Merge remote-tracking branch 'origin/fix-develop-segvs' into integration_2022_wk15

openair1/PHY/INIT/nr_init_ue.c

@@ -313,11 +313,26 @@ int init_nr_ue_signal(PHY_VARS_NR_UE *ue, int nb_connected_gNB)
   }
 
   // DLSCH
-  for (gNB_id = 0; gNB_id < ue->n_connected_gNB; gNB_id++) {
+  for (gNB_id = 0; gNB_id < ue->n_connected_gNB+1; gNB_id++) {
     for (th_id=0; th_id<RX_NB_TH_MAX; th_id++) {
       ue->pdsch_vars[th_id][gNB_id] = (NR_UE_PDSCH *)malloc16_clear(sizeof(NR_UE_PDSCH));
     }
+    for (th_id=0; th_id<RX_NB_TH_MAX; th_id++) {
+      phy_init_nr_ue_PDSCH( ue->pdsch_vars[th_id][gNB_id], fp );
+    }
+
+    int nb_codewords = NR_MAX_NB_LAYERS > 4 ? 2 : 1;
+    for (th_id=0; th_id<RX_NB_TH_MAX; th_id++) {
+      for (i=0; i<nb_codewords; i++) {
+        ue->pdsch_vars[th_id][gNB_id]->llr[i] = (int16_t *)malloc16_clear( (8*(3*8*8448))*sizeof(int16_t) );//Q_m = 8 bits/Sym, Code_Rate=3, Number of Segments =8, Circular Buffer K_cb = 8448
+      }
+      for (i=0; i<NR_MAX_NB_LAYERS; i++) {
+        ue->pdsch_vars[th_id][gNB_id]->layer_llr[i] = (int16_t *)malloc16_clear( (8*(3*8*8448))*sizeof(int16_t) );//Q_m = 8 bits/Sym, Code_Rate=3, Number of Segments =8, Circular Buffer K_cb = 8448
+      }
+    }
+  }
 
+  for (gNB_id = 0; gNB_id < ue->n_connected_gNB; gNB_id++) {
     for (th_id=0; th_id<RX_NB_TH_MAX; th_id++) {
       ue->pdcch_vars[th_id][gNB_id] = (NR_UE_PDCCH *)malloc16_clear(sizeof(NR_UE_PDCCH));
     }
@@ -344,21 +359,6 @@ int init_nr_ue_signal(PHY_VARS_NR_UE *ue, int nb_connected_gNB)
       ue->nr_srs_info->srs_estimated_channel_time_shifted[i] = (int32_t *) malloc16_clear(fp->ofdm_symbol_size*MAX_NUM_NR_SRS_SYMBOLS*sizeof(int32_t));
     }
 
-
-    for (th_id=0; th_id<RX_NB_TH_MAX; th_id++) {
-      phy_init_nr_ue_PDSCH( ue->pdsch_vars[th_id][gNB_id], fp );
-    }
-
-    int nb_codewords = NR_MAX_NB_LAYERS > 4 ? 2 : 1;
-    for (th_id=0; th_id<RX_NB_TH_MAX; th_id++) {
-      for (i=0; i<nb_codewords; i++) {
-        ue->pdsch_vars[th_id][gNB_id]->llr[i] = (int16_t *)malloc16_clear( (8*(3*8*8448))*sizeof(int16_t) );//Q_m = 8 bits/Sym, Code_Rate=3, Number of Segments =8, Circular Buffer K_cb = 8448
-      }
-      for (i=0; i<NR_MAX_NB_LAYERS; i++) {
-        ue->pdsch_vars[th_id][gNB_id]->layer_llr[i] = (int16_t *)malloc16_clear( (8*(3*8*8448))*sizeof(int16_t) );//Q_m = 8 bits/Sym, Code_Rate=3, Number of Segments =8, Circular Buffer K_cb = 8448
-      }
-    }
-
     // 100 PRBs * 12 REs/PRB * 4 PDCCH SYMBOLS * 2 LLRs/RE
     for (th_id=0; th_id<RX_NB_TH_MAX; th_id++) {
       ue->pdcch_vars[th_id][gNB_id]->llr                 = (int16_t *)malloc16_clear( 2*4*100*12*sizeof(uint16_t) );
@@ -471,16 +471,18 @@ void term_nr_ue_signal(PHY_VARS_NR_UE *ue, int nb_connected_gNB)
   }
   free_and_zero(ue->nr_gold_pdsch[0]);
 
