This commit fixes a bug caused by global variables in the gNB_MAC_INST
structure used to transfer MAC scheduler's results to the PHY layer.
When using several threads in the PHY layer, it may happen that the
MAC scheduler is called a second time by the second PHY thread for the
next slot before the content of the global variables in the gNB_MAC_INST
structure used to store the result of the MAC scheduler for the current
slot have been consumed by the first thread, leading to unpredictable
results (all of them certainly wrong).
The solution is to introduce a small memory management module for the
interface between the PHY threads and the MAC scheduler. The MAC
scheduler now fills the data structures passed to it, and not a
global variable anymore. The PHY layer manages the structures and
ensures that no data is overwritten before being fully consumed by
the various threads of the PHY layer.
Some care has to be taken because several PHY threads will access the
same structures at the same time for a given slot (in read mode, so
it's okay). A mechanism of reference counting has thus been introduced.
Only one CC is supported. This restriction should not be too hard
to remove if/when needed. (The code before this commit does not seem
to work for more than one CC anyway.)
integration_2023_w13
See merge request oai/openairinterface5g!2037
* !1945 Free RNTI related structures at PHY after UE disconnection
* !2011 Creation of new pipeline LTE-TDD-2x2-container
* !1854 Handle reconfigurationWithSync
* !2019 bugfix: LDPC decoding in the gNB: clear d properly
* !2038 Fix DL 2x2 errors
* !2039 Adding test-case in XML to reset USRP in LTE-TDD pipeline
* !2008 Implement NR-UE Deregistration procedure
* !2023 Rework PDCP OAI API to allow custom SRB data forwarding (e.g., in F1)
* !2015 Update documentation: CI, general README.md, overview of doc README, various improvements
The GTP module is common for 4G and 5G. It hardcodes the functions to
call when creating S1U and NGU tunnels. The S1U case is both applicable
for LTE and NR (NSA), which until now was not a problem as the PDCP data
forwarding functions where called pdcp_data_req. The issue is that in this
branch, the NR PDCP functions have changed to nr_pdcp_data_req_drb().
When setting up a GTP tunnel, we now pass the PDCP data forwarding
functions into GTP, since the caller knows whether it is in 4G or 5G,
but GTP itself does not.
In the UE, we wish to handle ^C in the UE to send a message before
terminating, so we allow custom wait handlers to implement such
functionality. Use itti_wait_tasks_unblock() to unblock
itti_wait_tasks_end().
Also, implement itti_wait_tasks_end() using semaphores, as signals are
not async-signal-safe (and therefore should not be used in a signal
handler).
nr_pdcp_entity_recv_sdu() processes PDCP SDUs, and forwards the PDU at the
end of the function, while having the PDCP lock. This can lead to
deadlocks, e.g., when another thread goes through RLC and tries to
access the PDCP while the PDCP tries to lock the RLC.
This commit changes the logic: instead of having recv_sdu() forward the
PDU directly, it fills a buffer with the PDU (thus the name change to
process_sdu()), and returns. We then unlock the PDCP, before forwarding
the PDU, thus eliminating potential deadlocks.
Creates function that have nr_pdcp_ prepended to their name, and change
the signature of some functions for the 5G PDCP:
- nr_pdcp_layer_init()
- nr_pdcp_data_req()
- nr_pdcp_remove_UE()
- nr_pdcp_config_set_security()
- nr_pdcp_remove_UE()
- Make separate functions for nr_pdcp_data_req_srb/drb()
Cedric Roux