Add support for Makefile build for micro-controller platforms

PiperOrigin-RevId: 202716942

3 files changed, 94 insertions, 8 deletions

diff --git a/tensorflow/contrib/lite/Makefile b/tensorflow/contrib/lite/Makefile
index 2b6997146e..a616138d33 100644
--- a/tensorflow/contrib/lite/Makefile
+++ b/tensorflow/contrib/lite/Makefile
@@ -17,7 +17,29 @@ else
 	endif
 endif
 
-ARCH := $(shell if [[ $(shell uname -m) =~ i[345678]86 ]]; then echo x86_32; else echo $(shell uname -m); fi)
+HOST_ARCH := $(shell if [[ $(shell uname -m) =~ i[345678]86 ]]; then echo x86_32; else echo $(shell uname -m); fi)
+
+# Self-hosting
+TARGET_ARCH := ${HOST_ARCH}
+
+# Cross compiling
+ifeq ($(CROSS),rpi)
+  TARGET_ARCH := armv7l
+  TARGET_TOOLCHAIN_PREFIX := arm-linux-gnueabihf-
+endif
+
+ifeq ($(CROSS),riscv)
+  TARGET_ARCH := riscv
+  TARGET_TOOLCHAIN_PREFIX := riscv32-unknown-elf-
+endif
+ifeq ($(CROSS),stm32f7)
+  TARGET_ARCH := armf7
+  TARGET_TOOLCHAIN_PREFIX := arm-none-eabi-
+endif
+ifeq ($(CROSS),stm32f1)
+  TARGET_ARCH := armm1
+  TARGET_TOOLCHAIN_PREFIX := arm-none-eabi-
+endif
 
 # Where compiled objects are stored.
 OBJDIR := $(MAKEFILE_DIR)/gen/obj/
@@ -25,11 +47,46 @@ BINDIR := $(MAKEFILE_DIR)/gen/bin/
 LIBDIR := $(MAKEFILE_DIR)/gen/lib/
 GENDIR := $(MAKEFILE_DIR)/gen/obj/
 
+LIBS :=
+ifeq ($(TARGET_ARCH),x86_64)
+        CXXFLAGS += -fPIC -DGEMMLOWP_ALLOW_SLOW_SCALAR_FALLBACK -pthread # -msse4.2
+endif
+
+ifeq ($(TARGET_ARCH),armv7l)
+        CXXFLAGS += -mfpu=neon -pthread -fPIC
+	LIBS += -ldl
+endif
+
+ifeq ($(TARGET_ARCH),riscv)
+#        CXXFLAGS += -march=gap8
+        CXXFLAGS += -DTFLITE_MCU
+	LIBS += -ldl
+	BUILD_TYPE := micro
+endif
+
+ifeq ($(TARGET_ARCH),armf7)
+        CXXFLAGS += -DGEMMLOWP_ALLOW_SLOW_SCALAR_FALLBACK -DTFLITE_MCU
+        CXXFLAGS += -fno-rtti -fmessage-length=0 -fno-exceptions -fno-builtin -ffunction-sections -fdata-sections
+        CXXFLAGS += -funsigned-char -MMD
+        CXXFLAGS += -mcpu=cortex-m7 -mthumb -mfpu=fpv5-sp-d16 -mfloat-abi=softfp
+        CXXFLAGS += '-std=gnu++11' '-fno-rtti' '-Wvla' '-c' '-Wall' '-Wextra' '-Wno-unused-parameter' '-Wno-missing-field-initializers' '-fmessage-length=0' '-fno-exceptions' '-fno-builtin' '-ffunction-sections' '-fdata-sections' '-funsigned-char' '-MMD' '-fno-delete-null-pointer-checks' '-fomit-frame-pointer' '-Os'
+	LIBS += -ldl
+	BUILD_TYPE := micro
+endif
+ifeq ($(TARGET_ARCH),armm1)
+        CXXFLAGS += -DGEMMLOWP_ALLOW_SLOW_SCALAR_FALLBACK -mcpu=cortex-m1 -mthumb -DTFLITE_MCU
+        CXXFLAGS += -fno-rtti -fmessage-length=0 -fno-exceptions -fno-builtin -ffunction-sections -fdata-sections
+        CXXFLAGS += -funsigned-char -MMD
+	LIBS += -ldl
+endif
+
 # Settings for the host compiler.
-CXX := $(CC_PREFIX)gcc
-CXXFLAGS := --std=c++11 -O3 -DNDEBUG
-CC := $(CC_PREFIX)gcc
-CCFLAGS := -O3 -DNDEBUG
+CXX := $(CC_PREFIX) ${TARGET_TOOLCHAIN_PREFIX}g++
+CXXFLAGS += --std=c++11 -O3 -DNDEBUG
+CCFLAGS := ${CXXFLAGS}
+CC := $(CC_PREFIX) ${TARGET_TOOLCHAIN_PREFIX}gcc
+AR := $(CC_PREFIX) ${TARGET_TOOLCHAIN_PREFIX}ar
+CFLAGS :=
 LDOPTS :=
 LDOPTS += -L/usr/local/lib
 ARFLAGS := -r
@@ -48,7 +105,7 @@ INCLUDES := \
 # override local versions in the source tree.
 INCLUDES += -I/usr/local/include
 
