**** BEGIN LOGGING AT Thu Aug 26 02:59:57 2010 Aug 26 05:23:15 gena2x: hi, here's patch to reduce HZ value on s3c24xx: http://downloads.tuxfamily.org/linuxrx1950/patches/random/0001-Use-100-as-HZ-value-on-S3C24XX.patch Aug 26 05:24:01 it's against 2.6.36-rc and doesn't apply even for 2.6.35, but it's pretty simple, so you can adapt it for 2.6.34 Aug 26 05:25:14 gena2x: btw, it doesn't affect on power consumption on my PDA, but you should test it with your gta02 :) Aug 26 07:25:42 <\marco> hi all Aug 26 07:26:38 <\marco> I've just reflashed my FR with the last shr-u but I've the same problem.. Aug 26 07:26:47 <\marco> opkg upgrade says: Aug 26 07:26:55 <\marco> Configuring util-linux-ng-readprofile. Aug 26 07:26:55 <\marco> update-alternatives: Error: cannot register alternative readprofile to /sbin/readprofile since it is already registered to /usr/sbin/readprofile Aug 26 07:27:16 update-alternatives --remove readprofile ../../bin/busybox Aug 26 07:27:19 and opkg upgrade agin Aug 26 07:28:12 <\marco> thank you JaMa|GoNe,it works :) Aug 26 07:29:11 <\marco> I had tried update-alternatives but my syntax was wrong :) Aug 26 07:31:09 \marco: http://lists.shr-project.org/pipermail/shr-user/2010-August/006042.html :) Aug 26 07:45:29 <\marco> JaMa|GoNe: :| very strange.. I've search through the mailing list but .. don't know.. maybe I was drunk :D.. thanks again :) Aug 26 07:53:11 good morning everybody Aug 26 07:54:52 I vaguely remember being able to boot my Freerunner into a "rescue" installation of Linux, but I can't seem to find how I did that. I booted from NOR, but it still boots SHR Aug 26 07:55:23 my freerunner is completely out of storage space, so unpacking /etc/dev.tar doesn't work any more, rendering my Freerunner unbootable Aug 26 08:15:04 <\marco> is there a python module to easily interface with /sys ? Aug 26 08:15:55 \marco: libomhacks is a C library, would you need python bindings? Aug 26 08:16:32 <\marco> oh.. I don't know libomhacks :) Aug 26 08:17:46 <\marco> lindi-: is there a python binding? Aug 26 08:18:32 \marco: not yet at least Aug 26 08:20:20 <\marco> well, no problem :) It isn't a primary need for my little project.. I'll use a brutal echo with os.system :D Aug 26 08:22:37 \marco: use brutal os.system("om ...") Aug 26 08:22:49 \marco: that way you don't need to hardcode /sys paths Aug 26 08:23:48 <\marco> lindi-: nice! Aug 26 08:23:56 <\marco> lindi-: thank you :) Aug 26 08:25:03 <\marco> I'm writing a little "dialog" to set usb-mode ( network, storage, host mode ) Aug 26 08:27:57 <\marco> lindi-: mhm..I reading the om help.. is it possible to read the usb-mode status? Aug 26 08:28:57 <\marco> you can set it.. but how you can read the current mode? Aug 26 08:30:34 \marco: just "om usb mode" Aug 26 08:30:45 <\marco> and... setting usb mode isn't working.. maybe it is linked to previous path? Aug 26 08:30:57 \marco: what kernel? Aug 26 08:31:06 <\marco> indeed Aug 26 08:31:15 <\marco> 2.6.32 Aug 26 08:31:25 <\marco> it is searching for /sys/devices/platform/s3c-ohci/usb_mode Aug 26 08:31:56 <\marco> the new path is /sys/devices/platform/s3c2410-ohci/usb_mode Aug 26 08:33:01 \marco: ah good catch, I'll add it as one of the alternatives it tries Aug 26 08:33:25 <\marco> lindi-: thank you :) Aug 26 08:36:21 \marco: is there something else that does not work? Aug 26 08:37:08 <\marco> lindi-: mhm let me check Aug 26 08:38:16 <\marco> lindi-: usb charger mode Aug 26 08:38:40 <\marco> it looks to /sys/class/i2c-adapter/i2c-0/0-0073/neo1973-pm-host.0/hostmode Aug 26 08:38:46 <\marco> and the new path is: Aug 26 08:39:11 <\marco> /sys/devices/platform/s3c2440-i2c/i2c-0/0-0073/pcf50633-gpio/reg-fixed-voltage.2/gta02-pm-usbhost.0/power_on Aug 26 08:40:27 \marco: power_on? Aug 26 08:40:33 interesting path Aug 26 08:40:44 <\marco> well, I'm looking to usb-script Aug 26 08:40:50 <\marco> maybe I'm wrong Aug 26 08:40:53 \marco: what's usb-script? Aug 26 08:41:01 <\marco> usbmode script sorry Aug 26 08:41:02 <\marco> :D Aug 26 08:41:08 ah you are writing one? Aug 26 08:41:11 <\marco> on wiki.openmoko.org Aug 26 08:41:15 <\marco> nono Aug 26 08:41:21 ah, got a direct link? Aug 26 08:41:57 <\marco> http://wiki.openmoko.org/wiki/Neo1973_USB_host#USB-Mode-Script_.282.6.28.2B_kernels.29 Aug 26 08:42:18 <\marco> the universal script contains all paths Aug 26 08:42:29 interesting Aug 26 08:43:29 <\marco> resume-reason ins't working too Aug 26 08:44:09 \marco: I only tested 2.6.32 from git.openmoko.org and it didn't have usb host or resume_reason. SHR added those, it's nice to hear about these issues Aug 26 08:44:22 <\marco> :) Aug 26 08:45:05 <\marco> the new path for resume_reason is: /sys/devices/platform/s3c2440-i2c/i2c-0/0-0073/resume_reason Aug 26 08:46:04 still at "om usb charger-mode" ... Aug 26 08:46:34 <\marco> eheh don't worry ^_^ Aug 26 08:46:39 \marco: but there were 2 resume_reasons before.. Aug 26 08:46:54 \marco: the 2nd was used in FSO and it's not available anymore Aug 26 08:47:23 <\marco> I've searched through /sys with find and I've only one resume_reasons Aug 26 08:47:35 ./fsousaged/src/plugins/lowlevel_openmoko/plugin.vala:internal const string SYSFS_RESUME_REASON_PATH = "/class/i2c-adapter/i2c-0/0-0073/neo1973-resume.0/resume_reason"; Aug 26 08:47:38 ./fsousaged/src/plugins/lowlevel_openmoko/plugin.vala:internal const string SYSFS_RESUME_SUBREASON_PATH = "/class/i2c-adapter/i2c-0/0-0073/resume_reason"; Aug 26 08:48:13 <\marco> yep, that one Aug 26 08:48:14 \marco: what about "om usb charger-limit"? Aug 26 08:48:29 <\marco> it's working :) Aug 26 08:49:00 \marco: good Aug 26 08:49:16 and subreason is used when REASON == "EINT09_PMU" Aug 26 08:49:21 <\marco> the others interface seem working Aug 26 08:49:25 \marco: you can see "om battery consumption" change when regulating charging limit? Aug 26 08:49:50 \marco: there are two resume_reason sysfs nodes, do you have both? Aug 26 08:50:03 \marco: 1) /sys/class/i2c-adapter/i2c-0/0-0073/neo1973-resume.0/resume_reason Aug 26 08:50:13 \marco: 2) /sys/class/i2c-adapter/i2c-0/0-0073/resume_reason Aug 26 08:50:40 <\marco> lindi-: I've only one resume_reason Aug 26 08:50:51 \marco: with only one resume_reason we can't really work well Aug 26 08:51:22 but I'll fix the other one still Aug 26 08:51:41 <\marco> lindi-: om battery consumption change when I regulate chargin limit :) Aug 26 08:52:07 \marco: good Aug 26 08:52:15 the one which is available is not even used by FSO.. because 10:48:03 < JaMa|GoNe> and subreason is used when REASON == "EINT09_PMU" Aug 26 08:52:19 \marco: what do you read in /sys/devices/platform/s3c2440-i2c/i2c-0/0-0073/resume_reason exactly? Aug 26 08:52:32 JaMa|GoNe: omhacks tries to use both Aug 26 08:53:16 <\marco> well, sometimes it has strange value.. relative values ( -27375 and similar ) Aug 26 08:53:17 lindi-: what can you read from subreason? fso checks only 2 options (of EINT09_PMU) Aug 26 08:53:20 intmap2.insert( "0000000200", ResumeReason.LowBattery ); Aug 26 08:53:23 intmap2.insert( "0002000000", ResumeReason.PowerKey ); Aug 26 08:53:36 <\marco> lindi-: 0400000000 Aug 26 08:54:18 JaMa|GoNe: button, usb_connect, usb_disconnect, rtc_alarm Aug 26 08:54:35 <\marco> 0400000000 is usb_connect, I suppose Aug 26 08:54:43 \marco: looks ok Aug 26 08:54:49 <\marco> good :) Aug 26 08:55:24 <\marco> lindi-: what about battery consumption oscillating from positive and negative values? Aug 26 08:55:45 <\marco> it's related to charging? Aug 26 08:56:18 \marco: please run while true; do om battery consumption; sleep 1; done for two minutes and put the output to http://paste.debian.net Aug 26 08:57:14 JaMa|GoNe: omhacks probably could use lowbattery check Aug 26 08:57:22 JaMa|GoNe: have you tested that it works? Aug 26 08:57:23 <\marco> lindi-: my FR is completely charged.. could be the reason? Aug 26 08:57:47 \marco: yes but I can tell you more if I see the numbers :) Aug 26 08:59:09 <\marco> lol... just a moment.. the nice feature that suspend the FR when it's full charged has strike Aug 26 08:59:10 anarsoul: hey. Aug 26 08:59:35 anarsoul: i've failed to bring up my ar6000, so didn't test your patch yet :( Aug 26 09:00:46 xrmx: did you ar6000 quest were successfull? Aug 26 09:01:12 \marco: I think we'll do a new release of omhacks today. please let me know if you find any other problems Aug 26 09:01:27 xrmx: for me, all components seem in place, but i see no network device. Aug 26 09:01:49 xrmx: something to power on? Aug 26 09:01:52 gena2x, nope, compiled with -DDEBUG but no debug messages, probably i've done something wrong but no idea Aug 26 09:02:26 i can bind/unbind sdi driver, i see ar module loading. Aug 26 09:02:27 gena2x, i have the device, it is powered on, it's something at a higher layer or subtle in the driver because i can't get a dhcp address Aug 26 09:02:44 hm, how did you power it? Aug 26 09:02:50 <\marco> lindi-: ok! :) let me finish the battery consumption test Aug 26 09:02:56 any sysfs node? Aug 26 09:03:09 should i do something with rfkill? Aug 26 09:05:48 gena2x, openmoko-panel Aug 26 09:06:15 lindi-: haven't tested it yet (as I don't have use for it without main reason available in 2.6.32..) Aug 26 09:06:40 xrmx: which kernel you using? Aug 26 09:07:17 JaMa|GoNe: maybe omhacks should read it from /dev/mem? ;-) Aug 26 09:07:21 gena2x, 2.6.29 Aug 26 09:07:41 :) Aug 26 09:08:03 xrmx: ah, sorry. my troubles are definely .34-specific :( Aug 26 09:08:05 <\marco> lindi-: http://paste.debian.net/85918 but seems ok.. ( and the led is orange ) Aug 26 09:08:23 lindi-: porting old resume_reason to 2.6.32 seems better way, but if you can do it with /dev/mem :) Aug 26 09:08:55 <\marco> maybe I've read the consumption while it was switching from orange to blue Aug 26 09:09:06 do wifi works .32? any what is power-up sequence? Aug 26 09:09:27 s/any/anyone knowns/ Aug 26 09:09:27 gena2x meant: do wifi works .32? anyone knowns what is power-up sequence? Aug 26 09:09:54 gena2x: :( Aug 26 09:10:11 