**** BEGIN LOGGING AT Sat Sep 03 02:59:57 2005 Sep 03 03:34:19 03koen 07org.oe.dev * r8296491e... 10/packages/linux/handhelds-pxa_2.4.19-rmk6-pxa1-hh40.bb: handhelds-pxa 2.4.19-hh40: lower default preference Sep 03 04:12:31 morning Sep 03 04:13:07 hi all Sep 03 04:13:46 (OffTopic) - i am trying to map a device memory into my process data space - has anyone done that before ? Sep 03 04:49:06 RP: ping Sep 03 04:51:17 \adfk Sep 03 04:52:19 mickeyl: ping Sep 03 04:57:03 lardman: pong Sep 03 04:57:50 RP: I am seeing charging/power issues Sep 03 04:58:07 RP: With your kernel, not my cpufreq patched one that is Sep 03 04:58:24 RP: I was just wondering whether they had all been fixed, etc. Sep 03 04:59:56 What kind of issues? Sep 03 05:01:48 not very methodical, but this is what I saw: http://people.bath.ac.uk/enpsgp/Zaurus/cpufreq/power_issues.txt Sep 03 05:02:01 I just noted it down on a bit of paper I had handy Sep 03 05:03:26 The crux of it is that the LED has to be on for it to be charging (no matter what sys info reckons it's doing), and it looks like it underestimates the amount it has charged; perhaps... Sep 03 05:04:54 Yes, the LED needs to be on for it to be charging Sep 03 05:05:08 The sysfs file should reflect that. If it doesn't, its a bug Sep 03 05:05:19 Don't confuse AC in with charging though Sep 03 05:05:29 ACIN != charging Sep 03 05:05:31 Yes, I'm not sure how GPE decides to tell the sys info applet, etc. that it is charging Sep 03 05:05:59 Yes, I understand that, though in that case, perhaps it's the logic which decides when to start charging (i.e. what battery level) Sep 03 05:06:04 I suspect GPE associates ACIN with Charging which is wrong Sep 03 05:06:12 I'll look into that Sep 03 05:06:41 I suppose another possibility is that my battery is knackered, but I can't see why it would be as it has been working fine Sep 03 05:07:20 lardman: The Charge_Full interrupt the devices use is a bit strange and I'm not convinced it works right either Sep 03 05:07:35 How recent is your kernel? Sep 03 05:08:00 I built it straight after you added the fix to the machine recognition the other day Sep 03 05:08:13 Not much has changed since then... Sep 03 05:08:15 so R19 iirc Sep 03 05:09:41 I suspect our battery voltage reading and percentage reporting is just badly wrong Sep 03 05:10:03 You battery will be fully charged. The test would be to see how long it lasts, ignoring the strange percentages Sep 03 05:10:38 Ah, okay, do you have a battery percentage recording script somewhere, so I can grab the percentages too (for interest's sake)? Sep 03 05:11:02 hrw had one... Sep 03 05:11:19 I was writing one to run in octave so I could make the cpu do something to test cpufreq Sep 03 05:11:54 lardman: if you turn on the msglevel parameter in /sys/devices/xxx/corgi-pm you will get a log to syslog Sep 03 05:12:02 echo 2 to it Sep 03 05:12:18 cool Sep 03 05:13:24 Now to see how/whether GPE performs any power management, and if so, how to switch it off Sep 03 05:16:07 Someone really needs to spend time figuring out how to reliably read the battery voltage and how to convert it into an accurate percentage. Sep 03 05:16:15 The problem is its a time sink Sep 03 05:16:52 I'd be interested to look into it, but will be busy until thre end of the year finishing up my PhD. I'll try to take a look though (as a break ;) ) Sep 03 05:18:02 My Z reckons it has 8% battery now; and wifi is up and running, I'll see how long it lasts Sep 03 05:19:29 the msglevel thing only writes when the level changes? Or every x minutes? Sep 03 05:20:39 Damn power management, it's turning the backlight off because it thinks it has low power, I'll have to try to ssh in so I can see if it's still alive Sep 03 05:25:54 lardman: I'd love to know why the battery readings are so wrong... Sep 03 05:27:21 RP: me too :) Sep 03 05:27:48 Is there any trick to getting usb network connection working with this kernel? Sep 03 05:28:04 modprobing the right modules I think Sep 03 05:28:10 Ah, yes Sep 03 05:31:24 Okay, I admit defeat, I can't find the module, any clues? Sep 03 05:32:34 03ph5 07org.oe.dev * r9fd3eb06... 