**** BEGIN LOGGING AT Thu Jul 28 02:59:57 2011 Jul 28 03:47:23 ls Jul 28 03:47:27 sorry Jul 28 03:47:30 wrong window ;) Jul 28 03:49:48 http://t.co/ca31My7 **** ENDING LOGGING AT Thu Jul 28 04:47:30 2011 **** BEGIN LOGGING AT Thu Jul 28 04:42:15 2011 Jul 28 05:08:13 When you boot fresh and open conky for example - are you using any swap? If so, then how much? Jul 28 05:08:35 Is it normal that I'm using just shy of 70megs? Jul 28 06:26:42 fauxfone: likely .. i have about 200MB swap in use usually Jul 28 06:27:11 dont care right after a reboot, normal usage matters Jul 28 07:06:11 cehteh: well, swap isn't supposed to be in use that often, right? I mean, if the design of a device and OS requires swap to be used just to boot.... Jul 28 07:08:00 Erm, IIRC, it's been designed to aggressively swap whatever it can, so there's memory for new programs Jul 28 07:08:26 aand the defaults aren't great Jul 28 07:09:18 ah Jul 28 07:09:23 have you changed your swap behavior? Jul 28 07:09:33 Me? yea Jul 28 07:09:39 ah Jul 28 07:09:40 mayb I should Jul 28 07:09:45 you've a thread on it? Jul 28 07:09:54 Me? No. Others have Jul 28 07:10:05 I've changed swap behavior, transitions, OC'd it... Jul 28 07:10:24 it runs *really* nicely now. Only feels sluggish when loading a full topic on Slashdot Jul 28 07:10:34 how big of usability difference? Jul 28 07:10:45 Enough, I'd say Jul 28 07:11:01 worried about OC damage? Jul 28 07:11:21 I know it will happen sooner or later. But in the short term? Not really Jul 28 07:11:37 I did a custom profile, undervolted everything just above crashing Jul 28 07:11:55 and it's 100% stable for now. If it stops being stable, I'll know something's failing... Jul 28 07:12:21 and well... it'll be time to change stuff and or get a new device. But having done it for ~9 months so far... Jul 28 07:13:02 Also, remember, it's not like a desktop: it's only OC'd for literally seconds or minutes each day. So while it might only last a month of constant on time at those speeds... that could easily be 5 years Jul 28 07:20:34 unless you watch a movie each day Jul 28 07:20:49 or do lots of phonecalls Jul 28 07:21:00 then what happens? Jul 28 07:21:08 why can't it be made to not overclock during over those things? Jul 28 07:21:14 or something Jul 28 07:21:14 then CPU clocks at full speed Jul 28 07:21:46 Yes Jul 28 07:22:28 browser showing a page with JS in background is also a good way to keep the CPU busy Jul 28 07:22:32 isnt the CPU OC'ed by default, IIRC the stock speed is 500 Jul 28 07:23:03 You can do whatever you want, honestly. Some people set up desktop-widgets with profiles Jul 28 07:23:18 hiemanshu: yes Jul 28 07:23:41 anyway Jul 28 07:23:45 ~omap-oc Jul 28 07:23:46 rumour has it, omap-oc is http://mg.pov.lt/maemo-irclog/%23maemo.2010-08-01.log.html#t2010-08-01T22:16:05 read that!, or and this http://mg.pov.lt/maemo-irclog/%23maemo.2011-03-11.log.html#t2011-03-11T03:04:11 Jul 28 07:25:06 Well, remember that the CPU will switch between min and max freqs on the fly Jul 28 07:25:21 min being 0; it can actually sleep for small fractions of a second Jul 28 07:26:17 not during calls, on video playback, or on browser with JS running mad - AFAIK Jul 28 07:27:01 yeah, but you dont need to OC unless you are doing something really really cpu intensive for a long while, ex. compiling, and compiling on the phone is a stupid idea anyways Jul 28 07:27:28 I, personally found that OCing and dropping the sample-rate to 150ms from 300ms really increased responsiveness Jul 28 07:27:30 and on stock kernel attaching USB nails the CPU to always running, tests about CPU clock in this situation, on PK, are still TBD Jul 28 07:28:33 Conky says that it drops to 500mhz min with usb enabled on PK Jul 28 07:30:37 so this still isn't fixed yet? Jul 28 07:30:56 err, fixed for charger Jul 