nx_js_jit_keystone.nx

nx_js_jit_keystone.nx -- P5 BASELINE-JIT KEYSTONE (SOTA ladder, after P4 ICs; mirrors how the VM arc itself began: nx_js_vm_bench proved 376x BEFORE the compiler was built). QUESTION: can the sovereign stack EMIT x86-64 at runtime and EXECUTE it as the third engine tier, and which class band does native code hit on the SAME kernel the SOTA bench measures? Machinery already probed GREEN by nx_nxe_exec_probe (RWX mmap prot=7; SysV: arg0 rdi, ret rax; cast fn-ptr call). This emits the K1 dispatch kernel (sum 0..N-1, a REAL counted loop) as 18 bytes of x86-64: 31 c0 xor eax, eax ; sum = 0 31 c9 xor ecx, ecx ; i = 0 48 39 f9 loop: cmp rcx, rdi ; i vs N 7d 08 jge end 48 01 c8 add rax, rcx ; sum += i 48 ff c1 inc rcx ; i++ eb f3 jmp loop c3 end: ret and races it against the SAME semantics through the bytecode VM (compile_run) and the tree-walker. HONEST FRAME: this is a KEYSTONE -- a hand-templated kernel proving emission+execution+the band -- NOT a JIT. The Sparkplug-tier build (per-opcode x86 templates off our bytecode) rides this proof. expect_exit: 0 license_tier: ORIGINAL