |
freemangordon | hmm,scratch that, I guess it is userspce that shall clear the buffers | 09:41 |
freemangordon | Wizzup: check your mail, I will take care of omap DDX | 09:51 |
freemangordon | good, xrandr refuses to rotate to portrait | 10:03 |
freemangordon | on n900 that is | 10:03 |
freemangordon | Wizzup: ping | 11:19 |
freemangordon | Wizzup: omap ddx and leste-config are fixed to support both n900 and mapphone displays, so, the only missing thing is the kernel | 12:01 |
Wizzup | ok | 12:38 |
Wizzup | freemangordon: pong | 12:38 |
Wizzup | freemangordon: so this new patch replaces the older patch? | 12:41 |
Wizzup | freemangordon: droid3-dev is running on image-builder-32bit | 12:55 |
freemangordon | Wizzup: yes | 13:09 |
freemangordon | old one has a bug | 13:09 |
Wizzup | good | 13:11 |
Wizzup | I puled it in | 13:11 |
Wizzup | I will make a omap-linux repo today | 13:11 |
Wizzup | (maybe just rename droid4-linux) | 13:11 |
Wizzup | still I will need to do the dtb stuff that I talked about yesterday but that should be ok | 13:11 |
freemangordon | also, this patch is a 'proper' patch (with description, sign-off, etc) | 13:12 |
freemangordon | ok | 13:12 |
Wizzup | freemangordon: no readdir errors in the new build log | 13:41 |
Wizzup | so that's good news | 13:41 |
wunderw2 | Hello :) | 13:56 |
Wizzup | hi | 13:56 |
wunderw2 | I am connected via XChat on N900 :) | 13:56 |
Wizzup | wunderw: we likely fixed the latest image bug, we'll know in a few hours | 14:03 |
wunderw | I'm waiting for my second N900, I will test on that :) | 14:03 |
wunderw | On my first, I have Fanoush's Boot Menu, where I can't add Leste item | 14:04 |
wunderw | I will use U-Boot on the second | 14:04 |
bencoh | silly question, but why not install a proper up-to-date uboot on the first one as well? | 14:05 |
wunderw | I need to have BackupMenu | 14:05 |
bencoh | I might be wrong, but I think it works there as well | 14:05 |
wunderw | It works best with Fanoush's Boot Menu | 14:05 |
Wizzup | well, for leste in the future we will probably require upstream u-boot | 14:05 |
Wizzup | for booting zImage | 14:05 |
Wizzup | like in the next week or so :) | 14:05 |
bencoh | oh | 14:05 |
bencoh | Wizzup: is it tested with fremantle as well? | 14:06 |
Wizzup | although I suppose I could make a special n900 postinst | 14:06 |
Wizzup | bencoh: yeah | 14:06 |
Wizzup | pretty sure pali's CI uses fremantle kernel | 14:06 |
Pali | yes! CI in upstream u-boot uses fremantle kernel | 14:08 |
bencoh | :) | 14:28 |
freemangordon | Wizzup: what about other images? | 14:42 |
freemangordon | I see the latest images is still built on 65bit builder | 14:46 |
freemangordon | *64bit | 14:47 |
Wizzup | freemangordon: first I want to see if droid3-dev is ok | 14:57 |
Wizzup | freemangordon: I already switched the others over to the 32 bit builder | 14:57 |
Wizzup | we just need to hit build | 14:57 |
Wizzup | but let me verify that it is ok first | 14:57 |
Wizzup | I fixed the ssh problem that made it fail, but the bionic image is building now so will test that one... | 15:19 |
freemangordon | ok | 16:14 |
Wizzup | I will also look at making the virtual machine i/o faster | 16:15 |
Wizzup | (it takes a bit too long imho) | 16:16 |
freemangordon | mhm | 16:17 |
crab | Wizzup: for the uboot change, does that mean a potential reflash / reinstall? | 16:33 |
crab | im fairly sure im not using the fanoush boot menu but i have a fairly adhoc handrolled weird install on my n900. | 16:33 |
Wizzup | crab: I'll look at doing some postinst for the n900 specifically where we append the dtb and make a uImage | 16:33 |
crab | let me see if i can try and figure out what i have and where i got it from | 16:33 |
crab | im pretty confident editing things and making them work, im most concerned with an install that requires the least work going forward. | 16:34 |
crab | and im more than happy to just do a proper reinstall if thats the best way to achieve that. | 16:35 |
crab | ill admit that when i first installed leste it went on its own partition but i now think its so superior to all the other shit on here i should just ONLY use it. | 16:36 |
Wizzup | I don't know yet, I'll let you know | 16:36 |
crab | thanks. | 16:36 |
