linux-kernel - Re: [PATCH v2 09/10] crypto: caam - add crypto

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <VI1PR04MB44452FF06F35075413CF87F88C350@VI1PR04MB4445.eurprd04.prod.outlook.com>
Date:   Mon, 13 Jan 2020 09:48:11 +0000
From:   Iuliana Prodan <iuliana.prodan@....com>
To:     Horia Geanta <horia.geanta@....com>,
        Herbert Xu <herbert@...dor.apana.org.au>,
        Aymen Sghaier <aymen.sghaier@....com>
CC:     "David S. Miller" <davem@...emloft.net>,
        "linux-crypto@...r.kernel.org" <linux-crypto@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        dl-linux-imx <linux-imx@....com>
Subject: Re: [PATCH v2 09/10] crypto: caam - add crypto_engine support for RSA
 algorithms

On 1/10/2020 10:46 AM, Horia Geanta wrote:
> On 1/3/2020 3:03 AM, Iuliana Prodan wrote:
>> +static int akcipher_enqueue_req(struct device *jrdev, u32 *desc,
>> +				void (*cbk)(struct device *jrdev, u32 *desc,
>> +					    u32 err, void *context),
>> +				struct akcipher_request *req,
>> +				struct rsa_edesc *edesc)
>> +{
>> +	struct caam_drv_private_jr *jrpriv = dev_get_drvdata(jrdev);
>> +	struct crypto_akcipher *tfm = crypto_akcipher_reqtfm(req);
>> +	struct caam_rsa_ctx *ctx = akcipher_tfm_ctx(tfm);
>> +	struct caam_rsa_key *key = &ctx->key;
>> +	int ret;
>> +
>> +	if (req->base.flags & CRYPTO_TFM_REQ_MAY_BACKLOG)
>> +		return crypto_transfer_akcipher_request_to_engine(jrpriv->engine,
>> +								  req);
> Resource leak in case transfer fails.
> 
>> +	else
>> +		ret = caam_jr_enqueue(jrdev, desc, cbk, &edesc->jrentry);
> What's the problem with transferring all requests to crypto engine?
> 
I'll address all your comments in v3.

Regarding the transfer request to crypto-engine: if sending all requests 
to crypto-engine, multibuffer tests, for non-backlogging requests fail 
after only 10 requests, since crypto-engine queue has 10 entries.
Here's an example:
root@...6qpdlsolox:~# insmod tcrypt.ko mode=422 num_mb=1024
insmod: ERROR: could not insert module tcrypt.ko: Resource temporarily 
unavailable
root@...6qpdlsolox:~#
root@...6qpdlsolox:~# dmesg
...
testing speed of multibuffer sha1 (sha1-caam)
tcrypt: test  0 (   16 byte blocks,   16 bytes per update,   1 updates):
tcrypt: concurrent request 11 error -28
tcrypt: concurrent request 13 error -28
tcrypt: concurrent request 14 error -28
tcrypt: concurrent request 16 error -28
tcrypt: concurrent request 18 error -28
tcrypt: concurrent request 20 error -28
tcrypt: concurrent request 22 error -28
tcrypt: concurrent request 24 error -28
tcrypt: concurrent request 26 error -28
tcrypt: concurrent request 28 error -28
tcrypt: concurrent request 30 error -28
tcrypt: concurrent request 32 error -28
tcrypt: concurrent request 34 error -28

tcrypt: concurrent request 1020 error -28
tcrypt: concurrent request 1022 error -28
tcrypt: At least one hashing failed ret=-28
root@...6qpdlsolox:~#

If sending just the backlog request to crypto-engine, and non-blocking 
directly to CAAM, these tests have a better chance to pass since JR has 
1024 entries.

Will need to work/update crypto-engine: set queue length when initialize 
crypto-engine, and remove serialization of requests in crypto-engine. 
But, until then, I would like to have a backlogging solution in CAAM driver.

Iulia

>> +
>> +	if (ret != -EINPROGRESS) {
>> +		switch (key->priv_form) {
>> +		case FORM1:
>> +			rsa_priv_f1_unmap(jrdev, edesc, req);
>> +			break;
>> +		case FORM2:
>> +			rsa_priv_f2_unmap(jrdev, edesc, req);
>> +			break;
>> +		case FORM3:
>> +			rsa_priv_f3_unmap(jrdev, edesc, req);
>> +			break;
>> +		default:
>> +			rsa_pub_unmap(jrdev, edesc, req);
>> +		}
>> +		rsa_io_unmap(jrdev, edesc, req);
>> +		kfree(edesc);
>> +	}
>> +
>> +	return ret;
>> +}
>> +
>>   static int caam_rsa_enc(struct akcipher_request *req)
>>   {
>>   	struct crypto_akcipher *tfm = crypto_akcipher_reqtfm(req);
>>   	struct caam_rsa_ctx *ctx = akcipher_tfm_ctx(tfm);
>> +	struct caam_rsa_req_ctx *req_ctx = akcipher_request_ctx(req);
>>   	struct caam_rsa_key *key = &ctx->key;
>>   	struct device *jrdev = ctx->dev;
>>   	struct rsa_edesc *edesc;
>> @@ -637,13 +726,9 @@ static int caam_rsa_enc(struct akcipher_request *req)
>>   	/* Initialize Job Descriptor */
>>   	init_rsa_pub_desc(edesc->hw_desc, &edesc->pdb.pub);
>>   
>> -	ret = caam_jr_enqueue(jrdev, edesc->hw_desc, rsa_pub_done,
>> -			      &edesc->jrentry);
>> -	if (ret == -EINPROGRESS)
>> -		return ret;
>> -
>> -	rsa_pub_unmap(jrdev, edesc, req);
>> -
>> +	req_ctx->akcipher_op_done = rsa_pub_done;
> This initialization could be moved into akcipher_enqueue_req().
> 
>> +	return akcipher_enqueue_req(jrdev, edesc->hw_desc, rsa_pub_done,
>> +				    req, edesc);
> edesc, edesc->hw_desc parameters not needed - can be deduced internally
> via req -> req_ctx -> edesc- > hw_desc.
> 
> Horia
>