-  for (int gNB_id = 0; gNB_id < ue->n_connected_gNB; gNB_id++) {
+  for (int gNB_id = 0; gNB_id < ue->n_connected_gNB+1; gNB_id++) {
 
     // PDSCH
     for (int th_id = 0; th_id < RX_NB_TH_MAX; th_id++) {
-
       free_and_zero(ue->pdsch_vars[th_id][gNB_id]->llr_shifts);
       free_and_zero(ue->pdsch_vars[th_id][gNB_id]->llr128_2ndstream);
       phy_term_nr_ue__PDSCH(ue->pdsch_vars[th_id][gNB_id], fp);
       free_and_zero(ue->pdsch_vars[th_id][gNB_id]);
     }
+  }
+  
+  for (int gNB_id = 0; gNB_id < ue->n_connected_gNB; gNB_id++) {
 
     for (int th_id = 0; th_id < RX_NB_TH_MAX; th_id++) {
       for (int i = 0; i < fp->nb_antennas_rx; i++) {

openair1/PHY/LTE_UE_TRANSPORT/transport_ue.h

@@ -155,8 +155,6 @@ typedef struct {
   uint8_t h[MAX_NUM_CHANNEL_BITS];
   /// Scrambled "b"-sequences (for definition see 36-211 V8.6 2009-03, p.14)
   uint8_t b_tilde[MAX_NUM_CHANNEL_BITS];
-  /// Modulated "d"-sequences (for definition see 36-211 V8.6 2009-03, p.14)
-  int32_t d[MAX_NUM_RE];
   /// Transform-coded "z"-sequences (for definition see 36-211 V8.6 2009-03, p.14-15)
   int32_t z[MAX_NUM_RE];
   /// "q" sequences for CQI/PMI (for definition see 36-212 V8.6 2009-03, p.27)

openair1/PHY/LTE_UE_TRANSPORT/ulsch_modulation.c

@@ -42,7 +42,7 @@
 
 //#define DEBUG_ULSCH_MODULATION
 
-void dft_lte(int32_t *z,int32_t *d, int32_t Msc_PUSCH, uint8_t Nsymb)
+void dft_lte(int32_t *z,struct complex16 *input, int32_t Msc_PUSCH, uint8_t Nsymb)
 {
 
 #if defined(__x86_64__) || defined(__i386__)
@@ -66,7 +66,7 @@ void dft_lte(int32_t *z,int32_t *d, int32_t Msc_PUSCH, uint8_t Nsymb)
 #endif
   //  printf("Doing lte_dft for Msc_PUSCH %d\n",Msc_PUSCH);
 
-  d0 = (uint32_t *)d;
+  d0 = (uint32_t *)input;
   d1 = d0+Msc_PUSCH;
   d2 = d1+Msc_PUSCH;
   d3 = d2+Msc_PUSCH;
@@ -476,7 +476,8 @@ void ulsch_modulation(int32_t **txdataF,
   // Modulation
 
   ulsch_Msymb = G/Q_m;
-
+  /// Modulated "d"-sequences (for definition see 36-211 V8.6 2009-03, p.14)
+  struct complex16 d[MAX_NUM_RE] __attribute__((aligned(32)));
   if(ulsch->cooperation_flag == 2)
     // For Distributed Alamouti Scheme in Collabrative Communication
   {
@@ -488,14 +489,14 @@ void ulsch_modulation(int32_t **txdataF,
 
