**** BEGIN LOGGING AT Wed Nov 01 02:59:57 2006 Nov 01 03:56:39 hey raduga Nov 01 08:23:24 hello all Nov 01 08:34:19 morniong Nov 01 09:00:42 morning all Nov 01 09:04:50 morning all Nov 01 09:04:55 hey RP Nov 01 09:29:59 morning crew Nov 01 09:38:32 hi ppl Nov 01 09:53:51 ljp: congrats, one year of Qtopia without any GPL licensed release, patch or rsync version :) Nov 01 09:54:24 ljp: I predict you will answer it is perfect when it needs to updates and I have to say I must agree Nov 01 09:58:43 having a little trouble building -cross packages when oe is setup to use an external toolchain Nov 01 09:59:12 I thinks this may be because of the overrides for CC CXX CPP etc in local.conf Nov 01 09:59:46 looking at cross.bbclass, I see no attempt to reset these values Nov 01 09:59:59 tkp: how do you override the cross toolchain? Nov 01 10:00:12 compare it with native.bblass, and there is a whole section entitled "# set the compiler as well. It could have been set to something else" Nov 01 10:00:53 I have not specifically overridden the cross toolchain... I have added export CC etc to local conf, as per the wiki Nov 01 10:01:18 so for example, I have export CC = "${CCACHE}i586-stv-linux-gcc-3.4.4 ${HOST_CC_ARCH}" Nov 01 10:01:31 which is surely wrong for building -cross packages no? Nov 01 10:04:28 as suspected, if I edit cross.bbclass, adding in the section which resets the compiler etc ( http://rafb.net/paste/results/NdEb0x51.html ) - taken from native.bbclass, then -cross packages build ok again Nov 01 10:04:38 Morning Nov 01 10:05:15 so, am I right in thinking that to build -cross packages, it should be using the hosts toolchain? Nov 01 10:07:35 tkp: That's what I would expect. I expect the -cross packages to run on the development host, generating output to be run on the target. So they must be compiled using the (dev) host toolchain. Nov 01 10:08:16 right... so my fix would seem sane then Nov 01 10:18:56 tkp: it looks sane to m Nov 01 10:19:42 I'll post it up to the bugtracker so other people can vet it and incorporate it into the repo. Nov 01 10:19:48 tkp: but I don't know if the other classes have assumptions on the toolchain. There are quite some number of variables that influence the build of the toolchain (which libc, which float support, etc) Nov 01 10:21:13 likewise: but this isn't todo with building the actual toolchain itself, it's more about using the right toolchain for the right job and clearly it is wrong to use the cross toolchain to build cross packages Nov 01 10:22:12 tkp: yes, that looks wrong to me, but I am not a (core) OE dev, so posting it on the bugtracker looks sane to do Nov 01 10:39:23 My bitbake always hangs (full processor load) when I enable INHERIT += "rootfs_ipk". Seeing if I can pin-point this... Nov 01 10:48:31 <_law_> morning Nov 01 10:50:31 03zecke123 * r626 10/ (13 files in 12 dirs): BitBake: Tag 1.6.2 and 1.7.2 of BitBake Nov 01 10:56:42 morning Nov 01 10:57:13 03zecke123 * r627 10bitbake/MANIFEST: bitbake/trunk/bitbake: Add more files to the MANIFEST and make it work again Nov 01 10:57:53 meta-toolchain generate toolchain-cross to build for target device. *-native *-cross are built with host toolchain Nov 01 10:58:55 someone ported network driver from 2.4 to 2.6? Nov 01 11:06:47 okay BitBake 1.6.2 and 1.7.2 snapshot is released Nov 01 11:06:58 ~lart monotone Nov 01 11:06:58 * ibot --purges monotone Nov 01 11:07:23 zecke: Sounds good :) Nov 01 11:07:34 zecke: will you send release announce to mailing lists? Nov 01 11:08:16 hrw: sure, I don't know when though :} Nov 01 11:08:16 RP: I have placed 1.7.2 in the bitbake-unstable product, I hope you don#t mind Nov 01 11:09:39 zecke: Not at all - Its not what I'd call stable yet :) Nov 01 11:09:40 init/main.c:91:2: error: #error Sorry, your GCC is too recent for kernel 2.4 Nov 01 11:09:42 shit. Nov 01 11:10:04 can I compile 2.4.33 with gcc 3.3.x? Nov 01 11:14:50 I can. its building Nov 01 11:22:26 hrw: got 2.4 source for 802.11g drivers? Nov 01 11:23:20 got Nov 01 11:25:16 and built Nov 01 11:28:10 but I knew before that 2.4 driver is building and is working in Zaurus Nov 01 11:28:45 /dev/hda6 88G 84G 4,0G 96% /home Nov 01 11:28:52 fsck... 4G only Nov 01 11:31:22 any monotone geeks alive? Nov 01 11:33:00 XorA: on #monotone@oftc yes Nov 01 11:55:22 hi mickeyl Nov 01 11:59:06 mickeyl: you still manage keys on monotone.openembedded.org Nov 01 11:59:09 mickeyl: ? Nov 01 12:01:27 i try not to, but yes, i could do something Nov 01 12:01:45 mickeyl: wolfson@openembedded.org seems to have disappeared Nov 01 12:01:54 did we got private ml? Nov 01 12:05:10 hrw: hold on 5 minutes Nov 01 12:05:16 XorA: i'll check asap Nov 01 12:05:26 mickeyl: thx mickeyl Nov 01 12:06:37 mickeyl: no hurries, if your too busy Ill bug koen when he reappears Nov 01 12:07:33 hrw: you should get mail in a couple of seconds Nov 01 12:08:15 thx Nov 01 12:08:15 np Nov 01 12:11:13 mickeyl: subscribed other core devs? Nov 01 12:11:21 i think so Nov 01 12:11:31 ok Nov 01 12:12:19 mickeyl: do you know who from OE/OZ/Ang team is capable of working on porting driver from 2.4 to 2.6? Nov 01 12:12:27 XorA: sorry, i can't find the proper mtn for the server atm. its been a while... please bug koen Nov 01 12:12:36 mickeyl: cool Nov 01 12:12:45 ~cabal Nov 01 12:12:48 There is no cabal Nov 01 12:12:50 :-D Nov 01 12:13:31 hrw: RP,zecke,woglinde,ph5,psokolsovsky Nov 01 12:13:37 surely more Nov 01 12:13:49 ok Nov 01 12:14:13 we can get two cf 802.11g cards