uvos | rafael2k: sphone works fine here | 00:31 |
---|---|---|
uvos | rafael2k: but i noticed that the default config file is wrong i forgot to load ui-calls-manager-gtk so you cant hangup a call (any call not just recents) | 00:31 |
uvos | rafael2k: ill fix that soon | 00:31 |
uvos | rafael2k: in the mean time you can just add the module to your sphone config file | 00:32 |
freemangordon | Wizzup: if we don;t want daemon to restart on upgrade, there is no need for hacks etc | 07:50 |
freemangordon | see https://manpages.debian.org/testing/debhelper/dh_installinit.1.en.html | 07:51 |
freemangordon | see -r | 07:51 |
freemangordon | hmm, dsme already does that | 07:59 |
freemangordon | https://github.com/maemo-leste/dsme/commit/2c7d8cf4a83a64a9a605c9822f8325231737de28 | 07:59 |
freemangordon | maybe I am missing the issue | 08:00 |
rafael2k | hey, sorry, ended up sleeping yesterday | 08:12 |
rafael2k | uvos: tks (for sphone info - I will try it out) | 08:12 |
rafael2k | Wizzup: my pp keyboard should arrive in the next days | 08:13 |
rafael2k | Wizzup: sound is still clogged with pa here, pa is shit, I'll check mobian setup | 08:13 |
humpelstilzchen[ | oh can we go without pa? :) | 09:43 |
rafael2k | humpelstilzchen[: I'd like too, but some software rely on it, like sphone and other I think | 09:58 |
humpelstilzchen[ | rafael2k: I wonder if apulse emulation helps | 09:59 |
rafael2k | lets try to fix pa first... seems to be some detail | 10:00 |
humpelstilzchen[ | oh, dead project. What a pitty | 10:00 |
sicelo | keep pa. you probably just need a simple config or sth. | 10:07 |
Wizzup | freemangordon: I don't think we want to fork every daemon we don't want to restart, but yes, that is a potential solution for ofono and such | 10:14 |
Wizzup | humpelstilzchen[: rafael2k: maemo needs pa | 10:15 |
freemangordon | sure I didn;t mean to fork, but we shall fix our daemons we don't want to be restarted | 10:15 |
freemangordon | my point was more than dsme is already ok in that regard | 10:15 |
freemangordon | BTW, what happens if ofono cashes? | 10:16 |
freemangordon | *crashes | 10:16 |
Wizzup | not sure, either it stays down or it gets started again via dbus activation | 10:17 |
freemangordon | tmlind: hmm, I think omap ddx commit 478fcc45e0b9d93dbe1a2c1f842441af529a3c04 is the reason why we see dma_buf leaks | 10:20 |
freemangordon | "[PATCH] omap: Fix missing usage count decrease in OMAPDRI2DestroyBuffer" | 10:20 |
freemangordon | it decreases usage count, but doesn;t check if it become zero afterwards | 10:21 |
freemangordon | Wizzup: could you have a look too, please? | 10:21 |
Wizzup | I mean it sounds very sensible | 10:22 |
freemangordon | ok, will fix the logic to my understanding, lets see what tmlind will say | 10:22 |
rafael2k | Wizzup: ok! pa stays! | 10:23 |
freemangordon | :) | 10:24 |
rafael2k | Wizzup: I can get sound, but there is some samplerate mismatch... interesting that what triggered the issue was the more recent alsa driver | 10:24 |
rafael2k | some new kernel behavior... dunno yet | 10:25 |
freemangordon | Wizzup: I guess this is the reason for CMA failures on n900 too | 10:26 |
freemangordon | ugh, we have a leak on rotation too | 10:28 |
Wizzup | freemangordon: yes, I think so @ rotation | 10:31 |
freemangordon | but this is another leak | 10:31 |
Wizzup | ok | 10:36 |
sicelo | remind me again what's your lowest mW rate on Droid 4? with my battery script in background, i'm averaging 100mW atm. I hope/suppose that's not excessively bad | 11:09 |
sicelo | @4V ... so around 25mA. i think i'm okay with that :-) | 11:10 |
Wizzup | I get less I think | 11:16 |
sicelo | i should think so, yes. no doubt my script has an impact (i don't even know if using sleep and while loop in the script is a good idea for cpu) | 11:21 |
