core_portme.c for ZedBoard

core_portme.c

/* 
	File: core_portme.c
*/
/*
	Author : Shay Gal-On, EEMBC
	Legal : TODO!
*/ 
#include <stdio.h>
#include <stdlib.h>
#include "coremark.h"
#if CALLGRIND_RUN
#include <valgrind/callgrind.h>
#endif

#define TIMER_DEVICE_ID		XPAR_XSCUTIMER_0_DEVICE_ID
#define TIMER_LOAD_VALUE	0xFFFFFFE
#define START_COUNTER		0
#define STOP_COUNTER		1

#define NSECS_PER_SEC 		667000000/2
#define CORETIMETYPE 		ee_u32
#define TIMER_RES_DIVIDER 	40
#define EE_TICKS_PER_SEC 	(NSECS_PER_SEC / TIMER_RES_DIVIDER)

typedef ee_u32 CORE_TICKS;

#include "xil_printf.h"
#include "xscutimer.h"
#include "xscugic.h"
#include "xparameters.h"

ee_u32 GetTimerValue(ee_u32 TimerIntrId, ee_u16 Mode);
static XScuTimer Timer;//timer



#if (MEM_METHOD==MEM_MALLOC)
#include <malloc.h>
/* Function: portable_malloc
	Provide malloc() functionality in a platform specific way.
*/
void *portable_malloc(size_t size) {
	return malloc(size);
}
/* Function: portable_free
	Provide free() functionality in a platform specific way.
*/
void portable_free(void *p) {
	free(p);
}
#else
void *portable_malloc(size_t size) {
	return NULL;
}
void portable_free(void *p) {
	p=NULL;
}
#endif


#if (SEED_METHOD==SEED_VOLATILE)
#if VALIDATION_RUN
	volatile ee_s32 seed1_volatile=0x3415;
	volatile ee_s32 seed2_volatile=0x3415;
	volatile ee_s32 seed3_volatile=0x66;
#endif
#if PERFORMANCE_RUN
	volatile ee_s32 seed1_volatile=0x0;
	volatile ee_s32 seed2_volatile=0x0;
	volatile ee_s32 seed3_volatile=0x66;
#endif
#if PROFILE_RUN
	volatile ee_s32 seed1_volatile=0x8;
	volatile ee_s32 seed2_volatile=0x8;
	volatile ee_s32 seed3_volatile=0x8;
#endif
	volatile ee_s32 seed4_volatile=ITERATIONS;
	volatile ee_s32 seed5_volatile=0;
#endif
/* Porting: Timing functions
	How to capture time and convert to seconds must be ported to whatever is supported by the platform.
	e.g. Read value from on board RTC, read value from cpu clock cycles performance counter etc. 
	Sample implementation for standard time.h and windows.h definitions included.
*/
/* Define: TIMER_RES_DIVIDER
	Divider to trade off timer resolution and total time that can be measured.

	Use lower values to increase resolution, but make sure that overflow does not occur.
	If there are issues with the return value overflowing, increase this value.
	*/
#if USE_CLOCK
	#define NSECS_PER_SEC CLOCKS_PER_SEC
	#define EE_TIMER_TICKER_RATE 1000
	#define CORETIMETYPE clock_t 
	#define GETMYTIME(_t) (*_t=clock())
	#define MYTIMEDIFF(fin,ini) ((fin)-(ini))
	#define TIMER_RES_DIVIDER 1
	#define SAMPLE_TIME_IMPLEMENTATION 1
#elif defined(_MSC_VER)
	#define NSECS_PER_SEC 10000000
	#define EE_TIMER_TICKER_RATE 1000
	#define CORETIMETYPE FILETIME
	#define GETMYTIME(_t) GetSystemTimeAsFileTime(_t)
	#define MYTIMEDIFF(fin,ini) (((*(__int64*)&fin)-(*(__int64*)&ini))/TIMER_RES_DIVIDER)
	/* setting to millisces resolution by default with MSDEV */
	#ifndef TIMER_RES_DIVIDER
	#define TIMER_RES_DIVIDER 1000
	#endif
	#define SAMPLE_TIME_IMPLEMENTATION 1
#elif HAS_TIME_H
	#define NSECS_PER_SEC 1000000000
	#define EE_TIMER_TICKER_RATE 1000
	// #define CORETIMETYPE struct timespec
	#define CORETIMETYPE ee_u32
	#define GETMYTIME(_t) clock_gettime(CLOCK_REALTIME,_t)
	#define MYTIMEDIFF(fin,ini) ((fin.tv_sec-ini.tv_sec)*(NSECS_PER_SEC/TIMER_RES_DIVIDER)+(fin.tv_nsec-ini.tv_nsec)/TIMER_RES_DIVIDER)
	/* setting to 1/1000 of a second resolution by default with linux */
	#ifndef TIMER_RES_DIVIDER
	#define TIMER_RES_DIVIDER 1000000
	#endif
	#define SAMPLE_TIME_IMPLEMENTATION 1
#else
//	#define CORETIMETYPE ee_u32
//	#define GETMYTIME(_t) clock_gettime(CLOCK_REALTIME,_t)
	#define SAMPLE_TIME_IMPLEMENTATION 1
//	#define SAMPLE_TIME_IMPLEMENTATION 0
#endif
#define EE_TICKS_PER_SEC (NSECS_PER_SEC / TIMER_RES_DIVIDER)

