/* -*- Mode: C; c-basic-offset:4 ; indent-tabs-mode:nil ; -*- */
/*
* (C) 2007 by Argonne National Laboratory.
* See COPYRIGHT in top-level directory.
*/
#ifndef PMI2_H_INCLUDED
#define PMI2_H_INCLUDED
#ifndef USE_PMI2_API
#error This header file defines the PMI2 API, but PMI2 was not selected
#endif
#define PMI2_MAX_KEYLEN 64
#define PMI2_MAX_VALLEN 1024
#define PMI2_MAX_ATTRVALUE 1024
#define PMI2_ID_NULL -1
#if defined(__cplusplus)
extern "C" {
#endif
/*D
PMI2_CONSTANTS - PMI2 definitions
Error Codes:
+ PMI2_SUCCESS - operation completed successfully
. PMI2_FAIL - operation failed
. PMI2_ERR_NOMEM - input buffer not large enough
. PMI2_ERR_INIT - PMI not initialized
. PMI2_ERR_INVALID_ARG - invalid argument
. PMI2_ERR_INVALID_KEY - invalid key argument
. PMI2_ERR_INVALID_KEY_LENGTH - invalid key length argument
. PMI2_ERR_INVALID_VAL - invalid val argument
. PMI2_ERR_INVALID_VAL_LENGTH - invalid val length argument
. PMI2_ERR_INVALID_LENGTH - invalid length argument
. PMI2_ERR_INVALID_NUM_ARGS - invalid number of arguments
. PMI2_ERR_INVALID_ARGS - invalid args argument
. PMI2_ERR_INVALID_NUM_PARSED - invalid num_parsed length argument
. PMI2_ERR_INVALID_KEYVALP - invalid keyvalp argument
. PMI2_ERR_INVALID_SIZE - invalid size argument
- PMI2_ERR_OTHER - other unspecified error
D*/
#define PMI2_SUCCESS 0
#define PMI2_FAIL -1
#define PMI2_ERR_INIT 1
#define PMI2_ERR_NOMEM 2
#define PMI2_ERR_INVALID_ARG 3
#define PMI2_ERR_INVALID_KEY 4
#define PMI2_ERR_INVALID_KEY_LENGTH 5
#define PMI2_ERR_INVALID_VAL 6
#define PMI2_ERR_INVALID_VAL_LENGTH 7
#define PMI2_ERR_INVALID_LENGTH 8
#define PMI2_ERR_INVALID_NUM_ARGS 9
#define PMI2_ERR_INVALID_ARGS 10
#define PMI2_ERR_INVALID_NUM_PARSED 11
#define PMI2_ERR_INVALID_KEYVALP 12
#define PMI2_ERR_INVALID_SIZE 13
#define PMI2_ERR_OTHER 14
/* This is here to allow spawn multiple functions to compile. This
needs to be removed once those functions are fixed for pmi2 */
typedef struct PMI_keyval_t {
char *key;
char *val;
} PMI_keyval_t;
/*@
PMI2_Connect_comm_t - connection structure used when connecting to other jobs
Fields:
+ read - Read from a connection to the leader of the job to which
this process will be connecting. Returns 0 on success or an MPI
error code on failure.
. write - Write to a connection to the leader of the job to which
this process will be connecting. Returns 0 on success or an MPI
error code on failure.
. ctx - An anonymous pointer to data that may be used by the read
and write members.
- isMaster - Indicates which process is the "master"; may have the
values 1 (is the master), 0 (is not the master), or -1 (neither is
designated as the master). The two processes must agree on which
process is the master, or both must select -1 (neither is the
master).
Notes:
A typical implementation of these functions will use the read and
write calls on a pre-established file descriptor (fd) between the
two leading processes. This will be needed only if the PMI server
cannot access the KVS spaces of another job (this may happen, for
example, if each mpiexec creates the KVS spaces for the processes
that it manages).
