maki/openairinterface5g
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Debug removed and first fix added for false RA caused by UL scheduling on PRACH resources

Browse Source
This commit is contained in:
Louis Adrien Dufrene
2019-01-16 08:58:53 +01:00
parent c638d16a04
commit 0aff104545
2 changed files with 14 additions and 111 deletions
Download Patch File Download Diff File Expand all files Collapse all files
openair2/LAYER2/MAC/eNB_scheduler_fairRR.c
+1 -1
View File
@@ -2582,7 +2582,7 @@ schedule_ulsch_fairRR(module_id_t module_idP, frame_t frameP,
frame_parms->prach_config_common.prach_ConfigInfo.prach_ConfigIndex,
frame_parms->prach_config_common.prach_ConfigInfo.prach_FreqOffset,
0,//tdd_mapindex
frameP); //Nf
frameP); //Nf --> shouldn't it be sched_frame ???
ulsch_ue_select[CC_id].list[ulsch_ue_select[CC_id].ue_num].nb_rb = 6;
ulsch_ue_select[CC_id].list[ulsch_ue_select[CC_id].ue_num].UE_id = -1;
ulsch_ue_select[CC_id].ue_num++;
openair2/LAYER2/MAC/eNB_scheduler_ulsch.c
+13 -110
View File
@@ -133,31 +133,6 @@ rx_sdu(const module_id_t enb_mod_idP,
memset(rx_lcids, 0, NB_RB_MAX * sizeof(unsigned char));
memset(rx_lengths, 0, NB_RB_MAX * sizeof(unsigned short));
// LAD
/*
if (UE_id == -1) {
LOG_E(MAC, "Step 1\n");
LOG_W(MAC, "[MAC] UE_id = -1 ; RNTI = %x ; frame = %d ; subframe = %d ; sdu_length = %d ; sdu = %d\n",
rntiP,
frameP,
subframeP,
sdu_lenP,
*sduP);
}
*/
// LAD
/*
first_rb = UE_list->UE_template[CC_idP][UE_id].first_rb_ul[harq_pid];
LOG_W(MAC, "[MAC] UE_id = %d : first_rb = %d ; scheduled_ul_bytes = %d ; TBS_UL = %d ; frame = %d ; subframe = %d\n",
UE_id,
first_rb,
UE_list->UE_template[CC_idP][UE_id].scheduled_ul_bytes,
UE_list->UE_template[CC_idP][UE_id].TBS_UL[harq_pid],
frameP,
subframeP);
*/
start_meas(&mac->rx_ulsch_sdu);
VCD_SIGNAL_DUMPER_DUMP_FUNCTION_BY_NAME(VCD_SIGNAL_DUMPER_FUNCTIONS_RX_SDU, 1);
@@ -206,17 +181,7 @@ rx_sdu(const module_id_t enb_mod_idP,
mac_eNB_rrc_ul_in_sync(enb_mod_idP, CC_idP, frameP, subframeP, current_rnti);
}
/* update bytes to schedule */
// LAD
/*
LOG_E(MAC, "Step 2\n");
LOG_W(MAC, "[MAC] UE_id != -1 and sduP != NULL : first_rb = %d ; scheduled_ul_bytes = %d ; TBS_UL = %d\n",
first_rb,
UE_list->UE_template[CC_idP][UE_id].scheduled_ul_bytes,
UE_list->UE_template[CC_idP][UE_id].TBS_UL[harq_pid]);
*/
/* Update bytes to schedule */
UE_list->UE_template[CC_idP][UE_id].scheduled_ul_bytes -= UE_list->UE_template[CC_idP][UE_id].TBS_UL[harq_pid];
if (UE_list->UE_template[CC_idP][UE_id].scheduled_ul_bytes < 0) {
@@ -310,15 +275,6 @@ rx_sdu(const module_id_t enb_mod_idP,
first_rb = ra[RA_id].msg3_first_rb;
// LAD
/*
LOG_E(MAC, "Step 3\n");
LOG_W(MAC, "[MAC] UE_id == -1 : first_rb = %d ; scheduled_ul_bytes = %d ; TBS_UL = %d\n",
first_rb,
UE_list->UE_template[CC_idP][UE_id].scheduled_ul_bytes,
UE_list->UE_template[CC_idP][UE_id].TBS_UL[harq_pid]);
*/
if (sduP == NULL) { // we've got an error on Msg3
LOG_W(MAC, "[eNB %d] CC_id %d, RA %d ULSCH in error in round %d/%d\n",
enb_mod_idP,
@@ -333,26 +289,6 @@ rx_sdu(const module_id_t enb_mod_idP,
// first_rb = UE_list->UE_template[CC_idP][UE_id].first_rb_ul[harq_pid]; // UE_id = -1 !!!!