 
         //UE1, -x1*
-        ((int16_t*)&ulsch->d[i])[0] = (ulsch->b_tilde[j] == 1)  ? (gain_lin_QPSK) : -gain_lin_QPSK;
-        ((int16_t*)&ulsch->d[i])[1] = (ulsch->b_tilde[j+1] == 1)? (-gain_lin_QPSK) : gain_lin_QPSK;
+        d[i].r = (ulsch->b_tilde[j] == 1)  ? (gain_lin_QPSK) : -gain_lin_QPSK;
+        d[i].i = (ulsch->b_tilde[j+1] == 1)? (-gain_lin_QPSK) : gain_lin_QPSK;
         //      if (i<Msc_PUSCH)
         //  printf("input %d (%p): %d,%d\n", i,&ulsch->d[i],((int16_t*)&ulsch->d[i])[0],((int16_t*)&ulsch->d[i])[1]);
 
         // UE1, x0*
-        ((int16_t*)&ulsch->d[i+1])[0] = (ulsch->b_tilde[j-2] == 1)  ? (-gain_lin_QPSK) : gain_lin_QPSK;
-        ((int16_t*)&ulsch->d[i+1])[1] = (ulsch->b_tilde[j-1] == 1)? (gain_lin_QPSK) : -gain_lin_QPSK;
+        d[i+1].r = (ulsch->b_tilde[j-2] == 1)  ? (-gain_lin_QPSK) : gain_lin_QPSK;
+        d[i+1].i = (ulsch->b_tilde[j-1] == 1)? (gain_lin_QPSK) : -gain_lin_QPSK;
 
         break;
 
@@ -521,8 +522,8 @@ void ulsch_modulation(int32_t **txdataF,
           qam16_table_offset_im+=1;
 
 
-        ((int16_t*)&ulsch->d[i])[0]=-(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_re])>>15);
-        ((int16_t*)&ulsch->d[i])[1]=(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_im])>>15);
+        d[i].r =-(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_re])>>15);
+        d[i].i =(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_im])>>15);
 
         //UE1,x0*
         qam16_table_offset_re = 0;
@@ -544,8 +545,8 @@ void ulsch_modulation(int32_t **txdataF,
 
         //    ((int16_t*)&ulsch->d[i+1])[0]=-(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_re])>>15);
         //    ((int16_t*)&ulsch->d[i+1])[1]=(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_im])>>15);
-        ((int16_t*)&ulsch->d[i+1])[0]=(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_re])>>15);
-        ((int16_t*)&ulsch->d[i+1])[1]=-(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_im])>>15);
+        d[i+1].r=(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_re])>>15);
+        d[i+1].i=-(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_im])>>15);
 
 
         break;
@@ -578,8 +579,8 @@ void ulsch_modulation(int32_t **txdataF,
           qam64_table_offset_im+=1;
 
 
-        ((int16_t*)&ulsch->d[i])[0]=-(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_re])>>15);
-        ((int16_t*)&ulsch->d[i])[1]=(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_im])>>15);
+        d[i].r=-(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_re])>>15);
+        d[i].i=(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_im])>>15);
 
         //UE1,x0*
         qam64_table_offset_re = 0;
@@ -605,8 +606,8 @@ void ulsch_modulation(int32_t **txdataF,
           qam64_table_offset_im+=1;
 
 
-        ((int16_t*)&ulsch->d[i+1])[0]=(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_re])>>15);
-        ((int16_t*)&ulsch->d[i+1])[1]=-(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_im])>>15);
+        d[i+1].r=(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_re])>>15);
+        d[i+1].i=-(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_im])>>15);
 
         break;
 
@@ -621,8 +622,8 @@ void ulsch_modulation(int32_t **txdataF,
       case 2:
         // TODO: this has to be updated!!!
 
