**** BEGIN LOGGING AT Wed Aug 08 02:59:58 2012 Aug 08 04:44:39 http://blog.mikeasoft.com/2007/06/30/im-in-ur-treo-linuxin-ur-phonez/ anybody know if the source for this treo kernel is available anywhere? Aug 08 04:48:38 it was 5 years ago so my memory's a bit shaky, but I think I was using the vanilla kernel sources at the time Aug 08 04:48:48 I think everything necessary for the 650 was already in the kernel tree Aug 08 04:52:11 ah, so theres a chance i could just build for it on my current system Aug 08 04:52:26 Elleo: but what configs would i have to turn on? :S Aug 08 04:52:41 my memory isn't that good I'm afraid ;) Aug 08 04:53:07 neither is the internets, half the pages on it are 404, lol Aug 08 04:53:18 clever: http://www.simiolabs.com/developer/wiki/index.php?title=Extracting_the_config_file_from_an_existing_kernel Aug 08 04:53:21 most of the blogs just link to a dead page Aug 08 04:53:28 was just about to google for that Aug 08 04:54:14 and i probly have a dozen copies of the script :) Aug 08 04:54:22 extract-ikconfig: Cannot find kernel config. Aug 08 04:54:23 :( Aug 08 04:54:28 ah :/ Aug 08 04:54:43 I expect a fair amount would have changed by now anyway Aug 08 04:55:04 i was just wondering how hard it would be to shove android into that old thing Aug 08 04:55:31 I expect it'd be pretty much impossible Aug 08 04:55:47 the memory constraints alone would probably cause serious issues Aug 08 04:55:47 why? Aug 08 04:56:36 the 650 only has 32mb of ram Aug 08 04:56:40 ah Aug 08 04:56:51 wikipedia was very vague on what was ram and what was flash Aug 08 04:57:04 what about the 700? Aug 08 04:57:09 not to mention all the hardware integration work you'd need to do to get anything along the lines of phone/data working Aug 08 04:57:40 I don't know if anything's changed, but when I was hacking on the 650 (5 years ago) no-one had really got anything running on the 700 Aug 08 04:58:04 i believe the bootloader 'worked', but none of the kernels i tried did anything usefull Aug 08 05:00:14 *beams cocoboot over* Aug 08 05:01:07 hmmm, sd card, *looks* Aug 08 05:02:40 Elleo: is the sd card reader in the treo just gpio? Aug 08 05:03:08 clever: these are old configs, but could be a starting point https://github.com/hackndev/linux-hnd/tree/master/arch/arm/configs Aug 08 05:03:49 mozzwald: ah nice Aug 08 05:04:03 *checks local tree* Aug 08 05:04:34 clever: I really couldn't say anything about the treo with any certainty at this point I'm afraid; shortly after that 650 hack I switched to a different phone platform and gave my treo away Aug 08 05:04:42 missing, but i can copy and tweak Aug 08 05:05:10 Elleo: all ive found so far are SDHD cards, and i was wondering if the support issue would be fixed by linux Aug 08 05:05:32 depends on where the SD controller is implemented Aug 08 05:07:26 dang, 5 CF cards and 2 SDHD cards, where did they all go? lol Aug 08 05:10:37 ah, i see why those configs arent in my android-x86 fork of linux, hnd have their own fork! Aug 08 05:13:51 CONFIG_FB_PXA seems to be the display Aug 08 05:15:55 built into the cpu :) Aug 08 05:17:13 hmmm cocoboot says 32mb ram for the 700p Aug 08 05:17:28 opening the lcd registers kills it, oops! Aug 08 05:19:27 do you happen to remember if the PXA cpu supports page flipping in the gpu area? Aug 08 05:21:45 reguarding SDHD support, it seems the cpu has a built-in SD card controller :S Aug 08 10:31:31 clever: the internal SD thingie can do SDHC as well, no prob Aug 08 10:31:42 it only passes commands to the SD card, nothing else anyway Aug 08 18:40:49 Marex: ah, i was wondering Aug 08 18:51:30 clever: the SD controller is basically only a bit better SPI Aug 08 18:51:42 there's nothing much more to it Aug 08 18:52:29 so i could make linux use an SDHD card, but cocoboot would have issues Aug 08 18:52:40 yea, palmos is dumb Aug 08 18:52:50 not really cocoboot ... cocoboot just calls palmos Aug 08 18:53:08 is cocoboot open source? Aug 08 18:55:03 i was thinking, add a delay between loading the kernel, and running it Aug 08 18:55:06 so id have time to swap cards Aug 08 18:55:54 clever: it is ... but why not rather use initrd what waits until you replace the cards ? Aug 08 18:56:20 that works too, just program it to look for a certain cart Aug 08 18:56:29 clever: or wait for a keypress Aug 08 18:56:29 then you have the choice to put them on the same card Aug 08 18:56:46 same card won't work, palmos won't pick sdhc Aug 08 18:56:55 you would need non-hd Aug 08 18:57:12 you can choose between non-hd or nonhd+sdhd Aug 08 19:00:57 HD ? Aug 08 19:01:13 btw yes, you'd need two cards Aug 08 19:01:34 yeah, if you wanted to use an SDHD card you can setup the initrd for split mode Aug 08 19:01:50 but the same initrd should work fine on a single normal SD card, it would just find the files right away Aug 08 19:02:09 so it depends mainly on how much space you would want Aug 08 19:02:19 oh this is what you mean Aug 08 19:02:26 yes ... whatever fits your needs Aug 08 19:02:38 ive already done the same thing to android, for other reasons Aug 08 19:02:57 the kernel lacked xfs, so once the initrd was up, i had to switch to an mp3 (fat32) to get the rest of the file Aug 08 19:03:22 mp3 player was the only 'usb stick' i had laying arround Aug 08 19:03:27 xfs? yuck yuck Aug 08 19:04:09 its a usb stick i made years ago with live gentoo Aug 08 19:04:26 now its a stick with live gentoo and live android Aug 08 19:09:14 Marex: figured out why libertas wasn't working on my TX. I didn't compile the pcmcia module.. silly me. still working on the pda_power/power_supply module problem. Aug 08 19:11:30 also got my hands on a tungsten c but I can't get keyboard to work. kernel complains failed to request gpio18 for col0 Aug 08 19:15:13 oh well Aug 08 19:15:22 too old kernel maybe ... I dunno how much got mainline either Aug 08 19:18:57 you still using your tc? Aug 08 19:19:57 I still have it ... I now use DENX M28EVK mostly Aug 08 19:20:28 and eventually, I'd like to get to hacking on the BD SabreLite MX6Quad board Aug 08 19:20:43 you know ... quadcore arm ... it's interesting Aug 08 19:23:02 yes, more cores is very interesting :) **** ENDING LOGGING AT Wed Aug 08 20:13:49 2012 **** BEGIN LOGGING AT Wed Aug 08 20:14:16 2012 **** ENDING LOGGING AT Wed Aug 08 20:41:31 2012 **** BEGIN LOGGING AT Wed Aug 08 20:42:22 2012 Aug 08 20:55:12 hmmm, the PXA cpu has jtag debuger in it, i wonder if the pins are broken out Aug 08 20:59:03 clever: on what device ? Aug 08 20:59:09 LD? yes ... TX? I think yes Aug 08 20:59:15 but they're very small Aug 08 20:59:24 only TC has holes for a standard ARM14 connector Aug 08 20:59:36 Marex: treo 700p Aug 08 20:59:48 likely ... on a very small and hard to solder pads Aug 08 20:59:53 ive already cracked the thing open once to read the exact number on the cpu Aug 08 21:00:14 you could have red it via the ID register Aug 08 21:00:18 CPUID or JTAGID or whatever Aug 08 21:00:18 any tips on where to find them? Aug 08 21:00:28 cocoboot only says 27x Aug 08 21:00:30 I'd have one, but you'd dislike it Aug 08 21:00:38 the lasered code says 272 i believe Aug 08 21:00:41 which is enough anyway Aug 08 21:00:56 there's not much difference. see the datasheet Aug 08 21:01:05 the lcd registers in cocoboot just crash the system Aug 08 21:01:07 btw the CPUID register does tell you the same Aug 08 21:01:13 that;s possible Aug 08 21:01:36 ah, i see, yeah the cpuid isnt decoded Aug 08 21:01:43 0x69054117 Aug 08 21:02:11 yea Aug 08 21:02:19 crosscheck with the marvell pxa27x datasheet Aug 08 21:02:19 Intel_PXA270_Developers_Manual is that the right datasheet? Aug 08 21:02:26 works as well Aug 08 21:02:46 http://www.marvell.com/application-processors/pxa-family/assets/pxa_27x_dev_man.pdf Aug 08 21:03:08 0x69 == i, vendor (intel) Aug 08 21:03:11 how lovely and so simple :'-) Aug 08 21:03:26 intel 0x054117 hmmm Aug 08 21:03:37 just look up the ID ... maybe try 0x6905_4117 Aug 08 21:03:47 I think they state then with underscore in the middle for some reason Aug 08 21:04:09 ARM* Architecture Version 5TE = 0b0000_0101 Aug 08 21:04:16 that would be the 0x05 byte Aug 08 21:05:14 intel arm5te 0x4117 hmmm Aug 08 21:05:48 then 3 bits for core g and another 3 bits for core r Aug 08 21:06:29 final 4 bits are the stepping, off the charts! Aug 08 21:07:09 chart doesnt show stepping for 0b1111 Aug 08 21:07:19 page 72 table 4 Aug 08 21:07:32 pxa27x rev c5 Aug 08 21:08:19 ah, so my pdf was a little too old :) Aug 08 21:08:56 and this one says final 3 bits are stepping, not 4 Aug 08 21:10:20 Marex: is there any chance of damaging the lcd if the controller is mis-configured? Aug 08 21:15:17 naw Aug 08 21:17:46 weird Aug 08 21:17:53 i put in an sd card and now i cant find cocoboot Aug 08 21:18:04 ah, there it is Aug 08 21:18:57 ok, kernel#1 fail Aug 08 21:19:04 instant reboot Aug 08 21:19:41 i guess export CROSS_COMPILE=arm-linux-gnu- isnt the right choice Aug 08 21:20:14 it should be ... what toolchian do you use ? Aug 08 21:20:25 and how big is the zImage ? Aug 08 21:21:21 c2d ~ # crossdev -s1 arm-linux-gnu Aug 08 21:21:30 gentoo script to build everything with 1 command Aug 08 21:22:40 zImage is ... cocoboot reset the phone Aug 08 21:22:55 1778KB i think Aug 08 21:23:14 though the kernel didnt compile until i picked a cpu type from a list *looks* Aug 08 21:23:56 i had to go into Intel PXA2xx/PXA3xx Implementations and pick something Aug 08 21:24:06 CompuLab CM-X255/CM-X270 modules is my current selection Aug 08 21:24:33 PXA based Palm PDAs may be better, on a second look Aug 08 21:25:09 heh, check that and things like Palm Tungsten|C and Palm Treo 680 come out of nowhere! :D Aug 08 21:25:25 well of course Aug 08 21:25:36 disable the compulab stuff Aug 08 21:25:44 i did :) Aug 08 21:25:55 Kconfig always hides what i want, one menu out of sight Aug 08 21:26:26 do i want arm or armeb? Aug 08 21:30:29 armel Aug 08 21:30:38 *looks* Aug 08 21:30:51 certainly not big-endian arm Aug 08 21:31:13 and 'arm' defaults to what? Aug 08 21:31:24 the kernel i tried was neither, just 'arm' Aug 08 21:33:56 armel Aug 08 21:34:18 maybe it was just the palm option then Aug 08 21:34:27 armeb is IXP and you'd have to select that ... the kernel should manage the compiler flags for that on it's own Aug 08 21:34:50 clever: of course, you need to enable your machine so you support the machine ID passed by the bootloader to kernel Aug 08 21:35:16 i'll try with palm enabled on the 'arm' compiler, if that doesnt work i'll give armel a try Aug 08 21:35:56 machineid d052, does the kernel check that and abort if it doesnt match? Aug 08 21:37:21 your compiler should be arm-linux-gnueabi- ... but then, given linux doesn't pull in even the libgcc, you should be fine with almsot any working toolchain Aug 08 21:37:27 clever: yes Aug 08 21:37:42 what's your gcc version and binutils version btw ? Aug 08 21:38:08 yeah, -s1 on crossdev says to only build binutils and gcc, no libc Aug 08 21:38:19 arm-linux-gnu-gcc (Gentoo 4.6.3 p1.5, pie-0.5.2) 4.6.3 Aug 08 21:39:11 -rwxr-xr-x 1 root root 1.8M Aug 8 18:39 zImage Aug 08 21:39:36 ok, that one hard-resets it too Aug 08 21:40:10 c2d ~ # crossdev -s1 armel-linux-gnueabi Aug 08 21:40:35 time to make another toolchain, i think i have 3 cross compilers now, lol Aug 08 21:41:34 clever: is the machine ID even supported in your kernel ? Aug 08 21:41:50 you have treo 700, machID Aug 08 21:42:07 check arch/arm/tools/...something... and grep it for the ID Aug 08 21:42:17 the ID is in decimal in there, so you have to convert the bases Aug 08 21:42:29 s/...something.../mach-types Aug 08 21:42:38 clever@c2d ~/build/kernel $ grep -r d052 -i arch/arm/ Aug 08 21:42:38 clever@c2d ~/build/kernel $ arm Aug 08 21:42:44 no hits under arch/arm Aug 08 21:42:44 decimal! Aug 08 21:42:54 ah Aug 08 21:43:57 d052 ? Aug 08 21:43:59 really ? Aug 08 21:44:07 that's not machine ID Aug 08 21:44:12 that's device ID of the treo! Aug 08 21:44:31 # git grep TREO arch/arm/tools/ Aug 08 21:44:32 arch/arm/tools/mach-types:treo680 MACH_TREO680 TREO680 1230 Aug 08 21:44:43 there's not even treo700 mach ID any longer Aug 08 21:44:45 cocoboot shows machine: H102 on the 650 Aug 08 21:45:09 let me try this kernel on the 650 Aug 08 21:45:20 Treo650 isn't supported either it seems Aug 08 21:45:24 only 680 Aug 08 21:45:29 Sleep_Walker: ^ ? Aug 08 21:45:38 locks the 650 up Aug 08 21:45:53 the hardware differs, you know ... Aug 08 21:45:57 yeah Aug 08 21:46:04 just trying it anyways Aug 08 21:46:09 i dont have any root/initrd, but i was expecting atleast a fbcon Aug 08 21:46:12 and (!) so does the machine ID Aug 08 21:46:24 no way ... it takes some more work to get there Aug 08 21:46:50 the kernel i got from the site got that far atleast Aug 08 21:47:45 # XXX: This is a cut-down version of the file; it contains only machines that Aug 08 21:47:45 # XXX: are in mainline or have been submitted to the machine database within Aug 08 21:47:45 # XXX: the last 12 months. If your entry is missing please email rmk at Aug 08 21:48:23 might have been hacked in some way ... so ... you can pick the sources and forward port it into mainline linux Aug 08 21:48:46 700p is on the list, it links back to http://www.hackndev.com/ Aug 08 21:49:16 https://github.com/hackndev this ? Aug 08 21:49:22 http://www.arm.linux.org.uk/developer/machines/list.php?id=1421 Aug 08 21:50:17 yea, what kernel did boot for you ? Aug 08 21:50:28 the one i got from the blog, *looks* Aug 08 21:50:49 http://blog.mikeasoft.com/2007/06/30/im-in-ur-treo-linuxin-ur-phonez/ Aug 08 21:51:04 'Overwrite the zImage with my version. This is a more up-to-date...' Aug 08 21:51:30 that one had fully working gui, partialy working audio, and touchscreen/keyboard/sd Aug 08 21:51:55 With much help from the chaps in #hackndev I managed to get Linux installed on my Treo 650: Aug 08 21:52:10 so what treo are you trying to get this working on ? Aug 08 21:52:41 the 700p, was expecting the newer model to run a bit faster/more ram Aug 08 21:53:01 faster? no ... Aug 08 21:53:10 more ram? I don't think so, maybe ... Aug 08 21:53:19 so the treo650 kernel booted on it ? Aug 08 21:54:21 i dont believe ive ever had anything work on the 700 Aug 08 21:54:52 ive manualy added http://www.arm.linux.org.uk/developer/machines/list.php?id=1421 to arch/arm/tools/mach-types, i'll look around to see if anything else is needed Aug 08 21:54:55 you just said you did boot a kernel on it Aug 08 21:55:14 that kernel worked on the 650 Aug 08 21:55:26 i have both models Aug 08 21:55:32 you need machine entry ... see arch/arm/mach_pxa/ and look for MACHINE_START() Aug 08 21:55:42 I see... the treo650 should be easier Aug 08 21:55:57 650 was also removed from the listing Aug 08 21:56:02 the 700 ... well, depends on how much time you want to invest Aug 08 21:56:10 yes, because it wasn't supported either Aug 08 21:56:36 just dont know what state the 650 or 700p where in when they got expired Aug 08 21:57:35 brb, lunch Aug 08 22:07:14 id like to try re-adding support first, just to see what state both devices are in Aug 08 22:07:54 from what i can see, it all hangs off CONFIG_TREO700P and CONFIG_TREO680 based stuff Aug 08 22:08:07 but every trace of the 700p is gone? like they just went thru and cleaned the house up Aug 08 22:08:26 may need to check the git logs and revert the change Aug 08 22:10:58 ./include/generated/mach-types.h:# define machine_is_treo680() (machine_arch_type == MACH_TYPE_TREO680) Aug 08 22:11:10 i do see the checks you mentioned Aug 08 22:11:30 ./include/generated/mach-types.h:#define MACH_TYPE_TREO680 1230 Aug 08 22:12:45 clever@c2d ~/build/kernel $ git log -SCONFIG_TREO700P|less Aug 08 22:12:47 * clever goes back to tv Aug 08 22:14:39 clever: well of course ... but not everything was pushed mainline I think Aug 08 22:14:48 and I don't think anyone had t700p in here Aug 08 22:18:42 its in the list on the website, so they must have had it at some point Aug 08 22:18:52 if git log finds nothing, i'll move down to the 650 Aug 08 22:19:12 also, if i just hack the machineid check, how much of the 650 code would maybe work on the 700p? Aug 08 22:19:19 any chance id atleast have lcd? Aug 08 22:20:20 dunno .. .try it Aug 08 22:20:40 and no, adding it to the list of machine IDs doesn't mean there ever was support for the device mainline Aug 08 22:21:07 btw. it's sufficient to adjust the machine id in MACHINE_START() or add one more MACHINE_START() entry Aug 08 22:23:25 but i'm still not sure what the machine id is, the 650 is h102 (258 dec) Aug 08 22:23:32 and i saw on the site that its 1230 Aug 08 22:23:47 wait no, 1230 is the primary key on the site Aug 08 22:23:51 nothing to do with hardware Aug 08 22:24:15 *puzzled* MACH_TYPE_TREO680 is also 1230? Aug 08 22:24:53 typeid 909 is the 650, http://www.arm.linux.org.uk/developer/machines/list.php?id=909 Aug 08 22:25:57 i'll need to read more of the code Aug 08 22:26:52 ok, if CONFIG_MACH_TREO680 then #define machine_is_treo680() (__machine_arch_type == MACH_TYPE_TREO680) Aug 08 22:27:26 __machine_arch_type feels like a compiler shortcut to read the hardware register, so maybe the typeid from that site is the hardware machine id? Aug 08 22:28:19 ./arch/arm/mach-pxa/palmtreo.c:MACHINE_START(TREO680, "Palm Treo 680") Aug 08 22:28:21 ah there it is :) Aug 08 22:31:30 mach ID is passed to the kernel by the bootloader Aug 08 22:31:52 so cocoboot should be finding it and passing it over Aug 08 22:32:05 http://www.arm.linux.org.uk/developer/machines/list.php?id=1421 for treo 700p Aug 08 22:32:14 I guess so, check the source Aug 08 22:32:27 it reports the machine Id on the homescreen, you can even change it Aug 08 22:32:37 1424 == 0x58d Aug 08 22:32:46 that input in the top left corner? Aug 08 22:33:09 the one that says 0 unknown device? Aug 08 22:35:43 that one I guess, yea Aug 08 22:35:51 so ... change that to pass the proper mach ID Aug 08 22:36:07 you can even change that to t680 one ... just to do a quick try Aug 08 22:37:25 i punched in 1230, it rebooted Aug 08 22:37:37 hex? Aug 08 22:38:07 rebooted! Aug 08 22:40:50 you compiled the t680 support in ? Aug 08 22:44:43 i believe the one on the card has t680 enabled with the plain 'arm' toolchain Aug 08 22:44:54 armel is still building the toolchain Aug 08 22:46:22 clever: you tried booting with 1230 ? Aug 08 22:46:32 btw try stripping down the kernel, 1.8meg seems a bit bulky already Aug 08 22:48:13 i'll do that once the next toolchain is ready Aug 08 23:33:11 Marex: any reason why 1.8mb is bulky, or is it just having trouble fitting in 32mb after unzip? Aug 08 23:46:19 naw, it has trouble finding continuous slab of memory in palmos probably Aug 08 23:46:25 I recall such issues Aug 08 23:52:04 ah Aug 08 23:52:12 makes some sense Aug 08 23:52:24 though the palmos malloc routines support ram defrag Aug 08 23:52:37 and half my tests are directly after a fresh reboot (due to the crash) Aug 08 23:53:11 -rw-r--r-- 1 clever clever 1.6M Aug 8 20:36 net/built-in.o Aug 08 23:53:20 bulk of the fat is in net, and i dont have any network working Aug 08 23:53:36 *nuke!* Aug 08 23:54:29 wireless off, no hardware in the device at all, zero loss Aug 08 23:55:04 -rw-r--r-- 1 clever clever 1.6M Aug 8 20:54 net/built-in.o Aug 08 23:55:06 :S Aug 08 23:57:13 -rw-r--r-- 1 clever clever 1.3M Aug 8 20:56 net/built-in.o Aug 08 23:58:05 -rwxr-xr-x 1 clever clever 1.5M Aug 8 20:57 arch/arm/boot/zImage Aug 09 00:00:25 forcing machineid 1230 doesnt help, reboots every time Aug 09 00:01:49 and i have a confirmation now Aug 09 00:01:54 http://www.arm.linux.org.uk/developer/machines/list.php?id=909 = t650 Aug 09 00:02:04 cocoboot shows 909 in that box :) Aug 09 00:02:27 1230 on the 650 device causes the lcd to glitch out Aug 09 00:03:53 leaving 909 on the 650 causes a plain lockup Aug 09 00:08:53 maybe the LCD is misconfigured ... but if it glitches Aug 09 00:08:56 hm Aug 09 00:11:43 the odd lines faded to white Aug 09 00:11:45 and then the even ones Aug 09 00:11:57 not sure, but maybe the lcd controller locked up and stopped scanning? Aug 09 00:12:22 you know how an lcd must be driven with ac? Aug 09 00:13:35 lcd registers do work on the 650, *makes notes* Aug 09 00:15:25 http://pastebin.com/zK72NEk7 Aug 09 00:15:52 so i can trick the t680 kernel into working on the t650 Aug 09 00:15:56 but it mis-configures the lcd Aug 09 00:16:27 08 21:03:52 < clever> leaving 909 on the 650 causes a plain lockup Aug 09 00:16:39 maybe the kernel knows its on the wrong device, and doesnt do anything? Aug 09 00:16:56 and in the other cases, fatal error, uncaught exception kind of thing Aug 09 00:19:10 Marex: i fudged the machineid, i told linux that 909 is the t680 Aug 09 00:19:22 now i dont need to fudge things in cocoboot, the lcd just glitches on defaults Aug 09 00:20:47 linux decides the device it runs on solely on machineID Aug 09 00:21:01 you need to pass machineID that the kernel supports Aug 09 00:21:08 mainline only supports t680 ... Aug 09 00:21:52 you can try the ancient HnD kernel too, it supports T650 Aug 09 00:22:04 i know, its just a temp fix so i can modify t680 support Aug 09 00:22:09 anything can be different on the t700p really ... you'd have to research it Aug 09 00:22:13 i'm using the android branch of the kernel, since android needs some special kernel stuff Aug 09 00:22:25 but i'll try the hnd branch too, maybe it will get me somehting Aug 09 00:22:29 well, that might also break things Aug 09 00:22:36 (android goo) Aug 09 00:22:39 yucky Aug 09 00:22:44 anyway ... gnight Aug 09 00:23:50 *reads lcd driver* Aug 09 00:31:26 *lost* Aug 09 00:36:48 https://github.com/hackndev/linux-hnd/tree/master/arch/arm/mach-pxa/palmt650 Aug 09 00:36:57 looks like some of the 650 support still exists, i'll give this a spin Aug 09 01:15:36 Marex: perfect, that branch worked on the first try Aug 09 01:15:50 i now have dmesg output complaining that it cant find root/init :) **** ENDING LOGGING AT Thu Aug 09 02:59:59 2012