[dpdk-dev] [PATCH v3 1/1] fbarray: get fbarrays from containerized secondary

Yasufumi Ogawa yasufum.o at gmail.com
Fri Jul 12 04:22:37 CEST 2019


On 2019/07/11 22:14, Burakov, Anatoly wrote:
> On 11-Jul-19 12:57 PM, Yasufumi Ogawa wrote:
>> On 2019/07/11 19:53, Burakov, Anatoly wrote:
>>> On 11-Jul-19 11:31 AM, yasufum.o at gmail.com wrote:
>>>> From: Yasufumi Ogawa <ogawa.yasufumi at lab.ntt.co.jp>
>>>>
>>> <...>
>>>
>>>> +    if (getpid() == 1) {
>>>> +        FILE *hn_fp;
>>>> +        hn_fp = fopen("/etc/hostname", "r");
>>>> +        if (hn_fp == NULL) {
>>>> +            RTE_LOG(ERR, EAL,
>>>> +                "Cannot open '/etc/hostname' for secondary\n");
>>>> +            return -1;
>>>> +        }
>>>> +
>>>> +        /* with docker, /etc/hostname just has one entry of 
>>>> hostname */
>>>> +        if (fscanf(hn_fp, "%s", proc_id) == EOF) {
>>>
>>> Apologies for not pointing this out earlier, but do i understand 
>>> correctly that there's no bounds checking here, and fscanf() will 
>>> write however many bytes it wants?
>> I understand "%s" is not appropriate. hostname is 12 bytes char and I 
>> thought proc_id[16] is enough, but it is unsafe. In addition, hostname 
>> can be defined by user with docker's option, so it should be enough 
>> for user defined name.
>>
>> How do you think expecting max 32 chars of hostname and set boundary 
>> "%32s" as following?
>>
>>      proc_id[33];  /* define proc id from hostname less than 33 bytes. */
>>      ...
>>      if (fscanf(hn_fp, "%32s", proc_id) == EOF) {
>>
> 
> As long as it takes NULL-termination into account as well, it should be 
> OK. I can't recall off the top of my head if %32s includes NULL 
> terminator (probably not?).
Do you agree if initialize with NULL chars to ensure proc_id is 
NULL-terminated? As tested on my environment, "%Ns" sets next of Nth 
char as NULL, but it seems more reliable.
     proc_id[33] = { 0 };

Yasufumi


More information about the dev mailing list