IRC log of #maemo-ssu for Friday, 2013-09-20

*** nox- has joined #maemo-ssu00:00
freemangordonsailus: no idea, I guess from meego00:07
*** NIN101 has quit IRC00:09
freemangordonoh, wait, vaux2 is for omap3isp00:09
freemangordoncameras are on VANA00:10
*** jon_y has quit IRC00:30
*** jon_y has joined #maemo-ssu00:30
sailusSame regulator.00:54
*** povbot has joined #maemo-ssu01:02
*** ChanServ sets mode: +v povbot01:02
*** DrCode has quit IRC01:10
*** amizraa has quit IRC01:10
Palifreemangordon: in 3.10-n900 repo I enabled CONFIG_OMAP3_L2_AUX_SECURE_SAVE_RESTORE and updated omap3-rom-rng (random number generator)01:12
*** freemangordon has quit IRC01:13
*** freemangordon has joined #maemo-ssu01:14
*** DrCode has joined #maemo-ssu01:16
DocScrutinizer05*sigh* freemangordon: not logged in to nickserv01:20
*** M4rtinK has quit IRC01:52
*** _nicolai_ has quit IRC01:57
*** xes has quit IRC02:04
*** MohammadAG has quit IRC02:05
*** MohammadAG has joined #maemo-ssu02:06
*** sunny_s has quit IRC02:17
*** int_ua has joined #maemo-ssu02:29
*** int_ua has quit IRC02:45
*** BCMM has quit IRC02:55
*** int_ua has joined #maemo-ssu02:58
*** Pali has quit IRC02:59
*** int_ua has quit IRC03:04
*** int_ua has joined #maemo-ssu03:06
*** int_ua has quit IRC03:27
*** dos1 has quit IRC04:11
*** Martix_ has joined #maemo-ssu04:51
*** jonwil has joined #maemo-ssu05:12
jonwilhi05:12
jonwil~seen nicolai05:14
infobotjonwil: i haven't seen 'nicolai'05:14
*** Martix_ has quit IRC05:32
*** amiconn has quit IRC05:55
*** amiconn_ has joined #maemo-ssu05:55
*** amiconn_ is now known as amiconn05:55
*** unclouded has quit IRC06:08
*** freemangordon has quit IRC06:15
*** freemangordon has joined #maemo-ssu06:15
*** unclouded has joined #maemo-ssu06:20
DocScrutinizer05moin jonwil06:23
jonwilhi06:23
DocScrutinizer05jonwil: you need to log in to nickserv, otherwise some IRC functions will not work for you06:25
DocScrutinizer05/ns info jonwil06:26
DocScrutinizer05:-o06:26
DocScrutinizer05[2013-09-20 05:26:02] [Notice] -NickServ- Last seen  : Oct 03 15:38:15 2012 (50 weeks, 1 day, 11:47:47 ago)06:26
DocScrutinizer05/ns help identify06:27
DocScrutinizer05/msg nickserv help identify06:28
DocScrutinizer05if your client doesn't support the /ns shortcut06:28
DocScrutinizer05/msg nickserv help password06:28
DocScrutinizer05/msg nickserv help setpass06:29
DocScrutinizer05scrap the line befor that06:30
jonwilI just ifentified06:30
jonwilidentified06:30
DocScrutinizer05good06:30
DocScrutinizer05then e.g. invite-exempts will apply06:31
DocScrutinizer05;-)06:31
*** nox- has quit IRC06:31
jonwilstill waiting for my Gentoo box to finish compiling :)06:31
jonwilfinish updating :)06:32
DocScrutinizer05hah06:34
DocScrutinizer05~listkeys cssu06:35
infobotFactoid search of 'cssu' by key (6): cssu-thumb ;; fapman+cssu ;; #maemo cssu-state ;; cssu ;; cssu-optional ;; cssu-devel.06:35
DocScrutinizer05~cssu-optional06:35
infobotextra, extra, read all about it, cssu-optional is http://maemo.org/community/maemo-developers/how_to_include_rewrites_of_closed_blobs/06:35
DocScrutinizer05~cssu-optional is also http://maemo.merlin1991.at/cssu/meetings/2012-05-14.txt06:39
infobotDocScrutinizer05: okay06:39
* DocScrutinizer05 ponders to set up an autoresponder for regex along the line "why .* not goes into CSSU"06:41
*** DrCode has quit IRC06:41
DocScrutinizer05especially for freemangordon and e*_06:41
DocScrutinizer05this constant bitching and spreading of FUD and made-up facts starts to seriously annoy me06:42
DocScrutinizer05http://talk.maemo.org/showpost.php?p=1375494&postcount=606:43
DocScrutinizer05extremely annoying example, filled with ~85% made-up and erratic assumptions: ivgalvez' rant in http://talk.maemo.org/showthread.php?t=91308 - now moved to another thread I can't locate06:46
*** DrCode has joined #maemo-ssu06:46
*** amiconn has quit IRC06:46
*** amiconn_ has joined #maemo-ssu06:46
*** amiconn_ is now known as amiconn06:46
DocScrutinizer05I mean, we have public meetings and policies decided upon during such meetings for some reason, eh?06:54
DocScrutinizer05not for some guys to wait until those who proposed and drafted the policy are annoyed to a degree where they leave, by constant ignorance, and then those guys try to push their way06:55
DocScrutinizer05that's an extremely rude behaviour against all those who took the time to come up with a *solutions* for itches that those ignorant guys uttered the loudest06:56
DocScrutinizer05or - to put it plain and straight - we have a policy how to include arbitrary packages into CSSU. If that policy doesn't meet your ideal, then too bad. It been set up for the benefit of all of us, incl users. You can't ignore it just for your own convenience06:58
DocScrutinizer05and for sure it's silly to try and convince the "distro" maintainer to ignore that policy, when it's been him who introduced it06:59
DocScrutinizer05merlin1991: ^^^ right?07:00
jonwilis http://talk.maemo.org/showthread.php?t=91399&highlight=ivgalvez the thread you wanted (re the rant from ivgalvez)?07:03
DocScrutinizer05YES! :-)) many thanks07:03
DocScrutinizer05>>First of all is lack of manpower<< so what? Ivgalvez to do the scheduler and tell devels what to work on?07:05
DocScrutinizer05>>Currently, the only development on Maemo OS is being done inside CSSU with a very small team of developers<< yep, but unrelated to FPTF  >>and it's difficult to think that they will be able to maintain both CSSU for N900 and a new port for the Neo900 at the same time.<< who says they shall or will do? Nobody suggested that07:06
DocScrutinizer05[removing redundant bits] >>But also there will be a clash of interests regarding development for N900 and Neo900<< UHUH! >>For example, open source replacements for closed applications are not entering CSSU<< well, see *my* rant above   >>we see that improved/upstream versions of packages are not allowed to be part of it<< ditto07:08
