From: Martin Rodriguez Reboredo <yakoyoku@gmail.com>
To: Qingsong Chen <changxian.cqs@antgroup.com>, linux-kernel@vger.kernel.org
Cc: 田洪亮 <tate.thl@antgroup.com>, "Miguel Ojeda" <ojeda@kernel.org>,
"Alex Gaynor" <alex.gaynor@gmail.com>,
"Wedson Almeida Filho" <wedsonaf@gmail.com>,
"Boqun Feng" <boqun.feng@gmail.com>,
"Gary Guo" <gary@garyguo.net>,
"Björn Roy Baron" <bjorn3_gh@protonmail.com>,
"Benno Lossin" <benno.lossin@proton.me>,
rust-for-linux@vger.kernel.org
Subject: Re: [PATCH v3 3/3] rust: kernel: add SgTable abstraction
Date: Mon, 12 Jun 2023 13:13:12 -0300 [thread overview]
Message-ID: <11fe7426-1d7b-78ff-f848-33ce6d7d3747@gmail.com> (raw)
In-Reply-To: <20230610104909.3202958-4-changxian.cqs@antgroup.com>
On 6/10/23 07:49, Qingsong Chen wrote:
> SgTable is similar to `struct sg_table`, consisted of
> scatterlist entries, and could be chained with each other.
>
> Signed-off-by: Qingsong Chen <changxian.cqs@antgroup.com>
> ---
> [...]
> +impl<'a, const N: usize> SgTable<'a, N> {
> + /// Construct a new initializer.
> + ///
> + /// # Errors
> + ///
> + /// The length of `entries` should exactly be the available size of [`SgTable<N>`],
> + /// or else an error is returned.
> + ///
> + /// If the table is `chainable`, the available size is `N - 1`, because one entry
> + /// should be reserved for chaining.
> + pub fn new(
> + entries: &'a [Pin<&mut [u8]>],
> + chainable: bool,
> + ) -> impl PinInit<SgTable<'a, N>, Error> {
> + build_assert!(N > 0);
> + // SAFETY: `slot` is valid while the closure is called, the `entries` are
> + // pinned and valid.
> + unsafe {
> + init::pin_init_from_closure(move |slot: *mut Self| {
> + let mut nr_entry = N;
> + if chainable {
> + nr_entry -= 1;
> + }
> + if nr_entry == 0 || entries.len() != nr_entry {
> + return Err(EINVAL);
> + }
> +
> + for i in 0..nr_entry {
> + // `slot` contains uninit memory, avoid creating a reference.
> + let opaque = addr_of!((*slot).entries[i].opaque);
> + let sgl = Opaque::raw_get(opaque);
> +
> + bindings::sg_set_buf(sgl, entries[i].as_ptr() as _, entries[i].len() as _);
> + if i + 1 == nr_entry {
> + (*sgl).page_link |= bindings::SG_END as u64;
> + (*sgl).page_link &= !(bindings::SG_CHAIN as u64);
> + }
> + }
> + Ok(())
> + })
> + }
> + }
> +
> + /// Chain two [`SgTable`] together.
> + ///
> + /// Transfer the last entry of this table as a chainable pointer to the
> + /// first entry of `sgt` SgTable.
> + ///
> + /// # Errors
> + ///
> + /// Return an error if this table is not chainable or has been chained.
> + pub fn chain_sgt<const M: usize>(&mut self, sgt: Pin<&mut SgTable<'_, M>>) -> Result {
> + if self.count() != N - 1 {
> + return Err(EINVAL);
> + }
> + self.entries[N - 2].unmark_end();
> +
> + // SAFETY: `sgt.entries` are initialized by the `new` constructor,
> + // so it's valid.
> + let next = unsafe { ScatterList::as_mut(sgt.entries[0].opaque.get()).unwrap() };
> + self.entries[N - 1].chain_sgl(next);
> + Ok(())
> + }
> +
> + /// Chain [`SgTable`] and [`ScatterList`] together.
> + ///
> + /// Transfer the last entry of this table as a chainable pointer to `sgl`
> + /// scatterlist.
> + ///
> + /// # Errors
> + ///
> + /// Return an error if this table is not chainable or has been chained.
> + pub fn chain_sgl(&mut self, sgl: Pin<&mut ScatterList<'_>>) -> Result {
> + if self.count() != N - 1 {
> + return Err(EINVAL);
> + }
> + self.entries[N - 2].unmark_end();
> + self.entries[N - 1].chain_sgl(sgl);
> + Ok(())
> + }
> +
> + /// Split the first table from chained scatterlist.
> + ///
> + /// # Errors
> + ///
> + /// Return an error if this table is not chainable or has not been chained.
> + pub fn split(&mut self) -> Result {
> + if !self.entries[N - 1].is_chain() {
> + return Err(EINVAL);
> + }
> + self.entries[N - 2].mark_end();
> + Ok(())
> + }
> +
> [...]
> +}
I'd suggest being more explicit in the `# Errors` section by
mentioning that `EINVAL` is returned, but do as you prefer.
Reviewed-by: Martin Rodriguez Reboredo <yakoyoku@gmail.com>
next prev parent reply other threads:[~2023-06-12 16:14 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-10 10:49 [PATCH v3 0/3] Rust scatterlist abstractions Qingsong Chen
2023-06-10 10:49 ` [PATCH v3 1/3] rust: kernel: add ScatterList abstraction Qingsong Chen
2023-06-12 15:08 ` Martin Rodriguez Reboredo
2023-06-10 10:49 ` [PATCH v3 2/3] rust: kernel: implement iterators for ScatterList Qingsong Chen
2023-06-12 15:10 ` Martin Rodriguez Reboredo
2023-06-10 10:49 ` [PATCH v3 3/3] rust: kernel: add SgTable abstraction Qingsong Chen
2023-06-12 16:13 ` Martin Rodriguez Reboredo [this message]
2023-06-10 15:33 ` [PATCH v3 0/3] Rust scatterlist abstractions Greg KH
2023-06-10 15:35 ` Greg KH
2023-06-12 3:49 ` Qingsong Chen
2023-06-12 5:38 ` Greg KH
2023-06-12 7:03 ` Qingsong Chen
2023-06-12 7:12 ` Greg KH
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=11fe7426-1d7b-78ff-f848-33ce6d7d3747@gmail.com \
--to=yakoyoku@gmail.com \
--cc=alex.gaynor@gmail.com \
--cc=benno.lossin@proton.me \
--cc=bjorn3_gh@protonmail.com \
--cc=boqun.feng@gmail.com \
--cc=changxian.cqs@antgroup.com \
--cc=gary@garyguo.net \
--cc=linux-kernel@vger.kernel.org \
--cc=ojeda@kernel.org \
--cc=rust-for-linux@vger.kernel.org \
--cc=tate.thl@antgroup.com \
--cc=wedsonaf@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).