**** BEGIN LOGGING AT Sat May 26 02:59:58 2012 May 26 09:59:37 ping freemangordon May 26 09:59:49 Pali, pong May 26 10:00:01 see log from yesterdat about bq2415x driver May 26 10:00:15 driver is working fine May 26 10:00:19 ok May 26 10:02:27 Pali, great May 26 10:02:50 Pali, could you merge those to kernel-cssu? May 26 10:03:02 ok, I will try May 26 10:03:46 BTW I put IBE bit setting patch in kernel-cssu, you may want to look at it May 26 10:04:02 That deffinitely shoulg go in KP May 26 10:04:16 Pali ^^^ May 26 10:04:32 ok May 26 10:04:43 what about to move from slow git garage to git gitorious? May 26 10:04:51 I am fine May 26 10:05:03 and have one git repository for kernel-power and kernel-cssu (with different branches)? May 26 10:05:08 actually it will be easier for me May 26 10:06:03 kernel-cssu is already on gitorious, better ask merlin1991 and MohammadAG what do they think about moving KP in CSSU project May 26 10:06:20 as a source code holder only May 26 10:07:00 otherwise I dont know how we can have branches split under different repos May 26 10:07:32 won't be easy at least May 26 10:08:20 DocScrutinizer51, ping May 26 10:23:12 freemangordon, I found problem, I forgot to blacklist bq2415x driver in my deb builds (I only blacklisted it on my n900) May 26 10:23:48 aaah, that seems the reason for my bootloop :D May 26 10:24:18 I will fix it in kernel-power git tree and then merge to kernel-cssu May 26 10:24:27 ok, thanks a lot May 26 10:25:41 Pali, BTW do we have the whole picture re BME replacement now? Do we miss something? May 26 10:26:33 bq2415x working fine for charging and boost mode May 26 10:27:01 we need only hald-addon-bme and libbmeipc for userspace May 26 10:27:17 ..and we have them semi-working May 26 10:30:53 Pali, actually libbmeipc needs only a correct temperature reading to be fully functional May 26 10:31:11 ok May 26 10:31:50 hald-addon-bme needs a little more work, i.e. charger support :) May 26 10:33:37 freemangordon, merged and commited May 26 10:33:41 but not tested yet May 26 10:34:00 Pali, thanks, will do it now May 26 10:34:30 I will build new +bq deb packages with blacklisted bq2415x_charger module May 26 10:35:06 i am going to rebuild kernel-cssu, you may want to try it, it is on -devel repo May 26 10:41:31 has kernel-cssu -bootimg and -modules package? May 26 10:41:40 yep May 26 10:41:42 if yes I can boot it from u-boot May 26 10:41:44 ok May 26 10:41:55 Pali, it is KP renamed May 26 10:42:04 ok May 26 10:46:20 freemangordon, it is possible to recompile all open source maemo core packages with thumb support when obs repository will work? May 26 10:47:39 NFC, but don't see a reason why it won't be. Do you know which gcc it will use? May 26 10:47:52 Pali ^^^ May 26 10:48:27 last gcc 4.6 or 4.7? May 26 10:49:09 hmm, sure? May 26 10:49:27 then it won't be that easy May 26 10:49:47 ok May 26 10:50:05 as most of the packages need some CFLAGS modifications May 26 10:50:28 and days ago I created program which read kernel version from mtd May 26 10:50:50 it is here: http://atrey.karlin.mff.cuni.cz/~pali/zimage-get-version.c May 26 10:51:00 how is that helpful? May 26 10:51:06 I tested it and worked fine on my device May 26 10:51:16 use it to ask user if he want to flash new kernel May 26 10:51:18 yeah, sure thing, but how that helps? May 26 10:52:01 Pali, that is why -thumb repos is May 26 10:52:06 IAUI May 26 10:52:11 *AIUI May 26 10:52:36 and I created patch (never tested) which change shell script which flashing kernel: http://atrey.karlin.mff.cuni.cz/~pali/fiasco-image-update.patch May 26 10:53:10 I still think that users should be asked before flashing kernel nto nand May 26 10:53:31 and all the stuff in -thumb repo will depend on kernel-feature-errata-xxx-workaround May 26 10:53:55 Pali, you have this functionality right now in KP (and KC) May 26 10:54:11 ok May 26 10:54:45 See, if kernel-cssu is going to make it in CSSU stble/testing, then I see the point May 26 10:55:14 But for thumb? DOn't think it will be usefull May 26 10:56:02 At least for now the userbase for thumb should be aware of the possible consequences May 26 10:56:25 i.e. new kernel, highly experimental, bla, bla May 26 10:56:59 Pali, BTW when you will start to use it? :P May 26 10:57:20 when you compile it with bq2415x driver (blacklisted) :) May 26 10:57:45 I don't mean kernel, but the userspace stuff May 26 10:59:10 maybe in july I will test thumb userspace (now I have exams month :) May 26 10:59:42 aah, ok May 26 11:00:37 Pali, BTW by testing I meant only to enable the repo, well , it seems I will have to search for another guinea pig :D May 26 11:01:33 is thumb repo ready? or it will be available when obs? May 26 11:03:21 Pali, it is May 26 11:04:12 with lots of stuff there, the only thing missing is dependency on kernel-feature-thumb in control files May 26 11:04:40 But that is not a problem so far, will add that soon May 26 11:05:05 ok May 26 11:05:24 so make sure you are running kernel-cssu before enabling it :D May 26 11:06:18 other question is, what about starting to play with upstream 3.x kernel on n900? May 26 11:06:44 Pali, I am thinking of that for a couple of weeks, but you know, lack of time May 26 11:07:15 And before doing that we should decide why May 26 11:07:21 yes, me to - no time May 26 11:07:41 what if 3.x kernel boots, which OS? May 26 11:07:52 maemo, nemo, ubuntu? May 26 11:08:43 the easiest will be ubuntu (or other arm os without special kernel like maemo) May 26 11:09:22 hm, awesome work guys (just red