Wizzup | you can get it from upower | 11:23 |
tmlind | freemangordon: ok sounds reasonable | 12:10 |
tmlind | sounds like something else might be wrong too if the calls are not paired | 12:12 |
freemangordon | tmlind: it starts with refcount of 1 | 12:13 |
tmlind | sicelo: i'm seeing about 75mW idle with 3g networking active | 12:13 |
tmlind | freemangordon | 12:14 |
tmlind | so are you saying it should free at 1 then? | 12:14 |
freemangordon | I think so | 12:15 |
freemangordon | well, it should free at 0 | 12:15 |
freemangordon | because this works for both front and back buffers | 12:16 |
freemangordon | but for front buffers it was checking if it was non-zero *before* decrementing | 12:16 |
freemangordon | which can never be true obviously | 12:17 |
tmlind | ok | 12:18 |
freemangordon | tmlind: do you say that OMAPDRI2DestroyBuffer shall be called more than once? | 12:21 |
freemangordon | or, that someone else shall decrement too? | 12:21 |
tmlind | freemangordon: i recall only increment was added that never got decremented earlier | 12:26 |
freemangordon | yes, but see https://github.com/maemo-leste/xf86-video-omap/blob/master/src/omap_dri2.c#L300 | 12:30 |
freemangordon | what happens on the last decrement? refrcnt becomes zero and then nothing | 12:31 |
freemangordon | unless I am missing the whole point | 12:32 |
tmlind | yeah i guess no free then | 12:41 |
tmlind | uvos: here are few gps ntp time injection examples dumped from los 14.1 after inject from gpstest | 13:24 |
tmlind | ntp_time_ms elapsd rt/2 ??? ?????????? rt/2 | 13:24 |
tmlind | GpsInterface_inject_time( 1549911987370, 65652, 76 ) AT+MPDTIME=287,1291573419,76 | 13:24 |
tmlind | GpsInterface_inject_time( 1642918214110, 91427, 290 ) AT+MPDTIME=308,4105031436,290 | 13:24 |
tmlind | GpsInterface_inject_time( 1642939067175, 51203, 14 ) AT+MPDTIME=308,4124400839,14 | 13:24 |
tmlind | GpsInterface_inject_time( 1642940045748, 55924, 21 ) AT+MPDTIME=308,4125358112,21 | 13:24 |
tmlind | ntp_time_ms = ntp time from server in milliseconds | 13:24 |
tmlind | elapsd = elapsed real time for the ntp query | 13:25 |
tmlind | rt/2 = round trip time / 2 | 13:25 |
tmlind | anybody have any idea what the two unknown values for MPDTIME are? | 13:25 |
tmlind | the longer one seems to be some time in milliseconds | 13:26 |
tmlind | i think los 14.1 is buggy, it seems to query ntp only once on bootup, after that all the gps time inject commands use the old ntp value.. | 13:26 |
tmlind | so the GpsInterface_inject_time() values only get updated after a reboot :) | 13:27 |
tmlind | not sure what the elapsed real time really means, here are some java examples: https://www.javatips.net/api/android.net.sntpclient | 13:28 |
tmlind | if anybody wants to trace other modem related commands, these commands with los 14.1 will work: | 13:30 |
tmlind | adb root | 13:30 |
tmlind | adb shell 'echo 0x7fffffff > /sys/module/ts27010mux/parameters/debug_level' | 13:30 |
tmlind | adb logcat -b all | 13:30 |
buZz | MPD is 'mobile platform device' , gee, TIL | 13:30 |
tmlind | won't work with the stock android | 13:31 |
tmlind | buZz: ok so one mystery less then :) | 13:31 |
tmlind | that elapsed real time is claimed to be in ms units too in the java examples, but seems more like system uptime to me? | 13:32 |
tmlind | yeah i guess mCachedNtpElapsedRealtime in the examples is how long the system has been up when doing the query so system clock can be used to calculate ntp time offset later on | 13:35 |
dsc_ | im starting to integrate telepathy into conversations | 13:38 |
Wizzup | :) | 13:50 |
tmlind | so looks like the GpsInterface_inject_time() values are from android boot, and describe the offset to system clock and don't directly related to the MPDTIME values as those are somehow based on current time | 13:53 |
