**** BEGIN LOGGING AT Sat Oct 06 03:00:02 2012 Oct 06 06:12:43 Is there anyone install Ruby 1.9 on N900 ? Oct 06 07:34:15 * archeyDevil is installing power-kernel and vim :) Oct 06 07:46:15 Gotta love deauth :D Oct 06 08:05:48 archeyDevil: oh you Oct 06 08:06:50 kerio: Hey Oct 06 08:07:41 Oh goody.. So easy to restore the kernel back :) Oct 06 08:08:15 Good thing I don't want to tho ^_^ Oct 06 08:50:39 <_PanzerSajt> Pali, Can you help me with u-boot? Oct 06 08:51:01 <_PanzerSajt> I can't boot into maemo because I used wrong parameters in the bootitems Oct 06 08:51:40 _PanzerSajt, stock nokia kernel is attached to uboot Oct 06 08:51:50 so you can still boot this kernel Oct 06 08:51:59 <_PanzerSajt> yes but the kernel-atags flag Oct 06 08:52:02 <_PanzerSajt> is wrong Oct 06 08:52:19 _PanzerSajt: can't you just flash xImage? Oct 06 08:52:24 *zImage Oct 06 08:52:35 _PanzerSajt, what version of uboot you have? Oct 06 08:52:46 can you open uboot console? Oct 06 08:52:52 <_PanzerSajt> yes Oct 06 08:53:08 <_PanzerSajt> 2012.04.01-01269 Oct 06 08:53:22 run this command: run attachboot Oct 06 08:53:49 <_PanzerSajt> it just tries to boot maemo Oct 06 08:53:57 <_PanzerSajt> but It hangs and starting kernel Oct 06 08:54:06 <_PanzerSajt> *at starting Oct 06 08:54:58 <_PanzerSajt> can I modify the address via setenv? Oct 06 08:55:14 what did you do? Oct 06 08:55:18 attachboot should work Oct 06 08:55:39 you rewritten attachboot env? or what? Oct 06 08:55:51 <_PanzerSajt> I tried to boot android but I haven't updated my .item files with ATAGADDR Oct 06 08:55:58 <_PanzerSajt> nothing rewritten Oct 06 08:56:08 what are atags? Oct 06 08:56:19 then maemo kernel should be bootable Oct 06 08:56:54 remove battery, wait 5 minutes, insert it and try again Oct 06 08:56:58 <_PanzerSajt> nitdroid boots fine Oct 06 08:57:06 <_PanzerSajt> booting ubuntu from sd works fine Oct 06 08:57:15 <_PanzerSajt> just maemo not Oct 06 08:57:15 can you charge battery in nitdroid/ubuntu? Oct 06 08:57:22 <_PanzerSajt> yes I can Oct 06 08:57:26 maybe battery is empty Oct 06 08:57:33 and maemo turing phone off Oct 06 08:57:38 <_PanzerSajt> no Oct 06 08:57:46 <_PanzerSajt> I have charged it 90% Oct 06 08:58:07 ok Oct 06 08:58:20 download zImage with framebuffer Oct 06 08:58:31 use mkimage to create uImage (from zImage) Oct 06 08:58:46 store it to sd card (on fat/ext4 partition) Oct 06 08:58:48 and boot it Oct 06 09:00:25 $ mkimage -A arm -O linux -T kernel -C none -a 80008000 -e 80008000 -n zImage-fb -d zImage-fb uImage Oct 06 09:00:43 this will convert zImage-fb to uImage ^^^ (run on host pc) Oct 06 09:00:54 uboot can boot only uImage(s) Oct 06 09:01:13 <_PanzerSajt> from where can I download the zImage? Oct 06 09:02:06 use this: http://mohammadag.xceleo.org/public/maemo/kernels/framebuffer/zImage-fb-omap1-pr13 Oct 06 09:03:12 uImage store to first fat32 partition on SD card Oct 06 09:04:16 and in uboot console run: Oct 06 09:04:20 setenv setup_omap_atag 1 Oct 06 09:04:22 run sdboot Oct 06 09:08:08 <_PanzerSajt> kernel panic Oct 06 09:08:24 can you write me full message? Oct 06 09:11:28 _PanzerSajt, maybe upload photo image somewhere, or rewrite screen to pastebin Oct 06 09:11:47 <_PanzerSajt> yes I'm uploading photo Oct 06 09:12:16 ok Oct 06 09:13:54 <_PanzerSajt> http://imageshack.us/a/img839/674/img2012100611100165.jpg Oct 06 09:15:21 oh, same error message as merlin1991 :-( Oct 06 09:16:03 _PanzerSajt, for unknown reason stock kernel booted from uboot on only some n900 cannot mount rootfs ubifs Oct 06 09:16:22 <_PanzerSajt> is it a hw or sw problem? Oct 06 09:16:27 sw problem Oct 06 09:16:34 <_PanzerSajt> oh that's fine Oct 06 09:16:52 you need to reflash only *kernel* to boot maemo Oct 06 09:16:53 Pali: still not solved? Oct 06 09:16:58 no Oct 06 09:17:06 hmm, any clue what happens? Oct 06 09:17:06 I cannot reproduce it on my n900 Oct 06 09:17:16 and I'm waiting for dumps from merlin1991 Oct 06 09:17:22 ok Oct 06 09:17:25 <_PanzerSajt> Pali, I can help you debug Oct 06 09:17:43 _PanzerSajt, can you boot rescue OS from uboot? Oct 06 09:18:02 I need atags dump from some kernel booted from uboot Oct 06 09:18:26 _PanzerSajt, also atags dumps from nitdorid/ubuntu are usefull too Oct 06 09:18:44 Pali: http://plugcomputer.org/plugforum/index.php?topic=1272.0 Oct 06 09:18:59 <_PanzerSajt> how can I get those dumps from ubuntu boot? Oct 06 09:20:03 see this thread: http://talk.maemo.org/showthread.php?t=84560 Oct 06 09:20:22 here is armel binary of dumpatags: http://talk.maemo.org/showpost.php?p=1214013&postcount=10 Oct 06 09:20:43 you need to run that on android/ubuntu/whatever started by uboot Oct 06 09:21:00 and upload dump to pastebin Oct 06 09:21:48 something needs to tell the ubi layer to attach mtd4 as ubi0 Oct 06 09:22:46 "rootfs" string is link for "mtd4" Oct 06 09:23:04 Pali: read what I posted few lines above Oct 06 09:23:20 freemangordon, I read Oct 06 09:23:24 ok :) Oct 06 09:23:37 but problem is that when you boot stock kernel without uboot it working fine! Oct 06 09:23:47 I know Oct 06 09:23:59 nolo -> kernel is OK, but nolo -> uboot -> kernel not Oct 06 09:24:09 so I do not think that something was not called Oct 06 09:24:16 you think there is some atag missing? Oct 06 09:24:33 maybe missing or bad atag from mtd parittion Oct 06 09:32:59 <_PanzerSajt> Pali, I can't run that binry Oct 06 09:33:26 <_PanzerSajt> and also when I cat /proc/atags I just get the u-boot boot parameters Oct 06 09:33:45 cat /proc/atags is usefull too Oct 06 09:34:15 use xxd or other tool which convert binary text to ascii Oct 06 09:37:10 here are other reports of this problem: http://talk.maemo.org/showpost.php?p=1247913&postcount=443 http://talk.maemo.org/showpost.php?p=1260514&postcount=495 Oct 06 09:37:13 * Sicelo also has that same issue.. can't boot stock maemo kernel with u-boot Oct 06 09:37:58 <_PanzerSajt> Pali, http://paste.ubuntu.com/1263476/ Oct 06 09:38:05 but ubooot-pr13 does it fine Oct 06 09:38:27 ah.. yes, /me is qhubekela on tmo Oct 06 09:38:52 wait, root=/dev/mmcblk0p3? Oct 06 09:39:56 <_PanzerSajt> kerio, yes but this is ubuntu running from SD Oct 06 09:42:30 which repo has xxd? Oct 06 09:42:53 vim Oct 06 09:43:01 vim depends on xxd Oct 06 09:43:08 (or gvim) Oct 06 09:43:18 ty Oct 06 09:43:28 _PanzerSajt, here is converted your dump http://pastebin.com/ADi7Tvgu Oct 06 09:43:36 there are missing *all* omap atags Oct 06 09:44:11 _PanzerSajt, you need to run "setenv setup_omap_atag 1" before starting system Oct 06 09:45:18 <_PanzerSajt> before starting ubuntu? Oct 06 09:45:41 yes Oct 06 09:45:43 in uboot console Oct 06 09:45:59 start uboot console and run: Oct 06 09:46:02 setenv setup_omap_atag 1 Oct 06 09:46:06 bootmenu Oct 06 09:47:57 Sicelo, do you have some system booted from uboot working? Oct 06 09:54:39 <_PanzerSajt> Pali, http://paste.ubuntu.com/1263508/ Oct 06 09:54:51 <_PanzerSajt> sorry but I can't acces pastebin Oct 06 09:55:23 no problem ubuntu has paste too :-) Oct 06 09:56:05 <_PanzerSajt> but I can't read your link Oct 06 09:56:15 <_PanzerSajt> http://pastebin.com/ADi7Tvgu Oct 06 09:59:09 here is converted pastebin: http://pastebin.com/bCczYFLm Oct 06 09:59:55 ok, here is copy (of second): http://paste.ubuntu.com/1263513/ Oct 06 10:00:30 rootfs is correct: Oct 06 10:00:32 001c:4f0b PARTITION: rootfs : size=0fb40000 offset=004c0000 mask=00000000 Oct 06 10:00:59 but there is abnormal mem tag: Oct 06 10:01:00 54410002 ATAG MEM size=00000000 start=90000000 Oct 06 10:04:07 <_PanzerSajt> how should it look like? Oct 06 10:05:06 size=0 is abnormal Oct 06 10:05:21 but it should work (kernel should ignore it) Oct 06 10:05:47 but I do not know where can be problem... Oct 06 10:06:17 more people wrote that booting from 2012.04-1 worked, but 2012.04-2 not Oct 06 10:06:23 I will look at changes Oct 06 10:10:58 <_PanzerSajt> Pali, I can't use maemo since 3-4 mounths due to this bug Oct 06 10:11:30 <_PanzerSajt> but I wouldn't like to resintall maemo and u-boot again Oct 06 10:11:49 <_PanzerSajt> because ubuntu is booting now just fine and it was really hard to achive Oct 06 10:14:29 in 2012.04-2 was enabled onenand support Oct 06 10:14:45 maybe uboot doing some initialization which can break maemo kernel? Oct 06 10:15:20 _PanzerSajt, I will compile same uboot version (2012.04-2) but with disabled onenand support Oct 06 10:17:00 _PanzerSajt, flasher-3.5 can send kernel/uboot image to NOLO and run it without flashing Oct 06 10:18:08 <_PanzerSajt> Pali, yes reflashing just the bootloader would be nice Oct 06 10:23:43 _PanzerSajt, here is uboot fisco image without onenand support: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/u-boot-2012.04-2+zImage-2.6.28-omap1.fiasco Oct 06 10:24:28 <_PanzerSajt> thanks I will try it later in the evening Oct 06 10:24:39 you can start it (without flashing to nand) with this command from hist pc: flasher-3.5 -F u-boot-2012.04-2+zImage-2.6.28-omap1.fiasco -l -b Oct 06 10:25:22 so you can test if working Oct 06 10:25:49 after reboot will be started original uboot Oct 06 10:26:21 _PanzerSajt, after testing write here results Oct 06 10:26:46 ping Sicelo, merlin1991 Oct 06 10:27:30 <_PanzerSajt> Pali, still the same Oct 06 10:27:44 <_PanzerSajt> your new u-boot starts but maemo doesn't Oct 06 10:27:47 <_PanzerSajt> bye Oct 06 10:27:49 ahhh... Oct 06 10:27:53 bye Oct 06 10:52:19 Pali: new KP52 deb? :3 Oct 06 10:52:47 kerio, why? you have prekp52 for testing... Oct 06 10:53:05 you said you'd make new debs with the updated modules Oct 06 10:54:28 yes, but later later Oct 06 10:54:42 * kerio is sad Oct 06 10:54:46 prekp52 version is working fine for me Oct 06 10:54:59 but uboot not! Oct 06 11:07:36 for testing, here are some old uboot fiasco images: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/ Oct 06 11:07:50 I need to know which was last "working" version Oct 06 11:08:23 _PanzerSajt, merlin1991, Sicelo, ... ^^^ Oct 06 11:45:29 <_PanzerSajt> Pali, I'm back Oct 06 11:45:58 <_PanzerSajt> can I boot maemo kernel just one time? Oct 06 11:46:13 <_PanzerSajt> like you freshly compiled u-boot Oct 06 11:53:45 yes Oct 06 11:54:30 _PanzerSajt, use that framebuffer zImage and run: flasher-3.5 -k zImage_file_name -l -b Oct 06 11:55:03 that framebuffer zImage is not fiasco image, but regular kernel image, so '-k' arg is needed (instead -F) Oct 06 11:55:48 _PanzerSajt, when you boot to maemo kernel, can you upload atags? Oct 06 11:56:49 <_PanzerSajt> ? Oct 06 11:58:09 <_PanzerSajt> sorry I don't understand Oct 06 11:58:27 use this kernel http://mohammadag.xceleo.org/public/maemo/kernels/framebuffer/zImage-fb-omap1-pr13 Oct 06 11:58:48 <_PanzerSajt> yes that's ok Oct 06 11:58:52 and run from host pc: flasher-3.5 -k zImage-fb-omap1-pr13 -l -b Oct 06 11:59:00 <_PanzerSajt> but _PanzerSajt, when you boot to maemo kernel, can you upload atags? Oct 06 11:59:19 xdd /proc/atags Oct 06 11:59:21 *xxd Oct 06 11:59:31 <_PanzerSajt> it wasn't working Oct 06 11:59:50 <_PanzerSajt> started booting but stopped Oct 06 12:00:04 _PanzerSajt, reason? Oct 06 12:00:09 kernel erro? Oct 06 12:00:28 <_PanzerSajt> I can't read Oct 06 12:00:30 <_PanzerSajt> too fast Oct 06 12:00:56 turn off phone (remove battery), wait some time, insert it and try again Oct 06 12:01:08 maybe bootreson was act dead Oct 06 12:02:15 <_PanzerSajt> couldn't create lockfile Oct 06 12:02:26 <_PanzerSajt> this was the last I could read Oct 06 12:02:40 <_PanzerSajt> and also there's no backlight Oct 06 12:02:54 you can enable R&D mode with disabled watchdogs which prevent rebooting: Oct 06 12:02:59 $ flasher-3.5 --enable-rd-mode --set-rd-flags=no-omap-wd,no-ext-wd,no-lifeguard-reset Oct 06 12:03:35 disabling R&D mode: $ flasher-3.5 --disable-rd-mode Oct 06 12:11:17 Pali: I'll gather the atags now Oct 06 12:11:30 merlin1991, ok Oct 06 12:12:04 merlin1991, can you test if one of these uboot version working? http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/ Oct 06 12:12:21 I will Oct 06 12:12:37 do they all have the stock maemo kernel attached? Oct 06 12:13:09 yes Oct 06 12:13:46 merlin1991, when you dump atags from rescueos/ubuntu/other_kernel first open uboot console and write this: Oct 06 12:13:47 setenv setup_omap_atag 1 Oct 06 12:13:49 bootmenu Oct 06 12:13:57 Pali: merlin1991, morning :P Oct 06 12:14:18 (this will force uboot to append omap atags and run bootmenu again) Oct 06 12:14:24 ZogG_laptop, hi Oct 06 12:15:43 Pali: can I put rescueos on an sd card to simply run sdboot? Oct 06 12:17:22 merlin1991, no because sdboot booting only kernel image Oct 06 12:18:56 merlin1991, this config item file should work: http://talk.maemo.org/showpost.php?p=1264989&postcount=507 Oct 06 12:20:36 Pali: it will take a bit longer, for reasons unkown to me my wlan is down :/ Oct 06 12:20:53 wtf? Oct 06 12:21:19 try: stop icd2 && start icd2 && stop wlancond && start wlancond Oct 06 12:22:29 <_PanzerSajt> Pali, I still can't boot maemo Oct 06 12:22:58 _PanzerSajt, do you have log? Oct 06 12:23:19 <_PanzerSajt> where and how can I acces log files? Oct 06 12:23:50 if you have ksyslogd installed, then log file is stored in /var/log/syslog Oct 06 12:24:06 (or sysklogd.. I do not know package name) Oct 06 12:25:29 Pali: nah it was the router Oct 06 12:25:36 fsckd router needs a reboot once in a while Oct 06 12:25:50 ok Oct 06 12:26:07 my router needs a kickin' every once in a while :( Oct 06 12:27:44 when it dies, i'll buy a routerboard one Oct 06 12:29:14 although the OS is closed Oct 06 12:33:49 Pali: do I need setup_omap_atag 1 if I have ITEM_OMAPATAG ? Oct 06 12:34:06 then no Oct 06 12:35:40 merlin1991, btw chess in n900 is open source: http://repository.maemo.org/pool/maemo5.0/free/o/osso-chess-ui/ Oct 06 12:35:53 ah shit Oct 06 12:36:04 so possible bugs can be fixed :-) Oct 06 12:36:05 which engine does it use? Oct 06 12:36:17 I remember checking on the source of some game and deducting that all are closed Oct 06 12:36:18 kerio, look at source code :-) Oct 06 12:36:53 https://wiki.maemo.org/Fremantle_closed_packages Oct 06 12:36:56 btw Pali you should update your u-boot-images archive, it still uses the old atag command in the item files Oct 06 12:37:01 only graphics & sounds are non free Oct 06 12:37:07 hm, babychess apparently? Oct 06 12:37:35 merlin1991, I will update them with new uboot version Oct 06 12:38:40 kerio, look at debian/control Oct 06 12:38:44 it depends on gnuchess Oct 06 12:39:01 OSSO Chess user interface. Required chess engine, GNU Chess, and OSSO games Start-up screen provided separately. Oct 06 12:39:20 but... it does have a chess engine built-in D: Oct 06 12:39:53 I see Oct 06 12:42:47 does it? afaik it uses gnuchess Oct 06 12:44:05 the source tarball has a chess engine in a subdir Oct 06 12:44:50 hmm Oct 06 12:46:31 anyway, the interface is of the built-in chess app is quite uncool, at least the graphics of the pieces. Oct 06 12:46:58 imho Oct 06 12:49:09 ffs where did my dumpatags go Oct 06 12:56:04 Pali: err which command in the uboot console lets me turn off the device? Oct 06 12:56:21 not possible Oct 06 12:56:31 only reset Oct 06 12:56:37 :/ Oct 06 12:57:52 merlin1991: you might get away with shel llevel commands Oct 06 12:58:03 GPIO etc Oct 06 12:58:59 if that's really important for you, I'd take the time to look options up for you, along the line of powering down the system hard Oct 06 12:59:30 Pali: -15 did not boot, 12 and 11 didn't even get to the kernel panic (I guess the config changed there from item_reuseatags to item_omapatags) Oct 06 12:59:33 DocScrutinizer05, do you know where is kernel code which power off n900? Oct 06 12:59:58 however note there's some cleaning up supposed to get done during graceful shutdown process, like configuring some peripherals to be properly off-state too Oct 06 13:00:08 merlin1991, in 12 and 11 there was change, but attached kernel should boot Oct 06 13:00:11 Pali: sorry, no Oct 06 13:01:25 a few commands to twl4030 will for sure shut down main system hard Oct 06 13:01:57 Pali: oops I ran the fb entry, will try 12 and 11 again Oct 06 13:01:58 but they won't care for LP5523 indicator LED, even cmt, PHY, etc Oct 06 13:02:19 there's no main switch? :o Oct 06 13:02:24 nope Oct 06 13:02:55 and no way to brutally tell bq24 to stop anything going from and to the battery? Oct 06 13:03:30 bq24 is a battery charger Oct 06 13:03:48 for modem it's a technical problem, other peripherals like LED and PHY are meant to keep config and even work during main system powered down Oct 06 13:04:18 kerio: there _is_ _no_ switch like that anywhare in swystem Oct 06 13:04:26 it does not control the main power supplies that get power from battery Oct 06 13:04:50 cmt *needs* direct connection to battery for technical reasons Oct 06 13:05:30 (impedance aka ESR of any such switch would spoil power for cmt) Oct 06 13:06:42 PHY (USB) needs power from battery to keep config which in turn is important to properly detect charger plugin Oct 06 13:07:24 it's possibly conceivable to blow the main power fuse. Oct 06 13:07:40 that however isn't really reversible. Oct 06 13:07:43 LP5523 LED is connected directly to battery since it's meant to blink even when system powered down or suspended, and that chip also has internal power-down mode Oct 06 13:08:01 oh, that's USB, on reflection Oct 06 13:08:15 main fuse? yep Oct 06 13:08:29 DocScrutinizer05: anyway, i think my weird-ish problems wrt act-dead with no charger are caused by BM Oct 06 13:08:32 but they're temporary Oct 06 13:08:37 no fuse on battery either (except internal one), same rationale as for switch Oct 06 13:08:47 Pali: yesterday's asm code was buggy :D http://pastebin.com/8LpMQmQ6 Oct 06 13:08:56 DocScrutinizer05, SpeedEvil, do you know design capacity of bl5j battery which was included in n900 pack? Oct 06 13:09:07 sure Oct 06 13:09:19 1350? I forget Oct 06 13:09:33 1320 Oct 06 13:09:41 printed on body Oct 06 13:09:47 freemangordon, wrote that nokia increased capacity of new bl5j batteries Oct 06 13:10:03 I read that, never seen it though Oct 06 13:10:19 all my batteries are 1320 Oct 06 13:11:20 <_PanzerSajt> my batteries allso. I have a 2.5 year old and a 1 year old Oct 06 13:11:42 hi _PanzerSajt :-) Oct 06 13:11:51 <_PanzerSajt> DocScrutinizer05, HY Oct 06 13:13:48 bme reporting 1293 Oct 06 13:15:03 that's since the whole situation about BSI is pretty undefined AIUI Oct 06 13:16:16 and rx51_battery driver ( https://gitorious.org/rx51-bme-replacement/rx51_battery/blobs/master/rx51_battery.c#line139 ) reporting between 1257-1292 Oct 06 13:17:01 DocScrutinizer05, it this ok for bme replacement? Oct 06 13:17:23 I think it's just good enough Oct 06 13:18:06 nothing _really_ depending on precise BSI/design-capacity reporting Oct 06 13:18:17 Pali: http://cdnm.at/~christian/maemo/atags/ Oct 06 13:18:43 merlin1991, thanks! Oct 06 13:19:03 merlin1991, btw both devices not bootabled from uboot? Oct 06 13:19:12 I also tested the attached images of 11 12 and 15 and they all failed Oct 06 13:19:20 Pali: yes those failed Oct 06 13:19:54 merlin1991, and do you have device where is uboot working? Oct 06 13:20:12 yes it has hw revision 2101 Oct 06 13:20:42 can you dump atags from it too (again one without uboot and one from rescue os)? Oct 06 13:20:55 okay, but lemme have lunch first :) Oct 06 13:21:29 ok, thanks :-) Oct 06 13:21:44 in both is this tag (from uboot): 54410002 ATAG MEM size=00000000 start=90000000 Oct 06 13:21:51 DocScrutinizer05, see ^^^ Oct 06 13:21:54 <_PanzerSajt> Pali, the rx51_battery is for battery management? what does it depend on? Oct 06 13:22:26 _PanzerSajt, rx51_battery is kernel driver which export battery desing capacity, battery voltage and temperature to /sys/ Oct 06 13:22:45 it is part of project which replace BME Oct 06 13:23:01 <_PanzerSajt> and do you have charging script/program? Oct 06 13:23:13 I have kernel driver for charging Oct 06 13:23:26 <_PanzerSajt> because now we have i2c script for charging the battery with ubuntu Oct 06 13:23:49 Pali: yes, this ATAg MEM line looks kinda weird Oct 06 13:23:58 <_PanzerSajt> http://bazaar.launchpad.net/~panzersajt/ubuntu-n900-accessories/trunk/revision/4#charge-automatic.sh Oct 06 13:24:29 \o/ my charge concept made it to buntkuh... *cough* Oct 06 13:24:29 Pali: can't rx51_battery driver grab the value from bq27200? Oct 06 13:24:58 kerio, no, rx51_battery read values from twl adc channels Oct 06 13:25:23 but you can use also bq27x00_battery driver which report other values Oct 06 13:25:28 kerio: bq27200 doesn't know about proper design-capacity Oct 06 13:25:35 DocScrutinizer05: *should* it? Oct 06 13:25:42 for bme replacement is needed both 3 drivers Oct 06 13:25:53 err, well, it *should* but evidently it doesn't Oct 06 13:26:08 but nothing cares about battery design Oct 06 13:26:26 _PanzerSajt, when I got some free time, I will try to use some recent kernel and try to use bq2415x kernel driver on it Oct 06 13:26:26 also how's the chip in N900 going to know design-cap of arbitrary plugged-in battery? Oct 06 13:26:42 no, i know, it doesn't know Oct 06 13:26:44 and it shouldn't know Oct 06 13:26:51 but why do we care? Oct 06 13:26:57 it's a useless metric Oct 06 13:27:21 kerio, because we need battery temperature Oct 06 13:27:33 <_PanzerSajt> Pali, but is that script safe? Oct 06 13:27:37 LOL, http://bazaar.launchpad.net/~panzersajt/ubuntu-n900-accessories/trunk/revision/4#charge-automatic.sh actually is my original script Oct 06 13:27:48 DocScrutinizer05 what does charge-automatic do exactly? Oct 06 13:27:49 <_PanzerSajt> DocScrutinizer05, how safe is it? :) Oct 06 13:27:56 safe enough Oct 06 13:28:04 it is ok for charging with wallcharger Oct 06 13:28:17 IOW it's absolutely safe, just maybe not that reliable Oct 06 13:28:18 <_PanzerSajt> NIN101, detects if the wallcharger is present Oct 06 13:28:27 Pali: battery temp is a function of the design capacity too? :O Oct 06 13:28:45 kerio, it is part of that driver Oct 06 13:28:54 oh, k Oct 06 13:29:01 <_PanzerSajt> DocScrutinizer05, do you mind that we use your script with out ubuntu project? Oct 06 13:29:09 <_PanzerSajt> *with our Oct 06 13:29:11 nah Oct 06 13:29:20 go ahead Oct 06 13:29:57 jacekowski: ping Oct 06 13:30:18 <_PanzerSajt> Pali, can you help skry? he is trying to bring 3.5 kernel to n900 Oct 06 13:30:28 you just could mention me, maybe point to http://talk.maemo.org/showthread.php?p=658278#post658278 Oct 06 13:30:39 _PanzerSajt: afaik he has 3.5 booting Oct 06 13:30:43 _PanzerSajt, now I'm overprojected :-) Oct 06 13:30:50 <_PanzerSajt> Pali, np Oct 06 13:30:53 bme, uboot, kernel-power, ... Oct 06 13:31:11 <_PanzerSajt> freemangordon, yes but he still strugles with charging Oct 06 13:31:13 DocScrutinizer05: hm, isn't charge21.sh "better"? Oct 06 13:31:18 sure Oct 06 13:31:23 I like when people make QT apps like hamster-filer with a "rotator" function, ;) Oct 06 13:31:42 _PanzerSajt: he just need to forward-port Pali's drivers Oct 06 13:31:57 "better" as in "more versatile, more reliable" Oct 06 13:32:12 <_PanzerSajt> freemangordon, so it will work? Oct 06 13:32:16 yes Oct 06 13:32:24 <_PanzerSajt> fine thanks Oct 06 13:32:51 you need to write to rx51 board code functions which connect isp and bq drivers together Oct 06 13:33:09 isp? Oct 06 13:33:16 in maemo kernel isp code is glued into musb, so I needed to change musb code... Oct 06 13:33:26 iirc isp is camera Oct 06 13:33:47 isp1704 kernel driver for charger detection Oct 06 13:33:54 aah, ok :) Oct 06 13:34:31 Pali: do you know anyone here but jacekowski who can review my assembler code Oct 06 13:34:43 (MUTEX_SET) Oct 06 13:34:52 DocScrutinizer05? Oct 06 13:35:03 ? Oct 06 13:35:07 hmm, yeah, why not :) Oct 06 13:35:16 http://pastebin.com/8LpMQmQ6 Oct 06 13:35:25 DocScrutinizer05: ^^^ Oct 06 13:35:33 maybe lxp? Oct 06 13:35:49 original part is for non-thumb code Oct 06 13:36:12 i need my thumb replacement reviewed Oct 06 13:36:34 aah, sorry I'm not really awake to do a proper review right now Oct 06 13:37:00 DocScrutinizer05: no hurry, but please do it when you're awake :) Oct 06 13:37:40 I'll try to not forget it, ping me again in a few hours if I did nevertheless Oct 06 13:37:41 this code is from berkeley db and I want to be sure I won't screw up address book :D Oct 06 13:37:53 ok Oct 06 13:39:01 Pali: (isp1704 kernel driver for charger detection) ... is highly critical code, since it will 'ruin' 1707 config when run in the wrong moment Oct 06 13:39:52 DocScrutinizer05, when I looked at that driver it register some otg notification, and usb code calling detection function Oct 06 13:40:12 Pali: gambling with this code can result in all sorts of highly unwanted weird effects Oct 06 13:42:39 incl system not booting up anymore, or not detecting fastcharger at all, or (most likely) breaking any established USB session used to do PC-Suite or AssRage or USB-networking or even charging from non-fastcharger powersources Oct 06 13:44:10 I'm not saying you shouldn't touch that code (au contraire it needs some love, urgently), but I suggest you closely inspect what and how this code is doing, when it may get called and when not, etc Oct 06 13:45:06 Pali: I added atags from attached kernel, stock kernel and rescueos for the device that works (2101) Oct 06 13:45:07 http://cdnm.at/~christian/maemo/atags/ Oct 06 13:45:25 ok, thanks! Oct 06 13:45:51 and the difference I could spot so far, is that on the one that works we have no MEM size=0 weird shit Oct 06 13:46:42 merlin1991: can you help with some gcc inline assembly? Oct 06 13:47:05 Pali: yes, basically it's the musb-hdrc code detection function that actually messes up the 1707 config Oct 06 13:47:09 freemangordon: you've got a billion times more expertise in that regard compared to me :D Oct 06 13:47:20 merlin1991: I need a review Oct 06 13:47:47 Pali: as a general rule you MUST NOT call the musb-hdrc detection code whenever any USB session got established Oct 06 13:48:04 merlin1991: otherwise code seems to work, but it is in very critical part of the system Oct 06 13:48:36 DocScrutinizer05, ok, in kp is that detection disabled when you enter to host mode Oct 06 13:48:58 Pali: USB sessions are not limited to hostmode Oct 06 13:49:02 so some code should be added to upstream kernel too... Oct 06 13:49:03 yes Oct 06 13:49:45 freemangordon: what exactly should I review? as I don't have the slightes clue how to read actual assembler code :D Oct 06 13:49:56 merlin1991, are you sure that attached_2101.txt is dump from kernel booted from uboot and rescueos_2101.txt is booted from uboot too? Oct 06 13:50:11 merlin1991: well, scratch that then :) Oct 06 13:50:20 Pali: I'm 100% certain, though I'm happy to retry Oct 06 13:50:45 merlin1991, ok, but uboot appending same mem atags for all kernels Oct 06 13:50:53 and there are two different... Oct 06 13:51:23 can you also boot other kernel (kernel-power, or framebuffer kernel) on that phone where booting from uboot working? Oct 06 13:51:54 Pali: the right thing to do is call that detection code unconditionally only once immediately after VBUS got detected (external power source), then check if that power source is fastcharger and store the info in a static. Then never again call the deetction function until VBUS=0 (power source detached) but rather return static result of initial probing when _anybody_ asks Oct 06 13:52:13 merlin1991, and did you used last uboot 2012.04-2? Oct 06 13:52:24 Pali: "immediately" here means not even ENUM may have started yet Oct 06 13:52:55 DocScrutinizer05, ok Oct 06 13:52:59 :-) Oct 06 13:53:07 Pali: actually I can't boot the fb kernel there Oct 06 13:53:37 Pali: is it nolo that passes ATAG MEM size=00000000 start=90000000 Oct 06 13:53:41 or the not attached stock kernel Oct 06 13:53:51 freemangordon, no Oct 06 13:54:03 but who? Oct 06 13:54:18 Pali: maemo kernel wise only the atached kernel boots Oct 06 13:54:31 freemangordon, uboot Oct 06 13:54:36 