for OE/OZ but need to get driver working Nov 01 12:20:29 is there somewhere can I view an up-to-the-minute version of the repo and/or repo diffs? Nov 01 12:21:07 hrw: please only accept it as 'give or take'. we will try, but we give no guarantees Nov 01 12:21:14 after all, that's how donations should be Nov 01 12:21:19 or devices on loan Nov 01 12:21:30 mickeyl: understandable Nov 01 12:21:44 tkp: oe website Nov 01 12:22:24 We can never say for certain we're going to do X, just that we will try our best Nov 01 12:22:25 mickeyl: customer waits with releasing them until OZ will supports those cards Nov 01 12:23:50 If they want guarantees, they'll have to pay someone. Even then putting timescales on things is tricky for kernel dev work like that :-/ Nov 01 12:24:21 I know and unserstand it Nov 01 12:24:28 hrw: oe website says that the last commit was made 20 hors ago, but I saw a commit message in this IRC room just a couple of hours ago Nov 01 12:25:15 then no other way, and we need to ask koen what broke Nov 01 12:26:02 mickeyl: what do you think about moving OZ website to ltg? sf.net is so lagged that I'm unable to edit any simple page Nov 01 12:26:41 * XorA wonders if oe mtn has a script to remove keys with no checkins Nov 01 12:27:18 hrw: +1 from me Nov 01 12:29:25 ' Your project has been submitted to the LTG Projects administrators. ' Nov 01 12:36:26 gracias LTG admin team Nov 01 12:43:46 hi Nov 01 12:45:20 hi nicolasfr Nov 01 12:46:29 hi hrw! I did not have time to go to post office yesterday (and closed today of course). I have to go the "basements" today to find a box, and pack everything tight. Will send saturday morning. Nov 01 12:49:46 nicolasfr: no problem Nov 01 12:50:57 morning all Nov 01 12:51:15 hi dirk Nov 01 12:51:24 hey Marcin Nov 01 12:59:37 do13_: returned for some hacking yet? Nov 01 13:00:05 XorA: hehe Nov 01 13:00:17 do13_: I heard you had a few miles of slab to lay Nov 01 13:00:47 XorA: right. Nov 01 13:05:58 psokolovsky: do you want to hack wifi driver? Nov 01 13:06:11 Hi! Nov 01 13:06:56 hrw, of course I want ;-). But wonder, would I have resources (time first of all) for that? Anyway, what's about it? Nov 01 13:08:01 psokolovsky: one company want to release 802.11g CF card. they want this card supported by OZ and other OE distros under 2.6 kernel. The problem is that only 2.4 driver exist. We will get 2 cards to get it working Nov 01 13:08:17 XorA: I had no choice. My wife complained. So I had to reduce my hacking time :) Nov 01 13:08:19 do13_: train her to hack Nov 01 13:08:30 XorA: better not Nov 01 13:11:38 "I always prefer OpenSource if the software is equally good or almost as good as the closed source alternative" Nov 01 13:11:38 hrw: I propose the wlan-ng stack. The best and well maintained wlan driver. :) Nov 01 13:11:51 wow, a true Free Software believer :) Nov 01 13:11:57 do13_: o yes Nov 01 13:12:13 hrw: what chipset are you talking BTW? Nov 01 13:13:25 Marwell 8385 Nov 01 13:15:42 hrw: wait a couple of weeks then ask the intel/marvel guys about support for Mainstone :-) Nov 01 13:16:46 XorA: and they will answer that it is supported. and show 2.4 driver Nov 01 13:17:48 KerwoodDerbyanyway, I have a Marvell wlan CF card and I have 2.6 driver source Nov 01 13:17:52 hmmm Nov 01 13:18:16 ~seen KerwoodDerby Nov 01 13:18:17 ? Nov 01 13:18:30 hrw: http://purl.rikers.org/%23oe/20060625.html.gz Nov 01 13:18:45 kerwoodderby was last seen on IRC in channel #oe, 114d 2h 6m 47s ago, saying: 'I've deleted the build directory, downloaded a new OE.db and rebuilt from scratch, and I still get this error'. Nov 01 13:19:45 18498 25 03:10< KerwoodDerby> anyway, I have a Marvell wlan CF card and I have 2.6 driver source 18499 25 03:13< Interrobang> KerwoodDerby: out of curiousity, what chipset does your card use? orinoco or anythign like that? Nov 01 13:19:49 hmm.. Nov 01 13:19:52 18500 25 03:13< KerwoodDerby> Marvell 8385 18501 25 03:13< emte> i belive ewi is down ... Nov 01 13:20:01 hrw: exactly Nov 01 13:20:08 ~seen kerwood Nov 01 13:20:17 kerwood was last seen on IRC in channel #opie, 2d 12h 20m 25s ago, saying: '~hail psokolovsky'. Nov 01 13:28:32 XorA: newer version of chipset is used in OLPC Nov 01 13:28:51 hrw: yeah I saw the Gnu people venting on that Nov 01 13:32:09 fsck.. Nov 01 13:32:15 I am trying to build some of the apps available in OE (GPE apps) for my desktop. Is there a conf-file appropriate for an Ubuntu edgy build? Nov 01 13:32:15 /home/hrw/devel/trisoft/linux-2.6/include/asm/processor.h:80: error: `CONFIG_X86_L1_CACHE_SHIFT' undeclared here Nov 01 13:32:34 mccarthy: MACHINE="native" DISTRO="generic" Nov 01 13:32:40 hrw thanks Nov 01 13:33:08 I include few asm/*.h to get such contants and it still break... Nov 01 13:33:55 hi Nov 01 13:33:59 hi CoreDump|home leoncamel Nov 01 13:35:08 hey. hrw. I read you blog recently .:) Nov 01 13:35:10 hrw, I got something wrong while bitbake opie-image Nov 01 13:35:12 hrw, I got this : dbus-glib.h:28:30: error: dbus/dbus-shared.h: No such file or directory Nov 01 13:35:18 hrw, while building OPIE. Nov 01 13:35:20 hrw, so , the opie is broken now ? Nov 01 13:35:51 gah, someone either DEFAULT_PREFERENCE = -1 in dbus-glib 0.72 or remove the one in dbus 0.94 Nov 01 13:35:57 leoncamel: thats question to psokolovsky as he maintain opie in OE now Nov 01 13:36:35 well, I wasn't yet nominated ;-) Nov 01 13:36:44 XorA, hmm, en ? what exactly ? Nov 01 13:37:37 psokolovsky, i nominated you .:) so, please tell me what