15:02.51 | *** join/#maemo-ssu infobot (~infobot@rikers.org) |
15:02.51 | *** topic/#maemo-ssu is Maemo Community Seamless Software Update "CSSU" channel, http://wiki.maemo.org/Community_SSU | Known bugs: http://j.mp/communityssu-bugs | Channel logs: http://mg.pov.lt/maemo-ssu-irclog/ | Sources: http://gitorious.org/community-ssu/ | Latest version (testing): 21.2011.38-1Tmaemo1.2 (stable): 21.2011.38-1Smaemo2.1 | meeting 14.03.2012 20:00 UTC, sry about missing the last one |
15:02.52 | *** mode/#maemo-ssu [+v infobot] by ChanServ |
15:23.48 | *** join/#maemo-ssu Sicelo (sicelo@members.bombshellz.net) |
15:23.48 | *** join/#maemo-ssu Sicelo (sicelo@unaffiliated/sicelo) |
15:28.43 | amiconn | DocScrutinizer51: There's an important difference between manufacturer and user optimisation though. |
15:29.21 | amiconn | The manufacturer needs to optimise in a way that it works for *all* devices. The user can optimise for a single device |
15:39.28 | *** join/#maemo-ssu Sicelo (sicelo@unaffiliated/sicelo) |
15:49.21 | Pali | ping merlin1991, freemangordon |
15:49.27 | merlin1991 | ping Pali |
15:49.47 | Pali | merlin1991, I updated my program kernel-version |
15:49.58 | Pali | now it use malloc() + read() |
15:50.04 | Pali | instead mmap |
15:50.41 | Pali | mtd kernel partition is 2MB, so it can be aloocated and readed by one read() call |
15:52.06 | merlin1991 | what does it report in case of installed uboot? |
15:52.25 | Pali | Version string not found |
15:52.48 | Pali | it ignore attached kernels to the end of some image |
15:53.22 | Pali | correctly, program try to find that kernel gz archive in first 2^16 bytes |
15:53.47 | Pali | and attached kernel to u-boot binary is at the end |
15:54.13 | merlin1991 | sure that's ok, but can we detect IF uboot is installed? |
15:54.40 | *** join/#maemo-ssu Sicelo (sicelo@unaffiliated/sicelo) |
15:54.58 | Pali | This program detect if zImage kernel is installed |
15:55.10 | Pali | so if not then some other bootloader is installed... |
15:55.31 | Pali | do we need to specify name of other image (e.g. u-boot)? |
15:55.56 | amiconn | To me that doesn't sound logical. If you only check the first 2^16 bytes, you only need to read those, not the whole 2^21 |
15:55.59 | merlin1991 | My idea of the cssu kernel flasher is a check if it is the right stock kernel or older cssu version --> flash cssu kernel, if not ask |
15:56.33 | merlin1991 | but if possible I'd like the ask part to be able to use an existing uboot in mtd3 and flash with that |
15:57.06 | merlin1991 | something along the lines of: we detected you have uboot installed, do you want to flash the cssu kernel or flash the cssu kernel with your uboot image or not flash at all |
15:57.32 | Pali | merlin1991, flashing u-boot with kernel is not simple... |
15:57.41 | merlin1991 | it's damn easy |
15:57.43 | Pali | you need to build compined u-boot image |
15:57.53 | Pali | and that is different as fiasco... |
15:58.03 | Pali | and we have more u-boot version |
15:58.17 | merlin1991 | you can go from fiasco to u-boot image |
15:58.30 | merlin1991 | and you can read the existing uboot and attach a new kernel |
15:58.38 | Pali | so we cannot prepaire 2 versions (one zImage and one uboot+zimage) |
15:58.58 | Pali | and for flashing: you cannot flash zImage |
15:59.08 | Pali | in mtd3 is some NOLO header (maybe with checksum) |
15:59.13 | merlin1991 | ffs |
15:59.13 | Pali | so you need to flash fiasco image |
15:59.21 | merlin1991 | I thought we have more in mtd3 |
15:59.27 | merlin1991 | damn |
15:59.36 | merlin1991 | ah well then it's a do you want to or not? question :D |
15:59.43 | Pali | and generating fiasco image is possible ony with SDK repo (with fiasco-gen package) |
16:00.48 | *** join/#maemo-ssu Sicelo (sicelo@unaffiliated/sicelo) |
16:01.13 | Pali | so dumping u-boot somehow from mtd3 attach to that new zImage, creating new fiasco image and flashing is hard and possible only in scratchbox with SDK repo |
16:02.00 | Pali | only we can do is ask question: do you want to flash new kernel image? |
