IRC log of #maemo-ssu for Sunday, 2014-04-20

*** _rd has joined #maemo-ssu00:02
*** LauRoman|Laptop has joined #maemo-ssu00:37
*** xes has joined #maemo-ssu00:51
*** LauRoman|Alt has joined #maemo-ssu01:01
*** LauRoman has quit IRC01:05
*** _rd has quit IRC01:27
*** sixwheeledbeast has left #maemo-ssu01:37
*** NIN101 has quit IRC02:00
*** dos1 has quit IRC03:02
*** arcean_ has quit IRC03:20
*** xes has quit IRC03:30
*** Milhouse has quit IRC04:48
*** nox- has quit IRC04:50
*** Milhouse has joined #maemo-ssu05:01
*** silviof1 has joined #maemo-ssu05:01
*** silviof has quit IRC05:03
*** Milhouse has quit IRC05:16
*** amiconn has quit IRC05:22
*** amiconn_ has joined #maemo-ssu05:22
*** amiconn_ is now known as amiconn05:22
*** Milhouse has joined #maemo-ssu05:30
*** LauRoman|Alt has quit IRC05:32
*** okias has quit IRC05:33
*** RedM has quit IRC06:05
*** Milhouse has quit IRC07:14
*** Milhouse has joined #maemo-ssu07:28
*** trx has joined #maemo-ssu07:33
*** trx has joined #maemo-ssu07:33
*** amizraa has quit IRC07:47
*** amizraa has joined #maemo-ssu07:48
*** Maxdamantus has quit IRC08:35
*** Maxdamantus has joined #maemo-ssu08:36
*** discopig has quit IRC09:18
*** discopig has joined #maemo-ssu09:20
*** sixwheeledbeast has joined #maemo-ssu10:02
*** _rd has joined #maemo-ssu10:07
*** _rd has quit IRC11:03
*** _rd has joined #maemo-ssu11:03
*** _rd has quit IRC11:12
*** _rd has joined #maemo-ssu11:49
*** NIN101 has joined #maemo-ssu11:52
freemangordonmerlin1991: ping11:59
freemangordonmerlin1991: could you help me with building glibc?11:59
*** _rd has quit IRC12:00
*** LauRoman has joined #maemo-ssu12:37
*** freemangordon has quit IRC12:55
*** freemangordon has joined #maemo-ssu12:57
*** _rd has joined #maemo-ssu12:59
*** M4rtinK has joined #maemo-ssu13:14
*** silviof1 is now known as silviof13:16
*** dos1 has joined #maemo-ssu13:17
*** arcean has joined #maemo-ssu13:33
*** dos1 has quit IRC13:45
*** dos1 has joined #maemo-ssu13:46
merlin1991freemangordon: what help do you need?14:02
freemangordonwell, i *somehow* built it14:03
freemangordonand it is in the -thumb repo already :)14:03
merlin1991btw freemangordon when you prepare an update, just upload everything but the .changes files and then push them in as last thing, that way you minimize the time the repo is in an inconsistent state to a min or so14:10
freemangordonmerlin1991: I know, but I want to test on the device first what I had put in the repo14:10
merlin1991dpkg -i is a thing ;)14:11
freemangordonsure, but it does not pull the dependencies14:13
freemangordonI wan't to know what apt-get upgrade will give me14:13
freemangordon/home/ivo/workspace/thumb/Tmaemo10/glibc/build-tree/glibc-2.5/configure --host=arm-none-linux-gnueabi --build=arm-none-linux-gnueabi --prefix=/usr --without-cvs --enable-add-ons=libidn,nptl ports  --without-selinux --with-headers=/home/ivo/workspace/thumb/Tmaemo10/glibc/debian/include --enable-kernel=2.6.0 --without-fp --with-tls --with-__thread --disable-profile --disable-force-install14:15
merlin1991that's from where?14:16
freemangordonconfig.log14:16
freemangordonin build-tree/armel-libc14:17
merlin1991that's one layer farther outside than we need14:17
merlin1991actually in the raw buildlog you see what's going on14:17
freemangordonwhere to check then?14:17
merlin1991iirc14:17
freemangordonok, will rebuild it14:17
* merlin1991 starts his scratchbox and goes looking14:18
freemangordonshall I export some ENV var with system headers location?14:19
merlin1991no the with-headers configure option does that for you14:19
freemangordonthe last -isystem I see is -isystem /home/ivo/workspace/thumb/Tmaemo10/glibc/debian/include14:20
freemangordonbut anyway, will follow the instructions in the README and will retry14:21
merlin1991that's not going to fix your toolchain14:21
*** _rd has quit IRC14:21
merlin1991please paste me the output of gcc -v and cpp -v14:21
freemangordon[sbox-FREMANTLE_ARMEL_GCC472: ~/workspace/thumb/Tmaemo10/glibc] > gcc -v14:22
freemangordonUsing built-in specs.14:22
freemangordonReading specs from /scratchbox/compilers/linaro-4.7-2012.07-fremantle-armv7a/gcc.specs14:22
