static void report_duplicated_guid(IN osm_sm_t * sm, osm_physp_t * p_physp,
osm_node_t * p_neighbor_node,
const uint8_t port_num)
{
osm_physp_t *p_old, *p_new;
osm_dr_path_t path;
p_old = p_physp->p_remote_physp;
p_new = osm_node_get_physp_ptr(p_neighbor_node, port_num);
OSM_LOG(sm->p_log, OSM_LOG_SYS | OSM_LOG_ERROR, "ERR 0D01: "
"Found duplicated node GUID.\n"
"Node 0x%" PRIx64 " port %u is reachable from remote node "
"0x%" PRIx64 " port %u and remote node 0x%" PRIx64 " port %u.\n"
"Paths are:\n",
cl_ntoh64(p_physp->p_node->node_info.node_guid),
p_physp->port_num,
p_old ? cl_ntoh64(p_old->p_node->node_info.node_guid) : 0,
p_old ? p_old->port_num : 0,
p_new ? cl_ntoh64(p_new->p_node->node_info.node_guid) : 0,
p_new ? p_new->port_num : 0);
osm_dump_dr_path_v2(sm->p_log, osm_physp_get_dr_path_ptr(p_physp),
FILE_ID, OSM_LOG_ERROR);
path = *osm_physp_get_dr_path_ptr(p_new);
if (osm_dr_path_extend(&path, port_num))
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D05: "
"DR path with hop count %d couldn't be extended\n",
path.hop_count);
osm_dump_dr_path_v2(sm->p_log, &path, FILE_ID, OSM_LOG_ERROR);
}
static void ni_rcv_get_port_info(IN osm_sm_t * sm, IN osm_node_t * node,
IN const osm_madw_t * madw)
{
osm_madw_context_t context;
osm_physp_t *physp;
ib_node_info_t *ni;
unsigned port;
ib_api_status_t status;
int mlnx_epi_supported = 0;
ni = ib_smp_get_payload_ptr(osm_madw_get_smp_ptr(madw));
port = ib_node_info_get_local_port_num(ni);
if (sm->p_subn->opt.fdr10)
mlnx_epi_supported = is_mlnx_ext_port_info_supported(ni->device_id);
physp = osm_node_get_physp_ptr(node, port);
if (!physp) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR OD1E: "
"Failed to find physp for port %d of Node GUID 0x%"
PRIx64 "\n", port,
cl_ntoh64(osm_node_get_node_guid(node)));
return;
}
context.pi_context.node_guid = osm_node_get_node_guid(node);
context.pi_context.port_guid = osm_physp_get_port_guid(physp);
context.pi_context.set_method = FALSE;
context.pi_context.light_sweep = FALSE;
context.pi_context.active_transition = FALSE;
context.pi_context.client_rereg = FALSE;
status = osm_req_get(sm, osm_physp_get_dr_path_ptr(physp),
IB_MAD_ATTR_PORT_INFO, cl_hton32(port),
TRUE, 0, CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR OD02: "
"Failure initiating PortInfo request (%s)\n",
ib_get_err_str(status));
if (mlnx_epi_supported) {
status = osm_req_get(sm,
osm_physp_get_dr_path_ptr(physp),
IB_MAD_ATTR_MLNX_EXTENDED_PORT_INFO,
cl_hton32(port),
TRUE, 0, CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D0B: "
"Failure initiating MLNX ExtPortInfo request (%s)\n",
ib_get_err_str(status));
}
}
static void query_sm_info(cl_map_item_t * item, void *cxt)
{
osm_madw_context_t context;
osm_remote_sm_t *r_sm = cl_item_obj(item, r_sm, map_item);
osm_sm_t *sm = cxt;
ib_api_status_t ret;
osm_port_t *p_port;
p_port= osm_get_port_by_guid(sm->p_subn, r_sm->smi.guid);
if (p_port == NULL) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3340: "
"No port object on given sm object\n");
return;
}
context.smi_context.port_guid = r_sm->smi.guid;
context.smi_context.set_method = FALSE;
context.smi_context.light_sweep = TRUE;
ret = osm_req_get(sm, osm_physp_get_dr_path_ptr(p_port->p_physp),
IB_MAD_ATTR_SM_INFO, 0, CL_DISP_MSGID_NONE, &context);
if (ret != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3314: "
"Failure requesting SMInfo (%s)\n",
ib_get_err_str(ret));
}
static void state_mgr_get_sw_info(IN cl_map_item_t * p_object, IN void *context)
{
osm_node_t *p_node;
osm_dr_path_t *p_dr_path;
osm_madw_context_t mad_context;
osm_switch_t *const p_sw = (osm_switch_t *) p_object;
osm_sm_t *sm = context;
ib_api_status_t status;
OSM_LOG_ENTER(sm->p_log);
p_node = p_sw->p_node;
p_dr_path =
osm_physp_get_dr_path_ptr(osm_node_get_physp_ptr(p_node, 0));
memset(&mad_context, 0, sizeof(mad_context));
mad_context.si_context.node_guid = osm_node_get_node_guid(p_node);
mad_context.si_context.set_method = FALSE;
mad_context.si_context.light_sweep = TRUE;
status = osm_req_get(sm, p_dr_path, IB_MAD_ATTR_SWITCH_INFO, 0,
OSM_MSG_LIGHT_SWEEP_FAIL, &mad_context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3304: "
"Request for SwitchInfo failed (%s)\n",
ib_get_err_str(status));
OSM_LOG_EXIT(sm->p_log);
}
static void ni_rcv_process_switch(IN osm_sm_t * sm, IN osm_node_t * p_node,
IN const osm_madw_t * p_madw)
{
ib_api_status_t status = IB_SUCCESS;
osm_physp_t *p_physp;
osm_madw_context_t context;
osm_dr_path_t *path;
ib_smp_t *p_smp;
OSM_LOG_ENTER(sm->p_log);
p_smp = osm_madw_get_smp_ptr(p_madw);
p_physp = osm_node_get_physp_ptr(p_node, 0);
/* update DR path of already initialized switch port 0 */
path = osm_physp_get_dr_path_ptr(p_physp);
osm_dr_path_init(path, p_smp->hop_count, p_smp->initial_path);
context.si_context.node_guid = osm_node_get_node_guid(p_node);
context.si_context.set_method = FALSE;
context.si_context.light_sweep = FALSE;
context.si_context.lft_top_change = FALSE;
/* Request a SwitchInfo attribute */
status = osm_req_get(sm, path, IB_MAD_ATTR_SWITCH_INFO, 0, TRUE, 0,
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
/* continue despite error */
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D06: "
"Failure initiating SwitchInfo request (%s)\n",
ib_get_err_str(status));
OSM_LOG_EXIT(sm->p_log);
}
/**********************************************************************
The plock must be held before calling this function.
**********************************************************************/
static void pi_rcv_process_switch_port0(IN osm_sm_t * sm,
IN osm_node_t * p_node,
IN osm_physp_t * p_physp,
IN ib_port_info_t * p_pi)
{
ib_api_status_t status;
osm_madw_context_t context;
uint8_t port, num_ports;
OSM_LOG_ENTER(sm->p_log);
if (p_physp->need_update)
sm->p_subn->ignore_existing_lfts = TRUE;
pi_rcv_check_and_fix_lid(sm->p_log, p_pi, p_physp);
/* Update the PortInfo attribute */
osm_physp_set_port_info(p_physp, p_pi, sm);
/* Determine if base switch port 0 */
if (p_node->sw &&
!ib_switch_info_is_enhanced_port0(&p_node->sw->switch_info))
/* PortState is not used on BSP0 but just in case it is DOWN */
p_physp->port_info = *p_pi;
/* Now, query PortInfo for the switch external ports */
num_ports = osm_node_get_num_physp(p_node);
context.pi_context.node_guid = osm_node_get_node_guid(p_node);
context.pi_context.port_guid = osm_physp_get_port_guid(p_physp);
context.pi_context.set_method = FALSE;
context.pi_context.light_sweep = FALSE;
context.pi_context.active_transition = FALSE;
context.pi_context.client_rereg = FALSE;
for (port = 1; port < num_ports; port++) {
status = osm_req_get(sm, osm_physp_get_dr_path_ptr(p_physp),
IB_MAD_ATTR_PORT_INFO, cl_hton32(port),
FALSE,
ib_port_info_get_m_key(&p_physp->port_info),
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0F16: "
"Failure initiating PortInfo request (%s)\n",
ib_get_err_str(status));
}
pi_rcv_process_endport(sm, p_physp, p_pi);
OSM_LOG_EXIT(sm->p_log);
}
static int set_lft_block(IN osm_switch_t *p_sw, IN osm_ucast_mgr_t *p_mgr,
IN uint16_t block_id_ho)
{
uint8_t block[IB_SMP_DATA_SIZE];
osm_madw_context_t context;
osm_dr_path_t *p_path;
ib_api_status_t status;
/*
Send linear forwarding table blocks to the switch
as long as the switch indicates it has blocks needing
configuration.
*/
if (!p_sw->new_lft) {
/* any routing should provide the new_lft */
CL_ASSERT(p_mgr->p_subn->opt.use_ucast_cache &&
p_mgr->cache_valid && !p_sw->need_update);
return -1;
}
p_path = osm_physp_get_dr_path_ptr(osm_node_get_physp_ptr(p_sw->p_node, 0));
context.lft_context.node_guid = osm_node_get_node_guid(p_sw->p_node);
context.lft_context.set_method = TRUE;
if (!osm_switch_get_lft_block(p_sw, block_id_ho, block) ||
(!p_sw->need_update && !p_mgr->p_subn->need_update &&
!memcmp(block, p_sw->new_lft + block_id_ho * IB_SMP_DATA_SIZE,
IB_SMP_DATA_SIZE)))
return 0;
OSM_LOG(p_mgr->p_log, OSM_LOG_DEBUG,
"Writing FT block %u to switch 0x%" PRIx64 "\n", block_id_ho,
cl_ntoh64(context.lft_context.node_guid));
status = osm_req_set(p_mgr->sm, p_path,
p_sw->new_lft + block_id_ho * IB_SMP_DATA_SIZE,
IB_SMP_DATA_SIZE, IB_MAD_ATTR_LIN_FWD_TBL,
cl_hton32(block_id_ho),
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS) {
OSM_LOG(p_mgr->p_log, OSM_LOG_ERROR, "ERR 3A05: "
"Sending linear fwd. tbl. block failed (%s)\n",
ib_get_err_str(status));
return -1;
}
return 0;
}
static void sm_state_mgr_send_master_sm_info_req(osm_sm_t * sm)
{
osm_madw_context_t context;
const osm_port_t *p_port;
ib_api_status_t status;
OSM_LOG_ENTER(sm->p_log);
memset(&context, 0, sizeof(context));
if (sm->p_subn->sm_state == IB_SMINFO_STATE_STANDBY) {
/*
* We are in STANDBY state - this means we need to poll the
* master SM (according to master_guid).
* Send a query of SubnGet(SMInfo) to the subn
* master_sm_base_lid object.
*/
p_port = osm_get_port_by_guid(sm->p_subn, sm->master_sm_guid);
} else {
/*
* We are not in STANDBY - this means we are in MASTER state -
* so we need to poll the SM that is saved in p_polling_sm
* under sm.
* Send a query of SubnGet(SMInfo) to that SM.
