IMMDevice::GetState() requires an out param and returns HRESULT, but the type of the return value is DEVICE_STATE

[docwilco]

Summary

The Rust IMMDevice::GetState takes an "out" parameter (pdwstate: *mut u32), and returns a value of type DEVICE_STATE:

    pub unsafe fn GetState(&self, pdwstate: *mut u32) -> DEVICE_STATE {
        (::windows_core::Interface::vtable(self).GetState)(::windows_core::Interface::as_raw(self), pdwstate)
    }

The VTable implementation for that is:

    pub GetState: unsafe extern "system" fn(*mut ::core::ffi::c_void, *mut u32) -> DEVICE_STATE,

This looks wrong to me, as the C++ side of things defines IMMDevice::GetState as:

HRESULT GetState(
  [out] DWORD *pdwState
);

This makes me think the Rust definition should be:

    pub unsafe fn GetState(&self) -> DEVICE_STATE {
        let mut result__ = core::mem::zeroed();
        (::windows_core::Interface::vtable(self).GetState)(::windows_core::Interface::as_raw(self), &mut result__).map(|| DEVICE_STATE(result__))
    }

With the VTable entry being:

    pub GetState: unsafe extern "system" fn(*mut core::ffi::c_void, *mut u32) -> windows_core::HRESULT,

I'm not super familiar with the windows-rs crate internals just yet, so I'm not sure if this is generated code where the generator should be updated, or if this is code that should now be updated manually?

Crate manifest

No response

Crate code

No response

[riverar]

We fixed this on the metadata side already, but haven't ingested that new metadata yet on the Rust side. See also: https://github.com/microsoft/win32metadata/issues/1856#issuecomment-2059966655

[kennykerr]

Yep, I'm just waiting for a release of the Win32 metadata.

[docwilco]

Oooh, shiny! Thank you! <3