Re: [dpdk-dev] [dpdk-stable] [PATCH v3] app/testpmd: fix support of hex string parser for flow API

[Ferruh Yigit <ferruh.yigit@intel.com>]

On 3/25/2019 9:25 AM, Zhao1, Wei wrote:
> Hi，Ferruh
> 
>> -----Original Message-----
>> From: Yigit, Ferruh
>> Sent: Monday, March 25, 2019 4:46 PM
>> To: Zhao1, Wei <wei.zhao1@intel.com>; dev@dpdk.org
>> Cc: stable@dpdk.org; stephen@networkplumber.org; Ananyev, Konstantin
>> <konstantin.ananyev@intel.com>
>> Subject: Re: [dpdk-stable] [PATCH v3] app/testpmd: fix support of hex string
>> parser for flow API
>>
>> On 3/25/2019 3:39 AM, Zhao1, Wei wrote:
>>> Hi,
>>>
>>>> -----Original Message-----
>>>> From: Yigit, Ferruh
>>>> Sent: Friday, March 22, 2019 10:56 PM
>>>> To: Zhao1, Wei <wei.zhao1@intel.com>; dev@dpdk.org
>>>> Cc: stable@dpdk.org; stephen@networkplumber.org; Ananyev,
>> Konstantin
>>>> <konstantin.ananyev@intel.com>
>>>> Subject: Re: [dpdk-stable] [PATCH v3] app/testpmd: fix support of hex
>>>> string parser for flow API
>>>>
>>>> On 3/22/2019 3:15 AM, Wei Zhao wrote:
>>>>> There is need for users to set configuration of HEX number for RSS
>>>>> key. The key byte should be pass down as hex number not as char
>>>>> string. This patch enable cmdline flow parse HEX number, in order to
>>>>> not using string which pass ASIC number.
>>>>>
>>>>> Fixes: f4d623f96119 ("app/testpmd: fix missing RSS fields in flow
>>>>> action")
>>>>> Cc: stable@dpdk.org
>>>>>
>>>>> Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
>>>>> Tested-by: Peng Yuan <yuan.peng@intel.com>
>>>>
>>>> <...>
>>>>
>>>>> @@ -4475,6 +4486,138 @@ parse_string(struct context *ctx, const
>>>>> struct
>>>> token *token,
>>>>>  	return -1;
>>>>>  }
>>>>>
>>>>> +static uint32_t
>>>>> +get_hex_val(char c)
>>>>> +{
>>>>> +	switch (c) {
>>>>> +	case '0': case '1': case '2': case '3': case '4': case '5':
>>>>> +	case '6': case '7': case '8': case '9':
>>>>> +		return c - '0';
>>>>> +	case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
>>>>> +		return c - 'A' + 10;
>>>>> +	case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
>>>>> +		return c - 'a' + 10;
>>>>> +	default:
>>>>> +		return 0;
>>>>> +	}
>>>>> +}
>>>>> +
>>>>> +static int
>>>>> +parse_hex_string(const char *src, uint8_t *dst, uint32_t *size) {
>>>>> +	const char *c;
>>>>> +	uint32_t i;
>>>>> +
>>>>> +	/* Check input parameters */
>>>>> +	if ((src == NULL) ||
>>>>> +		(dst == NULL) ||
>>>>> +		(size == NULL) ||
>>>>> +		(*size == 0))
>>>>> +		return -1;
>>>>> +	if ((*size & 1) != 0)
>>>>> +		return -1;
>>>>> +
>>>>> +	for (c = src, i = 0; i < *size; c++, i++) {
>>>>> +		if (isxdigit(*c))
>>>>> +			continue;
>>>>> +		else
>>>>> +			return -1;
>>>>> +	}
>>>>> +
>>>>> +	*size = *size / 2;
>>>>> +
>>>>> +	/* Convert chars to bytes */
>>>>> +	for (i = 0; i < *size; i++)
>>>>> +		dst[i] = get_hex_val(src[2 * i]) * 16 +
>>>>> +			get_hex_val(src[2 * i + 1]);
>>>>> +
>>>>> +	return 0;
>>>>> +}
>>>>
>>>> I can see this has been discussed already but what would you think
>>>> updating the 'parse_hex_string' something like following, it is less code to
>> maintain:
>>>>
>>>> static int
>>>> parse_hex_string(const char *src, uint8_t *dst, uint32_t *size) {
>>>>   int len;
>>>>   int i
>>>>   for (i = 0, len = 0; i < *size; i += 2) {
>>>>     char tmp[3];
>>>>     snprintf(tmp, 3, src + i);
>>>>     dst[len++] = strtoul(tmp, NULL, 16);
>>>>   }
>>>>   dst[len] = 0;
>>>>   *size = len;
>>>>   return 0;
>>>> }
>>>>
>>>> (indeed with better error checking on strtoul ;) )
>>>
>>>
>>> Why delete these check from parse_hex_string()?
>>
>> The point is using 'strtoul' instead of your functions, so that you won't need
>> 'get_hex_val()' at all, or won't need 'isxdigit()' because 'strtoul' will check it,
>> won't need size should be multiply of two restriction '(*size & 1)' because of
>> implementation change. Probably you will need NULL checks, but again point
>> is why not using 'strtoul' instead of writing your version of it?
> 
> Yes, we can use 'strtoul' , but my point is that  I think we need these check code even  if we use the code 'strtoul' .
> isxdigit(*c)) is need because  *c may be sring "0xrgh" which is not hex.
> If we use strtoul  will return 0 for that ,we can not distinguish between error or input is zero.  
> '(*size & 1) can be delete, I agree.
> 
> 	/* Check input parameters */
> 	if ((src == NULL) ||
> 	(dst == NULL) ||
> 		(size == NULL) ||
> 		(*size == 0))
> 		return -1;
> 	for (c = src, i = 0; i < *size; c++, i++) {
> 		if (isxdigit(*c))
> 			continue;
> 		else
> 			return -1;
> 	}
> 

Please feel free to keep/add whatever check is required, I wasn't suggesting the
final implementation.

'strtol' can detect invalid chars, by providing non-NULL to second argument, so
'isxdigit()' is not required but more checks needed.

Overall you previously mentioned 'strtol' can't be used, but it can be used and
I believe it is better way to go, but I am asking what do you think about it?
Checks and implementation details can be handled.

> 
>>>
>>> 	/* Check input parameters */
>>> 	if ((src == NULL) ||
>>> 	(dst == NULL) ||
>>> 		(size == NULL) ||
>>> 		(*size == 0))
>>> 		return -1;
>>> 	if ((*size & 1) != 0)
>>> 		return -1;
>>> 	for (c = src, i = 0; i < *size; c++, i++) {
>>> 		if (isxdigit(*c))
>>> 			continue;
>>> 		else
>>> 			return -1;
>>> 	}
>>>
>>>
>>>
>>>>
>>>> <...>
>>>>
>>>>> +	/* Output buffer is not necessarily NUL-terminated. */
>>>>> +	memcpy(buf, hex_tmp, hexlen);
>>>>> +	memset((uint8_t *)buf + len, 0x00, size - hexlen);
>>>>
>>>> Can't this overflow the 'buf'? since "len = 2 * hexlen"
>>>> I guess intention is "buf + hexlen"
>