**** BEGIN LOGGING AT Fri Sep 30 02:59:57 2011 Sep 30 03:06:16 SpeedEvil: speaking of bugs have you noticed, on the n900 that the gps would sometimes get stuck in a state where it claims to see 12 or so satellites at the minimum signal strength (6%?) and stop updating until it's turned off and on, or sometimes the whole device would have to be power cycled? Sep 30 03:11:47 Not seen that Sep 30 03:25:43 * balrog-k1n regrets clearing the last track after saving it :( Sep 30 04:26:26 finaly got my N900 put back together Sep 30 04:26:36 find a small piece of something got under the digitizer screen Sep 30 04:26:43 i cant blow it out Sep 30 04:26:45 :( Sep 30 05:59:35 http://www.linuxfordevices.com/c/a/News/ZTE-T98-and-Asus-Asus-ToughETBW11AA/ <-- hmm Sep 30 06:06:30 http://www.theregister.co.uk/2011/09/29/apple_claims_products_lame/ <-- rofl Sep 30 08:08:05 DocScrutinizer: lol, you're on metawatch as well ;) Sep 30 08:15:46 GeneralAntilles: What's the battery life like on your MetaWatch? Sep 30 08:17:06 oh, and GeneralAntilles is there Sep 30 08:18:42 DocScrutinizer: I got happy too early. The SIM failed again last night :( I guess I'll let my electronics friend have another go at hot air reflow. Or could there be another cause? Sep 30 08:49:18 ~aegis Sep 30 08:49:18 http://www.developer.nokia.com/Community/Wiki/Harmattan:Developer_Library/Developing_for_Harmattan/Harmattan_security/Security_guide , or "The purpose of this framework is: ... to make sure that the platform meets the requirements set by third party software that requires a safe execution environment.", or http://en.wikipedia.org/wiki/Trusted_Computing#Criticism, or http://en.qi-hardware.com/w/images/1/10/ME_382_LockedUpTechnology2.gif Sep 30 09:40:19 Meanwhile: Top 1% of iOS Game Developers Make a Third of All Revenue Sep 30 10:00:22 RST38h: something similar with tax returns iirc Sep 30 10:00:30 though not quite such extreme numbers Sep 30 10:01:44 http://www.theregister.co.uk/2011/09/30/software_copyright/ <-- this sounds way to much like business logic Sep 30 10:08:17 How do you exit the menu of Navit on Maemo on Nokia N900? (In particular, how do you use Navit at all, e.g. example maps.xml showing whether I can use multiple root XML elements, etc.? ) Sep 30 10:10:25 In particular, is there any way to use Navit without having to restart the program each time you change a setting? Sep 30 10:14:04 OK, I figured it out, it's uparrow, enter. Why? Sep 30 10:20:14 ~moo Sep 30 10:20:14 * infobot mooooooooo! I am cow, hear me moo, I weigh twice as much as you. I am cow, eating grass, methane gas comes out my ass Sep 30 10:21:20 c'mon it's Friday, so what's the completely weird possibly made up and for sure OT talk today? ;-D Sep 30 10:22:23 this comet MohammadAG mentioned destroyed the world a week ago and we just didn't notice yet? Sep 30 10:37:56 DocScrutinizer: I have solution for bq kernel module and BME Sep 30 10:43:15 Pali: Which is? Sep 30 10:43:17 Pali: nice Sep 30 10:43:24 bother to tell me? Sep 30 10:44:23 my idea was along a LD_PRELOAD lib around bme, to intercept and redirect fopen() and friends Sep 30 10:44:33 First is patch in bq module which export character device with ioctl for access bq registers Sep 30 10:44:41 only if it was to /dev/i2c-2 Sep 30 10:44:57 and I have simple library for LD_PRELOAD compatible with i2c-dev Sep 30 10:45:06 :-D Sep 30 10:45:16 Next is patch for kernel module i2c-dev which will use ioctl function in bq module Sep 30 10:45:22 In principle, that sounds not insane Sep 30 10:45:45 that i2-dev kernel module patch is not done yet Sep 30 10:46:04 :nod: could work Sep 30 10:46:33 take particular care about i2c-dev not building up too much overhead Sep 30 10:49:08 anyway my isea of just dealing with bme to redirect the access to /dev/i2c-2 to the proper /sys/bq27200/raw feels less invasive to me Sep 30 10:49:49 leaves the kernel space untainted, and only needs a LD_PRELOAD wrapper around BME Sep 30 10:49:57 He who writes the code makes the rules. Sep 30 10:50:05 :-) Sep 30 10:50:21 not in kernelspace, SpeedEvil Sep 30 10:50:45 And yes, such a patch will never, ever get upstream. Sep 30 10:50:54 yes, better is patched i2c-dev to use bq module for access reigsters Sep 30 10:50:57 Which raises some maintainance issues Sep 30 10:51:47 but BME and all battery our stuf is maemo related Sep 30 10:51:50 please check for "recursion termination conditions"! ;-D Sep 30 10:52:16 in pre build kernel-power v49 is that bq kernel patch included Sep 30 10:52:26 see commit: https://garage.maemo.org/plugins/ggit/browse.php/?p=kernel-power;a=commit;h=bac1e954cf97153134efb2d54ba11c50306a136e Sep 30 10:52:31 and honestly patching i2c-dev feels ... nasty Sep 30 10:52:42 patch: bq27x00-reg.diff Sep 30 10:54:03 as i2c-dev is a corner stone of the whole system, if there's any bug in it, we might even kill hw Sep 30 10:54:26 DocScrutinizer, the comet decided to turn back after it heard about Tizen Sep 30 10:54:51 Naah. Sep 30 10:54:53 so we must be sure that i2c patch will be bug-free Sep 30 10:54:59 last patch to i2c-dev introduced a bug so nasty nobody but me noticed it, and it wasn't even reproducable as it depended on a race condition Sep 30 10:55:11 It actually hit. The only casualties were 3500 nokians. Sep 30 10:56:19 noobish q: what does bq offer that bme doesn't? Sep 30 10:56:55 Accurate information about the battery state Sep 30 10:57:05 Pali: a LD_PRELOAD wrapper around bme, to make it access /sys/*/*/bq27200/raw rather than /dev/i2c-2, that would fix the issue without any tampering with kernel space Sep 30 10:57:35 wrapper is ready, see: https://code.launchpad.net/~pali/+junk/maemo_libbqioctl Sep 30 10:57:37 on that note, is Harmattan's bme better/more accurate? Sep 30 10:57:52 I tested it only with bq.sh script and i2cget Sep 30 10:58:03 MohammadAG: seems to be, yes Sep 30 10:59:20 wrapper library overwrite open, close and ioctl (which are exported by libc.so.6) Sep 30 10:59:34 Pali: great Sep 30 11:00:14 DocScrutinizer, can you look at that library? There can be some bugs... Sep 30 11:00:30 the link you just posted? Sep 30 11:01:09 Pali: ^^^ Sep 30 11:01:27 yes, maemo_libbqioctl Sep 30 11:01:37 will do, a pleasure Sep 30 11:15:22 Pali: I gather in bqread you get the register return value in struct bq27x00_reg_parms int ret, anyway you later on check this varable for <0. I don't get it how you'll get register values <0 from bq27200, or are you assuming an implicit cast of signed int8/16 to always positive int? Sep 30 11:16:14 see interface of reg_params in kernel: https://garage.maemo.org/plugins/ggit/browse.php/?p=kernel-power;a=blob;f=kernel-power-2.6.28/debian/patches/bq27x00-reg.diff;h=11a99503e121c9d75aa916250b4d756f22294bef;hb=bac1e954cf97153134efb2d54ba11c50306a136e Sep 30 11:16:50 value from bq register is casted to (32bit) signed int Sep 30 11:16:54 aiui the code will return dezimal 65534 for 0xFFFF, correct? Sep 30 11:17:15 should be Sep 30 11:17:21 :nod: Sep 30 11:17:41 I'm using function bq27x00_read in bq27x00_battery_reg_ioctl Sep 30 11:18:15 and in whole bq module is used function bq27x00_read Sep 30 11:28:38 hi Sep 30 11:28:54 one question: who manage the Planet Maemo feeds? Sep 30 11:33:25 I still fail to wrap my head around it completely, but I think it's not 100% the way I'd prefer, which is: bq27200.ko creates not only the plaintext "/sys/*/*/bq27200/current_now" and whatnot sysnodes, but also a plaintext(!) raw dump sysnode (see http://people.openmoko.org/joerg/battery/bq27k-detail dmp=(`cat /sys/devices/platform/s3c24xx_pwm.0/hdq.0/hdq/dump| tr -d '\n'`) ) -- then I'd assume the libbqioctl LD_PRELOAD wrapper around Sep 30 11:33:26 bme to intercept access to /dev/i2c-2 and map it to reads to that /sys/*/*/bq27200/raw (which is a plaintext file consisting of hex printed register values of all bq27200-chip registers) Sep 30 11:36:11 Pali: as I understand it you're doing the LD_PRELOAD for bme (et al) and intercept the open(), close(), ioctl() calls, but I fail to understand where to you redirect them. Seems they are not redirected to /sys/*/*/bq27200/raw (aka /sys/devices/platform/s3c24xx_pwm.0/hdq.0/hdq/dump ) plain text file Sep 30 11:37:25 ioctl call is redirected to ioctl call on /dev/bq27200-0 Sep 30 11:37:58 that character device is created by bq module Sep 30 11:43:09 yes. Sep 30 11:43:57 well it will probably work as well, though I find it's nicer to not create that new device /dev/bq27200-0 but rather that sysnode with plaintext Sep 30 11:44:21 but that's just me, your taste may vary on that Sep 30 11:46:50 btw if you'd augment libbqioctl to also handle bq24150 access, we could actually implement jrbme.ko proper kernel based charging handler and chip interface, according to http://maemo.cloud-7.de/bq24150-sysnode.spec.txt (redirect the ioctl calls to /sys/*/*/bq24150/r00..r05 ) Sep 30 11:47:18 does kernel-bfs have all kernel-power features? Sep 30 11:47:49 I think jacekowski already started on implementing such a kernel module to expose all those sysnodes for bq24150, but he stopped when he noticed it will collide with bme direct i2c-2 access Sep 30 11:48:06 Arkenoi: nfc what's kernel-bfs Sep 30 11:48:39 kernel with brain fuck scheduler, obviously. it is in extras-devel iirc Sep 30 11:49:15 hooray for new kernels for whatever-little-detail-changed from stock Sep 30 11:49:31 ¡ Sep 30 11:49:36 hehe Sep 30 11:49:50 * ruskie prefers the one he rolled himself with all the trimmings he wanted included Sep 30 11:51:02 afaik you can change scheduler even on the fly, so what's wrong with asking pali to include bfs scheduler to PK as an option? Sep 30 11:53:11 I mean there's been a reason why h-e-n stopped rolling a special h-e-n kernel and opted for integration of hostmode extensions to PK as quick as possible Sep 30 11:53:28 * ruskie still doesn't like PK Sep 30 11:53:56 so I guess you adore kernel-bfs then X-P Sep 30 11:54:01 nope Sep 30 11:54:04 as said... my own ;) Sep 30 11:54:16 I don't need overclocking and whatnot... just the same kernel with some stuff enabled Sep 30 11:54:46 * amiconn