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

Merge branch 'integration_2025_w36' into 'develop'

Browse Source 
Integration: `2025.w36`

* !3613 Speedup complex rotate for aarch64
* !3614 Ldpc encoder zc384 optimization
* !3625 gNB scheduler small fixes
* !3629 Fix for gNB max feedback time parameter
* !3619 Fix RA 2-Step MgsB Ack at UE
* !3626 Upgrade CI to Aerial 25-2

See merge request oai/openairinterface5g!3627
This commit is contained in:
Jaroslava Fiedlerova
2025-09-04 16:36:42 +00:00
parent f155ad7202 8e32225711
commit d372aa1b50
20 changed files with 614 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CMakeLists.txt
+1 -1
View File
@@ -170,7 +170,7 @@ endif()
if (NOT CROSS_COMPILE)
set(C_FLAGS_PROCESSOR "${C_FLAGS_PROCESSOR} -march=native")
else ()
set(C_FLAGS_PROCESSOR "${C_FLAGS_PROCESSOR} -march=armv8-a")
set(C_FLAGS_PROCESSOR "${C_FLAGS_PROCESSOR} -march=armv8.2-a")
endif()
# add autotools definitions that were maybe used!
ci-scripts/cls_containerize.py
+2 -2
View File
@@ -389,13 +389,13 @@ class Containerize():
elif image != 'ran-build':
cmd.run(f'sed -i -e "s#ran-build:latest#ran-build:{imageTag}#" docker/Dockerfile.{pattern}{self.dockerfileprefix}')
if image == 'oai-gnb-aerial':
cmd.run('cp -f /opt/nvidia-ipc/nvipc.src.2025.05.20.tar.gz .')
cmd.run('cp -f /opt/nvidia-ipc/nvipc_src.2025.08.27.tar.gz .')
logfile = f'{lSourcePath}/cmake_targets/log/{name}.docker.log'
ret = cmd.run(f'{self.cli} build {self.cliBuildOptions} --target {image} --tag {name}:{imageTag} --file docker/Dockerfile.{pattern}{self.dockerfileprefix} {option} . > {logfile} 2>&1', timeout=1200)
t = (name, archiveArtifact(cmd, ctx, logfile))
log_files.append(t)
if image == 'oai-gnb-aerial':
cmd.run('rm -f nvipc.src.2025.05.20.tar.gz')
cmd.run('rm -f nvipc_src.2025.08.27.tar.gz')
# check the status of the build
ret = cmd.run(f"{self.cli} image inspect --format=\'Size = {{{{.Size}}}} bytes\' {name}:{imageTag}")
if ret.returncode != 0:
ci-scripts/yaml_files/sa_gnb_aerial/docker-compose.yaml
+1 -1
View File
@@ -23,7 +23,7 @@ services:
- ../../../cmake_targets/share:/opt/cuBB/share
userns_mode: host
ipc: "shareable"
image: cubb-build:25-1
image: cubb-build:25-2
environment:
- cuBB_SDK=/opt/nvidia/cuBB
command: bash -c "sudo rm -rf /tmp/phy.log && sudo chmod +x /opt/nvidia/cuBB/aerial_l1_entrypoint.sh && /opt/nvidia/cuBB/aerial_l1_entrypoint.sh"
doc/Aerial_FAPI_Split_Tutorial.md
+2 -2
View File
@@ -230,10 +230,10 @@ After preparing the L1 software, the container needs to be committed to create a
In this file the RU MAC address needs to be specified before commiting the image.
```bash
~$ docker commit nv-cubb cubb-build:25-1
~$ docker commit nv-cubb cubb-build:25-2
~$ docker image ls
..
cubb-build 25-1 824156e0334c 2 weeks ago 23.9GB
cubb-build 25-2 824156e0334c 2 weeks ago 23.9GB
-..