28 07:31:00 ? Jul 28 07:33:05 Evidently not Jul 28 07:33:13 on charger? Jul 28 07:33:23 Yea Jul 28 07:33:28 cssu? Jul 28 07:33:36 Haven't installed. Jul 28 07:33:40 aah Jul 28 07:33:55 otoh cssu doesn't do anything to kernel afaik Jul 28 07:34:09 MohammadAG: ping? Jul 28 07:35:08 * DocScrutinizer loughs while noticing it was him who found that 500MHz CPU on charger idiocy Jul 28 07:35:51 isnt that just some stupid polling which keeps the cpu busy? Jul 28 07:35:57 and that's been like one or two weeks after I switched to PR1.3 ? Jul 28 07:36:18 cehteh: kernel patch Jul 28 07:36:38 which we all thought should get reverted, iirc Jul 28 07:38:04 yes Jul 28 07:38:17 cehteh: also see robbiethe1st 's comment, I guess >> it drops to 500mhz min<< means the min CPU clk freq is set to 500 Jul 28 07:38:28 well, I dunno about that Jul 28 07:38:33 it just never drops below it Jul 28 07:38:50 not on usb and charger .. Jul 28 07:39:06 well i dont care that much when it is connected Jul 28 07:39:17 there are more annoying things Jul 28 07:39:26 sure Jul 28 07:39:44 someome should get compcache running and packaged Jul 28 07:40:21 i wonder if ramz can be backported to the maemo5 kernel Jul 28 07:40:27 just to give fauxfone an idea about OC and what it means when I say ""you have to damn well know what you're doing and why you do OC"" Jul 28 07:40:41 yes Jul 28 07:41:25 well i have 250 500 and 750 as cpufreq, but anyone must know by himself what he does :P Jul 28 07:41:41 That's confusing because it sounds like there is some mysterious component... I've OC'd many computers in my day, just never a cell phone Jul 28 07:41:48 doing OC for the ImSoGeek and 25% more speed *in some random situations* usually isn't worth it Jul 28 07:42:03 ^agreed Jul 28 07:42:07 having 250 500 600 without OC and leaving 550 out is prolly a good idea for anyone who dont want to OC Jul 28 07:42:11 Which is why I asked about the usability experience Jul 28 07:43:01 Personally, I find that it doesn't get too hot with 950mhz under max load, so I've got that as my standard max freq; I can go up to 1.15ghz though if I need it Jul 28 07:43:14 the 550 state gives only latency when scaling back and is so close to the other frequencies, it makes no sense Jul 28 07:43:23 usability of OC is rather good, modulo the random crashes of gfx core. But damage done is out to anybody's guess Jul 28 07:43:45 robbiethe1st: there are a lot people who fried their devices in some or another way Jul 28 07:43:49 I meant added usability of the phone due to OC Jul 28 07:44:11 that however is marginal Jul 28 07:44:31 cehteh: Theoretically. I've seen at best a couple that are definitely CPU related; a number of potential corruption glitches, but nothing that a reflash wouldn't solve Jul 28 07:44:32 things like swappiness etc have WAY more impact Jul 28 07:44:36 actually my dsp got hung at 250mhz with too aggressive undervoltage Jul 28 07:45:01 robbiethe1st: there are really permantent damaged/destroyed phones .. so beware Jul 28 07:45:23 it will happen its just a matter of time and its very specific to any device Jul 28 07:45:30 robbiethe1st: nope, there are several permanent burnouts reported meanwhile Jul 28 07:45:38 for someone it may happen in a few hours, others may survive a few years Jul 28 07:46:09 Still, I'm not giving it up. Mainly because it works so well. If things start getting flakey, well, I'll know that there's a problem. it may be too late by then, but... Jul 28 07:46:10 600Mhz is actually OC .. Jul 28 07:46:26 in a sense it is Jul 28 07:46:30 see Jul 28 07:46:34 ~omap-oc Jul 28 07:46:34 [omap-oc] http://mg.pov.lt/maemo-irclog/%23maemo.2010-08-01.log.html#t2010-08-01T22:16:05 read that!, or and this http://mg.pov.lt/maemo-irclog/%23maemo.2011-03-11.log.html#t2011-03-11T03:04:11 Jul 28 07:46:37 2nd link Jul 28 