IRC log of #maemo-ssu for Thursday, 2012-11-01

*** sicelo^ has quit IRC00:19
*** sicelo^ has joined #maemo-ssu00:19
*** sicelo^ has joined #maemo-ssu00:19
*** MrPingu has joined #maemo-ssu00:49
merlin19913rd party c# api devs should be killed on the spot00:52
merlin1991I got an OutOfMemory Exception when calling some stuff00:52
merlin1991guess what I have 5 gigs of free ram, It's actually a nullpointerexception !00:53
* merlin1991 headdesks00:53
*** MrPingu has quit IRC00:59
*** MrPingu has joined #maemo-ssu01:00
PaliDocScrutinizer05, you are right. by default maemo using 192.168.2.15 and 192.168.2.14 ip addresses for USB network01:07
Palibut if somebody change it in /etc/network/intterfaces and maybe disable route, we need to calculate second ip addr somehow...01:08
Palimerlin1991 ^^^01:08
DocScrutinizer51thought as much01:09
merlin1991Pali: we should add an option to run a dhc client on the usb connection instead though01:10
merlin1991s/dhc/dhcp/01:11
infobotmerlin1991 meant: Pali: we should add an option to run a dhcp client on the usb connection instead though01:11
Palimerlin1991, this can be configured by some icd2 plugin01:11
merlin1991but the script would run regardless01:11
Palibut adding also configuration should be usefull (to have all at one place)01:12
merlin1991you mean ip gw and whatnot?01:12
Palithat option to run udhcpc instead ifup01:14
Palihm, I think that /etc/network/interfaces support udhcpc client01:14
DocScrutinizer51YAY THAT took a WHILE... I wondered why this shell expansion magic looked fishy01:15
DocScrutinizer51now I figure it might've been because of IPv601:15
Paliand if you call ifup usb0 it look at interfaces and run needed comamnds...01:15
*** Pali has quit IRC01:19
*** MrPingu has quit IRC01:20
*** MrPingu has joined #maemo-ssu01:23
*** MrPingu has quit IRC01:31
*** MrPingu has joined #maemo-ssu01:32
*** myuu_ has quit IRC01:34
*** myuu_ has joined #maemo-ssu01:35
*** NIN101 has quit IRC02:15
*** M4rtinK has quit IRC02:49
*** arcean has quit IRC03:31
*** dafox has quit IRC04:53
*** sicelo^ has quit IRC05:20
*** sicelo^ has joined #maemo-ssu05:21
*** amiconn_ has joined #maemo-ssu05:46
*** amiconn has quit IRC05:46
*** amiconn_ is now known as amiconn05:46
*** DocScrutinizer05 has quit IRC06:03
*** DocScrutinizer05 has joined #maemo-ssu06:03
*** dhbiker has quit IRC06:58
*** M13 has joined #maemo-ssu07:14
*** dhbiker has joined #maemo-ssu07:16
*** luf has joined #maemo-ssu08:01
lufHmmm, the really only supported network in bluez-4.99 is NAP :( I don't see why. I'll try to see into git if there is some comment on it.09:39
*** Pali has joined #maemo-ssu09:46
*** Lava_Croft has quit IRC10:00
kerioluf: ohai10:04
lufkerio: I red the specification and I see no reason to support the others. PANU is the client side and GN isn't very reasonable.10:08
keriomakes sense10:08
keriocan two NAP connect together, tho?10:08
lufNo. One have to be PANU ;)10:08
lufhttp://www.bluetooth.org/Building/HowTechnologyWorks/ProfilesAndProtocols/PAN.htm10:09
kerioso...10:09
lufSo PANU can be initalized using pand.10:09
kerioNAP too10:10
lufBut it's not necessary to run another daemon when bluez supported it.10:14
kerioi still have no idea of what os x wants to be able to click "connect" on the bluetooth PAN ui10:15
lufBut I'll take a look if  PANU can be supported as client side in bluez.10:15
lufIt's not on top of my TODO to be honest.10:16
lufkerio: let's use apple support ;)10:17
keriohahaha yeah right10:18
kerio"ok, tap the home button, go to settings and then internet tethering..."10:18
kerio"i don't have an iphone" "BLASPHEMER!"10:18
lufkerio: it sounds like enabling the NAP.10:19
keriobut i did D:10:19
kerioluf: anyway, brctl addbr pan0, then the dbus call to org.bluez.networkserver to enable NAP on pan0, right?10:20
lufNo.10:20
kerioD:10:20
lufYes. Sorry.10:21
keriok10:21
lufYou're right. I used another bridge name.10:21
lufThe connection are using bnep10:21
* DocScrutinizer05 yawns10:21
lufThen you also need to ifconfig to up the bridge interface with some IP.10:22
lufkerio: I know it's not good but I try to make a patch to support this without bridge :)10:23
kerionah, it's fine10:23
DocScrutinizer05where's the headline of all this that gets me excited?10:24
kerioluf: hm, there's no SDP either10:25
DocScrutinizer05\o/10:25
DocScrutinizer05not THAT excited, please!10:25
lufkerio: strange. I have no problem with it.10:26
lufdbus-send --system --type=method_call --dest=org.bluez /org/bluez/$(pidof bluetoothd)/hci0 org.bluez.NetworkServer.Register string:nap string:pan010:26
lufkerio: any log entry?10:27
keriothe problem is on osx's side only10:28
kerioit's just "connection" and then "disconnection"10:28
lufkerio: pand --connect from os x side?10:28
kerioos x doesn't use bluez10:28
kerioi wish it was that easy :)10:29
lufkerio: did you pair the phone and mac when PAN is enabled on N900?10:34
kerioyes10:34
lufkerio: http://hints.macworld.com/article.php?story=20070327142723231 it sounds easy :D10:37
freemangordon10:21 <luf> kerio: I know it's not good but I try to make a patch to support this without bridge :)10:42
freemangordoneither that or help to bring a kernel in CSSU10:43
luffreemangordon: I know. That's the reason I made a patch.10:43
lufAnd the patch is working ...10:44
