**** BEGIN LOGGING AT Wed Oct 24 03:00:01 2012 Oct 24 03:09:27 Hello Oct 24 03:09:38 (?) Oct 24 03:11:03 hello? Oct 24 03:11:07 Anybody.... Oct 24 03:11:55 ,,, Oct 24 03:11:57 ... Oct 24 03:12:13 :3: Oct 24 03:26:16 narwalrus: kinda the wrong time of the night Oct 24 03:28:11 this channel is mostly filled with european people so you'll get answers starting in 3 to 4 hours :) Oct 24 03:28:22 and I'm really late for bed, so gn :P Oct 24 03:45:46 Hello? Oct 24 07:28:12 vi_: compcache won't make it :(. No matter what I do, compression/decompression eats all of the cpu. Or I am too stupid and can't think of a better algo Oct 24 07:29:01 are you modifying it? Oct 24 07:29:14 ShadowJK: yes Oct 24 07:29:40 new pages go to compressed pool and to a queue to be written Oct 24 07:29:58 once the queue has 256 pages I write it (async) Oct 24 07:30:41 once there is less than one free page in allowed memory, I release 16 oldest pages from compressed pool Oct 24 07:32:12 ShadowJK: cannot understand what I am missing Oct 24 07:32:41 I am thinking to give up om compression, it just eats CPU with nothing gained Oct 24 07:34:23 ShadowJK: any thoughts? Oct 24 07:37:46 freemangordon: ping Oct 24 07:40:11 vi_: pong Oct 24 07:40:55 vi_: sorry, didn't see your queries, I was tweaking compcache for the last couple of days and didn't look at IRC Oct 24 07:41:12 vi_: red the backscroll Oct 24 07:41:14 *read Oct 24 07:41:16 freemangordon: heh. Thats cool. Oct 24 07:41:21 Yeah, I read it. Oct 24 07:41:23 no, it is not Oct 24 07:41:23 lame Oct 24 07:41:24 :( Oct 24 07:42:03 seems that the problem with comcache and n900 is that CPU is too weak, not the amount of RAM Oct 24 07:42:08 or both Oct 24 07:42:42 I thought it worked nicely on n8x0. Oct 24 07:42:50 freemangordon: check your msg. Oct 24 07:44:29 dunno if a lesser compression algo like snappy or lz4 would do anything? Oct 24 07:46:12 jaska: it uses lzo1 aiui Oct 24 07:46:20 which is supposed to be fast Oct 24 07:46:30 what about GPU acceleration? Oct 24 07:46:54 yeah lzo1 is pretty near to lz4 Oct 24 07:47:03 someone must have implemented some kind of compression algo that exploits the features of a GPU. Oct 24 07:47:05 naah, the problem is that compressed pool starts to trash once full Oct 24 07:47:55 and I just can't think of how to avoid that Oct 24 07:48:05 anyway, on my way to work, bbl Oct 24 07:48:39 bb Oct 24 07:49:28 freemangordon; well, what compression is it doing? lzop -1 like? Oct 24 07:51:33 freemangordon; is it actually burning cpu on compression, or does it possibly get stuck up releasing/allocating+compr once write throughput is less than swapout demand? Oct 24 07:52:25 * ShadowJK reads all the way to bottom and spots freemangordon has the same hunch, trashing Oct 24 08:12:02 ShadowJK: yeah, it seems like trashing Oct 24 08:12:09 and yes, it is lzo1 Oct 24 08:12:38 lzo1x_1_compress Oct 24 08:12:51 though I am thinking that my MM sucks Oct 24 08:13:41 the stuff you release from the pool, is it on swap partition too? Oct 24 08:13:58 either there or in the queue Oct 24 08:14:21 waiting to be written Oct 24 08:15:19 though I try to play it smart, i.e. I release those (16) pages from the compressed pool with oldest jiffies Oct 24 08:15:35 ans maybe this is one of the probles Oct 24 08:15:38 I mean so that a read of it doesn't pull it back to compressed pool Oct 24 08:15:50 it does not pull it back Oct 24 08:16:00 it just sits on the flash Oct 24 08:16:06 so that a read just pulls it into swap cache from flash Oct 24 08:16:13 no Oct 24 08:16:28 reading does not put stuff into te compressed pull Oct 24 08:16:32 only writing