Re: [PATCH v3] Add cancellation regression test for getpwuid_r.

[Carlos O'Donell <carlos at redhat dot com>]

On 12/08/2016 07:13 PM, Carlos O'Donell wrote:
> On 12/08/2016 02:02 PM, Mike Frysinger wrote:
>> On 08 Dec 2016 12:15, Carlos O'Donell wrote:
>>> +static void *
>>> +worker( void *arg )
>>
>> seems like this test case is mingling code from diff sources since it has
>> diff styles.  there's a number of non-GNU style errors in it -- this is
>> just the first one i saw.
> 
> I started with an uncontributed internal Red Hat reproducer, but all that
> was left by the time I was done was a few lines, which you're right, have
> all the wrong style.
> 
>>> +{
>>> +  int ret;
>>> +  unsigned int iter = 0;
>>> +  struct passwd pwbuf, *pw;
>>> +  uid_t uid;
>>> +  uid = geteuid();
>>> +  /* Use a reasonable sized buffer.  Note that _SC_GETPW_R_SIZE_MAX is
>>> +     just a hint and not any kind of maximum value.  */
>>> +  wbufsz = sysconf(_SC_GETPW_R_SIZE_MAX);
>>> +  if (wbufsz == -1)
>>> +    wbufsz = 1024;
>>> +  wbuf = xmalloc (wbufsz);
>>> +  pthread_cleanup_push (worker_free, wbuf);
>>> +  sem_post (&started);
>>> +  while (1)
>>
>> this func could do with a few newlines sprinkled around.  it's pretty
>> dense to read as-is.  actually the whole testcase is written this way :).
>>
>>> +  printf ("INFO: Previoulsy we would never get here\n");
>>
>> "Previously"
>>
>>> +  printf ("PASS: Cancelled getpwuid_r successfully"
>>
>> "Canceled"
>> -mike
>>
> 
> v3
> - Fix two spelling mistakes.
> - Fix GNU Coding Style violations.
> - Add more whitespace to make the test more readable.

I believe v3 addresses Florian and Mike's review.

I'll commit this next week if nobody objects.

-- 
Cheers,
Carlos.