qp-trie's count is incorrect and causes subtraction overflow in debug mode

[kevinaboos]

This weird error is revealed when trying to run any application on the shell in a debug build of Theseus.

I changed the impl Drop for AppCrateRef drop handler to print a bit more info:

impl Drop for AppCrateRef {
    fn drop(&mut self) {
        // trace!("### Dropping AppCrateRef {:?} from namespace {:?}", self.crate_ref, self.namespace.name());
        let crate_locked = self.crate_ref.lock_as_ref();
        // First, remove the actual crate from the namespace.
        if let Some(_removed_app_crate) = self.namespace.crate_tree().lock().remove(&crate_locked.crate_name) {
            // Second, remove all of the crate's global symbols from the namespace's symbol map.
            let mut symbol_map = self.namespace.symbol_map().lock();
            for sec_to_remove in crate_locked.global_sections_iter() {
                trace!("Removing sec_to_remove {:?}", sec_to_remove.deref());
                trace!("symbol_map count: {}\n{:?}", symbol_map.count(), symbol_map.deref());
                trace!("symbol_map had weak sec: {:?}", symbol_map.get(&sec_to_remove.name));
                trace!("symbol_map had strong sec: {:?}", symbol_map.get(&sec_to_remove.name).and_then(|w| w.upgrade()));

                match symbol_map.remove(&sec_to_remove.name) {
                    Some(_removed) => {
                        trace!("Removed symbol {}: {:?}", sec_to_remove.name, _removed.upgrade());
                    }
                    None => {
                        error!("NOTE: couldn't find old symbol {:?} in the old crate {:?} to remove from namespace {:?}.", sec_to_remove.name, crate_locked.crate_name, self.namespace.name());
                    }
                }
            }
        } else {
            error!("BUG: the dropped AppCrateRef {:?} could not be removed from namespace {:?}", self.crate_ref, self.namespace.name());
        }
    }
}

and the log shows some strange behavior surrounding Trie::count(): the count is 0 but there are two elements in the map (?)...

[D] kernel/mod_mgmt/src/lib.rs:816: load_crate_as_application(): trying to load application crate at "/namespaces/_applications/hello-2bf26f3c151abb0a.o"
[I] kernel/mod_mgmt/src/lib.rs:824: loaded new application crate: "hello-2bf26f3c151abb0a", num sections: 111, added 1 new symbols
[D] kernel/runqueue_round_robin/src/lib.rs:204: Adding task to runqueue_round_robin 1, TaskRef(Task { name: "hello-2bf26f3c151abb0a", id: 15, running_on: None, runstate: Blocked, pinned: None })
[D] kernel/spawn/src/lib.rs:610: task_wrapper [1]: "hello-2bf26f3c151abb0a" about to call task entry func 0x72c7790 {fn(alloc::vec::Vec<alloc::string::String>) -> isize} with arg []
[D] kernel/spawn/src/lib.rs:687: task_cleanup_success: "hello-2bf26f3c151abb0a" successfully exited with return value 0
[D] kernel/runqueue_round_robin/src/lib.rs:225: Removing task from runqueue_round_robin 1, TaskRef(Task { name: "hello-2bf26f3c151abb0a", id: 15, running_on: Some(1), runstate: Exited, pinned: None })
[I] applications/shell/src/lib.rs:1056: terminal: task [15] returned exit value: Some(0)
[T] kernel/task/src/lib.rs:970: [CPU 2] Task::drop(): hello-2bf26f3c151abb0a{15}
[T] kernel/mod_mgmt/src/lib.rs:429: Removing sec_to_remove LoadedSection { name: "hello::main::h8bbf4df10c7ebacc", typ: Text, parent: "<locked>", vaddr: v0x72C7790, size: 211, .. }
[T] kernel/mod_mgmt/src/lib.rs:430: symbol_map count: 0
{"shell::main::h0a3c693b3c290b52": (Weak), "hello::main::h8bbf4df10c7ebacc": (Weak)}
[T] kernel/mod_mgmt/src/lib.rs:431: symbol_map had weak sec: Some((Weak))
[T] kernel/mod_mgmt/src/lib.rs:432: symbol_map had strong sec: Some(LoadedSection { name: "hello::main::h8bbf4df10c7ebacc", typ: Text, parent: "<locked>", vaddr: v0x72C7790, size: 211, .. })
[T] kernel/panic_wrapper/src/lib.rs:32: at top of panic_wrapper: PanicInfo { payload: Any { .. }, message: Some(attempt to subtract with overflow), location: Location { file: "/home/kevin/.cargo/registry/src/github.com-1ecc6299db9ec823/qp-trie-0.8.0/src/trie.rs", line: 323, col: 13 }, can_unwind: true }

I'm not sure if this is a bug in qp-trie or an issue with how our new StrRef type implements Borrow<[u8]>, so this issue is a placeholder/TODO item until I can spend more time debugging it.

[kevinaboos]

Upon further examination of the qp-trie code, it seems like inserting or removing key-value pairs via the entry API does not update the Trie's inner count at all. I will verify this with a few test programs and then likely submit an issue on that repo.

[kevinaboos]

I filed sdleffler/qp-trie-rs#31 to address this.