Oct 24 08:17:05 (the normal swap cache) Oct 24 08:17:14 aah, yes Oct 24 08:17:57 on reading I am doing a normal READ bio if the page is not in compressed pool (or in write queue) Oct 24 08:18:15 so it behaves just like a normal swap Oct 24 08:18:53 but I think I have to get rid of the timestamping Oct 24 08:19:21 as finding the oldest pages takes looots of time Oct 24 08:20:11 and just free lets say 256 pages from the compressed pool once I am out of memory for compressed pages Oct 24 08:20:22 on a circular basis Oct 24 08:20:34 will try that when I am back home Oct 24 08:21:50 though maybe the best I can do is to not compress at all, but just implement what nokia did (for swap) in a better way Oct 24 08:21:52 256 pages sounds pretty huge Oct 24 08:22:03 1MiB Oct 24 08:22:18 this is what I use for eraseblock size Oct 24 08:22:33 though it can be changed of course Oct 24 08:24:05 though when I come to the point to implement background defragmentation, I wull have to lower that a bit :D Oct 24 08:25:37 in the meanwhile: a rpi is sitting on my desk, right in front of me :D:D:D Oct 24 08:32:59 rpi? Oct 24 08:35:22 raspberry pi, probably Oct 24 08:36:41 rpi2, a circle :) Oct 24 08:37:03 ah. Oct 24 08:37:14 freemangordon: I sure hope you got the 512MB version Oct 24 08:37:27 256MB is pretty low these days Oct 24 08:39:56 jon_y: naah, it is 256. Put it fits for the purpose, I am going to use that to make an embedded controller Oct 24 08:40:04 s/Put/But/ Oct 24 08:40:05 freemangordon meant: jon_y: naah, it is 256. But it fits for the purpose, I am going to use that to make an embedded controller Oct 24 08:41:12 freemangordon: control what? Oct 24 08:41:28 weather controller :) Oct 24 08:41:34 the universe Oct 24 08:41:37 well, not exactly controll, there will be a couple of sensors connected to tha Oct 24 08:41:48 *that Oct 24 08:41:55 missile guidance? Oct 24 08:41:59 hehe Oct 24 08:42:19 I'd be really happy with a $50 UAV Oct 24 08:42:25 naah, something like an alarms system Oct 24 08:42:31 ~UAV Oct 24 08:42:42 unmanned ariel vehicle Oct 24 08:42:49 :D Oct 24 08:42:56 that predator drone, but smaller Oct 24 08:43:16 loiter about for 10 minutes would be pretty cool Oct 24 08:43:44 freemangordon: going to stick a 3G modem and send you an SMS too if it finds something wrong? Oct 24 08:44:03 maybe Oct 24 08:44:43 the good news is that I will play with it as a part of the job I do for my company ;) Oct 24 08:44:53 nice Oct 24 08:44:54 Lucky! Oct 24 08:45:17 well, not exactly lucky, as it was me to propose rpi Oct 24 08:45:18 I have a rpi too, but it's sitting in a drawer until I can get some HDMI monitors Oct 24 08:45:27 * jon_y too lazy to setup bootp Oct 24 08:45:49 naah, no monitors, no network, just GPIO and kernel/libc on a bare metal Oct 24 08:46:09 freemangordon: could still be someone else who gets to do it Oct 24 08:46:09 how does the rpi boot up anyway? it's own uboot? Oct 24 08:46:26 deepy: no Oct 24 08:46:36 Alright let me rephrase then Oct 24 08:46:40 You could be working in a callcenter Oct 24 08:46:52 after all I am the principal SW engineer here :D Oct 24 08:47:00 callcenter now run by rpis Oct 24 08:47:13 script reading can fit in 256MB ram :) Oct 24 08:47:36 deepy: seems I missed your point Oct 24 08:48:00 Lucky as in, you get to do fun stuff! Oct 24 08:48:08 hopefully Oct 24 08:48:17 I'm stuck doing un-OOP java Oct 24 08:48:19 aah, yeah, got it Oct 24 08:48:37 jon_y: JAVA? poor you :( Oct 24 08:49:00 as much claim as it has to being OOP, it can be just as much in the opposite way Oct 24 08:49:07 by misusing oop Oct 24 08:49:33 yeah, you need to