Re: [dpdk-dev] [PATCH v3] app/proc-info: fix port mask parse issue

["Burakov, Anatoly" <anatoly.burakov@intel.com>]

On 16-Nov-18 10:15 AM, Burakov, Anatoly wrote:
> On 07-Nov-18 6:10 AM, Li Han wrote:
>> parse_portmask return type is int,but global variable
>> "enabled_port_mask" type is uint32_t.so in proc_info_parse_args
>> function,when parse_portmask return -1,"enabled_port_mask" will
>> get a huge value and "if (enabled_port_mask == 0)" will never happen.
>>
>> Fixes: 22561383ea17 ("app: replace dump_cfg by proc_info")
>> Signed-off-by: Li Han <han.li1@zte.com.cn>
>>
>> ---
>> v3:
>> *fix commit meassges issue
>> v2:
>> *fix typecast issue
>> ---
>>   app/proc-info/main.c | 9 +++------
>>   1 file changed, 3 insertions(+), 6 deletions(-)
>>
>> diff --git a/app/proc-info/main.c b/app/proc-info/main.c
>> index c20effa..650d599 100644
>> --- a/app/proc-info/main.c
>> +++ b/app/proc-info/main.c
>> @@ -37,7 +37,7 @@
>>   #define MAX_STRING_LEN 256
>>   /**< mask of enabled ports */
>> -static uint32_t enabled_port_mask;
>> +static uint64_t enabled_port_mask;
>>   /**< Enable stats. */
>>   static uint32_t enable_stats;
>>   /**< Enable xstats. */
>> @@ -90,7 +90,7 @@
>>   /*
>>    * Parse the portmask provided at run time.
>>    */
>> -static int
>> +static unsigned long
>>   parse_portmask(const char *portmask)
>>   {
>>       char *end = NULL;
>> @@ -103,12 +103,9 @@
>>       if ((portmask[0] == '\0') || (end == NULL) || (*end != '\0') ||
>>           (errno != 0)) {
>>           printf("%s ERROR parsing the port mask\n", __func__);
>> -        return -1;
>> +        return 0;
>>       }
>> -    if (pm == 0)
>> -        return -1;
>> -
>>       return pm;
>>   }
>>
> 
> Hi,
> 
> This fix appears wrong. If you're making the value uint64_t, you cannot 
> encode errors in the value. So, it's better to leave the return type as 
> int, return 0 or -1 on success/error, and store the parsed result in a 
> pointer passed to the function instead. Something like this:
> 
> static int
> parse_portmask(const char *portmask, uint64_t *mask)
> {
> ...
>      if (pm == 0)
>      return -1;
>      *mask = pm;
>      return 0;
> }
> 

Also, another thing to note. Unsigned long is 32-bit on 32-bit Linux, so 
if you're going to have uint64_t data, you should correct the parsing to 
use strtoull() instead.

-- 
Thanks,
Anatoly