10/packages/dbus/dbus-0.34/dbus-1.init: Sep 03 05:32:35 dbus-0.34: adapt init script to daemon name Sep 03 05:32:35 - /usr/bin/dbus-daemon-1 was changed to /usr/bin/dbus-daemon Sep 03 05:32:39 03ph5 07org.oe.dev * ra41eb54a... 10/packages/avahi/avahi_0.2.bb: avahi-0.2: start avahi-daemon after dbus-daemon Sep 03 05:33:34 lardman: g_ether ? Sep 03 05:34:48 RP: Hmm, I don't have one of those Sep 03 05:37:38 Which is not good as that means it wasn't included in gpe-image Sep 03 05:37:44 lardman: Looks like it doesn't make it into the images :-( Sep 03 05:37:53 It is called g_ether Sep 03 05:38:07 but isn't on my device either Sep 03 05:38:09 yep, I'm just trying to wget it from my feed Sep 03 05:39:12 one more random thing I'm having happen - my Z just locks up, the whole thing, including including the hw power key, then it will wake up after 30 sec or so (and immediately suspend if I'd pressed the hw on/off key) Sep 03 05:39:56 When does it just lock up? Under which circumstances? Sep 03 05:40:52 The fact it suspends suggests interrupts are getting processed but the scheduler is locked Sep 03 05:40:53 none in particular - I was using vi last night and it just locked up, then today it's done it twice since I've been talking to you - while I tas typing 'ifconfig usb0' and typing 'wget ...' Sep 03 05:41:18 I don't see it here... Sep 03 05:41:30 wouldn't the interrupt cause it to suspend straight away though? not wait until it becomes functional again? Sep 03 05:41:45 No. It gets fed through the input system Sep 03 05:41:55 and userspace thinking about it Sep 03 05:42:08 right Sep 03 05:42:40 Which wifi card? Sep 03 05:42:46 and does it happen without wifi? Sep 03 05:43:00 03ph5 07org.oe.dev * r06168994... 10/packages/avahi/avahi_0.2.bb: avahi-0.2: avahi system user creation fix Sep 03 05:43:26 wcf12, and it may just be with the card, I'll take it out in a minute and see if it happens again Sep 03 05:44:00 In fact it also happens without the card as I was using vi on a file on /media/cf last night when it happened Sep 03 05:44:51 No idea then I'm afraid Sep 03 05:45:13 that's fair enough Sep 03 05:45:32 Looks like it's suspended - not sure if that was a pm or a battery dead thing, but I can't get it to come back on Sep 03 05:45:40 I feel like I'm slowly totally destroying the 2.6 kernel :-/ Sep 03 05:45:49 :D Sep 03 05:46:25 * RP gets some food Sep 03 05:47:09 * lardman puts Z on charge in the hope that it will then be able to wake up again Sep 03 06:07:01 * RP returns Sep 03 06:07:55 Hi there Sep 03 06:08:42 I'm playing with oe for the first time, trying to build a rootfs for a geodegx Sep 03 06:09:02 what's that? Sep 03 06:09:32 RP: where are the power related files? In arch/arm/mach-pxa? Sep 03 06:09:53 "bitbake nano" dies on compiling glibc-2.3.5+cvs20050627-r0, function `pthread_initialize', undefined reference to `_res' plus some errno stuff... Sep 03 06:10:48 snickl: No idea I'm afraid, you'll have to see if any of the old hands know anything Sep 03 06:10:52 geodegx is a plain old half-i486/half-i586 PC Sep 03 06:11:26 with lots of integrated stuff Sep 03 06:11:54 lardman: yes, sharpsl_pm is the core, corgi_pm is the corgi specifics Sep 03 06:11:54 koen|train: train? you have wifi trains over there>? Sep 03 06:12:02 no gprs Sep 03 06:12:22 RP: cool. Is thare anyway of looking at the diffs, or has the move to monotone removed the history? Sep 03 06:12:39 * koen|train is enjoying his 1GB/month subscritption Sep 03 06:12:52 lardman: They won't tell you