-        ((int16_t*)&ulsch->d[i])[0] = (ulsch->b_tilde[j] == 1)  ? (-gain_lin_QPSK) : gain_lin_QPSK;
-        ((int16_t*)&ulsch->d[i])[1] = (ulsch->b_tilde[j+1] == 1)? (-gain_lin_QPSK) : gain_lin_QPSK;
+        d[i].r = (ulsch->b_tilde[j] == 1)  ? (-gain_lin_QPSK) : gain_lin_QPSK;
+        d[i].i = (ulsch->b_tilde[j+1] == 1)? (-gain_lin_QPSK) : gain_lin_QPSK;
         //        if (i<Msc_PUSCH)
         //    printf("input %d/%d Msc_PUSCH %d (%p): %d,%d\n", i,Msymb,Msc_PUSCH,&ulsch->d[i],((int16_t*)&ulsch->d[i])[0],((int16_t*)&ulsch->d[i])[1]);
 
@@ -646,8 +647,8 @@ void ulsch_modulation(int32_t **txdataF,
           qam16_table_offset_im+=1;
 
 
-        ((int16_t*)&ulsch->d[i])[0]=(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_re])>>15);
-        ((int16_t*)&ulsch->d[i])[1]=(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_im])>>15);
+        d[i].r=(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_re])>>15);
+        d[i].i=(int16_t)(((int32_t)amp*qam16_table[qam16_table_offset_im])>>15);
         //      printf("input(16qam) %d (%p): %d,%d\n", i,&ulsch->d[i],((int16_t*)&ulsch->d[i])[0],((int16_t*)&ulsch->d[i])[1]);
         break;
 
@@ -676,8 +677,8 @@ void ulsch_modulation(int32_t **txdataF,
           qam64_table_offset_im+=1;
 
 
-        ((int16_t*)&ulsch->d[i])[0]=(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_re])>>15);
-        ((int16_t*)&ulsch->d[i])[1]=(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_im])>>15);
+        d[i].r=(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_re])>>15);
+        d[i].i=(int16_t)(((int32_t)amp*qam64_table[qam64_table_offset_im])>>15);
 
         break;
 
@@ -688,7 +689,7 @@ void ulsch_modulation(int32_t **txdataF,
 
   // Transform Precoding
 
-  dft_lte(ulsch->z,ulsch->d,Msc_PUSCH,ulsch->Nsymb_pusch);
+  dft_lte(ulsch->z,d,Msc_PUSCH,ulsch->Nsymb_pusch);
 
   DevAssert(txdataF);
 

openair1/PHY/MODULATION/slot_fep.c

@@ -91,7 +91,7 @@ int slot_fep(PHY_VARS_UE *ue,
   }
 
   //  subframe_offset_F = frame_parms->ofdm_symbol_size * frame_parms->symbols_per_tti * (Ns>>1);
-
+ 
   if (l<0 || l>=7-frame_parms->Ncp) {
     printf("slot_fep: l must be between 0 and %d\n",7-frame_parms->Ncp);
     return(-1);

openair1/PHY/NR_TRANSPORT/nr_dlsch.c

@@ -103,7 +103,6 @@ void nr_generate_pdsch(processingData_L1tx_t *msgTx,
     uint16_t nb_re = ((12*rel15->NrOfSymbols)-nb_re_dmrs*dmrs_len-xOverhead)*rel15->rbSize*rel15->nrOfLayers;
     uint8_t Qm = rel15->qamModOrder[0];
     uint32_t encoded_length = nb_re*Qm;
-    uint32_t scrambled_output[rel15->NrOfCodewords][(encoded_length>>5)+1];
     int16_t mod_dmrs[n_dmrs<<1] __attribute__ ((aligned(16)));
 
     /* PTRS */
@@ -152,63 +151,61 @@ void nr_generate_pdsch(processingData_L1tx_t *msgTx,
     printf("\n");
 #endif
 