-LIBS := \
+LIBS += \
 -lstdc++ \
 -lpthread \
 -lm \
@@ -92,18 +149,21 @@ PROFILE_SUMMARIZER_SRCS := \
 
 CORE_CC_ALL_SRCS := \
 $(wildcard tensorflow/contrib/lite/*.cc) \
+$(wildcard tensorflow/contrib/lite/*.c)
+ifneq ($(BUILD_TYPE),micro)
+CORE_CC_ALL_SRCS += \
 $(wildcard tensorflow/contrib/lite/kernels/*.cc) \
 $(wildcard tensorflow/contrib/lite/kernels/internal/*.cc) \
 $(wildcard tensorflow/contrib/lite/kernels/internal/optimized/*.cc) \
 $(wildcard tensorflow/contrib/lite/kernels/internal/reference/*.cc) \
 $(PROFILER_SRCS) \
-$(wildcard tensorflow/contrib/lite/*.c) \
 $(wildcard tensorflow/contrib/lite/kernels/*.c) \
 $(wildcard tensorflow/contrib/lite/kernels/internal/*.c) \
 $(wildcard tensorflow/contrib/lite/kernels/internal/optimized/*.c) \
 $(wildcard tensorflow/contrib/lite/kernels/internal/reference/*.c) \
 $(wildcard tensorflow/contrib/lite/downloads/farmhash/src/farmhash.cc) \
 $(wildcard tensorflow/contrib/lite/downloads/fft2d/fftsg.c)
+endif
 # Remove any duplicates.
 CORE_CC_ALL_SRCS := $(sort $(CORE_CC_ALL_SRCS))
 CORE_CC_EXCLUDE_SRCS := \
@@ -113,6 +173,11 @@ $(wildcard tensorflow/contrib/lite/*/*/*test.cc) \
 $(wildcard tensorflow/contrib/lite/*/*/*/*test.cc) \
 $(wildcard tensorflow/contrib/lite/kernels/test_util.cc) \
 $(MINIMAL_SRCS)
+ifeq ($(BUILD_TYPE),micro)
+CORE_CC_EXCLUDE_SRCS += \
+tensorflow/contrib/lite/model.cc \
+tensorflow/contrib/lite/nnapi_delegate.cc
+endif
 # Filter out all the excluded files.
 TF_LITE_CC_SRCS := $(filter-out $(CORE_CC_EXCLUDE_SRCS), $(CORE_CC_ALL_SRCS))
 # File names of the intermediate files target compilation generates.
@@ -120,7 +185,6 @@ TF_LITE_CC_OBJS := $(addprefix $(OBJDIR), \
 $(patsubst %.cc,%.o,$(patsubst %.c,%.o,$(TF_LITE_CC_SRCS))))
 LIB_OBJS := $(TF_LITE_CC_OBJS)
 
-
 # Benchmark sources
 BENCHMARK_SRCS_DIR := tensorflow/contrib/lite/tools/benchmark
 BENCHMARK_ALL_SRCS := $(TFLITE_CC_SRCS) \
@@ -146,6 +210,9 @@ $(OBJDIR)%.o: %.c
 # The target that's compiled if there's no command-line arguments.
 all: $(LIB_PATH)  $(MINIMAL_PATH) $(BENCHMARK_BINARY)
 
+# The target that's compiled for micro-controllers
+micro: $(LIB_PATH)
+
 # Gathers together all the objects we've compiled into a single '.a' archive.
 $(LIB_PATH): $(LIB_OBJS)
 	@mkdir -p $(dir $@)
diff --git a/tensorflow/contrib/lite/allocation.cc b/tensorflow/contrib/lite/allocation.cc
index a4772731ec..c42622ff02 100644
--- a/tensorflow/contrib/lite/allocation.cc
+++ b/tensorflow/contrib/lite/allocation.cc
@@ -14,7 +14,9 @@ limitations under the License.
 ==============================================================================*/
 