uses PK for several features: hostmode, ipv6 and overclocking/ undervolting Sep 30 11:55:31 ruskie: overclocking doesn't add anything to kernel, it just changes a few numerical constants Sep 30 11:55:32 I stopped bothering with overclocking stuff 6 years ago Sep 30 11:56:05 ruskie: and it's not enabled in non-tampered PK, it's just *available* Sep 30 11:56:26 DocScrutinizer, kernel module for jrbe would be nice Sep 30 11:56:29 I thought it was enabled but just set to some default? Sep 30 11:56:38 Pali: indeed Sep 30 11:56:53 we could export functions to kernel instead some sysfs/dev nodes Sep 30 11:56:54 fbcon is another thing I want to have and have in my own kernel Sep 30 11:57:15 Pali: so you say I should finally finish my work on http://maemo.cloud-7.de/bq24150-sysnode.spec.txt ? Sep 30 11:57:20 :-) Sep 30 11:57:30 ruskie: fbcon is included in kernel-power Sep 30 11:57:51 ruskie: fine! Sep 30 11:58:43 DocScrutinizer, you can finish specification. I can help you with implementing kernel module, but I do not know bq24150 chip Sep 30 11:59:12 Pali: I guess ruskie knows that, and he just wants to let us all know what he would like in PK and what not. Not that this matters to PK config, as we for sure won't remove "fixes" just because ruskie doesn't consider them useful for him Sep 30 11:59:37 DocScrutinizer, what do you consider a "fix" ? Sep 30 11:59:52 Pali: we should ping jacekowski what's been the state of his work with bq24150 kernel module Sep 30 12:01:52 ok Sep 30 12:02:14 ruskie: see changelog of PK and you got it all - not going to debate this now with you, you're of course free to roll your own custom kernel whenever you like the way you like. I'm just saying cmmunity doesn't need another 392 kernel flavours in extras-devel, when everything can get integrated into PK, and enabled/disabled just like user decides to Sep 30 12:02:31 DocScrutinizer, agreed on that Sep 30 12:02:42 good thing I don't push things into extras anything ;) Sep 30 12:04:21 Pali: I'm not fluent in kernel hacking for maemo atm, but I for sure can help with completing the http://maemo.cloud-7.de/bq24150-sysnode.spec.txt and explaining any unclear detail on how kernel module shall work Sep 30 12:04:49 yes, i'd prefer kernel-bfs features to be integrated to p-k Sep 30 12:05:02 DocScrutinizer ok Sep 30 12:05:51 of course this would introduce a incompatibility into PK that still makes bme break whenever the bq24150.ko get loaded, unless bme is started LD_PRELOAD libbqioctl bme Sep 30 12:06:07 DocScrutinizer, can you look at this commit in android kernel? https://gitorious.org/android-n900/kernel-ng/commit/a7aa7f049e5be547a65247b7d88f6d248929d26c Sep 30 12:06:24 if rjbme.ko will be implemented we can stop bme and delete it Sep 30 12:06:44 I'd prefer to not touch android stuff except for most desperate state of mind Sep 30 12:06:59 that commit impelement something about bq charger Sep 30 12:08:01 Pali: the nice thing is user is free to either stop bme and use jrbme.ko to care about charging, or start bme and everything will work as usual, given bme got that libbqioctl wrapper Sep 30 12:09:13 yes, if kernel driver will be not static linked into kernel, it is possible stop bme and load module... unload module and start bme Sep 30 12:09:25 Pali: (andridiot bq charger) I'm quite confident my idea about how bq24150 works and shall get handled is good enough so I don't need any additional input from andridiot land that possibly spoils my day or my clear thoughts Sep 30 12:10:14 Pali: (unload module and start bme...) even with module loaded bme can work with that wrapper Sep 30 12:10:39 if we update/create new wrapper for that it is possible Sep 30 12:10:59 Pali: I got some RL topics on my calendar - c u l8r Sep 30 13:03:43 DocScrutinizer: remember my ranting about 2g/dual not working for me? Sep 30 13:03:57 I just tried with my sim in the n900 from my sister, and voila there it works Sep 30 13:04:55 My 2G is a bit dodgy Sep 30 13:05:12 hmm, what about trying a new SIM in *your* N900, after fresh full flash, as suggested? Sep 30 13:05:18 well I can't even get on 2g it just never switches and when I reopen the settings it's 3g again Sep 30 13:05:26 gonna do that soon Sep 30 13:05:38 need to back up stuff first :/ Sep 30 13:07:59 anyway I guess it's a hw defect in cellmo Sep 30 13:08:19 reflashing cellmo FW might help Sep 30 13:08:25 bbl Sep 30 13:10:54 Hi there! Sep 30 13:11:12 May I ask a question about maemo and meego here? Sep 30 13:11:36 Will I'll be able to install maemo packages in meego? Sep 30 13:13:41 hm just installed syslog to check and all I get is this: http://pastebin.com/55Ru1gX6 Sep 30 13:14:35 AVHard: no Sep 30 13:15:35 * AVHard start to dislike meego now Sep 30 13:18:58 ok, color me worried. The N800 suddenly thinks there is 14MB more free space on the internal storage Sep 30 13:33:22 figures, for some reason apt had cleaned out its cache but not downloaded updated lists. Do it really use 10MB tho?! Sep 30 13:49:04 pretty fsckng annoying that Nokia always sends his developer newsletter twice >:-( Sep 30 13:57:32 Jaffa, dunno Sep 30 13:57:44 Jaffa, haven't even gotten it to a full charge yet. Sep 30 13:58:27 But it arrived empty at about 1700 UTC yesterday Sep 30 13:59:24 I charged it until about 2100 UTC Sep 30 13:59:51 Got to about 66%, tried to charge it last night, but it apparently didn't want to take. Sep 30 13:59:55 22% this morning Sep 30 14:01:15 * GeneralAntilles1 buys some MLB playoff tickets Sep 30 14:05:43 GeneralAntilles, what device? Sep 30 14:20:46 crashanddie, digital MetaWatch. Sep 30 14:24:58 GeneralAntilles, clamp charger? Every day charging required? Sep 30 14:25:20 Clamp, dunno about every day Sep 30 14:25:27 I haven't yet had it 24 hours Sep 30 14:25:36 and didn't manage to get a complete charge on it yesterday Sep 30 14:25:48 It's likely to be highly usage dependent, though. Sep 30 14:25:54 I couldn't keep my hands off it. :P Sep 30 14:31:23 :) Sep 30 14:49:08 anyone got the n9 yet? Sep 30 14:49:53 * SpeedEvil checks pockets. Sep 30 14:49:55 Nope. Sep 30 14:50:13 damn Sep 30 14:50:15 lol Sep 30 14:54:38 pauly2: it sucks anyway Sep 30 14:56:57 lukw-jr: why it sucks? Sep 30 14:57:03 sry Sep 30 14:57:14 luke-jr: Sep 30 14:57:30 pauly2: no keyboard Sep 30 14:57:37 which is kinda the whole point Sep 30 14:59:39 luke-jr: can't afford one now anyways Sep 30 15:08:00 Jaffa, I'd say the analog isn't worth it, though. Sep 30 15:08:13 Jaffa, limited screen(s) resolution, no seconds hand, only 3 buttons. Sep 30 15:16:07 Is there a way to jury-rig the headset button to do something like rewind the player 10 seconds or so? Sep 30 15:16:47 Say, send a command to Panucci or Rhythmbox to do rewind and play. Sep 30 15:18:51 I listen to podcasts while outside, and I like to keep the volume low, so I miss some words when a car passes by. As things stand, rewinding requires taking the phone out, unlocking it (possibly entering the password), rewinding, and locking again. Sep 30 15:30:25 Pavel: http://talk.maemo.org/showthread.php?p=1006716#post1006716 might have sth for you, with a bit of tweaking the commands associated to the button events. Though it's long it's probably worth reading the whole thread Sep 30 15:30:53 Pavel: There's a thing to pause the payback on hadset press Sep 30 15:31:01 I forget the name, sorry Sep 30 15:31:12 [ANNOUNCE] headset-control : Make the headset button actually manipulate tracks. Sep 30 15:31:20 see link above Sep 30 15:33:48 Thanks. I'll give that a try. Sep 30 15:33:53 BBL Sep 30 15:58:03 quit Sep 30 16:16:56 any plans for CSSU to improve IM performance when connecting to network? with ~100 contacts it is a pain Sep 30 16:34:58 hello all! Sep 30 16:35:10 I can't seem to make my transitions.ini blur my background. what's up with that? Sep 30 16:35:24 i'm running CSSU and the theme customizer Sep 30 16:48:50 heya javispedro Sep 30 16:49:01 hi there. Sep 30 17:13:14 javispedro: hi Sep 30 17:24:34 DocScrutinizer: infobot joined! Sep 30 17:26:29 oops. i reflashed n900 and now nokia messaging vanished from service providers list Sep 30 17:27:11 ~part Sep 30 17:27:29 Leaving. (courtesy of docscrutinizer). Sep 30 17:27:38 DocScrutinizer: infobot joined! Sep 30 17:29:29 Leaving. (courtesy of docscrutinizer). Sep 30 17:29:38 DocScrutinizer: infobot joined! Sep 30 17:29:55 meh Sep 30 17:34:36 ok, one last test (sorry for the noise) Sep 30 17:34:44 Leaving. (courtesy of docscrutinizer). Sep 30 17:34:54 DocScrutinizer: infobot joined! Sep 30 17:35:04 better Sep 30 17:36:22 is there a way to remove the 'no sim' icon? Sep 30 17:36:24 it's irritating Sep 30 17:36:28 and unnecessary Sep 30 17:36:38 umm Sep 30 17:36:40 cover it with a piece of sticky paper? Sep 30 17:36:45 probably yes Sep 30 17:36:52 RST38h: i like it Sep 30 17:37:04 DocScrutinizer: I've been through the forums, and others also complain of the same thing Sep 30 17:37:07 but no solution thus far Sep 30 17:37:09 hahaha Sep 30 17:37:35 a piece of sticky paper huh? :) Sep 30 17:37:37 I had a bit more success not making the 'cell' as default when trying to call Sep 30 17:38:25 my buddy is going to bake my n900 in a little bit... i think he said tomorrow Sep 30 17:38:32 he found the perfect fitting usb port tho :) Sep 30 17:38:34 orangey, AIS recently got support for hidding icons Sep 30 17:38:43 AIS? Sep 30 17:38:54 advanced-interface-switcher Sep 30 17:39:11 You may hide wlan, gsm, or bluetooth icons completely Sep 30 17:39:27 orangey: has anybody tried tweaking that xml config file that determines which statusbar icons show up where, and which menu buttons are available? Sep 30 17:39:35 it's workaround for irritating icons-flashing, when You got more than one disabled Sep 30 17:40:05 Estel_: searching Sep 30 17:40:12 DocScrutinizer, I've given edit to my post about kernel bfs Sep 30 17:40:13 DocScrutinizer: Whatever AIS does, maybe that would work for this too.. Sep 30 17:40:44 There is now official annoucement thread now, with nice explanation what bfs-kernel is and why, by Idont Sep 30 17:41:19 macer: bake it? Sep 30 17:41:19 ...