**** BEGIN LOGGING AT Sat Mar 19 02:59:58 2016 Mar 19 08:24:18 hi Mar 19 08:51:13 jonwil: hi! Mar 19 08:51:41 * jonwil wonders if any more progress has been made on CSSU/FPTF/etc lately... Mar 19 08:52:38 jonwil: there is lots of activity on the kernel lately, we're ironing the last remaining rough edges Mar 19 08:53:44 nice one Mar 19 08:54:07 Will be good if you can run a stock kernel on a N900 with a full Fremantle userland Mar 19 08:54:10 and have everything work Mar 19 08:54:36 hmm? I think stick kernel already runs on n900 :) Mar 19 08:54:41 *stock Mar 19 08:55:15 although it would also be good to see some more userspace work :) Mar 19 08:55:18 i think the word he meant is 'mainline' Mar 19 08:55:23 But does Fremantle run on that? Mar 19 08:55:28 sure Mar 19 08:55:32 ~fptf Mar 19 08:55:33 i guess fptf is the Fremantle Porting Task Force, see http://talk.maemo.org/showthread.php?t=91308 Mar 19 08:56:06 Maxdamantus: http://talk.maemo.org/showpost.php?p=1493038&postcount=158 Mar 19 08:56:41 Maxdamantus: right now I play with 4.5-rc6 Mar 19 08:57:44 jonwil: yeah, but you know - 2 hands and 24 hours per day only :) Mar 19 08:59:07 yeah I meant mainline kernel :) Mar 19 09:00:13 I also have no time to do any coding work right now due to this http://brisbanemodeltrainshow.com.au/ Mar 19 09:00:49 I am exhibiting as part of the LEGO layout at that event and I have a lot of building to do :) Mar 19 09:01:45 sounds fun Mar 19 09:03:04 yeah it will be :) Mar 19 09:05:19 hehe, playing with lego for moneys? Mar 19 09:05:26 now that's the right idea ;) Mar 19 09:11:27 We get no money out of that Mar 19 09:11:39 Everyone in the club does it for fun Mar 19 09:11:54 Although we do raise money for charitable organizations at various points across the year Mar 19 09:11:56 oh well, still the right idea Mar 19 09:13:59 The centerpiece of the work I am doing is a high-tech science lab (along the lines of Oscorp or Stark Mar 19 09:14:12 Stark Industries Mar 19 09:14:42 This lab is being attacked by Doctor Octopus and is about to be saved by Hulk, Spider-Man, Iron Man, Thor and maybe others :) Mar 19 09:14:50 along with a police SWAT team Mar 19 09:16:50 and then comes the train and flattens everything and everyone, choo-choo mother-funkers! Mar 19 09:28:20 added Neo900 to https://www.pyra-handheld.com/wiki/index.php?title=Comparison_Chart Mar 19 09:28:49 maybe n900 too? Mar 19 09:28:54 nah, N900 is too obsolete Mar 19 09:29:06 no, it's not Mar 19 09:29:08 :) Mar 19 09:30:14 pandora is 1ghz and 512mb, n900 is 600-900mhz and 256mb Mar 19 09:30:25 exactly Mar 19 09:30:34 and Pandora is only on there because it was the direct predecessor Mar 19 09:30:39 and once fmg and friends tidy up the last bits we will be not-obsolete again Mar 19 09:30:46 Neo900 just only matches Pandora's specs :P Mar 19 09:31:19 but you are right, memory is way too small for gaming Mar 19 10:35:44 Is the goal to be able to use a mainline kernel on the N900 with no local patches or are there local patches that just cant go into mainline? Mar 19 10:40:11 jonwil: there are also patches which cannot go to mainline Mar 19 10:40:23 e.g. sgx drivers Mar 19 10:40:24 Like what? Mar 19 10:40:39 Yeah good point Mar 19 10:40:49 or also wl1251 bt coes mode Mar 19 10:40:57 *coex Mar 19 10:41:07 or wl1251 perm mac address Mar 19 10:41:19 those were rejected Mar 19 10:41:42 what was wrong with that WiFi stuff exactly? Mar 19 10:41:44 no idea about tidspbridge (freemangordon knows more) Mar 19 10:41:51 can they run as a loadable module? Mar 19 10:42:00 jonwil: stupidity of maintainers Mar 19 10:42:32 I'd rather use a slightly different kernel tree than compile external modules. Mar 19 10:42:45 about bt coex, they wanted from me to create some nonsense interface in cfg/nl 80211 stack Mar 19 10:43:22 every driver has either module param or sysfs node (wl1271 has too sysfs node) Mar 19 10:43:44 but wl1251 cannot use what wl1271 use... Mar 19 10:43:48 * Maxdamantus already always has a Dvorak defkeymap in his tree. Mar 19 10:44:26 about setting permanent mac address, they wanted that kernel reads it Mar 19 