sicelo | so when the modem's gone, even reloading the drivers that seem to be related does not cause it to reappear | 09:47 |
---|---|---|
sicelo | i *think* it completely disappears off USB. didn't find time to try and toggle the USB bus directly. should try that next time | 09:48 |
uvos | yeah try resetting the bus | 10:56 |
uvos | if that dosent work there are really only two options: 1. android dosent suffer from this problem, somehow 2. android can reset the modem via tty or via some gpio pin | 10:59 |
Wizzup | I did manage to get it back on usb btw | 11:20 |
sicelo | mine didn't. weird. and this time i'm almost sure it's not power related. the device was on charger | 11:22 |
sicelo | but i think we also need to find out why the icon is stuck on showing the bogus bars. modem is absent in ofono, so i guess there's a ModemRemoved signal we're not handling properly | 11:23 |
uvos | sphone will also have this problem | 11:24 |
uvos | it assumes the modem will stay for its lifetime | 11:24 |
uvos | it allso wont realise when its back, except by accident, since its name will likely be the same | 11:24 |
sicelo | i guess if the icon shows modem's gone, no one will mind too much that sphone doesn't know it is | 11:27 |
uvos | sure, its still a bug | 11:28 |
sicelo | yeah. i'll have a look at the icon thing sometime (but want to finish with the N900 ofono thing first) | 11:28 |
Wizzup | sicelo: how did you try to bring it back? | 11:32 |
Wizzup | uvos: yeah having that fixed would be nice. I think with the tp module this problem is gone at least | 11:32 |
Wizzup | (there might be other problems :p) | 11:32 |
sicelo | i tried to unload a couple of drivers i think are related, such as qmi_wwan, cdc_* | 11:33 |
Wizzup | so | 11:34 |
Wizzup | let me see if I can help | 11:35 |
Wizzup | uvos: from /sys/bus/platform/drivers, I think we want to reset omap-usb2 yeah? | 11:36 |
Wizzup | see ls /sys/bus/platform/drivers | grep usb | 11:36 |
uvos | probubly, but im not not sure about the topology off hand | 11:37 |
Wizzup | I'm trying to see if this can just be done via bind/unbind | 11:38 |
Wizzup | so /sys/bus/platform/drivers/phy-mapphone-mdm6600 can be bound/unbound but that is on a too high level | 11:40 |
Wizzup | unbinding cpcap-usb-phy.0 makes the kernel unhappy | 11:40 |
Wizzup | and unbinding 4a064800.ohci from ohci-platform makes the device very unhappy ;) | 11:42 |
Wizzup | sicelo: in any case I think the idea is to 'powercycle' the usb layer that the modem is on | 11:42 |
Wizzup | sicelo: ok.. | 11:46 |
Wizzup | try this: | 11:46 |
Wizzup | echo 4a064800.ohci > /sys/bus/platform/drivers/ohci-platform/unbind ; sleep 5 ; echo 4a064800.ohci > /sys/bus/platform/drivers/ohci-platform/bind | 11:46 |
sicelo | i'll do that next time, yes. for the current issue (which was from yesterday, ended up rebooting) | 12:12 |
sicelo | in the meantime, i'll make a DCEW widget that checks if ofono still sees modem | 12:50 |
Wizzup | sicelo: the status code should not be too hard to fix for this | 12:54 |
Wizzup | I imagine I just didn't monitor for the modem disappearing and reappearing | 12:54 |
Wizzup | probably worth fixing anyway | 12:55 |
sicelo | yes, i looked casually in the past ... there are some mentions of ModemAdded and ModemRemoved, but I didn't go as far as checking how/if it's handled | 12:55 |
Wizzup | there is code for modem_removed_cb and modem_added_cb | 12:57 |
norayr | in postmarketos podcast episode from 2022 the phosh or phoc developer says that he was using neo freerunner and then n900 till the time he started to work for librem on phosh. not leste, but n900 with maemo. | 15:30 |
Wizzup | makes sense | 15:38 |