happened exactly. Nov 01 13:37:45 leoncamel: the problem is dbus-glib 0.72 requires dbus 0.94 but dbus 0.94 is DEFAULT_PREFERENCE = -1 so it wont build Nov 01 13:38:12 whoever addes 0.72 and 0.94 needs to decide the relationships Nov 01 13:39:06 well, my answer would be "well OPIE uses dbus too? 8-)". so, I'm not the perfect eprson to ask yet ;-) Nov 01 13:39:24 roll on RP with versions DEPENDS Nov 01 13:39:56 leoncamel, anyway, OPIE in OE.dev builds for ansgrom distro (caveat: need to revesr a patch committed within last week). use that as reference Nov 01 13:40:02 XorA, the dependency is illegal ? right ? Nov 01 13:40:05 my guess is 0.94 should be allowed to build by default and wait for people to scream about breakage Nov 01 13:41:21 psokolovsky, BUT, the 0.72 depend on 0.94 is illegal. do you agree ? Nov 01 13:41:48 leoncamel: I dont understand why you say this is illegal? Nov 01 13:41:56 Finished Making Marvell Wlan Linux Driver Nov 01 13:42:25 XorA, OK. ignore it. So, I want to know, how to fix it ? Nov 01 13:42:31 ofcourse it is not true - I just hacked it very badly to get look how much is needed to change Nov 01 13:43:09 leoncamel, sure. that doens't happen in angstrom and I heard it has been fixed Nov 01 13:43:09 leoncamel: set a PREFFERED_VERSION for dbus-glib to 0.71, or remove DEFAULT_PREFERENCE = -1 from dbus 0.94 Nov 01 13:43:46 or set PREFFERED_VERSION for dbus to 0.94 Nov 01 13:44:25 XorA, OK. I got it. Nov 01 13:44:29 or wonder why opie wants dbus-glib :-) Nov 01 13:44:56 XorA, god knows. ;) Nov 01 13:46:19 I know Nov 01 13:46:30 bluez-utils instead of bluez-utils-nodbus Nov 01 13:49:05 ~hail hrw Nov 01 13:49:11 * ibot bows down to hrw and chants, "I'M NOT WORTHY!!" Nov 01 13:50:48 "Anyone who quotes me in their signature is an idiot -- Rusty Russell" Nov 01 13:50:49 :D Nov 01 13:56:56 After syncing OE I run into the following problem, does anyone have a clue where this was introduced? Nov 01 13:56:57 NOTE: Resolving missing task queue dependencies Nov 01 13:56:59 ERROR: No providers of runtime build target busybox Nov 01 14:04:54 hrw: http://www.geek-fr.com/lib/exe/fetch.php/progear.jpg ;) Nov 01 14:06:01 nicolasfr, what's that ? Nov 01 14:06:19 nicolasfr: nice package ;) Nov 01 14:06:42 a package for *laptop* ? Nov 01 14:06:53 webpad Nov 01 14:08:21 hehe, only box I found which fits! Nov 01 14:12:46 tkp: submit patches as attachments in bugzilla Nov 01 14:12:51 kerwood|afk: afk or alive? Nov 01 14:12:57 ah.. k. sorry Nov 01 14:17:20 tkp: 1543 (spyro) fixed Nov 01 14:17:46 03hrw 07org.oe.dev * rddb2619d... 10/ (1 packages/python/python-spyro_0.9.14.bb): python-spyro: fix SRC_URI - close #1543 Nov 01 14:24:00 any matchbox-desktop experts in da house? Nov 01 14:24:25 i can't get it to accept my --bg specification. it always comes up w/ the greenish/blueish backgound Nov 01 14:26:28 hrw: another src_uri to fix: http://web.mit.edu/~emin/www/source_code/pycodes/pycodes-1-1.tar.gz Nov 01 14:27:49 (python-pycodes) Nov 01 14:27:56 nicolasfr: author changed university? Nov 01 14:28:09 from berkley to mit Nov 01 14:29:01 dunno, this is the official page Nov 01 14:29:22 pushed Nov 01 14:29:36 I have some others... Nov 01 14:29:51 03hrw 07org.oe.dev * r95d2b2a9... 10/ (1 packages/python/python-pycodes_1.1.bb): python-pycodec: fix SRC_URI Nov 01 14:30:49 nicolasfr: prepare fix as patch Nov 01 14:31:12 ok, how should I do? Nov 01 14:31:37 nicolasfr: fix recipes, then 'mtn dif .' and check output Nov 01 14:32:00 ok Nov 01 14:37:16 OLPC has wifi on usb.. Nov 01 14:37:43 OLPC wifi is independant from the OLPC :-) Nov 01 14:37:56 XorA: 8385 also can be Nov 01 14:38:06 iiuc Nov 01 14:38:28 hrw: hurry up and get that zaurus mesh network running then :-) Nov 01 14:38:49 koen|gprs: morning Nov 01 14:39:00 koen|gprs: did you fiddle with monotone servers last night? Nov 01 14:39:08 hi koen Nov 01 14:42:25 hrw: alive! Nov 01 14:43:14 kerwood: in June you said that you have 2.6 driver for marvel 8385 cf card Nov 01 14:43:31 25 03:10< KerwoodDerby> anyway, I have a Marvell wlan CF card and I have 2.6 driver source Nov 01 14:43:37 * Crofton|home jumps from the window Garmin down %12 Nov 01 14:43:38 25 03:10< KerwoodDerby> anyway, I have a Marvell wlan CF card and I have 2.6 driver source Nov 01 14:43:42 25 03:10< KerwoodDerby> anyway, I have a Marvell wlan CF card and I have 2.6 driver source Nov 01 14:43:45 XorA: nope Nov 01 14:43:46 fsck Nov 01 14:43:54 ~lart pasting Nov 01 14:43:54 * ibot eats pasting's liver with some fava beans and a nice chianti Nov 01 14:44:19 kerwood: I have customer which has 8385 cf cards Nov 01 14:44:24 hrw: right, now I need to find the source of that claim... Nov 01 14:44:30 :) Nov 01 14:44:50 kerwood: it would be cool Nov 01 14:45:46 hrw: amazingly, this card was mentioned in the OZ comments -- iirc someone mentioned he had free Marvell cards for anyone who wanted to write support into OZ... Nov 01 14:48:12 kerwood: oesf? Nov 01 14:49:11 hrw: damn, I'm having trouble finding all these references -- perhaps it _was_ mentioned in oesf... Nov 01 14:54:29 when building my image with this prebuilt toolchain, I got lots of warnings like "NOTE: Couldn't find shared library provider for libpthread.so.0" at the do_package stage... I was expecting this - it is described on the wiki, and a possible solution to get the missing shared libs included in the image is also described: Nov 01 14:54:41 "To workaround this issue you could create a package for the metadata to