A memcpy() introduced to optimize for the case of no precoding corrupted RBs which are not allocated to PDSCH in the common txdataF array.
So instead fall back to memcpy() for each PDSCH RB.

With the merge of fix_NR_DLUL_PF, gNB_dlsch_ulsch_scheduler() has no
nb_periods_per_frame variable. There are two changes here:

1) calculate this locally for initializing tdd_beam_association

2) do not pass it into schedule_nr_mib(). Further, use the following:
   nb_periods_per_frame = slots_per_frame / slots_per_period
   slots_per_frame / nb_periods_per_frame =
     = slots_per_frame / (slots_per_frame / slots_per_period)
     = slots_per_period

Merge branch 'develop' of https://gitlab.eurecom.fr/oai/openairinterface5g into NR_FAPI_beamindex_SSB_RO

Ci new docker pipeline

See merge request oai/openairinterface5g!1095

Signed-off-by: Raphael Defosseux <raphael.defosseux@eurecom.fr>

In case of false positive PDSCH CRC check, a random MAC PDU is processed.
Currently pdu_len is unsigned 16 Bit, so we did break out of the processing loop only if we _exactly_ hit a residual pdu_len of 0, or we hit a LCID of DL_SCH_LCID_PADDING.
If we didn't hit either of these, pdu_len wrapped to a very positive value, and we continued processing the PDU even beyond its bounds!
This behavior was observed with SNR 1 dB and MCS 9.

So better make pdu_len signed and break out of the loop after processing at max. pdu_len Bytes!

Signed-off-by: Raphael Defosseux <raphael.defosseux@eurecom.fr>

Signed-off-by: Raphael Defosseux <raphael.defosseux@eurecom.fr>

Signed-off-by: Raphael Defosseux <raphael.defosseux@eurecom.fr>

issues found by valgrind

Signed-off-by: Raphael Defosseux <raphael.defosseux@eurecom.fr>