**** BEGIN LOGGING AT Sun Dec 13 02:59:57 2009 Dec 13 03:44:14 that'd be cool if you could power the sheeva over usb Dec 13 03:58:22 it's not quite that power-efficient :) Dec 13 04:00:06 :( Dec 13 04:00:12 oh well. Dec 13 10:56:17 Hi, I managed to break my plug using a dodgy kernel build (I built it myself and forgot to include ubifs, which is not part of th kirkwood kernel defaults). Anyhows, I am trying to restore my plug from the uboot prompt, loading a new kernel from USB drive, but having problems... MY problem is explained here: http://plugcomputer.org/plugforum/index.php?PHPSESSID=dc9a3efb5bd9c51a1d49e0ab4e9f4974&topic=1073.msg6583#new Dec 13 11:02:15 if you've got a kernel on the fat drive, you can boot straight off that particularly kernel Dec 13 11:03:15 should I have loaded into a different address? nand write.e 0x00800000 0x00100000 0x00400000 Dec 13 11:03:16 oh, never mind what i said Dec 13 11:04:05 after loading it and verifying into fatload usb 0 0x00800000 /uImage Dec 13 11:05:39 those offsets look right... Dec 13 11:06:40 and the ones in the post wrong? Will I have written over my rootfs with the numbers I was given? Dec 13 11:07:46 no, the ones you've got listed just above look right to me Dec 13 11:08:20 i'd probably boot off a workign kernel (from usb ie) and backup my plug first though just to be safe Dec 13 11:09:23 oh, i c what u mean... the addresses depend on where you did the fatload into Dec 13 11:10:48 fatload usb 0 address <--- nand write.e address Dec 13 11:19:39 how would I go about booting from the kernel on USB then? Dec 13 11:19:54 (tried those different addresses, and getting the same error Dec 13 11:21:12 well, if you've done a fatload into 0x00800000, then its simple a case of "bootm 0x00800000" Dec 13 11:21:42 ahh, simple as that. will it know where the rootfs is? Dec 13 11:21:47 i.e. on the nand? Dec 13 11:21:56 asuming the bootargs are set correctly, but i've you'd played with the environment vars in the uboot that shouldnt be a problem Dec 13 11:23:25 do a printenv and there should be a var called bootargs which tells the kernel where things liek rootfs are Dec 13 11:23:56 what i meant to say is "unless you'd played with the environment vars int he uboot..." Dec 13 11:28:00 I haven't touched the enviroment, it's working fine. However I couldn't boot, it just came up Starting Kernel and hung Dec 13 11:28:39 the error message does imply that erasing the nand will mark any bad blocks, is this something I might want to do? Dec 13 11:29:03 Failed write verify, nand offset 0x00100000. Suggest to erase block and repeat write operation! Dec 13 11:39:55 to be honest, im not 100% sure, i've only had mine a week... but if your getting "starting kernel" and nothing else its probably because bootargs isnt set... hit the reset button and go back to uboot... Dec 13 11:40:16 and then do a printenv bootcmd Dec 13 11:51:38 hmm, even though the verify was failing, the system now boots.... Dec 13 11:52:01 so it was writing a good kernel image, it just didn't realise it. Dec 13 11:53:08 good stuff... altho, it may be because there was a kernel loaded from the fatload... you might want to power it off completely and power it back up again, just to make sure it is reading the kernel from nand Dec 13 11:56:20 now I am in, I have written a kernel using the regular tools ( nandwrite and flash_eraseall ). I have also managed to get a full backup done too.. Should have done that before I started really :-) Dec 13 11:57:13 nice one :) Dec 13 12:16:36 yep, now working just fine, I have the ocf-linux kernel patch in there, so I have a /dev/crypto to support the new hardware crypto Dec 13 12:16:41 many thanks Dec 13 15:53:30 tbm: http://plugcomputer.org/plugforum/index.php?topic=997.msg6592#msg6592 Dec 13 15:55:46 yeah, thx **** ENDING LOGGING AT Mon Dec 14 02:59:56 2009