crab | oh i guess i should reboot the little dude as i just upgraded hildon wifi | 16:38 |
Wizzup | freemangordon: ok, new image has fixed theme | 16:39 |
lel | MerlijnWajer closed an issue: https://github.com/maemo-leste/bugtracker/issues/599 (Hildon fails to load theme on newer Droid 4 images) | 16:41 |
crab | my uboot is nearly 10 years old | 16:42 |
Wizzup | I'll see if we can make it work with a postinst | 16:43 |
Wizzup | probably can, just need to check | 16:43 |
Wizzup | uvos: btw on bionic it looks like the rgb led is somehow coupled to touchscreen leds, I have it blink when it is on and the touchscreen leds blink too | 16:43 |
Wizzup | (fresh image) | 16:43 |
Wizzup | freemangordon: got this on my d4: https://dpaste.com/9DY32AP3L | 17:41 |
Wizzup | this is still without config compact though | 17:42 |
freemangordon | Wizzup: did you do anything special to get that? | 18:08 |
freemangordon | OS error code 12: Cannot allocate memory | 18:09 |
freemangordon | hmm | 18:09 |
freemangordon | BTW my patch should not affect d4 | 18:09 |
freemangordon | on d4 we always go through DMM, so all buffers are linear | 18:09 |
freemangordon | so non-linear path is not taken | 18:10 |
freemangordon | neither is CMA used | 18:10 |
freemangordon | but yeah, I am not sure kernel can function properly without compaction | 18:10 |
freemangordon | eventually memory can get so fragmented that no pages/dma addresses buffers can be allocated | 18:13 |
freemangordon | omapdrm cannot allocate TILER memory | 18:18 |
freemangordon | 2 is TILFMT_32BIT | 18:18 |
freemangordon | for some reason tiler_reserve_2d() fails | 18:19 |
freemangordon | but still, without compaction this is to be expected | 18:20 |
Wizzup | freemangordon: no I just used it for days | 18:25 |
Wizzup | I think it also happened before I disable compaction | 18:25 |
Wizzup | but we'll see | 18:26 |
Wizzup | I am failing to boot on n900 with separate dtb atm :) | 18:26 |
freemangordon | TBH I don;t think this will work, but... | 18:27 |
Wizzup | what is 'this' ? | 18:27 |
freemangordon | re oops - I will not be surprised if we have memleaks here and there | 18:27 |
Wizzup | right | 18:28 |
freemangordon | this == "n900 does not boot" | 18:28 |
Wizzup | aha | 18:28 |
freemangordon | like: | 18:28 |
Wizzup | why do you think it will not work? | 18:28 |
freemangordon | not sure this was ever tested or intended to be supported in u-boot | 18:28 |
freemangordon | I think we can call Pali to the rescue :) | 18:28 |
freemangordon | have to cook dinner, ttyl | 18:29 |
Wizzup | we can also just append it with a preinstall file I think | 18:30 |
freemangordon | I think this is the right solution, demanding people to upgrade u-boot is not really polite | 18:30 |
Wizzup | I think having to upgrade u-boot is a non issue | 18:31 |
Wizzup | we also ask them to update kernel :) | 18:31 |
freemangordon | why is that? | 18:31 |
freemangordon | which kernel? | 18:31 |
Wizzup | linux kernel | 18:32 |
freemangordon | you can have u-boot with Nokia kernel | 18:32 |
Wizzup | nevermind, I don't think this is a useful discussion at all | 18:32 |
Wizzup | point is that we providing new and better sw | 18:32 |
freemangordon | yeah | 18:32 |
sicelo | When would the pre-install file run | 18:32 |
Wizzup | wanting to stick to something old doesn't make sense | 18:32 |
Wizzup | sicelo: well on postinst of the kernel package? | 18:32 |
Wizzup | it can just 'cat' the files | 18:32 |
Wizzup | freemangordon: I don't really want to use uImage either, which is also not supported on old u-boot | 18:33 |
freemangordon | zImage I guess | 18:33 |
Wizzup | this is a real PITA with multi-device kernel | 18:33 |
Wizzup | freemangordon: zImage is not supported yes, it only supports uImage | 18:33 |
freemangordon | you made a typo, thus my questio | 18:34 |
Wizzup | no | 18:34 |
Wizzup | old u-boot does not support not using uImage | 18:34 |
Wizzup | i.e. zImage | 18:34 |
freemangordon | ah, ok :) | 18:34 |
Wizzup | I agree it is a bit confusing | 18:34 |
freemangordon | I parsed wrongly | 18:34 |
freemangordon | yeah, ok | 18:34 |
freemangordon | anyway, bbl | 18:34 |
Wizzup | I might just be loading dtb on wrong addr or something | 18:35 |