-    /// scrambling
-    start_meas(dlsch_scrambling_stats);
     for (int q=0; q<rel15->NrOfCodewords; q++) {
-      memset((void*)scrambled_output[q], 0, ((encoded_length>>5)+1)*sizeof(uint32_t));
+      /// scrambling
+      start_meas(dlsch_scrambling_stats);
+      uint32_t scrambled_output[(encoded_length>>5)+4]; // modulator acces by 4 bytes in some cases
+      memset(scrambled_output, 0, sizeof(scrambled_output));
+      if ( encoded_length > rel15->rbSize * NR_SYMBOLS_PER_SLOT * NR_NB_SC_PER_RB * Qm * rel15->nrOfLayers) abort();
       nr_pdsch_codeword_scrambling(output,
                                    encoded_length,
                                    q,
                                    rel15->dataScramblingId,
                                    rel15->rnti,
-                                   scrambled_output[q]);
-    }
-    
-    stop_meas(dlsch_scrambling_stats);
+                                   scrambled_output);
 #ifdef DEBUG_DLSCH
-    printf("PDSCH scrambling:\n");
-    for (int i=0; i<encoded_length>>8; i++) {
-      for (int j=0; j<8; j++)
-	printf("0x%08x\t", scrambled_output[0][(i<<3)+j]);
-      printf("\n");
-    }
+      printf("PDSCH scrambling:\n");
+      for (int i=0; i<encoded_length>>8; i++) {
+        for (int j=0; j<8; j++)
+          printf("0x%08x\t", scrambled_output[(i<<3)+j]);
+        printf("\n");
+      }
 #endif
-    
-    /// Modulation
-    start_meas(dlsch_modulation_stats);
-    VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_gNB_PDSCH_MODULATION, 1);
-    for (int q=0; q<rel15->NrOfCodewords; q++)
-      nr_modulation(scrambled_output[q],
-		    encoded_length,
-		    Qm,
-		    mod_symbs[q]);
-    VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_gNB_PDSCH_MODULATION, 0);
-    stop_meas(dlsch_modulation_stats);
+      stop_meas(dlsch_scrambling_stats);
+      /// Modulation
+      start_meas(dlsch_modulation_stats);
+      nr_modulation(scrambled_output,
+                    encoded_length,
+                    Qm,
+                    mod_symbs[q]);
+      VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_gNB_PDSCH_MODULATION, 0);
+      stop_meas(dlsch_modulation_stats);
 #ifdef DEBUG_DLSCH
-    printf("PDSCH Modulation: Qm %d(%d)\n", Qm, nb_re);
-    for (int i=0; i<nb_re>>3; i++) {
-      for (int j=0; j<8; j++) {
-	printf("%d %d\t", mod_symbs[0][((i<<3)+j)<<1], mod_symbs[0][(((i<<3)+j)<<1)+1]);
+      printf("PDSCH Modulation: Qm %d(%d)\n", Qm, nb_re);
+      for (int i=0; i<nb_re>>3; i++) {
+        for (int j=0; j<8; j++) {
+          printf("%d %d\t", mod_symbs[0][((i<<3)+j)<<1], mod_symbs[0][(((i<<3)+j)<<1)+1]);
+        }
+        printf("\n");
       }
-      printf("\n");
-    }
 #endif
+    }
     
     start_meas(&gNB->dlsch_layer_mapping_stats); 
     /// Layer mapping
     nr_layer_mapping(mod_symbs,
-		     rel15->nrOfLayers,
-		     nb_re,
-		     tx_layers);
+                     rel15->nrOfLayers,
+                     nb_re,
+                     tx_layers);
 #ifdef DEBUG_DLSCH
     printf("Layer mapping (%d layers):\n", rel15->nrOfLayers);
     for (int l=0; l<rel15->nrOfLayers; l++)
       for (int i=0; i<(nb_re/rel15->nrOfLayers)>>3; i++) {
-	printf("layer %d, Re %d..%d : ",l,i<<3,(i<<3)+7);
-	for (int j=0; j<8; j++) {
-	  printf("l%d %d\t", tx_layers[l][((i<<3)+j)<<1], tx_layers[l][(((i<<3)+j)<<1)+1]);
-	}
-	printf("\n");
+        printf("layer %d, Re %d..%d : ",l,i<<3,(i<<3)+7);
+        for (int j=0; j<8; j++) {
+          printf("l%d %d\t", tx_layers[l][((i<<3)+j)<<1], tx_layers[l][(((i<<3)+j)<<1)+1]);
+        }
+        printf("\n");
       }
 #endif
 
