IRC log of #maemo-ssu for Friday, 2013-09-06

*** Martix has joined #maemo-ssu00:00
*** FlameReaper has quit IRC00:02
*** NIN101 has quit IRC00:06
*** Martix has quit IRC00:12
*** M4rtinK has joined #maemo-ssu00:28
*** dhbiker has joined #maemo-ssu00:46
*** discopig has joined #maemo-ssu00:50
*** discopig has quit IRC00:54
*** _rd has quit IRC00:58
*** Martix has joined #maemo-ssu01:00
*** sunny_s has quit IRC01:02
*** Martix has quit IRC01:04
*** jon_y_ has joined #maemo-ssu01:06
*** jon_y has quit IRC01:08
*** jon_y_ is now known as jon_y01:08
*** DrCode has quit IRC01:10
*** DrCode has joined #maemo-ssu01:14
*** BCMM has quit IRC01:16
*** dos1 has quit IRC01:27
*** wmarone has quit IRC01:30
*** discopig has joined #maemo-ssu01:52
DocScrutinizer05TU, PAH!02:02
DocScrutinizer05ETH Zurich maybe. Emphasis on "maybe"02:03
*** Pali has quit IRC02:05
*** M4rtinK has quit IRC02:07
*** dos1 has joined #maemo-ssu02:14
*** dos1 has quit IRC02:21
*** LauRoman has quit IRC03:12
*** DrCode has quit IRC03:37
*** DrCode has joined #maemo-ssu03:42
*** dafox has quit IRC04:03
*** nox- has quit IRC04:51
*** amiconn has quit IRC05:14
*** amiconn_ has joined #maemo-ssu05:14
*** amiconn_ is now known as amiconn05:14
*** DocScrutinizer05 has quit IRC06:02
*** DocScrutinizer05 has joined #maemo-ssu06:04
*** oldtopman has quit IRC06:16
*** Vlad_on_the_road has joined #maemo-ssu09:06
*** Vlad_on_the_road has quit IRC09:12
*** LauRoman has joined #maemo-ssu09:30
*** DrCode has quit IRC09:33
*** Pali has joined #maemo-ssu09:35
*** DrCode has joined #maemo-ssu09:44
FatPhilmerlin1991: good point. there are 2 lazy aspects - firstly just picking up on a heavily marketted trend language, and secondly just continuing with that bad choice.10:23
freemangordonFatPhil: in the kernel git trees you have around - what is the situation with powervr driver?10:25
FatPhilComputer languages should be tought the same way religions should be tought. A wide range of them, and with an emphasis on their similarities and differences. Learning *a* computer language is mostly useless, you end up with drones with no *understanding*10:25
FatPhilfreemangordon: I've checked my rover tree, and it's older than the latest released tree that KP is based on (no surprise, maintenance was outsourced later)10:26
freemangordonooh :(10:26
FatPhilNot by much, only a few patches worth, but it still came as a bit of a disappointment10:28
freemangordonFatPhil: powervr driver is the one we have 2 options for (in the context of the newer kernel) - forward port 1.3 or use the one from Nemo10:28
FatPhilI don't know Nemo10:28
freemangordonI already did the former, it seems to work. But it is dated and userland pieces are slow as hell10:29
FatPhilI do know that IMG's patch drops were the typical *SHIT* that you get from h/w verndors, and therefore I don't trust the code further than I can throw it.10:29
freemangordonFatPhil: what?!? you don;t know what nemo is?10:29
freemangordonFatPhil: https://wiki.merproject.org/wiki/Nemo10:30
freemangordonthe OS is unusable on N900 IMO, but there was a lot of HW adaptation done we can use10:30
FatPhilDoes Nemo come with a detailed commit history for the driver, or is it just a blob or a megapatch?10:34
freemangordonjust a blob AND a megapatch afaik10:34
freemangordonthe problem with powervr is that it depends on stuff in DSS driver10:35
freemangordonso if we upgrade powervr driver, DSS have to be upgraded as well10:35
keriodss?10:36
freemangordonwhich brings tons of problems, like X needs to be upgraded ;)10:36
freemangordonkerio: framebuffer etc10:36
freemangordonDisplay SubSystem iirc10:37
FatPhiland if you upgrade X, what else breaks? How big's the dependency tree?10:38
freemangordondammit, maybe we ca just hack clutter to use bc-cat instead of slow as hell texture upload and call it a day10:39
freemangordonFatPhil: NFC, never tried to upgrade X10:39
FatPhilI have a script that can split megapatches into whitespace-only changes, and actual real changes. That can sometimes help trying to work out what's changed.10:40
freemangordonFatPhil: I think our major problem is not dependency tree, but the patches(hacks) Nokia did10:40
FatPhilfor the powervr driver, most of the code was straight from IMG, we reported issues to them, and they'd drop a fixed version some time later10:41
freemangordonwell, X upgrade is not something I will start to do on my own10:41
FatPhilmaybe best to just stick with the old one, and only migrate forwards when necessary.10:42
freemangordonFatPhil: this is what I thought initially, but it seems migration on parts is not possible. Or I didn't get it as well :)10:43
freemangordonmaybe we can ask Skry for help, he is the one to bring ALARM on N90010:44
freemangordonI wonder whether he will be interested in the project, though I doubt given how much he like doc's attitude :D10:45
* freemangordon is out of cigarettes, going to buy some10:45
*** Sohil876 has joined #maemo-ssu10:57
*** Sohil876 has quit IRC11:04
*** DocScrutinizer05 has quit IRC11:13
*** DocScrutinizer05 has joined #maemo-ssu11:13
freemangordonPali: zimage-get-version /dev/mtd3ro returns "Version string not found"11:35
freemangordonPali: zimage-get-version -f /dev/mtd3ro returns "2.6.28.10-cssu3"11:36
freemangordonwhy -f is needed?11:36
Palido you have uboot?11:37
freemangordonPali: the real problem is that u-boot installer still says that the current version is unknown11:37
freemangordonno, I was about to install it11:37
Paliwithout -f it should write kernel version if there is no uboot11:37
freemangordonbut it does not do it11:38
freemangordonkernel-cssu3 here11:38
Paliuboot version is detected in fiasco-image-update11:38
Palifreemangordon, also try without -f again11:38
freemangordonI tried a couple of times11:38
Paliok11:38
Paliif without -f it does not write version, and you gave cssu3 kernel without uboot, then there is problem11:39
Paliplease create mtd3 dump and send me it11:39
freemangordonPali: what I see is that without -f it searches in the first 64KB only11:39
Paliyes11:39
freemangordonPali: with dd?11:39
Palibecause in 64Kb should be version header11:40
Paliuse cat11:40
freemangordonok11:40
Paliwith -f it looking in full image, so it will find also kernel if there is attached after uboot11:40
freemangordonI don;t have u-boot installed11:41
freemangordonthis is my devel device11:41
freemangordonI was about to install u-boot to test 3.10 :D11:41
Palineed to look at your mtd311:41
freemangordonok, sending it11:41
Palimaybe 64kB is not enough...11:42
freemangordonyep11:42
freemangordonPali: merlin1991.at/~freemangordon/u-boot/mtd3ro.zip11:44
freemangordonPali: why don't you just use 2MB always?11:46
Palibecause it detect also kernel if there is uboot11:46
freemangordonI meant enable -f as default11:47
freemangordonuh, oh, I have an idea11:48
freemangordonPali: what about do cat of first 2 megs in case of mtd device to an ordinary file and doing mmap on it?11:48
freemangordonthat way there'll be no failed allocation problems aiui11:49
Palicat also fail11:49
Paliproblem is that kernel does not have memory for allocating buffer when readin from nand11:49
freemangordonwhat? why?11:50
freemangordonoh, I see11:50
freemangordonshit :(11:50
Palimaybe cat also repeating if read failing11:50
freemangordonwhat about dd?11:50
Palibut now it should work, because app waiting and handling it11:50
freemangordonok11:50
Palidd with small bs maybe will work11:50
Paliif kernel is not tryting to read more into buffers11:51
freemangordonif you do a page at a time, it will always succeed11:51
freemangordoni.e. dd -bs 409611:51
*** dhbiker has quit IRC11:52
freemangordondd bs=4096 count=1 skip=N >> dumpfile11:53
freemangordonPali: you can just put that in a script wrapping the executable, so no code change will be needed11:54
freemangordonoh, you don;t mmap right now11:54
Palimmap on mtd not working before 2.6.2911:55
freemangordonYes, that is why you can copy mtd to an ordinary file and mmap it11:55
freemangordondd it page by page to a temp file, mmap the temp file and you're fine :)11:56
freemangordonaiui11:56
*** aap has quit IRC11:56
*** dhbiker has joined #maemo-ssu12:07
freemangordonPali: what I need to do to boot maemo from uSD. Sorry, I forgot :(12:12
