| uvos | Wizzup: here are the signalmaps for xt860 and xt875 parsed from dtb: | 01:10 |
|---|---|---|
| uvos | http://uvos.xyz/maserati/stockinfo/signalmaps/ | 01:11 |
| uvos | unfortionaly for you the modem wakeup gpio mdm6600_usb_rwkup is not in the xt860 signalmap | 01:12 |
| uvos | thats a bit perplexing | 01:12 |
| uvos | the other modem gpios are the same | 01:15 |
| Wizzup | uvos: thanks! | 01:24 |
| Wizzup | a lot of them look quite similar | 01:30 |
| Wizzup | to be expected of course | 01:30 |
| Wizzup | uvos: I don't think the resets happen so often that it we should not move the kernel to -devel, what do you think | 14:13 |
| Wizzup | freemangordon: do you need me to keep my droid on 1.9 ddk any longer or shall I migrate it? | 14:13 |
| freemangordon | Wizzup: I was just thinking about if you can gather any more useful info | 14:15 |
| freemangordon | btw, with ddk 1.9, do you see the same tearing as win 1.17? | 14:16 |
| freemangordon | it is easily visible in portrait | 14:16 |
| freemangordon | s/win/with | 14:17 |
| Wizzup | let me check | 14:24 |
| Wizzup | I don't think I see tearing with 1.17 in portrait mode | 14:25 |
| Wizzup | at least not when scrolling h-d | 14:25 |
| Wizzup | but I didn't move to triple buffering yet | 14:25 |
| Wizzup | freemangordon: ^ | 14:26 |
| uvos | there is tearing in 1.17 | 14:27 |
| uvos | its extreamly obvious | 14:27 |
| uvos | its just the other way than you expect | 14:27 |
| uvos | because the display is scanned from left to right | 14:27 |
| uvos | Wizzup: i think moveing it to devel is fine | 14:28 |
| uvos | devel is not stable by definition | 14:29 |
| Wizzup | uvos: I only see stutters | 14:29 |
| Wizzup | no tearing | 14:29 |
| Wizzup | I still have this: | 14:30 |
| Wizzup | # cat /etc/powervr.d/hildon-desktop.ini | 14:30 |
| Wizzup | [hildon-desktop] | 14:30 |
| Wizzup | WSEGL_UseHWSync=0 | 14:30 |
| Wizzup | (probably not relevant, but eh) | 14:30 |
| uvos | i have that too | 14:32 |
| uvos | Wizzup: http://uvos.xyz/maserati/tearingBackground.png | 14:33 |
| uvos | set that as your background | 14:33 |
| uvos | if you dont see tearing then (in portrait mode) ill belive you :P | 14:33 |
| bencoh | :) | 14:34 |
| Wizzup | yeah ok there is some | 14:35 |
| Wizzup | I guess I should try to set it on 1.19 as well then | 14:35 |
| Wizzup | freemangordon: yes I also see tearing on 1.9 | 14:37 |
| Wizzup | uvos: did you try to run cloudgps btw? | 15:13 |
| uvos | Wizzup: no | 15:44 |
| uvos | i wanted to | 15:45 |
| uvos | but could not install it | 15:45 |
| Wizzup | should be in the repo now | 15:46 |
| Wizzup | https://twitter.com/maemoleste/status/1465697460043464707 :) | 16:01 |
| tmlind_ | nice :) | 16:09 |
| tmlind_ | Wizzup: did you get the tsc2005 idled? i'm kind of behind on mails and irc.. | 16:10 |
| Wasmachineman_NL | what is this I heard about a Maemo-Leste IRC :^)? | 16:10 |
| Wizzup | tmlind_: I ported it to the pm runtime framework, but it doesn't idle yet | 16:11 |
| tmlind_ | ok. also, there should have been some better infrastructure merged to the input subsystem for idling devices earlier this year, sorry don't know the details, just recall seeing something about it | 16:11 |
| Wizzup | tmlind_: I'm not any further than what I wrote in my email, wondering if I need to look at suspending the irq or something | 16:11 |
| uvos | tmlind_: the problem isent ideling the interface persay | 16:12 |