@*/
typedef struct PMI2_Connect_comm {
int (*read) (void *buf, int maxlen, void *ctx);
int (*write) (const void *buf, int len, void *ctx);
void *ctx;
int isMaster;
} PMI2_Connect_comm_t;
struct MPIR_Info;
/*@
PMI2_Init - initialize the Process Manager Interface
Output Parameter:
+ spawned - spawned flag
. size - number of processes in the job
. rank - rank of this process in the job
- appnum - which executable is this on the mpiexec commandline
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
Initialize PMI for this process group. The value of spawned indicates whether
this process was created by 'PMI2_Spawn_multiple'. 'spawned' will be non-zero
iff this process group has a parent.
@*/
int PMI2_Init(int *spawned, int *size, int *rank, int *appnum);
/*@
PMI2_Finalize - finalize the Process Manager Interface
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
Finalize PMI for this job.
@*/
int PMI2_Finalize(void);
/*@
PMI2_Initialized - check if PMI has been initialized
Return values:
Non-zero if PMI2_Initialize has been called successfully, zero otherwise.
@*/
int PMI2_Initialized(void);
/*@
PMI2_Abort - abort the process group associated with this process
Input Parameters:
+ flag - non-zero if all processes in this job should abort, zero otherwise
- error_msg - error message to be printed
Return values:
If the abort succeeds this function will not return. Returns an MPI
error code otherwise.
@*/
int PMI2_Abort(int flag, const char msg[]);
/*@
PMI2_Spawn - spawn a new set of processes
Input Parameters:
+ count - count of commands
. cmds - array of command strings
. argcs - size of argv arrays for each command string
. argvs - array of argv arrays for each command string
. maxprocs - array of maximum processes to spawn for each command string
. info_keyval_sizes - array giving the number of elements in each of the
'info_keyval_vectors'
. info_keyval_vectors - array of keyval vector arrays
. preput_keyval_size - Number of elements in 'preput_keyval_vector'
. preput_keyval_vector - array of keyvals to be pre-put in the spawned keyval space
- jobIdSize - size of the buffer provided in jobId
Output Parameter:
+ jobId - job id of the spawned processes
- errors - array of errors for each command
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
This function spawns a set of processes into a new job. The 'count'
field refers to the size of the array parameters - 'cmd', 'argvs', 'maxprocs',
'info_keyval_sizes' and 'info_keyval_vectors'. The 'preput_keyval_size' refers
to the size of the 'preput_keyval_vector' array. The 'preput_keyval_vector'
contains keyval pairs that will be put in the keyval space of the newly
created job before the processes are started. The 'maxprocs' array
specifies the desired number of processes to create for each 'cmd' string.
The actual number of processes may be less than the numbers specified in
maxprocs. The acceptable number of processes spawned may be controlled by
``soft'' keyvals in the info arrays. The ``soft'' option is specified by
mpiexec in the MPI-2 standard. Environment variables may be passed to the
spawned processes through PMI implementation specific 'info_keyval' parameters.
@*/
int PMI2_Job_Spawn(int count, const char *cmds[],
int argcs[], const char **argvs[],
const int maxprocs[],
const int info_keyval_sizes[],
const struct MPIR_Info *info_keyval_vectors[],
int preput_keyval_size,
const struct MPIR_Info *preput_keyval_vector[],
char jobId[], int jobIdSize, int errors[]);
/*@
PMI2_Job_GetId - get job id of this job
Input parameters:
. jobid_size - size of buffer provided in jobid
Output parameters:
. jobid - the job id of this job
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
@*/
int PMI2_Job_GetId(char jobid[], int jobid_size);
/*@
PMI2_Job_Connect - connect to the parallel job with ID jobid
Input parameters:
. jobid - job id of the job to connect to
Output parameters:
. conn - connection structure used to exteblish communication with
the remote job
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
This just "registers" the other parallel job as part of a parallel
program, and is used in the PMI2_KVS_xxx routines (see below). This
is not a collective call and establishes a connection between all
processes that are connected to the calling processes (on the one
side) and that are connected to the named jobId on the other
side. Processes that are already connected may call this routine.
