**** BEGIN LOGGING AT Tue Jul 06 02:59:58 2010 **** BEGIN LOGGING AT Tue Jul 06 08:26:14 2010 Jul 06 08:26:17 there is problem with: http://logs.nslu2-linux.org/livelogs/openmoko-cdevel.txt Jul 06 08:27:43 one row is very brief log :-) Jul 06 08:27:54 yeah :) Jul 06 08:28:31 ThibG: ping Jul 06 08:37:07 gena2x, pong Jul 06 08:37:20 ThibG: oh, nice you around. Jul 06 08:37:30 ThibG: two things. Jul 06 08:38:38 ThibG: first old, as you collect glamo-mci bugs, i also have bug with rootdelay need to be set to 20second to boot. Jul 06 08:38:58 ThibG: but second is much more important for me. Jul 06 08:39:51 yes? Jul 06 08:40:08 ThibG: i reviewed sram timings for glamo vram. it turned out that they were lacking of 'page mode' with tuned sram, i can have glamo speed x2 and smooth 640x480 video. Jul 06 08:40:29 ThibG: but i also getting a problem with sd. Jul 06 08:40:57 ThibG: i reads something like this: "x0x0x0x0x0x0x0x0x0x0x0x0x0x0x0x0`������0000000000 �����������������u�u�u�u�u�u�u�u�u�u�u�u�u�u�u�t�t�t�t�t�t�t�t�t�t�t�t�t�t�t�t�;�;�;�;�;�;�;�;�;�;�;�" Jul 06 08:41:07 heh? Jul 06 08:41:11 ThibG: instead of binary file. Jul 06 08:41:33 ThibG: this look like some clocking, as same data repeated many times. Jul 06 08:41:48 without worker? Jul 06 08:41:51 ThibG: how do you think, it is possible to fix that. Jul 06 08:41:57 ThibG: yes, without. Jul 06 08:42:19 hm Jul 06 08:42:29 it look like 2byte repeated data. Jul 06 08:42:50 rtrtrtrtrtrtrtrtrtrtrtrtrtrtrtrbobobobobobobobobobobobobobobobotrtrtrtrtrtrtrtrtrtrtrtrtrtrtrtrererererererererererererererererpypypypypypypypypypypypypypypypyststststststststststststststststtitititititititititititititititirrrrrrrrrrrrrrrrutututututututututututututututut Jul 06 08:43:07 seem same amounts of data. Jul 06 08:43:17 just timing somewhere. Jul 06 08:44:53 ThibG: can you help with this issue? Jul 06 08:45:01 I wonder Jul 06 08:45:45 what did you change exactly? Jul 06 08:46:22 BANKCON1 register, fields 0:1 and 2:3. Jul 06 08:46:30 in s3c Jul 06 08:46:48 but glamo itself works without problems. Jul 06 08:47:17 i can boot qtmoko from nand, play video to framebuffer and so. Jul 06 08:48:30 So, only not working thing so far is sd. Jul 06 08:48:48 hm, are the datasheets available somewhere? Jul 06 08:48:55 sure. Jul 06 09:51:53 An Analysis of Power Consumption in a Smartphone *(Neo FreeRunner)* - http://ertos.nicta.com.au/publications/papers/Carroll_Heiser_10.pdf Jul 06 09:57:51 Martix_: wow, very interesting! Jul 06 10:01:13 Martix_: yeah! Jul 06 10:01:14 initially I was looking for leaked docs for our beloved media akcelerator :-) Jul 06 10:01:51 not found, but that satusfied my curiosity Jul 06 10:01:55 *satisfied Jul 06 10:03:19 for a few moment :-) Jul 06 10:04:19 gena2x: did you write your results for the glamo-speedup somewhere in a blog or something like that? Jul 06 10:04:38 Azog: wait. we are trying to fix sd right now. Jul 06 10:05:01 Azog: i'll write my results on community-ml. Jul 06 10:05:19 Martix_: was it you who discovered the inefficient choke used in a power converter for the LCD backlight? Jul 06 10:06:57 gena2x: no problem, i just wanted to show a friend how cool the FR could get *g* Jul 06 10:07:36 qtmoko from nand look much better right now... Jul 06 10:08:11 Martix_: am i reading it wrong or do you indeed never mention if the calypso deep sleep used? Jul 06 10:18:35 PaulFertser: its nit my paper ;-) Jul 06 10:18:49 s/nit/not/ Jul 06 10:18:50 Martix_ meant: PaulFertser: its not my paper ;-) Jul 06 10:19:21 Oh Jul 06 10:19:27 just used Google... Jul 06 10:26:49 PaulFertser: I think, its without deep sleep, because 1200 mAh * 3.7 V / 68.6 mW = 64,72 h Jul 06 10:27:18 with deep sleep suspend time can achieve 150 h Jul 06 10:27:39 Martix_: this paper may be wery useful to measure power consumption for overclocked cpu... Jul 06 10:39:25 there are many new questions, e.g. according to analysis Glamo is eating around 85 mW when idle, so how much power can we save by underclocking it Jul 06 10:40:32 !logs Jul 06 10:40:33 Channel logs for #openmoko-cdevel are archived at: Jul 06 10:40:34 http://hentges.net/tmp/logs/irc/%23openmoko-cdevel Jul 06 10:40:35 Live-logs are available at Jul 06 10:40:36 http://hentges.net/tmp/logs/irc/livelogs/%23openmoko-cdevel.livelog Jul 06 10:40:38 See ?? help-logs for usage instructions Jul 06 10:43:16 i miscalculated yesterday. we can write glamo with (640*480*2)*32=20Mb/s. Jul 06 10:43:39 while usual (640*480*2)*15=9.2Mb/s Jul 06 10:44:46 err. i mean copy :) Jul 06 10:45:50 DocScrutinizer: could you change link in topic to: http://hentges.net/tmp/logs/irc/livelogs/%23openmoko-cdevel.livelog ? it works and looks better :-) Jul 06 10:46:47 gena2x: with overclocked CPU or "standard" frequency? Jul 06 10:47:11 Martix_: both with 533/88. Jul 06 10:47:30 but with 400/100 this will be only higher. Jul 06 10:47:52 as 88 vs 100. Jul 06 10:51:34 what is overclocking limit of SDRAM clock? 125 Mhz? Jul 06 10:51:50 SDRAM limit according specs is 133mhz. Jul 06 10:51:55 without overclocking. Jul 06 10:52:11 and in fact it is not 'limited' Jul 06 10:52:22 in this way. Jul 06 10:52:32 so, why you running only 533/88 instead of 533/133? Jul 06 10:52:36 but 133/clk3 is specified as normal timing. Jul 06 10:52:51 not only sdram on bus. Jul 06 10:53:07 problems with PCLK? Jul 06 10:53:10 glamo also :) Jul 06 10:53:17 damn Glamo :-) Jul 06 10:53:32 i am unsure if PCLK is used at all. Jul 06 10:53:55 as a minimun i do not understand where it is used. Jul 06 10:54:14 just don't know. Jul 06 10:54:22 may be serials? Jul 06 11:08:45 gena2x: anything new to test? ;) Jul 06 11:09:15 now i am working on mmc issue. Jul 06 11:09:24 got idea what happens. Jul 06 11:09:40 i personally did test of mplayer on 640x480. Jul 06 11:09:58 640x360 (wide screen) is especially good. Jul 06 11:10:43 dos1: mpeg2 at 15fps with high bitrate and 85% cpu load (without sound) :) Jul 06 11:17:09 gena2x: ok, encoding videos :D Jul 06 11:17:26 dos1: mencoder $1 -ofps 20 -ovc lavc -lavcopts vcodec=mpeg2video:vhq:vbitrate=400 -vf rotate=1,scale=360:-2,eq2=1.1:1:0.01 -oac mp3lame -lameopts br=64 -o $1.moko Jul 06 11:17:28 :) Jul 06 11:17:34 gena2x: saw in logs ;) Jul 06 11:21:09 GNUtoo|laptop: hi Jul 06 11:21:18 why is Xorg hanging the whole phone Jul 06 11:21:28 when I booting newest bitbake Jul 06 11:21:29 ^^ Jul 06 11:21:33 uhm Jul 06 11:21:36 newest build Jul 06 11:24:04 dos1: seem at least workaround for mmc found. Jul 06 11:26:36 :) Jul 06 12:03:52 hmm Jul 06 12:04:20 320x240 mode works well on 2.6.32 Jul 06 12:04:32 i hope we won't really need it anyway thanks to gena2x :D Jul 06 12:05:08 dos1: :))) Jul 06 12:05:17 dos1: how's your battery life with .32? Jul 06 12:05:36 but if someone wants to try it, just setting it with xrandr, sending SIGHUP to enlightenment and setting 2 times smaller values in /etc/profile.d/elementary.sh Jul 06 