Wizzup | but using addr of initrd when not using initrd doesn't work | 18:35 |
Wizzup | so I am assuming it doesn't work | 18:35 |
sicelo | It should. Pinephone and librem5 are also u-boot, and they support external dtb just fine | 18:43 |
Wizzup | sicelo: u-boot supports it fine, that is not in question | 18:45 |
Wizzup | whether it works on the n900 with the defconfig is another question | 18:45 |
Wizzup | I am trying this: | 18:48 |
Wizzup | ext2load mmc 0:1 0x82008000 zImage | 18:48 |
Wizzup | ext2load mmc 0:1 0x84008000 omap3-n900.dtb | 18:48 |
Wizzup | setenv bootargs 'root=/dev/mmcblk0p2 rootwait console=ttyS2,115200 verbose earlyprintk debug' | 18:48 |
Wizzup | bootz 0x82008000 - 0x84008000 | 18:48 |
Wizzup | the annoying thing with doing it in postinst is that our kernel pkg really only has one postinst | 18:49 |
Wizzup | we have: CONFIG_ARM_ATAG_DTB_COMPAT_CMDLINE_FROM_BOOTLOADER=y | 18:50 |
Wizzup | so I would assume this would work | 18:50 |
Wizzup | err | 18:50 |
Wizzup | I guess there isn't even an option for not supporting getting dtb from bootloader in linux | 18:51 |
Wizzup | must be always on | 18:51 |
Wizzup | maybe having appended dtb on somehow causes kernel to read appended dtb garbage | 18:53 |
Wizzup | parazyd: ping, what do you think about having extra packages in droid4-linux's control file (like linux-image-n900 that just depends on the main kernel package) that have their own additional postinst to create zimage with dtb appended or even uImage, would that work? | 18:58 |
Wizzup | uvos: droid4 devel image boots fine now | 19:16 |
Wizzup | uvos: the wifi module doesn't probe though | 19:16 |
Wizzup | actually I think debian has an /etc/kernel postinst dir so we can just insert scripts there for n900- | 19:47 |
Wizzup | parazyd: ^^ | 19:47 |
Wizzup | what do you think? | 19:47 |
bencoh | it would probably make it simpler for fremantle user to try leste | 19:53 |
freemangordon | Wizzup: hmm, 32MiB for kernel? | 19:53 |
bencoh | the other option would be to push the latest uboot version to maemo/maemo-extras | 19:54 |
freemangordon | isn't 0x84008000 - 0x82008000 too much? | 19:54 |
freemangordon | oh, I guess no, zImage is compressed | 19:55 |
freemangordon | so it must be uncompressed somewhere I guess | 19:55 |
freemangordon | Wizzup: if you ask me, creating uImage with attached dtb while installing on-device is the best option | 19:57 |
freemangordon | there is postinst.d and preinst.d | 19:58 |
freemangordon | a simple script in postinst.d that cats zImage and dtb and calls mkimage should do the job | 19:59 |
freemangordon | that script should be part of leste-config-n900 I guess | 20:00 |
Wizzup | freemangordon: so the way to read those bootz args is | 20:03 |
Wizzup | kernel_addr initrd_addr ftd_addr | 20:03 |
Wizzup | freemangordon: since we have no initrd addr we specify - | 20:03 |
freemangordon | ok | 20:04 |
freemangordon | Wizzup: still, I think creating uImage in postinst should be way easier for everybody once done | 20:04 |
Wizzup | freemangordon: right /etc/kernel/postinst.d | 20:04 |
freemangordon | mhm | 20:04 |
Wizzup | yeah I think that makes sense | 20:04 |
freemangordon | do you know who executes the stuff there? | 20:05 |
freemangordon | apt? | 20:06 |
Wizzup | not sure | 20:06 |
Wizzup | freemangordon: debian/linux-image/DEBIAN/postinst:test -d /etc/kernel/postinst.d && run-parts --arg="5.15.2-00592-g63f8176258b7-dirty" --arg="/boot/vmlinuz-5.15.2-00592-g63f8176258b7-dirty" /etc/kernel/postinst.d | 20:06 |
Wizzup | so probably our kernel will | 20:06 |
freemangordon | https://kernel-team.pages.debian.net/kernel-handbook/ch-update-hooks.html | 20:07 |
freemangordon | 8.1. Kernel hooks | 20:07 |
Wizzup | keep in mind we do not use debian kernel, we have our own | 20:07 |
freemangordon | yeah | 20:07 |
Wizzup | but we use builddep from mainline | 20:07 |
freemangordon | that's why my question | 20:07 |
freemangordon | good that our kernel is compatible | 20:07 |
Wizzup | right | 20:08 |
bencoh | don't you use make deb-pkg (or whatever it is call nowdays)? | 20:08 |
Wizzup | so I will try to make a control file which has omap-linux-image as actual kernel and modules and then aliases that depend on, with optional postinst hooks | 20:08 |