luffreemangordon: I also have the kernel-module-bridge on my TODO ...10:45
freemangordonnaah, not another module ouside of the kernel10:46
freemangordonsee what happens with fcam and similar10:47
luffreemangordon: I don't understand. You mean to recompile the kernel with the module support?10:48
freemangordonDocScrutinizer05: maybe we should reconsider libc fallback strategy for pselect/ppoll10:48
freemangordonluf: we don't need another package containing kernel module for a specific kernel version10:48
luffreemangordon: ok now I also understand the second part of condition (after or) :D10:49
freemangordon:D10:50
*** MrPingu has quit IRC11:00
*** Lava_Croft has joined #maemo-ssu11:06
*** MrPingu has joined #maemo-ssu11:16
*** Lava_Croft has quit IRC11:17
*** Mihanizat0r has joined #maemo-ssu11:20
*** Lava_Croft has joined #maemo-ssu11:20
*** Lava_Croft has joined #maemo-ssu11:21
*** M13 has quit IRC11:22
*** dhbiker has quit IRC11:32
luffreemangordon: how can I help with kernel in CSSU?11:34
freemangordonluf: implement what zeq was supposed to do in glibc11:34
freemangordonppoll/pselect fallback path when there is no kernel support for ppoll/pseslect11:35
luf:)11:36
*** dhbiker has joined #maemo-ssu11:36
lufIs here some volunteer who wants write wiki pages about experimental things instead of me? :)11:39
freemangordonluf: you'd better ask that on TMO11:43
DocScrutinizer05freemangordon: what's the problem with the fallback strategy?11:50
freemangordonnoone to implement it :D11:50
DocScrutinizer05o.O11:51
DocScrutinizer05how hard could that be?11:51
DocScrutinizer05sounds like 10 c scrlines to me11:52
DocScrutinizer05ooh yeah, we had that nifty optimal way to do it, by changing the jump table addr11:53
freemangordonDocScrutinizer05: yeah, but zeq is mia again, and I am just too overprojected11:53
DocScrutinizer05well, we could go for a simplified stategy that's "terribly" suboptimal for the fallback case, but way simpler to implement11:53
DocScrutinizer05using a static boolean and a if then else in the routine11:54
DocScrutinizer05or "local" static function pointer11:55
DocScrutinizer05which works kinda same like a boolean plus if-then-else11:55
DocScrutinizer05such local static *function should be implementable with no more than 30 lines of code, var declarations and init code included11:57
freemangordonDocScrutinizer05: hmm, I am going to bit the bullet and clone glibc. I hate semi-finished stuff more than you do12:05
freemangordon*bite12:05
DocScrutinizer05freemangordon: :-)12:07
keriofreemangordon: =D12:08
kerioi won't be able to test it this weekend, tho12:08
freemangordonkerio: it won;t be ready this weeken tho :P12:08
keriothen it's all good12:09
DocScrutinizer05freemangordon: one tiny request: add a logger line to indicate when fallback kicks in12:09
DocScrutinizer05one-shot of course12:09
freemangordonDocScrutinizer05: ok12:09
DocScrutinizer05thanks12:10
freemangordonmakes sense12:10
freemangordonthough I am not sure where i should log that, having in mind the boot stage glibc is first loaded ;)12:10
DocScrutinizer05hehehe rrrright12:10
DocScrutinizer05well, I guess a call to logger will handle that12:11
DocScrutinizer05lemme check back12:11
freemangordonmaybe a better idea will be to add an additional line to /lib/libc6.so output12:12
DocScrutinizer05meh, logger(1), uses syslog(3)12:12
DocScrutinizer05I thought there'd be a logger(3) call12:12
freemangordonso it will be possible to get if we are in fallback mode (and log that) later, when there is syslog available12:13
freemangordonthe other option is to check if kernel exposes to userland interface to printk12:13
freemangordonanyway, will check the possibilities12:14
DocScrutinizer05nfc what happens when a process is calling syslog(3) on a system without syslogd running12:14
freemangordonme neither12:15
DocScrutinizer05I used to believe it'll log to dmesg ringbuffer then, but obviously that's incorrect assumption12:16
DocScrutinizer05otoh writing to /dev/log(?) will not automatically propagate to syslogd ?12:17
freemangordonDocScrutinizer05: the only thing that I can bet is working by the time glibc is initially loaded is printk12:18
freemangordonthough i will check what preinit does12:18
freemangordonmaybe touch /tmp/glbc_fallback will do the job aswell12:18
*** NIN101 has joined #maemo-ssu12:21
DocScrutinizer05ooh /dev/log is a socket12:21
DocScrutinizer05:-/12:21
freemangordonDocScrutinizer05: nevermind, I will find a sane way to indicate that we run in fallback mode. I started to like "touch /tmp/glibc_fallback" very much12:23
DocScrutinizer05hey, some idea I had some 10min ago as well, so it can't be that bad if both of us have same idea12:24
freemangordon:)12:24
DocScrutinizer05but you'd also need a unlink  /tmp/glibc_fallback then12:24
DocScrutinizer05as otherwise it might get sticky, depending on fs of tmp12:25
freemangordon"/tmp" is tmpfs12:25
freemangordonthat's for sure (on maemo)12:25
DocScrutinizer05hmm, ok12:26
freemangordonhmm, though if someone remounts it to have gcc or whatnot running...12:26
keriojust add a new function is_glibc_using_the_pselect_fallback_or_not()12:26
freemangordonDocScrutinizer05: yeah, unlink as well12:27
freemangordonkerio: My self-esteem is not high enough to mess with libc ABI :P12:27