12:06:02 after that it looks mostly the same, it's just awfully uglier and faster :P Jul 06 12:06:37 mrmoku: seems to be ok, i haven't notice any difference between 29 and 32 Jul 06 12:09:24 ~logs Jul 06 12:09:25 logs are on http://ibot.rikers.org/channel, where "channel" is replaced by the URL-encoded channel name, such as %23freenode for #freenode. Lines starting with spaces are not logged. logs are updated daily --- livelogs are http://logs.nslu2-linux.org/livelogs/openmoko-cdevel.txt and http://hentges.net/tmp/logs/irc/livelogs/#openmoko-cdevel.livelog Jul 06 12:12:41 hmm Jul 06 12:12:45 * mrmoku fears his bat is dying :/ Jul 06 12:12:54 now i see that in ventura there is no really fingerscroll Jul 06 12:13:14 it scrolls only after releasing finger Jul 06 12:13:40 so that wasn't really slowness! :D Jul 06 12:17:38 GNUtoo|laptop: its totaly broken now Jul 06 12:17:56 mrmoku: do you know if there is any WIP on buildhost? :D Jul 06 12:18:06 mrmoku: you know, i would like to build something ;D Jul 06 12:18:31 dos1: don't think so... JaMa is away... spaetz busy with RL and I'm not doing anything :P Jul 06 12:18:37 so if you don't keep it busy... Jul 06 12:31:14 mrmoku: i will have few cookies for you in few minutes ;) Jul 06 12:31:58 leviathan, I didn't change a thing for xorg yet Jul 06 12:31:59 i just have to fix and update my oe git checkout ;) Jul 06 12:32:06 * mrmoku just had 4 plates of potato soup... no need for cookies :P Jul 06 12:32:46 ~nf Jul 06 12:32:47 The #openmoko-cdevel Newsflash Bulletin Board. (continued at ~NF2. For help see ~NF-help) - - - Recommended URLs and channels(chanlogs etc):see ~RL - - - NEWSFLASH [2010-03-28 21:09 UTC]:buildhost got rebooted... new RAM will arrive beginning of next week Jul 06 12:32:54 GNUtoo|laptop: SHR (ewm) doesnt get up anymore Jul 06 12:33:01 aha Jul 06 12:33:08 heh Jul 06 12:33:11 mrmoku: btw, resource icons in lock screen are still not working :P Jul 06 12:33:19 yeah :/ Jul 06 12:33:26 GNUtoo|laptop: and with the kernel build from scratch' defconfig it doesnt even boot up Jul 06 12:33:43 leviathan, try removing udev Jul 06 12:33:56 GNUtoo|laptop: and the size of the kernel image is too big to flash Jul 06 12:34:25 ok Jul 06 12:36:22 leviathan, no one touched kernel Jul 06 12:36:30 leviathan, and I didn't touch xorg yet Jul 06 12:36:39 so maybe it's something else Jul 06 12:37:13 hmm Jul 06 12:37:16 strange Jul 06 12:37:21 anyway Jul 06 12:37:28 its broken Jul 06 12:37:36 just wanted to inform you ^^ Jul 06 12:37:47 anyway Jul 06 12:37:52 leviathan, ok maybe something else broke it Jul 06 12:38:00 where does zimage*.fastboot point? Jul 06 12:38:11 means the autobuild boot.img Jul 06 12:38:17 sd card? Jul 06 12:38:23 first or second partition? Jul 06 12:38:26 yes sdcard Jul 06 12:38:29 first partition Jul 06 12:38:35 hmm Jul 06 12:38:39 /dev/mmcblk0p1 Jul 06 12:38:45 or something like this Jul 06 12:38:54 I've got second partition in use Jul 06 12:38:55 hmm Jul 06 12:39:00 then Jul 06 12:39:20 I bet you didn't rm the devfs /etc/init.d things Jul 06 12:39:41 GNUtoo|laptop: does udev break bootup? Jul 06 12:39:44 wtf? Jul 06 12:39:45 why? Jul 06 12:39:56 leviathan, because we now use devtmpfs Jul 06 12:40:00 at least SHR does Jul 06 12:44:10 http://en.wikipedia.org/wiki/Personal_Package_Archive Jul 06 12:44:12 oops Jul 06 12:44:13 sorry Jul 06 12:44:51 mrmoku: libphone-ui-shr is using edje_edit_part_selected_state_set to display or hide resource icon Jul 06 12:45:14 mrmoku: i think we should use program in edje file and then edje_object_signal_emit in C code Jul 06 12:45:27 like it's done with battery Jul 06 12:45:54 and signal strength Jul 06 12:46:21 mrmoku: i'll try to fix it :) Jul 06 12:47:14 k Jul 06 12:47:26 don't remember if there was a reason for that... Jul 06 12:49:06 mrmoku: probably that macro thing in edje file Jul 06 12:50:00 mrmoku: but i'll try do that anyway within those macro Jul 06 13:34:33 SHR: 03seba.dos1 07libphone-ui-shr * r5a229ca749e6 10/src/view/idle-view.c: idle-view: first step of attemting to fix resource status icons Jul 06 13:51:12 mrmoku: fixed! :) Jul 06 13:52:00 SHR: 03seba.dos1 07libphone-ui-shr * r92e79a60cfb6 10/data/idle_screen/idle_screen.edc: idle-screen: fix displaying active resources Jul 06 14:07:53 SHR: 03seba.dos1 07libphone-ui-shr * rd2f8e3a8a1f5 10/data/idle_screen/ (Miso.ttf idle_screen.edc): idle-screen: add Miso font for SHR logo, fix font usage in other places Jul 06 14:08:10 hmm Jul 06 14:08:22 GNUtoo|laptop: perhaps that was it Jul 06 14:08:26 I'll rebuild Jul 06 14:09:05 ok Jul 06 14:10:32 dos1: nice :) build it :P Jul 06 14:11:00 mrmoku: fix for resources is already in repo :P Jul 06 14:11:05 :) Jul 06 14:11:06 mrmoku: last commit is building right now Jul 06 14:11:33 dos1: good thing to keep CI :) Jul 06 14:12:36 finished, upgrade :D Jul 06 14:13:36 * mrmoku upgrading Jul 06 14:15:40 mrmoku: http://patchwork.dev.bearstech.com/project/SHR/list/ ;] Jul 06 14:19:26 dos1: tested? Jul 06 14:19:33 mrmoku: yup Jul 06 14:19:38 good :) Jul 06 14:19:44 as I have no time to test now... will apply Jul 06 14:23:40 dos1: pushed Jul 06 14:23:44 SHR: 03seba.dos1 07libphone-ui-shr * rae17459d1b55 10/src/view/idle-view.c: idle-view: fix also alarm indicator Jul 06 14:23:48 mrmoku: thanks :) Jul 06 14:25:10 dos1: thank you Jul 06 14:27:45 dos1: as you anyway doing some fixwork, fix also fsogsmd not to hang it's dbus iface during suspend/resume in deep sleep mode. 8) Jul 06 14:28:13 Q-Master: hehe :D Jul 06 14:28:28 Q-Master: but i don't have #1024 fix on my neo Jul 06 14:29:05 dos1: That's another one fix you should apply. 8)) Jul 06 14:29:38 Q-Master: can you give me patch? :D Jul 06 14:29:51 8) Jul 06 14:35:10 dos1: are you coming to moscow by any chance? ;) Jul 06 14:39:43 Martix__: probably not much, because it already has dynamic gating turned on for everything Jul 06 14:51:53 Weiss: are you here? :) Jul 06 14:52:39 ish :) Jul 06 14:52:52 at work, so no datasheets right now :) Jul 06 14:53:12 (... but were you playing with S3C registers anyway, in which case you need the S3C2442 datasheet and not the Glamo ones?) Jul 06 14:54:11 Weiss: ok. it turned out that problem with mmc is general burst-memory read problem. Jul 06 14:54:32 Weiss: but memory write (copy) is now 19Mb/s :) Jul 06 14:54:45 :O Jul 06 14:54:56 Weiss: 640*480*32*2 Jul 06 14:55:17 PaulFertser: i think i don't unfortunately :x Jul 06 14:55:24 are you absolutely sure you're not just measuring the cache performance? Jul 06 14:55:35 (not trying to be nasty, just want to be sure..) Jul 06 14:55:49 Weiss: burst-read can be workarounded urgly but easily bu seeting timing before reading mmc. Jul 06 14:56:04 Weiss: first, i see screen changes :) full screen one :) Jul 06 14:56:14 gena2x: you mean, Glamo's MMC engine doesn't like burst transfers? Jul 06 14:56:41 Weiss; second, i am running same test two times - with