07:47:40 power-on-hours allowed at 600 are 23k, while for all other speeds below the chip is specified for 100k h lifetime Jul 28 07:47:53 Yes. we all know the theoretical amount Jul 28 07:48:12 I'm assuming it'd last about a month if running 100% of the time Jul 28 07:48:12 so you got a knee from 500->600 = ~1/4 Jul 28 07:48:12 well nokia applies also very conservative voltage settings .. i believe that my OC to 750Mhz and much reduced voltage doesnt to more harm than 600Mhz with standard voltage .. but i have absolutely no proof of that Jul 28 07:48:33 600->800 possibly = 1/10..1/100 Jul 28 07:48:37 But, running it for seconds or minutes every day... and I'm undervolting it... Jul 28 07:48:38 of lifetime Jul 28 07:48:49 of lifetime @ 600 Jul 28 07:48:49 I know it's lasted for 9 months so far, so... Jul 28 07:49:23 DocScrutinizer: is there a formula taking voltage and frequency into account for expected lifetime? Jul 28 07:49:23 Look; we've argued this before. I'm not giving it up. If it fails, it'll be my fault. Jul 28 07:49:39 yes Jul 28 07:49:45 see electromigration in wikipedia Jul 28 07:50:22 it smewhat suggests the factors and their multipliers Jul 28 07:50:38 And well, if -- when -- mine fails, I'll go find one with a broken screen, use that, and tell everyone what I got from mine. Jul 28 07:51:00 it's not voltage but current. Note please that current can't get reduced by undervolting Jul 28 07:51:35 Um, why not? lower voltage = lower current at the same resistance Jul 28 07:51:40 robbiethe1st: fair enough Jul 28 07:51:57 robbiethe1st: nope, there's SmartReflex Jul 28 07:52:13 Which will limit the maximum voltage, anyway Jul 28 07:52:36 SmartReflex basically implements a miniature LDO for each gate on chip, speaking simplified Jul 28 07:53:22 smart reflex isnt enabled normally Jul 28 07:53:26 so reducing core voltage by profiles just reduces the power and also the regulation headroom for those micro-LDOs Jul 28 07:54:04 well temperature comes also into account its not only current or? Jul 28 07:54:10 cehteh: the core voltage ramping part isn't enabled normally for some good reason. You can't switch off the micro-LDO property of smartreflex Jul 28 07:54:19 cehteh: yes Jul 28 07:54:30 see second link of ~omap-oc Jul 28 07:55:04 Though, um, wait... if that's true, how come overvolting allows for higher frequencies? if the internal LDO is regulating it, you shouldn't see much if any gain at all Jul 28 07:55:16 anyways with my current settings the device is very rarely and only for short times on 750 (gave some bias for switching) Jul 28 07:55:23 you don't Jul 28 07:55:43 undervolting definitely doesn't help for better OC Jul 28 07:55:43 and without the stinky 550 mhz level it falls directly back to 500Mhz Jul 28 07:55:54 it *only* reduces power consumption Jul 28 07:55:55 No, overvolting Jul 28 07:56:17 nope but less power means less temperature Jul 28 07:56:27 less voltage Jul 28 07:56:51 after all i am aware that this is highly nontrival and complex Jul 28 07:57:03 but well, so far it works for me :P Jul 28 07:57:18 overvolting gives micro LDOs more headroom so they can regulate to a higher voltage on output. Those are dynamically controlled to deliver optimum voltage to the gate they service, depending an a number of sensor inputs Jul 28 07:57:20 battery lasts longer, device is responsive and gets less warm Jul 28 07:58:20 cehteh: decent tests or it never happened Jul 28 07:58:33 blah :P Jul 28 07:59:08 send me a device which i can use to test Jul 28 09:49:15 hello / bonjour ! Jul 28 09:53:54 bonjour Fukakai Jul 28 09:55:07 I want to create a web server with my nokia n900, to access with my windows by 80 port ... do you think is it possible ? have you tutorials ? think you Jul 28 09:55:49 yes, it's possible. Jul 28 09:56:26 i have ever install