Palifreemangordon: now I found where is problem12:13
freemangordongood :)12:13
Paliwe should check if gzip signature is in first 64kB12:13
Paliand try to unpack full image12:13
Palinot only first 64kB :-)12:13
Palibecause it can be 0kB for unpacking12:14
freemangordonPali: please help, what I need to do to boot maemo from uSD12:15
Paliok12:15
Palihm, from uSD?12:15
freemangordonyep ;)12:15
PaliI have only working booting from eMMC12:15
Palibut maybe it will work also from uSD12:15
freemangordonwell, I did that back in the days I was playing with 3.512:16
freemangordonit works, but I can;t remember what partitions need to be on uSD12:16
freemangordonesp which one is mounted as /12:16
PaliI booted normally from ubifs rootfs12:17
Palijust created uboot bootmenu.d config file12:17
freemangordonI want to keep maemo on nand / eMMC intact, so I can use it to fix stuff12:17
Paliok12:18
Palilook at my scripts at: https://code.launchpad.net/~pali/+junk/maemo_M32GB12:18
Palithis copying and configuring maemo at new root /home/maemo/12:18
freemangordonhmm, can I still boot stock maemo?12:18
Palitry to change scripts to use /media/mmc<something>/ instead /home/maemo12:18
Palifreemangordon: you need cssu-devel12:19
Paliand it should work to boot both 2.6.28 and 3.10 kernels12:19
freemangordonPali: if I can boot to stock maemo after your scrip runs, I am just fine if the "new" maemo is in /home/maemo12:19
Paliyes, these scripts only copying maemo to /home/maemo and modifing only files in /home/maemo/12:20
Paliso everything in rootfs ubifs is not touched12:20
freemangordonPali: the point is that if I screw something in the init scripts(which I will, for sure) to have a tool to boot the device and fix it12:20
freemangordonok12:20
Paliyou need bootmenu for choosing between ubifs and emmc12:20
freemangordonis that in repos?12:21
Palibootmenu package is in repos12:21
Paliyou need to install it12:21
freemangordonstable?12:21
Paliyes12:21
freemangordonok, I will12:21
freemangordonwhy cssu-devel is needed?12:21
Palithere is only one problem, if you never had enabled R&D mode12:21
freemangordonit is enabled12:22
Pali(when R&D string is not in CAL)12:22
Palithis will be fixed in next version12:22
freemangordonthis is my devel device after all :D12:22
Palicssu-devel is needed because of maemo specific kernel interfaces12:22
freemangordonwhich package?12:22
Paliin cssu-devel are patched OSS applications to not fail if /proc/<something> not exists12:22
Palido not remember all packages...12:23
freemangordonhmm, I think this is already in -testing12:23
Palinot all12:23
Palimerlin did not included all12:23
freemangordonwell, I guess I can chroot to /home/maemo and apt-get upgrade from there12:23
freemangordonor not12:24
Paliand there is problem with mce (because of broken kernel drivers), this is fixed in 3.11 (not 3.10)12:24
Paliyes, you can12:24
Palialso you can boot stock kernel with /home/maemo12:24
Paliand upgrade it here12:24
Paliand no chroot with correct /home/user or /opt is needed12:24
freemangordonok, I have u-boot and bootmenu12:24
*** aap has joined #maemo-ssu12:25
freemangordonwhat else I need? wget your scripts and execute them?12:25
Palibzr (or maybe wget will work too)12:25
Palifirst run boot.sh12:26
Paliand then run init.sh12:26
freemangordonITEM_LINUXRC="maemo5/boot.sh" ?12:26
Palibut first check which version of upstart do you have12:26
Palifreemangordon: yes12:26
freemangordon0.3.8-68+0cssu612:26
freemangordonupstart ^^^12:27
Paliif you have new upstart which working correctly with ad not using genfstab, remove all genfstab from init.sh12:27
Palifreemangordon: not run boot.sh but M32GBboot.sh12:27
Paliboot.sh is really only init script for bootmenu12:28
freemangordonok12:28
freemangordonwhere I should put it?12:28
freemangordonboot.sh that is12:28
Palilook at file M32GBboot.sh12:28
freemangordonok12:28
Paliit copying to right place12:28
Palifreemangordon: remove also changes to /home/maemo5/etc/event.d/rcS-late12:29
Palifrom M32GBinit.sh12:29
freemangordonwhat changes?12:29
freemangordonok12:29
Palibecause new rcS-late does not need these hacks12:29
Paliand can fail with these old hacks12:29
Palireally read what these scripts doing...12:30
freemangordonok12:30
freemangordonPali: hmm, gcc4.7 compiles 3.10 just fine :)12:36
Palifreemangordon: yes gcc 4.7 is required for compiling12:37
freemangordonoh :D12:37
PaliI'm using arm-linux-gnueabi-gcc from ubuntu12:37
freemangordonI didn't try 4.2.112:37
Pali4.7 is needed for .arch sec extenstion for smc calls12:38
Paligcc from scratchbox not working (if failing somewhere in kernel header where is check for old gcc)12:39
freemangordonok, 3.10 built ok, modules installed12:40
Palifreemangordon: you must install all modules into one /lib/modules/<version>/ dir12:43
Paliwithout subfolder12:43
freemangordonsure12:43
Paliand run depmod for correct version12:44
freemangordonmake modules_install does that for me in SB12:44
Palibecause linux kernel makefile install modules into lot of subfolders...12:44
*** dafox has joined #maemo-ssu12:44
Palicp `find . -name "*.ko" ` /destdir/12:44
Paliok, now patch for zimage-get-version is there12:45
freemangordonhmm, no, why? make INSTALL_MOD_PATH=/blabla modules_install12:45
freemangordonthe bouns being .kos are being stripped12:46
freemangordon*bonus12:46
freemangordonPali: give me the binary/deb, so I can test it12:46
freemangordonor source, doesn;t matter12:46
Paliyou need to move all modules from /blabla/subdir/subdir into /blabla...12:47
freemangordonPali: what, why?12:47
Palibecause maemo doing: insmode /lib/modules/<version>/module_name.ko12:48
Pali*insmod12:48
freemangordoninsmod? not modprobe?12:48
Paliyes12:48
freemangordonoh, ok12:48
Paliyou must have all modules in /lib/modules/<version>/12:48
freemangordonok. got it12:48
Palifreemangordon: sources are here: https://code.launchpad.net/~pali/+junk/maemo_fiasco-image-update-ask12:49
Paliyou only need to compile zimage-get-version.c12:50
freemangordonhmm, no more kernel-cssu3 on the device :(12:50
Pali$ gcc zimage-get-version.c -O2 -lz -o zimage-get-version12:50
Paliyou can test it with any kernel :-)12:50
freemangordonwill have to reinstall it first, after your script finishes copying12:50
freemangordonPali: yes, but it was failing with cssu312:51
PaliI tested it with your mtd dump and it worked12:51
freemangordonwell, ok then, no need to test it :)12:51
jon_yit needs new binutils too right?12:53
jon_yoops12:53
jon_ywrong conversation12:53
*** mauroc has joined #maemo-ssu12:53
maurocHello everyone12:54
Palihi12:54
maurocHi Pali12:54
maurocNice to meet you12:54
freemangordonhi12:55
maurochello12:56
freemangordonPali: wait, why modules are not stripped?12:57
Paliwhere?12:57
Paliyou can strip modules without problem12:57
freemangordonin modules install_dir12:57
freemangordonyes, but make doesn't do it12:58
PaliI know that kernel-power doing manual strip and manual cp12:58
Pali$ make modules_install INSTALL_MOD_PATH=/target/$VERSION-modules/ INSTALL_MOD_STRIP=112:58
freemangordonyeah :)12:59
freemangordonthanks12:59
Palifreemangordon ^^^ this is part of my instal script12:59
Pali$ scp `find $VER-modules -name "*.ko"` root@$IP:/lib/modules/$STR/13:00
Paliand uboot item file needs only:13:01
PaliITEM_NAME="Maemo 5 with kernel $VER (Internal Nand)"13:01
PaliITEM_KERNEL="$VER"13:01
Palino atagg omap table, this was already removed from upstream kernel13:01
Paliso /proc/bootreason is fake static file13:01
freemangordonok13:01
freemangordonhmm, copying rootfs takes ages13:02
maurocexcuse me..I got a broken filesystem after the last cssu update I guess13:06
maurocit was a busybox update if I remember correctly13:06
freemangordonPali: hmm, nothing appeared in bootmenu menu13:08
Palifreemangordon, in maemo bootmenu?13:08
freemangordonyep13:08
Palidid you run also M32boot script?13:08
freemangordonshall I copy the .item by hand?13:08
freemangordonyes, I do13:08
freemangordonoh, no :D13:09
freemangordonhmm, "./M32GBboot.sh: line 9: [ 0: not found"13:10
freemangordonYAY, booted :D13:11
freemangordonwith 2.6.28 that is13:11
Pali[12:10:11] <freemangordon> hmm, "./M32GBboot.sh: line 9: [ 0: not found"13:12