install every needed library and use ${BOOTSTRAP_EXTRA_RDEPENDS} to make sure this package is installed when creating images." Nov 01 14:55:06 tkp: does that relate to this error I have here? Nov 01 14:55:08 NOTE: Resolving missing task queue dependencies Nov 01 14:55:10 ERROR: No providers of runtime build target busybox Nov 01 14:55:17 but what do I need to add to RDEPENDS for this new meapackage to get libpthread.so.0 installed in the image for example Nov 01 14:55:59 morning Nov 01 14:56:23 likewise: I have not seen that error before, but I don't think it's related Nov 01 14:56:48 kerwood: thx. it pointed me to good directrion Nov 01 14:57:00 XorA: for once I didn't fiddle with anything :) Nov 01 14:58:27 how can I create a package which pulls the shared libraries from my external toolchain? Nov 01 14:59:00 koen|gprs: wolfson@openembedded.org seems to be missing Nov 01 14:59:31 XorA: hrm Nov 01 14:59:43 XorA: who is hiding behind it? Nov 01 15:00:09 hrw: its a key for our OE mirror Nov 01 15:00:54 ah Nov 01 15:01:04 XorA: 64 bytes from 64.233.187.99: icmp_seq=1091 ttl=237 time=11074.044 ms Nov 01 15:01:09 which since last night stopped mirroring Nov 01 15:01:14 XorA: so I can't fix that right now Nov 01 15:01:28 XorA: try bugging mickeyl :) Nov 01 15:01:33 koen|gprs: its ok, no hurry, we arent ready to go public Nov 01 15:01:41 koen|gprs: mickeyl is 200% CPU load Nov 01 15:05:57 hrw: sorry, I was wrong. I don't have a 2.6 driver, just source for the 2.4 driver :(. Nov 01 15:05:57 hrw: Get it at: http://www.sparklan.com/download.php?support_id=13&ranload=1162393887 Nov 01 15:06:09 kerwood: 2.4 I have too Nov 01 15:07:39 kerwood: but maybe this one is newer Nov 01 15:09:00 its same Nov 01 15:10:20 anyway gumstix guys have nice patchset.. Nov 01 15:11:53 hrw: here's the mention of them on the OZ website: http://openzaurus.org/wordpress/developers#comment-3278 Nov 01 15:16:11 has anyone ever done a metapackage which pulls in shared libraries from an external toolchain? I can't find an example of how this might be done anywhere Nov 01 15:16:32 tkp: just have a do_install method Nov 01 15:16:43 tkp: more info only if you will write docs :) Nov 01 15:16:58 zecke: I'll write some docs when I get it working! Nov 01 15:17:32 I have a do_install() method... but how do I grab the libs from the toolchains lib dir? Nov 01 15:18:03 intall ${TOOLCHINE}/lib/libxx.so ${STAGING_LIBDIR} Nov 01 15:18:11 file://${PRE_BUILT}/lib/ Nov 01 15:18:12 right Nov 01 15:19:02 with the right args Nov 01 15:19:45 oh, so I only need to put thin in STAGING_LIBDIR, and shlibs will grab them from there? Nov 01 15:20:29 tkp: you wanting to package them as well? Nov 01 15:20:45 tkp: sorry that was for do stage Nov 01 15:21:10 install ${PRE_BUILD}/lib/libxx.so ${D}${libdir}/ Nov 01 15:21:14 for do_install Nov 01 15:21:43 or similar Nov 01 15:21:54 I was thinking of making a package and including it in ${BOOTSTRAP_EXTRA_RDEPENDS} as per the wiki... byut perhaps it would be better to add them to stage and let shlibs ggrab them from there so they are included in all other packages which need them Nov 01 15:23:05 although, I can't think of a way of adding it to DEPENDS for all packages that will need to use them... maybe I'll just put them in their own package after all Nov 01 15:23:09 tkp: dont make mistake I just did, ${STAGING} is for building against, ${D}/blah is for packaging Nov 01 15:23:17 tkp: sorry I caused confusion Nov 01 15:23:20 XorA: np Nov 01 15:24:10 RP: ping Nov 01 15:24:59 sirfred: pong Nov 01 15:25:13 sirfred: I saw the email, Haven't had time to look at tit yet though :-( Nov 01 15:25:44 RP: No problem. Just a question. Nov 01 15:26:32 RP: I'm making some tranfer test from host memory to imageon one, and also blitting from external to external, external to internal,... Nov 01 15:27:02 RP: Is there any way to know what bandwidth should we expect? Knowing how the imageon host interface works and so on? Nov 01 15:27:25 sirfred: We know the memory bus speed... Nov 01 15:27:43 RP: The zaurus bus? Nov 01 15:28:13 sirfred: PXA memory bus - the w100 is a memory mapped device Nov 01 15:28:46 RP: But I suppose that the internal and external memory of the imageon itself are accesed by the imageon in a different fashion. Nov 01 15:28:59 sirfred: You should be able to look at the memory control registers for that memory bank too. Nov 01 15:29:08 sirfred: Yes, the external memory is slower Nov 01 15:29:59 RP: Unfortunately, very slow after my first tests with the mplayer videoout driver I'm developing Nov 01 15:30:06 sirfred: The chip numbers for the external memory were available online somewhere. The access speed was set in some w100 registers too iirc Nov 01 15:30:31 RP: It would be great if you could provide me that information. Nov 01 15:31:26 sirfred: You have as much access to it as I do - the config information is what we have in w100fb.h and what w100fb.c sets it to in the Zaurus case Nov 01 15:32:09 RP: If you're interested, these are the numbers I've got Nov 01 15:32:38 (I) Copy from host to internal memory : 13.405156 MBps Nov 01 15:32:49 (I) Copy from host to external memory : 11.288802 MBps Nov 01 15:33:11 (I) HOSTDATA from host to external memory: 5.669435 MBps Nov 01 15:33:21 (I) HOSTDATA from host to internal memory: 5.669566 MBps Nov 01 15:33:37 (I) BLT from internal to internal memory: 143.381181 MBps Nov 01 15:33:52 (I) BLT from external to external memory: 18.056163 MBps Nov 01 15:34:04 (I) BLT from internal to external memory: 45.282603 MBps Nov 01 15:34:45 So using internal memory where possible is a big gain... Nov 01 15:34:50 I'm not sure if they make sense, but this seems to be what I have, using the default configuration of the imageon. Nov 01 15:35:03 RP: Of course, but for an overlay surface, it's not enough. Nov 01 15:35:23 sirfred: right :-( Nov 01 15:35:39 sirfred: So what you're saying is an overlay in external memory is too slow to be useful? Nov 01 15:35:51 RP: I might have figured out why the oss-mixer emulation does not change the volume for alsa-soc Nov 01 15:36:11 RP:That is what I'm thinking now, but perhaps it's a configuration matter. Nov 01 15:36:45 RP: Even if we were able to have hardware IDCT, needing to overlay in a external surface should be a problem. Nov 01 15:37:04 sirfred: I just don't know... Nov 01 15:37:20 hvontres|poodle: What do you think the problem is? Nov 01 15:37:24 RP: it looks like we currently have no Master Volume setting on the alsa side of things. Nov 01 15:38:06 ${libdir} points to /usr/lib... is there a var which points to /lib? Nov 01 15:38:11 RP: For example, I have a test program that creates an overlay from a source surface YUV420, and uses acceleration to rotate and scale it on the overlay surface. The size of the overlay surface is critical. Nov 01 15:38:51 I was playing with nmixer and setmixer (text based oss controls) and the only settings I get are "line" "igain" and "ogain". if I change "ogain", the alsa playback volume does change Nov 01 15:39:49 * XorA hides from SoC discussions Nov 01 15:40:03 SoC? Nov 01 15:40:15 RP: so If there is an easy way to map "Playback" to "Master Volume" the existing mixer controls should all work Nov 01 15:40:22 tkp: I forgot you, sorry :) Nov 01 15:40:31 hvontres|poodle: You need to ask our ALSA expert in training (XorA) ;-) Nov 01 15:40:38 tkp: no libdir does not point to lib, force it to install there Nov 01 15:40:39 RP: A source surface of 208x128, for example, gives me a rate of 152fps, but if I try to double its size scaling, the rate goes down to 50fps Nov 01 15:40:41 hvontres|poodle: Offhand I don't know Nov 01 15:40:50 tkp: normally one does not install into libs :) Nov 01 15:40:56 zecke: k, np... Nov 01 15:41:03 * hvontres|poodle sends the houds to find XorA Nov 01 15:41:07 tkp: mkdir /lib in ${D} when installing Nov 01 15:41:11 so, I'm going to have to set up all the .so.* symlinks manually too Nov 01 15:41:13 tkp: and use oe_libinstall Nov 01 15:41:32 sirfred: Keep in mind doubling the size increases the bandwidth by a factor of 4... Nov 01 15:41:36 oe_libinstall? what does that do different to just using install? Nov 01 15:41:41 XorA : any ideas? Nov 01 15:41:44 RP: Setting fastpllclk to 1 helps a little. And the best rate goes to 175fps Nov 01 15:42:03 sirfred: We should always use fastpllclk Nov 01 15:42:14 sirfred: It increases the w100's internal speed Nov 01 15:42:18 RP: Yes, but that's not host-imageon bandwidth. Nov 01 15:42:29 RP: Its imageon-imageon bandwidth, with its own memory Nov 01 15:42:40 hvontres|poodle: Ill talk to lrg about it Nov 01 15:42:41 sirfred: Does it imporve the external memory access? Nov 01 15:43:25 RP: Looking at the numbers, host-to-external is around 11MBps and internal-to-external is around 42 MBps Nov 01 15:43:26 OT: http://de.wikipedia.org/wiki/Bonsai <- the two favorite names in this article are? Nov 01 15:43:58 RP: But it seems to be not enough, anyway. Unless I'm making some horrible mistakes. Nov 01 15:44:23 XorA: Cool... I think this would help get some more people away from the dreadded crappix kernels :) Nov 01 15:44:32 sirfred: I haven't the time atm to think about this - I need to do other things. Perhaps if you're around this evening...? Nov 01 15:44:49 RP: Oh, excuse me. I don't want to disturb you. Nov 01 15:44:57 hvontres|poodle: I dont actually have an OSS mixer installed Nov 01 15:45:04 RP: Just wanted to know your opinion. When you have time. Nov 01 15:45:27 * Crofton is pleased with alsa on the OSK Nov 01 15:45:44 just don't try building the sound driver as a module.. Nov 01 15:45:52 sirfred: Offhand, I think you might be right and that bandwidth could be a problem :-/ Nov 01 15:47:43 XorA: I used nmixer and setmixer from the OZ feeds on my poodle. Nov 01 15:48:17 hvontres|poodle: I shall try and remember to check into it tomorrow, my corgi is at home and thats same codec as poodle Nov 01 15:48:31 koen|away: okay my Bug Monotir is called Ikada, the stuff we knew as tinderbox is now Hokidachi Nov 01 15:49:27 XorA: Thanks a lot Nov 01 15:49:48 oelibinstall... nice :) Nov 01 15:53:46 sirfred: I have a feeling if you can successfully downscale to 320x240 at a decent speed 99% of users would be happy Nov 01 15:53:55 sirfred: ipod video is afterall only that res Nov 01 15:54:32 zecke: Hokidachi remind Zauruses ;D Nov 01 15:54:57 hrw: they do? I will label my software as names of bonsai trees Nov 01 15:55:29 XorA: Well, I was testing with a 352x288 video scaling to 640x480 more or less. Nov 01 15:55:30 zecke: for me it reminded - both are .jp Nov 01 15:55:36 XorA: It's very slow, but I expected more. Nov 01 15:56:32 XorA: I'm not really sure if without scaling it's working fine, because I was not able to put sound to work in mplayer. Nov 01 15:56:53 XorA: So, I'm not sure if it's playing at 1:1 speed or near Nov 01 15:57:06 sirfred: well that would of course be uber cool, but to go from any size to 320x240 would be enough Nov 01 15:57:18 sirfred: whats the sound problem? Nov 01 15:58:15 XorA: Don't know. it just does not work using alsa. Nov 01 15:58:43 sirfred: odd, with OE .dev? I use that everyday Nov 01 15:59:02 sirfred: hmm, check what sample rate its trying to play Nov 01 15:59:25 sirfred: If its less than 44.1KHz then you tried a known bug Nov 01 15:59:34 s/tried/tripped/ Nov 01 15:59:35 XorA: In dmesg I can see a lot of lines like: Nov 01 15:59:42 <4>soc: DAI[15:0] failed to match rate Nov 01 15:59:49 sirfred: thats normal Nov 01 16:00:02 Buf finally, I can see: <4>soc: audio rate 44100 chn 2 Nov 01 16:00:02 <4>pop wq checking: Playback status: inactive waiting: yes Nov 01 16:00:12 Its a 44.1Khz afaik Nov 01 16:01:08 But I can hear nothing, I've checked with alsamixer to be sure the volume was not muted, and it wasn't. Nov 01 16:01:25 XorA: Using 2.6.17-r28 Nov 01 16:01:55 sirfred: you need to unmute hifi playback as well Nov 01 16:02:01 though zaurusd should do that Nov 01 16:02:38 XorA: Weird, now it's working. I was just trying with the fbdev vo driver, and started to sound. Nov 01 16:03:00 It works randomly. Now it doesn't Nov 01 16:03:57 Oh, I see that it was using oss driver, I don't know why. Forcing alsa with -ao alsa seems to work now. Nov 01 16:04:32 sirfred: /etc/mplayer/mplayer.conf forces oss Nov 01 16:04:45 now alsa is fully supported should really fix that Nov 01 16:05:18 XorA: Well, I can see that with my w100 driver it's slower than the fbdev one. Nov 01 16:05:40 As I need to make two copies, one from host to the internal memory and another one from the internal memory to the external to rotate. Nov 01 16:06:36 So, I'm getting 34-36% time in vo driver while the fbdev one is only 22%-24% Nov 01 16:07:36 The video is actually 384x288 Nov 01 16:18:59 sirfred: what I suspect most users would love to see is 640/480x480 divx play at 30fps at 320x240 without re-encode Nov 01 16:19:15 ie, the most common download format Nov 01 16:19:26 sirfred: if thats faster way of doing things Nov 01 16:19:39 I'm having a little trouble with oe_libinstall... in my source directory I have 4 files for libutil: libutil-2.4.so, libutil.a, libutil.so -> libutil.so.1 & libutil.so.1 -> libutil-2.4.so Nov 01 16:19:52 I run oe_libinstall like so: oe_libinstall -C ${PRE_BUILT}/lib libutil ${D}/lib Nov 01 16:20:05 and I end up with two sytmlinks only Nov 01 16:20:06 XorA: The problem is always the bandwidth. Nov 01 16:20:15 libutil.so -> libutil.so.1 & libutil.so.1 -> libutil-2.4.so Nov 01 16:20:16 sirfred: I know :-( Nov 01 16:20:22 XorA: We can not upload that amount of data to the imageon. Nov 01 16:20:43 damn ATI for not making imageon handle 640x480 properly Nov 01 16:21:05 XorA: Being able to use directly YUV surfaces instead of letting mplayer to convert to RGB, is saving 8 bits per pixel, when using YUV420 Nov 01 16:21:12 and I get "NOTE: stv-shlibs contains dangling symlink to /lib/libutil-2.4.so" Nov 01 16:21:34 how can I make oe_libinstall install the actuall lib and not just the symlinks? Nov 01 16:21:42 XorA: The original w100 driver only supports a type of scaling done directly in the video_ctrl register. Nov 01 16:22:08 An expansion that can make the video size 2x, 3x, or 4x, AFAIR Nov 01 16:25:20 XorA: I think they didn't expect overlay surfaces to be placed in external memory. Nov 01 16:26:21 sirfred: thats a pity Nov 01 16:26:25 So, we only can use external memory for that. We could try: Nov 01 16:26:25 host->external--(Rotate + Scale to 320x240)-->internal (overlay scaling to 2x, 2y) Nov 01 16:26:55 damn ati, and damn sharp for using broken ati :/ Nov 01 16:27:19 (also, damn you, for supporting sharp's decision by buying a zaurus with broken ati chip) Nov 01 16:27:24 :) Nov 01 16:27:36 sirfred: unfortuneately my W100 gather dust while I use my pxafb Nov 01 16:27:46 from what I learnt recently c7x0 was released for one reason - to show people CG silicon screen technology Nov 01 16:28:23 hrw: sigh Nov 01 16:28:32 I wonder how newer releases of the imageon perform Nov 01 16:28:34 time for me to go catch trains Nov 01 16:28:43 XorA|gone: cu Nov 01 16:28:55 xora: happy commuting! Nov 01 17:02:33 shit.. someone know how to recover file from google cache format? Nov 01 17:03:46 copy - paste? Nov 01 17:04:26 altho i am guessing you already tried that Nov 01 17:05:04 emte: binary file... Nov 01 17:08:23 cu Nov 01 17:11:39 hrw|afk, maybe try wget and pull the page recursivly Nov 01 17:37:25 I have this metapackage which installs every shared lib that I need on the target system, but when it comes to the do_package() stage, bitbake complaines about Multiple provides for some of the libs and asks me to set LEAD_SONAME. Nov 01 17:37:58 is it possible to set LEAD_SONAME to more than one lib name, or must I create a seperate package for each shared lib? Nov 01 18:06:13 is there any machines in OE that use the UCB1xOO CODEC has touchscreen support and wake-up support by touching the touchscreen? Nov 01 18:20:11 re Nov 01 18:20:20 Gerrath: waking by TS? never heard Nov 01 18:24:25 hrw, the UCB1400 audio codec supports it, I just wondered if anyone ever used it. Nov 01 18:27:53 hrw, I thought it would be a popular feature but I finding out that I'm wrong :-/ Nov 01 18:28:00 Gerrath: several years ago, I worked on a few systems that implemented TS wake, so it is possible with the UCB1400. Nov 01 18:28:28 Gerrath: I think most phones/PDAs don't implement it because you don't want to stick the device in your pocket and have it waking up. Nov 01 18:28:35 Gerrath: seems more common on industrial terminals. Nov 01 18:29:04 cbrake, do you remember or have examples of the correct settings for the UCB1400? The datasheet is a little bit slim in that subject. Nov 01 18:37:32 btw - looks like my updater.sh works ok on spitz too - but I did not yet rebooted Nov 01 18:38:07 booted ;) Nov 01 18:38:24 koen|gprs: psplash is rotated on spitz Nov 01 18:38:38 hrw: bug mallum :) Nov 01 18:40:13 opie sysinfo show linux kernel 2.6.18-17.3 ;D Nov 01 18:40:23 umbaumba 17.3 booted opie on spitz Nov 01 18:41:46 Gerrath: no, that was a few lives ago. Nov 01 18:47:43 zecke2: cool Nov 01 18:49:43 XorA|gone: re-added the key, I suspect it got lost in the regenerate_rosters cycle last week Nov 01 18:49:52 XorA|gone: syncing should work now Nov 01 18:50:26 koen|gprs: can you create database for openzaurus @ltg? Nov 01 18:52:01 hrw: sure, send me the instructions and I can do it right now Nov 01 18:52:25 koen|gprs: you never administrated mysql? Nov 01 18:52:48 I loathe mysql Nov 01 18:53:22 /dev/null has better data integrity Nov 01 18:53:34 koen|gprs: Too true Nov 01 18:55:34 03pfalcon 07org.oe.dev * r5a0b7138... 10/ (1 conf/distro/angstrom-2007.1.conf): Nov 01 18:55:34 angstrom: Upgrade linux-handhelds-2.6 for htcuniversal to 2.6.17-hh2. Nov 01 18:55:34 * Acked-By: goxboxlive Nov 01 18:55:34 * Approved by Koen Kooi. Nov 01 19:02:35 hi all - just wanted to double check - if I have some new packages and distros, and image, confs - should I submit those as bugs via bugzilla & attachments? Nov 01 19:02:50 that is what I gather from the wiki, but wanted to double check before I did Nov 01 19:02:58 guttata_: yes Nov 01 19:03:15 : thanks Nov 01 19:06:25 http://www.google.pl/search?q=openembedded+documentation Nov 01 19:06:37 looks shitty.. no links to documentation Nov 01 19:07:51 http://www.uv-ac.de/openembedded/ Nov 01 19:19:25 Hmm I need to install a directory called /var/omninames Nov 01 19:19:35 what var would be for /var/ Nov 01 19:20:35 Crofton: ${localstatedir} Nov 01 19:20:42 Crofton: see conf/bitbake.conf Nov 01 19:20:46 thanks Nov 01 19:27:58 psokolovsky: Also add +cvs${SRCDATE}" so we allways build form the cvs. Because cr2 is always comittiing in the head of cvs. Nov 01 19:28:55 goxboxlive, well we all commit to head, but not all the time what works immediately ;-) Nov 01 19:29:37 goxboxlive, CVS HEAD may be broken, etc. So, idea is to have stable tags in public stuff Nov 01 19:29:44 I am allways testing it before commiting. Nov 01 19:30:00 goxboxlive, you can add that locally Nov 01 19:30:20 yes u probably right, i have Nov 01 19:30:23 goxboxlive, or, ask Koen otherwise. I got ok for -hh2 upgrade, nothing else. Nov 01 19:30:58 np, i'll edit localy Nov 01 19:31:17 ok Nov 01 19:31:46 I'd like to avoid depending on floating srcdates as much as possible Nov 01 19:31:50 mickey|sports: hmm why would split(sep='\n') split on ','? Nov 01 19:32:10 zecke: because python isn't awk :) Nov 01 19:32:30 I have created a metapackage which uses oe_libinstall to install lots of missing shared libs. When I look in the install/ directory, I can see all the libs and symlinks. I can also see all the files in the rootfs/ directory. But when I flash with the image, lots of the libs are missing Nov 01 19:32:40 any ideas what might be going on? Nov 01 19:32:55 tkp: yes, you should take the "Not packaged..." messages seriously Nov 01 19:33:07 I suspect it's something todo with the LEAD_SONAME thing... Nov 01 19:33:11 tkp: no Nov 01 19:33:25 zecke: there were none of those message when I cooked my metapackage Nov 01 19:33:34 tkp: LEAD_SONAME comes if you install two libraries into the same package Nov 01 19:33:51 well, I have installed about 10 libs in this package Nov 01 19:34:09 tkp: reread the log message, update FILES, check the content of the package Nov 01 19:34:20 tkp: but rereading the log to see the message makes most sense IMHO Nov 01 19:34:39 zecke: I see all the files in install/... I thought that means they should be installed Nov 01 19:35:04 tkp: dunno, verify your hypothesis? Nov 01 19:35:07 only log message I get is NOTE: Multiple libraries (... + a list of all my libs) Nov 01 19:36:09 I don't understand how the files can be in install/ and in rootfs/ but not in the actual image Nov 01 19:36:10 tkp: no it is still not LEAD_SONAME Nov 01 19:36:24 tkp: well, then use the right image? Nov 01 19:36:35 heh.. yeah. I am. for sure Nov 01 19:36:49 tkp: or your image is too big and got truncated? Nov 01 19:37:00 no. the image has loads of space left Nov 01 19:38:32 tkp: so, if it is in rootfs and not in your image? Nov 01 19:38:46 tkp: mount your image loopback and take a look at the content Nov 01 19:39:03 bloody hell! There is no nethack.bb in OE? What kind of geeks are we sheesh Nov 01 19:39:49 CoreDump|home, lol, Ive actually got complaints about that before... Nov 01 19:39:50 hmm.. no... files are not in the ext2 file... Nov 01 19:39:57 * tkp deletes all existing image files and tries again Nov 01 19:39:58 tkp: so what? Nov 01 19:40:17 you can loopback mount pretty much everything... Nov 01 19:40:27 ~praise loop-mounts Nov 01 19:40:28 or use mtdram and you can mount any flash filesystem Nov 01 19:40:28 All hail loop-mounts! Nov 01 19:40:39 zecke: yes, I loopback mounted it and the files are not in there Nov 01 19:40:59 tkp: ah okay, and they are in the package? Nov 01 19:41:36 * tkp thinks that perhaps if an