through logs) May 26 11:09:22 but I created patch for 2.6.37+ kernel which add some maemo sysfs entries May 26 11:09:49 well, if we are going to make Maemo distribution with upgraded kernel, I am all for that. But that is not a project for only 2 men May 26 11:10:22 Pali, I've heard that You're commiting some patches to upstream kernel May 26 11:10:37 yes for bq27x00_battery driver May 26 11:10:52 what do You think about possibilities for interesting any outside-out-little-world kernel devs in N900 kernel hacking? May 26 11:11:05 Council could help with that May 26 11:11:15 some micro-fundraising for donating N900 May 26 11:11:18 for example May 26 11:11:42 * freemangordon is afk for lunch May 26 11:11:47 I'm thinking about people that are good kernel devs but not aware of awesomness that comes with linux computer in pocket May 26 11:11:59 bon apetit freemangordon :) May 26 11:12:38 of course I don't mean giveaways of N900's to anyone who is willing to take. rather, maybe You know some kernel devs that could be teased by N900 yet lack device... May 26 11:12:47 in a way Davy's from PhoneME was May 26 11:13:03 vi donated device and results are so far so good May 26 11:13:16 maybe similar way for kernel devs could be possible? May 26 11:13:33 kernel developers are most reglamented things we have May 26 11:14:12 we got maintainers - sometimes, even too much of them ;) various kind of devs - of course, could be more, but is OK... Although, only 2 kactive kernel devs May 26 11:14:44 ...You're both doing awesome work, like small army, but You could definitely use some working hands, I suppose. May 26 11:15:51 Not to mention, that we have 0 active USB kernel devs, as AFAIK hostmode kernel implementations are kinda frozen, since Paul's departure. May 26 11:16:01 but that's different topic May 26 11:17:21 I do not have other linux mobile devices (except n900), so I worked only on n900 kernel... I do not know other linux boards... but now I'm waiting for pre-ordered RPI May 26 11:17:52 for n900 kernel, the best will be to find other kernel devs, 2 people around n900 kernel is very small number May 26 11:18:27 how can council help with kernel? May 26 11:19:28 upstream n900 kernel status is on this wiki page: http://elinux.org/N900 May 26 11:19:53 some drivers are still not upstreamed May 26 11:20:59 I have contacts for some nokia people which worked on that (not upstreamed) and authors of camera and ir drivers will upstream it May 26 11:27:51 I see May 26 11:28:36 I meant - as I've written - that Council could help with organising N900's for donation, if You could be able to find kernel devs that would like to join our awesome little world ;) May 26 11:29:45 After all, N900 is still unchallenged in many things, yet, I'm sure many talented kernel devs have no or slight idea about it May 26 11:30:29 I used Davy and his PhoneMe as example (not related to kernel directly) - he was interested In hacking for N900 due to N900's people getting in contact with him May 26 11:30:49 so, vi donated device (N900) and things started to roll out fast. May 26 11:31:48 maybe we could tease some kernel devs with N900/Maemo awesomness ;) and if they would show real interest, Council can organize device for donation (to make them really "inside" our world and easy development of kernel) May 26 11:32:41 currently, we have 3 branches of kernel-power related things, plenthora of allmighty maintainers around it and... 2 "doers"... May 26 11:34:50 problem is that I do not know any other kernel devs interested in nokia products May 26 11:35:22 and worse thing that it is not possible to buy nokia n900 from nokia May 26 11:37:21 yea, I know, but maybe some devs would be interested in having full linux handheld in pocket May 26 11:37:30 No ideap it would require diplomat :P May 26 11:39:03 maybe we should just "advertise" on (virtual) places, where kernel devs usually hand out, that we're holding project that brings upstream patches to kernel of linux-powered handheld... May 26 11:39:15 and plan to bring upstream kernel to it as a whole... May 26 11:39:41 so we're providing devices for people interested in prolonged kernel development May 26 11:40:04 let's say, our mini-micro-nano version of Nokia's device seeding activity ;) May 26 11:40:24 just idea, NFC if it would work. May 26 11:41:30 after all, kernel devs downt fall from trees on autumn, and there is thin chance that they'll come here on their own, considering niche, lack of marketing and overall Nokia road to bankrupcy stance May 26 11:41:58 we would need to show that we're kinda independent bodyp that will live with or without nokians ;) May 26 11:42:07 and will keep our projects alive May 26 12:21:24 freemangordon: pong May 26 12:24:14 Estel_: this is actually a good plan (re announcing the whole "upstream Nokia kernel patches" idea on LKML et al) May 26 12:25:35 e.g. Paul Fertser (ex kernel maintainer of openmoko and main responsible of final maemo hostmode kernel) is not interested in fremantle as it's not mainstream enough for him May 26 12:27:05 but then otoh, you're rapidly approaching mer domain by this path May 26 12:38:10 ([2012-05-26 13:15:50] Not to mention, that we have 0 active USB kernel devs, as AFAIK hostmode kernel implementations are kinda frozen, since Paul's departure.) What is it you'd want to see getting developed in hostmode kernel? Plwase accept that all the original devels think it's feature complete and no rewarding path to go ahead with any kind of development regarding h-e-n patches