- lnet_ni_t *ni;
-
- LNET_LOCK();
-
- cfs_list_for_each_entry (ni, &the_lnet.ln_nis, ni_list) {
- ni->ni_status = NULL;
- }
-
- LNET_UNLOCK();
-
- LIBCFS_FREE(the_lnet.ln_ping_info,
- offsetof(lnet_ping_info_t,
- pi_ni[the_lnet.ln_ping_info->pi_nnis]));
- the_lnet.ln_ping_info = NULL;
- return;
-}
-
-int
-lnet_ping_target_init(void)
-{
- lnet_md_t md = {0};
- lnet_handle_me_t meh;
- lnet_process_id_t id;
- int rc;
- int rc2;
- int infosz;
-
- rc = lnet_create_ping_info();
- if (rc != 0)
- return rc;
-
- /* We can have a tiny EQ since we only need to see the unlink event on
- * teardown, which by definition is the last one! */
- rc = LNetEQAlloc(2, LNET_EQ_HANDLER_NONE, &the_lnet.ln_ping_target_eq);
- if (rc != 0) {
- CERROR("Can't allocate ping EQ: %d\n", rc);
- goto failed_0;
- }
-
- memset(&id, 0, sizeof(lnet_process_id_t));
- id.nid = LNET_NID_ANY;
- id.pid = LNET_PID_ANY;
-
- rc = LNetMEAttach(LNET_RESERVED_PORTAL, id,
- LNET_PROTO_PING_MATCHBITS, 0,
- LNET_UNLINK, LNET_INS_AFTER,
- &meh);
- if (rc != 0) {
- CERROR("Can't create ping ME: %d\n", rc);
- goto failed_1;
- }
-
- /* initialize md content */
- infosz = offsetof(lnet_ping_info_t,
- pi_ni[the_lnet.ln_ping_info->pi_nnis]);
- md.start = the_lnet.ln_ping_info;
- md.length = infosz;
- md.threshold = LNET_MD_THRESH_INF;
- md.max_size = 0;
- md.options = LNET_MD_OP_GET | LNET_MD_TRUNCATE |
- LNET_MD_MANAGE_REMOTE;
- md.user_ptr = NULL;
- md.eq_handle = the_lnet.ln_ping_target_eq;
-
- rc = LNetMDAttach(meh, md,
- LNET_RETAIN,
- &the_lnet.ln_ping_target_md);
- if (rc != 0) {
- CERROR("Can't attach ping MD: %d\n", rc);
- goto failed_2;
- }
-
- return 0;
-
- failed_2:
- rc2 = LNetMEUnlink(meh);
- LASSERT (rc2 == 0);
- failed_1:
- rc2 = LNetEQFree(the_lnet.ln_ping_target_eq);
- LASSERT (rc2 == 0);
- failed_0:
- lnet_destroy_ping_info();
- return rc;
-}
-
-void
-lnet_ping_target_fini(void)
-{
- lnet_event_t event;
- int rc;
- int which;
- int timeout_ms = 1000;
- cfs_sigset_t blocked = cfs_block_allsigs();
-
- LNetMDUnlink(the_lnet.ln_ping_target_md);
- /* NB md could be busy; this just starts the unlink */
-
- for (;;) {
- rc = LNetEQPoll(&the_lnet.ln_ping_target_eq, 1,
- timeout_ms, &event, &which);
-
- /* I expect overflow... */
- LASSERT (rc >= 0 || rc == -EOVERFLOW);
-
- if (rc == 0) {
- /* timed out: provide a diagnostic */
- CWARN("Still waiting for ping MD to unlink\n");
- timeout_ms *= 2;
- continue;
- }
-
- /* Got a valid event */
- if (event.unlinked)
- break;
- }
-
- rc = LNetEQFree(the_lnet.ln_ping_target_eq);
- LASSERT (rc == 0);
- lnet_destroy_ping_info();
- cfs_restore_sigs(blocked);