freemangordonrename spec cpp to old_cpp14:22
freemangordonCOLLECT_GCC=/scratchbox/compilers/linaro-4.7-2012.07-fremantle-armv7a/bin/sbox-arm-none-linux-gnueabi-gcc14:22
freemangordonCOLLECT_LTO_WRAPPER=/scratchbox/compilers/linaro-4.7-2012.07-fremantle-armv7a/bin/../libexec/gcc/arm-none-linux-gnueabi/4.7.2/lto-wrapper14:22
freemangordonTarget: arm-none-linux-gnueabi14:22
freemangordonConfigured with: ../../sources/gcc-linaro-4.7-2012.07/configure --host=i686-pc-linux-gnu --build=i686-pc-linux-gnu --target=arm-none-linux-gnueabi --prefix=/usr/src/toolchain_build/output/ --with-ppl=/usr/src/toolchain_build/output/ --with-build-sysroot=/usr/src/toolchain_build/output/arm-none-linux-gnueabi/libc --with-sysroot=/ --with-build-time-tools=/usr/src/toolchain_build/output/bin --enable-threads --disable-libmudflap --disable-libssp --disable-14:22
freemangordonlibgomp --disable-libstdcxx-pch --with-gnu-as --with-gnu-ld --enable-languages=c,c++ --enable-shared --enable-symvers=gnu --enable-__cxa_atexit --disable-nls --with-gmp=/usr/src/toolchain_build/output/ --with-mpfr=/usr/src/toolchain_build/output/ --with-mpc=/usr/src/toolchain_build/output/ --enable-cloog-backend=isl --with-ppl=/usr/src/toolchain_build/output/ --with-cloog=/usr/src/toolchain_build/output/ --with-host-libstdcxx='-lstdc++ -lsupc++ -lm' --14:22
freemangordonwith-arch=armv7-a --with-cpu=cortex-a8 --with-tune=cortex-a8 --with-fpu=neon --with-float=softfp --with-mode=arm14:22
freemangordonThread model: posix14:22
freemangordongcc version 4.7.2 20120701 (prerelease) (Linaro GCC 4.7-2012.07)14:22
merlin1991pastebin would have been nice ;)14:22
freemangordonok :)14:23
merlin1991and for cpp it's really nice because that prints even more14:23
freemangordonhttp://pastebin.com/ma61wNBe14:23
merlin1991and also please pastebin14:23
merlin1991/scratchbox/compilers/linaro-4.7-2012.07-fremantle-armv7a/gcc.spec14:24
DocScrutinizer05 WTF is wrong with nice/renice on maemo?14:24
DocScrutinizer05is it blocked via ssh?14:25
freemangordonmerlin1991: http://pastebin.com/1bbNUA1t14:25
freemangordonI guess I should follow README after all14:25
merlin1991yep14:26
merlin1991your spec file achives the same thing /usr/include is going to be the very first file to look for includes14:26
merlin1991and it contains the original maemo kernel headers -> no pselect14:26
freemangordonyeah :(14:27
freemangordonI was hoping gcc4.7.2 will ignore that14:27
merlin1991but freemangordon don't forget to revert your spec file in the end, some other stuff doesn't build without that assumption (I'm looking at Qt about to cut its throat)14:28
freemangordonanyway, could you remove glibc (and co) from the thumb repo?14:28
freemangordonso I would not need to increase the version14:28
freemangordonmerlin1991: ^^^14:29
merlin1991sure14:29
freemangordonthanks!14:29
merlin1991okay done14:32
freemangordonabout 6-7packages, right?14:32
merlin1991actually I don't know, the advantage of "reprepro -b $PATH removesrc fremantle glibc" is that it removes all binary packages built from the glibc sources :D14:33
freemangordon:D14:34
merlin1991hm and diffs are not enabled on the thumb repo, so can't check that way either14:35
merlin1991should I enable pdiff support for the future?14:35
freemangordonNFC what that is14:35
freemangordonhowever, if there are remnants it will spit on me when I try to reupload the same version14:36
merlin1991hm tbh it makes no sense to run pdiffs when the packages are uploaded one by one, the overhead would be higher than what is gained14:36
*** _rd has joined #maemo-ssu14:37
DocScrutinizer05changing priority works fine via htop F7/F8, but not via nice|renice no matter what I try. and I can't make childspin do anything else except printing dots14:37
DocScrutinizer05NB on stock kernel afaik14:37
merlin1991replace afaik with uname -a ;)14:38
DocScrutinizer05unless childspin is a version that somehow does some bugfixing to ppoll14:38
DocScrutinizer05meh, ok14:38
DocScrutinizer05~ # uname -a14:38
DocScrutinizer05Linux IroN900 2.6.28-omap1 #1 PREEMPT Fri Aug 6 11:50:00 EEST 2010 armv7l unknown14:38
