/* This header file defines standard types and provides
 * function prototypes used in the OpenMP to CIVLC transformation.
 */

#ifdef __CIVLC_OMP__
#else
#define __CIVLC_OMP__

#include<civlc.cvh>
#include<pointer.cvh>
#include<seq.cvh>
#include<concurrency.cvh>

/* *********************** Types *********************** */

typedef enum __OMP_var_status_{
  EMPTY, // local is empty
  FULL, // local is occupied, no writes to it have been made
  MODIFIED // local is occupied, writes have been made to it
} $omp_var_status;

typedef enum __OMP_loop_policy_{
  ROUND_ROBIN, // partition by round robin order
  RANDOM, // a random partition
  ALL // all possible partitions
} $omp_loop_policy;

/* global shared object, containing a reference to a shared variable. 
 * A handle type.
 */
typedef struct OMP_gshared {
  _Bool init[]; // which threads have joined
  void * original; // pointer to original variable
} * $omp_gshared;

/* local view of a shared object, belonging to a single
 * thread, with reference to the global object, and
 * a local copy and a status of the shared object. 
 * The type of the status variable is obtained from
 * the type of the original variable by replacing
 * all leaf nodes in the type tree with "int".
 * A handle type. 
 */
typedef struct OMP_shared {
  $omp_gshared gshared;
  // The thread id
  int tid;
  // Pointer to the local copy of the shared variable.
  // This provides the thread's "private view" of the variable.
  void * local;
  // Pointer to the local status variable
  void * status;
} * $omp_shared;

/* the worksharing information that a thread needs for executing 
 * a worksharing region. It contains the kind of the worksharing 
 * region, the location of the region, the status of the region 
 * and the subdomain (iterations/sections/task assigned to the thread).
 */
typedef struct OMP_work_record {
  int kind; // loop, barrier, sections, or single
  int location; // location in model of construct
  _Bool arrived; // has this thread arrived yet?
  //$domain loop_domain; // full loop domain; null if not loop
  $domain subdomain; // tasks this thread must do
  // reduction operation?
} $omp_work_record;

/* global team object, represents a team of threads executing
 * in a parallel region. A handle type. This is where all the
 * state needed to correctly execute a parallel region will
 * be stored. This includes a global barrier, and a worksharing 
 * queue for every thread.
 */
typedef struct OMP_gteam {
  /* scope in which data is allocated in heap */
  $scope scope;
  /* number of threads in team */
  int nthreads;
  /* which threads have joined this gteam */
  _Bool init[];
  /* work queues.  Length nthreads. For each thread,
   * a FIFO queue of work records */
  $omp_work_record work[][];
  /* the shared object data. */
  $omp_gshared shared[];
  $gbarrier gbarrier;
} * $omp_gteam;

/*
 * local object belonging to a single thread and referencing the global team object.  
 * A handle type. It also includes the local views of all shared data and a local barrier.
 */
typedef struct OMP_team {
  $omp_gteam gteam;
  $scope scope;
  int tid;
  $omp_shared shared[];
  $barrier barrier;
} * $omp_team;



/* *********************** Functions *********************** */

/* creates new global team object, allocating object in heap
 * in specified scope. Number of threads that will be in the
 * team is nthreads.
 */
$omp_gteam $omp_gteam_create($scope scope, int nthreads) {
  $omp_work_record empty[0];
  $omp_gteam result = ($omp_gteam)$malloc(scope, sizeof(struct OMP_gteam));
  _Bool f = $false;

  result->scope = scope;
  result->nthreads = nthreads;
  $seq_init(&result->init, nthreads, &f);
  $seq_init(&result->work, nthreads, &empty);
  $seq_init(&result->shared, 0, NULL);
  result->gbarrier = $gbarrier_create(scope, nthreads);
  return result;
}