Palithere is whitespace13:12
Pali6966 205b c2a0 "if [.."13:12
Paligoing to fix it13:13
freemangordonPali: going to have something to lunch, bbl13:14
Palifreemangordon: script M32GBboot.sh fixed on launchpad13:15
Paliok13:15
keriommh, i think that installing gnome on this netbook is a mistake13:20
*** LauRoman has quit IRC13:23
Palikerio: what about LXDE?13:25
kerioi just went with the default options13:25
keriodebian-installer apparently doesn't ask which DE you want :c13:25
*** sunny_s has joined #maemo-ssu13:26
kerioanyway, this netbook's hd has 1800 bad blocks apparently13:28
kerioso it's going to be fun13:28
*** mauroc has left #maemo-ssu13:34
FatPhilThe device I just put CSSU on does seem to have a whole bunch of issues, I have no idea if that's because of CSSU or because of the million things previously installed on it. SO I think a complete reflash and install from scratch would be best.13:36
FatPhilSpotted what I think was some flashlight app spawning in a tight loop this morning (so quickly I couldn't identify who was spawning the processes)13:37
FatPhilpowertop was listing ohmd constantly using 2% CPU too13:39
FatPhilI suspect that was because it was constantly being informed about new processes, or soemthing.13:39
keriooh god gnome shell is horrible13:41
jon_ykerio: v3?13:41
kerioyeah13:42
jon_yhmm, I tried it in Debian testing, it was alright there13:42
FatPhilHmmm, looks like it needs blue rectangles around all those icons: http://www.gnome.org/gnome-3/13:42
jon_ythen again, I had xwwm at work13:43
jon_ynobody knows better at work13:43
kerioit's just so fucking slow13:43
jon_ydoes it have 3D effects etc?13:44
keriothis is a netbook with a crappy cpu and no graphical effect13:44
kerioer, wtf did i write13:44
kerioand no graphics card13:44
jon_yI used it in a VM13:44
jon_yit was strangely OK13:44
jon_ywasn't laggy or anything13:44
jon_yjust a basic black/grey/white theme13:44
jon_yFatPhil: Debian gnome-3 did not look like that13:45
FatPhillooks like a kiosk, not a computer.13:45
jon_yif anything, it looked like a slightly polished gnome-213:46
jon_yFatPhil: it screams OSX if the dock thing is moved to the bottom13:46
FatPhilthis looks like a computer: http://dwm.suckless.org/screenshots/dwm-20120806.png13:46
jon_yFatPhil: my work VNC looks like that13:46
jon_ythough it is annoying that window focus follows the mouse13:47
jon_yFatPhil: looks like everyone wants tile today13:48
jon_yWin3.11 all over again :)13:48
jon_ytiled icons is now new again13:49
*** M4rtinK has joined #maemo-ssu14:08
*** arcean has joined #maemo-ssu14:21
*** sunny_s_ has joined #maemo-ssu14:26
*** sunny_s has quit IRC14:27
*** Martix has joined #maemo-ssu14:37
FatPhilTiled icons suffer from the flaw of not obeying the mantra "there are only 3 numbers in programming - 0, 1, and an arbitrary number"14:38
*** Martix has quit IRC14:40
freemangordonFatPhil: hmm, this is weird (issues on CSSU)14:41
keriook, gnome shell is actually not *that* bad14:43
kerioit's just too heavy for this netbook i think14:43
FatPhilfreemangordon: If I wanted to be able to get back any of the cool toys that were installed, I just need to backup /etc/apt/sources* and the list from dpkg -l ? (not gonna do a full backup, I have no interest of recreating this system exactly)14:45
*** kolp has joined #maemo-ssu14:46
*** kolp has quit IRC14:47
freemangordonPali: something is broken, the instant I press a key in bootmenu, the screen goes off14:50
Palifreemangordon: try to compile bootmenu from git: https://garage.maemo.org/projects/bootmenu/14:51
Paliand try to use this version14:52
freemangordonok14:52
Palior is bootmenu working?14:52
freemangordonyep, it does that only with 3.1014:52
Palitry version from git14:53
Palifreemangordon, right support for 3.x kernels is only in bootmenu git14:57
kerioi think i'm going to reinstall this14:58
kerioand make it a minimal install14:58
keriois xfce still good? i used to prefer it to lxde14:59
*** dhbiker has quit IRC15:04
freemangordonPali: "warning: remote HEAD refers to nonexistent ref, unable to checkout."15:09
freemangordonWTF?15:09
Palihm? garage git is broken?15:10
freemangordonbootmenu that is15:10
freemangordonyep15:10
Palifatal: https://vcs.maemo.org/git/bootmenu/info/refs not found: did you run git update-server-info on the server?15:10
freemangordonI run nothing on the server15:10
Palithis is what git clone show me now15:11
Paliit looks like git repo on server does not have configured post receive hook15:11
Pali$ git clone ssh://drop.maemo.org/git/bootmenu15:11
Palithis working ^^^15:11
freemangordonfor you, but not for me15:12
freemangordonPermission denied (publickey). ;)15:12
freemangordonok, I'll just get a snapshot15:13
PaliI will try something15:14
freemangordonshit, snapshot doesn't eork too15:14
freemangordon*work15:14
Palifreemangordon: it is really broken15:15
Pali'git update-server-info' must be called on server after each push15:16
Palivia git hook15:16
freemangordonPali: ok, if you say so :)15:16
freemangordonI am not much into the serverside of git15:16
*** arcean has quit IRC15:18
*** dhbiker has joined #maemo-ssu15:18
FatPhilgitweb there seems very b0rked15:19
*** arcean has joined #maemo-ssu15:25
*** mauroc has joined #maemo-ssu15:32
Paliinfo: problem with bootmenu garage repository solved15:34
Palifreemangordon: now I see that you need version from git where is fixed /dev/input/ device for keyboard and power button to work also with upstream kernel15:36
Palibefore bootmenu had hardcoded /dev/input/ paths for maemo kernel and upstrema kernel has switched order15:36
Palinow bootmenu looking for keypad & pwrbutton description, so it set correct /dev/input/ names15:37
freemangordonPali: "dh_clean: Sorry, but 5 is the highest compatibility level supported by this debhelper."15:38
freemangordonWTF?15:38
Palibroken scratchbox15:39
freemangordonno15:39
freemangordonit is fine, debian/compat is 615:39
Palifreemangordon: first run: ./configure n90015:39
freemangordoni simlinked by hand15:40
freemangordonhowever after I run ./configure n900 debian/compat still contains 615:40
Paliwhat you symlinked?15:40
Paliyou need to run configure15:40
freemangordonstuff in /debian15:40
Palibecause it doing something more15:40
freemangordonI did it15:40
Palinot only symlinks15:40
Paliok15:40
Palitry to edit debian/compat to 515:41
freemangordonsure15:41
freemangordon:D15:41
freemangordonthe point is you can't push that into autobuilder15:41
PaliI'm using madde for compiling and autobuilder has updated scratchbox to support debhelper 715:41
PaliI can, autobuilder has support for debhelper715:41
Paliit working fine :-)15:42
freemangordonoh, yes , it is on hathor15:42
Paliif you have still problems, copy only bootmenu.sh to your device :-)15:42
freemangordonno, it is ok after Y did echo 5> debian/compat15:43
Paliand compile watchdogs.c manually15:43
freemangordon*I15:43
Paliok, if it is working, I can change it also in git15:43
freemangordonI have to install it in "host" maemo, ain;t?15:43
Paliyou need to install bootmenu to ubifs15:43
freemangordonhmm:15:44
freemangordon"Preparing to replace bootmenu 1.12 (using bootmenu_1.12_armel.deb)"15:44
freemangordonyour debian/changelog not updated?15:44
freemangordonwell, not yours but bootmenu15:45
freemangordonPali: ^^^15:45
Paliyes, not updated15:45
freemangordonok15:45
Paliit is updated, but only on my HDD :-)15:45
Palidid not pushed it yet15:45
freemangordoncool :D15:45
freemangordonok, it works now15:46
Paliok15:46
freemangordonor not, it still tries to mount ubifs15:46
freemangordonWTF?15:46
Paliwho trying to mount ubifs?15:46
Paliyou need to use ubifs15:47
Palithere is stored maemo bootmenu15:47
freemangordon"VFS: Mounted root (ubifs filesystem) on device 0:1215:47
freemangordonand Kernel panic after that15:47
Paliso you need it15:48
freemangordonok, but why it panics?15:48
Palidid you build kernel with rx51_defconfig?15:48
freemangordonyes15:48
Palican you show me what you have in /etc/bootmenu.d?15:49
Palimaybe there is problem that unclean ubifs cannot be mounted15:49
PaliI see this error sometimes...15:49
Palitry to boot 2.6 kernel and shutdown phone from bootmenu15:49
