[U-Boot,v2,04/12] SPL: Add NOR flash booting support

SPL NOR flash booting support is quite simple. Only copying of the
images is needed.

On MPC5xxx we need to make sure to only use the standard memcpy()
implementation and not the MPC5xxx specific one. As the MPC5xxx
version has some complexity which is not needed for this SPL
booting.

Signed-off-by: Stefan Roese <sr@denx.de>
---
Changes in v2:
- Add option to skip copying of the mkimage header

 common/spl/Makefile  |  1 +
 common/spl/spl.c     |  5 ++++
 common/spl/spl_nor.c | 71 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 include/spl.h        |  3 +++
 4 files changed, 80 insertions(+)
 create mode 100644 common/spl/spl_nor.c

On Mon, Aug 27, 2012 at 12:50:59PM +0200, Stefan Roese wrote:
> SPL NOR flash booting support is quite simple. Only copying of the
> images is needed.
> 
> On MPC5xxx we need to make sure to only use the standard memcpy()
> implementation and not the MPC5xxx specific one. As the MPC5xxx
> version has some complexity which is not needed for this SPL
> booting.
> 
> Signed-off-by: Stefan Roese <sr@denx.de>

Added to my v5 series, thanks!

Hi Stefan,

2012/8/27 Stefan Roese <sr@denx.de>:
> SPL NOR flash booting support is quite simple. Only copying of the
> images is needed.
>
> On MPC5xxx we need to make sure to only use the standard memcpy()
> implementation and not the MPC5xxx specific one. As the MPC5xxx
> version has some complexity which is not needed for this SPL
> booting.
>
> Signed-off-by: Stefan Roese <sr@denx.de>
> ---
> Changes in v2:
> - Add option to skip copying of the mkimage header
>
>  common/spl/Makefile  |  1 +
>  common/spl/spl.c     |  5 ++++
>  common/spl/spl_nor.c | 71 ++++++++++++++++++++++++++++++++++++++++++++++++++++
>  include/spl.h        |  3 +++
>  4 files changed, 80 insertions(+)
>  create mode 100644 common/spl/spl_nor.c
>
> diff --git a/common/spl/Makefile b/common/spl/Makefile
> index b61b438..53a82c4 100644
> --- a/common/spl/Makefile
> +++ b/common/spl/Makefile
> @@ -15,6 +15,7 @@ LIB   = $(obj)libspl.o
>
>  ifdef CONFIG_SPL_BUILD
>  COBJS-$(CONFIG_SPL_FRAMEWORK) += spl.o
> +COBJS-$(CONFIG_SPL_NOR_SUPPORT) += spl_nor.o
>  COBJS-$(CONFIG_SPL_YMODEM_SUPPORT) += spl_ymodem.o
>  endif
>
> diff --git a/common/spl/spl.c b/common/spl/spl.c
> index 13bebbc..61936ee 100644
> --- a/common/spl/spl.c
> +++ b/common/spl/spl.c
> @@ -155,6 +155,11 @@ void board_init_r(gd_t *dummy1, ulong dummy2)
>                 spl_nand_load_image();
>                 break;
>  #endif
> +#ifdef CONFIG_SPL_NOR_SUPPORT
> +       case BOOT_DEVICE_NOR:
> +               spl_nor_load_image();
> +               break;
> +#endif
>  #ifdef CONFIG_SPL_YMODEM_SUPPORT
>         case BOOT_DEVICE_UART:
>                 spl_ymodem_load_image();
> diff --git a/common/spl/spl_nor.c b/common/spl/spl_nor.c
> new file mode 100644
> index 0000000..a1c13a4
> --- /dev/null
> +++ b/common/spl/spl_nor.c
> @@ -0,0 +1,71 @@
> +/*
> + * Copyright (C) 2012 Stefan Roese <sr@denx.de>
> + *
> + * See file CREDITS for list of people who contributed to this
> + * project.
> + *
> + * This program is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU General Public License as
> + * published by the Free Software Foundation; either version 2 of
> + * the License, or (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.         See the
> + * GNU General Public License for more details.
> + */
> +
> +#include <common.h>
> +#include <spl.h>
> +
> +/*
> + * Don't use the special MPC5xxx memcpy implementation, only use
> + * the standard one.
> + */
> +#if defined(CONFIG_MPC512X) || defined(CONFIG_MPC5200)
> +extern void *__memcpy(void *, const void *, size_t);
> +#define memcpy         __memcpy
> +#endif

can't we move this to arch-specifc spl code?
I think arch/CPU/SoC specific definitions should be avoided in new
generic frameworks.

how about:

__weak void *spl_memcpy(void *dst, const void *src, size_t size)
{
         return memcpy(dst, src, size);
}

and in arch/powerpc/cpu/mpc5xxx/spl_boot.c:

void *spl_memcpy(void *dst, const void *src, size_t size)
{
         return __memcpy(dst, src, size);
}

