From b52fe9795db249e7803aeef22c2ed6129a97aaeb Mon Sep 17 00:00:00 2001
From: Mike Pall <mike>
Date: Thu, 15 Aug 2024 00:07:34 +0200
Subject: [PATCH] Limit CSE for IR_CARG to fix loop optimizations.

Thanks to Peter Cawley.

(cherry picked from commit 3bdc6498c4c012a8fbf9cfa2756a5b07f56f1540)

`IR_CALLXS` for the vararg function contains `IR_CARG(fptr, ctid)` as
the second operand. The `loop_emit_phi()` scans only the first operand
of the IR, so the second is not marked as PHI. In this case, when the IR
appears in both the invariant and variant parts of the loop, CSE may
remove it and thus lead to incorrect emitting results.

This patch tweaks the CSE rules to avoid CSE across the `IR_LOOP`.

Sergey Kaplun:
* added the description and the test for the problem

Part of tarantool/tarantool#10199

Reviewed-by: Sergey Bronnikov <sergeyb@tarantool.org>
Reviewed-by: Maxim Kokryashkin <m.kokryashkin@tarantool.org>
Signed-off-by: Sergey Kaplun <skaplun@tarantool.org>
---
 src/lj_opt_fold.c                             | 11 ++++
 .../lj-1244-missing-phi-carg.test.lua         | 56 +++++++++++++++++++
 2 files changed, 67 insertions(+)
 create mode 100644 test/tarantool-tests/lj-1244-missing-phi-carg.test.lua

diff --git a/src/lj_opt_fold.c b/src/lj_opt_fold.c
index 2702f79f12..cd4395bbfa 100644
--- a/src/lj_opt_fold.c
+++ b/src/lj_opt_fold.c
@@ -2406,6 +2406,17 @@ LJFOLD(XSNEW any any)
 LJFOLD(BUFHDR any any)
 LJFOLDX(lj_ir_emit)
 
+/* -- Miscellaneous ------------------------------------------------------- */
+
+LJFOLD(CARG any any)
+LJFOLDF(cse_carg)
+{
+  TRef tr = lj_opt_cse(J);
+  if (tref_ref(tr) < J->chain[IR_LOOP])  /* CSE across loop? */
+    return EMITFOLD;  /* Raw emit. Assumes fins is left intact by CSE. */
+  return tr;
+}
+
 /* ------------------------------------------------------------------------ */
 
 /* Every entry in the generated hash table is a 32 bit pattern:
diff --git a/test/tarantool-tests/lj-1244-missing-phi-carg.test.lua b/test/tarantool-tests/lj-1244-missing-phi-carg.test.lua
new file mode 100644
index 0000000000..d498129901
--- /dev/null
+++ b/test/tarantool-tests/lj-1244-missing-phi-carg.test.lua
@@ -0,0 +1,56 @@
+local ffi = require('ffi')
+local table_new = require('table.new')
+
+-- Test file to demonstrate LuaJIT incorrect behaviour for
+-- recording the FFI call to the vararg function. See also:
+-- https://github.com/LuaJIT/LuaJIT/issues/1244.
+local tap = require('tap')
+local test = tap.test('lj-1244-missing-phi-carg'):skipcond({
+  ['Test requires JIT enabled'] = not jit.status(),
+})
+
+-- Loop unrolls into 2 iterations. Thus means that the loop is
+-- executed on trace on the 5th iteration (instead of the usual
+-- 4th). Run it even number of iterations to test both, so last is
+-- 6th.
+local NTESTS = 6
+
+test:plan(NTESTS)
+
+-- XXX: Hack with function's prototypes to avoid creation of
+-- custom functions to be loaded via FFI (vararg part will be just
+-- ignored).
+ffi.cdef[[
+  double sin(double, ...);
+  double cos(double, ...);
+]]
+
+local EXPECTED = {[0] = ffi.C.sin(0), ffi.C.cos(0)}
+
+-- Array of 2 functions.
+local fns = ffi.new('double (*[2])(double, ...)')
+fns[0] = ffi.C.cos
+fns[1] = ffi.C.sin
+
+-- Avoid reallocating the table on the trace.
+local result = table_new(8, 0)
+
+jit.opt.start('hotloop=1')
+
+local fn = fns[0]
+-- The first result is `cos()`.
+for i = 1, NTESTS do
+  result[i] = fn(0)
+  fn = fns[i % 2]
+  -- The call persists in the invariant part of the loop as well.
+  -- Hence, XLOAD (part of the IR_CARG -- function to be called)
+  -- should be marked as PHI, but it isn't due to CSE.
+  fn(0)
+end
+
+for i = 1, NTESTS do
+  test:is(result[i], EXPECTED[i % 2],
+          ('correct result on iteration %d'):format(i))
+end
+
+test:done(true)