Rename hooks should affect names passed to subsequent hooks; rename hook for a thing should be called before any hook that receives data referring to that thing

Open
  • Is something wrong?
#3,540 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
25/100
Issue type
Bug
Clarity
Mostly clear
Activity status
Active
Tech stack
c, rust
Domain
compilers

Research direction

Start with the reproduction: a demo.h with an anonymous union inside in6_addr, a ParseCallbacks impl overriding item_name and field_name, and debug output showing the order of calls. Find where bindgen invokes item_name and field_name in the parse and codegen passes, and check whether the renamed name is used when field_name runs. Done means field_name receives in6_addr_u and item_name runs once for that type, with a regression test added to the bindgen test suite.

Written by the indexing model from the issue text.

Description

If you build and run this test program, using bindgen 0.73...

use bindgen::callbacks::{ParseCallbacks, ItemInfo, ItemKind, FieldInfo};

#[derive(Debug)]
struct DemoCallbacks;

impl ParseCallbacks for DemoCallbacks {
    fn item_name(&self, info: ItemInfo<'_>) -> Option<String> {
        if info.kind == ItemKind::Type {
            eprintln!("item_name:  {:?}", &info);
            if let Some(trimmed) = info.name.strip_suffix("__bindgen_ty_1") {
                return Some(format!("{trimmed}_u"));
            }
        }
        None
    }
    fn field_name(&self, info: FieldInfo<'_>) -> Option<String> {
        eprintln!("field_name: {:?}", info);
        None
    }
}

fn main() {
    bindgen::builder()
        .header("demo.h")
        .parse_callbacks(Box::new(DemoCallbacks))
        .allowlist_type("in6_addr")
        .generate()
        .expect("parsing failed")
        .write(std::io::stdout().lock())
        .expect("write error");
}

... with this file as demo.h in the working directory:

struct in6_addr {
  union {
    unsigned char  u6_addr8[16];
    unsigned short u6_addr16[8];
    unsigned int   u6_addr32[8];
  } in6_u;
};

... you will get this output:

item_name:  ItemInfo { name: "in6_addr", kind: Type }
field_name: FieldInfo { type_name: "in6_addr", field_name: "in6_u", field_type_name: Some("union (unnamed at demo.h:2:3)") }
item_name:  ItemInfo { name: "in6_addr__bindgen_ty_1", kind: Type }
field_name: FieldInfo { type_name: "in6_addr_u", field_name: "u6_addr8", field_type_name: None }
field_name: FieldInfo { type_name: "in6_addr_u", field_name: "u6_addr16", field_type_name: None }
field_name: FieldInfo { type_name: "in6_addr_u", field_name: "u6_addr32", field_type_name: None }
item_name:  ItemInfo { name: "in6_addr__bindgen_ty_1", kind: Type }
/* automatically generated by rust-bindgen 0.73.2 */

#[repr(C)]
#[derive(Copy, Clone)]
pub struct in6_addr {
    pub in6_u: in6_addr_u,
}
#[repr(C)]
#[derive(Copy, Clone)]
pub union in6_addr_u {
    pub u6_addr8: [::std::os::raw::c_uchar; 16usize],
    pub u6_addr16: [::std::os::raw::c_ushort; 8usize],
    pub u6_addr32: [::std::os::raw::c_uint; 8usize],
}
// ... size/alignment/offset assertions omitted ...

The generated bindings (everything after the "automatically generated by ..." line) correctly reflect the renamed anonymous union in the input, but the calls to field_name do not. This is the key part of the debug spew:

field_name: FieldInfo { type_name: "in6_addr", field_name: "in6_u", field_type_name: Some("union (unnamed at demo.h:2:3)") }
item_name:  ItemInfo { name: "in6_addr__bindgen_ty_1", kind: Type }

These calls should have happened in the opposite order, and the FieldInfo passed to the field_name hook should have had field_type_name: Some("in6_addr_u"). Also, item_name should not have been called a second time for this type.

This is just one example instance of a general principle: in all cases, whenever bindgen finds a thing that has or needs a name, the relevant rename hook should have a chance to act on that thing before any operation that refers to that thing happens, and all the hooks for subsequent operations should see adjusted names, not the originals, and especially not placeholders like "union (unnamed at demo.h.2:3)".

Dominant language
Rust
Stars
5.3k
Forks
846
Avg merge
2d 6h
Merged PRs (30d)
11

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from rust-lang/rust-bindgen

All issues in rust-lang/rust-bindgen

Similar issues

More Rust issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.