freemangordonor API12:28
freemangordonDocScrutinizer05: on the other hand... is that really needed12:28
freemangordonyou can always call __syscall(...) and check if the kernel supports that or not12:29
freemangordoni.e. instead of messing with glibc, we can have a simple helper binary12:29
freemangordoncalled later enough in the boot process, when we are sure we have syslg running12:30
freemangordontoldya I am smart :D:D:D12:30
*** NIN101 has quit IRC12:33
*** NIN101 has joined #maemo-ssu12:34
*** dhbiker has quit IRC12:37
*** dhbiker has joined #maemo-ssu12:38
*** M4rtinK has joined #maemo-ssu12:43
DocScrutinizer05you can also any time check if that *function == kernel_pselect or *function == glibc_psel_emu13:00
DocScrutinizer05adding according function zo glibc API shouldn't break anything aiui13:01
*** Mihanizat0r has quit IRC13:11
*** lizardo has joined #maemo-ssu13:46
*** Milhouse has quit IRC13:59
*** arcean has joined #maemo-ssu14:21
*** MrPingu has quit IRC14:31
*** MrPingu has joined #maemo-ssu14:31
*** MrPingu has quit IRC14:35
*** MrPingu has joined #maemo-ssu14:35
*** MrPingu has quit IRC14:40
*** MrPingu has joined #maemo-ssu14:40
*** MrPingu has quit IRC14:43
*** MrPingu has joined #maemo-ssu14:43
*** Pali has quit IRC15:06
freemangordonDocScrutinizer05: http://pastebin.com/SWmPpMXc15:46
freemangordonLOL15:46
freemangordonthose guys are are saving me a lot of work :D15:47
DocScrutinizer05damn, 43 lines15:47
DocScrutinizer05;-P15:47
freemangordonnaah, this is what is currently in glibc source15:47
freemangordoni didn;t touch it15:47
freemangordonthe point is:15:48
freemangordonif (result == -1 && errno == ENOSYS)15:48
freemangordonresult = __generic_ppoll (fds, nfds, timeout, sigmask);15:48
freemangordonthe fallback is already there :D:D:D15:48
DocScrutinizer05yeah, in 43 lines15:49
DocScrutinizer05though, subtracting whitelines and comments, we might get under 3015:50
jon_ymagic happens in INLINE_SYSCALL?15:51
freemangordonDocScrutinizer05: this is the code we currently run, what to substract?15:57
freemangordonthough you miss the #include at the end15:58
freemangordonjon_y: no magic here15:58
DocScrutinizer05freemangordon: nevermind, just kidding15:58
freemangordonaah, ok15:58
freemangordonso, we just need __NR_ppoll defined and everyone will be happy15:59
DocScrutinizer05[2012-11-01 10:57:14] <DocScrutinizer05> such local static *function should be implementable with no more than 30 lines of code, var declarations and init code included15:59
freemangordonaah, ok, i've missed that15:59
freemangordon:D15:59
jon_ycombine all your vars into a memblob accessed by hard coded offsets :)16:00
jon_ywill save you a few lines16:00
jon_ychar *blob[256]; .... (int *)(&blob[34])....16:01
*** Pali has joined #maemo-ssu16:04
*** luf has quit IRC16:13
keriojon_y: plenty of empty lines there16:47
keriofreemangordon: does that mean that glibc already takes advantage of pselect?16:49
keriowell, pselect/ppoll16:49
jon_ykerio: ?16:49
jon_ywhat's with the length limit anyway?16:49
keriojon_y: just a joke :)16:49
jon_yok16:50
jon_ywell, it's not inline, so 30 lines aren't much16:50
*** dafox has joined #maemo-ssu16:56
*** M4rtinK has quit IRC17:18
freemangordonkerio: that means that there is no need to patch glibc further17:21
freemangordonit only needs to be build agains the correct kernel headesr17:21
freemangordon*headers17:21
kerioi see17:21
keriowait, what do you mean built against the correct kernel header?17:22
freemangordonheaders from a kernel that implements ppoll/pselect17:22
keriohm17:22
keriowould it be compatible with the stock kernel afterwards, though?17:22
freemangordonread the backscroll17:22
keriodude, if i asked it means that i didn't understand that from the backscroll17:23
keriofor all i know it could be a fallback for compile-time17:23
kerioi mean, it's not, but it could be less runtimy17:23
freemangordonit is a runtime fallback17:24
keriowhy does it need different kernel headers then?17:24
freemangordonfor compile time, to define  __NR_ppoll17:24
keriok17:25
freemangordonthis is not defined in stock, as the call is not implemented17:25
freemangordon(not defined afaik)17:25
keriowhat kernel headers is the current glibc compiled against, right now?17:25
freemangordon2.6.17 or something17:25
freemangordonor even older17:26
DocScrutinizer05freemangordon: brilliant job!17:27
freemangordonDocScrutinizer05: which one?!?17:28
DocScrutinizer05spotting the right 43 lines in glib17:28
freemangordonaah, ok :D17:28
freemangordonI need o spot those for pselect too17:29
freemangordon*to17:29
freemangordonbut that will be after I am back home17:29
freemangordonbye, bbl17:29
*** Pali has quit IRC17:30
*** luf has joined #maemo-ssu17:45
luffreemangordon: why don't you like bridge for BT networking?17:47
freemangordontherei is no bridge in stock kernel18:18
freemangordonluf: ^^^18:18
*** Pali has joined #maemo-ssu18:24
*** dhbiker has quit IRC18:41
freemangordonDocScrutinizer05: the code for pselect is the same as for ppol18:41
freemangordon*ppoll18:42
*** luf has quit IRC18:51
freemangordonguys, I need some advice. I am going to rebuild glibc against kernel headers with ppoll/pselsect support. Now, there are 2 options - I either do install kernel headers in /usr/include or pass directory with kernel headers ti glibc make19:01