10:44:57 but mac address is stored in CAL, so kernel would need to have CAL driver + access to correct NAND partition Mar 19 10:45:03 thats sounds stupid too Mar 19 10:45:36 can mac be set as a boot param? Mar 19 10:46:00 no Mar 19 10:46:18 this was rejected when raspbeery pi wanted to upstream this approach Mar 19 10:46:23 lol Mar 19 10:46:43 ehm.. this was rejected for rpi, not for n900 Mar 19 10:46:55 but I did not tried something like that for wl1251/n900 Mar 19 10:47:17 that way uboot might be the cure Mar 19 10:47:28 Sounds like the people who maintain the Linux WiFi driver stack are stupid for not seeing that there are WiFi chips out there that do things differently to the way most "normal" chips work Mar 19 10:47:29 as it has a bit more lax policy Mar 19 10:47:50 wl1271 has mac address stored in the first few bytes in firmware Mar 19 10:48:09 in firmware file, generated by some tool Mar 19 10:48:44 I was thinking about using request_firmware() interface too for wl1251, but request firmware and mac address separately Mar 19 10:50:02 but because udev project is dead, and systemd decided to drop support for run-time firmware files (files which are not phisically stored in VFS, but generated when kernel ask) I would mean that we need to write our tools for loading firmwares Mar 19 10:50:26 also wl1251 NVS firmware file needs to be generated on runtime (use some bits from CAL) Mar 19 10:50:47 pali: check eudev Mar 19 10:50:54 it's systemd free udev fork Mar 19 10:51:16 Pali: eudev isn't dead. Mar 19 10:51:28 I refuse to use a systemd OS Mar 19 10:51:28 KotCzarny: I can check, but such thing IIRC udev did not supported Mar 19 10:51:42 pali, still, you can probably send them a patch Mar 19 10:52:01 yes, but that means to write that code Mar 19 10:52:19 and for n900 case it would be easier to write specific tool (not part of udev) Mar 19 10:52:32 which would directly interact with CAL Mar 19 10:53:11 Luke-Jr: do you know if (e)udev has support for generating firmware by external application? Mar 19 10:53:41 Our IRC channel is #gentoo-udev on Freenode Mar 19 10:53:49 https://wiki.gentoo.org/wiki/Project:Eudev Mar 19 10:53:56 ok, I will ask Mar 19 10:54:01 Pali: I would imagine so. It's just udev Mar 19 10:57:38 anyway, we still needs to patch kernel, because it always try to load firmware directly from VFS without userspace interaction Mar 19 10:57:57 and when that direct-read fails, then it try userspace (via netlink) Mar 19 10:58:10 pali, can inkernel driver be modified to load special firmware which would be prepared on first boot by userland? Mar 19 10:58:51 KotCzarny: kernel loads firmware by request_firmware("filename") call Mar 19 10:59:26 and kernel first try to read "filename" from VFS and if it fails then it fallback asking for userspace Mar 19 10:59:26 so at first boot kernel load fails, system boots, userland prepares firmware, on reboot/module reload things work as usual Mar 19 10:59:40 ah right Mar 19 10:59:43 isn't there a firmware loader helper tool option in the kernel Mar 19 10:59:47 or was that deprecated Mar 19 10:59:49 I was thinking about it, but this is big hack Mar 19 11:00:09 pali, yup, blobs Mar 19 11:00:11 you need to overwrite firmware in /lib/firmware Mar 19 11:00:17 which is in UBIFS/NAND Mar 19 11:00:28 I will avoid such thing... Mar 19 11:00:39 but its first-boot time thing Mar 19 11:00:53 Wizzup: there is that userspace fallback Mar 19 11:01:15 either via netlink or sysfs Mar 19 11:01:24 (maybe sysfs was deleted) Mar 19 11:02:15 pali, how about generating random mac on boot (via some device related algo) and then changing it in userland later? Mar 19 11:02:22 that would be another option Mar 19 11:02:40 KotCzarny: permanent mac address cannot be modified Mar 19 11:02:48 from userspace Mar 19 11:03:00 kernel already generates random one Mar 19 11:03:14 and userspace set "temporary" mac address from CAL Mar 19 11:03:33 uhum Mar 19 11:03:41 this is what wl1251-cal tool is doing Mar 19 11:04:09 read mac from CAL and set it as temporary (spoofed) mac address Mar 19 11:04:09 then i guess first-boot firmware gen