check the underlying class type to use the interface, that kind of retard Oct 24 08:50:01 I knew a guy who was really into C, he liked to call Object-Oriented Programmers for OOPs Oct 24 08:50:19 java, like C++ is pretty tame, if used properly Oct 24 08:50:26 yeah, I prefer plain C too Oct 24 08:50:35 I wouldn't know, I suck at C. But I can do Python Oct 24 08:50:53 I do perl too, beautiful language if used right Oct 24 08:51:23 sort of the glue code for in-between application linkages Oct 24 08:51:52 wanted to pick up python, but couldn't find a use for it at work Oct 24 08:51:53 I've seen perl, I don't think you can call it beautiful Oct 24 08:51:59 web development! Oct 24 08:52:06 deepy: is it easy to write C extensions for python? Oct 24 08:52:13 err, in a way, yes Oct 24 08:52:25 swig and cython are the best ways I hear Oct 24 08:52:39 java has javah, produce C headers for java classes Oct 24 08:52:46 But I heard it's really easy to screw up the C part and then you'll have segfaulting code Oct 24 08:52:54 perl has xs Oct 24 08:53:01 yep, all of them have the same issue Oct 24 08:53:05 Never really understood if that was a problem with C, Python, Cython or the programmer Oct 24 08:53:18 so, as a C programmer, you have the responsibility Oct 24 08:53:25 all the weight and power of it Oct 24 08:53:52 what does python use to generate stubs? Oct 24 08:54:06 stubs? Oct 24 08:54:11 or are you supposed to use the python headers and write from there? Oct 24 08:55:08 in java, you could tell it that some methods are supposed to come from the C JNI extensions Oct 24 08:55:20 I'm not entirely sure, but I think there might be the possibility that Cython or SWIG comes with a tool Oct 24 08:55:25 so it generates some stub headers, you use those headers to implement your functions Oct 24 08:55:37 I was wondering if python has something like that Oct 24 08:55:48 Cython or CPython? Oct 24 08:55:54 Cython Oct 24 08:56:02 never heard of Cython Oct 24 08:56:26 I think SWIG might go the stubs approach Oct 24 08:56:51 hmm, not sure about swig Oct 24 08:56:58 http://www.swig.org/Doc2.0/SWIGDocumentation.html#SWIG_nn2 Oct 24 08:57:26 http://www.cython.org/ :: Cython C-Extensions for Python Oct 24 08:58:02 I guess it isn't too different from perl xs http://docs.cython.org/src/tutorial/clibraries.html Oct 24 08:58:13 perl xs2h tool Oct 24 09:06:59 I wish I could contribute more to the conversation, but I relaly have no knowledge of C :D Oct 24 09:07:06 or perl, apart from the fact that knowing perl can secure a job Oct 24 09:13:25 I did php too, I won't say it's a a very elegant language Oct 24 09:13:41 maybe it was the crappy code base I had to pick up Oct 24 09:15:51 deepy: if you do pick up C, promise to avoid using an IDE Oct 24 09:16:07 use vi/notepad and the compiler directly Oct 24 09:16:31 helps you learn how the whole code links together into the running system Oct 24 09:17:12 also, look up "Linkers and Loaders" in google, primer to understanding running code in an OS environment Oct 24 09:17:53 avoidin IDE :e? Oct 24 09:18:53 deepy: avoid using a GUI programming like eclipse or MSVC Oct 24 09:19:19 use the tools directly, avoid all the junk obscuring your view Oct 24 09:19:37 or getting locked into a particular IDE Oct 24 11:28:28 http://www.raspberrypi.org/archives/2221 Oct 24 11:29:24 indeed Oct 24 11:29:43 does the pi GPU exist in other devices? Oct 24 14:19:53 jon_y: is this a general advice, or only applicable for *learning* c? Oct 24 14:20:31 DocScrutinizer05: I can't say for other languages Oct 24 14:20:38 jon_y: tbh I even appreciate syntax highlighting for