From fbf2c9a2fba42a48112606a28b196605f9227e04 Mon Sep 17 00:00:00 2001
From: whitequark <whitequark@whitequark.org>
Date: Fri, 25 May 2018 02:18:13 +0000
Subject: [PATCH] compiler: hoist loads of kernel invariants to function entry
 block.

Addresses intraprocedural cases of #1007.
---
 artiq/compiler/module.py                      |  2 +
 artiq/compiler/transforms/__init__.py         |  1 +
 artiq/compiler/transforms/constant_hoister.py | 44 +++++++++++++++++++
 artiq/test/lit/constant_hoisting/device_db.py |  8 ++++
 .../lit/constant_hoisting/invariant_load.py   | 25 +++++++++++
 5 files changed, 80 insertions(+)
 create mode 100644 artiq/compiler/transforms/constant_hoister.py
 create mode 100644 artiq/test/lit/constant_hoisting/device_db.py
 create mode 100644 artiq/test/lit/constant_hoisting/invariant_load.py

diff --git a/artiq/compiler/module.py b/artiq/compiler/module.py
index 8e7091b81..cea86cddb 100644
--- a/artiq/compiler/module.py
+++ b/artiq/compiler/module.py
@@ -61,6 +61,7 @@ class Module:
         dead_code_eliminator = transforms.DeadCodeEliminator(engine=self.engine)
         local_access_validator = validators.LocalAccessValidator(engine=self.engine)
         local_demoter = transforms.LocalDemoter()
+        constant_hoister = transforms.ConstantHoister()
         devirtualization = analyses.Devirtualization()
         interleaver = transforms.Interleaver(engine=self.engine)
         invariant_detection = analyses.InvariantDetection(engine=self.engine)
@@ -79,6 +80,7 @@ class Module:
         interleaver.process(self.artiq_ir)
         local_access_validator.process(self.artiq_ir)
         local_demoter.process(self.artiq_ir)
+        constant_hoister.process(self.artiq_ir)
         if remarks:
             invariant_detection.process(self.artiq_ir)
 
diff --git a/artiq/compiler/transforms/__init__.py b/artiq/compiler/transforms/__init__.py
index 64ee25c46..5696be95f 100644
--- a/artiq/compiler/transforms/__init__.py
+++ b/artiq/compiler/transforms/__init__.py
@@ -6,6 +6,7 @@ from .iodelay_estimator import IODelayEstimator
 from .artiq_ir_generator import ARTIQIRGenerator
 from .dead_code_eliminator import DeadCodeEliminator
 from .local_demoter import LocalDemoter
+from .constant_hoister import ConstantHoister
 from .interleaver import Interleaver
 from .typedtree_printer import TypedtreePrinter
 from .llvm_ir_generator import LLVMIRGenerator
diff --git a/artiq/compiler/transforms/constant_hoister.py b/artiq/compiler/transforms/constant_hoister.py
new file mode 100644
index 000000000..0484b1c39
--- /dev/null
+++ b/artiq/compiler/transforms/constant_hoister.py
@@ -0,0 +1,44 @@
+"""
+:class:`ConstantHoister` is a code motion transform:
+it moves any invariant loads to the earliest point where
+they may be executed.
+"""
+
+from .. import types, ir
+
+class ConstantHoister:
+    def process(self, functions):
+        for func in functions:
+            self.process_function(func)
+
+    def process_function(self, func):
+        entry = func.entry()
+        worklist = set(func.instructions())
+        moved = set()
+        while len(worklist) > 0:
+            insn = worklist.pop()
+
+            if (isinstance(insn, ir.GetAttr) and insn not in moved and
+                    types.is_instance(insn.object().type) and
+                    insn.attr in insn.object().type.constant_attributes):
+                has_variant_operands = False
+                index_in_entry = 0
+                for operand in insn.operands:
+                    if isinstance(operand, ir.Argument):
+                        pass
+                    elif isinstance(operand, ir.Instruction) and operand.basic_block == entry:
+                        index_in_entry = entry.index(operand) + 1
+                    else:
+                        has_variant_operands = True
+                        break
+
+                if has_variant_operands:
+                    continue
+
+                insn.remove_from_parent()
+                entry.instructions.insert(index_in_entry, insn)
+                moved.add(insn)
+                print(insn)
+
+                for use in insn.uses:
+                    worklist.add(use)
diff --git a/artiq/test/lit/constant_hoisting/device_db.py b/artiq/test/lit/constant_hoisting/device_db.py
new file mode 100644
index 000000000..e39c83c09
--- /dev/null
+++ b/artiq/test/lit/constant_hoisting/device_db.py
@@ -0,0 +1,8 @@
+device_db = {
+    "core": {
+        "type": "local",
+        "module": "artiq.coredevice.core",
+        "class": "Core",
+        "arguments": {"host": None, "ref_period": 1e-9}
+    }
+}
diff --git a/artiq/test/lit/constant_hoisting/invariant_load.py b/artiq/test/lit/constant_hoisting/invariant_load.py
new file mode 100644
index 000000000..62fac4202
--- /dev/null
+++ b/artiq/test/lit/constant_hoisting/invariant_load.py
@@ -0,0 +1,25 @@
+# RUN: env ARTIQ_DUMP_IR=%t ARTIQ_IR_NO_LOC=1 %python -m artiq.compiler.testbench.embedding +compile %s
+# RUN: OutputCheck %s --file-to-check=%t.txt
+
+from artiq.language.core import *
+from artiq.language.types import *
+
+# CHECK-L:   %LOC.self.FLD.foo = numpy.int32 getattr('foo') <instance testbench.c> %ARG.self
+# CHECK-L: for.head:
+
+class c:
+    kernel_invariants = {"foo"}
+
+    def __init__(self):
+        self.foo = 1
+
+    @kernel
+    def run(self):
+        for _ in range(10):
+            core_log(1.0 * self.foo)
+
+i = c()
+
+@kernel
+def entrypoint():
+    i.run()