Aha. I missed that the seed is injected into buf before passing it
into half_md4_transform. I was looking at it as just the empty buffer
before the first iteration of the loop (or, in my case, since I was
testing with a 6 char filename, the only iteration).

I will repeat my experiment with that and see if I can tease it out.

Thanks, Ted!

On Mon, Oct 11, 2021 at 1:20 PM Theodore Ts'o [off-list ref] wrote:

On Mon, Oct 11, 2021 at 08:30:36AM -0700, Avi Deitcher wrote:

quoted

Does someone know how this is constructed and used?

On Mon, Oct 4, 2021 at 12:57 AM Avi Deitcher [off-list ref] wrote:

quoted

Hi Andreas,

I had looked in __ext4fs_dirhash(). Yes, it does reference the seed -
and create a default if none is there at the filesystem level - but it
doesn't appear to use it, in that function. hinfo is populated in the
function - hash, minor-hash, seed - but it never uses the seed to
manipulate the hash.

The seed is used to initialize the buf array, so long as the seed is
not all zero's.  If it is all zeros, then the default seed is used
instead (right above this bit of code:

        if (hinfo->seed) {
                for (i = 0; i < 4; i++) {
                        if (hinfo->seed[i]) {
                                memcpy(buf, hinfo->seed, sizeof(buf));
                                break;
                        }
                }
        }

The legacy hash doesn't use the seed, yes.  But for the other hash
types (hash_version), they mix the filename (in different ways
depending on the hash type.  For example, for half md4:

        case DX_HASH_HALF_MD4:
                p = name;
                while (len > 0) {
                        (*str2hashbuf)(p, len, in, 8);
                        half_md4_transform(buf, in);
                                           ^^^
                        len -= 32;
                        p += 32;
                }
                minor_hash = buf[2];
                hash = buf[1];
                break;

When the hash seed is different, that means the initial state of the
buf array will different, and this influences the resulting hash.

Cheers,

                                        - Ted

-- 
Avi Deitcher
avi@deitcher.net
Follow me http://twitter.com/avideitcher
Read me http://blog.atomicinc.com