sorry, my question was where did u-boot get them Oct 06 13:54:46 from sdrc code Oct 06 13:54:55 or I do not know Oct 06 13:54:59 trying to find it now Oct 06 13:55:02 Pali: also yes it's 2012.04-2 Oct 06 13:56:02 Pali: can you just make a quick'n'dirty hack and check if (ATAG MEM size=00000000) do not pass atag Oct 06 13:56:11 merlin1991, I will build uboot image which force 2 static mem tags (same which report nolo) Oct 06 13:56:26 DocScrutinizer05, you can look at current implementation: http://git.infradead.org/battery-2.6.git/blob/HEAD:/drivers/power/isp1704_charger.c Oct 06 13:57:07 hmm, is it possible that 2204 is with 512MB ram :P Oct 06 13:57:25 :D:D:D Oct 06 13:58:10 Pali: seems one of the differences between HW revs is memory chip Oct 06 13:58:14 does the u-boot implementation for booting the attached image differ from booting a menu entry? Oct 06 13:58:27 no Oct 06 13:58:38 why do I have different mem tags then? oO Oct 06 13:58:44 I have no idea Oct 06 13:58:57 revs above 2101 have faster memory, for sure Oct 06 13:59:03 but mem atags are read via function get_sdr_cs_size when uboot is starting Oct 06 13:59:07 :o Oct 06 13:59:08 really? Oct 06 13:59:11 * kerio has 2104 Oct 06 13:59:26 DocScrutinizer05: it's quite the fragile chip, isn't it Oct 06 13:59:33 Pali: where is that (link to gitorious please) Oct 06 13:59:42 I want to look at the code Oct 06 13:59:48 Pali: http://git.infradead.org/battery-2.6.git/blob/HEAD:/drivers/power/isp1704_charger.c#l103 is doing the damage when any data is actually being transferred Oct 06 14:00:08 freemangordon, file arch/arm/cpu/armv7/omap3/sdrc.c function dram_init_banksize Oct 06 14:00:17 Pali: http://git.infradead.org/battery-2.6.git/blob/HEAD:/drivers/power/isp1704_charger.c#l132 should do for most 'usual' cases Oct 06 14:00:27 Pali: I just managed to get the fb kernel to boot by changing the .item file for it back to ITEM_REUSEATAGS Oct 06 14:00:51 Pali: still it's forbidden to call this function when any session already established Oct 06 14:01:39 YAY!!! http://git.infradead.org/battery-2.6.git/blob/HEAD:/drivers/power/isp1704_charger.c#l143 Oct 06 14:01:55 Pali: which branch is that on gitorious? Oct 06 14:02:25 freemangordon, https://gitorious.org/u-boot-shr/u-boot/blobs/pali/arch/arm/cpu/armv7/omap3/sdrc.c#line232 Oct 06 14:02:30 thanks Oct 06 14:03:21 Pali: nvm the images not booting on the u-boot where the attached image worked (I had a typo in the .item file ITEM_OMAPATAGS instead of ITEM_OMAPATAG) :/ Oct 06 14:03:37 unlike isp1704_charger_type() the freaking isp1704_charger_verify() does NOT care about restoring sane original state of 1707 config Oct 06 14:03:47 merlin1991, ok Oct 06 14:03:58 DocScrutinizer05, I will look at that driver later Oct 06 14:04:13 first we need working u-boot, then we can booting other kernel :-) Oct 06 14:04:27 and maemo kernel does not have this driver Oct 06 14:05:11 Pali: do you have any idea where is number of DRAM banks calculated? Oct 06 14:05:26 in rx51 config file Oct 06 14:05:27 or i have to clone the code :D Oct 06 14:05:28 it is 2 Oct 06 14:05:48 well, that should not be hardcoded AIUI Oct 06 14:05:59 as we have atleast 2 revs with only one bank Oct 06 14:06:06 2101 and 2204 Oct 06 14:06:15 2104 is with 2 banks Oct 06 14:06:22 https://gitorious.org/u-boot-shr/u-boot/blobs/pali/include/configs/nokia_rx51.h Oct 06 14:06:43 freemangordon, see dumps: http://talk.maemo.org/showthread.php?t=84560 Oct 06 14:08:59 this is 2104: http://talk.maemo.org/showpost.php?p=1216012&postcount=22 and has 2 banks Oct 06 14:09:09 this is 2104 and has 1 bank: http://talk.maemo.org/showpost.php?p=1214052&postcount=14 Oct 06 14:09:20 that is what I am saying Oct 06 14:09:32 2104 is with 2 banks Oct 06 14:09:55 2104 has both (1 or 2) Oct 06 14:10:01 aah Oct 06 14:10:20 well, it seems I should have to clone the code Oct 06 14:10:25 freemangordon: ie my 2104 has only 1 bank (if I count the MEM atags as an educated guess for mem banks) Oct 06 14:10:34 you can Oct 06 14:11:45 and it seems like u-boot substidutes another weird MEM atag when there is only 1 in the original set Oct 06 14:12:51 uboot totally ignore MEM tag from nolo Oct 06 14:13:18 may I ask why? Oct 06 14:14:27 merlin1991: because u-boot should do the same as nolo Oct 06 14:14:29 because uboot is bootloader which initialize hw, and uboot has also code for sdrc Oct 06 14:15:02 and adding code which parse nolo tags are use it before uboot is initialized is hard Oct 06 14:15:07 Pali: clonning, will look at the code Oct 06 14:15:11 ok Oct 06 14:15:34 freemangordon hit the point Oct 06 14:17:02 merlin1991, here is uboot image with patch with skip all mem size=0 tags: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/u-boot-new.fiasco Oct 06 14:17:47 merlin1991, can you ran command bdinfo in uboot console Oct 06 14:18:02 it write info about mem Oct 06 14:18:10 also Pali my bad about the rescueos on u-boot Oct 06 14:18:13 * merlin1991 facepalms Oct 06 14:18:22 I somehow scpd the wrong txt Oct 06 14:18:32 AIUI mem banks should get read out of some SPD area of storage chips, I.E. get this info from hw, not at compile time from any board config file Oct 06 14:18:47 merlin1991, ok update it Oct 06 14:19:51 btw that's exactly the reason why flasher-3.5 fails when you don't have correct hw-rev Oct 06 14:20:18 (which kinda would render my prev assumption moot) Oct 06 14:20:25 Pali: updated http://cdnm.at/~christian/maemo/atags/rescueos_2101.txt Oct 06 14:20:30 Pali: sorry i disappeared.. band connectivity here. Oct 06 14:20:32 now it's in line with the other atags on 2101 Oct 06 14:21:01 what's the "best" hw rev, i wonder Oct 06 14:21:07 the joy of SoC hw revisions Oct 06 14:21:16 merlin1991, ok this is fine Oct 06 14:21:25 um, for us noobs.. what's this about having 1 or 2 memory banks.. ? i have 2204. Oct 06 14:21:32 kerio: there's no 'best' hw-rev Oct 06 14:22:01 revs above 2101 have faster memory, for sure Oct 06 14:22:07 those with 2 banks have one unpopulated bank, i would assume. is it? Oct 06 14:22:13 merlin1991, can you test that new uboot image? Oct 06 14:22:19 currently flashing Oct 06 14:22:20 there's certainly some kind of partial ordering between hw revs :) Oct 06 14:22:39 merlin1991, ok, but you do not need to flash uboot Oct 06 14:22:45 Sicelo: your CPU has different possible OneNAND/RAM piggyback POP, and those chips have different addressing modes Oct 06 14:23:26 Pali: bdinfo gives me the bogus banks Oct 06 14:23:42 Sicelo: like your usual RAM stick in PC might have 8 chips with 1Gbit each, or 4 chips with 2GBit Oct 06 14:23:54 bot are 1Gbyte Oct 06 14:23:57 both* Oct 06 14:24:12 Pali: also new u-boot still does not boot Oct 06 14:24:27 can you dump atags? Oct 06 14:24:33 on my way :D Oct 06 14:24:36 if mem=0 is still here? Oct 06 14:24:37 ok Oct 06 14:25:36 DocScrutinizer05, why then kernel started from uboot cannot mount ubifs rootfs but same kernel started by nolo can? Oct 06 14:26:02 atags (with partition layout) are ok Oct 06 14:26:39 Pali: https://gitorious.org/u-boot-shr/u-boot/blobs/pali/arch/arm/lib/bootm.c#line130 Oct 06 14:27:16 CONFIG_NR_DRAM_BANKS is 2, no matter how many actual banks we have Oct 06 14:27:17 freemangordon, I know Oct 06 14:27:25 is it (weak)? Oct 06 14:27:34 here I added line to skip size=0 Oct 06 14:27:36 can we move that to board code? Oct 06 14:27:38 ok Oct 06 14:27:43 and merlin testing new code Oct 06 14:27:57 not possible to move, because "static void" Oct 06 14:28:11 http://cdnm.at/~christian/maemo/atags/rescueos_2104_new.txt Oct 06 14:28:35 much better :D Oct 06 14:28:37 ok, there is no mem size=0 :-) Oct 06 14:28:52 Pali: but it really seems like a bug Oct 06 14:28:58 you should upstream that Oct 06 14:29:07 yes Oct 06 14:29:08 merlin1991: does it boot? Oct 06 14:29:18 nope Oct 06 14:30:52 but the atags are now 1 : 1 comapred to stock (except the rescueos specific entries regarding cmdline and initrd Oct 06 14:32:19 merlin1991: it does not boot, because it cannot mount rootfs? Oct 06 14:32:43 * merlin1991 tries the fb kernel Oct 06 14:33:26 yep ubifs fails for rootfs Oct 06 14:34:21 merlin1991, I added -9 and -10 uboot images: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/ Oct 06 14:34:23 did someone compared kernel cmdline? Oct 06 14:34:35 they are old version from my git Oct 06 14:34:42 *compare Oct 06 14:34:54 both are marked to 2012.04 Oct 06 14:36:34 Pali: -10 boots Oct 06 14:36:51 merlin1991, so -10 boots and -11 not? Oct 06 14:37:04 * merlin1991 tries -11 Oct 06 14:38:27 -11 is stuck at starting kernel Oct 06 14:39:22 ok, Oct 06 14:39:29 going to generate diff Oct 06 14:39:34 and source code Oct 06 14:39:50 Pali: share it please (diff) Oct 06 14:39:58 yes, I upload it Oct 06 14:40:02 <_PanzerSajt> Pali, how about the bug? Oct 06 14:40:21 wtf Oct 06 14:40:38 merlin1991, what? Oct 06 14:40:54 Pali: where? :D:D:D Oct 06 14:41:21 hm after trying -11 -10 didn't want to work untill I pulled the battery once Oct 06 14:43:02 now I flashed -11 pulled the battery once and it booted Oct 06 14:43:03 WTF? Oct 06 14:43:40 merlin1991: does it boot if you power it down? Oct 06 14:43:43 (-11) Oct 06 14:44:27 yep Oct 06 14:45:00 merlin1991, test if -12 is booting too Oct 06 14:45:18 I'm working my way towards 15 again :D Oct 06 14:46:10 btw Pali where are those -x revisions from, tags in git? Oct 06 14:46:13 -15 is last version (2012.04-2) but recompiled Oct 06 14:46:32 they are my internal numbering in mu pc Oct 06 14:46:37 hehe Oct 06 14:46:48 it is big mess :-) Oct 06 14:47:00 * freemangordon prays for Pali's PC Oct 06 14:47:25 Pali: 12 didn't boot, even after I removed the batter and tried it Oct 06 14:47:47 * merlin1991 flashes back 11 Oct 06 14:48:07 * freemangordon is waiting for .diff :P Oct 06 14:48:14 I uploaded tarballs here: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/ Oct 06 14:48:38 11 and 12 Oct 06 14:48:58 okay 11 does not boot emideatly after flashing, but boots on the next try Oct 06 14:50:50 now I see that -12 has included atags in uboot code Oct 06 14:53:28 hm the diff isn't even that big, but something in there must have a sideeffect Oct 06 14:56:53 Pali: could you compile me a recent uboot without the thumb errata workaround? Oct 06 14:57:06 merlin1991, ok Oct 06 14:57:33 or better compile a -12 without that part Oct 06 14:57:55 that's the only thing in the diff I've spotted that makes a huge difference Oct 06 14:59:55 hehe, yeah, trsutzone crap probably doesn't get reset by 'normal reboot' Oct 06 15:00:06 Pali: can you dump nolo atags from u-boot? Oct 06 15:00:19 freemangordon, yes, I have patch for it Oct 06 15:00:30 now going to compile image for merlin Oct 06 15:00:37 and then will find that patch Oct 06 15:00:51 there is not much difference between -11 and -12, besides atags Oct 06 15:01:10 and the thumb errata code ;) Oct 06 15:01:29 merlin1991: yes, but that should not prevent rootfs from mounting :D:D:D Oct 06 15:01:40 who knows why kernel goes nuts? Oct 06 15:01:58 since atags with the last patch from pali are literally the same as on stock kernel from nolo Oct 06 15:02:10 besides their order Oct 06 15:02:12 in that patch are only: removed code which copying nolo atags and replaced by static atags and added code for thumb Oct 06 15:02:23 I honestly pray for thumb-workaround not turning out to be the culprit Oct 06 15:02:32 DocScrutinizer05: it is not Oct 06 15:02:41 freemangordon: we will see ;) Oct 06 15:02:44 hehe Oct 06 15:03:21 merlin1991, image is here: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/ Oct 06 15:03:51 btw Pali do I need anything special to build u-boot from that source? Oct 06 15:03:58 or is a simple scratchbox setup enough? Oct 06 15:04:03 it is Oct 06 15:04:20 I'm using directly ubuntu gcc arm cross compiler Oct 06 15:04:41 dpkg-buipdpackage -b -aarmel without madde and without scratchbox Oct 06 15:04:48 i was using 4.6.2 last time i built it, not sure if 4.2 will do the job Oct 06 15:04:51 *dpkg-buildpackage Oct 06 15:05:32 Pali: why not using -mthumb? Oct 06 15:05:46 because uboot has no thumb support Oct 06 15:05:57 it has own flag -marm Oct 06 15:06:10 what does that mean? (no thumb support) Oct 06 15:06:20 wow, so it's not USB and PHY / twl4030 that caused recent problems with KP51xx (not properly) booting, but actually something around trustzone? Oct 06 15:06:42 uboot build system cannot compile with thumb Oct 06 15:06:46 ok Oct 06 15:07:05 when I tried to change all -marm to -mthumb code, it builds but not worked... Oct 06 15:07:14 in qemu is crashed Oct 06 15:07:24 aah, this is different ;) Oct 06 15:07:34 :( Oct 06 15:07:44 okay disabled thumb didn't help anything Oct 06 15:07:57 I don't even get a msg from the framebuffer kernel Oct 06 15:07:59 pheeew Oct 06 15:08:03 toldya :P Oct 06 15:08:13 you removed battery? Oct 06 15:08:21 not yet Oct 06 15:08:27 to "reset" "sticky" trustzone crap? Oct 06 15:08:58 I mean, the culprit sits in hw from prev 'rogue' kernel doing 'bad thing' to trustzone Oct 06 15:09:13 not in new kernel failing to do the right thing Oct 06 15:09:17 well removed batter does not help Oct 06 15:09:29 :-S Oct 06 15:09:29 freemangordon, patch is here: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/dumpatags.patch Oct 06 15:09:37 still don't get past the starting kernel part Oct 06 15:09:37 bisect time Oct 06 15:09:38 but maybe it needs some changes to work... Oct 06 15:10:16 DocScrutinizer05: no, it is only one patch Oct 06 15:10:42 bisect only helps with a patch series ;) Oct 06 15:10:57 Pali: your default: is commented Oct 06 15:10:58 so pretty please remove this one patch and recompile the very same source on very same toolchain with absolutely same parameters Oct 06 15:11:27 Pali: what if there is some atag u-boot does not know about it? Oct 06 15:11:41 DocScrutinizer05: we narrowed it down to this patch, pre patch version works Oct 06 15:11:58 yes, and I suggest to double-check Oct 06 15:11:59 