+ many posts of people complaining about it being unable to install alongside kp48/49, lol Sep 30 17:41:26 soldering, probably Sep 30 17:41:37 he said the only way to do it would be to reflow it Sep 30 17:42:07 it's great how people who don't know what kernel is - even on basic level go and read annoucement about bleeding-edge one with many warnings... Sep 30 17:42:10 so he is going to use a solder epoxy and use hot air Sep 30 17:42:11 hey, anyone know how the wifi auto connect works on the N900? I want to find a workarround for it not autoconnecting to WPA2-enterprise Sep 30 17:42:32 then, go ahead and install it on top of kp49, lol2 Sep 30 17:43:07 solder epoxy? wtf, solder with some de-oxidizing agent in form of epoxy putty? Sep 30 17:43:21 i forgot how he said it heh Sep 30 17:43:30 hardening only, when it's hot? Sep 30 17:43:35 Estel_: I went offline after posting the announcement thread, only to find out there were about two pages full of issues when I came back today o.O Sep 30 17:43:42 never heard about it Sep 30 17:43:51 he is using hot air on it Sep 30 17:43:55 Yea, saw that :) btw, nice to see You here Sep 30 17:43:59 to get the solder to make the port stick Sep 30 17:44:08 Seems like some didn't knew about the difference between kernel-bfs-flasher and kernel-bfs-bootimg :-/ Sep 30 17:44:20 Macer, I just wonder what epoxy-solder is Sep 30 17:44:47 yea, but teying to install it alongside kp is even more lol Sep 30 17:44:57 anyway, it will settle down after few days Sep 30 17:44:59 maybe i got the wording wrong.. but it would assume it is a thick paste like substance Sep 30 17:45:08 joy of making annoucements Sep 30 17:45:10 that will melt and harden when heated and cooled Sep 30 17:45:15 yep. There was/is one valid issue between the posts though Sep 30 17:45:39 It seems that Maemo will reboot several times when trying to boot a different kernel (with a different extraversion) Sep 30 17:45:46 You got trishas trying to use it, failing, complaining for x pages, then, finally, annoucement reach desired audience Sep 30 17:45:58 yea, nticed it Sep 30 17:46:10 After succesfully booting it, it will boot every time flawlessly though Sep 30 17:46:21 really strange, but not exlusive to kernel-bfs AFAIK Sep 30 17:46:22 I wonder why it looks like that Sep 30 17:47:00 me too, will look into it later. Probably has something to do with the modules not being load or so. At least, that's my only guess right now Sep 30 17:47:27 Macer, if it is something like solder + soldering paste n form of putty, but becoming real solder after heating to ~400 C degrees, that would be great Sep 30 17:47:30 WAAAAAAAAAAAAAAAAAAAAAAH!!!!!!!!!!!! Sep 30 17:47:37 >>Kernel-bfs reintroduces the original workaround but makes it optional. This way you can choose to accept the risks associated with removing the safety mechanism and use both bq27x00_battery and bme like in kernel-power v45-v47. To enable this old behaviour, execute "echo 1 > /sys/kernel/n900/i2c-dev_break_ioctl". You could put that command in a startup script for convenience's sake. Please refer to this post for more information regarding Sep 30 17:47:39 the workaround.<< Sep 30 17:47:48 Estel_: heh Sep 30 17:48:05 DocScrutinizer: Hehe, I knew you would post something like that ;) Sep 30 17:48:08 never let fools mess with your kernel! Sep 30 17:48:11 Macer, it would allow to repair broken pds Sep 30 17:48:18 hahaha, I knw it also Sep 30 17:48:19 Estel_: not only does it not do what I want, it adds an extra icon! Sep 30 17:48:37 I knpw You will love me after that, but I was one requesting it :P Sep 30 17:48:50 Estel_: well. as long as my usb port works ;) Sep 30 17:48:52 orangey, hahaha, how come? Sep 30 17:49:00 i haven't had my n900 for like 3 months Sep 30 17:49:02 Estel_: the AIS icon! Sep 30 17:49:14 icon?! ais doesnt have icon Sep 30 17:49:16 or more Sep 30 17:49:32 idont: why don't you also implement a self destruct sysnode? echo 1 >/sys/BS/boom Sep 30 17:49:38 Estel_: my status area now has a new icon.. a bunch of stacked bars Sep 30 17:49:40 DocScrutinizer: I agree that the the safety mechanism in i2c-dev shouldn't be disabled, but some people are willing to take the risk. Anyway, i2c-dev_break_ioctl is 0 by default of course Sep 30 17:49:51 DocScrutinizer, but You must admit that there is no confirmed report of any fault causesd by it Sep 30 17:49:52 I for one will not echo 1 ;-) Sep 30 17:50:06 Estel_: BS Sep 30 17:50:22 and I will echo 1. I was using it for ages without single glitch. Sep 30 17:50:34 I agree with rationale to disable it by default Sep 30 17:50:43 but freedom of choice is best Sep 30 17:50:47 DocScrutinizer: Please don't forget that I actually *agree* with you opinion. Sep 30 17:51:06 Estel_: there's no way to postmortem-analyze it's been a i2c collision on any arbitrary bus on any arbitrary chip on any register that made the hw go hamuk Sep 30 17:51:11 I would not cry here, if something goes wrong. Will just come and post info about it ;) Sep 30 17:51:14 Estel_: +1 Sep 30 17:51:22 @ freedom of choice ;) Sep 30 17:51:57 DocScrutinizer, but how come one can use it extensively for months without any collisions? Sep 30 17:52:10 I can't belive it's pure luck Sep 30 17:52:17 so patches like this one add to a general "linux sucks and is instable" notion that is impossible to fix, unless you radically forbid any such bullshit ever hits the daylight Sep 30 17:52:27 there must be something deeper about it Sep 30 17:52:43 DocScrutinizer, one enabling thia should know the risk Sep 30 17:53:03 DocScrutinizer: OK, now that AIS didn't work, where is this XML file that decides status bar icons? Sep 30 17:53:14 if I build airplane on my own and go using it, I'm also knowing risk and it's ok - as long as I can't crash on someone head Sep 30 17:53:19 Jaffa, scratch all the battery talk. That was phone battery, not watch battery. . . . Sep 30 17:53:57 DocScrutinizer, anyway, You have any idea what symptoms could prove conflict? Sep 30 17:53:59 Estel_: unless you understand why it happens, when it happens, and why it has to be FORBIDDEN to ever get to any machine, I can't teach you here about reentrant-save race-free kernel design Sep 30 17:54:03 I am blown away by how modern a phone can remain when it's open. my n900 doesn't show its age at all! Sep 30 17:54:27 except modest won't mark my replied emails as replied.. boo Sep 30 17:54:32 I would love to report here even smallest things that may be related to echoing 1 it Sep 30 17:55:05 Estel_, echo one where? Sep 30 17:55:08 Estel_: you probably won't even notice, and if you do you never would think it's related Sep 30 17:55:41 I understand, but what the risk is, in Your opinion? what may happen? Sep 30 17:56:05 can be a 1000 things, starting from wrong backlight brightness (readout of ALS got fsckd) to battery exploding (bq24150 programmed to charge battery to 4.45V) Sep 30 17:56:10 mece, I mean kernel-bfs feature to enable bq27x00_battery co-existence with bme Sep 30 17:56:34 ok, I'll look for any hardware issues. *any* Sep 30 17:56:37 Estel_, ok, not my area :) carry on. Sep 30 17:56:41 mece: I've reintroduced the i2c-battery patch in kernel-bfs, but with some sysfs code wrapped around it (disabled by default) Sep 30 17:57:11 if something, even smallest thing goes wrong, I'll be first to report it. ven if it's not related. Sep 30 17:57:24 but just plain suspection ;) Sep 30 17:57:27 idont: this patch is a) idiotic, and b) there are better ways to achieve same result Sep 30 17:57:42 DocScrutinizer, what ways? Sep 30 17:57:49 * Estel_ is very interested Sep 30 17:57:51 DocScrutinizer: a) I agree and b) They're not finished yet AFAIK Sep 30 17:57:59 and c) bq27200.ko driver is poorly written/designed anyway Sep 30 17:58:03 Estel_: redirecting ioctls Sep 30 17:58:33 DocScrutinizer: I'm not using the bq27x00_battery module Sep 30 17:58:45 Well, only to test if the patch was working :P Sep 30 17:59:04 idont: Just looking at the announcement. Is the experience markedly better than the current kernel power? Sep 30 17:59:05 idont: e.g bq27200.ko could open-read-close i2c-dev instead of continuously occupy it Sep 30 17:59:27 orangey, it depends Sep 30 17:59:28 GeneralAntilles: eh hehe, sorry for the confusion :P Sep 30 17:59:40 Well, it certainly makes sens Sep 30 17:59:41 e Sep 30 17:59:47 Estel_: on? Sep 30 17:59:48 DocScrutinizer, so why no one rewrite it for us and even submit changes upstream? Sep 30 17:59:53 Watch battery life combined with currently SSID seems a little arbitrary. Sep 30 18:00:04 because nobody needs that driver so far Sep 30 18:00:12 orangey, on use case. BFS is sheulder, as You already know. Sep 30 18:00:22 sheulder? Sep 30 18:00:29 right, scheduler Sep 30 18:00:31 It is reportedly known to work better on non-server machines Sep 30 18:00:31 yes, I know Sep 30 18:00:39 so it's a complete BS to introduce a bug into i2c-dev to enable that driver coexist with bme Sep 30 18:00:41 but its harsd to measure in hard numbers Sep 30 18:00:45 orangey: I would say to try it out for yourself :). Results seem to be mixed (ranging from "impressive" to "no difference"). At leat I'm happy to use it :-) Sep 30 18:00:49 Idont quoted some graphs Sep 30 18:00:58 idont: what's your experience? Sep 30 18:01:15 orangey: improvement, but I'm biased towards BFS' design Sep 30 18:02:00 idont: why would you want to keep bme running and instead fuckup kernel i2c-dev when bq27200.ko works pretty ok in an unpatched version of kernel, as long as bme got stopped? Sep 30 18:02:16 Estel_: I'm not sure which graphs you're talking about. Could you point me to them? (I'm sorry if I forgot something) Sep 30 18:02:24 I've put in enough phone time today, I think. I'll try it in a couple of weeks when things are a bit more free for me Sep 30 18:02:39 DocScrutinizer: Any hints re that XML for the status bar icons? All I can find is a thread discussing it on diablo Sep 30 18:02:47 DocScrutinizer: Does everything still work with bme stopped? Sep 30 18:03:13 does everything still work with i2c fucked up? Sep 30 18:03:14 could