openshwprojects
Some more fixes to usage of gmtime
After recent changes to fix usage of time_t, there were 4 other places, which should be addressed:
max@max:~/LN882H/20240326/OpenBK7231T_App$ find . -name "*.[ch]" | xargs grep -C 2 -n "gmtime((time_t"
./src/driver/drv_max72xx_clock.c-71-
./src/driver/drv_max72xx_clock.c-72- // NOTE: on windows, you need _USE_32BIT_TIME_T
./src/driver/drv_max72xx_clock.c:73: ltm = gmtime((time_t*)&g_ntpTime);
./src/driver/drv_max72xx_clock.c-74-
./src/driver/drv_max72xx_clock.c-75- if (ltm == 0) {
--
./src/driver/drv_ntp_events.c-172-
./src/driver/drv_ntp_events.c-173- // NOTE: on windows, you need _USE_32BIT_TIME_T
./src/driver/drv_ntp_events.c:174: ltm = gmtime((time_t*)&runTime);
./src/driver/drv_ntp_events.c-175-
./src/driver/drv_ntp_events.c-176- if (ltm == 0) {
--
./src/driver/drv_ntp_events.c-313- uint8_t hour_b, minute_b;
./src/driver/drv_ntp_events.c-314- int sunflags = 0;
./src/driver/drv_ntp_events.c:315: struct tm *ltm = gmtime((time_t*) &ntp_eventsTime);
./src/driver/drv_ntp_events.c-316-#endif
./src/driver/drv_ntp_events.c-317-
--
./src/httpserver/json_interface.c-655- struct tm* ltm;
./src/httpserver/json_interface.c-656- int ntpTime = NTP_GetCurrentTime() - g_secondsElapsed;
./src/httpserver/json_interface.c:657: ltm = gmtime((time_t*)&ntpTime);
./src/httpserver/json_interface.c-658-
./src/httpserver/json_interface.c-659- if (ltm != 0) {
max@max:~/LN882H/20240326/OpenBK7231T_App$
The first one (in /src/driver/drv_max72xx_clock.c) is just a forgotten removal of the (now obsolete) typecast.
The second and third in src/driver/drv_ntp_events.c will need some consideration: "runtime" is a parameter ("void NTP_RunEventsForSecond(unsigned int runTime) {") so it would need to alter the function call to use "time_t runTime" or we can just use the working typecast to a time_t var here (like in the "quick and dirty solution)
An alternate "fix" could be to define "ntp_eventsTime" as time_t, too, this will resolve both issues (since "runTime" is called with "ntp_eventsTime", so we can change both to "time_t". On the other hand, all calculations with times are based on "unsigned int", so they would need attention, to. So I used the first approach, to cast "unsigned in" to time_t, which is of same or even larger type, so casting shold be o.k.
For the third occurrence I propose something like my "solution" from issue again: adding a test, if NTP is synced - this avoids calling it when NTP is not synced (resulting in a negative "ntpTime") define ntpTime as time_t her, too (doing the casting inside the calculation and we can fix using the local time instead of UTC time at the same time ;-)