to kernel May 26 12:52:10 Interderesting. Someone has been trying to login to my nick. Naughty person, at least they only tried once might have been a mistake. May 26 12:55:38 DocScrutinizer, I looked into what powertop does to spit this bullshit in KP. Well, it seems to me that even under stock kernel powertop spits bullshit. May 26 12:56:01 o.O May 26 12:56:06 :-P May 26 12:56:14 i.e. it maps Cx state to frequency, which is a total nonsense May 26 12:56:24 IAUI May 26 12:56:53 it maps Cx to freq? how so? May 26 12:57:14 check the first lines, i.e C0 is on the smae line as 600 MHz, and so on May 26 12:59:16 you're missing the strong separator line between col3 and col4 May 26 12:59:22 they're unrelated May 26 13:00:01 optically messy, but technically OK May 26 13:00:11 yeah, sure, but most(including me) will be left with the impression for that May 26 13:00:46 i.e. that time spent in Cx is somehow related to the frequency May 26 13:00:51 sure thing, but a ticket against gfx designers, not against powertop core code and functionality May 26 13:01:15 I agree it's extremely irritating May 26 13:01:32 DocScrutinizer, BTW powertop (not the maemo one) is OSS, are you aware of anyone trying it? May 26 13:03:03 Adn there is some additional code on TI site, to be more compatible with OMAP May 26 13:03:15 *And May 26 13:10:14 freemangordon: (foss powertop) it seems it's massively x86 centric May 26 13:10:32 freemangordon: powertop|sed "s/\( *C[0-5] | *[0-9\.%]* | *[^ |]* *\)|/\1#/" May 26 13:13:21 DocScrutinizer, yeah, sure thing, but it can be used as a base, most of the stuff will be there (i.e. process parsing), OMAP specific stuff coulld be easily added, after all all that powertop does is to read sysfs nodes May 26 13:13:43 yup May 26 13:14:41 I'm wondering if we could find enough of FOSS code in maemo powertop binary to request Nokia to disclose the GPL source (if the original powertop is actually GPL) May 26 13:14:54 BTW the pug to open maemo powertop is assigned to qgil, we can ask Estel_ to pester him to give us the source May 26 13:15:03 s/pug/bug/ May 26 13:15:05 freemangordon meant: BTW the bug to open maemo powertop is assigned to qgil, we can ask Estel_ to pester him to give us the source May 26 13:15:07 :-D May 26 13:15:17 why not May 26 13:15:19 great minds think alike, it seems May 26 13:15:26 ;-) May 26 13:15:56 yeah, sure :D May 26 13:18:56 DocScrutinizer, I dont think "the legal" approach will work. And most-probably (and it is mentioned in a thread on TMO) maemo powertop is a re-write. So we could asc qgil to do us a favor, nothing more May 26 13:19:01 *ask May 26 13:37:21 Pali, kernel-cssu with charger support and thumb errata workaround is in cssu-devel repo May 26 13:37:48 freemangordon, ok I will install -bootimg package May 26 14:17:38 freemangordon, uname -a: Linux Pali-Nokia-N900 2.6.28.10-cssu1 #1 PREEMPT Sat May 26 09:16:14 EDT 2012 armv7l GNU/Linux May 26 14:23:17 :-) May 26 14:35:32 freemangordon, I do not know if you read jonwil post about hald-addon-bme, so here is: http://lists.maemo.org/pipermail/maemo-developers/2011-June/028452.html May 26 14:49:13 does somebody know how to convert Gauss-Kruger coords to Lat/Long May 26 15:13:02 sure, Pali May 26 15:13:53 DocScrutinizer, is for that simple math formula? May 26 15:14:11 http://lists.openmoko.org/pipermail/openmoko-kernel/2008-June/002987.html May 26 15:14:20 some background May 26 15:15:35 http://www.google.de/search?q=gauss-krueger+conversion May 26 15:19:47 http://www.moenk.de/index.php?serendipity[subpage]=downloadmanager&thiscat=4&file=16 May 26 15:19:56 GK_nach_GPS.xls May 26 15:50:40 DocScrutinizer, re kernel hostmode patches - I may be wrong, but I remember You was talking about lack of tuning clock's things in hostmode ekrnel patches, that result in MAX 4MB/s transfer via hostmode May 26 15:51:16 +, long time ago, we were talking about something hostmode related, and You told me that it should be done in kernel side, but since Paul departure, no USB ekrnel developer + it's niche... but I can't remember what was that about May 26 15:51:28 of course take all of this with grianj of salt, as I'm not knowledgeable about hostmode internals May 26 15:51:56 ah, probably, it was about idle power draw May 26 15:52:01 which can be ugly workarounded outside kernel patches May 26 15:52:09 but proper solution would be in kernel implementation - IIRC May 26 15:52:23 so i had impression, that there is some work left in kernel side, that cant be accomplished May 26 15:52:33 + don't get me wrong, I agree that hostmode is - generally - feature complete May 26 15:53:49 freemangordon, this thing You've though I could ask Qgil to open - powertop for Maemo - does we have enough documented open bits, or it's still pending work? May 26 16:15:55 I'm learning how to properly use steam-mop in housework cleaning... It's more complicated that any shell-scripting! May 26 16:31:40 Estel_ : Is it on purpose config file of ereswap is resetted after reflash? May 26 16:32:49 honestly? No, but /home/.ereswap should be touched via reswap? 