@*/
int PMI2_Job_Connect(const char jobid[], PMI2_Connect_comm_t * conn);
/*@
PMI2_Job_Disconnect - disconnects from the job with ID jobid
Input parameters:
. jobid - job id of the job to connect to
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
@*/
int PMI2_Job_Disconnect(const char jobid[]);
/*@
PMI2_KVS_Put - put a key/value pair in the keyval space for this job
Input Parameters:
+ key - key
- value - value
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
If multiple PMI2_KVS_Put calls are made with the same key between
calls to PMI2_KVS_Fence, the behavior is undefined. That is, the
value returned by PMI2_KVS_Get for that key after the PMI2_KVS_Fence
is not defined.
@*/
int PMI2_KVS_Put(const char key[], const char value[]);
/*@
PMI2_KVS_Fence - commit all PMI2_KVS_Put calls made before this fence
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
This is a collective call across the job. It has semantics that are
similar to those for MPI_Win_fence and hence is most easily
implemented as a barrier across all of the processes in the job.
Specifically, all PMI2_KVS_Put operations performed by any process in
the same job must be visible to all processes (by using PMI2_KVS_Get)
after PMI2_KVS_Fence completes. However, a PMI implementation could
make this a lazy operation by not waiting for all processes to enter
their corresponding PMI2_KVS_Fence until some process issues a
PMI2_KVS_Get. This might be appropriate for some wide-area
implementations.
@*/
int PMI2_KVS_Fence(void);
/*@
PMI2_KVS_Get - returns the value associated with key in the key-value
space associated with the job ID jobid
Input Parameters:
+ jobid - the job id identifying the key-value space in which to look
for key. If jobid is NULL, look in the key-value space of this job.
. src_pmi_id - the pmi id of the process which put this keypair. This
is just a hint to the server. PMI2_ID_NULL should be passed if no
hint is provided.
. key - key
- maxvalue - size of the buffer provided in value
Output Parameters:
+ value - value associated with key
- vallen - length of the returned value, or, if the length is longer
than maxvalue, the negative of the required length is returned
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
@*/
int PMI2_KVS_Get(const char *jobid, int src_pmi_id, const char key[], char value[],
int maxvalue, int *vallen);
/*@
PMI2_Info_GetNodeAttr - returns the value of the attribute associated
with this node
Input Parameters:
+ name - name of the node attribute
. valuelen - size of the buffer provided in value
- waitfor - if non-zero, the function will not return until the
attribute is available
Output Parameters:
+ value - value of the attribute
- found - non-zero indicates that the attribute was found
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
This provides a way, when combined with PMI2_Info_PutNodeAttr, for
processes on the same node to share information without requiring a
more general barrier across the entire job.
If waitfor is non-zero, the function will never return with found
set to zero.
Predefined attributes:
+ memPoolType - If the process manager allocated a shared memory
pool for the MPI processes in this job and on this node, return
the type of that pool. Types include sysv, anonmmap and ntshm.
. memSYSVid - Return the SYSV memory segment id if the memory pool
type is sysv. Returned as a string.
. memAnonMMAPfd - Return the FD of the anonymous mmap segment. The
FD is returned as a string.
- memNTName - Return the name of the Windows NT shared memory
segment, file mapping object backed by system paging
file. Returned as a string.
@*/
int PMI2_Info_GetNodeAttr(const char name[], char value[], int valuelen, int *found,
int waitfor);
/*@
PMI2_Info_GetNodeAttrIntArray - returns the value of the attribute associated
with this node. The value must be an array of integers.
Input Parameters:
+ name - name of the node attribute
- arraylen - number of elements in array
Output Parameters:
+ array - value of attribute
. outlen - number of elements returned
- found - non-zero if attribute was found
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
Notice that, unlike PMI2_Info_GetNodeAttr, this function does not
have a waitfor parameter, and will return immediately with found=0
if the attribute was not found.
Predefined array attribute names:
+ localRanksCount - Return the number of local ranks that will be
returned by the key localRanks.
. localRanks - Return the ranks in MPI_COMM_WORLD of the processes
that are running on this node.
- cartCoords - Return the Cartesian coordinates of this process in
the underlying network topology. The coordinates are indexed from
zero. Value only if the Job attribute for physTopology includes
cartesian.
