Merge #4971

4971: Fix undefined behavior in fixed point number conversion r=def- a=Robyt3

See #4970:

```
/home/runner/work/ddnet/ddnet/src/base/math.h:50:11: runtime error: left shift of negative value -32
    #0 0x198d955 in i2fx(int) /home/runner/work/ddnet/ddnet/src/base/math.h:50:11
    #1 0x197ebdf in CLayerQuads::NewQuad(int, int, int, int) /home/runner/work/ddnet/ddnet/src/game/editor/layer_quads.cpp:47:22
    #2 0x18ef8d3 in CEditor::Init() /home/runner/work/ddnet/ddnet/src/game/editor/editor.cpp:6360:18
    #3 0xa8e5be in CClient::Run() /home/runner/work/ddnet/ddnet/src/engine/client/client.cpp:2917:13
    #4 0xaf6726 in main /home/runner/work/ddnet/ddnet/src/engine/client/client.cpp:4458:11
    #5 0x7fadd9e610b2 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x240b2)
    #6 0x43cefd in _start (/home/runner/work/ddnet/ddnet/san/DDNet+0x43cefd)
```

Upstream teeworlds/teeworlds#3143.

## Checklist

- [X] Tested the change ingame (on upstream)
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [X] Changed no physics that affect existing maps
- [X] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)


Co-authored-by: Robert Müller <robytemueller@gmail.com>

src/base/math.h

@@ -38,20 +38,31 @@ constexpr inline T mix(const T a, const T b, TB amount)
  return a + (b - a) * amount;
 }
 
-inline float random_float() { return rand() / (float)(RAND_MAX); }
+inline float random_float()
+{
+ return rand() / (float)(RAND_MAX);
+}
+
+constexpr int fxpscale = 1 << 10;
 
 // float to fixed
-constexpr inline int f2fx(float v) { return (int)(v * (float)(1 << 10)); }
-constexpr inline float fx2f(int v) { return v * (1.0f / (1 << 10)); }
+constexpr inline int f2fx(float v)
+{
+ return (int)(v * fxpscale);
+}
+constexpr inline float fx2f(int v)
+{
+ return v / (float)fxpscale;
+}
 
 // int to fixed
-inline int i2fx(int v)
+constexpr inline int i2fx(int v)
 {
- return v << 10;
+ return v * fxpscale;
 }
-inline int fx2i(int v)
+constexpr inline int fx2i(int v)
 {
- return v >> 10;
+ return v / fxpscale;
 }
 
 inline int gcd(int a, int b)
@@ -70,19 +81,32 @@ class fxp
  int value;
 
 public:
- void set(int v) { value = v; }
- int get() const { return value; }
+ void set(int v)
+ {
+ value = v;
+ }
+ int get() const
+ {
+ return value;
+ }
  fxp &operator=(int v)
  {
- value = v << 10;
+ value = i2fx(v);
  return *this;
  }
  fxp &operator=(float v)
  {
- value = (int)(v * (float)(1 << 10));
+ value = f2fx(v);
  return *this;
  }
- operator float() const { return value / (float)(1 << 10); }
+ operator int() const
+ {
+ return fx2i(value);
+ }
+ operator float() const
+ {
+ return fx2f(value);
+ }
 };
 
 constexpr float pi = 3.1415926535897932384626433f;