Fix race condition on double RA during RRC reestablishment
Sometimes, especially in poor channel conditions, a UE can trigger
multiple RA procedures during RRC reestablishment, causing
reconfigCellGroup to be overwritten with a CellGroup that already has
spCellConfig removed, leading to an assertion failure in
update_cellgroup_for_reestablishment(). See logs below
[NR_MAC] 456.13 UE 991d: Received Ack of Msg4. CBRA procedure succeeded (UE Connected)
[NR_MAC] Adding new UE context with RNTI 0x991d
[NR_PHY] [RAPROC] 461.9 Initiating RA procedure with preamble 8, energy 48.0 dB (I0 274, thres 130), delay 21 start symbol 2 freq index 0
[NR_MAC] 461.9 UE RA-RNTI 0081 TC-RNTI ecd8: initiating RA procedure
[NR_MAC] UE ecd8: Sync beam index 0
[NR_MAC] UE ecd8: Msg3 scheduled at 462.3 (461.18 TDA 2) start 0 RBs 8
[NR_MAC] UE ecd8: 461.18 Generating RA-Msg2 DCI, RA RNTI 0x81, state 1, preamble_index(RAPID) 8, timing_offset = 21 (estimated distance 820.3 [m])
[NR_MAC] 462. 3 PUSCH with TC_RNTI 0xecd8 received correctly
[NR_MAC] Received a MAC CE for C-RNTI with 991d
[NR_MAC] Remove NR rnti 0xecd8
[NR_MAC] 462. 3 RA with C-RNTI 991d complete
[NR_PHY] [RAPROC] 471.19 Initiating RA procedure with preamble 6, energy 51.0 dB (I0 276, thres 130), delay 21 start symbol 2 freq index 0
[NR_MAC] 471.19 UE RA-RNTI 010d TC-RNTI 022d: initiating RA procedure
[NR_MAC] UE 022d: Sync beam index 0
[NR_MAC] UE 022d: Msg3 scheduled at 472.13 (472.8 TDA 2) start 0 RBs 8
[NR_MAC] UE 022d: 472.8 Generating RA-Msg2 DCI, RA RNTI 0x10d, state 1, preamble_index(RAPID) 6, timing_offset = 21 (estimated distance 820.3 [m])
[NR_MAC] 472.13 PUSCH with TC_RNTI 0x022d received correctly
[NR_MAC] Received a MAC CE for C-RNTI with 991d
[NR_MAC] Remove NR rnti 0x022d
[NR_MAC] 472.13 RA with C-RNTI 991d complete
[NR_RRC] [UL] (cellID 1, UE ID 1 RNTI 991d) Received RRCReestablishmentComplete
[NR_RRC] UE 1 Processing NR_RRCReestablishmentComplete from UE
[NR_MAC] UE 991d: gNB-DU Configuration Query received, will include CellGroupConfig in response
Assertion (new_CellGroup->spCellConfig) failed!
In update_cellgroup_for_reestablishment() /phy/openairinterface5g/openair2/LAYER2/NR_MAC_gNB/mac_rrc_dl_handler.c:666
This MR solves it by skipping the reconfigCellGroup overwrite when it
already holds a valid config with spCellConfig, and by replacing the
fatal assertion on missing spCellConfig with an early return.
T: macpdu2wireshark: gentle exit (and some other minor work in the T tracer)
When outputting to a file, typing ctrl+c exits the program without
calling fclose(), leading to loss of traces in the output file.
Let's intercept ctrl+c and exit gently, calling fclose().
Also include a small fix for compilation of _check_vcd, some work on
extract and hacks/plot-ofdm, minor stuff.
Add a simple GTP unit test
Add a simple unit test for GTP that as of now tests the main
functionality, i.e., for a single UE, set up a bearer between two
endpoints and checking that traffic sending works. This is a start to
extend it further, e.g., with API changes, extensions to support new
functionality such as IPv6, and refactorings.