tmlind | uvos: did you figure some solution to the pte_offset_map_lock() no longer being inlined to build pvr modules? | 17:21 |
tmlind | actually __pte_offset_map_lock() no longer being inlined i mean | 17:22 |
tmlind | hmm maybe follow_pte() can be used instead, will check | 17:41 |
freemangordon | Wizzup: uvos: we have to set some env vars on gles2 devices for glimagesink to work, see https://gstreamer.freedesktop.org/documentation/gl/gstgldisplay.html?gi-language=c#GstGLDisplay | 18:29 |
freemangordon | GST_GL_PLATFORM=egl GST_GL_API=gles2 | 18:30 |
freemangordon | I am not sure what is the proper way to do that - per device seems an overkill to me | 18:30 |
freemangordon | do you know some standard way to detect if GLX/destop GL is supported? | 18:31 |
arno11 | Wizzup: sicelo: i made another PR for libcmtspeechdata: i found new values to improve/control latency. Now latency is really stable and similar for 2G and 3G. | 18:36 |
Wizzup | great | 18:38 |
arno11 | sound was sometimes a bit distorted when the latency was too low | 18:38 |
arno11 | now latency stays @100-150ms | 18:39 |
Wizzup | cooking atm, will merge in 30-45mins | 18:40 |
arno11 | cool thx | 18:40 |
arno11 | IMO calls are 98% working now :) | 18:44 |
uvos | tmlind: no all i had to do is fix the ps_vma->vm_flags acesses to use the new vm_flags_set function | 18:44 |
uvos | after that it compiled and worked perfectly fine | 18:44 |
uvos | this was ontop of 6.6-rc4 | 18:44 |
uvos | freemangordon: the only way to know if dekstop gl is supported is to try and create a context unfortionatly | 18:45 |
uvos | i gues we could write a tiny utility that dose that and sets the envvar | 18:46 |
uvos | tmlind: ok thats not true | 19:03 |
uvos | i examined the maemo-6.6 tree and these are the changes realted to pvr: | 19:04 |
uvos | https://github.com/IMbackK/droid4-linux/tree/6.6-rc4 | 19:04 |
uvos | so i just exported that symbol | 19:04 |
Wizzup | arno11: perfect, I need to set up another n900 because my dev n900 screen is unhappy | 19:14 |
Wizzup | freemangordon: we can do this in /etc/profile.d | 19:14 |
Wizzup | freemangordon: per device | 19:14 |
Wizzup | like /etc/profile.d/qt-maemo-sh which we have for each device | 19:15 |
Wizzup | freemangordon: so just make /etc/profile.d/gst-maemo.sh(.leste) and ensure we have the right value for the device(s) | 19:15 |
arno11 | Wizzup: ah ok | 19:24 |
Wizzup | arno11: but can merge in the meantime | 19:54 |
arno11 | Wizzup: oh i forgot i made another PR for Pinephone UCM few weeks ago | 19:58 |
arno11 | earpiece default volume is too loud according to teyrok from TMO | 19:59 |
arno11 | the fix has been already tested by him | 19:59 |
arno11 | (the PR is in leste-config) | 20:03 |
Wizzup | ok, you made it now? :) | 20:18 |
arno11 | ? lol nope sept 18 | 20:19 |
freemangordon | Wizzup: how did you manage to build mawf-gst-renderer? | 20:20 |
Wizzup | arno11: ok, let me try to find it all | 20:22 |
Wizzup | freemangordon: uh | 20:22 |
Wizzup | let me check | 20:22 |
Wizzup | the amd64 one is failing? | 20:22 |
freemangordon | arm64 as well | 20:23 |
freemangordon | on make check | 20:23 |
Wizzup | freemangordon: probably disabled tests | 20:23 |
freemangordon | how? | 20:23 |
freemangordon | like, I am using the same rules file | 20:23 |
Wizzup | debian/rules: hacky way to disable tests in CI | 20:23 |
Wizzup | do you have that commit? | 20:23 |
Wizzup | dd9c1a0532ac0ffa900fa9177d8e10150a03f70e | 20:23 |
* freemangordon checks | 20:23 | |
Wizzup | arno11: do you perhaps have a list of MR/PRs for me to check? | 20:24 |
freemangordon | Wizzup: argh | 20:24 |
freemangordon | maemo/chimaera was not pushed to master :( | 20:25 |