anarsoul: i didn't resign atm :) Aug 26 09:10:37 <\marco> gena2x: my wifi is working Aug 26 09:11:07 gena2x: for me wifi works too on .32 and .34 Aug 26 09:11:17 aha, thanks. Aug 26 09:11:33 what about power-up sequence? Aug 26 09:11:40 some sys node? Aug 26 09:11:45 gena2x: for me too, powered up by fsousaged (rfkill) Aug 26 09:12:21 gena2x: ok :) Aug 26 09:13:51 something like debian-gta02:/sys/bus/platform/devices/gta02-pm-wlan.0/rfkill/rfkill1# echo 1 > state ? Aug 26 09:14:48 or it should happen automagically? Aug 26 09:15:40 <\marco> lindi-: thanks again for the great utility :) I'll let you know if I find any other problems :) Aug 26 09:15:57 <\marco> gena2x: use om wifi power 1 Aug 26 09:16:22 <\marco> gena2x: opkg intall omhacks Aug 26 09:17:18 \marco: my om seem has no support for this on .34 Aug 26 09:18:02 <\marco> lol :D Aug 26 09:18:05 <\marco> lindi-: ping Aug 26 09:18:32 it wants to touch /sys/bus/platform/drivers/s3c2440-sdi/s3c2440-sdi Aug 26 09:18:32 <\marco> maybe path is changed again in .34 Aug 26 09:19:37 but true path is echo s3c2440-sdi >/sys/bus/platform/drivers/s3c-sdi/unbind Aug 26 09:19:49 or bind, as appropriate Aug 26 09:20:09 radekp: did you done something special to make it work on .34? Aug 26 09:20:26 <\marco> gena2x: what about iliwi? it works? Aug 26 09:20:53 <\marco> I don't have that path too.. but iliwi it's working Aug 26 09:22:09 <\marco> lindi-: om wifi power seems to work ( it returns 0 ) but with strace I see a no such file... indeed setting power to 1 don't work.. Aug 26 09:22:25 ili-what? i'm on debian :) Aug 26 09:22:25 <\marco> lindi-: sorry to not have found it before :/ Aug 26 09:22:48 <\marco> gena2x: oh.. :D sorry I suppose you were on shr-u :P Aug 26 09:28:32 \marco: iliwi is using fsousaged and that's using rfkill Aug 26 09:29:18 <\marco> JaMa|GoNe: oh.. I see :) Aug 26 09:29:54 JaMa|GoNe: sorry, i'm not really understand about rfkill, how to 'use' it? i've read that rfkill is kernel subsystem to control radio emitters, but how to apply this to our wifi? **** BEGIN LOGGING AT Thu Aug 26 09:33:24 2010 Aug 26 09:33:25 gena2x: I don't know internals either.. but you can check fso source.. Aug 26 09:33:47 gena2x: but iirc it starts with /dev/rfkill Aug 26 09:34:22 JaMa|GoNe: ah, ok. thanks... Aug 26 09:34:25 fsodeviced/src/plugins/kernel26_rfkill/plugin.vala Aug 26 09:34:35 \marco: yes I don't know how wifi power control should be done in 2.6.32 Aug 26 09:35:01 \marco: fso uses rfkill but I don't know if it works. see the bugs about increased power consumption and all that Aug 26 09:37:45 <\marco> lindi-: :D eheh.. I agree, it' a good idea to wait a little :) Aug 26 09:39:37 <\marco> I've to go.. have a good day guys! :) Aug 26 09:39:40 <\marco> bye! Aug 26 09:43:04 JaMa|GoNe: thanks, rfkill helped to bring it little further. Aug 26 09:48:47 gena2x: any progress? Aug 26 09:49:56 anarsoul: still no wifi even without patch :( Aug 26 09:50:25 i can try to test in .32, but want .34. Aug 26 09:59:19 gena2x: maybe it make sense to test on .32? at least we'll know if these patches bring some improvement (reduced cpu usage and or higher download/upload speeds) Aug 26 10:00:59 but of course, it's better to test on .34 Aug 26 10:01:49 hm. you think it will bring higher speed? Aug 26 10:02:44 i expected only less cpu usage :) Aug 26 10:03:23 well, now s3cmci uses cpu for transfers, and under high cpu load (i.e. while browser renders page) it can't transfer much data Aug 26 10:03:44 ah, in this way :) Aug 26 10:04:59 my poor ar6000 just hangs somewhere while modprobe... Aug 26 10:05:29 with patch? Aug 26 10:05:57 anarsoul: nope, nope. i'll say than i'll start testing it. Aug 26 10:06:46 hm. i guess my english is much worse than ar6000 code... Aug 26 10:10:12 xrmx: btw, it successfully shows me tons of debugging data with EXTRA_CFLAGS=-DDEBUG and 1 commenting in code. Aug 26 10:15:56 gena2x, compiling 2.6.34 right now, will check with that Aug 26 10:16:21 xrmx, nono. it should work with any kernel version. Aug 26 10:16:37 and in .34 it can't even start for me. Aug 26 10:18:12 gena2x, this is what i have in my 2.6.29 kernel http://pastebin.com/vrUkSUcq Aug 26 10:22:50 radekp: if you have wifi working on .34, maybe you'd like to test some patches that may improve wifi performance (or at least reduce cpu usage during wifi transfers)? Aug 26 10:23:38 radekp: or just say how did you bring up ar6000 on .34 ;) Aug 26 10:24:25 anarsoul: hi, yes, sure i want Aug 26 10:24:25 xrmx: hm. did you do make clean? Aug 26 10:24:44 gena2x, yes, i build a debian kernel Aug 26 10:24:58 radekp: 0021 and 0022 from http://downloads.tuxfamily.org/linuxrx1950/patches/bjdooks-next-samsung/25_aug_2010/ Aug 26 10:25:33 gena2x: just git clone mine tree http://github.com/radekp/linux-2.6 with config, checkout branch qtmoko-v25 and build Aug 26 10:25:55 gena2x: it should work, you'll get same kernel as is in qtmoko v25 release Aug 26 10:26:27 gena2x: and it's the same kernel as in SHR - only different config and without glamo drm patches Aug 26 10:26:49 radekp: ok, thanks. i'll start with .config diffing... Aug 26 10:26:50 anarsoul: oki, i'll try Aug 26 10:26:54 radekp: it's also necessary to enable CONFIG_MMC_S3C_DMA in config Aug 26 10:27:02 anarsoul: oki Aug 26 10:27:25 radekp: please let me know about results :) Aug 26 10:29:41 ahh cool there is also patch for libts on that page Aug 26 10:32:22 gena2x, do you have some docs to cross-build a kernel "the SHR way"? Aug 26 10:35:48 ThibG: strange question. are you sure it is for me? :) usual cross-build kernel should work with shr. i bet you have to recompile whole shr to get proper shr kernel/module packages. shr patches are at: http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-openmoko-2.6.32, also shr uses kms config instead of fb config. Aug 26 10:36:16 hm ok Aug 26 10:43:08 anarsoul, what about http://downloads.tuxfamily.org/linuxrx1950/patches/bjdooks-next-samsung/25_aug_2010/0014-mmc-slow-fix.patch ? Aug 26 10:46:14 ThibG: SHR way is just "bb virtual/kernel" Aug 26 10:46:38 hm ok Aug 26 10:52:47 xrmx: it's not required Aug 26 10:53:13 xrmx: when DMA for s3cmci enabled I don't need it Aug 26 10:54:57 xrmx: it fixes card detection timeout on my rx1950, however I don't think that this fix will be accepted upstream :) Aug 26 10:55:20 anarsoul, :) Aug 26 10:59:50 anarsoul: i have now test environment ready :) will try netcat over wifi - do you think it's good test? Aug 26 11:00:57 radekp: should be good enought Aug 26 11:00:57 anarsoul: i am sending /dev/zero from freerunner to my PC, without your patch it eats ~28% of CPU Aug 26 11:01:22 and what's transfer rate? Aug 26 11:06:58 anybody using a panasonic microsd on fr? Aug 26 11:16:29 anybody seen GNUtoo :) Aug 26 11:16:58 in case you don't see it in the other log, I accomplished the guide Aug 26 11:17:14 GNUtoo: http://htc-linux.org/wiki/index.php?title=Rootfs/Userfriendly Aug 26 11:20:18 radekp: and what's cpu usage with patch? Aug 26 11:23:04 GNUtoo: I just added a ready made example initrd that has the exact example /init from the wiki included Aug 26 11:28:02 anarsoul: sorry had phone call, will test now Aug 26 11:28:36 anarsoul, booted with your s3c dma patch Aug 26 11:29:18 xrmx: nice, what about wifi? :) Aug 26 11:33:16 anarsoul, i don't have the device, i have a wsod but ssh works Aug 26 11:34:34 xrmx: some kernel misconfiguration? Aug 26 11:35:34 xrmx:are you on 2-4-2? then use http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-openmoko-2.6.32/0021-glamo-display-fix-WSOD-for-242-timming.patch Aug 26 11:35:40 anarsoul, the wifi is compiled for sure, my dmesg http://pastebin.com/gJDifihm Aug 26 11:35:54 xrmx: I'm talking about wsod :) Aug 26 11:36:00 gena2x, i am Aug 26 11:36:18 does someone knows where i can find 2-4-2 timing Qi version ? Aug 26 11:36:43 [ 1.350000] mmc0: new SDIO card at address 0001 Aug 26 11:36:46 anarsoul, i had the screen output, then it blanked, and wsod when i touched the screen Aug 26 11:36:47 xrmx: sdio card was detected Aug 26 11:37:07 GarthPS: yes, check list, it had links to appropriate qi by dos. Aug 26 11:37:45 gena2x: I am, thanks for specifying dos did Aug 26 11:38:58 gena2x, i can't find the patch on openembedded 2.6.34 openmoko kernel Aug 26 11:39:32 xrmx: I really doubt that my patch can cause wsod Aug 26 11:40:19 anarsoul, i'm not saying that, it's just that i have something more urgent before caring about wifi :) Aug 26 11:42:25 however, 183-185, 188-190 lines in your dmesg are strange Aug 26 11:43:06 xrmx: hm... it should be same. Aug 26 11:43:57 xrmx: let me check... Aug 26 11:44:40 gena2x, i don't have mfd/glamo/glamo-display.c Aug 26 11:45:32 xrmx: wait a bit... Aug 26 11:46:30 xrmx: it's in drivers/video/glamo-fb.c Aug 26 11:47:25 xrmx: i'll push it to web now... Aug 26 11:47:50 gena2x, it's around line 310 right? Aug 26 11:47:51 anarsoul: i dont have wifi after applying the patches too Aug 26 11:48:11 :( can you show your dmesg with and without patches? Aug 26 11:48:46 anarsoul: sure Aug 26 11:51:28 xrmx: http://www.bsdmn.com/openmoko/glamo/glamo242ws_34.diff Aug 26 11:51:55 radekp: can you please add this ^^^ to openmoko too? Aug 26 11:51:59 tfu Aug 26 11:52:04 to qtmoko of course :) Aug 26 11:52:39 gena2x: are you sure that dos did glamo timmings? not cpu overclock? Aug 26 11:53:22 GarthPS: yes... let me try to search... it did it in irc for sure, but i think posted to community too Aug 26 11:53:55 gena2x: tahnks . sorry to bother you Aug 26 11:54:13 gena2x, doesn't apply, the code is 360 lines before Aug 26 11:54:23 damn Aug 26 11:54:26 sorry Aug 26 11:54:32 anarsoul: http://activationrecord.net/radekp/pub/dmesg_orig.txt http://activationrecord.net/radekp/pub/dmesg_dma.txt Aug 26 11:54:45 gena2x, anyway no hurry i'll fix it manually Aug 26 11:54:47 gena2x: will test it now Aug 26 11:54:48 i forgot, i have ts patch also, which is before Aug 26 11:55:04 i have damn zoo of patches... Aug 26 