freemangordon | "hook scripts for boot loaders must be named using the prefix zz- " | 20:08 |
Wizzup | bencoh: sure, with patches | 20:08 |
freemangordon | Wizzup: wait, why is that | 20:08 |
Wizzup | freemangordon: it is not a hook for bootloader | 20:08 |
freemangordon | Wizzup: still, I don;t think this script belongs to the kernel package | 20:09 |
freemangordon | neither we need special kernel package for n900 | 20:09 |
freemangordon | I think it should be like: | 20:09 |
freemangordon | 1.one kernel package to rule them all | 20:10 |
freemangordon | 2. leste-config-n900 with postinst script that cats zImage+dtb and calls mkimage | 20:10 |
freemangordon | thats all | 20:10 |
freemangordon | and yea, I think this is hook script for bootloader | 20:11 |
freemangordon | unless I dont; understand what do they mean by "bootloader hook" | 20:12 |
bencoh | updating bootloader config in a special way maybe | 20:12 |
freemangordon | yeah, maybe | 20:12 |
freemangordon | not importeant really | 20:12 |
freemangordon | *important | 20:13 |
uvos | <Wizzup> uvos: btw I still see hangs sometimes on droid3, do you see them as well? | 20:13 |
uvos | yes lots | 20:13 |
uvos | allways hangs mid drawing frames | 20:13 |
uvos | <Wizzup> uvos: btw on bionic it looks like the rgb led is somehow coupled to touchscreen leds, I have it blink when it is on and the touchscreen leds blink too | 20:13 |
uvos | known issue, the ts buttons are connected - somewhere idk where | 20:13 |
freemangordon | uvos: what version is omap ddx? | 20:13 |
uvos | they share a regulator with the pannel and rgb led | 20:13 |
Wizzup | uvos: could be pvr clock rate? | 20:14 |
uvos | and are set on by the vendor kernel | 20:14 |
Wizzup | mhm | 20:14 |
uvos | so any time the mainline kernel sets eitehr of those on | 20:14 |
uvos | the ts buttons light up | 20:15 |
uvos | Wizzup: well the pvr clock rate is the same as on the 3.0 vendor kernel | 20:15 |
uvos | its possible that moto used a different rate on the 2.6 kernel for solana | 20:15 |
uvos | we can check that | 20:15 |
uvos | but idk why | 20:15 |
uvos | all the omap4 variants besides *70 have the same rating for sgx | 20:16 |
Wizzup | ok | 20:16 |
Wizzup | just not sure what else could cause it | 20:16 |
uvos | well d4 also hangs | 20:17 |
uvos | if you load sgx esspecaly | 20:17 |
parazyd | Wizzup: I have to think about that a bit re: postinst | 20:17 |
uvos | so maybe d3 just shows it sooner for some reason | 20:17 |
Wizzup | parazyd: we touched on it some after my msg | 20:17 |
uvos | Wizzup: could be ram accuatlly | 20:17 |
uvos | we dont touch the dram controller at all | 20:17 |
Wizzup | right | 20:17 |
uvos | the vendork kenrel might be changing things | 20:18 |
uvos | depending on other clocks | 20:18 |
Wizzup | yeah | 20:18 |
freemangordon | Wizzup: what do you think about my suggestion re leste-config/postinst? | 20:18 |
uvos | freemangordon: sec let me boot the thing | 20:18 |
freemangordon | ok | 20:18 |
Wizzup | freemangordon: sorry, how was it different from mine, maybe I missed it | 20:18 |
freemangordon | uvos: the point is this https://github.com/maemo-leste/xf86-video-omap/commit/472f6e725cf9e3b6206c89b7f3cc5bebd90beb54 | 20:19 |
Wizzup | 20:10 < freemangordon> 1.one kernel package to rule them all | 20:19 |
Wizzup | 20:10 < freemangordon> 2. leste-config-n900 with postinst script that cats zImage+dtb and calls mkimage | 20:19 |
freemangordon | if you don't have that, device may seem hung | 20:19 |
freemangordon | Wizzup: yes | 20:19 |
Wizzup | I suppose that could work too, but this allows potentially to have one without the other | 20:19 |
uvos | freemangordon: its not that | 20:19 |
freemangordon | uvos: ok | 20:19 |
uvos | it dosent seam hung | 20:19 |
uvos | it watchdog rebotos | 20:19 |
uvos | *reboots | 20:19 |
freemangordon | ah, ok | 20:20 |
freemangordon | Wizzup: how's that different to your suggestion? | 20:20 |
freemangordon | how do you know you run on n900? | 20:20 |
Wizzup | freemangordon: I was suggesting to have extra pkgs in control file of our omap-linux debian/control | 20:21 |
freemangordon | exactly, another package | 20:21 |
Wizzup | freemangordon: like linux-image-n900 | 20:21 |
Wizzup | same repo, just another virtual | 20:21 |