FatPhilWho's the CSSU lead? I have a few hacks for user-space packages, and I'll happily share them with the wider community15:49
Paliand boot 3.10 kernel what happen15:49
freemangordonPali: ok15:50
PaliFatPhil: post patches somewhere :-) (No idea if gitorious working again), and I can look at it15:50
DocScrutinizer05FatPhil: merlin1991 is the dude with the CSSU hat on15:51
freemangordonFatPhil: merlin1991 is the maintainer, but it really doesn't matter to whom you'll send the patches15:51
DocScrutinizer05merlin1991: technically mohammadAG15:51
freemangordonDocScrutinizer05: come on :)15:52
DocScrutinizer05in daily struggle merlin1991 and me15:52
freemangordonmag is not in anything maemo for the last 2 years or so15:52
DocScrutinizer05yes, but he's still project lead of CSSU15:52
DocScrutinizer05technically15:53
freemangordonok, ok :)15:53
DocScrutinizer05what it's in RL is another story ;-D15:53
freemangordonshit, the backover on my devel device got broken!15:54
freemangordonbackcover even15:54
DocScrutinizer05the last official assignment is: merlin1991:CSSU project lead together with MohammadAG.  CSSU-T 100% MohammadAG + 50% merlin1991.  CSSU-S: 100% chemist + 50% merlin199115:55
DocScrutinizer05at least that's what I negotiated with them, and still in effect unless they negotiated sth new that I didn't get aware of15:57
DocScrutinizer05so far moh, merlin and chemist always listened to my advice in 100% of cases, since I like to discuss a lot publicly but rarely ever give advice in private query. Thechnically though all three are free to ignore me16:00
DocScrutinizer05it's a meritocracy, which some e*_ never wrapped his head around16:01
freemangordonPali: do I have to run your scripts again after reinstalling bootmenu?16:01
Palino16:02
Paliyou mustnt run it again16:02
freemangordonhmm, where did my 3.10 .item file disappeared?16:04
freemangordonWTF?16:04
freemangordonPali: no joy, seems it can't mount rootfs16:08
Palifreemangordon all items files are on ubifs not on new rootfs in /home/16:09
Palifreemangordon: rootfs is corrupted?16:09
freemangordonyes, that part seems to be ok, I have bootmenu entry to boot from eMMC, but when I select it, the kernel panics16:10
freemangordonno, 2.6.28 boots just fine16:10
Paliplease provide what you have in /etc/bootmenu.d/16:10
freemangordon-rw-r--r--    1 root     root           78 Sep  6  2013 20-Maemo5-kernel-3.10.0+.item16:11
freemangordon-rw-r--r--    1 root     root          117 Jul 29  2012 20-Maemo5-kernel-cssu-2.6.28.10-cssu3.item16:11
freemangordon-rwxr-xr-x    1 root     root          234 Sep  6  2013 eMMC-Maemo5.item16:11
freemangordonPali: shall I install upstart from cssu-devel?16:13
Paliyes, you need it16:13
freemangordonok16:13
Paliboot to maemo stored in /home/maemo5 with 2.6 kernel16:14
Paliinstall all cssu-devel packages here16:14
Paliand try to reboot to 3.10 kernel16:14
freemangordonthat is what I am doing16:14
jon_y3.10 is working??16:16
jon_yhmm, guess not16:17
Palijon_y: I'm able to boot it16:17
jon_ymount problems?16:17
Palibut gsm/3g stack not working, camera not working, and lot of maemo apps using 100% cpu16:17
jon_yI'm guessing there are no proprietary nokia blobs?16:17
PaliI'm using nokia blobs...16:17
jon_ywhat blob modules does it require?16:18
Palisomething working (like closed GUI apps) and someting not (mce daemon, camera)16:18
freemangordonall of them :D16:18
Palido you mean kernel binary blobs?16:18
jon_yI mean kernel wise16:18
jon_yyeah16:18
Palikernel is full open16:18
jon_yany binary only kernel modules?16:18
Palino16:18
jon_yok, good16:18
Palithis is good, that n900 does not using any binary kernel modules (like n810)16:19
jon_yunlike Andy too16:19
jon_ywhat is keeping gsm from working?16:19
freemangordonPali: WTF is hildon-welcome in cssu-devel?16:20
jon_yor rather, what is the main challenge?16:20
Palifreemangordon: yes, not fail if /proc/<nokia_name> not exists16:20
jon_ywell, there is back porting nokia stuff16:20
freemangordonoh16:20
Palijon_y: kernel drivers and closed nokia phone applications not working16:20
jon_yrefuse to work with linux kernel version change?16:21
jon_yhmm, kernel drivers aren't ported properly yet?16:22
freemangordonmost of them are, but not all16:22
jon_ycool, I hope you are pushing this upstream16:22
freemangordonyes16:22
freemangordonPali is the upstream man :)16:22
jon_yvery nice16:23
jon_yhmm, missed the Linux 3.11 bus :)16:23
FatPhilWhen talking to a whole range of ex-Nokians recently, sooooo many have mentioned Pali, he's quite notorious (in a good way)16:23
jon_yooh16:23
jon_ywhat did they think of elop? :)16:24
freemangordonjon_y: are you serious?16:24
jon_yabout?16:24
*** dafox has quit IRC16:24
freemangordonwhat would you think of an asshole that fires you?16:24
FatPhilThe MS trojan horse? Exactly what we thought about him 3 years ago. Everything has gone as predicted.16:24
jon_yI need to hear their own words :)16:24
FatPhilDon't quote me on this...16:25
jon_ythey said "trojan horse"?16:25
jon_yok16:25
FatPhilWell, I did.16:25
freemangordonFatPhil: did you make some folloup discussion re SR?16:25
freemangordon*followup16:26
FatPhilfreemangordon - nothing since out privmsg chat, I think.16:26
jon_yseriously, praise be to Pali16:26
jon_yworking on the kernel and all16:26
jon_yespecially uboot16:26
FatPhilI need to look at your patches again. What I saw looked well thought out. We need more insight into the Si versions, to be sure.16:27
freemangordonok, I'll gadly answer whatever question needed, to the extent of what I can remember :D16:28
FatPhilWould CSSU get confused if I were to have a RD image rather than a PR image on the device? I feel a bit lost without the R&D binaries!16:33
freemangordonPali: do I have to upgrade any of this http://pastebin.com/LAkfVyYi ?16:34
freemangordonFatPhil: it requires PR1.3 ;)16:34
jon_yR&D battery?16:34
jon_yoh wait sorry16:34
Palifreemangordon: not needed16:35
freemangordonok, rebooting16:35
FatPhilI have a pair of 36-2 images, for example.16:36
PaliFatPhil: what is RD image?16:36
FatPhilPali: what developers in Nokia used to install.16:36
freemangordonPali: still no joy :(16:36
PaliFatPhil: and difference between normal and RD?16:37
freemangordon"Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000200"16:37
freemangordonthere is mtdoops, but I doubt it will be of much help16:38
FatPhilPali: I can do a dpkg -l and diff the two16:38
FatPhil(next week)16:38
Paliok16:39
FatPhilPali: we had testing tools called "toast" and "insanity", and performance/endurance testing tools.16:39
freemangordonhmm run toast on OC device, please :D16:39
Palifreemangordon: can you start usb console from bootmenu16:39
freemangordoncan I?16:40
Paliand try to boot manually via usb console?16:40
freemangordonI am new to bootmenu, sorry16:40
Paliconnect phone to linux machine, and press shift when in bootmenu16:40
freemangordonok16:40
Paliit export usb ethernet card and assign ip address via dhcp16:40
Paliafter that you shold be able to ssh to n90016:41
keriowho runs the dhcpd?16:41
keriothe n900?16:41
Paliyes16:41
Palikerio: not dhcpd, but dnsmasq16:41
keriomh, why didn't you go with 192.168.2.14/15 as usual?16:41
kerioand yeah, dnsmasq working as a dhcp server16:41
Paliit will use that :-)16:41
Palibut this is easier as static ip address configuration16:42
Palidhcp is auto configuration16:42
freemangordonseems g_ether is not loaded16:43
freemangordonyep, it is not16:44
Paliit using g_nokia16:44
freemangordondoesn't matter, it is still not loaded16:44
freemangordonnothing happens when I connect the cable16:44
freemangordonhmm, could it be something wrong with my modules dir?16:44
Paliwhat bootmenu wrote?16:45
Palion screen?16:45
freemangordonUSB networking is up16:45
freemangordonetc.16:45
Palicheck on computer what you see in dmesg and in lsusb16:45
freemangordonnothing :)16:46
PaliFatPhil: do you mean these tools? https://maemo.gitorious.org/maemo-tools16:46
Palifreemangordon: boot into 2.6 kernel and check if you have g_nokia.ko16:46
freemangordonok16:46
Paliand also try to rerun depmod16:46
freemangordonok16:47
FatPhilPali: those and more16:47
Palidepmod -a <kernel version>16:47
freemangordonok16:47