| Wizzup | tmlind_: the other problem I ran into with regards to idle tests (I have a list on the github issue) is that the n900 panics and resets quite often (like the boot problem) | 16:12 |
| Wizzup | Wasmachineman_NL: yes, we're all here :) | 16:12 |
| tmlind_ | Wizzup: ok weird | 16:12 |
| uvos | tmlind_: the stuff merged was about force ideling interfaces iirc | 16:12 |
| uvos | tmlind_: its about it blocking omap idle while not even open | 16:12 |
| uvos | ie just probed | 16:12 |
| Wasmachineman_NL | Wizzup even on stock Maemo my N900 crashes and reboots every once in a while lol | 16:12 |
| tmlind_ | uvos: ok, do you have the commit id for that patch? | 16:13 |
| Wizzup | this is the tsc2005 one https://github.com/maemo-leste/droid4-linux/commit/d43f08d6b99d142a28a3dd97481ce3adec0e7638 | 16:13 |
| uvos | tmlind_: not off hand | 16:13 |
| uvos | tmlind_: i also just remember the discussion on the ml not if it was merged even | 16:13 |
| Wizzup | Wasmachineman_NL: mine doesn't really, after a long time of use, but I hope I can switch to leste in 2021q1 | 16:14 |
| Wasmachineman_NL | Wizzup i'm impatiently awaiting the Pro1 M-L port | 16:14 |
| tmlind_ | uvos: ok | 16:14 |
| Wasmachineman_NL | until then, still using my ye olde N900 I bought in early 2015 | 16:15 |
| Wasmachineman_NL | holy shit imagine this, 2015 is almost 7 years ago | 16:15 |
| Wizzup | Wasmachineman_NL: that's probably some time out still, we're focussing on droid4/n900/droid3/bionic atm | 16:15 |
| Wasmachineman_NL | I gathered | 16:15 |
| Wizzup | parazyd has my fx tec pro | 16:15 |
| Wasmachineman_NL | I was one of the first people to get a Pro1 and I hate it | 16:15 |
| Wasmachineman_NL | mostly because Android | 16:16 |
| Wasmachineman_NL | FUCK android. | 16:16 |
| Wizzup | tmlind_: in any case I think we're pretty close to idling on n900, the omap3-isp seems to block idle as well mostly | 16:16 |
| Wizzup | here's the list https://github.com/maemo-leste/bugtracker/issues/545#issuecomment-980458784 | 16:16 |
| Wasmachineman_NL | (says the one who's had Android running on his Xperia X1 in 2010 lol) | 16:16 |
| Wizzup | tmlind_: if I blacklist omap3-isp, then the n900 panics usually right after everything is loaded/started | 16:17 |
| Wasmachineman_NL | omap3-isp sounds like something for the image processing sensor | 16:17 |
| Wizzup | (and the panic is in the idle code somewhere, same as the random boot problem) | 16:17 |
| Wizzup | Wasmachineman_NL: yes, iiuc | 16:17 |
| uvos | havin omap3-isp loaded is pretty optimistic no? | 16:17 |
| Wasmachineman_NL | why is the ISP being used in idle anyway? | 16:17 |
| uvos | what do we need it for | 16:17 |
| Wizzup | uvos: it's just in the defconfig I think | 16:17 |
| Wizzup | Wasmachineman_NL: well it's complicated, we don't use it now, but the module is loaded, and when it is loaded it should not block phone idle / use power | 16:18 |
| tmlind_ | Wizzup: if isp is really needed, this might be related to the beagleboard timer issue | 16:18 |
| Wasmachineman_NL | Wizzup so when it's loaded it causes excessive idle power use | 16:19 |
| Wasmachineman_NL | got it. | 16:19 |
| Wasmachineman_NL | and when it's unloaded it freaks out the OS causing it to crash and burn | 16:19 |
| Wasmachineman_NL | or rather, reboot | 16:19 |
| Wizzup | tmlind_: how quickly does this manifest on the beagleboard? | 16:19 |
| Wasmachineman_NL | ? | 16:19 |
| tmlind_ | Wizzup: well beagleboard did not boot, sounds like this is something else actually | 16:19 |
| Wizzup | Wasmachineman_NL: there's various problems with the latest kernel for the n900 atm, this is just one of them | 16:19 |