*/
p_port = sm->p_polling_sm->p_port;
}
if (p_port == NULL) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3203: "
"No port object for GUID 0x%016" PRIx64 "\n",
cl_ntoh64(sm->master_sm_guid));
goto Exit;
}
context.smi_context.port_guid = p_port->guid;
context.smi_context.set_method = FALSE;
status = osm_req_get(sm, osm_physp_get_dr_path_ptr(p_port->p_physp),
IB_MAD_ATTR_SM_INFO, 0, CL_DISP_MSGID_NONE,
&context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3204: "
"Failure requesting SMInfo (%s)\n",
ib_get_err_str(status));
Exit:
OSM_LOG_EXIT(sm->p_log);
}
static qos_mad_item_t *osm_qos_mad_create(IN osm_sm_t * sm,
IN osm_physp_t * p,
IN uint32_t data_size,
IN uint8_t * p_data,
IN ib_net16_t attr_id,
IN uint32_t attr_mod)
{
qos_mad_item_t *p_mad;
osm_madw_context_t context;
osm_madw_t *p_madw;
osm_node_t *p_node;
p_node = osm_physp_get_node_ptr(p);
switch (attr_id){
case IB_MAD_ATTR_SLVL_TABLE:
context.slvl_context.node_guid = osm_node_get_node_guid(p_node);
context.slvl_context.port_guid = osm_physp_get_port_guid(p);
context.slvl_context.set_method = TRUE;
break;
case IB_MAD_ATTR_VL_ARBITRATION:
context.vla_context.node_guid = osm_node_get_node_guid(p_node);
context.vla_context.port_guid = osm_physp_get_port_guid(p);
context.vla_context.set_method = TRUE;
break;
default:
return NULL;
}
p_mad = (qos_mad_item_t *) malloc(sizeof(*p_mad));
if (!p_mad)
return NULL;
memset(p_mad, 0, sizeof(*p_mad));
p_madw = osm_prepare_req_set(sm, osm_physp_get_dr_path_ptr(p),
p_data, data_size,
attr_id, cl_hton32(attr_mod),
CL_DISP_MSGID_NONE, &context);
if (p_madw == NULL) {
free(p_mad);
return NULL;
}
p_mad->p_madw = p_madw;
return p_mad;
}
/**********************************************************************
Initiate a remote port info request for the given physical port
**********************************************************************/
static void state_mgr_get_remote_port_info(IN osm_sm_t * sm,
IN osm_physp_t * p_physp)
{
osm_dr_path_t *p_dr_path;
osm_dr_path_t rem_node_dr_path;
osm_madw_context_t mad_context;
ib_api_status_t status;
OSM_LOG_ENTER(sm->p_log);
/* generate a dr path leaving on the physp to the remote node */
p_dr_path = osm_physp_get_dr_path_ptr(p_physp);
memcpy(&rem_node_dr_path, p_dr_path, sizeof(osm_dr_path_t));
if (osm_dr_path_extend(&rem_node_dr_path, osm_physp_get_port_num(p_physp))) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 332D: "
"DR path with hop count %d couldn't be extended "
"so skipping PortInfo query\n",
p_dr_path->hop_count);
goto Exit;
}
memset(&mad_context, 0, sizeof(mad_context));
mad_context.pi_context.node_guid =
osm_node_get_node_guid(osm_physp_get_node_ptr(p_physp));
mad_context.pi_context.port_guid = p_physp->port_guid;
mad_context.pi_context.set_method = FALSE;
mad_context.pi_context.light_sweep = TRUE;
mad_context.pi_context.active_transition = FALSE;
/* note that with some negative logic - if the query failed it means
* that there is no point in going to heavy sweep */
status = osm_req_get(sm, &rem_node_dr_path, IB_MAD_ATTR_PORT_INFO, 0,
CL_DISP_MSGID_NONE, &mad_context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 332E: "
"Request for PortInfo failed (%s)\n",
ib_get_err_str(status));
Exit:
OSM_LOG_EXIT(sm->p_log);
}
/**********************************************************************
The plock must be held before calling this function.
**********************************************************************/
static void si_rcv_get_fwd_tbl(IN osm_sm_t * sm, IN osm_switch_t * p_sw)
{
osm_madw_context_t context;
osm_dr_path_t *p_dr_path;
osm_physp_t *p_physp;
osm_node_t *p_node;
uint32_t block_id_ho;
uint32_t max_block_id_ho;
ib_api_status_t status = IB_SUCCESS;
OSM_LOG_ENTER(sm->p_log);
CL_ASSERT(p_sw);
p_node = p_sw->p_node;
CL_ASSERT(osm_node_get_type(p_node) == IB_NODE_TYPE_SWITCH);
context.lft_context.node_guid = osm_node_get_node_guid(p_node);
context.lft_context.set_method = FALSE;
max_block_id_ho = osm_switch_get_max_block_id_in_use(p_sw);
p_physp = osm_node_get_physp_ptr(p_node, 0);
p_dr_path = osm_physp_get_dr_path_ptr(p_physp);
for (block_id_ho = 0; block_id_ho <= max_block_id_ho; block_id_ho++) {
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Retrieving FT block %u\n", block_id_ho);
status = osm_req_get(sm, p_dr_path, IB_MAD_ATTR_LIN_FWD_TBL,
cl_hton32(block_id_ho),
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
/* continue the loop despite the error */
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3603: "
"Failure initiating PortInfo request (%s)\n",
ib_get_err_str(status));
}
OSM_LOG_EXIT(sm->p_log);
}
static void state_mgr_update_node_desc(IN cl_map_item_t * obj, IN void *context)
{
osm_madw_context_t mad_context;
osm_node_t *p_node = (osm_node_t *) obj;
osm_sm_t *sm = context;
osm_physp_t *p_physp = NULL;
unsigned i, num_ports;
ib_api_status_t status;
OSM_LOG_ENTER(sm->p_log);
CL_ASSERT(p_node);
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Updating NodeDesc for 0x%016" PRIx64 "\n",
cl_ntoh64(osm_node_get_node_guid(p_node)));
/* get a physp to request from. */
num_ports = osm_node_get_num_physp(p_node);
for (i = 0; i < num_ports; i++)
if ((p_physp = osm_node_get_physp_ptr(p_node, i)))
break;
if (!p_physp) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 331C: "
"Failed to find any valid physical port object.\n");
goto exit;
}
mad_context.nd_context.node_guid = osm_node_get_node_guid(p_node);
status = osm_req_get(sm, osm_physp_get_dr_path_ptr(p_physp),
IB_MAD_ATTR_NODE_DESC, 0, CL_DISP_MSGID_NONE,
&mad_context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR,
"ERR 331B: Failure initiating NodeDescription request "
"(%s)\n", ib_get_err_str(status));
exit:
OSM_LOG_EXIT(sm->p_log);
}
/**********************************************************************
The plock must be held before calling this function.
**********************************************************************/
void osm_req_get_node_desc(IN osm_sm_t * sm, osm_physp_t * p_physp)
{
ib_api_status_t status = IB_SUCCESS;
osm_madw_context_t context;
OSM_LOG_ENTER(sm->p_log);
context.nd_context.node_guid =
osm_node_get_node_guid(osm_physp_get_node_ptr(p_physp));
status = osm_req_get(sm, osm_physp_get_dr_path_ptr(p_physp),
IB_MAD_ATTR_NODE_DESC, 0, TRUE, 0,
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D03: "
"Failure initiating NodeDescription request (%s)\n",
ib_get_err_str(status));
OSM_LOG_EXIT(sm->p_log);
}
static void
__osm_pi_rcv_set_sm(IN osm_sm_t * sm,
IN osm_physp_t * const p_physp)
{
osm_bind_handle_t h_bind;
osm_dr_path_t *p_dr_path;
OSM_LOG_ENTER(sm->p_log);
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Setting IS_SM bit in port attributes\n");
p_dr_path = osm_physp_get_dr_path_ptr(p_physp);
h_bind = osm_dr_path_get_bind_handle(p_dr_path);
/*
The 'IS_SM' bit isn't already set, so set it.
*/
osm_vendor_set_sm(h_bind, TRUE);
OSM_LOG_EXIT(sm->p_log);
}
static ib_api_status_t sl2vl_update_table(osm_sm_t * sm,
osm_physp_t * p, uint8_t in_port,
uint8_t out_port,
unsigned force_update,
const ib_slvl_table_t * sl2vl_table)
{
osm_madw_context_t context;
ib_slvl_table_t tbl, *p_tbl;
osm_node_t *p_node = osm_physp_get_node_ptr(p);
uint32_t attr_mod;
unsigned vl_mask;
uint8_t vl1, vl2;
int i;
vl_mask = (1 << (ib_port_info_get_op_vls(&p->port_info) - 1)) - 1;
for (i = 0; i < IB_MAX_NUM_VLS / 2; i++) {
vl1 = sl2vl_table->raw_vl_by_sl[i] >> 4;
vl2 = sl2vl_table->raw_vl_by_sl[i] & 0xf;
if (vl1 != 15)
vl1 &= vl_mask;
if (vl2 != 15)
vl2 &= vl_mask;
tbl.raw_vl_by_sl[i] = (vl1 << 4) | vl2;
}
if (!force_update && (p_tbl = osm_physp_get_slvl_tbl(p, in_port)) &&
!memcmp(p_tbl, &tbl, sizeof(tbl)))
return IB_SUCCESS;
context.slvl_context.node_guid = osm_node_get_node_guid(p_node);
context.slvl_context.port_guid = osm_physp_get_port_guid(p);
context.slvl_context.set_method = TRUE;
attr_mod = in_port << 8 | out_port;
return osm_req_set(sm, osm_physp_get_dr_path_ptr(p),
(uint8_t *) & tbl, sizeof(tbl),
IB_MAD_ATTR_SLVL_TABLE,
cl_hton32(attr_mod), CL_DISP_MSGID_NONE, &context);
}
/*
* QoS primitives
*/
static ib_api_status_t vlarb_update_table_block(osm_sm_t * sm,
osm_physp_t * p,
uint8_t port_num,
unsigned force_update,
const ib_vl_arb_table_t *
table_block,
unsigned block_length,
unsigned block_num)
{
ib_vl_arb_table_t block;
osm_madw_context_t context;
uint32_t attr_mod;
unsigned vl_mask, i;
vl_mask = (1 << (ib_port_info_get_op_vls(&p->port_info) - 1)) - 1;
memset(&block, 0, sizeof(block));
memcpy(&block, table_block, block_length * sizeof(block.vl_entry[0]));
for (i = 0; i < block_length; i++)
block.vl_entry[i].vl &= vl_mask;
if (!force_update &&
!memcmp(&p->vl_arb[block_num], &block,
block_length * sizeof(block.vl_entry[0])))
return IB_SUCCESS;
context.vla_context.node_guid =
osm_node_get_node_guid(osm_physp_get_node_ptr(p));
context.vla_context.port_guid = osm_physp_get_port_guid(p);
context.vla_context.set_method = TRUE;
attr_mod = ((block_num + 1) << 16) | port_num;
return osm_req_set(sm, osm_physp_get_dr_path_ptr(p),
(uint8_t *) & block, sizeof(block),
IB_MAD_ATTR_VL_ARBITRATION,
cl_hton32(attr_mod), CL_DISP_MSGID_NONE, &context);
}
static void requery_dup_node_info(IN osm_sm_t * sm, osm_physp_t * p_physp,
unsigned count)
{
osm_madw_context_t context;
osm_dr_path_t path;
cl_status_t status;
if (!p_physp->p_remote_physp) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D0D: "
"DR path couldn't be extended due to NULL remote physp\n");
return;
}
path = *osm_physp_get_dr_path_ptr(p_physp->p_remote_physp);
if (osm_dr_path_extend(&path, p_physp->p_remote_physp->port_num)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D08: "
"DR path with hop count %d couldn't be extended\n",
path.hop_count);
return;
}
context.ni_context.node_guid =
p_physp->p_remote_physp->p_node->node_info.port_guid;
context.ni_context.port_num = p_physp->p_remote_physp->port_num;
context.ni_context.dup_node_guid = p_physp->p_node->node_info.node_guid;
context.ni_context.dup_port_num = p_physp->port_num;
context.ni_context.dup_count = count;
status = osm_req_get(sm, &path, IB_MAD_ATTR_NODE_INFO, 0,
TRUE, 0, CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D02: "
"Failure initiating NodeInfo request (%s)\n",
ib_get_err_str(status));
}
/**********************************************************************
* Send SubnSet(SMInfo) SMP with HANDOVER attribute to the
* remote_sm indicated.
