curl / Mailing Lists / curl-library / Single Mail


Re: curl_multi_perform sometimes does not return while its threadus es1 00% cpu resources

From: <>
Date: Thu, 13 Dec 2018 19:54:07 +0900

thank you for your comments
Yes, your assumption is correct.
And I confirmed that time_t is defined as unsigned long in windows, which I think is equivalent to HAVE TIME T UNSIGNED.

Regarding an example source code, it's hard for me to pick up and post them and it(= "compare(i, t->key) never gets negative") occurs once in thousand times after simply calling curl_easy_perform()

I don't use libcurl multi-threaded, I mean, there are many other threads in this process, but the only one thread calls curl_easy_perform().

Thank you in advance.


2018/12/13 17:53、Daniel Stenberg <>のメール:

>> On Thu, 13 Dec 2018, "" wrote:
>> I confirmed that time_t in windows is defined as unsigned
> This document contradicts your statement:
> "If _USE_32BIT_TIME_T is defined, time_t is a long integer. If not
> defined, it is a 64-bit integer."
> Signed, in both cases.
>> , but still I’m investigating whether HAVE_TIME_T_UNSIGNED is done in windows
>> or is necessary or not in windows
> It is necessary to define for *any* system using an unsigned time_t. If
> Windows is such a system, it should be defined.
> But... as I mentioned above I don't think it is, and libcurl is used *WIDELY*
> on Windows already by lots of users who don't experience this problem you
> describe so I don't think your problem is as easily explained as this.
> I would rather urge you to provide an example source code that reproduces the
> problem. Do you use libcurl multi-threaded? Do you use the appropriate thread
> mutex callbacks (if any) ?
> --
> /

Received on 2018-12-13