11:55:31 gena2x: push those to some git repo.. please Aug 26 11:55:58 no wait, let me check, ts should be in glamo-core... Aug 26 11:55:58 JaMa|GoNe, agree, that is a concern for me too Aug 26 11:56:37 radekp: can you enable s3c24xx dma debugging (it's kernel option) and try again? Aug 26 11:57:20 radekp: CONFIG_S3C2410_DMA_DEBUG Aug 26 11:57:31 anarsoul: oki sure Aug 26 11:58:08 xrmx, hm.... Aug 26 11:58:50 xrmx: are you building shr kernel? Aug 26 11:59:05 xrmx: all ok with my patch for fb, lines are proper Aug 26 11:59:26 xrmx: may be something just different in your kernel version. Aug 26 11:59:32 gena2x, nope, i'm building om-gta02-2.6.34 plus random patches collected Aug 26 12:00:05 gena2x, the list is on openmoko-kernel Aug 26 12:00:12 patch should be glamofb_cmd_mode Aug 26 12:00:26 gena2x, it's around line 310 here Aug 26 12:00:38 few lines after '/* RGB interface needs vsync/hsync */' Aug 26 12:00:49 between 2 reg_writes Aug 26 12:00:51 gena2x, i'm already building Aug 26 12:01:17 anarsoul: i can also increase loglevel, should i? Aug 26 12:01:48 yes, please Aug 26 12:02:57 gena2x, http://pastebin.com/SuvNH0eh Aug 26 12:03:45 xrmx: possible it fixes not only 2-4-2, but also various 4-4-4 wsods. Aug 26 12:04:20 hm, but why our code is so different? Aug 26 12:04:26 i am checking this... Aug 26 12:06:00 gena2x, mine match openmoko git Aug 26 12:06:42 radekp: it aborted DMA for some reason (result == 2 == S3C2410_RES_ABORT), i.e. driver called s3c2410_dma_ctrl with DMAOP_FLUSH Aug 26 12:08:47 and I suspect it was called from s3cmci_prepare_dma :( Aug 26 12:08:53 anarsoul: http://activationrecord.net/radekp/pub/dmesg_dma_debug.txt Aug 26 12:10:31 radekp: try to comment out s3c2410_dma_ctrl with S3C2410_DMAOP_FLUSH in s3cmci_prepare_dma Aug 26 12:11:09 anarsoul: oki Aug 26 12:11:31 stop, don't do that Aug 26 12:11:35 oki :) Aug 26 12:12:05 radekp: insert printk-s before each s3c2410_dma_ctrl with DMAOP_FLUSH in args Aug 26 12:12:20 one is in finalize_request, second is in s3cmci_prepare_dma Aug 26 12:12:37 * anarsoul 'd like to know who flushes dma queue Aug 26 12:14:11 xrmx: damn, seem i missed some non-trivial merge in my local git some time ago. Aug 26 12:14:48 xrmx: so things clean. Aug 26 12:15:18 radekp: btw, did you reverted this patch? http://people.freedesktop.org/~rm/2.6.34-patches/0004-Add-patch-from-debian-kernel-that-should-help-with-W.patch Aug 26 12:17:33 anarsoul: i have this patch included Aug 26 12:17:59 ok Aug 26 12:26:10 radekp: that would be nice to get some debug info from ar6000 driver too Aug 26 12:26:55 anarsoul: i have inserted the printks in s3cmci.c http://pastebin.com/NcSWRyXM Aug 26 12:27:20 anarsoul: those line with >>>>>>>>>>> show the flush line Aug 26 12:27:43 xrmx: i have to learn git better, but now such thing may happen :( Aug 26 12:27:46 anarsoul: ehmm once more this is useless :) Aug 26 12:28:07 mmm, what about line numbers? :) Aug 26 12:28:13 or function names Aug 26 12:28:19 gena2x, yeah, some git skills help :) Aug 26 12:29:37 anarsoul: http://pastebin.com/aWNkYcbA Aug 26 12:30:01 btw, that's pretty hard to debug driver remotely :( Aug 26 12:30:02 anarsoul: after >>>>>>>>>>>>>> is function name Aug 26 12:30:07 anarsoul: yup :( Aug 26 12:30:48 radekp: please add mrq->data->error to printk in finalize_request Aug 26 12:33:49 with 2.6.34 /sys/bus/platform/drivers/s3c2440-sdi is now s3c-sdi Aug 26 12:34:27 yes Aug 26 12:34:39 anarsoul: -110 Aug 26 12:34:52 xrmx: but you device is still s3c2440-sdi Aug 26 12:35:12 xrmx: so you should bing s3c2440-sdi to s3c-sdi Aug 26 12:35:15 -ETIMEDOUT Aug 26 12:35:16 :) Aug 26 12:35:17 *bind Aug 26 12:35:59 gena2x, echo s3c2440-sdi > /sys/bus/platform/drivers/s3c-sdi/bind Aug 26 12:35:59 -bash: echo: write error: No such device Aug 26 12:36:11 xrmx: so it is already bind :) Aug 26 12:36:18 xrmx: try unbind first :) Aug 26 12:36:27 xrmx: it is normal for it to fail Aug 26 12:36:53 lindi-: nothing normal. just it is already bind. Aug 26 12:37:15 now i just have to bringup wifi, do you om usually? Aug 26 12:37:19 *use Aug 26 12:37:22 radekp: can you trace where this error come from? Aug 26 12:37:35 xrmx: try rfkill Aug 26 12:38:02 rfkill list Aug 26 12:38:04 gena2x, with your patch, no more wsod but screen stays blank Aug 26 12:38:17 xrmx: what? Aug 26 12:38:17 anarsoul: hmm i am not sure, never did something like this, how can i do it? Aug 26 12:38:28 xrmx: om screen power 1? Aug 26 12:38:31 anarsoul: printks? or is there something more advanced? Aug 26 12:38:54 gena2x, wsod Aug 26 12:39:02 xrmx: but as i already told, i can't bring up ar6000 :( Aug 26 12:39:03 printks should be enough, there's only 3 -ETIMEDOUT in drivers/mmc/core/* Aug 26 12:39:13 in sd_ops.c, in mmc_ops.c, in sdio_ops.c Aug 26 12:39:47 put printk with __func__ and __LINE__ near each of them Aug 26 12:39:52 ahh oki, i'll do Aug 26 12:39:54 xrmx: ah, pure om-gta02-2.6.34 misses some patches also Aug 26 12:40:28 xrmx: but first try blank/unblank few times Aug 26 12:41:08 xrmx: http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-openmoko-2.6.32/0020-mfd-glamo-Enable-FIFO-stage-for-the-LCD-engine-s-mem.patch Aug 26 12:41:20 gena2x, nope still white Aug 26 12:41:29 xrmx: http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-openmoko-2.6.32/0004-ar6000_delay.patch.patch Aug 26 12:41:49 xrmx: http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-openmoko-2.6.32/0016-jbt6k74-fix-WS.patch Aug 26 12:42:09 xrmx: last one is specially important to fix wsods. Aug 26 12:42:28 xrmx: btw, are you on shr or debian? kms or fb? Aug 26 12:43:20 gena2x, debian, fb, this patches http://people.freedesktop.org/~rm/2.6.34-patches/ plus the mdelay(1) we just added Aug 26 12:44:15 xrmx: ah, you have ar6000_delay Aug 26 12:44:32 xrmx: but miss http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-openmoko-2.6.32/0016-jbt6k74-fix-WS.patch Aug 26 12:44:38 add it Aug 26 12:44:55 anarsoul: i think it comes from sdio_ops.c Aug 26 12:45:06 radekp: let's try :) Aug 26 12:45:33 gena2x, i have the enable fifo stage Aug 26 12:46:02 xrmx: FIFO stage is also must have Aug 26 12:46:29 xrmx: i don't see it on list Aug 26 12:46:34 radekp: I doubt, because it attaches card with no pb Aug 26 12:46:41 xrmx: which number? Aug 26 12:46:51 [ 2.175000] mmc0: new SDIO card at address 0001 Aug 26 12:47:10 i.e. mmc_send_io_op_cond succeed Aug 26 12:47:29 gena2x, it's already in 2.6.34 git Aug 26 12:48:25 anarsoul: i have added printk to all those .c files before -ETIMEDOUT and got only the one from sdio_ops.c in printk Aug 26 12:48:37 ah, I see Aug 26 12:49:27 anarsoul: i have also the others but they are after the mrq->data->error != 0 check Aug 26 12:50:01 so, -ETIMEDOUT came from mmc_send_io_op_cond Aug 26 12:50:21 can you find it's references in ar6000 driver? Aug 26 12:50:30 xrmx: yes, right. see it, but others should be missing Aug 26 12:50:33 yes, ok, i'll try Aug 26 12:51:17 xrmx: FIFO patch were missing in .34 for a while, i forgot that larsc added it. Aug 26 12:51:17 anarsoul: hmm no such reference in ar6000 directory Aug 26 12:51:35 gena2x, it looks like jbt6k74 patch depends on other changes, it removes an mdelay i don't have Aug 26 12:52:06 it should _add_ delay Aug 26 12:52:10 it's ok Aug 26 12:52:12 where was FIFO added in .34? Aug 26 12:52:35 JaMa|GoNe: it's about this: http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-openmoko-2.6.32/0020-mfd-glamo-Enable-FIFO-stage-for-the-LCD-engine-s-mem.patch Aug 26 12:53:06 gena2x, yes, but the first hunk removes mdelay(120) to add mdelay(200) and i don't have that mdelay(120) Aug 26 12:53:12 yes and it still apply on .34 head from git.om Aug 26 12:53:13 JaMa|GoNe: it's already exist on om-gta02-2.6.34 Aug 26 12:53:45 JaMa|GoNe: http://git.openmoko.org/?p=kernel.git;a=blob;f=drivers/video/glamo-fb.c;h=3f8ec8d466abeffdeebbea97ae3ae38358097237;hb=3f6725d43021d2d7597027e36020df5b44d0667e Aug 26 12:54:00 xrmx: that's because one kms patch was adding mdelay(120) first Aug 26 12:54:33 xrmx: but that patch has not only mdelay change, but other important thing Aug 26 12:54:40 JaMa|GoNe, does introduce other changes i should double check? Aug 26 12:55:47 xrmx: you may get it at http://www.bsdmn.com/openmoko/jbtThibGblankfix+tscheck+voltageallow+1resumefix34.patch Aug 26 12:56:00 this is original patch location :) Aug 26 12:56:18 but apply changes only for drivers/video/backlight/jbt6k74.c Aug 26 12:56:19 gena2x: ah, I see, you're pointing to 2.6.32 OE patches, not 2.6.34.. Aug 26 12:56:54 JaMa|GoNe: yes, xrmx gathering .34 patches. i just point which are missing Aug 26 12:57:15 gena2x: and there is still http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/linux/linux-openmoko-2.6.34/0001-glamo-display-Enable-FIFO-stage-for-the-LCD-engine-s.patch Aug 26 12:58:12 JaMa|GoNe: as i pointed, base branch already has this change (if it is om-gta02-2.6.34) Aug 26 13:00:54 gena2x: no it does not Aug 26 13:01:04 gena2x: check carefully which file is changed Aug 26 13:01:54 wow Aug 26 13:01:56 hm. Aug 26 13:02:09 so your change is for kms? :) Aug 26 13:03:01 radekp: I really don't know how to debug it remotely :) so gta02 kernel hackers has more work to do now (I have no gta02 to test patch unfortunately) Aug 26 13:03:20 gena2x: yes Aug 26 13:03:40 JaMa|GoNe: ok, anyway xrmx need fb version now :) Aug 26 13:04:12 gena2x, the fb is already in git Aug 26 13:04:27 yeah. Aug 26 13:04:43 so, your wsod gone after ThibG fix? Aug 26 13:04:52 radekp: btw, please test that these patches did not break sound support Aug 26 13:04:54 gena2x, compiling! Aug 26 13:05:30 * gena2x going to bring up web interface for git... Aug 26 13:05:33 anarsoul: oki, i can test sound Aug 26 13:16:41 gena2x: I hope you'll get some time to find out why timeout happens with DMA-enabled s3cmci when you get your wifi working on .34 :) Aug 26 13:18:01 anarsoul: heh, i already have negative amounts of time for almost anything :( Aug 26 13:19:19 i don't know