@*/
int PMI2_Info_GetNodeAttrIntArray(const char name[], int array[], int arraylen, int *outlen,
int *found);
/*@
PMI2_Info_PutNodeAttr - stores the value of the named attribute
associated with this node
Input Parameters:
+ name - name of the node attribute
- value - the value of the attribute
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Notes:
For example, it might be used to share segment ids with other
processes on the same SMP node.
@*/
int PMI2_Info_PutNodeAttr(const char name[], const char value[]);
/*@
PMI2_Info_GetJobAttr - returns the value of the attribute associated
with this job
Input Parameters:
+ name - name of the job attribute
- valuelen - size of the buffer provided in value
Output Parameters:
+ value - value of the attribute
- found - non-zero indicates that the attribute was found
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
@*/
int PMI2_Info_GetJobAttr(const char name[], char value[], int valuelen, int *found);
/*@
PMI2_Info_GetJobAttrIntArray - returns the value of the attribute associated
with this job. The value must be an array of integers.
Input Parameters:
+ name - name of the job attribute
- arraylen - number of elements in array
Output Parameters:
+ array - value of attribute
. outlen - number of elements returned
- found - non-zero if attribute was found
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
Predefined array attribute names:
+ universeSize - The size of the "universe" (defined for the MPI
attribute MPI_UNIVERSE_SIZE
. hasNameServ - The value hasNameServ is true if the PMI2 environment
supports the name service operations (publish, lookup, and
unpublish).
. physTopology - Return the topology of the underlying network. The
valid topology types include cartesian, hierarchical, complete,
kautz, hypercube; additional types may be added as necessary. If
the type is hierarchical, then additional attributes may be
queried to determine the details of the topology. For example, a
typical cluster has a hierarchical physical topology, consisting
of two levels of complete networks - the switched Ethernet or
Infiniband and the SMP nodes. Other systems, such as IBM BlueGene,
have one level that is cartesian (and in virtual node mode, have a
single-level physical topology).
. physTopologyLevels - Return a string describing the topology type
for each level of the underlying network. Only valid if the
physTopology is hierarchical. The value is a comma-separated list
of physical topology types (except for hierarchical). The levels
are ordered starting at the top, with the network closest to the
processes last. The lower level networks may connect only a subset
of processes. For example, for a cartesian mesh of SMPs, the value
is cartesian,complete. All processes are connected by the
cartesian part of this, but for each complete network, only the
processes on the same node are connected.
. cartDims - Return a string of comma-separated values describing
the dimensions of the Cartesian topology. This must be consistent
with the value of cartCoords that may be returned by
PMI2_Info_GetNodeAttrIntArray.
These job attributes are just a start, but they provide both an
example of the sort of external data that is available through the
PMI interface and how extensions can be added within the same API
and wire protocol. For example, adding more complex network
topologies requires only adding new keys, not new routines.
. isHeterogeneous - The value isHeterogeneous is true if the
processes belonging to the job are running on nodes with different
underlying data models.
@*/
int PMI2_Info_GetJobAttrIntArray(const char name[], int array[], int arraylen, int *outlen,
int *found);
/*@
PMI2_Nameserv_publish - publish a name
Input parameters:
+ service_name - string representing the service being published
. info_ptr -
- port - string representing the port on which to contact the service
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
@*/
int PMI2_Nameserv_publish(const char service_name[], const struct MPIR_Info *info_ptr,
const char port[]);
/*@
PMI2_Nameserv_lookup - lookup a service by name
Input parameters:
+ service_name - string representing the service being published
. info_ptr -
- portLen - size of buffer provided in port
Output parameters:
. port - string representing the port on which to contact the service
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
@*/
int PMI2_Nameserv_lookup(const char service_name[], const struct MPIR_Info *info_ptr,
char port[], int portLen);
/*@
PMI2_Nameserv_unpublish - unpublish a name
Input parameters:
+ service_name - string representing the service being unpublished
- info_ptr -
Return values:
Returns 'MPI_SUCCESS' on success and an MPI error code on failure.
@*/
int PMI2_Nameserv_unpublish(const char service_name[], const struct MPIR_Info *info_ptr);
#if defined(__cplusplus)
}
#endif
#endif /* PMI2_H_INCLUDED */