Wizzup | like we do with many things | 20:21 |
freemangordon | and you may have one without the other | 20:21 |
Wizzup | I don't think so | 20:21 |
freemangordon | the same if it is in the config | 20:21 |
freemangordon | ok | 20:21 |
Wizzup | well yes you could *uninstall* linux-image-n900 | 20:21 |
Wizzup | that would uninstall the postinst | 20:21 |
Wizzup | but that is more clear that is breaks things than leste-config-n900 imho | 20:21 |
freemangordon | yeah, you have a point | 20:22 |
freemangordon | Wizzup: so hildon-meta-n900 will depend on linux-image-n900? | 20:23 |
freemangordon | IOW - which package will pull linux-image-n900? | 20:24 |
Wizzup | freemangordon: right that is already the case now wrt meta | 20:24 |
freemangordon | ok | 20:37 |
Wizzup | so I will try that then | 20:38 |
freemangordon | ok | 20:39 |
sicelo | interesting - on Fremantle, it was said the bcm2048 integrated radio *requires* the bt to work. just re-read the short information sheet (https://pdf.dzsc.com/20090227/200902072254092347.pdf) and it states "FM system can operate independently from Bluetooth" ... can't wait to get my N900 fixed. maybe there's a way to make the FM transmitter work | 20:52 |
sicelo | s/transmitter/receiver/ | 20:54 |
freemangordon | it is the same chip | 20:55 |
freemangordon | and I guess it needs FW to be loaded | 20:55 |
Wizzup | uvos: any idea why wl module is not loaded for droid3? | 20:58 |
sicelo | the FW is for bt. you think the bt side must really be up? | 20:58 |
Wizzup | uvos: seems kinda weird | 20:58 |
Wizzup | freemangordon: do you want me to keep gdb attached to X to provide a backtrace when it crashes again? | 21:03 |
Wizzup | freemangordon: although I guess it seems like kernel problem | 21:03 |
Wizzup | freemangordon: or leak like you said | 21:03 |
freemangordon | Wizzup: do you still see crashes? | 21:05 |
freemangordon | or you mean that WARN()? | 21:05 |
Wizzup | well what I pasted caused X to crash | 21:07 |
freemangordon | ah | 21:09 |
freemangordon | yeah, it is good to have backtrace | 21:09 |
freemangordon | maybe this can be handled gracefully | 21:10 |
Wizzup | freemangordon: ok | 21:12 |
Wizzup | freemangordon: I think I will have to make those virtual packages in a different repo | 21:50 |
freemangordon | Wizzup: why is that? | 21:54 |
Wizzup | freemangordon: scripts/packaging/builddep seems to create control files or something | 21:56 |
Wizzup | it's overriding whatever else I have | 21:56 |
Wizzup | and end result is I don't even get a .deb file for the other pkgs in my control file | 21:56 |
Wizzup | dpkg-genchanges: warning: package linux-image-droid4 in control file but not in files list | 21:56 |
Wizzup | dpkg-genchanges: warning: package linux-image-droid3 in control file but not in files list | 21:56 |
Wizzup | dpkg-genchanges: warning: package linux-image-bionic in control file but not in files list | 21:56 |
Wizzup | dpkg-genchanges: warning: package linux-image-n900 in control file but not in files list | 21:56 |
Wizzup | freemangordon: let me see if I am forgetting something else | 22:00 |
freemangordon | Wizzup: or, maybe do as I suggested | 22:01 |
Wizzup | freemangordon: so they seem to call some custom bash create_package function that calls dpkg-gencontrol | 22:02 |
Wizzup | freemangordon: right that is no different from the virtuals though right? | 22:02 |
uvos__ | [20:58] <Wizzup> uvos: any idea why wl module is not loaded for droid3? | 22:03 |
Wizzup | I don't think leste-config is the right place for this | 22:03 |
uvos__ | its not? | 22:03 |
uvos__ | in what kernel | 22:03 |
Wizzup | uvos__: not for me, is it loaded for you? | 22:03 |
Wizzup | uvos__: just latest image | 22:03 |
freemangordon | Wizzup: up to you, it is just that it will be easier | 22:04 |
Wizzup | freemangordon: ok, let me take a few minutes to figure out what is easiest | 22:04 |
Wizzup | so I was thinking of creating maemo-kernel-virtuals repos with just linux-image-n900 with depends on kernel package, but I'd really much rather have it in the kernel package | 22:05 |
Wizzup | I could patch scripts/packaging/builddep | 22:05 |
Wizzup | but it's not really pretty | 22:05 |
Wizzup | brb | 22:06 |