and without changes. Jul 06 14:56:51 Weiss: and third is mplayer :) Jul 06 14:57:16 Weiss: i see mplayer mplays 640x480x15 without flicker. Jul 06 14:58:02 Weiss: i mean burst write is ok, but burst read nok. Jul 06 14:58:09 Weiss: and glamo is unrelated. Jul 06 14:58:24 Weiss: but sram chip in glamo is related. Jul 06 14:58:44 Weiss: so, timings i want to have/set are that sram chip timings. Jul 06 15:01:44 gena2x: so, you fiddled with registers in glamo-core.c? Jul 06 15:01:56 Weiss: no. no need. Jul 06 15:02:31 Weiss: as far as i undertand things, s3c controls transfer to glamos sram, not glamo. Jul 06 15:02:47 Weiss: but i may be wrong of course. Jul 06 15:03:03 ah, S3C registers? Jul 06 15:03:15 Weiss: not registers. Jul 06 15:03:20 Weiss: glamo has sram. Jul 06 15:03:26 Weiss: sram has timings. Jul 06 15:03:42 Weiss: need to set s3c to match that timings. Jul 06 15:03:59 Weiss: so, i want to know somehow that timings. is this is possible? Jul 06 15:04:33 I think there might be some stuff about that in the datasheets.. I'll check Jul 06 15:04:50 (and report back sooner or later depending on whether or not my new modem has been delivered yet...) Jul 06 15:05:54 Weiss: alternatively (i you think telling timings violating something), you can just provide me with correct s3c settings for different clock speeds. Jul 06 15:06:20 i'll figure timings on my own out of this settings :) Jul 06 15:07:39 Weiss: hope i'll see your report. Jul 06 15:10:38 I think telling you the relevant timings is allowed - it's kind of the idea of the NDA that I can do that Jul 06 15:11:10 since I'll be telling you from my own knowledge having read the documentation, not by quoting from the docs Jul 06 15:11:40 hmm. i think not. Jul 06 15:11:55 D means discrosure. Jul 06 15:12:05 hm... Jul 06 15:12:08 i don't know. Jul 06 15:12:22 there's an explicit allowance in OM's NDA for using the information to write FOSS code, however Jul 06 15:12:22 it would be good if it is possiblem. Jul 06 15:13:20 ok, this in fact not needed asap Jul 06 15:13:33 * Weiss does wonder how much of SMedia's requirement for an NDA is to protect them from accusations of wrong/poor/sloppy documentation Jul 06 15:15:25 (also less of a problem since SMedia doesn't seem to exist as a company any more...) Jul 06 15:15:34 not sure what difference that makes Jul 06 15:16:13 no difference. docs may have 3rd party secrets so related stuff. Jul 06 15:16:16 imo. Jul 06 15:16:28 s/so/or/ Jul 06 15:16:41 s/stuff/sh.t/ Jul 06 15:17:10 hehe Jul 06 15:17:42 wikileaks..... ;] Jul 06 15:19:28 but my case may be a bit different... as i want to setup other device... so don't know. Jul 06 15:21:05 oh.. I hope this happens.. Jul 06 15:21:18 cz_jc: which bit of it? :) Jul 06 15:21:28 Weiss, everything. it just sounds crazy :) Jul 06 15:21:47 safe to say I will NOT be deliberately leaking any NDA'd documentation.. it'd be far too obvious who did it for a start.. Jul 06 15:21:59 but I will do everything I can to help gena2x speed things up within the allowances of the NDA :D Jul 06 15:22:13 Weiss, it's like concorde pilots saying 'well.. we can't climb any higher' and then a year later 'oh right, we have reheats' hehee Jul 06 15:22:25 (and maybe send him my spare debug board if the speedups are real...) Jul 06 15:22:48 Weiss: cool. Jul 06 15:23:12 Weiss, great :) Jul 06 15:23:16 Weiss: i can provide you with uboot, kernel and mplayer. where to get my dboard? Jul 06 15:23:18 :) Jul 06 15:23:54 cz_jc: hehe... there are so many reasons Concorde shouldn't work... it gets off the ground by pumping all the fuel to the back of the plane to tilt it upwards, and slows down for landing by splatting itself against the atmosphere like a spaceship ... and aligns shockwaves inside its engines to make the airflow work supersonic... Jul 06 15:24:53 Weiss, well, it's just a matter of setting the trim pumps to 'aft' and engine rating mode to 'approach'. Then you release reheat guard and you can press those reheat buttons under throttle lever :) Jul 06 15:25:42 it can't literaly climb from about 30 000 ft higher without reheat :) Jul 06 15:25:55 (if loaded for normal trips) Jul 06 15:26:47 cz_jc: aha.. Glamo register 0x1f44: "RR Olympus-Snecma 593 reheat option" Jul 06 15:28:12 hehe Jul 06 15:29:01 Weiss, shame concorde isn't operating any more though Jul 06 15:29:02 gena2x: so... can i test something? :D Jul 06 15:29:14 I wonder why these settings weren't found earlier... makes me a little nervous Jul 06 15:29:20 * mrmoku wonders if that reheat option would keep the thing hot enough to fight WSoD ;) Jul 06 15:29:20 yeah me too Jul 06 15:29:54 dos1: wait a bit. i am still working on mmc. Jul 06 15:30:11 I remember Andy mentioned that it'd go a bit faster, but with significant risk of problems, but I don't think the improvements were anything like this Jul 06 15:30:27 dos1: this timings are not so easy. Jul 06 15:30:51 gena2x: since yesterday, this is all without DMA, right? Jul 06 15:31:04 it sounds way good to be true Jul 06 15:31:06 Weiss: he had debug kernel, so anything were slow without glamo. also - your patch, it may be extremly important. Jul 06 15:31:30 Weiss: tried both dma and non-dma. Jul 06 15:31:42 Weiss: dma seems a bit slower this time :) Jul 06 15:32:01 Weiss: something like 1.05/1.2 or so Jul 06 15:32:05 DMA is about CPU usage though, not speed... if we can do pixmap migration with no CPU impact, that's very significant.. Jul 06 15:33:41 gena2x, do I understand correctly you overclocked glamos SRAM and synchronized timings with it properly ? Jul 06 15:33:48 Weiss: anyway, mem bus is busy :) Jul 06 15:33:58 cz_jc: no overcloks this time. Jul 06 15:34:11 gena2x, wow Jul 06 15:34:33 gena2x: busy, but not 100% busy? (since main RAM is so much faster then VRAM?) Jul 06 15:34:50 previously, the memory bus was 100% logjammed even with a relative trickle to Glamo Jul 06 15:34:57 Weiss: in fact i think that vram should be faster than ram. Jul 06 15:35:16 Weiss: as vram is sram, and dram is dram... Jul 06 15:35:53 gena2x, glamo vram is faster than normal ram ? Jul 06 15:35:55 but VRAM is behind a layer of Glamo doom? Jul 06 15:36:33 cz_jc: this is theory -> sram is faster than sdram. sram is used in cpu L2 caches. Jul 06 15:36:45 cz_jc: but i do not know actual timings. Jul 06 15:37:24 gena2x, sure.. because it's a bunch of registers that don't need refreshing. It 'should' work 1:1 with CPU instruction cycle Jul 06 15:37:59 vram can't do 1:1, and can't be <4 clocks Jul 06 15:38:02 hmmm... Jul 06 15:38:07 it can in fact :) Jul 06 15:38:14 but. Jul 06 15:38:17 but. Jul 06 15:38:19 how? :D Jul 06 15:38:44 it needs sincronization between glamo and cpu, which introduces delay. Jul 06 15:39:05 that stands to reason :) Jul 06 15:39:29 gena2x, so you load data to a temporary buffer and transfer it via dma when the time is right ? Jul 06 15:39:55 