anything. All the history is in my patch archive though Sep 03 06:13:06 http://www.rpsys.net/openzaurus/patches/ Sep 03 06:13:20 lardman: fwiw openembedded.bkbits.net is still up Sep 03 06:13:22 RP: You based the numbers etc. on the original stuff from the 2.4.18 Sharp kernel? Sep 03 06:13:37 Before it became sharpsl_pm.patch it was known as corgi_power Sep 03 06:13:53 lardman: correct, although I have changed a lot of things now... Sep 03 06:14:08 including the power translation table which was a recent change Sep 03 06:14:15 RP: :) that's why I was after the diffs, so I could understand more quickly Sep 03 06:14:43 lardman: They just say patch -rx replaced with patch -ry Sep 03 06:14:55 RP: Okay Sep 03 06:19:43 when running make menuconfig, what do I have to pass in addition of tell it the arch? -arch=pxa ? Sep 03 06:20:20 export ARCH=arm before running it Sep 03 06:20:26 Thanks Sep 03 06:20:47 There are probably better ways to do it but I know that works... Sep 03 06:20:58 google found my probem, but no solution unfortunately...: http://www.handhelds.org/hypermail/oe/24/2490.html Sep 03 06:22:54 http://sourceware.redhat.com/ml/libc-alpha/2005-04/msg00096.html Sep 03 06:23:37 A fair few hits searching for something like "undefined reference to `_h_errno' "; not sure if any of it will help mind you... Sep 03 06:28:14 03ph5 07org.oe.dev * rec654721... 10/ (2 files in 2 dirs): Sep 03 06:28:14 ipaq-pxa270: updates to machine configuration and kernel defconfig Sep 03 06:28:14 - This closes bug #235. Sep 03 06:40:42 Does anyone here use Linux on a laptop? If so, do you use userspace cpufreq control, and if so, which of cpuspeed, cpufreqd or powernowd is recommended? Sep 03 06:40:44 lardman: You hit bull's eye with that url! The redhat bugzilla has a lot to say on that issue, but it seems even glibc specialists bailed out on that :( The easy solution seems to be to disable TLS for non-NPTL build for machines < i686... Sep 03 06:41:09 snickl: just luck, glad to have helped a bit Sep 03 06:41:18 :) Sep 03 06:41:43 please do add a bug to the bugtracker though, with a link to the url etc., just in case someone else comes across it Sep 03 06:42:01 ack Sep 03 06:42:15 actually, I don't know where bugs for your device would go, but it would be good to make a note somewhere anyway Sep 03 06:42:50 geodegx is a AMD thing, right? Sep 03 06:47:32 03mickeyl 07org.oe.dev * r796b9972... 10/packages/qt/ (3 files): qt/x11: add qt-x11-free-4.0.1 and make it work Sep 03 06:48:29 koen|train: I remember "national (semi) geode", but it's possible that AMD bought it since then. Sep 03 07:15:53 03mickeyl 07org.oe.dev * r945d2682... 10/packages/qt/qt-x11-free-4.0.1/ (3 files): qt4/x11: add patches necessary for clean cross compilation Sep 03 07:31:31 03hrw 07org.oe.dev * rcbbc5114... 10/packages/sqlite/ (4 files in 2 dirs): added sqlite 3.2.5 Sep 03 07:36:59 Time for a WinXP reboot, bbiam Sep 03 08:11:29 03ph5 07org.oe.dev * r0e8a4fab... 10/packages/x11/ (files/errordb-keysymdb-path-fix.patch x11_cvs.bb): Sep 03 08:11:29 x11-cvs: make libx11 look for XErrorDB and XKeysymDB in the right place Sep 03 08:11:29 - This closes bug #292. Sep 03 08:19:52 Can anyone see anything in particular which is wrong with this patch? (all but the new file get rejected for some reason): http://people.bath.ac.uk/enpsgp/Zaurus/cpufreq/cpufreq.patch Sep 03 08:20:43 I think it must be something to do with the format of the single file multi-patch, or something like that - I copied the individual patches which I created into the one file based on another one I found which worked Sep 03 08:42:30 hi Sep 03 08:42:47 ~seen hrw|gone Sep 03 08:42:52 hrw|gone was last seen on IRC in channel #oe, 1d 7h 43m 15s ago, saying: 'hi'. Sep 03 08:43:29 ~thx Sep 03 08:43:30 np Sep 03 08:43:35 ~np Sep 03 08:43:36 [np] \"no