-}
-
-int
-lnet_ping (lnet_process_id_t id, int timeout_ms, lnet_process_id_t *ids, int n_ids)
-{
- lnet_handle_eq_t eqh;
- lnet_handle_md_t mdh;
- lnet_event_t event;
- lnet_md_t md = {0};
- int which;
- int unlinked = 0;
- int replied = 0;
- const int a_long_time = 60000; /* mS */
- int infosz = offsetof(lnet_ping_info_t, pi_ni[n_ids]);
- lnet_ping_info_t *info;
- lnet_process_id_t tmpid;
- int i;
- int nob;
- int rc;
- int rc2;
- cfs_sigset_t blocked;
-
- if (n_ids <= 0 ||
- id.nid == LNET_NID_ANY ||
- timeout_ms > 500000 || /* arbitrary limit! */
- n_ids > 20) /* arbitrary limit! */
- return -EINVAL;
-
- if (id.pid == LNET_PID_ANY)
- id.pid = LUSTRE_SRV_LNET_PID;
-
- LIBCFS_ALLOC(info, infosz);
- if (info == NULL)
- return -ENOMEM;
-
- /* NB 2 events max (including any unlink event) */
- rc = LNetEQAlloc(2, LNET_EQ_HANDLER_NONE, &eqh);
- if (rc != 0) {
- CERROR("Can't allocate EQ: %d\n", rc);
- goto out_0;
- }
-
- /* initialize md content */
- md.start = info;
- md.length = infosz;
- md.threshold = 2; /*GET/REPLY*/
- md.max_size = 0;
- md.options = LNET_MD_TRUNCATE;
- md.user_ptr = NULL;
- md.eq_handle = eqh;
-
- rc = LNetMDBind(md, LNET_UNLINK, &mdh);
- if (rc != 0) {
- CERROR("Can't bind MD: %d\n", rc);
- goto out_1;
- }
-
- rc = LNetGet(LNET_NID_ANY, mdh, id,
- LNET_RESERVED_PORTAL,
- LNET_PROTO_PING_MATCHBITS, 0);
-
- if (rc != 0) {
- /* Don't CERROR; this could be deliberate! */
-
- rc2 = LNetMDUnlink(mdh);
- LASSERT (rc2 == 0);
-
- /* NB must wait for the UNLINK event below... */
- unlinked = 1;
- timeout_ms = a_long_time;
- }
-
- do {
- /* MUST block for unlink to complete */
- if (unlinked)
- blocked = cfs_block_allsigs();
-
- rc2 = LNetEQPoll(&eqh, 1, timeout_ms, &event, &which);
-
- if (unlinked)
- cfs_restore_sigs(blocked);
-
- CDEBUG(D_NET, "poll %d(%d %d)%s\n", rc2,
- (rc2 <= 0) ? -1 : event.type,
- (rc2 <= 0) ? -1 : event.status,
- (rc2 > 0 && event.unlinked) ? " unlinked" : "");
-
- LASSERT (rc2 != -EOVERFLOW); /* can't miss anything */
-
- if (rc2 <= 0 || event.status != 0) {
- /* timeout or error */
- if (!replied && rc == 0)
- rc = (rc2 < 0) ? rc2 :
- (rc2 == 0) ? -ETIMEDOUT :
- event.status;
-
- if (!unlinked) {
- /* Ensure completion in finite time... */
- LNetMDUnlink(mdh);
- /* No assertion (racing with network) */
- unlinked = 1;
- timeout_ms = a_long_time;
- } else if (rc2 == 0) {
- /* timed out waiting for unlink */
- CWARN("ping %s: late network completion\n",
- libcfs_id2str(id));
- }
- } else if (event.type == LNET_EVENT_REPLY) {
- replied = 1;
- rc = event.mlength;
- }
-
- } while (rc2 <= 0 || !event.unlinked);
