Re: [Qemu-devel] [PATCH v3 3/3] slirp: set mainloop timeout with more precise value

[Jan Kiszka]

On 2013-08-21 10:07, liu ping fan wrote:
> On Wed, Aug 21, 2013 at 3:36 PM, Alex Bligh <address@hidden> wrote:
>>
>>
>> --On 21 August 2013 10:15:52 +0800 Liu Ping Fan <address@hidden> wrote:
>>
>>> -void slirp_update_timeout(uint32_t *timeout)
>>> +static void slirp_update_timeout(uint32_t *timeout)
>>>  {
>>> -    if (!QTAILQ_EMPTY(&slirp_instances)) {
>>> -        *timeout = MIN(1000, *timeout);
>>
>>
>> If you are putting things in macros, you might as well change that
> 
> TIMEOUT_FAST/SLOW have definite meaning, and used more than one place
> in the code. For 1000ms, I do not know this magic value's meaning, but
> whatever, it just occurs once. So there is no trouble to read the
> code.

You could name it ONE_SEC or so. Can be done as trivial patch on top.

IIRC, slirp requires regular polling for the aging of certain requests
like DNS.

Jan

> 
>> 1000 as well, and hopefully comment why that particular magic value
>> is there.
>>
>>
>>> +    Slirp *slirp;
>>> +    uint32_t t;
>>> +
>>> +    *timeout = MIN(1000, *timeout);
>>> +    if (*timeout <= TIMEOUT_FAST) {
>>> +        return;
>>> +    }
>>> +    t = *timeout;
>>
>>
>>
>>
>> --
>> Alex Bligh

-- 
Siemens AG, Corporate Technology, CT RTC ITP SES-DE
Corporate Competence Center Embedded Linux