shellscript in e.g. kate or mcedit Oct 24 14:20:45 for C/C++, it's definitely applicable Oct 24 14:21:16 well, point is, get rid of the IDE doing all the compiler config, syntax highlighting is fine Oct 24 14:21:27 jon_y: I'm not asking abiut which language, but if this advice is meant for experienced programmers as well Oct 24 14:21:30 call the compiler yourself, write a makefile or a buildscript Oct 24 14:21:40 get familiar with the tools that you use Oct 24 14:22:05 once you get the underlying layers, feel free to use an IDE if it improves your performance Oct 24 14:22:19 the important part is laerning what the tools actually do Oct 24 14:22:31 I think a proper IDE, esp editor with auto-completion, ctags, compile-error feedback to source etc can increase my productivity by factor 2 .. 10 Oct 24 14:22:32 before treating everything as a blackbox Oct 24 14:22:54 well, when starting, be sure to learn about your tools Oct 24 14:22:54 yep, 100% ACK Oct 24 14:23:52 I find it painful to lurk on programming forums or mailing list where newbies keep asking about setting up their favorite IDEs to workaround some compile error that their getting Oct 24 14:24:08 they don't even understand the error in the first place Oct 24 14:24:17 or how to fix it Oct 24 14:24:33 espeicially linker errors Oct 24 14:25:01 yep Oct 24 14:25:32 I'm even more radical in that regard, suggesting people start to learn about what's programming by coding in assembler Oct 24 14:25:42 things like learning to use readelf, nm, c++filt can save you a lot of troubles Oct 24 14:25:56 yeah, I'm fine with starting with asm :) Oct 24 14:26:15 * jon_y actually started with C and made his way downwards to the hardware Oct 24 14:26:47 once you know how the processor works, you appreciate what the compiler can do for you, and you never will abuse it Oct 24 14:26:47 * jon_y now does java, perl, php, C, C++ Oct 24 14:27:28 yup, getting the basics right can save you from years of unlearning Oct 24 14:27:40 I've seen C++ programs (texteditor) where each single character been a c++ object Oct 24 14:27:50 wat. Oct 24 14:27:51 figuer efficentcy of such concept! :-/ Oct 24 14:28:30 luckily I didn't have much to unlearn Java when somebody recommended I start with Eclipse Oct 24 14:28:47 quickly looked up the tools to understand the class loader error messages Oct 24 14:29:13 it kind of gets mysterious when IDE builds fine, but get an error running Oct 24 14:29:36 especially one as cryptic as telling you X class can't be loaded Oct 24 14:35:06 jon_y: check out Qt3 textedit box for richtext. iirc it needs factor 1000 of RAM for opening any size of plain ascii text. Fun bit: on *closing* the textedit box, this even *doubles* Oct 24 14:35:58 jon_y: I've seen this in twinklephone, on function "show logfile" Oct 24 14:36:34 maybe it's only been factor 100, but anyway - insane Oct 24 14:36:47 double?? Oct 24 14:37:02 crazy stuff Oct 24 14:37:11 and actually closing the display allocated same amount of RAM *again* Oct 24 14:37:36 why did it do something like that? Oct 24 14:37:51 really bad memleak? Oct 24 14:38:01 * DocScrutinizer05 suspects destructors of all objects got called and thus something cracy happened Oct 24 14:38:14 crazy* Oct 24 14:38:31 should be freeing in destructors, not allocating Oct 24 14:39:08 blame Qt3 Oct 24 14:39:54 I try to be very conservative with C++ Oct 24 14:41:34 I argued long time with chief developer Michel deBoer of twinklephone, about replacing that crap by a system("less") call, but he refused to touch that code, argument "less is not posix, thus not available on some of the target