adding printf regression test

tests/regression/basic/main.cpp

@@ -26,7 +26,7 @@ vx_device_h device = nullptr;
 vx_buffer_h staging_buf = nullptr;
 
 static void show_usage() {
-   std::cout << "Vortex Driver Test." << std::endl;
+   std::cout << "Vortex Test." << std::endl;
    std::cout << "Usage: [-t testno][-k: kernel][-n words][-h: help]" << std::endl;
 }
 
@@ -224,15 +224,13 @@ int main(int argc, char *argv[]) {
     count = 1;
   }
 
-  //kernel_run(count, 1, 1, test, 4, 4);
-
   // open device connection
   std::cout << "open device connection" << std::endl;
   RT_CHECK(vx_dev_open(&device));
   
   unsigned max_cores;
   RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_CORES, &max_cores));
-  uint32_t num_points = 1 * count;
+  uint32_t num_points = count;
   uint32_t num_blocks = (num_points * sizeof(int32_t) + 63) / 64;
   uint32_t buf_size   = num_blocks * 64;
 
@@ -245,7 +243,7 @@ int main(int argc, char *argv[]) {
   RT_CHECK(vx_alloc_dev_mem(device, buf_size, &value));
   kernel_arg.dst_ptr = value;
 
-  kernel_arg.count = count;
+  kernel_arg.count = num_points;
 
   std::cout << "dev_src=" << std::hex << kernel_arg.src_ptr << std::endl;
   std::cout << "dev_dst=" << std::hex << kernel_arg.dst_ptr << std::endl;

tests/regression/demo/main.cpp

@@ -23,7 +23,7 @@ vx_device_h device = nullptr;
 vx_buffer_h staging_buf = nullptr;
 
 static void show_usage() {
-   std::cout << "Vortex Driver Test." << std::endl;
+   std::cout << "Vortex Test." << std::endl;
    std::cout << "Usage: [-k: kernel] [-n words] [-h: help]" << std::endl;
 }
 

tests/regression/dogfood/main.cpp

@@ -88,7 +88,7 @@ vx_buffer_h src2_buf = nullptr;
 vx_buffer_h dst_buf  = nullptr;
 
 static void show_usage() {
-   std::cout << "Vortex Driver Test." << std::endl;
+   std::cout << "Vortex Test." << std::endl;
    std::cout << "Usage: [-t:testid] [-s:testid] [-e:testid] [-k: kernel] [-n words] [-c] [-h: help]" << std::endl;
 }
 

tests/regression/io_addr/Makefile

@@ -2,7 +2,7 @@ RISCV_TOOLCHAIN_PATH ?= /opt/riscv-gnu-toolchain
 VORTEX_DRV_PATH ?= $(realpath ../../../driver)
 VORTEX_RT_PATH ?= $(realpath ../../../runtime)
 
-OPTS ?= -n1
+OPTS ?= -n16
 
 VX_CC  = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-gcc
 VX_CXX = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-g++

tests/regression/io_addr/main.cpp

@@ -32,7 +32,7 @@ vx_device_h device = nullptr;
 vx_buffer_h staging_buf = nullptr;
 
 static void show_usage() {
-   std::cout << "Vortex Driver Test." << std::endl;
+   std::cout << "Vortex Test." << std::endl;
    std::cout << "Usage: [-k: kernel] [-n words] [-h: help]" << std::endl;
 }
 
@@ -148,13 +148,7 @@ int main(int argc, char *argv[]) {
   std::cout << "open device connection" << std::endl;  
   RT_CHECK(vx_dev_open(&device));
 
-  unsigned max_cores, max_warps, max_threads;
-  RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_CORES, &max_cores));
-  RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_WARPS, &max_warps));
-  RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_THREADS, &max_threads));
-
-  uint32_t num_tasks  = max_cores * max_warps * max_threads;
-  uint32_t num_points = count * num_tasks;
+  uint32_t num_points = count;
 
   RT_CHECK(vx_alloc_dev_mem(device, NUM_ADDRS * sizeof(uint32_t), &usr_test_mem));
 

tests/regression/mstress/main.cpp

@@ -75,7 +75,7 @@ vx_device_h device = nullptr;
 vx_buffer_h staging_buf = nullptr;
 
 static void show_usage() {
-   std::cout << "Vortex Driver Test." << std::endl;
+   std::cout << "Vortex Test." << std::endl;
    std::cout << "Usage: [-k: kernel] [-n words] [-h: help]" << std::endl;
 }
 