how this is possible from physics point of view, but i have pure fact in from of me :( Aug 26 13:20:08 :( Aug 26 13:21:06 yeah, i want to have as a minimum 3 copy of myself. Aug 26 13:21:35 anarsoul: here is how mmc_send_io_op_cond() is called: http://pastebin.com/err6RPmQ Aug 26 13:22:27 and 2 copies of radeks. to have 1 copy which will finish up v26 ;) Aug 26 13:28:11 radekp: I can try to optimize s3cmci to use pio on short requests (performing 4-bytes DMA requests is too expencive, it's easier to perform PIO in that case) Aug 26 13:28:15 maybe that will help Aug 26 13:28:16 :) Aug 26 13:28:26 will post patch tonight Aug 26 13:28:36 anarsoul: oki, i can test again Aug 26 13:28:50 radekp: please test that dma patch didn't break sound Aug 26 13:29:10 (it works for me, but who knows how it's on fr) Aug 26 13:32:10 anarsoul: sound is ok Aug 26 13:32:19 anarsoul: both gsm and audio Aug 26 13:34:00 cool Aug 26 13:34:33 anarsoul, radekp: hey guy, i missed, whats result? working? cpu decreased? speed increased? Aug 26 13:35:02 gena2x: wifi does not init with dma patch Aug 26 13:35:03 gena2x: card detection failed with -ETIMEDOUT Aug 26 13:35:22 aha, ok... Aug 26 13:35:32 i mean nok :) Aug 26 13:35:41 anarsoul: btw is this DMA thing also used for reading from SD card? Aug 26 13:35:51 radekp: no Aug 26 13:35:57 sd card is glamo Aug 26 13:36:04 via glamo Aug 26 13:36:16 ahh ok Aug 26 13:36:18 but ThibG thought once to implement dma in glamo :) Aug 26 13:36:45 huh? Aug 26 13:36:47 really? Aug 26 13:36:49 gena2x: i was reading about it - i think there was some problem with it Aug 26 13:36:56 ThibG: need logs? ;) Aug 26 13:37:01 hm, maybe Aug 26 13:37:28 radekp: yep Aug 26 13:37:31 for gta01 Aug 26 13:37:44 radekp: i tested dma speed (read same letter about 2-4-2 timings) Aug 26 13:37:46 gta02 uses glamo-mci to access sd card Aug 26 13:38:01 at the end of letter, you'll find dma performance tests. Aug 26 13:38:23 or may be just conclusions out of tests. Aug 26 13:38:58 radekp: but main conclusion is that dma seem not slower than pio. Aug 26 13:39:51 ThibG: but do you think idea is too bad to think of? Aug 26 13:40:14 I just don't know how to achieve that Aug 26 13:40:30 ThibG: to archieve is easy. Aug 26 13:40:49 i have sample dma implementation Aug 26 13:41:14 with some flaws like bad resourse allocation, but working. Aug 26 13:41:20 please implement DMA for mci :p Aug 26 13:41:27 hehe :) Aug 26 13:41:52 only needed thing is replace that one small memcpy :) Aug 26 13:42:33 gena2x: dma is not slower than pio for large blocks, but for small blocks (i.e. 4-32 bytes) dma op is too expensive Aug 26 13:42:57 anarsoul: but this is not case of sd transfers i think :) Aug 26 13:43:13 they are in blocks of minimum 512, yeah? Aug 26 13:43:17 it's easier to read 4 bytes from fifo that setup dma engine (write at least source, dest, size, and start reg) Aug 26 13:43:24 gena2x: nope, there's small requests too Aug 26 13:43:40 last time i measured it were big series of blocks. Aug 26 13:43:43 ThibG, what do you think of opening a git tree based on Linus' one for upstreaming glamo stuff? Aug 26 13:44:05 gena2x: card detection/configuration are small requests :) Aug 26 13:44:09 ThibG already upstreaming glamo without tree :) Aug 26 13:44:12 xrmx, what's the point? Aug 26 13:44:23 ThibG, so other people can help? Aug 26 13:44:30 anarsoul: ah, detection... Aug 26 13:44:42 xrmx, I use om-gta02-2.6.34 for that Aug 26 13:44:51 working/cleaning there Aug 26 13:44:56 and upstreaming Aug 26 13:45:30 ThibG, isn't it out of date both wrt upstream status and your patches? Aug 26 13:45:32 /last change on om-gta02-2.6.34 were 2010-07-12 Aug 26 13:45:50 not *that* out of date wrt upstream Aug 26 13:46:02 as for my patches... well, if someone applied them, it wouldn't be the case :p Aug 26 13:46:07 gena2x: btw, did you noticed my 100hz patch? :) Aug 26 13:46:16 ThibG, upstream has some changes to s3c* stuff which are not in 2.6.34 Aug 26 13:46:28 radekp: please try this patch: http://downloads.tuxfamily.org/linuxrx1950/patches/random/s3cmci-pio-for-small-reqs.patch Aug 26 13:46:36 anarsoul: oki Aug 26 13:46:37 anarsoul: yeah. it should decrease negative value of my time even more :( Aug 26 13:46:41 xrmx, why not, then Aug 26 13:47:05 driver will use dma only for requests larger than 64 bytes Aug 26 13:47:22 anarsoul: as i want to do full testing of 100Hz system Aug 26 13:47:47 gena2x: as I said, it doesn't impact on power consumption on my PDA Aug 26 13:47:59 anarsoul: i want to lmbench it Aug 26 13:48:04 anarsoul, should we update config? Aug 26 13:48:14 xrmx: for what? Aug 26 13:48:33 anarsoul, do we need S3C_...PIO=y no? Aug 26 13:48:58 PIO should be disabled if you want DMA transfers Aug 26 13:48:59 :) Aug 26 13:49:50 anarsoul, PIODMA off too? Aug 26 13:49:55 xrmx: yep Aug 26 13:50:06 it's choice, just run menuconfig and you'll see Aug 26 13:50:11 you can choose only one Aug 26 13:50:50 PIO/DMA or PIODMA, last one allows to choose pio/dma via pdata Aug 26 13:51:16 (and last one is useless) Aug 26 13:54:29 radekp: patch is only compile-tested (has no usb cable for pda with me), so maybe it doesn't work at all :)) Aug 26 13:56:39 anarsoul: nevermind, will try in a minute or two Aug 26 14:05:42 gena2x: tested your WS patch with 2.6.34 and xserver-xorg-video-glamo Aug 26 14:05:53 gena2x: xrandr -o 1 now works - great Aug 26 14:06:05 i had to post it month ago... Aug 26 14:06:24 qvga switch stuff should also work flawless Aug 26 14:06:42 i mean _more_ flawless Aug 26 14:07:36 gena2x: will it work with 2.6.29 kernel too? Aug 26 14:08:25 radekp: i think it should. in fact, it fixes issue identified as switching from gcc-4.1 to gcc-4.4, but may be this is not all. Aug 26 14:08:40 never tried it on .29. will be interesting to test. Aug 26 14:08:46 * radekp will try Aug 26 14:08:59 heh. poor v26 :) Aug 26 14:09:28 btw, is there any gta01-owner? :) Aug 26 14:09:48 i have one at home :) Aug 26 14:10:10 i am not owner but i have it borrowed from Pavel Machek Aug 26 14:10:21 anarsoul: now wifi is working Aug 26 14:10:27 with this patch? Aug 26 14:10:44 radekp: please test cpu load Aug 26 14:11:03 iirc, it was ~30% with pio-only, am I right? Aug 26 14:11:19 yup Aug 26 14:12:00 radekp: btw, please show dmesg :) Aug 26 14:13:01 anarsoul: http://pastebin.com/VAsf1qmV Aug 26 14:13:34 anarsoul: i hope applied all those patches ok Aug 26 14:14:17 hmm, it seems that you disabled dma debug stuff... so we can't see whether dma is used Aug 26 14:14:44 as it uses irq Aug 26 14:14:51 you may see /proc/interrups Aug 26 14:17:00 radekp: soo, what's cpu usage now? :) Aug 26 14:18:40 btw, there's still errors: mmc0:0001: error -110 reading SDIO_CCCR_INTx Aug 26 14:21:02 anarsoul: i will reenable debug Aug 26 14:21:27 and also Aug 26 14:21:34 in s3cmci.c Aug 26 14:21:44 increase S3CMCI_MIN_DMA_LEN to 64 Aug 26 14:22:21 anarsoul: compiled 100HZ patch, booting in 2 mins :) Aug 26 14:22:33 gena2x: is it faster or slower? Aug 26 14:22:54 oh Aug 26 14:23:02 sorry, misunderstood you Aug 26 14:23:03 :) Aug 26 14:24:21 hm. Aug 26 14:24:29 look like it is working Aug 26 14:25:00 this is goood Aug 26 14:25:57 radekp, do you have ts working? Aug 26 14:27:16 anarsoul: yeah, cool Aug 26 14:27:28 anarsoul: i times less timer interrupts. Aug 26 14:27:28 gena2x: ? Aug 26 14:27:37 s/i /2 / Aug 26 14:27:38 gena2x meant: anarsoul: 2 times less timer interrupts. Aug 26 14:27:51 does it have impact on power consumption? Aug 26 14:27:59 anarsoul: my dmesg now http://pastebin.com/ffMv184G Aug 26 14:28:18 xrmx: yes, in qtmoko it works Aug 26 14:28:43 radekp: what about cpu usage during wifi transfer? Aug 26 14:28:54 does wifi work at all? Aug 26 14:29:31 xrmx: here ts works too. with X11 Aug 26 14:29:40 gena2x, here don't Aug 26 14:29:56 xrmx: patch your tslib Aug 26 14:30:12 or kernel Aug 26 14:30:21 gena2x, it works with 2.6.29 Aug 26 14:30:53 .34 and .29 have different drivers it is (1) Aug 26 14:31:08 and .32/.34 also base on evdev it is (2) Aug 26 14:31:34 becuase of different driver you'll get jitter. (i have utgly fix for it) Aug 26 14:31:55 because of evdev, you loose ability to use tslib. Aug 26 14:32:20 gena2x, that may be Aug 26 14:32:23 anarsoul: i dont see any evident difference Aug 26 14:32:39 this were short story about ts Aug 26 14:32:41 :) Aug 26 14:32:46 radekp: because there's no DMA transfers :) Aug 26 14:32:51 xrmx: tslib works with qtmoko - i dont know about x - probably not Aug 26 14:33:02 anarsoul: ahh why not? Aug 26 14:33:10 radekp: you have patch in kernel to add pressure :) Aug 26 14:33:11 radekp: S3CMCI_MIN_DMA_LEN is too high Aug 26 14:33:16 try 32 or 16 Aug 26 14:33:23 in s3cmci.c Aug 26 14:33:30 radekp: if tslib works somewhere, it works everywhere. Aug 26 14:34:09 gena2x: i have tried much with Xorg - maybe i misconfigured it Aug 26 14:34:30 i require almost nothing to configure Aug 26 14:34:40 radekp: or even 8, that requires some testing to find out which value is better Aug 26 14:34:45 Section "InputDevice" Aug 26 14:34:46 Identifier "Touchscreen" Aug 26 14:34:46 Driver "tslib" Aug 26 14:34:46 Option "Device" "/dev/input/event1" Aug 26 14:34:46 Option "EmulateRightButton" "1" Aug 26 14:36:14 radekp: btw, how do you measure cpu usage? Aug 26 14:37:25 anarsoul: just with top Aug 26 14:37:42 gena2x: will try that config Aug 26 14:38:11 gena2x, i've added this here and it is working http://lists.alioth.debian.org/pipermail/pkg-fso-maint/2010-August/003413.html Aug 26 14:38:40 xrmx: sure. but you'll have no right button Aug 26 14:38:42 :) Aug 26 14:39:05 and also i do not know how is with filtering in evdev ;) Aug 26 14:39:16 gena2x, just discovered that :) Aug 26 14:39:24 radekp: 100 - id? Aug 26 14:39:30 or 100 - id - wa? :) Aug 26 14:39:47 xrmx: we can name this 'manual calibration' :) Aug 26 14:40:47 anarsoul: i just did top and checked ar6000_io process Aug 26 14:41:02 anarsoul: it eats still around 30% of CPU Aug 26 14:41:27 