integration_2022_wk31b
See merge request oai/openairinterface5g!1649
MR !1624 NR PUCCH2 fix
MR !1627 fix for pusch dtx in phytest mode
MR !1634 Minor optimization in function init_symbol_rotation
MR !1637 gtp: small fix: set E if there is an extension header
MR !1641 Resolve AMF ip address using FQDN
delete files rxsigF0.m, txsig0.m, ldpctest_BG_1_Zc_384_rate_1-3_block_length_8448_maxit_5.txt
MR !1593 Reworking of BWP handling at MAC (initial MR)
MR !1652 fix for 3-digit MNC in asn1_msg.c
integration_2022_wk30b
See merge request oai/openairinterface5g!1630
MR !1594 Schedule BWP switching to the first active BWP when RA with Msg3 carrying DCCH or DTCH message
MR !1628 Fix a segmentation fault when there is no spCellConfigDedicated with maxMIMO_Layers parameter
MR !1528 CSI Feedback
MR !1631 Fix floating point for SRS
MR !1632 gtp pgu session header
MR !1633 Improve SIB1 length to support maximum value specified in 3GPP TS 38.331 section 5.2.1
MR !1463 Support arbitrary number of retransmissions (up to 8)
integration_2022_wk30
See merge request oai/openairinterface5g!1626
MR !1616 NR SA tutorials v2
MR !1620 nr_pdcp: nia1 implementation
MR !1622 CI: do not trigger tests if nothing is tagged NSA
MR !1623 Fix for ssb subcarrier offset in case of f<3GHz
MR !1606 code maintenance
MR !1568 Add undefined behavior sanitizer in build_oai, fix some errors
MR !1587 Dynamically build L2sim proxy
The previous version would trigger sub-pipelines (LTE-2x2, ...) if no MR
had the NSA tag (without a commit, etc). Of course, this would make all
pipelines subsequently fail (as not enough arguments, such as the
commit, were given). In this commit, we read the output of `jq` from
JSON, which will result in an empty list if no MR is tagged with the NSA
label (and the output in the console will be nicer as well).
In the previous version, not all fields, and in particular the
`eNB_index`, were not initialized, leading to errors when they were used
(e.g., the PDCP code checks that eNB_index == 0).
integration_2022_wk26
See merge request oai/openairinterface5g!1605
MR !1602 Use (std) bool instead of custom bool types
MR !1600 Discard RTX if RI changed
MR !1598 Set correct master clock rate for N320
MR !1584 enablement of statistics
MR !1603 Fix decoding multiple ULSCH in the same slot in the case of a ULSCH fails to be detected
MR !1590 Improvements and AVX2 in cpx_rotate_vector()
OAI is riddled with boolean_t, FALSE, TRUE, etc. This makes no sense, as
there is a standard bool type, and could lead to nasty bugs if a
definition of a bool is non-standard (0 == false, true == !false). This
commit removes all non-standard bools in the whole project.
integration_2022_wk25
See merge request oai/openairinterface5g!1596
MR !1549: Remove ssb subcarrier offset from config file
MR !1589: Putting all CM code in one file
MR !1592: Support Fedora 36
MR !1492: Harmonization of PUCCH Resources
MR !1581: F1: Interoperability with Accelleran CU
MR !1599: nr rlc: minor: update tests
integration_2022_wk24
See merge request oai/openairinterface5g!1591
MR !1382: Semi-automatic code generator for LDPC decoder for AVX2 and AVX512
MR !1521: 5G NR performance improvements for low SNR conditions
MR !1579: pthread_create, RT scheduler, and address sanitizer are incompatible in Ubuntu 18.04
MR !1588: small fix in csi configuration
MR !1583: Encoding/decoding CellGroup at MAC and fix RA with Msg3 carrying DCCH or DTCH messages
The CI compiles the physims using Asan. This in turn compiles Asan into
the LDPC decoder generator files. Since they are run during build, there
are two options:
1) trigger the build using capability SYS_PTRACE to allow running asan
when executing the generators.
2) disable asan in the generators.
The first option seems not feasible, as the compilation of the ldpctest
does not succeed (in fact, for a reason I don't understand, ninja seems
to trigger linking and is then blocked indefinitely). Therefore, this
commit simple disables asan in the generators.
The files openair1/PHY/CODING/nrLDPC_cnProc.h and
openair1/PHY/CODING/nrLDPC_decoder/nrLDPC_cnProc.h have the same name,
and also the same include guard/define. This commit renames the define
in the latter to prevent this name clash.
Physical simulators are built using address sanitizer. During the build
process, we use generators that are themselves built during the build
process. They are also asan, and we therefore need to give the
SYS_PTRACE capability during the build process to have them run
correctly.
Robert Schmidt