**********************************************************************/
static void state_mgr_send_handover(IN osm_sm_t * sm, IN osm_remote_sm_t * p_sm)
{
uint8_t payload[IB_SMP_DATA_SIZE];
ib_sm_info_t *p_smi = (ib_sm_info_t *) payload;
osm_madw_context_t context;
const osm_port_t *p_port;
ib_api_status_t status;
OSM_LOG_ENTER(sm->p_log);
/*
* Send a query of SubnSet(SMInfo) HANDOVER to the remote sm given.
*/
memset(&context, 0, sizeof(context));
p_port = p_sm->p_port;
if (p_port == NULL) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3316: "
"No port object on given remote_sm object\n");
goto Exit;
}
/* update the master_guid in the sm_state_mgr object according to */
/* the guid of the port where the new Master SM should reside. */
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Handing over mastership. Updating sm_state_mgr master_guid: %016"
PRIx64 " (node %s)\n", cl_ntoh64(p_port->guid),
p_port->p_node ? p_port->p_node->print_desc : "UNKNOWN");
sm->master_sm_guid = p_port->guid;
context.smi_context.port_guid = p_port->guid;
context.smi_context.set_method = TRUE;
p_smi->guid = sm->p_subn->sm_port_guid;
p_smi->act_count = cl_hton32(sm->p_subn->p_osm->stats.qp0_mads_sent);
p_smi->pri_state = (uint8_t) (sm->p_subn->sm_state |
sm->p_subn->opt.sm_priority << 4);
/*
* Return 0 for the SM key unless we authenticate the requester
* as the master SM.
*/
if (ib_sminfo_get_state(&p_sm->smi) == IB_SMINFO_STATE_MASTER) {
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Responding to master SM with real sm_key\n");
p_smi->sm_key = sm->p_subn->opt.sm_key;
} else {
/* The requester is not authenticated as master - set sm_key to zero */
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Responding to SM not master with zero sm_key\n");
p_smi->sm_key = 0;
}
status = osm_req_set(sm, osm_physp_get_dr_path_ptr(p_port->p_physp),
payload, sizeof(payload), IB_MAD_ATTR_SM_INFO,
IB_SMINFO_ATTR_MOD_HANDOVER, CL_DISP_MSGID_NONE,
&context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3317: "
"Failure requesting SMInfo (%s)\n",
ib_get_err_str(status));
Exit:
OSM_LOG_EXIT(sm->p_log);
}
/**********************************************************************
The plock must NOT be held before calling this function.
**********************************************************************/
static void ni_rcv_process_new(IN osm_sm_t * sm, IN const osm_madw_t * p_madw)
{
osm_node_t *p_node;
osm_node_t *p_node_check;
osm_port_t *p_port;
osm_port_t *p_port_check;
osm_router_t *p_rtr = NULL;
osm_router_t *p_rtr_check;
cl_qmap_t *p_rtr_guid_tbl;
ib_node_info_t *p_ni;
ib_smp_t *p_smp;
osm_ni_context_t *p_ni_context;
osm_alias_guid_t *p_alias_guid, *p_alias_guid_check;
uint8_t port_num;
OSM_LOG_ENTER(sm->p_log);
p_smp = osm_madw_get_smp_ptr(p_madw);
p_ni = ib_smp_get_payload_ptr(p_smp);
p_ni_context = osm_madw_get_ni_context_ptr(p_madw);
port_num = ib_node_info_get_local_port_num(p_ni);
osm_dump_smp_dr_path_v2(sm->p_log, p_smp, FILE_ID, OSM_LOG_VERBOSE);
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Discovered new %s node,"
"\n\t\t\t\tGUID 0x%" PRIx64 ", TID 0x%" PRIx64 "\n",
ib_get_node_type_str(p_ni->node_type),
cl_ntoh64(p_ni->node_guid), cl_ntoh64(p_smp->trans_id));
if (PF(port_num > p_ni->num_ports)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D0A: "
"New %s node GUID 0x%" PRIx64 "is non-compliant and "
"is being ignored since the "
"local port num %u > num ports %u\n",
ib_get_node_type_str(p_ni->node_type),
cl_ntoh64(p_ni->node_guid), port_num,
p_ni->num_ports);
goto Exit;
}
p_node = osm_node_new(p_madw);
if (PF(p_node == NULL)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D07: "
"Unable to create new node object\n");
goto Exit;
}
/*
Create a new port object to represent this node's physical
ports in the port table.
*/
p_port = osm_port_new(p_ni, p_node);
if (PF(p_port == NULL)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D14: "
"Unable to create new port object\n");
osm_node_delete(&p_node);
goto Exit;
}
/*
Add the new port object to the database.
*/
p_port_check =
(osm_port_t *) cl_qmap_insert(&sm->p_subn->port_guid_tbl,
p_ni->port_guid, &p_port->map_item);
if (PF(p_port_check != p_port)) {
/*
We should never be here!
Somehow, this port GUID already exists in the table.
*/
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D15: "
"Duplicate Port GUID 0x%" PRIx64
"! Found by the two directed routes:\n",
cl_ntoh64(p_ni->port_guid));
osm_dump_dr_path_v2(sm->p_log,
osm_physp_get_dr_path_ptr(p_port->p_physp),
FILE_ID, OSM_LOG_ERROR);
osm_dump_dr_path_v2(sm->p_log,
osm_physp_get_dr_path_ptr(p_port_check->
p_physp),
FILE_ID, OSM_LOG_ERROR);
osm_port_delete(&p_port);
osm_node_delete(&p_node);
goto Exit;
}
p_alias_guid = osm_alias_guid_new(p_ni->port_guid,
p_port);
if (PF(!p_alias_guid)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D18: "
"alias guid memory allocation failed"
" for port GUID 0x%" PRIx64 "\n",
cl_ntoh64(p_ni->port_guid));
goto alias_done2;
}
/* insert into alias guid table */
p_alias_guid_check =
(osm_alias_guid_t *) cl_qmap_insert(&sm->p_subn->alias_port_guid_tbl,
p_alias_guid->alias_guid,
&p_alias_guid->map_item);
if (p_alias_guid_check != p_alias_guid) {
/* alias GUID is a duplicate */
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D19: "
"Duplicate alias port GUID 0x%" PRIx64 "\n",
cl_ntoh64(p_ni->port_guid));
osm_alias_guid_delete(&p_alias_guid);
}
alias_done2:
/* If we are a master, then this means the port is new on the subnet.
Mark it as new - need to send trap 64 on these ports.
The condition that we are master is true, since if we are in discovering
state (meaning we woke up from standby or we are just initializing),
then these ports may be new to us, but are not new on the subnet.
If we are master, then the subnet as we know it is the updated one,
and any new ports we encounter should cause trap 64. C14-72.1.1 */
if (sm->p_subn->sm_state == IB_SMINFO_STATE_MASTER)
p_port->is_new = 1;
/* If there were RouterInfo or other router attribute,
this would be elsewhere */
if (p_ni->node_type == IB_NODE_TYPE_ROUTER) {
if (PF((p_rtr = osm_router_new(p_port)) == NULL))
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D1A: "
"Unable to create new router object\n");
else {
p_rtr_guid_tbl = &sm->p_subn->rtr_guid_tbl;
p_rtr_check =
(osm_router_t *) cl_qmap_insert(p_rtr_guid_tbl,
p_ni->port_guid,
&p_rtr->map_item);
if (PF(p_rtr_check != p_rtr))
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D1B: "
"Unable to add port GUID:0x%016" PRIx64
" to router table\n",
cl_ntoh64(p_ni->port_guid));
}
}
p_node_check =
(osm_node_t *) cl_qmap_insert(&sm->p_subn->node_guid_tbl,
p_ni->node_guid, &p_node->map_item);
if (PF(p_node_check != p_node)) {
/*
This node must have been inserted by another thread.
This is unexpected, but is not an error.
We can simply clean-up, since the other thread will
see this processing through to completion.
*/
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Discovery race detected at node 0x%" PRIx64 "\n",
cl_ntoh64(p_ni->node_guid));
osm_node_delete(&p_node);
p_node = p_node_check;
ni_rcv_set_links(sm, p_node, port_num, p_ni_context);
goto Exit;
} else
ni_rcv_set_links(sm, p_node, port_num, p_ni_context);
p_node->discovery_count++;
ni_rcv_get_node_desc(sm, p_node, p_madw);
switch (p_ni->node_type) {
case IB_NODE_TYPE_CA:
case IB_NODE_TYPE_ROUTER:
ni_rcv_process_new_ca_or_router(sm, p_node, p_madw);
break;
case IB_NODE_TYPE_SWITCH:
ni_rcv_process_new_switch(sm, p_node, p_madw);
break;
default:
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D16: "
"Unknown node type %u with GUID 0x%" PRIx64 "\n",
p_ni->node_type, cl_ntoh64(p_ni->node_guid));
break;
}
Exit:
OSM_LOG_EXIT(sm->p_log);
}
static void pi_rcv_process_endport(IN osm_sm_t * sm, IN osm_physp_t * p_physp,
IN const ib_port_info_t * p_pi)
{
osm_madw_context_t context;
ib_api_status_t status;
ib_net64_t port_guid;
int extended;
uint8_t rate, mtu, mpb;
unsigned data_vls;
cl_qmap_t *p_sm_tbl;
osm_remote_sm_t *p_sm;
OSM_LOG_ENTER(sm->p_log);
port_guid = osm_physp_get_port_guid(p_physp);
/* HACK extended port 0 should be handled too! */
if (osm_physp_get_port_num(p_physp) != 0 &&
ib_port_info_get_port_state(p_pi) != IB_LINK_DOWN) {
/* track the minimal endport MTU, rate, and operational VLs */
mtu = ib_port_info_get_mtu_cap(p_pi);
if (mtu < sm->p_subn->min_ca_mtu) {
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Setting endport minimal MTU to:%u defined by port:0x%"
PRIx64 "\n", mtu, cl_ntoh64(port_guid));
sm->p_subn->min_ca_mtu = mtu;
}
extended = p_pi->capability_mask & IB_PORT_CAP_HAS_EXT_SPEEDS;
rate = ib_port_info_compute_rate(p_pi, extended);
if (ib_path_compare_rates(rate, sm->p_subn->min_ca_rate) < 0) {
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Setting endport minimal rate to:%u defined by port:0x%"
PRIx64 "\n", rate, cl_ntoh64(port_guid));
sm->p_subn->min_ca_rate = rate;
}
data_vls = 1U << (ib_port_info_get_vl_cap(p_pi) - 1);
if (data_vls > 1U << (sm->p_subn->opt.max_op_vls - 1))
data_vls = 1U << (sm->p_subn->opt.max_op_vls - 1);
if (data_vls >= IB_MAX_NUM_VLS)
data_vls = IB_MAX_NUM_VLS - 1;
if ((uint8_t)data_vls < sm->p_subn->min_data_vls) {
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Setting endport minimal data VLs to:%u defined by port:0x%"
PRIx64 "\n", data_vls, cl_ntoh64(port_guid));
sm->p_subn->min_data_vls = data_vls;
}
}
/* Check M_Key vs M_Key protect, can we control the port ? */
mpb = ib_port_info_get_mpb(p_pi);
if (mpb > 0 && p_pi->m_key == 0) {
OSM_LOG(sm->p_log, OSM_LOG_INFO,
"Port 0x%" PRIx64 " has unknown M_Key, protection level %u\n",
cl_ntoh64(port_guid), mpb);
}
if (port_guid != sm->p_subn->sm_port_guid) {
p_sm_tbl = &sm->p_subn->sm_guid_tbl;
if (p_pi->capability_mask & IB_PORT_CAP_IS_SM) {
/*
* Before querying the SM - we want to make sure we
* clean its state, so if the querying fails we
* recognize that this SM is not active.
