Wizzup | freemangordon: hmm, I believe that if we pass all the hidden SSIDs as arguments to the dbus scan call, maybe the hidden ap stuff can work over dbus too | 00:50 |
---|---|---|
Wizzup | I think if we want to see why the cmd one fails vs the dbus one, we can look at these calls: | 00:52 |
Wizzup | ret = wpa_supplicant_trigger_scan(wpa_s, scan_params); | 00:52 |
Wizzup | in both wpa_supplicant/scan.c and wpa_supplicant/dbus/dbus_new_handlers.c | 00:53 |
Wizzup | could maybe use wpa_scan_clone_params | 00:54 |
freemangordon | Wizzup: hmm, not sure I understand what you wrote ^^^, but in order wpa_supplicant to 'see' hidden wlans one need to pass them in Scan method, IIRC. That's how I do it for a single ssid: https://github.com/maemo-leste/connui-internet/blob/076ff8117d805b1485f95b5ad6c3a87adae9ce53/src/settings/easy-wlan.c#L960 | 07:31 |
parazyd | Wizzup: The N900 wifi iface doesn't even go UP | 10:15 |
parazyd | Unrelated to wpa_supplicant | 10:15 |
parazyd | Downgrading wpasupplicant and rebooting doesn't help either | 10:28 |
stano_ | kernel:[15572.118377] cpcap_battery cpcap_battery.0: Battery empty at 3162mV, powering off what voltage should give me a low-voltage beep? | 10:29 |
stano_ | i'll add my own beep again, keep missing the low-voltage point to begin recharging for calibration :( | 10:32 |
parazyd | I've no idea what changed besides wpasupplicant | 10:37 |
rafael2k | hi | 10:49 |
rafael2k | now I'm completelly locked-out of my maemo on the pinephone | 10:49 |
rafael2k | no telephony and no wifi, with the latest update | 10:49 |
parazyd | In what sense? | 10:49 |
rafael2k | :/ | 10:49 |
parazyd | You should have usbnet | 10:49 |
rafael2k | latest icd something related to wifi ended breaking everything | 10:49 |
parazyd | rafael2k: In -devel, right? | 10:50 |
rafael2k | parazyd: yeap | 10:50 |
parazyd | ok | 10:50 |
parazyd | usbnet should work so you can ssh in | 10:51 |
rafael2k | tks, I'll use it | 10:51 |
parazyd | https://leste.maemo.org/Status/USB_Peripheral | 10:51 |
rafael2k | neat | 10:53 |
rafael2k | worked like a charm! | 10:53 |
rafael2k | having already ip set really helped - tks | 10:53 |
rafael2k | ; ) | 10:53 |
rafael2k | I compiled the kgx terminal in a debian sid chroot | 10:58 |
rafael2k | nice terminal, runs fine in maemo too | 10:58 |
rafael2k | it would compile on bullseye too (not on buster, unfortunatelly) | 10:59 |
rafael2k | btw, I see many people forcing GLES in the pinephone (for gstreamer playback, for example) | 11:02 |
rafael2k | There are certainly some showstopper bugs in OpenGL (not ES) implementation, this is why enabling X accel creates trouble | 11:04 |
uvos | there are shome showstopper bugs in glamors gles support too | 11:04 |
uvos | so idk if that helps us any | 11:04 |
uvos | stano_: the battery low notification is based on a charge percent, so it fails on d4 if its not calibrated | 11:05 |
Wizzup | freemangordon: adding the networks and then using 'scan' from the control interface also works | 11:07 |
rafael2k | uvos: right | 11:07 |
stano_ | ah ty uvos :) | 11:07 |
Wizzup | rafael2k: so -devel broke wifi for you? | 11:07 |
stano_ | i add a voltage based beep again | 11:07 |
rafael2k | Wizzup: yeap | 11:07 |
Wizzup | the nightmare continous | 11:08 |
Wizzup | at least in -devel it's OK to break wifi | 11:08 |
uvos | hehe | 11:08 |
rafael2k | :P | 11:08 |
rafael2k | I can bring it up "by hand" | 11:09 |
rafael2k | but the UI shows empty network list now | 11:10 |
freemangordon | what device is that? | 11:13 |
uvos | pp | 11:14 |
freemangordon | ok | 11:14 |
freemangordon | well, not "ok" , but still :) | 11:14 |
Wizzup | let me check my d4 | 11:20 |
Wizzup | rafael2k: does the kgx terminal have virtual keyboart support like osso-xterm does? | 11:23 |
Wizzup | ok, let's see | 11:26 |
Wizzup | ok, I made a mistake :) | 11:26 |
Wizzup | yeah, this is terrible, so because wpa_supplicant randomly returns failure on the d4 and pp even though it works (for scanning), we cannot rely on its exit state being proper | 11:28 |
