Re: [PATCH v8 5/5] rust: add dynamic ID pool abstraction for bitmap

[Boqun Feng <boqun.feng@gmail.com>]

On Tue, May 20, 2025 at 05:42:51AM -0700, Alice Ryhl wrote:
> On Mon, May 19, 2025 at 10:21 PM Boqun Feng <boqun.feng@gmail.com> wrote:
> >
> > On Mon, May 19, 2025 at 08:46:37PM -0700, Alice Ryhl wrote:
> > > On Mon, May 19, 2025 at 4:56 PM Boqun Feng <boqun.feng@gmail.com> wrote:
> > > >
> > > > On Tue, May 20, 2025 at 12:51:07AM +0200, Jann Horn wrote:
> > > > > On Mon, May 19, 2025 at 6:20 PM Burak Emir <bqe@google.com> wrote:
> > > > > > This is a port of the Binder data structure introduced in commit
> > > > > > 15d9da3f818c ("binder: use bitmap for faster descriptor lookup") to
> > > > > > Rust.
> > > > >
> > > > > Stupid high-level side comment:
> > > > >
> > > > > That commit looks like it changed a simple linear rbtree scan (which
> > > > > is O(n) with slow steps) into a bitmap thing. A more elegant option
> > > > > might have been to use an augmented rbtree, reducing the O(n) rbtree
> > > > > scan to an O(log n) rbtree lookup, just like how finding a free area
> > > >
> > > > I think RBTree::cursor_lower_bound() [1] does exactly what you said
> > >
> > > We need the smallest ID without a value, not the smallest ID in use.
> > >
> >
> > Ok, but it shouldn't be hard to write a Rust function that search that,
> > right? My point was mostly the Rust rbtree binding can do O(log n)
> > search. I have no idea about "even so, should we try something like Jann
> > suggested". And I think your other reply basically says no.
> 
> We would need to store additional data in the r/b tree to know whether
> to go left or right, so it would be somewhat tricky. We don't have an

Hmm... I'm confused, I thought you can implement a search like that by
doing what RBTree::raw_entry() does except that when Ordering::Equal you
always go left or right (depending on whether you want to get an unused
ID less or greater than a key value), i.e. you always search until you
get an Vacant entry. Why do you need store additional data for that?
Maybe I'm missing something here?

Regards,
Boqun

> implementation of that in Rust.
> 
> Alice