freemangordon, good question Oct 06 15:12:26 just to rule out any form of PEBKAC Oct 06 15:12:36 freemangordon, uncomment default Oct 06 15:12:41 Pali: please, apply that patch with uncommenetd default: and give that image to merlin1991 to test Oct 06 15:12:44 :-) Oct 06 15:13:01 DocScrutinizer05: bootloader should ot depend on the user :P Oct 06 15:13:06 which version -11 or -12? Oct 06 15:13:07 *not Oct 06 15:13:21 Pali: does not matter aiui Oct 06 15:13:25 but maybe on refetching last cmdline to build it, from shell history Oct 06 15:13:27 whatever Oct 06 15:13:37 a typo somewhere Oct 06 15:14:00 we just need a confirmation that all needed atags are passed to the kernel in the same way as nolo passes them Oct 06 15:14:06 first thing I do in such a case is DOUBLE-check Oct 06 15:14:55 Pali: do you expect me to apply the patch build it? Oct 06 15:15:01 preferrably with other user doing check on another machine Oct 06 15:15:37 (the build, not the texting on device) Oct 06 15:15:49 testing* Oct 06 15:15:57 DocScrutinizer05: I am almost sure we have some atag we're missing. The one u-boot is not aware of Oct 06 15:16:09 sounds plausible Oct 06 15:16:30 build is here: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/ Oct 06 15:16:45 hmm, ATAG_CMDLINE Oct 06 15:17:08 I missed ATAG_CNDLINE in some of the pastes of merlin1991 Oct 06 15:17:12 we using cmdline from kernel image Oct 06 15:17:25 aah, yes Oct 06 15:17:51 Pali: where does that dump the tags to? Oct 06 15:17:58 to screen Oct 06 15:18:04 still it might e.g mess up parsing of other ATAG parameters Oct 06 15:18:06 Pali: do you know what this atag-dumper will do in case of unknown tag? Oct 06 15:18:24 it write unknown Oct 06 15:18:29 or it write hexdump Oct 06 15:18:48 do we have hexdumps? Oct 06 15:18:48 "unknown tag" for unknown tag Oct 06 15:18:52 I'd love to see raw hexdump of whole buffer area for both cases Oct 06 15:18:57 yeah Oct 06 15:19:02 and hexdump if there is something unknown in omap table Oct 06 15:19:45 DocScrutinizer05: I only had kernel cmdlines for rescueos, but stock kernel and attached that did boot (on 2101) has no cmdline atag Oct 06 15:19:51 e.g figure CMDLINE terminated by 0x0a instead 0x0d Oct 06 15:20:17 DocScrutinizer05: it is build-in cmdline that is used Oct 06 15:20:23 might result in next parameter not get parsed Oct 06 15:20:38 not the one passed from bootloader Oct 06 15:20:48 there's shitloads of tags Oct 06 15:20:54 they don't even fit on the screen Oct 06 15:21:26 merlin1991: use your second device to capture a video Oct 06 15:21:32 or third :P Oct 06 15:21:34 :D:D Oct 06 15:21:50 Pali: I am not kidding ;) Oct 06 15:22:03 freemangordon: you got my idea that missing proper EOL termination of ATAG N-1 will ruin ATAG N? Oct 06 15:22:09 it is funny, what you can do with 3 n900 :-) Oct 06 15:22:23 DocScrutinizer05: I am not sure atags are eol-ed Oct 06 15:22:35 they are terminated I'd hope Oct 06 15:22:42 but we really need hexdump Oct 06 15:22:45 or is this a static array? Oct 06 15:22:51 af on the n900 camera sucks ass Oct 06 15:22:51 afaik Oct 06 15:22:53 yep, hexdump FTW Oct 06 15:23:28 Pali: do we have hexdups for ok abd bad cases? Oct 06 15:23:34 *and Oct 06 15:24:21 and syslog/dmesg log from non-booting system Oct 06 15:24:33 http://paste.ubuntu.com/1263508/ Oct 06 15:25:42 this is BAD by _PanzerSajt ^^^ Oct 06 15:26:01 booting ubuntu Oct 06 15:26:10 * DocScrutinizer05 points at OOPS area Oct 06 15:26:40 mtd_oops Oct 06 15:27:21 aka mtd2: 00040000 00020000 "log" Oct 06 15:27:38 DocScrutinizer05: ? Oct 06 15:28:02 dunno if there's a simple lightweight way to exploit it for uBootlog Oct 06 15:28:39 aiui there's a kernel module mtd_log.ko or mtd_oops.ko Oct 06 15:29:21 might be worthwhile to rip a few sourcode lines out there and implant it to uBoot Oct 06 15:29:39 strictly for R&D! Oct 06 15:31:15 http://tomoyo.sourceforge.jp/cgi-bin/lxr/source/drivers/mtd/mtdoops.c Oct 06 15:32:11 kernel ooops == kernel panic? Oct 06 15:33:48 basically yes Oct 06 15:34:03 http://tomoyo.sourceforge.jp/cgi-bin/lxr/source/drivers/mtd/mtdoops.c#L55 Oct 06 15:34:59 merlin1991: a kernel OOPS allows kernel to continue business, while a panic is supposed to kill off kernel for good Oct 06 15:35:19 Pali: 32wd_to? Oct 06 15:37:52 Pali: anything else I can do to help you? Oct 06 15:38:15 Pali: I 32wd_to in atags on my device Oct 06 15:38:21 *I have Oct 06 15:39:09 for bootreason? Oct 06 15:39:29 no, as omap atag Oct 06 15:39:39 which value? Oct 06 15:39:41 freemangordon, which omap atag? Oct 06 15:39:52 it should be bootreason reset by watchdog Oct 06 15:40:03 no, boot-mode normal Oct 06 15:40:24 I am reading a hexdump Oct 06 15:40:24 merlin1991, provides hexdump (or xxd) of /proc/atags Oct 06 15:40:38 freemangordon, boot-mode is property and normal is value Oct 06 15:41:07 it is in omap tag "version" in omap table Oct 06 15:41:07 yes, but I have property 32wd_to, don't know how to check the value Oct 06 15:41:12 yes Oct 06 15:41:25 it should be bootreason Oct 06 15:41:33 bootreason is omap tag in omap table Oct 06 15:41:41 "version" is also omap tag in omap table Oct 06 15:41:52 Pali: from stock kernel, rescueoss or where? Oct 06 15:41:59 Pali: actually not sure, as it is between rootfs and product Oct 06 15:42:03 "version" tag has two properties: 1) name 2) value Oct 06 15:42:27 rootfs is part of omap tag layout Oct 06 15:42:30 * DocScrutinizer05 waves Oct 06 15:42:39 merlin1991, as DocScrutinizer05 from all Oct 06 15:42:48 DocScrutinizer05, do you look at hexdumps? Oct 06 15:43:15 err, nope, I don't have any hexdumps Oct 06 15:43:41 merlin1991 will dump it Oct 06 15:44:05 I guess I'll be afk when he does Oct 06 15:44:43 freemangordon, open some merlin text dump and look why is bootreason between rootfs and product :-) Oct 06 15:44:47 http://cdnm.at/~christian/maemo/atags/rescueos_2204.txt Oct 06 15:44:58 aah now I see what you meant Oct 06 15:45:02 Pali: which version of uboot? Oct 06 15:45:10 Pali: aah, ok :D Oct 06 15:45:44 merlin1991, one uboot which can boot device, one uboot which cannot boot and one for direct boot (without uboot) Oct 06 15:46:18 http://pastebin.ca/2245676 o/ cya Oct 06 15:46:37 no uBoot, normal stock kernel and CSSU Oct 06 15:46:46 o/ Oct 06 15:47:50 well, that's 2101 Oct 06 15:48:04 the hw revision that works :D Oct 06 15:48:39 do you need a 2104? Oct 06 15:49:04 hmm, my secon device is 2204 Oct 06 15:49:11 *second Oct 06 15:49:20 http://pastebin.ca/2245679 2104 Oct 06 15:49:32 2104 with kp51 Oct 06 15:51:33 I honestly don't get the catch how ATAG needs battery removal to recover Oct 06 15:51:57 is a reboot not even executing bootloader at all? Oct 06 15:52:22 (a sidenote without hope for answer, since I'm afk now) Oct 06 15:53:40 freemangordon, merlin1991, any clue why ubifs cannot be mounted? Oct 06 15:53:55 nah Oct 06 15:53:57 I think we are sure now that atags are ok Oct 06 15:54:01 or not? Oct 06 15:54:19 apart from the mem 0 tags they were equal to the ones on stock maemo kernel Oct 06 15:55:11 can you test that uboot which deleting mem 0 again (with removing battery)? Oct 06 15:55:12 so it's trustzone *again*? Oct 06 15:55:15 ubimkvol /dev/ubi0 -m -N rootfs? Oct 06 15:55:43 DocScrutinizer05: actually for ie -11 it's enough if I let if fail the first boot and just boot again Oct 06 15:55:48 DocScrutinizer05, what from trustzonoe? Oct 06 15:56:02 * DocScrutinizer05 recalls some TZ related method to (write/rw-)lock NAND partitions Oct 06 15:56:38 is this info/code public? Oct 06 15:56:41 Pali: merlin1991: atags are ordered in a different way Oct 06 15:56:52 probably only get reset by full reboot (security shit) Oct 06 15:57:01 freemangordon, order is not important Oct 06 15:57:18 and order in omap table is same Oct 06 15:57:49 opam yes Oct 06 15:57:52 *omap Oct 06 15:58:09 but working version has also different order Oct 06 15:58:11 <_PanzerSajt> Pali, what was the problem with my paste? Oct 06 15:58:19 Pali: ok Oct 06 15:58:21 _PanzerSajt, no Oct 06 15:58:32 _PanzerSajt, we did not find where is problem yet... Oct 06 15:58:37 <_PanzerSajt> oh sorry Oct 06 15:58:52 <_PanzerSajt> can I help you in any way? Oct 06 15:59:00 ubifs_get_sb cannot open "ubi0:rootfs" error -19 Oct 06 15:59:13 Pali: ^^ with the patched u-boot that only has one mem atag Oct 06 16:00:09 same here: http://imageshack.us/a/img839/674/img2012100611100165.jpg Oct 06 16:00:23 this is screen by _PanzerSajt Oct 06 16:00:36 I'd love to see the part about avaiable partitions Oct 06 16:00:37 :D Oct 06 16:00:40 and this is last unpatched version of u-boot Oct 06 16:01:23 there is no available partition Oct 06 16:01:46 <_PanzerSajt> Pali, how can I flash the original kernel and/or u-boot Oct 06 16:02:05 <_PanzerSajt> but only kernel not the full rootfs Oct 06 16:02:23 _PanzerSajt: you can get the Fiasco image and flash with "flasher-3.5 -F $fiascoimage -f --flash-only=kernel Oct 06 16:02:41 merlin1991: is there any way to see kernel cmdline? Oct 06 16:03:16 Pali: did you possibly update your arm compiler or binutils between 11 and 12 ? Oct 06 16:03:36 no, because I compiled both version today Oct 06 16:03:51 on same pc with same tools Oct 06 16:03:54 merlin1991: see my urgent suggestion about DOUBLEcheck Oct 06 16:04:08 * merlin1991 starts to seriously wonder WHY 11 works and 12 not Oct 06 16:04:42 maybe we're down to *real* bughunting now, means "find the fangling mempointer" Oct 06 16:04:48 dangling* Oct 06 16:05:23 inser one char into a version string that only gets printed out -> no more work Oct 06 16:05:25 real fun Oct 06 16:05:35 merlin1991, can you give me (again) link for atags from -11 and atags from -12 Oct 06 16:06:03 <_PanzerSajt> reflashing kernel only --> bootloop Oct 06 16:06:16 err did you kill the modules? Oct 06 16:06:32 did you use the right fiasco image? Oct 06 16:06:39 Pali: are we sure nand is powered up? Oct 06 16:06:46 <_PanzerSajt> the fiasco image is the right one Oct 06 16:07:00 freemangordon, both version -11 and -12 does not have nand code enabled Oct 06 16:07:15 _PanzerSajt: which version is the fiasco image? Oct 06 16:07:18 onenand code is enabled in -15 (= last version) Oct 06 16:07:23 Pali: naah, i meant the power to the chip Oct 06 16:07:36 merlin1991, I have something... Oct 06 16:07:43 last version has included onenand support Oct 06 16:07:43 ... Oct 06 16:07:53 you can read something from nand Oct 06 16:07:56 in uboot console Oct 06 16:08:01 can you try if it works? Oct 06 16:08:07 sure Oct 06 16:08:09 how? Oct 06 16:08:19 there should be command onennad Oct 06 16:08:21 <_PanzerSajt> merlin1991, RX-51_2009SE_20.2010.36-2_PR_COMBINED_MR0_ARM.bin Oct 06 16:08:35 hm that's the right one Oct 06 16:08:57 Pali: did you do make clean? Oct 06 16:09:06 for -12 Oct 06 16:09:10 yes Oct 06 16:09:20 dpkg-buildpackage build it in separate dir Oct 06 16:09:23 _PanzerSajt: you could flash http://atrey.karlin.mff.cuni.cz/~pali/u-boot/old/u-boot-11.fiasco that's the u-boot version that still works on my device Oct 06 16:09:59 Pali: so I should use -15 and see if I can use onennand ? Oct 06 16:10:01 <_PanzerSajt> merlin1991, sure what is the exact command? Oct 06 16:10:14 merlin1991, yes Oct 06 16:10:16 flasher-3.5 -F u-boot-11.fiasco -f -R Oct 06 16:10:31 hm, i realized that i don't need bluetooth DUN at all Oct 06 16:10:37 and i'd rather have bluetooth PAN working Oct 06 16:10:46 but it's a fucking mess Oct 06 16:10:55 as vi_ once said Oct 06 16:11:25 hm onenand info gives me "onenand0" Oct 06 16:12:35 onenand --help Oct 06 16:12:38 or something Oct 06 16:13:15 <_PanzerSajt> merlin1991, still no good Oct 06 16:13:16 merlin1991, in uboot is also support for ubifs, but due to big size is disabled Oct 06 16:13:36 I can enabled it, but image will be without stock kernel Oct 06 16:14:01 Pali: the usage info is so obscure, that I don't know what I should use :D Oct 06 16:14:03 _PanzerSajt, maybe you have corrupted rootfs/maemo5 Oct 06 16:14:29 merlin1991, i will run uboot in qemu and look at it Oct 06 16:15:05 usage info? onenand? where? how? Oct 06 16:15:12 DocScrutinizer05: in uboot Oct 06 16:15:15 aaah Oct 06 16:15:42 uBoot, the wanna-be-OS Oct 06 16:16:06 ;-) Oct 06 16:16:33 wait until they stop supporting the actual boot of the other OSes =) Oct 06 16:16:44 hehehehehe Oct 06 16:16:50 lol Oct 06 16:17:05 wait until they integrate systemd Oct 06 16:17:12 will probably say that people who want to boot other OSes are "not their target audience, uboot is not for them" Oct 06 16:17:57 actually uBoot and systemd should go together nicely, regarding this attitude ;-) Oct 06 16:18:59 if uBoot was any "sane", they already had included proper DOS shell ;-P Oct 06 16:19:46 <_PanzerSajt> I'm giving up. Have to reflash whole emmc, but not anytime soon Oct 06 16:19:58 uh? Oct 06 16:20:00 DocScrutinizer05: it's still way better than multiboot ;) Oct 06 16:20:09 will reflashing COMBINEd not suffice? Oct 06 16:20:26 merlin1991: no doubt Oct 06 16:20:32 <_PanzerSajt> but I prefer clean installation Oct 06 16:20:46 then yes, you're in for full reflash Oct 06 16:21:04 <_PanzerSajt> and I don't have anything important on my n900 Oct 06 16:21:06 hi Oct 06 16:21:24 * DocScrutinizer05 is afk again Oct 06 16:21:39 Is there sombody who developed something for n900? Oct 06 16:23:14 m4g0g: quite a few in here developed *something* for n900 Oct 06 16:23:30 like, an "unhealthy affection" for example... Oct 06 16:24:00 :3 Oct 06 16:24:08 it's not unhealthy! Oct 06 16:24:51 I try to develop one app for n900 and have question: how to test project via qt-creator which depends on library Oct 06 16:25:28 when I try to run app from qt-creator on my n900 I get: Oct 06 16:25:30 m4g0g: you can install the library to the madde sysroot Oct 06 16:25:36 dpkg-shlibdeps: failure: no dependency information found for /home/magog/Programming/QtSDK/Maemo/4.6.2/sysroots/fremantle-arm-sysroot-20.2010.36-2-slim/usr/lib/libkqoauth.so.0 (used by