freemangordonwhich one sounds better?19:02
*** dhbiker has joined #maemo-ssu19:04
*** dhbiker_ has joined #maemo-ssu19:04
*** dhbiker has quit IRC19:05
*** dhbiker has joined #maemo-ssu19:16
*** M4rtinK has joined #maemo-ssu19:17
freemangordonmerlin1991: how shall we proceed with glibc in CSSU, it seems glibc does a fallback for ppoll/pselect if the kernel does not support them19:18
merlin1991test that :)19:18
freemangordonI am buiding a kernel with patch from zeq right now to be able to compile glibc wit ppoll/pselsect support, will test glibc first on kernel with no support of course :D19:19
freemangordonbut ly looking at the code the test will pass19:20
freemangordon*by19:20
freemangordonthen what?19:20
merlin1991what does it need @ build time for glibc to build that?19:20
freemangordonkernel headers installed in /usr/include19:20
merlin1991is it enough to fiddle with the kernel headers a little or do you need  full blown ppoll kernel?19:21
freemangordon(of kernel with ppoll/pselect support)19:21
freemangordonwell, you need only the headers19:21
freemangordonbut in /usr/include19:21
freemangordonand so far the only way that could happen is make install_headers from the kernel build tree19:22
freemangordon(the only way I found)19:22
freemangordonmerlin1991: make headers_install ARCH=arm INSTALL_HDR_PATH=/usr/include19:24
merlin1991is there a huge difference in the headers or is it trivial to reproduce by hand?19:27
*** MrPingu has quit IRC19:34
Palifreemangordon, can you send me yours patched conversation css?19:35
*** arcean has quit IRC19:40
freemangordonPali: sorry, I forgot19:42
freemangordonwill do it in a couple of minutes19:42
freemangordonmerlin1991: should be trivial, though I didn't check. here is the patch in the kernel http://www.snewbury.org.uk/maemo/sigmask/tif_restore_sigmask-syscalls-arm.patch19:43
DocScrutinizer05freemangordon: do a diff? ;-)19:48
DocScrutinizer05(of kernel-headers)19:49
freemangordonPali: http://merlin1991.at/~freemangordon/css.zip19:50
freemangordonI hope i didn't missed a file19:50
freemangordonDocScrutinizer05: i don;t have the original headers anymore19:50
DocScrutinizer05pretty poor handling19:51
DocScrutinizer05;-)19:51
DocScrutinizer05a cp -l /usr/include could've saved you some work and options19:52
freemangordonhttp://pastebin.com/uxJRxxLJ19:52
freemangordontha patch as < 100 lines19:53
freemangordonDocScrutinizer05: ^^^19:54
DocScrutinizer05i'd say <1019:54
freemangordonthis is not the whole patch19:54
freemangordononly headres part19:54
freemangordon*headers19:55
*** Pali has quit IRC19:55
*** dhbiker has quit IRC20:01
freemangordonwell, new glibc with "stock" kernel-cssu3 boots, lets see if the test binary still hangs20:01
freemangordonthu fuck20:08
freemangordonsymbol pselect, version GLIBC_2.4 not defined in file libc.so.6 with link time reference20:08
freemangordon:(20:08
freemangordonWTF? pselect is not exported?20:13
*** dhbiker has joined #maemo-ssu20:15
*** crashanddie has joined #maemo-ssu20:24
*** crashanddie has joined #maemo-ssu20:24
*** ekze has quit IRC20:25
*** Pali has joined #maemo-ssu20:40
*** M4rtinK has quit IRC20:51
*** DocScrutinizer05 has quit IRC20:51
*** DocScrutinizer05 has joined #maemo-ssu20:55
*** DocScrutinizer05 has quit IRC20:56
*** DocScrutinizer05 has joined #maemo-ssu21:12
*** NIN101 has quit IRC21:15
freemangordonI am stupid, I installed kernel headers in wring directory21:15
freemangordonthe correct cmd is:21:15
freemangordonmake headers_install ARCH=arm INSTALL_HDR_PATH=/usr21:15
DocScrutinizer05my lenovo thinkpad T500 is stupid, it rather emergency shuts-down system due to CPU overtemp rather than cranking up fan speed from 2800rpm to at leas 4900rpm :-((21:24
DocScrutinizer05now how stupid is *that*, eh?21:25
ShadowJKIf laptops didn't shutdown, they wouldn't sell any of those stupid cooling pads21:25
*** polemo has joined #maemo-ssu21:26
DocScrutinizer05I'm thoroughly pissed right now21:26
*** ekze has joined #maemo-ssu21:32
*** ekze-nyan has joined #maemo-ssu21:32
*** nox- has joined #maemo-ssu21:34
*** ekze has quit IRC21:36
*** ekze-nyan is now known as ekze21:36
*** ekze has joined #maemo-ssu21:37
*** polemo has quit IRC21:39
*** crashanddie has quit IRC21:48
freemangordonDocScrutinizer05: are you sure you fan is ok?21:50
freemangordon*your21:50
DocScrutinizer05well, if controlled manually, it spins up to 500021:50
freemangordonhmm21:51
freemangordonand who controls it otherwise? BIOS?21:51
DocScrutinizer05modprobe -r thinkpad_acpi; modprobe thinkpad_acpi fan_control=1; echo level full-speed > /proc/acpi/ibm/fan21:51
DocScrutinizer05BIOS21:51
DocScrutinizer05that weird comprocessor21:51
DocScrutinizer05co-processor21:52
freemangordonisn't there some fan policy settings there?21:52
DocScrutinizer05that also does kbd mapping and scanning21:52
DocScrutinizer05yep, there's some fan setting I think, though I'm sure I cranked that up to as high fan as possible21:52
freemangordoncheck if there is something like "target CPU temp" and put that as low as possible21:54
freemangordonok, *working* glibc is built :)21:55
freemangordonI am really impressed maemo was able to boot without ppoll/pselect21:55
