Panic/return false on overflow in no_threads read/try_read impl

[asder8215]

As per discussion with Mark in #153555, it's possible for no_threads impl of RwLock to trigger a silent overflow on RwLock::read/RwLock::try_read if we try to acquire more than isize::MAX read locks. This PR adds an explicit panic/return false when our read lock counter is at isize::MAX for RwLock::read/RwLock::try_read; the message is similar to that of sys/sync/rwlock/futex.rs here.

[rustbot]

r? @Mark-Simulacrum

rustbot has assigned @Mark-Simulacrum.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: @ChrisDenton, libs
• @ChrisDenton, libs expanded to 8 candidates

[Mark-Simulacrum]

Let's write this as m.checked_add(1).expect("rwlock overflowed read locks"), and same with below.

I would also move the subtraction in the unlock code to do the same. It's not as necessary, but I think it would be good to confirm we are in fact holding a read lock and should be cheap for this implementation.

[asder8215]

Done.

For RwLock::read_unlock, I used mem::replace, just like in RwLock::write_unlock,and checked if the old value is greater than 0 to see if it was read locked, asserting an error if it wasn't. I added an error msg for RwLock::write and RwLock::downgrade.

However, I do wonder if it's better if we use an assertion check to see if the lock is read/write locked, and then set the RwLock mode accordingly instead of mem::replace since we probably shouldn't mem::replace the mode value with 0 if in RwLock::read_unlock/RwLock::write_unlock if it was write locked/read locked.

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.