Skip to contentSkip to conversation

Can I read env from RAM in uboot script?

U-Boot John Stile 4 participants 10 messages Jan 26, 2013 – Jan 30, 2013 Arm/Microchip (AT91) SoC support

lore
Share & tools
lore

John Stile

author
#0 lore raw

Is it possible to have uboot read it's environment from a RAM address, rather than NAND?
OR
Can uboot's scripting support load variables from a RAM address?

My NAND layout has redundant halves:
0
uboot
ubootenv
kernel
fs
128
uboot
ubootenv
kernel
fs
256

My firmware update strategy will update the non-booted side of NAND.

It is easy to hack at91bootstrap, to load a uboots env area at some RAM address, just as it loads uboot at JUMP_ADDR, but how do I get uboot to use this preloaded uboot-env?

Since uboot takes its environment area address at compile time, I wonder if uboot could be made to read it from a RAM address (written there by at91bootstrap), durring the env_relocate_spec()?

So far I have traced:
cpu/arm926ejs/start.S calls start_armboot() lib_arm/board.c start_armboot() calls env_relocate() ./common/env_common.c env_relocate() calls env_relocate_spec() My u-boot.map indicates my env_relocate_spec() comes from env_nand.o ./common/env_nand.c has env_relocate_spec() has ifdefs for ENV_IS_EMBEDDED, but my config is not. and CFG_ENV_IS_IN_NAND which I am currently configured for.

So where is a good point of attack, or is there one?

#1 lore raw

Dear John Stile,

Quoted 4 lines · In message <1359165410.7974.114.camel@genx> you wrote:

Is it possible to have uboot read it's environment from a RAM address, rather than NAND?
OR
Can uboot's scripting support load variables from a RAM address?

Yes, all this can be done. And easily. See for example the "env import" command.

Best regards,

Wolfgang Denk

Signature · 7 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Software suppliers are trying to make their  software  packages  more
``user-friendly''.  .  .  .  Their best approach, so far, has been to
take all the old brochures, and stamp the words, ``user-friendly'' on
the cover.                                               - Bill Gates
author
#2 lore raw
Quoted 11 lines · Wolfgang Denk

Dear John Stile,

Quoted 4 lines · In message <1359165410.7974.114.camel@genx> you wrote:

Is it possible to have uboot read it's environment from a RAM address, rather than NAND?
OR
Can uboot's scripting support load variables from a RAM address?

Yes, all this can be done. And easily. See for example the "env import" command.

That is a great idea.
What version of uboot introduced 'env import'?
I'm stuck on u-boot-1.3.4.
Is there a patch?

#3 lore raw

Hello John,

Quoted 8 lines · John Stile
Quoted 7 lines · Wolfgang Denk

Dear John Stile,

Quoted 4 lines · In message <1359165410.7974.114.camel@genx> you wrote:

Is it possible to have uboot read it's environment from a RAM address, rather than NAND?
OR
Can uboot's scripting support load variables from a RAM address?

Quoted 8 lines
Quoted 3 lines

Yes, all this can be done. And easily. See for example the "env import" command.

That is a great idea.
What version of uboot introduced 'env import'?
I'm stuck on u-boot-1.3.4.
Is there a patch?

search the mailinglist, this is asking for a visit to
http://www.uboat.net/special/archiv/ or a response the like. Try to upgrade if possible...

Regards,
Jeroen

#4 lore raw
Quoted 13 lines · John Stile
Quoted 11 lines · Wolfgang Denk
Quoted 10 lines

Dear John Stile,

Quoted 4 lines · In message <1359165410.7974.114.camel@genx> you wrote:

Is it possible to have uboot read it's environment from a RAM address, rather than NAND?
OR
Can uboot's scripting support load variables from a RAM address?

Yes, all this can be done. And easily. See for example the "env import" command.

Quoted 8 lines
Quoted 5 lines

That is a great idea.
What version of uboot introduced 'env import'?
I'm stuck on u-boot-1.3.4.
Is there a patch?

search the mailinglist, this is asking for a visit to
http://www.uboat.net/special/archiv/ or a response the like. Try to upgrade if possible...

for completeness sake, Wolfgang will likely tell you it is museum material, since it is almost 5 years old. No idea what the site is about, just popped up with "u-boot museum" instead of Wolfgang's replies.