platforms" Oct 24 14:42:19 (long story short) Oct 24 14:42:36 how about letting $PAGER do it, if not, just dump out everything to stdout Oct 24 14:43:03 twinklephone is a Qt GUI app Oct 24 14:43:03 maybe I don't want any paging Oct 24 14:43:43 "show logfile" been a auxilary function under extras menu in that GUI Oct 24 14:44:07 noboy usually would use it Oct 24 14:44:12 nobody* Oct 24 14:44:43 its supposed to put it into a textedit box? Oct 24 14:44:50 install twinklephone and test it ;-) Oct 24 14:44:58 yep Oct 24 14:46:34 doesn't QT have a decent textedit box with scrollers attached? Oct 24 14:46:42 maybe qt3 doesn't Oct 24 14:47:43 well, I guess that's exactly the one twinklephone used Oct 24 14:48:42 I've done wxwidgets and SWT, I don't think QT is that stupid Oct 24 14:49:04 I mean just some scroller and textbox shouldn't be doubling memory Oct 24 14:50:46 doubling? the doubling happened on close, on top of an already kicking factor of 50, 100, or 1000 (can't recall anymore - probably it's been too insane to keep details in memory) Oct 24 14:51:26 I.E for opening 100k of logfile text, Qt allocated let's say 5MB of RAM Oct 24 14:51:29 maybe 50 Oct 24 14:51:37 I can't imagine a scrollable text box doing that unless something went crazy wrong Oct 24 14:51:59 I checked it Oct 24 14:52:22 I found it and discussed it and Michel and others confirmed it Oct 24 14:52:35 * jon_y be avoiding qt3 Oct 24 14:52:58 valgrind better be talking Oct 24 14:54:09 twinklephone still available for installation on virtually all common distro repositories Oct 24 14:54:29 will need qt3 libs but so what Oct 24 14:55:11 you're able to reproduce my findings in less than 10 minutes Oct 24 14:58:08 and I honestly don't think that's particularly twinklephone specific problem, I'm quite sure it's genuine Qt3 textedit box problem Oct 24 14:59:23 I sure hope with all that newfangledness in QT4 and C++11 fixes it Oct 24 15:00:38 if they stop implementig each isngle char as a RTF object, I'm sure it will improve ;-P (only handwaving) Oct 24 15:01:19 s/RTF/rich text/ Oct 24 15:01:19 DocScrutinizer05 meant: if they stop implementig each isngle char as a rich text object, I'm sure it will improve ;-P (only handwaving) Oct 24 15:05:47 but honestly, what else could explain the need of way beyond 10bytes of RAM to store one byte (or even word) of char? Oct 24 15:14:26 http://tech.groups.yahoo.com/group/twinklephone/message/998 Oct 24 15:14:53 >> It seems for 1Meg of log text, there may be more than 40Meg(!) of ram eaten up<< Oct 24 15:17:48 >> jr@msupdate:~/.twinkle> #now "delete"ing log view (button) --- r@msupdate:~/.twinkle> #this _increased_ mem-usage by another 10Mb !!! edit-history of QT textedit? 1MB of log := 40MB VSZ. try with 25MB logfile ;-) << Oct 24 15:22:40 I've been incorrect though on my prevous claims: it's not allocating double RAM size on closing, but rather on deleting the log Oct 24 15:34:56 vim. Oct 24 15:34:59 That is all. Oct 24 15:35:17 that is just a tad off-topic though Oct 24 15:35:40 we're not short on proper cmdline tools to display content of a textfile Oct 24 15:36:15 this however is about idiocy in Qt eating 40 bytes of ram per byte of text to display in textedit widget Oct 24 15:41:30 I was responding to your desire for an IDE. Oct 24 15:42:38 aaah Oct 24 16:05:29 DocScrutinizer05: how do i make qt behave? Oct 24 16:05:39 nfc Oct 24 16:06:25 probably depends on your requirements aka definition of 2behave" Oct 24 16:06:31 "behave" Oct 24 16:07:10 * DocScrutinizer05 idly wonders why - since lenovo T500 - the shift key seems to not behave more often than not Oct 24 