lighthttpd and it work's in local, but I can't access with the phone IP ... Jul 28 09:56:41 Ah - that's a routing problem then Jul 28 09:56:49 what do you mean by 'by the phone ip' Jul 28 09:56:57 Over the phone net? Jul 28 09:57:36 i have a widget where i can see IP gprs0 10.62.27... Jul 28 09:57:52 NAT :) same boat as i am Jul 28 09:57:59 Ah Jul 28 09:58:05 Fukakai: that's not a real IP Jul 28 09:58:20 99.999% of phone providers put you behind a NAT. Jul 28 09:58:29 You can never make inbound connections to the phoe. Jul 28 09:58:32 phone Jul 28 09:58:51 I've try with this IP, and with the IP I have find to my www.mon-ip.com/ Jul 28 09:58:55 This is - generally - a good thing - as 'background radiation' from probes ad stuff would eat your battery life. Jul 28 09:59:12 But is annoying if you want to. Jul 28 09:59:26 openvpn ftw Jul 28 10:00:23 i can be reached from the www cehteh by using openvpn? Jul 28 10:00:37 Sicelo: Sort-of Jul 28 10:00:45 Sicelo: you setup a VPN to a server. Jul 28 10:00:57 you need some endpoint reachable from www Jul 28 10:00:59 Then you make that server route inbound connections down the VPN Jul 28 10:01:17 thanks. :) /me will look into it Jul 28 10:01:23 It might be worthwhile at this point to ask what you're trying to do - there may be other ways that are simpler. Jul 28 10:02:08 XD sorry i don't understand all, it's new for me ^^ Jul 28 10:02:19 well, i always use Reverse SSH for remote access.. but would like to show off an http server ;) Jul 28 10:02:29 Ah. Jul 28 10:02:42 You can do webserver over that pipe Jul 28 10:02:44 I think Jul 28 10:03:15 Can't you make a reverse ssh connection do SOCKS proxying somehow? Jul 28 10:03:38 Nevermind the obvious route of simply using a reverse ssh on port 80 Jul 28 10:04:51 ok, i see Jul 28 10:05:26 It would be nice if there were some way to 'sign' your accesses. Jul 28 10:05:27 Fukakai, what about doing it on wireless at home , wireless will give you an ip for sure Jul 28 10:05:33 problem in that case, at least aiui, reverse ssh suffers from fact that the person accessing my webserver will have to have login rights to the server first. :/ Jul 28 10:05:43 So that you could have connections only connecting if the connector had your key Jul 28 10:05:44 and configuring the router should be easy Jul 28 10:08:46 pavi, i have a router and an ip yes, but i want the server can stay up all the time XD in the street ... Jul 28 10:10:21 On the go mobile server , thats ambitious :) Jul 28 10:10:41 yes XD Jul 28 10:11:14 i'm going to try the reverse ssh solution, thinks for help :D Jul 28 10:24:10 you will find it on the wiki Jul 28 10:25:57 Sicelo: Err - no Jul 28 10:26:14 Sicelo: You setup ssh to port forward any connection on port 80 to the phoes port 80 Jul 28 10:26:25 No auth required Jul 28 10:29:51 ... but i don't understand, i need to ping my phone to configure ssh, isn't it ? Jul 28 10:30:41 You need to configure your phone so it has a ssh tunnel always logged in to the server Jul 28 10:31:21 oh wow! the power of ignorance. thanks SpeedEvil Jul 28 10:35:20 sorry, i understand, i'm going to do that :) Jul 28 10:47:20 o_O Jul 28 11:05:34 hmm other than the FM chip in the N900 anyone know of any USB FM tuners that work (well?) on linux? Jul 28 11:07:06 No clue; I haven't needed any... Jul 28 11:07:15 hell, I don't use the one on my N900, let alone anything else Jul 28 11:07:30 Why not use your N900 as your FM tuner; route audio over usb network? Jul 28 11:07:33 wl1271? Jul 28 11:07:54 why would anybod listen to FM radio Jul 28 11:08:05 you can get same stuff over interweb Jul 28 11:08:28 jacekowski, connection drops, media player doesn't retry but goes to new item in the playlist etc. Jul 28 11:08:37 i.e. new tech not as reliable as old tech Jul 28 11:09:00 mgedmin: you can change that Jul 28 