no. syncronization is special line as i understand - nwait. Jul 06 15:40:57 but i am hm.. not a hardware engineer :) Jul 06 15:41:36 nwait is the line which Glamo hammered on, which slowed the system down.. Jul 06 15:41:51 gena2x, I think I know what you mean, that one causes the CPU to release the entire memory bus and wait till its back ready, right ? Jul 06 15:42:30 (so basically, writes to DRAM and all mmapped stuff are inhibited) Jul 06 15:42:42 i think sdram has no nwait. Jul 06 15:43:10 gena2x, no the memory chip doesn't care.. s3c2442 does :) Jul 06 15:43:39 it is per-bank setup. Jul 06 15:44:20 gena2x, really? my original understanding was the whole external bus was in halt Jul 06 15:44:35 I like to be proven wrong :) Jul 06 15:45:13 i am really do not have full understanding too, sorry. Jul 06 15:45:36 but i have facts about speed :) Jul 06 15:45:47 oooh ! Jul 06 15:45:49 I see it :) Jul 06 15:45:56 gena2x, you are right indeed Jul 06 15:46:01 there is only one nWait pin Jul 06 15:46:02 but Jul 06 15:46:19 it can be set, per bank, which ones does the nWAIT pin affect :) Jul 06 15:46:29 we all knew that making glamo faster is impossible. he just didn't know that :D Jul 06 15:46:52 haha. Jul 06 15:47:05 i just alway were not sure in that. Jul 06 15:47:07 dos1, hehe that about sums it up Jul 06 15:47:15 you may read my 2009 mails :) Jul 06 15:50:16 gena2x, you are a bearer of news too good to be true :) Jul 06 15:50:51 gena2x, all I can do at this point is wish you good luck so that a revolution is made :) Jul 06 15:51:21 mrmoku: hmm, seems like idle-view doesn't even register for getting WakeupTimeChanged signal for alarms Jul 06 15:51:25 i hope today. Jul 06 15:51:50 mrmoku: there is only idle_screen_view_update_call function, which seems to be never called at all Jul 06 15:51:50 can't wait :)) Jul 06 15:51:57 i'll publish one of two things - proper kernel or mmc workaround. Jul 06 15:52:56 s/update_call/update_alarm/ Jul 06 15:52:56 dos1 meant: mrmoku: there is only idle_screen_view_update_alarm function, which seems to be never called at all Jul 06 15:54:08 s/two things/two things today/ Jul 06 15:54:09 gena2x meant: i'll publish one of two things today - proper kernel or mmc workaround. Jul 06 15:54:53 gena2x, sure.. take your time.. and thanks for all this reasearch :) Jul 06 15:55:57 mrmoku: but the same with idle_screen_view_update_call Jul 06 15:57:05 larsc, your thought about the state of glamo for upstreaming? Jul 06 15:59:48 uhm.. don't know. Jul 06 16:01:39 ok Jul 06 16:01:55 well, time to let reviewers tell, I guess! Jul 06 16:02:38 ok, time to prepare and publish kernel. Jul 06 16:02:42 :) Jul 06 16:10:18 mickeyl: is it ok that pdp.registration is still "home" even after deactivating context? Jul 06 16:32:53 hi,does someone know flash uttilities well,I've this issue: Jul 06 16:33:22 flash_info /dev/mtd3 => Device /dev/mtd3 has 0 erase regions Jul 06 16:36:46 dos1: yes. once being registered to a PDP cell, it stays like that until you forcefully unregister (which is not necessary) Jul 06 16:37:05 it's like being registered vs. having an ongoing phonecall Jul 06 16:37:32 mrmoku: ^^^ so i guess pdp.registation isn't a good indicator for GPRS connection in idle-view ;D Jul 06 17:02:12 dos1: showing a possibility of GPRS is a good thing also, like on SE blue bar over the GSM bars Jul 06 17:03:04 Q-Master: for that i think we should force PDP registration on GSM registration Jul 06 17:03:13 mickeyl: ^^^^ does that have any downsides? Jul 06 17:04:46 mickeyl: fix the fsogsmd plz. I'm tired of restarting it. 8( On 2.6.32 dbus locks happens too often. Jul 06 17:05:10 Q-Master: i would, if i had any idea where the problem is. Jul 06 17:05:20 8( Jul 06 17:05:34 mickeyl: it is somehow connected with deep sleep. Jul 06 17:05:35 mickeyl: what about the problems in my last ticket? Jul 06 17:05:50 Q-Master: well, i don't think so Jul 06 17:05:57 mickeyl: where i finally tried using the flowcontrolled node Jul 06 17:06:14 Q-Master: i think i also have dbus locks, and i'm not using deep sleep Jul 06 17:06:35 dos1: with deep sleep it happens much faster then. Jul 06 17:08:34 ok Jul 06 17:09:00 when freeze happens with 2.6.32, Neo responds to pings, but i can't log into ssh Jul 06 17:10:44 dos1: I meant exactly fsogsmd's dbus lockups. It is seeing rings, but doesn't signal to other sw on them and doesn't receive any dbus calls. Jul 06 17:11:13 Q-Master: yup, i had that few times Jul 06 17:11:29 but i have lots of hard freezes of whole system... Jul 06 17:11:43 if someone has a bit time, i't be useful to enable dbus verbose debugging Jul 06 17:12:17 fsogsmd locks happens to me mostly overnight Jul 06 17:12:32 but i haven't check logs for it yet Jul 06 17:19:10 Q-Master: i'm not sure it doesn't receive any calls. I think i straced it once and it received dbus messages, it just never answered. Jul 06 17:22:20 PaulFertser: it is definitely receiving RINGs, but seems not signalling other sw on them Jul 06 17:22:49 mickeyl: I'll be able to do that tomorrow evening. Jul 06 17:23:52 Q-Master: i was stracing fsogsmd and i think i remember seeing it receiving my dbus calls over the socket Jul 06 17:24:23 ~lart abuse teams of the majority of dns registrars and isps Jul 06 17:24:30 PaulFertser: it might receive it over the socket, but seems not seeing it in main loop then Jul 06 17:24:34 * apt beats abuse teams of the majority of dns registrars and isps severely about the head and shoulders with a rubber chicken Jul 06 17:24:49 ~lart abuse teams of the majority of dns registrars and isps once again Jul 06 17:24:49 * apt hooks into a hydrant and hoses abuse teams of the majority of dns registrars and isps once again down Jul 06 17:24:58 ~lart abuse teams of the majority of dns registrars and isps and again Jul 06 17:24:59 * apt cuts abuse teams of the majority of dns registrars and isps and again into thin stripes Jul 06 17:25:50 Q-Master: there's no concurrency there. It's mainloop that calls anything that can read() afaik Jul 06 17:27:37 ~lart Hard Freeze of Death on 2.6.32 kernel Jul 06 17:27:37 * apt throws a AN/M-8 smoke grenade at Hard Freeze of Death on 2.6.32 kernel Jul 06 17:27:50 HFoD ;] Jul 06 17:28:03 dos1: how hard is that? Jul 06 17:28:12 dos1: does ramconsole show kernel panic? Jul 06 17:28:38 lindi-: ping works, it seems to happen at heavy load Jul 06 17:29:11 lindi-: it may be wifi connected, but i'm not sure (maybe just using wifi makes it more probably to happen) Jul 06 17:29:30 lindi-: how to check ramconsole? Jul 06 17:35:55 lindi-: well, it just looks like everything get a HUUGE slowdown and in few seconds it goes into state when it's just frozen Jul 06 17:36:42 lindi-: sometimes cursor in vala-terminal is still blinking for some time, but i can't do anything Jul 06 17:37:09 lindi-: and even screen is not dimmed or undimmed Jul 06 17:37:22 dos1: he