This class is based on a template that receives as argument the
rlc_am_*_tx/rx entities, so that those are different for LTE and NR.
Moved code from rlc_am_lte/nr entities so that they use the new base class.
For Ubuntu 18.04: Linking the static srsran_rf to libuhd
also requires to link Boost's "system" library.
Newer systems do not require this special handling.
relevant for the UE when it receives 2nd reconfiguration it could
be that the bearers are added again. in this case the handling is fine
and everything will work as expected. No reason to log error.
add default in switch instead of using the the END state directly
that is checked before entering the switch. Coverity correctly
reports this as dead code.
fix printf format string on 32bit arches like ARMv7
FAILED: lib/src/phy/ue/CMakeFiles/srsran_ue.dir/ue_sync.c.o
/usr/bin/gcc -DASSERTS_ENABLED -DENABLE_TIMEPROF -DENABLE_TTCN3 -DHAVE_BACKWARD -DHAVE_MBEDTLS -DHAVE_PCSC -I/usr/include/PCSC -I/usr/include/SoapySDR -Ilib/include -I../lib/include -Wno-unused-but-set-variable -Wall -Wno-comment -Wno-write-strings -Wno-unused-result -Wformat -Wmissing-field-initializers -Wtype-limits -std=c99 -fno-strict-aliasing -D_GNU_SOURCE -O3 -fno-trapping-math -fno-math-errno -DBUILD_TYPE_RELEASE -DIS_ARM -DHAVE_NEON -mfloat-abi=hard -mfpu=neon -fvisibility=hidden -fdiagnostics-color=always -Werror --param large-function-growth=1600 -O3 -DNDEBUG -fPIC -MD -MT lib/src/phy/ue/CMakeFiles/srsran_ue.dir/ue_sync.c.o -MF lib/src/phy/ue/CMakeFiles/srsran_ue.dir/ue_sync.c.o.d -o lib/src/phy/ue/CMakeFiles/srsran_ue.dir/ue_sync.c.o -c ../lib/src/phy/ue/ue_sync.c
In file included from ../lib/include/srsran/srsran.h:31,
from ../lib/src/phy/ue/ue_sync.c:13:
../lib/src/phy/ue/ue_sync.c: In function ‘srsran_ue_sync_set_tti_from_timestamp’:
../lib/include/srsran/phy/utils/debug.h:59:23: error: format ‘%lu’ expects argument of type ‘long unsigned int’, but argument 3 has type ‘uint64_t’ {aka ‘long long unsigned int’} [-Werror=format=]
fprintf(stdout, "[DEBUG]: " _fmt "\n", ##__VA_ARGS__); \
^~~~~~~~~~~
../lib/src/phy/ue/ue_sync.c:1044:3: note: in expansion of macro ‘DEBUG’
DEBUG("rx time with 3gpp base in ms %lu\n", time_3gpp_msecs);
^~~~~
../lib/src/phy/ue/ue_sync.c:1044:41: note: format string is defined here
DEBUG("rx time with 3gpp base in ms %lu\n", time_3gpp_msecs);
~~^
%llu
cc1: all warnings being treated as errors
CID 369560 Missing unlock, in dft_fftw.c
CID 363710 Logically dead code, in resampler.c
CID 55097 Result is not floating point, in chest_test_dl.c
CID 351020 Result is not floating point, in chest_nbiot_test_dl.c
Fix several Coverity issues
This commit adrresses the following code issues found by Coverity:
CID 339785 Division by zero, in fading_channel_test.c
CID 355272 Division by zero, in awgn_channel_test.c
CID 355277 NULL ptr dereference, in awgn_channel_test.c
CID 359663 NULL ptr dereference, in delay_channel_test.c
CID 369544 NULL ptr dereference, in chest_dl_nbiot.c
CID 373326 Resource Leak, in dft_fftw.c
CID 373329 Resource Leak, in dft_fftw.c
CID 372878 Division by zero, in sliv_test.c
CID 372871 Division by zero, in dmrs_pdcch.c
CID 370622 Negative loop bound, in csr_rs.c
CID 370624 Negative loop bound, in csr_rs.c
CID 370626 Negative loop bound, in csr_rs.c
CID's 369568, 369594 NULL ptr dereference, in ch_awgn.c
CID 369540 Logically dead code, in refsignal_dl.c
CID 369608 Logically dead code, in refsignal_ul.c
CIDs 366291, 366296, 366297 Out-of-bounds access, in zc_sequence.c
CID 372209 Division by zero, in cqi.c
CID 370992 Uninitialized pointer read, in pdcch_test.c
CID 373334 Integer overflow, in ue_sync.h
CID 370993, 370995 Undefined division, in pdcch_test.c
CID 370994 Undefined division, in ssb_decode_test.c
CIDs 353368 353364 353365 359673 353366 353367
Explicit null dereferenced, in psss_file_test.c
CID 371865 Unchecked return value, in rf_uhd_imp.cc
CID 363810 Undefined division, in ldpc_rm_chain_test.c
CID 372209, 372211, 372213, 372216 Undefined modulo, in cqi.c
CID 339834 Array compared against 0, in chest_dl.c
CID 369589 Out of bounds access, in dmrs_pucch.c
CID 371681 Out of bounds access, in ue_sync_nr_test.c
CIDs 370761, 370825 Copy-paste error, in ssb.c
CID 369599 out of bounds read, in dmrs_pdcch_test.c
CID 363795 out of bounds read, in ldpc_dec_c_avx2_flood.c
CID 363805 out of bounds read, in ldpc_dec_c_avx2long_flood.c
CID 363821 out of bounds read, in ldpc_dec_c_flood.c
until the correct gain settings are documented and potential
mis-configuration are handled. It seems that the default
Tx/rx gains values are not working with this function.
