thread 'main' panicked at '"__mbstate_t_(anonymous_struct_at_foo_cc_2_3)" is not a valid Ident'

Question

thread 'main' panicked at '"__mbstate_t_(anonymous_struct_at_foo_cc_2_3)" is not a valid Ident'

glandium opened this issue 2 years ago · comments

This started happening on clang-trunk with llvm/llvm-project@19e984e

Input C/C++ Header

typedef struct {
  struct {};
} __mbstate_t;

Bindgen Invocation

bindgen::Builder::default()
    .header("foo.cc")
    .generate()
    .unwrap()

Actual Results

thread 'main' panicked at '"__mbstate_t_(anonymous_struct_at_foo_cc_2_3)" is not a valid Ident', /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.46/src/fallback.rs:756:9
stack backtrace:
   0: rust_begin_unwind
             at /rustc/a55dd71d5fb0ec5a6a3a9e8c27b2127ba491ce52/library/std/src/panicking.rs:584:5
   1: core::panicking::panic_fmt
             at /rustc/a55dd71d5fb0ec5a6a3a9e8c27b2127ba491ce52/library/core/src/panicking.rs:142:14
   2: proc_macro2::fallback::validate_ident
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.46/src/fallback.rs:756:9
   3: proc_macro2::fallback::Ident::_new
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.46/src/fallback.rs:698:9
   4: proc_macro2::fallback::Ident::new
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.46/src/fallback.rs:708:9
   5: proc_macro2::Ident::new
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.46/src/lib.rs:974:21
   6: bindgen::ir::context::BindgenContext::rust_ident_raw
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/ir/context.rs:861:9
   7: bindgen::ir::context::BindgenContext::rust_ident
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/ir/context.rs:853:9
   8: <bindgen::ir::comp::CompInfo as bindgen::codegen::CodeGenerator>::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:1786:31
   9: <bindgen::ir::ty::Type as bindgen::codegen::CodeGenerator>::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:814:39
  10: <bindgen::ir::item::Item as bindgen::codegen::CodeGenerator>::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:500:17
  11: <bindgen::ir::comp::CompInfo as bindgen::codegen::CodeGenerator>::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:2115:13
  12: <bindgen::ir::ty::Type as bindgen::codegen::CodeGenerator>::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:814:39
  13: <bindgen::ir::item::Item as bindgen::codegen::CodeGenerator>::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:500:17
  14: <bindgen::ir::module::Module as bindgen::codegen::CodeGenerator>::codegen::{{closure}}
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:523:21
  15: <bindgen::ir::module::Module as bindgen::codegen::CodeGenerator>::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:553:13
  16: <bindgen::ir::item::Item as bindgen::codegen::CodeGenerator>::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:491:17
  17: bindgen::codegen::codegen::{{closure}}
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:4364:9
  18: bindgen::ir::context::BindgenContext::gen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/ir/context.rs:1173:19
  19: bindgen::codegen::codegen
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/codegen/mod.rs:4328:5
  20: bindgen::Bindings::generate
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/lib.rs:2438:32
  21: bindgen::Builder::generate
             at /home/glandium/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.60.1/src/lib.rs:1502:9

Emilio Cobos Álvarez · Answer 1 · Tue Oct 18 2022 07:06:03 GMT+0800 (China Standard Time)

This is due to this behavior change here, which changes the behavior of libclang:

llvm/llvm-project@19e984e#diff-cde6fae42b53073a85e0901697f60ea80262705f86643b7b27524c7bd24c6f80

It seems they changed an API that was returning an empty string to return a formatted thing that we can't really easily parse or identify as an anonymous struct.

@glandium I think this is worth an upstream bug report, could you do that if you have a build locally? Or do you want me to?

Mike Hommey · Answer 2 · Tue Oct 18 2022 07:50:04 GMT+0800 (China Standard Time)

Better if you do because I don't really know what bindgen would rather have available.

Emilio Cobos Álvarez · Answer 3 · Tue Oct 18 2022 08:22:36 GMT+0800 (China Standard Time)

@glandium can you check if e69e0df, or 9763aa5 (on top) pass tests with llvm trunk? I can check tomorrow otherwise.

Mike Hommey · Answer 4 · Tue Oct 18 2022 08:43:19 GMT+0800 (China Standard Time)

Tests seem to pass, but they also pass with current master...

However, for the testcase, with e69e0df + clang-trunk:

#[repr(C)]
#[derive(Debug, Copy, Clone)]
pub struct __mbstate_t {
    pub _address: u8,
}
#[repr(C)]
#[derive(Debug, Copy, Clone)]
pub struct __mbstate_t__bindgen_ty_1 {
    pub _address: u8,
}

With e69e0df + clang 14:

#[repr(C)]
#[derive(Debug, Copy, Clone)]
pub struct __mbstate_t {
    pub __bindgen_anon_1: __mbstate_t__bindgen_ty_1,
}
#[repr(C)]
#[derive(Debug, Copy, Clone)]
pub struct __mbstate_t__bindgen_ty_1 {
    pub _address: u8,
}

Mike Hommey · Answer 5 · Tue Oct 18 2022 08:55:39 GMT+0800 (China Standard Time)

Tests seem to pass, but they also pass with current master...

scrap that, I didn't run them correctly. They do fail with master and fail less with your patches, but there are still 39 failing (vs. 87 without the patches), with errors that align with the previous comment.

Patrick Walton · Answer 6 · Fri Oct 21 2022 09:07:27 GMT+0800 (China Standard Time)

#2316 is a fix.

Trevor Gross · Answer 7 · Fri Nov 11 2022 16:16:17 GMT+0800 (China Standard Time)

It looks like #2316 was canceled but #2319 resolved it instead? I am wondering why I am still hitting this issue on 0.61.0

  thread 'main' panicked at '"type-parameter-0-1" is not a valid Ident', /home/default/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.47/src/fallback.rs:756:9
  stack backtrace:
     0: rust_begin_unwind
               at /rustc/a8314ef7d0ec7b75c336af2c9857bfaf43002bfc/library/std/src/panicking.rs:584:5
     1: core::panicking::panic_fmt
               at /rustc/a8314ef7d0ec7b75c336af2c9857bfaf43002bfc/library/core/src/panicking.rs:142:14
     2: proc_macro2::fallback::validate_ident
               at /home/default/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.47/src/fallback.rs:756:9
     3: proc_macro2::fallback::Ident::_new
               at /home/default/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.47/src/fallback.rs:698:9
     4: proc_macro2::fallback::Ident::new
               at /home/default/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.47/src/fallback.rs:708:9
     5: proc_macro2::imp::Ident::new
               at /home/default/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.47/src/wrapper.rs:701:50
     6: proc_macro2::Ident::new
               at /home/default/.cargo/registry/src/github.com-1ecc6299db9ec823/proc-macro2-1.0.47/src/lib.rs:974:21
     7: bindgen::ir::context::BindgenContext::rust_ident_raw
               at /home/default/.cargo/registry/src/github.com-1ecc6299db9ec823/bindgen-0.61.0/ir/context.rs:859:9

Trevor Gross · Answer 8 · Fri Nov 11 2022 16:18:20 GMT+0800 (China Standard Time)

Oops, I misread the release log, looks like there actually hasn't been one since the patch.

Please ignore me, hopefully the next release isn't too far