means you should be using watchdog and it'll reboot the device automatically and then you'll retrive the ramconsole from dmesg or with the special tool (i do not remember the details). Jul 06 17:37:42 and after some time (~15 seconds) even cursor stops blinking Jul 06 17:38:24 PaulFertser: yup, i understood that, but i still don't know how to do that :) Jul 06 17:38:48 dos1: he'll tell you some apt-get command Jul 06 17:38:51 dos1: :) Jul 06 17:38:56 hehe ;D Jul 06 17:39:05 dos1: the idea is that you should install the watchdog software Jul 06 17:41:12 larsc, GLAMO_GPIO_REG_GEN1? Did you forget to commit a header? Jul 06 17:41:34 don't know Jul 06 17:41:56 , but i don't think so Jul 06 17:41:58 dos1: http://docs.openmoko.org/trac/ticket/2135 Jul 06 17:43:01 hmm Jul 06 17:43:16 shouldn't it be applied to default kernel? Jul 06 17:43:39 dos1: the opinions differ Jul 06 17:50:57 <[Rui]> hi all Jul 06 17:52:08 SHR: 03seba.dos1 07shr-wizard * rb04c841861b0 10/python/shr-wizard.py: python: always display scrollbar Jul 06 17:53:21 mickeyl: have to rebuild dbus for that IIRC? Jul 06 18:11:53 hi, how do I uninstall android on my freerunner Jul 06 18:11:54 ? Jul 06 18:12:05 I want/need shr on flash Jul 06 18:12:16 for now it's only on sdcard Jul 06 18:12:36 dfu-util doesn't work Jul 06 18:12:40 for instance Jul 06 18:12:59 dfu-util -d 0x1d50:0x5119 -a u-boot -R -D qi-ubi-s3c2442-1.0.2-r3+gitr0+c38b062a609f1442e6a9e13005cfbdfd59a5ac0d.udfu Jul 06 18:13:14 fails and say : No such Alternate Setting: "u-boot" Jul 06 18:13:42 I had it only for trying and as a backup system in case my phone didn't work on shr Jul 06 18:13:50 but now I want to get rid of it temporarly Jul 06 18:14:00 in order to try mtd-utils command on shr's flash Jul 06 18:14:02 GNUtoo|laptop: you just start nor u-boot Jul 06 18:14:09 yes and then? Jul 06 18:14:11 GNUtoo|laptop: and use that particular command Jul 06 18:14:46 PaulFertser, was it a question? Jul 06 18:15:04 GNUtoo|laptop: no :) dfu-util is supposed to work like that. -a u-boot should be available. Jul 06 18:15:23 it's not Jul 06 18:15:26 I tried -a qi Jul 06 18:15:30 and it'sthe same Jul 06 18:15:36 android messed up the partition scheme Jul 06 18:16:11 dos1, OMG a commit! Jul 06 18:16:38 PaulFertser, I just wanted to try SHR for trying mtd-utils commands Jul 06 18:16:43 dos1, you just passed me (in the race for the top-committer) Jul 06 18:16:56 PaulFertser, maybe you know the suposed output of the commands on standard GNU/Linux? Jul 06 18:17:04 TAsn: :D Jul 06 18:17:17 GNUtoo|laptop: no, with u-boot the environment is always created a new every time you start it Jul 06 18:17:30 ah? Jul 06 18:17:37 I run uboot for nor Jul 06 18:17:41 dos1, I'll be back! Jul 06 18:17:42 GNUtoo|laptop: with nor u-boot there's no persistent env Jul 06 18:17:46 on nand there is the special android qi Jul 06 18:17:54 GNUtoo|laptop: i understand what you say Jul 06 18:18:03 ok Jul 06 18:18:30 GNUtoo|laptop: probably you've got an incompatible dfu-util version or you specify wrong VID:PID or something like that. Jul 06 18:18:34 basically under my htcdream I have that: flash_info /dev/mtd3 => Device /dev/mtd3 has 0 erase regions so I wanted to try on SHR Jul 06 18:18:39 ok Jul 06 18:18:43 I've that Jul 06 18:18:44 dos1, what do we need fbdev for? Jul 06 18:18:56 but I used the -d arguemntn too Jul 06 18:18:58 TAsn: just in feed, for testing Jul 06 18:19:01 and same result Jul 06 18:19:26 oh feed. Jul 06 18:19:40 PaulFertser, I'll compile dfu-utils Jul 06 18:19:48 GNUtoo|laptop: well... you can boot from SD and flash qi with "flasherase_all" + "nandwrite" combo Jul 06 18:19:54 ok Jul 06 18:20:03 but I'll have to re-do the partition scheme? Jul 06 18:20:38 GNUtoo|laptop: no, once you flash proper Qi, it'll use the right scheme on every boot. Jul 06 18:20:50 ah ok thanks a lot Jul 06 18:20:51 GNUtoo|laptop: because it's sorta hardcoded in Qi Jul 06 18:21:31 TAsn: hmm, http://shr-project.org/trac/ticket/763? Jul 06 18:21:40 TAsn: i don't understand this ticket ;x Jul 06 18:22:14 ok thanks a lot!!! Jul 06 18:22:35 Always welcome Jul 06 18:22:52 mrmoku: yes Jul 06 18:23:13 dos1, the main issue is fixed Jul 06 18:23:23 it was a combo box to choose fields Jul 06 18:23:27 mickeyl: ok, will try - if Q-Master does not beat me :) Jul 06 18:23:32 in the contacts app Jul 06 18:23:44 but what's still missing is a dialog to creating new fields Jul 06 18:24:04 mrmoku: *nod* i'm glad to fix things but these days i have no room for a serious debugging session :/ Jul 06 18:24:59 mrmoku: if you'll have something compiled, give it also to me ;) Jul 06 18:24:59 mickeyl: sure Jul 06 18:25:06 dos1: ok Jul 06 18:25:11 larsc, you've changed GLAMO_REG_GPIO_GEN1 to GLAMO_GPIO_REG_GEN1 but the later is undefined Jul 06 18:25:15 hehe Jul 06 18:26:34 mrmoku: http://shr-project.org/trac/ticket/1071 - isn't it already implemented? Jul 06 18:27:00 AHAH Game is starting soon, the less interesting one, but still. Jul 06 18:27:42 dos1: somehow at least partly, yes Jul 06 18:28:08 TAsn: interesting... to see whom we'll play on sun ;) Jul 06 18:28:37 mrmoku, Netherlands... Jul 06 18:28:39 PaulFertser, compiling dfu-utils and trying it gave the same results...trying from shr Jul 06 18:28:46 TAsn: yeah, think so too Jul 06 18:29:10 TAsn: what about http://shr-project.org/trac/ticket/716? Jul 06 18:29:23 so...from shr Jul 06 18:29:28 root@om-gta02 ~ # om battery consumption Jul 06 18:29:28 -81750 Jul 06 18:29:30 if it fails I'll have some issues Jul 06 18:29:57 lindi-: ^^ is that normal consumption? Jul 06 18:30:08 GNUtoo|laptop: i had that issue with nor u-boot not providing "alternatives" once. I'm not sure what i did to it, but it started working. Probably replugged usb or something... Jul 06 18:30:21 ok Jul 06 18:30:46 Could not unlock MTD device: /dev/mtd7 Jul 06 18:30:51 ah it behave like my g1 Jul 06 18:30:53 strange Jul 06 18:31:21 dos1, read the last comment by mrmoku Jul 06 18:31:29 dos1: ready to test? Jul 06 18:31:32 * soltys yawns Jul 06 18:32:01 mrmoku: that's charging Jul 06 18:32:15 <[Rui]> dos1: 716 is definitely implemented already Jul 06 18:32:18 lindi-: hmm... same when I unplug... have to wait a bit? Jul 06 18:32:38 GNUtoo|laptop: why mtd7? Jul 06 18:32:43 gena2x: yup :) Jul 06 18:32:46 ahh Jul 06 18:32:47 PaulFertser, it was a try Jul 06 18:32:48 lindi-: thanks :) Jul 06 18:32:49 GNUtoo|laptop: you're supposed to flash Qi to /dev/mtd1 Jul 06 18:32:57 <[Rui]> dos1: well, most of it... only wifi is missing but then... Jul 06 18:33:01 168937 Jul 06 18:33:12 <[Rui]> dos1: my only complaint about iliwi is that it occupies *a*lot* of memory Jul 06 18:33:13 GNUtoo|laptop: and erase it first, then nandwrite -p (i suggest you do _not_ use -m) Jul 06 18:33:20 PaulFertser, I only want SHR on flash because I want to know why flash_* didn't work on htcdream Jul 06 18:33:24 ok Jul 06 18:33:26 [Rui], :( Jul 06 18:33:51 <[Rui]> TAsn: about as much as elmdentica, last I compared Jul 06 18:34:44 rebooting Jul 06 18:34:45 ok Jul 06 18:34:52 PaulFertser, what -m does? Jul 06 18:34:59 [Rui], pardon me for my ignorance, but how much does elmdentica take? :P Jul 06 18:35:13 GNUtoo|laptop: mark bad blocks whenever found Jul 06 18:35:20 GNUtoo|laptop: disasterous when you do not erase first. Jul 06 18:35:20 yes I know Jul 06 18:35:23 <[Rui]> TAsn: one minute Jul 06 18:35:24 ah ok Jul 06 18:35:31 PaulFertser, I did that on the htcdream Jul 06 18:35:32 mrmoku: http://shr-project.org/trac/ticket/688 Jul 06 18:35:36 what's disastrous? Jul 06 18:35:38 mrmoku: am i right in last comment? Jul 06 18:35:46 I wasn't able to recover my nand Jul 06 18:35:57 GNUtoo|laptop: marks all blocks as bad :) Jul 06 18:36:35 ah ok Jul 06 18:36:39 maybe that was my isse Jul 06 18:36:43 how do I recover from that Jul 06 18:36:45 GNUtoo|laptop: possible to recover, at least with gta02 :) Jul 06 18:36:45 ? Jul 06 18:36:50 ah ok Jul 06 18:36:56 btw I half bricked gta02 now Jul 06 18:37:08 it reboot endlessly Jul 06 18:37:11 with nand Jul 06 18:37:32 maybe I should have flashed to /dev/mtd0? Jul 06 18:37:35 GNUtoo|laptop: gta02 is unbrickable Jul 06 18:37:40 GNUtoo|laptop: mtd0 is NOR Jul 06 18:37:42 <[Rui]> TAsn: VIRT(35644) RES(16M) SHR(6272) Jul 06 18:37:43 but mtd0 is partition scheme Jul 06 18:37:45 You can't flash to it easily Jul 06 18:37:46 ij Jul 06 18:37:47 ok Jul 06 18:37:57 I know it's unbrickable Jul 06 18:38:02 that's why I say half Jul 06 18:38:06 GNUtoo|laptop: the partition scheme is not kept anywhere at all when you use Qi. Jul 06 18:39:06 ok Jul 06 18:39:23 it's passed as kernel command line arguments then? Jul 06 18:39:46 <[Rui]> TAsn: well, iliwi either got thinner or elmdentica got fatter, it's now a 50% ratio, +- Jul 06 18:39:52 now it seem bricked for real...screen doesn't come back on Jul 06 18:39:55 I'll try serial Jul 06 18:41:09 <[Rui]> TAsn: well, 50% +- or RES, others doesn't fare so well. VIRT(24344) RES(7503) SHR(5020) Jul 06 18:42:47 maybe battery got discharged Jul 06 18:42:50 but it was on usb Jul 06 18:42:54 but battery was low Jul 06 18:43:03 GNUtoo|laptop: Qi generates partition table on boot on the fly, and passes to the kernel, right Jul 06 18:43:15 [Rui], fat old elmdentica ;P Jul 06 18:43:19 You probably leak :P Jul 06 18:43:38 ah screen is here again Jul 06 18:45:16 too bad I reboted and it's gone again Jul 06 18:45:16 <[Rui]> TAsn: there's probably a leak or two, although I'm careful, I'm not paranoid Jul 06 18:45:18 hmmm Jul 06 18:45:23 I'll charge the battery Jul 06 18:45:36 [Rui], :P Jul 06 18:45:47 I'm always ultra paranoid (well usually) Jul 06 18:46:03 ThibG: that was not intentional Jul 06 18:46:17 I'm now on uboot Jul 06 18:46:22 does it charge the phone? Jul 06 18:46:39 I've put the power adapter on Jul 06 18:46:47 btw I can't use power adapter + serial Jul 06 18:46:58 if I do I got garbage on the serial console Jul 06 18:47:05 GNUtoo|laptop: i think nor u-boot should charge with adapter Jul 06 18:47:08 <[Rui]> TAsn: however, I think EFL leaks more :) Jul 06 18:47:08 <[Rui]> I got scared running valgrind Jul 06 18:47:11 <[Rui]> TAsn: iliwi links against more libraries? :) Jul 06 18:47:12 ok thanks Jul 06 18:47:24 so... Jul 06 18:47:42 <[Rui]> TAsn: I remember leaving some possible leakes for later to refer to, but I can't remember if I did close them or not. Jul 06 18:47:43 PaulFertser, do you know how to recover this bad block things? Jul 06 18:47:51 [Rui], I also thought efl leaks more Jul 06 18:47:54 because I had one bad block Jul 06 18:47:54 <[Rui]> TAsn: one of these days I might run valgrind again Jul 06 18:47:56 GNUtoo|laptop: on gta02 yes Jul 06 18:48:02 mrmoku, told you :P Jul 06 18:48:03 ok tell me Jul 06 18:48:06 GNUtoo|laptop: if it's only one, probably it should indeed be there Jul 06 18:48:12 [Rui], it's not just a matter of running valgrind Jul 06 18:48:18 <[Rui]> TAsn: «thought» meaning you don't anymore? Jul 06 18:48:25 you have to first compile efl with -O0 Jul 06 18:48:31 <[Rui]> TAsn: it helps a lot, for catching stuff Jul 06 18:48:35 PaulFertser, maybe it works on htcdream too Jul 06 18:48:39 GNUtoo|laptop: if the majority of the blocks got marked due to mistake (not erasing first) then yes, i'll tell you the method :) Jul 06 18:48:51 [Rui], raster claims it barely :P Jul 06 18:49:02 GNUtoo|laptop: u-boot has "scrub" command that erases the NAND fully. Including factory information about the bad blocks. Jul 06 18:49:10 [Rui], thought = raster made me think again. :P Jul 06 18:49:20 PaulFertser, ah ok Jul 06 18:49:35 <[Rui]> TAsn: well, maybe I caught it at a bad time, but valgrind complained a lot about efl stuff when I ran it Jul 06 18:49:36 PaulFertser, so not usable on dream,fastboot doesn't erase fully Jul 06 18:51:01 <[Rui]> well, it doesn't complain about efl anymore as I load the timeline, there must have been fixes :) Jul 06 18:51:10 [Rui], yeah, but valgrind doesn't like -O2 Jul 06 18:51:17 Which you used (in efl) Jul 06 18:51:32 <[Rui]> TAsn: I don't know wheter I did or not in the RPMs, probably did Jul 06 18:52:02 surely did. Jul 06 18:52:04 GNUtoo|laptop: it also depends on the way bad blocks are marked. If it's bbt (bad block table), you can erase it with flash_eraseall most probably. If it's OOB, it might be more tricky. Jul 06 18:52:45 I won't use flash_eraseall on dream Jul 06 18:52:53 because I don't think it has 2 bootloaders Jul 06 18:53:09 I'll ask about goldcarding and unbrickability maybe Jul 06 18:54:08 GNUtoo|laptop: eraseall erases only the mtd partition you ask to erase. Jul 06 18:54:14 TAsn: if you are not paranoid it doesn't mean that THEY're not watching ya... 8)) Jul 06 18:54:47 "Just because you're paranoid Doesn't mean they're not after you" Jul 06 18:54:50 ah ok thanks Jul 06 18:54:57 Nirvana! Jul 06 18:55:04 then I'll try it Jul 06 18:55:48 GNUtoo|laptop: but i'm afraid it doesn't touch OOB, so if your bad block info is in there, it won't help. Jul 06 18:56:37 GNUtoo|laptop: i think there should be the way, just some research is needed Jul 06 18:57:01 ok Jul 06 18:59:24 PaulFertser, thanks it did the trick,I've the previous error now Jul 06 18:59:26 [ 244.483032] block 313 is bad Jul 06 18:59:38 I was told it came from jffs2 Jul 06 18:59:55 I did that to mount: Jul 06 18:59:55 "hey wait i've got a new complaint; forever in debt to your priceless advice" Jul 06 19:00:01 <[Rui]> well, I compiled elmdentica with -O0 and so far I can see that selecting from sqlite3 and displaying the merged timelines doesn't cause complaints Jul 06 19:00:04 mount /dev/mtdblock3 ./mnt/ -t yaffs2 Jul 06 19:00:37 <[Rui]> there are a few libjson complaints, though Jul 06 19:00:38 PaulFertser, thanks a lot for the help!!! Jul 06 19:00:47 <[Rui]> and the libjson guy doesn't answer me :( Jul 06 19:01:10 