Arithmetic crash occurs

[HeeillWang]

I executed fuzzing, and found arithmetic overflow bug on date.rs. Tested on 0.3.22 but would be reproduced on latest version. Here I omits detailed reproduction steps, as those cases looks quite obvious... Please let me know anyone needs more details.

Case-1 : NumericalStdDuration impl for u64

Thread '<unnamed>' panicked at 'attempt to multiply with overflow', time-0.3.22/src/ext.rs:223
Thread '<unnamed>' panicked at 'attempt to multiply with overflow', time-0.3.22/src/ext.rs:227
Thread '<unnamed>' panicked at 'attempt to multiply with overflow', time-0.3.22/src/ext.rs:231
Thread '<unnamed>' panicked at 'attempt to multiply with overflow', time-0.3.22/src/ext.rs:235

time/time/src/ext.rs

Lines 222 to 236 in c96bb1a

	fn std_minutes(self) -> StdDuration {
	StdDuration::from_secs(self * Second::per(Minute) as Self)
	}
	fn std_hours(self) -> StdDuration {
	StdDuration::from_secs(self * Second::per(Hour) as Self)
	}
	fn std_days(self) -> StdDuration {
	StdDuration::from_secs(self * Second::per(Day) as Self)
	}
	fn std_weeks(self) -> StdDuration {
	StdDuration::from_secs(self * Second::per(Week) as Self)
	}

Case-2

Thread '<unnamed>' panicked at 'attempt to divide by zero', time-0.3.22/src/duration.rs:1294

time/time/src/duration.rs

Lines 1348 to 1379 in c96bb1a

	macro_rules! duration_mul_div_int {
	($($type:ty),+) => {$(
	impl Mul<$type> for Duration {
	type Output = Self;
	fn mul(self, rhs: $type) -> Self::Output {
	Self::nanoseconds_i128(
	self.whole_nanoseconds()
	.checked_mul(rhs as _)
	.expect("overflow when multiplying duration")
	)
	}
	}
	impl Mul<Duration> for $type {
	type Output = Duration;
	fn mul(self, rhs: Duration) -> Self::Output {
	rhs * self
	}
	}
	impl Div<$type> for Duration {
	type Output = Self;
	fn div(self, rhs: $type) -> Self::Output {
	Self::nanoseconds_i128(self.whole_nanoseconds() / rhs as i128)
	}
	}
	)+};
	}
	duration_mul_div_int![i8, i16, i32, u8, u16, u32];

Expected patch

• Add checking logic with assert statement or checked operations
• Explicitly mention panic condition on docs.

[jhpratt]

Case 1 results in wrapping in release mode, which is an issue. I pushed a commit to avoid this by using checked multiplication.

Case 2 is the same as #623, as panicking on division by zero is very much intentional.

Closing as case 1 has been resolved.