freemangordon | neither is creating yet another repo to hold just one package | 22:07 |
freemangordon | unless by repo you mean git repo | 22:08 |
freemangordon | not debian repo | 22:08 |
Wizzup | I meant git repo yes | 22:09 |
Wizzup | we can do it in leste-config it just feels a bit weird | 22:09 |
Wizzup | but I don't mind having the postinst in there | 22:09 |
Wizzup | I can just do that now I suppose, but still, we will need something to provide linux-image-droid4 and so on virtual packages still I think | 22:10 |
Wizzup | unless we are just going to change all dependencies to linux-image-omap in hildon-meta etc | 22:10 |
Wizzup | uvos__: btw I think reboots hang because one tty process becomes unkillable due to serial oops | 22:11 |
Wizzup | uvos__: the S0 one | 22:11 |
freemangordon | Wizzup: I think it makes sense to change all deps to linux-image-omap | 22:11 |
uvos__ | yes i know | 22:11 |
uvos__ | Wizzup: for some reason my wl12xx dose probe on d3 | 22:12 |
uvos__ | but wifi dosent work at all | 22:12 |
Wizzup | freemangordon: hmm ok | 22:12 |
uvos__ | can scann etc | 22:12 |
uvos__ | (also older kernel image) | 22:12 |
uvos__ | another think i notice | 22:13 |
uvos__ | is that d3 has alot more pixmap corruption | 22:13 |
Wizzup | freemangordon: let's try that then | 22:14 |
freemangordon | ok | 22:15 |
freemangordon | Wizzup: how do you think, how critical is rotation on n900? IMO abook is with higher prio, no? | 22:17 |
Wizzup | I think I agree yes | 22:18 |
freemangordon | ok | 22:18 |
Wizzup | yeah definitely | 22:18 |
uvos__ | x refuses to rotate right | 22:19 |
uvos__ | it dosent crash | 22:19 |
Wizzup | yes | 22:19 |
uvos__ | ok | 22:19 |
freemangordon | so VRFB will wait after abook | 22:19 |
uvos__ | then yeah | 22:19 |
freemangordon | but, I think what I can spend some time on is PVR driver fences | 22:19 |
Wizzup | having abook further along would help with our end of january conversations goals | 22:20 |
freemangordon | for some reason current code sets the fence too early | 22:20 |
freemangordon | Wizzup: yeah | 22:20 |
Wizzup | I need to start pushing really hard on telepathy in the next two weeks or we won't make those | 22:20 |
Wizzup | spent too much on kernel stuff relatively :p | 22:20 |
Wizzup | too much time that is | 22:20 |
freemangordon | I'll suspend kernel work until we have at least semi-usable abook | 22:20 |
freemangordon | same here | 22:20 |
uvos__ | Wizzup: also on bringup devices that really dont matter :P | 22:21 |
uvos__ | (not that i blame you, its fun) | 22:21 |
uvos__ | i now know why wifi dosent work on my d3 | 22:21 |
uvos__ | maserati-callibarte failed and hosed the firmware | 22:22 |
uvos__ | appearently it dosent work on d3 for some reason | 22:22 |
Wizzup | uvos__: yeah | 22:22 |
Wizzup | uvos__: ah I see, yeah, probably that never normally ran on mine because I used bionic sdcard initially | 22:22 |
freemangordon | one more thing - maybe we shall make a freeze week. and if there are no critical bugs, all that is in -devel shall be pushed to stable | 22:22 |
freemangordon | freeze like - no new stuff in -devel | 22:22 |
freemangordon | as I am under the impression that -stable lags too much behind -devel | 22:23 |
freemangordon | might be wrong though, didn;t compared | 22:23 |
uvos__ | well freezing devel is not the point of devel for us is it? | 22:23 |
Wizzup | one of the reasons is we are making big changes | 22:23 |
freemangordon | understood | 22:24 |
freemangordon | but we are at the end of that | 22:24 |
freemangordon | more or less | 22:24 |
uvos__ | i think another big(ish) problem is that stable has been less stable than devel in the past | 22:24 |
uvos__ | mostly because none of us use/test it | 22:24 |
freemangordon | uvos__: another reason to make -stable equal to -devel | 22:24 |
Wizzup | freemangordon: so do I make leste-config-n900 depend on u-boot-tools for mkimage? | 22:25 |
freemangordon | mhm | 22:25 |
Wizzup | k | 22:25 |
freemangordon | uvos__: I am talking about very short freeze (week or so), to see if there are any critical bugs | 22:26 |
freemangordon | I am not saying we shall stop development for that period :) | 22:26 |
freemangordon | so, at the end of freeze, we push -devel to -stable and then build whatever is new in -devel | 22:27 |