```
docker/Dockerfile.build.ubuntu.cross-arm64
+2 -11
View File
@@ -37,20 +37,11 @@ RUN /bin/sh oaienv && \
mkdir -p log ran_build/build ran_build/build-cross && \
cd ran_build/build && \
cmake ../../.. -GNinja && \
echo "====== Start of log for host-tools ======" && \
ninja ldpc_generators generate_T > ../../log/host-tools.txt 2>&1 && \
echo "====== End of log for host-tools ======" && \
ninja ldpc_generators generate_T && \
cd ../build-cross/ && \
# install missing libyaml-cpp-dev for arm64
apt install -y libyaml-cpp-dev:arm64 && \
cmake ../../.. -GNinja -DENABLE_TELNETSRV=ON \
-DCMAKE_TOOLCHAIN_FILE=../../../cmake_targets/cross-arm.cmake -DNATIVE_DIR=../build \
-DCMAKE_C_FLAGS="-Werror" -DCMAKE_CXX_FLAGS="-Werror" && \
echo "====== Start of log for cross-build executables ======" && \
# making always passing so we can analyze the all.txt file
ninja -k10 dlsim ulsim ldpctest polartest smallblocktest nr_pbchsim nr_dlschsim nr_ulschsim nr_dlsim nr_ulsim nr_pucchsim nr_prachsim lte-softmodem nr-softmodem nr-cuup oairu lte-uesoftmodem nr-uesoftmodem params_libconfig coding rfsimulator telnetsrv > ../../log/all.txt 2>&1; \
# displaying errors and warnings
grep -E -A3 "warning:|error:" ../../log/all.txt || true && \
# will fail only if string "errors:" is found
grep -E "error:" -vqz ../../log/all.txt && \
echo "====== End of log for cross-build executables ======"
ninja -k10 dlsim ulsim ldpctest polartest smallblocktest nr_pbchsim nr_dlschsim nr_ulschsim nr_dlsim nr_ulsim nr_pucchsim nr_prachsim lte-softmodem nr-softmodem nr-cuup oairu lte-uesoftmodem nr-uesoftmodem params_libconfig coding rfsimulator telnetsrv
docker/Dockerfile.gNB.aerial.ubuntu
+1 -1
View File
@@ -45,7 +45,7 @@ WORKDIR /oai-ran
COPY . .
RUN /bin/sh oaienv && \
tar -xvzf nvipc* && \
tar -xvzf nvipc_src.*.tar.gz && \
cd nvipc_src.* && \
rm -rf build && mkdir build && cd build && \
cmake .. -DNVIPC_DPDK_ENABLE=OFF -DNVIPC_DOCA_ENABLE=OFF -DNVIPC_CUDA_ENABLE=OFF -DENABLE_SLT_RSP=ON && \
docker/Dockerfile.gNB.aerial.ubuntu.sanitize-address
+1 -1
View File
@@ -45,7 +45,7 @@ WORKDIR /oai-ran
COPY . .
RUN /bin/sh oaienv && \
tar -xvzf nvipc* && \
tar -xvzf nvipc_src.*.tar.gz && \
cd nvipc_src.* && \
rm -rf build && mkdir build && cd build && \
cmake .. -DNVIPC_DPDK_ENABLE=OFF -DNVIPC_DOCA_ENABLE=OFF -DNVIPC_CUDA_ENABLE=OFF -DENABLE_SLT_RSP=ON && \
openair1/PHY/CODING/nrLDPC_encoder/ldpc384_alignr_byte_128.c
+153
View File
File diff suppressed because one or more lines are too long
openair1/PHY/CODING/nrLDPC_encoder/ldpc384_simd512_alignr_byte.c
+151
View File
File diff suppressed because one or more lines are too long
openair1/PHY/CODING/nrLDPC_encoder/ldpc384_simd512_permutex_byte.c
+218
View File
File diff suppressed because one or more lines are too long
openair1/PHY/CODING/nrLDPC_encoder/ldpc_encode_parity_check.c
+22 -3
View File
@@ -36,13 +36,27 @@
#include "assertions.h"
#include "common/utils/LOG/log.h"
#define USE_PERMUTEX
#ifdef __aarch64__
#define USE_ALIGNR
#endif
#ifdef __AVX512F__
#if defined(__AVX512VBMI__) && defined(USE_PERMUTEX)
// For AVX512 machines, use an AVX512 version of the encoder for Zc=384 only for now. This is used almost exclusively for
// high-throughput cases. The version with "permutex2var" instruction uses less memory (i.e. 1/64th of the memory to store the
// input), but uses more reads instead of creating 64 shifts of the input with memcpy
#include "ldpc384_simd512_permutex_byte.c"
#else
#include "ldpc384_simd512_byte.c"
#endif
#else
#include "ldpc384_byte.c"
#endif
#ifdef USE_ALIGNR
#include "ldpc384_alignr_byte_128.c"
#else
#include "ldpc384_byte_128.c"
#endif
#include "ldpc352_byte.c"
#include "ldpc352_byte_128.c"
#include "ldpc320_byte.c"
@@ -96,8 +110,13 @@ static void encode_parity_check_part_optim(uint8_t *cc, uint8_t *d, short BG,sho
memcpy(&c[2 * i1 * Zc], &cc[i1 * Zc], Zc * sizeof(unsigned char));
memcpy(&c[(2 * i1 + 1) * Zc], &cc[i1 * Zc], Zc * sizeof(unsigned char));
}
for (int i1 = 1; i1 < simd_size; i1++) {
memcpy(&c[(2 * ncols * Zc * i1)], &c[i1], (2 * ncols * Zc * sizeof(unsigned char)) - i1);
#if (defined(USE_PERMUTEX) && defined(__AVX512VBMI__)) || defined(USE_ALIGNR)
if (Zc < 384 || BG == 2)
#endif
{
for (int i1 = 1; i1 < simd_size; i1++) {
memcpy(&c[(2 * ncols * Zc * i1)], &c[i1], (2 * ncols * Zc * sizeof(unsigned char)) - i1);
}
}
if (BG == 1) {
switch (Zc) {
openair1/PHY/TOOLS/tools_defs.h
+45 -1
View File
@@ -828,6 +828,47 @@ static inline void rotate_cpx_vector(const c16_t *const x, const c16_t *const al
// log2_amp - increase the output amplitude by a factor 2^log2_amp (default is 0)
// WARNING: log2_amp>0 can cause overflow!!
