|
|
@@ -68,18 +68,25 @@ failed, using an uninitialised `np` — so the loop ran ~forever inside
|
|
|
`ParamInfo` is guarded. Also grew the AST arena/pool (`caeeeb5`) — the
|
|
|
arena/pool data lines are compact `z` runs, so this is nearly free.
|
|
|
|
|
|
-## What remains for the session re-emit
|
|
|
-- **Correctness:** `MaxProc`/`MaxSym` must be ≥ the session's proc/var
|
|
|
- counts (the compiler has hundreds of procs; 64 is too small), so
|
|
|
- procs beyond the cap are currently skipped.
|
|
|
-- **Performance:** `FindSym`/`FindProc` are linear scans, and `Lower`
|
|
|
- looks up per node — O(nodes × table size). A whole-session re-emit is
|
|
|
- very slow. A hash index (or reusing `SymTab` for module-level names)
|
|
|
- would fix it.
|
|
|
-- **Size-dependent behaviour:** with `MaxProc = 1024` the same run
|
|
|
- **segfaults** (in `CopySpan` via `ParamInfo`) — a size/layout-sensitive
|
|
|
- latent bug, the same family as the earlier `ASTDUMP` fragility.
|
|
|
-
|
|
|
-So the session re-emit is capacity-fixable but needs (a) a fast lookup
|
|
|
-and (b) the size-dependent bug isolated. The grammar rewrite remains the
|
|
|
-alternative flip path.
|
|
|
+## Hash index attempted — also trips V3 miscompiles
|
|
|
+Implemented an open-addressing name index (`HashName`/`HashGetSym`/
|
|
|
+`HashGetProc`) for `FindSym`/`FindProc` in `Lower`. Results:
|
|
|
+
|
|
|
+- With the index, the session re-emit on the compiler's sources became
|
|
|
+ **fast** (`MaxProc=512` completed in <300 s; before it hung).
|
|
|
+- But the tables must hold the whole session (the compiler has **713**
|
|
|
+ functions, so `MaxProc ≥ 713`), and at `MaxProc ∈ {768,1024}` the same
|
|
|
+ run is **slow/segfaults** (`CopySpan` via `ParamInfo`) — the
|
|
|
+ size/layout-sensitive bug again.
|
|
|
+- Worse, the hash itself makes the **self-host fail**: with the index in
|
|
|
+ `Lower`, `m2s2` (built by `m1s1`) produces no output on the compiler
|
|
|
+ input (stage 3). So the index triggers *another* latent V3 miscompile.
|
|
|
+
|
|
|
+Reverted to the green state (`a9a8ecd`). Suite 204/204.
|
|
|
+
|
|
|
+## Conclusion
|
|
|
+Three separate latent V3 bugs now block the whole-session re-emit:
|
|
|
+(a) the `ProcInfo` garbage-`np` loop (fixed), (b) a size/layout-sensitive
|
|
|
+miscompile at large tables, (c) a miscompile triggered by the hash code.
|
|
|
+Until (b)/(c) are isolated, the session re-emit is impractical, and the
|
|
|
+**grammar rewrite** remains the more predictable flip path.
|
