Commit 41de51b1 authored by Robert Schmidt's avatar Robert Schmidt

Format to dynamically handle UL SCHED

parent 5d1d2bcd
......@@ -1276,7 +1276,7 @@ typedef struct {
/**
* definition of a scheduling algorithm implementation used in the
* default scheduler
* default DL scheduler
*/
typedef struct {
char *name;
......@@ -1287,6 +1287,19 @@ typedef struct {
void *data;
} default_sched_dl_algo_t;
/**
* definition of a scheduling algorithm implementation used in the
* default UL scheduler
*/
typedef struct {
char *name;
void *(*setup)(void);
void (*unset)(void **);
int (*run)(
module_id_t, int, int, int, int, int, UE_list_t *, int, int, contig_rbs_t *, void *);
void *data;
} default_sched_ul_algo_t;
/*! \brief eNB common channels */
typedef struct {
int physCellId;
......@@ -1447,6 +1460,7 @@ typedef struct eNB_MAC_INST_s {
SCHEDULER_MODES scheduler_mode;
/// scheduling algorithm used in default scheduler
default_sched_dl_algo_t dl_algo;
default_sched_ul_algo_t ul_algo;
int32_t puSch10xSnr;
int32_t puCch10xSnr;
......
......@@ -133,6 +133,11 @@ void mac_top_init_eNB(void)
AssertFatal(d, "%s(): no scheduler algo '%s' found\n", __func__, s);
mac[i]->dl_algo = *(default_sched_dl_algo_t *) d;
mac[i]->dl_algo.data = mac[i]->dl_algo.setup();
s = "round_robin_ul";
d = dlsym(NULL, s);
AssertFatal(d, "%s(): no scheduler algo '%s' found\n", __func__, s);
mac[i]->ul_algo = *(default_sched_ul_algo_t *) d;
mac[i]->ul_algo.data = mac[i]->ul_algo.setup();
init_UE_info(&mac[i]->UE_info);
init_slice_info(&mac[i]->slice_info);
}
......
......@@ -818,17 +818,28 @@ int pp_find_rb_table_index(int approximate) {
return p + 1;
}
int g_start_ue_ul = -1;
int round_robin_ul(module_id_t Mod_id,
int CC_id,
int frame,
int subframe,
int sched_frame,
int sched_subframe,
UE_list_t *UE_list,
int max_num_ue,
int num_contig_rb,
contig_rbs_t *rbs) {
void *rr_ul_setup(void) {
void *data = malloc(sizeof(int));
*(int *) data = 0;
AssertFatal(data, "could not allocate data in %s()\n", __func__);
return data;
}
void rr_ul_unset(void **data) {
if (*data)
free(*data);
*data = NULL;
}
int rr_ul_run(module_id_t Mod_id,
int CC_id,
int frame,
int subframe,
int sched_frame,
int sched_subframe,
UE_list_t *UE_list,
int max_num_ue,
int num_contig_rb,
contig_rbs_t *rbs,
void *data) {
AssertFatal(num_contig_rb <= 2, "cannot handle more than two contiguous RB regions\n");
UE_info_t *UE_info = &RC.mac[Mod_id]->UE_info;
const int max_rb = num_contig_rb > 1 ? MAX(rbs[0].length, rbs[1].length) : rbs[0].length;
......@@ -968,9 +979,10 @@ int round_robin_ul(module_id_t Mod_id,
end = la > lb ? 2 : -1;
}
if (g_start_ue_ul == -1)
g_start_ue_ul = UE_list->head;
int sUE_id = g_start_ue_ul;
int *start_ue = data;
if (*start_ue == -1)
*start_ue = UE_list->head;
int sUE_id = *start_ue;
int rb_idx_given[MAX_MOBILES_PER_ENB];
memset(rb_idx_given, 0, sizeof(rb_idx_given));
......@@ -1049,10 +1061,17 @@ int round_robin_ul(module_id_t Mod_id,
}
/* just start with the next UE next time */
g_start_ue_ul = next_ue_list_looped(UE_list, g_start_ue_ul);
*start_ue = next_ue_list_looped(UE_list, *start_ue);
return rbs[0].length + (num_contig_rb > 1 ? rbs[1].length : 0);
}
default_sched_ul_algo_t round_robin_ul = {
.name = "round_robin_ul",
.setup = rr_ul_setup,
.unset = rr_ul_unset,
.run = rr_ul_run,
.data = NULL
};
void ulsch_scheduler_pre_processor(module_id_t Mod_id,
int CC_id,
......@@ -1060,9 +1079,10 @@ void ulsch_scheduler_pre_processor(module_id_t Mod_id,
sub_frame_t subframeP,
int sched_frameP,
unsigned char sched_subframeP) {
UE_info_t *UE_info = &RC.mac[Mod_id]->UE_info;
const int N_RB_UL = to_prb(RC.mac[Mod_id]->common_channels[CC_id].ul_Bandwidth);
COMMON_channels_t *cc = &RC.mac[Mod_id]->common_channels[CC_id];
eNB_MAC_INST *mac = RC.mac[Mod_id];
UE_info_t *UE_info = &mac->UE_info;
const int N_RB_UL = to_prb(mac->common_channels[CC_id].ul_Bandwidth);
COMMON_channels_t *cc = &mac->common_channels[CC_id];
UE_list_t UE_to_sched;
UE_to_sched.head = -1;
......@@ -1119,16 +1139,17 @@ void ulsch_scheduler_pre_processor(module_id_t Mod_id,
}
}
round_robin_ul(Mod_id,
CC_id,
frameP,
subframeP,
sched_frameP,
sched_subframeP,
&UE_to_sched,
4, // max_num_ue
n_contig,
rbs);
mac->ul_algo.run(Mod_id,
CC_id,
frameP,
subframeP,
sched_frameP,
sched_subframeP,
&UE_to_sched,
4, // max_num_ue
n_contig,
rbs,
mac->ul_algo.data);
// the following block is meant for validation of the pre-processor to check
// whether all UE allocations are non-overlapping and is not necessary for
......@@ -1147,7 +1168,7 @@ void ulsch_scheduler_pre_processor(module_id_t Mod_id,
continue;
print = 1;
uint8_t harq_pid = subframe2harqpid(&RC.mac[Mod_id]->common_channels[CC_id],
uint8_t harq_pid = subframe2harqpid(&mac->common_channels[CC_id],
sched_frameP, sched_subframeP);
LOG_D(MAC, "%4d.%d UE%d %d RBs (index %d) at start %d, pre MCS %d %s\n",
frameP,
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment