srsRAN/lib/src/phy/dft/dft_fftw.c

400 lines
11 KiB
C

/*
* Copyright 2013-2019 Software Radio Systems Limited
*
* This file is part of srsLTE.
*
* srsLTE is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero General Public License as
* published by the Free Software Foundation, either version 3 of
* the License, or (at your option) any later version.
*
* srsLTE is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Affero General Public License for more details.
*
* A copy of the GNU Affero General Public License can be found in
* the LICENSE file in the top-level directory of this distribution
* and at http://www.gnu.org/licenses/.
*
*/
#include "srslte/srslte.h"
#include <complex.h>
#include <fftw3.h>
#include <math.h>
#include <pwd.h>
#include <string.h>
#include <unistd.h>
#include "srslte/phy/dft/dft.h"
#include "srslte/phy/utils/vector.h"
#define dft_ceil(a, b) ((a - 1) / b + 1)
#define dft_floor(a, b) (a / b)
#define FFTW_WISDOM_FILE "%s/.srslte_fftwisdom"
static int get_fftw_wisdom_file(char* full_path, uint32_t n)
{
const char* homedir = NULL;
if ((homedir = getenv("HOME")) == NULL) {
homedir = getpwuid(getuid())->pw_dir;
}
return snprintf(full_path, n, FFTW_WISDOM_FILE, homedir);
}
#ifdef FFTW_WISDOM_FILE
#define FFTW_TYPE FFTW_MEASURE
#else
#define FFTW_TYPE 0
#endif
static pthread_mutex_t fft_mutex = PTHREAD_MUTEX_INITIALIZER;
// This function is called in the beggining of any executable where it is linked
__attribute__((constructor)) static void srslte_dft_load()
{
#ifdef FFTW_WISDOM_FILE
char full_path[256];
get_fftw_wisdom_file(full_path, sizeof(full_path));
fftwf_import_wisdom_from_filename(full_path);
#else
printf("Warning: FFTW Wisdom file not defined\n");
#endif
}
// This function is called in the ending of any executable where it is linked
__attribute__((destructor)) static void srslte_dft_exit()
{
#ifdef FFTW_WISDOM_FILE
char full_path[256];
get_fftw_wisdom_file(full_path, sizeof(full_path));
fftwf_export_wisdom_to_filename(full_path);
#endif
fftwf_cleanup();
}
int srslte_dft_plan(srslte_dft_plan_t* plan, const int dft_points, srslte_dft_dir_t dir, srslte_dft_mode_t mode)
{
bzero(plan, sizeof(srslte_dft_plan_t));
if (mode == SRSLTE_DFT_COMPLEX) {
return srslte_dft_plan_c(plan, dft_points, dir);
} else {
return srslte_dft_plan_r(plan, dft_points, dir);
}
return 0;
}
int srslte_dft_replan(srslte_dft_plan_t* plan, const int new_dft_points)
{
if (new_dft_points <= plan->init_size) {
if (plan->mode == SRSLTE_DFT_COMPLEX) {
return srslte_dft_replan_c(plan, new_dft_points);
} else {
return srslte_dft_replan_r(plan, new_dft_points);
}
} else {
ERROR("DFT: Error calling replan: new_dft_points (%d) must be lower or equal "
"dft_size passed initially (%d)\n",
new_dft_points,
plan->init_size);
return -1;
}
}
static void allocate(srslte_dft_plan_t* plan, int size_in, int size_out, int len)
{
plan->in = fftwf_malloc(size_in * len);
plan->out = fftwf_malloc(size_out * len);
}
int srslte_dft_replan_guru_c(srslte_dft_plan_t* plan,
const int new_dft_points,
cf_t* in_buffer,
cf_t* out_buffer,
int istride,
int ostride,
int how_many,
int idist,
int odist)
{
int sign = (plan->forward) ? FFTW_FORWARD : FFTW_BACKWARD;
const fftwf_iodim iodim = {new_dft_points, istride, ostride};
