Hi Bruno,

Bruno Haible via GNU coreutils General Discussion <[email protected]>
writes:

> Pádraig Brady wrote:
>> > On Slackware 15 (32-bit x86), one test fails:
>> > 
>> > FAIL: tests/cksum/b2sum
>> 
>> This is more fallout from a recent change to use idx_t.
>> 
>> I.e., we're now assigning intmax_t to idx_t,
>> hence the error mismatch on 32 bit.
>> 
>> The attached should address this.
>
> I confirm that the newest snapshot passes all tests on Slackware 15.
>
> But do I understand things correctly, that on 32-bit systems, previously
> 'cksum', in the cases HASH_ALGO_BLAKE2 || HASH_ALGO_CKSUM, did support
> input files larger than 4 GiB, whereas now it's limited to input files
> smaller than 2 GiB ? Or, what else was the effect of changing the
> variable 'digest_length' from uintmax_t to idx_t?

In this case 'digest_length' refers to the number of bits in the digest,
e.g., 224, 256, 384, or 512 for SHA-2. It is not related to the size of
the input which should be practically unlimited. I guess there is a
limit on the message size for SHA-2, 2^64 - 1 bits for SHA-256 and
2^128 - 1 bits for SHA-512, but I think we don't have to worry about
that for some time. :)

Collin

Reply via email to