Unsigned global set to -1

[e00E]

Input C/C++ Header

unsigned long long g_107 = 18446744073709551615UL;

Bindgen Invocation

bindgen failing.h -o failing.rs

Actual Results

Bindgen generates

pub const g_107: ::std::os::raw::c_ulonglong = -1;

which when compiled with rustc --crate-type lib --test -o out failing2.rs gives

error[E0600]: cannot apply unary operator `-` to type `u64`
 --> failing2.rs:3:48
  |
3 | pub const g_107: ::std::os::raw::c_ulonglong = -1;
  |                                                ^^

error: aborting due to previous error

Expected Results

Not sure, but 18446744073709551615 does not look like -1 in the first place.
Found with csmith fuzzing.

[pepyakin]

Not sure, but 18446744073709551615 does not look like -1 in the first place.
Found with csmith fuzzing.

In fact it is!
hex(18446744073709551615) = 0xFFFF_FFFF_FFFF_FFFF which is -1 in signed interpretation. See Two's complement.

This happens because bindgen doesn't check for a signedness of type, it just emits as signed var with int_expr.

[highfive]

Hi! If you have any questions regarding this issue, feel free to make a comment here, or ask it in the #servo channel in IRC.

If you intend to work on this issue, then add @highfive: assign me to your comment, and I'll assign this to you. 😄

[fitzgen]

Thanks for the bug report @e00E ! Did you use C-Smith and C-Reduce?

This should be easy enough to fix by making signed and unsigned versions of the codegen helpers that @pepyakin linked to.

[e00E]

I did use csmith but not C-reduce. I manually cut out everything except the part responsible for the error.

[Jlonghi]

Hi I'd like to start contributing and take this on as my first bug!
@highfive: assign me

[highfive]

Hey @josh0588! Thanks for your interest in working on this issue. It's now assigned to you!

[fitzgen]

@josh0588 if you run into any questions, don't hesitate to ask :)

[Jlonghi]

@fitzgen thanks, it appears there already is a signed and unsigned version of the codegen helpers 'uint_ expr' and 'int_expr? but the match construct only calls the int_expr

[fitzgen]

@josh0588 ah, I forgot about that! Looks like we jsut need to call it at the right time then :)