PaliFatPhil: arent these tools in SDK or TOOLS repository?16:47
freemangordonsp-endurance?16:47
Palifreemangordon: you should see in kernel framebuffer if drivers were loaded16:49
FatPhilWOh - that's a list of familiar names: https://gitorious.org/+maemo-tools-developers16:50
Palifreemangordon: g_nokia.ko is broken in 3.11, here is my patch: http://pastebin.com/6JrzkiqY16:52
FatPhilI'll overlap with Leonid here at work for a while, I'll see if I can get him back involved16:52
Palipatch should enable usb networking16:52
freemangordonPali: framebuffer scrolls too fast, I guess I can just tweak boot.sh to start /bin/sh on the device16:57
Paliremopile kernel with above patch ^^^16:58
Paliand try again usb networking16:58
freemangordonok16:59
*** mkaindl has joined #maemo-ssu17:03
*** sunny_s_ has quit IRC17:04
*** sunny_s has joined #maemo-ssu17:08
freemangordonPali: hmm, where i should put modules?17:15
freemangordonon ubifs?17:15
Paliyes on ubifs and on /home/maemo5/lib too17:15
freemangordonok, that explains WTF is going on :D17:15
Paliyou did not copied modules to ubifs?17:16
* freemangordon hides17:16
freemangordonyep, I didn;t, I was expacting copying them to maemo5 to be enough17:17
Paliyou need modules on ubifs which using preinit and bootmenu17:17
freemangordonyeah, now I get it17:17
PaliSDHC drivers, ext3, ..17:17
freemangordontoldya, I am new to bootmenu17:17
freemangordonyes, yes17:17
Paliif you do not want to copy all modules, run grep with insmod and modprobe on /bootmenu.sh and /usr/sbin/preinit17:18
freemangordonI guess this needs initrd17:18
Palifreemangordon: booting now?17:31
freemangordonyep, but reboots on X startup17:31
freemangordongoing to do depmod in maemo5 too17:32
*** iDont has joined #maemo-ssu17:33
Palifreemangordon, look into syslog, why device rebooted17:34
freemangordondidn't help, going to check in syslog17:34
freemangordonyeah :D17:34
Palialso try to correct shutdown17:35
Paliand power up17:35
freemangordonhmm?17:35
Palisometimes 3.10 kernel does not initialize all HW17:35
freemangordonaah17:35
freemangordonok17:35
Paliand for example snd missing17:35
Pali(maybe other parts too)17:35
Palibut look into systlog17:35
freemangordonwell, I'll try to fix whatever is possible once I have it up and running17:35
freemangordoni guess it is "init: sgx pre-start process (1476) terminated with status 1" to blame17:40
freemangordonseems powervr doesn;t get loaded17:40
Palifreemangordon: try to comment some parts in event.d sgx script17:43
Palithere is some echo > lines17:43
freemangordonyep, I remember this was needed with 3.5 too17:43
jon_ybtw, anybody know why kernel reboot with terminal on sometimes only?17:44
jon_ysometimes it doesn't display any messages17:44
jon_yI'm curious why this happens17:45
*** mauroc has left #maemo-ssu17:56
*** oldtopman has joined #maemo-ssu17:59
freemangordonPali: could you verify on your device which parts I need to clean from sgx script?18:07
iDontPali, freemangordon: has one of you seen this patch before? http://git.openembedded.org/openembedded/plain/recipes/linux/linux-2.6.28/nokia900/0001-Fix-CPU-frequency-driver-so-that-it-loads-before-the.patch18:08
freemangordoniDont: no, do we really need it?18:08
iDontThere's a bug that prevents the kernel from booting when the mmc modules are built-in and cpu_freq is enabled as well. With that patch applied, I can boot Maemo from the eMMC without help of the NAND (i.e. directly calling /sbin/init on the eMMC from the kernel cmdline)18:09
freemangordonoh, nice18:09
iDontbut...18:09
iDontit breaks overclocking. Or more precise: the code that sets the max freq to 600Mhz. I.e. when both that patch and the overlocking patch are enabled, the n900 will boot at 1150MHz18:10
freemangordonI see18:10
freemangordonwell, I guess OC patch needs to be fixed too18:10
iDontI can check if I might be able to do that, but I'm not sure if I have the skills to do so.18:11
freemangordoniDont: I'll look at it18:11
iDont(on a sidenote: I've backported mtdswap from Harmattan and am now using the rootfs on the NAND as swap)18:12
freemangordonhow fast it is?18:12
DocScrutinizer05FatPhil: (list familar) http://wiki.maemo.org/Documentation/devtools/maemo518:12
*** NIN101 has joined #maemo-ssu18:13
iDontthe eMMC  is currently the bottleneck, I haven't tweaked any I/O, VM, or whatever yet. So I can't speak of the performance of mtdswap yet. I'm not sure how to measure it either tbh18:14
freemangordonhow does it feels when you load memory hungry applications18:14
freemangordon?18:14
freemangordonhmm, eMMC is slow? why is that?18:15
iDontOn my previous installation I tweaked some settings (e.g. nr_requests) that made the device feel snappier. I haven't done that yet with whole Maemo on the eMMC18:16
iDontit's not like the eMMC magically got any slower :D18:16
freemangordon:)18:17
iDontanyway, I can't say anything about the performance yet for that reason. I also haven't installed any applications yet or anything18:18
iDontjust running a clean Maemo (+CSSU-thumb) on the eMMC with mtdswap on the nand18:19
Palifreemangordon: nothing is changed on my device in sgx script18:20
keriois mtdswap worth it?18:20
kerioi'm running swap on uSD right now18:20
freemangordonI added || true after the echo lines, but still no joy, going to check syslog again18:20
freemangordonPali: ^^^18:20
Palifreemangordon: no || true, comment lines18:21
PaliI remember that chaning these sys values caused something bad18:21
iDontkerio:  I used to run swap on uSD too, it's great. But my new SD card is really slow in random i/o compared to my previous (way smaller) one. So I thought: why not use that other bus for swap18:21
freemangordonPali: tried that before adding || true, no difference18:21
Paliso then problem is somewhere esle18:21
Paliwhat is in syslog?18:22
freemangordon:nod:18:22
kerioiDont: well, for instance because swap is written a lot and nand only has a limited amount of writes18:22
freemangordonPali: do I need bme-replacement?18:22
Palifreemangordon, yes18:23
PaliBME not working18:23
Paliand is maybe reboot reason18:23
freemangordonhald-addon-bme: unable to send connect req to BME18:23
freemangordonPali: I guess I need KP52, ain't?18:24
Pali3.10 has support for bme replacement too18:24
freemangordonhmm, scratch that18:24
freemangordonwhere to get that from, gitorious?18:25
Pali~bme-replacement18:25
infobotbme-replacement is, like, http://atrey.karlin.mff.cuni.cz/~pali/rx51-bme-replacement  http://atrey.karlin.mff.cuni.cz/~pali/projects/maemo/bme-replacement.html  See also: http://wiki.maemo.org/Bme_replacement . Please, use wiki page to report bugs/problems and/or solutions to them!18:25
iDontkerio: I'm fully aware of that, there's not much I can add on that topic18:25
iDontkerio: the question is if we'll reach that limit in the device's lifetime. If you have (on anybody else) can shed some light on that, that'd be great18:27
iDontwoops, that sentence went wrong18:27
kerioit depends on the usage, i reckon18:27
jon_yI put mine in cold storage18:27
jon_yshould I get a BL-5J now?18:27
jon_yshould I take out the battery when not in use?18:28
freemangordonPali: I guess I have to remove bq modules from the blacklis18:29
iDontkerio: for now it's just a fun little project to run this exotic setup :-). Who knows, maybe I'll switch back to my old uSD (or buy a faster, larger one) in the future18:29
Palifreemangordon: no18:29
keriojon_y: who cares?18:29
keriojust buy a neo900 when it comes out :D18:29
Palifreemangordon: just install deb packages18:29
jon_ykerio: sure, but I want the original to survive :)18:29
freemangordonPali: is there any problem if I boot cssu3 with bme-replacement?18:29
freemangordonas long as I don;t try to charge18:29
freemangordonhmm, "Package kernel-feature-rx51-battery is not installed."18:30
Paliif you have kernel with bq drivers, it should work18:30
freemangordonwell, I guess I need KP5218:30
jon_yshow your grandkids you had a CELLPHONE18:30
jon_ynot just a stupid watch18:30
Paliinstall kp5218:30
freemangordonPali: cssu3 is kp51r118:30
freemangordonI will18:30
keriojon_y: *pair of glasses18:31
jon_yI was looking in the news today18:31
