From: Rafal Somla
Date: September 10 2010 11:57am
Subject: Re: bzr commit into mysql-5.5 branch (Georgi.Kodinov:3082) WL#5366
List-Archive: http://lists.mysql.com/commits/117998
Message-Id: <4C8A1D31.7090100@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

Hello Joro,

I think I spot another problem with your code. I discovered it when I tried 
to use similar trick in my own code...

Georgi Kodinov wrote:
> #At file:///home/kgeorge/mysql/work/wl5366-5.5/ based on revid:georgi.kodinov@stripped
...
> +
> +  do
> +  {
> +    rc= getpwnam_r(name, pwd_arg, buf, buf_size, &ret);
> +
> +    if (rc == ERANGE)
> +    {
> +      /*
> +        The buffer specified by the buffer and bufsize argu-
> +        ments was insufficiently sized to store the result.
> +        The caller should retry with a larger buffer.
> +      */
> +      buf= (char *) ((*to_free) ? 
> +        realloc(*to_free, buf_size * 2) :
> +        malloc(buf_size * 2));
> +
> +      if (!buf)
> +      {
> +        free(*to_free);
> +        *to_free= NULL;
> +        return NULL;
> +      }
> +
> +      *to_free= buf;
> +      buf_size*= 2;
> +      continue;

I think the intention here is that the loop will repeat, so that a call to 
getpwnam_r() with a bigger buffer will be made again. But continue will move 
to the end of the loop and terminate it, since condition is always false...

> +    }
> +  } while (0);
> +
> +  if (rc)
> +  {
> +    free(to_free);
> +    *to_free= NULL;
> +    return NULL;
> +  }
> +
> +  return ret;
> +#else
> +  return NULL;
> +#endif
> +}

Rafal