| Wizzup | tmlind_: yes, the idle oops is the same as during boot | 16:20 |
| Wizzup | but if I don't block omap3isp later on, the system doesn't panic, but it also doesn't idle, so yeah... | 16:20 |
| Wasmachineman_NL | does liberachat webIRC support custom themes? | 16:20 |
| Wizzup | (it might still crash on boot) | 16:20 |
| Wizzup | Wasmachineman_NL: sorry, don't know, you might want to get a better irc client anyway, I just linked it from the tweet for people to find us easily | 16:21 |
| Wasmachineman_NL | that's where I came from :^) | 16:21 |
| Wizzup | I know | 16:21 |
| Wasmachineman_NL | might break out one of my Precisions and shitpost- I mean IRC from there | 16:21 |
| tmlind_ | Wizzup: so do you need to do anything special on n900 to make it oops? or just let it sit idle? | 16:22 |
| Wizzup | tmlind_: for me just booting with the defconfig makes it oops every now and then. I can plug in my serial now and see how quickly/often I can make it crash | 16:24 |
| Wizzup | It's not entirely reproducible, but happens frequently enough for me | 16:24 |
| tmlind_ | Wizzup: ok i'll leave 5.15.2 kernel running idle for a while, i don't really have any modules loaded though | 16:24 |
| Wizzup | I don't think that causes trouble for me, oddly enough | 16:25 |
| Wizzup | I also have had that idle for a while in /bin/bash with just panel loaded | 16:25 |
| Wizzup | (and off mode enabled) | 16:25 |
| Wizzup | the emails in "Oops while booting 5.15.2 on Nokia N900" contain the traces, if they are of any us | 16:26 |
| Wizzup | use* | 16:26 |
| tmlind_ | yeah ok | 16:26 |
| tmlind_ | freemangordon: no idea about the 6ms flip :) | 16:26 |
| Wizzup | on the upside the droid3 hangs (that I had all the time on 5.11) are gone mostly it seems | 16:27 |
| Wasmachineman_NL | bit of a stupid question: are Kernel-Power or overclocking a thing with Maemo-Leste? | 16:29 |
| uvos | i run the droid at 1.4GHz | 16:30 |
| uvos | beond that no not really | 16:30 |
| Wizzup | uvos: jesus | 16:30 |
| Wizzup | :D | 16:30 |
| Wizzup | Wasmachineman_NL: power kernel is not a thing, mainline linux is more power than power kernel ever was | 16:30 |
| Wizzup | Wasmachineman_NL: currently we are trying to make everything work reliably, like the phone not resetting at night or in your pocket, once that is all going well, someone could look at overclocking, but not currently | 16:31 |
| Wasmachineman_NL | interesting. | 16:31 |
| uvos | i also ran the memory card interface at 50MB/s for a while - need to make that a nicer patch one of these days | 16:31 |
| tmlind_ | uvos: if you make a patch, please do it for the sdhci-omap.c, just update the dts to use sdhci compatible for your mmc to test | 16:34 |
| tmlind_ | Wizzup: the omap3_l3_app_irq means there was some invalid access somewhere, in this case it looks like the read instruction was not valid? | 16:35 |
| tmlind_ | that sounds like memory problem.. maybe run memtester on that device? | 16:36 |
| tmlind_ | uvos: sdhci-omap assuming you're already on v5.16-rc :) | 16:36 |
| Wizzup | tmlind_: I think freemangordon had the same problems on his n900 wrt it not always booting, but he doesn't have the serial to confirm the problem | 16:37 |
| Wizzup | https://dpaste.com/8H8XKFN7G this one just happened while booting | 16:39 |
| Wizzup | sorry, raw here: https://dpaste.com/8H8XKFN7G.txt | 16:40 |
| Wizzup | the 5.1 kernel does not suffer from this problem AFAICT | 16:40 |
