Compiler warning on MinGW gcc 8.1.0

[man4]

I'm getting the following compiler warning on MinGW, gcc 8.1.0 (which didn't appear in gcc 7.3.0):

misc.cpp: In function 'int WinProcGroup::get_group(size_t)':
misc.cpp:241:77: warning: cast between incompatible function types from 'FARPROC' {aka 'long long int (*)()'} to 'fun1_t' {aka 'bool (*)(_LOGICAL_PROCESSOR_RELATIONSHIP, _SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*, long unsigned int*)'} [-Wcast-function-type]
   auto fun1 = (fun1_t)GetProcAddress(k32, "GetLogicalProcessorInformationEx");
                                                                             ^
misc.cpp: In function 'void WinProcGroup::bindThisThread(size_t)':
misc.cpp:309:71: warning: cast between incompatible function types from 'FARPROC' {aka 'long long int (*)()'} to 'fun2_t' {aka 'bool (*)(short unsigned int, _GROUP_AFFINITY*)'} [-Wcast-function-type]
   auto fun2 = (fun2_t)GetProcAddress(k32, "GetNumaNodeProcessorMaskEx");
                                                                       ^
misc.cpp:310:67: warning: cast between incompatible function types from 'FARPROC' {aka 'long long int (*)()'} to 'fun3_t' {aka 'bool (*)(void*, const _GROUP_AFFINITY*, _GROUP_AFFINITY*)'} [-Wcast-function-type]
   auto fun3 = (fun3_t)GetProcAddress(k32, "SetThreadGroupAffinity");

According to the release notes for gcc 8, the option -Wextra is responsible for this warning.

[Hanamuke]

That should be easy to fix but i don't have an msvc compiler, maybe @mstembera can help ?

[mstembera]

@Hanamuke These are Mingw/gcc 8.1.0 warnings not msvc.

[syzygy1]

I think it's a rather stupid warning of gcc. I don't think there is any "right" way to do the cast that does not result in bogus warnings. (Feel free to prove me wrong, though!)

It is possible (at least in C) to cast first to void (*)(void), then to the desired type, but in my view that is just a hack.

[WOnder93]

In C++, I think one should use static_cast<...>(...) or even reinterpret_cast<...>(...) in this situation.

[snicolet]

The gcc 8 release notes ( https://gcc.gnu.org/gcc-8/changes.html ) indicate that there is a new flag -Wold-style-cast to get help for how to deal with casting problems: "The -Wold-style-cast diagnostic can now emit fix-it hints telling you when you can use a static_cast, const_cast, or reinterpret_cast."

==> @man4 , @mstembera If you have access to gcc 8, could you try to edit the makefile and compile the current master with -Wold-style-cast to see the diagnostic the compiler emits? Thanks in advance.

[man4]

Here are the relevant warnings:

misc.cpp:241:77: warning: use of old-style cast to 'fun1_t' {aka 'bool (*)(enum _LOGICAL_PROCESSOR_RELATIONSHIP, struct _SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*, long unsigned int*)'} [-Wold-style-cast]
   auto fun1 = (fun1_t)GetProcAddress(k32, "GetLogicalProcessorInformationEx");
                                                                             ^
               --------
               reinterpret_cast<fun1_t> (                                     )

misc.cpp:309:71: warning: use of old-style cast to 'fun2_t' {aka 'bool (*)(short unsigned int, struct _GROUP_AFFINITY*)'} [-Wold-style-cast]
   auto fun2 = (fun2_t)GetProcAddress(k32, "GetNumaNodeProcessorMaskEx");
                                                                       ^
               --------
               reinterpret_cast<fun2_t> (                               )

misc.cpp:310:67: warning: use of old-style cast to 'fun3_t' {aka 'bool (*)(void*, const struct _GROUP_AFFINITY*, struct _GROUP_AFFINITY*)'} [-Wold-style-cast]
   auto fun3 = (fun3_t)GetProcAddress(k32, "SetThreadGroupAffinity");
                                                                   ^
               --------
               reinterpret_cast<fun3_t> (                           )

I've attached the full stderr output too (which contains warnings for all old-style casts):
out.txt

I tried naively changing all the casts to reinterpret_cast<...>(...) but the incompatible function type warning still appears. Indeed, the warning seems unrelated to the fact that old-style casting was used.

[man4]

It is possible (at least in C) to cast first to void (*)(void), then to the desired type, but in my view that is just a hack.

Changing (fun1_t) to (fun1_t)(void(*)()) (and similarly for the other casts) does remove the warnings in gcc 8, but I agree with @syzygy1 that this looks ugly.