/* destroys the global team object. All shared objects
 * associated to the team must have been destroyed before
 * calling this function.
 */
void $omp_gteam_destroy($omp_gteam gteam) {
  int nthreads = gteam->nthreads;

  $assert $seq_length(&gteam->shared)==0 :
          "shared objects must be deallocated before freeing gteam";
  for (int i=0; i<nthreads; i++) {
    int numRecords = $seq_length(&gteam->work[i]);

    $assert numRecords == 0 :
            "Thread %d still has %d queued worksharing events",
            i, numRecords;
  }
  $free(gteam->gbarrier);
  $free(gteam);
}


/* creates new local team object for a specific thread. */
$omp_team $omp_team_create($scope scope, $omp_gteam gteam, int tid) {
  $omp_team result = ($omp_team)$malloc(scope, sizeof(struct OMP_team));

  $assert !gteam->init[tid] :
          "Thread %d has already joined gteam",
          tid;
  gteam->init[tid] = $true;
  result->gteam = gteam;
  result->scope = scope;
  result->tid = tid;
  $seq_init(result->shared, 0, NULL);
  result->barrier = $barrier_create(scope, gteam->gbarrier, tid);
  return result;
}

/* destroys the local team object */
void $omp_team_destroy($omp_team team) {
  int numShared = $seq_length(team->shared);

  for(int i = 0; i < numShared; i++){
    $free(team->shared[i]);
  }
  $free(team->barrier);
  $free(team);
}

/* creates new global shared object, associated to the given
 * global team. A pointer to the shared variable that this
 * object corresponds to is given.
 */
$omp_gshared $omp_gshared_create($omp_gteam gteam,
                                 void *original) {
  $omp_gshared result =
    ($omp_gshared)$malloc(gteam->scope, sizeof(struct OMP_gshared));
  _Bool f = $false;
 
  $seq_init(result->init, gteam->nthreads, &f);
  result->original = original;
  //result->status = status;
  return result;
}

/* destroys the global shared object, copying the content
 * to the original variable.
 */
void $omp_gshared_destroy($omp_gshared gshared) {
  $free(gshared);
}

/* creates a local shared object, returning handle to it.  
 * The local copy of the shared object is initialized
 * by copying the values from the original variable referenced to
 * by the gshared object. The created shared object is appended
 * to the shared queue of the $omp_team object. */
$omp_shared $omp_shared_create($omp_team team,
                               $omp_gshared gshared, void *local, void *status) {
  $omp_shared result =
    ($omp_shared)$malloc(team->scope, sizeof(struct OMP_shared));

  $assert !gshared->init[team->tid] :
          "Thread %d has already created its local copy for %p.\n", 
          team->tid, gshared;
  result->gshared = gshared;
  result->tid = team->tid;
  result->local = local;
  result->status = status;
  $seq_append(&team->shared, &result, 1);
  return result;
}

/* destroys the local shared object */
void $omp_shared_destroy($omp_shared shared) {
  $free(shared);
}

/* called by a thread to read a shared object.
 * ref is a pointer into the local copy of the shared variable.
 * The result of the read is stored in the
 * memory unit pointed to by result.
 * assumes ref is a pointer to a scalar.
 */
void $omp_read($omp_shared shared, void *result, void *ref) {
  int tid = shared->tid;
  int *status_ref = (int*)$translate_ptr(ref, shared->status);
  int status = *status_ref;

  if(status == EMPTY){
    void *local = $translate_ptr(ref, shared->local);
 
    $copy(local, ref); // copy shared to local
    *status_ref = FULL; // set status to FULL
  }
  // read local
  $copy(result, ref);
}

/* called by a thread to write to the shared object.
 * ref is a pointer into the local copy of the shared variable.
 * The value to be written is taken
 * from the memory unit pointed to by value.
 * assumes ref is a pointer to a scalar.
 */
void $omp_write($omp_shared shared, void *ref, void *value) {
  int tid = shared->tid;
  int *status_ref = (int*)$translate_ptr(ref, shared->status);
  
