On Fri, 21 Jul 2023 at 10:09, Philippe Mathieu-Daudé <phi...@linaro.org> wrote: > > On 20/7/23 17:58, Peter Maydell wrote: > > In the m48t59 device we almost always use 64-bit arithmetic when > > dealing with time_t deltas. The one exception is in set_alarm(), > > which currently uses a plain 'int' to hold the difference between two > > time_t values. Switch to int64_t instead to avoid any possible > > overflow issues. > > > > Signed-off-by: Peter Maydell <peter.mayd...@linaro.org> > > --- > > hw/rtc/m48t59.c | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > Similarly: > > -- >8 -- > --- a/hw/rtc/m48t59.c > +++ b/hw/rtc/m48t59.c > @@ -88,7 +88,7 @@ static M48txxInfo m48txx_sysbus_info[] = { > static void alarm_cb (void *opaque) > { > struct tm tm; > - uint64_t next_time; > + int64_t next_time; > M48t59State *NVRAM = opaque;
Why? next_time should always be positive here, I think. -- PMM