Fix overflow when negating minimum int16_t value in precoding
This closes#955
It prevents undefined behavior caused by the minimum int16_t value
(-32768) in two's complement representation. Because possible negation
by multiplying with -1 during precoding cause an overflows since +32768
cannot be represented in 16-bit signed integers and end up again in
-32768 because of overflow, here reveald by using the SIMD functions.
Workaround we use a slightly smaller number than 1 in Q15 Format to
prevent this error. Q15 representation of (integer value -32767) -1 +
2^-15 = -0.999969482421875
remove specific directory for LTE SIM management tools
This is very old code has never been integrated with the "main" OAI
cmake infrastructure and more recent libraries.
This MR integrates the three binaries for LTE virtual sim management in
the regular build process and log system
Remove --usim-test command line option
Remove the --usim-test option: to my knowledge, it's not used by users,
and is misleadingly placed in various places where there should not be a
difference w.r.t. USIM configuration.
When making the changes, I assumed that --usim-test would be 0 (the
default), which is the case in all executables when not specified in
options except for the nr_dlsim, nr_pbchsmi, nr_ulsim simulators.
The only exception is the initialization of PDCP from the MAC, which
does not seem to be correct (the PDCP is in the CU, so it does not make
sense to call it from the MAC which is in the DU).
TS 38.213 4.1 says coreset 0 is present only if kssb < 24 for FR1 and <
12 for FR2. This commit sets get_sib1 flag based on kssb.
Also, setting of mac state is taken out of function nr_ue_decode_mib().
Remove the --usim-test option: to my knowledge, it's not used by users,
and is misleadingly placed in various places where there should not be a
difference w.r.t. USIM configuration.
When making the changes, I assumed that --usim-test would be 0 (the
default), which is the case in all executables when not specified in
options except for the nr_dlsim, nr_pbchsmi, nr_ulsim simulators.
The only exception is the initialization of PDCP from the MAC, which is
not correct (the PDCP is in the CU, so it does not make sense to call it
from the MAC which is in the DU). Instead, always initialize from main,
even in the NSA case (NSA was covered by the incorrect initialization
from MAC).
This in turn let to some more cleanup around function du_rlc_data_req().
After receiving GTP packets at the DU, they were given to the PDCP to
enqueue the packet at RLC. This does not work anymore, since the PDCP is
not initialized; it also does not seem necessary, as this enqueue
functionality is necessary to decouple PDCP and RLC to avoid deadlocks
in monolithic, but in split-mode, this cannot happen. Instead, call
directly into the RLC when receiving GTP packets.
bugfix: avoid segmentation fault when Redcap tries to connect to gNB
bugfix: avoid segmentation fault when Redcap tries to connect to gNB
as NR_ServingCellConfig_t scd is NULL when Redcap does RA
NR UE: Enable additional PDU session
- fix a bug with wraparound of buffer_remain_ep
- Add --extra-pdu-id command line option
Testing procedure:
1. Start cn in openairinterface5g/doc/tutorial_resources/oai-cn5g with
docker compose up
2. Start gnb:
sudo ./nr-softmodem -O ../../targets/PROJECTS/GENERIC-NR-5GC/CONF/gnb.sa.band78.fr1.106PRB.usrpb210.conf --gNBs.[0].min_rxtxtime 6 --rfsim --rfsimulator.server_address server
3. Monitor extra interfaces being created:
watch -n 0.5 "ifconfig | grep tun"
4. Start UE:
sudo ./nr-uesoftmodem -C 3619200000 -r 106 --numerology 1 --ssb 516 --band 78 --rfsim --uicc0.imsi 001010000000001 --extra-pdu-id 11
5. An extra tun interface oaitun_ue1p11 should be created. Observe in
the terminal from step 3
6. Ping the core network from both interfaces, e.g., ping -I
oaitun_ue1p11 10.0.0.1, ping -I oaitun_ue1 10.0.0.1. This should
generate traffic on the gNB on different LCIDs.
Measurement gaps
Scheduling measurement gaps to provide sufficient time for the UE to
re-tune its transceiver to the target carrier, complete the set of
measurements and then re-tune its transceiver back to the original
carrier.
fix(RLC): Count dropped DL SDUs by the RLC
When a downlink SDU was discarded by the RLC Acknowledged Mode entity
due to a full transmit buffer (tx_maxsize), the drop statistics were not
being updated. The txpdu_dd_pkts and txpdu_dd_bytes counters remained at
0. It should be noted that while the statistics are named txpdu_dd_*,
the drop at this stage occurs at the SDU level.
speedup NR RLC RX for AM
rx_list is replaced by an array of size 2^(SN size - 1) which is 131072
for an SN size of 18 bits, leading to 1048576 bytes of memory used.
(Less for smaller SN sizes.)
A lot of RX processing was to run through rx_list. When the list was big
(with some non-ideal channel), it can take some time. Removing this list
should improve things. Checking with benchmark_nr_rlc_am_entity a
speedup of more thant x10 is achieved (except for the case 1024 which is
surprisingly faster than the others in the develop branch).
The only thing that may be slower is when looking for cur->next where
the SN is different, in which case we need to look for all SN one by one
until we find something or reach the limit (rx_highest_status). These
are the functions get_first_rx_pdu() and get_next_rx_pdu().
When a downlink SDU was discarded by the RLC Acknowledged Mode entity due to a full transmit buffer (tx_maxsize), the drop statistics were not being updated.
The `txpdu_dd_pkts` and `txpdu_dd_bytes` counters remained at 0.
It should be noted that while the statistics are named `txpdu_dd_*`, the drop at this stage occurs at the SDU level.
The changes done in this commit are to contribute the supporting changes
for the existing "Initial support for RedCap" feature in gNB.
In this patch, handle and store the RedCap related IE's present in "NR
UE Capability Information" msg which are
- 'supportOfRedCap-r17'
- 'supportOf16DRB-RedCap-r17'
- 'longSN-RedCap-r17'
- 'am-WithLongSN-RedCap-r17'
use 'longSN-RedCap-r17' and 'am-WithLongSN-RedCap-r17' IEs to set the SN
length to either 12 Bit or 18 Bit accordingly.
Add some helper functions that will be used by the MAC handler to fill
the right MeasGap configuration based on Measurement Configuration sent
by the CU, in the next commit.
Improvements to the FHI for enabling further use cases
These changes to the FHI were primarily thought to enable the integration of
the intel ACC ( !3344 ) in parallel of the FHI.
I preferred to make a separate MR because I believe this topic is orthogonal
enough to be handled separately and because I wanted to bring even further
improvements.
The improvements that it actually brings are:
- Enable compiling and running xran F release with DPDK 21+
- Enable including a BBDEV device in addition of the ethernet device for the
FHI when performing EAL init, which is necessary for using the intel ACC
with the FHI.
- Fix an issue in usage of pkgconfig in xran E and F release. This enables to
reliably build xran with the version of DPDK requested to pkgconfig while
before this fix xran was always selecting the headers installed in /usr/local.
Split out type and function definitions for a cleaner separation for
some consumers of these functions. Concretely, nr_mac_common.h includes
various ASN.1 NR RRC headers, but not all consumers of these APIs might
need these headers. For instance, the next commit will use these
functions in the driver for 7.2 (it needs to compute PRACH length), but
the driver itself does not depend on nor does it need the ASN.1 NR RRC
headers. Defining a new header avoids the problem.
continuously update SIB19 information for rfsim NTN LEO scenarios
Contains mainly these changes:
- gNB: add function nr_update_sib19() allowing to update SIB19
information
- rfsimulator: update SIB19 contents every frame for LEO channel
simulations
- NR UE: improve initial timing advance calculation from SIB19
Before, we only considered ta_Common_r17 and the SAT position from SIB19
to compute the initial timing advance value. This is good enough if the
satellite does not move too fast (e.g. GEO satellite).