DocScrutinizer05>>but a port to a different hardware will require to rewrite as many closed source components as possible.<< says who?07:09
DocScrutinizer05>>Because it's not just enough to run an unmodified Fremantle image on the Neo900 with a shiny new kernel.<< again, says who?07:09
DocScrutinizer05>>Using closed source applications from Nokia in a non Nokia device, would be illegal<< bullshit07:10
DocScrutinizer05>>and even if Nokia might not care about it now, it's a clear risk to the project as a C&D could be received at any moment<< and here it starts to get bizarre07:10
DocScrutinizer05I'll rather stop my review, or some chanop might kick me for extreme spamming and ranting07:11
jonwilI see no reason why replacement work cant be done in a way that benefits both CSSU and Neo90007:16
jonwilfor example any MCE replacement that may show up could go into CSSU AND be changed for any changes hardware on Neo90007:18
DocScrutinizer05jonwil: absolutely my (even initial) idea07:19
DocScrutinizer05but some guys take _everything_ they can use to pervert and invert it to the opposite and use it against something unrelated, just because it can get used this way, after transmutation07:20
jonwilI would even say that we could potentially make ofono or fsogsmd talk the right dbus interfaces for replacing the cellular services daemon (like we talked about before) but make it support both an Option backend (for the Neo900) and the existing ISI backend (for the N900) so it too could be used in CSSU07:20
DocScrutinizer05of course, I thought we already made clear we'll do exactly that ;-)07:21
jonwilThe hard part there is figuring out all the DBUS interfaces07:22
DocScrutinizer05but that doesn't matter to ivgalvez who is pissed about CSSU not forcefeeding new alternative distro to all users07:22
DocScrutinizer05when it was for ivgalvez, even the wallpaper was nailed down in MP07:23
DocScrutinizer05"hey, let's introduce our own Non-Nokia theme!"07:23
DocScrutinizer05"let's ship it via CSSU!"07:23
DocScrutinizer05instead of fixing the idiocy of MP, those guys plan to exploit the concept for their own purposes, doing exactly same like Nokia did, just in the opposite direction07:24
DocScrutinizer05"Nokia foced chess-game onto all users' devices - new CSSU will delete it from all isers' devices" SUUUURE!07:25
wmaroneboy, you jumped down his throat07:26
jonwilhey, I like the chess game (except for the fact that it shows just how much I SUCK at chess when even the easy setting is far too hard for me to beat :)07:26
DocScrutinizer05"Nokia forced camera-ui onto user, CSSU now forces camera-ui2 onto all users" SUUUUURE07:26
jonwilso yeah today I plan to work on playing with pulseaudio-module-nokia-voice and on MCE stuff (mce stuff will need to wait for my Gentoo box to finish updating :)07:27
DocScrutinizer05FIX DAT MP, FINALLY! nuff said07:27
DocScrutinizer05idiotic bitching about "what goes into CSSU MP, and what doesn't"07:28
jonwilI wonder why Nokia went with the mp-blah concept in the first place07:28
DocScrutinizer05easier for them07:28
DocScrutinizer05they sold a product07:28
DocScrutinizer05no idea about FOSS concepts and spirit07:28
jonwilyeah they wanted a fixed "firmware" that contain a fixed set of packages with updates just pulling in a new mp-blah package07:29
DocScrutinizer05basically Nokia exploited and abused linux as much as they legally could07:29
jonwilCompared to how it is on some phones running a Linux kernel (like my old Motorola Z6) we have it so good on the N900.07:33
jonwilThe Z6 only ran Motorola signed kernels07:33
DocScrutinizer05:nod: sure07:34
jonwilwell until someone found an exploit in the bootloader that is :)07:34
DocScrutinizer05but that doesn't make MP concept any better or any more worthy to be kept and possibly even augmented, instead of finally fixed07:35
jonwilnope07:35
jonwilwe should definatly fix it07:35
DocScrutinizer05yes07:35
jonwiland get rid of it for good07:35
DocScrutinizer05yes07:35
DocScrutinizer05paricularly since I'm fed up with taking the flames for allegedly "not letting in" alternative FOSS replacements "into CSSU", when in fact I'm only defending user from augmenting resp perpetuating MP concept07:37
jonwilall of the bits related to package management and installing are 100% FOSS so there is nothing standing in the way of replacing that crap07:38
DocScrutinizer05it fells like me standing at a illegal landfill and getting bashed by those who don't understand why they can't dump their trash there as well07:38
DocScrutinizer05feels* even07:39
freemangordonDocScrutinizer05: morning. do you have time, oscilloscope and a spare n900 you can disassemble? I am out of ideas what to do with SSI/ISP, I think both suffer from the same problem07:44
freemangordonIt is either lack of some power supply or some clock07:45
DocScrutinizer05jonwil: what do you think about radically removing all DEPENDS in MP, and instead using pre/post-install script of MP to run apt-get install commands?07:45
DocScrutinizer05freemangordon: sorry, no scope07:46
freemangordonbut I just can't figure out what exactly is missing :(07:46
freemangordondammit07:46
freemangordonI have a 10MHz 20 years old russian one :D07:46
DocScrutinizer05freemangordon: I wound up my lab some 6 years ago07:47
DocScrutinizer05technology moved too fast for my lab to keep pace with it, thus it became more and more obsolete07:47
freemangordonwe'll have to find someone who can trace what is going on in the real life (HW level that is)07:47
DocScrutinizer05and in the end I agreed with Werner Almesberger and Andy Green that human brain is the best lab tool of all07:48