*/
p_sm =
(osm_remote_sm_t *) cl_qmap_get(p_sm_tbl,
port_guid);
if (p_sm != (osm_remote_sm_t *) cl_qmap_end(p_sm_tbl))
/* clean it up */
p_sm->smi.pri_state =
0xF0 & p_sm->smi.pri_state;
if (sm->p_subn->opt.ignore_other_sm)
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Ignoring SM on port 0x%" PRIx64 "\n",
cl_ntoh64(port_guid));
else {
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Detected another SM. Requesting SMInfo "
"from port 0x%" PRIx64 "\n",
cl_ntoh64(port_guid));
/*
This port indicates it's an SM and
it's not our own port.
Acquire the SMInfo Attribute.
*/
memset(&context, 0, sizeof(context));
context.smi_context.set_method = FALSE;
context.smi_context.port_guid = port_guid;
status = osm_req_get(sm,
osm_physp_get_dr_path_ptr
(p_physp),
IB_MAD_ATTR_SM_INFO, 0,
FALSE,
ib_port_info_get_m_key(&p_physp->port_info),
CL_DISP_MSGID_NONE,
&context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR,
"ERR 0F05: "
"Failure requesting SMInfo (%s) "
"from port 0x%" PRIx64 "\n",
ib_get_err_str(status),
cl_ntoh64(port_guid));
}
} else {
p_sm =
(osm_remote_sm_t *) cl_qmap_remove(p_sm_tbl,
port_guid);
if (p_sm != (osm_remote_sm_t *) cl_qmap_end(p_sm_tbl))
free(p_sm);
}
}
OSM_LOG_EXIT(sm->p_log);
}
/**********************************************************************
The plock must be held before calling this function.
**********************************************************************/
static void ni_rcv_process_existing_ca_or_router(IN osm_sm_t * sm,
IN osm_node_t * p_node,
IN const osm_madw_t * p_madw)
{
ib_node_info_t *p_ni;
ib_smp_t *p_smp;
osm_port_t *p_port;
osm_port_t *p_port_check;
uint8_t port_num;
osm_dr_path_t *p_dr_path;
osm_alias_guid_t *p_alias_guid, *p_alias_guid_check;
OSM_LOG_ENTER(sm->p_log);
p_smp = osm_madw_get_smp_ptr(p_madw);
p_ni = ib_smp_get_payload_ptr(p_smp);
port_num = ib_node_info_get_local_port_num(p_ni);
/*
Determine if we have encountered this node through a
previously undiscovered port. If so, build the new
port object.
*/
p_port = osm_get_port_by_guid(sm->p_subn, p_ni->port_guid);
if (!p_port) {
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Creating new port object with GUID 0x%" PRIx64 "\n",
cl_ntoh64(p_ni->port_guid));
osm_node_init_physp(p_node, port_num, p_madw);
p_port = osm_port_new(p_ni, p_node);
if (PF(p_port == NULL)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D04: "
"Unable to create new port object\n");
goto Exit;
}
/*
Add the new port object to the database.
*/
p_port_check =
(osm_port_t *) cl_qmap_insert(&sm->p_subn->port_guid_tbl,
p_ni->port_guid,
&p_port->map_item);
if (PF(p_port_check != p_port)) {
/*
We should never be here!
Somehow, this port GUID already exists in the table.
*/
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D12: "
"Port 0x%" PRIx64 " already in the database!\n",
cl_ntoh64(p_ni->port_guid));
osm_port_delete(&p_port);
goto Exit;
}
p_alias_guid = osm_alias_guid_new(p_ni->port_guid,
p_port);
if (PF(!p_alias_guid)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D11: "
"alias guid memory allocation failed"
" for port GUID 0x%" PRIx64 "\n",
cl_ntoh64(p_ni->port_guid));
goto alias_done;
}
/* insert into alias guid table */
p_alias_guid_check =
(osm_alias_guid_t *) cl_qmap_insert(&sm->p_subn->alias_port_guid_tbl,
p_alias_guid->alias_guid,
&p_alias_guid->map_item);
if (p_alias_guid_check != p_alias_guid) {
/* alias GUID is a duplicate */
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D13: "
"Duplicate alias port GUID 0x%" PRIx64 "\n",
cl_ntoh64(p_ni->port_guid));
osm_alias_guid_delete(&p_alias_guid);
osm_port_delete(&p_port);
goto Exit;
}
alias_done:
/* If we are a master, then this means the port is new on the subnet.
Mark it as new - need to send trap 64 for these ports.
The condition that we are master is true, since if we are in discovering
state (meaning we woke up from standby or we are just initializing),
then these ports may be new to us, but are not new on the subnet.
If we are master, then the subnet as we know it is the updated one,
and any new ports we encounter should cause trap 64. C14-72.1.1 */
if (sm->p_subn->sm_state == IB_SMINFO_STATE_MASTER)
p_port->is_new = 1;
} else {
osm_physp_t *p_physp = osm_node_get_physp_ptr(p_node, port_num);
if (PF(p_physp == NULL)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D1C: "
"No physical port found for node GUID 0x%"
PRIx64 " port %u. Might be duplicate port GUID\n",
cl_ntoh64(p_node->node_info.node_guid),
port_num);
goto Exit;
}
/*
Update the DR Path to the port,
in case the old one is no longer available.
*/
p_dr_path = osm_physp_get_dr_path_ptr(p_physp);
osm_dr_path_init(p_dr_path, p_smp->hop_count,
p_smp->initial_path);
}
ni_rcv_get_port_info(sm, p_node, p_madw);
Exit:
OSM_LOG_EXIT(sm->p_log);
}
int osm_ucast_mgr_set_fwd_table(IN osm_ucast_mgr_t * const p_mgr,
IN osm_switch_t * const p_sw)
{
osm_node_t *p_node;
osm_dr_path_t *p_path;
osm_madw_context_t context;
ib_api_status_t status;
ib_switch_info_t si;
uint16_t block_id_ho = 0;
uint8_t block[IB_SMP_DATA_SIZE];
boolean_t set_swinfo_require = FALSE;
uint16_t lin_top;
uint8_t life_state;
CL_ASSERT(p_mgr);
OSM_LOG_ENTER(p_mgr->p_log);
CL_ASSERT(p_sw);
p_node = p_sw->p_node;
CL_ASSERT(p_node);
p_path = osm_physp_get_dr_path_ptr(osm_node_get_physp_ptr(p_node, 0));
/*
Set the top of the unicast forwarding table.
*/
si = p_sw->switch_info;
lin_top = cl_hton16(p_sw->max_lid_ho);
if (lin_top != si.lin_top) {
set_swinfo_require = TRUE;
si.lin_top = lin_top;
}
/* check to see if the change state bit is on. If it is - then we
need to clear it. */
if (ib_switch_info_get_state_change(&si))
life_state = ((p_mgr->p_subn->opt.packet_life_time << 3)
| (si.life_state & IB_SWITCH_PSC)) & 0xfc;
else
life_state = (p_mgr->p_subn->opt.packet_life_time << 3) & 0xf8;
if ((life_state != si.life_state)
|| ib_switch_info_get_state_change(&si)) {
set_swinfo_require = TRUE;
si.life_state = life_state;
}
if (set_swinfo_require) {
OSM_LOG(p_mgr->p_log, OSM_LOG_DEBUG,
"Setting switch FT top to LID %u\n", p_sw->max_lid_ho);
context.si_context.light_sweep = FALSE;
context.si_context.node_guid = osm_node_get_node_guid(p_node);
context.si_context.set_method = TRUE;
status = osm_req_set(p_mgr->sm, p_path, (uint8_t *) & si,
sizeof(si),
IB_MAD_ATTR_SWITCH_INFO,
0, CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(p_mgr->p_log, OSM_LOG_ERROR, "ERR 3A06: "
"Sending SwitchInfo attribute failed (%s)\n",
ib_get_err_str(status));
}
/*
Send linear forwarding table blocks to the switch
as long as the switch indicates it has blocks needing
configuration.
*/
context.lft_context.node_guid = osm_node_get_node_guid(p_node);
context.lft_context.set_method = TRUE;
if (!p_sw->new_lft) {
/* any routing should provide the new_lft */
CL_ASSERT(p_mgr->p_subn->opt.use_ucast_cache &&
p_mgr->cache_valid && !p_sw->need_update);
goto Exit;
}
for (block_id_ho = 0;
osm_switch_get_lft_block(p_sw, block_id_ho, block);
block_id_ho++) {
if (!p_sw->need_update &&
!memcmp(block,
p_sw->new_lft + block_id_ho * IB_SMP_DATA_SIZE,
IB_SMP_DATA_SIZE))
continue;
OSM_LOG(p_mgr->p_log, OSM_LOG_DEBUG,
"Writing FT block %u\n", block_id_ho);
status = osm_req_set(p_mgr->sm, p_path,
p_sw->new_lft + block_id_ho * IB_SMP_DATA_SIZE,
sizeof(block),
IB_MAD_ATTR_LIN_FWD_TBL,
cl_hton32(block_id_ho),
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(p_mgr->p_log, OSM_LOG_ERROR, "ERR 3A05: "
"Sending linear fwd. tbl. block failed (%s)\n",
ib_get_err_str(status));
}
Exit:
OSM_LOG_EXIT(p_mgr->p_log);
return 0;
}
static void ucast_mgr_set_fwd_top(IN cl_map_item_t * p_map_item,
IN void *cxt)
{
osm_ucast_mgr_t *p_mgr = cxt;
osm_switch_t * p_sw = (osm_switch_t *) p_map_item;
osm_node_t *p_node;
osm_dr_path_t *p_path;
osm_madw_context_t context;
ib_api_status_t status;
ib_switch_info_t si;
boolean_t set_swinfo_require = FALSE;
uint16_t lin_top;
uint8_t life_state;
CL_ASSERT(p_mgr);
OSM_LOG_ENTER(p_mgr->p_log);
CL_ASSERT(p_sw);
p_node = p_sw->p_node;
CL_ASSERT(p_node);
if (p_mgr->max_lid < p_sw->max_lid_ho)
p_mgr->max_lid = p_sw->max_lid_ho;
p_path = osm_physp_get_dr_path_ptr(osm_node_get_physp_ptr(p_node, 0));
/*
Set the top of the unicast forwarding table.