A number of supporting cleanup has been done as well.
remove cblas and atlas dependancy
With new Fedora, we have again the cblas.h place not working
cblas+atlas dependancy is annoying for years
i start this MR to remove this dependancy
first commit: only cleaning: remove dead code, make static functions,
simplify code
after this commit, we see we use blas only in: conjugate_transpose()
that is a quite trivial function H_hermH_plus_sigma2I() a bit more
complex, but still possible to recode mutl_matrix_matrix_col_based() i
think we have the same elsewhere in OAI
atlas is used only in: lin_eq_solver() in don't know if replacement is
hard
We can observe all is related to MMSE, that is also coded in OAI 5G, so
maybe the solution is to use 5G code in 4G
Since nothing seems to be used, we can remove it.
For unit testing, allow to remove endpoints such that we could reuse a
port/IP combination.
The function does not take care of endpoints that are still "dangling",
it is the user's responsibility to remove all existing TEIDs.
Add a simple unit test that tests sending unidirectional traffic. So
far, it merely replicates what is already in nr-cuup-load-test, but it's
a single executable (so easier to handle), and could be more easily
extended to further tests (e.g., IPv6).
A couple of ideas for refactoring/improvements have been added in
comments in the test.
When outputting to a file, typing ctrl+c exits the program without calling
fclose(), leading to loss of traces in the output file.
Let's intercept ctrl+c and exit gently, calling fclose().
Since the parent commit, USE_MMSE is not defined. Thus, the
corresponding code is commented out. Since there are no linker problems,
it is obvious that this code is never called, and we can remove it.
This reverts commit 16c333b1a3.
I tried to build everything in Ubuntu images, but
- this introduced a regression (ignores $BUILD_OPTION)
- made that we don't use -Werror (when we should)
I tried to fix the use of the first, but it would involve more changes
than I prefer. On the second, we cannot enable -Werror, as FlexRIC does
not build warning-free (in Release mode). We thus have to build only the
targets that we require, and therefore we can keep using build_oai,
which is the simplest option (cf. the first point).
Integration: `2026.w07`
* !3899 Solving issues caused by invalid RRC configuration
* !3905 Add undocumented Aerial ERROR.indication vendor specific error codes
* !3896 Add fallbacks for ARMv8a to make it compile
* !3734 vrtsim: add CIR DB playback with CLI flags
* !3886 Integrate GPU-Accelerated Channel Simulation + CI Test Pipeline
* !3907 RRC QoS refactor: PDU Session Modify support, SDAP fixes, and code improvements
* !3891 Fix for split 7.2 DU configurations including DU & RU ethernet addresses
* !3913 Remove VCD signal dumper from NR code
* !3783 RRC: Multi-cell handling architecture
* !3916 CI: Rename RAN-SA-2x2-Module-CN5G pipeline
* !3922 Fix and update tracy tag to v0.13.1
* !3917 Fix SRS power computation
* !3915 Cleanup Cmake: remove some double compilation
* !3883 Improvements in handling beamforming inputs
* !3906 fix(ci): remove git reset command from Jenkinsfile
* !3894 Reimplement BladeRF radio driver, make it work for 5G
Closes#845
See merge request oai/openairinterface5g!3908
fix(ci): remove git reset command from Jenkinsfile
- By default, the variable eNB_CommitID is set as 'develop' in the
config
- When OAI-FLEXRIC-RAN-Integration-Test is triggered by upstream FlexRIC
push events, it takes develop
- Triggered by RAN takes commit SHA where origin/SHA does not work
Below are issues seen after this week's merge to develop:
https://jenkins-oai.eurecom.fr/view/RAN/job/RAN-Ubuntu-ARM-Image-Builder/4103/pipeline-overview/?selected-node=44
08:43:07 + git reset --hard origin/9777d23173a740de7cf982659fae13ad6627d3aa
08:43:07 fatal: ambiguous argument 'origin/9777d23173a740de7cf982659fae13ad6627d3aa': unknown revision or path not in the working tree.
