All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Alejandro Vallejo" <alejandro.vallejo@cloud.com>
To: "Teddy Astie" <teddy.astie@vates.tech>, <xen-devel@lists.xenproject.org>
Cc: "Anthony PERARD" <anthony.perard@vates.tech>,
	"Yann Dirson" <yann.dirson@vates.tech>
Subject: Re: [RFC PATCH 07/25] tools/xenbindgen: Add support for structs in TOML specs
Date: Mon, 25 Nov 2024 17:07:57 +0000	[thread overview]
Message-ID: <D5VFAZ01ONWK.2H3VDANBDRQR5@cloud.com> (raw)
In-Reply-To: <8dcd8297-d9d9-4106-ba6d-eefd5df6f69a@vates.tech>

On Mon Nov 25, 2024 at 12:39 PM GMT, Teddy Astie wrote:
> Hi Alejandro,
>
> Le 15/11/2024 à 12:51, Alejandro Vallejo a écrit :
> > Signed-off-by: Alejandro Vallejo <alejandro.vallejo@cloud.com>
> > ---
> >   tools/rust/xenbindgen/src/c_lang.rs | 56 ++++++++++++++++++++++++-
> >   tools/rust/xenbindgen/src/spec.rs   | 64 ++++++++++++++++++++++++++++-
> >   2 files changed, 117 insertions(+), 3 deletions(-)
> >
> > diff --git a/tools/rust/xenbindgen/src/c_lang.rs b/tools/rust/xenbindgen/src/c_lang.rs
> > index f05e36bb362f..597e0ed41362 100644
> > --- a/tools/rust/xenbindgen/src/c_lang.rs
> > +++ b/tools/rust/xenbindgen/src/c_lang.rs
> > @@ -17,9 +17,10 @@
> >
> >   use std::fmt::Write;
> >
> > -use crate::spec::OutFileDef;
> > +use crate::spec::{OutFileDef, StructDef, Typ};
> >
> >   use convert_case::{Case, Casing};
> > +use log::{debug, trace};
> >
> >   /// An abstract indentation level. 0 is no indentation, 1 is [`INDENT_WIDTH`]
> >   /// and so on.
> > @@ -29,6 +30,39 @@ struct Indentation(usize);
> >   /// Default width of each level of indentation
> >   const INDENT_WIDTH: usize = 4;
> >
> > +/// Create a C-compatible struct field. Without the terminating semicolon.
> > +fn structfield(typ: &Typ, name: &str) -> String {
> > +    match typ {
> > +        Typ::Ptr(x) => {
> > +            let t: &Typ = x;
> > +            format!(
> > +                "XEN_GUEST_HANDLE_64({}) {name}",
> > +                match t {
> > +                    Typ::U8 => "uint8",
> > +                    Typ::U16 => "uint16",
> > +                    Typ::U32 => "uint32",
> > +                    Typ::U64 => "uint64_aligned_t",
> > +                    Typ::I8 => "int8",
> > +                    Typ::I16 => "int16",
> > +                    Typ::I32 => "int32",
> > +                    Typ::I64 => "int64_aligned_t",
> > +                    _ => panic!("foo {t:?}"),
> > +                }
> > +            )
> > +        }
> > +        Typ::Struct(x) => format!("struct {x} {name}"),
> > +        Typ::Array(x, len) => format!("{}{name}[{len}]", structfield(x, "")),
> > +        Typ::U8 => format!("uint8_t {name}"),
> > +        Typ::U16 => format!("uint16_t {name}"),
> > +        Typ::U32 => format!("uint32_t {name}"),
> > +        Typ::U64 => format!("uint64_aligned_t {name}"),
> > +        Typ::I8 => format!("int8_t {name}"),
> > +        Typ::I16 => format!("int16_t {name}"),
> > +        Typ::I32 => format!("int32_t {name}"),
> > +        Typ::I64 => format!("int64_aligned_t {name}"),
> > +    }
> > +}
> > +
>
> I think _t are missing in the Ptr cases (we are currently generating
> XEN_GUEST_HANDLE_64(uint8) which I don't think is valid).

It is intentional. The handles are presently missing those _t in Xen's public
headers, but that's something I'll be changing in the interest of sanity. That
way we can just recurse to the inner type.


> Aside that, wouldn't it be better to have a separate function for
> converting the type to its C representation ?
>
> Something like
>
> impl Typ { // or blanket trait
>      fn c_repr(&self) -> String {
>          match self {
>              /* ... */
>          }
>      }
> }

That's roughhly what typ_rs() does, and indeed what typ_c() used to do. There's
a complication though...

>
> fn structfield(typ: &Typ, name: &str) -> String {
>      format!("{} {name}", typ.c_repr());
> }
>
> We can also consider Typ::Struct or Typ::Array cases to call recursively
> to c_repr on its inner type to get its representation.
>
> That way, we can have XEN_GUEST_HANDLE_64(struct something).