*/
si = p_sw->switch_info;
lin_top = cl_hton16(p_sw->max_lid_ho);
if (lin_top != si.lin_top) {
set_swinfo_require = TRUE;
si.lin_top = lin_top;
}
/* check to see if the change state bit is on. If it is - then we
need to clear it. */
if (ib_switch_info_get_state_change(&si))
life_state = ((p_mgr->p_subn->opt.packet_life_time << 3)
| (si.life_state & IB_SWITCH_PSC)) & 0xfc;
else
life_state = (p_mgr->p_subn->opt.packet_life_time << 3) & 0xf8;
if (life_state != si.life_state || ib_switch_info_get_state_change(&si)) {
set_swinfo_require = TRUE;
si.life_state = life_state;
}
if (set_swinfo_require) {
OSM_LOG(p_mgr->p_log, OSM_LOG_DEBUG,
"Setting switch FT top to LID %u\n", p_sw->max_lid_ho);
context.si_context.light_sweep = FALSE;
context.si_context.node_guid = osm_node_get_node_guid(p_node);
context.si_context.set_method = TRUE;
status = osm_req_set(p_mgr->sm, p_path, (uint8_t *) & si,
sizeof(si), IB_MAD_ATTR_SWITCH_INFO,
0, CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(p_mgr->p_log, OSM_LOG_ERROR, "ERR 3A06: "
"Sending SwitchInfo attribute failed (%s)\n",
ib_get_err_str(status));
}
OSM_LOG_EXIT(p_mgr->p_log);
}
/**********************************************************************
The plock must be held before calling this function.
**********************************************************************/
static void ni_rcv_set_links(IN osm_sm_t * sm, osm_node_t * p_node,
const uint8_t port_num,
const osm_ni_context_t * p_ni_context)
{
osm_node_t *p_neighbor_node;
osm_physp_t *p_physp, *p_remote_physp;
OSM_LOG_ENTER(sm->p_log);
/*
A special case exists in which the node we're trying to
link is our own node. In this case, the guid value in
the ni_context will be zero.
*/
if (p_ni_context->node_guid == 0) {
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Nothing to link for our own node 0x%" PRIx64 "\n",
cl_ntoh64(osm_node_get_node_guid(p_node)));
goto _exit;
}
p_neighbor_node = osm_get_node_by_guid(sm->p_subn,
p_ni_context->node_guid);
if (PF(!p_neighbor_node)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 0D10: "
"Unexpected removal of neighbor node 0x%" PRIx64 "\n",
cl_ntoh64(p_ni_context->node_guid));
goto _exit;
}
/* When setting the link, ports on both
sides of the link should be initialized */
CL_ASSERT(osm_node_link_has_valid_ports(p_node, port_num,
p_neighbor_node,
p_ni_context->port_num));
if (osm_node_link_exists(p_node, port_num,
p_neighbor_node, p_ni_context->port_num)) {
OSM_LOG(sm->p_log, OSM_LOG_DEBUG, "Link already exists\n");
goto _exit;
}
p_physp = osm_node_get_physp_ptr(p_node, port_num);
if (!p_physp) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR OD0E: "
"Failed to find physp for port %d of Node GUID 0x%"
PRIx64 "\n", port_num,
cl_ntoh64(osm_node_get_node_guid(p_node)));
goto _exit;
}
/*
* If the link went UP, after we already discovered it, we shouldn't
* set the link between the ports and resweep.
*/
if (osm_physp_get_port_state(p_physp) == IB_LINK_DOWN &&
p_node->physp_discovered[port_num]) {
/* Link down on another side. Don't create a link*/
p_node->physp_discovered[port_num] = 0;
sm->p_subn->force_heavy_sweep = TRUE;
goto _exit;
}
if (osm_node_has_any_link(p_node, port_num) &&
sm->p_subn->force_heavy_sweep == FALSE &&
(!p_ni_context->dup_count ||
(p_ni_context->dup_node_guid == osm_node_get_node_guid(p_node) &&
p_ni_context->dup_port_num == port_num))) {
/*
Uh oh...
This could be reconnected ports, but also duplicated GUID
(2 nodes have the same guid) or a 12x link with lane reversal
that is not configured correctly.
We will try to recover by querying NodeInfo again.
In order to catch even fast port moving to new location(s)
and back we will count up to 5.
Some crazy reconnections (newly created switch loop right
before targeted CA) will not be catched this way. So in worst
case - report GUID duplication and request new discovery.
When switch node is targeted NodeInfo querying will be done
in opposite order, this is much stronger check, unfortunately
it is impossible with CAs.
*/
p_physp = osm_node_get_physp_ptr(p_node, port_num);
if (!p_physp) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR OD0F: "
"Failed to find physp for port %d of Node GUID 0x%"
PRIx64 "\n", port_num,
cl_ntoh64(osm_node_get_node_guid(p_node)));
goto _exit;
}
if (p_ni_context->dup_count > 5) {
report_duplicated_guid(sm, p_physp, p_neighbor_node,
p_ni_context->port_num);
sm->p_subn->force_heavy_sweep = TRUE;
} else if (p_node->sw)
requery_dup_node_info(sm, p_physp->p_remote_physp,
p_ni_context->dup_count + 1);
else
requery_dup_node_info(sm, p_physp,
p_ni_context->dup_count + 1);
}
/*
When there are only two nodes with exact same guids (connected back
to back) - the previous check for duplicated guid will not catch
them. But the link will be from the port to itself...
Enhanced Port 0 is an exception to this
*/
if (osm_node_get_node_guid(p_node) == p_ni_context->node_guid &&
port_num == p_ni_context->port_num &&
port_num != 0 && cl_qmap_count(&sm->p_subn->sw_guid_tbl) == 0) {
OSM_LOG(sm->p_log, OSM_LOG_VERBOSE,
"Duplicate GUID found by link from a port to itself:"
"node 0x%" PRIx64 ", port number %u\n",
cl_ntoh64(osm_node_get_node_guid(p_node)), port_num);
p_physp = osm_node_get_physp_ptr(p_node, port_num);
if (!p_physp) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR OD1D: "
"Failed to find physp for port %d of Node GUID 0x%"
PRIx64 "\n", port_num,
cl_ntoh64(osm_node_get_node_guid(p_node)));
goto _exit;
}
osm_dump_dr_path_v2(sm->p_log, osm_physp_get_dr_path_ptr(p_physp),
FILE_ID, OSM_LOG_VERBOSE);
if (sm->p_subn->opt.exit_on_fatal == TRUE) {
osm_log_v2(sm->p_log, OSM_LOG_SYS, FILE_ID,
"Errors on subnet. Duplicate GUID found "
"by link from a port to itself. "
"See verbose opensm.log for more details\n");
exit(1);
}
}
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Creating new link between:\n\t\t\t\tnode 0x%" PRIx64
", port number %u and\n\t\t\t\tnode 0x%" PRIx64
", port number %u\n",
cl_ntoh64(osm_node_get_node_guid(p_node)), port_num,
cl_ntoh64(p_ni_context->node_guid), p_ni_context->port_num);
if (sm->ucast_mgr.cache_valid)
osm_ucast_cache_check_new_link(&sm->ucast_mgr, p_node, port_num,
p_neighbor_node,
p_ni_context->port_num);
p_physp = osm_node_get_physp_ptr(p_node, port_num);
p_remote_physp = osm_node_get_physp_ptr(p_neighbor_node,
p_ni_context->port_num);
if (!p_physp || !p_remote_physp)
goto _exit;
osm_node_link(p_node, port_num, p_neighbor_node, p_ni_context->port_num);
osm_db_neighbor_set(sm->p_subn->p_neighbor,
cl_ntoh64(osm_physp_get_port_guid(p_physp)),
port_num,
cl_ntoh64(osm_physp_get_port_guid(p_remote_physp)),
p_ni_context->port_num);
osm_db_neighbor_set(sm->p_subn->p_neighbor,
cl_ntoh64(osm_physp_get_port_guid(p_remote_physp)),
p_ni_context->port_num,
cl_ntoh64(osm_physp_get_port_guid(p_physp)),
port_num);
_exit:
OSM_LOG_EXIT(sm->p_log);
}
static void ucast_mgr_set_fwd_top(IN cl_map_item_t * p_map_item,
IN void *cxt)
{
osm_ucast_mgr_t *p_mgr = cxt;
osm_switch_t * p_sw = (osm_switch_t *) p_map_item;
osm_node_t *p_node;
osm_physp_t *p_physp;
osm_dr_path_t *p_path;
osm_madw_context_t context;
ib_api_status_t status;
ib_switch_info_t si;
boolean_t set_swinfo_require = FALSE;
uint16_t lin_top;
uint8_t life_state;
CL_ASSERT(p_mgr);
OSM_LOG_ENTER(p_mgr->p_log);
CL_ASSERT(p_sw && p_sw->max_lid_ho);
p_node = p_sw->p_node;
CL_ASSERT(p_node);
if (p_mgr->max_lid < p_sw->max_lid_ho)
p_mgr->max_lid = p_sw->max_lid_ho;
p_physp = osm_node_get_physp_ptr(p_node, 0);
CL_ASSERT(p_physp);
p_path = osm_physp_get_dr_path_ptr(p_physp);
/*
Set the top of the unicast forwarding table.
*/
si = p_sw->switch_info;
lin_top = cl_hton16(p_sw->max_lid_ho);
if (lin_top != si.lin_top) {
set_swinfo_require = TRUE;
si.lin_top = lin_top;
context.si_context.lft_top_change = TRUE;
} else
context.si_context.lft_top_change = FALSE;
life_state = si.life_state;
ib_switch_info_set_life_time(&si, p_mgr->p_subn->opt.packet_life_time);
if (life_state != si.life_state)
set_swinfo_require = TRUE;
if (set_swinfo_require) {
OSM_LOG(p_mgr->p_log, OSM_LOG_DEBUG,
"Setting switch FT top to LID %u\n", p_sw->max_lid_ho);
context.si_context.light_sweep = FALSE;
context.si_context.node_guid = osm_node_get_node_guid(p_node);
context.si_context.set_method = TRUE;
status = osm_req_set(p_mgr->sm, p_path, (uint8_t *) & si,
sizeof(si), IB_MAD_ATTR_SWITCH_INFO,
0, FALSE,
ib_port_info_get_m_key(&p_physp->port_info),
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(p_mgr->p_log, OSM_LOG_ERROR, "ERR 3A06: "
"Sending SwitchInfo attribute failed (%s)\n",
ib_get_err_str(status));
}
OSM_LOG_EXIT(p_mgr->p_log);
}
/**********************************************************************
The plock must be held before calling this function.
