Run a block through a frame and step, and compile blocks to AArch64 that call the same steps (021)

- run() keeps a block's operand stack and open loops in a Frame; step()
  runs one instruction (branches and loops too, one dispatch), and
  enter_each, enter_loop, leave_loop and iterate are their own functions.
  The executor is the same machine code, 7% fewer instructions.
- The verifier records each instruction's stack height (from its block's
  start), so compiled code knows each slot's place.
- module-jit.c (AArch64 Linux; elsewhere, or PURE_JIT=0, none): a
  module's handlers', views' and folds' blocks compiled when it loads.
  Branches and loops are branches; every other instruction calls the
  executor's own step, so both run the same meaning. A loop that is not
  opened lands where its iteration would have ended; an empty loop
  iterates in place.
- test-module recompiles a module whose code a test changes after load.
- A search (100,000 nodes, 1,000,000 edges, core only) is 150 ms
  compiled, against 122 ms executed: a call of step for every instruction
  costs more than the executor's switch; the hot instructions are written
  as code next.

1 files changed, 2 insertions, 0 deletions

diff --git a/core/module.h b/core/module.h
index 9562358..9b425b2 100644
--- a/core/module.h
+++ b/core/module.h
@@ -255,6 +255,8 @@ typedef struct {
      * (numbers by value, texts by bytes), each key once, and keys do not change. */
     uint8_t state_key[PURE_MODULE_STATES];
     bool state_indexed[PURE_MODULE_STATES]; /* list: an index state refers to it (set at load) */
+    uint8_t *height;    /* the operand stack's height at each instruction, from its block's start (the verifier's) */
+    struct PureJit *jit; /* its blocks compiled (module-jit.c, docs/design/021), or NULL */
     PureInputShape input[PURE_MODULE_INPUTS];
     PureField field[PURE_MODULE_FIELDS];
     PureValue initial[PURE_MODULE_STATES];