Regards,
Jeroen

author
#5 lore raw
Quoted 17 lines · Jeroen Hofstee
Quoted 17 lines · John Stile
Quoted 16 lines · Wolfgang Denk
Quoted 10 lines

Dear John Stile,

Quoted 4 lines · In message <1359165410.7974.114.camel@genx> you wrote:

Is it possible to have uboot read it's environment from a RAM address, rather than NAND?
OR
Can uboot's scripting support load variables from a RAM address?

Yes, all this can be done. And easily. See for example the "env import" command.

That is a great idea.
What version of uboot introduced 'env import'?
I'm stuck on u-boot-1.3.4.
Is there a patch?

Quoted 8 lines
Quoted 3 lines

search the mailinglist, this is asking for a visit to
http://www.uboat.net/special/archiv/ or a response the like. Try to upgrade if possible...

for completeness sake, Wolfgang will likely tell you it is museum material, since it is almost 5 years old. No idea what the site is about, just popped up with "u-boot museum" instead of Wolfgang's replies.

I saw that.

I tried to update u-boot-1.3.4 to u-boot-2013.

I changed at91bootstrap's JUMP_ADDR from 0x23F00000 to 0x21F00000, as advised by previous posts.

Now I am getting the wrong board type error from the kernel. my board is an AT91SAM9G20EK_2MMC

I am having a hard time setting:
#define MACH_TYPE_AT91SAM9G20EK_2MMC 2288
insted of
#define MACH_TYPE_AT91SAM9G20EK 1624

In had to change buildroot's .config from:
BR2_TARGET_UBOOT_BOARDNAME="at91sam9g20ek_2mmc_nandflash" to BR2_TARGET_UBOOT_BOARDNAME="at91sam9g20ek_nandflash" whihch I suspect is the problem.

I'm trying to change the buildroots output/build/uboot-custom manually, and rerun the build, but no luck yet.

My kernel is built with CONFIG_MACH_AT91SAM9G20EK_2MMC=y

So how do I force u-boot to agree?

#6 lore raw

Hi John Stile,

On 01/29/2013 08:25 AM, John Stile wrote:
[snip]

Quoted 21 lines

I saw that.

I tried to update u-boot-1.3.4 to u-boot-2013.

I changed at91bootstrap's JUMP_ADDR from 0x23F00000 to 0x21F00000, as advised by previous posts.

Now I am getting the wrong board type error from the kernel. my board is an AT91SAM9G20EK_2MMC

I am having a hard time setting:
#define MACH_TYPE_AT91SAM9G20EK_2MMC 2288
insted of
#define MACH_TYPE_AT91SAM9G20EK 1624

In had to change buildroot's .config from:
BR2_TARGET_UBOOT_BOARDNAME="at91sam9g20ek_2mmc_nandflash" to BR2_TARGET_UBOOT_BOARDNAME="at91sam9g20ek_nandflash" whihch I suspect is the problem.

Quoted 6 lines

I'm trying to change the buildroots output/build/uboot-custom manually, and rerun the build, but no luck yet.

My kernel is built with CONFIG_MACH_AT91SAM9G20EK_2MMC=y

So how do I force u-boot to agree?

Please have test with the following patch (add at91sam9g20ek 2mmc nandflash boot support) and check whether it will solve you problem.

You should use at91sam9g20ek_2mmc_nandflash to configure the u-boot

------------------------------------------------

(quoted diff) +0 -0

b/board/atmel/at91sam9260ek/at91sam9260ek.c
index 2555672..ce475df 100644

board/atmel/at91sam9260ek/at91sam9260ek.c +5 -0
@@ -157,6 +157,10 @@ int board_early_init_f(void)
 
  int board_init(void)
  {
+#ifdef CONFIG_AT91SAM9G20EK_2MMC
+       /* arch number of AT91SAM9260EK-Board */
+       gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9G20EK_2MMC;
+#else
  #ifdef CONFIG_AT91SAM9G20EK
         /* arch number of AT91SAM9260EK-Board */
         gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9G20EK;
@@ -164,6 +168,7 @@ int board_init(void)
         /* arch number of AT91SAM9260EK-Board */
         gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9260EK;
  #endif
+#endif
         /* adress of boot parameters */
         gd->bd->bi_boot_params = CONFIG_SYS_SDRAM_BASE + 0x100;
 
boards.cfg +0 -0
index e4b0d44..fa863dc 100644
@@ -95,6 +95,7 @@ at91sam9g10ek_nandflash      arm         arm926ejs 

at91sam9261ek atmel
at91sam9g20ek_dataflash_cs0 arm arm926ejs at91sam9260ek
atmel at91
at91sam9260ek:AT91SAM9G20,SYS_USE_DATAFLASH_CS0
at91sam9g20ek_dataflash_cs1 arm arm926ejs at91sam9260ek
atmel at91
at91sam9260ek:AT91SAM9G20,SYS_USE_DATAFLASH_CS1
at91sam9g20ek_nandflash arm arm926ejs at91sam9260ek
atmel at91 at91sam9260ek:AT91SAM9G20,SYS_USE_NANDFLASH
+at91sam9g20ek_2mmc_nandflash arm arm926ejs at91sam9260ek
atmel at91
at91sam9260ek:AT91SAM9G20,AT91SAM9G20EK_2MMC,SYS_USE_NANDFLASH
at91sam9m10g45ek_nandflash arm arm926ejs at91sam9m10g45ek
atmel at91
at91sam9m10g45ek:AT91SAM9M10G45,SYS_USE_NANDFLASH

  at91sam9rlek_dataflash       arm         arm926ejs   at91sam9rlek 
    atmel          at91        at91sam9rlek:AT91SAM9RL,SYS_USE_DATAFLASH
  at91sam9rlek_nandflash       arm         arm926ejs   at91sam9rlek 
    atmel          at91        at91sam9rlek:AT91SAM9RL,SYS_USE_NANDFLASH

---------------------------------------------------------------------------

Best Regards,
Bo Shen

author
#7 lore raw
Quoted 49 lines · Bo Shen

You should use at91sam9g20ek_2mmc_nandflash to configure the u-boot

------------------------------------------------

board/atmel/at91sam9260ek/at91sam9260ek.c +5 -0
index 2555672..ce475df 100644
@@ -157,6 +157,10 @@ int board_early_init_f(void)
 
  int board_init(void)
  {
+#ifdef CONFIG_AT91SAM9G20EK_2MMC
+       /* arch number of AT91SAM9260EK-Board */
+       gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9G20EK_2MMC;
+#else
  #ifdef CONFIG_AT91SAM9G20EK
         /* arch number of AT91SAM9260EK-Board */
         gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9G20EK;
@@ -164,6 +168,7 @@ int board_init(void)
         /* arch number of AT91SAM9260EK-Board */
         gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9260EK;
  #endif
+#endif
         /* adress of boot parameters */
         gd->bd->bi_boot_params = CONFIG_SYS_SDRAM_BASE + 0x100;
 
boards.cfg +0 -0
index e4b0d44..fa863dc 100644
@@ -95,6 +95,7 @@ at91sam9g10ek_nandflash      arm         arm926ejs 

at91sam9261ek atmel
at91sam9g20ek_dataflash_cs0 arm arm926ejs at91sam9260ek
atmel at91
at91sam9260ek:AT91SAM9G20,SYS_USE_DATAFLASH_CS0
at91sam9g20ek_dataflash_cs1 arm arm926ejs at91sam9260ek
atmel at91
at91sam9260ek:AT91SAM9G20,SYS_USE_DATAFLASH_CS1
at91sam9g20ek_nandflash arm arm926ejs at91sam9260ek
atmel at91
at91sam9260ek:AT91SAM9G20,SYS_USE_NANDFLASH
+at91sam9g20ek_2mmc_nandflash arm arm926ejs at91sam9260ek
atmel at91
at91sam9260ek:AT91SAM9G20,AT91SAM9G20EK_2MMC,SYS_USE_NANDFLASH
at91sam9m10g45ek_nandflash arm arm926ejs
at91sam9m10g45ek
atmel at91

Quoted 10 lines

at91sam9m10g45ek:AT91SAM9M10G45,SYS_USE_NANDFLASH
at91sam9rlek_dataflash arm arm926ejs at91sam9rlek
atmel at91
at91sam9rlek:AT91SAM9RL,SYS_USE_DATAFLASH
at91sam9rlek_nandflash arm arm926ejs at91sam9rlek
atmel at91
at91sam9rlek:AT91SAM9RL,SYS_USE_NANDFLASH
---------------------------------------------------------------------------

That worked! Very nice! But, I had to apply the patches manually. Email may have messed up the patch sets with new-lines and indentation. Can this be added to the main stream? I am on to learning 'env import'. Thank you.