Wizzup | I added a fallback, in case the control interface scan does not work, then scan using dbus (for the n900) | 11:29 |
Wizzup | but since we cannot rely on its result, I made it not report failure | 11:29 |
uvos | if it happens on pp too then we need to report this | 11:29 |
uvos | as thats a tottaly unrelated driver.... | 11:29 |
Wizzup | So many fun things to report to wpa_supplicant ;) | 11:29 |
uvos | yeah i thought the wierdness was just wlcore bing buggy | 11:30 |
Wizzup | I don't think so | 11:30 |
uvos | and wlcore is buggy but if it happens on pp too | 11:30 |
Wizzup | I don't know how to fix this anymore without breaking either hidden aps, or one of the devices | 11:30 |
uvos | then thats not it | 11:31 |
Wizzup | suggestions welcome | 11:32 |
rafael2k | Wizzup: no vkb in kgx | 11:32 |
parazyd | Wizzup: Perhaps https://iwd.wiki.kernel.org/ :p | 11:32 |
rafael2k | btw, debian kernel now supports pp | 11:32 |
Wizzup | probably doesn't even work on the n900 at all | 11:32 |
Wizzup | freemangordon: I think for our icd scan purposes we also want to scan for wildcard ssid, not just specific hidden ones | 11:33 |
rafael2k | (in unstable) | 11:33 |
Wizzup | parazyd: I'd rather not pull in more untested stuff tbh | 11:33 |
uvos | iwd is hardly untested | 11:34 |
uvos | but its support for really old stuff might be lacking yeah | 11:34 |
rafael2k | repackage debian's kernel instead of keeping our own seems a good option too | 11:34 |
Wizzup | parazyd: wext might get us out of this mess | 11:35 |
parazyd | Feel free to add it | 11:35 |
Wizzup | maybe we should just drop n900 support or something | 11:39 |
parazyd | lol | 11:40 |
Wizzup | really | 11:40 |
Wizzup | I don't have the energy for this stuff anymore atm | 11:40 |
* Wizzup bbl | 11:40 | |
parazyd | So just revert | 11:40 |
parazyd | Drop support for hidden aps | 11:41 |
uvos | dopping n900 seams a bit extreem rn, we might want consider it in the future | 11:41 |
uvos | but not now | 11:41 |
Wizzup | ok, pulled in a cross compiler over my 4g sim, let's see if I can build a kernel with wext | 11:55 |
Wizzup | was more data than I wanted to spend on it though ;) | 11:55 |
Wizzup | rafael2k: please update and restart icd2 | 12:00 |
Wizzup | rafael2k: let me know if it works again for you | 12:00 |
rafael2k | ok | 12:07 |
Wizzup | my n900 still doesn't have a static mac, weird | 12:09 |
mighty17[m] | well acc to downstream pwm_max_duty is 1200 which is weird | 12:21 |
mighty17[m] | and it is -ve somehow https://github.com/LineageOS/android_kernel_ti_omap4/blob/cm-14.1/drivers/video/omap2/displays/panel-ltn.c#L74 | 12:21 |
rafael2k | Wizzup: working again! | 12:22 |
Wizzup | once this is in stable we should probably tweet about it or send a mail to the ml, to inform our users | 12:28 |
Wizzup | tmlind: uvos: if I look at trying to port the n900 5.1 pvr to newer kernel, should I aim for something stable? | 12:39 |
Wizzup | since we seem to be settling on that as well for the d4 | 12:39 |
Wizzup | (or do we actually want someone to bisect the pvr traces for d4) | 12:40 |
uvos | i dont think forward porting n900s pvr is a good idea | 12:42 |
uvos | judging by what ddk1.9 dose its ugly and quite some work | 12:42 |
uvos | and any time is better invested in fixing glamor | 12:43 |
uvos | fixing 5.11 ddk1.9 on d4 brings us no benefit really | 12:43 |
uvos | 5.11 is eol as far as mainline support is concerned | 12:44 |
Wizzup | hm, ok | 12:46 |
uvos | really the glamor on ddk1.17 is a showstopper problem in lots of ways | 12:46 |
uvos | and we might need to drop everything and work on that | 12:47 |
uvos | since its keeping the n900 on an ainchent kernel with no pm and is starting to hurt the mapphones too | 12:47 |
uvos | and its causing lots of mainiance headaches | 12:47 |
parazyd | uvos: imho it'd be best to document the current state and bugs, because I'm not even sure where to start on this | 12:48 |