freemangordonesp when it comes to battery usage optimizations07:48
DocScrutinizer05ETM07:48
freemangordon~etm07:49
DocScrutinizer05and patched kernel drivers, with lots of kprintf07:49
DocScrutinizer05embedded trace macrocell, see OMAP07:49
freemangordonDocScrutinizer05: hmm, no, I want to see what comes in and goes out of the peripherals, not OMAP07:50
DocScrutinizer05a nifty little ARM IP block07:50
freemangordonfor example - is camera mux really in high when it should be07:50
DocScrutinizer05err wut? isn't it like 2what goes out there comes in here" and vice versa?07:51
freemangordonis anything flowing through SSI bus and what is the frequency?07:51
freemangordonetc07:51
DocScrutinizer05well, I bet the frequency is what it got set to07:51
freemangordonOMAP can't be used to monitor RAPU<->GAZOO for example07:52
DocScrutinizer05and what's flowing thru it shows up in the buffer07:52
DocScrutinizer05err, actually the APE OMAP can't, that's true07:52
freemangordonDocScrutinizer05: The point is that IMO everything is set up correctly07:52
freemangordonbut the shit still don;t work07:53
DocScrutinizer05well, then what's missing is the right initialization sequence07:53
freemangordonwhich is done by the drivers, that work with older kernels07:53
DocScrutinizer05no, init isn't done by the kernel drivers for all I know07:53
freemangordonso it must be something in that new facny hwmod and/or clock framework07:53
freemangordonfancy even07:54
DocScrutinizer05it's a complex thing done in higher level daemons07:54
DocScrutinizer05as well you could assume DHCP is done by the NIC kernel driver07:54
freemangordonDocScrutinizer05: it looks like the whole power domains are shut down when they should not be07:54
freemangordonanyway, attachig a scope will be much easier/faster than that ETM IMO07:55
DocScrutinizer05ISI is a rather complex multichannel protocol afaik07:55
*** oldtopman has quit IRC07:55
freemangordonDocScrutinizer05: I am not getting interrupts from the SSI module, what has ISI to do with that07:56
freemangordonthe same for ISP07:56
freemangordonit is like saying that it is libdigicam to blame for no interrupts from ISP :P07:57
DocScrutinizer05well07:57
* freemangordon think who could have a scope and spare device to play with...07:57
freemangordonthinks*07:57
DocScrutinizer05patch the SSI driver in fremantle 2.6.28 and trace the first action it is doing07:58
freemangordonDocScrutinizer05: the driver that doesn't work in 3.10 works in 3.507:58
DocScrutinizer05(scope) have you even checked if there are any accessible contact points to attach a probe to?07:59
freemangordon(SSI that is)07:59
freemangordonsee the bottom of rx51 schematics07:59
jonwilTime to have some food then work on pulseaudio-module-nokia-voice stuff :)08:00
DocScrutinizer05I don't even know what you would want to verify or log, on SSI, with a scope08:00
freemangordonpower supplies and clocks08:01
freemangordonthe usual stuff08:01
freemangordonDocScrutinizer05: I feel stuck, can;t think of any other way to continue08:01
DocScrutinizer05ok, so let's assume I did, and i'm telling you "clock is there"08:01
freemangordonwhat frequency it is?08:02
DocScrutinizer05AHA08:02
DocScrutinizer05does that matter?08:02
freemangordonyes08:02
DocScrutinizer05why?08:02
freemangordonbecause (in case of SSI) we talk at 55Mbs for example08:02
freemangordonis if we have iface cclock at 110, it won;t work08:03
DocScrutinizer05as long as it's not way too high, it shouldn't make a lot of difference, except for max bandwidth08:03
freemangordonsame for the cameras, they expect 9.6 Mhz08:04
freemangordon(programmed to do so)08:04
DocScrutinizer05when you're afraid that clock is not N but 2*N, then set clock to N/16 for a start, and see if it works then08:04
freemangordonalready did :(08:04
freemangordonI am afraid that either plls are incorrectly programmed or powered down08:05
DocScrutinizer05that should be able to get checked08:05
freemangordonboth ssi and ISP use their own pll/divider noone else uses08:05
DocScrutinizer05PLLs have some registers where you can check if they are locked for example08:06
freemangordonor rather divider chain08:06
DocScrutinizer05seems you have a clue, a trace to follow, already08:06
freemangordonDocScrutinizer05: sure, but attaching a scope is faster way to check what is wring/missing08:06
DocScrutinizer05no08:06
DocScrutinizer05it only shows you THAT *sth* is missing (or not)08:07
DocScrutinizer05it doesn't reveal WHAT is missing08:07
DocScrutinizer05how about JTAG?08:07
freemangordonwell, if a power supply is missing, it will show which power supply is that08:07
freemangordonDocScrutinizer05: would'n jtag require a special SW?08:08
DocScrutinizer05err08:08
freemangordonI know what jtag is08:08
DocScrutinizer05on your JTAG probe, sure08:08
freemangordon:nod:08:08
freemangordonwhich I don;t have :)08:08
DocScrutinizer05just thinking aloud08:08
freemangordonah08:08
DocScrutinizer05also iirc you *always* can _read_ GPIO08:09
DocScrutinizer05even when not in GPIO mode08:09
freemangordonyes, but is that really propagated to the outside world? is output buffer enabled? etc...08:09
DocScrutinizer05yes08:10
freemangordonanyway, have to run08:10
freemangordonwill think about what to do08:10
DocScrutinizer05the GPIO input afaik is always connected to the pin, and reading the register always shows the actual state of the pin08:10
freemangordonhmm, iirc it is connected through a buffer08:11
DocScrutinizer05not input08:11
freemangordonwill check in trmo08:11
freemangordon*trm08:11
DocScrutinizer05:nod:08:11
freemangordonhowever, gpios won;t help for the power supplies08:11
DocScrutinizer05check in RL as well, with sth simple like e.g. some UART CTS or RTS or whatever08:11