* Make filename const in filesink
* Sine generation returns the next phase
* Avoid malloc/free in radio class
* Implement Tx gain in ZMQ
* Initial ratio RT gain test
* UHD: use timed Tx gain commands to align changes to subframes
* Minor improvement in test_radio_rt_gain
* Fix compilation
* Check RF gain thread id before joining
* Remove redundant zero initialization.
Co-authored-by: Fabian Eckermann <fabian@srs.io>
* Fix a race condition when accessing the NR PHY cfg by the RRC and phy workers.
Rework how the phy cfg is handled, now workers have their own copy that gets updated after a reconfig moving it out of the state class.
* Default initialize sf_len member in sf_worker for consistency.
* Asynchronous NR PHY configuration
* Fix compilation
* Corrected method override and fix unitialised value
* Added carrier equal comparison to avoid aligment byte padding comparison
Co-authored-by: faluco <borja.ferrer@softwareradiosystems.com>
This was done to avoid integrity issues, when the UE's RRC erroneously sent
measurement reports while the re-establishment was already in progress.
As errously sending PDCP PDUs on DRBs can cause issues as well, this was
disabled too.
Move emergency handlers header file into the support folder.
Refactored signal handling:
- Remove the dependency with the running static variable in the header file.
- Move implementations down to cc files.
- Allow specifying a new signal handler that will be used to stop the applications.
- Move signal handling files to support.
since absolute freq point a members have been removed from the srslte_nr_carrier_t
struct (they are now actual frequencies) the abs_point_a values are now in
the RRC cell structs
this is to avoid invalid configs that later on cause segfaults, etc.
e.g.:
0 0x7f397adf64a3 (/lib/x86_64-linux-gnu/libc.so.6+0xbb4a3)
1 0x7f397cc8c6ce (/usr/lib/x86_64-linux-gnu/libasan.so.4+0x796ce)
2 0x55b921d61707 in srsran_vec_cf_copy /mnt/data/jenkins/workspace/srslte_ogt_trial_builder_x86-ubuntu1804-asan/srsLTE/lib/src/phy/utils/vector.c:226
3 0x55b921d3b1d4 in ssb_demodulate /mnt/data/jenkins/workspace/srslte_ogt_trial_builder_x86-ubuntu1804-asan/srsLTE/lib/src/phy/sync/ssb.c:632
4 0x55b921d3d8ce in srsran_ssb_csi_measure /mnt/data/jenkins/workspace/srslte_ogt_trial_builder_x86-ubuntu1804-asan/srsLTE/lib/src/phy/sync/ssb.c:905
5 0x55b9215dc20f in srsue::nr::cc_worker::measure_csi() /mnt/data/jenkins/workspace/srslte_ogt_trial_builder_x86-ubuntu1804-asan/srsLTE/srsue/src/phy/nr/cc_worker.cc:373
6 0x55b9215dd216 in srsue::nr::cc_worker::work_dl() /mnt/data/jenkins/workspace/srslte_ogt_trial_builder_x86-ubuntu1804-asan/srsLTE/srsue/src/phy/nr/cc_worker.cc:483
7 0x55b9215f9c69 in srsue::nr::sf_worker::work_imp() /mnt/data/jenkins/workspace/srslte_ogt_trial_builder_x86-ubuntu1804-asan/srsLTE/srsue/src/phy/nr/sf_worker.cc:78
8 0x55b921c47cc6 in srsran::thread_pool::worker::run_thread() /mnt/data/jenkins/workspace/srslte_ogt_trial_builder_x86-ubuntu1804-asan/srsLTE/lib/src/common/thread_pool.cc:48
9 0x55b92152ea0b in srsran:🧵:thread_function_entry(void*) /mnt/data/jenkins/workspace/srslte_ogt_trial_builder_x86-ubuntu1804-asan/srsLTE/lib/include/srsran/common/threads.h:103
10 0x7f397c9fb6da in start_thread (/lib/x86_64-linux-gnu/libpthread.so.0+0x76da)
11 0x7f397ae5c71e in __clone (/lib/x86_64-linux-gnu/libc.so.6+0x12171e)
under some circumstances it could happen that the RLC is configured
when SDUs are already being written to the queue. The resize
operation of the underlying container would fail in this case.
Make sure to empty the queue before doing the resize.
* Improve reliability of timestamp to tti conversion
Difftime is not reliable as it might use 32 bit calculation, depending on the system. This leads to wrong frame numbers and subframe indices.
I encountered this Issue when testing the pssch_ue exampel on my system and the conversion from timestamp to frame number and sf idx was wrong.
* Improve GNSS Sync
Added loop to check for GNSS alignment while syncing.
If the received GNSS signal is weak, synchronization errors might occur while syncing.
* decouple EUTRA and NR stack classes
* implement dummy X2 interface with control and data plane methods
* implement eNB time source interface that PHY calls
we've had a few runs in the CI where opening the specified RF device
failed but the eNB/UE still continued to run, just picking the next available
run. This led to false-positive tests.
The policy should be that whenever the user specified a RF device to
be openend, and this device fails, the whole process should fail and
the application should exit.
The auto-detection mode is still available but only if no device name
is specified at all.
the Msg3 UL grant requires the TTI in which the RAR has been received
to calcualte the correct timing. There was a race between PHY and Stack
thread.
This patch circumvents the issue by removing a PHY state member that only holds
the RAR Rx timing. In the new interface the RA proc passes the Rx TTI
to the PHY again when setting the UL grant so the PHY can calculate the
correct timing without any state.
* the old srsran_cell_cfg_t is only carried for SIB params
* the SIB scheduling, etc needs to be ported and integrated into NR data structures
* disbable SIB test for the moment