anyone suggest something robust for time lapse video on n900? Sep 30 18:03:50 DocScrutinizer: irrelevant IMO, the risk is optional. Sep 30 18:03:57 BS Sep 30 18:04:00 ? Sep 30 18:04:37 iDontp i mean one quoted from bfs reasdme Sep 30 18:04:39 readme Sep 30 18:05:05 "courtesy of Serge Belyshev" Sep 30 18:05:12 to end users (and here all those who don't do kernel dev are end users) nothing is ever "optional" as there will be packages that request bq27200 module and will "take the risk" on behalf of "end user" Sep 30 18:05:17 idont, sorry for typo Sep 30 18:05:25 np, one moment Sep 30 18:05:43 it is rather comparing numbers proving inefficiency of cfa sheulder Sep 30 18:05:51 Estel_: graphs should be somewhere here: http://ck.kolivas.org/patches/bfs/ Sep 30 18:07:28 idont, bme functions doesn't work with bme disables, obviously Sep 30 18:07:29 DocScrutinizer: I disagree, the user has to echo 1 himself and therefore willfully accepts the risks. I'm assuming no software would do this on behalf of the user (that would be an unacceptable risk IMO) Sep 30 18:07:42 HAHAHAHAHAHA Sep 30 18:07:50 so no emergency shutdown on low battery Sep 30 18:08:04 (= possible risk of filesysten damage) Sep 30 18:08:11 Estel_: BS Sep 30 18:08:13 no charging via pc Sep 30 18:08:19 @500mA Sep 30 18:08:31 (only 100mA, if any) Sep 30 18:08:34 etc Sep 30 18:08:40 Estel_: I know, it was more of a rhetorical question to DocScrutinizer ;-). That's why the patch is included. Sep 30 18:08:40 * DocScrutinizer out Sep 30 18:09:01 DocScrutinizer, I don't like when someone try to babysit me as user, by denying me to use some option... Sep 30 18:09:09 ...if risk is well described Sep 30 18:09:11 this is too fucking nuts for my nerves Sep 30 18:09:13 DocScrutinizer: Please know that the patch will be removed as soon as a proper fix for bq27x00_battery + bme will be finished Sep 30 18:09:15 i.e with all warninga Sep 30 18:09:39 DocScrutinizer: I'm sorry, it is not my intention to get your nerves up :-( Sep 30 18:09:54 DocScrutinizer, no offense, but when You introduced hotswapping battery... Sep 30 18:10:08 Estel_: I'm just saying one more sentence: you got NFC what's the risk, and you mess with other people's hardware in a way I deprecate deeply Sep 30 18:10:18 via stopping bme and fooling chip to emergency charge Sep 30 18:10:33 I'm not meesing with other people hardware in any way Sep 30 18:10:50 talking like that, aircrack-ng should be forbidden Sep 30 18:11:09 cause it make 50% of user to mess with other people hardware Sep 30 18:11:17 in direct way Sep 30 18:11:40 thatnk for proof of NFC Sep 30 18:11:46 thanks even Sep 30 18:11:49 I agrwe that I'm not specialist, so I accept risk twice - the one I understan, and the horrible one I dont understand Sep 30 18:11:57 it's my choice Sep 30 18:12:24 as long as it is not enabled by default, nor adveristed as safe, but warned, its ok Sep 30 18:12:42 hope you're expert enough to strace each app you install for doing that magic echo 1 >selfdestruct Sep 30 18:12:54 Your topic about hotswapping battery can fuck up people filesystem if fails, and You have written warninga in red around it Sep 30 18:13:07 and it's ok ;) Sep 30 18:13:21 thanks for again proving NFC Sep 30 18:13:35 removing battery is sth different to stopping bme Sep 30 18:13:39 Any hints on where I can configure my statusbar icons to remove the 'no sim' icon? Sep 30 18:13:44 and I'm definitely out now Sep 30 18:14:07 time lapse ... anyneo Sep 30 18:14:09 ? Sep 30 18:14:14 DocScrutinizer: Hope all is well, I'll be out in a minute too Sep 30 18:14:53 look. Description on kernel-bfs annoucement is clear and understandable to any person installing it Sep 30 18:14:59 what is the name of the area in which the clock and 'nosim' icons are? Is that 'status bar'? Sep 30 18:17:30 with all due respect for DocScrutinizer, I just don't agree here and I'm quite surprised by this over-reaction Sep 30 18:17:47 no, nothing is well! one fool writing a bs bq27200 driver, andother fool writes a patch to make it work with broken brainfsckd bme and that patch introduces risk of severe hw damage. the next fool reverts the rejection of that whole stinking pile of BS and tells a nice story of "taking the risk" (for WHAT?????) - and tomorrow an adridiot copycats it and serves it as hot pie to cyanogen, from where same fool#1 will import it next week as " Sep 30 18:17:49 certified to work in andridiot" - january it goes upstream Sep 30 18:17:53 no, nothing OK Sep 30 18:18:31 DocScrutinizer, You're not forced to use it, yep? Sep 30 18:18:53 for me, freedom of choice is much more important. Sep 30 18:19:18 Your ex'ple with selfdestruct echo - why not, anyone can use vi_ russian roulette script Sep 30 18:19:32 Estel_: once fool #6 sends it upstream I'm forced to use it Sep 30 18:19:43 as far is it's described properly, it's no more bad thing. People can damage their devices psychically... Sep 30 18:19:50 for posterity, it is: /etc/hildon-desktop/status-menu.plugins Sep 30 18:19:55 so advocate for not using it upstream Sep 30 18:19:58 and I already been forced to use it when it went into PK without proper review Sep 30 18:20:28 erm, You were not forced to update. And in bfs-kernel it'a disabled by default Sep 30 18:20:33 The risk of hardware damage is real - the kernel is talking over I2C to the power regulators on the chips dozens of times a second. Sep 30 18:20:45 Race conditions that can corrupt these writes can kill chips. Sep 30 18:20:48 now, You're excited by f€t, that people are able to enable it, if they like to Sep 30 18:20:55 no one forcing anyone to use it Sep 30 18:21:07 SpeedEvil: please take over, my blood pressure destabilizes Sep 30 18:21:17 SpeedEvil, exactly, that is how I understand risk Sep 30 18:21:49 DocScrutinizer, SpeedEvil: Looks like it segfaults on my phone, but I am running the power user kernel (since I need to use the WiFi hotspot app). Sep 30 18:22:13 SpeedEvil, if You dont mind explaining for a while without nervous reactions ;) Sep 30 18:22:17 Estel_: NOBODY IS ALLOWED TO PUBLISH SHIT WHEN THERE ARE BETTER ALTERNATIVES Sep 30 18:22:17 anyone test meego 1.3 yet? Sep 30 18:22:28 Really? Sep 30 18:22:31 there are no alternatives working already Sep 30 18:22:50 and everyone can publish shit, without forcing others to use it (microsoft) Sep 30 18:23:11 there's no excuse for own incompetence in "others are free to reject it" Sep 30 18:23:27 not if hw damage is the alternative Sep 30 18:23:51 dropping phone into bathtube is also risk Sep 30 18:24:11 would You penalty people for using it while taking bath, or in bathroom at all? Sep 30 18:24:11 Estel_: you're close to a kick for trolling Sep 30 18:24:30 interesting, cause I'm performing civilized discussion. Sep 30 18:24:39 You're free to dissagree Sep 30 18:24:57 no you use purposely stupid analogies Sep 30 18:25:08 But if Yout think emotions are overhelming You, won't You mind going off-line for a while? I don't like being terrorized Sep 30 18:25:08 Internet suddenly got disconnected :S Sep 30 18:25:16 :P Sep 30 18:25:57 nobody may sell a bath radio to stick to your bathtub with two suckers, not even when there's a writing in red on it about "DANGER! Do Not Let Drop Into Water!" Sep 30 18:26:30 DocScrutinizer, with all due respect again, scarring someone by kick in IRC channel, judt bcauise You don't agree or think she/he is talking shit, is much worse for Open Spirit than enabling people to echo 1 some dangerous path Sep 30 18:27:11 of course this would ban hair dryera Sep 30 18:27:15 dryers* Sep 30 18:27:28 You realizep how many people dry hairs in bathtube? Sep 30 18:27:34 :o Sep 30 18:27:46 so Sep 30 18:27:50 Estel_: I don't think I got kicked, my internet connection stopped as a whole (all lights lighted up on the modem, never seen that before) Sep 30 18:27:50 refusing to see the difference between stupid behaviour and using stupidly behaving appliances/other stuff is considered trolling here Sep 30 18:27:56 anyone checked out meego ce1.3 on n900 yet? Sep 30 18:28:02 for me, Your analogies - in this certain matter - are stupid, but I'm not trying to terrorize You with "kick for trolling" Sep 30 18:28:04 how is the performance? Sep 30 18:28:11 t7^, I just asked in meego Sep 30 18:28:23 stskeeps says there's a decent speed improvement Sep 30 18:28:28 oh Sep 30 18:28:31 they even say to give it a try on my class2 microsd Sep 30 18:28:31 thanks Sazpaimon Sep 30 18:28:41 really? Sep 30 18:28:49 yeah Sep 30 18:28:49 hmm i lost my class 10 card!! Sep 30 18:28:49 screwit Sep 30 18:28:51 Sorry DocScrutinizer, You may don't like me from now on, but I consider You behaviour as clearly terrorizing me with possible channel operator rights abuse. Sep 30 18:28:53 dropped it somewhere Sep 30 18:29:02 i'll get a class 4 now Sep 30 18:29:04 and try meeg Sep 30 18:29:11 something I don't like on civilized discuassion. Sep 30 18:29:20 class 10 isnt as good as a class 5 or 6 for an OS install imo Sep 30 18:29:31 the higher the class, the worse the random read/write speeds get Sep 30 18:29:38 Estel_: you're free to consider that as whatever you like Sep 30 18:29:45 I may dissagree with You, but I would defend Your right to speak Your opinion (in civilized manner) to my last blood drop ;) Sep 30 18:30:02 if you're running an OS on the card, random read/write is very important Sep 30 18:30:10 and now sorry gentlemen, I got better things to do than to teach fools about risk management in sw development Sep 30 18:30:11 the class only applies to sequential read and write Sep 30 18:30:38 DocScrutinizer, You already stated that You're very emotional about all of this. You really don't think You need to calm down a little? Sep 30 18:30:41 I've found class 6 cards have the best tradeoff between sequential and random read/write speeds Sep 30 18:31:16 I thick channel operator abuse is something none of us want here. Sep 30 18:32:28 DocScrutinizer: The patch is very device specific (i.e. bme -> n900). I don't think it is realistic to even expect this patch to ever end up upstream. Otherwise, I would agree with