freemangordonbtw I can always read /dev/mem for mux and gpios08:12
freemangordonno need to do it in the kernel08:12
DocScrutinizer05:nod:08:12
freemangordonhmm, maybe I should do it, while searching for a girl with a washing machine :)08:13
freemangordon(someone with a scope)08:13
freemangordonanyway, bye08:13
DocScrutinizer05why would you need a scope for power lines?08:13
DocScrutinizer05also, which power lines would that be?08:14
*** LauRoman has quit IRC08:15
DocScrutinizer05honestly, when some LDO isn't powered up I'd expect that to a) show immediately since every single one powers quite a number of components, and b) you should be able to easily tell the fact from reading the config of the LDO directly from LDO's config registers, and see if those values suggest it's powered or not08:18
DocScrutinizer05I'd rather trust in that than in any clumsy probing of a impossible to reach trace on the PCB with a scope08:19
*** luf has joined #maemo-ssu08:30
*** piscodig is now known as discopig09:01
*** LaoLang_cool has joined #maemo-ssu09:47
*** DrCode has quit IRC09:49
*** DrCode has joined #maemo-ssu09:51
*** Martix_ has joined #maemo-ssu10:42
*** amizraa has joined #maemo-ssu10:43
*** FlameReaper has joined #maemo-ssu10:58
*** dos1 has joined #maemo-ssu11:11
*** FlameReaper has quit IRC11:11
*** FlameReaper has joined #maemo-ssu11:12
*** FlameReaper has quit IRC11:13
*** FlameReaper has joined #maemo-ssu11:14
*** FReaper has joined #maemo-ssu11:17
*** FlameReaper has quit IRC11:18
*** dhbiker has quit IRC11:20
*** dhbiker has joined #maemo-ssu11:21
*** dos1 has quit IRC11:26
jonwilok, investigating module-nokia-voice didn't get me anywhere11:31
*** FlameReaper has joined #maemo-ssu11:33
*** FReaper has quit IRC11:35
*** FReaper has joined #maemo-ssu11:37
*** FlameReaper has quit IRC11:41
*** Pali has joined #maemo-ssu11:44
*** DrCode has quit IRC11:51
jonwilanyone here any good with debian packaging?11:54
*** BCMM has joined #maemo-ssu12:12
*** FlameReaper has joined #maemo-ssu12:13
*** FReaper has quit IRC12:16
*** Martix_ has quit IRC12:18
*** FlameReaper has quit IRC12:20
*** M4rtinK has joined #maemo-ssu12:24
*** BCMM has quit IRC13:00
Palijonwil: me?13:18
jonwil:P13:19
jonwilwrong channel :P13:19
*** discopig has quit IRC13:32
*** BCMM has joined #maemo-ssu13:36
*** discopig has joined #maemo-ssu13:43
*** discopig is now known as Guest2350413:43
*** freemangordon has quit IRC13:50
*** freemangordon has joined #maemo-ssu13:51
*** dos1 has joined #maemo-ssu14:21
*** arcean has joined #maemo-ssu14:29
*** dhbiker has quit IRC14:49
dos1<DocScrutinizer05> jonwil: what do you think about radically removing all DEPENDS in MP, and instead using pre/post-install script of MP to run apt-get install commands?14:50
dos1it's not as simple as that, you can't run another apt-get/dpkg when one instance is running14:50
dos1I think the correct way to go would be chaging dependencies to recommendations, or something like that14:52
dos1but it might be necessary to check and maybe fix apt configuration (dunno about HAM) and checking it's impact on other packages14:53
dos1s/it's/its/14:53
infobotdos1 meant: but it might be necessary to check and maybe fix apt configuration (dunno about HAM) and checking its impact on other packages14:53
Palinot possible14:53
Paliyou cannot run dpkg/apt from pre/post scripts14:53
dos1Pali: just said that :P14:54
Palithe only one way to deal with dependences is metapackage14:54
dos1yes, that's for sure14:55
dos1but some dependencies shouldn't be hard ones14:55
dos1I guess everything is now in DEPENDS?14:56
jonwilThe right way is to fix H-A-M and apt-get and dpkg and etc so that we can do things the way Debian proper does it14:57
jonwilthey dont have metapackages or other crap14:57
dos1in Debian, there's no "one set of complete system"14:57
dos1Ubuntu is actually a bit closer to Maemo in this regard, with their ubuntu-desktop, kubuntu-desktop, xubuntu-desktop etc. metapackages14:58
dos1but well, maybe it's possible to use tasksel somehow14:59
dos1err, those ubuntu metapkgs are provided by tasksel :)14:59
dos1so I quess that's the answer - it's as close to Debian as we can get15:00
jonwilwhat we really need is some way to say "this package is sticky and shouldn't be uninstalled automatically" (IIRC apt-get/dpkg/etc supports that but the Maemo version is missing that functionality)15:01
jonwilThat way we can mark packages are "must keep" and stop them from being uninstalled unless the user asks for them to be uninstalled...15:01
dos1this "stickiness" is meant to be used by user, not by distro maintainers15:01
jonwilno, there is some other feature that lets you mark something as "system" or something15:01
dos1oh, that one15:02
dos1it's for showing big warning "you're trying to remove essential package, type 'Yes I know what I'm doing' to continue" :P15:02
jonwilyes, the one that makes sure that whatever the apt-get command for removing unnecessary packages is wont remove them automatically15:03
dos1autoremove is actually another thing15:03
jonwiloh ok15:03
dos1proper dependencies alone should do it just fine15:03
dos1the proper way to handle current metapkg mess would be to use tasks together with package priorities15:06
dos1https://wiki.debian.org/tasksel15:06
*** M13 has joined #maemo-ssu15:06
jonwilok, be back soon, need to go get some food before the shops shut :)15:06
jonwilwill stay idle in IRC and read scrollback when I get back :P15:07
dos1this way you're free to replace any package you want, for instance with its free replacement15:07
dos1and still you have control under dependences of whole system and you can bring back original state with few commands15:08