#if SAMPLE_TIME_IMPLEMENTATION
/** Define Host specific (POSIX), or target specific global time variables. */
static CORETIMETYPE start_time_val, stop_time_val;

/* Function: start_time
	This function will be called right before starting the timed portion of the benchmark.

	Implementation may be capturing a system timer (as implemented in the example code) 
	or zeroing some system parameters - e.g. setting the cpu clocks cycles to 0.
*/
void start_time(void) {
//	GETMYTIME(&start_time_val );
	start_time_val = GetTimerValue (TIMER_DEVICE_ID, START_COUNTER);

#if CALLGRIND_RUN
	CALLGRIND_START_INSTRUMENTATION
#endif
#if MICA
    asm volatile("int3");/*1 */
#endif
}
/* Function: stop_time
	This function will be called right after ending the timed portion of the benchmark.

	Implementation may be capturing a system timer (as implemented in the example code) 
	or other system parameters - e.g. reading the current value of cpu cycles counter.
*/
void stop_time(void) {
#if CALLGRIND_RUN
	 CALLGRIND_STOP_INSTRUMENTATION 
#endif
#if MICA
    asm volatile("int3");/*1 */
#endif
	// GETMYTIME(&stop_time_val );
	stop_time_val = GetTimerValue (TIMER_DEVICE_ID, STOP_COUNTER);
}
/* Function: get_time
	Return an abstract "ticks" number that signifies time on the system.
	
	Actual value returned may be cpu cycles, milliseconds or any other value,
	as long as it can be converted to seconds by <time_in_secs>.
	This methodology is taken to accomodate any hardware or simulated platform.
	The sample implementation returns millisecs by default, 
	and the resolution is controlled by <TIMER_RES_DIVIDER>
*/
CORE_TICKS get_time(void) {
	// CORE_TICKS elapsed=(CORE_TICKS)(MYTIMEDIFF(stop_time_val, start_time_val));
	CORE_TICKS elapsed = start_time_val - stop_time_val;
	return elapsed;
}
/* Function: time_in_secs
	Convert the value returned by get_time to seconds.

	The <secs_ret> type is used to accomodate systems with no support for floating point.
	Default implementation implemented by the EE_TICKS_PER_SEC macro above.
*/
secs_ret time_in_secs(CORE_TICKS ticks) {
	secs_ret retval=((secs_ret)ticks) / (secs_ret)EE_TICKS_PER_SEC;
	return retval;
}
#else 
#error "Please implement timing functionality in core_portme.c"
#endif /* SAMPLE_TIME_IMPLEMENTATION */

ee_u32 default_num_contexts=MULTITHREAD;