jon_ysamsung watch selling nearly the same price as a regular cell18:31
jon_ywhat are you smoking samsung18:31
jon_yI guess some people have lots of monies18:32
FatPhilsmaller = more expensive18:33
DocScrutinizer05jon_y: store device with battery removed. Store batteries in a cool (but not freezing) place and recharge them once or twice a year, preferably to max 75%18:33
jon_yok, should I get a new battery now?18:33
keriojon_y: well18:33
jon_ythe current battery has been there since I bought it18:33
kerioi heard that some crazy lunatics are going to spend 700€ on a replacement board for a 2009 phone made for nerds18:33
keriocould you believe that?18:34
keriomadness, i say18:34
jon_ykerio: who said anything about that? :)18:34
keriospending the same amount for a fancy new tech-wristwatch doesn't seem that bad, tbh18:35
jon_yI am resigning my hope of actually getting the Neo900 :|18:35
keriowhy? it's going to happen18:35
jon_y700€ is way too much18:35
*** sunny_s has quit IRC18:35
jon_yplus tax and shipping after that18:35
jon_yyeah, sort of the reason I'm keeping the n900 in cold storage now18:36
jon_ywhile it is still working18:36
FatPhilI don't understand how early development is going to be funded. Promises to part with 500e are worth nothing if the device never appears.18:36
jon_yI can hardy afford 250€18:37
jon_ythem currency conversions are real killers18:37
jon_ytime to save up, or wait until year end promotion18:37
DocScrutinizer05FatPhil: we never will reach 500EUR tag18:38
DocScrutinizer05FatPhil: early development gonna funded by some fundraiser, or preorders without warranty to ever deliver something18:38
DocScrutinizer05(the latter being a fundraiser donation with added value in fact)18:39
freemangordonPali: seems KP52 bootimg does not provide kernel-feature-rx51-battery18:39
Palifreemangordon, yes18:39
jon_yI could probably get the N900 if it didn't come with killer taxes :(18:39
DocScrutinizer05FatPhil: we need a humble ~15k EUR for prototypes18:39
Palibecause of feature problems which you described18:39
freemangordonPali: aah, yes18:40
Paliinstall kernel-power kernel-power-modules and kernel-power-flasher18:40
DocScrutinizer05some more if it shall happen fast18:40
freemangordonso only the flasher18:40
freemangordoni'll have to reinstall u-boot after that18:41
*** dos1 has joined #maemo-ssu18:41
freemangordonhmm, no :)18:41
freemangordonnice18:41
jon_ywhat was that flasher replacement?18:42
jon_ythe one that ask if you want to flash the kernel?18:42
freemangordonPali: well, at least now it stays on the running dots :)18:44
freemangordonoh, no18:45
freemangordonBOOTED :D18:45
kerio:D18:45
jon_yanybody remember what the flasher replacement was?18:46
jon_ythe flasher that runs on the n90018:46
jon_ywill allow you to fake *-flasher package installs18:46
freemangordonPali: ok, now I have a test device to work on, will see what I can do18:48
Palifreemangordon: nice :-)18:48
Palinow you need to stop hal and mce18:48
Paliand other daemons which using 100%CPU18:48
freemangordonBTW DSP doesn't work too, I guess it needs Harm codec nodes and updated libgstdsp18:48
jon_yah yes fiasco-image-update-ask18:49
Palifirst we need dsp kernel module :D18:49
freemangordonhmm, only hulda is using 100% CPU18:49
Palistop hulda18:49
freemangordonPali: WHAT? no DSP module? why, it is upstreamed iirc, in staging18:49
Palinot working anymore18:49
freemangordonoh18:49
freemangordonok18:49
Palibecause omap now using multiboard kernel image18:50
Paliand that dsp module using omap specific functions which is illegal now18:50
jon_yPali: will uboot still work on 3.1x images?18:51
Palifreemangordon: led not working in 3.10, it will be fixed in 3.1118:51
freemangordonok18:51
Palijon_y: yes18:51
jon_yok cool18:51
Palibut you need to compile uboot with gcc4.718:51
Palifor that18:51
jon_ydoes your uboot offering in the repo cover that?18:52
Paliuboot binary in extras is already compiled with gcc4.7, so no problem there18:52
freemangordonFatPhil: could you repeat what was changed in kernel API causing that 100% CPU usage?18:52
jon_yPali: awesome, thanks18:52
Palifreemangordon, I had problems also with mce and dsme and hal18:53
Palican you try to "stop ke-recv-extra"18:53
Pali(this is that hulda daemon)18:53
Paliand check if this is only one bad daemon?18:53
freemangordonPali: hmm, already rebooted to "stock" will check later18:54
Palihulda ke-recv-extras is open source and source code is on gitorious18:54
Paliok18:54
freemangordonhave to make something for my daugther to eat, bbl18:54
FatPhilfreemangordon - what we saw was a poll() on a sysfs/procfs file with POLL_IN flag19:16
FatPhilvirtual files always have data, so that returns immediately19:16
FatPhilIf you want _interesting_ data, then use POLL_PRI19:16
FatPhilthat limits it to data that the system actually wants you to have, rather than being prepared to let you have.19:17
freemangordonFatPhil: thanks19:19
freemangordonPali: "stop ke-recv-extra"does nothing - job not changed19:19
Palithen try: /etc/init.d/hulda stop19:20
freemangordonthat one did the job19:20
freemangordonPali: rtcom-call-ui segfaults19:21
freemangordonbut that's all I see19:21
Palifreemangordon: see commit 1af3557abdef34ee036a6de4cb79e24468544b8d in linux kernel tree19:21
PaliI reverted that commit in v3.10-n90019:22
freemangordonoh, nice19:22
Palihttps://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=1af3557abdef34ee036a6de4cb79e24468544b8d19:22
PaliFatPhil: this is that commit which causing our error? ^^^19:23
freemangordonok, so we don;t have the problam described by FatPhil19:23
Palithis commit is reverted in my git tree, but proper solution should be to fix daemons...19:23
PaliI reverted it in d8d5cab26e0f347dbce63ffcede6efaecbd6bfcd19:24
Palifreemangordon: I know rtcom crashing19:25
Paliand no idea why19:25
freemangordondo you know why?19:25
FatPhilPali - some commit between .29 and .32, yes - and that looks like the one19:25
freemangordonPali: ok19:25
FatPhilproper solution is to fix daemons, indeed.19:25
freemangordonPali: BTW by "leds broken" you mean display leds?19:26
Palithat 3 color led on left side19:26
freemangordonah19:26
Palilp53...19:26
freemangordonhmm, display brightness is at min19:27
Palikernel developers decided to change API19:27
Paliand totally broke mce...19:27
freemangordonPali: any idea how to stop rtcom-call-ui from respawning?19:27
PaliI wrote about this problem and they added API back in 3.1119:27
Palibut I did not tested it19:27
Palifreemangordon: rename rtcom-call-ui binary19:28
FatPhil freemangordon is it dsme that respawns it?19:28
Palido not remember who and why it still trying to respawn19:28
freemangordonFatPhil: yes19:28
PaliI think not19:28
FatPhilThere's a config file listing the tasks hat should be respawned19:28
freemangordoniiuc19:28
PaliI was not able to find dsmetool command which starting it19:29
Paliand if dsme is respawning that binary you should see info from dsme in syslog19:29
Palifreemangordon: what not working is sscd daemon which talking with cellular chip19:30
Paliin print some debug info that it is unable to power on it19:30
freemangordonPali: :nod:19:31
FatPhilindeed DSME tells you what it's doing in syslog, which is useful19:31
FatPhilI seem to have two of them running, one launched by maemo-launcher, the other by init19:36
freemangordon"/usr/share/applications/hildon/rtcom-call-ui.desktop" :)19:37
freemangordonX-Maemo-Prestarted=always19:37
freemangordonPali: comment ^^^19:38
Paliah :-) I forgot about this19:39
freemangordonseems light sensor does not work either19:42
*** dos1 has quit IRC19:42
freemangordonwhy it takes ages to boot?!?19:42
keriowelcome to the wonderful world of shitty I/O19:43
freemangordonkerio: with 2.6.28 it botts in 30-40 seconds, with 3.10 it need >90 secobds19:44
keriosame-ish config?19:44
Palifreemangordon: non working cellular chip19:45
Palithere is gui for PIN which waiting...19:45
freemangordonoh, ok19:45
FatPhilfreemangordon: 100% cpu daemons might be slowing boot down19:51
FatPhildoes top say anything?19:51
*** iDont has quit IRC19:51
freemangordonyeah, hulda :)19:51
FatPhilanyone know to change windows in irssi from a ssh/session on the n900?19:52