0_o May 26 16:33:07 someone already suggested to move adjust-rcS-late and update-rcS-late out of there, because it *isn't* touched after reflash May 26 16:33:20 soudns reasonable, as after reflasdh, you're using - at least for a while - standard flash partition... May 26 16:33:26 but, still, I haven't had this in mind May 26 16:47:35 Let me explain, I reflashed combine only, all apps are gone. Then reinstall ereswap and voila, config file is overwritten May 26 17:36:52 well, obviously, during reinstall, config files are overwritten May 26 17:57:39 Pali, yeah i read it. Maybe we could ask jonwil to continue my work (la addon and libbmeipc) May 26 17:57:52 *hal addon May 26 17:58:11 but jonwil is offline May 26 17:58:29 Pali, we can PM him for example :D May 26 17:58:42 ok :) May 26 18:00:03 Estel_, I don't understandyour question (re qgil), but there is two-years old bug re opensourcing powertop, and qgil is assigned that bug May 26 18:00:46 Pali, now you have the correct kernel, you may enable -thumb repo and do apt-get update/upgrade May 26 18:01:17 and tell me (or note somewhere) what apt-get thinks about what space will be freed May 26 18:01:53 freemangordon, I will play with -thumb later May 26 18:02:52 * freemangordon wishes javispedro was here, AIUI he thinks he has a piece of SW that will crash if thumb-compiled May 26 18:03:09 dosbox? May 26 18:03:50 Pali, sure, just don't forget to write down how many MBs are freed May 26 18:04:15 ok May 26 18:04:49 freemangordon, you can try enable -thumb repo in qemu if you need to know MBs... May 26 18:04:57 qemu? May 26 18:05:04 n900 qemu images are part of madde May 26 18:05:04 you mean in SB? May 26 18:05:11 no madde here May 26 18:05:16 no real system n900 qemu images May 26 18:05:25 ok, besides Qt creator May 26 18:06:20 it is good for testing if maemo system will be bootable after core changes (e.g kernel, bootloader, libraries or init scripts...) May 26 18:06:49 Pali, I can reflash my second device and do it, but if you are goint to do it, then there is no sense May 26 18:07:04 Pali, I am always testing on the device :) May 26 18:07:29 I will play with -thumb next week May 26 18:08:01 * freemangordon heads for another coffee, hoping it will make his typing skills better May 26 18:08:07 madde also working on windows May 26 18:09:13 and about qemu runtime is wiki page: http://wiki.maemo.org/MADDE/QEMU_runtime May 26 18:15:43 evening chaps May 26 18:17:33 hi May 26 18:19:07 Pali, maybe it is a good idea to send a mail on maemo-devels mailing list re hal addon and libbmeipc FOSS replacement and their current status May 26 18:20:16 ok, write mail. you did initial code May 26 18:20:41 I am not sure I am subscibed :D. Let me check May 26 18:22:13 I have an idea... May 26 18:22:42 with android, a whole new slew of frequency governers have been invented May 26 18:24:01 vi___, android does not work like linux AFAIK, they don't use the same PM methods like OMAP kernel does May 26 18:25:05 i.e. you're either off or full on, no PM domains managements. And that is why android suck on battery May 26 18:25:26 freemangordon: I am talking about frequency scaling governer like ondemand and conservative. May 26 18:25:32 might be wrong, but I think felipe contreras wrote something about that on his blog May 26 18:25:43 vi___, I got your point May 26 18:25:57 just saying that AFAIK it is a bit different May 26 18:26:06 ok May 26 18:26:21 freemangordon: well what I was going to say was May 26 18:26:37 there is this governer that behaves like ondemand upto a set frequency May 26 18:26:49 the behaves like conservative above that frequency May 26 18:27:08 vi___, http://felipec.wordpress.com/2012/01/04/android-vs-maemo-power-managment-static-vs-dynamic/ May 26 18:27:27 vi___, but what will happen then with the whole race-to-idle concept? May 26 18:27:40 freemangordon: I think it is overated May 26 18:27:49 might be ;) May 26 18:27:59 when you have the device in your pocket doing nothing, maybe May 26 18:28:08 when you are listening to an mp3 May 26 18:28:17 you have a constant small cpu load May 26 18:28:29 and the processor NEVER sleeps May 26 18:28:36 incorrect May 26 18:28:40 it sleeps May 26 18:28:45 well, ok. It does a little May 26 18:29:14 if you run powertop when playing an mp3, ondemand May 26 18:30:03 if you run powertop when playing an mp3 with 250-600 compared to 250-900 May 26 18:30:08 you will see no differance May 26 18:30:38 no difference in? May 26 18:30:45 Cx states? May 26 18:30:47 t5ime spent in c4 May 26 18:30:56 or c3 May 26 18:30:58 or c2 May 26 18:30:59 or c1 May 26 18:31:24 hmm, that is strange May 26 18:31:50 you've just thrown race-to-idle through the window :D May 26 18:31:57 yup May 26 18:32:04 got any other paradigms I can shatter? May 26 18:32:23 you wait till you hear what I have to say about qauntum mecanics May 26 18:32:30 :D:D:D May 26 18:32:47 you can start with thumb2 in that regard :P May 26 18:32:53 bwahaha May 26 18:32:57 way out my league May 26 18:33:21 anyway May 26 18:33:29 assuming I am correct about the above May 26 18:33:49 yeah, I got the idea May 26 18:34:05 it would make more sense to have the CPU jump to a maximum of say 600MHz (the best power/speed) point. May 26 18:34:22 "stay on 5509600 for as long as possible" May 26 18:34:30 and if it needs more, then to scale up the available freqs, just like conservative May 26 18:34:40 s/5509600/550-600/ May 26 18:34:40 freemangordon meant: "stay on 550-600 for as long as possible" May 26 18:34:53 makes sense May 26 18:35:32 do yu have any hard data (powertop) while laying one and the same mp3, under one and the same conditions? May 26 18:35:54 not on me but I could make it and pastebinnit May 26 18:36:03 it just does not makes sense that we decode the same stream for the same time under 600 and under 900 May 26 18:36:21 there is something more here May 26 18:36:42 perhaps the sampling rate is too low? May 26 18:37:04 NFC May 26 18:37:59 hmm, it