tests/regression/printf/Makefile

@@ -0,0 +1,70 @@
+RISCV_TOOLCHAIN_PATH ?= /opt/riscv-gnu-toolchain
+VORTEX_DRV_PATH ?= $(realpath ../../../driver)
+VORTEX_RT_PATH ?= $(realpath ../../../runtime)
+
+OPTS ?= -n4
+
+VX_CC  = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-gcc
+VX_CXX = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-g++
+VX_DP  = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-objdump
+VX_CP  = $(RISCV_TOOLCHAIN_PATH)/bin/riscv32-unknown-elf-objcopy
+
+VX_CFLAGS += -march=rv32imf -mabi=ilp32f -O3 -Wstack-usage=1024 -ffreestanding -nostartfiles -fdata-sections -ffunction-sections
+VX_CFLAGS += -I$(VORTEX_RT_PATH)/include -I$(VORTEX_RT_PATH)/../hw
+
+VX_LDFLAGS += -Wl,-Bstatic,-T,$(VORTEX_RT_PATH)/linker/vx_link.ld -Wl,--gc-sections $(VORTEX_RT_PATH)/libvortexrt.a
+
+VX_SRCS = kernel.c
+
+#CXXFLAGS += -std=c++11 -O2 -Wall -Wextra -pedantic -Wfatal-errors
+CXXFLAGS += -std=c++11 -O0 -g -Wall -Wextra -pedantic -Wfatal-errors
+
+CXXFLAGS += -I$(VORTEX_DRV_PATH)/include
+
+LDFLAGS += -L$(VORTEX_DRV_PATH)/stub -lvortex
+
+PROJECT = printf
+
+SRCS = main.cpp
+
+all: $(PROJECT) kernel.bin kernel.dump
+ 
+kernel.dump: kernel.elf
+	$(VX_DP) -D kernel.elf > kernel.dump
+
+kernel.bin: kernel.elf
+	$(VX_CP) -O binary kernel.elf kernel.bin
+
+kernel.elf: $(VX_SRCS)
+	$(VX_CC) $(VX_CFLAGS) $(VX_SRCS) $(VX_LDFLAGS) -o kernel.elf
+
+$(PROJECT): $(SRCS)
+	$(CXX) $(CXXFLAGS) $^ $(LDFLAGS) -o $@
+
+run-simx: $(PROJECT) kernel.bin   
+	LD_LIBRARY_PATH=$(POCL_RT_PATH)/lib:$(VORTEX_DRV_PATH)/simx:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
+	
+run-fpga: $(PROJECT) kernel.bin   
+	LD_LIBRARY_PATH=$(POCL_RT_PATH)/lib:$(VORTEX_DRV_PATH)/opae:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
+
+run-asesim: $(PROJECT) kernel.bin   
+	LD_LIBRARY_PATH=$(POCL_RT_PATH)/lib:$(VORTEX_DRV_PATH)/opae/ase:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
+	
+run-vlsim: $(PROJECT) kernel.bin   
+	LD_LIBRARY_PATH=$(POCL_RT_PATH)/lib:$(VORTEX_DRV_PATH)/opae/vlsim:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
+
+run-rtlsim: $(PROJECT) kernel.bin   
+	LD_LIBRARY_PATH=$(POCL_RT_PATH)/lib:$(VORTEX_DRV_PATH)/rtlsim:$(LD_LIBRARY_PATH) ./$(PROJECT) $(OPTS)
+
+.depend: $(SRCS)
+	$(CXX) $(CXXFLAGS) -MM $^ > .depend;
+
+clean:
+	rm -rf $(PROJECT) *.o .depend
+
+clean-all: clean
+	rm -rf *.elf *.bin *.dump
+
+ifneq ($(MAKECMDGOALS),clean)
+    -include .depend
+endif

tests/regression/printf/common.h

@@ -0,0 +1,11 @@
+#ifndef _COMMON_H_
+#define _COMMON_H_
+
+#define KERNEL_ARG_DEV_MEM_ADDR 0x7ffff000
+
+struct kernel_arg_t {
+  uint32_t num_points;
+  uint32_t src_ptr;
+};
+
+#endif

tests/regression/printf/kernel.c

@@ -0,0 +1,15 @@
+#include <stdint.h>
+#include <vx_intrinsics.h>
+#include <vx_spawn.h>
+#include "common.h"
+
+void kernel_body(int task_id, void* arg) {
+	struct kernel_arg_t* _arg = (struct kernel_arg_t*)(arg);
+	int* src_ptr = (int*)_arg->src_ptr;
+	vx_printf("task=%d, value=%d\n", task_id, src_ptr[task_id]);
+}
+
+void main() {
+	struct kernel_arg_t* arg = (struct kernel_arg_t*)KERNEL_ARG_DEV_MEM_ADDR;
+	vx_spawn_tasks(arg->num_points, kernel_body, arg);
+}