**********************************************************************/
static void si_rcv_get_mcast_fwd_tbl(IN osm_sm_t * sm, IN osm_switch_t * p_sw)
{
osm_madw_context_t context;
osm_dr_path_t *p_dr_path;
osm_physp_t *p_physp;
osm_node_t *p_node;
osm_mcast_tbl_t *p_tbl;
uint32_t block_id_ho;
uint32_t max_block_id_ho;
uint32_t position;
uint32_t max_position;
uint32_t attr_mod_ho;
ib_api_status_t status = IB_SUCCESS;
OSM_LOG_ENTER(sm->p_log);
CL_ASSERT(p_sw);
p_node = p_sw->p_node;
CL_ASSERT(osm_node_get_type(p_node) == IB_NODE_TYPE_SWITCH);
if (osm_switch_get_mcast_fwd_tbl_size(p_sw) == 0) {
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Multicast not supported by switch 0x%016" PRIx64 "\n",
cl_ntoh64(osm_node_get_node_guid(p_node)));
goto Exit;
}
context.mft_context.node_guid = osm_node_get_node_guid(p_node);
context.mft_context.set_method = FALSE;
p_tbl = osm_switch_get_mcast_tbl_ptr(p_sw);
max_block_id_ho = osm_mcast_tbl_get_max_block(p_tbl);
if (max_block_id_ho > IB_MCAST_MAX_BLOCK_ID) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3609: "
"Out-of-range mcast block size = %u on switch 0x%016"
PRIx64 "\n", max_block_id_ho,
cl_ntoh64(osm_node_get_node_guid(p_node)));
goto Exit;
}
max_position = osm_mcast_tbl_get_max_position(p_tbl);
CL_ASSERT(max_position <= IB_MCAST_POSITION_MAX);
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Max MFT block = %u, Max position = %u\n", max_block_id_ho,
max_position);
p_physp = osm_node_get_physp_ptr(p_node, 0);
p_dr_path = osm_physp_get_dr_path_ptr(p_physp);
for (block_id_ho = 0; block_id_ho <= max_block_id_ho; block_id_ho++) {
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Retrieving MFT block %u\n", block_id_ho);
for (position = 0; position <= max_position; position++) {
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Retrieving MFT position %u\n", position);
attr_mod_ho =
block_id_ho | position << IB_MCAST_POSITION_SHIFT;
status =
osm_req_get(sm, p_dr_path,
IB_MAD_ATTR_MCAST_FWD_TBL,
cl_hton32(attr_mod_ho),
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
/* continue the loop despite the error */
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3607: "
"Failure initiating PortInfo request (%s)\n",
ib_get_err_str(status));
}
}
Exit:
OSM_LOG_EXIT(sm->p_log);
}
static boolean_t
__osm_link_mgr_set_physp_pi(osm_sm_t * sm,
IN osm_physp_t * const p_physp,
IN uint8_t const port_state)
{
uint8_t payload[IB_SMP_DATA_SIZE];
ib_port_info_t *const p_pi = (ib_port_info_t *) payload;
const ib_port_info_t *p_old_pi;
osm_madw_context_t context;
osm_node_t *p_node;
ib_api_status_t status;
uint8_t port_num;
uint8_t mtu;
uint8_t op_vls;
boolean_t esp0 = FALSE;
boolean_t send_set = FALSE;
osm_physp_t *p_remote_physp;
OSM_LOG_ENTER(sm->p_log);
p_node = osm_physp_get_node_ptr(p_physp);
port_num = osm_physp_get_port_num(p_physp);
if (port_num == 0) {
/*
CAs don't have a port 0, and for switch port 0,
we need to check if this is enhanced or base port 0.
For base port 0 the following parameters are not valid (p822, table 145).
*/
if (!p_node->sw) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 4201: "
"Cannot find switch by guid: 0x%" PRIx64 "\n",
cl_ntoh64(p_node->node_info.node_guid));
goto Exit;
}
if (ib_switch_info_is_enhanced_port0(&p_node->sw->switch_info)
== FALSE) {
/* This means the switch doesn't support enhanced port 0.
Can skip it. */
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Skipping port 0, GUID 0x%016" PRIx64 "\n",
cl_ntoh64(osm_physp_get_port_guid(p_physp)));
goto Exit;
}
esp0 = TRUE;
}
/*
PAST THIS POINT WE ARE HANDLING EITHER A NON PORT 0 OR ENHANCED PORT 0
*/
p_old_pi = &p_physp->port_info;
memset(payload, 0, IB_SMP_DATA_SIZE);
memcpy(payload, p_old_pi, sizeof(ib_port_info_t));
/*
Should never write back a value that is bigger then 3 in
the PortPhysicalState field - so can not simply copy!
Actually we want to write there:
port physical state - no change,
link down default state = polling
port state - as requested.
*/
p_pi->state_info2 = 0x02;
ib_port_info_set_port_state(p_pi, port_state);
if (ib_port_info_get_link_down_def_state(p_pi) !=
ib_port_info_get_link_down_def_state(p_old_pi))
send_set = TRUE;
/* didn't get PortInfo before */
if (!ib_port_info_get_port_state(p_old_pi))
send_set = TRUE;
/* we only change port fields if we do not change state */
if (port_state == IB_LINK_NO_CHANGE) {
/* The following fields are relevant only for CA port, router, or Enh. SP0 */
if (osm_node_get_type(p_node) != IB_NODE_TYPE_SWITCH ||
port_num == 0) {
p_pi->m_key = sm->p_subn->opt.m_key;
if (memcmp(&p_pi->m_key, &p_old_pi->m_key,
sizeof(p_pi->m_key)))
send_set = TRUE;
p_pi->subnet_prefix = sm->p_subn->opt.subnet_prefix;
if (memcmp(&p_pi->subnet_prefix,
&p_old_pi->subnet_prefix,
sizeof(p_pi->subnet_prefix)))
send_set = TRUE;
p_pi->base_lid = osm_physp_get_base_lid(p_physp);
if (memcmp(&p_pi->base_lid, &p_old_pi->base_lid,
sizeof(p_pi->base_lid)))
send_set = TRUE;
/* we are initializing the ports with our local sm_base_lid */
p_pi->master_sm_base_lid = sm->p_subn->sm_base_lid;
if (memcmp(&p_pi->master_sm_base_lid,
&p_old_pi->master_sm_base_lid,
sizeof(p_pi->master_sm_base_lid)))
send_set = TRUE;
p_pi->m_key_lease_period =
sm->p_subn->opt.m_key_lease_period;
if (memcmp(&p_pi->m_key_lease_period,
&p_old_pi->m_key_lease_period,
sizeof(p_pi->m_key_lease_period)))
send_set = TRUE;
if (esp0 == FALSE)
p_pi->mkey_lmc = sm->p_subn->opt.lmc;
else {
if (sm->p_subn->opt.lmc_esp0)
p_pi->mkey_lmc = sm->p_subn->opt.lmc;
else
p_pi->mkey_lmc = 0;
}
if (memcmp(&p_pi->mkey_lmc, &p_old_pi->mkey_lmc,
sizeof(p_pi->mkey_lmc)))
send_set = TRUE;
ib_port_info_set_timeout(p_pi,
sm->p_subn->opt.
subnet_timeout);
if (ib_port_info_get_timeout(p_pi) !=
ib_port_info_get_timeout(p_old_pi))
send_set = TRUE;
}
/*
Several timeout mechanisms:
*/
p_remote_physp = osm_physp_get_remote(p_physp);
if (port_num != 0 && p_remote_physp) {
if (osm_node_get_type(osm_physp_get_node_ptr(p_physp))
== IB_NODE_TYPE_ROUTER) {
ib_port_info_set_hoq_lifetime(p_pi,
sm->p_subn->
opt.
leaf_head_of_queue_lifetime);
} else
if (osm_node_get_type
(osm_physp_get_node_ptr(p_physp)) ==
IB_NODE_TYPE_SWITCH) {
/* Is remote end CA or router (a leaf port) ? */
if (osm_node_get_type
(osm_physp_get_node_ptr(p_remote_physp)) !=
IB_NODE_TYPE_SWITCH) {
ib_port_info_set_hoq_lifetime(p_pi,
sm->
p_subn->
opt.
leaf_head_of_queue_lifetime);
ib_port_info_set_vl_stall_count(p_pi,
sm->
p_subn->
opt.
leaf_vl_stall_count);
} else {
ib_port_info_set_hoq_lifetime(p_pi,
sm->
p_subn->
opt.
head_of_queue_lifetime);
ib_port_info_set_vl_stall_count(p_pi,
sm->
p_subn->
opt.
vl_stall_count);
}
}
if (ib_port_info_get_hoq_lifetime(p_pi) !=
ib_port_info_get_hoq_lifetime(p_old_pi) ||
ib_port_info_get_vl_stall_count(p_pi) !=
ib_port_info_get_vl_stall_count(p_old_pi))
send_set = TRUE;
}
ib_port_info_set_phy_and_overrun_err_thd(p_pi,
sm->p_subn->opt.
local_phy_errors_threshold,
sm->p_subn->opt.
overrun_errors_threshold);
if (memcmp(&p_pi->error_threshold, &p_old_pi->error_threshold,
sizeof(p_pi->error_threshold)))
send_set = TRUE;
/*
Set the easy common parameters for all port types,
then determine the neighbor MTU.
*/
p_pi->link_width_enabled = p_old_pi->link_width_supported;
if (memcmp(&p_pi->link_width_enabled,
&p_old_pi->link_width_enabled,
sizeof(p_pi->link_width_enabled)))
send_set = TRUE;
if (sm->p_subn->opt.force_link_speed &&
(sm->p_subn->opt.force_link_speed != 15 ||
ib_port_info_get_link_speed_enabled(p_pi) !=
ib_port_info_get_link_speed_sup(p_pi))) {
ib_port_info_set_link_speed_enabled(p_pi,
sm->p_subn->opt.
force_link_speed);
if (memcmp(&p_pi->link_speed, &p_old_pi->link_speed,
sizeof(p_pi->link_speed)))
send_set = TRUE;
}
/* calc new op_vls and mtu */
op_vls =
osm_physp_calc_link_op_vls(sm->p_log, sm->p_subn, p_physp);
mtu = osm_physp_calc_link_mtu(sm->p_log, p_physp);
ib_port_info_set_neighbor_mtu(p_pi, mtu);
if (ib_port_info_get_neighbor_mtu(p_pi) !=
ib_port_info_get_neighbor_mtu(p_old_pi))
send_set = TRUE;
ib_port_info_set_op_vls(p_pi, op_vls);
if (ib_port_info_get_op_vls(p_pi) !=
ib_port_info_get_op_vls(p_old_pi))
send_set = TRUE;
/* provide the vl_high_limit from the qos mgr */
if (sm->p_subn->opt.qos &&
p_physp->vl_high_limit != p_old_pi->vl_high_limit) {
send_set = TRUE;
p_pi->vl_high_limit = p_physp->vl_high_limit;
}
}
if (port_state != IB_LINK_NO_CHANGE &&
port_state != ib_port_info_get_port_state(p_old_pi)) {
send_set = TRUE;
if (port_state == IB_LINK_ACTIVE)
context.pi_context.active_transition = TRUE;
else
context.pi_context.active_transition = FALSE;
}
context.pi_context.node_guid = osm_node_get_node_guid(p_node);
context.pi_context.port_guid = osm_physp_get_port_guid(p_physp);
context.pi_context.set_method = TRUE;
context.pi_context.light_sweep = FALSE;
/* We need to send the PortInfoSet request with the new sm_lid
in the following cases:
1. There is a change in the values (send_set == TRUE)
2. This is a switch external port (so it wasn't handled yet by
osm_lid_mgr) and first_time_master_sweep flag on the subnet is TRUE,
which means the SM just became master, and it then needs to send at
PortInfoSet to every port.