merlin1991hm that should lock up even with the fixed glibc (which is in cssu-testing)14:39
merlin1991essentially childspin should stop printing at some points when it misses a signal14:39
DocScrutinizer05~ # ls -l `which childspin `14:39
DocScrutinizer05-rwxr-xr-x    1 root     staff        6826 Jul 11  2013 /usr/local/bin/childspin14:39
DocScrutinizer05shall I run it some longer?14:40
freemangordonDocScrutinizer05: start HAM14:40
freemangordonand do "refresh"14:40
DocScrutinizer05what ever thou say. Sire14:41
merlin1991yeah a refresh by ham should put enough load on the system to make childspin miss some signals14:41
DocScrutinizer05a "while true; do :; done" in a shell, reniced via htop to -18, brought childspin almost to a halt (like delays of seconds between bursts of points), but nothing else14:44
DocScrutinizer05on -19 chilspin (at 10) stalls completely14:45
merlin1991well the interesting bit is if childspin runs again once you stop your busy loop14:46
freemangordon:nod:14:46
freemangordonDocScrutinizer05: also - start a couple of childspin processes14:47
freemangordonok, with fixed gcc.specs childspin looks waaay better :)14:47
*** _rd has quit IRC14:48
DocScrutinizer05FUCK YOU KDE!14:49
DocScrutinizer05steling focus14:50
DocScrutinizer05on irc highlight14:50
DocScrutinizer05no, childspin stalled for good14:50
freemangordonhmm?14:50
merlin1991freemangordon: I should probably patch the buildscripts for glibc so it overwrites the gcc specs aswell, there is a -specs switch after all14:50
freemangordonmerlin1991: lets push a new -testing first14:51
DocScrutinizer05but it also doesn't print out any diagnostics which iirc it is supposed to do14:51
freemangordonDocScrutinizer05: no, it just stalls14:51
DocScrutinizer05oooh?14:51
freemangordonyep14:51
freemangordonmerlin1991: and pester chem|st to finally put a new -stable out14:52
DocScrutinizer05nah, I'm almost sure it's supposed to print some diags14:52
freemangordonDocScrutinizer05: I am 100% sure it just stalls on missed signal :)14:52
*** _rd has joined #maemo-ssu14:52
DocScrutinizer05mhm14:52
freemangordoncheck the code14:52
DocScrutinizer05can you pls point me to src?14:53
* freemangordon is trying to find it14:53
freemangordonhmm, it seems it is no longer available in inet14:54
freemangordon:(14:54
DocScrutinizer05FUCK!14:54
freemangordonI need to bootup my SDK vmware image14:54
freemangordonDocScrutinizer05: don;t want to waste time on that now, will do it later14:55
DocScrutinizer05http://mg.pov.lt/maemo-ssu-irclog/%23maemo-ssu.2012-08-04.log.html#t2012-08-04T20:44:59  http://mg.pov.lt/maemo-ssu-irclog/%23maemo-ssu.2012-08-04.log.html#t2012-08-04T20:57:49  and yes you're right http://mg.pov.lt/maemo-ssu-irclog/%23maemo-ssu.2012-08-04.log.html#t2012-08-04T21:28:5014:59
freemangordonsee :P15:01
freemangordonok, fixed glibc is in  the thumb repo15:01
DocScrutinizer05now for the FSCKNG nice/renice issue15:10
DocScrutinizer05WTF is wrong with it?15:10
DocScrutinizer05freemangordon: btw it didn't need HAM to make childpsin lock, I think it locked before I started HAM, when I had the "whilce true; do :; done" process running at niceness -18 or -1915:11
DocScrutinizer05merlin1991: loocks like you got the binary on your server. So I guess you also have the sourcecode, right?15:12
DocScrutinizer05I wonder if the source was n some ML or ticket-attachment15:13
DocScrutinizer05freemangordon: btw re FPTF: could you eventually try to build a "complete" fremantle just from sources? according to http://talk.maemo.org/showthread.php?p=1420525#post142052515:16
*** _rd has quit IRC15:17
merlin1991the source was a bugtracker attachment iirc on the ubuntu bt15:29
merlin1991I have the .c file on my pc, but not on this laptop, can share it when I get back home15:30
* merlin1991 goes to install jdk 6 because his university lives way in the past15:33
merlin1991in a time before security mattered15:33