-
- if (!replied) {
- if (rc >= 0)
- CWARN("%s: Unexpected rc >= 0 but no reply!\n",
- libcfs_id2str(id));
- rc = -EIO;
- goto out_1;
- }
-
- nob = rc;
- LASSERT (nob >= 0 && nob <= infosz);
-
- rc = -EPROTO; /* if I can't parse... */
-
- if (nob < 8) {
- /* can't check magic/version */
- CERROR("%s: ping info too short %d\n",
- libcfs_id2str(id), nob);
- goto out_1;
- }
-
- if (info->pi_magic == __swab32(LNET_PROTO_PING_MAGIC)) {
- lnet_swap_pinginfo(info);
- } else if (info->pi_magic != LNET_PROTO_PING_MAGIC) {
- CERROR("%s: Unexpected magic %08x\n",
- libcfs_id2str(id), info->pi_magic);
- goto out_1;
- }
-
- if (info->pi_version != LNET_PROTO_PING_VERSION) {
- CERROR("%s: Unexpected version 0x%x\n",
- libcfs_id2str(id), info->pi_version);
- goto out_1;
- }
-
- if (nob < offsetof(lnet_ping_info_t, pi_ni[0])) {
- CERROR("%s: Short reply %d(%d min)\n", libcfs_id2str(id),
- nob, (int)offsetof(lnet_ping_info_t, pi_ni[0]));
- goto out_1;
- }
-
- if (info->pi_nnis < n_ids)
- n_ids = info->pi_nnis;
-
- if (nob < offsetof(lnet_ping_info_t, pi_ni[n_ids])) {
- CERROR("%s: Short reply %d(%d expected)\n", libcfs_id2str(id),
- nob, (int)offsetof(lnet_ping_info_t, pi_ni[n_ids]));
- goto out_1;
- }
-
- rc = -EFAULT; /* If I SEGV... */
-
- for (i = 0; i < n_ids; i++) {
- tmpid.pid = info->pi_pid;
- tmpid.nid = info->pi_ni[i].ns_nid;
-#ifdef __KERNEL__
- if (cfs_copy_to_user(&ids[i], &tmpid, sizeof(tmpid)))
- goto out_1;
-#else
- ids[i] = tmpid;
-#endif
- }
- rc = info->pi_nnis;
+ lnet_handle_eq_t eqh;
+ lnet_handle_md_t mdh;
+ lnet_event_t event;
+ lnet_md_t md = {0};
+ int which;
+ int unlinked = 0;
+ int replied = 0;
+ const int a_long_time = 60000; /* mS */
+ int infosz;
+ lnet_ping_info_t *info;
+ lnet_process_id_t tmpid;
+ int i;
+ int nob;
+ int rc;
+ int rc2;
+ sigset_t blocked;
+
+ infosz = offsetof(lnet_ping_info_t, pi_ni[n_ids]);
+
+ if (n_ids <= 0 ||
+ id.nid == LNET_NID_ANY ||
+ timeout_ms > 500000 || /* arbitrary limit! */
+ n_ids > 20) /* arbitrary limit! */
+ return -EINVAL;
+
+ if (id.pid == LNET_PID_ANY)
+ id.pid = LNET_PID_LUSTRE;
+
+ LIBCFS_ALLOC(info, infosz);
+ if (info == NULL)
+ return -ENOMEM;
+
+ /* NB 2 events max (including any unlink event) */
+ rc = LNetEQAlloc(2, LNET_EQ_HANDLER_NONE, &eqh);
+ if (rc != 0) {
+ CERROR("Can't allocate EQ: %d\n", rc);
+ goto out_0;
+ }
+
+ /* initialize md content */
+ md.start = info;
+ md.length = infosz;
+ md.threshold = 2; /*GET/REPLY*/
+ md.max_size = 0;
+ md.options = LNET_MD_TRUNCATE;
+ md.user_ptr = NULL;
+ md.eq_handle = eqh;
+
+ rc = LNetMDBind(md, LNET_UNLINK, &mdh);
+ if (rc != 0) {