freemangordon60% of the stuff was not working, but still21:56
DocScrutinizer05WTF?? since boot my laptop isn't aware of winter time anymore21:57
freemangordonok, chldspin locked with "stock" kernel-cssu, lets see how it will be with patched21:58
DocScrutinizer05freemangordon: btw you know you can speed up locking of childspin by factor 1000 simply by running a reniced -15 do while true;:;done loop in another shell22:01
freemangordonI am just running HAM :D:D:D22:02
DocScrutinizer05lol22:02
freemangordonyeah, and start to rotate landskape->portrait->...22:02
freemangordon*landscape22:02
freemangordonafter 3-4 times it locks22:02
DocScrutinizer05same prio tasks don't jelp anything22:03
DocScrutinizer05help*22:03
DocScrutinizer05H-D is higher prio, HAM prolly isn't22:03
freemangordonmaybe, IDK but it works22:03
DocScrutinizer05by random, yeah22:04
DocScrutinizer05my method though yields immediate reproducable reliable scriptable results22:06
*** Pali has quit IRC22:07
freemangordonyou'll have a chance to test it :)22:07
DocScrutinizer05if you're doing real testing you don't want a bug to show up _eventually_ since your supposed fix might actually only extend the period until the bug shows up22:09
freemangordonwell, without fix it takes 10-15 seconds22:10
DocScrutinizer05if the patch e.g. reduces statistical probability by factor 10, your test wouldn't show the bug still is there unless you do 30..40 rotations. My method would kick in after 1s instead after 0.1s22:10
freemangordonok, give me the command22:10
DocScrutinizer05run a reniced busy loop in another (sub)shell22:11
DocScrutinizer05that's all iirc22:11
DocScrutinizer05reniced to higher prio22:11
DocScrutinizer05lesser niceness22:11
DocScrutinizer05so the childspin process actually is seeing premetions22:12
DocScrutinizer05pre-emptions22:12
DocScrutinizer05if that's a word at all22:12
freemangordoncould you give me the exact script, I am fighting with while...22:13
DocScrutinizer05while true; do :; done22:13
freemangordonworks22:14
DocScrutinizer05you could also do a for ((i=0; i<100000; i++)); do :; done22:15
DocScrutinizer05which has the advantage you don't need to kill it22:15
freemangordonok22:15
DocScrutinizer05alas that's bash22:15
freemangordonbad loop variable :)22:15
freemangordonanyway, while loop does the job too22:17
DocScrutinizer05try this: while [ $i -lt 10000 ]; do i=$((i + 1)); done22:17
freemangordoni=0; while [ $i -lt 10000 ]; do i=$((i + 1)); done22:18
freemangordoni'll run that 1000000 times22:18
freemangordonis that enough to lock chikdspin on your device?22:19
freemangordon*chldspin22:19
DocScrutinizer05I think so, it should lock device immediately22:19
freemangordondevice?22:19
DocScrutinizer05childspin22:19
freemangordonaah, ok :)22:19
freemangordonno, it works22:19
DocScrutinizer05maybe you need one or 2 0 more22:20
freemangordondid you run in xterm or in ssh session?22:20
DocScrutinizer05err ssh22:20
freemangordonok, me too22:20
freemangordoncannot lock it, even with 100000022:20
DocScrutinizer05lemme try to reproduce it22:20
freemangordonok22:20
freemangordon2 chldspin processes in 2 consoles, third one runing 1000000 while...22:24
freemangordonnone locked22:24
freemangordon:)22:24
kerionow do that with a forkbomb22:24
freemangordonwhy?22:24
kerioto see if it locks up :P22:25
keriohmm, i wonder if it actually does22:25
*** wmarone_ has joined #maemo-ssu22:25
*** wmarone has quit IRC22:25
DocScrutinizer05I friggin don't get how to renice22:27
DocScrutinizer05no matter what I do, the result is a niceness of 022:27
freemangordonrenice -1522:27
DocScrutinizer05doesn't do anything here22:27
freemangordonwhat?22:27
DocScrutinizer05what's the result of nice after your renice?22:28
freemangordonyou are trying to renice current sh?22:29
freemangordondo it from another console22:29
freemangordonyou'll see < int top22:29
DocScrutinizer05halebop:~ # renice -10 $$22:29
DocScrutinizer056224 (process ID) old priority 0, new priority -1022:29
DocScrutinizer05halebop:~ # nice22:29
DocScrutinizer05-1022:29
freemangordoni.e. renice -15 $pidof22:30
freemangordonyeah22:30
DocScrutinizer05~ # renice -15 $$22:30
DocScrutinizer05~ # nice22:30
DocScrutinizer05022:30
*** arcean has joined #maemo-ssu22:30
freemangordonmy device is overheating, stopping for a while :D22:30
DocScrutinizer05IroN900:~# renice -15 $$22:31
DocScrutinizer05IroN900:~# nice22:31
DocScrutinizer05022:31
DocScrutinizer05(bash ^^^)22:31
freemangordonhehe22:31
freemangordonhmm, that is strange22:31
DocScrutinizer05IroN900:~# renice -n -15 $$22:32
DocScrutinizer05IroN900:~# nice22:32
DocScrutinizer05022:32
freemangordonseems nice cannot get niceness :D22:32
freemangordonanyway, in top the process is marked with '<'22:32
DocScrutinizer05top doesn't show a changed niceness either22:32
freemangordonwhat ps | grep sh shows22:33
DocScrutinizer05dang, now it does22:34
DocScrutinizer05IroN900:~# ./childspin22:34
DocScrutinizer05......22:34
DocScrutinizer05that's it22:34
freemangordondo you want new glibc and the kernel, to confirm it fixes the bug?22:35
DocScrutinizer05the number of dots is 'live'22:35
freemangordonyeah, got it22:35
DocScrutinizer05not today, sorry. I'm soooooo late with breakfast22:36
freemangordonhehe22:36
DocScrutinizer05and tomorrow is robotnik again22:36