Wizzup | freemangordon: hmm leste-config does something weird with .leste files I am not sure if this will work... | 22:27 |
Wizzup | parazyd: ping | 22:27 |
parazyd | pong | 22:28 |
Wizzup | parazyd: hi, so | 22:29 |
Wizzup | parazyd: I want to add a file that ends up in /etc/kernel/postinst.d | 22:29 |
Wizzup | to create a uImage for the n900 and freemangordon would like it in leste-config | 22:29 |
Wizzup | can I do that without making the file end in .leste and that kind of stuff? | 22:29 |
Wizzup | I fear that might mess with run-parts | 22:29 |
Wizzup | this would only be for leste-config-n900 | 22:30 |
parazyd | You can keep it in leste-config but then you have to install it by hand, not with the configfile infrastructure. | 22:30 |
Wizzup | so just add another line to debian/leste-config-n900.install ? | 22:30 |
parazyd | Otherwise, it does have to have the .leste suffix. | 22:30 |
parazyd | Yeah | 22:30 |
Wizzup | hmm | 22:30 |
parazyd | Remember that debian/gen_displace will overwrite them | 22:31 |
Wizzup | I am fine with whatever works really, so if you say that's the way to go then I'll do that | 22:31 |
parazyd | And it also uses a wildcard, so I'm not exactly sure how to add another file. | 22:31 |
parazyd | Probably if you just put it on top it'll work | 22:32 |
parazyd | Why not make the uImage from kernel postinst? | 22:32 |
Wizzup | if you can show me how I'll gladly do it | 22:32 |
Wizzup | I can't figure out how to work with scripts/packaging/builddeb | 22:32 |
Wizzup | it just does it's own thing and ignores everything else | 22:33 |
Wizzup | or so it seems to me | 22:33 |
Wizzup | I tried to add extra packages to the debian/control file of the kernel package, and then a debian/*.install file for it, it just gets ignored | 22:33 |
Wizzup | and the .deb files are also not generated | 22:33 |
Wizzup | I have a working postinst.d, I just need to be able to place it somewhere | 22:34 |
parazyd | Why are all the patches gone from here? https://github.com/maemo-leste/droid4-linux/tree/maemo/beowulf-devel/debian/patches | 22:34 |
parazyd | https://github.com/maemo-leste/droid4-linux/commit/c8eb20cb4f6c0572e76cf0cfd008af59d501ab7c#diff-e33e320ccfa50e52c4e29b3c27e54d78892ba05b36f2328d7b532f6539a3a276 | 22:35 |
Wizzup | parazyd: patches to kernel? we carry them in git | 22:35 |
parazyd | You killed the quirks patch | 22:35 |
Wizzup | yes | 22:35 |
parazyd | Is it on the main branch? | 22:35 |
Wizzup | it is in the kernel branch afaik yes | 22:35 |
Wizzup | yes for sure | 22:35 |
Wizzup | see 9539db972bd8cbf9653a1b3d4f949561fa49b70e | 22:36 |
Wizzup | in https://github.com/maemo-leste/droid4-linux/commits/wip/n900/maemo-5.15-cleaned-up | 22:36 |
parazyd | ah good | 22:37 |
parazyd | https://github.com/maemo-leste/droid4-linux/blob/wip/n900%2Fmaemo-5.15-cleaned-up/scripts/package/builddeb#L208 | 22:37 |
Wizzup | it manually calls dpkg-deb and stuff | 22:37 |
parazyd | So this is how the postinst is made | 22:37 |
Wizzup | that is a generic postinst | 22:37 |
Wizzup | we want one specifically that only runs on the n900 | 22:37 |
Wizzup | which is why there is a run-parts in the postinst for /etc/kernel/postinst.d per normal debian kernel stuff | 22:37 |
parazyd | https://github.com/maemo-leste/droid4-linux/blob/wip/n900%2Fmaemo-5.15-cleaned-up/scripts/package/builddeb#L192 | 22:37 |
parazyd | Just add it there | 22:37 |
freemangordon | parazyd: /etc/kernel/postinst.d | 22:37 |
parazyd | Wizzup: linux-image-n900.postinst should be the filename | 22:38 |
parazyd | Or whatever your virtual is called | 22:38 |
Wizzup | how will I prevent this from being installed with all other packages? | 22:38 |
Wizzup | like I said, no other .deb files are created when I tried this | 22:38 |
parazyd | What do you have in your debian/control file? | 22:39 |
parazyd | And is it respected? | 22:39 |
Wizzup | I removed it, but it was just basic packages like linux-image-n900 with Depends: on linux-image-omap | 22:39 |
Wizzup | I build locally in my lxc container so it never made it online | 22:39 |
Wizzup | I got this: | 22:39 |
Wizzup | dpkg-genchanges: warning: package linux-image-n900 in control file but not in files list | 22:39 |