*** LauRoman|Laptop has joined #maemo-ssu15:36
*** trx has quit IRC15:45
*** delphi has joined #maemo-ssu15:45
*** LauRoman|Laptop has quit IRC15:46
*** okias has joined #maemo-ssu15:49
freemangordonDocScrutinizer05: I am not sure where one should start from, I more a developer than a distro maintainer :)16:39
freemangordonanyway, I guess this is not a one man show project16:39
*** arcean_ has joined #maemo-ssu16:50
*** arcean has quit IRC16:50
DocScrutinizer05building a system from scratch isn't THAT hard, gentoo users do it twice a week17:13
DocScrutinizer05I heard17:13
ApicB-)17:13
DocScrutinizer05note that I have NFC about that stuff, but I suspect it's as complicated as "git checkout *; make all"17:15
DocScrutinizer05if it turns out to be significantly more complex than that, nobody is asking you to fix that as a one man show, we need to meet and join forces to come up with concepts how to fix it17:17
DocScrutinizer05(we=developer community)17:17
DocScrutinizer05basically we're actually at the beginning of giving birth to a "new" distro17:18
DocScrutinizer05called maybe freemantle, or community-fremantle-<arch>, where <arch> for now is "n900" and very soon "neo900"17:19
DocScrutinizer05NB that such arch would be much more open to contributions by anybody whol likes to contribute than CSSU is by definition17:21
DocScrutinizer05while I think community-fremantle-n900 should stay app/abi-compatible to stock fremantle / cssu, it has much more liberty to accept alternative packages for any stock core nokia app you can think of, and it has the opportunity to come up with pdecent dependency management from beginning17:23
DocScrutinizer05we wouldn't even need to keep HAM compatibility in repo, since when somebody wants to use HAM s/he's free to use CSSU17:25
freemangordonDocScrutinizer05: so, how do you imagine the stuff should work - a flashable image or internet repo or a combination?17:27
DocScrutinizer05and as long as HAM can get tweaked to work on community-fremantle-*, users can still install maemo-extras apps on that community-fremantle-* system17:27
*** LauRoman|Alt has joined #maemo-ssu17:28
DocScrutinizer05first as simple an approach as possible, maybe a tarball on uSD to get expanded under a working stock fremantle installation17:28
freemangordonchroot?17:28
DocScrutinizer05flashable image sounds awesome17:29
*** LauRoman has quit IRC17:29
DocScrutinizer05no chroot, brute force replace the old system17:29
freemangordondon;t get it17:29
DocScrutinizer05maybe chroot into something like backupmenu to do so17:29
DocScrutinizer05ok, first: backup menu archive to restore via BM17:30
freemangordonwhat about SGX drivers?17:30
DocScrutinizer05we'll see17:30
freemangordonthose are closed after all17:30
DocScrutinizer05does it work without them? do we need to copy the binary from a stock fremantle?17:30
freemangordonbut honestly, I don;t want to give up on acceleration17:30
freemangordonhildon-desktop won;t work afaik17:30
DocScrutinizer05that's exactly what this experiment is all about17:31
DocScrutinizer05find out about what's missing17:31
freemangordonhmm, I see17:31
freemangordonmaybe quemu will do the job17:31
freemangordon*qemu17:31
DocScrutinizer05possibly17:31
freemangordonok, will see what I can do17:31
DocScrutinizer05first approach it would, I guess17:32
DocScrutinizer05when it works under qemu, we can go nuke our beloved N900s ;-)17:32
freemangordonafter issuing a new -thumb and fixing the extras promotion(again)17:32
DocScrutinizer05with community-fremantle17:32
DocScrutinizer05on additional benefit, you're free to build the whole damn thing as thumb release ;-)17:33
DocScrutinizer05also don't forget to refer to http://www.omappedia.com/wiki/Maemo_Getting_Started to ease your burden with that stuff17:34
* merlin1991 opts for upgrade "ALL" the packages to sane stable upstream releases17:34
bencohupstream being ?17:35
merlin1991so the start would be to bootstrap a toolchain and dpkg (perl)17:35
bencohupstream for all "external" projects ?17:35
DocScrutinizer05merlin1991: makes some sense, as long as they have no N900 specific tweaks17:35
merlin1991yes17:35
DocScrutinizer05sudo comes to mind17:35
DocScrutinizer05but there#s prolly 100 times as much that safely could get upgraded17:36
