[PATCH] imd: make it work on big-endian machine

[Antony Pavlov]

On Fri, 20 May 2016 00:05:59 +0300
Antony Pavlov <antonynpavlov at gmail.com> wrote:

> The commit
> 
>     commit 5e335773e0814173b40873a891964a7858d64688
>     Author: Sascha Hauer <s.hauer at pengutronix.de>
>     Date:   Tue Mar 29 10:06:46 2016 +0200
> 
>         imd: use struct imd_header * as argument
> 
> introduces additional imd type checks like this
> 
>     if (!imd_is_string(imd->type))
>            return NULL;
> 
> These checks work incorrectly on any big-endian machine
> because imd->type field needs addition convertion
                                         ^^^^^^^^^^

Hmm, It looks like I have to fix a typo: "convertion" -> "conversion".
I'll resend  v2 version of the patch in a few minutes.


> to little-endian byteorder before use.
> 
> Here is the imd command output on big-endian qemu-malta:
> 
>     barebox:/ imd /dev/nor0.barebox
>     release: <NULL>
>     build: <NULL>
> 
> This patch fixes the problem by adding necessary convertion
> via imd_read_type().
> 
> Signed-off-by: Antony Pavlov <antonynpavlov at gmail.com>
> ---
>  common/imd.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/common/imd.c b/common/imd.c
> index 159b73a..a8a10fc 100644
> --- a/common/imd.c
> +++ b/common/imd.c
> @@ -212,7 +212,7 @@ const char *imd_string_data(struct imd_header *imd, int index)
>  	int len = imd_read_length(imd);
>  	char *p = (char *)(imd + 1);
>  
> -	if (!imd_is_string(imd->type))
> +	if (!imd_is_string(imd_read_type(imd)))
>  		return NULL;
>  
>  	for (i = 0; total < len; total += l, p += l) {
> @@ -239,7 +239,7 @@ char *imd_concat_strings(struct imd_header *imd)
>  	char *str;
>  	char *data = (char *)(imd + 1);
>  
> -	if (!imd_is_string(imd->type))
> +	if (!imd_is_string(imd_read_type(imd)))
>  		return NULL;
>  
>  	str = malloc(len);
> -- 
> 2.7.0
> 


-- 
-- 
Best regards,
  Antony Pavlov