When triggered through upstream FlexRIC (as a PUSH Event), it is taking
the branch as shown in the logs:
https://jenkins-oai.eurecom.fr/job/OAI-FLEXRIC-RAN-Integration-Test/3832/pipeline-overview/?selected-node=47
13:45:27 + git reset --hard origin/develop
13:45:27 HEAD is now at 9777d23173 Merge branch 'integration_2026_w06' into 'develop'
Cleanup Cmake: remove some double compilation
Harmonize some cmake code to avoid double compilation, notably around
polar and smallblock. In develop, we just add their sources to other
targets, whereas now we define STATIC libraries (further changed to
shared library in !3856) to avoid double compilation.
With this command line:
cmake .. -GNinja -DENABLE_TESTS=ON && ninja
shows that before we compile 12631 files, where after it is 12583
there is additional supporting cleanup (move code, remove headers,
types, ...).
there is more harmonization to do, I limited myself to some easy wins.
For instance, we do the same with PDCP/SDAP, and this should be
refactored into STATIC libraries as well.
Fix SRS power computation
The SRS symbols have been remapped recently in
srs_estimated_channel_freq, and therefore the calculation of the SRS
power has to retrieve the symbols from different positions.
CI: Rename RAN-SA-2x2-Module-CN5G pipeline
RAN-SA-2x2-Module-CN5G performs tests not only with 2x2 but also with a
4x4 MIMO configuration. The pipeline name has been generalized to
RAN-SA-Multi-Antenna-CN5G to cover multiple antenna configurations.
RRC: Multi-cell handling architecture
This MR implements a foundational multi-cell handling architecture for
RRC by separating DU management from cell management and establishing a
hierarchical cell organization within each DU container. The changes
enable proper multi-cell support where each DU can serve multiple cells,
addressing the previous limitation where DU and cell management were
tightly coupled with the assumption of one cell per DU.
Key Changes:
- Data Structure Separation: Separated for DU containers (DU-specific
info) from cell containers (cell-specific data)
- Global DU and Cell Trees: Cells are now stored in global RB trees in
RRC, similarly to DUs
- Du-specific cell seq_arr: within each DU container (du->cells) is
stored a seq_arr of cells belonging to the DU
- UE Cell Association Framework: Introduced serving_cells array to track
PCell and SCells for each UE with servCellIndex indexing
- Cell Management Library: Extracted cell management functions into
rrc_cell_management.c/h for better testability and maintainability
Other relevant changes:
- F1AP Refactoring: cell list definitions, heap allocation to prevent
stack overflow
- 5G-S-TMSI Helpers: Common helper functions for 3GPP TS 23.003
compliant construction
- Integration Tests: Comprehensive test suite for DU lookup, cell
lookup, and UE cell association
- Cell Duplication Checks: Validation in F1 DU configuration update to
prevent duplicate cells
- Documentation: Updated RRC developer documentation with multi-cell
architecture details
Validated with RFsim in re-establishment and handover scenarios.
openair2/COMMON/rrc_messages_types.h has mostly 4G types. The only 5G
(gNB) RRC type is gNB_RrcConfigurationReq, which we can move to a header
in the corresponding directory (where it is used).
This allows to separate 5G gNB RRC module from the 4G types in
rrc_messages_types.h in the next commit.
reversing bits is not NR specific. Think of it like this: LTE code might
use reverse_bits*() as well, but in the old location, it would pull in a
bunch of NR dependencies that are superfluous.
Remove VCD signal dumper from NR code
We don't test VCD in 5G NR code (and haven't tested it in 4G for years).
Remove the VCD traces in 5G code to avoid any ambiguity. 4G code remains
in place, as well as the tools to analyze existing traces.
Move the module cmake code to its subdirectory for better organization,
and remove superfluous include_directories() that are not recommended
to be used (as it populates all other targets with those directories,
most of which don't need it anyway).
Slightly clean up the module by removing headers to unused dependencies.
The GTP module has a dependency to 4G and 5G RLC, which should not exist
(the RLC modules have a dependency on GTP, not the other way around),
but in this commit, make as little modifications as possible.
Robert Schmidt