16:02.06 | merlin1991 | yeah |
16:02.12 | merlin1991 | well better than flashing without asking |
16:02.44 | Pali | and for cssu kernel I suggest NOT to change version string |
16:03.26 | Pali | we have more kernel packages in extras-devel repo (also joikuspot) which has modules agains default kernel |
16:03.58 | Pali | and if we does not change kernel interface (only bugfix patches) there will be no compatibility problem |
16:05.03 | Pali | merlin1991, I suggest to create $PACKAGE which dpkg-divert /sbin/fiasco-image-update |
16:05.37 | Pali | and use some new wrapper for that fiasco-* binary which will ask if you want to flash kernel if new version is not same (or similar) as old |
16:06.09 | Pali | this will also ask for any kernel updates (not only cssu), so users will see if some package will want to flash kernel |
16:06.24 | merlin1991 | can we do it somehow to only have it with cssu updates? |
16:06.51 | Pali | we can include that $PACKAGE as predepends for cssu kernel |
16:07.03 | Pali | and package can go to cssu repo |
16:09.38 | *** join/#maemo-ssu lardman (~simon@Maemo/community/contributor/lardman) |
16:11.33 | *** join/#maemo-ssu ekze (nyan@2002:5046:e474:1::3) |
16:14.04 | *** join/#maemo-ssu Sicelo (sicelo@unaffiliated/sicelo) |
16:33.22 | freemangordon | merlin1991, the tool to create .po files from .mo file (l10n) is called msgunfmt, please remember that :D. |
16:34.08 | merlin1991 | noting it down :) |
16:34.21 | Pali | stored in IRC log :-) |
16:34.24 | merlin1991 | is it a simple tool as in can anybody do it? |
16:34.25 | freemangordon | going to RE osso-pdfviewerv l10n |
16:34.42 | merlin1991 | because then I can just do it, and leave you the more interesting tasks ;) |
16:34.49 | Pali | freemangordon, if you have time, can you also RE tvout l10n? |
16:34.51 | freemangordon | you have it in every distribution, gettext tool |
16:35.22 | Pali | we have some new untranslated strings in cssu tvout control panel |
16:35.24 | freemangordon | at least it is I have it in scratchbox |
16:35.37 | freemangordon | and in ubuntu too |
16:35.48 | freemangordon | pretty damn standatr tool |
16:36.43 | Pali | merlin1991, did you check if new version of kernel-version.c program get correct version of yours kernels? |
16:36.56 | freemangordon | merlin1991, http://www.gnu.org/savannah-checkouts/gnu/gettext/manual/html_node/msgunfmt-Invocation.html |
16:37.24 | merlin1991 | freemangordon: that looks quite straightforward |
16:37.30 | merlin1991 | Pali: no I didn't find the time yet |
16:37.43 | merlin1991 | Pali: did you update the armel build? |
16:38.55 | Pali | yes |
16:38.58 | freemangordon | merlin1991, if you can do l10n instead of me, I will really appreciate that, I am running out of time now and tomorrow will travel to countryside. On the other hand I almost made a promise on TMO to do it for the next update |
16:39.13 | merlin1991 | I can do it |
16:39.25 | freemangordon | I know you can :p |
16:39.41 | freemangordon | The question is will you :) |
16:39.47 | merlin1991 | okay I WILL do it :D |
16:39.51 | freemangordon | :D |
16:39.55 | freemangordon | great |
16:42.33 | merlin1991 | Pali I can check tomorrow morning, gotta leave now :/ |
16:42.43 | Pali | ok |
16:43.07 | Pali | let me know if it working fine with wifi kp version... |
16:43.22 | freemangordon | Pali, why the hell kernel-power 50 .deb is missing in autobuilder? |
16:43.38 | freemangordon | but all other packages are there |
16:43.40 | Pali | fucking extras and autobuilder |
16:43.53 | freemangordon | man, thats insane |
16:46.02 | freemangordon | Pali, any sign of X-Fade? |
16:46.23 | Pali | I tried ping him today again |
16:46.32 | freemangordon | I’ll do that now |
17:03.12 | *** join/#maemo-ssu M4rtinK (~M4rtinK@98.138.broadband7.iol.cz) |
17:05.37 | *** join/#maemo-ssu NIN101 (~NIN@206.253.166.69) |
17:26.02 | merlin1991 | damn you should have said that 3 hours ago |
17:26.13 | merlin1991 | I had him fix the -testing repo |