*** freemangordon has quit IRC15:11
*** freemangordon has joined #maemo-ssu15:11
*** FlameReaper has joined #maemo-ssu15:17
*** Mihanizat0r has joined #maemo-ssu15:22
*** M13 has quit IRC15:26
*** amizraa has quit IRC15:35
*** amizraa has joined #maemo-ssu15:40
jonwilok, back16:00
jonwiltime to cook some Nachos then get into MCE work :)16:03
DocScrutinizer05meh, c'mon16:11
*** kolp has joined #maemo-ssu16:12
jonwilmeh what?16:13
dos1meh, nachos16:13
dos1;)16:13
DocScrutinizer05I can't run apt from within apt? Oh! then how about exec, or about shell scripts, or about batch, or about "echo $cmd > /etc/crontab" Ooops we got no cron, but we got alarmd16:13
DocScrutinizer05afk for coffee16:14
dos1IMO that would be worse than what's there at the moment16:14
DocScrutinizer05nothing can be worse than nokia/cssu MP16:15
dos1even if it's not worse, then for sure it's not better either :P16:16
DocScrutinizer05it has one killer feature, it doesn't create hard dependencies that forbid uninstalling of packages16:17
DocScrutinizer05resp uninstall MP when e.g. installing procps16:18
dos1I'm for doing it "the debian way", with tasksel and/or metapackages done properly16:18
dos1which doesn't create hard dependencies either16:18
DocScrutinizer05the OK all for it, grat!16:18
DocScrutinizer05great even16:18
DocScrutinizer05we just finally MUST sanitize that shit that Nokia inherited to us16:19
DocScrutinizer05it's like a psycho virus16:19
DocScrutinizer05needs to get nuked from God's great earth16:20
*** dhbiker has joined #maemo-ssu16:20
DocScrutinizer05since everybody has another notion what needs wnd what mustn't get into the MP16:20
dos1actually, it shouldn't be hard to modify current metapkgs to behave correctly16:21
dos1what's more challenging is proper upgrade path16:21
dos1it may or may not be a bit harder, it needs evaluating16:22
DocScrutinizer05first things first. Please come up with a proper sanitized MP16:23
DocScrutinizer05since obviously nobody likes the concept merlin1991 and me introduced how to maybe live *with* the old MP16:24
jonwilI think the idea that a metapackage forces exactly one and only one specific version of <package> is a very bad idea going forward16:24
DocScrutinizer05ergo we need to nuke MP completely, or transform it into something bearable16:25
dos1MP shouldn't depend on specific versions16:25
jonwilDebian packaging is something I know nothing about so I will let the gurus nut it out :)16:26
dos1that's right16:26
dos1and it should depend on as little as possible to get bare system working16:26
DocScrutinizer05*nokia* could do that MP shite, but a *community* impossibly can prosper with that abomination16:26
dos1the rest should be in recommends, which is installed by default anyway16:26
dos1but it's not hard dependency16:26
dos1exactly like tasks are done in Debian16:27
dos1actually, I've googled a bit and it turns out that tasksel now is implemented using nothing else than metapackages :P16:27
jonwilWe have 100% source code to all things that matter for package management on N900 Fremantle16:27
jonwiland we can change it any way we like16:27
DocScrutinizer05or maybe let's think why we need MP. Fix the crap that depends on it?16:29
DocScrutinizer05so we actually *can* purge the MP and still everybody happy16:29
dos1yup, deleting mp should be possible16:30
jonwilAnyone know if there are any packages on Fremantle that relate to cryptography and security (in particular, to encryption for internet traffic like SSL) that are currently closed source?16:30
dos1depending on mp is very lazy way to depend on particular version of system16:30
DocScrutinizer05I mean, one big problem always been: apt-get upgrade could create inconsistencies. But that been just because packages that depend on each other had no such dependencies declared in their .deb16:31
dos1you know, needs PR1.2 or greater - depends on mp version 1.2 or later16:31
dos1we shouldn't need to modify anything in package management, maybe except of HAM16:31
jonwilWith all the noise about security and cryptography lately (especially the whole NSA leaks thing) I dont want to have important-for-security-but-closed-source packages on my device, things I cant audit or play with or update to a later version with fixes for security issues...16:31
dos1rather packages are to be fixed16:32
dos1even closed ones can still be repackaged, so that shouldn't be a big issue16:32
DocScrutinizer05jonwil: I don't think we got such pkgs16:32
DocScrutinizer05dos1: exactly16:32
DocScrutinizer05but THAT is really a bit of work, I guess16:33
jonwilwifi security bits are closed but WiFi security is so weak anyway that you shouldn't be relying on it for actual security16:33
DocScrutinizer05jonwil: particularly you couldn't fix it16:33
jonwilmight take a look at the list of closed pkgs and look for anything that might be security-sensitive :)16:33
DocScrutinizer05nice plan16:34
DocScrutinizer05go ahead!16:34
DocScrutinizer05dos1: then otoh, how many packages might be there on maemo that have no big sister on debian, with proper dependencies?16:35
dos1I have no idea16:35
DocScrutinizer05wouldn't cloning the debian dependencies be pretty simple16:36
DocScrutinizer05we still can keep the friggin MP for the core system, the real *core*16:37
DocScrutinizer05like d-bus, alarmd, etc16:37
dos1those should be hard depends in MP IMO16:37
dos1and soft ones for the rest16:37
*** Guest23504 has quit IRC16:38
DocScrutinizer05I dunno if our apt supports that16:38
dos1unless it's very clippled down, it should16:38
DocScrutinizer05I mean, it doesn't even support "CONFLICTS"16:38
dos1crippled*16:38
DocScrutinizer05or is that just HAM?16:38
dos1I think it's just HAM16:38
DocScrutinizer05I think that's just HAM that blows chunks16:39
dos1I never noticed any symptoms of apt being dumbed down on Maemo16:39
DocScrutinizer05Nokia crippled apt and mohammadag reverted it ;-P16:39