hm, please check overall cpu usage, not ar6000 Aug 26 14:41:41 and it would be nice if you can check transfer rate too Aug 26 14:42:33 (id + wa) is cpu idle time in that case Aug 26 14:42:53 xrmx: want to test my dejitter-patch ? :) Aug 26 14:43:17 gena2x, i'm back to 2.6.29 Aug 26 14:43:30 hehe :) Aug 26 14:44:26 anarsoul: hmm maybe i am measuring it wrong, is there any better tool then top? Aug 26 14:45:05 hmm i will try time Aug 26 14:45:14 radekp: top is OK, but measure overall cpu load (or cpu idle), it's third line in top Aug 26 14:45:28 radekp, /bin/time Aug 26 14:47:09 anarsoul: overall cpu with S3CMCI_MIN_DMA_LEN 16 is like 5% user 15%system Aug 26 14:47:15 anarsoul: i will check without dma now Aug 26 14:47:30 radekp: just disable DMA in config Aug 26 14:47:35 no need to revert patches :) Aug 26 14:49:56 time will bot measure time spent in irq Aug 26 14:50:48 or overall system slowdown because of dma trasfer Aug 26 14:51:26 you may notice overall system slowdown by cpu usage of some constant small part cpu consumer Aug 26 14:51:45 perfert thing is to notice how much cpu eat top itself. Aug 26 14:56:08 anarsoul: hmm still i can see any difference, the cpu is about the same, but i am getting afwuly hungry now ;-) Aug 26 14:56:16 :) Aug 26 14:56:35 anarsoul: gotta go home for some food because i forgot money at home Aug 26 14:57:16 radekp: ok, bye :) Aug 26 14:57:36 anarsoul: see you Aug 26 16:07:01 onen|openBmap: openBmap program looks nice :) Aug 26 16:30:36 gena2x: so, what's your conclusion about 100hz patch? does it helps to reduce power consumption? :) Aug 26 16:30:52 anarsoul: i did some tests Aug 26 16:31:07 my main target is performance Aug 26 16:31:22 so i didn't measure power consumption - as it is very wrong in my kernel. Aug 26 16:31:37 tests i did Aug 26 16:33:25 I see Aug 26 16:34:05 very not of very high quality this time, but it's clearly seen that under load benefit is not really significant and close to test error, but in average it is 1-2% better than 200hz version Aug 26 16:35:16 hi onen|openBmap Aug 26 16:35:38 GNUtoo|laptop: hi, wassup? good progresses with the g1? ;-) Aug 26 16:35:40 but i think this is not true multiprocess tests Aug 26 16:35:52 onen|openBmap, we have official images Aug 26 16:35:57 but some blockers remains Aug 26 16:36:02 like buffer underrun for alsa Aug 26 16:36:04 and wifi Aug 26 16:36:07 * onen|openBmap stays without voice Aug 26 16:36:09 but I've high hopes Aug 26 16:36:13 wifi means PSM Aug 26 16:36:26 anarsoul: but i vote for 2% :) Aug 26 16:36:31 GNUtoo|laptop: PSM? Aug 26 16:36:42 next will be GPS integration; then keyboard; then alsa Aug 26 16:36:49 GNUtoo|laptop: so you want to add support for gpsd? Aug 26 16:36:50 PSM means power saving mode Aug 26 16:36:58 onen|openBmap, in openBmap Aug 26 16:37:05 basically things are like this: Aug 26 16:37:08 anarsoul: as it is overall. also so far by feelings, i can't feel difference in usage Aug 26 16:37:13 * onen|openBmap is listening to GNUtoo|laptop Aug 26 16:37:24 fso used gypsy Aug 26 16:37:34 gena2x: and no difference in power consumption on rx1950 :( Aug 26 16:37:38 and fso-gpsd translated it in gpsd compatible thing Aug 26 16:37:40 so I think patch is pretty useless Aug 26 16:37:47 anarsoul: i think 2% is much Aug 26 16:38:03 gypsy is dbus Aug 26 16:38:07 and gpsd tcp Aug 26 16:38:09 so... Aug 26 16:38:11 now Aug 26 16:38:19 gena2x: it's close to measurement error Aug 26 16:38:20 :) Aug 26 16:38:31 ogpsd is the gypsy implementation, it's part of frameworkd Aug 26 16:38:33 anarsoul: of course, i run several times Aug 26 16:38:41 the new one in cornucopia is fsotltd Aug 26 16:38:45 it's not complete Aug 26 16:38:47 no GPGGA Aug 26 16:38:50 only time for now Aug 26 16:38:51 anarsoul: and can easily see it is not really error Aug 26 16:39:02 ogpsd and fsotldt provide the same thing Aug 26 16:39:16 GNUtoo|laptop: GPGGA? Aug 26 16:39:17 the dbus gypsy API won't be implemented in cornucopia Aug 26 16:39:22 yes gps fix Aug 26 16:39:28 only time is implemented Aug 26 16:39:45 more will come, expect at least 1 month for implementing it Aug 26 16:39:51 mickey|zzZZzz, is very busy Aug 26 16:39:54 so.... Aug 26 16:39:55 radekp: hm, it seems that my last patch disabled dma usage at all :( Aug 26 16:40:07 now in fsotdldt the gypsy API won't be implemented Aug 26 16:40:10 gpsd will be Aug 26 16:40:17 and a new API will replace GYPSY Aug 26 16:40:30 and the dream has no ogpsd support Aug 26 16:40:34 so.... Aug 26 16:40:39 I want openBmap on my dream Aug 26 16:41:01 because I want to map gsm towers and use the location provider to get my posision then Aug 26 16:41:08 gpsd works on htcdream Aug 26 16:41:14 gpsd means the official one Aug 26 16:41:18 GNUtoo|laptop: do I get it right. the future is, low layer is gpsd compatible, and on top of it, a new dbus api, which differs from what my client uses (gypsy). is tha tcorrect? Aug 26 16:41:36 onen|openBmap, nearly correct Aug 26 16:41:54 the only thing that could be uncorrect is who is top of it Aug 26 16:42:03 I don't remember what mickeyl said about that Aug 26 16:42:07 I'll look at logs Aug 26 16:43:11 GNUtoo|laptop: ok. it does not make a big difference I guess. The new dbus api will most probably have signals, and gpsd doesn't (IIRC). so I guess the new api will be in the long term more interesting. Aug 26 16:43:23 GNUtoo|laptop: nevertheless, for now you have gpsd. Aug 26 16:43:30 yes Aug 26 16:43:47 so, I really want openBmap to work on my htcdream Aug 26 16:43:47 GNUtoo|laptop: so: I did some work on making the core app ready for plugins, but not on the gps side. Aug 26 16:43:48 now Aug 26 16:43:55 ok Aug 26 16:43:58 so I wondered Aug 26 16:44:03 if I do a gpsd version Aug 26 16:44:15 1) can I send patches somewhere Aug 26 16:44:27 2) there will be a problem with the server where the data are sent Aug 26 16:44:32 the 2 worries me Aug 26 16:44:45 because the website says something about contacting you before writing customs clients Aug 26 16:44:52 GNUtoo|laptop: I am very unsatisfied with the gps part, as it is not modular. I have been thinking about making it so. then a gpsd implementation would be loaded simply by a matter of specifying in config file (to keep things simple, at least in a first place) Aug 26 16:45:19 onen|openBmap, that would be so nice Aug 26 16:45:20 btw I tried openBmapGTK Aug 26 16:45:23 it looks great Aug 26 16:45:29 but I didn't try the most interesting part Aug 26 16:45:33 the location provider Aug 26 16:45:36 GNUtoo|laptop: thanks! Aug 26 16:45:40 I'd like to have that too Aug 26 16:45:49 onen|openBmap, thanks me when it'll be done Aug 26 16:45:56 GNUtoo|laptop: the location provider has been done by baruch, and he stopped before it gets usable Aug 26 16:46:06 ah so it's not usable yet? Aug 26 16:46:09 GNUtoo|laptop: the python implementation is from me but does not work Aug 26 16:46:22 ah, is there some system that works Aug 26 16:46:32 for taking infos from the database? Aug 26 16:46:34 GNUtoo|laptop: but I plan to work again on it Aug 26 16:46:38 nice!!! Aug 26 16:46:56 onen|openBmap, do you want to ear about GPS on anti-vendor ports? Aug 26 16:47:02 GNUtoo|laptop: don't hold your breath, time has been scarce last monthes :-( Aug 26 16:47:13 ok Aug 26 16:47:45 GNUtoo|laptop: " thanks me when it'll be done" ? Aug 26 16:48:12 GNUtoo|laptop: on anti-vendor ports? Aug 26 16:48:34 you thanked me for the future work or for congratulating you on how openBmap looks? Aug 26 16:49:05 onen|openBmap, some people like me try to port SHR to newer phones, like for example: Aug 26 16:49:08 *htcdream Aug 26 16:49:11 *palm pre Aug 26 16:49:12 *n900 Aug 26 16:49:19 GNUtoo|laptop: a *simple* implementation won't be difficult for a location service. my python implementation was almost ready. but a lot has to be done by hand (db update, for one) Aug 26 16:49:26 palm is against it, nokia too etc.... Aug 26 16:49:33 so that's why it's called anti-vendor Aug 26 16:49:38 they make proprietary parts Aug 26 16:49:41 to discourage that Aug 26 16:49:46 GNUtoo|laptop: for thanking me. but now I thank you for the great work on the dream ;-) Aug 26 16:49:51 ah ok Aug 26 16:49:55 thanks Aug 26 16:50:06 about GPS on newer phones: Aug 26 16:50:11 htcdream has GPS Aug 26 16:50:29 palm pre: not yet but should be easy as it speaks the same protocol than the modem according to morphis Aug 26 16:50:34 n900: no one tried Aug 26 16:50:38 GNUtoo|laptop: back to your initial question: Aug 26 16:51:02 so about htcdream: I get fixes even inside buildings but, it's less good than om-gta02 and doesn't have apgs Aug 26 16:51:20 GNUtoo|laptop: 1. you can send patches to me. Or if you want you can get git writing access to our sourceforge git tree? Aug 26 16:51:31 less good means that I went into the street with one on each hand, and it was more continuous on om-gta02 than on dream Aug 26 16:51:48 onen|openBmap, I'll send patches first Aug 26 16:52:14 GNUtoo|laptop: 2. no problem. we pay a lot of attention to have high quality data. and it starts with tracability in the sense of: knowing which client of which version has sent the data Aug 26 16:52:18 onen|openBmap, I send patches to your mail? Aug 26 16:52:21 or to the openmoko mailing list? Aug 26 16:52:35 for 2 I will modify the client Aug 26 16:52:36 GNUtoo|laptop: it allows fixing/deleting bad data afterwards, if a bug is discovered Aug 26 16:52:37 so... Aug 26 16:52:43 ah ok Aug 26 16:52:44 nice Aug 26 16:52:51 GNUtoo|laptop: to my email is fine Aug 26 16:52:59 your email is? Aug 26 16:53:06 GNUtoo|laptop: on ML if you prefer Aug 26 16:53:12 ok Aug 26 16:53:18 it would get more review Aug 26 16:53:26 and tell people that there is some activity Aug 26 16:53:28 GNUtoo|laptop: go for it. Aug 26 16:53:33 GNUtoo|laptop: I'm out for a while. keep trying with the Userfriendly system. hope you can get it working Aug 26 16:53:51 but I've a big issue Aug 26 16:53:54 dcordes, ok Aug 26 16:53:54 GNUtoo|laptop: for 2. If you send patches to me, then you won't have a custom client. it will be in mine Aug 26 16:54:03 ok Aug 26 16:54:06 my issue is this one Aug 26 16:54:18 I go to different places Aug 26 16:54:21 I can map them Aug 26 16:54:30 hmmm Aug 26 16:54:47 basically my goal is to be able to use a locator instead of the GPS Aug 26 16:54:51 because the GPS: Aug 26 16:54:55 *get slow fixes Aug 26 16:55:00 *consume a lot of power Aug 26 16:55:16 GNUtoo|laptop: this part asking to contact us is so that we don't get data from code which uses the id and version number of another code base. that's all Aug 26 16:55:18 mapping without having the possibility to use the data is not great Aug 26 16:55:23 ok Aug 26 16:55:46 that's my issue(mapping without the possibility to use the data) Aug 26 16:56:03 so... Aug 26 16:56:07 GNUtoo|laptop: as said earlier, locator is something we want, but so far, time has been lacking Aug 26 16:56:08 I wonder what to do Aug 26 16:56:13 indeed Aug 26 16:56:21 is it hard to do? Aug 26 16:56:48 GNUtoo|laptop: and I cannot give you some expected delivery date. I did in the past for other things, and I feel bad because they are still not there yet Aug 26 16:57:08 ok Aug 26 16:57:20 maybe I could help a bit Aug 26 16:57:26 but I would do something simple Aug 26 16:57:27 GNUtoo|laptop: not at all. sqlite file with anonymised data gets updated everytime data gets incorporated. this sqlite file is on our webiste Aug 26 16:57:29 and not complete Aug 26 16:57:37 like something someone can piggyback on Aug 26 16:58:00 basically I would care about something offline Aug 26 16:58:03 I've 3g for now Aug 26 16:58:08 GNUtoo|laptop: you can have a look at my python locator code. basically, it looks for the cellid, if not the lac, if not present the country Aug 26 16:58:09 but I've not 3g all the time Aug 26 16:58:17 ok Aug 26 16:58:26 it's in git? Aug 26 16:58:34 GNUtoo|laptop: yes Aug 26 16:58:47 GNUtoo|laptop: it is more of a proof of concept Aug 26 16:58:56 ok Aug 26 16:59:11 GNUtoo|laptop: no api has been thought as the final target is geoclue which has an api Aug 26 16:59:24 it's the vala one? Aug 26 16:59:25 GNUtoo|laptop: but my purpose was to have sth simple, which does the trick Aug 26 16:59:53 I know vala a bit Aug 26 16:59:59 GNUtoo|laptop: for "mass market", I would want to have some way to update the db on the phone. My idea from the beginning is to use offline db, no web api Aug 26 17:00:00 s/a bit/a little bit Aug 26 17:00:05 GNUtoo|laptop: python Aug 26 17:00:11 ok Aug 26 17:00:15 wha'ts that then: Aug 26 17:00:23 openBmap-locator-vala Aug 26 17:00:48 GNUtoo|laptop: that's the work from baruch, I talked about earlier Aug 26 17:00:56 ah ok Aug 26 17:01:22 GNUtoo|laptop: IIRC it does not use the sqlite provided by the website, but builts a db from raw data Aug 26 17:01:34 GNUtoo|laptop: I chose sqlite because it makes things very simple. Aug 26 17:01:36 ah ok Aug 26 17:01:39 indeed Aug 26 17:01:42 and it works offline Aug 26 17:01:53 do you use sqlalchemy? Aug 26 17:02:44 GNUtoo|laptop: well my python code uses a local sqlite file. it may need some rework to make it compatible to the sqlite from our website. but that was the idea behind it Aug 26 17:02:59 GNUtoo|laptop: sqlalchemy. no. doesn't know Aug 26 17:03:11 ok Aug 26 17:03:36 maybe I'll try the sqlite from website when I'll have some time Aug 26 17:03:44 dream already takes a lot of time Aug 26 17:04:18 GNUtoo|laptop: I have chosen simpel things also because if possible, I would like to be able to port to other platforms (meego now, but maemo in the past). and packages, tools, may or may not be available Aug 26 17:04:28 GNUtoo|laptop: even android has support for some python, and sqlite Aug 26 17:04:36 ok Aug 26 17:04:55 GNUtoo|laptop: the sqlite is very simple Aug 26 17:05:00 ok Aug 26 17:05:02 GNUtoo|laptop: ping me if you need any more input Aug 26 17:05:07 GNUtoo|laptop: any time Aug 26 17:05:09 ok Aug 26 17:05:11 thanks a lot Aug 26 17:05:20 GNUtoo|laptop: he, I did nothing so far ;-) Aug 26 17:05:22 is your git big? Aug 26 17:05:30 how many GB? Aug 26 17:05:43 thanks for the ping permission then Aug 26 17:05:54 GNUtoo|laptop: I want to try to release a new version of the client. I am not sure wha twill be the next step, but it might be the location service. we'll see Aug 26 17:06:04 ok Aug 26 17:06:10 basically there are 2 projects Aug 26 17:06:16 openBmap and cell-locator Aug 26 17:06:21 GNUtoo|laptop: our git. no I don't think so. it should be very small Aug 26 17:06:31 I choosethe one whith a location provider Aug 26 17:06:41 because it might be like a game Aug 26 17:06:48 but if no one can use the db, it's not fun Aug 26 17:06:54 GNUtoo|laptop: cell locator? url? Aug 26 17:07:09 sorry Aug 26 17:07:12 cellhunter Aug 26 17:07:13 http://ch.omoco.de/cellhunter/ Aug 26 17:07:17 bad memory Aug 26 17:07:32 GNUtoo|laptop: that's waht nick (openbmap founder) keeps repeating to me: as long as we don t use the data, it is no fun for users Aug 26 17:07:44 indeed Aug 26 17:07:52 and even better: Aug 26 17:07:52 GNUtoo|laptop: we have history with this project Aug 26 17:07:59 imagine people with n900 Aug 26 17:08:04 that have no GPS Aug 26 17:08:08 GNUtoo|laptop: you should ask to mickey|zzZZzz what he thinks Aug 26 17:08:08 they take a freerunner Aug 26 17:08:12 the map things Aug 26 17:08:17 GNUtoo|laptop: because I am of course completely biased ;-) Aug 26 17:08:18 and they got a GPS with openbmap Aug 26 17:08:23 without a working GPS Aug 26 17:08:29 ok Aug 26 17:08:39 he told me to ogpsd it Aug 26 17:08:52 when I wanted to implement the gypsy for htcdream but didn't succeed Aug 26 17:09:03 xrmx: radekp: here's updated s3cmci-dma patchset: http://downloads.tuxfamily.org/linuxrx1950/patches/random/s3cmci-dma/ Aug 26 17:09:04 because roughly I've got a gps activator Aug 26 17:09:08 that registers to fso Aug 26 17:09:17 so if ogpsd does the same Aug 26 17:09:19 .... Aug 26 17:09:26 radekp: last patch (that enables mixed pio/dma xfers) was buggy, I fixed it Aug 26 17:09:29 :) Aug 26 17:09:32 GNUtoo|laptop: an now imagine this: as meego is supposed to use geoclue, people have gsm based location service on the n900 and other phones. not bad neither ;-) Aug 26 17:09:45 indeed Aug 26 17:09:55 * GNUtoo|laptop wikipedias geoclue Aug 26 17:10:32 GNUtoo|laptop: simply put: location framework, with time spent on api, and you can provide a location plugin Aug 26 17:10:34 wow nice Aug 26 17:10:48 we should definitely talk to mickey|zzZZzz Aug 26 17:11:01 GNUtoo|laptop: this had been discussed back in essen in the fso shr meeting! Aug 26 17:11:10 Aug 25 15:28:18 one is a new location API Aug 26 17:11:11 Aug 25 15:28:18 taking into account what we learned when working with gypsy Aug 26 17:11:11 Aug 25 15:28:18 and geoclue et. al. Aug 26 17:11:22 ah I wasn't there Aug 26 17:11:30 GNUtoo|laptop: for me this would be the way to go. and so that is the reason of very simple approach of my python code Aug 26 17:11:39 GNUtoo|laptop: just give sth quickliy, which works Aug 26 17:11:42 indeed that would be great Aug 26 17:11:49 and wich is cross-platform Aug 26 17:12:26 onen|openBmap, btw just a quesiton on openBmap: Aug 26 17:12:27 GNUtoo|laptop: what i like, is that it would work on meego, but also on laptop with 3g keys Aug 26 17:13:06 while I use it, it's passive or active, does it scan the gsm providers etc...in other words, can I still use my phone and make calls Aug 26 17:13:29 GNUtoo|laptop: short yes. (of course ;-) ) Aug 26 17:13:34 nice Aug 26 17:13:51 so it maps only one GSM provider? Aug 26 17:13:55 GNUtoo|laptop: long: when you are in a call, gsm data is garbage, the client won't log while in a call Aug 26 17:14:05 ok Aug 26 17:15:00 GNUtoo|laptop: only one yes. because unicity of an antenna (a cell to be precise) is country code (MCC), provider (MNC), LAC (some kind of group) and cell id Aug 26 17:15:19 anarsoul, have patch 1 and patch2 changed? Aug 26 17:15:32 GNUtoo|laptop: and the modem only gives data for one mcc + mnc Aug 26 17:15:51 GNUtoo|laptop: but for multiple cellls in different lacs (called neighbour cells) Aug 26 17:16:16 GNUtoo|laptop: at a time, you are only connected to a single cell Aug 26 17:17:26 GNUtoo|laptop: by the way. don't use irc log if you want to get in touch with me. I don't use it :-) Aug 26 17:17:42 ok Aug 26 17:17:54 ? Aug 26 17:18:02 irc log? Aug 26 17:20:05 GNUtoo|laptop: I know that some people have irc on all the time, and when back home, they look at what has been told since they left. I don't do it. On irc expect me to read only things you get an acknoldgment from me ;-) Aug 26 17:20:17 GNUtoo|laptop: about the gpsd way of working Aug 26 17:20:26 ok Aug 26 17:20:35 I only connect when I'm near the computer too Aug 26 17:20:37 but I keep logs Aug 26 17:20:53 because people with a lot of knowledge say something Aug 26 17:21:01 and sometime you don't get it the first time Aug 26 17:21:06 then you re-read the logs Aug 26 17:21:09 and understand Aug 26 17:21:16 GNUtoo|laptop: I have very little knowledge, correct me if wrong: gpsd has no way of a event like a signal you subscribe to, for fix, etc/ Aug 26 17:21:17 ? Aug 26 17:21:37 I think you can suscribe a client to it Aug 26 17:21:49 I'm very new to gpsd Aug 26 17:21:58 GNUtoo|laptop: ok Aug 26 17:22:07 lindi-, sent me a gpsd working program Aug 26 17:22:20 else I would have been confused between old api,new api etc... Aug 26 17:22:30 I had a NMEA printer working tough Aug 26 17:22:36 but I don't know with which API Aug 26 17:23:30 GNUtoo|laptop: ok. Aug 26 17:24:24 I sent a patch to support gpsd in openbmap ages ago :) Aug 26 17:24:39 (I think!) Aug 26 17:28:26 lindi-: hi. It may be very much possible. I have some patches, I wanted to add, when next release is out. but next release is still not out. I know it is very bad for contributors. I can only apologise :-(h Aug 26 17:28:41 lindi-: for gpsd does not ring me a bell though... Aug 