merlin1991I'd go for shotgun testing17:37
merlin1991upgrade everything and then backstep with things that don't work17:37
DocScrutinizer05unlike CSSU, for fptf-fremantle (community-fremantle is too many chars, and ambiguous) the root policy is: when we have a newer *compatible* version somewhere upstream, we probably want to use that17:38
DocScrutinizer05(shotgun) we might discuss that later. Not exactly trivial to decide what makes sense for this procedure17:39
DocScrutinizer05maybe a good approach would be to compare the sources of original Nokia (foss) stuff with what's in linus or whatever public gitorous(?), and if there's a 100% match for any past public version with what Nokia used, we can safely assume the package has no fremantle specific tweaks done by Nokia17:41
DocScrutinizer05so such package probably is safe for upgrade to latest17:42
DocScrutinizer05if OTOH ther +are* Nokia tweaks then odds are those were done to make package work under fremantle, and somebody should investigate what's it all about17:43
DocScrutinizer05makes sense?17:43
freemangordonI suspect we have too much NEON optimizations to simply upgrade to upstream17:45
*** nox- has joined #maemo-ssu17:45
freemangordonwe == maemo517:45
DocScrutinizer05EEEEW WUT?17:45
*** arcean__ has joined #maemo-ssu17:45
DocScrutinizer05never been aware we had *any* neon optimizations17:46
*** arcean_ has quit IRC17:46
DocScrutinizer05where do you expect/know to find those NEON bits? (pkg/binary/lib)17:47
DocScrutinizer05HD? mafw and friends? Qt?17:48
DocScrutinizer05actually what the heck +does* neon? isn't it a vector ALU or sth like that? similar to SIMD or whatsthename?17:48
DocScrutinizer05>>NEON technology is a 128-bit SIMD (Single Instruction, Multiple Data) architecture extension for the ARM Cortex™-A series processors, designed to provide ...<<17:50
DocScrutinizer05so yeah, hard to tell where it gets used, but the otoh I bet on it being wrapped into a lib17:51
DocScrutinizer05then*17:51
DocScrutinizer05when such optimizations didn't make it upstream, the we should ask "why not?"17:54
DocScrutinizer05then*17:54
DocScrutinizer05GRR17:54
DocScrutinizer05I mean, even servers come with Cortex-A now, I heard17:55
DocScrutinizer05so also with Neon, no?17:55
DocScrutinizer05it's not like Neon was an extremely exotic niche case to support or rather ignore upstream17:56
DocScrutinizer05I could understand when upstream isn't interested in H-E-N/hostmode kernel patches ;-). But Neon...17:57
DocScrutinizer05but yeah, those patches might have been done to Nokia fremantle branch first, and when they got merged upstream there already been quite a number of other changes as well already. So those packages where Nokia did the optimization might fail the diff-to-mainline test, despite they are compatible to later upstream versions that got neon support merged from fremantle to upstream18:01
DocScrutinizer05maybe we could check the merge where the Nokia patches got into upstream, to see if the merge been complete or somebody messed with it so the results in upstream are not compatible to what they were taken from in fremantle18:03
*** arcean__ has quit IRC18:10
*** arcean has joined #maemo-ssu18:10
*** MohammadAG has quit IRC18:15
*** MohammadAG has joined #maemo-ssu18:16
*** _rd has joined #maemo-ssu18:48
*** BCMM has joined #maemo-ssu18:58
*** LauRoman|Alt has quit IRC19:13
*** arcean_ has joined #maemo-ssu20:03
*** arcean has quit IRC20:06
*** arcean__ has joined #maemo-ssu20:11
*** arcean__ is now known as arcean20:11
*** arcean_ has quit IRC20:13
*** arcean_ has joined #maemo-ssu20:50
*** arcean has quit IRC20:52
*** _rd has quit IRC20:54
*** _rd has joined #maemo-ssu21:36
*** RedW has joined #maemo-ssu21:55
*** arcean_ is now known as arcean22:23
*** dhbiker has quit IRC22:35
*** dhbiker has joined #maemo-ssu22:36
*** _rd has quit IRC22:37
*** LauRoman has joined #maemo-ssu22:45
*** _rd has joined #maemo-ssu22:52
*** BCMM has quit IRC23:00
*** _rd has quit IRC23:57

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