DocScrutinizer05Nokia forgot that apt is FOSS when designing their ovi store16:40
DocScrutinizer05;-P16:40
*** luf has quit IRC16:40
dos1well, apt (proper, debian one :P) is pretty flexible16:41
dos1you can configure it to treat recommends as dependences, or suggestions, or whatever16:42
DocScrutinizer05prolly one of the reasons for dropping maemo and doing all things completely different in meego/harm16:42
DocScrutinizer05Nokia hoped to sanitize themselves from ovi-store16:42
DocScrutinizer05on fremantle they realized that they fscked up16:43
dos1what was so fscked up with it?16:43
*** arcean_ has joined #maemo-ssu16:43
*** DocScrutinizer06 has joined #maemo-ssu16:45
*** DocScrutinizer06 has quit IRC16:45
*** DocScrutinizer06 has joined #maemo-ssu16:45
*** DocScrutinizer05 has quit IRC16:45
*** DocScrutinizer06 is now known as DocScrutinizer0516:45
DocScrutinizer05what been my last post that made it?16:46
*** arcean has quit IRC16:46
DocScrutinizer05~ping16:47
infobot~pong16:47
*** FatPhil_ has quit IRC16:48
*** freemangordon has quit IRC16:48
dos1[15:43] <DocScrutinizer05> on fremantle they realized that they fscked up16:48
dos1[15:43] <dos1> what was so fscked up with it?16:48
*** freemangordon has joined #maemo-ssu16:48
*** FatPhil has joined #maemo-ssu16:49
dos1and then joins, leaves and your question :P16:49
jonwildid we find a solution to the problem of the kernel maintainer who insists that only Nokia can change Nokia drivers even though Nokia have long abandoned Linux and probably dont even care that the driver in question still exists?16:49
dos1let's send them copy of g_nokia with our patches named g_maemo16:50
dos1so only maemo devs will be able to change it :P16:50
*** oldtopman has joined #maemo-ssu16:51
DocScrutinizer05that's why they *had* to come up with aegis, and HARM16:53
DocScrutinizer05since when are packages owned by one maintainer, in kernel upstream?16:54
DocScrutinizer05dos1' idea sounds a feasible approach indeed16:55
*** BCMM has quit IRC16:56
jonwilIts not that they are owned by one maintainer, its that the lead developer (Nokia in this case) gets to decide how "their" driver works and that making wholesale changes without buy-in from the lead developer isn't acceptable16:56
dos1well, Nokia is not g_nokia maintainer nor lead developer anymore16:57
jonwilOne needs to convince the kernel guys of that16:57
jonwilconvince them that Nokia have abandoned g_nokia16:57
DocScrutinizer05well, it's not the first nor the only case where more than one driver mutually exclusive service the same resource16:57
DocScrutinizer05CSSU happily can switch to g_maemo16:58
DocScrutinizer05and g_maemo can coexist upstream alongside g_nokia16:59
DocScrutinizer05or did I miss sth?17:00
DocScrutinizer05while g_nokia bitrots and g_maemo obviously serves same purpose but gets constant maintenance, upstream maintainers eventually will drop g_nokia17:02
DocScrutinizer05dispute aolved17:03
DocScrutinizer05solved even17:03
DocScrutinizer05particularly when no freak on this planet is using g_nokia in any bew distro/release17:04
DocScrutinizer05new*17:04
DocScrutinizer05isn't that how FOSS and linux works?17:04
DocScrutinizer05I think upstream not allowing somebody $random-hacker to wreck g_nokia is a sane thing17:06
DocScrutinizer05we would get into endless battles17:06
DocScrutinizer05ruining linux at large17:06
DocScrutinizer05but just like nobody may wreck g_nokia, nobody may forbid a concurrent g_maemo17:07
DocScrutinizer05that's the FOSS way17:07
* DocScrutinizer05 wonders why he didn't think that way when bitching against lis302 brainfxored upstream drivers17:09
*** freemangordon has quit IRC17:13
*** freemangordon has joined #maemo-ssu17:13
*** cybertheque has joined #maemo-ssu17:20
cyberthequePerhaps what I want is SSU related...17:21
cyberthequeNokia-N900:/# /usr/bin/flasher --query-rd-mode --local17:22
cyberthequeflasher v2.8.2 (Jan  8 2010)17:22
cyberthequeUsing flashing protocol Mk II.17:22
cyberthequeFound device RX-51, hardware revision 210117:22
cyberthequeMethod is not supported in the current mode17:22
cyberthequeHow to enable 'update' mode in the shell?17:22
cyberthequeThanks...17:22
*** freemangordon has quit IRC17:23
*** discopig has joined #maemo-ssu17:23
*** discopig has quit IRC17:23
*** discopig has joined #maemo-ssu17:23
*** FlameReaper has quit IRC17:23
*** FlameReaper has joined #maemo-ssu17:28
*** FlameReaper has joined #maemo-ssu17:29
DocScrutinizer05you might try to start softupd17:33
DocScrutinizer05I dunno exactly but something is needed in addition to flasher to access the local NAND/CAL. However there are more direct simple ways to check for rd-mode than via flasher17:34
*** FlameReaper has quit IRC17:34
*** arcean_ has quit IRC17:34
DocScrutinizer05NOT for setting R&D mode though17:34
DocScrutinizer05anyway check out KP package and how it manages to flash new kernel locally17:35
*** FlameReaper has joined #maemo-ssu17:35
DocScrutinizer05you need to do similar "tricks" to get flasher --query-rd-mode --local working17:36
cyberthequek - will investigate.  my intention is to  hopefully enter the usb update mode from a shell command while logged in without needing to press the 'u' key and power cycle.  the mainboard is wired up on a test bench (no display) to a serial port interface so it is dangerous to physically manipulate it for flashing...17:36
jonwilQuestion, is there a reason why libc6 is not in CSSU?17:37
DocScrutinizer05you don't need to press the "U" key, NOLO will check for (external USB-attached) flasher to send a "Hello, here I am!" and that will serve as well for entering flashing mode17:38
cyberthequek - i will plug in the usb and see -- thanks17:38
DocScrutinizer05jonwil: well, prolly becasue it's considered high risk17:38