seems the second coffie has no effect on my typing skills, maybe i should try with a AJck Daniels? May 26 18:38:28 perhaps mixed with a little cola? May 26 18:38:43 derp May 26 18:38:48 naah, only ice May 26 18:38:54 maemos pastebinit is depreciated May 26 18:39:05 it should work AFAIK May 26 18:39:07 freemangordon: then you simply must meet my good friend single malt May 26 18:40:19 THIS API HAS BEEN DISABLED. Please use Pastebin's new API. http://pastebin.com/api May 26 18:40:26 son of a gun May 26 18:40:36 pastebinit was damned useful May 26 18:41:12 WTF, it was FB first, now pastebin. May 26 18:41:29 smells like conspiracy May 26 18:41:35 :D May 26 18:44:25 MohammadAG, ping May 26 18:47:28 vi___, BTW i got a better idea - use ondemand's behaviour when unlocked, limit the frequencies to 500-600 when locked May 26 18:48:10 something like batterypatch :P May 26 18:48:42 lols May 26 18:48:48 ... May 26 18:48:53 that way you can listen to your mp3s with the best power ratio May 26 18:49:12 freemangordon: actually the best way I have found to listen to mp3 is... May 26 18:49:18 with conservative governer May 26 18:49:29 limits 250-900 May 26 18:49:36 up threshold 67 May 26 18:49:41 down threshold 65 May 26 18:49:56 the n900 hangs around 250 May 26 18:50:03 touching 500 from time to time May 26 18:50:30 as oppose to hanging around 250 tehn whizzing upto 900 from time to time May 26 18:50:37 and here is the ball breaker May 26 18:51:11 there also seems to be no effect on the c states used in comparison to ondemand May 26 18:51:35 vi___, that does not make sense May 26 18:51:39 :( May 26 18:51:41 freemangordon: no May 26 18:52:55 vi___, it might be that we don't stay idle for long enough to enter a different c state? May 26 18:52:58 tis a noodle scratcher to be sure May 26 18:53:25 so locking to 900 should affect c states May 26 18:53:28 freemangordon: I am not sure how the C state change algorithm works May 26 18:53:46 freemangordon: I imagine so. May 26 18:55:00 from what I know every c-state has a set of power domains to turn off and a latency to enter/exit. Not sure how the decisions are taken though May 26 18:56:21 well the good news is, debian wheezy pastebinit will straitup install with no fiddling May 26 18:56:39 just wget the deb, dpkg -i that mutha, job done May 26 18:56:40 it works? May 26 18:57:00 iirc the audio buffers are so tiny the cpu can't really sleep? May 26 18:57:40 although when playing with ffmpeg's decoders, aac and mp3 used far less power than default codecs May 26 18:59:22 hmm. But then if we lock the max frequency there should be a change in c-states picture. May 26 18:59:43 not that it will be more efficient May 26 19:01:11 ooh, 67% C2 playing mp3 with default media player May 26 19:01:50 31% C0, 80% 250MHz May 26 19:02:04 does not look good May 26 19:02:23 no C4 at all? May 26 19:03:35 no May 26 19:04:01 great, which kernol do you use? May 26 19:04:11 this is stock May 26 19:05:08 http://slexy.org/view/s2MeI6rWt1 May 26 19:05:17 let me check how is PK with 250-805 May 26 19:05:43 avg 5ms in c2 May 26 19:06:01 ^ open media player + mp3 + pk50 + ssh over wifi (no activity other than playing mp3, screen locked) May 26 19:06:18 ^locked at 900 MHz May 26 19:08:12 http://slexy.org/view/s27UIVhKkl May 26 19:08:23 ^ the same at locked @ 250 May 26 19:08:28 I stand corrected May 26 19:08:36 there is a differance May 26 19:08:57 yeah, and it is major May 26 19:09:16 15.6 vs 48.1 in C0 May 26 19:09:18 i thiknk you need 500ms sleep time to enter c4 May 26 19:09:30 and audio buffer isn't big enough to sleep 500ms :) May 26 19:09:48 yeah :) May 26 19:10:24 but that sems to be a problem of mafw-gst-renderer not providing long enough queue May 26 19:11:40 vi___, could you try the same mp3 via gst-launch and a big output queue placed before audiosink May 26 19:11:48 sure May 26 19:11:51 ... May 26 19:11:59 wot codes i need to write? May 26 19:12:09 aah, let me do it May 26 19:12:27 I am somewhat unfamiliar with gst-launch May 26 19:12:38 ok, I will do it May 26 19:13:11 hey, myt N900 sleeps properly when playing music May 26 19:13:21 you need about 2000ms hardware buffer, in 500ms chunks, so that sound hw irq rate is 2 per second or so May 26 19:13:22 but I use ogg ;) May 26 19:13:27 dunno if the hw is capable of that May 26 19:13:33 software buffers do little May 26 19:13:35 it properly enters c2 and c3 May 26 19:13:39 sometimes, even c4 a little May 26 19:13:48 so my results are *not8 in line with vi_"s May 26 19:13:56 BTw, I got few news, 1 good and 2 bad May 26 19:14:08 Estel_: ya... May 26 19:14:15 good is that i know what was causing C! bug - as i expected, goddamned fm radio. I jsut got call today while listening to it, and it got stuck May 26 19:14:18 is it about my n950? May 26 19:14:32 of cours,e bug appear then May 26 19:14:35 2nd new,s bad one May 26 19:15:06 as addition to keeping cpu in C1, borked fm radio also disables microphone (softwareish) - it's either disabled at all, or expect bluetooth - anyway, both in call and while using dictaphone, it doesn't work May 26 19:15:12 messing with alsa doesn';t help, only restart May 26 19:15:41 3th news, also bad, or even worse - whiole fm radio dies out in bad way, it not only keep cpu in c1 state and diusables microphone, it also have good chance to screw Your rootfs during shutdown May 26 19:15:45 I've no idea WTF... May 26 19:15:58 But second time in this week I had to restore rootfs from backup, *only* rootfs May 26 19:16:07 and everytime it happen, it's when FM