uvos | https://github.com/maemo-leste/bugtracker/issues/524 | 12:49 |
uvos | no one worked on this so this is up to date | 12:49 |
Wizzup | uvos: I think what would be good to add is what components to build, etc, to continue work on this | 12:51 |
uvos | right i just started writing sutch a comment for the bug | 12:52 |
Wizzup | parazyd: the n9xx linux refers to zimage, but I think we actually need uImage? | 12:53 |
Wizzup | parazyd: how do I make that, I don't see it in the rules file | 12:53 |
Wizzup | parazyd: is it LOADADDR=0x40008000 ? | 12:54 |
parazyd | Wizzup: 0ccb5d063c2a96490f666580747456fdb9ff1666 | 12:56 |
parazyd | git log --follow -p scripts/package/builddeb | 12:57 |
Wizzup | oh, it's appended dtb as well... | 13:00 |
parazyd | Yes | 13:00 |
Wizzup | "just add it yourself" surely is taking a lot of time now :-D | 13:00 |
parazyd | uImage is kernel+dtb | 13:01 |
parazyd | cat zImage omap3-n900.dtb > uImage | 13:01 |
parazyd | err | 13:01 |
parazyd | cat zImage omap3-n900.dtb > zImage | 13:01 |
parazyd | mkimage -A arm -O linux -T kernel -C none -a 80008000 -e 80008000 -n zImage -d zImage uImage | 13:02 |
uvos | https://github.com/maemo-leste/bugtracker/issues/524 | 13:06 |
parazyd | Thanks | 13:07 |
Wizzup | uvos: iirc you didn't want to go the hybris approach and suggested to fix the remaining bugs that fmg ran into, no? | 13:10 |
uvos | hybris? | 13:11 |
uvos | you mean hybrid? | 13:11 |
uvos | yes im wrting up the pvr+pvr path right now | 13:11 |
uvos | glamor-hybris is (might be) usefull with either path | 13:12 |
Wizzup | sicelo: CONFIG_WEXT works fine with scanning using wpa_cli | 13:12 |
Wizzup | parazyd: please enable it in kernel and rebuild, then we can fix all this stuff once and for all, until it breaks next time | 13:12 |
Wizzup | we will need to make sure that wpa_supplicant is started differently (via the dbus service) | 13:13 |
Wizzup | which is already a divert, so I don't know how that works | 13:13 |
Wizzup | thanks | 13:14 |
Wizzup | it just needs CONFIG_WEXT, nothing else | 13:14 |
freemangordon | Wizzup: so, n900+wext+hidden wlan support works? | 13:14 |
Wizzup | freemangordon: going to test hidden wlan now, but at least it's not insanely buggy like the other setup | 13:15 |
freemangordon | mh | 13:15 |
freemangordon | mhm | 13:15 |
uvos | https://github.com/maemo-leste/bugtracker/issues/524 | 13:18 |
Wizzup | freemangordon: nope, no working hidden ap with wext | 13:20 |
uvos | hehe | 13:20 |
Wizzup | freemangordon: well, connecting works | 13:22 |
Wizzup | if set up properly | 13:22 |
Wizzup | but the easy-wlan way didn't work for me | 13:22 |
Wizzup | maybe it's because I had other hidden wlans in gconf | 13:22 |
Wizzup | parazyd: any idea how we can modify the /usr/share/dbus-1/system-services/fi.w1.wpa_supplicant1.service to be n900 specific? | 13:23 |
Wizzup | it's already a divert | 13:23 |
parazyd | What package does it come from? | 13:26 |
Wizzup | freemangordon: I see this: Jul 16 13:27:35 localhost iap_conndlg 2.88+0m5[2718]: Unable to receive reply from wpa_supplicant! | 13:27 |
Wizzup | parazyd: I think libicd-network-wpasupplicant | 13:27 |
parazyd | Wizzup: Looks like you'd need some device identifier in the postinst | 13:36 |
Wizzup | we don't have that I think | 13:40 |
Wizzup | parazyd: how about a separate package only for this divert, and one can be for all devices, and one if for n900 component and replaces the one for all devices? | 13:41 |
parazyd | one sec | 13:44 |
parazyd | Yeah ok, that can work | 13:45 |
parazyd | Build it from the same git repo | 13:45 |
parazyd | Then in hildon-connectivity-wlan you can have a dep | 13:45 |
Wizzup | well, it would have the same source package | 13:45 |
parazyd | wpasup-divert-n900 || wpasup-divert-common | 13:45 |
Wizzup | ok | 13:46 |
Wizzup | well, I'll test it locally, but until we have wext we don't want to push it out anyway | 13:46 |
freemangordon | parazyd: hmm, wait, what about "provides:? | 13:53 |