*/
if (osm_node_get_type(p_node) == IB_NODE_TYPE_SWITCH && port_num
&& sm->p_subn->first_time_master_sweep == TRUE)
send_set = TRUE;
if (send_set)
status = osm_req_set(sm, osm_physp_get_dr_path_ptr(p_physp),
payload, sizeof(payload),
IB_MAD_ATTR_PORT_INFO,
cl_hton32(port_num),
CL_DISP_MSGID_NONE, &context);
Exit:
OSM_LOG_EXIT(sm->p_log);
return send_set;
}
static int link_mgr_set_physp_pi(osm_sm_t * sm, IN osm_physp_t * p_physp,
IN uint8_t port_state)
{
uint8_t payload[IB_SMP_DATA_SIZE], payload2[IB_SMP_DATA_SIZE];
ib_port_info_t *p_pi = (ib_port_info_t *) payload;
ib_mlnx_ext_port_info_t *p_epi = (ib_mlnx_ext_port_info_t *) payload2;
const ib_port_info_t *p_old_pi;
const ib_mlnx_ext_port_info_t *p_old_epi;
osm_madw_context_t context;
osm_node_t *p_node;
ib_api_status_t status;
uint8_t port_num, mtu, op_vls, smsl = OSM_DEFAULT_SL;
boolean_t esp0 = FALSE, send_set = FALSE, send_set2 = FALSE;
osm_physp_t *p_remote_physp, *physp0;
int qdr_change = 0, fdr10_change = 0;
int ret = 0;
ib_net32_t attr_mod, cap_mask;
OSM_LOG_ENTER(sm->p_log);
p_node = osm_physp_get_node_ptr(p_physp);
p_old_pi = &p_physp->port_info;
port_num = osm_physp_get_port_num(p_physp);
if (port_num == 0) {
/*
CAs don't have a port 0, and for switch port 0,
we need to check if this is enhanced or base port 0.
For base port 0 the following parameters are not valid
(IBA 1.2.1 p.830 table 146).
*/
if (!p_node->sw) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 4201: "
"Cannot find switch by guid: 0x%" PRIx64 "\n",
cl_ntoh64(p_node->node_info.node_guid));
goto Exit;
}
if (ib_switch_info_is_enhanced_port0(&p_node->sw->switch_info)
== FALSE) {
/* Even for base port 0 we might have to set smsl
(if we are using lash routing) */
smsl = link_mgr_get_smsl(sm, p_physp);
if (smsl != ib_port_info_get_master_smsl(p_old_pi)) {
send_set = TRUE;
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Setting SMSL to %d on port 0 GUID 0x%016"
PRIx64 "\n", smsl,
cl_ntoh64(osm_physp_get_port_guid
(p_physp)));
} else {
/* This means the switch doesn't support
enhanced port 0 and we don't need to
change SMSL. Can skip it. */
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Skipping port 0, GUID 0x%016" PRIx64
"\n",
cl_ntoh64(osm_physp_get_port_guid
(p_physp)));
goto Exit;
}
} else
esp0 = TRUE;
}
memcpy(payload, p_old_pi, sizeof(ib_port_info_t));
/*
Should never write back a value that is bigger then 3 in
the PortPhysicalState field - so can not simply copy!
Actually we want to write there:
port physical state - no change,
link down default state = polling
port state - as requested.
*/
p_pi->state_info2 = 0x02;
ib_port_info_set_port_state(p_pi, port_state);
/* Check whether this is base port0 smsl handling only */
if (port_num == 0 && esp0 == FALSE) {
ib_port_info_set_master_smsl(p_pi, smsl);
goto Send;
}
/*
PAST THIS POINT WE ARE HANDLING EITHER A NON PORT 0 OR ENHANCED PORT 0
*/
if (ib_port_info_get_link_down_def_state(p_pi) !=
ib_port_info_get_link_down_def_state(p_old_pi))
send_set = TRUE;
/* didn't get PortInfo before */
if (!ib_port_info_get_port_state(p_old_pi))
send_set = TRUE;
/* we only change port fields if we do not change state */
if (port_state == IB_LINK_NO_CHANGE) {
/* The following fields are relevant only for CA port, router, or Enh. SP0 */
if (osm_node_get_type(p_node) != IB_NODE_TYPE_SWITCH ||
port_num == 0) {
p_pi->m_key = sm->p_subn->opt.m_key;
if (memcmp(&p_pi->m_key, &p_old_pi->m_key,
sizeof(p_pi->m_key)))
send_set = TRUE;
p_pi->subnet_prefix = sm->p_subn->opt.subnet_prefix;
if (memcmp(&p_pi->subnet_prefix,
&p_old_pi->subnet_prefix,
sizeof(p_pi->subnet_prefix)))
send_set = TRUE;
p_pi->base_lid = osm_physp_get_base_lid(p_physp);
if (memcmp(&p_pi->base_lid, &p_old_pi->base_lid,
sizeof(p_pi->base_lid)))
send_set = TRUE;
/* we are initializing the ports with our local sm_base_lid */
p_pi->master_sm_base_lid = sm->p_subn->sm_base_lid;
if (memcmp(&p_pi->master_sm_base_lid,
&p_old_pi->master_sm_base_lid,
sizeof(p_pi->master_sm_base_lid)))
send_set = TRUE;
smsl = link_mgr_get_smsl(sm, p_physp);
if (smsl != ib_port_info_get_master_smsl(p_old_pi)) {
ib_port_info_set_master_smsl(p_pi, smsl);
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Setting SMSL to %d on GUID 0x%016"
PRIx64 ", port %d\n", smsl,
cl_ntoh64(osm_physp_get_port_guid
(p_physp)), port_num);
send_set = TRUE;
}
p_pi->m_key_lease_period =
sm->p_subn->opt.m_key_lease_period;
if (memcmp(&p_pi->m_key_lease_period,
&p_old_pi->m_key_lease_period,
sizeof(p_pi->m_key_lease_period)))
send_set = TRUE;
/* M_KeyProtectBits are currently always zero */
p_pi->mkey_lmc = 0;
if (esp0 == FALSE || sm->p_subn->opt.lmc_esp0)
ib_port_info_set_lmc(p_pi, sm->p_subn->opt.lmc);
if (ib_port_info_get_lmc(p_old_pi) !=
ib_port_info_get_lmc(p_pi) ||
ib_port_info_get_mpb(p_old_pi) !=
ib_port_info_get_mpb(p_pi))
send_set = TRUE;
ib_port_info_set_timeout(p_pi,
sm->p_subn->opt.
subnet_timeout);
if (ib_port_info_get_timeout(p_pi) !=
ib_port_info_get_timeout(p_old_pi))
send_set = TRUE;
}
/*
Several timeout mechanisms:
*/
p_remote_physp = osm_physp_get_remote(p_physp);
if (port_num != 0 && p_remote_physp) {
if (osm_node_get_type(osm_physp_get_node_ptr(p_physp))
== IB_NODE_TYPE_ROUTER) {
ib_port_info_set_hoq_lifetime(p_pi,
sm->p_subn->
opt.
leaf_head_of_queue_lifetime);
} else
if (osm_node_get_type
(osm_physp_get_node_ptr(p_physp)) ==
IB_NODE_TYPE_SWITCH) {
/* Is remote end CA or router (a leaf port) ? */
if (osm_node_get_type
(osm_physp_get_node_ptr(p_remote_physp)) !=
IB_NODE_TYPE_SWITCH) {
ib_port_info_set_hoq_lifetime(p_pi,
sm->
p_subn->
opt.
leaf_head_of_queue_lifetime);
ib_port_info_set_vl_stall_count(p_pi,
sm->
p_subn->
opt.
leaf_vl_stall_count);
} else {
ib_port_info_set_hoq_lifetime(p_pi,
sm->
p_subn->
opt.
head_of_queue_lifetime);
ib_port_info_set_vl_stall_count(p_pi,
sm->
p_subn->
opt.
vl_stall_count);
}
}
if (ib_port_info_get_hoq_lifetime(p_pi) !=
ib_port_info_get_hoq_lifetime(p_old_pi) ||
ib_port_info_get_vl_stall_count(p_pi) !=
ib_port_info_get_vl_stall_count(p_old_pi))
send_set = TRUE;
}
ib_port_info_set_phy_and_overrun_err_thd(p_pi,
sm->p_subn->opt.
local_phy_errors_threshold,
sm->p_subn->opt.
overrun_errors_threshold);
if (memcmp(&p_pi->error_threshold, &p_old_pi->error_threshold,
sizeof(p_pi->error_threshold)))
send_set = TRUE;
/*
Set the easy common parameters for all port types,
then determine the neighbor MTU.
*/
p_pi->link_width_enabled = p_old_pi->link_width_supported;
if (memcmp(&p_pi->link_width_enabled,
&p_old_pi->link_width_enabled,
sizeof(p_pi->link_width_enabled)))
send_set = TRUE;
if (sm->p_subn->opt.force_link_speed &&
(sm->p_subn->opt.force_link_speed != 15 ||
ib_port_info_get_link_speed_enabled(p_pi) !=
ib_port_info_get_link_speed_sup(p_pi))) {
ib_port_info_set_link_speed_enabled(p_pi,
sm->p_subn->opt.
