Difference between revisions of "BeagleBoardNAND"
(→Hardware: BeagleBoard RevC5 memory increase) |
(→U-Boot: Note these instructions only apply to v2011.12 or OLDER) |
||
Line 85: | Line 85: | ||
Note: The command ''nandecc hw'' is essential here! X-Loader is started by OMAP3 boot ROM. This uses HW ECC while reading the NAND, so while writing, we have to use OMAP3 HW ECC, too. If you don't use HW ECC boot ROM [[BeagleBoardRecovery|can't boot from NAND]] any more. | Note: The command ''nandecc hw'' is essential here! X-Loader is started by OMAP3 boot ROM. This uses HW ECC while reading the NAND, so while writing, we have to use OMAP3 HW ECC, too. If you don't use HW ECC boot ROM [[BeagleBoardRecovery|can't boot from NAND]] any more. | ||
− | ==U-Boot== | + | ==U-Boot prior to v2011.12== |
[http://elinux.org/BeagleBoard#U-Boot Build] or download [http://code.google.com/p/beagleboard/downloads/list binary] (flash-uboot.bin) U-Boot. Put it at first (boot) FAT partition of MMC/SD card and boot from card. Then start boot from card and use the following to write U-Boot to NAND: | [http://elinux.org/BeagleBoard#U-Boot Build] or download [http://code.google.com/p/beagleboard/downloads/list binary] (flash-uboot.bin) U-Boot. Put it at first (boot) FAT partition of MMC/SD card and boot from card. Then start boot from card and use the following to write U-Boot to NAND: |
Revision as of 06:59, 27 January 2014
This page is about using (booting/running from) NAND memory on BeagleBoard. Parts of this page are inspired by Steve's flash procedure document.
Contents
Hardware
OMAP3530 has 256MB NAND flash in PoP (PoP: Package-On-Package implementation for Memory Stacking) configuration.
- BeagleBoard has 256MB of Micron NAND (512MB as of Rev C5)
- EVM 128MB of Samsung OneNAND or 128MB of Micron NAND
- Zoom MDK also uses the Micron NAND
Software
The following software parts can be stored and booted/run from NAND:
- X-Loader
- U-Boot (+ environment/configuration data)
- Linux kernel
- Linux file system
The memory partitioning for this as used on BeagleBoard:
0x00000000-0x00080000 : "X-Loader" 0x00080000-0x00260000 : "U-Boot" 0x00260000-0x00280000 : "U-Boot Env" 0x00280000-0x00680000 : "Kernel" 0x00680000-0x10000000 : "File System"
To be able to write something to (empty) NAND, you first need to boot from another source, e.g. MMC/SD card boot. In addition to the files you need for MMC/SD card boot (MLO & U-Boot), put the files you want to flash into the first FAT partition of MMC/SD card. Then you can read them from there and write them to NAND.
Note: If something goes wrong writing the initial X-Loader, your board might not boot any more without pressing the user button. See BeagleBoard recovery article how to fix this.
X-Loader
Build or download binary (x-load_revc_v3.bin.ift) X-Loader. Put it at the first (boot) FAT partition of MMC/SD card and boot from the card. Then start boot from card and use the following to write X-Loader to NAND:
...40T....... Texas Instruments X-Loader 1.41 Starting on with MMC Reading boot sector 147424 Bytes Read from MMC Starting OS Bootloader from MMC... U-Boot 1.3.3-00411-g76fe13c-dirty (Jul 12 2008 - 17:12:05) OMAP3530-GP rev 2, CPU-OPP2 L3-165MHz OMAP3 Beagle Board + LPDDR/NAND DRAM: 128 MB NAND: 256 MiB In: serial Out: serial Err: serial Hit any key to stop autoboot: 0 OMAP3 beagleboard.org # mmc init OMAP3 beagleboard.org # fatload mmc 0:1 80000000 x-load_revc_v3.bin.ift reading x-load.bin.ift_for_NAND 9808 bytes read OMAP3 beagleboard.org # nand unlock (Note - if you are using a more recent version of U-Boot, nand unlock is missing simply skip this step. i.e. U-boot 2009.08) device 0 whole chip nand_unlock: start: 00000000, length: 268435456! NAND flash successfully unlocked OMAP3 beagleboard.org # nandecc hw OMAP3 beagleboard.org # nand erase 0 80000 NAND erase: device 0 offset 0x0, size 0x80000 Erasing at 0x60000 -- 100% complete. OK OMAP3 beagleboard.org # nand write 80000000 0 80000 NAND write: device 0 offset 0x0, size 0x80000 524288 bytes written: OK OMAP3 beagleboard.org #
Note: The command nandecc hw is essential here! X-Loader is started by OMAP3 boot ROM. This uses HW ECC while reading the NAND, so while writing, we have to use OMAP3 HW ECC, too. If you don't use HW ECC boot ROM can't boot from NAND any more.
