[U-Boot] __mmc_get_env_addr() is not being called?

[Jason Liu]

2011/12/16 Stefano Babic <sbabic at denx.de>:
> On 16/12/2011 06:54, Shawn Guo wrote:
>> Hi,
>>
>
> Hi Shawn,
>
>> I'm running v2011.12-rc1 (with one missing usdhc patch applied) on
>> mx6qarm2 board, and seeing a problem.  When I install u-boot on a
>> blank SD card, it boots fine with messge "*** Warning - bad CRC,
>> using default environment" seen as expected.  But once I run command
>> 'save' to write envs to the card, the card stop booting.
>>
>> I tracked the issue a little bit, and found __mmc_get_env_addr() does
>> not get called at all.  Therefore when I run 'save' for the first time,
>> the envs are written to offset 0 of the card, which in turn overwrites
>> the boot image.
>>
>> I have not figured out why it does.  But the following change fixes
>> the problem for me.
>>
>> diff --git a/common/env_mmc.c b/common/env_mmc.c
>> index 8441c77..3832fe4 100644
>> --- a/common/env_mmc.c
>> +++ b/common/env_mmc.c
>> @@ -46,13 +46,11 @@ DECLARE_GLOBAL_DATA_PTR;
>>  #define CONFIG_ENV_OFFSET 0
>>  #endif
>>
>> -static int __mmc_get_env_addr(struct mmc *mmc, u32 *env_addr)
>> +static int mmc_get_env_addr(struct mmc *mmc, u32 *env_addr)
>>  {
>>         *env_addr = CONFIG_ENV_OFFSET;
>>         return 0;
>>  }
>> -int mmc_get_env_addr(struct mmc *mmc, u32 *env_addr)
>> -       __attribute__((weak, alias("__mmc_get_env_addr")));
>>
>> Any comment on why the change fixes the problem is appreciated.
>
> The function is declared weak, this means that a specific architecture
> could implement its own version. The reason why it does not work is that
> there is a specific implementation in board/freescale/sdhc_boot.c, and
> the mmc_get_env_addr() in this file is taken instead of the default when
> CONFIG_ENV_IS_IN_MMC (as in our case) is set. Instead of your proposal,
> we should surround mmc_get_env_addr() in sdhc_boot.c with something like
> #ifndef CONFIG_MX6Q, or check if we can reuse this function, that really
> does not use CONFIG_ENV_OFFSET at all.
>
> I put Jason (board maintainer) in CC - Jason, what do mind ? Do you get
> the same issue when you tested the board ?

Stefano, I saw the issue.
Fabio's patch can fix the i.mx28evk issue, but can't fix the following
platforms:

11    144  include/configs/efikamx.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC
  12     86  include/configs/mx51evk.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC
  13     58  include/configs/mx53ard.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC
  14     68  include/configs/mx53evk.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC
  15     51  include/configs/mx53loco.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC
  16     58  include/configs/mx53smd.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC
  17     48  include/configs/mx6qarm2.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC
  18     48  include/configs/mx6qsabrelite.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC
  20    105  include/configs/vision2.h <<CONFIG_FSL_ESDHC>>
             #define CONFIG_FSL_ESDHC

The commit  97039ab98 (env_mmc: Allow board code to override the
environment address)
break the i.mx5/6 boards support.

 BR,
Jason

>
> Best regards,
> Stefano Babic
>
> --
> =====================================================================
> DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
> HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
> Phone: +49-8142-66989-0 Fax: +49-8142-66989-80  Email: office at denx.de
> =====================================================================
> _______________________________________________
> U-Boot mailing list
> U-Boot at lists.denx.de
> http://lists.denx.de/mailman/listinfo/u-boot