I believe there is a Uboot command that will give you a list. Yup :
DM355 EVM # nand bad
Device 0 bad blocks:
00520000
00b60000
00c20000
01620000
01ea0000
020a0000
02360000
(plus more)
In general if you are updating an EVM then you shouldn't have to worry
about bad blocks in the kernel space. I saw this problem on our own
boards when we had a few NAND that had bad blocks in very bad spots.
I use devrocket to create a new filesystem - the platform image "mode"
allows you to pick and choose what packages to use. Otherwise you will
have to build it manually from either the NFS filesystem or the minimal
one (that is in NAND already). This is actually the only thing that I
use devrocket for.
Once you have a filesystem, you need to tar it up, and get it on the
board. I generally put it on an SD card and mount the card. Then, you
need to erase the filesystem that isn't being used - and by this I mean
that the NAND flash is partitioned into 5 sections:
root@(none):~# cat /proc/mtd
dev: size erasesize name
mtd0: 003c0000 00020000 "bootloader"
mtd1: 00040000 00020000 "params"
mtd2: 00400000 00020000 "kernel"
mtd3: 20000000 00020000 "filesystem1"
mtd4: 1f800000 00020000 "filesystem2"
You will be booting (and running) either filesystem1 or 2. Erase the
alternate one with 'flash_eraseall /dev/mtd2'. Then mount it. I use
YAFFS2 but JFFS2 works as well.
mount -t yaffs2 /dev/mtdblock4 /mnt/nand
Then your NAND partition is just like any other mounted storage. Untar
your new filesystem to the new mount point (/mnt/nand).
Then when you reboot, update your bootargs like so:
bootargs=root=/dev/mtdblock4 rootfstype=yaffs2 console=ttyS0,115200n8
mem=100M
I usually try to make sure that I always have a good working filesystem
in one of the partitions so I can recover the board if the new
filesystem is broken.
Good luck!
Steve
Albert Burbea wrote:
Thanks Stephen,
now, how do you get the "bad block list" BEFORE flashing the kernel?
Also, how do you build a new filesystem, and flash it on the NAND?
Thanks
Albert
On 8/8/08, *Stephen Berry* <[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>> wrote:
This would seem to be a pretty common thing to do, but for some
reason it wasn't described anywhere that I could find...
If you want to be able to update your kernel or uboot image in
NAND (I'm using a DM355 platform, but this should work for others)
*while* your system is running and without the use of a JTAG
device, these are the commands that I use to do this:
/usr/sbin/flash_eraseall -j /dev/mtd2
/usr/sbin/nandwrite -p /dev/mtd2 uImage
This example updates the kernel. It is assumed that /dev/mtd2 is
your NAND kernel partition. Uboot would be /dev/mtd0.
Also if you see any bad blocks during the erase operation you'll
need to make sure they don't occur in the middle of your kernel
code. It seems that the bootloader doesn't understand bad blocks
and will NOT skip them. So you need to have a contiguous set of
good blocks for your kernel.
I've been able to save some bad boards in the past by moving the
kernel away from the bad block area and changing the uboot 'nboot'
command to match that offset.
Enjoy,
Steve
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
<mailto:[email protected]>
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
--
Albert Burbea
Harishonim 8
Ramat Gan 52502, Israel
Tel/Fax + 972-3-7526016
Mobile: +972-52-3541842
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