freemangordonanyway, i'll upload glibc to -devel22:36
DocScrutinizer05cool stuff. I'll most likely give it a try tomorrow22:37
freemangordonyou'll need new kernel too ;)22:37
DocScrutinizer05come up with a proper testing script22:37
DocScrutinizer05yeah, got that22:37
freemangordonwho, me?22:37
DocScrutinizer05I'll...22:37
freemangordonaah, ok22:37
freemangordonproper testing should be:22:37
freemangordoninstall new glibc22:37
freemangordonsee if it locks (it locks)22:38
DocScrutinizer05wonder how I could swap glib without rebooting, durin a script22:38
freemangordoninstall new kernel22:38
freemangordondpkg -i22:38
*** deepy is now known as deepy\SIGSEGV22:38
freemangordonnewly launched processes will lod new lib22:38
DocScrutinizer05there's still be the old instance in RAM?22:38
DocScrutinizer05do they?22:38
freemangordondeffinitely22:38
DocScrutinizer05then it will suffice to switch some symlinks ;-)22:39
freemangordonmaybe22:39
freemangordonyou can chroot as well22:39
DocScrutinizer05well, dpkg doesn't do much more22:39
DocScrutinizer05nice22:39
DocScrutinizer05chroot, I like that idea22:39
DocScrutinizer05though setting up a chroot is a biach22:40
freemangordonyeah22:40
freemangordon"/proc /sys /dev ... umm..."22:40
freemangordon"/tmp"22:40
freemangordonwhat else22:41
DocScrutinizer05nah, just "renaming" /lib/glib* will suffice then22:41
freemangordonmaybe22:41
DocScrutinizer05could do that with mount22:41
freemangordonit it will become clear very fast22:41
DocScrutinizer05bindmount the new glib file over the old one, only for this childspin process22:42
freemangordonmount -o bind works over files?22:42
freemangordonlol22:42
DocScrutinizer05supposed to22:42
DocScrutinizer05also supposed to support per-process stuff22:43
DocScrutinizer05man mount22:43
freemangordonnice22:43
freemangordonno, thanks, I don;t have 4 spare hours :D22:43
DocScrutinizer05kudos to javispedro who tought me22:43
freemangordonok, glibc is in cssu-devel22:44
freemangordonmerlin1991: ^^^22:44
DocScrutinizer05bindmount /dev/foo, for mce ;-)22:44
DocScrutinizer05dunno though if messybox ever heard of that22:44
*** arcean has quit IRC22:46
*** arcean has joined #maemo-ssu22:47
DocScrutinizer05or fstab entry is:22:48
DocScrutinizer05                     /olddir /newdir none bind22:48
DocScrutinizer05              After this call the same contents is accessible in two places.  One can also remount a single file (on a single file)22:48
DocScrutinizer05man 8 mount|less "+/The bind mounts."22:50
merlin1991freemangordon: nice!22:50
freemangordonmerlin1991: now what :P22:51
merlin1991i vanish from irc so that I don't have to deal with a new kernel in cssu ;)22:51
freemangordonbut,but...22:51
freemangordonyou're the only one remaining22:51
merlin1991that's the plan ;)22:52
freemangordonthe last man standing :D22:52
DocScrutinizer05the nice thing: the new glibc will work with stock kernel as well22:52
freemangordonyep. exactly in the same way as the old one22:52
merlin1991yep though it still makes sense to patch the friggin kernel to fix the problem22:52
DocScrutinizer05sure22:52
freemangordonit makes no sense to have new libc without patching the friggin kernel22:53
DocScrutinizer05just that we can't ship it as mandatory, best we can get is optional22:53
freemangordonoptional in CSSU?22:53
freemangordonWTF?22:53
DocScrutinizer05yep22:53
freemangordonkernel22:53
merlin1991do we have any assesment on the patches in kernel-power by now?22:53
freemangordonno22:53
merlin1991well22:53
freemangordonbut i'll try to prepare one22:53
merlin1991does the patch apply to vanilla kernel?22:54
freemangordonvanilla?22:54
freemangordonas in stock?22:54
merlin1991aka omap122:54
freemangordonthat's not vanilla :P22:54
DocScrutinizer05I'm still thinking we should support KP and just recommend it, while keeping CSSU in a state where it still works with stock kernel as well, just not taking benefit of kernel patches then22:54
merlin1991DocScrutinizer05: well libc works with stock kernel22:55
DocScrutinizer05for glib that's a pretty easy approach, since glib works with both kernels22:55
freemangordonDocScrutinizer05: since when bug fixes are optional? isn't that the whole point to have CSSU?22:55
merlin1991and we have the patch from pali for the fiasco flasher that gives you a yes/no option for Do I really want a kernel22:55
merlin1991also nod to fmg this is a valid bug in the kernel22:55
freemangordonguys, that makes no sense22:55
DocScrutinizer05I'm fine with that, if it really works22:55
freemangordonhave in mind one of the users of ppoll/pselect is cmt22:56
freemangordoni saw that when i booted maemo with no ppoll/pselect in libc :D22:56
DocScrutinizer05hehe22:56
DocScrutinizer05as long as users can opt out of new kernel, and cssu won't explode into their face with stock kernel, I got no problems with that approach22:57
freemangordonNov  1 21:35:01 Nokia-N900 cellular: ssc[1873]: **** SV=00, HW=RX-51 rev 2204 SW= ****22:58
freemangordonNov  1 21:35:01 Nokia-N900 init: sscd main process (1873) terminated with status 12722:58
freemangordonNov  1 21:35:01 Nokia-N900 init: sscd main process ended, respawning22:58
freemangordonNov  1 21:35:01 Nokia-N900 cellular: ssc[1878]: **** ssc-daemon 0.5.26 sscd NORMAL started ****22:58