problem\", or \"null pointer\", or \"no pants\", or non-deterministically polynomial, or ¡ningún problema!, or \"Now Playing\", an abbreviation used by those lame mIRC/WinAmp plugins Sep 03 08:47:17 hrw: do you have a battery power tracking script handy? Sep 03 08:48:35 lardman: what script? Sep 03 08:48:53 lardman: I used "logread -f>/media/cf/battery.log" Sep 03 08:50:31 hrw: oh, I thought you had whipped up a special one Sep 03 08:50:45 no Sep 03 08:51:04 RP made all work with kernel output parsing Sep 03 08:51:23 okay Sep 03 08:52:30 this was with kernel which gave all battery infos in dmesg Sep 03 08:52:47 now I would rather log changes in /sys/*/batter_percentage Sep 03 08:53:45 Onto a different topic... Is it possible to make patch more verbose so I can see why my patch is failing? Sep 03 08:54:00 that is the patch running inside of bitbake Sep 03 08:55:55 you use patcher or quilt? Sep 03 08:56:03 quilt say where does patch fail Sep 03 08:56:51 patcher does not Sep 03 08:57:03 probably patcher then, it just says that hunks have failed and which file Sep 03 08:57:43 so its quilt Sep 03 08:57:55 oh right Sep 03 08:58:04 it cant tell more as it use patch which give exactly that info Sep 03 08:58:23 I'm an idiot Sep 03 08:58:47 it was bloody notepad adding LF and I forgot to run dos2unix Sep 03 08:59:01 (serves me right for trying to use Windows I suppose) Sep 03 08:59:05 ;) Sep 03 08:59:15 Odd though, the error message gave no hint that that was the problem Sep 03 08:59:16 use normal editor under windows then Sep 03 08:59:34 even Wordpad tends to add them as they're the default Sep 03 08:59:34 ConText, Crimson Editor, GVim Sep 03 08:59:58 I may go back to using vi via ssh... ;) Sep 03 09:00:11 use gvim under windows :) Sep 03 09:00:49 RP: you still about? Sep 03 09:01:09 hrw: have you seen any more power/charging issues with your c860? Sep 03 09:02:09 not currently - but didnt tested 2.6.13-mm1-r1 yet Sep 03 09:02:17 RP: I wanted to ask about whether altering the CPU speed will break things - like the LCD, CF, etc. Sep 03 09:02:27 currently I have 2.6.13-rc6-mm2-r20 on my husky Sep 03 09:02:29 hrw: I'm still on 2.6.12-rc5-mm1 Sep 03 09:02:45 same here, but I've got all sorts of problems with the battery charging Sep 03 09:03:41 hrw: If you're not seeing any issues, I may have to check and see whether my battery has died all of a sudden - any ideas which contact is which on it? Sep 03 09:05:20 its: +,test,- but I forgot from which side (reverse then in nokia phones) Sep 03 09:05:47 that's okay, as long as it's the outer two my voltmeter won't care Sep 03 09:06:04 check oesf anyway Sep 03 09:06:06 Actually, it probably says it on the battery doesn't it now that I come to think about it Sep 03 09:11:06 cu Sep 03 09:38:04 lardman: I would expect to see issues with CF, LCD Sep 03 09:38:26 Well, CF. THe LCD should be ok as I think the w100 clock is independent Sep 03 09:44:44 Yeah, I just saw the comment at the top of the patch I applied (which applied and had built successfully I hasten to add) Sep 03 09:45:27 How much work would it be to sort the CF (and SD if needs be) timing out? Sep 03 09:49:08 Someone had started to add cpufreq hooks to the code if I remember rightly Sep 03 09:49:29 Have you tried the cpufreq patch? Sep 03 09:49:48 I've not tested the kernel yet Sep 03 10:03:59 time to go shopping, bye all Sep 03 10:28:45 hmm Sep 03 10:28:55 ok, i think i've got the hinge problem Sep 03 10:29:07 ~lart QPEApplication hiding qDebug Sep 03 10:29:07 * ibot beats QPEApplication hiding qDebug into protomatter with the andromeda galaxy Sep 03 10:36:49 btw.. corsair.. the makefile is so bad its amusing. one of the 40-odd main.o dependencies lines: "main.o: /opt/Qtopia/include/qcursor.h /opt/Qtopia/include/qfont.h" .. mixed in with a bunch of host-system