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

Simplify PRACH beam handling, remove race

Browse Source 
Simplify by removing a small malloc and memcpy() the beams. This should
also remove the data race between ru and gNB/CU-DU
This commit is contained in:
Laurent THOMAS
2025-10-30 11:54:30 +01:00
committed by Robert Schmidt
parent cffb07d798
commit d005336cb8
4 changed files with 15 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files
executables/nr-ru.c
+1 -1
View File
@@ -1222,7 +1222,7 @@ void *ru_thread(void *param)
for (int prach_oc = 0; prach_oc < p->lower.num_prach_ocas; prach_oc++) {
int prachStartSymbol = p->lower.prachStartSymbol + prach_oc * N_dur;
int beam_id = p->lower.beam ? p->lower.beam[prach_oc] : 0;
int beam_id = p->beams[prach_oc];
//comment FK: the standard 38.211 section 5.3.2 has one extra term +14*N_RA_slot. This is because there prachStartSymbol is given wrt to start of the 15kHz slot or 60kHz slot. Here we work slot based, so this function is anyway only called in slots where there is PRACH. Its up to the MAC to schedule another PRACH PDU in the case there are there N_RA_slot \in {0,1}.
rx_nr_prach_ru(ru,
p->lower.fmt, // could also use format