11:09:32 i possibly see way how that could sell Jul 28 11:09:49 i'll have to check costs of making something like that and maybe make profit on it Jul 28 11:12:08 jacekowski, lag Jul 28 11:12:12 is the main reason Jul 28 11:12:18 it has a slight lag compared to radio Jul 28 11:12:39 and since the distance of where I would like to run the fm tuner is like 5m away from the other radio this would be highly annoying Jul 28 11:13:00 and some of us don't have access to data that easily :( Jul 28 11:18:56 the one device I found is sadly only a newer revision which is not supported while the older one is Jul 28 12:41:02 jacekowski: maybe because FMRX works like 40h while internet radio over GPRS works maybe 3h ? Jul 28 12:47:07 pavi: is bq27200 module still auto-loaded on PK? Jul 28 12:47:32 DocScrutinizer, wrong person or typo ? Jul 28 12:47:58 err, aren't you maintaining powerkernel? Jul 28 12:48:05 I have no idea what PK is or bq* is Jul 28 12:48:07 oooh pali ? Jul 28 12:48:10 DocScrutinizer, no way Jul 28 12:48:18 sorry Jul 28 12:48:20 I am a dumb newbie Jul 28 12:48:26 mental typo Jul 28 12:49:01 lol Jul 28 12:56:59 DocScrutinizer, Pali's on Jul 28 13:18:45 andre__: the server cert of b.m.o is popping up as outdated... in my browser... Jul 28 13:19:32 Pali: 14:47 < DocScrutinizer> pavi: is bq27200 module still auto-loaded on PK? Jul 28 13:22:21 I think yes. I was not removed board-rx51 patch Jul 28 13:23:54 chem|st, okay? file a bug or ping X-Fade or so? Jul 28 13:23:59 nothing I can change. Jul 28 13:24:46 X-Fade: ping, read above Jul 28 13:24:54 andre__: thanks Jul 28 13:28:21 DocScrutinizer: and in KP is this patch (so maybe bme can live together with bq) https://garage.maemo.org/plugins/ggit/browse.php/?p=kernel-power;a=blob;f=kernel-power-2.6.28/debian/patches/i2c-battery.diff;h=5e59c8920591879683deae6a9ce3e4ce9ae86c3d;hb=HEAD Jul 28 13:34:59 Pali: this is an extremely critical patch, that has potential to break a _lot_ of things, if there are any races or reentrance issues or other unforeseen side effects. I'd not dare to use it without review by at least 10 first grade kernel hacker cracks Jul 28 13:39:06 Pali: the worst thing about it is you won't know for which symptoms to look, it could be everything from kernel oops to endlessly looping code, to collisions on I2C bus that cause chips to misbehave or even break Jul 28 13:42:42 this patch was in KP since v45 Jul 28 13:44:33 I and other people dit not report any problems Jul 28 13:45:22 ooh, I know of a lot of problem reports with unclear symptoms and unknown reasons Jul 28 13:47:14 and for this class of patch not even a problem free use on a million device-hours means anything. Such patch needs source level evaluation and validation, to make sure there are no side effects and possible race conditions Jul 28 13:47:31 so I just noticed that my ipv6 isnt working anymore Jul 28 13:47:47 gprs1 isnt coming up Jul 28 13:48:16 you need to check and proove there can't happen collisions on concurrent access to I2C bus Jul 28 13:48:55 you need to check all systems that access I2C driver that they all don't rely on the -EBUSY behaviour Jul 28 13:49:28 uh Jul 28 13:49:30 what Jul 28 13:49:43 plus you need to have a concept to make sure nobody in future will run into problems when relying on -EBUSY here Jul 28 13:50:53 oh you must not be talking to me Jul 28 13:51:08 Sazpaimon_: indeed, I was answering to Pali Jul 28 13:51:15 ok Jul 28 13:51:38 sorry for being a bit slow Jul 28 13:52:48 i wonder if tmobile disabled my ipv6 address Jul 28 13:53:19 not sure why that would happen though Jul 28 13:53:22 this kernel patch was added by Matan in v45 and I did not removed any added patch in >=v47 Jul 28 13:54:10 Pali: you're changing the API semantics, and you're gambling with hardware management which needs evaluation