nr_fill_indication() fills CRC and RX requests, to be sent from L1 to
L2. Before this commit, nr_fill_indication() is called when
- we finally decoded LDPC (or not)
- low signal on PUSCH
This might happen at the same time, i.e., nr_fill_indication() might try
to fill both CRC and RX, in the same list, from different threads. This
can lead to this assertion:
Assertion (crc->rnti == rx->rnti) failed!
In handle_nr_ulsch() ../../../openair2/NR_PHY_INTERFACE/NR_IF_Module.c:190
mis-match between CRC RNTI e071 and RX RNTI 5e3b
e.g., e071 is low energy, and while 5e3b related message are filled in
one thread, e071 is being put into the message structure as well.
At least that is my understanding. I could not actually reproduce this
assertion; to be seen if it still happens.
Do not assert if the list of phy statistics is full; instead return
NULL.
In most places, this is handled appropriately (we do not store anything
if we can't). In nr_decode_pucch0(), the return value of get_phy_stats()
is used a little bit all over the function, and a constant
if (uci_stats)
uci_stats->variable++;
would look ugly, so make uci_stats point to the stack if we cannot store
in the phy_stats.
the uci_stats variable (pointing to per-UE statistics for UCI) is used
in function nr_decode_pucch0() to store information which is important
beyond statistics, e.g., uci_stats->pucch0_thres as a threshold (which
comes from somewhere else)
This commit refactors to use original or intermediate variables to not
mix uci_stats usage with actual processing.
Also, group uci_stats a bit more.
UBsan fails with errors without this commit. The commit changes:
- avoid two possible integer overflows in nr_dlsch_channel_level() and nr_dlsch_channel_level_median()
- nr_dlsch_channel_level_median() is simplified
- use MAX_ANT in avg to handle more than 4x4 config
- deleted: ../../../executables/rfsim.c
- removed unused command line options to set SINR and SNR
- removed unused functions to get SINR and SNR
- removed unused RFSIM struct members
PUSCH decoding is started multiple times in nr_ulsch_procedures() for
each ULSCH segment, and stopped once inside
phy_procedures_gNB_uespec_RX(). Not only is it not clear if that works
with the API, it also might measure more than just PUSCH decoding.
Remove the PUSCH decoding, it is misleading. Check L1 RX processing
instead.
This commit introduces a separate thread (l1_tx_thread) that processes
gNB DL/TX slots. It receives a message from the ru_thread when a new
slot started, and starts processing.
The DL part of the scheduler is run in the l1_tx_thread. Therefore,
only call UL indication (for scheduler) in UL slots. The UL indication
previously triggered the DL scheduler -- hence, it had to be called in
every slot. Now, since the DL scheduler is moved into the DL thread, we
don't have to call the scheduler in every slot anymore.
The reorder thread is removed, as reordering with this scheme is not
necessary anymore.
The main advantage of this version is that the TX can be scheduled
earlier (sl_ahead). Further, there can no longer be race conditions in
the TX path, since the scheduler/L2, TX L1 processing and the RU tx
function are all execute in the same thread (L1_tX_thread).
Also, the scheduler prepares PRACH/PUSCH/PUCCH FAPI PDUs. As of this
commit, the scheduler runs in the TX processing chain, whereas these
PDUs are destined for RX processing. Thus, too avoid data races,
instead of triggering the RX processing at the same time as TX
processing in the RU thread, this commit changes the code to trigger RX
processing after the scheduler has been run in TX processing to avoid
data races.
Finally, we synchronize RU and L1 threads. This is important for
rfsimulator, since unlike most radios, rfsimulator can run slower or
faster, depending on I/O and CPU (it does not have a stable "tick").
Co-authored-by: kiran <saikiran@iitj.ac.in>
The UL_INFO_mutex is used in only one place, and therefore useless.
Also, the scheduler uses a lock internally to prevent concurrent access.
Hence, the UL_INFO_mutex is not needed.
Remove kill_gNB_proc(), as it is now dead code.
- Rate matching performed by card
- Due to rate matching on card, small TBs are encoded correctly, so the
limitation of encoding small TBs in CPU can be lifted (all TBs encoded
on T2), see a later commits
francescomani