#ifdef __aarch64__
if (output_shift == 15) { // allows specific NEON instruction
int16x8_t ar = (int16x8_t)vdupq_n_s16(alpha->r);
int16x8_t ai = (int16x8_t)vdupq_n_s16(alpha->i);
int16x8_t *y_128 = (int16x8_t *)y;
int16x8_t *x_128 = (int16x8_t *)x;
for (uint32_t i = 0; i < (N >> 2); i++) {
// Split interleaved -> separate real/imag
int16x8_t br = vuzp1q_s16(x_128[i], x_128[i]);
int16x8_t bi = vuzp2q_s16(x_128[i], x_128[i]);
// Start with the two “diagonal” products using high-half, doubling, sat:
// x = round( (2*ar*br) / 2^16 ), y = round( (2*ar*bi) / 2^16 )
int16x8_t real = vqdmulhq_s16(ar, br);
int16x8_t imag = vqdmulhq_s16(ar, bi);
// real -= round( (2*ai*bi) / 2^16 )
real = vqrdmlshq_s16(real, ai, bi);
// imag += round( (2*ai*br) / 2^16 )
imag = vqrdmlahq_s16(imag, ai, br);
// Re-interleave [real, imag]
int16x8x2_t z = vzipq_s16(real, imag);
y_128[i] = z.val[0];
/*
printf("y : (%d %d) (%d %d) (%d %d) (%d %d)\n",
vgetq_lane_s16(y_128[i],0),
vgetq_lane_s16(y_128[i],1),
vgetq_lane_s16(y_128[i],2),
vgetq_lane_s16(y_128[i],3),
vgetq_lane_s16(y_128[i],4),
vgetq_lane_s16(y_128[i],5),
vgetq_lane_s16(y_128[i],6),
vgetq_lane_s16(y_128[i],7));*/
}
} else {
#endif
uint32_t i; // loop counter
simd_q15_t *y_128, alpha_128;
@@ -859,7 +900,10 @@ static inline void rotate_cpx_vector(const c16_t *const x, const c16_t *const al
shift));
// print_ints("y_128[0]=", &y_128[0]);
}
#if defined(__x86__) || defined(__x86_64__)
#ifdef __aarch64__
}
#endif //__aarch64__
#if defined(__x86_64__) || defined(__i386__)
}
#endif
}
openair1/SIMULATION/NR_PHY/dlsim.c
+1
View File
@@ -1032,6 +1032,7 @@ printf("%d\n", slot);
reset_meas(&gNB->dlsch_rate_matching_stats);
reset_meas(&gNB->dlsch_segmentation_stats);
reset_meas(&gNB->dlsch_modulation_stats);
reset_meas(&gNB->dlsch_precoding_stats);
reset_meas(&gNB->dlsch_encoding_stats);
reset_meas(&gNB->dci_generation_stats);
reset_meas(&gNB->tinput);
openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_RA.c
+8 -4
View File
@@ -856,6 +856,8 @@ static void nr_generate_Msg3_retransmission(module_id_t module_idP,
}
if (rbStart > (sched_pusch.bwp_info.bwpSize - ra->msg3_nb_rb)) {
// cannot find free vrb_map for msg3 retransmission in this slot
reset_beam_status(&nr_mac->beam_info, sched_frame, sched_slot, UE->UE_beam_index, slots_frame, beam_ul.new_beam);
reset_beam_status(&nr_mac->beam_info, frame, slot, UE->UE_beam_index, slots_frame, beam_dci.new_beam);
return;
}
@@ -915,10 +917,11 @@ static void nr_generate_Msg3_retransmission(module_id_t module_idP,
ss,
coreset,
&UE->UE_sched_ctrl.sched_pdcch,
true,
0);
if (CCEIndex < 0) {
LOG_E(NR_MAC, "UE %04x cannot find free CCE!\n", UE->rnti);
reset_beam_status(&nr_mac->beam_info, sched_frame, sched_slot, UE->UE_beam_index, slots_frame, beam_ul.new_beam);
reset_beam_status(&nr_mac->beam_info, frame, slot, UE->UE_beam_index, slots_frame, beam_dci.new_beam);
return;
