Some panics found by afl.rs

[StevenJiang1110]

I have used afl.rs to fuzz this crate, and I found several cases will crash. The version I work on is version 0.2.24.

There are 2 about unreachable.

let date_ = time::Date::try_from_ymd(-36 ,11 ,1).unwrap();
let _ = time::Date::weekday(date_);

let date_ = time::Date::try_from_yo(-26 ,96).unwrap();
let _ = time::Date::weekday(date_);

And there are several about arithmetic overflow.

let _ = time::UtcOffset::west_minutes(32768);

let _ = time::util::weeks_in_year(-102);

let date_ = time::Date::try_from_yo(-31 ,136).unwrap();
let date2_ = time::Date::next_day(date_);
let _ = time::Date::iso_year_week(date2_);

let date_ = time::Date::try_from_yo(-60, 64).unwrap();
let date2_ = time::Date::previous_day(date_);
let _ = time::Date::week(date2_);

let date_ = time::Date::try_from_yo(-208 ,99).unwrap();
let date2_ = time::Date::try_with_hms_milli(date_, 22, 48,48 ,48).unwrap();
let _ = time::PrimitiveDateTime::week(date2_);

I hope you can check if these are really bugs that need to be fixed. Thanks a lot.

[jhpratt]

I'll verify these once I'm on my laptop, but on the assumption everything you've said is true, these are mostly bugs. The only one that would not be is the construction of the UtcOffset with an out of bounds minute value; this is explicitly mentioned in the documentation, so a panic is valid behavior.

Would you mind sharing the code used to find these? I've never used a fuzzer myself. This would also let me run it on the current head, where significant changes have taken place across the board.

[jhpratt]

Verifying the following as bugs in 0.2.24:

• date!(-36-11-01).weekday()
• date!(-26-96).weekday()
• util::weeks_in_year(-102)
• date!(-31-137).weekday()
• date!(-31-137).week()
• date!(-60-63).iso_year_week()
• date!(-208-99).week()

I'll be investigating these further, of course. First up is to see if these bugs exist on main; I know one won't, as I (apparently incorrectly) assumed valid data. Needless to say this should be interesting.

[jhpratt]

Fixed in 0540d19.

[jhpratt]

v0.2.25 has been released with this patch. I'd also like to report that this issue does not exist on main.

[StevenJiang1110]

Thanks for your prompt reply. That's very helpful.
I would like to issue a pr about the fuzzing codes to find the bugs I found(I'm working on ddl now, so maybe several days later). They are using afl.rs.