/* Function: portable_init
	Target specific initialization code 
	Test for some common mistakes.
*/
void portable_init(core_portable *p, int *argc, char *argv[])
{
#if PRINT_ARGS
	int i;
	for (i=0; i<*argc; i++) {
		ee_printf("Arg[%d]=%s\n",i,argv[i]);
	}
#endif
	if (sizeof(ee_ptr_int) != sizeof(ee_u8 *)) {
		ee_printf("ERROR! Please define ee_ptr_int to a type that holds a pointer!\n");
	}
	if (sizeof(ee_u32) != 4) {
		ee_printf("ERROR! Please define ee_u32 to a 32b unsigned type!\n");
	}
#if (MAIN_HAS_NOARGC && (SEED_METHOD==SEED_ARG))
	ee_printf("ERROR! Main has no argc, but SEED_METHOD defined to SEED_ARG!\n");
#endif
	
#if (MULTITHREAD>1) && (SEED_METHOD==SEED_ARG)
	{
		int nargs=*argc,i;
		if ((nargs>1) && (*argv[1]=='M')) {
			default_num_contexts=parseval(argv[1]+1);
			if (default_num_contexts>MULTITHREAD)
				default_num_contexts=MULTITHREAD;
			/* Shift args since first arg is directed to the portable part and not to coremark main */
			--nargs;
			for (i=1; i<nargs; i++)
				argv[i]=argv[i+1];
			*argc=nargs;
		}
	}
#endif /* sample of potential platform specific init via command line, reset the number of contexts being used if first argument is M<n>*/
	p->portable_id=1;
}
/* Function: portable_fini
	Target specific final code 
*/
void portable_fini(core_portable *p)
{
	p->portable_id=0;
}

#if (MULTITHREAD>1)

/* Function: core_start_parallel
	Start benchmarking in a parallel context.
	
	Three implementations are provided, one using pthreads, one using fork and shared mem, and one using fork and sockets.
	Other implementations using MCAPI or other standards can easily be devised.
*/
/* Function: core_stop_parallel
	Stop a parallel context execution of coremark, and gather the results.
	
	Three implementations are provided, one using pthreads, one using fork and shared mem, and one using fork and sockets.
	Other implementations using MCAPI or other standards can easily be devised.
*/
#if USE_PTHREAD
ee_u8 core_start_parallel(core_results *res) {
	return (ee_u8)pthread_create(&(res->port.thread),NULL,iterate,(void *)res);
}
ee_u8 core_stop_parallel(core_results *res) {
	void *retval;
	return (ee_u8)pthread_join(res->port.thread,&retval);
}
#elif USE_FORK
static int key_id=0;
ee_u8 core_start_parallel(core_results *res) {
	key_t key=4321+key_id;
	key_id++;
	res->port.pid=fork();
	res->port.shmid=shmget(key, 8, IPC_CREAT | 0666);
	if (res->port.shmid<0) {
		ee_printf("ERROR in shmget!\n");
	}
	if (res->port.pid==0) {
		iterate(res);
		res->port.shm=shmat(res->port.shmid, NULL, 0);
		/* copy the validation values to the shared memory area  and quit*/
		if (res->port.shm == (char *) -1) {
			ee_printf("ERROR in child shmat!\n");
		} else {
			memcpy(res->port.shm,&(res->crc),8);
			shmdt(res->port.shm);
		}
		exit(0);
	}
	return 1;
}
ee_u8 core_stop_parallel(core_results *res) {
	int status;
	pid_t wpid = waitpid(res->port.pid,&status,WUNTRACED);
	if (wpid != res->port.pid) {
		ee_printf("ERROR waiting for child.\n");
		if (errno == ECHILD) ee_printf("errno=No such child %d\n",res->port.pid);
		if (errno == EINTR) ee_printf("errno=Interrupted\n");
		return 0;
	}
	/* after process is done, get the values from the shared memory area */
	res->port.shm=shmat(res->port.shmid, NULL, 0);
	if (res->port.shm == (char *) -1) {
		ee_printf("ERROR in parent shmat!\n");
		return 0;
	} 
	memcpy(&(res->crc),res->port.shm,8);
	shmdt(res->port.shm);
	return 1;
}
#elif USE_SOCKET
static int key_id=0;
ee_u8 core_start_parallel(core_results *res) {
	int bound, buffer_length=8;
	res->port.sa.sin_family = AF_INET;
	res->port.sa.sin_addr.s_addr = htonl(0x7F000001);
	res->port.sa.sin_port = htons(7654+key_id);
	key_id++;
	res->port.pid=fork();
	if (res->port.pid==0) { /* benchmark child */
		iterate(res);
		res->port.sock = socket(PF_INET, SOCK_DGRAM, IPPROTO_UDP);
		if (-1 == res->port.sock) /* if socket failed to initialize, exit */   {
			ee_printf("Error Creating Socket");
		} else {
			int bytes_sent = sendto(res->port.sock, &(res->crc), buffer_length, 0,(struct sockaddr*)&(res->port.sa), sizeof (struct sockaddr_in));
			if (bytes_sent < 0)
				ee_printf("Error sending packet: %s\n", strerror(errno));
			close(res->port.sock); /* close the socket */
		}
		exit(0);
	} 
	/* parent process, open the socket */
	res->port.sock = socket(PF_INET, SOCK_DGRAM, IPPROTO_UDP);
	bound = bind(res->port.sock,(struct sockaddr*)&(res->port.sa), sizeof(struct sockaddr));
	if (bound < 0)
		ee_printf("bind(): %s\n",strerror(errno));
	return 1;
}
ee_u8 core_stop_parallel(core_results *res) {
	int status;
	int fromlen=sizeof(struct sockaddr);
	int recsize = recvfrom(res->port.sock, &(res->crc), 8, 0, (struct sockaddr*)&(res->port.sa), &fromlen);
	if (recsize < 0) {
		ee_printf("Error in receive: %s\n", strerror(errno));
		return 0;
	}
	pid_t wpid = waitpid(res->port.pid,&status,WUNTRACED);
	if (wpid != res->port.pid) {
		ee_printf("ERROR waiting for child.\n");
		if (errno == ECHILD) ee_printf("errno=No such child %d\n",res->port.pid);
		if (errno == EINTR) ee_printf("errno=Interrupted\n");
		return 0;
	}
	return 1;
}
#else /* no standard multicore implementation */
#error "Please implement multicore functionality in core_portme.c to use multiple contexts."
#endif /* multithread implementations */
#endif