16:50:43 vi_: ping Oct 24 17:01:21 ShadowJK: freeing compressed memory pool on circular basis fixes the trashing Oct 24 17:07:19 hmm Oct 24 17:47:14 vi__: hi Oct 24 17:47:34 i have some POC ramzswap for you :) Oct 24 17:48:14 here http://dox.bg/files/dw?a=cbd27e2587 Oct 24 17:49:07 use "/sbin/insmod /lib/modules/2.6.28.10-cssu3/ramzswap.ko memlimit_kb=16000 backing_swap=/dev/$your_swap_device" Oct 24 17:49:36 and have in mind that it is not intended to be used more than lets say 10 minutes :D Oct 24 17:50:05 freemangordon: does the backing swap have to be exclusive? Oct 24 17:50:26 freemangordon: also, do you have to enable the swap somehow? Oct 24 17:50:51 vi__: as once you run out of continious (256k in size) blocks on the BSWAP device you are doomed, no defragmentation so far Oct 24 17:51:04 kerio: exclusive? Oct 24 17:51:19 swapon is your friend Oct 24 17:51:36 freemangordon: can (or must) the backing device be enabled for normal swap too? Oct 24 17:51:59 kerio: of course no Oct 24 17:52:00 swapon what? /dev/ramzswap0 or something? Oct 24 17:52:18 swapon /dev/ramzswap0 Oct 24 17:52:25 i see Oct 24 17:52:39 vi__: have in mind that module is for the thumb mernel Oct 24 17:52:39 so it's to be used as the single swap device Oct 24 17:52:43 yes Oct 24 17:52:50 single? Oct 24 17:53:01 no, you can add as much swap devices as you wish Oct 24 17:53:04 i see Oct 24 17:53:15 but it makes no sense Oct 24 17:53:23 do you get the full-ish size of the backing swap partition with your module? Oct 24 17:53:43 yes. minus 1 eraseblock Oct 24 17:53:54 wtf is an eraseblock? Oct 24 17:54:12 the block you write at once Oct 24 17:54:27 i see Oct 24 17:54:34 i.e. 128*4096 in that particular case Oct 24 17:54:40 i bet NAND swap would be really fast :3 Oct 24 17:55:13 unfortunately we don;t have enough free NAND to use it for swap Oct 24 17:55:23 sure we do Oct 24 17:55:35 install fremantle on the eMMC Oct 24 17:55:38 well, it is occupied by rootfs Oct 24 17:56:14 anyway, I am going AFK for some half an hour, bbl Oct 24 18:10:54 * vi__ rading backscroll Oct 24 18:14:41 vi__: qt doesn't follow the keyboard settings :c Oct 24 18:15:22 qt? Oct 24 18:16:23 qt! Oct 24 18:16:40 quite terrifying Oct 24 18:22:41 insmod in t-minus 15s Oct 24 18:25:05 hehe Oct 24 18:29:51 crash! Oct 24 18:30:14 yay! Oct 24 18:32:31 I hope he does not insmod fro the startup scripts Oct 24 18:34:20 for some reason sometimes ramzswap.ko is not copied ok via ssh, such crashes happen sometimes here too, I need to copy the module again :( Oct 24 18:38:29 copy? Oct 24 18:38:54 aah Ssh Oct 24 18:39:07 aka scp Oct 24 18:39:25 well. Oct 24 18:39:31 that didnt work. Oct 24 18:39:44 crashed so hard I had to tug the battery. Oct 24 18:39:52 vi__: donload the file again Oct 24 18:39:58 *download Oct 24 18:40:13 the file is fine. Oct 24 18:40:16 not before doing a move to save away the old one, to compare Oct 24 18:40:27 hmm, that should not happen Oct 24 18:40:27 module inserts fine Oct 24 18:40:31 aah, ok Oct 24 18:40:35 but? Oct 24 18:40:54 err, it crashes? Oct 24 18:41:03 aka PANIC Oct 24 18:41:09 or OOPS Oct 24 18:41:11 or harder Oct 24 18:41:16 seems no, lets wat vi__ to explain Oct 24 18:41:18 I activate ramz with backingswap on sd card with a higher priority than the emmc. Oct 24 18:41:26 ok Oct 24 18:41:41 then it grinds to a halt. Oct 24 18:41:49 how big is the partition? Oct 24 18:41:49 had to pull the battery. Oct 24 18:42:07 freemangordon: ~600mb Oct 24 18:42:14 that's fine Oct 24 18:42:29 vi__: that should not happen (tm) Oct 24 18:42:49 anything in syslog? Oct 24 18:43:21 you should see something like "ramzswap: