CPU/Recompiler: Truncate ARM64 multiplication results in LO

Clear the upper product bits after extracting HI so an MFLO register rename cannot carry them into a guest fastmem address.
This commit is contained in:
Stenzek
2026-10-10 21:47:42 +10:00
parent 885d5bbdde
commit 6db5f053d9
+1
View File
@@ -1547,6 +1547,7 @@ void CPU::ARM64Recompiler::Compile_mult(CompileFlags cf, bool sign)
(sign) ? armAsm->smull(lo.X(), rs, rt) : armAsm->umull(lo.X(), rs, rt);
armAsm->lsr(hi.X(), lo.X(), 32);
armAsm->mov(lo, lo); // Clear upper product bits before LO can be renamed to a GPR.
}
void CPU::ARM64Recompiler::Compile_mult(CompileFlags cf)