radio get stuck May 26 19:16:18 BTW, it's *fiorst* two times ever, when I have to restore rootfs May 26 19:16:38 from recovery console, rootfs is accesible and files look OK, although Maemo doesn't even play 5 dots May 26 19:16:55 after restoring rootfs from bacup, it automagically works again May 26 19:17:10 I have no freakin idea how borked FM Radio app is able to fuck rootfs that should be unfuckable, but... May 26 19:17:15 just a word of warning May 26 19:17:40 vi___, I've no idea, mines sleep wel as soon as i've switched from using Easy Debian's xbindkeys to Maemo's xbindkeys May 26 19:18:17 Estel_: I have always used maemos xbindkeys May 26 19:19:08 even for the middle/right click thing May 26 19:20:18 for middle-right click you're probably using maemo's xdotool? May 26 19:20:25 invoked via xbindkeys? May 26 19:20:31 yup May 26 19:20:41 same here May 26 19:20:44 hey May 26 19:20:49 I like easy debian May 26 19:20:54 have You managed to make maemo's xbindkeys to start with boot? May 26 19:20:59 but only for some things May 26 19:21:06 Estel_: yes May 26 19:21:10 same here, but easy debians xbindkeys is borked, it keeps CPU in C3 and use 200+ wakeups when idle May 26 19:21:17 vi___, how?! I can't manage to do it May 26 19:21:23 every way I tried via etc/event.d failed May 26 19:21:28 Estel_: I know :) May 26 19:21:42 it took me a long time to work out May 26 19:21:58 i tried it alone, via osso-xterm 'xbindkeys', wrapped inside shell script, run-standalone.sh, everything May 26 19:22:04 I needed it for mounting/unmounting special loop back files for holding maps and stuff May 26 19:22:05 so share Your ancient knowledge, pls ;) May 26 19:22:14 that I do not use anymore May 26 19:22:26 you need to use the X11 startup files May 26 19:23:23 /etc/X11/Xsession.dcat /etc/X11/Xsession.post/80fullscreen May 26 19:23:23 #!/bin/sh May 26 19:23:23 xbindkeys -f /home/user/.xbindkeysrc May 26 19:23:24 cat /etc/X11/Xsession.post/80fullscreen May 26 19:23:24 #!/bin/sh May 26 19:23:27 xbindkeys -f /home/user/.xbindkeysrc May 26 19:23:34 oops, double post May 26 19:23:43 Estel_: do you see? May 26 19:24:51 yea May 26 19:24:53 thanks a lot May 26 19:25:18 but it need to be in two places? May 26 19:25:34 ? May 26 19:25:41 BTW, I just realized I've lsot 2nd time my work on fixing eject.sh by blueled May 26 19:25:42 I accidentially posted it twice May 26 19:25:46 * Estel_ is going berseker May 26 19:25:56 no no May 26 19:26:01 /etc/X11/Xsession.dcat /etc/X11/Xsession.post/80fullscreen May 26 19:26:16 those 2 files May 26 19:26:25 ok ignore EVERYTHING i have posted May 26 19:26:30 =) May 26 19:26:32 I will get it right this time May 26 19:26:35 cat /etc/X11/Xsession.post/80fullscreen May 26 19:26:35 #!/bin/sh May 26 19:26:35 xbindkeys -f /home/user/.xbindkeysrc May 26 19:26:45 ah, simple as that May 26 19:26:46 so I have a file /etc/X11/Xsession.post/80fullscreen May 26 19:27:05 that contains the script: May 26 19:27:06 #!/bin/sh May 26 19:27:12 xbindkeys -f /home/user/.xbindkeysrc May 26 19:27:18 roger that May 26 19:27:20 thanks a lot May 26 19:27:24 literally that simple May 26 19:27:34 This is how I start most of my scripts May 26 19:27:48 configure frequencies and DSP speeds etc May 26 19:27:55 so I have no need for kernel config May 26 19:28:01 wuuuuut? My N900 isn't booting again May 26 19:28:04 lols May 26 19:28:26 what the fuckin fuck? May 26 19:28:36 Estel_: sounds like you have someting fundamentally broken May 26 19:28:48 like nand or something May 26 19:28:59 sounds like I've managed to get something fundamentaly broken in 15 minutes time from recovering backup and restoring things May 26 19:29:19 nope, this time I expect that i've fucked something.. wlel, what I've did... May 26 19:29:22 Estel_: perhaps disable watchdogs in R&D mode. May 26 19:29:39 I would rather expect that my rcS-late gone apeshit May 26 19:29:40 that would eliminate a lot of possibilities May 26 19:29:47 but indeed You're right also May 26 19:30:14 will try to edit rcS-late via vi and check wtf May 26 19:30:30 maybe my own script from ereswpa, that is working for others, misbehaved with my rcS-late May 26 19:30:33 some forgotten mod or something May 26 19:30:43 I've checked it bvriefly - maybe TOO briefly May 26 19:31:36 did you install batterypatch? May 26 19:31:53 very funny May 26 19:32:09 LOL May 26 19:32:20 BTw I'm using You, vi May 26 19:32:23 to edit my rcS-late May 26 19:32:30 ha! May 26 19:32:44 does anybody have unmodified one and could pastebin it? I would use it as reference May 26 19:33:07 because overall it looks legit, but maybe sed did something small... May 26 19:33:18 what is it you need? May 26 19:33:21 what file? May 26 19:33:45 erm, who remember command to delete whole line in vi May 26 19:33:50 dd May 26 19:33:51 /etc/even.d/rcS-late May 26 19:33:58 ah, thanks May 26 19:34:08 ?etc/event.d/rcS-late May 26 19:34:09 I mean May 26 19:34:11 fuck May 26 19:34:18 http://slexy.org/view/s20wjbyjyg May 26 19:34:19 /etc/event.d/rcS-late May 26 19:34:21 thanks May 26 19:34:25 this time I made it without typo May 26 19:34:39 Estel_ May 26 19:34:44 dpkg -S to search for the files May 26 19:34:48 then you know the package May 26 19:34:52 you can download the package and unpack it May 26 19:35:03 and it will contain an unmodified then. May 26 19:35:48 NIN101: protip May 26 19:36:35 brb, gotta fix meh motorcycle. May 26 19:36:50 Nin101, thanks, but when my N900 isn;'t booting May 26 19:36:53 I cant ;) May 26 19:36:59 oh-huh May 26 19:37:03 editing rcS-late