FatPhiloh - I got rid of the toolbar, so can't o "esc"19:52
freemangordonPali: "/usr/sbin/sscd -f --wait-cmt-startup -d 3"19:53
kerioFatPhil: HAH19:54
freemangordonspits lots of info in syslog19:54
kerio~vi-kbd19:54
infoboti guess vi-kbd is http://wiki.maemo.org/Remapping_keyboard/user_vis_awesome_kbdmapping19:54
keriopress esc19:54
FatPhileasy - hang up the screen and start in another xterminal!19:54
FatPhilwith the "esc" button19:55
keriowhy do you people use screen19:55
keriotmux is so much better :<19:55
FatPhilfuck I'm being eaten to death by fucking mozies19:55
keriowtf is a mozy?19:56
FatPhilbeen using screen since ~5 years before tmux xisted - inertial19:56
FatPhilmosquito19:56
keriooic19:57
keriobe a real man and eat them back19:57
freemangordonPali: hmm, it is weird that cellmo doesn't respond20:00
sixwheeledbeasthey all sorry to disturb, having trouble with new maemo-certman not working is there anyway to debug?20:02
sixwheeledbeastWorks on one device but not another...20:02
freemangordonsixwheeledbeast: "not working" as in?20:03
sixwheeledbeastsearching... for hours20:03
freemangordoncheck in syslog20:03
freemangordonif it succeeds to connect to supl.nokia.com20:04
DocScrutinizer05Pali: lp5523 API is brainfscked in maemo. it's a minimalistic API that doesn't expose all the capabilities of the lp5523 chip20:12
DocScrutinizer05Pali: however it's possible to augment the API in a way that keeps backward compatibility20:13
PaliDocScrutinizer05: it is restored in 3.1120:13
DocScrutinizer05fine, but is it also augmented?20:13
DocScrutinizer05lp5523 allows double the number of program steps ber engine, compared to lp5521 which it's based on. The lp5523 driver completely hides/ignores this feature since it would need some sysfs node that allows partitioning the program storage and asign chunks of available memory to each engine20:15
DocScrutinizer05s/ ber/ per/20:15
infobotDocScrutinizer05 meant: lp5523 allows double the number of program steps per engine, compared to lp5521 which it's based on. The lp5523 driver completely hides/ignores this feature since it would need some sysfs node that allows partitioning the program storage and asign chunks ...20:15
kerioDocScrutinizer05: what would that allow?20:16
DocScrutinizer05it would allow way longer programs for the engines20:16
DocScrutinizer05another point is the muxing of the 9 LED outputs to the 3 engines that's hardcoded in Nokia's lp5523.ko20:17
*** wmarone has joined #maemo-ssu20:18
DocScrutinizer05i.e. the API supports only 2 engines controlling the 3 color indicator LED, while 3rd engine is assigned to kbf LEDs hardcoded - I call it abuse of that versatile engine for a silly ramp-up/down that as well could get done by mce process20:19
DocScrutinizer05the lp5523.ko API is a pretty silly minimal effort and uninspired augment of the lp5521.ko API20:22
DocScrutinizer05it however *could* get sanitized/salvaged20:23
DocScrutinizer05without breaking it20:23
DocScrutinizer05that's what i hope for in new kernels20:23
FatPhilyou don't want the heartbeat controlled by mce20:25
DocScrutinizer05huh?20:26
FatPhilthe ramp-up and ramp-down20:26
FatPhilthat's justa simple program for the led controller20:26
DocScrutinizer05what got ramp-up/down to do with heartbeat?20:26
DocScrutinizer05I know how it works, I analyzed it and basicaly wrote the wiki page about it20:27
FatPhilthat's what it was called - the slowly pulsing led20:27
DocScrutinizer05unrelated to kbd-backlight20:29
DocScrutinizer05which engine3 is abused for, while mce could do same thing under sw-control, writing new brightness levels to led_kb:1/brightness to :6/brightness20:30
DocScrutinizer05basically the engines do exactly same,just without help of APE20:30
*** dos1 has joined #maemo-ssu20:32
DocScrutinizer05if mce process would take care about those 0..500ms ramping, every once in 10 min when kbd backlight needs to light up or shut down, then we could free up engine3 for general pupose20:32
*** dos1 has quit IRC20:33
*** arcean_ has joined #maemo-ssu20:33
DocScrutinizer05hell, forget mce process, I can create a shell script working more than fine for that, in as short as 5 minutes20:33
*** dos1 has joined #maemo-ssu20:33
FatPhilwhat general purpose though. Why are you so enamourd with the idea of offloding a atask a picocontroller can do ontoa 500'Hz geneal purpose CPU?20:34
DocScrutinizer05the major fsckup in API is: the ledmux is for indicator LED only, using lowercase r, g, b to assign LED to engine1 and uppercase R, G, B to assign to engine2. No way to assign to engine3 since engine3 is reserved for kbd LED ramping20:35
*** arcean has quit IRC20:36
DocScrutinizer05FatPhil: try creating a rainbow LED-pattern for indicator LED!20:36
DocScrutinizer05near impossible since only 2 engines available20:36
FatPhilI wrote a perl script which would turn the kbd leds into Knight Rider's "KITT" 4 years ago, I don't nneed lessons in how easy it is to control20:36
FatPhilImpossible for me to see - I'm a dichromat20:37
DocScrutinizer05solution: augment the mux sysnode to take old style "rGb" assignments as well as some new syntax like e.g. "1:1,2:1,3:3,4:1,"... to assign each of the 9 outputs to one of the three engines20:38
DocScrutinizer05FatPhil: I wrote a bash script for same purpose20:38
DocScrutinizer05fine that you don't need lessons ;-)20:39
FatPhilforgive the typos, I've got about 1min lag here!20:39
DocScrutinizer05np, foget the typos20:40
DocScrutinizer05the point about engine3 is: it's hardcoded to be used for kbd only. that's not what you expect from a kernel driver and its API20:41
DocScrutinizer05the API is broken by design20:41
DocScrutinizer05though fixable20:42
DocScrutinizer05all it needs to fix it is an augmented syntax for the muxing node, plus some additional nodes to partiton/manage the program storage20:43
DocScrutinizer05http://wiki.maemo.org/LED_patterns#LEDsToMap20:47
FatPhilI agree that if it's not hard wired, that's the imposition of policy, which is bad20:50
FatPhilhard-wired as in electrionicall, no choice20:50
FatPhilhowever, the completely free absraction would be chaos to manage, if &20:50
DocScrutinizer05in lp5521 there been one engine with 16 steps max prog storage. in http://wiki.maemo.org/LED_patterns#Engine_Pattern that's kept though the LP5523 has a total of 96 steps assignable arbitrarily to the 3 engines20:50
*** Vlad_on_the_road has joined #maemo-ssu20:51
DocScrutinizer05since driver "abuses" two(?) program steps for a mux-pattern table, they couldn't go for a fixed 32 steps per engine either20:52
DocScrutinizer05so they rather decided to not support at all the additional 48(-2) program steps20:53
DocScrutinizer05the 16step/engine limit is hardcoded to the driver20:54
DocScrutinizer05they could have gone for 30 or even 31 steps / engine, but noooo...20:54
DocScrutinizer05OOPS scratch that about mux sysnode, I confused that with MCE syntax21:02
DocScrutinizer05IroN900:/sys/class/i2c-adapter/i2c-2/2-0032/engine3_leds is perfectly sane, as are engine1_leds and engine2_leds21:04
sixwheeledbeastfreemangordon: issue fixed thanks, seems I had a faulty sim card.21:07
sixwheeledbeastamazing how different the locations are between the two devices now I am fixed. 10-15 miles different between two devices on the same network21:10
ShadowJKthey're talking to different towers?21:11
sixwheeledbeastI suppose so but I have at least three closer than the 15 mile away one.21:18
*** BCMM has joined #maemo-ssu21:23
*** BCMM has quit IRC21:55
*** Vlad_on_the_road has quit IRC21:59
*** Vlad_on_the_road has joined #maemo-ssu22:13
DocScrutinizer05don't you think FatPhil last post was epic? ;-D22:18
DocScrutinizer05(lp5523.ko) so what it needs most to sanitize is a sysnode that allows partitioning of memory to the three engines. 3 ranges out of 0..95, plus a 4th range that stays unused, for e.g. mux lookup table22:25
DocScrutinizer05plus a second sysnode that would allow direct raw r/w-access to that memory22:26
DocScrutinizer05since I wouldn't like to hack stuff like mux-LUT via defining a temporary range for engineX and then loading a "program" to that range22:27
DocScrutinizer05that direct access should cover complete chip's addr range, incl the variables' registers22:30