would be the best (apart from doing cal/nand read) Mar 19 11:04:49 I will wait for answer from #gentoo-udev Mar 19 11:05:37 if that is supported from udev, I will try it and prepare kernel patches to read mac address via request_firmware (like wl1271) Mar 19 11:15:25 jonwil: can you make yours wl1251-cal tool more verbose? Mar 19 11:15:33 like original nokia's one? Mar 19 11:15:42 Feel free to update the source code on CSSU, I have no time to do it Mar 19 11:15:54 ok Mar 19 11:16:01 then I will look at it Mar 19 11:16:28 jonwil: but is your doing 100% same thing as nokia's one? Mar 19 11:16:46 Its doing the same job but in a different way Mar 19 11:16:56 ok, that should be enough Mar 19 11:17:46 It gets the country to use from the CAL rather than the cellular modem Mar 19 11:18:40 It also uses lower level libcal calls rather than whatever higher level calls (sysinfod I think) the stock binary uses Mar 19 11:19:03 13:46:18] this was rejected when raspbeery pi wanted to upstream this approach Mar 19 11:19:04 what about putting it in dt? Mar 19 11:19:23 L29Ah: what put to DT? Mar 19 11:20:20 the mac address Mar 19 11:20:29 but mac address is not static Mar 19 11:20:34 what Mar 19 11:20:48 it is different for each n900... Mar 19 11:20:54 and is stored in CAL Mar 19 11:21:04 how you want to store it into DT?? Mar 19 11:21:12 generate dt with an ad-hoc tool Mar 19 11:21:18 dt overlay, even Mar 19 11:21:36 but wl1251 driver is loaded even before userspace start Mar 19 11:21:40 so what Mar 19 11:21:49 we have userspace tool which read mac from CAL Mar 19 11:22:06 and no, I'm not going to create CAL driver for kernel Mar 19 11:22:07 generate it once before using the kernel that requires it Mar 19 11:22:32 kernel boots, loads wl1251, start userspace Mar 19 11:22:51 kernel boots, gets its dt stuff, loads wl1251, start userspace Mar 19 11:22:52 kernel needs it before it can start tools for generating them Mar 19 11:22:59 you have a kernel now Mar 19 11:23:04 that doesn't need nothing Mar 19 11:23:09 the original one Mar 19 11:23:31 I do not understand how to want to solve this chicken and egg problem Mar 19 11:23:39 and anyway you can just bail out and say user to generate it in dmesg Mar 19 11:23:57 if he didn't do so earlier Mar 19 11:24:01 I do not want to add another hack Mar 19 11:24:06 also random mac would do :] Mar 19 11:24:10 we already have one with temporary addresses Mar 19 11:24:21 random permament mac is already uses Mar 19 11:24:34 I want to solve this problem in better way Mar 19 11:25:44 isn't the mac stored at some fixed offset of the mtd? Mar 19 11:25:51 no Mar 19 12:07:24 pali, first-boot solution then or eudev one Mar 19 13:13:26 Pali: anyway keep in mind Neo900 is using wl1851 aka WG7351 Mar 19 13:13:48 DocScrutinizer05: uff.. do we have driver for it? Mar 19 13:14:14 well, I bet we have, TI supports Linux drivers for all their WiFi modules afaik Mar 19 13:15:16 ok Mar 19 13:16:05 http://processors.wiki.ti.com/index.php/WL18xx Mar 19 13:16:16 jonwil: original nokia's wl1251 uses com.nokia.phone.net for country code Mar 19 13:16:26 this is missing in your implementation... Mar 19 13:16:48 yes, I mentioned my implementation uses cal rather than cellular modem Mar 19 13:17:11 goal was to make my clone dependent only on libcal and nothing else Mar 19 13:17:17 well also libnl Mar 19 13:18:09 what we _don't_ have is support for the advanced modes (injection etc) Mar 19 13:18:09 but not on dbus or libsysinfo or higher level stuff Mar 19 13:21:17 *if* somebody could come up with a http://intelligraphics.com/igx-wireless-presents-stealth-sniffer/ to RE, we could get Mar 19 13:22:47 TI cooperated with intelligraphics for those special firmwares and drivers Mar 19 13:23:52 jonwil: are you able to add that code which ask dbus for country code? Mar 19 13:24:49 no, I specifically wanted to use country code from cal only and not have dependency between wl1251-cal and cellular radio stuff Mar 19 13:24:51 Pali: btw Pyra uses same module ;-) Mar 19 13:25:45 jonwil: but this will break wifi connections for n900 which have country code too "restricted" stored