 #include <fcntl.h>
+#ifndef TFLITE_MCU
 #include <sys/mman.h>
+#endif
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <unistd.h>
@@ -27,10 +29,13 @@ limitations under the License.
 #include "tensorflow/contrib/lite/allocation.h"
 #include "tensorflow/contrib/lite/context.h"
 #include "tensorflow/contrib/lite/error_reporter.h"
+#ifndef TFLITE_MCU
 #include "tensorflow/contrib/lite/nnapi_delegate.h"
+#endif
 
 namespace tflite {
 
+#ifndef TFLITE_MCU
 MMAPAllocation::MMAPAllocation(const char* filename,
                                ErrorReporter* error_reporter)
     : Allocation(error_reporter), mmapped_buffer_(MAP_FAILED) {
@@ -111,6 +116,7 @@ MemoryAllocation::MemoryAllocation(const void* ptr, size_t num_bytes,
   buffer_ = ptr;
   buffer_size_bytes_ = num_bytes;
 }
+#endif
 
 MemoryAllocation::~MemoryAllocation() {}
 
diff --git a/tensorflow/contrib/lite/interpreter.cc b/tensorflow/contrib/lite/interpreter.cc
index 62a0b1ff08..4b3ba5df10 100644
--- a/tensorflow/contrib/lite/interpreter.cc
+++ b/tensorflow/contrib/lite/interpreter.cc
@@ -24,15 +24,22 @@ limitations under the License.
 #include "tensorflow/contrib/lite/context.h"
 #include "tensorflow/contrib/lite/error_reporter.h"
 #include "tensorflow/contrib/lite/graph_info.h"
+#ifndef TFLITE_MCU
 #include "tensorflow/contrib/lite/kernels/eigen_support.h"
 #include "tensorflow/contrib/lite/kernels/gemm_support.h"
+#endif
 #include "tensorflow/contrib/lite/memory_planner.h"
+#ifndef TFLITE_MCU
 #include "tensorflow/contrib/lite/nnapi_delegate.h"
+#endif
 #include "tensorflow/contrib/lite/profiling/profiler.h"
 #include "tensorflow/contrib/lite/schema/schema_generated.h"
 #include "tensorflow/contrib/lite/util.h"
 
 namespace tflite {
+#ifdef TFLITE_MCU
+class NNAPIDelegate {};
+#endif
 
 namespace {
 
@@ -557,6 +564,7 @@ TfLiteStatus Interpreter::Invoke() {
   }
 
   TfLiteStatus status = kTfLiteOk;
+#ifndef TFLITE_MCU
   if (nnapi_delegate_) {
     if (next_execution_plan_index_to_prepare_ == execution_plan_.size()) {
       TF_LITE_ENSURE_OK(&context_, nnapi_delegate_->Invoke(this));
@@ -570,6 +578,7 @@ TfLiteStatus Interpreter::Invoke() {
       return kTfLiteError;
     }
   }
+#endif
 
   // Invocations are always done in node order.
   // Note that calling Invoke repeatedly will cause the original memory plan to
@@ -826,6 +835,7 @@ TfLiteStatus Interpreter::ResizeTensorImpl(TfLiteTensor* tensor,
 }
 
 void Interpreter::UseNNAPI(bool enable) {
+#ifndef TFLITE_MCU
   // TODO(aselle): This is a workaround for finding if NNAPI exists.
   // We also need to make sure getLibraryHandle() is renamed to be NNAPI
   // prefixed.
@@ -835,6 +845,7 @@ void Interpreter::UseNNAPI(bool enable) {
   } else if (!nnapi_delegate_) {
     nnapi_delegate_.reset(new NNAPIDelegate);
   }
+#endif
 }
 
 void Interpreter::SetNumThreads(int num_threads) {
@@ -842,8 +853,10 @@ void Interpreter::SetNumThreads(int num_threads) {
 
   // TODO(ahentz): find a way to avoid this. It causes gemmlowp and eigen to
   // be required in order to compile the framework.
+#ifndef TFLITE_MCU
   gemm_support::SetNumThreads(&context_, num_threads);
   eigen_support::SetNumThreads(&context_, num_threads);
+#endif
 }
 
 TfLiteStatus Interpreter::ModifyGraphWithDelegate(TfLiteDelegate* delegate,