Subtraction overflow in BitVec::last_one() and last_zero()

[VilleHallivuori]

It seems that there is subraction overflow in BitVec last_one()/last_zero() on cases where there are not matches. Counting works properly.

Example to reproduce:

#[test]
fn test_bitvec_last_ones() {
    let bv = bitvec![0];
    assert_eq!(0, bv.count_ones());
    println!("bv={:?}", bv);
    assert_eq!(None, bv.last_one()); // <-- Causes underflow
}

Result:

---- test_bitvec_last_ones stdout ----
bv=BitVec<bitvec::order::Lsb0, usize> { addr: 0x7f7fcc0009a0, head: 000000, bits: 1, capacity: 64 } [0]
thread 'test_bitvec_last_ones' panicked at 'attempt to subtract with overflow', /home/vhallivu/.cargo/registry/src/github.com-1ecc6299db9ec823/bitvec-0.22.0/src/slice/specialization.rs:200:21
stack backtrace:
   0: rust_begin_unwind
             at /rustc/2fd73fabe469357a12c2c974c140f67e7cdd76d0/library/std/src/panicking.rs:493:5
   1: core::panicking::panic_fmt
             at /rustc/2fd73fabe469357a12c2c974c140f67e7cdd76d0/library/core/src/panicking.rs:92:14
   2: core::panicking::panic
             at /rustc/2fd73fabe469357a12c2c974c140f67e7cdd76d0/library/core/src/panicking.rs:50:5
   3: bitvec::slice::specialization::<impl bitvec::slice::BitSlice<bitvec::order::Lsb0,T>>::sp_iter_ones_last
             at /home/vhallivu/.cargo/registry/src/github.com-1ecc6299db9ec823/bitvec-0.22.0/src/slice/specialization.rs:200:6
   4: <bitvec::slice::iter::IterOnes<O,T> as core::iter::traits::double_ended::DoubleEndedIterator>::next_back
             at /home/vhallivu/.cargo/registry/src/github.com-1ecc6299db9ec823/bitvec-0.22.0/src/slice/iter.rs:2405:4
   5: bitvec::slice::BitSlice<O,T>::last_one
             at /home/vhallivu/.cargo/registry/src/github.com-1ecc6299db9ec823/bitvec-0.22.0/src/slice.rs:1209:3
   6: bitvec_test::test_bitvec_last_ones
             at ./src/main.rs:12:22
   7: bitvec_test::test_bitvec_last_ones::{{closure}}
             at ./src/main.rs:8:1
   8: core::ops::function::FnOnce::call_once
             at /home/vhallivu/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/ops/function.rs:227:5
   9: core::ops::function::FnOnce::call_once
             at /rustc/2fd73fabe469357a12c2c974c140f67e7cdd76d0/library/core/src/ops/function.rs:227:5
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.


failures:
    test_bitvec_last_ones

[myrrlyn]

... Huh.

This crash makes complete sense, but the fix absolutely does not.

I am now issuing backports for all the minor series currently live. Thank you for the report! Please re-open this issue if you encounter this again.