Re: [dpdk-dev] [EXT] [dpdk-dev v1] test/cryptodev: fix incomplete data length

[Anoob Joseph <anoobj@marvell.com>]

Hi Kai,

Also, couple of nits. Please check inline.

Thanks,
Anoob

> -----Original Message-----
> From: dev <dev-bounces@dpdk.org> On Behalf Of Anoob Joseph
> Sent: Monday, November 8, 2021 9:36 AM
> To: Kai Ji <kai.ji@intel.com>; dev@dpdk.org; Akhil Goyal
> <gakhil@marvell.com>
> Cc: pablo.de.lara.guarch@intel.com; adamx.dybkowski@intel.com;
> roy.fan.zhang@intel.com
> Subject: Re: [dpdk-dev] [EXT] [dpdk-dev v1] test/cryptodev: fix incomplete
> data length
> 
> Hi Kai,
> 
> Patch looks good. Wondering if we need same fix in functions such as "
> test_zuc_auth_cipher()".
> 
> We were also hitting this issue when we enabled few additional features in
> Marvell PMDs. Upon investigation, we realized that this issue would come up
> for certain packet size combinations if the padded lengths are not same. We
> observed the issue only with test_mixed_auth_cipher(), which is getting
> addressed with this patch. Just wondering if you have checked whether
> other places also would need a fix.
> 
> Thanks,
> Anoob
> 
> > -----Original Message-----
> > From: dev <dev-bounces@dpdk.org> On Behalf Of Kai Ji
> > Sent: Friday, November 5, 2021 9:12 PM
> > To: dev@dpdk.org
> > Cc: Kai Ji <kai.ji@intel.com>; pablo.de.lara.guarch@intel.com;
> > adamx.dybkowski@intel.com
> > Subject: [EXT] [dpdk-dev] [dpdk-dev v1] test/cryptodev: fix incomplete
> > data length
> >
> > External Email
> >
> > ----------------------------------------------------------------------
> > This patch fixes incorrect data lengths computation in cryptodev unit test.
> > Previously some data lengths were incorrectly set, which was
> > insensitive for crypto op unit tets but is critical for raw data path
> > API unit tests. The patch addressed the issue by setting the correct data
> lengths for some tests.
> >
> > Fixes: 681f540da52b ("cryptodev: do not use AAD in wireless
> > algorithms")
> > Cc: pablo.de.lara.guarch@intel.com
> >
> > Fixes: e847fc512817 ("test/crypto: add encrypted digest case for
> > AES-CTR-
> > CMAC")
> > Cc: adamx.dybkowski@intel.com
> >
> > Signed-off-by: Kai Ji <kai.ji@intel.com>
> > ---
> >  app/test/test_cryptodev.c | 16 +++++++++-------
> >  1 file changed, 9 insertions(+), 7 deletions(-)
> >
> > diff --git a/app/test/test_cryptodev.c b/app/test/test_cryptodev.c
> > index
> > 52457596e2..b926412742 100644
> > --- a/app/test/test_cryptodev.c
> > +++ b/app/test/test_cryptodev.c
> > @@ -4102,9 +4102,9 @@ test_kasumi_decryption(const struct
> > kasumi_test_data *tdata)
> >
> >  	/* Create KASUMI operation */
> >  	retval = create_wireless_algo_cipher_operation(tdata-
> > >cipher_iv.data,
> > -					tdata->cipher_iv.len,
> > -					tdata->ciphertext.len,
> > -					tdata->validCipherOffsetInBits.len);
> > +			tdata->cipher_iv.len,
> > +			RTE_ALIGN_CEIL(tdata->validCipherLenInBits.len, 8),
> > +			tdata->validCipherOffsetInBits.len);
> >  	if (retval < 0)
> >  		return retval;
> >
> > @@ -7335,6 +7335,7 @@ test_mixed_auth_cipher(const struct
> > mixed_cipher_auth_test_data *tdata,
> >  	unsigned int plaintext_len;
> >  	unsigned int ciphertext_pad_len;
> >  	unsigned int ciphertext_len;
> > +	unsigned int data_len;
> >
> >  	struct rte_cryptodev_info dev_info;
> >  	struct rte_crypto_op *op;
> > @@ -7395,21 +7396,22 @@ test_mixed_auth_cipher(const struct
> > mixed_cipher_auth_test_data *tdata,
> >  	plaintext_len = ceil_byte_length(tdata->plaintext.len_bits);
> >  	ciphertext_pad_len = RTE_ALIGN_CEIL(ciphertext_len, 16);
> >  	plaintext_pad_len = RTE_ALIGN_CEIL(plaintext_len, 16);
> > +	data_len = RTE_MAX(ciphertext_pad_len, plaintext_pad_len);

[Anoob] Isn't ciphertext_pad_len guaranteed to be the larger one of the two? Do we need another variable and the RTE_MAX?
 
> >
> >  	if (verify) {
> >  		ciphertext = (uint8_t *)rte_pktmbuf_append(ut_params-
> > >ibuf,
> > -				ciphertext_pad_len);
> > +				data_len);
> >  		memcpy(ciphertext, tdata->ciphertext.data, ciphertext_len);
> >  		if (op_mode == OUT_OF_PLACE)
> > -			rte_pktmbuf_append(ut_params->obuf,
> > ciphertext_pad_len);
> > +			rte_pktmbuf_append(ut_params->obuf, data_len);
> >  		debug_hexdump(stdout, "ciphertext:", ciphertext,
> >  				ciphertext_len);
> >  	} else {
> >  		plaintext = (uint8_t *)rte_pktmbuf_append(ut_params-
> > >ibuf,
> > -				plaintext_pad_len);
> > +				data_len);
> >  		memcpy(plaintext, tdata->plaintext.data, plaintext_len);
> >  		if (op_mode == OUT_OF_PLACE)
> > -			rte_pktmbuf_append(ut_params->obuf,
> > plaintext_pad_len);
> > +			rte_pktmbuf_append(ut_params->obuf, data_len);

[Anoob] Now that more things are common across the branches, can we move out some bits outside the if condition? Like, the above line is definitely same and be kept outside condition. The append call prior to this can also be kept common if we can rename the local variable.
 
> >  		debug_hexdump(stdout, "plaintext:", plaintext,
> plaintext_len);
> >  	}
> >
> > --
> > 2.17.1