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.)
After this commit, only the RNTI is used as an identifier in the MAC.
Further, it removes some module_id, but a lot of them remain (the goal
to remove all is still far from us)
This struct is used to store the uplink 5G nFAPI
messages between the proxy and the gNB MAC layer.
The original OAI implementation utilizes a similarly
named variable (UL_info) to store the uplink 5G nFAPI
messages between the PHY and MAC layer. This commit
will utilize the original UL_info variable prior to
calling the handle functions. These changes allow us
to make minimal changes to the nFAPI handle functions
and modifies the NR_UL_Indication parent function instead.
This requires some updates to the nFAPI code as well.
We also have declared the def_dci_rel15 extern variable
consistently throughout the code.
In this commit, the gNB is still crashing. However, it seems
that the NSA UE is sending the proper responses back, but there
some sort of incorrect management of the def_dci_pdu_rel15
variable. If we memset this array to 0, we can see incorrect
0 assignments to elements of the UCI indications. Hopefully
fixing this issue will result in less gNB crashes.
1. If matched rnti does not exist between crc_ind pdus and rx_ind puds, the ind or unmatched pdus will be requeued.
2. Sending crc_ind and rx_ind togetehr with crc_ind first in nrUE.
1. Changed hard coded preamble index(63) to soft coded.
2. Active UE will not be considered in gNB scheduging RA.
3. Added condition to process dci indication to handle multiple UEs.
4. Bug fix in mutex_lock (unlock to lock). It should be lock rather than unlock.
5. nr_ue_scheduler update not to entering into getting IP traffic to be transmitted in noS1 mode.
6. log update in checking the number of rx_ind or crc
How? Well, we had to properly allocate the
ul_config_req. We also had to update the
ue_get_rach function. There were several
places where the functionality would check
for a particular mode (ra, sa, phy_test, etc).
In some of these places we had to add NSA mode.
In the VNF, they start the scheduler after 3 sec,
this is sort of a hack, might want to revisit.
This commit also includes removal of printfs.
Cedric Roux