[U-Boot] [PATCH v2 2/3] fit: allow fit to call hardware accelerated hash

[Ben Whitten]

Hi Tom,

> Subject: Re: [U-Boot] [PATCH v2 2/3] fit: allow fit to call hardware accelerated
> hash
> 
> On Wed, Jun 06, 2018 at 05:03:57PM +0100, Ben Whitten wrote:
> 
> > Move to calling the abstraction which allows for hardware acceleration.
> > We also remove unneeded defines and only include objects if required.
> >
> > Signed-off-by: Ben Whitten <ben.whitten at lairdtech.com>
> [snip]
> > diff --git a/common/image-fit.c b/common/image-fit.c
> > index 8c15ed1..7d8c961 100644
> > --- a/common/image-fit.c
> > +++ b/common/image-fit.c
> > @@ -1082,26 +1082,17 @@ int fit_set_timestamp(void *fit, int noffset,
> time_t timestamp)
> >  int calculate_hash(const void *data, int data_len, const char *algo,
> >  			uint8_t *value, int *value_len)
> >  {
> > -	if (IMAGE_ENABLE_CRC32 && strcmp(algo, "crc32") == 0) {
> > -		*((uint32_t *)value) = crc32_wd(0, data, data_len,
> > -							CHUNKSZ_CRC32);
> > -		*((uint32_t *)value) = cpu_to_uimage(*((uint32_t *)value));
> > -		*value_len = 4;
> > -	} else if (IMAGE_ENABLE_SHA1 && strcmp(algo, "sha1") == 0) {
> > -		sha1_csum_wd((unsigned char *)data, data_len,
> > -			     (unsigned char *)value, CHUNKSZ_SHA1);
> > -		*value_len = 20;
> > -	} else if (IMAGE_ENABLE_SHA256 && strcmp(algo, "sha256") == 0) {
> > -		sha256_csum_wd((unsigned char *)data, data_len,
> > -			       (unsigned char *)value, CHUNKSZ_SHA256);
> > -		*value_len = SHA256_SUM_LEN;
> > -	} else if (IMAGE_ENABLE_MD5 && strcmp(algo, "md5") == 0) {
> > -		md5_wd((unsigned char *)data, data_len, value,
> CHUNKSZ_MD5);
> > -		*value_len = 16;
> > -	} else {
> > +	struct hash_algo *hash_algo;
> > +	int ret;
> > +
> > +	ret = hash_lookup_algo(algo, &hash_algo);
> 
> This is unfortunately not code equivalent.  There's a little bit missing
> to switch this over to using SPL_CRC32_SUPPORT as well.  A good test
> would be to make sure that a sunxi platform still builds after this (it
> doesn't currently with the series) and use buildman -Ssdel to compare
> before/after sizes and make sure it's a tiny change in size related to
> the code switch and not a huge change due to droppped (or added!)
> support.  Thanks!

I see that I should be including the crc32.o file regardless in non SPL builds,
now the sunxi board build passes locally.
I tested a fit image with all hashes enabled and they all passed so atleast the
calling of the functions seems to work. Did you spot another inequivalence?
I am un-familiar with the buildman tool, but I shall give it a go as you suggest.

Thanks,
Ben