DocScrutinizer05~tell cyberthequeabout flashing17:39
DocScrutinizer05~tell cybertheque about flashing17:39
DocScrutinizer05cybertheque: there's a reason why the howto says "start flasher, *then* plug in USB to N900"17:40
DocScrutinizer05both ROMBL and NOLO will check for a fraction of a second if a "hello" message comes in via USB17:41
DocScrutinizer05if there does, they will enter flashing mode, just as if "U" been pressed17:42
DocScrutinizer05actually ROMBL not even checks for "U" (in case you want / need to do coldflash)17:42
cyberthequethanks, works17:42
cyberthequeConsole message seen is:17:42
cyberthequeNokia-N900:/# [44798.477172] g_nokia gadget: high speed config #1: nokia117:43
DocScrutinizer05btw your description of what you do sounds intriguing. May I ask...?17:43
dos1~rombl17:43
cyberthequeflasher-3.5 connects17:43
DocScrutinizer05ROM base BOOTLOADER17:43
DocScrutinizer05or stage-0 bootloader17:43
dos1thx17:43
DocScrutinizer05the only bootloader you impossibly can ruin or mess up17:44
*** FlameReaper has quit IRC17:44
DocScrutinizer05it supports coldflashing to flash a new XLOADER and NOLO, if you ever manage to break them17:44
cyberthequei have a series of tests using the stock kernel with initrd on boot - need to load (not flash) these and view console messages.  At some point I will need to thoroughly understand the bootloader however and appreciate your comments.17:45
DocScrutinizer05dos1: check your PC's syslog, when plugging in a powered-down N900 to USB. First thing you see is ROMBL greeting17:45
*** FReaper has joined #maemo-ssu17:46
cyberthequeexactly - i have installed a libusb dummy for ROMBL17:46
dos1I have to trust your words :)17:46
DocScrutinizer05second must be NOLO then, then 3rd which is linux kernel17:47
dos1xloader and nolo have to be signed, I guess?17:48
Palionly xloader is signed17:49
Palinolo is not signed on n90017:49
*** arcean_ has joined #maemo-ssu17:50
DocScrutinizer05xloader17:50
DocScrutinizer05NOLO not17:50
DocScrutinizer05it would be xloader to check NOLO signature - it doesn't17:51
DocScrutinizer05ROMBL however checks xloader's signature17:51
dos1interesting17:51
DocScrutinizer05particularly interesting if N900 xloader could boot up N917:52
Palino, because n9 has another signature17:52
Palialso different n9 hw revs have different signature17:53
DocScrutinizer05oooh right :-/17:54
Palimaybe different n900 hw revs have different signing key?17:55
DocScrutinizer05well, I doubt they have different ROMBL signature, since frequently you don't change ROMBL at all. But for sure 3630 SoC has a different ROMBL than 343017:55
DocScrutinizer05the key is 2hardcoded" to the ROMBL17:56
DocScrutinizer05and I doubt N900 hw revs have changed ROMBL though they *may* have done this17:57
Palijonwil: if you have patches for libc6 which should be in CSSU, ask merlin for new repo and push changes17:57
cyberthequeI can query the device and reboot it but not load to it in this state. i don't suppose the ssu ever needs to simply load images rather than flashing them...17:57
DocScrutinizer05you could test by coldflashing with overriding hw code to another version17:57
Palicybertheque: I did not read full backlog, but do you want to boot n900 into "update" emmc mode?17:58
jonwilThe CSSU libc thing was in reference to things Nokia bundled with libc i.e. timezones17:58
cyberthequei just need to load images, not update (flash) them, without pressing the 'u' key and plugging/unpluggin the usb cable17:58
jonwili.e. the idea was to put Fremantle libc with newest timezone files in CSSU17:58
DocScrutinizer05cybertheque: loading images to RAM can be done by NOLO only, you need to physically reboot the system to give NOLO full control over the hw17:59
DocScrutinizer05in linux you'd do same via e.g. kexec17:59
DocScrutinizer05_not_ via NOLO load-to-ram mechanism18:00
cyberthequewhat do you suggest?  short of wiring a momentary contact switch to the 'u' key on the keyboard that i can press at a distance, is there a chance of doing it in software?18:00
DocScrutinizer05I elaborated on it some lines above - you don't need the 'U' key18:01
cyberthequei will re-read, but i believe that i did what you said, and i cannot load, only flash or query18:02
DocScrutinizer05you just need to make sure you _first_ start flasher, only _then_ plug N900 USB to PC18:02
Palicybertheque: if you want to load images, you need to connect n900 via usb18:02
cyberthequeit is connected by usb18:02
Paliand then use desktop flasher-3.518:02
cyberthequedid that - start flasher-3.5 and then plug in the usb18:02
Paliflasher-3.5 -k <kernel_image> -l -b18:03
cyberthequewhile the n900 is booted into the o/s18:03
Palicybertheque: use linux version, open keyboard and press U18:03
Paliand after that connect usb18:03
DocScrutinizer05DAMN Pali ;-P18:03
cyberthequewhole idea is to avoid the need to press 'u'18:03
DocScrutinizer05^^^18:03
Paliok, then you need to start flasher and insert usb cable18:03
Paliand use linux version of flasher18:04
Paliand make sure that your battery is charged to full18:04
cyberthequeindeed - is the linux version different in capabilities then?  btw, the mainboard is powered externally (not depending on a battery)18:04
DocScrutinizer05or simply, read *full*18:04
DocScrutinizer05~flash18:04
infobotextra, extra, read all about it, maemo-flashing is http://wiki.maemo.org/Updating_the_tablet_firmware18:04
Paliwindows and mac systems has very slow usb enumeration, so you need to press U because win/mac does not see device on usb18:05
DocScrutinizer05yes, the linux version runs in an environment that's better understood18:05
DocScrutinizer05Pali: I don't even know if flasher relies on ENUM or just sends a hello-sequence to all USB endpoints all the time, to detect device18:07