| Wizzup | still, I can run memtester if you'd like | 16:41 |
| Wizzup | (looks like the same as in my first email) | 16:42 |
| Wizzup | happened 1 out of 4 times atm | 16:43 |
| Wizzup | I assume we're not missing some errata in the config... | 16:43 |
| Wizzup | we have ARM_ERRATA_430973 in 5.15, but not in 5.1, weird.. | 16:45 |
| tmlind_ | yeah a missing errata handling could explain something like that | 16:46 |
| tmlind_ | or applying a wrong errata too i guess | 16:47 |
| Wizzup | a bunch are enabled looking at .config https://dpaste.com/8TPGCNKLA.txt - but this just omap2plus_defconfig | 16:51 |
| Wizzup | btw: 5.1 did have 430973 applied, but only that one | 16:51 |
| Wizzup | I guess I could disable all except 430973 but that might give problems for the droid4 and others, but just to see what happens | 16:55 |
| Wizzup | I'm a bit doubtful this is the problem though, but can try | 16:55 |
| tmlind_ | Wizzup: see also this old thread: https://linux-arm-kernel.infradead.narkive.com/Esh7UIgf/arm-errata-430973-on-multi-platform-kernels-was-omap3-n900-add-microphone-bias-voltages#post2 | 16:59 |
| Wizzup | tmlind_: so we have this in .config of the kernel that has this problem | 17:00 |
| Wizzup | but I guess u-boot also has it, but that should not be a problem according to that thread | 17:01 |
| tmlind_ | i forgot the details, but i thought we got it all fixed up earlier | 17:01 |
| Wizzup | right, it looks like both 5.1 and 5.15 have this applied and 5.1 doesn't seem to have the problem, but our build of that kernel *only* has 430973 applied, our 5.15 has 7 erratas applied, the majority for A9 and A15 and one for an older model, and just one for A8 | 17:02 |
| Wizzup | when I say applied I mean that it's in the .config as =y | 17:03 |
| Wizzup | so I can remove them from the .config and rebuild, but I would be a bit surprised if the errata's for A9 are applied on A8 cpu | 17:03 |
| tmlind_ | yeah a9 errata are not applied for a8 | 17:04 |
| tmlind_ | Wizzup: so does your old working kernel say anything about the BPIALL workaround? | 17:06 |
| Wizzup | I tried to grep for that (BPIALL) and didn't find anything, so I'm not sure which commit or config you refer to | 17:07 |
| tmlind_ | well i wonder if all the spectre-v2 patches somehow made the issue reappear | 17:09 |
| Wizzup | ah | 17:09 |
| Wizzup | 5.1 and 5.15 both do seem to have CPU_SPECTRE=y in their .config, but I don't think that enables spectre workarounds per se | 17:10 |
| tmlind_ | yeah those patches are in v5.1 already | 17:11 |
| tmlind_ | for reference, here's a better link to the old arm errata thread: https://lore.kernel.org/linux-omap/20150403183942.GV10805@atomide.com/ | 17:16 |
| tmlind_ | bbl | 17:17 |
| Wizzup | bye | 17:17 |
| Wizzup | hmm I wonder if we need HSI_CHAR in .config to fix the n900 dma problem in nokia-modem | 17:18 |
| * Wizzup tries | 17:20 | |
| Wizzup | So the 5.1 kernel is quite different in options, it's also build as thumb2 kernel whereas I don't think we build our 5.15 as thumb kernel | 17:23 |
| Wizzup | it even had SMP disabled | 17:24 |
| bencoh | SMP shouldn't affect n900 much | 17:25 |
| bencoh | well, it might, but ... | 17:25 |
| Wizzup | probably not, just looking at some obvious differences | 17:25 |
| bencoh | thumb2 on the other hand .... | 17:25 |
| Wizzup | yeah | 17:25 |
| bencoh | there were two thumb2-related (iirc) errata for n900 | 17:26 |
| bencoh | and I think one of the fix belonged to kernelspace | 17:26 |
| bencoh | so you should probably check it's enabled | 17:26 |