> +
> +void spl_nor_load_image(void)
> +{
> +       /*
> +        * Loading of the payload to SDRAM is done with skipping of
> +        * the mkimage header in this SPL NOR driver
> +        */
> +       spl_image.flags |= SPL_COPY_PAYLOAD_ONLY;
> +
> +       if (spl_start_uboot()) {
> +               /*
> +                * Load real U-Boot from its location in NOR flash to its
> +                * defined location in SDRAM
> +                */
> +               spl_parse_image_header(
> +                       (const struct image_header *)CONFIG_SYS_UBOOT_BASE);
> +
> +               memcpy((void *)spl_image.load_addr,
> +                      (void *)(CONFIG_SYS_UBOOT_BASE +
> +                               sizeof(struct image_header)),
> +                      spl_image.size);
> +       } else {
> +               /*
> +                * Load Linux from its location in NOR flash to its defined
> +                * location in SDRAM
> +                */
> +               spl_parse_image_header(
> +                       (const struct image_header *)CONFIG_SYS_OS_BASE);
> +
> +               memcpy((void *)spl_image.load_addr,
> +                      (void *)(CONFIG_SYS_OS_BASE +
> +                               sizeof(struct image_header)),
> +                      spl_image.size);
> +
> +               /*
> +                * Copy DT blob (fdt) to SDRAM. Passing pointer to flash
> +                * doesn't work (16 KiB should be enough for DT)
> +                */
> +               memcpy((void *)CONFIG_SYS_SPL_ARGS_ADDR,
> +                      (void *)(CONFIG_SYS_FDT_BASE),
> +                      (16 << 10));
> +       }
> +}
> diff --git a/include/spl.h b/include/spl.h
> index 673c4a4..4a76239 100644
> --- a/include/spl.h
> +++ b/include/spl.h
> @@ -60,6 +60,9 @@ void spl_display_print(void);
>  /* NAND SPL functions */
>  void spl_nand_load_image(void);
>
> +/* NOR SPL functions */
> +void spl_nor_load_image(void);
> +
>  /* MMC SPL functions */
>  void spl_mmc_load_image(void);
>
> --
> 1.7.12
>
> _______________________________________________
> U-Boot mailing list
> U-Boot@lists.denx.de
> http://lists.denx.de/mailman/listinfo/u-boot

On 08/27/2012 10:29 AM, Daniel Schwierzeck wrote:
> Hi Stefan,
> 
> 2012/8/27 Stefan Roese <sr@denx.de>:
>> SPL NOR flash booting support is quite simple. Only copying of the
>> images is needed.
>>
>> On MPC5xxx we need to make sure to only use the standard memcpy()
>> implementation and not the MPC5xxx specific one. As the MPC5xxx
>> version has some complexity which is not needed for this SPL
>> booting.
>>
>> Signed-off-by: Stefan Roese <sr@denx.de>
[snip]
>> +/*
>> + * Don't use the special MPC5xxx memcpy implementation, only use
>> + * the standard one.
>> + */
>> +#if defined(CONFIG_MPC512X) || defined(CONFIG_MPC5200)
>> +extern void *__memcpy(void *, const void *, size_t);
>> +#define memcpy         __memcpy
>> +#endif
> 
> can't we move this to arch-specifc spl code?
> I think arch/CPU/SoC specific definitions should be avoided in new
> generic frameworks.

This is a good point.  And I think we can further tweak this by
modifying arch/powerpc/lib/Makefile to just not build memcpy_5200.c for
CONFIG_SPL_BUILD.  So lets go with that, and I'll just fixup my local copy.

On 08/27/2012 07:59 PM, Tom Rini wrote:
> On 08/27/2012 10:29 AM, Daniel Schwierzeck wrote:
>> Hi Stefan,
>>
>> 2012/8/27 Stefan Roese <sr@denx.de>:
>>> SPL NOR flash booting support is quite simple. Only copying of the
>>> images is needed.
>>>
>>> On MPC5xxx we need to make sure to only use the standard memcpy()
>>> implementation and not the MPC5xxx specific one. As the MPC5xxx
>>> version has some complexity which is not needed for this SPL
>>> booting.
>>>
>>> Signed-off-by: Stefan Roese <sr@denx.de>
> [snip]
>>> +/*
>>> + * Don't use the special MPC5xxx memcpy implementation, only use
>>> + * the standard one.
>>> + */
>>> +#if defined(CONFIG_MPC512X) || defined(CONFIG_MPC5200)
>>> +extern void *__memcpy(void *, const void *, size_t);
>>> +#define memcpy         __memcpy
>>> +#endif
>>
>> can't we move this to arch-specifc spl code?
>> I think arch/CPU/SoC specific definitions should be avoided in new
>> generic frameworks.
> 
> This is a good point.  And I think we can further tweak this by
> modifying arch/powerpc/lib/Makefile to just not build memcpy_5200.c for
> CONFIG_SPL_BUILD.  So lets go with that, and I'll just fixup my local copy.

Good idea. I'll change the powerpc Makefile accordingly.

Thanks,
Stefan