Re: [PATCH v4 07/12] Input: xbox_gip - Add security implementation from xone

[Vicki Pfau <vi@endrift.com>]

Hi Eric,

On 8/24/26 2:49 PM, Eric Biggers wrote:
> On Fri, Aug 21, 2026 at 07:11:31PM -0700, Vicki Pfau wrote:
>> +int gip_security_start_handshake(struct gip_security *security)
>> +{
>> +	struct shash_desc *shash_transcript, *shash_prf;
>> +
>> +	if (!security->shash_transcript) {
>> +		shash_transcript = gip_security_alloc_shash("sha256");
>> +		if (IS_ERR(shash_transcript))
>> +			return PTR_ERR(shash_transcript);
>> +
>> +		security->shash_transcript = shash_transcript;
>> +	}
>> +
>> +	if (!security->shash_prf) {
>> +		shash_prf = gip_security_alloc_shash("hmac(sha256)");
>> +		if (IS_ERR(shash_prf)) {
>> +			crypto_free_shash(shash_transcript->tfm);
>> +			kfree(shash_transcript);
>> +			security->shash_transcript = NULL;
>> +			return PTR_ERR(shash_prf);
>> +		}
>> +
>> +		security->shash_prf = shash_prf;
>> +	}
> 
> There's no need to use crypto_shash.  Just call the SHA-256 and
> HMAC-SHA256 functions (<crypto/sha2.h>) directly.

This code was originally written by someone else more familiar with these functions than me, but it looks like shash_transcript gets finalized repeatedly without being reset. See gip_security_get_transcript, which exports the state, finalizes it, then rolls back the state before finalizing it. I don't think this can be done with the basic sha2 functions. I might be able to do it with the HMAC though. I'd need to familiarize myself with the API first.> 
> - Eric

Vicki