| bencoh | ah right, maemo.org is down :( | 17:27 |
| Wizzup | I think we went over the erratas earlier, but maybe I missed something | 17:27 |
| bencoh | just in case: https://webcache.googleusercontent.com/search?q=cache:gDfkAwFC5P8J:https://wiki.maemo.org/Community_SSU/Thumb | 17:28 |
| Wizzup | I think the latter (65 something) is worked around in gcc/ld etc, so we should have that fixed in all modern userspace | 17:35 |
| bencoh | indeed | 17:35 |
| Wizzup | what I can try to do is pick all of spinals patches and config changes to 5.1 and apply them to 5.15 | 17:35 |
| bencoh | the question is whether 5.15 includes a fix for the first one | 17:36 |
| bencoh | and whether it is enabled | 17:36 |
| Wizzup | yes to both @ 430973 | 17:36 |
| bencoh | ah, alright :) | 17:36 |
| Wizzup | 5.1 also has SLUB as allocator, not SLAB | 17:38 |
| Wizzup | to bad there's such a big difference hehe | 17:38 |
| Wizzup | sicelo: do you want to try apt dist-upgrade on beowulf-experimental ? | 17:42 |
| Wizzup | sicelo: hm, wait, I am not sure if the old packages are being removed now as well, I'll have to check that | 17:43 |
| * freemangordon wonders what has happened with Tomi | 17:44 | |
| Wizzup | sicelo: yeah hold off for now | 17:45 |
| Wizzup | sicelo: well whatever I did, I broke it atm, so I'll need to revert some stuff | 17:58 |
| Wizzup | I don't know why -experimental doesn't want to dist-upgrade to new pvr anymore, sigh | 19:21 |
| _inky | i wish i could buy every maemo leste a pinephone. not only because i own one, seriously. but because pinephone is a device that attracts people to mobile linux. and maemo could have more share, therefore, more developers. | 22:29 |
| _inky | i like hildon desktop so much. it is so underappreciated. | 22:29 |
| _inky | it is very useful when comparing to kde plasma or phosh. very. | 22:29 |
| Wizzup | mhm | 22:31 |
| Wizzup | we'll get there. | 22:31 |
| Wizzup | the keyboard addon will help loads I think | 22:32 |
| uvos | but, but we like our strange, rare, near obsolete nokia/motorola devices | 22:32 |
| Wizzup | why not both | 22:34 |
| uvos | well i dont disagree that we should try and have the pp be supported as a first class citizen, not sure how to improve the situation tho | 22:36 |
| Wizzup | what situation? | 22:36 |
| Wizzup | I spoke with tl lim in the very early days and he was up for doing a run with leste as well, but especially then we weren't ready | 22:36 |
| uvos | pp lagging in implemented features and stability compeard to mapphones. | 22:37 |
| Wizzup | either someone will come along, or we'll do it when mapphone is usable | 22:43 |
| uvos | speaking of lagging in features | 22:44 |
| uvos | https://github.com/maemo-leste/leste-config/pull/26 <-> https://github.com/maemo-leste/libmatchbox2/pull/8 these 2 want review | 22:45 |
| uvos | i accidentaly "updated" h-d today and now gnome-contacts is broken again ;( | 22:46 |
| uvos | Wizzup: btw what kind of power managament are you haveing on d4 atm with 5.15 | 22:48 |
| uvos | im having a hard time getting under 200mW its probubly just something misbehaving on this unit as the bionic still idles fine at 90 ish or sometimes lower | 22:49 |
| Wizzup | uvos: I'd have to check but it seems not too bad really | 23:04 |
| Wizzup | I think easily 20h with modem on | 23:04 |
| Wizzup | also haven't seen a lot of resets now | 23:04 |
| Wizzup | uvos: check @ PRs, will try to check later tonight | 23:07 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