GNUtoo|laptop: do you mean anything works now? Or should i try to suggest anything else? Jul 06 19:01:22 PaulFertser, I'll explain: Jul 06 19:01:23 <[Rui]> also libcurl Jul 06 19:01:41 at the beginning I had that error: Jul 06 19:01:47 [ 244.483032] block 313 is bad Jul 06 19:01:53 then I bricked even more my flash Jul 06 19:02:23 Q-Master, that's not the correct quote! And furthermore, I am paranoid! Jul 06 19:02:23 with the -m command Jul 06 19:02:37 now I've recovered to the state before -m Jul 06 19:02:42 TAsn: mine _is_ correct Jul 06 19:02:42 but I still can't mount it Jul 06 19:02:48 I was told my flash was dead Jul 06 19:02:56 but I'm not shure it's true Jul 06 19:03:13 TAsn: "sit and drink pennyroyal tea. Still a life is inside of me" Jul 06 19:03:18 maybe I should mark only this block as bad Jul 06 19:03:21 and reflash? Jul 06 19:03:50 PaulFertser, yep. Jul 06 19:04:04 GNUtoo|laptop: hm, so that's something yaffs2 tells you and hence refuses to mount? Jul 06 19:04:28 that's what I was told Jul 06 19:04:33 I'm not shure if it's true Jul 06 19:04:46 LOL Jul 06 19:04:48 looser! Jul 06 19:04:54 http://pastebin.com/LYjyEDdf Jul 06 19:04:58 that's what I have Jul 06 19:05:00 such an idiot for not passing before. Jul 06 19:05:19 I'll grep yaffs2 sources Jul 06 19:06:05 yaffs_guts.c: (TSTR("block %d is bad" TENDSTR), blk)); Jul 06 19:06:10 so I think so Jul 06 19:09:34 that's in a if block Jul 06 19:09:41 } else if (state == YAFFS_BLOCK_STATE_DEAD) { Jul 06 19:09:58 and Jul 06 19:10:01 if (state == YAFFS_BLOCK_STATE_DEAD) { Jul 06 19:10:15 basically it scans the blocks Jul 06 19:10:20 /* Scan all the blocks to determine their state */ Jul 06 19:10:23 and do that: Jul 06 19:10:27 if (state == YAFFS_BLOCK_STATE_DEAD) { Jul 06 19:10:37 GNUtoo|laptop: but it should continue mounting after skipping it, i guess... Jul 06 19:10:45 <[Rui]> TAsn: well, I can't find leaks Jul 06 19:10:48 it doesn't Jul 06 19:11:08 mount: wrong fs type, bad option, bad superblock on /dev/mtdblock3,... Jul 06 19:11:54 GNUtoo|laptop: i wonder what will happen if you try jffs2 right after flasherase_all Jul 06 19:12:02 I'll try Jul 06 19:12:50 PaulFertser, lol Jul 06 19:12:52 it works Jul 06 19:13:03 GNUtoo|laptop: heh :) Jul 06 19:13:10 GNUtoo|laptop: not sure what is the moral :) Jul 06 19:13:18 maybe my image is corupted? Jul 06 19:13:35 Or possibly some yaffs odditie Jul 06 19:13:47 maybe Jul 06 19:21:09 leviathan, hi Jul 06 19:21:19 leviathan, I am recovering my nand Jul 06 19:21:53 GNUtoo|laptop: that last sentence sounds great \o/ Jul 06 19:22:07 yes Jul 06 19:22:11 I was pointed to: Jul 06 19:22:18 http://lists.infradead.org/pipermail/linux-mtd/2009-May/025462.html Jul 06 19:22:28 in an android channel Jul 06 19:22:51 <[Rui]> TAsn: and it still complains about loads of EFL stuff. Jul 06 19:22:54 I meant "great" in an ironic sense. Jul 06 19:23:55 ok Jul 06 19:24:20 [ 1744.493865] msm_nand_read_oob 2e40000 0 1c failed -74, corrected 0 Jul 06 19:24:41 I've such things before the block 313 is bad Jul 06 19:24:45 Hmm Jul 06 19:24:51 ok Jul 06 19:34:04 PaulFertser, nandtest /dev/mtd3 returns no error Jul 06 19:34:11 ah I didn't see that Jul 06 19:34:17 ECC failures Jul 06 19:34:20 4 Jul 06 19:34:57 GNUtoo|laptop: if i really wanted to make that work, i'd try to find out several things: Jul 06 19:35:26 1. If hardware or software ecc should be used, and whether its consistently set in nand driver and fs driver and everywhere that access nand Jul 06 19:35:39 ok Jul 06 19:35:51 2. Where does the kernel store information about bad blocks. Is it BBT (where?) or is it OOB. Jul 06 19:36:08 3. If it's OOB, i'd find out the ioctls allowing me to fully erase it. Jul 06 19:36:45 And then i'd try afresh. Jul 06 19:36:59 ok Jul 06 19:37:44 GNUtoo|laptop: there's Werner's utility that might be useful as a reference. Jul 06 19:38:01 GNUtoo|laptop: http://svn.openmoko.org/developers/werner/badnand/ Jul 06 19:38:11 thanks a lot!!! Jul 06 19:39:36 :D Jul 06 19:39:53 Q-Master: you! made me listen to Nirvana again! Jul 06 19:40:38 It's in fact ain't that bad, as i was feeling good enough beforehand. Jul 06 19:42:35 hi all Jul 06 19:42:38 MEMREADOOB: Invalid argument Jul 06 19:43:03 so maybe it's not oob Jul 06 19:43:23 I disappeared for some time because I don't have much free time to follow shr mailing lists and channel too but once in a while I download SHR, give it a try and report here Jul 06 19:43:36 and on my openmoko.org page Jul 06 19:44:06 I tested latest shr unstable and iliwi is improving a lot, now I am able to connect to my wpa2 wifi network without problem and download the latest updates Jul 06 19:44:33 nice Jul 06 19:44:48 now we only need a GUI package management for the masses Jul 06 19:45:23 an important thing it lacks now that should be installed by default is SHR-installer Jul 06 19:45:29 dos1, ^^ Jul 06 19:45:44 so people could easily download and install new applications Jul 06 19:46:07 graziano: opkg is not easy enough? Jul 06 19:46:16 graziano: did you test it? Jul 06 19:47:31 opkg is great for me but not for the masses Jul 06 19:47:38 btw Jul 06 19:47:45 GNUtoo|laptop, is right Jul 06 19:47:47 leviathan, how's the bad block handled? Jul 06 19:48:23 dos1, what's the exact name of the package? Jul 06 19:48:32 graziano: shr-installer Jul 06 19:48:36 GNUtoo|laptop: opkg is utter crap for anybody. But it's similar enough to apt-get, so the masses shouldn't have any problems i guess. Jul 06 19:49:00 lol I know Jul 06 19:49:11 for the utter crap Jul 06 19:49:17 it's not very developed Jul 06 19:49:24 what is inside factory NAND partition (mtd5)? Jul 06 19:49:25 but you can report bugs Jul 06 19:49:46 Martix__, I'm talking about my htcdream Jul 06 19:49:54 PaulFertser, as we both agreed months ago you are not a good user experience person to test with ;) Jul 06 19:50:03 I'll unbrick my freerunner later Jul 06 19:50:16 GNUtoo|laptop: Im talking about my Neo FreeRunner :-) Jul 06 19:50:24 ok Jul 06 19:50:30 * graziano installs shr-installer Jul 06 19:50:44 and Im curious what is inside mtd5 Jul 06 19:50:47 no icon Jul 06 19:50:53 is it normal or should I reboot? Jul 06 19:51:19 graziano: guess refresh of icons is broken Jul 06 19:51:51 ehi mrmoku :)) Jul 06 19:52:06 will try rebooting Jul 06 19:53:43 true Jul 06 19:53:51 rebooted, now the icon is there Jul 06 19:53:59 Martix__: mac addresses, imea, serial Jul 06 19:53:59 oh fuck Jul 06 19:54:00 ! Jul 06 19:54:02 it's NICE! Jul 06 19:54:02 :) Jul 06 19:54:11 * graziano looks dos1 Jul 06 19:54:20 graziano: i tend to forget about the things i do not like ;) Jul 06 19:54:39 ;) Jul 06 19:54:39 leviathan, hi how does htcdream nand works: Jul 06 19:54:52 *ecc: how does it works Jul 06 19:55:32 PaulFertser: is it used anyway in current distributions? Jul 06 19:56:03 Martix__: it is Jul 06 19:56:41 Martix__: it's used by the bootloader to set mac address