ramzswap_schedule_write queue[128]" Oct 24 18:43:47 there are some more debug messages Oct 24 18:44:23 I dont have syslog installed. Oct 24 18:44:38 was the device freshly booted? Oct 24 18:44:49 freemangordon: yes. Oct 24 18:44:59 installing remote syslog is a nice idea Oct 24 18:45:22 vi__: did you mess /proc/sys/vm/... settings? Oct 24 18:45:27 *mess with Oct 24 18:45:31 I found that incredibly helpful occasionally Oct 24 18:45:32 DocScrutinizer05: remote syslog? :o Oct 24 18:45:45 log to remoteloghost:513 Oct 24 18:46:09 oic Oct 24 18:47:29 vi__: try the opposite, i.e. swapon $sdcard_swap, swapoff /dev/mmcblk0p3 Oct 24 18:47:43 and use /dev/mmcblk1p3 as backingswap Oct 24 18:47:48 oops Oct 24 18:47:56 mmcblk0p3 :) Oct 24 18:47:59 freemangordon: why would that change things Oct 24 18:48:26 maybe his SD card is too slow Oct 24 18:48:36 or swap partition is badly aligned Oct 24 18:48:39 or... Oct 24 18:50:57 it is a class 4 32gb sandisk. Oct 24 18:51:12 I mess with the vm settings all the time Oct 24 18:54:20 vi__: try to use eMMC swap Oct 24 18:54:36 as backingswap device Oct 24 18:56:49 freemangordon: why? Oct 24 18:58:27 I cannot insmod ramzswap with swap already on sd. Oct 24 18:58:43 it causes an insta-crash Oct 24 18:58:59 it has always been like that, even with the old one. Oct 24 18:59:25 vi__: something is wrong with that module you have, let me reupload it Oct 24 19:04:19 vi__: http://merlin1991.at/~freemangordon/ramzswap/ Oct 24 19:05:25 freemangordon: this is for the cssu kernel only? Oct 24 19:07:54 kerio: yes, I think I already told that Oct 24 19:08:05 vi__: the module is for kernel-cssu3 Oct 24 19:11:30 community ssu installation failed because internet connection was lost. how can I continue it? Oct 24 19:12:28 <_Lena> Hello, can someone help me whith the BME / OTG stuff of the n900 ? Oct 24 19:12:53 with the what now? Oct 24 19:12:57 akls2: run the update from HAM Oct 24 19:13:04 kerio, Oct 24 19:13:06 (Application Manager) Oct 24 19:13:07 kerio, oh! thanks Oct 24 19:13:16 ye-ye, I know :D Oct 24 19:13:28 akls2: what the enabler does is set things up so that HAM can install CSSU properly Oct 24 19:13:39 after you ran the enabler once, you just need to do the system upgrade from HAM Oct 24 19:13:45 kerio, apt-get upgrade is going to work too? Oct 24 19:13:47 _Lena: install h-e-n Oct 24 19:13:51 akls2: use HAM Oct 24 19:13:59 ok Oct 24 19:14:11 <_Lena> yes i did that allready and i had a keyboard working Oct 24 19:14:22 <_Lena> but now it is not charging any more Oct 24 19:14:30 akls2: if you need to ask if you can use apt-get, you can't use it Oct 24 19:14:35 <_Lena> i did a coldflash and no change Oct 24 19:14:53 _Lena: ...woah Oct 24 19:15:32 DocScrutinizer05: *poke* Oct 24 19:16:11 shhhh! Oct 24 19:18:13 _Lena: coldflash is *deprecated* Oct 24 19:19:01 _Lena: coldflash is mostly useless Oct 24 19:19:04 _Lena: ...since virtually nobody has an idea why (s)he would need it, due to knowledge about what coldflash actually does is not really widespread Oct 24 19:19:42 further: h-e-n doesn't impair normal battery charging Oct 24 19:19:46 <_Lena> i Thought it was overwriting the coontnt of the nand Oct 24 19:20:06 _Lena: actually it's not exactly doing that Oct 24 19:20:40 coldflash is needed *only* when NOLO doesn't show up on power-up Oct 24 19:20:47 <_Lena> does it reset the register of the bq* chip Oct 24 19:20:57 NOLO is the bootloader that shows the dim NOKIA screen Oct 24 19:21:02 no Oct 24 19:21:19 nothing is resetting bq* registers Oct 24 19:21:46 <_Lena> ups, i got confused whithe the term "coldflash" i think.... Oct 24 19:21:58