  $copy(ref, value);
  *status_ref = MODIFIED;
}

/* applies the associative operator
 * specified by op to the local copy and the corresponding shared copy, 
 * and writes the result back to the shared copy.
 * This happens in one atomic step. Example: you can
 * use this to add some value to a shared variable,
 * using CIVL_SUM for op.
 * assumes local is a pointer to a scalar.
 */
void $omp_apply_assoc($omp_shared shared,
                      $operation op,
                      void *local){
  $atomic{
    void *shared_ref = $translate_ptr(local, shared->gshared->original);

    $apply(shared_ref, op, local, shared_ref);
  }
}

/* performs an OpenMP flush operation on the shared object
 */
void $omp_flush($omp_shared shared, void *ref) {
  // need to drill down into all leaf nodes of the object
  // being flushed...
  // also, it should be ok to flush a memory unit if you are not
  // the owner but you also have no reads or writes to that variable

  $omp_gshared gshared = shared->gshared;
  int tid = shared->tid;
  void *refs[];
  int numRefs;
  
  // get all leaf node pointers
  $leaf_node_ptrs(&refs, ref);
  numRefs = $seq_length(&refs);
  for(int i = 0; i < numRefs; i++){
    void *leaf = refs[i];
    int *leaf_status = (int *)$translate_ptr(leaf, shared->status);
    void *leaf_local = (int *)$translate_ptr(leaf, shared->local);
    void *leaf_shared = (int *)$translate_ptr(leaf, gshared->original);
    
    switch(*leaf_status){
      case EMPTY:
        break;
      case MODIFIED:
        $copy(leaf_shared, leaf_local);
      case FULL:
        *leaf_status = EMPTY;
        $set_default(leaf_local);
        break;
    }
  }
}

/* performs an OpenMP flush operation on all shared
 * objects. This is the default in OpenMP if no argument
 * is specified for a flush construct.
 */
void $omp_flush_all($omp_team team) {
   int num_shared = $seq_length(team->shared);

   for (int i=0; i<num_shared; i++) {
     $omp_shared shared = team->shared[i];
     
     $omp_flush(shared, shared->local);
   }
}

/* performs a barrier only. Note however that usually
 * (always?) a barrier is accompanied by a flush-all,
 * so $omp_barrier_and_flush should be used instead.
 */
void $omp_barrier($omp_team team){
  $barrier_call(team->barrier);
}

/* combines a barrier and a flush on all shared objects
 * owned by the team. Implicit in many OpenMP worksharing
 * constructs.
 */
void $omp_barrier_and_flush($omp_team team) {
  // this is a collective operation: all members of team call
  $barrier_call(team->barrier);
  $omp_flush_all(team);
  $barrier_call(team->barrier);
}

// To be continued: needs a way to specify non-rectangular 
// domains.
/* called by a thread when it reaches an omp for loop,
 * this function returns the subset of the loop domain
 * specifying the iterations that this thread will execute,
 * according to the given policy.
 * The dimension of the domain returned equals the
 * dimension of the given domain loop_dom.
 */
$domain $omp_arrive_loop($omp_team, $domain loop_dom, int policy);

/* called by a thread when it reaches an omp sections
 * construct, this function returns the subset of the
 * integers 0..numSections-1 specifying the indexes of
 * the sections that this thread will execute. The sections
 * are numbered from 0 in increasing order.
 */
$domain(1) $omp_arrive_sections($omp_team team, int numSections){
  int low = team->tid, high = numSections - 1, step = team->gteam->nthreads;
  $range range = low .. high # step;
  $domain(1) dom = ($domain(1)) {range};
  
  return dom;
}

/* called by a thread when it reaches on omp single
 * construct, returns the thread ID of the thread that
 * will execute the single construct.
 */
int $omp_arrive_single($omp_team team){
  return 0;
}

#endif