for the usb networking. Jul 06 19:57:00 GNUtoo|laptop: you can learn from the datasheet Jul 06 19:57:05 Martix__: and shr bootscripts are capable of that as well Jul 06 19:57:15 how can I restore it? Jul 06 19:57:18 GNUtoo|laptop: that the ecc functionality is very poor Jul 06 19:57:18 leviathan, ah there are public datasheets? Jul 06 19:57:25 GNUtoo|laptop: no Jul 06 19:57:27 ok Jul 06 19:57:29 so tell me Jul 06 19:57:39 so I don't have to read something leaked Jul 06 19:57:40 ok Jul 06 19:57:46 I've 4 ecc errors Jul 06 19:57:46 mmm iliwi did not save my wpa2 password Jul 06 19:57:49 with nandtest Jul 06 19:58:09 note for the dev if he's here. iliwi should ask me to remember the wifi pass Jul 06 19:58:15 I am a lazy guy :) Jul 06 19:58:15 Martix__: with flasherase_all, then mkfs.ext2, then copy the contents from someone else's device Jul 06 19:58:19 GNUtoo|laptop: basicaly there is ECC there Jul 06 19:58:25 but you can forget about it Jul 06 19:58:30 its not of ANY use Jul 06 19:58:34 Martix__: or even better dump it from some other device, then flasherase_all, then nandwrite it. Jul 06 19:58:49 or better: qualcomms hardware is crap Jul 06 19:58:51 -.- Jul 06 19:59:13 leviathan, ok Jul 06 19:59:18 PaulFertser: thats what I doing now, but mac addresses, imea, serial are unique numbers Jul 06 19:59:38 Martix__: you don't have to, i have this partition borked and i live with it :D Jul 06 19:59:40 leviathan, for bad block it uses oob? Jul 06 19:59:42 so I should hexedit this image first :-) Jul 06 19:59:49 Martix__: IMEA and serial are written on the sticker inside the battery compartment. Also noone uses that data anyway. Jul 06 19:59:52 Martix__: just your USB MAC will be random at each boot Jul 06 19:59:58 leviathan, basically PaulFertser asked me that: Jul 06 20:00:19 Martix__: you can edit it after having flashed as well. Jul 06 20:00:25 PaulFertser: and USB MAC? Jul 06 20:00:33 http://pastebin.com/eNqfnByw Jul 06 20:01:03 Martix__: who cares if it'll be the same as mine? Chances we'll need to plug our devices at the same time in the same PC are very low. Jul 06 20:01:23 :-) Jul 06 20:08:30 updated: http://wiki.openmoko.org/wiki/User:Graziano Jul 06 20:08:33 feel free to edit as usual Jul 06 20:13:05 I'll try to disable ecc Jul 06 20:13:15 bye Jul 06 20:14:26 GNUtoo|laptop: some kind of ECC should be there to work properly Jul 06 20:14:37 GNUtoo|laptop: so you need Software ECC consistently everywhere Jul 06 20:17:24 hi everyone Jul 06 20:22:40 ok Jul 06 20:22:49 I'll make ARCH=arm xconfig Jul 06 20:35:26 btw where can I find a kernel small enough for uboot for my freerunner? Jul 06 20:35:48 GNUtoo|laptop: what's the size limit? Jul 06 20:35:53 PaulFertser: factory partition seems like corrupted ext2, thats strange Jul 06 20:36:00 lindi-, 2M I think Jul 06 20:36:07 GNUtoo|laptop: what creates that limit? Jul 06 20:36:13 lindi-, uboot Jul 06 20:36:18 I don't have more details Jul 06 20:36:19 GNUtoo|laptop: are you sure? Jul 06 20:36:30 I'll look in the wiki Jul 06 20:36:37 i wouldn't trust the wiki :( Jul 06 20:36:39 GNUtoo|laptop: you can increase it Jul 06 20:36:46 ok I'll try Jul 06 20:36:51 more is on wiki Jul 06 20:36:52 as soon as I get commandline on nor uboot Jul 06 20:37:04 but ttyACM0 was not found Jul 06 20:37:10 that's why I didn't try that yet Jul 06 20:38:04 Martix__: it is ext2 Jul 06 20:38:12 GNUtoo|laptop: no Jul 06 20:38:31 GNUtoo|laptop: the limit is the command in u-boot env that says "fetch first 2M from NAND". Jul 06 20:38:36 ah ok Jul 06 20:38:37 GNUtoo|laptop: one can easily make it 3M Jul 06 20:38:38 that's all Jul 06 20:38:42 ok Jul 06 20:38:48 nice to know Jul 06 20:39:00 I tought uboot was not great because of that lol Jul 06 20:39:42 well, it would be nice if it could automaticaly determine kernel size, like Qi :P Jul 06 20:40:14 when I mount /dev/mtdblock3 and do a ls I've Jul 06 20:40:20 ./lost+found Jul 06 20:40:28 PaulFertser: but I cant mount it, fsck didnt help even I dont need it :-) Jul 06 20:40:36 so I bet nandwrite failed Jul 06 20:40:39 somewhat Jul 06 20:40:55 GNUtoo|laptop: can you mount mtd5? Jul 06 20:41:28 Martix__: that's why i say you should flasherase_all it, and flash anew. Jul 06 20:41:38 Martix_, same issue Jul 06 20:42:20 PaulFertser: I am trying to mount backuped image :-) Jul 06 20:43:29 ah nice /dev/ttyACM0 works on freerunner Jul 06 20:45:33 Martix__: -o loop? Jul 06 20:45:39 sure Jul 06 20:51:03 it has ext2 errors, and after fsck: mount: Stale NFS file handle Jul 06 20:51:24 Martix__: SHR? Jul 06 20:51:36 Fedora 13 :-) Jul 06 20:51:45 on openmoko? Jul 06 20:51:50 no Jul 06 20:52:06 off topic then :) Jul 06 20:52:24 wait scping to FR :-) Jul 06 20:52:46 Martix__: i don't think ext2 runs on mtd devices Jul 06 20:52:57 or am I missing something here? Jul 06 20:53:16 factory partition in mtd5 is strange :-) Jul 06 20:53:19 I am, something Jul 06 20:53:24 lindi-: it runs suboptimally, the guys decided that's a good choice for the factory partition Jul 06 20:53:42 $ sudo file -sL /dev/mtd* | nc paste.dyndns.org 1234 Jul 06 20:53:43 http://paste.debian.net/79838/ Jul 06 20:55:05 lol, the kernel is DOS executable (device driver) :D Jul 06 21:05:22 http://pastebin.com/Dfsf1G3M Jul 06 21:05:23 hmmm Jul 06 21:13:30 lindi-, not shure...maybe it can but only in readonly Jul 06 21:20:10 nothing work Jul 06 21:20:23 neither dream or freerunner Jul 06 21:20:53 dream: I write files to the yaffs2 filesystem with nandwirte or after mounting it, they don't stay Jul 06 21:21:38 freerunner: I can't load a file from sd Jul 06 21:22:35 GNUtoo|laptop: btw I have a same issue like you with: "No such Alternate Setting: "u-boot"" and its not related to Android Jul 06 21:23:01 Martix__, ok how can I load qi to flash it? Jul 06 21:23:14 I want to load it from sd Jul 06 21:23:17 then flash it Jul 06 21:23:20 all from uboot Jul 06 21:26:01 dream: [ 1158.145111] Partially written block 84 detected Jul 06 21:26:04 what does that mean? Jul 06 21:28:01 GNUtoo|laptop: you should load image from uSD to RAM, its same command set as u-boot does when loading kernel Jul 06 21:28:21 GNUtoo|laptop: mmcinit, ext2load... Jul 06 21:28:22 Martix__, yes but ext2ls doesn't seem to work Jul 06 21:28:46 why ext2ls? you dont know filename? Jul 06 21:29:06 it was just a check Jul 06 21:29:10 I'll rename the file Jul 06 21:29:41 and then use nand write Jul 06 21:29:57 write it from RAM to NAND Jul 06 21:32:07 ext2load mmc 1 0x32000000 qi Jul 06 21:32:13 data abort Jul 06 21:32:14 pc : [<33fa1f98>] lr : [<33fa253c>] Jul 06 21:32:15 etc... Jul 06 21:32:55 ahh Jul 06 21:32:58 mmc errors Jul 06 21:33:07 I'll change mmc card Jul 06 21:41:30 GNUtoo|laptop: I hope to fix it in cooperation with daniel walker Jul 06 21:41:51 ok Jul 07 01:21:42 does FSO2 has an eventd anymore? Jul 07 01:23:57 does FSO still work? **** ENDING LOGGING AT Wed Jul 07 02:59:56 2010