cyberthequeok - linux flasher is next then.  so to clarify, while booted into the running o/s on the n900, i will be able to enter the loader's usb 'flashing' or 'update' mode, load images and reboot into them?  or can i perhaps simply switch the n900 to the 'update' mode while in the booted o/s, reboot, and have it talk to the external flasher to permit loading images at that point?18:07
DocScrutinizer05no18:08
Paliit doing infinite loop with some sleep() and asking kernel for enumerate devices and read usb device list18:08
DocScrutinizer05Pali: mhm, thanks18:08
Palimy 0xFFFF flasher doing same18:08
DocScrutinizer05:-)18:08
Palithere is no way to tell kernel: sleep process until device with this device id is present in usb...18:08
DocScrutinizer05cybertheque: for the 5th time: start flasher on PC, then plug in the powered down(!!!) N900, so it runs NOLO which will check for any hello-msg on USB sent by flasher. When such msg is detected during the sub-second time window, it enters flashing mode (resp "load-to-ram" mode)18:10
DocScrutinizer05no matter if U pressed or not18:10
cyberthequeok - powered down state - that was the missing ingredient.  thanks18:10
Palicybertheque: if host (desktop computer) send some hello message to n900, it will enter into NOLO flashing mode18:11
cyberthequegot it - was confused that it could do it while in the running state18:11
Palibut, there is "update" userspace mode18:12
Paliand in this runlevel you can flash eMMC18:12
DocScrutinizer05yes, but that probably can't load to RAM and boot that kernel18:12
Palino18:12
Paliin "update" mode is kernel already running18:12
DocScrutinizer05exactly18:12
Paliit is just classic linux runlevel18:12
DocScrutinizer05thus: use kexec when you plan to do such tricks from running system18:13
PaliI was just confused about "update" mode...18:13
DocScrutinizer05yeah18:13
DocScrutinizer05~xyawn18:13
infobothmm... xyawn is nap18:13
DocScrutinizer05DANG!!18:13
DocScrutinizer05~xyawn18:13
infoboti guess xyawn is big coffee18:13
DocScrutinizer05better18:13
Palithere is also "test" and "normal" runlevel modes18:13
cyberthequeok, did not work with the windows flasher, so will try with linux flasher next (just for the record)18:14
Palicybertheque: you can try my 0xFFFF flasher on linux (if flasher-3.5 will not work)18:14
DocScrutinizer05yeah, we toldya it will most likely not work with windows flasher18:14
cyberthequeok, thanks Pali - will try both18:15
Pali~0xFFFF18:15
DocScrutinizer05if it would work on all platforms, Nokia hadn't invented that clumsy "press and hold 'U'" trick18:15
Paliinfobot: 0xFFFF is https://gitorious.org/0xffff/18:15
infobotPali: okay18:15
*** FReaper has quit IRC18:22
cyberthequeok (don't flame me for using windows yet again) but i found a trick that worked.  1. start flasher-3.5 on windows (usb unplugged); 2. plug in usb with n900 in powerd down state; 3: n900 boots to o/s; 4: use power switch to power cycle n900; 5. n900 boots into a mode (have a complete console trace of it) that accepts the loading of the images18:22
cyberthequeand now i know from viewing the console log why my initrd didn't work before -- too large18:23
dos1I guess you could make a shortcut from 2 to 418:24
dos1plug usb to already powered on n900 and then power cycle18:25
cyberthequewill try18:25
*** NIN101 has joined #maemo-ssu18:25
cyberthequedos1: required two power cycles with the shortcut and the device usb is enumerated in pc-suite mode on first power cycle so i guess the five step procedure is easier for me18:33
*** BCMM has joined #maemo-ssu18:33
dos1strange18:33
cyberthequetried the 5 step procedure a few more times - works every time for me.  good to know - will save a lot of wear and tear on 30 ga wire harness to the board.18:37
*** freemangordon has joined #maemo-ssu18:40
*** arcean_ has quit IRC18:40
*** sixwheeledbeast has left #maemo-ssu18:40
cyberthequeBTW, for the record, running the stock o/s there is a time-out on the serial port of about five seconds after which the first input character is not recognized; i have gotten into the habit of typing the backspace key prior to any text input delayed more than five seconds.18:40
*** sixwheeledbeast has joined #maemo-ssu18:42
DocScrutinizer05known issue, UART power-down timer18:44
DocScrutinizer05you can change the power-down timeout by writing to /sys/*/*/*/usart/power/timeout or similar sysnode18:44
cyberthequek - tnx18:44
*** LauRoman has joined #maemo-ssu18:45
*** arcean_ has joined #maemo-ssu18:52
cyberthequefor the record: Nokia-N900:~# cat /sys/devices/platform/serial8250.0/sleep_timeout18:52
cybertheque1018:52
cyberthequeNokia-N900:~# echo "20" > /sys/devices/platform/serial8250.0/sleep_timeout18:52
cyberthequeNokia-N900:~# cat /sys/devices/platform/serial8250.0/sleep_timeout18:52
cybertheque2018:52
*** BCMM has quit IRC20:12
*** arcean_ has quit IRC20:16
*** arcean_ has joined #maemo-ssu20:16
*** arcean_ has quit IRC20:27
*** arcean_ has joined #maemo-ssu20:28
*** xes has joined #maemo-ssu20:33
*** oldtopman has quit IRC20:45
*** Vlad_on_the_road has joined #maemo-ssu21:07
*** int_ua has joined #maemo-ssu22:05
*** arcean_ has quit IRC22:07
*** int_ua has quit IRC22:14
*** int_ua has joined #maemo-ssu22:14
*** Vlad_on_the_road has quit IRC22:27
*** Vlad_on_the_road has joined #maemo-ssu22:27
*** jonwil has quit IRC22:33
*** _rd has joined #maemo-ssu22:36
*** Mihanizat0r has quit IRC23:07
*** sunny_s has joined #maemo-ssu23:11
*** MohammadAG has quit IRC23:15
*** MohammadAG has joined #maemo-ssu23:22
*** xes_ has joined #maemo-ssu23:48
*** xes has quit IRC23:50
*** xes_ has quit IRC23:50
*** xes has joined #maemo-ssu23:50
*** int_ua has quit IRC23:51
*** sunny_s has quit IRC23:55

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!