UL is now being scheduled but not decoded ok yet

This commit is contained in:
Florian Kaltenberger
2020-01-23 12:07:47 +01:00
committed by Florian Kaltenberger
parent 2605e3f18a
commit 00f0e0ccb8
11 changed files with 108 additions and 79 deletions
+3 -4
View File
@@ -356,7 +356,7 @@ void processSlotTX( PHY_VARS_NR_UE *UE, UE_nr_rxtx_proc_t *proc) {
nr_scheduled_response_t scheduled_response;
// program PUSCH. this should actually be done by the MAC upon reception of an UL DCI
if (proc->nr_tti_tx == NR_UPLINK_SLOT || UE->frame_parms.frame_type == FDD){
if (proc->nr_tti_tx == 8 || UE->frame_parms.frame_type == FDD){
dcireq.module_id = UE->Mod_id;
dcireq.gNB_index = 0;
@@ -383,7 +383,7 @@ void processSlotTX( PHY_VARS_NR_UE *UE, UE_nr_rxtx_proc_t *proc) {
rvidx = 0;
//------------------------------------------------------------------------------//
scheduled_response.ul_config->sfn_slot = NR_UPLINK_SLOT;
scheduled_response.ul_config->sfn_slot = 8;
scheduled_response.ul_config->number_pdus = 1;
scheduled_response.ul_config->ul_config_list[0].pdu_type = FAPI_NR_UL_CONFIG_TYPE_PUSCH;
scheduled_response.ul_config->ul_config_list[0].ulsch_config_pdu.rnti = n_rnti;
@@ -735,7 +735,6 @@ void *UE_thread(void *arg) {
readBlockSize,
UE->frame_parms.nb_antennas_rx),"");
if (slot_nr==18)
AssertFatal( writeBlockSize ==
UE->rfdevice.trx_write_func(&UE->rfdevice,
timestamp+
@@ -745,7 +744,7 @@ if (slot_nr==18)
txp,
writeBlockSize,
UE->frame_parms.nb_antennas_tx,
4),"");
1),"");
if( slot_nr==(nb_slot_frame-1)) {
// read in first symbol of next frame and adjust for timing drift