ee_u32 GetTimerValue(ee_u32 TimerIntrId,ee_u16 Mode)

{

    int                 Status;
    XScuTimer_Config    *ConfigPtr;
    volatile ee_u32     CntValue  = 0;
    XScuTimer           *TimerInstancePtr = &Timer;

    if (Mode == 0) {
      // Initialize the Private Timer so that it is ready to use
      ConfigPtr = XScuTimer_LookupConfig(TimerIntrId);
      Status = XScuTimer_CfgInitialize (TimerInstancePtr, ConfigPtr,
    		  ConfigPtr->BaseAddr);
      if (Status != XST_SUCCESS) {
    	  print ("XST_FAILURE\n");
          return XST_FAILURE;
      }

      // Load the timer prescaler register.
      XScuTimer_SetPrescaler (TimerInstancePtr, TIMER_RES_DIVIDER);
      // Load the timer counter register.
      XScuTimer_LoadTimer (TimerInstancePtr, TIMER_LOAD_VALUE);
      // Start the timer counter and read start value
      XScuTimer_Start(TimerInstancePtr);
      CntValue = XScuTimer_GetCounterValue (TimerInstancePtr);
    } else {
       //  Read stop value and stop the timer counter
       CntValue = XScuTimer_GetCounterValue (TimerInstancePtr);
       XScuTimer_Stop(TimerInstancePtr);
    }

    return CntValue;
}

core_portme.h

/* File: core_portme.h */

/*
	Author : Shay Gal-On, EEMBC
	Legal : TODO!
*/ 
/* Topic: Description
	This file contains configuration constants required to execute on different platforms
*/

#include <stdio.h>

#ifndef CORE_PORTME_H
#define CORE_PORTME_H
/************************/
/* Data types and settings */
/************************/
/* Configuration: HAS_FLOAT 
	Define to 1 if the platform supports floating point.
*/
#ifndef HAS_FLOAT 
#define HAS_FLOAT 1
#endif
/* Configuration: HAS_TIME_H
	Define to 1 if platform has the time.h header file,
	and implementation of functions thereof.
*/
#ifndef HAS_TIME_H
#define HAS_TIME_H 0
#endif
/* Configuration: USE_CLOCK
	Define to 1 if platform has the time.h header file,
	and implementation of functions thereof.
*/
#ifndef USE_CLOCK
#define USE_CLOCK 0
#endif
/* Configuration: HAS_STDIO
	Define to 1 if the platform has stdio.h.
*/
#ifndef HAS_STDIO
#define HAS_STDIO 1
#endif
/* Configuration: HAS_PRINTF
	Define to 1 if the platform has stdio.h and implements the printf function.
*/
#ifndef HAS_PRINTF
#define HAS_PRINTF 1
#endif