DocScrutinizer05much like every sane i2c-chip driver should expose such a /raw sysnode22:30
DocScrutinizer05for N900 board config, the lp5523.ko driver should include a board specific check of the /raw sysnode for the self-destruction bit not getting set (the IO pin that is connected to SoC GPIO, usually is input, but could get configured to output Vbat to the SoC if not blocked by kernel driver)22:34
DocScrutinizer05actually one of the few hw bugs I found in N90022:35
*** bsdmaniak has joined #maemo-ssu22:36
* DocScrutinizer05 idly wonders what the SoC would do when +Vbat gets applied to one of the (1.8V?) GPIO22:38
DocScrutinizer05ShadowJK: ^^^22:38
ShadowJKvent magic smoke?22:39
DocScrutinizer05you forgot 2blue"22:39
DocScrutinizer05"blue" even22:39
DocScrutinizer05it's not exactly as simple to predict. The LP5523 has a rather modest fan-out, and the SoC has clamping diodes on all pins22:40
DocScrutinizer05so, i guess first thing to happen is a rise in Vio(1V8)22:42
DocScrutinizer05depending on fan-out vs current current consumption on Vio (I adore aliterations), either the output of LP5523 goes down to 1V8, or the clamp diode smokes up, or the whole CPU/SoC gets hit by lightning22:45
*** sunny_s has joined #maemo-ssu23:01
DocScrutinizer05Pali: would you be interested in me providing aughmented/fixed API spec and possibly untested patches for lp5523.ko?23:02
PaliDocScrutinizer05: maybe you should inform about lp API TI developers of lp driver23:03
PaliI really do not have time for it :-(23:04
DocScrutinizer05I don't think TI developed that driver23:04
PaliI'm happy that TI developers added back old API support23:04
Paliyes, TI developers created new API in 3.1023:04
DocScrutinizer05the chip is LXP aka Phillips iirc, the driver been definitely tailored to fit N900 needs only23:05
DocScrutinizer05ooh, so it has new API?23:05
freemangordonPali: doesn't ssi need some clock to be registered in the board code?23:05
DocScrutinizer05that been my initial question actually23:05
Palihttps://lkml.org/lkml/2013/8/5/4623:06
PaliDocScrutinizer05: yes, there is new api for lp552323:06
DocScrutinizer05thanks :-D23:06
Paliit using genertic linux firmware functions for loading program for lp chip23:07
Paliin above mail are linux git commits which introduced that API23:07
Palifreemangordon: no idea23:07
freemangordonPali: omap_ssi uses ssi_ssr_fck and ssi_ick23:07
freemangordonthose 2 are used by bridgedriver too23:08
PaliI have really nothing about ssi/hsi23:08
Palifeel free to patch any ssi code23:08
freemangordonmy point is - shouldn't those be somehow registered or whatever in the board code?23:08
PaliDocScrutinizer05: and https://lkml.org/lkml/2013/8/13/70323:09
Palifreemangordon: no idea23:09
DocScrutinizer05EEEEK mipi hsi?23:10
freemangordonPali: those are registered in cclock3xxx_data.c, so it is ok23:15
*** BCMM has joined #maemo-ssu23:16
freemangordonPali: BTW, which hsi/ssi/etc patches did you use?23:17
Palithose from you and skry23:18
Palinow I found 0xFFFF flasher in ubuntu repository :D funny that my SW appeared here and I not know about it23:24
Palihttp://packages.ubuntu.com/saucy/0xffff23:24
*** bsdmaniak has quit IRC23:24
freemangordon:D23:25
freemangordonhttp://changelogs.ubuntu.com/changelogs/pool/universe/0/0xffff/0xffff_0.6~git20130406-1/0xffff.copyright23:27
freemangordon40423:27
freemangordonhttp://changelogs.ubuntu.com/changelogs/pool/universe/0/0xffff/0xffff_0.6~git20130406-1/copyright23:27
*** LauRoman has joined #maemo-ssu23:28
freemangordonPali: maybe we should ask for some help on #nemomobile, or the author for hsi/ssi/cmt driver23:29
Palifreemangordon: look here: http://elinux.org/N90023:29
PaliSre trying to upstream hsi/ssi/cmt patches23:30
Palihttps://lkml.org/lkml/2013/8/11/6723:30
freemangordonooh, nice23:30
Pali"There is currently work going on to get the omap-ssi driver into the mainline kernel. This work may be ready for 3.13 or 3.14."23:31
freemangordon:)23:31
*** arcean_ has quit IRC23:33
DocScrutinizer05Pali: you tested that new lp5523.ko driver?23:36
PaliDocScrutinizer05: not yet, this is reason why I did not answer that email23:36
Palisorry I did not have time...23:36
DocScrutinizer05:nod:23:36
DocScrutinizer05np23:36
DocScrutinizer05just asking since this determines my proofread-mode23:36
PaliDocScrutinizer05: maybe you should ask questions about lp drivers that TI developers...23:37
Palinow when they refactored lp driver23:37
DocScrutinizer05when it's known to work, I search the bugs at my end. otherwise I ponder if the code might be buggy23:37
freemangordonPali: WTF is DT?23:38
Palidevice tree23:38
freemangordonok, thanks23:39
DocScrutinizer05hsi drivers for omap4 might be interesting, since there's one location where the WONTFIX omap4 SiErr hits23:39
Palifreemangordon: this is that device tree which comes from Sun and OpenFirmware23:40
freemangordonPali: not that I know anything about it :)23:41
DocScrutinizer05the nasty point: when OMAP$ misses an IRQ on HSI, then the far end has to do sth about it which is not anything conforming HSI protocol specs23:41
Paliit is used for replacing board peripheral files with static device data23:41
freemangordonooh, so no more board files?23:42
Palino more static data for describing hw23:42
Palibut do not know if this means removing board data completly...23:43
freemangordonyeah, got it. what is used, some scripts?23:43
DocScrutinizer05amazing!23:43
DocScrutinizer05how does that work?23:43
DocScrutinizer05parameters handed to the build process?23:43
Palifreemangordon, device tree is language in which is this info written23:44
freemangordonhttp://www.devicetree.org/Main_Page23:44
PaliDocScrutinizer05: board hw id is already set by bootloader, so kernel know type of hw23:45
DocScrutinizer05WOW23:45
DocScrutinizer05one (kernel) for all (platforms)23:46
Palifor all arm platforms23:47
Palithis is what linux kernel trying...23:47
* DocScrutinizer05 wonders how sane this idea is, regarding overhead and cruft, aka dead code/data23:47
Paliomap2 and omap3 has already support for this multiarch23:47
Paliof course you can choose which components you want to compile23:48
DocScrutinizer05aah23:48
Palibut kernel developers want something like one kernel image for all x8623:48
Paliand because of this, DSP and infrared drivers was disabled :-( and are not compiling anymore23:49
Palibecause using omap specific functions23:49
DocScrutinizer05yeah sure, so SSE extension support in kernel for x386 engines23:49
Palii386 support is dropped23:49
DocScrutinizer05lol23:49
Paliand for SSE you can check cpuid23:50
Paliyou need i486 at least for last 3.x kernel23:50
DocScrutinizer05that doesn't help much for code in kernel or not, you can check CPU ID only when kernel already got loaded, incl SSE23:50
dos1armv4t vs. armv7te? :P23:50
dos1if you want one kernel for Neo Freerunner and N900, you'll have to either use armv4t, or drop Freerunner support :D23:51
DocScrutinizer05lololol23:51
dos1but one kernel for N900, Neo900 and GTA04... why not?23:52
DocScrutinizer05sure23:52
DocScrutinizer05it's same platform23:52
Palinot possible, n900 kernel area is too small for all that code :D23:53
ShadowJKI think the devicetree thing originally came from Linus getting pissed off at BeagleBoard's 5 variants, and everything beagleboard-like, like N900, etc23:53
Palidevicetree was already used in kernel for sun and powerpc arch23:53
Palithey decided that devicetree can be used also for arm :-)23:54
Palibecause ACPI is evil23:54
*** nox- has joined #maemo-ssu23:54
PaliACPI DSDT tables are nonsense23:54
dos1Pali: then u-boot and kernel on SD? ;]23:54
Pali:D23:55
DocScrutinizer05:-S  https://github.com/milokim/lp55xx/blob/master/drivers/leds/leds-lp5523.c#L22323:55
PaliDocScrutinizer05: now when driver is (again?) maintained by developers (TI), you should write all your objections about it...23:56
DocScrutinizer05will do23:56
PaliI was very surprised that they immediately created wraper for old API compatiblity....23:57
PaliI will resend SMC/PPA patches again and will prepare board data and functions for bq2415x charger for upstreaming23:59

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