debian/moonreader/opt/moonreader/bin/moonreader). Oct 06 16:25:36 dh_shlibdeps: command returned error code 512 Oct 06 16:25:36 make: *** [binary-arch] Error 1 Oct 06 16:26:05 if you want to test on device you can simply create an overrides file for shlibdpes Oct 06 16:26:09 * merlin1991 checks how that worked Oct 06 16:26:44 RST38h: it's hard not to feel affection for the best phone ever made Oct 06 16:26:57 see? that is what I mean Oct 06 16:27:06 merlin1991: any chance to capture boot video? Oct 06 16:27:11 Is there any howtos or manual about this magic with "creating an overrides file for shlibdpes" ? Oct 06 16:27:17 m4g0g: Do not use qt-creator. Use command line. Oct 06 16:27:23 m4g0g: there is http://man.he.net/man1/dpkg-shlibdeps Oct 06 16:27:27 Also, do not use Madde. Use MaemoSDK Oct 06 16:27:36 merlin1991: to see what cmdline tries kernel to execute and what mtd devices it sees Oct 06 16:27:45 RST38h: what do you mean? Oct 06 16:27:54 I mean what I have said. Oct 06 16:28:21 m4g0g: and http://www.debian.org/doc/debian-policy/ch-sharedlibs.html#s-sharedlibs-shlibdeps Oct 06 16:29:58 m4g0g: basically it's enough to have a shlibs.local file in your packaging dir (qt_packaging_something) with the name of the library in it Oct 06 16:30:19 it tells dpkg-shlibdpes to ignore all dependencies for this library Oct 06 16:30:35 obviously not the "right way (tm)" but good enough for quick tests Oct 06 16:31:25 m4g0g: so in your case you need to have libkqoauth in your shlibs.local file Oct 06 16:31:28 merlin1991: ? Oct 06 16:31:56 freemangordon: I'll try Oct 06 16:34:53 freemangordon: well: http://cdnm.at/~christian/maemo/12100003.mp4 Oct 06 16:35:00 :) Oct 06 16:36:21 freemangordon, what about start patching kernel? Oct 06 16:36:41 prink everyhwere where is return -ESOMETHING in ubifs Oct 06 16:36:56 merlin1991, just for test, can you try to boot kernel-power? Oct 06 16:37:11 hehe have to install that first :D Oct 06 16:39:18 Pali: which package did adjust the bootscript to use uname -a for the modules dir? Oct 06 16:40:05 kernel-power (-flasher maybe?) Oct 06 16:41:29 merlin1991: this video is missing first line from boot process :( Oct 06 16:41:33 *lines Oct 06 16:41:45 can we get dmesg somehow? Oct 06 16:42:34 freemangordon, connect to serial console and set that R&D flag which route dmesg output to serial consle... Oct 06 16:42:51 or start patching kernel... Oct 06 16:42:57 Pali: and serial console is under the battery? Oct 06 16:43:03 port even Oct 06 16:43:18 freemangordon, no idea... Oct 06 16:43:31 I'm using it in qemu Oct 06 16:43:39 but that's cheating! Oct 06 16:45:10 okay kp51 installed, reflashing u-boot Oct 06 16:46:27 nope does not work Oct 06 16:50:22 Pali: can you do some memcmp between nolo passed atags and uboot atags? and print differences Oct 06 16:50:50 diff is ok Oct 06 16:50:56 binary dump I do not have Oct 06 16:51:13 in uboot code Oct 06 16:51:37 freemangordon, not easy Oct 06 16:51:47 because atags must start at one address Oct 06 16:52:21 can't you dup nolo atags to some buffer Oct 06 16:52:45 merlin1991: so. I add this line in shlibs.local : libkqoauth 0 kqoauth Oct 06 16:53:04 you sure the library version is 0 ? Oct 06 16:53:07 install libkqoauth-0.96 on n900 Oct 06 16:53:33 http://codepad.org/Le3tfJl6 Oct 06 16:53:40 but got this error Oct 06 16:54:02 * freemangordon is afk Oct 06 16:55:08 just build it with the line lbkqoauth (nothign else) in shlibs.local Oct 06 16:55:47 m4g0g: how did you install the lib on the n900? Oct 06 16:55:55 via apt / dpkg or just dropped the binary? Oct 06 16:56:00 apt-get install libkqoauth Oct 06 16:56:14 libkqoauth-dev? Oct 06 16:56:30 you think I need dev? Oct 06 16:56:32 WTF Oct 06 16:56:42 there is kqoauth and libkqoauth in extras-devel Oct 06 16:56:43 lol Oct 06 16:56:58 no Oct 06 16:57:06 libkqouath and libqoauth Oct 06 16:57:11 m4g0g: well if you install libkqoauth your shlinbs.local file should be libkqoauth 0 libkqoauth Oct 06 16:57:20 yep Oct 06 16:57:32 and not only kqoauth in the end Oct 06 16:57:45 I am testing Oct 06 16:58:27 just sayin: http://maemo.org/packages/view/kqoauth/ and http://maemo.org/packages/view/libkqoauth/ Oct 06 16:58:47 The same guy! repacakged a higher version in a different package name Oct 06 16:58:48 wtf! Oct 06 16:59:06 thx Oct 06 16:59:09 all works fine Oct 06 16:59:28 merlin1991: there is different packages Oct 06 16:59:37 the second depends on qca2 Oct 06 16:59:45 but the first doesn't Oct 06 17:27:40 Pali: http://git.infradead.org/battery-2.6.git/blob/HEAD:/drivers/power/isp1704_charger.c#l458 has some hint about what's the problem, though the comment doesn't really provide the proper solution Oct 06 17:27:50 merlin1991, can you test this uboot image? http://atrey.karlin.mff.cuni.cz/~pali/u-boot/u-boot-new.bin Oct 06 17:28:00 (it is not fiasco! but kernel) Oct 06 17:28:16 arf I just booted into windows Oct 06 17:28:18 * merlin1991 reboots Oct 06 17:29:16 this is my last version + I disabled generating atags code - it reuse full from nolo Oct 06 17:29:25 so you can boot only maemo Oct 06 17:30:08 why "only" maemo? Oct 06 17:30:21 because it pass all atags from nolo Oct 06 17:30:30 and other kernels don't like that? Oct 06 17:30:31 not possible to modify it (e.g cmdline...) Oct 06 17:30:51 oh Oct 06 17:31:08 I need to know if problem is in that code (which generating atags) or not Oct 06 17:31:10 ARF Oct 06 17:31:34 thats all what can I do... Oct 06 17:32:34 that new one is entirely broken, I don't get past the Nokia text Oct 06 17:33:44 actually the stock kernel refues to boot now too Oct 06 17:34:04 I included MAG kernel with framebuffer Oct 06 17:34:20 i never get to the bootmenu Oct 06 17:34:25 ah... Oct 06 17:34:47 ok, I will patch -15 Oct 06 17:35:02 might have been a weird issue though Oct 06 17:35:20 I flashed the zImage and it didn't get past Nokia text either Oct 06 17:35:29 then pulled the battery for a longer time and it works Oct 06 17:35:34 flashing your u-boot again Oct 06 17:37:37 and booting atatched kernel? Oct 06 17:38:05 nah, seems like my inital statement was right, I don't get past the nolo screen with this version of uboot Oct 06 17:39:45 merlin1991, try this: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/u-boot-new2.fiasco Oct 06 17:40:02 it is -15 with disabled atags generating Oct 06 17:42:48 same kernel panic Oct 06 17:43:43 so atags are OK for sure! Oct 06 17:43:53 freemangordon, DocScrutinizer06, ^^^ Oct 06 17:43:58 hmm Oct 06 17:44:13 where the fuck is the problem then? the diff 11 to 12 wasn't exactly huge Oct 06 17:44:17 * DocScrutinizer06 doesn't know what to say or what to do Oct 06 17:44:36 happy dangling pointer hunting! ;-P Oct 06 17:46:19 it's already time for advanced bisecting then, commenting out 50% of the already small (aiui) patch that makes diff between 11 and 12 Oct 06 17:48:01 yo Oct 06 17:48:07 if you end with commeting out *all* of that patch and just replacing some NOP-function codeline (that doesn't get optimized out by compiler nevertheless!!) and it still breaks, you know what time it is Oct 06 17:48:20 * vi__ PPSSSSHHHT Oct 06 17:48:29 * vi__ cracks a beer Oct 06 17:48:51 peanut butter and jelly time? Oct 06 17:48:57 DocScrutinizer06: what time would that be, grab your hair dance around and light the n900 on fire time? Oct 06 17:48:59 good plan Oct 06 17:49:10 nah Oct 06 17:49:12 happy dangling pointer hunting! ;-P Oct 06 17:49:16 ohai vi__ Oct 06 17:49:17 time Oct 06 17:49:26 yo Oct 06 17:49:32 the weirdest thing is, that it works on 2101 Oct 06 17:49:40 vi__: found a sensible way to do PAN? Oct 06 17:49:50 hmm Oct 06 17:50:01 no, I am fixing motorcycles this weekend. Oct 06 17:50:06 hm Oct 06 17:50:22 so maybe the dangling pointer nukes something along proper setup of NAND RAM banks? Oct 06 17:50:26 brb, BT KB Oct 06 17:54:47 kerio: bt pan. Oct 06 17:54:56 bt pan. Oct 06 17:55:20 what I want is an IP network over BT. Oct 06 17:55:27 d Oct 06 17:55:29 merlin1991, try this: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/u-boot-new3.fiasco Oct 06 17:55:39 ffs btkb suck my nuts Oct 06 17:56:02 it has also disabled thumb code + disabled unused code (which was used for generating omap atags) Oct 06 17:56:05 vi__: there's a pc-connectivity-manager that fucks with your bluetooth but gives you a PAN, in theory Oct 06 17:57:30 o Oct 06 18:00:10 ffs Oct 06 18:00:14 goddamit Oct 06 18:00:25 jesus titty fucking christ Oct 06 18:00:35 POS bt keyboard Oct 06 18:01:19 now my n900 kb thinks it has dthe same geometry as the BT keyvboard. Oct 06 18:01:36 so none of the keys work. Oct 06 18:01:47 FML Oct 06 18:03:16 how do you reset the keyboard geometry Oct 06 18:03:21 hahahaha Oct 06 18:03:28 setxkbsomething Oct 06 18:03:48 that's what you get for not using a HID keyboard and not using extkbd Oct 06 18:03:51 setxkbmap Oct 06 18:04:04 vi__: never ever execute setxkbmap Oct 06 18:04:10 ...wat Oct 06 18:04:16 it screwes internal kbd layout Oct 06 18:04:23 ...WAT Oct 06 18:04:32 kerio: ^^^ Oct 06 18:04:34 i used to do that all the time when i were remapping my kbd Oct 06 18:04:34 setxkbmap only sets keymap, not the geometry. Oct 06 18:04:57 vi__: have in mind I am axtkbd author ;) Oct 06 18:05:07 and I played a lot with this stuff Oct 06 18:05:14 *extkbd Oct 06 18:05:40 freemangordon: what hal magic must I execute to get my kb back? Oct 06 18:05:54 vi__: which one? external or internal Oct 06 18:06:23 iirc there should be hildonsetxkbmap or something Oct 06 18:06:43 n900 keyboard Oct 06 18:06:44 there is control panel applet for it Oct 06 18:06:47 Pali: what now, it is not atags? Oct 06 18:06:58 freemangordon, waiting for merlin result Oct 06 18:07:02 vi__: yeah, cpl should do the job too Oct 06 18:07:04 n900 keyboard is now mapped the same as my external bt keyboard Oct 06 18:07:06 paroneayea: ok Oct 06 18:07:15 I need last test from merlin1991 Oct 06 18:07:18 vi__: that is because of setxkbmap Oct 06 18:07:28 Pali: ^^^ Oct 06 18:07:57 freemangordon, problem is not in atag struture Oct 06 18:07:59 it is ok Oct 06 18:08:06 yeah, i got that Oct 06 18:08:08 but problem can be in other code... Oct 06 18:08:23 overwriting some nand init data Oct 06 18:08:25 ? Oct 06 18:08:32 do not know... Oct 06 18:08:37 ping merlin1991 Oct 06 18:08:39 but seems logical Oct 06 18:08:41 are you here? Oct 06 18:09:03 kerio: does geyes 308e look any good? Oct 06 18:11:27 obviously merlin1991 is not here :D Oct 06 18:11:35 I need a few mins :D Oct 06 18:11:42 ok :-) Oct 06 18:13:30 so i have to reboot to reset kb geometry. Oct 06 18:13:33 ffs Oct 06 18:13:42 f everything. Oct 06 18:14:40 Pali: is reuse_atags call needed? Oct 06 18:15:29 yes, because this is used for bootreason, bootmode and hwrev Oct 06 18:15:31 scratch that Oct 06 18:15:35 yeah, Oct 06 18:15:45 but I builded last image which has commented that code Oct 06 18:16:14 Pali: hmm, recursion? Oct 06 18:16:31 is that allowed? Oct 06 18:16:40 no Oct 06 18:16:53 it will failed to boot and uboot will freeze Oct 06 18:17:07 aah, my bad, reuse_atags calls reuse_omap_atags, not itself Oct 06 18:17:42 kerio: look up 'geyes 308e' Oct 06 18:18:25 kerio: have you installed libicd-network-dummy? Oct 06 18:18:30 ke Oct 06 18:19:12 I installed pc-connectivity-manager but I have literally no idea htf to work it. Oct 06 18:20:05 vi__: iirc it is broken on fremantle Oct 06 18:20:33 PC-C-M? Oct 06 18:20:41 Pali: still same bug Oct 06 18:20:55 :-( Oct 06 18:22:57 vi__: i have libicd-network-null Oct 06 18:23:24 Sicelo: I installed that but it made my list of networks empty. Oct 06 18:23:45 I uninstalled it and they all came back. Oct 06 18:23:52 any tips? Oct 06 18:24:05 something's wrong with your setup.. Oct 06 18:24:18 vi__: damn, that's one sexy keyboard Oct 06 18:24:21 :-/ Oct 06 18:24:38 are the keys fullsize? Oct 06 18:24:51 kerio: I cannot work out if it is SPP or HID though. Oct 06 18:25:01 I think it is fullsize. Oct 06 18:25:09 oh, you don't have it Oct 06 18:25:10 :c Oct 06 18:25:57 was thinking of getting it, if it is HID and HID will eliminate all my BT KB issues (due to having SPP shit) Oct 06 18:27:19 In Theory™ installing extkbd will make HID bluetooth keyboards work flawlessly Oct 06 18:27:38 ...I have extkb installed for my SPP. Oct 06 18:27:49 huh? why? Oct 06 18:27:56 It does not work without extkb. Oct 06 18:27:57 i didn't know extkbd supported them Oct 06 18:28:15 it does not Oct 06 18:28:18 oh, but you also have to run kbdd manually, right Oct 06 18:28:27 or whatever the daemon is called Oct 06 18:28:29 yes Oct 06 18:28:34 like a fucking dog Oct 06 18:28:48 haha Oct 06 18:29:03 xterm is the only APPZ i need. Oct 06 18:29:07 no, HID keyboards appear in the extkbd list and you connect to them via the standard bluetooth menu Oct 06 18:29:41 the SPP KB appears as 'ser keyboard' in extkb. Oct 06 18:29:51 I assume ser=serial. Oct 06 18:29:53 oh, ok Oct 06 18:29:55 so what's the problem? Oct 06 18:30:09 my geometry goes to cock. Oct 06 18:30:28 just like yer mom herp derp Oct 06 18:30:29 my n900 KB takes obn the ext kb geometry and nothing works. Oct 06 18:30:41 aaah. Oct 06 18:30:43 Pali: are you sure your build tree is ok? Oct 06 18:30:47 you win this round... Oct 06 18:30:53 do you do any manual setting with setxkbmap? Oct 06 18:30:54 yes Oct 06 18:31:04 nope Oct 06 18:31:21 hm Oct 06 18:31:30 \_o_/ Oct 06 18:31:39 Pali: if all of the code is commented, I don;t see why it shouldn't work Oct 06 18:32:03 merlin1991: did you do battery removal? Oct 06 18:32:16 EE troubleshooting steps: Oct 06 18:32:31 1. is it switched on? Oct 06 18:32:37 2. is it plugged in? Oct 06 18:32:50 3. is it plugged in the right way around? Oct 06 18:33:36 freemangordon, here is patch agains git: http://pastebin.com/mgBTwgiE Oct 06 18:34:11 kerio: do you use pastebinit? Oct 06 18:35:21 no Oct 06 18:35:35 oh well. Oct 06 18:35:53 Pali: hmm, that won't work Oct 06 18:36:02 unless you manually checked for the lack of nested comments