Move the aec_rdft* files to a more proper place beneath APM and make them thread-safe.

webrtc/modules/audio_processing/aec/aec_core_neon.cc

 /*
  *  Copyright (c) 2014 The WebRTC project authors. All Rights Reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS.  All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
 /*
  * The core AEC algorithm, neon version of speed-critical functions.
  *
  * Based on aec_core_sse2.c.
  */
 
 #include <arm_neon.h>
 #include <math.h>
 #include <string.h>  // memset
 
 extern "C" {
 #include "webrtc/common_audio/signal_processing/include/signal_processing_library.h"
 }
 #include "webrtc/modules/audio_processing/aec/aec_common.h"
 #include "webrtc/modules/audio_processing/aec/aec_core_optimized_methods.h"
-#include "webrtc/modules/audio_processing/aec/aec_rdft.h"
+#include "webrtc/modules/audio_processing/utility/ooura_fft.h"
 
 namespace webrtc {
 
 enum { kShiftExponentIntoTopMantissa = 8 };
 enum { kFloatExponentShift = 23 };
 
 __inline static float MulRe(float aRe, float aIm, float bRe, float bIm) {
   return aRe * bRe - aIm * bIm;
 }
 
@@ skipping 140 matching lines @@
       ef[1][i] *= abs_ef;
     }
 
     // Stepsize factor
     ef[0][i] *= mu;
     ef[1][i] *= mu;
   }
 }
 
 static void FilterAdaptationNEON(
+    const OouraFft& ooura_fft,
     int num_partitions,
     int x_fft_buf_block_pos,
     float x_fft_buf[2][kExtendedNumPartitions * PART_LEN1],
     float e_fft[2][PART_LEN1],
     float h_fft_buf[2][kExtendedNumPartitions * PART_LEN1]) {
   float fft[PART_LEN2];
   int i;
   for (i = 0; i < num_partitions; i++) {
     int xPos = (i + x_fft_buf_block_pos) * PART_LEN1;
     int pos = i * PART_LEN1;
@@ skipping 21 matching lines @@
       const float32x4x2_t g_n_h = vzipq_f32(e, f);
       // Store
       vst1q_f32(&fft[2 * j + 0], g_n_h.val[0]);
       vst1q_f32(&fft[2 * j + 4], g_n_h.val[1]);
     }
     // ... and fixup the first imaginary entry.
     fft[1] =
         MulRe(x_fft_buf[0][xPos + PART_LEN], -x_fft_buf[1][xPos + PART_LEN],
               e_fft[0][PART_LEN], e_fft[1][PART_LEN]);
 
-    aec_rdft_inverse_128(fft);
+    ooura_fft.InverseFft(fft);
     memset(fft + PART_LEN, 0, sizeof(float) * PART_LEN);
 
     // fft scaling
     {
       const float scale = 2.0f / PART_LEN2;
       const float32x4_t scale_ps = vmovq_n_f32(scale);
       for (j = 0; j < PART_LEN; j += 4) {
         const float32x4_t fft_ps = vld1q_f32(&fft[j]);
         const float32x4_t fft_scale = vmulq_f32(fft_ps, scale_ps);
         vst1q_f32(&fft[j], fft_scale);
       }
     }
-    aec_rdft_forward_128(fft);
+    ooura_fft.Fft(fft);
 
     {
       const float wt1 = h_fft_buf[1][pos];
       h_fft_buf[0][pos + PART_LEN] += fft[1];
       for (j = 0; j < PART_LEN; j += 4) {
         float32x4_t wtBuf_re = vld1q_f32(&h_fft_buf[0][pos + j]);
         float32x4_t wtBuf_im = vld1q_f32(&h_fft_buf[1][pos + j]);
         const float32x4_t fft0 = vld1q_f32(&fft[2 * j + 0]);
         const float32x4_t fft4 = vld1q_f32(&fft[2 * j + 4]);
         const float32x4x2_t fft_re_im = vuzpq_f32(fft0, fft4);
@@ skipping 475 matching lines @@
   WebRtcAec_FilterAdaptation = FilterAdaptationNEON;
   WebRtcAec_Overdrive = OverdriveNEON;
   WebRtcAec_Suppress = SuppressNEON;
   WebRtcAec_ComputeCoherence = ComputeCoherenceNEON;
   WebRtcAec_UpdateCoherenceSpectra = UpdateCoherenceSpectraNEON;
   WebRtcAec_StoreAsComplex = StoreAsComplexNEON;
   WebRtcAec_PartitionDelay = PartitionDelayNEON;
   WebRtcAec_WindowData = WindowDataNEON;
 }
 }  // namespace webrtc