image of the same name exists in deploy/images then it will not be overwritten when I recook the image.bb Nov 01 19:42:04 zecke: yes... they are in the package... or at least, they are in package/install/ Nov 01 19:42:28 tkp: but as the images have time in the name, it is unlikely that it will try to overwrite it Nov 01 19:42:40 my image only has the date Nov 01 19:44:21 zecke: are there any docs for your tiny trees? Nov 01 19:45:24 koen|gprs: what kind of docs? Nov 01 19:45:30 re Nov 01 19:45:47 koen|gprs: besides the german wikipedia pages? Nov 01 19:45:58 I finally enter a way to amd64. just sold cpu which is in this box. Nov 01 19:46:29 hehe Nov 01 19:46:37 hrmpf... deleted the image, recooked it and mounted via loopback, but still there are missing files :( Nov 01 19:46:41 hrw: GPUs are powerful enough to run linux and start irssi? Nov 01 19:47:10 zecke: usage, aims, known bugs, etc Nov 01 19:47:44 koen|gprs: yes, they are marked confidential and have not been written to paper :) Nov 01 19:47:48 that is a good excuse :) Nov 01 19:47:55 zecke: I use NPU now Nov 01 19:48:45 zecke: ah, trolltech hired you Nov 01 19:49:21 koen|gprs: no, they are too stupid to make me an offer :) Nov 01 19:49:40 koen|gprs: Nov 01 19:49:40 http://bugs.openembedded.org/buglist.cgi?bug_file_loc=&bug_file_loc_type=allwordssubstr&bug_id=&bugidtype=include&chfieldfrom=10%3A00&chfieldto=12%3A00&chfieldvalue=&email1=&email2=&emailassigned_to1=1&emailassigned_to2=1&emailcc2=1&emailqa_contact2=1&emailreporter2=1&emailtype1=substring&emailtype2=substring&field0-0-0=noop&long_desc=&long_desc_type=substring&query_format=advanced&remaction=&short_desc=&short_desc_type=allwordssubstr&type0-0-0=noop&value0-0-0=&ctype Nov 01 19:51:30 zecke: ok, got it... if an image already exists of the same name, it will not be overwritten... much like with .ipkg files Nov 01 19:51:42 time to add a timestamp to my image name Nov 01 19:51:43 tkp: hmm Nov 01 20:16:44 koen|gprs: have you read daniel's comment? Nov 01 20:16:56 koen|gprs: http://blogs.qtdeveloper.net/archives/2006/10/23/the-death-of-opie-is-greatly-exaggerated/ the second one :) Nov 01 20:18:13 ~curse ipkg for being ... and .... and .. Nov 01 20:18:15 May you be reincarnated as a Windows XP administrator, ipkg for being ... and .... and .. ! Nov 01 20:18:58 zecke: the infamous sharp lobotomy Nov 01 20:19:38 koen|gprs: What is wrong with Closed Source? :) Nov 01 20:20:15 zecke: that's a trick question :) Nov 01 20:20:32 zecke: it assumes there is a source Nov 01 20:20:49 koen|gprs: I was attempted to write I never did Open Source :) Nov 01 20:21:10 'Free Software" Nov 01 20:23:35 good info: mc from OZ 3.5.4.1 works under umbaumba 17.3 Nov 01 20:24:56 ~lart ipkg again. Nov 01 20:24:59 * ibot takes out a cattle prod and gives ipkg again. a good jolt Nov 01 20:25:34 sometimes I think angstrom should invent its own package format Nov 01 20:25:52 koen|gprs: msi Nov 01 20:25:58 just to stop users from thinking ".ipk == ipk, so it *must* work" Nov 01 20:25:58 koen|gprs: angstrom NEED to have one main point of contact and it should be listed on webpage Nov 01 20:26:06 koen|gprs: apk? Nov 01 20:26:17 .dab :) Nov 01 20:26:33 .shar? :) Nov 01 20:26:43 dumb angstrom bufons? Nov 01 20:27:47 hrw: you mean like http://www.angstrom-distribution.org/contact ? Nov 01 20:28:29 'night all Nov 01 20:28:40 koen|gprs: no Nov 01 20:28:46 hi koen|away Nov 01 20:28:55 koen|gprs: it lack *main* point of contact - one person Nov 01 20:34:02 cu tomorrow Nov 01 20:43:31 koen|gprs: cheers for that, I shall restart our syncer tomorrow Nov 01 21:06:33 nite guys Nov 01 21:08:34 03pfalcon 07org.oe.dev * r420b7125... 10/ (1 packages/linux/linux-handhelds-2.6/h2200/defconfig): Nov 01 21:08:34 h2200: New defconfig for 2.6.17-hh2. Nov 01 21:08:34 * Based on Michal Panczyk's submission, with some more tweaks. Nov 01 21:08:34 * Tested to compile, but needs run testing. Nov 01 21:16:14 03kristoffer 07org.oe.dev * r287486df... 10/ (1 MAINTAINERS): MAINTAINERS : Added myself for the good of all man kind Nov 01 21:16:28 I assume if I want a WRT54 I should get the GL to be safe? Nov 01 21:36:31 03kristoffer 07org.oe.dev * rb125865c... 10/ (1 conf/distro/jlime-donkey.conf): Nov 01 21:36:31 conf/distro/jlime-donkey.conf: Update to distro file Nov 01 21:36:31 * Increase optimization to -O2 as 4.1.1 seems stable enough Nov 01 21:36:31 * Prepare for using DISTRO_EXTRA_RDEPENDS Nov 01 22:32:53 Night Nov 01 23:41:59 03rpurdie 07org.oe.dev * r3c603fa4... 10/ (16 files in 3 dirs): linux-oz-2.6: Update to a 2.6.18+git kernel Nov 02 02:13:15 03pfalcon 07org.oe.dev * r05d362f2... 10/ (1 conf/machine/h2200.conf): Nov 02 02:13:15 h2200: Remove kernel as explicit dependency. Nov 02 02:13:15 * task-base depends on kernel, so that's how it is pulled in. Nov 02 02:13:15 * Other machines don't have such explciit dependency. Nov 02 02:13:15 * Part of general clean up of configs for hh.org-based machines. Nov 02 02:24:54 03pfalcon 07org.oe.dev * r7b0b1b3e... 10/ (1 conf/machine/h2200.conf): Nov 02 02:24:54 h2200: Remove mtd-utils utils from MACHINE_EXTRA_RDEPENDS Nov 02 02:24:54 * mtd-utils are not h2200-specific, so have no place in its machine config. Nov 02 02:24:54 * Should be really be part of MACHINE_FEATURES, or, much better, part Nov 02 02:24:54 of debug/devel images only (it's really unsafe to by default install Nov 02 02:24:55 utils which can trash flash into the user images). **** ENDING LOGGING AT Thu Nov 02 02:59:57 2006