If email messed it up once, I guess it could mess it up again, but I generated patches using:
diff -Naur before/ after/ > diff.patch

atmel/at91sam9260ek/at91sam9260ek.c 2013-01-29 09:10:46.000000000 -0800 renamed from atmel/at91sam9260ek/at91sam9260ek.c.orig 2013-01-29 09:09:35.000000000 -0800 +5 -0
@@ -157,6 +157,10 @@
 
 int board_init(void)
 {
+#ifdef CONFIG_AT91SAM9G20EK_2MMC
+       /* arch number of AT91SAM9260EK-Board */
+       gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9G20EK_2MMC;
+#else
 #ifdef CONFIG_AT91SAM9G20EK
        /* arch number of AT91SAM9260EK-Board */
        gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9G20EK;
@@ -164,6 +168,7 @@
        /* arch number of AT91SAM9260EK-Board */
        gd->bd->bi_arch_number = MACH_TYPE_AT91SAM9260EK;
 #endif
+#endif
        /* adress of boot parameters */
        gd->bd->bi_boot_params = CONFIG_SYS_SDRAM_BASE + 0x100;
 
boards.cfg 2013-01-29 08:58:21.000000000 -0800 renamed from boards.cfg.orig 2013-01-29 08:56:47.000000000 -0800 +1 -0
@@ -94,6 +94,7 @@
 at91sam9g10ek_nandflash      arm         arm926ejs   at91sam9261ek       atmel          at91        at91sam9261ek:AT91SAM9G10,SYS_USE_NANDFLASH
 at91sam9g20ek_dataflash_cs0  arm         arm926ejs   at91sam9260ek       atmel          at91        at91sam9260ek:AT91SAM9G20,SYS_USE_DATAFLASH_CS0
 at91sam9g20ek_dataflash_cs1  arm         arm926ejs   at91sam9260ek       atmel          at91        at91sam9260ek:AT91SAM9G20,SYS_USE_DATAFLASH_CS1
+at91sam9g20ek_2mmc_nandflash arm         arm926ejs   at91sam9260ek       atmel          at91        at91sam9260ek:AT91SAM9G20,AT91SAM9G20EK_2MMC,SYS_USE_NANDFLASH
 at91sam9g20ek_nandflash      arm         arm926ejs   at91sam9260ek       atmel          at91        at91sam9260ek:AT91SAM9G20,SYS_USE_NANDFLASH
 at91sam9m10g45ek_nandflash   arm         arm926ejs   at91sam9m10g45ek    atmel          at91        at91sam9m10g45ek:AT91SAM9M10G45,SYS_USE_NANDFLASH
 at91sam9rlek_dataflash       arm         arm926ejs   at91sam9rlek        atmel          at91        at91sam9rlek:AT91SAM9RL,SYS_USE_DATAFLASH
#8 lore raw

Hi John Stile,

Quoted 4 lines · John Stile

That worked! Very nice! But, I had to apply the patches manually. Email may have messed up the patch sets with new-lines and indentation. Can this be added to the main stream? I am on to learning 'env import'. Thank you.

I will prepare the patch and send to u-boot mainline.

Quoted 3 lines

If email messed it up once, I guess it could mess it up again, but I generated patches using:
diff -Naur before/ after/> diff.patch

I think this can not be applied to u-boot mainline directly, because it missed comments and author, and etc.

Best Regards,
Bo Shen

#9 lore raw
In reply to Jeroen Hofstee

Dear Jeroen Hofstee,

Quoted 11 lines · In message <[email protected]> you wrote:
Quoted 7 lines
Quoted 4 lines

I'm stuck on u-boot-1.3.4.
Is there a patch?

search the mailinglist, this is asking for a visit to
http://www.uboat.net/special/archiv/ or a response the like. Try to upgrade if possible...

for completeness sake, Wolfgang will likely tell you it is museum material, since it is almost 5 years old. No idea what the site is about, just popped up with "u-boot museum" instead of Wolfgang's replies.

No surprise. Museums were not invented yet by the time when v1.3.4 was current ;-)

Best regards,

Wolfgang Denk

Signature · 6 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
It is impractical for  the  standard  to  attempt  to  constrain  the
behavior  of code that does not obey the constraints of the standard.
                                                          - Doug Gwyn