that these kind of patches should be killed with fire (for the greater good) Sep 30 18:33:03 anyway, DocScrutinizer, I just like Yout to know that I still respect You and You work, even when I totally diss_ree with You here, and I think Your behavior (by calling idont fool and trying to terrorize me, by possible kick from channel) is tot!ly out of line here.is Sep 30 18:33:20 I also hope that all is well Sep 30 18:33:31 out for a while, ping me if needed. Sep 30 18:33:46 Let's call it a day, I think everyone has said what they've wanted to say Sep 30 18:33:50 I would like ods file open with gnumeric but in file manager I have to select this application for each file I added application/ods=gnumeric.desktop to /usr/share/applications/defaults.list also I have this http://codepad.org/GVHsjBgi what shall I do? Sep 30 18:33:55 agree Sep 30 18:34:22 I tired this topic but doesn't work for me http://talk.maemo.org/showthread.php?t=9349 Sep 30 18:38:20 Sep 30 18:39:12 no suggestion ? Sep 30 18:40:14 mintux1, what is not working for You exactly? Sep 30 18:40:49 dbus-switchboard is thing You should use to change MIME correlations Sep 30 18:49:54 Estel_: when I open file manager and I want to open ods file it ask for a application to open. I would like when I press it . it open in gnumeric automatically not select gnumeric in list each time Sep 30 18:50:47 and that is how dbus-switchboard is supported to work AFAIK - at least if gnumeric support opening files like that Sep 30 18:51:09 in dbus-switchboard, You need to set gnumeric as handler for mimetype of Your choice. Sep 30 18:53:28 Estel_: I want that but I could not Sep 30 18:53:40 it doesn't work Sep 30 18:54:00 could You tell us what exactly does not work, particullary? Sep 30 18:54:06 i.e. what You did. Sep 30 18:54:12 and what results were? Sep 30 18:55:30 Estel_: I added "application/ods=gnumeric.desktop" to /usr/share/applications/defaults.list according of this topic http://talk.maemo.org/showthread.php?t=9349 also I changed this file to http://codepad.org/GVHsjBgi after open file manager and then ask me to select app again Sep 30 18:56:21 I would like when open a ods file in file manager.it open in gnumeric. just this Sep 30 19:02:05 mintux1, insteasd of following this topic Sep 30 19:02:12 install dbus-switchboard Sep 30 19:02:24 from extras-devel (usual warnings etc) Sep 30 19:02:35 and use it to achieve Your goal Sep 30 19:04:30 on slightly different topic, could anyone point me to /sbin or /bin maemo default script, for enabling mass-storage mode? Sep 30 19:05:37 for unknown reasons, sometimes my N900 (after long uptime, without echoing 1 bfs-kernel dangerous feature) ;) doesn't react to connecting it to pc, while it react for connecting it to charger. USB port ks in perfect condition, seems software-related. Sep 30 19:06:22 the same way, many times, after disconnecting from mass-storage, my MyDocs isn't mounted - so I use script for manual mounting, if such thing happen. Sep 30 19:06:56 Would like to use same method for kicking it into usb mass storage mode, when it deny it Sep 30 19:07:16 something to do with ke-recv Sep 30 19:09:27 Estel_: what do you think why access to bq27200 via i2cget is safe, while bq27200.ko module either rebots or kills the hardware? Only because the bq27200 module is poorly written and the patch is making that even worse. There are better ways to implement the bq27200.ko so it acts exactly like i2cget, and this will allow loading of the module without stopping bme and without causing reboots. That is why this patch is *absolute bullshit* - it Sep 30 19:09:28 doesn't fix things but makes them worse, and there's no excuse whatsoever to proliferate it to users other than the one who wrote that crap, NO MATTER how many warning tags you attach to this dangerous version, no matter how many padlocks you add Sep 30 19:21:49 and the patch to implement that is only marginally more complex than the one that needs this echo 1>foobar lock, damn probably incl this safety lock the whole thing is more complex than a proper fix. The key is (I said this before, maybe it went unnoticed with all my shouting) that the bq27200.ko module mustn't open the i2c-dev exclusively on loading, but simply opens it immediately before it reads a value from bq-chip, then closes it again. Sep 30 19:21:50 If I2C-bus is occupied in that very moment (the sutuation where now a collision would occur, yielding bogus values at very least), the bq27200.ko module should return -EAGAIN for the read() trying to read out a value from sysfs node. Immediately after (trying) read from bq-chip the module shall close and free the i2c-bus again. BME can handle occasional failed reads, and it's up to your app reading from /sys/bq27200/* to do the same then. Sep 30 19:21:51 Result: peaceful coexistence, no kernel reboots, no collisions killing hardware; patch: ~20 lines Sep 30 19:25:19 there is no justification whatsoever to do it the dangerous way like in kernel-cfs, not even when there are additional safety measures implemented Sep 30 19:26:33 DocScrutinizer: But why would pali then go through all the trouble of intercepting and redirecting ioctls, when the solution is that simple? Sep 30 19:26:47 *Do note that I haven't looked through the bq27x00_battery module's code* Sep 30 19:27:08 and of course you are free to use that hw-killing kernel anyway, and I'm far from teachering you. But that kernel shouldn't be available for general public Sep 30 19:27:16 I'm not entirely sure there's such a concept as non-exclusive access by a kernel module.. Sep 30 19:27:36 idont: because that'S the *clean* way, while my suggestion is a dirty hack still, but a working dirty hack Sep 30 19:27:58 DocScrutinizer: fair enough Sep 30 19:28:02 ShadowJK: probably there isn't, indeed Sep 30 19:30:10 ShadowJK: nevertheless during module_init the bq27200 somehow opens/allocates the bq27200 chip and that makes /dev/i2c-2 reject access to that chip. The code inside bq27200.ko doing this (and reverting it on module unload) must be exploitable somehow Sep 30 19:30:46 It's part of the thing that makes no two kernel modules being able to hold the same device simultaneously Sep 30 19:30:56 yes Sep 30 19:31:25 DocScrutinizer: what about a seperate kernel build with the patch enabled, only available over at kernel-bfs' garage page and thus not easily accessible by the general public. Combining this with the said arguments (the patch being optional, disabled by default, and device-specific), would you find it in your heart to have peace with the subject, just until the proper fix has been finished? Sep 30 19:31:56 yes, of course Sep 30 19:33:20 Okay, I'll make that happen right away then. I'm really sorry for getting your nerves up, that wasn't my intention of course. Sep 30 19:33:45 there are more dirty kernels around than anybody could count - the key is not to make them available to ignorant users Sep 30 19:34:24 idont: I am sorry to get THAT upset about it Sep 30 19:34:50 whole aegis situation that probably been caused by events like that made it worse for me Sep 30 19:35:02 yeah :/ Sep 30 19:35:52 DocScrutinizer: no problem, you've got your reasons :-) Sep 30 19:36:36 I want ssh service run when n900 come up Sep 30 19:37:02 mintux1: it does, in standard config Sep 30 19:37:19 but in mine I should run it manually Sep 30 19:37:38 mhm, so you probably added that on purpose Sep 30 19:38:11 yeah maybe you installed that OpenSSH selector or something Sep 30 19:38:15 I forgot the name of that Sep 30 19:38:20 but standard it's always running Sep 30 19:40:33 so have you heard of that Meltemi OS for Nokia low-end phones? Sep 30 19:40:54 sure Sep 30 19:41:10 actually that's a step in the right direction Sep 30 19:41:15 sounds like maemo-light Sep 30 19:41:17 now they should ditch WP7 Sep 30 19:41:29 slowly but surely Sep 30 19:41:31 Estel_: I tried dbus-switchboard but it doesn't work too :-( Sep 30 19:44:08 Apps on Nokia Store had 2.5 times more downloads in the second quarter of 2011 than did apps on Apples App Store Sep 30 20:05:26 hi therep I was off, just went through logs Sep 30 20:06:12 DocScrutinizer, I absolutely get Your reasons and agree with it, and I would be first to advocate inclusion of Your "dirty" hack for opening and closing i2c access Sep 30 20:06:24 when something like that would be available Sep 30 20:06:45 unfortunatelly, I'm unable to write that ~20 lines of code You mentioned. Sep 30 20:06:52 aiui idont will look into it Sep 30 20:06:58 or not, dunno Sep 30 20:07:44 DocScrutinizer, if You would write such small patch for bq27x00_battery.ko I would be eternally grateful. Sep 30 20:08:17 fact is, for now "nobody" needs the bq27200 kernel module anyway, so best option is to not even include it to kernel, or simply ignore and not load it Sep 30 20:08:34 Anyway, when I suggested 412b to use such a i2c-like way of gathering real battery info in his advanced-power-monitor, he answered that it would be too much reaource intensive Sep 30 20:09:12 so I'm not 100% sure, if You're right that it's so simple solution Sep 30 20:09:35 thats why I think iDont should not delete this option, until proper fix is introduced. Sep 30 20:09:58 the simplest solution for now is to use i2cget, as is done in my err *cough* shadowjk's bw27200.sh script Sep 30 20:10:14 lol Sep 30 20:10:53 Wish someone would just make the replacement battery applets access bq27200 through i2c-dev instead of sysfs :P Sep 30 20:11:03 as stated above, for now nobody needs /sys/bq27200/* and it better stays that way as long as we got no proper module implementation Sep 30 20:11:38 Estel_: the i2c-battery-sysfs patch will be disabled in the repo version of kernel-bfs. An exact same build with the patch enabled will be available from kernel-bfs' garage page. This seems to be the best compromise until we've got a proper fix. Sep 30 20:12:24 oooh there acrually is "somebody" who needs the bq27200/ sysnodes? replacement-battery-applet? too bad, see what happens when you publish APIs based on broken design Sep 30 20:12:56 I see idont . But isn't it much more hassle for You, and restraining users freedom a little? You know, forcing to use garage instead repo, for something *DISABLED* by default, that someone can enable if she/he wish Sep 30 20:13:33 DocScrutinizer, You have not answered my questin about resource-intensivity of Your approach to act like i2cget Sep 30 20:13:59 I don't see any impact on resource demand Sep 30 20:14:13 as stated above ;) 412b deny to use that way for gathering info for adevanced-power-monitor, as it seems to be reource heavy Sep 30 20:14:34 after all you'd not want to read out bq27200 more frequently than maybe max every other second Sep 30 20:14:38 You know, applet must pool such data quite often. Anyway, im just quoting Sep 30 20:14:47 thought about same thing Sep 30 20:14:49 so the additional open/close is negligible Sep 30 20:14:59 but I'm not coder, so I rely on what he said. Sep 30 20:15:31 this concern suggests he's doing sth terribly wrong Sep 30 20:15:57 idont, for me using garage page isn't a problem, but I still don't get the thing with restraining people to access something that is clearly stated as dangerous, cause possible hardware damage etc and depreciated. Sep 30 20:16:04 maybe Sep 30 20:16:17 * Estel_ will try to find appropriate post Sep 30 20:16:33 a system("i2cdump 2 0x55") every 5 seconds is no problem regarding any resource demand Sep 30 20:17:37 neither is a readout from a *proper* /sys/bq27200/capacity sysnode Sep 30 20:19:01 Estel_: Yes, it is a bit more hassle for me. I'm even putting my personal opinion aside. However, if there are such strong feelings against including a patch like this, then I'm willing to just that. Those really interested in the patch won't mind downloading the debs manually. Sep 30 20:19:03 It's very interesting, and probably important Sep 30 20:19:32 if the app fails to implemet proper refresh cycle management (I.E. have a timer to call once it every 30s to update battery state, only when visible) then that's the problem of that app - not any resource hungry access to bq27200 Sep 30 20:20:34 idont: even on extras-devel apps that deliberately risk to kill the hardware are not allowed afaik Sep 30 20:20:43 I'll try to talk about it with 412b. Anyway, I think it would be best to include proper fix that ShadowJK is suggesting Sep 30 20:20:58 hey, rootsh is allowed ;) Sep 30 20:21:07 using proper command You can kill device Sep 30 20:21:14 so what? that isn't dangerous to hw in any way Sep 30 20:21:17 like vi_ russian roulette Sep 30 20:21:38 overclocking too much is dangerous Sep 30 20:21:44 Estel_: I'm immediately stopping to answer this. It's not leading anywhere Sep 30 20:21:57 really, dont treat people like idiots, even if they are. Sep 30 20:22:00 no problem. Sep 30 20:22:22 though I agree the combination of PK_with_OC plus the OC-GUI should be banned as well Sep 30 20:22:49 I'm just talking that it is IMO not good to deny something to people willing taking risk, just beause YOU tyhink it should be banned. Sep 30 20:23:10 You know, joy of Open Source is that no one should dictate what is ok. Sep 30 20:23:28 bfs sheulder wasn't accepted upstream, but there are people using it. Sep 30 20:23:48 the problem with OC stuff is you can't exactly tell who's the culprit and also there are claims it serves a certain purpose that **can not reached otherwise**, which is not the case for the bq27200.ko patch Sep 30 20:23:59 I just don't like when one person try to restrain others, when they don't pose danger to others. Sep 30 20:24:52 currently, bq patch serves purpose of serving as data source for battery status menu applet, that can show extremely precise data about battery state, when fuel gauge is calibrated properly. Sep 30 20:25:18 I would be first to advocate replacing this module with proper fix, if anyone dare to implement that. Sep 30 20:25:30 I would, but I don't have enough knowledge :( Sep 30 20:25:37 There are those who takes risks, and then you see them asking for help with very little usable battery life. There are those who don't take risks, and still benefit from the usual usable battery life Sep 30 20:25:39 You say that it is 20 line of code Sep 30 20:25:54 for bq27200.ko there's a documented (since 2h ;-D) alternative way that has no risks for hardware, and risk for hardware is a dirty collateral damage of that bq27200, while for OC it's more or less the purpose. So the two cases differ quite a bit Sep 30 20:26:16 but You "waste" 100 lines on irc, just to restrain others for using shitty, bad, nut only one availabe solution ATM Sep 30 20:26:30 ok, agreed. Sep 30 20:26:59 but that alternative way wasn't implemented. I would love to test it *nd advocate it, it is written already Sep 30 20:27:36 I really dont like current solution either, but it seems no one knowledgeable want to fix it - by fizxing bq module or implementing workaround by i2cget Sep 30 20:28:07 I would 100% agree with You advocating to ban currwnt solution, if, at the meantime, You would public other, safe solution. Sep 30 20:28:48 before that, it's just banning without giving alternative solution. Mind that not everybody is coder able to implement proper way of soing it. Sep 30 20:29:21 You'rw giving us great, alternative solution and I appreciate that, but without implementation for it, it is not replacement for current bad, shitty solution. Sep 30 20:29:35 I hope You get my point without blood pressure variations ;) Sep 30 20:29:53 my blood pressure is all OK :-D Sep 30 20:30:47 DocScrutinizer: aegis Sep 30 20:30:50 * hiemanshu watches the BP rise Sep 30 20:30:54 *sigh* Sep 30 20:31:31 Doc: Why are you wasting time arguing with this guy? Sep 30 20:31:36 Estel_: Whilst I fully agree with you, the truth is that there indeed are many ignorant users. Even though I personally think it should not be done, maintaining a seperate build seems the best compromise (available to those who want it/more difficult to get for the more ignorant people) Sep 30 20:31:37 btw psycho_oreos , thanks for hint about mass storage mode Sep 30 20:31:56 idont, it's fine with me Sep 30 20:32:06 Estel_: alas I'm not set up here to do kernel development right atm, if only for missing free storage on disk and missing bucks for a fresh HDD Sep 30 20:32:28 RST38h, does I feel some "elitarish" thinkingfrom Your side? Sep 30 20:32:42 I see DocScrutinizer Sep 30 20:32:42 Estel_: Glad to hear that :). Gosh, I really started some discussion by implementing a feature I don't even use myself :-P Sep 30 20:32:49 hahaha Sep 30 20:32:56 joy of mainaining something Sep 30 20:33:00 No, Estel_ you feels regularish trolololo Sep 30 20:33:17 Estel_, did you find the answer from that hint? :) Sep 30 20:33:40 psycho_oreos, unfortunatelly not yet. It's strangw cause reboot always fix that Sep 30 20:34:06 RST38h, sorry for my english. I mean that You sounded like someone talking from "high elite position" Sep 30 20:34:27 anyway, RST38h I don't think dissing someone is a way of civilized, meritocratic discussion. Sep 30 20:34:37 even if You think I'm talking shit. Sep 30 20:35:09 I would just retain my right to disagree about patch, that I think is shitty written myself :D Sep 30 20:35:11 Estel:It is only meritocratic if you have proven your merit Sep 30 20:35:40 IMO i've proved that think disabled by default, only enabled if user do some echo, knowing all the risk - is ok Sep 30 20:35:48 I'm enjoying freedom of choice. Sep 30 20:35:56 But, we may disagree here. Sep 30 20:35:59 But it does not really matter. What matters is that you have not been able to let go of the topic for the last 3+ hours or so. Sep 30 20:35:59 -bfs7 has been pushed to Maemo's autobuilder, it should pop up in Maemo's repositories in about half an hour or so. -bfs6 (with the i2c patch enabled) can still be found at kernel-bfs' garage page Sep 30 20:36:22 Estel_: we all got your point of user being free to do what he likes, but see - even idont changed his mind somewhat and agreed to not publish it via extras-devel (the "power-users' extras repo" nowadays). His reasoning is completely in line with my concerns Sep 30 20:36:22 we have been talking \out this, then I was out for a hour or so... Sep 30 20:36:53 RST38: It is hard to let a topic go if there's still a disagreement. I suffer from that too Sep 30 20:37:10 hehe, he agrees because outcry after this, stating that he don't agree with it, just to make things less agressive ;) Sep 30 20:37:17 RST38h: ^ Sep 30 20:37:50 RST38h, we can not agree, but I don't feel like dissing You and naming like troll etc. If we're into meritocratic, dissing someone is pure Sophism. Sep 30 20:38:02 but, we can just let go that pathetic matter? Sep 30 20:38:38 I got my personal opinion about such uncivilized behavior. It does not make me not want to co-operate with You in future, anyway ;) Sep 30 20:39:05 Still, my young son is calling me, so I'm off. Sep 30 20:39:52 so bye for now, and peace please. We don't need to mock up each other, IMO. Sep 30 20:39:52 fine, so problem solved, everybody happy (more or less), todo list for the next 5 years done and agreed upon, let's have some RL - it's WEEKEND fellas! Sep 30 20:40:03 yea ;) Sep 30 20:42:48 Hehe, I've been asked to come down to a friend's house about two hours ago :-P. Seems like I should get going too. Have a nice weekend all! Sep 30 20:43:10 cya idoru Sep 30 20:43:14 idont Sep 30 20:44:15 RST38h: doesn't help to shout or ignore people, if you want to convince them Sep 30 20:45:22 so as estel obviously didn't troll on purpose (though it looked to me like he did, for a short moment), I prefer to try and get the matter discussed to an end Sep 30 20:46:13 btw estel been quite helpful on a number of tmo threads Sep 30 20:47:11 so while his merits clearly aren't based on sw development experience, he gained some merits for common sense and a helping attitude Sep 30 20:47:56 the problem here is to explain the idea of "poisonous code" Sep 30 20:48:14 to somebody not into sw development that much Sep 30 20:51:56 it's obviously not easy to explain why rat poison that looks like cookies mustn't be sold in a supermarket, no matter how explicit and bold the warning printed on the box, and how hard it is to open the box Sep 30 20:53:31 and that analogy only covers the end user aspect, not the developer side, when packages like communtiy-battery-applet (or whatever the name) depend on that poisonous feature Sep 30 23:19:30 ping idont Sep 30 23:19:46 ~seen idont Sep 30 23:19:59 idont <~iDont@j138215.upc-j.chello.nl> was last seen on IRC in channel #maemo, 2h 37m 9s ago, saying: 'Hehe, I've been asked to come down to a friend's house about two hours ago :-P. Seems like I should get going too. Have a nice weekend all!'