ctest: check incompatible pointer types for public fields.

[mbyx]

Description

Attempts to fix #5196 by adding a test for improper field types into the C side of the field_type test.

Sources

Checklist

• Relevant tests in libc-test/semver have been updated
• No placeholder or unstable values like *LAST or *MAX are
  included (see #3131)
• Tested locally (cd libc-test && cargo test --target mytarget);
  especially relevant for platforms that may not be checked in CI

[mbyx]

Also, I'm not sure, but I think this was already tested in a way? Because while working on it I got the following result:

cargo:warning=/tmp/.tmpQcOH17/incompatible_ptr_types.out.c:66:9: error: assignment to 'ctest_field_ty__Foo__a' {aka 'float *'} from incompatible pointer type 'int32_t *' {aka 'int *'} [-Werror=incompatible-pointer-types]
cargo:warning=   66 |     ptr = &b->a;
cargo:warning=      |         ^
cargo:warning=/tmp/.tmpQcOH17/incompatible_ptr_types.out.c:69:12: error: returning 'int32_t *' {aka 'int *'} from a function with incompatible return type 'ctest_field_ty__Foo__a' {aka 'float *'} [-Werror=incompatible-pointer-types]
cargo:warning=   69 |     return &b->a;
cargo:warning=      |            ^~~~~
cargo:warning=At top level:

For the specific test case at least, the correct warning would have popped up even if the extra test lines had not been added. I might not be understanding the requirements here though.

Oh and if that is the case, the reason for it not detecting before may be that the field was not public, as non public fields can't be checked. I think, it's been a while since I looked at ctest :)

[tgross35]

Huh; I thought I saw a case related to #1020 where we had an integer as a field but a union is the correct type, and no skips. Now I can't find that, and you're right that it seems covered.

Maybe I was looking at union fields - do you know if we test those?

[mbyx]

Yup, we do test those. The populate_field_ptr_tests in template.rs creates the field_ptr test for all public struct and union fields that haven't been manually skipped. And the field_ptr test creates a pointer to that field that is returned by the function, while using both the actual and the Rust translated types, so it should catch all instances of this warning.

[tgross35]

You're right, sorry for all the confusion here. It looks like tests don't cover this, would you mind adding them? Could just be ctest-test/src/mismatched-union-field-ty.{c,rs} and ctest-test/src/mismatched-struct-field-ty.{c,rs} to make sure compilation fails.