Wolfgang Denk wrote:

In message 43ED3544.3040007@quartics.com you wrote:

...

I am using latest u-boot with an ARM Integrator/CP board (16MB flash). Everything is fine except that when I save the environment in the flash and reboot the system, u-boot restarts with the environment variables it was compiled with, not what has been saved in the flash sector. I used

...

...

Any clues anyone??

Sounds as if your system was misconfigured.

Best regards,

Wolfgang Denk

Yes, indeed. What source file/configuration define do I need to look at to see the cause of this problem??

Regards, Umar

Wolfgang Denk wrote:

In message 43ED4459.3060007@quartics.com you wrote:

...

...

Sounds as if your system was misconfigured.

...

...

Yes, indeed. What source file/configuration define do I need to look at to see the cause of this problem??

In the first step: your board configuration file(s), i. e. include/config/<board>.h and board/<board>/config.mk

Best regards,

Wolfgang Denk

Hi,

I have made some progress on this issue. I thought that fully describing the scenario might help in deciphering the solution.

I am working without the ARM JTAG tools. In order to get U-Boot up and flashed, I had to rely on the ARM bootPROM monitor that comes built-in in the Integrator/CP board. So to get U-Boot flashed, I take the U-Boot binary and run the following command on it:

arm-linux-objcopy --change-address 0x24000000 -I binary -O srec u-boot.bin my-u-boot.srec

The command above changes the base address fro 0x1000000 (SDRAM) to 0x24000000 (first sector of the flash chip). I send this srec file to the monitor which flashes the file to the flash chip. This allows me to flip a switch on the board and the ARM bootPROM monitor to run the image at location 0x24000000.

The problem I am encountering is that U-Boot does not load the saved environment variables from flash but rather uses the ones that have been compiled-in with the code. I found out this is happening because of the following code in include/environment.h:

# if (CFG_ENV_ADDR >= CFG_MONITOR_BASE) && \ (CFG_ENV_ADDR+CFG_ENV_SIZE) <= (CFG_MONITOR_BASE + CFG_MONITOR_LEN) # define ENV_IS_EMBEDDED 1 # endif

So, in common/env_flash.c, I put the following at the beginning:

#undef ENV_IS_EMBEDDED

This changes the environment address to 0x24f80000, which is the correct address. However, when I print the environment using the "printenv" command I get:

Integrator-CP # printenv bo´ cm´ cp´ x2´ 80´ 0 ´ 7f´ 0´ 00´ 0;´ oo´

An "md" command on that address reveals the following memory contents:

Integrator-CP # md 0x24f80000 24f80000: 00b4d8f9 00b46f62 00b46d63 00b47063 ....bo..cm..cp.. 24f80010: 00b43278 00b43038 00b42030 00b46637 x2..80..0 ..7f.. 24f80020: 00b43020 00b43030 00b43b30 00b46f6f 0..00..0;..oo.. 24f80030: 00b46200 00b46474 00b47961 00b46200 .b..td..ay...b.. 24f80040: 00b47264 00b43d65 00b43034 00b47473 dr..e=..40..st.. 24f80050: 00b43d6e 00b46972 00b47300 00b4756f n=..ri...s..ou.. 24f80060: 00b46573 00b46c61 00b46474 00b43d72 se..al..td..r=.. 24f80070: 00b46972 00b47600 00b46669 00b4006e ri...v..if..n... 24f80080: 00b4746f 00b47367 00b46f6f 00b4642f ot..gs..oo../d.. 24f80090: 00b46d2f 00b46c62 00b4306b 00b42077 /m..bl..k0..w .. 24f800a0: 00b4746f 00b47974 00b46a3d 00b43273 ot..ty..=j..s2.. 24f800b0: 00b46474 00b47472 00b47261 00b4616c td..rt..ar..la.. 24f800c0: 00b4302e 00b46178 00b43030 00b47830 .0..xa..00..0x.. 24f800d0: 00b43030 00b46a28 00b43273 00b46f63 00..(j..s2..co.. 24f800e0: 00b46c6f 00b47474 00b4414d 00b47069 ol..tt..MA..ip.. 24f800f0: 00b46e6f 00b40000 00b40000 00b40000 on..............

However, I know that this data is not being read correctly because if I define ENV_IS_EMBEDDED again, recompile, and reload then I get:

Integrator-CP # md 0x24f80000 24f80000: 3eb0d8f9 746f6f62 3d646d63 30207063 ...>bootcmd=cp 0 24f80010: 30343278 30303038 78302030 30636637 x24080000 0x7fc0 24f80020: 31783020 30303030 62203b30 6d746f6f 0x100000; bootm 24f80030: 6f6f6200 6c656474 323d7961 75616200 .bootdelay=2.bau 24f80040: 74617264 38333d65 00303034 69647473 drate=38400.stdi 24f80050: 65733d6e 6c616972 64747300 3d74756f n=serial.stdout= 24f80060: 69726573 73006c61 72656474 65733d72 serial.stderr=se 24f80070: 6c616972 72657600 3d796669 6f62006e rial.verify=n.bo 24f80080: 7261746f 723d7367 3d746f6f 7665642f otargs=root=/dev 24f80090: 64746d2f 636f6c62 7220306b 6f722077 /mtdblock0 rw ro 24f800a0: 7366746f 65707974 66666a3d 6d203273 otfstype=jffs2 m 24f800b0: 61706474 3d737472 666d7261 6873616c tdparts=armflash 24f800c0: 303a302e 30306178 40303030 63327830 .0:0xa00000@0x2c 24f800d0: 30303030 66666a28 20293273 736e6f63 0000(jffs2) cons 24f800e0: 3d656c6f 41797474 2030414d 6e3d7069 ole=ttyAMA0 ip=n 24f800f0: 00656e6f 00000000 00000000 00000000 one.............

So the actual data is still stored in flash in perfect condition but defining ENV_IS_EMBEDDED causes U-Boot to start eating the data when it reads it.

Any help on why this is happening is appreciated...

Regards, Umar