}
@@ -1507,7 +1510,7 @@ static void nr_generate_Msg2(module_id_t module_idP,
}
uint8_t aggregation_level;
int CCEIndex = get_cce_index(nr_mac, CC_id, slotP, 0, &aggregation_level, beam.idx, ss, coreset, &sched_ctrl->sched_pdcch, true, 0);
int CCEIndex = get_cce_index(nr_mac, CC_id, slotP, 0, &aggregation_level, beam.idx, ss, coreset, &sched_ctrl->sched_pdcch, 0);
if (CCEIndex < 0) {
LOG_W(NR_MAC, "UE %04x: %d.%d cannot find free CCE for Msg2!\n", UE->rnti, frameP, slotP);
@@ -1707,7 +1710,6 @@ static void nr_generate_Msg4_MsgB(module_id_t module_idP,
ss,
coreset,
&sched_ctrl->sched_pdcch,
true,
0);
if (CCEIndex < 0) {
@@ -1735,8 +1737,10 @@ static void nr_generate_Msg4_MsgB(module_id_t module_idP,
TYPE_TC_RNTI_,
coreset->controlResourceSetId,
false);
if (!msg4_tda.valid_tda)
if (!msg4_tda.valid_tda) {
reset_beam_status(&nr_mac->beam_info, frameP, slotP, UE->UE_beam_index, n_slots_frame, beam.new_beam);
return;
}
NR_pdsch_dmrs_t dmrs_info = get_dl_dmrs_params(scc, dl_bwp, &msg4_tda, 1);
bwp_info_t bwp_info = get_pdsch_bwp_start_size(nr_mac, UE);
openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_dlsch.c
+1 -2
View File
@@ -551,7 +551,6 @@ static bool allocate_dl_retransmission(module_id_t module_id,
sched_ctrl->search_space,
sched_ctrl->coreset,
&sched_ctrl->sched_pdcch,
false,
sched_ctrl->pdcch_cl_adjust);
if (CCEIndex<0) {
sched_ctrl->dl_cce_fail++;
@@ -798,6 +797,7 @@ static void pf_dl(module_id_t module_id,
rbStart,
max_rbSize,
rbStop);
reset_beam_status(&mac->beam_info, frame, slot, iterator->UE->UE_beam_index, slots_per_frame, beam.new_beam);
iterator++;
continue;
}
@@ -811,7 +811,6 @@ static void pf_dl(module_id_t module_id,
sched_ctrl->search_space,
sched_ctrl->coreset,
&sched_ctrl->sched_pdcch,
false,
sched_ctrl->pdcch_cl_adjust);
if (CCEIndex < 0) {
sched_ctrl->dl_cce_fail++;
openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_phytest.c
-2
View File
@@ -139,7 +139,6 @@ void nr_preprocessor_phytest(module_id_t module_id, frame_t frame, slot_t slot)
sched_ctrl->search_space,
sched_ctrl->coreset,
&sched_ctrl->sched_pdcch,
false,
0);
AssertFatal(CCEIndex >= 0, "Could not find CCE for UE %04x\n", UE->rnti);
@@ -286,7 +285,6 @@ void nr_ul_preprocessor_phytest(gNB_MAC_INST *nr_mac, post_process_pusch_t *pp_p
sched_ctrl->search_space,
sched_ctrl->coreset,
&sched_ctrl->sched_pdcch,
false,
0);
if (CCEIndex < 0) {
LOG_E(MAC, "%s(): CCE list not empty, couldn't schedule PUSCH\n", __func__);
openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_primitives.c
+2 -3
View File
@@ -572,10 +572,9 @@ int get_cce_index(const gNB_MAC_INST *nrmac,
const NR_SearchSpace_t *ss,
const NR_ControlResourceSet_t *coreset,
NR_sched_pdcch_t *sched_pdcch,
bool is_common,
float pdcch_cl_adjust)
{
const uint32_t Y = is_common ? 0 : get_Y(ss, slot, rnti);
const uint32_t Y = get_Y(ss, slot, rnti);
uint8_t nr_of_candidates;