17:26.26 | merlin1991 | (there was a bogus entry in Sources.gz) |
17:34.49 | *** join/#maemo-ssu arcean (~arcean@aacx243.neoplus.adsl.tpnet.pl) |
17:44.11 | *** join/#maemo-ssu M4rtinK (~M4rtinK@ip-85-160-58-221.eurotel.cz) |
18:30.19 | *** join/#maemo-ssu M4rtinK (~M4rtinK@ip-85-160-58-221.eurotel.cz) |
18:39.58 | *** join/#maemo-ssu arcean_ (~arcean@aaeq238.neoplus.adsl.tpnet.pl) |
19:23.49 | *** join/#maemo-ssu Sicelo- (sicelo@gateway/shell/globalshellz/x-vvoqzzggvysnyyvf) |
19:28.45 | DocScrutinizer | ok, seems you all decided to go the mad hackers way and by all means force kernel into CSSU |
19:29.14 | Sicelo- | latest cssu comes with kernel? |
19:29.30 | DocScrutinizer | I'm afraid it will |
19:30.08 | DocScrutinizer | no really, I'm fine with that, if anybody could explain to me how that fits into CSSU manifest concept |
19:30.28 | *** join/#maemo-ssu LinuxCode (~LinuxCode@fedora/LinuxCode) |
19:31.38 | DocScrutinizer | CSSU has no dependencies on kernel. KErnel has no dependencies on any CSSU patch. And PK got published via extras repo since ages. 3 times no check, no argument why we want that stuff in CSSU |
19:32.55 | DocScrutinizer | but obviously it's again en vogue to forcefeed another unrelated stuff to users of CSSU |
19:34.11 | NIN101 | forcing kernels? sucks. |
19:40.15 | DocScrutinizer | yeah, and when we get some security patch to kernel, will we force a NIL update of CSSU to those that opted out of CSSU-kernel? |
19:40.52 | DocScrutinizer | or will that kernel patch only ship with next stable? |
19:41.43 | DocScrutinizer | and what we gonna do about the 57 flavours of uBoot-kernels? |
19:42.04 | Pali | DocScrutinizer, see log |
19:42.25 | DocScrutinizer | I had a cursory look at log and threw up |
19:42.47 | Pali | I'm writing program which ask user which has non nokia kernel flashed if he want to flash new version |
19:43.12 | Pali | now it write current version flashed in mtd3 |
19:43.20 | DocScrutinizer | uhuh, and those who got Nokia kernel? those have no choice or what? |
19:43.57 | Pali | we should not change kernel string |
19:44.10 | DocScrutinizer | you should not touch kernel at all |
19:44.18 | Pali | and we include bugfixes, so reason why to flash it |
19:44.47 | DocScrutinizer | I come over anybody touching my kernel without me asking for it, with fire |
19:45.17 | NIN101 | yep. |
19:45.42 | DocScrutinizer | NO MATTER which kernel I decided to use that very minute |
19:47.44 | DocScrutinizer | my daily phone actually *is* running stock kernel, and I damn well know why |
19:49.33 | DocScrutinizer | you can't even know if the system which is running CSSU been actually booted from kernel in mtd3 |
19:50.17 | DocScrutinizer | could have been booted from a kernel via vlasher&ramload, could be booted from mmc via kexec, with kernel God-knows-where |
19:50.33 | DocScrutinizer | s/vlasher/flasher/ |
19:51.13 | DocScrutinizer | maybe some cool haxor patches his NOLO to make it boot from mtd4 |
19:52.02 | Pali | but old version of nokia kernel in mtd3 can be updated to one which will have fixed security bugs |
19:52.09 | DocScrutinizer | honestly keep your fingers from mandatory kernel updates unless you want a shitsorm hitting your inbox |
19:53.27 | DocScrutinizer | who verified that "bugfixed" version, who audited it, who signed it off? |
19:54.20 | DocScrutinizer | who tested it still works with e.g. uBoot, if only for mere limitaions of available free space in mtd3? |
19:57.54 | DocScrutinizer | sorry, but in my book kernel is clearly where CSSU leaves own terrain and goes hamuk abroad - IOW there's a clear red line which CSSU mustn't cross |
19:58.58 | DocScrutinizer | I'm all happy with a CSSU-kernel as a associated but independent project |
20:00.16 | DocScrutinizer | but CSSU as we know it is about userland, not kernel domain |
20:01.28 | DocScrutinizer | even flasher has an option to not flash kernel, or only flash kernel |
20:02.06 | DocScrutinizer | kernel is not userland, and userland is not kernel. full stop. |