parazyd | So if you have that, you'd create linux-image-n900.postinst and do whatever you need in there. | 22:39 |
Wizzup | well I had linux-image-n900.install | 22:40 |
Wizzup | and it was ignored entirely | 22:40 |
parazyd | That's different | 22:40 |
parazyd | It's probably generated by the deb scripts | 22:40 |
parazyd | (So are postinsts, but we can edit them here) | 22:40 |
Wizzup | but we still won't get a .deb file | 22:41 |
Wizzup | unless we manually call create_package | 22:41 |
Wizzup | which feels way more hacky than adding a single postinst hook from another place | 22:41 |
Wizzup | it's not great either way hehe | 22:41 |
Wizzup | at least my understanding is there still won't be a .deb file | 22:42 |
parazyd | leste-config is a good alternative if you think this is messy. | 22:42 |
Wizzup | from reading scripts/package/builddep | 22:42 |
Wizzup | well it is what I tried first, but then I failed to achieve it and tried an alternative | 22:42 |
Wizzup | ok, let me see if I can somehow make it work with leste-config then | 22:42 |
Wizzup | honestly I think I'll just end up making a separate package for all of this | 22:43 |
Wizzup | (with its own repo) | 22:43 |
Wizzup | that way I won't have to hack around the displace stuff either | 22:43 |
Wizzup | parazyd: hmm looks like gen_displace just overrides .install files | 22:47 |
Wizzup | grep 'Package: ' control | cut -d' ' -f2 | while read -r pkg; do printf "%s/* /\n" "$pkg" > "${pkg}.install" | 22:47 |
parazyd | 22:31 <parazyd> And it also uses a wildcard, so I'm not exactly sure how to add another file. | 22:47 |
parazyd | 22:32 <parazyd> Probably if you just put it on top it'll work | 22:47 |
parazyd | 22:31 <parazyd> Remember that debian/gen_displace will overwrite them | 22:47 |
Wizzup | yes I understand | 22:47 |
Wizzup | but if it just does > then how will a wildcard help? | 22:48 |
parazyd | I don't understand the question? | 22:48 |
parazyd | Just edit the script and add what you want | 22:48 |
parazyd | rip matrix | 22:48 |
Wizzup | oh, you want me to edit gen_displace | 22:49 |
parazyd | Right | 22:49 |
parazyd | Sorry, I wasn't clear | 22:49 |
Wizzup | I'm just going to make a separate package for this I think, it's all too complicate for wanting something so simple :D | 22:49 |
parazyd | That's ok too | 22:51 |
parazyd | Probably could create some kernel-config package | 22:51 |
parazyd | This way we can have control for all devices with a repo | 22:52 |
Wizzup | ok, I'll do that then, but start it off simple | 22:52 |
parazyd | *nod* | 22:52 |
Wizzup | btw we're also planning to change hildon-meta per-device packages | 22:52 |
Wizzup | e.g. linux-image-n900 will become linux-image-omap | 22:52 |
Wizzup | (and also have maemo-kernel-config-n900) | 22:52 |
parazyd | Sounds good | 22:52 |
Wizzup | great, thanks! | 22:53 |
uvos__ | freemangordon: http://uvos.xyz/maserati/h-mime | 22:55 |
uvos__ | see error, also this seams a bit verbose, this gets printed quite often | 22:56 |
uvos__ | heh | 23:10 |
uvos__ | so for some reason probeing wl12xx/udev informing userspace is slower on d3 | 23:10 |
uvos__ | this triped up ti-utils | 23:10 |
Wizzup | I guess we'll need some replaces: lines for the kernel upgrade(s) as well | 23:10 |
Wizzup | uvos__: heh... | 23:10 |
uvos__ | so maserati-callibrate works on d3 now | 23:11 |
uvos__ | ill upgrade now and see if with the new kernel image it dosent probe like on yours | 23:11 |
Wizzup | ok, cool | 23:13 |
freemangordon | uvos__: those are harmless | 23:17 |
freemangordon | afaik | 23:17 |
uvos__ | Wizzup: ti-utils-wilink6 lacks a devel branch | 23:20 |
uvos__ | Wizzup: shal i push the fix to stable immidatly? | 23:21 |
uvos__ | or make one | 23:21 |
Wizzup | let's do -devel | 23:21 |
uvos__ | Wizzup: so i just dist-upgraded my solana and wifi works fine | 23:54 |
uvos__ | interestingly i get quite superior throughput vs d43 | 23:55 |
uvos__ | *d4 | 23:55 |
uvos__ | attenas/impeadance matched traces are probubly better | 23:56 |
uvos__ | the serial oops dosent happen when shutting down from emergency shell | 23:57 |
uvos__ | so some deamon closing the device causes the oops | 23:57 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!