. Sep 30 23:20:48 ping ShadowJK Sep 30 23:21:10 hello Sep 30 23:21:45 hi there. Could You please tell me, if one of symtoms from bq27200 module ioccuping i2c exclusively Sep 30 23:21:55 is that i2cget can't access it? Sep 30 23:22:01 yes Sep 30 23:22:17 damn, something went wrong with mainstream kernel-bfs update Sep 30 23:22:44 heh. Is bq27..... .ko loaded? Sep 30 23:22:49 it seems that hotfix 'forced' by DocScrutinizer for disabling dangerous paths completely, did exactly reverse Sep 30 23:22:58 yea, from start, without echoing magic Sep 30 23:23:16 Well... Sep 30 23:23:16 it just works out of the box (flawlessly, but stil...):P Sep 30 23:23:54 the funniest thing is that in bfs6 it worked well, but in bfs7 aimed to kill it, it work out of box Sep 30 23:24:04 If bq2xxxx.ko is loaded, then nothing from userspace can access the same address on i2c Sep 30 23:24:25 (except if it was opened before the kernel module was loaded) Sep 30 23:24:43 I know I may be asking noobish question, but how bme can work in such conditions? i5's out of userspace? Sep 30 23:25:18 bme gets started before bq27200.ko is loaded Sep 30 23:25:30 What happens is: bme starts, opens i2c-dev for bq27200 address. Then, the kernel module is loaded and acquires bq27200 address. Sep 30 23:25:39 stop bme; start bme -> problems start Sep 30 23:26:47 ShadowJK: actually that's another bug in that module, no? Shouldn't it either refuse to load, or kill all filehandles that are open at i2c-2 ? Sep 30 23:27:25 heh Sep 30 23:27:41 It's probably not the responsibility of the module Sep 30 23:27:55 i understand but one thing makes me wonder Sep 30 23:28:10 while using kp46-47, with module autoloaded Sep 30 23:28:22 and trying to hotswap DocScrutinizer way Sep 30 23:28:28 i was disabling and enabling bme Sep 30 23:28:33 probably it should make /dev/i2c-2 return -EBLOCKED or sth for all ioctl operations to the addr 0x55 Sep 30 23:28:46 without any problems, with bq...0.ko loaded Sep 30 23:28:57 how it's even possible? Sep 30 23:29:04 should't i get instant reboot? Sep 30 23:29:13 or infamous patch prevent that? Sep 30 23:29:50 those had the check commented out so no open() to any I2C-2 would fail ever Sep 30 23:30:10 I hope we are correct with diagnosis of problem, because I warned people in bfs thread about current version behavior. Otherwise, I'm spreading good-will FUD ;P Sep 30 23:30:20 that'S been the infamous patch, yes Sep 30 23:30:27 Frankly, it all needs a reading of the omap3 i2c driver, to see what happens if i2c-dev and bqxxxx.ko tries to submit requests simultaneously. IF it ends up as two simultaneous bitbangs, then very very bad, and random, things can happen. If the driver has arbitration and queueing, then at most the worst that happens is that the device that is being accessed gets confused (which in the case of bq27200 isn't entirely catastrophic)... Sep 30 23:31:07 and by "submit requests simultaneously", I do not mean "open device simultaneously" Sep 30 23:31:10 :P Sep 30 23:31:24 :P any way of checking that? Sep 30 23:31:27 yes Sep 30 23:32:46 ShadowJK: scheduler further mitigates the real-life problems, but it doesn't cure the basic problem. Also note carefully that this nasty nasty evil hack disabled locking for ALL i2c-dev Sep 30 23:32:51 I would love to check it once for all and settle this. If it really result in simultaneous bitbangs, I would love to not use it and advocate not doing so everywhere. Sep 30 23:33:31 do we have any i2c bus on any behind-battery test pads..? Sep 30 23:33:42 ...erm, so maybe little rewrite, to make it disabling lock only for particular i2c-dev ? Sep 30 23:34:03 so odds are there are concurrent and thus colliding accesses to e.g PMU cpu regulators Sep 30 23:34:33 I recall something related ti i2c and testpads, but DocScrutinizer probably will know it for sure Sep 30 23:34:35 I don't really understand the kernel i2c stack enough Sep 30 23:35:39 ShadowJK: the patch was pretty simple to understand: it basically just commented out the >> return(-EAGAIN); << line Sep 30 23:35:48 from i2c-dev Sep 30 23:35:50 iirc Sep 30 23:35:55 so, DocScrutinizer, would it be good thing to rewrite this path, to allow it only disabling lock for certain i2c-devs? Sep 30 23:36:07 Yeah I mean I don't have a grasp of the rest of the i2c stack :P Sep 30 23:36:57 no, it shouldn't mess with i2c-locking mechanism at all. The bq27200 driver needs moving around of a few lines - those who lock and unlock the i2c bus Sep 30 23:37:14 I'm trying to accure information how this could be done right way... Maybe I would be able to buggy someone for fix, if it is not heavy work Sep 30 23:37:48 probably author of bq27200.ko is the one you should pester about it Sep 30 23:37:53 I wonder if colin.stephane ir freemangordon could write fix for it. Colin seemed especially interested in bq module. Sep 30 23:38:08 this is maemo-specific thing, isnt it? Sep 30 23:38:12 due to bme? Sep 30 23:38:24 I think fmg is competent to implement that fix, yes Sep 30 23:38:58 I really regret that I'm not competent Sep 30 23:39:34 would be great to end this f*ckin saga. I was never dissed so many times due to some line-commented-out thing :P Sep 30 23:39:56 Estel_: that'S the problem, it's not particularly maemo specific. The patch applies to any linux, and will "work" everywhere, while you see similar "problems" like the bme-lock of i2c in 1000s of shades on arbitrary platforms Sep 30 23:40:14 I see. Sep 30 23:40:34 erm, so our fix may be even included mainstream? great ;) Sep 30 23:40:43 we just need it done :( Sep 30 23:42:12 so, if I understand it correctly, there are two options for us inside Maemo - fix this module, or buggy advanced-power-monitor author to implement i2cget method of accessing info, with correct opening and closing i2c, yes? Sep 30 23:42:19 so any semi-(in)competent "devel" anywhere facing a similar problem of i2c bus locked by a module and he wants to access the chip directly, could decide to port that """fix""" to another platform, and eventually you bet the knowledge of collision problems won't migrate with the patch Sep 30 23:42:44 yes Sep 30 23:42:52 exactly, those are the 2 options Sep 30 23:43:22 which one seems to be quicker/better for us, particullary? Sep 30 23:43:31 by logic i think second one? Sep 30 23:43:40 fix bq27200 module is the better way Sep 30 23:43:50 that would make bq module unneeded at all? Sep 30 23:43:57 so, logic failed eh ;P Sep 30 23:44:36 for widespread linux, of course it's better, that's why asked about "us" in Maemo, particullary. Sep 30 23:44:43 using i2cget or i2cdump in advanced-power-monitor is the much faster way though Sep 30 23:45:18 and will work even on stock kernel Sep 30 23:45:24 I will buggy 412b then, and meanwhile also buggy colin and freeman Sep 30 23:45:46 the worst thing is that I don't understand difference between i2cget method and i2cdump one Sep 30 23:46:16 err, the one fetches one register, the dump does a dump of all registers Sep 30 23:46:25 and I don't have heart to ask overprojected guys like You and ShadowJK to talk with him. So I'll advocate some method without knowing it well :/ Sep 30 23:46:51 which one is less resource-intensive for thing that need refresh once for few seconds? Sep 30 23:47:00 just send him here, we are all available for a short chat Sep 30 23:47:13 Yea, I know. Will try. Sep 30 23:47:38 DocScrutinizer, basically it's all design around the premise that if you have a bq27200.ko driver, then everythign that needs information from bq27200 hardware should talk to that driver. So there are poor (or none at all) locking methods for the driver in the kernel to use... Sep 30 23:48:04 if this applet needs only one value from the chip (e.g. percentage of capacity) then i2cget is the method of choice Sep 30 23:49:25 I hope that jrbme will save day for us, someday ;) Sep 30 23:49:39 if it needs more than one value (percentage, mAh, time-to-empty), then I'd suggest i2cdump, like used in http://people.openmoko.org/joerg/battery/bq27k-detail Sep 30 23:49:58 by the way, any chances for another dump of Your development booston, with fix for idle current draw? Sep 30 23:50:32 it seems that booston / kill it / boostoff doesn't work 100%, so connecting briefly to charger is a must Sep 30 23:50:46 yup, I'm hoping every day to work on it, and every evening I find myself having "wasted" my time on other things Sep 30 23:51:02 quite inconveinent during travel, so even quick dirty fix would be muchmappreciated Sep 30 23:51:04 "fix this module" vs "i2cget method" are incompatible. As I see it, there are two choices: 1) everyone uses i2cget/i2c-dev. 2) everyone uses the kernel module's /sys/...../bq27x00.ko/* Sep 30 23:51:12 much appreciated* Sep 30 23:51:41 ShadowJK, true, but without jrbme we can't achieve 2) Sep 30 23:51:48 yes Sep 30 23:51:59 ShadowJK: you missed my "specification" as of some hours ago here in this chan? Sep 30 23:53:26 DocScrutinizer, nice to know. Would test it with pleasure, so don't wait for posting it in hostmode thread, when You got something Sep 30 23:53:29 I suggested that bq27200.ko is doing a compromise between "normal" mode which locks i2c bus, and evil-nasty-patch mode that doesn't lock at all Sep 30 23:54:17 aka opening and closing every time, yep? Sep 30 23:54:30 how about we just have a daemon that writes to a tmpfs a replica of the /sys/..../bq27200/* files? ;P Sep 30 23:54:45 ShadowJK: basically