in CAL Mar 19 13:26:25 country code will be set to too restricted even if you are in different country Mar 19 13:39:01 ((country code from CAL)) please don't! I'd have *lots* of fun with such stuff, >50% of my N900 are not exactly compatible in CAL with my location Mar 19 13:40:19 I'd rather prefer a quite user friendly setting for country Mar 19 13:43:51 (with right overriding left) CAL -> maemo settings "system language" -> maemo settings "Country" -> GSM country (if logged in) -> WLAN country info from associated AP -> **augmented settings** "WLAN country regulation domain" (with default) Mar 19 13:54:51 dang, wrong module Mar 19 13:56:21 WL1837MODGIMOC Mar 19 14:07:13 http://www.ti.com/product/wl1837mod Mar 19 14:13:43 http://www.digikey.com/product-detail/en/texas-instruments/WL1837MODGIMOCT/296-43667-1-ND/5864739 Mar 19 14:14:33 http://www.ti.com/lit/ds/symlink/wl1837mod.pdf Mar 19 14:15:12 Pali: ^^^ Mar 19 14:19:19 http://www.ti.com/product/WL1837MOD/toolssoftware Mar 19 14:20:49 http://www.ti.com/tool/wilink8-wifi-nlcp Mar 19 14:25:52 LOL >>Product of work done by tens of developers.<< Mar 19 14:27:31 TENS Mar 19 14:40:36 OVER9 Mar 19 14:41:08 000 laughs Mar 19 15:46:15 Pali: >> The WL1837MODCOM8I EVB includes the following features: [...] WLAN and Bluetooth , BLE, and ANT cores that are software- and hardware-compatible with prior WL127x, WL128x, and BL6450 offerings for smooth migration to device << Mar 19 15:48:16 http://www.ti.com/lit/ug/swru382a/swru382a.pdf Mar 19 17:53:18 Pali: http://paste.opensuse.org/6657073 Mar 19 17:54:30 great! Mar 19 19:08:35 jonwil: I updated wl1251-cal code Mar 19 19:08:56 but country code is still missing... Mar 19 19:11:57 Luke-Jr: it is not supported in gentoo udev Mar 19 19:13:28 eudev is also used in slack, and probably other systemdless distros Mar 19 19:15:25 i use devtmpfs because i'm a bad boy Mar 19 19:15:42 and also because this is a very tiny embedded system Mar 19 19:15:42 also Mar 19 19:16:00 but for dynamic things udev is still nice Mar 19 19:27:39 hey follow hacker gals and guys: >> The WiLink 8 device's WLAN capability to capture and register precise arrival time of the connected AP’s beacon is used to achieve ultra-precise synchronization between multiple connected audio devices.<< (from http://www.ti.com/tool/TIDC-WL1837MOD-AUDIO-MULTIROOM-CAPE ) This gives me some other ideas of possible usecases - indoor location etc comes to mind Mar 19 19:31:02 or other usecases with high precision sync requirements. e.g. electron flash snyc to camera, when both are controlled from same WLAN signal with a delay between trigger and shot in the range of a maybe 20ms so both can schedule a high precision timer event to exactly same point in time Mar 19 23:18:41 Sounds neat: https://en.wikipedia.org/wiki/TiMidity%2B%2B Curious about its Gtk2 GUI... Mar 19 23:20:33 What does timidity-daemon actually do? Take up the role of mod-player-back-end for media-player-front-end? Mar 19 23:37:12 probably workaround to existing software that needs the classical MIDI API. So more or less... Mar 19 23:37:40 For somebody: in timidity-interfaces-extra, move *.desktop from /usr/share/applications/ to /usr/share/applications/hildon/ Mar 19 23:44:40 Also, probably update timidity in maemo repositories from 2.13.2 to 2.14.0? Hopefully, it's not that much different... Mar 19 23:52:19 Pali: How do I join mstardict project? It's related to http://talk.maemo.org/showpost.php?p=1456805&postcount=167 I already clicked [Request to join] once at https://garage.maemo.org/projects/mstardict/ Mar 19 23:52:37 It was, maybe, week or even more ago... Mar 19 23:53:51 Oksana: probably some administrator must accept your request Mar 19 23:54:32 "If you would like to contribute to this project by becoming a developer, contact one of the project admins, designated in bold text below." Mar 19 23:54:40 I'm not project admin :-( Mar 20 00:08:21 Okay, so I should gmail the admin?.. Mar 20 00:16:18 yes Mar 20 01:01:11 anyone awake? Mar 20 01:30:50 Not much. And yes, I gmailed, just now **** ENDING LOGGING AT Sun Mar 20 02:59:58 2016