/* Configuration: CORE_TICKS
	Define type of return from the timing functions.
 */
#if defined(_MSC_VER)
#include <windows.h>
typedef size_t CORE_TICKS;
#elif HAS_TIME_H
#include <time.h>
typedef clock_t CORE_TICKS;
#else
// #error "Please define type of CORE_TICKS and implement start_time, end_time get_time and time_in_secs functions!"
#endif

/* Definitions: COMPILER_VERSION, COMPILER_FLAGS, MEM_LOCATION
	Initialize these strings per platform
*/
#ifndef COMPILER_VERSION 
 #ifdef __GNUC__
 #define COMPILER_VERSION "GCC"__VERSION__
 #else
 #define COMPILER_VERSION "Please put compiler version here (e.g. gcc 4.1)"
 #endif
#endif
#ifndef COMPILER_FLAGS 
 #define COMPILER_FLAGS FLAGS_STR /* "Please put compiler flags here (e.g. -o3)" */
#endif
#ifndef MEM_LOCATION 
 #define MEM_LOCATION "Please put data memory location here\n\t\t\t(e.g. code in flash, data on heap etc)"
 #define MEM_LOCATION_UNSPEC 1
#endif

/* Data Types:
	To avoid compiler issues, define the data types that need ot be used for 8b, 16b and 32b in <core_portme.h>.
	
	*Imprtant*:
	ee_ptr_int needs to be the data type used to hold pointers, otherwise coremark may fail!!!
*/
typedef signed short ee_s16;
typedef unsigned short ee_u16;
typedef signed int ee_s32;
typedef double ee_f32;
typedef unsigned char ee_u8;
typedef unsigned int ee_u32;
typedef ee_u32 ee_ptr_int;
typedef size_t ee_size_t;
typedef ee_u32 CORE_TICKS;

/* align_mem:
	This macro is used to align an offset to point to a 32b value. It is used in the Matrix algorithm to initialize the input memory blocks.
*/
#define align_mem(x) (void *)(4 + (((ee_ptr_int)(x) - 1) & ~3))

/* Configuration: SEED_METHOD
	Defines method to get seed values that cannot be computed at compile time.
	
	Valid values:
	SEED_ARG - from command line.
	SEED_FUNC - from a system function.
	SEED_VOLATILE - from volatile variables.
*/
#define SEED_METHOD SEED_VOLATILE

// #ifndef SEED_METHOD
// #define SEED_METHOD SEED_ARG
// #endif

/* Configuration: MEM_METHOD
	Defines method to get a block of memry.
	
	Valid values:
	MEM_MALLOC - for platforms that implement malloc and have malloc.h.
	MEM_STATIC - to use a static memory array.
	MEM_STACK - to allocate the data block on the stack (NYI).
*/
#define MEM_METHOD MEM_STACK
// #ifndef MEM_METHOD
// #define MEM_METHOD MEM_MALLOC
// #endif

/* Configuration: MULTITHREAD
	Define for parallel execution 
	
	Valid values:
	1 - only one context (default).
	N>1 - will execute N copies in parallel.
	
	Note: 
	If this flag is defined to more then 1, an implementation for launching parallel contexts must be defined.
	
	Two sample implementations are provided. Use <USE_PTHREAD> or <USE_FORK> to enable them.
	
	It is valid to have a different implementation of <core_start_parallel> and <core_end_parallel> in <core_portme.c>,
	to fit a particular architecture. 
*/
#ifndef MULTITHREAD
#define MULTITHREAD 1
#endif

/* Configuration: USE_PTHREAD
	Sample implementation for launching parallel contexts 
	This implementation uses pthread_thread_create and pthread_join.
	