arno11 | Wizzup: yep let me check | 20:25 |
freemangordon | still, why it is failing on 64bits | 20:26 |
freemangordon | here in VM there is no issue | 20:26 |
Wizzup | freemangordon: sorry | 20:26 |
freemangordon | anyway, I have to fix the tests | 20:26 |
freemangordon | but I hate when it does not fail here and fails on CI :) | 20:27 |
arno11 | Wizzup: https://github.com/maemo-leste/libcmtspeechdata/pull/2/commits/3a78c894b630e1ef1cfbd3a2abaeb3ce9e0684a6 | 20:32 |
arno11 | https://github.com/maemo-leste/libcmtspeechdata/pull/2/commits/2806096990b2cb86c088aa368c033d066eac6664 | 20:32 |
arno11 | https://github.com/maemo-leste/libcmtspeechdata/pull/2/commits/ee953a1d7b8fc448925ab8beb93405d0d96c4420 | 20:32 |
Wizzup | freemangordon: might very well be the host kernel of the builder | 20:33 |
Wizzup | or some qemu specific bug like before | 20:33 |
arno11 | and the Pinephone one: https://github.com/maemo-leste/leste-config/pull/37/commits/f6ff97675845c8b8d3f7bd72a3aeb556f4b021a2 | 20:34 |
Wizzup | arno11: shall I do a new build for libmtspeechdata? | 20:35 |
arno11 | yes i think | 20:35 |
Wizzup | ok | 20:35 |
arno11 | thx man :) | 20:36 |
Wizzup | thank you :) | 20:36 |
freemangordon | hmm: https://phoenix.maemo.org/job/mafw-gst-renderer-binaries/architecture=arm64,label=arm64/11/consoleText | 20:48 |
freemangordon | "No factory for glimagesink!" | 20:48 |
Wizzup | arno11: libcmtspeechdata in repo has your fixes | 20:54 |
Wizzup | building leste-config now | 20:54 |
Wizzup | also for -devel | 20:54 |
freemangordon | arno11: patch that allows us to use hrtimer for IRTX on n900 was just pushed to upstream 6.1 | 20:57 |
freemangordon | it is also in 6.6 | 20:57 |
Wizzup | arno11: the chrt commands will still need to be integrated somehow yeah? | 20:57 |
freemangordon | so we shall either move to 6.6 or rebase our 6.1 | 20:57 |
Wizzup | I don't think we can move to 6.6 yet | 20:58 |
freemangordon | this https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/queue-6.1/drivers-clocksource-timer-ti-dm-don-t-call-clk_get_r.patch?id=cc8743ec5bb4cd059b3df5f787c658a570a8a6e3 | 20:59 |
freemangordon | Wizzup: sure | 20:59 |
arno11 | Wizzup: yes for chrt stuff, and it needs to be set after boot and once the related apps are loaded | 20:59 |
freemangordon | but we shall rebase our 6.1 | 20:59 |
Wizzup | ok | 20:59 |
arno11 | freemangordon: super cool ! | 20:59 |
arno11 | Wizzup: thx again | 21:00 |
tmlind | uvos: ok i'll take a look when i get a chance | 21:00 |
freemangordon | arno11: I will fix ir-rx51 until upstream decides on PWM atomic functions etc and pwm-ir-tx is fixed | 21:01 |
Wizzup | freemangordon: do you want me to add these env vars to leste-config? | 21:01 |
freemangordon | Wizzup: no, I want to avoid that | 21:02 |
Wizzup | I can set it for the n900 and mapphone devices | 21:02 |
Wizzup | why? | 21:02 |
Wizzup | It seems so much better than any kind of runtime detection? | 21:02 |
freemangordon | will try to implement runtime detection as uvos proposed | 21:02 |
Wizzup | firefox can't even get it right :) | 21:02 |
freemangordon | actually I think I can make it better ;) | 21:02 |
Wizzup | ok, seems like a lot of work for no reason, but up to you :P | 21:03 |
freemangordon | gst provides gst_gl_display_new_with_type() | 21:05 |
freemangordon | so I can try until it fails | 21:05 |
freemangordon | umm... succeeds :) | 21:05 |
arno11 | freemangordon: ok | 21:06 |
Wizzup | freemangordon: how will that go with llvmpipe glx? | 21:12 |
Wizzup | or the gl emulation that we have | 21:12 |
freemangordon | will try es2 first | 21:13 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!