ra[RA_id].msg3_round++;
// LAD
/*
LOG_E(MAC, "Step 4\n");
LOG_W(MAC, "[MAC] [UEINFO1] UE_id = %d ; RNTI_ue_template = %x ; RNTI_sdu = %x\n",
UE_id,
UE_list->UE_template[CC_idP][UE_id].rnti,
current_rnti);
// LAD
LOG_W(MAC, "[MAC] UE_id == -1 and sduP == NULL : first_rb = %d ; scheduled_ul_bytes = %d ; TBS_UL = %d\n",
first_rb,
UE_list->UE_template[CC_idP][UE_id].scheduled_ul_bytes,
UE_list->UE_template[CC_idP][UE_id].TBS_UL[harq_pid]);
// LAD
LOG_W(MAC, "[MAC] [RAPROC] msg3_subframe = %d ; msg3_frame = %d\n",
ra[RA_id].Msg3_subframe,
ra[RA_id].Msg3_frame);
*/
/* Prepare handling of retransmission */
get_Msg3allocret(&mac->common_channels[CC_idP],
ra[RA_id].Msg3_subframe,
@@ -360,13 +296,6 @@ rx_sdu(const module_id_t enb_mod_idP,
&ra[RA_id].Msg3_frame,
&ra[RA_id].Msg3_subframe);
// LAD
/*
LOG_W(MAC, "[MAC] [RAPROC] After update: msg3_subframe = %d ; msg3_frame = %d\n",
ra[RA_id].Msg3_subframe,
ra[RA_id].Msg3_frame);
*/
add_msg3(enb_mod_idP, CC_idP, &ra[RA_id], frameP, subframeP);
}
@@ -421,17 +350,6 @@ rx_sdu(const module_id_t enb_mod_idP,
if (UE_id != -1) {
UE_list->UE_sched_ctrl[UE_id].round_UL[CC_idP][harq_pid] = 0;
}
// LAD
/*
LOG_E(MAC, "Step 5\n");
LOG_W(MAC, "[MAC] [UEINFO2] UE_id = %d ; RNTI_ue_template = %x ; RNTI_sdu = %x ; frame = %d ; subframe = %d\n",
UE_id,
UE_list->UE_template[CC_idP][UE_id].rnti,
current_rnti,
frameP,
subframeP);
*/
/* Control element */
for (int i = 0; i < num_ce; i++) {
@@ -1278,42 +1196,27 @@ schedule_ulsch(module_id_t module_idP,
LTE_DL_FRAME_PARMS *frame_parms = &(RC.eNB[module_idP][CC_id]->frame_parms);
RA_t *ra_ptr = cc->ra;
/* Louis-Adrien: Only set for FDD (for the moment)
* Hard coded for prach-ConfigIndex = 0 and prach-Freqoffset = 2
* ToDo: The PRACH resources should be added with modularity (here?)
/* From Louis-Adrien to François:
* The comment bloc below is to configure with a command line.
* I took it from the equivalent part in the fairRR scheduler (around line 2578 in eNB_scheduler_fairRR.c).
* As said in the meeting, it seems to work only for small TBS.
* The cause of false RA still present with this fix is to investigate.
*
* Note: in the get_prach_prb_offset() function below, the last argument is frameP in eNB_scheduler_fairRR.c
* I think it should be sched_frame instead. This parameter has only impacts in case TDD and preamble format 4.
* To confirm.
*/
/*
if (cc->tdd_Config == NULL) { // FDD
if (((sched_frame %2) == 0) && sched_subframe == 1) { // RACH frame and subframe
if (first_rb[CC_id] < 8) {
n_rb_ul_tab = to_prb(cc->ul_Bandwidth); // return total number of PRB
if (n_rb_ul_tab >= 8) {
first_rb[CC_id] = 8;
} else {
return;
}
}
}
}
*/
if (is_prach_subframe(frame_parms, sched_frame, sched_subframe) == 1) {
start_rb = get_prach_prb_offset(frame_parms,
frame_parms->prach_config_common.prach_ConfigInfo.prach_ConfigIndex,
frame_parms->prach_config_common.prach_ConfigInfo.prach_FreqOffset,
0, // tdd_mapindex
frameP); // Nf --> shouldn't it be sched_frame ???
sched_frame); // Nf
first_rb[CC_id] = start_rb + nb_rb;
// LAD
LOG_W(MAC, "[MAC] Config Index = %d ; Freq_offset = %d ; first_rb = %d ; subframe = %d ; sched_subframe = %d\n",
frame_parms->prach_config_common.prach_ConfigInfo.prach_ConfigIndex,
frame_parms->prach_config_common.prach_ConfigInfo.prach_FreqOffset,
first_rb[CC_id],
subframeP,
sched_subframe);
}
*/
/*
* Check if RA (Msg3) is active in this subframeP, if so skip the PRB used for Msg3
@@ -1326,7 +1229,7 @@ schedule_ulsch(module_id_t module_idP,
first_rb[CC_id] = ra_ptr->msg3_first_rb + ra_ptr->msg3_nb_rb;
}
/* Louis-Adrien: I couldn't find an interdiction of multiple Msg3 scheduling
* on the same resources. Also the performance improvement of breaking is low,
* on the same time resources. Also the performance improvement of breaking is low,
* since we will loop until the end, most of the time.
* I'm letting the break as a reminder, in case of misunderstanding the spec.
*/