	Valid values:
	0 - Do not use pthreads API.
	1 - Use pthreads API
	
	Note: 
	This flag only matters if MULTITHREAD has been defined to a value greater then 1.
*/
#ifndef USE_PTHREAD
#define USE_PTHREAD 0
#endif

/* Configuration: USE_FORK
	Sample implementation for launching parallel contexts 
	This implementation uses fork, waitpid, shmget,shmat and shmdt.
	
	Valid values:
	0 - Do not use fork API.
	1 - Use fork API
	
	Note: 
	This flag only matters if MULTITHREAD has been defined to a value greater then 1.
*/
#ifndef USE_FORK
#define USE_FORK 0
#endif

/* Configuration: USE_SOCKET
	Sample implementation for launching parallel contexts 
	This implementation uses fork, socket, sendto and recvfrom
	
	Valid values:
	0 - Do not use fork and sockets API.
	1 - Use fork and sockets API
	
	Note: 
	This flag only matters if MULTITHREAD has been defined to a value greater then 1.
*/
#ifndef USE_SOCKET
#define USE_SOCKET 0
#endif

/* Configuration: MAIN_HAS_NOARGC
	Needed if platform does not support getting arguments to main. 
	
	Valid values:
	0 - argc/argv to main is supported
	1 - argc/argv to main is not supported
*/
// #ifndef MAIN_HAS_NOARGC
// #define MAIN_HAS_NOARGC 0
// #endif

/* Configuration: MAIN_HAS_NORETURN
	Needed if platform does not support returning a value from main. 
	
	Valid values:
	0 - main returns an int, and return value will be 0.
	1 - platform does not support returning a value from main
*/
#ifndef MAIN_HAS_NORETURN
#define MAIN_HAS_NORETURN 0
#endif

/* Variable: default_num_contexts
	Number of contexts to spawn in multicore context.
	Override this global value to change number of contexts used.
	
	Note:
	This value may not be set higher then the <MULTITHREAD> define.
	
	To experiment, you can set the <MULTITHREAD> define to the highest value expected, and use argc/argv in the <portable_init> to set this value from the command line.
*/
extern ee_u32 default_num_contexts;

#if (MULTITHREAD>1)
#if USE_PTHREAD
	#include <pthread.h>
	#define PARALLEL_METHOD "PThreads"
#elif USE_FORK
	#include <unistd.h>
	#include <errno.h>
	#include <sys/wait.h>
	#include <sys/shm.h>
	#include <string.h> /* for memcpy */
	#define PARALLEL_METHOD "Fork"
#elif USE_SOCKET
	#include <sys/types.h>
	#include <sys/socket.h>
	#include <netinet/in.h>
	#include <arpa/inet.h>
	#include <sys/wait.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <string.h>
	#include <unistd.h>
	#include <errno.h>
	#define PARALLEL_METHOD "Sockets"
#else
	#define PARALLEL_METHOD "Proprietary"
	#error "Please implement multicore functionality in core_portme.c to use multiple contexts."
#endif /* Method for multithreading */
#endif /* MULTITHREAD > 1 */

typedef struct CORE_PORTABLE_S {
#if (MULTITHREAD>1)
	#if USE_PTHREAD
	pthread_t thread;
	#elif USE_FORK
	pid_t pid;
	int shmid;
	void *shm;
	#elif USE_SOCKET
	pid_t pid;
	int sock;
	struct sockaddr_in sa;
	#endif /* Method for multithreading */
#endif /* MULTITHREAD>1 */
	ee_u8	portable_id;
} core_portable;

/* target specific init/fini */
void portable_init(core_portable *p, int *argc, char *argv[]);
void portable_fini(core_portable *p);

#if (SEED_METHOD==SEED_VOLATILE)
 #if (VALIDATION_RUN || PERFORMANCE_RUN || PROFILE_RUN)
  #define RUN_TYPE_FLAG 1
 #else
  #if (TOTAL_DATA_SIZE==1200)
   #define PROFILE_RUN 1
  #else
   #define PERFORMANCE_RUN 1
  #endif
 #endif
#endif /* SEED_METHOD==SEED_VOLATILE */

#endif /* CORE_PORTME_H */