fixed it May 26 19:37:03 sorry, didn't have context :-) May 26 19:37:15 BTW, vi, you've modified rcS-late, some remnants from additional mounts are there ;) May 26 19:37:21 NIN101, no problem, thanks for tip anyway May 26 19:37:29 Estel_: yeah... May 26 19:37:34 just ignore that May 26 19:37:39 now, what I've had fucked in my few pro commands that work for everyone else except the author May 26 19:37:48 be glad I had the common sense to comment it when I did it. May 26 19:37:58 I haven't had rain of reports about unable to boot after installing ereswap, so it seems I'm only one affected, WTF May 26 19:38:06 ;) May 26 19:38:15 sounds like bad juju May 26 19:38:21 yup, definatley bad juju May 26 19:38:29 I just deleted some lines that shouldn't cause any problems and replaced it with vanilla rcS-late content, which, BTW, I know form memory now :p May 26 19:39:02 you must make a pentagram of n900s connected by a token ring ethernet and sacrifice an n950 in the middle May 26 19:39:22 ;P May 26 19:39:33 for now, I managed to boot it no problem with 200 MB swap ;P May 26 19:39:40 it is the only way to appease the angry gods that smite your boot sequance May 26 19:40:02 yea, wil ldo it as soon as I call my son, as i should did it 45 minutes afo, fuck the alck of dumb phone May 26 19:40:06 brb May 26 19:40:12 s/alck/lack/ May 26 19:40:32 Heh when my n900 breaks I am going back to a symbian May 26 19:41:59 if my breaks - and I really doubt it - I'm going another N900 May 26 19:42:23 s/my/mine/ May 26 19:42:24 Estel_ meant: if mine breaks - and I really doubt it - I'm going another N900 May 26 19:45:31 ok, now i've serious question May 26 19:45:47 vanilla rcS-late contains line: May 26 19:46:30 /sbin/swapon -a May 26 19:46:34 I've replaced it with: May 26 19:50:28 http://sebsauvage.net/paste/?92f8a16ed86860f9#jtkNwnaeL88P8gtj7EPLLHhNwcUHDxubq3ZNrHc0cLY= May 26 19:51:06 that was what I've replaced it with May 26 19:51:14 now, what is so FCKD in it May 26 19:51:17 that it works for everyone else, but fails on me May 26 19:51:42 i9.e. result in system failing to display 5 dots and kept forever with black screen and backlight on May 26 19:51:52 (+keyboard backlight turning on when you expand keyboard) May 26 19:52:06 it can';t be even called reboot loop, as it just sits like that May 26 19:52:14 vi, do you have ereswap installed? May 26 19:52:31 s/vi/vi___/ May 26 19:52:31 Estel_ meant: vi___, do you have ereswap installed? May 26 19:55:46 strange May 26 19:56:05 now when i think about it, it seems to me that I had same settings yesterday, and shutdown/poweron many times, already May 26 19:56:12 trying to make xbindkeys working properly May 26 19:56:48 but today, it made device non-bootable, and removing this line made it bootable again May 26 19:56:52 something is wrong here May 26 20:17:57 ping vi___ May 26 20:30:47 * Estel_ investigates why few lines from zerobin above make device unbootable, while it worked just yesterday May 26 20:56:29 merlin1991, ping May 26 21:07:25 AAAAARGH May 26 21:07:37 it seems, that command responsible for making device unbootable was May 26 21:07:40 blkid -g May 26 21:07:44 WTF I ask, wtf? May 26 21:07:57 without it, it works (although, defeats the purpose) May 26 21:08:15 why the hellish hell blkid -g make device unbootable?! May 26 21:14:18 DocScrutinizer, ping May 26 21:14:27 DocScrutinizer51, ping May 26 21:16:42 could anyone tell me any possible reason why simple command: May 26 21:16:45 blkid -g May 26 21:17:01 put in rcS-late, just before enabling swap, worked for few days flawlessly... May 26 21:17:21 although now, beyong any doubt, make whole rcS-late go apeshit and device staying before 5 dots forever? May 26 21:17:36 it's driving me insane May 26 21:19:10 mhm May 26 21:19:29 Estel_: you made this re-swap thingy? May 26 21:19:39 yea, few days ago May 26 21:19:42 it's working fine for everyone May 26 21:19:48 nope not for me :) May 26 21:19:52 but ijh my device, blkid -g part made it unbootable May 26 21:19:54 hm? May 26 21:20:01 wewhat doesn't work for You there? May 26 21:20:08 my 2nd swap is a swapfile not a blockdevice May 26 21:20:37 I see. May 26 21:20:43 anyways i was thinking about rewriting it for me May 26 21:20:44 well, it wasn't intended to work with swapfiles May 26 21:20:48 but May 26 21:20:55 you missed the dependency on sysstat too May 26 21:20:57 if Yuo manage to make it compatible with both, surely send a patch May 26 21:21:05 is there a git somewhere? May 26 21:21:12 honestly, it's dependency on busybox-power May 26 21:21:25 as systat is in non-default repo May 26 21:21:31 its in tools May 26 21:21:40 but, someone already suggested way of collecting those data, that doesn't involve iostat May 26 21:21:48 unfortunelly, no git yet May 26 21:21:52 but it didnt installed busybox-power for me yesterday either just brole May 26 21:21:54 broke May 26 21:21:59 it was my first package that amde it into repo May 26 21:22:06 ok May 26 21:22:24 for sure i'll try to use way that doesn't require busybox-power May 26 21:22:25 busybox power has also a swapon which supports swap priorities? May 26 21:22:31 afaik yes May 26 21:22:41 busybox power has everything like it *should* look in busybox May 26 21:22:57 = upstream version (in practice) + all tools that should come with busybox May 26 21:22:58 you can read the stuff systat gives you plainly from /proc or /sys .. its just some more fiddeleing May 26 21:23:05 yea May 26 21:23:08 that could make this dependency opsolete May 26 21:23:47 well i like