What does mktime overflow with the parameter?

Question

According to standard, mktime should perform normalisation - when you want to add a minute and a half to struct tm, you add 90 seconds to tm_sec and call mktime, ignoring its return value.

What I didn't find in the standard is if the parameter is normalised even if it cannot be represented in time_t (for example having the year set to 2100).

So, is this code safe?

struct tm future;
memset(&future, 0, sizeof(future));
future.tm_mon = 1;
future.tm_sec = 90;   //I want this to be normalised by mktime
future.tm_year = 200; //but this can cause troubles

mktime(&future);

//future.tm_sec should be 30
//future.tm_min should be 1
//future.tm_year should be still 200

Answer 1

The C standard has this to say about mktime():

On successful completion, the values of the tm_wday and tm_yday components of the structure are set appropriately, and the other components are set to represent the specified calendar time, but with their values forced to the ranges indicated above; the final value of tm_mday is not set until tm_mon and tm_year are determined.

So you can only rely on the normalisation being performed if the function completed successfully.

Answer 2

You need to check the return value of mktime. You should not ignore the return value from library functions! (exception for printf and a (not so) few others).

If it returns (time_t)-1 the result is not trustworthy.

See an example at http://codepad.org/KTZwUHt0

The same code on my computer prints

time_t is 64 bits long
normalized to 2100-01-31 T 00:01:30