Now we also consider ta_CommonDrift_r17, ta_CommonDriftVariant_r17 and
the SAT velocity together with the epoch time from SIB19. This improves
the accuracy of the initial TA computation, esp. for LEO satellite
scenarios.
Various fixes for UE handling, HO doc updates
- On release request, search both attach and connected list to free a UE.
- on DU disconnect, only release the correspnonding UE contexts.
- When no AMF for the UE, auto-release the UE (NGAP would also send an
answer, but better check first)
- implement a "grace period" before the DU/MAC automatically releases a
UE: this is to avoid cases in which (for whatever reason) the UE
release request is not honored by the CU. In order to not fill up UE
contexts unnecessarily, the DU will eventually (after 10s) drop the UE
context
- some clarification in HO tutorial
When the RA procedure fails at Msg4, we trigger UL failure and send a
message to the CU to request release. There was a bug in that upon the
ensueing release command, we did not search in the attach list (only
connected list). However, the UE failing at Msg4 does not have
transitioned to the connected list yet, thus not freeing the context.
Fixes: 8f62baefa6 ("Use list of UEs during RA instead of array of RA
processes)"
It might happen that the CU does not answer a release request. In that
case, the MAC would never free the UE context, potentially blocking the
DU. Avoid this by introducing a "grace period" after which a UE context
is automatically freed at the MAC (and RLC), currently at 10s after the
30s UL failure timer.
rx_list is replaced by an array of size 2^(SN size - 1)
which is 131072 for an SN size of 18 bits, leading to
1048576 bytes of memory used. (Less for smaller SN sizes.)
A lot of RX processing was to run through rx_list. When the list was
big (with some non-ideal channel), it can take some time. Removing this
list should improve things. Checking with benchmark_nr_rlc_am_entity
a speedup of more thant x10 is achieved (except for the case 1024 which
is surprisingly faster than the others in the develop branch).
The only thing that may be slower is when looking for cur->next where
the SN is different, in which case we need to look for all SN one by
one until we find something or reach the limit (rx_highest_status).
These are the functions get_first_rx_pdu() and get_next_rx_pdu().
NR RRC Reject UE
Improvements in procedures to handle RRCReject message both at gNB and
UE. At gNB we send RRC Reject if the UE cannot be given resources to go
into connected mode.
NR UE: enable CSI report for SISO case
Correct precoded_sinr_dB wrongly set to 0 for SISO case, also correct
ri_restriction is 0 when calculating CQI bit length for SISO.
Update/refine gNB MAC stats, update L1
Only applicable with -q option.
Print a number of statistics, e.g., rx_ulsch_sdu to track UL MAC stats
timing. Remove unused time_meas_t definitions, and fix the name for the
total gNB scheduler time.
Print the RU stats in nrL1_stats.log, as some are already there, and do
not "pollute" the periodic logs when -q is enabled.
I used this to see why there are sometimes late slots with Aerial. I
used the above stats to get a general overview. This additional patch
shows that at least in RFsim, the RLC indication in rx ulsch sdu can
take up to 20us on my machine. Given that we can have multiple PDUs in
UL (10?), this can amount to a sizeable amount of time. While this is
for another MR, I thought the cleanup would be nice to have.
Refactor F1 UE context messages, add unit tests
Add encoder+decoder+util functions for F1 UE context messages, and use
them in the stack. The overall goal is to unit test encoder+decoder (to
be sure that what we send is what we receive), and remove any memory
leaks related to F1 handling. Messages that have a unit test:
- F1 UE context setup request/response
- F1 UE context modification request/response
- F1 UE context release request/command/complete
This MR removes almost all remaining memory leaks that are observable in
a "standard" configuration, running SISO in RFsim. In RFsim (without
channel modelling), there are only two memory leaks shown in asan, one
in L1, and one in NGAP.
Robert Schmidt