the idea to measure the write traffic .. but wont it be enough to measure it on the primary device and if expired recycle both swaps May 26 21:24:21 and then perhaps have both swaps at the same priority (long time ago i tested that i should test that again) May 26 21:24:45 no, for sure not. May 26 21:24:54 first of all, having swap in eMMC enabled decreases performance May 26 21:24:59 if swaps have the same pri then the kernel distributes the load on them (i have a slow µSD card btw) May 26 21:25:04 no matter if it's part of swap pair, or 32754355 swap devices qworking paralelly May 26 21:25:11 yea, in theory May 26 21:25:19 in practice, flash based RAId 0 isn't good idea ;) May 26 21:25:26 yes i know my tests 2 years ago wherent good May 26 21:25:26 and, it collides with request from optfs May 26 21:25:33 i hoped that changed meanwhile May 26 21:25:52 optfs on eMMC also require constant access, and flash media doens't handle colliding request very well May 26 21:26:07 swap on even slowest microSd is still better performance-wise May 26 21:26:10 a shame that compcache/zram is broken too :/ May 26 21:26:29 in fact I've no idea why ramzswap doesn't give us benefits May 26 21:26:33 i have a lot data on microsd (formatted as ext4) May 26 21:26:58 so there is quite some load on it .. especially maps for monav .. May 26 21:27:41 I see. May 26 21:27:53 Well, best way is to have one medium "static" and put swap there May 26 21:28:02 kernel has bug with swapfiles on vfat atleast May 26 21:28:03 if You have microSd good enough and reliable, You can put optfs there May 26 21:28:11 but honestly May 26 21:28:15 i would mvoe monav things to emmc May 26 21:28:28 remember, that swap on emmc wear it out, and it's non-replaceable medium May 26 21:28:35 microSd's are cheap and replaceable May 26 21:28:42 ShadowJK, do you have idea why May 26 21:28:45 blkid -g May 26 21:28:50 as aprt of rcS-late May 26 21:28:52 yes .. but its a squashfs bigger than 2GB :) May 26 21:28:55 = unbootable device? May 26 21:29:06 when does emmc get mounted May 26 21:29:13 and when does emmc driver get loaded? :) May 26 21:29:22 * ShadowJK bets it's related May 26 21:29:22 and on the other hand laptop_mode = 600 ... gives quite some good avoidance for writes May 26 21:29:37 ShadowJK, but blkid does have anything to do with eMMC? 0_o May 26 21:29:43 oh? May 26 21:29:47 cehteh, it's a suicude, professional one May 26 21:29:54 either way .. i customized my device a bit so i cant complain that it is incompatible with some stock packages May 26 21:30:19 laptopm mode 600? I wouldn't want to be in Your skin, after You hit any serious need for writing swaps May 26 21:30:21 there is no much write load on my emmc too May 26 21:30:38 + using file as swap is always slower May 26 21:30:47 it's not May 26 21:30:49 it isnt since ages May 26 21:30:50 sounds quite like a bit of masochism performance wise May 26 21:30:56 ok, sorry May 26 21:30:59 in 2.2 kernels or so it was May 26 21:31:05 But if you have a swap partition you can be sure of the alignment May 26 21:31:05 I always though onj our flash media it is. May 26 21:31:19 well currently i dont use it either May 26 21:31:21 acknowledged May 26 21:31:29 only using the stock swap partition May 26 21:31:46 poor is Your eMMC :) May 26 21:31:48 Basically it asks where the blocks belonging to that file is, and uses block level access bypassing the filesystem May 26 21:32:15 the write load is not *that* much from swap .. May 26 21:33:05 even if you write a fee hundred MB per day (which is prolly not the case) emmc will last quite long May 26 21:38:17 ShadowJK, little question if You don't mind May 26 21:38:37 do You happen to know if blkid -g is required for "just after boot" situation? May 26 21:38:50 no idea May 26 21:38:56 one line down I'm doing little grep from blkid info May 26 21:39:07 it was mean't to clear blkid's data before May 26 21:39:14 suprisingly, invoking blkid works well May 26 21:39:30 *only* -g (clearing) makes everything go apeshit May 26 21:39:46 and only if invoked from rcS-late May 26 21:40:01 BTW, it seems to work for other people, which is very strange IMO May 26 21:40:14 I could almost swear it worked for me yesterday, too May 26 21:40:24 voodoo programing May 26 21:43:25 -g May 26 21:43:27 Perform a garbage collection pass on the blkid cache to remove devices which no longer exist. May 26 21:43:48 perhaps it is relying on this cache? May 26 21:44:15 why do you need to -g anyway/ May 26 21:44:16 ? May 26 21:44:33 also, swapfile==pain in the ass May 26 21:44:44 so you have swapfile on usd? May 26 21:45:13 you have to umount swapfile before you can connect n900 by usb mass storage mode May 26 21:45:25 so you need to put it on a seperate partiton on USB May 26 21:45:40 if that is the case you may as well just use a swap partition... May 26 21:45:47 unless I have missed something May 26 22:05:03 vi___, no, not m,e with swapfiles May 26 22:05:26 i need garbage collection pass, or I can have data in blkid belonging to long time non-existing swap partitions May 26 22:05:51 but I'm not sure if it's the case also, when device just booted May 27 00:01:18 I've uploaded a new version of the restinpeace package today in the morning, but it's not showing on the device while it's showing as the latest on http://maemo.org/packages/view/restinpeace/ May 27 00:01:54 is it something wrong with my package cache or with the server? May 27 00:02:17 What version do you see on the device? **** ENDING LOGGING AT Sun May 27 02:59:59 2012