26 17:29:26 anarsoul: i have moved from work to home, which patches should i apply? Aug 26 17:29:51 * onen|openBmap plans to spend little but on a regular basis time on openBmap project... Aug 26 17:30:01 anarsoul: 0001, 0002, 0003 or some more? Aug 26 17:32:37 radekp: 0001-0003 Aug 26 17:32:49 actually Aug 26 17:32:56 0001-0002 should be already applied Aug 26 17:33:03 and 0003 replaces last one Aug 26 17:38:00 lindi-, nice Aug 26 17:38:20 I'll go bye Aug 26 17:40:19 me too bye Aug 26 17:45:03 anarsoul, patch 3 applies fine with: Hunk #1 succeeded at 333 with fuzz 1 (offset -22 lines). Aug 26 17:45:57 anarsoul: hmm i dont have wifi device again Aug 26 17:46:51 xrmx: that's just description in Kconfig, after removing the Kconfig part from the patch it applies ok Aug 26 17:50:44 radekp: increase S3CMCI_MIN_DMA_LEN to 64 Aug 26 17:51:37 anarsoul: oki Aug 26 17:58:40 anarsoul: hmm still no wifi device Aug 26 17:58:59 =( Aug 26 17:59:31 you can increase it to 128, but I doubt that it'll help :( Aug 26 17:59:40 can you post dmesg? Aug 26 18:02:35 sure, should i enable dma debug? Aug 26 18:02:38 yep Aug 26 18:08:19 anarsoul: http://pastebin.com/REH88eCP Aug 26 18:10:25 radekp: hm, no errors this time Aug 26 18:12:48 anarsoul: this version was with DMA_LEN 128 Aug 26 18:13:24 and still does not work? Aug 26 18:16:09 anarsoul: no, still cant see wifi interface Aug 26 18:16:24 :( Aug 26 18:16:36 then I doubt I can fix it without device Aug 26 18:24:35 radekp: can you trace ar6000 init? Aug 26 18:24:53 just try to enable ar6000 debug info Aug 26 18:25:08 anarsoul: oki Aug 26 18:25:09 xrmx and gena2x should know how to do this Aug 26 18:26:12 and -DDEBUG to ar6000/Makefile EXTRA_CFLAGS Aug 26 18:26:32 this will bring you some debug info... Aug 26 18:28:02 gena2x: oki, thanks Aug 26 18:28:31 but i am unsure how much it will help you :) Aug 26 18:28:49 gena2x: it will show where init fails Aug 26 18:29:03 it has many different kinds of debug info. you'll have to read code Aug 26 18:29:24 I have no ar6000 unfortunately Aug 26 18:29:25 anarsoul: hehe. you'll need to know cryptic ar6000 code to understand it. Aug 26 18:29:45 and it will take a long time for me to clone another kernel tree :) Aug 26 18:29:48 anarsoul: i can route you shell via internet if you want :) Aug 26 18:30:31 gena2x: no, thanks, I can lookup sources via gitweb if I'll need Aug 26 18:30:51 i mean shell to ar6000 device :) Aug 26 18:32:00 hi lindi- , can I try your gpsd patch? Aug 26 18:32:05 anarsoul: cloning a tree is fast if you have another handy for a --reference Aug 26 18:32:08 :) Aug 26 18:32:25 gena2x: I think that's pretty useless without physical access to device :) Aug 26 18:32:41 anarsoul: why? you can testwhatever you want :) Aug 26 18:33:02 gena2x: I can't reflash kernel :) Aug 26 18:33:11 gena2x: why not? Aug 26 18:33:14 anarsoul: ^ Aug 26 18:33:36 gena2x: it has sd, you just copy kernel and modules and that's all Aug 26 18:33:41 damn anarsoul ^ Aug 26 18:33:43 :) Aug 26 18:34:11 He's just lazy, can't you see that? ;) Aug 26 18:34:15 :) Aug 26 18:34:42 PaulF: partially true Aug 26 18:34:44 i'm just joking. but in fact, such solution possible Aug 26 18:34:52 onen|openBmap: "Subject: openbmap-logger: gpsd support To: onen.om@free.fr Date: Thu, 29 Oct 2009 23:07:38 +0200" Aug 26 18:34:56 anarsoul: btw, it would be interesting to know what are the RL usecases you found for your PDAs. Aug 26 18:35:15 lindi-: ok. mea culpa then :( Aug 26 18:35:31 lindi-: not lost, but in the todo list. Aug 26 18:35:36 lindi-: sorry about that Aug 26 18:36:15 PaulF: with linux? :) none, just started to work on rootfs Aug 26 18:36:31 PaulF: with winmo I'm using PDA as mp3-player and book reader Aug 26 18:36:45 GNUtoo|htcdream: http://paste.debian.net/86016/ Aug 26 18:36:45 PaulF: and, of course, fun from kernel hacking:) Aug 26 18:37:43 PaulF: but why you're asking? Aug 26 18:37:44 :) Aug 26 18:38:11 PaulF: paul is searching for meaning of live Aug 26 18:38:18 :) Aug 26 18:38:26 anarsoul: i can't find a decent usecase neither for my FR nor for hx4700 i can borrow from a friend. Aug 26 18:38:40 heh, i told you. Aug 26 18:38:43 anarsoul: but in fact gena2x really gave a more correct and accurate answer. Aug 26 18:38:50 _fact_ Aug 26 18:38:59 * lindi- has many useful usecases for his FRs Aug 26 18:39:02 PaulF: you don't read books? :) Aug 26 18:39:40 lindi-: hm , in fact would be interesting to know. Aug 26 18:40:17 anarsoul: better to get real book. i have large screen and much better resolution, need no batteries and feel sweet Aug 26 18:40:22 anarsoul: i like reading books but i'm always in search of some good non-depressing ones. So far Ray Bradbury is my favourite in this genre. Aug 26 18:40:32 s/i have/it has/ Aug 26 18:40:33 gena2x meant: anarsoul: better to get real book. it has large screen and much better resolution, need no batteries and feel sweet Aug 26 18:41:13 And yes, FR's screen is a bit too limited. But i've read "Обитаемый остров" on it recently, it was ok (the book is great of course). Aug 26 18:42:14 PaulF: so, you have at least one use case :) Aug 26 18:42:28 anarsoul: also my beloved emacs UI for the telephony functionality. Aug 26 18:42:58 But fso is not stable enough and that's somewhat spoiling the fun :/ Aug 26 18:43:44 gena2x: well.. making and receiving calls, irc, ssh, emergency web browsing, listening to podcasts, editing openstreetmap Aug 26 18:43:44 anarsoul: hmm damn ar6000 writes so many messages that i cant see start of the log Aug 26 18:43:59 radekp: just put whole dmesg somewhere Aug 26 18:44:00 :) Aug 26 18:44:08 radekp: :) Aug 26 18:44:51 (better as file) Aug 26 18:46:06 lindi-: I even forgot that we had a discussion about your patch. I am very sorry. Aug 26 18:46:54 lindi-: this might be of interest to gnutoo. I have to go through your patch again to see how long this would take me to add it to the codebase Aug 26 18:50:11 onen|openBmap: well it wasn't terribly useful for me Aug 26 18:50:18 onen|openBmap: I mean, I could live without openbmap Aug 26 18:51:13 PaulF: main problem for me is that there's no linux-based environment for devices with qvga screen :( Aug 26 18:51:21 anarsoul: http://activationrecord.net/radekp/pub/dmesg.txt Aug 26 18:51:51 anarsoul: do you mean you can't find a WM of choice? Aug 26 18:51:59 anarsoul: try qtmoko - it works in qvga Aug 26 18:52:48 [ 3.220000] Failed to read register table Aug 26 18:52:49 [ 3.225000] Failed to connect to WMI CONTROL service status:-1 Aug 26 18:53:12 radekp: tried, it doesn't work well on devices with 32mb of ram :)) Aug 26 18:53:52 anarsoul: hmm that's not much, maybe with uclibc - e.g. on openwrt rootfs it could work Aug 26 18:54:46 radekp: qt libraries are too huge, afaik they're about 12-15mb Aug 26 18:56:48 radekp: btw, xorg+matchbox works just fine, even without swap Aug 26 18:57:08 but there's no apps to fit into qvga :) Aug 26 18:57:27 no browser, im messanger, etc Aug 26 19:00:15 32mb is disaster. even 128 is few, but 32... Aug 26 19:01:05 gena2x: yep Aug 26 19:02:16 http://iki.fi/lindi/linux9110-beta3.tar.gz is/was my "distro" for nokia communicator that has only 2MB of RAM Aug 26 19:02:33 wowwowwow Aug 26 19:02:39 and x86 sillyness made most of it unusable Aug 26 19:03:21 it spoke TCP/IP over serial port and used NBD to mount /usr and swap over network Aug 26 19:04:03 * radekp had qtopia2 running on mine Siemens SX1 Aug 26 19:04:35 i also had compiled qtmoko with uclibc, but i dont remember how big the libriaries were Aug 26 19:04:36 afaik sx1 has 16mb of ram Aug 26 19:04:37 :) Aug 26 19:05:20 ZaPPaS`: hey. Do you by any chance have something nice to say to me? ;) Aug 26 19:05:38 ZaPPaS`: have you seen Michael Albinus's letter on the ML? Aug 26 19:05:49 anarsoul: 16 MB is plenty :) Aug 26 19:07:19 btw can i measure FR's current with ammeter easily? Aug 26 19:07:42 i soldered two wires to my old nokia battery Aug 26 19:07:48 radekp: hmmm, I think there should be software way :) Aug 26 19:08:12 then attach one to mulimeter Aug 26 19:08:26 and just attach it to battepy pads Aug 26 19:08:44 * radekp will try Aug 26 19:08:48 i mean battery-holder pads and twist a bit Aug 26 19:09:08 i hope it wont explode or something, i have zero EE skills Aug 26 19:09:13 i used wires from old 40pin IDE cable Aug 26 19:09:19 gena2x: radekp: there's no way to measure current via software?? Aug 26 19:09:30 on fr Aug 26 19:09:32 check carefully + and - Aug 26 19:09:40 anarsoul: om battery consumption prints it in uA Aug 26 19:09:52 anarsoul: there is, but you need battery driver initialized i think Aug 26 19:10:02 anarsoul: it has, but it lags (10s), so multimeter is much more reliable Aug 26 19:10:20 radekp: basically soldering nokia batteries is safe because you'd be soldering to the pads on the PCB, not to the elements themselves. Aug 26 19:10:48 lindi, did you find your gpsd patch? Aug 26 19:11:00 GNUtoo|htcdream: I think I pasted you a link Aug 26 19:11:06 hi btw Aug 26 19:11:08 GNUtoo|htcdream: 21:35 Aug 26 19:11:09 ah? Aug 26 19:11:14 PaulF: what about GTA02 battery? Aug 26 19:11:14 ah ok Aug 26 19:11:19 GNUtoo|htcdream: http://paste.debian.net/86016/ Aug 26 19:11:24 thanks Aug 26 19:11:36 radekp: same. But i think lindi-'s method works decently as well. Aug 26 19:11:47 anarsoul: the question is about how to reliably measure the current in suspend. Aug 26 19:11:51 my phone ran out of battery Aug 26 19:12:10 that s why I didn t saw it Aug 26 19:12:15 thanks a lot Aug 26 19:12:22 PaulF: yep you need to be careful and not trust it blindly Aug 26 19:15:32 radekp: i attached a battery without soldering to it btw, i used some duct tape to secure the wires on pads and it worked ~ok. Aug 26 19:25:21 PaulF: i were unable to measure suspend current with anything except multimeter Aug 26 19:25:55 radekp: if you want to try kill power consumption bug, mulimeter is much better Aug 26 19:26:15 radekp: is it instantly shows you all needed values Aug 