freemangordonNov  1 21:35:04 Nokia-N900 init: csd main process (1876) terminated with status 14222:58
freemangordonNov  1 21:35:04 Nokia-N900 init: csd main process ended, respawning22:58
freemangordonthis happens when there is no ppoll/pselect22:59
DocScrutinizer05after all KP is rather thoroughly used/"tested" by a few users. I still think it's not ready for mandatory primetime, but if users have a chance to opt out...22:59
freemangordondraw your conclusions22:59
freemangordonDocScrutinizer05: I still don;t think all of the patches in KP should go in CSSU kernel22:59
freemangordonthough, on the other hand if those are modules there should be no problem23:00
DocScrutinizer05freemangordon: the problem is there's no alternative to having the one and only kernel of your system. You can't use 2 kernels, and you can't add features of whatever homebrew kernel you are using simply to the new mandatory KP23:00
DocScrutinizer05so users need a working way to opt out of whatever we ship for kernel23:01
freemangordonwell, I think opt-out should be given only to those who don't use stock23:02
DocScrutinizer05if there's an opt-out and CSSU will still work as it does today, with stock/whatever kernel... no problem23:02
merlin1991comone doc, give the man a break#23:02
DocScrutinizer05nope, some may use stock kernel for some good reason23:02
merlin1991s/comone/come on/23:02
infobotmerlin1991 meant: come on doc, give the man a break#23:02
freemangordonDocScrutinizer05: they have the option to not upgrade23:03
DocScrutinizer05no, they don't23:03
merlin1991why?23:03
DocScrutinizer05since CSSU is meant to be mandatory23:03
freemangordonit is either mandatory as a whole or not at all23:03
DocScrutinizer05hmmmm23:04
DocScrutinizer05that's exactly the point23:04
freemangordonyou can;t define "this part of CSSU is mandatory, but this is not"23:04
DocScrutinizer05sure we can23:04
DocScrutinizer05that's what optional is for23:04
merlin1991and we also can eat small kittens23:04
freemangordonfixing bugs should not be optional23:04
*** lizardo has quit IRC23:04
merlin1991also in no way cssu is "meant to be mandatory" as long as autobuilder can't use the repo23:04
DocScrutinizer05installing a new kernel though should23:04
freemangordondisagree, if that is what is needed to fix a bug23:05
merlin1991we need some stats on how many users are on plain cssu as opposed to -thumb, I kinda got the feeling that our biggest userbase is on -thumb23:05
DocScrutinizer05kernel is a very sensitive topic to astonishingly many users23:05
merlin1991bteween DocScrutinizer05 we have over 700 users of the -thumb flavour, which basically ships kp23:06
freemangordonDocScrutinizer05: most of the users don't have a clue what kernel is and who put that on their devices23:06
DocScrutinizer05I got the feeling that the whole debate here is about whether we should offer opt-out to users of stock kernel or not - which is kinda infantile23:06
freemangordonif CSSU aims stock system, then we should not offer opt-out to stock kernel users23:07
DocScrutinizer05freemangordon: so they won't have a problem with "install new leete kernel? (recommended!) [y/n]"23:08
freemangordonthey have the option to not upgrade23:08
freemangordonthey will, I told you, imagine there are users who don't have a clue what kernel is23:08
freemangordonthey'll get frightened by that question23:08
freemangordonand will never upgrade23:09
DocScrutinizer05nope, I'm not supporting this, since it deprives users of freedom of choice for no reason other than cssu maintainers defining KP as mandatory23:09
freemangordonthey have the freedom to not upgrade23:09
DocScrutinizer05nope23:09
DocScrutinizer05no good argument23:09
freemangordonor stay with PR 1.323:09
freemangordonyes it is, CSSU is supposed to continue what Nokia did23:10
freemangordonand Nokia shipped new kernel with every PR23:10
DocScrutinizer05"you got the freedom to GTFOAD" is no politically correct statement ever23:10
freemangordonI am developer, not politician23:10
DocScrutinizer05that's why I tell you about politics23:10
freemangordonand "technicaly" this is the correct POV23:10
freemangordonand trust me, I know better than ordinary Joe the answer of "install new leete kernel? (recommended!) [y/n]"23:12
freemangordonyou know it better, too23:12
keriois the answer "yes"?23:12
freemangordonyes23:12
kerioi answered correctly, yay ^_^23:12
freemangordongood, now change your nick23:12
freemangordonto Joe23:12
*** arcean has quit IRC23:13
kerionuh uh, we clearly showed that you know exactly as well as me the answer of "install new leete kernel? (recommended!) [y/n]"23:13
DocScrutinizer05well, there are just 3 points I'll not drop: a) keep CSSU (userland) compatibility to stock kernel, so it doesn't explode when other kernels are used, b) keep as much freedom of choice for users as possible since it's insane to say "you don't want to decide that, so we removed that option for you to increase your comfort", and c) keep CSSU (userland) compatibility to stock kernel, so it doesn't explode when other kernels are used,23:13
freemangordonDocScrutinizer05: my sister, her boyfriend and a friend of mine (girl) have n900 and CSSU. None of them have FC what kernel is and do they have to install it23:14