const fftwf_iodim howmany_dims = {how_many, idist, odist};
pthread_mutex_lock(&fft_mutex);
/* Destroy current plan */
fftwf_destroy_plan(plan->p);
plan->p = fftwf_plan_guru_dft(1, &iodim, 1, &howmany_dims, in_buffer, out_buffer, sign, FFTW_TYPE);
pthread_mutex_unlock(&fft_mutex);
if (!plan->p) {
return -1;
}
plan->size = new_dft_points;
plan->init_size = plan->size;
return 0;
}
int srslte_dft_replan_c(srslte_dft_plan_t* plan, const int new_dft_points)
{
int sign = (plan->dir == SRSLTE_DFT_FORWARD) ? FFTW_FORWARD : FFTW_BACKWARD;
pthread_mutex_lock(&fft_mutex);
if (plan->p) {
fftwf_destroy_plan(plan->p);
plan->p = NULL;
}
plan->p = fftwf_plan_dft_1d(new_dft_points, plan->in, plan->out, sign, FFTW_TYPE);
pthread_mutex_unlock(&fft_mutex);
if (!plan->p) {
return -1;
}
plan->size = new_dft_points;
return 0;
}
int srslte_dft_plan_guru_c(srslte_dft_plan_t* plan,
const int dft_points,
srslte_dft_dir_t dir,
cf_t* in_buffer,
cf_t* out_buffer,
int istride,
int ostride,
int how_many,
int idist,
int odist)
{
int sign = (dir == SRSLTE_DFT_FORWARD) ? FFTW_FORWARD : FFTW_BACKWARD;
const fftwf_iodim iodim = {dft_points, istride, ostride};
const fftwf_iodim howmany_dims = {how_many, idist, odist};
pthread_mutex_lock(&fft_mutex);
plan->p = fftwf_plan_guru_dft(1, &iodim, 1, &howmany_dims, in_buffer, out_buffer, sign, FFTW_TYPE);
if (!plan->p) {
return -1;
}
pthread_mutex_unlock(&fft_mutex);
plan->size = dft_points;
plan->init_size = plan->size;
plan->mode = SRSLTE_DFT_COMPLEX;
plan->dir = dir;
plan->forward = (dir == SRSLTE_DFT_FORWARD) ? true : false;
plan->mirror = false;
plan->db = false;
plan->norm = false;
plan->dc = false;
plan->is_guru = true;
return 0;
}
int srslte_dft_plan_c(srslte_dft_plan_t* plan, const int dft_points, srslte_dft_dir_t dir)
{
allocate(plan, sizeof(fftwf_complex), sizeof(fftwf_complex), dft_points);
pthread_mutex_lock(&fft_mutex);
int sign = (dir == SRSLTE_DFT_FORWARD) ? FFTW_FORWARD : FFTW_BACKWARD;
plan->p = fftwf_plan_dft_1d(dft_points, plan->in, plan->out, sign, FFTW_TYPE);
pthread_mutex_unlock(&fft_mutex);
if (!plan->p) {
return -1;
}
plan->size = dft_points;
plan->init_size = plan->size;
plan->mode = SRSLTE_DFT_COMPLEX;
plan->dir = dir;
plan->forward = (dir == SRSLTE_DFT_FORWARD) ? true : false;
plan->mirror = false;
plan->db = false;
plan->norm = false;
plan->dc = false;
plan->is_guru = false;
return 0;
}
int srslte_dft_replan_r(srslte_dft_plan_t* plan, const int new_dft_points)
{
int sign = (plan->dir == SRSLTE_DFT_FORWARD) ? FFTW_R2HC : FFTW_HC2R;
pthread_mutex_lock(&fft_mutex);
if (plan->p) {
fftwf_destroy_plan(plan->p);
plan->p = NULL;
}
plan->p = fftwf_plan_r2r_1d(new_dft_points, plan->in, plan->out, sign, FFTW_TYPE);
pthread_mutex_unlock(&fft_mutex);
if (!plan->p) {
return -1;
}
plan->size = new_dft_points;
return 0;
}
int srslte_dft_plan_r(srslte_dft_plan_t* plan, const int dft_points, srslte_dft_dir_t dir)
{
allocate(plan, sizeof(float), sizeof(float), dft_points);
int sign = (dir == SRSLTE_DFT_FORWARD) ? FFTW_R2HC : FFTW_HC2R;
pthread_mutex_lock(&fft_mutex);
plan->p = fftwf_plan_r2r_1d(dft_points, plan->in, plan->out, sign, FFTW_TYPE);
pthread_mutex_unlock(&fft_mutex);
if (!plan->p) {
return -1;
}
plan->size = dft_points;
plan->init_size = plan->size;