U-Boot prior to v2011.12
Build or download binary (flash-uboot.bin) U-Boot. Put it at first (boot) FAT partition of MMC/SD card and boot from card. Then start boot from card and use the following to write U-Boot to NAND:
OMAP3 beagleboard.org # mmc init OMAP3 beagleboard.org # fatload mmc 0:1 80000000 u-boot.bin reading u-boot.bin 147424 bytes read OMAP3 beagleboard.org # nand unlock (Note - if you are using a more recent version of U-Boot, nand unlock is missing simply skip this step. i.e. U-boot 2009.08) device 0 whole chip nand_unlock: start: 00000000, length: 268435456! NAND flash successfully unlocked OMAP3 beagleboard.org # nandecc sw OMAP3 beagleboard.org # nand erase 80000 160000 NAND erase: device 0 offset 0x80000, size 0x160000 Erasing at 0x1c0000 -- 100% complete. OK OMAP3 beagleboard.org # nand write 80000000 80000 160000 NAND write: device 0 offset 0x80000, size 0x160000 1441792 bytes written: OK OMAP3 beagleboard.org #
Note: You can use the same u-boot.bin you use to boot from MMC/SD card for NAND. There are no differences between U-Boot used for MMC/SD card boot and NAND boot.
Note: Here, you don't need the nandecc hw option. X-Loader which loads & starts U-Boot is able to understand SW ECC written by U-Boot.
Kernel
While X-Loader and U-Boot can be written only by U-Boot, for kernel and file system there are two ways to write them to NAND: Either by U-Boot (similar way as for X-Loader and U-Boot above) or from running kernel (e.g. booted from MMC card).
Note: X-Loader and U-Boot can't be written from already running kernel, too, because from kernel point of view X-loader and U-Boot NAND partitions are marked as write only. See omap3beagle_nand_partitions[] configuration structure in kernel's arch/arm/mach-omap2 directory.
Writing kernel with U-Boot
OMAP3 beagleboard.org # mmc init OMAP3 beagleboard.org # fatload mmc 0:1 80000000 uImage reading uImage OMAP3 beagleboard.org # nandecc sw OMAP3 beagleboard.org # nand erase 280000 400000 NAND erase: device 0 offset 0x280000, size 0x400000 Erasing at 0x660000 -- 100% complete. OK OMAP3 beagleboard.org # nand write 80000000 280000 400000 NAND write: device 0 offset 0x280000, size 0x400000 4194304 bytes written: OK OMAP3 beagleboard.org #
Once you did this, use U-Boot commands to boot kernel (uImage) from NAND:
nand read 80000000 280000 400000 ; bootm 80000000
These, you can e.g. store as bootcmd and your board will automagically boot uImage from NAND.
Writing kernel with kernel
Once you have a kernel booted, e.g. from MMC card, you can use it to write the kernel itself (uImage) to NAND and then switch from MMC boot to kernel NAND boot. For this, observe the kernel's boot messages. These should have something like:
... omap2-nand driver initializing NAND device: Manufacturer ID: 0x2c, Chip ID: 0xba (Micron NAND 256MiB 1,8V 16-bit) cmdlinepart partition parsing not available Creating 5 MTD partitions on "omap2-nand": 0x00000000-0x00080000 : "X-Loader" 0x00080000-0x00260000 : "U-Boot" 0x00260000-0x00280000 : "U-Boot Env" 0x00280000-0x00680000 : "Kernel" 0x00680000-0x10000000 : "File System" ...
At kernel's prompt command cat /proc/mtd will give you similar output:
root@beagleboard:~# cat /proc/mtd dev: size erasesize name mtd0: 00080000 00020000 "X-Loader" mtd1: 001e0000 00020000 "U-Boot" mtd2: 00020000 00020000 "U-Boot Env" mtd3: 00400000 00020000 "Kernel" mtd4: 0f980000 00020000 "File System"
While the first three partitions (X-Loader, U-Boot and U-Boot Env) are read only from kernel point of view, Kernel and File System partition can be written by the kernel itself. To do this, you need MTD User modules in your kernel's root file system.
