On 16/04/25 12:33, Herbert Xu wrote:
> On Fri, Apr 11, 2025 at 02:43:22PM +0530, T Pratham wrote:
>> + // Need to do a timeout to ensure mutex gets unlocked if DMA callback fails for any reason
>> + ret = wait_for_completion_timeout(&actx->aes_compl, msecs_to_jiffies(DTHE_DMA_TIMEOUT_MS));
> You cannot wait here. The preferred method of indicating completion
> is through an interrupt. If you hardware can't do that, then you need
> to use crypto_engine and poll in that thread.
Sure, will change the strategy and update this.
>> +static int dthe_aes_crypt(struct skcipher_request *req, int enc)
>> +{
>> + struct dthe_tfm_ctx *ctx = crypto_skcipher_ctx(crypto_skcipher_reqtfm(req));
>> +
>> + /*
>> + * If data is not a multiple of AES_BLOCK_SIZE, need to return -EINVAL
>> + * If data length input is zero, no need to do any operation.
>> + */
>> + if (req->cryptlen % AES_BLOCK_SIZE) {
>> + skcipher_request_complete(req, -EINVAL);
> The completion function must not be called unless you first
> returned EINPROGRESS.
>
> PS Please set the bit KERN_DRIVER_ONLY in cra_flags.
KERN_DRIVER_ONLY bit is set.
>
> Cheers,
Thanks.
Regards
T Pratham <t-pratham@xxxxxx>