Initially structfield() was typ_c() (like the Rust backend). Then arrays
happened... Size and typename surround the name of the field (e.g: uint8_t
handle[16]) so I stopped doing it like that because I thought I couldn't.

I have since then noticed I can cheat! The following two fields are identical.
Except the first one is a heck of a lot simpler to generate.

  __typeof__(uint8_t[16]) handle;
  uint8_t handle[16];

My latest branch simplifies all this by s/structfield/typ_c/ and using that
typeof trick.

>
> Cheers
>
> Teddy
>
>
>
> Teddy Astie | Vates XCP-ng Developer
>
> XCP-ng & Xen Orchestra - Vates solutions
>
> web: https://vates.tech

Cheers,
Alejandro


  reply	other threads:[~2024-11-25 17:08 UTC|newest]

Thread overview: 51+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-11-15 11:51 [RFC PATCH 00/25] Introduce xenbindgen to autogen hypercall structs Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 01/25] xen/domctl: Refine grant_opts into max_grant_version Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 02/25] xen/domctl: Replace altp2m_opts with altp2m_mode Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 03/25] tools/xenbindgen: Introduce a Xen hypercall IDL generator Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 04/25] tools/xenbindgen: Add a TOML spec reader Alejandro Vallejo
2024-11-25 15:13   ` Teddy Astie
2024-11-25 16:51     ` Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 05/25] tools/xenbindgen: Add basic plumbing for the C backend Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 06/25] tools/xenbindgen: Add xenbindgen's Cargo.lock file Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 07/25] tools/xenbindgen: Add support for structs in TOML specs Alejandro Vallejo
2024-11-25 12:39   ` Teddy Astie
2024-11-25 17:07     ` Alejandro Vallejo [this message]
2024-11-25 15:03   ` Teddy Astie
2024-11-25 17:16     ` Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 08/25] tools/xenbindgen: Add support for enums " Alejandro Vallejo
2024-11-25 16:39   ` Teddy Astie
2024-11-25 17:18     ` Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 09/25] tools/xenbindgen: Add support for bitmaps " Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 10/25] tools/xenbindgen: Add support for includes in the " Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 11/25] tools/xenbindgen: Validate ABI rules at generation time Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 12/25] xen: Replace sysctl/readconsole with autogenerated version Alejandro Vallejo
2024-11-25 12:05   ` Jan Beulich
2024-11-25 18:51     ` Alejandro Vallejo
2024-11-26  9:40       ` Jan Beulich
2024-11-26 12:27         ` Alejandro Vallejo
2024-11-26 13:20           ` Jan Beulich
2024-11-26 14:36             ` Alejandro Vallejo
2024-11-26 16:30               ` Jan Beulich
2024-11-26 14:39             ` Teddy Astie
2024-11-26 16:28               ` Jan Beulich
2024-11-15 11:51 ` [RFC PATCH 13/25] xen: Replace hand-crafted altp2m_mode descriptions with autogenerated ones Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 14/25] xen: Replace common bitmaps in domctl.createdomain with autogenerated versions Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 15/25] xen/arm: Replace hand-crafted xen_arch_domainconfig with autogenerated one Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 16/25] xen/x86: " Alejandro Vallejo
2024-11-25 12:09   ` Jan Beulich
2024-11-25 18:53     ` Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 17/25] xen/ppc: Replace empty " Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 18/25] xen/riscv: " Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 19/25] xen: Replace hand-crafted domctl/createdomain with autogenerated version Alejandro Vallejo
2024-12-04 14:48   ` Teddy Astie
2024-11-15 11:51 ` [RFC PATCH 20/25] tools/xen-sys: Create a crate with autogenerated Rust constructs Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 21/25] tools/xenbindgen: Add Rust backend to xenbindgen Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 22/25] tools/xen-sys: Add autogenerated Rust files Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 23/25] licence: Add Unicode-DFS-2016 to the list of licences Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 24/25] tools/rust: Add deny.toml Alejandro Vallejo
2024-11-15 11:51 ` [RFC PATCH 25/25] ci: Add a CI checker for Rust-related helpful properties Alejandro Vallejo
2024-11-21 17:46 ` [RFC PATCH 00/25] Introduce xenbindgen to autogen hypercall structs Anthony PERARD
2024-11-22 10:52   ` Teddy Astie
2024-11-22 13:26     ` Alejandro Vallejo
2024-11-22 13:12   ` Alejandro Vallejo
2024-11-22 16:34     ` Anthony PERARD

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=D5VFAZ01ONWK.2H3VDANBDRQR5@cloud.com \
    --to=alejandro.vallejo@cloud.com \
    --cc=anthony.perard@vates.tech \
    --cc=teddy.astie@vates.tech \
    --cc=xen-devel@lists.xenproject.org \
    --cc=yann.dirson@vates.tech \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.