26 19:26:17 gena2x: do you mean bq27k measurements taken right after resume didn't match what you measured? Aug 26 19:27:19 PaulF: i were unable to notice suspend (reduced) current. may be i did someting wrong, may be i had huge current Aug 26 19:28:14 PaulF: but direct battery measure is much easier, and you can see it instantly, no need to parse some files etc Aug 26 19:28:36 radekp: ah, also i have to use 'unfused' input of multimeter Aug 26 19:28:39 gena2x: sure Aug 26 19:47:33 what version of qi should i use with shr-testing? Aug 26 19:58:09 djp_: supposedly any that can boot the kernel for you Aug 26 19:58:12 djp_: i'd use the latest of course Aug 26 19:59:29 PaulF: ok, will do. i've had an issue with shr-testing (4 May) whereby i can't use the terminal or sms messaging. basically the keyboard doesn't work with either Aug 26 19:59:51 djp_: most probably not related to the bootloader. Aug 26 20:00:00 djp_: did usb networking work? Aug 26 20:00:11 PaulF: yes Aug 26 20:00:27 djp_: how did the keyboard not work then? Aug 26 20:00:55 PaulF: i mean the fr virtual keyboard... foget the name... Aug 26 20:01:01 forget even Aug 26 20:01:02 Illume Aug 26 20:01:14 djp_: but did it show up or something? Aug 26 20:01:37 PaulF: yeah, showed up and appeared to accept my input, but wouldn't appear in terminal Aug 26 20:02:09 djp_: and when you try the latest shr-u, then even usb networking doesn't work for you? Aug 26 20:02:20 hmm my ammeter is only up to 200mA then it cut's of the line Aug 26 20:02:49 radekp: surely it has no additional input for bigger currents? Aug 26 20:02:59 PaulF: yup 10A Aug 26 20:03:11 PaulF: i believe so yes. usb networking is fine. Aug 26 20:03:59 djp_: so what doesn't work then? Aug 26 20:04:17 PaulF: the virtual keyboard Aug 26 20:04:44 PaulF: that you use your stylus to enter with Aug 26 20:05:02 djp_: so both with shr-u and shr-t? Aug 26 20:05:03 PaulF: so i can't compose an sms when out and about Aug 26 20:05:26 why did i get an error if nothing failed? "NOTE: Tasks Summary: Attempted 8836 tasks of which 6278 didn't need to be rerun and 0 failed. make: *** [image] Error 1" Aug 26 20:05:57 PaulF: shr-t. shr-u is fine, however i can't open settings under shr-u Aug 26 20:06:23 PaulF: was hoping that testing was a little more stable, thus my attempt to install it... Aug 26 20:06:28 djp_: probably you do not need settings? ;) Does everything else works? Aug 26 20:07:03 PaulF: i wanted to import my contacts. i usually went via settings/other... Aug 26 20:07:15 real 115m38.606s Aug 26 20:07:25 djp_: how exactly do settings not work? Aug 26 20:07:26 glad I didn't sit and wait for that build :) Aug 26 20:07:58 djp_: (i imported my sim contacts with the help of some mdbus2 command to fetch those and with some vim to apply opimd-cli in a trivial way) Aug 26 20:08:07 PaulF: when i tap on settings with my stylus, it shows the wording settings at the bottom of the screen but nothing opens up Aug 26 20:08:41 djp_: i guess you should let dos1 or JaMa|GoNe or mrmoku|away know about it. Aug 26 20:08:52 djp_: they can test and see if it's reproducible, and fix. Aug 26 20:09:03 PaulF: ok. will do. Aug 26 20:09:10 PaulF: thanks. Aug 26 20:09:15 djp_: what if you start settings from the console, do you see anything interesting? Aug 26 20:09:26 djp_: i guess that's shr-settings Aug 26 20:10:07 PaulF: i will try that. currently attempting to install qtmoko Aug 26 20:10:50 PaulF: installed qtmoko earlier and it gave me a modprobe fatal error! Aug 26 20:11:11 * djp_ is not having much luck... Aug 26 20:11:28 djp_: it's a little bit funny how you have tried different variants and they all failed Aug 26 20:11:46 djp_: probably with qtmoko your problem was not using the corresponding kernel Aug 26 20:11:53 PaulF: i' Aug 26 20:11:57 question, if I were to be a build maintainer (eg for shr-t or the fabled shr-s), would I need to do all the builds locally then upload them, or would just fiddle with the git branches? Aug 26 20:12:09 PaulF: i'm reflashing... see what happens... Aug 26 20:12:46 incidentally, any sign of spaetz (the shr-t maintainer) recently? Aug 26 20:13:34 tim_abell: merge/cherry-pick stuff from shr-u branch to shr-t branch, build&test images on shr buildhost and if ok, sync the results to public feeds Aug 26 20:13:45 PaulF: same thing... modprobe error... Aug 26 20:13:50 excellent, ta JaMa|GoNe Aug 26 20:14:31 I figure there's no point volunteering till I have at least managed to do a few builds locally and got them flashed onto my fr and working Aug 26 20:15:10 I'm thinking a howto be a distro maintainer wiki page would be good... Aug 26 20:15:15 PaulF: using images from here... https://sourceforge.net/projects/qtmoko/files/ Aug 26 20:15:24 djp_: surely theer's no conflict between a version of rootfs and kernel you're using? Aug 26 20:16:03 tim_abell: yup and keeping carefull eye on commits to http://git.openembedded.org/cgit.cgi/openembedded/log/ is good start Aug 26 20:16:19 tim_abell: to know what new is there and what is dangerous to merge etc Aug 26 20:16:43 I already keep up with the mailing lists, which is a start Aug 26 20:17:04 tim_abell: then subscribe also oe-commits :) Aug 26 20:17:08 PaulF: using rootfs= qtmoko-debian-v24.jffs2 and kernel= uImage-v24.bin Aug 26 20:17:37 JaMa|GoNe, is that the same as the git log? if so I'll just read the history when I'm looking to pull stuff in Aug 26 20:17:49 yes it's the same Aug 26 20:17:54 djp_: does that modprobe error mean you can't even ssh in? Aug 26 20:18:01 ah ha http://www.shr-project.org/trac/wiki/shr-testing2009 Aug 26 20:18:06 djp_: and can't get no dmesg and anything? Aug 26 20:19:16 PaulF: yeah... it hangs at boot Aug 26 20:19:45 djp_: strange, v24 worked for everybody... Aug 26 20:19:45 PaulF: just repeats the message... though on initial boot i can use the recalibration screen and set the time... Aug 26 20:19:56 tim_abell: maybe you'll be also interested in this http://wiki.openembedded.net/index.php/Testing Aug 26 20:20:11 PaulF: do you think i need to clear my nand? Aug 26 20:20:31 PaulF: that's where i'm attempting to install qtmoko Aug 26 20:20:44 djp_: i'm not sure what to suggest providing you're flashing the kernel and the rootfs properly. Aug 26 20:20:46 gnite all Aug 26 20:20:57 JaMa|GoNe: have nice dreams Aug 26 20:21:03 :) Aug 26 20:21:06 PaulF: it worked previously... it's weird Aug 26 20:21:28 htc dreams... Aug 26 20:21:42 ta JaMa|GoNe I'll read up on that Aug 26 20:24:55 JaMa|GoNe: how do i rotate screen correctly, also with the input? Aug 26 20:25:28 oooh, grr, now the input is rotated but not the screen :)) Aug 26 20:25:41 vanous: xrandr Aug 26 20:25:54 what else??? Aug 26 20:26:07 JaMa|GoNe: input too? Aug 26 20:26:44 thanks a lot lindi- !!! I'm rebasing the patch Aug 26 20:29:03 * radekp had the same problem with 2.6.34 - input was rotated but screen not, quite different to use it this way ;-) Aug 26 20:30:19 radekp: hehe :) what an exercise Aug 26 20:30:26 started making some notes at http://www.shr-project.org/trac/wiki/How%20to%20join?version=12#Contribute Aug 26 20:30:42 sorry, http://www.shr-project.org/trac/wiki/How%20to%20join#Contribute Aug 26 20:33:57 self._session.query('sadmoyq') Aug 26 20:33:58 AttributeError: gps instance has no attribute 'query' Aug 26 20:34:00 hmmm Aug 26 20:34:14 the API has changed? Aug 26 20:34:22 I already saw python code doing that Aug 26 20:34:28 and it didn't work Aug 26 20:35:29 JaMa|GoNe: sorry to bother you :) , xrandr -o 1 should do it, or? Aug 26 20:38:49 PaulF: i have lowered the value in pcf50633_bl_set_brightness_limit and i can boot under 200mA now Aug 26 20:45:34 JaMa|GoNe (and anyone else interested), how's that for a first draft of a new page on maintenance? http://www.shr-project.org/trac/wiki/ShrMaintainerHowTo Aug 26 20:46:02 I figure while I want to be a maintainer, I also want to make it as easy as possible for anyone else to do it too Aug 26 20:46:16 i've been reading a lot about how debian do it and i like their approach Aug 26 20:57:09 hmm, loading tslib in xorg.conf doesn't help... xrandr gets display output/input 90deg off Aug 26 21:00:37 radekp: please, test dma with ar6000_delay patch reverted :) maybe it causes timeouts Aug 26 21:01:42 anarsoul: oki, but it will take some time if i make it today... Aug 26 21:04:55 radekp: that's ok, I'm not hurry :) Aug 26 21:12:05 hmm, low battery here, but no help... with xinput i can correct the output/input for normal (portrait) mode, but not get it to work in landscape... Aug 26 21:20:47 * pespin booted successfully wing-linux (angstrom+android) into htc-artemis (omap-850) :) Aug 26 21:21:08 is there some shr for this platform? it would be great! :) Aug 26 21:29:38 hm. is talpadk on IRC? Aug 26 21:29:56 g'night all Aug 26 21:30:16 I don't quite find out how this matchbox panel docking thing in his ffportrait matchbox theme would work Aug 26 22:02:42 ah, found out how to dock mb-panel there Aug 26 22:03:41 but why the fuck does the mb-applet-menu-launcher's menu have about 90 px big menu entries? Aug 26 22:06:41 and where is decided into which category to put programs? Aug 26 22:07:14 neon is in "Accessories"... neither /usr/share/menu/neon nor /usr/share/applications/neon.desktop do even contain that word Aug 26 22:21:05 hi dcordes Aug 26 22:21:16 I'll look if I've loop support Aug 26 22:53:55 <[Rui]> hi all Aug 26 22:54:11 <[Rui]> can anyone bump omnewrotate? Aug 27 01:17:15 i'm trying http://build.shr-project.org/shr-unstable/images/om-gta02/shr-full-eglibc-ipk--20100823-om-gta02.rootfs.tar.gz on uSD Aug 27 01:17:24 but it keeps asking for the PIN Aug 27 01:18:55 i did an opkg update and upgrade but the problem is still there Aug 27 01:19:10 reboot. the same Aug 27 01:19:28 any clues? Aug 27 01:28:10 your pin is wrong ? Aug 27 01:32:14 i typed the pin several times Aug 27 01:32:45 then rebooted to an old shr-testing and it works fine **** ENDING LOGGING AT Fri Aug 27 02:59:57 2010