linux-kernel - Re: [PATCH v5 00/26] Generic `Allocator` support for Rust

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Zr4MEPlbYGrx6BbO@cassiopeiae>
Date: Thu, 15 Aug 2024 16:09:20 +0200
From: Danilo Krummrich <dakr@...nel.org>
To: Benno Lossin <benno.lossin@...ton.me>
Cc: Alice Ryhl <aliceryhl@...gle.com>, Boqun Feng <boqun.feng@...il.com>,
	ojeda@...nel.org, alex.gaynor@...il.com, wedsonaf@...il.com,
	gary@...yguo.net, bjorn3_gh@...tonmail.com, a.hindborg@...sung.com,
	akpm@...ux-foundation.org, daniel.almeida@...labora.com,
	faith.ekstrand@...labora.com, boris.brezillon@...labora.com,
	lina@...hilina.net, mcanal@...lia.com, zhiw@...dia.com,
	cjia@...dia.com, jhubbard@...dia.com, airlied@...hat.com,
	ajanulgu@...hat.com, lyude@...hat.com, linux-kernel@...r.kernel.org,
	rust-for-linux@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [PATCH v5 00/26] Generic `Allocator` support for Rust

On Thu, Aug 15, 2024 at 01:39:05PM +0000, Benno Lossin wrote:
> On 15.08.24 15:33, Danilo Krummrich wrote:
> > On Thu, Aug 15, 2024 at 02:34:50PM +0200, Alice Ryhl wrote:
> >> On Thu, Aug 15, 2024 at 2:33 PM Danilo Krummrich <dakr@...nel.org> wrote:
> >>>
> >>> On Thu, Aug 15, 2024 at 11:20:32AM +0200, Alice Ryhl wrote:
> >>>> On Thu, Aug 15, 2024 at 4:52 AM Danilo Krummrich <dakr@...nel.org> wrote:
> >>>>>
> >>>>> On Wed, Aug 14, 2024 at 12:32:15PM -0700, Boqun Feng wrote:
> >>>>>> Hi Danilo,
> >>>>>>
> >>>>>> I'm trying to put your series on rust-dev, but I hit a few conflicts due
> >>>>>> to the conflict with `Box::drop_contents`, which has been in rust-dev
> >>>>>> for a while. And the conflict is not that trivial for me to resolve.
> >>>>>> So just a head-up, that's a requirement for me to put it on rust-dev for
> >>>>>> more tests from my end ;-)
> >>>>>
> >>>>> I rebased everything and you can fetch them from [1].
> >>>>>
> >>>>> I resolved the following conflicts:
> >>>>>
> >>>>>   - for `Box`, implement
> >>>>>     - `drop_contents`
> >>>>>     - `manually_drop_contents` [2]
> >>>>
> >>>> Not sure I like this name. It sounds like something that runs the
> >>>> destructor, but it does the exact opposite.
> >>>
> >>> I thought it kinda makes sense, since it's analogous to `ManuallyDrop::new`.
> >>>
> >>> What about `Box::forget_contents` instead?
> >>
> >> One option is `into_manually_drop`. This uses the convention of using
> >> the `into_*` prefix for conversions that take ownership of the
> >> original value.
> > 
> > The signature of the current `Box::manually_drop_contents` is the same as for
> > `Box::drop_contents`, namely
> > `fn manually_drop_contents(this: Self) -> Box<MaybeUninit<T>, A>`.
> > 
> > `into_manually_drop` seems misleading for for returning a
> > `Box<MaybeUninit<T>, A>`.
> > 
> > I still think `forget_contents` hits it quite well. Just as `drop_contents`
> > drops the value, `forget_contents` makes the `Box` forget the value.
> 
> I think `forget_contents` sounds good. Can you please add some more docs
> to that function though? Like an example and change "Manually drops the
> contents, but keeps the allocation" to "Forgets the contents (does not
> run the destructor), but keeps the allocation.".

I can't really think of a good real world example other than moving out of a
`Box`, can you? Otherwise, maybe it just shouldn't be public?

> 
> Another thing that I spotted while looking at the patch, `move_out`
> doesn't need the `transmute_copy`, you should be able to just call
> `read` on the pointer.

While technically it's the same I thought `transmute_copy` is considered better,
since it has less stict safety requirements?

For `transmute_copy` we only need to say, that dst has the same type as src,
whereas for `read` the pointer must be valid for reads, properly aligned and
point to an initialized value.

> 
> ---
> Cheers,
> Benno
>