[U-Boot] [PATCH] mkimage: Fix 'Unknown OMAP image type - 5'

[Dirk Behme]

On 08.08.2011 21:05, Albert ARIBAUD wrote:
> Hi Dirk,
>
> Le 05/08/2011 20:42, Dirk Behme a écrit :
>> From: Dirk Behme<dirk.behme at googlemail.com>
>>
>> Using mkimage with e.g.
>>
>> tools/mkimage -A arm -T firmware -O u-boot -d u-boot.bin foo.img
>>
>> gives a warning
>>
>> "Unknown OMAP image type - 5"
>>
>> while it seems that the image itself is created successfully.
>>
>> This does come from the patch "mkimage: Add OMAP boot image support".
>>
>> Reordering the init_xx_image_type() sequence does make this
>> message go away.
>>
>> Signed-off-by: Dirk Behme<dirk.behme at googlemail.com>
>> CC: John Rigby<john.rigby at linaro.org>
>> CC: Aneesh V<aneesh at ti.com>
>> CC: Sandeep Paulraj<s-paulraj at ti.com>
>>
>> ---
>>
>> This is reproducable with the recent mainline mkimage where the
>> patch "mkimage: Add OMAP boot image support" is applied:
>>
>> http://git.denx.de/cgi-bin/gitweb.cgi?p=u-boot.git;a=commit;h=3decb14abe76d244ba98fd158ef95f89e7e37d70
>>
>>
>> tools/mkimage.c | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> Index: u-boot.git/tools/mkimage.c
>> ===================================================================
>> --- u-boot.git.orig/tools/mkimage.c
>> +++ u-boot.git/tools/mkimage.c
>> @@ -156,12 +156,12 @@ main (int argc, char **argv)
>> init_imx_image_type ();
>> /* Init FIT image generation/list support */
>> init_fit_image_type ();
>> - /* Init TI OMAP Boot image generation/list support */
>> - init_omap_image_type();
>> /* Init Default image generation/list support */
>> init_default_image_type ();
>> /* Init Davinci UBL support */
>> init_ubl_image_type();
>> + /* Init TI OMAP Boot image generation/list support */
>> + init_omap_image_type();
>>
>> params.cmdname = *argv;
>> params.addr = params.ep = 0;
>> _______________________________________________
>> U-Boot mailing list
>> U-Boot at lists.denx.de
>> http://lists.denx.de/mailman/listinfo/u-boot
>
> Any idea why reordering fixes the issue?

No, unfortunately not. I looked at the other patches touching this 
section of mkimage.c and it looked like everybody is adding new init 
functions below the existing ones. And this helped ;)

I hoped that John or Aneesh would comment on this...

> Seems to me like init
> functions are not / should not be dependent on order, so the "fix"
> seems fragile to me, at least as long as we cannot add a good
> explanation.

Yes, I agree.

John? Aneesh? Any idea?

Best regards

Dirk