[U-Boot-Users] environment not being used

[U-Boot-Users] environment not being used

[Shamile Khan]

Hi,

I am using U-boot on our FPGA based board. Currently, its not possible
to execute instructions directly from flash on our board since our flash
device is configured as an 8-bit device and it is connected to a 32-bit
processor bus. 

To get around this, we have a small program which is stored in FPGA
bitstream and runs in FPGA on-chip memory (called BRAM). This program
copies u-boot binary image from flash to sdram and jumps to it. This
allows U-boot to run. However the problem is that the environment
variables saved in flash are not being used even though U-boot has been
configured with the CFG_ENV_IS_IN_FLASH option turned on. 
CFG_FLASH_BASE has been set to the flash base address of u-boot and when
I use saveenv the environment is saved at the proper offset
CFG_ENV_OFFSET. The only thing thats a bit different for my U-boot
configuration is that CFG_MONITOR_BASE is set to the sdram base address
where U-boot starts executing from instead of being set to
CFG_FLASH_BASE.

Any ideas why the environment variables in flash do not get used?

Thanks,
Shamile 

[U-Boot-Users] environment not being used

[Wolfgang Denk]

In message <1075423717.18221.233.camel@localhost.localdomain> you wrote:
> 
> To get around this, we have a small program which is stored in FPGA
> bitstream and runs in FPGA on-chip memory (called BRAM). This program
> copies u-boot binary image from flash to sdram and jumps to it. This
> allows U-boot to run. However the problem is that the environment
> variables saved in flash are not being used even though U-boot has been
> configured with the CFG_ENV_IS_IN_FLASH option turned on. 
> CFG_FLASH_BASE has been set to the flash base address of u-boot and when
> I use saveenv the environment is saved at the proper offset
> CFG_ENV_OFFSET. The only thing thats a bit different for my U-boot
> configuration is that CFG_MONITOR_BASE is set to the sdram base address
> where U-boot starts executing from instead of being set to
> CFG_FLASH_BASE.

Normally U-Boot attempts to read the environment when  starting  from
the same position in memory where it will save it to when you issue a
"saveenv"  command.  You  will  have  to  modify U-Boot such that the
initialization reads from the "fixed" copy in  RAM,  while  "saveenv"
saves to the address in flash, performing the 16-bit-expansion needed
for  your  setup. It seems obvious that a plain copy as usuall cannot
work on your system.

Best regards,

Wolfgang Denk

-- 
See us @ Embedded World, Nuremberg, Feb 17 - 19,  Hall 12.0 Booth 440
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
Drun'? 'm not drun'! You woudn' dare call m' drun' if I was sober!
                                     - Terry Pratchett, _Men at Arms_

[U-Boot-Users] environment not being used

[Shamile Khan]

Hi,

Just as a follow up, I conclude that U-boot treats flash reads exactly
as sdram reads i.e. 32-bit accesses. So its not just about the
environment variables. Most of the flash-based U-boot functionality
wouldn't work for us unless we modify all the commands that have access
to flash e.g md, imiinfo, bootm etc. A hardware fix (an 8-bit bus to our
byte-mode flash chip) seems like the right way to go.

Shamile

On Sat, 2004-01-31 at 12:12, Wolfgang Denk wrote:
> In message <1075423717.18221.233.camel@localhost.localdomain> you wrote:
> > 
> > To get around this, we have a small program which is stored in FPGA
> > bitstream and runs in FPGA on-chip memory (called BRAM). This program
> > copies u-boot binary image from flash to sdram and jumps to it. This
> > allows U-boot to run. However the problem is that the environment
> > variables saved in flash are not being used even though U-boot has been
> > configured with the CFG_ENV_IS_IN_FLASH option turned on. 
> > CFG_FLASH_BASE has been set to the flash base address of u-boot and when
> > I use saveenv the environment is saved at the proper offset
> > CFG_ENV_OFFSET. The only thing thats a bit different for my U-boot
> > configuration is that CFG_MONITOR_BASE is set to the sdram base address
> > where U-boot starts executing from instead of being set to
> > CFG_FLASH_BASE.
> 
> Normally U-Boot attempts to read the environment when  starting  from
> the same position in memory where it will save it to when you issue a
> "saveenv"  command.  You  will  have  to  modify U-Boot such that the
> initialization reads from the "fixed" copy in  RAM,  while  "saveenv"
> saves to the address in flash, performing the 16-bit-expansion needed
> for  your  setup. It seems obvious that a plain copy as usuall cannot
> work on your system.
> 
> Best regards,
> 
> Wolfgang Denk
> 
> -- 
> See us @ Embedded World, Nuremberg, Feb 17 - 19,  Hall 12.0 Booth 440
> Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
> Drun'? 'm not drun'! You woudn' dare call m' drun' if I was sober!
>                                      - Terry Pratchett, _Men at Arms_
> 

[U-Boot-Users] environment not being used

[Wolfgang Denk]

In message <1075855660.7664.95.camel@localhost.localdomain> you wrote:
> 
> Just as a follow up, I conclude that U-boot treats flash reads exactly
> as sdram reads i.e. 32-bit accesses. So its not just about the

This is not correct.

U-Boot accesses the flash just the way how you configure your  memory
controller for flash accesses.

> to flash e.g md, imiinfo, bootm etc. A hardware fix (an 8-bit bus to our
> byte-mode flash chip) seems like the right way to go.

Either you or me must  be  missing  something.  Why  don't  you  just
configure U-Boot for a 8 bit flash device then?

Best regards,

Wolfgang Denk

-- 
See us @ Embedded World, Nuremberg, Feb 17 - 19,  Hall 12.0 Booth 440
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
Real Programmers always confuse Christmas and Halloween because
OCT 31 == DEC 25 !  - Andrew Rutherford (andrewr at ucs.adelaide.edu.au)

[U-Boot-Users] environment not being used

[Shamile Khan]

On Wed, 2004-02-04 at 15:30, Wolfgang Denk wrote:
> In message <1075855660.7664.95.camel@localhost.localdomain> you wrote:
> > 
> > Just as a follow up, I conclude that U-boot treats flash reads exactly
> > as sdram reads i.e. 32-bit accesses. So its not just about the
> 
> This is not correct.
> 
> U-Boot accesses the flash just the way how you configure your  memory
> controller for flash accesses.

Exactly. The memory controller is in FPGA. Its been configured to have a
data bus width of 8 bits to the flash memory device. Consequently if a
32 bit access is requested, the first 8 bits are fine and the rest are
garbage.

> 
> > to flash e.g md, imiinfo, bootm etc. A hardware fix (an 8-bit bus to our
> > byte-mode flash chip) seems like the right way to go.
> 
> Either you or me must  be  missing  something.  Why  don't  you  just
> configure U-Boot for a 8 bit flash device then?

I do have the flash device configured as an 8-bit device and flash
memory writes are for byte mode in the flash driver. So writes to flash
are fine as long as I use cp.b instead of just cp. 

As far as reads are concerned, since the flash device is in byte mode I
can only access 8 bits in one flash memory read cycle. However U-boot
commands like md, bootm, iminfo, crc32 etc access 32 bits in one flash
memory read cycle (md.b is an exception).

So those commands would work if I have for example 4 byte-mode flash
devices and the memory controller concatenates the data thats read from
those devices. But I only have one byte-mode flash device.

> Best regards,
> 
> Wolfgang Denk
> 
> -- 
> See us @ Embedded World, Nuremberg, Feb 17 - 19,  Hall 12.0 Booth 440
> Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
> Real Programmers always confuse Christmas and Halloween because
> OCT 31 == DEC 25 !  - Andrew Rutherford (andrewr at ucs.adelaide.edu.au)
> 

[U-Boot-Users] environment not being used

[Wolfgang Denk]

In message <1075936515.1553.21.camel@localhost.localdomain> you wrote:
>
> > U-Boot accesses the flash just the way how you configure your  memory
> > controller for flash accesses.
> 
> Exactly. The memory controller is in FPGA. Its been configured to have a
> data bus width of 8 bits to the flash memory device. Consequently if a
> 32 bit access is requested, the first 8 bits are fine and the rest are
> garbage.

Then your memory controller is broken.

Best regards,

Wolfgang Denk

-- 
See us @ Embedded World, Nuremberg, Feb 17 - 19,  Hall 12.0 Booth 440
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
To the systems programmer,  users  and  applications  serve  only  to
provide a test load.