@@ -227,7 +224,7 @@ void nr_generate_pdsch(processingData_L1tx_t *msgTx,
 
 #ifdef DEBUG_DLSCH_MAPPING
     printf("PDSCH resource mapping started (start SC %d\tstart symbol %d\tN_PRB %d\tnb_re %d,nb_layers %d)\n",
-	   start_sc, rel15->StartSymbolIndex, rel15->rbSize, nb_re,rel15->nrOfLayers);
+           start_sc, rel15->StartSymbolIndex, rel15->rbSize, nb_re,rel15->nrOfLayers);
 #endif
 
     start_meas(&gNB->dlsch_resource_mapping_stats);

openair1/PHY/NR_UE_TRANSPORT/sss_nr.c

@@ -558,6 +558,9 @@ int rx_sss_nr(PHY_VARS_NR_UE *ue, UE_nr_rxtx_proc_t *proc, int32_t *tot_metric,
   }
   //#endif
 
+  if (Nid1==N_ID_1_NUMBER)
+    return -1;
+  
   int re = 0;
   int im = 0;
   d = (int16_t *)&d_sss[Nid2][Nid1];

openair1/PHY/TOOLS/oai_dfts.c

@@ -7082,43 +7082,40 @@ static inline void dft12f(simd_q15_t *x0,
 
   simd_q15_t tmp_dft12[12];
 
-  simd_q15_t *tmp_dft12_ptr = &tmp_dft12[0];
-
   // msg("dft12\n");
 
   bfly4_tw1(x0,
             x3,
             x6,
             x9,
-            tmp_dft12_ptr,
-            tmp_dft12_ptr+3,
-            tmp_dft12_ptr+6,
-            tmp_dft12_ptr+9);
-
+            tmp_dft12,
+            tmp_dft12+3,
+            tmp_dft12+6,
+            tmp_dft12+9);
 
   bfly4_tw1(x1,
             x4,
             x7,
             x10,
-            tmp_dft12_ptr+1,
-            tmp_dft12_ptr+4,
-            tmp_dft12_ptr+7,
-            tmp_dft12_ptr+10);
+            tmp_dft12+1,
+            tmp_dft12+4,
+            tmp_dft12+7,
+            tmp_dft12+10);
 
 
   bfly4_tw1(x2,
             x5,
             x8,
             x11,
-            tmp_dft12_ptr+2,
-            tmp_dft12_ptr+5,
-            tmp_dft12_ptr+8,
-            tmp_dft12_ptr+11);
+            tmp_dft12+2,
+            tmp_dft12+5,
+            tmp_dft12+8,
+            tmp_dft12+11);
 
   //  k2=0;
-  bfly3_tw1(tmp_dft12_ptr,
-            tmp_dft12_ptr+1,
-            tmp_dft12_ptr+2,
+  bfly3_tw1(tmp_dft12,
+            tmp_dft12+1,
+            tmp_dft12+2,
             y0,
             y4,
             y8);
@@ -7126,9 +7123,9 @@ static inline void dft12f(simd_q15_t *x0,
 
 
   //  k2=1;
-  bfly3(tmp_dft12_ptr+3,
-        tmp_dft12_ptr+4,
-        tmp_dft12_ptr+5,
+  bfly3(tmp_dft12+3,
+        tmp_dft12+4,
+        tmp_dft12+5,
         y1,
         y5,
         y9,
@@ -7138,9 +7135,9 @@ static inline void dft12f(simd_q15_t *x0,
 
 
   //  k2=2;
-  bfly3(tmp_dft12_ptr+6,
-        tmp_dft12_ptr+7,
-        tmp_dft12_ptr+8,
+  bfly3(tmp_dft12+6,
+        tmp_dft12+7,
+        tmp_dft12+8,
         y2,
         y6,
         y10,
@@ -7148,9 +7145,9 @@ static inline void dft12f(simd_q15_t *x0,
         W4_12);
 