bq27200.ko could exactly mimic i2cdump Sep 30 23:54:59 it's a major rewrite.. Sep 30 23:55:02 Estel_: yep Sep 30 23:55:28 hey, and the other way? rewrite of i2cdump to become new bq? Sep 30 23:55:37 probably even worse rewrite? Sep 30 23:55:40 hehehe Sep 30 23:55:44 no Sep 30 23:56:05 i2cdump is a userland app, not a kernel module Sep 30 23:56:46 I know, but I have no idea about how hard is to re-code one into another. Joys of talking with non-coder, I guess. Sep 30 23:56:48 IF we could create files in /sys, then it actually could get done by a modified i2cdump Sep 30 23:57:13 Like, give me half an hour and I could have a script that uses i2cget to create a directory structure that looks exactly like the /sys/...../bq one created by the bq27200 kernel module... But to make bq27200 kernel module lock and release resources constantly, is a big undertaking I think, as there's just not the supporting infrastructure available inside the kernel as far as I'm aware Sep 30 23:57:44 (but my directory structure copy wouldn't reside in /sys obviously) Sep 30 23:57:55 that's the point Sep 30 23:58:07 I think only kernel can create files in /sys Sep 30 23:59:01 but I can't see why we can't wrap some sort of i2cdump executable with a module framework Sep 30 23:59:23 hey, I've found pot by 412b about our ugly fast method, could You look int it? http://talk.maemo.org/showpost.php?p=1078288&postcount=703 Sep 30 23:59:53 of course I still preffer Your guys method of fixing bq module, so keep brainstorming ;) Oct 01 00:01:57 I like Your ideas, especially that actual bq module lack many usefull info that for example, bq27200.sh provides Oct 01 00:02:07 all edv, vdq, etc Oct 01 00:02:37 so baking something more usefull around i2cdump would be awesome. Oct 01 00:02:51 or at least around i2cdump method Oct 01 00:05:10 don't forget i2cdump works on STOCK kernel as well! Oct 01 00:05:28 exactly Oct 01 00:06:01 I can't follow that guy's concerns about additional overhead Oct 01 00:06:16 I would love if all of this discussed here would result not only in theoretical ideas (although great ones). Really, some ideas seem very appealing. Oct 01 00:06:28 Would *love* to see them living. Oct 01 00:06:58 I think the point is about async nature of APM monitors, whatever it is. Oct 01 00:07:08 Would You care to post here, anyway? Oct 01 00:07:21 412b is quite motivated and fast developer Oct 01 00:07:35 he implement things really fast, if enough rationale given Oct 01 00:07:36 any python dev around? how would you call a shell executable and store the stdout to some python variable? Oct 01 00:08:44 so, fixing APM would be quick fix for now, and something around i2cdump method and bq module would be ultimate fix for long-term Oct 01 00:09:15 yes Oct 01 00:10:32 brb for a while, would disconnect (checking bleeding edge drivers wifi with bfs kernel) Oct 01 00:10:50 the ultimate "fix" though is pali's project to LD_PRELOAD wrap bme to intercept direct calls to /dev/i2c-2 and map them to proper readout of sysnode values Oct 01 00:11:13 * DocScrutinizer away for a beer - finally Oct 01 00:13:44 which btw is a TYPICAL cssu project, as it needs a patch to bme (renaming/moving away, then creating a script at original location that does the LD_PRELOAD stuff) Oct 01 00:13:49 damn, it's true that bleeding edge wifi drivers doesnt work with bfs kernel... Should be compiled against bfs headers, probably. Oct 01 00:14:14 bbl Oct 01 00:14:17 o/ Oct 01 00:15:41 BTW, maybe I'm biased with bfs sheulder as well, but IMO performance is better. Seems visible especially in resource- heavy Easy Debian programs Oct 01 00:16:17 going back to our topic, does Pali method fix our all concerns with jrbme, i.e. emergency shutdowns, charging from pc etc? Oct 01 00:16:34 and dozens of Maemo elements that talk with bme? Oct 01 00:17:01 You said once, that it would need enormous rewrite of many things, some closed, to replace bme Oct 01 00:24:44 brb Oct 01 00:36:51 hello all Oct 01 00:37:03 I am trying to run some commands when my phone is in a call Oct 01 00:37:04 orangey, Good evening! Oct 01 00:37:10 hi Ken-Young! Oct 01 00:37:26 I followed this thread: http://talk.maemo.org/showthread.php?t=54039 Oct 01 00:37:49 and entered this :/usr/sbin/speedadaptfix * * com.nokia.mce.signal sig_call_state_ind active * Oct 01 00:38:02 into a file in /etc/dbus-scripts.d Oct 01 00:38:25 however, there is no apparent relief. the program appears not to execute at all. How can I debug such a thing? Oct 01 00:42:45 dbus-scripts needs to be executed Oct 01 00:43:12 this is in my ps ax: /usr/sbin/dbus-scripts --session Oct 01 00:43:55 ok so maybe the session isn't picking it up, maybe the signals are sent through system instead of session Oct 01 00:44:05 you know what.. just tried system, and that worked! Oct 01 00:44:49 there you go Oct 01 00:49:27 hah Oct 01 00:49:48 hmm. and how do I make it so that the system one loads in addition to session? Oct 01 00:50:03 that's one of the flaws of dbus-scripts app Oct 01 00:50:07 ? Oct 01 00:50:19 honestly, the n900's handling of SIP is disgraceful Oct 01 00:50:30 I think there's no concept in it for that Oct 01 00:50:30 I've never seen a device handle it so badly Oct 01 00:51:05 hehe - i've seen few devices ever hndle sip Oct 01 00:51:17 but what's your problem? Oct 01 00:51:44 DocScrutinizer: my n810 was marvelous. all of my androids are fine. hell, even every generation of iphone can do it Oct 01 00:52:03 it works fine now - I just went backward a bit after upgrading to the latest CSSU Oct 01 00:52:12 but in pr 1/1.1, it was unusable. Oct 01 00:52:24 in pr 1.2 it was mostly unusable, but occasionally usable Oct 01 00:52:38 my N810 and N900 differ only marginally regarding handling of SIP Oct 01 00:52:43 what exactly doesn't work for You when SIP'ing? Oct 01 00:52:54 starting pr 1.3, it could be used. adding power kernel made it again unusable. following that thread made it usable again Oct 01 00:52:57 I didn't think cssu would touch any of the SIP parts.. Oct 01 00:53:06 Estel_: choppy sound Oct 01 00:53:06 I got never any problems with SIP in N900 Oct 01 00:53:14 Estel_: you use kp? Oct 01 00:53:17 orangey, there is known bug for that Oct 01 00:53:29 Estel_: of course there is. you'll note it's unresolved and open Oct 01 00:53:40 of course (currently, bfs-kernel, but that isn't important for this certain case) Oct 01 00:53:41 more or less one of the devs said 'wontfix' Oct 01 00:53:48 yes. Oct 01 00:53:56 anyway, I have it working well. Oct 01 00:53:57 Still, I never encountered it ;) Oct 01 00:54:13 for me SIP is perfectly clear Oct 01 00:54:26 but as with all bleeding edge, sometimes you go backward. in this case, the 'backward' is that somehow dbus system events aren't running anymore Oct 01 00:55:30 not sure how to make it automatically load system eitehr Oct 01 00:56:50 hmm. looks like the system service no longer starts Oct 01 00:58:07 ok gentlemans, time for little sleep here. It's 3 AM, after all Oct 01 00:58:24 bb everyone Oct 01 01:03:16 sometimes I do want to pull out my hair Oct 01 01:03:58 i did update-rc.d dbus-scripts default Oct 01 01:04:01 and still it doesn't seem to run Oct 01 01:11:07 * orangey screams Oct 01 01:26:13 /etc/init.d/dbus-scripts start Oct 01 01:26:13 oh yeah.. upstart. Oct 01 01:26:24 DocScrutinizerI know that's how to start it! Oct 01 01:26:28 I want it to auto-start on boot Oct 01 01:27:58 so if that works and there's a proper link to a runlevel directory... Oct 01 01:28:14 hi DocScrutinizer51 Oct 01 01:28:17 it should start, no? Oct 01 01:28:22 DocScrutinizer51: all I can tell you is that it doesn't work from rc Oct 01 01:28:25 hi hy Oct 01 01:28:37 see http://talk.maemo.org/showthread.php?t=35567 Oct 01 01:29:19 hi i wondering do u guys know a site that has cheap lcd and digitizer for n900 Oct 01 01:29:21 I won't, too much hassle on N900 in a pub Oct 01 01:30:26 you using xchat for irc ? Oct 01 01:33:38 orangey: you're right Oct 01 01:34:20 i've installed power kernel 42 from extras, ok to install 48 over it from devel ? and does 48 work ok with bleeding edge Oct 01 01:34:24 ~ $ ls /etc/rc?.d/S*bus* Oct 01 01:34:33 DocScrutinizer51: mourn with me Oct 01 01:34:37 only shows dbus in 4 5 Oct 01 01:34:59 + is it ok to install via manager or always best with apt-get ? Oct 01 01:35:27 HyperSnyper: xchat here Oct 01 01:35:40 i couldn't figure the colours out Oct 01 01:35:44 :P Oct 01 01:35:48 didn't liek default Oct 01 01:36:12 got distracted by other things though, sos din't look at it for long Oct 01 01:36:13 heh Oct 01 01:36:24 see my files under xchat at Oct 01 01:36:50 DocScrutinizer: got it working. just edited /etc/event.d/dbus-scripts to include --session --system Oct 01 01:37:15 ~jrtools Oct 01 01:37:15 methinks jrtools is http://wiki.maemo.org/User:Joerg_rw/tools Oct 01 01:37:35 cool Oct 01 01:37:52 orangey: didn't know this works Oct 01 01:38:09 gnite! Oct 01 01:38:10 DocScrutinizer: seems to Oct 01 01:38:28 my next move is going to be nitdroid Oct 01 01:38:31 should be exciting Oct 01 01:38:32 ok Oct 01 01:39:10 orangey: you'll hate it Oct 01 01:39:21 DocScrutinizer51: which part? android? Oct 01 01:39:31 yoh Oct 01 01:39:46 also see /topic Oct 01 01:39:46 there are 3 things that would make it a no-brainer to switch to android: 1) no google; 2) conversations; 3) almost all FLOSS Oct 01 01:40:10 topic doesn't allude to android Oct 01 01:40:17 ohh Oct 01 01:40:35 damn, forgot the ibot key Oct 01 01:41:01 there are no reason ever to switch to android Oct 01 01:41:11 ~listkeys multiboot Oct 01 01:41:11 that PoS is unusable garbage Oct 01 01:41:12 Factoid search of 'multiboot' by key (2): multiboot ;; maemo-multiboot. Oct 01 01:41:38 ~maemo-multiboot Oct 01 01:41:38 somebody said maemo-multiboot was deprecated, and a horrible hack. PROBLEMS WITH NITDROID/MULTIBOOT? reflash rootfs&kernel aka COMBINED Oct 01 01:47:17 HyperSnyper: actually http://wiki.maemo.org/User:Joerg_rw/ Oct 01 01:47:56 dunno if colors are in there Oct 01 01:48:07 if not, droplme a note **** ENDING LOGGING AT Sat Oct 01 02:59:56 2011