osmo-trx/Transceiver52M/arch/common/convolve_base.c

150 lines
3.6 KiB
C
Raw Normal View History

/*
* Convolution
* Copyright (C) 2012, 2013 Thomas Tsou <tom@tsou.cc>
*
* SPDX-License-Identifier: LGPL-2.1+
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library 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
* Lesser General Public License for more details.
*/
#include <malloc.h>
#include <string.h>
#include <stdio.h>
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
/* Base multiply and accumulate complex-real */
static void mac_real(const float *x, const float *h, float *y)
{
y[0] += x[0] * h[0];
y[1] += x[1] * h[0];
}
/* Base multiply and accumulate complex-complex */
static void mac_cmplx(const float *x, const float *h, float *y)
{
y[0] += x[0] * h[0] - x[1] * h[1];
y[1] += x[0] * h[1] + x[1] * h[0];
}
/* Base vector complex-complex multiply and accumulate */
static void mac_real_vec_n(const float *x, const float *h, float *y,
int len)
{
for (int i=0; i<len; i++)
mac_real(&x[2 * i], &h[2 * i], y);
}
/* Base vector complex-complex multiply and accumulate */
static void mac_cmplx_vec_n(const float *x, const float *h, float *y,
int len)
{
for (int i=0; i<len; i++)
mac_cmplx(&x[2 * i], &h[2 * i], y);
}
/* Base complex-real convolution */
int _base_convolve_real(const float *x, int x_len,
const float *h, int h_len,
float *y, int y_len,
int start, int len)
{
for (int i = 0; i < len; i++) {
mac_real_vec_n(&x[2 * (i - (h_len - 1) + start)],
h,
&y[2 * i], h_len);
}
return len;
}
/* Base complex-complex convolution */
int _base_convolve_complex(const float *x, int x_len,
const float *h, int h_len,
float *y, int y_len,
int start, int len)
{
for (int i = 0; i < len; i++) {
mac_cmplx_vec_n(&x[2 * (i - (h_len - 1) + start)],
h,
&y[2 * i],
h_len);
}
return len;
}
/* Buffer validity checks */
int bounds_check(int x_len, int h_len, int y_len,
int start, int len)
{
if ((x_len < 1) || (h_len < 1) ||
(y_len < 1) || (len < 1)) {
fprintf(stderr, "Convolve: Invalid input\n");
return -1;
}
if ((start + len > x_len) || (len > y_len) || (x_len < h_len)) {
fprintf(stderr, "Convolve: Boundary exception\n");
fprintf(stderr, "start: %i, len: %i, x: %i, h: %i, y: %i\n",
start, len, x_len, h_len, y_len);
return -1;
}
return 0;
}
/* API: Non-aligned (no SSE) complex-real */
int base_convolve_real(const float *x, int x_len,
const float *h, int h_len,
float *y, int y_len,
int start, int len)
{
if (bounds_check(x_len, h_len, y_len, start, len) < 0)
return -1;
memset(y, 0, len * 2 * sizeof(float));
return _base_convolve_real(x, x_len,
h, h_len,
y, y_len,
start, len);
}
/* API: Non-aligned (no SSE) complex-complex */
int base_convolve_complex(const float *x, int x_len,
const float *h, int h_len,
float *y, int y_len,
int start, int len)
{
if (bounds_check(x_len, h_len, y_len, start, len) < 0)
return -1;
memset(y, 0, len * 2 * sizeof(float));
return _base_convolve_complex(x, x_len,
h, h_len,
y, y_len,
start, len);
}
/* Aligned filter tap allocation */
SigProcLib: Improve Vector buffer allocation mess Original issue: In order to use SSE instructions, 16-byte aligned memory chunks are needed, and C++ version < C++11 doesn't provide for a native new/delete store. For that reason, memalign() must be used in the implementation of convolve_h_alloc() for some buffers. On the other side, The C++ code relies on C++ "new T[]" operator to allocate a chunk of memory containing an array of class instances. As classes are complex types, they cannot be allocated through C structures (calling malloc). Experimentally can be seen too that it's unreliable and the process will crash during startup if malloc() is used and then a Complex<> deferred from it. Previous implementation allowed for use of convolve_h_alloc or new[] based on how the (signal)Vector is called, because then the buffer is not going to be managed internally. But that's unreliable since resize() calling resize() on it could use "delete" operator on a malloc'ed buffer, and end up having a new new[] allocated buffer. It was also found that some of the callers were actually leaking memory through ASan (because the buffer is not managed by the Vector instance). IMHO best option would be to rewrite all this code using C structures and malloc/free exclusively, since it would make all this cod eeasier to maintain. But for now, let's extend the Vector class to allow specifying an external alloc/free function and let the Vector instance take care of the ownership of the buffer in all scenarios. Change-Id: Ie484a4762a7f77fe1b105188ea03a6f025730b82
2018-12-03 16:46:04 +00:00
void *convolve_h_alloc(size_t len)
{
#ifdef HAVE_SSE3
return memalign(16, len * 2 * sizeof(float));
#else
return malloc(len * 2 * sizeof(float));
#endif
}