[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAH5fLgjaAB+6aqNaPXRxfKBzOjVxBNq5_Gj6ntG7rMM1ToMNSQ@mail.gmail.com>
Date: Mon, 22 Apr 2024 08:31:11 +0200
From: Alice Ryhl <aliceryhl@...gle.com>
To: David Laight <David.Laight@...lab.com>
Cc: "boqun.feng@...il.com" <boqun.feng@...il.com>, "a.hindborg@...sung.com" <a.hindborg@...sung.com>,
"akpm@...ux-foundation.org" <akpm@...ux-foundation.org>, "alex.gaynor@...il.com" <alex.gaynor@...il.com>,
"arnd@...db.de" <arnd@...db.de>, "arve@...roid.com" <arve@...roid.com>,
"benno.lossin@...ton.me" <benno.lossin@...ton.me>,
"bjorn3_gh@...tonmail.com" <bjorn3_gh@...tonmail.com>, "brauner@...nel.org" <brauner@...nel.org>,
"cmllamas@...gle.com" <cmllamas@...gle.com>, "gary@...yguo.net" <gary@...yguo.net>,
"gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
"joel@...lfernandes.org" <joel@...lfernandes.org>, "keescook@...omium.org" <keescook@...omium.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>, "linux-mm@...ck.org" <linux-mm@...ck.org>,
"maco@...roid.com" <maco@...roid.com>, "ojeda@...nel.org" <ojeda@...nel.org>,
"rust-for-linux@...r.kernel.org" <rust-for-linux@...r.kernel.org>, "surenb@...gle.com" <surenb@...gle.com>,
"tkjos@...roid.com" <tkjos@...roid.com>, "viro@...iv.linux.org.uk" <viro@...iv.linux.org.uk>,
"wedsonaf@...il.com" <wedsonaf@...il.com>, "willy@...radead.org" <willy@...radead.org>
Subject: Re: [PATCH v5 1/4] rust: uaccess: add userspace pointers
On Sun, Apr 21, 2024 at 9:49 PM David Laight <David.Laight@...lab.com> wrote:
>
> From: Alice Ryhl
> > Sent: 21 April 2024 19:38
> >
> > On Sun, Apr 21, 2024 at 8:08 PM David Laight <David.Laight@...lab.com> wrote:
> > >
> > > Should you be implementing 'struct iov_iter' ?
> > >
> > > Even if it means creating an IO_UBUF for ioctls?
> > > (Although that might take some 'fettling' for read+write for ioctls.)
> >
> > That seems to be intended for when you have several chunks of memory
> > in userspace that you want to treat as one contiguous chunk. That's
> > not something I need in the Android Binder driver.
>
> It also transparently supports in-kernel users and some other cases.
>
> I think there is a patch intended for 6.10 that removes the 'read'
> and 'write' driver 'ops' and requires that drivers support 'read_iter'
> and 'write_iter'.
Binder uses an ioctl, not read/write, so even if Binder could use it,
there's no real advantage for Binder to do so. But it does sound like
something we want to support eventually in the kernel. However, I've
spent a long time on polishing this API, and it fits my needs well. I
don't want to start over now that I am almost finished.
So, I think that this could be a good follow-up to this patch that
someone else is welcome to submit.
Alice
Powered by blists - more mailing lists