In this example we mount boot (FAT) partition of MMC card (using a dual boot card) to read kernel image (uImage) from. If you have network connection in your kernel, you can use this, too. Or you put uImage in your root file system. Goal is to have access to uImage from running kernel to be able to write it to NAND.
root@beagleboard:~# mkdir -p /mnt/fat root@beagleboard:~# mount /dev/mmcblk0p1 /mnt/fat/ root@beagleboard:~# ls -la /mnt/fat -rwxr-xr-x 1 root root 16740 Jul 7 17:28 mlo -rwxr-xr-x 1 root root 717116 Jul 24 2008 u-boot.bin -rwxr-xr-x 1 root root 2106940 Jul 26 2008 uImage root@beagleboard:~# cp /mnt/fat/uImage . root@beagleboard:~# ls -la -rwxr-xr-x 1 root root 2106940 Jul 22 00:30 uImage root@beagleboard:~# flash_eraseall /dev/mtd3 Erasing 128 Kibyte @ 3e0000 -- 96 % complete. root@beagleboard:~# nandwrite -p /dev/mtd3 uImage Writing data to block 0 at offset 0x0 <...> Writing data to block 20 at offset 0x280000 root@beagleboard:~#
File system
As with kernel, while X-Loader and U-Boot can be written only by U-Boot, for file system there are two ways to write them to NAND: Either by U-Boot (similar way as for X-Loader and U-Boot above) or from running kernel (e.g. booted from MMC card). A lot of users report they have issues with writing (root) file system with U-Boot. Main issue is that U-Boot has to write file system exactly in format kernel expects. If there are minor incompatibilities, kernel will later not be able to read file system written by U-Boot.
So, while we document here how to write file system with U-Boot, recommended way is to write (root) file system by kernel itself. With this, it is ensured that kernel writes a file system it will later be able to read.
Writing file system with U-Boot
This way is not recommended. See above.
OMAP3 beagleboard.org # mmc init OMAP3 beagleboard.org # fatload mmc 0:1 80000000 rootfs.jffs2 reading rootfs.jffs2 12976128 bytes read OMAP3 beagleboard.org # nand unlock device 0 whole chip nand_unlock: start: 00000000, length: 268435456! NAND flash successfully unlocked OMAP3 beagleboard.org # nandecc sw OMAP3 beagleboard.org # nand erase 680000 F980000 NAND erase: device 0 offset 0x680000, size 0xf980000 Erasing at 0xffe0000 -- 100% complete. OK OMAP3 beagleboard.org # nand write.jffs2 80000000 680000 ${file_size} NAND write: device 0 offset 0x680000, size 0xc60000 Writing data at 0x12df800 -- 100% complete. 12976128 bytes written: OK OMAP3 beagleboard.org #
For more info see Steve's flash procedure.
Writing file system with kernel
This is the recommended way. See above.
First, we boot kernel with (root) file system on SD card, write (root) file system using file system image at SD card to Beagle's NAND with running kernel. After this is done, we switch kernel's boot arguments to take root file system from NAND instead of SD card, then.
To be able to manipulate/erase/write NAND from kernel's user space, we need MTD Utils (e.g. flash_eraseall).
If you don't have them already, you can get them
- using the angstrom demo, you can install via
opkg install mtd-utils
- cross compiling them your self (good cross compile exercise)
For file system in Beagle's NAND, we use JFFS2. As JFFS2 is part of the standard git kernel, only thing is to configure kernel to be able to use JFFS2 is to enable in make menuconfig (check if already enabled):
CONFIG_JFFS2_FS=y CONFIG_JFFS2_FS_DEBUG=0 CONFIG_JFFS2_FS_WRITEBUFFER=y CONFIG_JFFS2_ZLIB=y CONFIG_JFFS2_RTIME=y
Having kernel supporting JFFS2 and MTD Utils, we now first erase file system partition and create JFFS2 into it:
root@beagleboard:~# cat /proc/mtd dev: size erasesize name mtd0: 00080000 00020000 "X-Loader" mtd1: 001e0000 00020000 "U-Boot" mtd2: 00020000 00020000 "U-Boot Env" mtd3: 00400000 00020000 "Kernel" mtd4: 0f980000 00020000 "File System" root@beagleboard:~# flash_eraseall -j /dev/mtd4 Erasing 128 Kibyte @ f960000 -- 99 % complete. Cleanmarker written at f960000.
Then, we can mount "File system" partition:
root@beagleboard:~# cd /mnt root@beagleboard:~# mkdir nand root@beagleboard:~# mount -t jffs2 /dev/mtdblock4 /mnt/nand
and extract the root file system image to it:
root@beagleboard:~# cd nand root@beagleboard:~# tar xfz <where_ever_your_root_fs_image_is_at_sd_card>/rootfs.tar.gz . ... wait ... root@beagleboard:~# cd .. root@beagleboard:~# sync root@beagleboard:~# umount nand
Now, you should reboot your board and edit bootargs in U-Boot to configure root fs in NAND:
... root=/dev/mtdblock4 rootfstype=jffs2 ...