Oct 06 18:36:17 use #if 0 Oct 06 18:36:21 Pali: what about die_id and cpuinfo? Oct 06 18:36:26 that can be nested Oct 06 18:36:46 it is only print someting... Oct 06 18:36:50 it is possible those are messing the rings Oct 06 18:36:52 I know Oct 06 18:36:58 but still Oct 06 18:36:59 to anyone who uses pastebinit on their n900, the deb from the debian sid repo installs. This fixes the breakage with pastebin.com Oct 06 18:37:19 dieid_num_r() is in -11 too Oct 06 18:37:30 print_cpuinfo() is only in -12 Oct 06 18:37:53 comment that Oct 06 18:40:19 kerio: hah, nested comments :-D Something I appreciated in modula Oct 06 18:40:24 iirc Oct 06 18:41:26 updated: http://atrey.karlin.mff.cuni.cz/~pali/u-boot/u-boot-new3.fiasco Oct 06 18:41:44 merlin1991 ^^ Oct 06 18:41:54 is there a KB key to go from tasknav to the desktop? Oct 06 18:42:01 sry Pali can't test it now Oct 06 18:42:10 ok Oct 06 18:42:17 hm wait if I grab my notebook I could Oct 06 18:42:18 vi__: it can be configured to use ctrl+bksp to do that too Oct 06 18:42:20 * merlin1991 goes searching Oct 06 18:42:31 *h-d Oct 06 18:42:32 kerio: but indeed #if 0 are nice to use for that Oct 06 18:42:55 :3 :3 :3 Oct 06 18:43:14 ??? Oct 06 18:43:23 i got my ass bitten more than once for disabling code with /* */ Oct 06 18:43:30 :-/ :-S :-\ Oct 06 18:43:47 because there *was* a multiline comment inside Oct 06 18:43:58 haha Oct 06 18:44:15 kerio: lol Oct 06 18:44:26 general rule: don't span comments over several lines, NEVAR Oct 06 18:44:45 I enjoy ctrl+bckspace as alt-tab too much. Oct 06 18:44:57 but that clashes with the general rule of wrapping lines at 79chars Oct 06 18:45:13 how's that? Oct 06 18:45:34 you have to make big comments on a single line Oct 06 18:45:40 /* you can comment*/ Oct 06 18:45:43 /*as many lines as you want*/ Oct 06 18:45:47 oic Oct 06 18:46:12 how will that help with disabling code via /* */? Oct 06 18:46:42 How well does android use an external KB? Oct 06 18:47:28 well, the droid/milestone has a physical kbd Oct 06 18:47:37 merlin1991, you must write that now uboot is working!! Oct 06 18:47:38 :D Oct 06 18:47:58 Pali: even if it was a lie? Oct 06 18:48:07 also, who will trade me 2 mint condition n900 for an n950? Oct 06 18:48:15 d Oct 06 18:48:19 WTF Oct 06 18:48:23 s Oct 06 18:48:36 so i have just isolated the bug. Oct 06 18:48:39 vi__: i will trade you two mints Oct 06 18:49:14 The BT KB only works properly when the n900 KB is open... Oct 06 18:49:20 haha Oct 06 18:49:21 neat Oct 06 18:49:39 i *did* hear something like that, yeah Oct 06 18:49:46 Pali: same old :/ Oct 06 18:49:48 OH Oct 06 18:49:54 NOW YOU SAY. Oct 06 18:50:18 well it didn't do shit for me Oct 06 18:50:26 yeah Oct 06 18:50:30 so i classified that under "tmo advice" Oct 06 18:50:38 you with yopur magic HID keyboard Oct 06 18:50:49 u jelly Oct 06 18:50:51 all working keys and blowjobs. Oct 06 18:51:10 and to be fair, two greenfn+key combos don't work Oct 06 18:51:17 the ones for € and for yen Oct 06 18:51:59 freemangordon, I do not know where is problem... Oct 06 18:52:18 brb' need moar booze. Oct 06 18:52:25 I'm closing uboot tabs in konsole... Oct 06 18:52:39 (( how will that help with disabling code via /* */?)) well, same rules apply for this case as well Oct 06 18:52:40 today was epic fail :-( Oct 06 18:52:48 freemangordon: ? Oct 06 18:53:07 Pali: no it was not. Oct 06 18:53:18 yeah, vi__ got his keyboard working properly! Oct 06 18:53:27 You have eliminated several possibilities. Oct 06 18:53:43 kerio: my kb is still utter fail. Oct 06 18:53:48 I eliminated everything Oct 06 18:54:08 cool, so now you don't have any problems left Oct 06 18:54:32 and still uboot not working... Oct 06 18:54:41 no, uboot is eliminated too Oct 06 18:54:51 >:D Oct 06 18:55:41 Pali: see backscroll, all my comments about dangling pointers Oct 06 18:55:56 Pali: absolutely not anything unheard Oct 06 18:56:22 Pali: arbitrary change in code triggers sleeping bug elsewhere Oct 06 18:56:56 pointer pointing to random location, nuking different shit in every new build Oct 06 18:56:57 Pali: id email you a beer and a packet of smokes... Oct 06 18:57:28 ...stupid startrek replicator Y U NO EXIST YET?? Oct 06 18:57:40 Pali: alas this class of bugs is almost impossible to track down Oct 06 18:57:54 Pali: try lint etc Oct 06 18:58:30 try proper warning settings with gcc Oct 06 18:58:50 another option, merlin1991, can you compile uboot on your pc? Oct 06 18:59:03 not atm, tomorrow Oct 06 18:59:08 ok Oct 06 18:59:25 freemangordon, what about start patching kernel ubi driver? Oct 06 19:00:01 I do not see what we can do next... Oct 06 19:01:17 probably won't lead to anywhere, since whatever the culprit, the result -19 in ubi driver is a lateral effect I'd bet Oct 06 19:01:38 valgrind dat shit Oct 06 19:01:50 but yeah, we should do it nevertheless Oct 06 19:03:19 jacekowski: I need some help. I rewrote some gcc inline assembly to be thumb compatible, but as it is a very critical piece of code I need someone to review it Oct 06 19:03:35 Pali: other approach: try your 11 source with new build, see if it still works (or already change of build date causes same or other errors). Then apply minimal changes to that 11 src, until you can trigger error by a single-char edit Oct 06 19:03:38 it should probably work.. my N900 is one of those that can't boot stock kernel with this uboot. all kernels i had (omap-fb, kp, nemo) wouldn't. but latest nemo kernel does Oct 06 19:04:18 so it does seem plausible that kernel itself is source of problem, i think Oct 06 19:04:18 jacekowski: http://pastebin.com/8LpMQmQ6 Oct 06 19:05:15 jacekowski: ARM mode inline assembly is the original code Oct 06 19:06:13 jacekowski: the part under #ifdef __thumb__ is my thumb-compatible replacement. I need to be sure both do one and the same thing. Code works by the way Oct 06 19:07:06 Pali: I have 2204 here, tomorrow will start to play with u-boot. is everything commited? Oct 06 19:07:26 I did not commited any code Oct 06 19:07:40 seems that my last change working only in qemu Oct 06 19:08:04 (I rewritten assembler code, but on merlin device not worked) Oct 06 19:08:07 can you do the, do some "2204bug" branch so I can clone it Oct 06 19:08:24 s/the/this/ Oct 06 19:08:25 freemangordon meant: can you do this, do some "2204bug" branch so I can clone it Oct 06 19:09:03 freemangordon, use branch pali Oct 06 19:09:18 + maybe above patch from pastebin which disable anything... Oct 06 19:12:58 Pali: ok Oct 06 19:13:39 hehe, I should just send my device to pali via parcel :P Oct 06 19:14:31 there is one someting: NOLO loading kernel image at some random address (is different in qemu and on my n900) Oct 06 19:14:37 merlin1991: RDP Oct 06 19:14:45 and uboot loading kernel at same address always Oct 06 19:14:47 or was it RDA? Oct 06 19:15:28 Pali: brilliant spotting Oct 06 19:16:00 freemangordon: why do you go back to .arm at some point? Oct 06 19:16:10 so maybe nolo on other hwrev loading kernel at some other address to fix something... Oct 06 19:17:19 are you sure it's even uBoot that fails, and *not* kernel during early init phase? Oct 06 19:17:47 I really do not know where is problem... Oct 06 19:17:59 we have uboot image which working Oct 06 19:18:06 that's -11 :D Oct 06 19:18:13 maybe the dangling pointer is in kernel, and uBoot found a nice (and constant) addr to load kernel where it always crashes Oct 06 19:18:14 and with some changes not working (-12) Oct 06 19:18:49 u-boot always loading kernel at 80008000 Oct 06 19:18:50 DocScrutinizer06: +1 Oct 06 19:19:11 but when you buidling uImage, you can specify different address Oct 06 19:19:40 so next step already clear, no? Oct 06 19:20:04 DocScrutinizer06, so write other address where to load kernel Oct 06 19:20:10 buold a supposedly not working 12 rev, and change that location to something slightly different Oct 06 19:21:02 DocScrutinizer06, I think we should need why ubi failed Oct 06 19:21:20 it is because ubi cannot read nand memory? Oct 06 19:21:37 or it is because that it read something different? Oct 06 19:21:50 (e.g signature not correct?) Oct 06 19:22:04 or it fail because nand is locked? Oct 06 19:22:09 yep Oct 06 19:22:25 <_PanzerSajt> Pali, still no soulion? Oct 06 19:22:32 <_PanzerSajt> *solution? Oct 06 19:22:41 or because ubi want to allocate big memory and it is not possible? (uboot bad initialized ram) Oct 06 19:22:42 _PanzerSajt: this is a nasty one Oct 06 19:23:48 Pali: or simply because somebody nuked a few bytes in variables or program text RAM area of ubi Oct 06 19:24:25 DocScrutinizer06, kernel image before is started was OK Oct 06 19:24:37 uboot checking for checksum if is correct before booting image Oct 06 19:24:57 uImage is zImage + info about load address + checksum + name + ... Oct 06 19:25:31 so we are sure that uboot does not deleted/changed attached kernel image Oct 06 19:25:42 Pali: I talked about kernel nuking itself Oct 06 19:25:54 ok Oct 06 19:26:01 but then the -19 is from uBoot ubi AIUI Oct 06 19:26:09 ? Oct 06 19:26:12 is it? Oct 06 19:26:15 <_PanzerSajt> Pali, DocScrutinizer06 merlin1991 if you don't fell offended may I ask your age? Mine is 18. I'm asking this because I would like to dig deeper with programming and linux and I would like to know whether fall behind or my knowledge is adequate to my age? Oct 06 19:26:20 DocScrutinizer06, yes Oct 06 19:26:26 if not, then we don't know if uBoot is any bad Oct 06 19:26:33 check my nick ;) Oct 06 19:27:12 while I'm like 3 times as old Oct 06 19:27:49 so I couldn't judge about your coding knowlege and if it's adequate for your age, I forgot I ever been as young as you're Oct 06 19:28:36 <_PanzerSajt> I hope that I can measure my knowledge Oct 06 19:29:15 <_PanzerSajt> but this year I have to apply to university and I would like to choose which is suitable for me Oct 06 19:30:23 Pali: ((yes)) so kernel won't ever get loaded when error strikes, so no way we got any problems with kernel not liking the addr where it got loaded Oct 06 19:30:42 s/loaded/loaded+started/ Oct 06 19:30:43 DocScrutinizer06 meant: Pali: ((yes)) so kernel won't ever get loaded+started when error strikes, so no way we got any problems with kernel not liking the addr where it got loaded Oct 06 19:30:59 _PanzerSajt, look at Korotkevich. He is about 18 and has 6 gold medals from IOI :D It does not depends on your age :P Oct 06 19:31:35 <_PanzerSajt> yes but you should collet as many knowledge as you can when you're young Oct 06 19:31:47 definitely Oct 06 19:32:59 where knowledge here means useful knowledge, the ability to solve problems, strategies for doing so. A particular language is the useless class of knowledge Oct 06 19:33:24 learning about concept of multitasking or OO is useful Oct 06 19:34:21 coding something in assembler first, then do same thing in a high level language, that's useful Oct 06 19:34:47 learning about structures is useful Oct 06 19:35:42 to be honest, i'm not a big fan of inline assembly Oct 06 19:36:01 jacekowski: doesn't matter in this special case Oct 06 19:36:15 but well, Oct 06 19:36:15 he just asked for a review Oct 06 19:36:36 <_PanzerSajt> DocScrutinizer06, yes multitasking and assembly these are just big question marks for me Oct 06 19:36:48 <_PanzerSajt> :( Oct 06 19:37:46 in yur age I started learning about that stuff Oct 06 19:40:21 Pali: wait, what now was that -19 ubi error? Oct 06 19:40:24 DocScrutinizer06: yeah, but you're ooooooooold Oct 06 19:40:53 freemangordon: where that code came from? Oct 06 19:41:27 DocScrutinizer05, error is here: http://imageshack.us/a/img839/674/img2012100611100165.jpg Oct 06 19:41:28 was it "uBoot: ubi -19, can't load kernel" or was it "kernel: -19 can't mount rootfs" ? Oct 06 19:41:46 the kernel is already loaded at that point, clearly Oct 06 19:42:02 the latter iirc.. Oct 06 19:42:41 so it *may* be a bug aka dangling pointer in _kernel_ Oct 06 19:42:42 it says 'Kernel panic, not syncing.. Oct 06 19:43:40 Nemo guys fixed this in their kernel.. which never booted with this uboot until the very latest one Oct 06 19:43:49 I read "please append a correct 'root=' boot option" Oct 06 19:44:07 freemangordon: it looks ok, Oct 06 19:44:10 sounds to me like kernel cmdline is incorrect Oct 06 19:44:33 it does say that too Oct 06 19:44:39 let me get the full ine Oct 06 19:45:52 freemangordon: switch to ARM is ok, and then switch back to thumb is ok Oct 06 19:46:08 freemangordon: other than that it's the same code Oct 06 19:46:33 I see "ubifs error -19" which is before that root= error Oct 06 19:46:35 why do we need inline ASM there? Oct 06 19:46:56 freemangordon: though, i'm not sure how GCC is dealing with it Oct 06 19:47:24 yes Oct 06 19:47:30 ubifs_get_sb: -19 Oct 06 19:47:44 kernel panics because it can't mount that Oct 06 19:48:00 anybody looked into ubifs_get_sb and checked what -19 means? Oct 06 19:48:10 DocScrutinizer05, cmdline is included in zImage Oct 06 19:48:17 uboot not adding any commandline Oct 06 19:48:24 for maemo kernel Oct 06 19:48:44 <_PanzerSajt> Pali, DocScrutinizer05 thanks for the help and good luck with u-boot ;) Oct 06 19:48:54 :-) Oct 06 19:49:02 freemangordon: i would run it through disassembler to confirm that gcc is not making a mess out of it, but other than that, it looks ok Oct 06 19:49:25 <_PanzerSajt> bye Oct 06 19:49:27 good advice Oct 06 19:51:05 find the line with prink(*"cannot open* in ubifs_get_sb Oct 06 19:51:15 check what error=-19 means Oct 06 19:52:22 #define ENODEV 19 Oct 06 19:52:32 in errno.h Oct 06 19:53:57 it's not the definition that counts, but the few lines of code before the ubifs_get_sb(){...