tests/regression/printf/main.cpp

@@ -0,0 +1,149 @@
+#include <iostream>
+#include <unistd.h>
+#include <string.h>
+#include <vortex.h>
+#include "common.h"
+
+#define RT_CHECK(_expr)                                         \
+   do {                                                         \
+     int _ret = _expr;                                          \
+     if (0 == _ret)                                             \
+       break;                                                   \
+     printf("Error: '%s' returned %d!\n", #_expr, (int)_ret);   \
+	 cleanup();			                                              \
+     exit(-1);                                                  \
+   } while (false)
+
+///////////////////////////////////////////////////////////////////////////////
+
+const char* kernel_file = "kernel.bin";
+uint32_t count = 4;
+
+vx_device_h device = nullptr;
+vx_buffer_h staging_buf = nullptr;
+
+static void show_usage() {
+   std::cout << "Vortex Test." << std::endl;
+   std::cout << "Usage: [-k: kernel] [-n words] [-h: help]" << std::endl;
+}
+
+static void parse_args(int argc, char **argv) {
+  int c;
+  while ((c = getopt(argc, argv, "n:k:h?")) != -1) {
+    switch (c) {
+    case 'n':
+      count = atoi(optarg);
+      break;
+    case 'k':
+      kernel_file = optarg;
+      break;
+    case 'h':
+    case '?': {
+      show_usage();
+      exit(0);
+    } break;
+    default:
+      show_usage();
+      exit(-1);
+    }
+  }
+}
+
+void cleanup() {
+  if (staging_buf) {
+    vx_buf_release(staging_buf);
+  }
+  if (device) {
+    vx_dev_close(device);
+  }
+}
+
+int run_test(const kernel_arg_t& kernel_arg,
+             uint32_t buf_size, 
+             uint32_t num_points) {
+  // start device
+  std::cout << "start device" << std::endl;
+  RT_CHECK(vx_start(device));
+
+  // wait for completion
+  std::cout << "wait for completion" << std::endl;
+  RT_CHECK(vx_ready_wait(device, -1));
+
+  return 0;
+}
+
+int main(int argc, char *argv[]) {
+  size_t value; 
+  kernel_arg_t kernel_arg;
+  
+  // parse command arguments
+  parse_args(argc, argv);
+
+  if (count == 0) {
+    count = 1;
+  }
+
+  // open device connection
+  std::cout << "open device connection" << std::endl;  
+  RT_CHECK(vx_dev_open(&device));
+
+  unsigned max_cores, max_warps, max_threads;
+  RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_CORES, &max_cores));
+  RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_WARPS, &max_warps));
+  RT_CHECK(vx_dev_caps(device, VX_CAPS_MAX_THREADS, &max_threads));
+
+  uint32_t num_points = count;
+  uint32_t buf_size = count * sizeof(int32_t);
+
+  std::cout << "number of points: " << count << std::endl;
+  std::cout << "buffer size: " << buf_size << " bytes" << std::endl;
+
+  // upload program
+  std::cout << "upload program" << std::endl;  
+  RT_CHECK(vx_upload_kernel_file(device, kernel_file));
+
+  // allocate device memory
+  std::cout << "allocate device memory" << std::endl;  
+
+  RT_CHECK(vx_alloc_dev_mem(device, buf_size, &value));
+  kernel_arg.src_ptr = value;
+
+  kernel_arg.num_points = num_points;
+
+  std::cout << "dev_src=" << std::hex << kernel_arg.src_ptr << std::endl;
+  
+  // allocate shared memory  
+  std::cout << "allocate shared memory" << std::endl;    
+  uint32_t alloc_size = std::max<uint32_t>(buf_size, sizeof(kernel_arg_t));
+  RT_CHECK(vx_alloc_shared_mem(device, alloc_size, &staging_buf));
+  
+  // upload kernel argument
+  std::cout << "upload kernel argument" << std::endl;
+  {
+    auto buf_ptr = (void*)vx_host_ptr(staging_buf);
+    memcpy(buf_ptr, &kernel_arg, sizeof(kernel_arg_t));
+    RT_CHECK(vx_copy_to_dev(staging_buf, KERNEL_ARG_DEV_MEM_ADDR, sizeof(kernel_arg_t), 0));
+  }
+
+  // upload source buffer0
+  {
+    auto buf_ptr = (float*)vx_host_ptr(staging_buf);
+    for (uint32_t i = 0; i < num_points; ++i) {
+      buf_ptr[i] = i-1;
+    }
+  }
+  std::cout << "upload source buffer" << std::endl;      
+  RT_CHECK(vx_copy_to_dev(staging_buf, kernel_arg.src_ptr, buf_size, 0));  
+
+  // run tests
+  std::cout << "run tests" << std::endl;
+  RT_CHECK(run_test(kernel_arg, buf_size, num_points));
+
+  // cleanup
+  std::cout << "cleanup" << std::endl;  
+  cleanup();
+
+  std::cout << "PASSED!" << std::endl;
+
+  return 0;
+}