/*
* Copyright (c) 2014, 2017 embedded brains GmbH. All rights reserved.
*
* embedded brains GmbH
* Dornierstr. 4
* 82178 Puchheim
* Germany
* <rtems@embedded-brains.de>
*
* The license and distribution terms for this file may be
* found in the file LICENSE in this distribution or at
* http://www.rtems.org/license/LICENSE.
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include <rtems/counter.h>
#include <rtems.h>
#include <stdio.h>
#include <stdlib.h>
#include <inttypes.h>
#include <alloca.h>
#include "tmacros.h"
#define FUNCTION_LEVELS 16
#define SAMPLES 123
#define CPU_COUNT 32
const char rtems_test_name[] = "TMCONTEXT 1";
static rtems_counter_ticks t[SAMPLES];
static volatile int prevent_optimization;
static size_t cache_line_size;
static size_t data_size;
static volatile int *main_data;
static Context_Control ctx;
static int dirty_data_cache(volatile int *data, size_t n, size_t clsz, int j)
{
size_t m = n / sizeof(*data);
size_t k = clsz / sizeof(*data);
size_t i;
for (i = 0; i < m; i += k) {
data[i] = i + j;
}
return i + j;
}
static int call_at_level(
int start,
int fl,
int s,
bool dirty
)
{
int prevent_optimization;;
prevent_optimization = start + fl;
if (fl == start) {
/*
* Some architectures like the SPARC have register windows. A side-effect
* of this context switch is that we start with a fresh window set. On
* architectures like ARM or PowerPC this context switch has no effect.
*/
_Context_Switch(&ctx, &ctx);
}
if (fl > 0) {
call_at_level(
start,
fl - 1,
s,
dirty
);
} else {
char *volatile space;
rtems_counter_ticks a;
rtems_counter_ticks b;
if (dirty) {
dirty_data_cache(main_data, data_size, cache_line_size, fl);
rtems_cache_invalidate_entire_instruction();
}
a = rtems_counter_read();
/* Ensure that we use an untouched stack area */
space = alloca(1024);
(void) space;
_Context_Switch(&ctx, &ctx);
b = rtems_counter_read();
t[s] = rtems_counter_difference(b, a);
}
return prevent_optimization;
}
static void load_task(rtems_task_argument arg)
{
volatile int *load_data = (volatile int *) arg;
size_t n = data_size;
size_t clsz = cache_line_size;
int j = (int) rtems_get_current_processor();
while (true) {
j = dirty_data_cache(load_data, n, clsz, j);
}
}
static int cmp(const void *ap, const void *bp)
{
const rtems_counter_ticks *a = ap;
const rtems_counter_ticks *b = bp;
return *a - *b;
}
static void sort_t(void)
{
qsort(&t[0], SAMPLES, sizeof(t[0]), cmp);
}
static void test_by_function_level(int fl, bool dirty)
{
RTEMS_INTERRUPT_LOCK_DECLARE(, lock)
rtems_interrupt_lock_context lock_context;
int s;
uint64_t min;
uint64_t q1;
uint64_t q2;
uint64_t q3;
uint64_t max;
fl += prevent_optimization;
rtems_interrupt_lock_initialize(&lock, "test");
rtems_interrupt_lock_acquire(&lock, &lock_context);
for (s = 0; s < SAMPLES; ++s) {
call_at_level(fl, fl, s, dirty);
}
rtems_interrupt_lock_release(&lock, &lock_context);
rtems_interrupt_lock_destroy(&lock);
sort_t();
min = t[0];
q1 = t[(1 * SAMPLES) / 4];
q2 = t[SAMPLES / 2];
q3 = t[(3 * SAMPLES) / 4];
max = t[SAMPLES - 1];
printf(
" <Sample functionNestLevel=\"%i\">\n"
" <Min unit=\"ns\">%" PRIu64 "</Min>"
"<Q1 unit=\"ns\">%" PRIu64 "</Q1>"
"<Q2 unit=\"ns\">%" PRIu64 "</Q2>"
"<Q3 unit=\"ns\">%" PRIu64 "</Q3>"
"<Max unit=\"ns\">%" PRIu64 "</Max>\n"
" </Sample>\n",
fl,
rtems_counter_ticks_to_nanoseconds(min),
rtems_counter_ticks_to_nanoseconds(q1),
rtems_counter_ticks_to_nanoseconds(q2),
rtems_counter_ticks_to_nanoseconds(q3),
rtems_counter_ticks_to_nanoseconds(max)
);
}
static void test(bool dirty, uint32_t load)
{
int fl;
printf(
" <ContextSwitchTest environment=\"%s\"",
dirty ? "dirty" : "normal"
);
if (load > 0) {
printf(" load=\"%" PRIu32 "\"", load);
}
printf(">\n");
for (fl = 0; fl < FUNCTION_LEVELS; ++fl) {
test_by_function_level(fl, dirty);
}
printf(" </ContextSwitchTest>\n");
}
static void Init(rtems_task_argument arg)
{
uint32_t load = 0;
TEST_BEGIN();
printf("<Test>\n");
cache_line_size = rtems_cache_get_data_line_size();
if (cache_line_size == 0) {
cache_line_size = 32;
}
data_size = rtems_cache_get_data_cache_size(0);
if (data_size == 0) {
data_size = cache_line_size;
}
main_data = malloc(data_size);
rtems_test_assert(main_data != NULL);
test(false, load);
test(true, load);
for (load = 1; load < rtems_get_processor_count(); ++load) {
rtems_status_code sc;
rtems_id id;
volatile int *load_data = NULL;
load_data = malloc(data_size);
if (load_data == NULL) {
load_data = main_data;
}
sc = rtems_task_create(
rtems_build_name('L', 'O', 'A', 'D'),
1,
RTEMS_MINIMUM_STACK_SIZE,
RTEMS_DEFAULT_MODES,
RTEMS_DEFAULT_ATTRIBUTES,
&id
);
rtems_test_assert(sc == RTEMS_SUCCESSFUL);
sc = rtems_task_start(id, load_task, (rtems_task_argument) load_data);
rtems_test_assert(sc == RTEMS_SUCCESSFUL);
test(true, load);
}
printf("</Test>\n");
TEST_END();
rtems_test_exit(0);
}
/*
* Do not use a clock driver, since this will disturb the test in the "normal"
* environment.
*/
#define CONFIGURE_APPLICATION_DOES_NOT_NEED_CLOCK_DRIVER
#define CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER
#define CONFIGURE_MAXIMUM_TASKS (1 + CPU_COUNT)
#define CONFIGURE_INIT_TASK_STACK_SIZE (32 * 1024)
#define CONFIGURE_MAXIMUM_PROCESSORS CPU_COUNT
#define CONFIGURE_RTEMS_INIT_TASKS_TABLE
#define CONFIGURE_INIT
#include <rtems/confdefs.h>