   //  k2=3;
-  bfly3(tmp_dft12_ptr+9,
-        tmp_dft12_ptr+10,
-        tmp_dft12_ptr+11,
+  bfly3(tmp_dft12+9,
+        tmp_dft12+10,
+        tmp_dft12+11,
         y3,
         y7,
         y11,
@@ -10606,15 +10603,43 @@ int dfts_autoinit(void)
 
 #ifndef MR_MAIN
 
-void dft(uint8_t sizeidx, int16_t *sigF,int16_t *sig,unsigned char scale_flag){
-	AssertFatal((sizeidx>=0 && sizeidx<(int)DFT_SIZE_IDXTABLESIZE),"Invalid dft size index %i\n",sizeidx);
-	dft_ftab[sizeidx](sigF,sig,scale_flag);
+void dft(uint8_t sizeidx, int16_t *input,int16_t *output,unsigned char scale_flag){
+	AssertFatal((sizeidx >= 0 && sizeidx<DFT_SIZE_IDXTABLESIZE),"Invalid dft size index %i\n",sizeidx);
+        int algn=0xF;
+        #ifdef __AVX2__
+        if ( (dft_ftab[sizeidx].size%3) != 0 ) // there is no AVX2 implementation for multiples of 3 DFTs
+          algn=0x1F;
+        #endif 
+        AssertFatal(((intptr_t)output&algn)==0,"Buffers should be aligned %p",output);
+        if (((intptr_t)input)&algn) {
+          LOG_D(PHY, "DFT called with input not aligned, add a memcpy, size %d\n", sizeidx);
+          int sz=dft_ftab[sizeidx].size;
+          if (sizeidx==DFT_12) // This case does 8 DFTs in //
+            sz*=8;
+          int16_t tmp[sz*2] __attribute__ ((aligned(32))); // input and output are not in right type (int16_t instead of c16_t)
+          memcpy(tmp, input, sizeof tmp);
+          dft_ftab[sizeidx].func(tmp,output,scale_flag);
+        } else
+          dft_ftab[sizeidx].func(input,output,scale_flag);
 };
 
-void idft(uint8_t sizeidx, int16_t *sigF,int16_t *sig,unsigned char scale_flag){
-	AssertFatal((sizeidx>=0 && sizeidx<(int)IDFT_SIZE_IDXTABLESIZE),"Invalid idft size index %i\n",sizeidx);
-	idft_ftab[sizeidx](sigF,sig,scale_flag);
+void idft(uint8_t sizeidx, int16_t *input,int16_t *output,unsigned char scale_flag){
+	AssertFatal((sizeidx>=0 && sizeidx<DFT_SIZE_IDXTABLESIZE),"Invalid idft size index %i\n",sizeidx);
+        int algn=0xF;
+        #ifdef __AVX2__
+          algn=0x1F;
+        #endif
+        AssertFatal( ((intptr_t)output&algn)==0,"Buffers should be 16 bytes aligned %p",output);
+        if (((intptr_t)input)&algn ) {  
+          LOG_D(PHY, "DFT called with input not aligned, add a memcpy\n");
+          int sz=idft_ftab[sizeidx].size;
+          int16_t tmp[sz*2] __attribute__ ((aligned(32))); // input and output are not in right type (int16_t instead of c16_t)
+          memcpy(tmp, input, sizeof tmp);
+          dft_ftab[sizeidx].func(tmp,output,scale_flag);
+        } else
+          idft_ftab[sizeidx].func(input,output,scale_flag);
 };
+
 #endif
 
 /*---------------------------------------------------------------------------------------*/

openair2/LAYER2/NR_MAC_gNB/mac_proto.h

@@ -519,4 +519,7 @@ bool nr_find_nb_rb(uint16_t Qm,
 void nr_sr_reporting(int Mod_idP, frame_t frameP, sub_frame_t slotP);
 
 void dump_mac_stats(gNB_MAC_INST *gNB, char *output, int strlen, bool reset_rsrp);
+
+void process_CellGroup(NR_CellGroupConfig_t *CellGroup, NR_UE_sched_ctrl_t *sched_ctrl);
+
 #endif /*__LAYER2_NR_MAC_PROTO_H__*/