} decides to printk(... ENODEV...) Oct 06 19:53:59 NIN101, it is possible to use maemo kernel with your rescue os? Oct 06 19:54:19 hmm Oct 06 19:55:07 jacekowski: it is berkeley db Oct 06 19:55:11 maybe we should try to boot maemo kernel with initfs and here try to mount ubifs manually Oct 06 19:55:22 i.e. adressbook db Oct 06 19:55:26 and rescue os is good candicate Oct 06 19:55:51 sounds like a smart plan Oct 06 19:56:24 I am not sure, maybe it even is. Oct 06 19:56:33 jacekowski: (as you see)this part is mutex stuff, so I should be extremely careful with it :) Oct 06 19:56:52 damn netsplit :( Oct 06 19:56:56 still i'd first look up in sourcecode what exactly failed to cause this -19/ENODEV error to get thrown Oct 06 19:57:59 jacekowski: I run that through IDA and all sems well, I just need someone to reasure me it is fine :D Oct 06 19:58:32 gtg gn all. Oct 06 19:58:35 :q Oct 06 19:58:42 o/ Oct 06 20:00:14 http://mxr.maemo.org/fremantle/source/kernel/fs/ubifs/super.c#2004 Oct 06 20:00:22 DocScrutinizer05 ^^ Oct 06 20:00:36 hah, I was just searching Oct 06 20:01:46 and it returns from http://mxr.maemo.org/fremantle/source/kernel/fs/ubifs/super.c#1847 Oct 06 20:02:09 http://mxr.maemo.org/fremantle/source/kernel/fs/ubifs/super.c#1800 Oct 06 20:02:13 yeah Oct 06 20:02:21 seems I looked at this one before Oct 06 20:02:28 same issue Oct 06 20:02:38 several months ago Oct 06 20:03:20 Pali: most probably oneNAND is not loaded Oct 06 20:03:32 thus mtd does not exists Oct 06 20:03:38 and it can return from here http://mxr.maemo.org/fremantle/source/kernel/drivers/mtd/ubi/kapi.c#232 or http://mxr.maemo.org/fremantle/source/kernel/drivers/mtd/ubi/kapi.c#247 or http://mxr.maemo.org/fremantle/source/kernel/drivers/mtd/ubi/kapi.c#249 Oct 06 20:06:13 I think it failing somewhere here: http://mxr.maemo.org/fremantle/source/kernel/drivers/mtd/ubi/kapi.c#102 Oct 06 20:08:43 Pali: ubi = ubi_get_device(ubi_num);? Oct 06 20:10:51 only here is ubi code assining non null value: http://mxr.maemo.org/fremantle/source/kernel/drivers/mtd/ubi/build.c#888 Oct 06 20:11:12 btw, in dmesg we should see this output: http://mxr.maemo.org/fremantle/source/kernel/drivers/mtd/ubi/build.c#867 Oct 06 20:13:02 http://mxr.maemo.org/fremantle/source/kernel/fs/ubifs/super.c#1847 -> http://mxr.maemo.org/fremantle/source/kernel/drivers/mtd/ubi/kapi.c#211 Oct 06 20:13:05 so yeah, Oct 06 20:13:24 and I got a so terrible deja-vu, I have to stop this immediately Oct 06 20:13:34 DocScrutinizer05, it is possible to export tty console via USB? Oct 06 20:13:55 theoretically yes Oct 06 20:14:08 g_nokia.ko can do it Oct 06 20:14:17 iirc OM FR uBoot worked like that Oct 06 20:14:20 I remember that it working Oct 06 20:14:52 Uboot can export console via usb too... but there is problem that kernel musb driver crashing if uboot enabled that usb console Oct 06 20:15:03 so I disabled tty console in uboot Oct 06 20:15:18 sure, we wanna debig kernel anyway Oct 06 20:15:41 so no use in uBoot console Oct 06 20:15:44 it is possible to compile g_nokia into kernel image and set default output to usb console tty? Oct 06 20:16:00 kernel usually does that via cmdline parameter Oct 06 20:16:23 console=tty Oct 06 20:16:36 :nod: Oct 06 20:16:46 ttyACM or whatever Oct 06 20:17:05 but what will kernel do if driver g_nokia is not loaded? Oct 06 20:17:19 FUCK Oct 06 20:17:24 i can't sync via bluetooth anymore Oct 06 20:17:28 I don't think this works via g_nokia Oct 06 20:17:33 what the fucking fuck Oct 06 20:17:59 it's a genuine kernel thing Oct 06 20:18:21 musb_hdrc is monolithic Oct 06 20:19:18 the "console tty" probably is just an EndPoint on a standard ENUM session on USB, established by musb_hdrc and relatives Oct 06 20:19:54 merlin1991, freemangordon, can you try (maybe next day) to start rescue os, but with maemo kernel? Oct 06 20:20:18 it needs to repack initrd image (add maemo kernel modules) Oct 06 20:20:33 and try to mount ubifs rootfs here Oct 06 20:20:41 and see full dmesg! Oct 06 20:20:56 I think this is key for problem Oct 06 20:22:26 we goddamn had exactly same -19 error in same ubifs, looked same sourcecode lines, some months ago, in this very channel 8or #maemo-ssu) Oct 06 20:23:06 or I'm losing my mind Oct 06 20:23:06 DocScrutinizer05, I think I remember something... Oct 06 20:23:34 somebody wrote about non bootable maemo kernel... Oct 06 20:24:05 so it was this uboot problem too... Oct 06 20:24:13 :nod: Oct 06 20:24:31 this appeared in uboot 2012.04 so 6 month ago Oct 06 20:24:37 :nod: Oct 06 20:25:45 GAAAAAAAAAAAAAH WHAT THE FUCK IS WRONG WITH THIS Oct 06 20:26:03 ouch my ears Oct 06 20:26:15 the syncml is listed in sdptool Oct 06 20:27:02 ??? Oct 06 20:27:12 all of a sudden, isync doesn't work Oct 06 20:27:23 (contacts/calendar synchronization via bluetooth syncml) Oct 06 20:27:26 *burp* Oct 06 20:27:41 CSSU-T ? Oct 06 20:27:53 no modifications lately Oct 06 20:28:01 *shrug* Oct 06 20:28:09 then I'm out now Oct 06 20:29:29 i can restore an old backup, but i literally made no modifications to bluetooth or syncing since there Oct 06 20:29:31 *then Oct 06 20:29:55 people still use the n900? i have not touched me n800 in ages. sucks that nokia basicly let the n900 just up and die Oct 06 20:31:12 I'm using mine regularly for poscass Oct 06 20:32:08 I'm using it as my primary phone Oct 06 20:32:15 I use my n800 as a clock. leave good old xdial clock runnign 24/7 Oct 06 20:33:01 I recently moved my landline to my n900. Oct 06 20:34:29 for some defintions anyway Oct 06 20:34:29 'primary calling phone that presents a geographic I'd' Oct 06 20:34:51 oh ok, apparently bluetooth-dun is required to make isync realize that it's talking to a n900 Oct 06 20:35:17 shamus: it's my primary (and only) phone, and i recently bought a second n900 to replace the first Oct 06 20:35:47 how are the specs? i always thouhg the older n800 and n810 were underpowered Oct 06 20:36:16 bad-ish Oct 06 20:36:21 too little ram Oct 06 20:36:36 but every other "smartphone" sucks balls Oct 06 20:36:48 true all smartphones suck Oct 06 20:37:03 persionaly i still use dumbphones Oct 06 20:37:09 and an android tablet Oct 06 20:37:18 yeah, android sucks too Oct 06 20:37:37 ics is not that bad have yet to try jellybean Oct 06 20:37:53 shamus: ever tried using fremantle for a while? Oct 06 20:38:05 but i wil admit the premison system sucks balls Oct 06 20:38:10 nope Oct 06 20:38:27 my tab came preloaded with ice cream samwitch Oct 06 20:38:56 i'd grab a used n900 if you can spare a bit of cache Oct 06 20:38:58 *cash Oct 06 20:39:33 im guessign they have slowly droped in cost due to there age Oct 06 20:39:46 yeah, you can grab one for 100-150 Oct 06 20:40:09 apparently, you've dropped spelling due to age as well? Oct 06 20:40:15 still shoked me how much amazon riped me off on the n800 when i got it. was about 2 years ago and still about 120 ush Oct 06 20:41:09 was pissed that the n900 cam out less then a few months after i got my n800 Oct 06 20:41:23 but i was already passed the return period Oct 06 20:42:19 I have a nexus 7, running jellybean. Oct 06 20:42:39 it annoys me constantly with the subtle breakages. Oct 06 20:44:44 I've been lobbying the company to get me a Note 2 Oct 06 20:44:55 haha Oct 06 20:45:01 lobby for a n900! Oct 06 20:45:08 I've got 3 N900s Oct 06 20:45:12 they're way underpowered. Oct 06 20:45:16 :c Oct 06 20:45:42 doens't the note 2 have 10 times the amount of ram of an n900? Oct 06 20:45:45 beowulf FTW ;-P Oct 06 20:46:48 it's also about 10 times as bulky Oct 06 20:46:53 >:C Oct 06 20:47:02 and yeah, ram is a big issue on the n900 Oct 06 20:47:06 cssu-thumb helps, though Oct 06 20:47:11 get another 20 N900 and you blow your scrubby Note2 outa the water Oct 06 20:47:34 ;-) Oct 06 20:47:39 DocScrutinizer05: get another 20 N900s so you can try resoldering more RAM on one Oct 06 20:47:48 DocScrutinizer05: you mean in power requirements? Oct 06 20:47:49 haha Oct 06 20:48:44 kerio: toldya there's no way to solder more RAM to N900 Oct 06 20:48:57 you can desolder the ram that's already there though Oct 06 20:49:05 unless you build a piggyback board Oct 06 20:52:28 could be scarier I have seen someone replace the CPU on a eepc. that shit is just crazy who in there right mine replaces a soder ball aray at home Oct 06 20:53:05 was on hack a day a while back Oct 06 20:55:52 it's still unclear if POP is desolderable at all, or maybe they used a drop of acrylic glue to stick the upper-POP to lowerPOP Oct 06 20:56:33 after all you wouldn't want the RAM/NAND come off the CPU during reflow ;-P Oct 06 20:58:10 you rather exploit the MMC bus 1 by adding a mux and place some RAM in parallel to the eMMC Oct 06 20:58:46 not exactly genuine RAM for what CPU sees of it, but makes for a nice swap I bet Oct 06 20:58:49 DocScrutinizer05: whatevs, just figure out your EE thing and build me a n900 with a quad-core i7 pls Oct 06 20:59:08 *BURRRP* Oct 06 20:59:19 silly thing: swap on nand Oct 06 20:59:23 would it be possible, somehow? Oct 06 20:59:38 probly not Oct 06 20:59:40 to be honest, I have a dual core HT atom CPU, it's not bad. Oct 06 20:59:41 not worth it Oct 06 20:59:41 without nuking the nand to shit Oct 06 21:00:24 isn't it much faster than the emmc or the usd? Oct 06 21:00:25 there's not more than 256MB NAND anyway Oct 06 21:00:49 and no, it's not THAT much faster Oct 06 21:01:00 http://hackaday.com/2012/08/10/swapping-out-eee-pc-bga-chip-for-1-6-ghz-upgrade/ Oct 06 21:01:20 if you ask my thats just crazy Oct 06 21:01:25 you still need to flash-erase whole pages, which kills your swap Oct 06 21:01:32 performance-wise Oct 06 21:01:42 shamus: what's the point, though? Oct 06 21:01:45 and also regarding wear Oct 06 21:02:02 im guessing they did it becasue they could Oct 06 21:02:13 do they realy need a reasion Oct 06 21:02:18 DocScrutinizer05: would it be easier to replace the whole SoC? Oct 06 21:02:23 yes Oct 06 21:02:27 waaaay easier Oct 06 21:02:37 cool, let's do that then Oct 06 21:02:51 has anyone run Maemo on any other SoC than the n900? Oct 06 21:02:55 we already had that topic finalized some 4 weeks ago Oct 06 21:03:30 we just need an OMAP3430 with better specs and better OneNAND POP Oct 06 21:03:41 and somebody uttered "palm pre" Oct 06 21:03:48 neat Oct 06 21:03:52 as a possuble source for such a critter Oct 06 21:04:33 but then, there's cert for xloader Oct 06 21:05:05 so unless we get uBoot to properly boot maemo, we're probably still screwed Oct 06 21:05:12 now what would be funny/sad is somone trying to get Maemo to run on a palm Vx lol Oct 06 21:05:33 would a 3630 work fine too? Oct 06 21:05:41 (palm pre 2) Oct 06 21:05:42 prolly not Oct 06 21:05:46 damn Oct 06 21:06:07 I guess their electric specs don't match Oct 06 21:06:28 the pre's is 500/600mhz like the n900's Oct 06 21:06:29 uneducated guess Oct 06 21:06:36 the pre plus has half a gb of ram, though Oct 06 21:06:42 which would be ZOMGSOGOOD Oct 06 21:07:29 kerio: download "datasheets" of omap3430 and 3630 from TI and check Oct 06 21:08:03 if the pins are all the same names, odds are they could be just a drop-in replacement Oct 06 21:09:04 if you find a devel board with chip socket, that claims to take both 3430 and 3630, then as well odds are good Oct 06 21:11:15 and *particularly* if you check kernel sources and there's no diff in noardconfig parameters for 3430 and 3630 boards, you could assume the peripherals are all the same, and thus maybe chips are compatible Oct 06 21:11:29 board-config even Oct 06 21:24:15 freemangordon: only thing i don't like about it is the fact that there are thumb alternatives you are supposed to use instead Oct 06 21:24:34 freemangordon: because swp + 2 switches have HUGE penalty Oct 06 21:26:43 jacekowski: can't that code be written in C? Oct 06 21:26:54 freemangordon: swp and swpb is deprecated in arm6 Oct 06 21:27:03 any code can be rewritten in C, doesn't mean it'll be faster. Oct 06 21:27:06 kerio: not something like that Oct 06 21:27:16 why not Oct 06 21:27:16 ? Oct 06 21:27:37 kerio: because it's thumb to arm switch and back to thumb for once Oct 06 21:28:12 kerio: and it's using swp instruction and you can't make gcc emit that instruction Oct 06 21:28:18 for two reasons Oct 06 21:28:24 it's deprecated Oct 06 21:28:36 and C has no understanding of mutexes Oct 06 21:29:06 what SWP is doing is swapping two memory locations as one atomic operation Oct 06 21:30:50 there are compiler specific ways of doing things Oct 06 21:31:24 but those use memory barriers Oct 06 21:31:31 which is the way it should have been done Oct 06 21:32:10 GCC is pretty shit Oct 06 21:32:16 :c Oct 06 21:32:19 don't say that! Oct 06 21:32:28 it's decent on x86 Oct 06 21:32:39 but other platforms, it makes me cry Oct 06 21:32:50 i thought linaro wasn't that bad Oct 06 21:33:42 linaro has not much to do with gcc Oct 06 21:34:20 only platform where gcc generates acceptable code is x86 Oct 06 21:34:38 and even then it's pretty bad when compared with ICC Oct 06 21:34:56 only "good" thing is, that MSVC is just as bad as gcc Oct 06 21:35:00 heh Oct 06 21:35:08 how does clang compare? Oct 06 21:35:23 clang is bit strange thing to compare it directly Oct 06 21:35:43 clang/llvm as a whole? Oct 06 21:36:02 yeah Oct 06 21:36:17 no, i mean, how does it compare to clang/llvm as a whole? :) Oct 06 21:36:56 well, it has improved a lot Oct 06 21:37:24 which one? gcc or llvm? Oct 06 21:37:28 llvm Oct 06 21:37:35 but gcc is still better? Oct 06 21:37:38 but thing is, llvm has plenty of overheads Oct 06 21:37:40 i'm talking about arm Oct 06 21:37:46 dunno about arm Oct 06 21:38:16 to be honest, there is no good all round compiler for arm Oct 06 21:38:23 :( Oct 06 21:38:30 i've seen code generated by codesorcery Oct 06 21:38:32 hmm Oct 06 21:38:36 or codesomething Oct 06 21:38:38 keil Oct 06 21:38:47 and arm arm compiler Oct 06 21:38:52 and all had their problems Oct 06 21:39:07 though, arm compiler seemed to do least amount of fuck ups Oct 06 21:41:38 jacekowski: ldrex / strex won;t do the job Oct 06 21:42:01 I am using 4.7.2 and can use sync intrinstics Oct 06 21:42:24 but they implement mutex by themselves, not MUTEX_SET Oct 06 21:42:54 yeah, well, ideally you would replace whole mutex implementation fully in thumb Oct 06 21:43:30 