force_link_speed);
if (memcmp(&p_pi->link_speed, &p_old_pi->link_speed,
sizeof(p_pi->link_speed))) {
send_set = TRUE;
/* Determine whether QDR in LSE is being changed */
if ((ib_port_info_get_link_speed_enabled(p_pi) &
IB_LINK_SPEED_ACTIVE_10 &&
!(ib_port_info_get_link_speed_enabled(p_old_pi) &
IB_LINK_SPEED_ACTIVE_10)) ||
((!(ib_port_info_get_link_speed_enabled(p_pi) &
IB_LINK_SPEED_ACTIVE_10) &&
ib_port_info_get_link_speed_enabled(p_old_pi) &
IB_LINK_SPEED_ACTIVE_10)))
qdr_change = 1;
}
}
if (sm->p_subn->opt.fdr10 &&
p_physp->ext_port_info.link_speed_supported & FDR10) {
if (sm->p_subn->opt.fdr10 == 1) { /* enable */
if (!(p_physp->ext_port_info.link_speed_enabled & FDR10))
fdr10_change = 1;
} else { /* disable */
if (p_physp->ext_port_info.link_speed_enabled & FDR10)
fdr10_change = 1;
}
if (fdr10_change) {
p_old_epi = &p_physp->ext_port_info;
memcpy(payload2, p_old_epi,
sizeof(ib_mlnx_ext_port_info_t));
p_epi->state_change_enable = 0x01;
if (sm->p_subn->opt.fdr10 == 1)
p_epi->link_speed_enabled = FDR10;
else
p_epi->link_speed_enabled = 0;
send_set2 = TRUE;
}
}
if (osm_node_get_type(p_physp->p_node) == IB_NODE_TYPE_SWITCH) {
physp0 = osm_node_get_physp_ptr(p_physp->p_node, 0);
cap_mask = physp0->port_info.capability_mask;
} else
cap_mask = p_pi->capability_mask;
if (!(cap_mask & IB_PORT_CAP_HAS_EXT_SPEEDS))
qdr_change = 0;
/* Do peer ports support extended link speeds ? */
if (port_num != 0 && p_remote_physp) {
osm_physp_t *rphysp0;
ib_net32_t rem_cap_mask;
if (osm_node_get_type(p_remote_physp->p_node) ==
IB_NODE_TYPE_SWITCH) {
rphysp0 = osm_node_get_physp_ptr(p_remote_physp->p_node, 0);
rem_cap_mask = rphysp0->port_info.capability_mask;
} else
rem_cap_mask = p_remote_physp->port_info.capability_mask;
if (cap_mask & IB_PORT_CAP_HAS_EXT_SPEEDS &&
rem_cap_mask & IB_PORT_CAP_HAS_EXT_SPEEDS) {
if (sm->p_subn->opt.force_link_speed_ext &&
(sm->p_subn->opt.force_link_speed_ext != IB_LINK_SPEED_EXT_SET_LSES ||
p_pi->link_speed_ext_enabled !=
ib_port_info_get_link_speed_sup(p_pi))) {
p_pi->link_speed_ext_enabled = sm->p_subn->opt.force_link_speed_ext;
if (memcmp(&p_pi->link_speed_ext_enabled,
&p_old_pi->link_speed_ext_enabled,
sizeof(p_pi->link_speed_ext_enabled)))
send_set = TRUE;
}
}
}
/* calc new op_vls and mtu */
op_vls =
osm_physp_calc_link_op_vls(sm->p_log, sm->p_subn, p_physp);
mtu = osm_physp_calc_link_mtu(sm->p_log, p_physp);
ib_port_info_set_neighbor_mtu(p_pi, mtu);
if (ib_port_info_get_neighbor_mtu(p_pi) !=
ib_port_info_get_neighbor_mtu(p_old_pi))
send_set = TRUE;
ib_port_info_set_op_vls(p_pi, op_vls);
if (ib_port_info_get_op_vls(p_pi) !=
ib_port_info_get_op_vls(p_old_pi))
send_set = TRUE;
/* provide the vl_high_limit from the qos mgr */
if (sm->p_subn->opt.qos &&
p_physp->vl_high_limit != p_old_pi->vl_high_limit) {
send_set = TRUE;
p_pi->vl_high_limit = p_physp->vl_high_limit;
}
}
Send:
if (port_state != IB_LINK_NO_CHANGE &&
port_state != ib_port_info_get_port_state(p_old_pi)) {
send_set = TRUE;
if (port_state == IB_LINK_ACTIVE)
context.pi_context.active_transition = TRUE;
else
context.pi_context.active_transition = FALSE;
}
context.pi_context.node_guid = osm_node_get_node_guid(p_node);
context.pi_context.port_guid = osm_physp_get_port_guid(p_physp);
context.pi_context.set_method = TRUE;
context.pi_context.light_sweep = FALSE;
/* We need to send the PortInfoSet request with the new sm_lid
in the following cases:
1. There is a change in the values (send_set == TRUE)
2. This is a switch external port (so it wasn't handled yet by
osm_lid_mgr) and first_time_master_sweep flag on the subnet is TRUE,
which means the SM just became master, and it then needs to send at
PortInfoSet to every port.
*/
if (osm_node_get_type(p_node) == IB_NODE_TYPE_SWITCH && port_num
&& sm->p_subn->first_time_master_sweep == TRUE)
send_set = TRUE;
if (!send_set)
goto Exit;
attr_mod = cl_hton32(port_num);
if (qdr_change)
attr_mod |= cl_hton32(1 << 31); /* AM SMSupportExtendedSpeeds */
status = osm_req_set(sm, osm_physp_get_dr_path_ptr(p_physp),
payload, sizeof(payload), IB_MAD_ATTR_PORT_INFO,
attr_mod, CL_DISP_MSGID_NONE, &context);
if (status)
ret = -1;
if (send_set2) {
status = osm_req_set(sm, osm_physp_get_dr_path_ptr(p_physp),
payload2, sizeof(payload2),
IB_MAD_ATTR_MLNX_EXTENDED_PORT_INFO,
cl_hton32(port_num),
CL_DISP_MSGID_NONE, &context);
if (status)
ret = -1;
}
Exit:
OSM_LOG_EXIT(sm->p_log);
return ret;
}
/**********************************************************************
Sweeps the node 1 hop away.
This sets off a "chain reaction" that causes discovery of the subnet.
Used when there is suspicion that something on the subnet has changed.
**********************************************************************/
static ib_api_status_t state_mgr_sweep_hop_1(IN osm_sm_t * sm)
{
ib_api_status_t status = IB_SUCCESS;
osm_bind_handle_t h_bind;
osm_madw_context_t context;
osm_node_t *p_node;
osm_port_t *p_port;
osm_physp_t *p_physp;
osm_dr_path_t *p_dr_path;
osm_dr_path_t hop_1_path;
ib_net64_t port_guid;
uint8_t port_num;
uint8_t path_array[IB_SUBNET_PATH_HOPS_MAX];
uint8_t num_ports;
osm_physp_t *p_ext_physp;
OSM_LOG_ENTER(sm->p_log);
/*
* First, get our own port and node objects.
*/
port_guid = sm->p_subn->sm_port_guid;
CL_ASSERT(port_guid);
/* Set the in_sweep_hop_0 flag in subn to be FALSE.
* This will indicate the sweeping to continue beyond the
* the current node.
* This is relevant for the case of SM on switch, since in the
* switch info we need to signal that the sweeping should
* continue through the switch. */
sm->p_subn->in_sweep_hop_0 = FALSE;
p_port = osm_get_port_by_guid(sm->p_subn, port_guid);
if (!p_port) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3310: "
"No SM port object\n");
status = IB_ERROR;
goto Exit;
}
p_node = p_port->p_node;
CL_ASSERT(p_node);
port_num = ib_node_info_get_local_port_num(&p_node->node_info);
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"Probing hop 1 on local port %u\n", port_num);
p_physp = osm_node_get_physp_ptr(p_node, port_num);
CL_ASSERT(p_physp);
p_dr_path = osm_physp_get_dr_path_ptr(p_physp);
h_bind = osm_dr_path_get_bind_handle(p_dr_path);
CL_ASSERT(h_bind != OSM_BIND_INVALID_HANDLE);
memset(path_array, 0, sizeof(path_array));
/* the hop_1 operations depend on the type of our node.
* Currently - legal nodes that can host SM are SW and CA */
switch (osm_node_get_type(p_node)) {
case IB_NODE_TYPE_CA:
case IB_NODE_TYPE_ROUTER:
memset(&context, 0, sizeof(context));
context.ni_context.node_guid = osm_node_get_node_guid(p_node);
context.ni_context.port_num = port_num;
path_array[1] = port_num;
osm_dr_path_init(&hop_1_path, h_bind, 1, path_array);
status = osm_req_get(sm, &hop_1_path, IB_MAD_ATTR_NODE_INFO, 0,
CL_DISP_MSGID_NONE, &context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3311: "
"Request for NodeInfo failed (%s)\n",
ib_get_err_str(status));
break;
case IB_NODE_TYPE_SWITCH:
/* Need to go over all the ports of the switch, and send a
* node_info from them. This doesn't include the port 0 of the
* switch, which hosts the SM.
* Note: We'll send another switchInfo on port 0, since if no
* ports are connected, we still want to get some response, and
* have the subnet come up.
*/
num_ports = osm_node_get_num_physp(p_node);
for (port_num = 1; port_num < num_ports; port_num++) {
/* go through the port only if the port is not DOWN */
p_ext_physp = osm_node_get_physp_ptr(p_node, port_num);
if (p_ext_physp && ib_port_info_get_port_state
(&(p_ext_physp->port_info)) > IB_LINK_DOWN) {
memset(&context, 0, sizeof(context));
context.ni_context.node_guid =
osm_node_get_node_guid(p_node);
context.ni_context.port_num = port_num;
path_array[1] = port_num;
osm_dr_path_init(&hop_1_path, h_bind, 1,
path_array);
status = osm_req_get(sm, &hop_1_path,
IB_MAD_ATTR_NODE_INFO, 0,
CL_DISP_MSGID_NONE,
&context);
if (status != IB_SUCCESS)
OSM_LOG(sm->p_log, OSM_LOG_ERROR,
"ERR 3312: "
"Request for NodeInfo failed (%s)\n",
ib_get_err_str(status));
}
}
break;
default:
OSM_LOG(sm->p_log, OSM_LOG_ERROR,
"ERR 3313: Unknown node type %d (%s)\n",
osm_node_get_type(p_node), p_node->print_desc);
}
Exit:
OSM_LOG_EXIT(sm->p_log);
return status;
}
/**********************************************************************
Initiates a lightweight sweep of the subnet.
Used during normal sweeps after the subnet is up.
**********************************************************************/
static ib_api_status_t state_mgr_light_sweep_start(IN osm_sm_t * sm)
{
ib_api_status_t status = IB_SUCCESS;
osm_bind_handle_t h_bind;
cl_qmap_t *p_sw_tbl;
cl_map_item_t *p_next;
osm_node_t *p_node;
osm_physp_t *p_physp;
uint8_t port_num;
OSM_LOG_ENTER(sm->p_log);
p_sw_tbl = &sm->p_subn->sw_guid_tbl;
/*
* First, get the bind handle.
*/
h_bind = osm_sm_mad_ctrl_get_bind_handle(&sm->mad_ctrl);
if (h_bind == OSM_BIND_INVALID_HANDLE) {
OSM_LOG(sm->p_log, OSM_LOG_DEBUG,
"No bound ports. Deferring sweep...\n");
status = IB_INVALID_STATE;
goto _exit;
}
OSM_LOG_MSG_BOX(sm->p_log, OSM_LOG_VERBOSE, "INITIATING LIGHT SWEEP");
CL_PLOCK_ACQUIRE(sm->p_lock);
cl_qmap_apply_func(p_sw_tbl, state_mgr_get_sw_info, sm);
CL_PLOCK_RELEASE(sm->p_lock);
CL_PLOCK_ACQUIRE(sm->p_lock);
cl_qmap_apply_func(&sm->p_subn->node_guid_tbl, state_mgr_get_node_desc,
sm);
CL_PLOCK_RELEASE(sm->p_lock);
/* now scan the list of physical ports that were not down but have no remote port */
CL_PLOCK_ACQUIRE(sm->p_lock);
p_next = cl_qmap_head(&sm->p_subn->node_guid_tbl);
while (p_next != cl_qmap_end(&sm->p_subn->node_guid_tbl)) {
p_node = (osm_node_t *) p_next;
p_next = cl_qmap_next(p_next);
for (port_num = 1; port_num < osm_node_get_num_physp(p_node);
port_num++) {
p_physp = osm_node_get_physp_ptr(p_node, port_num);
if (p_physp && (osm_physp_get_port_state(p_physp) !=
IB_LINK_DOWN)
&& !osm_physp_get_remote(p_physp)) {
OSM_LOG(sm->p_log, OSM_LOG_ERROR, "ERR 3315: "
"Unknown remote side for node 0x%016"
PRIx64
" (%s) port %u. Adding to light sweep sampling list\n",
cl_ntoh64(osm_node_get_node_guid
(p_node)),
p_node->print_desc, port_num);
osm_dump_dr_path(sm->p_log,
osm_physp_get_dr_path_ptr
(p_physp), OSM_LOG_ERROR);
state_mgr_get_remote_port_info(sm, p_physp);
}
}
}
cl_qmap_apply_func(&sm->p_subn->sm_guid_tbl, query_sm_info, sm);
CL_PLOCK_RELEASE(sm->p_lock);
_exit:
OSM_LOG_EXIT(sm->p_log);
return status;
}
