#250 linux x86_64 clang warnings

[Edward-san]

Hi, I decided to move from gcc to clang compilation, as it's easier to compile and it uses a lot of new features gcc doesn't have yet officially (like for example the address/undefined/thread sanitizer or more c++11 features).

I compiled the v2 branch with the same options + c99 standard + pedantic and I got these results in the attached file.

[Edward-san]

[Simon Howard]

Thanks. None of these look particularly serious but they should probably be fixed at some point.

[Edward-san]

I'm sure the "'GameMission_t' -> 'GameMode_t'", the "memset" and the "MF_SPECTRAL" are issues to be concerned if some assumptions are wrongly done.

Apart from that, I recompiled everything with the ub sanitizer, which instruments the code in order to report runtime undefined behaviors. I was surprised to see the result when I ran chocolate-doom with doom v1.9 shareware version and let the game play all the demos... it's in the new attached file. How could doom work with these ub things?

[Edward-san]

[Simon Howard]

The ub warnings don't actually seem that big an issue. Undefined behaviors are just things that aren't explicitly specified by the C standard, but some behaviors are so common across all compilers and CPUs that they can be reasonably assumed. Left shift of a negative value is one example. The C standard says:

"The result of E1 << E2 is E1 left-shifted E2 bit positions; vacated bits are filled with zeros. If E1 has an unsigned type, the value of the result is E1|2E2, reduced modulo one more than the maximum value representable in the result type. If E1 has a signed type and nonnegative value, and E1|2E2 is representable in the result type, then that is the resulting value; otherwise, the behavior is undefined."

In practise though, all modern CPUs do the sensible thing. Integer overflows are another similar example.

[Edward-san]

Ping. Someone at doomworld noted a buffer overflow, which is a lot similar to the memset warnings I got with clang. Uploading a list of the relevant warnings in the v2 branch.

[Edward-san]

Attached the 'memset' fix from prboom+ code. I don't understand why prboom uses '&mousearray' as the pointer, instead of 'mousearray', which is a pointer...