[dpdk-dev] [PATCH v5 2/2] test/test: add unit test for CRC computation

[De Lara Guarch, Pablo]

Hi Jasvinder,


> -----Original Message-----
> From: Singh, Jasvinder
> Sent: Tuesday, March 21, 2017 2:46 PM
> To: dev at dpdk.org
> Cc: olivier.matz at 6wind.com; Doherty, Declan; De Lara Guarch, Pablo
> Subject: [PATCH v5 2/2] test/test: add unit test for CRC computation
> 
> This patch provides a set of tests for verifying the functional
> correctness of 16-bit and 32-bit CRC APIs.
> 
> Signed-off-by: Jasvinder Singh <jasvinder.singh at intel.com>
> ---
>  test/test/Makefile   |   2 +
>  test/test/test_crc.c | 223
> +++++++++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 225 insertions(+)
>  create mode 100644 test/test/test_crc.c
> 

> diff --git a/test/test/test_crc.c b/test/test/test_crc.c
> new file mode 100644
> index 0000000..2eb0bff
> --- /dev/null
> +++ b/test/test/test_crc.c
> @@ -0,0 +1,223 @@


> +
> +#include "test.h"
> +
> +#define CRC_VEC_LEN				32

Unnecessary extra tab.

> +#define CRC32_VEC_LEN1			1512
> +#define CRC32_VEC_LEN2			348
> +#define CRC16_VEC_LEN1			12
> +#define CRC16_VEC_LEN2			2
> +#define LINE_LEN 75

I would align this to the other values above.

...

> +
> +/* 16-bit CRC test vector 1*/

Extra space before "*/".

> +static const uint8_t crc16_vec1[CRC16_VEC_LEN1] = {
> +	0x0D, 0x01, 0x01, 0x23, 0x45, 0x67,
> +	0x89, 0x01, 0x23, 0x45, 0x00, 0x01,
> +};
> +
> +/* 16-bit CRC test vector 2*/

Same here.

> +static const uint8_t crc16_vec2[CRC16_VEC_LEN2] = {
> +	0x03, 0x3f,
> +};
> +/** CRC results */
> +uint32_t crc32_vec_res, crc32_vec1_res, crc32_vec2_res;
> +uint32_t crc16_vec_res, crc16_vec1_res, crc16_vec2_res;
> +
> +static int
> +crc_calc(const uint8_t *vec,
> +	uint32_t vec_len,
> +	enum rte_net_crc_type type)
> +{
> +	/* compute CRC */
> +	uint32_t ret = rte_net_crc_calc(vec, vec_len, type);
> +
> +	/* dump data on console*/

Same here.

> +	rte_hexdump(stdout, NULL, vec, vec_len);

I would use TEST_HEXDUMP, which dumps the stream only if debug is enabled (to avoid too much output).

> +
> +	return  ret;
> +}
> +
> +static void
> +crc_calc_scalar(void)
> +{
> +	uint32_t i;
> +	enum rte_net_crc_type type;
> +	uint8_t *test_data;
> +
> +	/* 32-bit ethernet CRC: scalar result */
> +	type = RTE_NET_CRC32_ETH;
> +	crc32_vec_res = crc_calc(crc_vec,
> +						CRC_VEC_LEN,
> +						type);

Remove unnecessary tabs (more of this in the rest of the file).

> +
> +	/* 32-bit ethernet CRC scalar result*/
> +	test_data = rte_zmalloc(NULL, CRC32_VEC_LEN1, 0);
> +
> +	for (i = 0; i < CRC32_VEC_LEN1; i += 12)
> +	rte_memcpy(&test_data[i], crc32_vec1, 12);

Wrong indentation (more of this in the rest of the file).
Also, why not just one memcpy call, with CRC32_VEC_LEN1.
A comment would be good to explain why.

...

> +static int
> +test_crc(void) {

Looks strange to start the parameters in a new line.
I would say you can start in the same line as the function name.

> +	uint32_t i;
> +	enum rte_net_crc_type type;
> +	uint8_t *test_data;
> +	uint32_t ret;
> +
> +	/* set crc scalar mode */

Probably the references of "crc" in comments should be uppercase "CRC".

> +	rte_net_crc_set_alg(RTE_NET_CRC_SCALAR);
> +	crc_calc_scalar();
> +
> +	/* set crc sse4.2 mode */
> +	rte_net_crc_set_alg(RTE_NET_CRC_SSE42);
> +
> +	/* 32-bit ethernet CRC: Test 1 */
> +	type = RTE_NET_CRC32_ETH;
> +
> +	ret = crc_calc(crc_vec,
> +		CRC_VEC_LEN,
> +		type);
> +	if (ret != crc32_vec_res) {
> +		printf("test_crc(32-bit): test1 failed !\n");

Extra space before "!" (keep consistency with the other printfs).

> +		return -1;
> +	}
> +
> +	/* 32-bit ethernet CRC: Test 2 */
> +	test_data = rte_zmalloc(NULL, CRC32_VEC_LEN1, 0);

Free the memory after the test (also for other calls).