int agg_level_search_order[NUM_PDCCH_AGG_LEVELS];
@@ -2518,7 +2517,7 @@ void set_max_fb_time(NR_UE_UL_BWP_t *UL_BWP, const NR_UE_DL_BWP_t *DL_BWP)
{
UL_BWP->max_fb_time = 8; // default value
// take the maximum in dl_DataToUL_ACK list
if (DL_BWP->dci_format != NR_DL_DCI_FORMAT_1_0 && UL_BWP->pucch_Config) {
if (UL_BWP->pucch_Config) {
const struct NR_PUCCH_Config__dl_DataToUL_ACK *fb_times = UL_BWP->pucch_Config->dl_DataToUL_ACK;
for (int i = 0; i < fb_times->list.count; i++) {
if(*fb_times->list.array[i] > UL_BWP->max_fb_time)
openair2/LAYER2/NR_MAC_gNB/gNB_scheduler_ulsch.c
+2 -5
View File
@@ -1876,7 +1876,6 @@ static bool allocate_ul_retransmission(gNB_MAC_INST *nrmac,
sched_ctrl->search_space,
sched_ctrl->coreset,
&sched_ctrl->sched_pdcch,
false,
sched_ctrl->pdcch_cl_adjust);
if (CCEIndex<0) {
LOG_D(NR_MAC, "[UE %04x][%4d.%2d] no free CCE for retransmission UL DCI UE\n", UE->rnti, frame, slot);
@@ -2125,7 +2124,6 @@ static int pf_ul(gNB_MAC_INST *nrmac,
sched_ctrl->search_space,
sched_ctrl->coreset,
&sched_ctrl->sched_pdcch,
false,
sched_ctrl->pdcch_cl_adjust);
if (CCEIndex < 0) {
@@ -2444,7 +2442,7 @@ void post_process_ulsch(gNB_MAC_INST *nr_mac, post_process_pusch_t *pusch, NR_UE
sched_ctrl->last_ul_slot = sched_pusch->slot;
LOG_D(NR_MAC,
"ULSCH/PUSCH: %4d.%2d RNTI %04x UL sched %4d.%2d DCI L %d start %2d RBS %3d startSymbol %2d nb_symbol %2d dmrs_pos %x MCS "
"ULSCH/PUSCH: %4d.%2d RNTI %04x UL sched %4d.%2d DCI L %d start %2d RBS %3d TDA %2d dmrs_pos %x MCS "
"Table %2d MCS %2d nrOfLayers %2d num_dmrs_cdm_grps_no_data %2d TBS %4d HARQ PID %2d round %d RV %d NDI %d est %6d sched "
"%6d est BSR %6d TPC %d\n",
frame,
@@ -2455,8 +2453,7 @@ void post_process_ulsch(gNB_MAC_INST *nr_mac, post_process_pusch_t *pusch, NR_UE
sched_ctrl->aggregation_level,
sched_pusch->rbStart,
sched_pusch->rbSize,
sched_pusch->tda_info.startSymbolIndex,
sched_pusch->tda_info.nrOfSymbols,
sched_pusch->time_domain_allocation,
sched_pusch->dmrs_info.ul_dmrs_symb_pos,
current_BWP->mcs_table,
sched_pusch->mcs,
openair2/LAYER2/NR_MAC_gNB/mac_proto.h
-1
View File
@@ -431,7 +431,6 @@ int get_cce_index(const gNB_MAC_INST *nrmac,
const NR_SearchSpace_t *ss,
const NR_ControlResourceSet_t *coreset,
NR_sched_pdcch_t *sched_pdcch,
bool is_common,
float pdcch_cl_adjust);
bool nr_find_nb_rb(uint16_t Qm,
openair2/NR_UE_PHY_INTERFACE/NR_IF_Module.c
+1 -1
View File
@@ -1190,7 +1190,7 @@ void update_harq_status(NR_UE_MAC_INST_t *mac, uint8_t harq_pid, uint8_t ack_nac
if (current_harq->active) {
LOG_D(PHY,"Updating harq_status for harq_id %d, ack/nak %d\n", harq_pid, current_harq->ack);
// we can prepare feedback for MSG4 in advance
if (mac->ra.ra_state == nrRA_WAIT_CONTENTION_RESOLUTION)
if (mac->ra.ra_state == nrRA_WAIT_CONTENTION_RESOLUTION || mac->ra.ra_state == nrRA_WAIT_MSGB)
prepare_msg4_msgb_feedback(mac, harq_pid, ack_nack);
else {
current_harq->ack = ack_nack;