plan->mode = SRSLTE_REAL;
plan->dir = dir;
plan->forward = (dir == SRSLTE_DFT_FORWARD) ? true : false;
plan->mirror = false;
plan->db = false;
plan->norm = false;
plan->dc = false;
return 0;
}
void srslte_dft_plan_set_mirror(srslte_dft_plan_t* plan, bool val)
{
plan->mirror = val;
}
void srslte_dft_plan_set_db(srslte_dft_plan_t* plan, bool val)
{
plan->db = val;
}
void srslte_dft_plan_set_norm(srslte_dft_plan_t* plan, bool val)
{
plan->norm = val;
}
void srslte_dft_plan_set_dc(srslte_dft_plan_t* plan, bool val)
{
plan->dc = val;
}
static void copy_pre(uint8_t* dst, uint8_t* src, int size_d, int len, bool forward, bool mirror, bool dc)
{
int offset = dc ? 1 : 0;
if (mirror && !forward) {
int hlen = dft_floor(len, 2);
memset(dst, 0, size_d * offset);
memcpy(&dst[size_d * offset], &src[size_d * hlen], size_d * (len - hlen - offset));
memcpy(&dst[(len - hlen) * size_d], src, size_d * hlen);
} else {
memcpy(dst, src, size_d * len);
}
}
static void copy_post(uint8_t* dst, uint8_t* src, int size_d, int len, bool forward, bool mirror, bool dc)
{
int offset = dc ? 1 : 0;
if (mirror && forward) {
int hlen = dft_ceil(len, 2);
memcpy(dst, &src[size_d * hlen], size_d * (len - hlen));
memcpy(&dst[(len - hlen) * size_d], &src[size_d * offset], size_d * (hlen - offset));
} else {
memcpy(dst, src, size_d * len);
}
}
void srslte_dft_run(srslte_dft_plan_t* plan, const void* in, void* out)
{
if (plan->mode == SRSLTE_DFT_COMPLEX) {
srslte_dft_run_c(plan, in, out);
} else {
srslte_dft_run_r(plan, in, out);
}
}
void srslte_dft_run_c_zerocopy(srslte_dft_plan_t* plan, const cf_t* in, cf_t* out)
{
fftwf_execute_dft(plan->p, (cf_t*)in, out);
}
void srslte_dft_run_c(srslte_dft_plan_t* plan, const cf_t* in, cf_t* out)
{
float norm;
int i;
fftwf_complex* f_out = plan->out;
copy_pre((uint8_t*)plan->in, (uint8_t*)in, sizeof(cf_t), plan->size, plan->forward, plan->mirror, plan->dc);
fftwf_execute(plan->p);
if (plan->norm) {
norm = 1.0 / sqrtf(plan->size);
srslte_vec_sc_prod_cfc(f_out, norm, f_out, plan->size);
}
if (plan->db) {
for (i = 0; i < plan->size; i++) {
f_out[i] = srslte_convert_power_to_dB(f_out[i]);
}
}
copy_post((uint8_t*)out, (uint8_t*)plan->out, sizeof(cf_t), plan->size, plan->forward, plan->mirror, plan->dc);
}
void srslte_dft_run_guru_c(srslte_dft_plan_t* plan)
{
if (plan->is_guru == true) {
fftwf_execute(plan->p);
} else {
ERROR("srslte_dft_run_guru_c: the selected plan is not guru!\n");
}
}
void srslte_dft_run_r(srslte_dft_plan_t* plan, const float* in, float* out)
{
float norm;
int i;
int len = plan->size;
float* f_out = plan->out;
memcpy(plan->in, in, sizeof(float) * plan->size);
fftwf_execute(plan->p);
if (plan->norm) {
norm = 1.0 / plan->size;
srslte_vec_sc_prod_fff(f_out, norm, f_out, plan->size);
}
if (plan->db) {
for (i = 0; i < len; i++) {
f_out[i] = srslte_convert_power_to_dB(f_out[i]);
}
}
memcpy(out, plan->out, sizeof(float) * plan->size);
}
void srslte_dft_plan_free(srslte_dft_plan_t* plan)
{
if (!plan)
return;
if (!plan->size)
return;
pthread_mutex_lock(&fft_mutex);
if (!plan->is_guru) {
if (plan->in)
fftwf_free(plan->in);
if (plan->out)
fftwf_free(plan->out);
}
if (plan->p)
fftwf_destroy_plan(plan->p);
pthread_mutex_unlock(&fft_mutex);
bzero(plan, sizeof(srslte_dft_plan_t));
}