Pali | Hello! Same situation with N900 U-Boot port again happens... On ML are some patches waiting for more than half of year without answer and now are trying to remove N900 code again. | 15:20 |
Pali | Wizzup: I added you to CC | 15:21 |
sicelo | :-( | 15:34 |
Wizzup | Pali: yeah I need to work on the dm stuff | 15:50 |
Wizzup | will reply later today | 15:55 |
Pali | note that I have already WIP patches for dm-keyboard and dm-video | 15:56 |
Pali | I have not sent them yet because other N900 patches are waiting on the list for review (for 1/2 year) | 15:57 |
Wizzup | iirc this was about usb | 15:58 |
Pali | yes, now I recalled... | 16:00 |
Pali | there is missing dm-usb, dm-video and dm-keyboard | 16:00 |
Pali | and Tom started talking about dm-usb in dm-video thread | 16:01 |
Wizzup | yeah I volunteered for dm-usb | 16:08 |
reinob | I have a question regarding the flashing of u-boot replacing the maemo5 kernel. I'm reading https://leste.maemo.org/Nokia_N900 and see that I can flash u-boot-2013.04-2.bin, which will replace the existing u-boot/kernel with the new u-boot. But I don't know which default options (bootmenu.scr?) will be used. | 18:37 |
reinob | Can I assume that if something bootable (the leste image) is in the SD card it will boot? | 18:38 |
reinob | or do I need to store a bootmenu.scr on the SD card? | 18:39 |
reinob | (my ultimate goal would be to wipe mmcblk1 (Mydocs and /opt) and use it with Leste, so that booting only requires mtdblock* and the sd card itself) | 18:44 |
freemangordon | reinob: don;t you already have u-boot? | 18:53 |
freemangordon | if you have, there is no ned to flash u-boot | 18:53 |
freemangordon | *need | 18:53 |
reinob | yup. But I want it to be independent of mmcblk1. As far as I know currently it is reading bootmenu.scr from Mydocs (or is it just a copy, and the actual one is stored together with u-boot on mtd3, or whereever the kernel was?) | 18:54 |
freemangordon | unless I am missing something as I didn't really read that wiki page | 18:54 |
freemangordon | ah, I see | 18:54 |
freemangordon | Pali: ^^^ | 18:54 |
reinob | I'd like to at least format "Mydocs" as ext4 to use it from within Leste, but I fear that by doing this U-boot won't be able to boot anything. | 18:56 |
Pali | reinob: Hi! U-Boot reads bootmenu.scr from MyDocs. If it does not exist then it use default bootmenu (which is compiled in U-Boot). | 18:56 |
Pali | New version of U-Boot first reads bootmenu.scr from SD card. If it does not exist then fallback to MyDocs | 18:56 |
reinob | OK! so if I flash the new u-boot and prepare a suitable bootmenu.scr on the SD card (first partition? or where exactly), then it should be OK | 18:57 |
Pali | so if you want to be independent of eMMC, just move/rename bootmenu.scr from all eMMC partitions | 18:57 |
Pali | reinob: only first partition of SD card and only first partition of eMMC | 19:00 |
Pali | partition has to be either fat or ext2/3/4 | 19:00 |
Pali | relevant code: https://source.denx.de/u-boot/u-boot/-/blob/master/include/configs/nokia_rx51.h#L144-152 | 19:01 |
reinob | thx! | 19:01 |
reinob | BTW does overclocking work with the Leste kernel? I tried playing with scaling_max_freq but no dice.. | 19:02 |
freemangordon | no, it does not :) | 19:14 |
reinob | OK! :). I'll see if in the coming days I have some time to play with this.. and hope I can stay a bit in touch.. | 19:18 |
tmlind | heh i think i figured out the MPDTIME.. current_gps_time = ntp_time converted to gps_time based on system uptime, current_gps_time / 0xffffffff is the 308 part, current_gps_time % 0xffffffff is the remainder in ms | 19:54 |
* tmlind gives it a try | 19:55 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!