headers.. sigh. Sep 03 10:37:08 egad Sep 03 10:37:18 (and yes, if I get it working I'll be submitting the .bb to you and a new makefile to the author) Sep 03 10:37:33 does that for every .o he builds. Sep 03 10:38:04 is that generated from a qmake/tmake .pro or something? Sep 03 10:38:12 300-line makefile, of which 270 lines or so are devoted to being brain damaged. Sep 03 10:38:36 dunno where he genned it, but its right out of sf cvs.. Sep 03 10:39:02 i mean, is there a .pro in the tree next to the makefile? :P Sep 03 10:39:23 and the install forces you to put the sdk in /opt/Qtopia etc.. Sep 03 10:39:24 nope Sep 03 10:40:32 is there a trick to building qt2 apps in OE btw? can't seem to get it finding arm-linux/qt2/include :( Sep 03 10:40:50 theres probably a .bbclass you should inherit Sep 03 10:40:54 * kergoth hasnt messed with oe in months Sep 03 10:47:55 inherit qmake Sep 03 10:49:03 mickeyl: I noticed the hinge code didn't seem to be working. Since I fixed the touchscreen, I can rotate but not automatically Sep 03 10:49:17 I gave up trying to get qDebug to work :-( Sep 03 10:49:38 heh Sep 03 10:49:46 it's simple Sep 03 10:50:01 just submit QT_DEBUG or remove three lines in QPEApplication Sep 03 10:51:15 I tried that and it still didn't seem to do anything. I probably miscompiled or something... Sep 03 10:51:21 anyway, I'm fixing the hinge issue as we speak... Sep 03 10:53:07 mickeyl: excellent :) Sep 03 10:53:24 * pH5 learns that ipkg remove --force-depends busybox is not a good idea. Sep 03 10:54:13 mickeyl: thanks. Sep 03 10:58:00 RP: fix commited in opie HEAD. only thing to rebuild is libopiecore2 Sep 03 11:07:03 mickey|shower: Ok, thanks :) Sep 03 11:34:55 03nail 07org.oe.dev * r9b7968f0... 10/packages/samba/samba_3.0.20.bb: Samba: upstream upgrade to 3.0.20 Sep 03 11:49:27 waves Sep 03 13:36:02 03mickeyl 07org.oe.dev * rb812c2ab... 10/packages/qte/ (qte-2.3.10/improve-calibration-r0.patch qte_2.3.10.bb): qte-2.3.10: integrate tslib RMK calibration algorithm to calibrated mouse handler. patch courtesy Richard Purdie. Sep 03 15:24:04 03nail 07org.oe.dev * rbf64f2b4... 10/packages/meta/ucslugc-packages.bb: uclugc packages: Remove gphoto2 from ucslugc. Breaks autobuild Sep 03 15:38:06 anyone know of gps and/or altimeter hw that works with any OE setup? Sep 03 21:38:08 Is there a way to specify DEFAULT_PREFERENCE (or something) only when using uclibc? Sep 03 21:38:54 it would seem samba 3.0.20 breaks on uclibc, but not on glibc Sep 03 23:08:45 hi Sep 03 23:10:07 anyone home? :) Sep 03 23:10:39 I have read some info about the wifi driver for CWB1000 Sep 03 23:10:47 anyone get more info about that? :) Sep 04 01:37:55 03nail 07org.oe.dev * r4f44dda8... 10/packages/meta/ucslugc-packages.bb: ucslugc packages: Move mpd to broken packages. Doesn't build on ucslugc Sep 04 01:37:59 03nail 07org.oe.dev * r2587f48c... 10/packages/icecast/icecast_2.2.0.bb: Fix duplicate DEPENDS Sep 04 01:38:03 03nail 07org.oe.dev * r65c90e7d... 10/packages/meta/openslug-packages.bb: Openslug-packages: Add icecast to openslug feed Sep 04 01:38:07 03nail 07org.oe.dev * r8bf8798a... 10/packages/meta/ucslugc-packages.bb: ucslugc packages: Reinstate gphoto2 and mpd to ucslugc feed. Should work now Sep 04 01:38:11 03nail 07org.oe.dev * rcaba6347... 10/packages/musicpd/mpd_0.11.5.bb: mpd: add libiconv DEPENDS Sep 04 01:38:15 03nail 07org.oe.dev * ra2b8c7ff... 10/conf/distro/ucslugc.conf: ucslugc.conf: Pin samba version to 3.0.14a, since 3.0.20 breaks in ucslugc Sep 04 01:38:19 03nail 07org.oe.dev * r472eae40... 10/conf/distro/openslug-packages.conf: openslug-packages: Add libiconv, needed for mpd Sep 04 01:38:23 03nail 07org.oe.dev * rf967ac0a... 10/conf/distro/ucslugc.conf: ucslugc.conf: Fix samba preferred version Sep 04 01:48:00 hello ! **** ENDING LOGGING AT Sun Sep 04 02:59:56 2005