on an abstract level for correctness, see problem of proof of race condition free design that can't be done by tests Jul 28 13:55:01 Pali: I'm not blaming anybody. I'm just urgently suggesting not to keep this patch, it's pretty dangerous Jul 28 13:56:23 Ok, I ask Matan about his patch Jul 28 14:17:24 basically this patch seems to allow two processes access the same register in same chip on same I2C bus concurrently, which can never work but is happening rarely but definitely *will* happen eventually. This can cause any arbitrary problem, from garbled transmission on I2C bus causing chip malfunctions and breakage, to kernel panics or whatever Jul 28 14:20:08 any patch implementing this at very least needs to come with an abstract evaluation explaining all the issues that have been considered, what's the way they would happen, and why the author thinks they can not happen. Jul 28 14:24:19 bare minimum is a bold warning "not been evaluated and confirmed for any side effects - expect complete system breakdown when using this patch" Jul 28 14:33:22 oooh Matan Jul 28 14:33:47 last I heard was Matan gone trolling Jul 28 14:36:59 anyway it's duty of package maintainer to evaluate a patch and decide whether or not to include it to the package (here: to power-kernel). Matan can explain his patch, but it's finally up to the maintainer (which I gather was titan for this patch) to include or reject the patch, ideally after discussing it with some other guys involved Jul 28 14:38:42 and yes, I actually heard reports about weird behaviour of (iirc) bme not working correctly on PK even after rmmod of bq27200 module. Now I see a possible reason Jul 28 14:39:54 anyway note that this patch has impact on *all* I2C subsystems Jul 28 14:40:34 which is enough on its own to reject it Jul 28 14:41:55 Does someone know of the top of the head if there's a cli tool packaged for maemo for reading exif data from images? Jul 28 14:42:21 also, does someone know which package contains 'file'? Jul 28 14:42:31 usually file Jul 28 14:42:48 oh, indeed. Jul 28 14:43:32 ah nevermind Jul 28 14:43:39 I could just use apt-cache right away next time -.- Jul 28 14:44:40 lol Jul 28 14:54:12 Also, individual apps can, if they wish, request either exclusive or non-exclusive access, and force override of exclusive access by others... So it's not like there aren't already enough options available to userspace, you don't need more kludges in kernel... Jul 28 15:00:12 ShadowJK: ack. The bq27200 module is borked, and the patch to i2c-dev tries to fix that (well to bq27200.ko's excuse I have to mention it's basically bme that is borked :-D ) Jul 28 15:00:27 Hi! Anybody repo admins here? A wrong package version went into stable. Jul 28 15:01:02 ShadowJK: anyway patching i2c-dev - WTF :-O Jul 28 15:01:08 Well, in MeeGo bme actually uses the kernel module.. But the kernel module doesn't export everything, so I like to unload bme and kernel module anyway Jul 28 15:01:16 yes, wtf :) Jul 28 15:01:43 The headless-chicken tendencies like these are why I've never dared touch powerkernel Jul 28 15:01:46 or cssu ;) Jul 28 15:02:06 I can well feel with you Jul 28 15:03:09 while we are at it, let's patch a special priority for pulseaudio into scheduler ;-P Jul 28 15:03:21 * ShadowJK is reminded of the people complaining their filesystems corrupt in 3 days Jul 28 15:03:47 if pulseaudio runs at RT, it's essentially a special priority anyway Jul 28 15:04:04 "I added 87 patches to the fs driver, and still it's not working" ? ;-) Jul 28 15:04:15 ah it's not RT Jul 28 15:04:25 The problem is, that a wrong version of ghostscript went into stable. So cups (also in stable) is not installable. Jul 28 15:04:49 DocScrutinizer, more like "stupid unstable shit, I overclocked to 1500 and it STILL crashes" ;p Jul 28 15:05:15 mase_76: HARD to fix Jul 28 15:05:51 But not impossible I think. Jul 28 15:05:53 mase_76: btw what is "stable"? Jul 28 15:06:07 extras Jul 28 15:06:17 ok Jul 28 15:06:30 file a ticket, ping x-fade Jul 28 15:07:06 ping the maintainer of the buggy package (ghostscript?) Jul 28 15:07:30 I'm from Debian, so I call it stable, sorry! I am the maintainer. Jul 28 15:07:44 aah even better Jul 28 15:07:48 ping x-fade Jul 28 15:07:52 he can help Jul 28 15:08:37 no need to file a ticket ;-) Jul 28 15:09:33 re bme coexistence with other things, if you really want to do that, you probably want to create a LD_PRELOAD library to intercept the i2c requests bme sends, take the data from the kernel module in /sys/ and feed it back to bme. That would be the safe option Jul 28 15:10:25 YEAH! Jul 28 15:10:55 IF anybody would care to untergo this useless exercise ;-D Jul 28 15:11:31 under* Jul 28 15:14:08 ok, so *hotfix* - what to do? Copy the old aka stock i2c-dev.ko to powerkernel? Jul 28 15:14:21 prbably won't fly due to vermagic Jul 28 15:14:35 unless you force it Jul 28 15:15:08 otoh I'd be surprised to find i2c-dev being a module at all :-S Jul 28 15:15:57 as the CPU controls SmartReflex via I2C, so it will be needed pretty early in boot Jul 28 15:16:25 needs a fixed PK version then Jul 28 15:19:42 yoh, no *i2c*.ko Jul 28 15:19:49 :-/ Jul 28 15:20:34 Pali: 2 devels with an idea of kernel suggest this patch MUST go, urgently Jul 28 15:21:50 it's dangerous to the point of potentially damaging hardware Jul 28 15:24:33 ok, I drop this patch Jul 28 15:24:42 Pali: you are PK maintainer, so it's up to you to decide. You don't need Matan's or even titan's OK Jul 28 15:25:49 solution with LD_PRELOAD for bme sounds good, but who know what all is needed in new library? Jul 28 15:26:53 Pali: first question is: why do we need it at all? Jul 28 15:27:21 nobody is depending on bq27200 module Jul 28 15:27:49 (btw you obviously need to drop this module as well, for autoload on boot) Jul 28 15:28:15 the module can exist, just can't load it on boot Jul 28 15:28:54 for LD_PRELOAD we need to override the calls that open and read/write to I2C. I gather that'S normal fopen and friends Jul 28 15:29:11 ShadowJK: that's what I meant to say Jul 28 15:29:18 iirc something tries to make use out of the module, I've seen people complain "lshal" has multiple battery voltage measurements Jul 28 15:29:28 it's all ioctl actually :/ Jul 28 15:29:35 yes, i know. but I think that bq is usefull hor host mode (when bme is stopped) Jul 28 15:29:37 ooh Jul 28 15:29:49 ok, so s/fopen/ioctl/ Jul 28 15:30:04 Pali: not really Jul 28 15:30:54 at least h-e-n itself relies on i2ctools, I.E. it probably will fail when this module is loaded Jul 28 15:31:08 I wonder how that data was taken from /sys and stuffed into hal Jul 28 15:31:41 ShadowJK: MohammadAG 's battery-applet rewrite? Jul 28 15:31:49 not sure Jul 28 15:32:41 ah well, as I don't use cssu or pk I'm not affected by all the dangerous stuff anyway :P Jul 28 15:33:20 Pali: please include a reference to my rant above to the commit dropping this patch, otherwise in 2 weeks somebody else checks it in again Jul 28 15:34:42 if this stuff breaks, and two things are bitbanging addresses onto the i2c bus at the same time, and stuff explodes/crashes/dies, they'll have no clue what just happened Jul 28 15:35:00 indeed - LOL Jul 28 15:35:38 is this a problem of cssu too? or just KP? Jul 28 15:35:46 just PK Jul 28 15:35:56 KP whatever Jul 28 15:36:35 * DocScrutinizer hates Inverse Polish Notation in natural speech Jul 28 15:38:45 the car police fast down drives street empty the Jul 28 15:39:12 did you learn forth or what? Jul 28 15:39:16 or did you meet yoda? Jul 28 15:39:21 kernel power Jul 28 15:40:00 object:kernel, property:power Jul 28 15:40:47 is just powerkernel for me each