and having in mind berkeley deb uses swp/b even in it's latest code, I don't dare to rewrite mutex implementation in such a critical databse Oct 06 21:43:32 but thing about bdb, in most cases there is no real need for mutexes in it Oct 06 21:43:38 s/deb/db/ Oct 06 21:43:38 freemangordon meant: and having in mind berkeley db uses swp/b even in it's latest code, I don't dare to rewrite mutex implementation in such a critical databse Oct 06 21:43:57 in mose use cases* Oct 06 21:44:28 jacekowski: there is no penalty for going thumb/arm/thumb afaik Oct 06 21:44:51 i'm pretty sure it causes pipeline flush Oct 06 21:45:30 most probably. but it is the same wuth gcc intrinstics, they put data memory barriers, i've already tried them Oct 06 21:45:41 and used IDA to check the code after that Oct 06 21:45:59 but i think swp has higher penalty than mode switch Oct 06 21:46:18 i know I can write in assembly, as memory barriers are for luti-cpu, but still Oct 06 21:46:42 mode shwitch should not have any penalty besides what bx does Oct 06 21:47:14 branch misprediction has pretty hefty penalty Oct 06 21:47:41 and normal running system you get very very high hitrate on that Oct 06 21:48:04 but branch misprediction means that full pipeline is thrown out Oct 06 21:48:21 and cortex-a8 has 13 stage pipeline Oct 06 21:48:49 you mean bx will flush the pipeline? Oct 06 21:49:22 but aiui there is no way to mispredict it, as it is always taken Oct 06 21:49:54 BX can have condition Oct 06 21:49:58 and abyway BTB is flushed on every context switch :D:D:D Oct 06 21:50:16 well, it can, but it has not in the current code Oct 06 21:50:39 ahm, you are talking about bx in your code Oct 06 21:50:48 i though we are talking generally about bx Oct 06 21:50:49 yeah Oct 06 21:50:52 no Oct 06 21:51:05 i am talking about my code Oct 06 21:51:06 but in your case it will cause flush Oct 06 21:51:13 will it? Oct 06 21:51:19 because of swpb? Oct 06 21:51:22 it's just predicting location Oct 06 21:51:41 no, because instruction decoder will be running in ARM mode Oct 06 21:51:50 aah, i see Oct 06 21:51:53 so prefetched instructions will be decoded as arm Oct 06 21:52:05 yes, you're right Oct 06 21:52:10 and then you switch to thumb and only then processor realises that it's decoded them wrong Oct 06 21:52:44 correct. damn, i'll have to rewrite that with lrdex/strex Oct 06 21:53:01 but thing is, penalty on that is only part of it Oct 06 21:53:13 swp is not penalty free as well Oct 06 21:53:23 i'm not 100% sure if it's cached Oct 06 21:53:34 yeah, but it is a legacy from the origoinal code Oct 06 21:53:49 jacekowski: i'm convinced :D Oct 06 21:54:12 SWP causes processor wide bus lock Oct 06 21:54:45 though, on n900 with single core it doesn't matter so much Oct 06 21:55:11 no matter that code will work in it's current form, I hate when developers do it in the easy (read ineffective)way Oct 06 21:55:31 but on multicore system it will stop all cores Oct 06 21:56:21 yeah, i know that.hmm, actually i can rewrite that with only 3 instruction Oct 06 21:56:24 and may cause cache lines to be flushed Oct 06 21:56:31 ldrex,mov, strex Oct 06 21:57:02 no memory barrier as we're on single core Oct 06 21:57:18 on x86 there is memory snooping for that, but i'm not sure how ARM implements that kind of stuff exactly Oct 06 21:57:29 and I am not trying to upstream that :D Oct 06 21:57:42 How can I reset n900 to factory default Oct 06 21:57:48 grayhat: flash it Oct 06 21:57:51 anyways Oct 06 21:57:53 sleep to do Oct 06 21:57:55 jacekowski: ldrex/strex Oct 06 21:58:10 jacekowski: i dont know how to do that Oct 06 21:58:14 freemangordon: i'm talking hardware side Oct 06 21:58:16 ~flashing Oct 06 21:58:16 methinks maemo-flashing is http://wiki.maemo.org/Updating_the_tablet_firmware Oct 06 21:58:27 freemangordon: as in what happens if one core updates memory location that is cached in another core Oct 06 21:58:27 jacekowski: aah, ok Oct 06 21:58:41 there is cache protocol for that Oct 06 21:58:42 Thanks y'all Oct 06 21:59:04 * freemangordon has forgotten the name Oct 06 21:59:18 john? Oct 06 21:59:20 freemangordon: on x86 all cores snoop memory accesses and will update or in some cases invalidate specific lines Oct 06 22:00:16 anyways Oct 06 22:00:18 sleep time Oct 06 22:00:24 MESI Oct 06 22:00:28 jacekowski: ^^^ Oct 06 22:00:32 night Oct 06 22:01:31 freemangordon: NUMA makes it a lot more fun than that Oct 06 22:01:35 but anyways Oct 06 22:01:37 sleep time Oct 06 22:01:51 :) Oct 06 22:32:14 is uboot working or not? there seems to be a lot of talk about its problems Oct 06 22:34:38 it works Oct 06 22:36:03 Skry: or better said - if it works, it works :D Oct 06 22:40:30 or even better: compare your hw-rev with the list of those who are known to work Oct 06 22:45:03 aka 2101 works everything else is a gamble :D Oct 06 22:46:13 2204 Oct 06 22:46:19 ...swell Oct 06 22:54:44 so what do 2204 owners use for a boot loader? Oct 06 22:57:56 freemangordon: so very true, fortunately I've personally never had any problems with it. Oct 06 23:06:39 Where do people get non-2101 units? I've personally handled 6 N900s, and all of them were r2101. Oct 06 23:07:57 how do I check which I have? Oct 06 23:07:58 got two Oct 06 23:10:17 Hurrian, mine came from amazon Oct 06 23:11:07 entitled, cat /proc/component_version Oct 06 23:15:53 2101 and... 2101 Oct 06 23:15:57 * nox- 2204 Oct 06 23:16:28 good i never tried uboot i guess... Oct 06 23:16:39 I have a 2101 ad 2104 a 2204 and a 4th where I don't know the hw revision :D Oct 06 23:16:44 nox-, you use any kind of multi boot? Oct 06 23:16:59 nox-: actually u-boot prior to some very specific version works on all of them Oct 06 23:17:04 marainein, no Oct 06 23:17:24 merlin1991, oh... Oct 06 23:17:58 ie the uboot-pr1.3 package in extras devel works on all device I own Oct 06 23:18:16 and which one is broken? Oct 06 23:18:35 the latest and greatest by pali :D Oct 06 23:18:43 oh well... Oct 06 23:21:37 /proc/cpuinfo... in features the one im using daily has thumbee and the one in reserve does not? Oct 06 23:22:15 Nothing currently uses ThumbEE. It's just a CPU feature seen by Linux. Oct 06 23:22:19 due to software I've installed 8 suppose, all should support thumbee Oct 06 23:23:05 as it's the same armv7 revision Oct 06 23:24:28 Again, it /supports/ ThumbEE, but nothing uses it Oct 06 23:24:45 anyone replaced the digitizer on their n900? are all those chinese ones as good as any other? Oct 06 23:24:52 It's for faster Java, but I don't recall even Android devices using it. Oct 06 23:25:03 entitled, the chinese digitizers are crap. Oct 06 23:25:32 IIRC there was this one seller on eBay with surplus N900 ones, but I can't find it anymore. Oct 06 23:25:57 friend's n900 has broken touch sensor, only works in part of the screen Oct 06 23:26:11 and mine is pretty scratched up Oct 06 23:27:02 ouch, Oct 06 23:27:26 these things are quite squishy. They pop up on eBay sometimes, though Oct 06 23:27:37 it's pretty clear which ones are Nokia and which ones are ripoffs Oct 06 23:31:57 he got it from uk vendor but with the price being so low it's most likely one of those chinese copies Oct 06 23:33:10 these things are pretty pricey, 30-50 EUR a pop. Oct 06 23:33:54 ye Oct 06 23:50:28 re cpuinfo: what I find a tad concerning is the >>BogoMIPS : 249.96<< on PK46 device, while stock kernel gives >>BogoMIPS : 499.92<< Oct 06 23:52:39 WTH?! the other one with stock kernel and CSSU-TH also has >>BogoMIPS : 249.96<< Oct 06 23:52:50 btw all three 2101 Oct 07 00:05:19 btw, on a sidenote: Oct 07 00:05:21 t900:~# uptime Oct 07 00:05:23 02:03:09 up 190 days, 2:58, load average: 0.00, 0.00, 0.00 Oct 07 00:05:41 :-) Oct 07 00:07:07 since I got other N900 for battery charging, I won't probably boot it that often anynore ;-D Oct 07 00:08:17 so one thing I can confirm: WLAN driver, kernel, and sshd are rock solid ;-) Oct 07 00:09:43 I can hardly get through a day without having to reboot the N950 to get back lost modem connectivity. Oct 07 00:10:02 no modem on t900 Oct 07 00:10:07 no SIM Oct 07 00:10:13 or rather... Oct 07 00:10:24 I want a Jolla device. :< Oct 07 00:10:37 '' Oct 07 00:10:40 probably Oct 07 00:10:54 a sim that's not valid anymore, and thus not registering Oct 07 00:12:52 DocScrutinizer05, measurements are taken at boot Oct 07 00:13:05 dang, *#0000# shows "V 3.2010.02-8" on t900 Oct 07 00:13:38 Hurrian: :nod: Oct 07 00:14:11 still doesn't explain it completely to me Oct 07 00:14:17 If you edit /etc/pmconfig, you can get it to bump up. Oct 07 00:14:48 I think for some devices, it ramps up the CPU OPP earlier, and for some devices later, resulting in 250 and 500 MHz BOGOMIPS Oct 07 00:15:40 yeah, but for both systems they run same kernel and same /etc/pmconfig Oct 07 00:16:05 I wonder if one of them booted on battery while the other on charger Oct 07 00:21:12 MEH, CSSU-TH comes with PK, no? Oct 07 00:21:29 KP Oct 07 00:22:52 A rebranded KP (the one on the newest -thumb comes with 51r1) Oct 07 00:23:51 so, after testing the thumb device with booting from battery, this boils down to "on KP: BogoMIPS : 249.96. On stock kernel: BogoMIPS : 499.92" Oct 07 00:24:06 weird Oct 07 00:24:34 does (say) md5sum /dev/zero go the same speed Oct 07 00:24:46 hmm Oct 07 00:25:11 how would I test speed of md5sum /dev/zero ;-) Oct 07 00:25:56 dd if=/sev/zero bs=1024 count=102400 Oct 07 00:25:59 OE something Oct 07 00:28:11 IroN900:~# time md5sum /dev/mtd4ro Oct 07 00:28:12 b23b5d09162b92c0284923a7f628d2a5 /dev/mtd4ro Oct 07 00:28:14 real 0m0.154s Oct 07 00:28:31 t900:~# time md5sum /dev/mtd4ro Oct 07 00:28:32 b23b5d09162b92c0284923a7f628d2a5 /dev/mtd4ro Oct 07 00:28:34 real 0m 0.22s Oct 07 00:29:03 add another couple of zeros, I guess Oct 07 00:29:24 grr, scratch that Oct 07 00:29:30 it's probably not significant those Oct 07 00:29:56 then Oct 07 00:30:23 did it have thumb patches? Oct 07 00:31:52 real 0m0.317 Oct 07 00:31:55 busybox version seems to differ Oct 07 00:32:16 Skry: you need to call this at least 4..5 times Oct 07 00:32:26 first time is moot Oct 07 00:32:48 format of my md5sum output differs between the two devices Oct 07 00:33:25 Isn't MTD4 full of zeros? Oct 07 00:33:34 (and roughly 2MB) Oct 07 00:33:58 /dev/zero is a pure CPU test Oct 07 00:34:31 IroN900:~# time busybox md5sum /dev/mtd4ro Oct 07 00:34:32 b23b5d09162b92c0284923a7f628d2a5 /dev/mtd4ro Oct 07 00:34:34 (empty line) Oct 07 00:34:35 real 0m0.173s (.174 .170 .173) Oct 07 00:35:41 t900:~# time busybox md5sum /dev/mtd4ro Oct 07 00:35:42 b23b5d09162b92c0284923a7f628d2a5 /dev/mtd4ro Oct 07 00:35:44 real 0m 0.24s (.20 .24 .23 .23 .19) Oct 07 00:36:44 BusyBox v1.10.2 (Debian 3:1.10.2.legal-1osso26+0m5) vs BusyBox v1.10.2 (Debian 3:1.10.2.legal-1osso31+0cssu0) Oct 07 00:38:45 so the one busybox (0cssu0, CSSU-T) is faster, has an empty line, and has 3 decimals Oct 07 00:39:08 while the other is significantly slower, has no empty line and only 2 decimals Oct 07 00:39:11 :-/ Oct 07 00:40:15 DocScrutinizer05: ever think of making a completely new distro for the n900? Oct 07 00:40:16 heh Oct 07 00:40:37 using the kde based stuff? or is the n900 too slow to handle it Oct 07 00:41:14 I don't think kde has many apps that I'd like to use on N900 Oct 07 00:42:00 GGon: its slow but usable after some tweaking Oct 07 00:42:02 or rather, under KDE I'd miss a lot of apps which aren't available there Oct 07 00:42:43 like what? Oct 07 00:42:54 telepathy? :) Oct 07 00:42:56 dialer, ... Oct 07 00:43:02 SpeedEvil: lol Oct 07 00:43:12 still no easy way to access the modem for voice ? Oct 07 00:43:23 no Oct 07 00:43:25 wow Oct 07 00:43:50 i figured that would have been figured out by now Oct 07 00:44:14 then again i haven't used my n900 for voice comm in forever since the modem is broken anyways heh Oct 07 00:44:25 i've been using it like a glorified nit Oct 07 00:44:53 GGon, you just need to setup ofono. Oct 07 00:45:05 Hurrian: i thought they were still having problems with ofono Oct 07 00:45:18 with regards to the sound or something Oct 07 00:45:25 It works for voice calls and texts and data, if you're using the 2.6.37-N900 Adaptation kernel. Oct 07 00:45:46 There's no dialer app, so if you're going to really use it for that purpose then don't bother. Oct 07 00:45:59 is a dialer app that difficult to make in QT? Oct 07 00:46:05 You may want to try building the Nemo dialer though. It's in Qt. Oct 07 00:46:41 just thought it would be interesting to see something running activ Oct 07 00:46:51 that's what the kde touchy ui is called isn't it? Oct 07 00:47:19 i haven't seen a real tablet or phone use it tho... so it must not exist :) Oct 07 00:47:45 jk. i know people got it to half ass run on a transformer but i know i couldn't figure out how to get it on mine. or there wasn't an easy image to just flash to it in order to get it working Oct 07 00:47:58 Plasma Active? Oct 07 00:48:03 It runs on the N900. Oct 07 00:58:42 someone? Oct 07 00:59:43 nope, he's not here right now! Oct 07 01:06:22 i have o problem with u-boot Oct 07 01:06:44 restart looping Oct 07 01:06:58 no usb cable at hand Oct 07 01:07:01 help Oct 07 01:38:06 foo... i got n9 Oct 07 01:43:49 fuck Oct 07 01:43:53 really Oct 07 01:47:07 that was weird Oct 07 01:48:08 n900 was fully charged, I turned it off and then I get this weird bzzzzz sound and it blinked purple. Oct 07 01:48:31 was connected to the charger at that moment Oct 07 01:48:42 seems to be ok tho Oct 07 01:48:54 * ketas curses at the security features Oct 07 01:48:57 flickered purple? Oct 07 01:49:11 ketas++ Oct 07 01:50:03 i got used n9 and accidentally runned track and protect... now WHAT Oct 07 01:50:39 no idea what that is. Oct 07 01:50:48 I think t calls the ninjas. Oct 07 01:51:09 maybe should use #meego ? Oct 07 01:51:39 #harmattam Oct 07 01:51:41 n Oct 07 01:52:01 is more on topic for n9 Oct 07 02:52:59 is there an auto installer for thumb **** ENDING LOGGING AT Sun Oct 07 02:59:58 2012