freemangordon | so both wpasup-divert-n900 and wpasup-divert-common will "Provides: wpasup-divert" | 13:54 |
freemangordon | and hildon-connectivity-wlan should depend on wpasup-divert | 13:54 |
parazyd | That's just an additional stance and isn't deterministic | 13:54 |
parazyd | Because the common will be available on n900 too | 13:55 |
parazyd | You can work around this by: Depends: divert-n900 || divert-common | 13:55 |
parazyd | dpkg always picks the first available | 13:55 |
freemangordon | ah, ok | 13:55 |
parazyd | And since -n900 is only in the n900 component, this works | 13:55 |
freemangordon | ok, got it | 13:55 |
freemangordon | yeah, makes sense | 13:55 |
freemangordon | thanks for the lesson :) | 13:56 |
parazyd | np :) | 13:56 |
parazyd | We also already use this for bionic | 13:56 |
parazyd | In leste-config I think | 13:56 |
freemangordon | BTW, any progress with addressbook? | 13:56 |
parazyd | Not yet, I'm getting ready to travel, so next week I'll have time for IDA | 13:56 |
freemangordon | great | 13:57 |
parazyd | Also fixing up the pinephone now | 13:57 |
freemangordon | I am going to travel too, will be MIA till next month | 13:57 |
parazyd | *nod* | 13:57 |
freemangordon | Wizzup: re "Unable to receive reply from wpa_supplicant!" - maybe plugin scans while conndlgs wants to get hidden ap properties | 13:58 |
freemangordon | just rey once more | 13:58 |
freemangordon | *try | 13:58 |
freemangordon | maybe we shall increase the timeout or the number of retries | 13:59 |
freemangordon | OTOH if you see that only once, then it actually worked | 14:01 |
freemangordon | https://github.com/maemo-leste/connui-internet/blob/master/src/settings/easy-wlan.c#L1396 | 14:01 |
freemangordon | it tries a couple of times with timeout that's incremented by 500 ms for each try | 14:02 |
Wizzup | freemangordon: I saw it about four times | 14:18 |
Wizzup | freemangordon: anything you'd like to add to uvos' comments here? https://github.com/maemo-leste/bugtracker/issues/524 | 14:19 |
Wizzup | parazyd: alt. we could just not enable nl80211 on the n900 kernel | 14:48 |
Wizzup | then the wext fall back would work | 14:48 |
Wizzup | and we could do that until we fix nl80211 in wpa on n900 | 14:48 |
Wizzup | what do you think? | 14:48 |
parazyd | I'm fine with whatever you choose | 14:49 |
lel | MerlijnWajer opened an issue: https://github.com/maemo-leste/bugtracker/issues/559 (wpa_supplicant bugs to look into) | 14:49 |
Wizzup | I am not sure if it is possible to disable nl80211 in kernel but keep wext | 14:49 |
Wizzup | but if it is, that might be easier than splitting up the divert stuff | 14:50 |
parazyd | The divert isn't really difficult | 14:50 |
Wizzup | would you be up for doing it for me? | 14:52 |
parazyd | Sure, just give me the n900 contents | 14:56 |
parazyd | Wizzup: ^ | 15:01 |
Wizzup | just put wext before nl80211 | 15:02 |
Wizzup | in the wpa_supplicant line | 15:02 |
Wizzup | that's all really :) | 15:02 |
parazyd | ok | 15:02 |
Wizzup | freemangordon: I asked in #linux-wireless and there's someone asking about host mode on the n900, since he apparently did the kernel work way back for it :p | 15:03 |
stano_ | is there something i can tell cpcap or whatever calibrates battery that it has just booted from an empty battery and can start counting the total charging power? | 17:31 |
stano_ | coulombs, whatever | 17:31 |
uvos | stano_: no | 20:31 |
Wizzup | #linux-wireless is helpful so far with n900 wifi | 22:09 |
sicelo | Nice. | 22:52 |
sicelo | What are some interesting points? | 22:52 |
sicelo | I would like to see n900 wifi working properly in nl80211, even with iwd if possible | 22:53 |
sicelo | ah, the person who talked about n900 host mode must be Paul? Yes, he, Doc, and a few others were responsible for making it happen | 23:13 |
sicelo | Wizzup: if it's not too much to ask, please share the log of that wireless chat | 23:22 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!