Floats leave through a different gate
Going int → float was ferry-then-convert. Coming back, the order flips: convert first, on the FPU, then ferry the result out. Here's chop(x), which truncates an f32 to s32:
0: trunc.w.s ft0, fa0 # float → whole number, still FPU-side
4: mfc1 v0, ft0 # ferry the bits out to the integer bank
8: nop # hazard pad on the way out, too
c: jr ra
10: nop
trunc.w.s — truncate to word, from single — does the arithmetic: it chops toward zero, exactly the semantics of a C cast ((s32)3.9f is 3, (s32)-3.9f is -3). That matching semantics is why a bare cast compiles to trunc and nothing else.mfc1 — move from coprocessor 1 — is the return ferry, mtc1's mirror. Raw bits out, unchanged, into v0.- The trailing
nop is the crossing toll again, paid on this side of the border this time.
So the two crossings, as reflexes: mtc1 + cvt.s.w = int in, trunc.w.s + mfc1 = int out. Spot either pair and you know a cast (or silent promotion) sits at that point in the C — one of the most load-bearing fingerprints in float-heavy game code.
In the target, the cast wraps computed float work: one .s operation runs first, and its result — not a bare argument — takes the exit. Mind what the operation is and what it consumes.
Your task
Write func_80060eb4 to reproduce the target assembly.