freemangordonand while they have a person to ask, lots of users don't23:15
kerioDocScrutinizer05: are a) and c) supposed to be the same?23:15
merlin1991freemangordon: what kind of reality disortion field do you posess that you could convince 2 woman to use a n900? :D23:15
DocScrutinizer05yes23:15
DocScrutinizer05kerio: ^23:15
kerioDocScrutinizer05: k23:15
freemangordonmerlin1991: you won;t believe it :D23:15
keriomerlin1991: come on, the n900 is somewhat sexy23:16
freemangordonmerlin1991: actually they are hery,very happy with it23:16
merlin1991if I look at $average woman in my social circle, I could probably sell them a n9, but a n900 would never go through23:16
merlin1991yeah but the initial hurdle to get them use it for a week or so is almost impossible :D23:17
DocScrutinizer05merlin1991: his dark force has to be scary huge, when he is able to convince users to use N900 but they will get scared and confused by install new leete kernel? (recinstall new leete kernel? (recommended!) [y/n]"ommended!) [y/n]"23:17
DocScrutinizer05meh23:17
DocScrutinizer05THAT would even confuse *me*23:17
merlin1991heh we should make that flash question a loop so it comes back whenever you type n :D23:17
freemangordonDocScrutinizer05: your mobile phone is not supposed to ask such questions23:17
DocScrutinizer05uhuh23:17
DocScrutinizer05so what about the standard requester that pops up every time you install a extras package in HAM?23:18
freemangordonusers are used to it, this will be new stuff, read scary stuff23:18
DocScrutinizer05bullshit, sorry23:19
freemangordon(the above question)23:19
DocScrutinizer05installing CSSU might be scary, this question for sure isn't23:19
DocScrutinizer05installing a new kernel without even letting me know thous is FUCKING SCARY23:20
freemangordonDocScrutinizer05: look at the userbase. geeks (at least most of them) are already on -thumb.23:20
DocScrutinizer05I never look at the userbase, since nobody ever seen it23:21
DocScrutinizer05all your assumptions about userbase are handwaving23:21
freemangordonthe remaining (if ever heard of CSSU) will get scared. Thin about CSSU-stable too23:21
freemangordonno, merlin1991 has statistics for -thumb23:21
DocScrutinizer05so what?23:22
DocScrutinizer05does that tell you about their potential to feel scared by a really nice and simple polite question with a suggested way to answer it?23:22
freemangordonI gave you an examle about real people I know23:23
freemangordonthey will get scared23:23
DocScrutinizer05I#ll stop/leave this discussion now, since I already posted a) b) c) and nothing more left to do here, but some breakfast to eat before I go to sleep23:23
freemangordonhehe23:24
freemangordon:D23:24
freemangordonmerlin1991: so, the same question again23:24
freemangordonwhat now?23:24
freemangordon:P23:24
* freemangordon hopes merlin1991 will not leave IRC23:24
merlin1991we'll need that asessment anyway23:24
freemangordonthat's for sure23:25
merlin1991nfc how to properly write the word :P23:25
freemangordonok, lets continue when we have it, ok?23:25
merlin1991that sounds good23:25
freemangordonassessment23:25
merlin1991thanks :)23:25
* freemangordon chacks with google23:25
freemangordon*checks23:26
freemangordonyep,  assessment23:26
merlin1991dict.leo.org is quite good, though it's a german<>english dictionary23:26
DocScrutinizer05on4e last suggestion: finally get rid of this rotten idea of MP, CSSU needs to strive to reduce shit forced down user's throat by MP, not improve on evil concept invented by Nokia for reasons of laziness23:26
freemangordonDocScrutinizer05: that means new HAM23:26
merlin1991we'd have to hook into ham AND apt then23:27
freemangordonor red pill mode enabled by default?23:27
DocScrutinizer05if it does mean new HAM, then we should put new HAM on top of our ToDo list - duh, I thought it's already there since several months23:27
freemangordonDocScrutinizer05: that gains nothing to users23:28
freemangordonunlike bugfixes and such23:28
freemangordonafter all n900 IS NOT DESKTOP23:28
DocScrutinizer05sorry taht's just more random nonsense23:29
DocScrutinizer05I'm a user too, and I recently helped another user how to get back stock camera-ui23:29
freemangordonand what was his/her problem?23:30
DocScrutinizer05and we had plenty of complaints about CBSMS widhget23:30
DocScrutinizer05the probelm was FUCKING23:30
DocScrutinizer05MP23:30
freemangordonas long as we do something there will be complains23:30
freemangordonit is inevitable and you know that pretty well23:31
DocScrutinizer05you're not arguing anymore, you're spreading useless statements23:31
DocScrutinizer05I'm out23:31
keriodon't fight :(23:36
Skryhave to agree with Doc's point b, and what comes to removing options etc, I have just one word, Gnome3.23:41
freemangordonSkry: Gnome3 brings bugs, does not fix them, unlike the matter of our discussion23:43
Skrythough I also understand that to average Joe this kind of stuff is at least quite confusing23:43
Skryfreemangordon: yeah, targeted that one more to Doc but anyways, I guess this is a matter of the amount of average Joes involved . Not saying anything to either direction, I'm sure you all figure out the best solution, or a solution at least :)23:47

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