20:04.14 | DocScrutinizer | users don't appreciate kernel updates, unless they do it all under their own control, at the time they like, to the kernel they carefully picked or even built themselves |
20:06.56 | DocScrutinizer | and let's look at it from the diametrically other side: are you maintaining two locations for powerkernel then, or will CSSU refer tpo stuff outside CSSU repo, or will all the OC fanbois be forced to switch to CSSU then? If the latter, have fun with bugtrac! :-/ |
20:07.37 | LinuxCode | I agree with the kernel update statement |
20:07.50 | LinuxCode | statement |
20:08.20 | LinuxCode | wouldnt mind a userland tool, that informs me of an update though |
20:08.34 | DocScrutinizer | sure, nothing against such a tool |
20:08.37 | LinuxCode | maybe informing of the security issues/other bug fixes |
20:09.10 | DocScrutinizer | :nod: |
20:10.28 | DocScrutinizer | if it wasn't such unbearable crap regarding usability, that tool already had a name though: HAM |
20:11.40 | Raimu | DocScrutinizer: Your argument on keeping KP outta mainline CSSU is very, very convincing. |
20:12.02 | DocScrutinizer | and if Pali could get a hold of x-fade, to fix the lockup in extras repo, then nobody was *that* eager to get powerkernel into CSSU |
20:35.14 | DocScrutinizer | ( [2012-03-16 16:29:21] <amiconn> The manufacturer needs to optimise in a way that it works for *all* devices. The user can optimise for a single device ) sure, but the problem with user "optimizing" is your device won't differ that much from average due to immutable properties (there are no golden devices that are 3 times as fast as the worst case) - what manuf optimizes for is operation in a wide range of "environmental" conditions, |
20:35.15 | DocScrutinizer | like extreme heat, cold, low battery, fresh battery, different GSM network peculiarities in different countries etc pp. The problem with user trying to compete against that is mainly that the user regularly only has limited knowledge about the fringe cases where the optimization actually needs to grip |
20:38.52 | DocScrutinizer | very usually that user "optimization" is done on a WFM basis, without *any* proper evaluation of even yield on the primary optimization goal (is it *really* faster / more power economic / whatever?) In 99.9% of cases this "optimization" is based on perceived 1337ness of doing such things on your own, plus a placebo effect that frequently comes in when testing stuff like standby times or performance without proper tools |
20:42.28 | *** join/#maemo-ssu NIN101 (~NIN@206.253.166.69) |
20:47.19 | DocScrutinizer | a simple analogy: you might want to 'optimize' an electric motor. You reduce voltage so it consumes less power. All fine, it gets less warm, less wear, consumes less power. But what if the motor is regulated to keep a certain rotations per minute? you still can reduce voltage, but that will cause rotation speed to go down, and the regulation kicks in making the motor draw more electric current to compensate. Result: the consumed energy |
20:47.22 | DocScrutinizer | is probably higher due to motor not operation at optimum conditions, plus the higher current drawn will cause increased wear on the collector & brushes |
20:48.13 | *** join/#maemo-ssu trbs (~trbs@2001:470:d2ad:1:4a5b:39ff:fe7d:1623) |
20:49.49 | DocScrutinizer | IOW: it's hard to do a proper optimization when you have no complete knowledge of how the system works. On a system complex as a SoC only the chip manufacturer has sufficient knowledge |
21:10.17 | *** join/#maemo-ssu Sicelo (sicelo@members.bombshellz.net) |
21:10.17 | *** join/#maemo-ssu Sicelo (sicelo@unaffiliated/sicelo) |
21:10.51 | *** join/#maemo-ssu dafox (~dafox@ip51cc571d.speed.planet.nl) |
21:13.01 | *** join/#maemo-ssu bsdmaniak (~bsdmaniak@std93-20-88-120-139-80.fbx.proxad.net) |
21:47.29 | *** join/#maemo-ssu nox- (noident@p579002A8.dip.t-dialin.net) |
21:47.43 | *** join/#maemo-ssu nox- (noident@freebsd/developer/nox) |
22:18.53 | *** join/#maemo-ssu M4rtinK (~M4rtinK@mail.melf.eu) |