day, short PK Jul 28 15:42:39 as kernel-hm is hostmode kernel, and kernel-fremantel-1.3-foo-bla is stock kernel Jul 28 15:42:55 Ok, I disabled i2c-battery.diff in my git repo and recompiling kernel deb packages Jul 28 15:43:11 also disabled autoloading bq Jul 28 15:43:28 check for etc/modules (?) to disable auto-modprobing of bq27200 on boot Jul 28 15:43:34 aah, k :-D Jul 28 15:43:47 Pali: many thanks for taking care Jul 28 15:44:09 bq autoloading was done in board-rx51*.c Jul 28 15:44:19 ok, nfi here Jul 28 15:45:10 kernel-power (never?) did not touched /etc/modules so users can add/remove lines Jul 28 15:45:37 just stamp the patch "DEPRECATED! breaks API, may kill your hardware" Jul 28 15:46:14 that also explains a bit Jul 28 15:46:16 :-D Jul 28 15:46:30 ( board-rx51*.c ) Jul 28 15:48:29 several weeks ago there were 2 users same day in this channel that has issues with PK, regarding some odd behaviour of bme IIRC. I wasn't able to talk them thru disabling bq27200 :-P Jul 28 15:48:59 s/that has/ who had/. Jul 28 15:56:16 I'm still not maintainer of kernel-power-modules, kernel-power-flasher, ... packages Jul 28 15:56:32 I'm still on pending list on maemo.org package interface Jul 28 15:56:42 Pali: this has to change, definitely Jul 28 15:56:50 who can give me permitions? Jul 28 15:56:57 maintainer of kernel should be same as for modules etc Jul 28 15:57:20 how did you receive mainainership for PK? Jul 28 15:57:24 I got permitions only for package kernel-power and kernel-power-source Jul 28 15:57:43 I do not know. One day I found my name on the list Jul 28 15:57:50 I'd guess the process would be the same as for PK Jul 28 15:57:55 LOL Jul 28 15:58:21 see: http://maemo.org/packages/view/kernel-power and http://maemo.org/packages/view/kernel-power-modules Jul 28 15:58:33 and also other kernel-power-* packages Jul 28 15:58:35 after 2000 years, another virgin birth :-D Jul 28 15:59:27 Pali: ask MohammadAG - I think he knows a friggin lot about it Jul 28 15:59:52 MohammadAG: ping ^^^ Jul 28 16:00:10 MohammadAG: (10 lines backscoll) Jul 28 16:00:54 o/ bbl Jul 28 16:01:12 and maemo.org package interface has still bug - it does not see new version v47 of kernel-power Jul 28 16:01:27 x-fade Jul 28 16:02:20 I tried contact him, but without responce (IRC and mail) Jul 28 16:03:31 I also reported bug: https://bugs.maemo.org/show_bug.cgi?id=12284 Jul 28 16:03:32 04Bug 12284: kernel-power in maemo.org package interface Jul 28 16:05:02 the SSL cert expired. Jul 28 16:11:36 ~seen X-Fade Jul 28 16:11:40 x-fade is currently on #maemo (9d 16h 28m 30s) #meego (9d 16h 28m 30s). Has said a total of 57 messages. Is idling for 1d 7h 36m 16s, last said: 'arfoll: Hehe, cool stuff.'. Jul 28 16:12:34 Pali: recently you need luck to get a hold of X-Fade on IRC, but occasionally it may happen Jul 28 16:14:25 unless you also got on his ignore list ;-D Jul 28 16:21:17 what things are included in the n900 package? i've bought one, but there is no disc in its package Jul 28 16:29:01 what things are included in the n900 package? i've bought one, but there is no disc in its package Jul 28 16:34:17 TimmyT1: from memory: n900, battery, usb charger, handsfree, user guide Jul 28 16:35:48 oh yeah, and tv-out cable Jul 28 16:37:51 TimmyT1: so no disc... any electronic user guide is preinstalled on the n900 itself Jul 28 16:38:13 villager: Gh0sty: anything else? this pic shows that it has another thing : http://wapreview.com/images/N900Accessories.jpg Jul 28 16:39:21 under the phone Jul 28 16:40:21 TimmyT1: let's see... wall charger, usb charger, extra earbuds for handsfree, n900, cleaning cloth, handsfree, manual, adapter for old-style nokia charger to new micro-usb, user guide, tv-out cable Jul 28 16:42:21 yup I think that's all on that picture Jul 28 16:42:23 heh lets move here :p Jul 28 16:42:34