Hi Wolfgang,
On Thu, 18 Apr 2013 16:39:09 +0200, Wolfgang Denk wd@denx.de wrote:
So how about changing the element type of output in the definition of hash_func_ws, adapting the corresponding implementations sha1_csum_wd, sha256_csum_wd and crc32_wd_buf, and adapting the output argument of the sole call to hash_func_ws, that is, the local "u8 output[HASH_MAX_DIGEST_SIZE];" in hash.c? Then we'be done with alignment.
OK, but that would be a totally different approach (which appears to be a good one, here).
Indeed; I would suggest splitting this change in two independent ones:
- fix the endianness issue: change the endianness of crc through the use of htonl() but leave the existing memcpy() in place as it is, even though it is not speed-optimized. That's what Simon's patch does if the HOSTCC part is ignored;
- fix the unalignment issue by changing parameter 'output' of function type 'hash_func_ws' from u8* to u32* and adjusting the rest of the code accordingly -- which would lead to replacing the crc32 final memcpy() with a single indirect assignment.
These two changes could be submitted either separately, or as a series.
Amicalement,