+ CERROR("Can't bind MD: %d\n", rc);
+ goto out_1;
+ }
+
+ rc = LNetGet(LNET_NID_ANY, mdh, id,
+ LNET_RESERVED_PORTAL,
+ LNET_PROTO_PING_MATCHBITS, 0);
+
+ if (rc != 0) {
+ /* Don't CERROR; this could be deliberate! */
+
+ rc2 = LNetMDUnlink(mdh);
+ LASSERT(rc2 == 0);
+
+ /* NB must wait for the UNLINK event below... */
+ unlinked = 1;
+ timeout_ms = a_long_time;
+ }
+
+ do {
+ /* MUST block for unlink to complete */
+ if (unlinked)
+ blocked = cfs_block_allsigs();
+
+ rc2 = LNetEQPoll(&eqh, 1, timeout_ms, &event, &which);
+
+ if (unlinked)
+ cfs_restore_sigs(blocked);
+
+ CDEBUG(D_NET, "poll %d(%d %d)%s\n", rc2,
+ (rc2 <= 0) ? -1 : event.type,
+ (rc2 <= 0) ? -1 : event.status,
+ (rc2 > 0 && event.unlinked) ? " unlinked" : "");
+
+ LASSERT(rc2 != -EOVERFLOW); /* can't miss anything */
+
+ if (rc2 <= 0 || event.status != 0) {
+ /* timeout or error */
+ if (!replied && rc == 0)
+ rc = (rc2 < 0) ? rc2 :
+ (rc2 == 0) ? -ETIMEDOUT :
+ event.status;
+
+ if (!unlinked) {
+ /* Ensure completion in finite time... */
+ LNetMDUnlink(mdh);
+ /* No assertion (racing with network) */
+ unlinked = 1;
+ timeout_ms = a_long_time;
+ } else if (rc2 == 0) {
+ /* timed out waiting for unlink */
+ CWARN("ping %s: late network completion\n",
+ libcfs_id2str(id));
+ }
+ } else if (event.type == LNET_EVENT_REPLY) {
+ replied = 1;
+ rc = event.mlength;
+ }
+
+ } while (rc2 <= 0 || !event.unlinked);
+
+ if (!replied) {
+ if (rc >= 0)
+ CWARN("%s: Unexpected rc >= 0 but no reply!\n",
+ libcfs_id2str(id));
+ rc = -EIO;
+ goto out_1;
+ }
+
+ nob = rc;
+ LASSERT(nob >= 0 && nob <= infosz);
+
+ rc = -EPROTO; /* if I can't parse... */
+
+ if (nob < 8) {
+ /* can't check magic/version */
+ CERROR("%s: ping info too short %d\n",
+ libcfs_id2str(id), nob);
+ goto out_1;
+ }
+
+ if (info->pi_magic == __swab32(LNET_PROTO_PING_MAGIC)) {
+ lnet_swap_pinginfo(info);
+ } else if (info->pi_magic != LNET_PROTO_PING_MAGIC) {
+ CERROR("%s: Unexpected magic %08x\n",
+ libcfs_id2str(id), info->pi_magic);
+ goto out_1;
+ }
+
+ if ((info->pi_features & LNET_PING_FEAT_NI_STATUS) == 0) {
+ CERROR("%s: ping w/o NI status: 0x%x\n",
+ libcfs_id2str(id), info->pi_features);
+ goto out_1;
+ }
+
+ if (nob < offsetof(lnet_ping_info_t, pi_ni[0])) {
+ CERROR("%s: Short reply %d(%d min)\n", libcfs_id2str(id),
+ nob, (int)offsetof(lnet_ping_info_t, pi_ni[0]));
+ goto out_1;
+ }
+
+ if (info->pi_nnis < n_ids)
+ n_ids = info->pi_nnis;
+
+ if (nob < offsetof(lnet_ping_info_t, pi_ni[n_ids])) {
+ CERROR("%s: Short reply %d(%d expected)\n", libcfs_id2str(id),
+ nob, (int)offsetof(lnet_ping_info_t, pi_ni[n_ids]));
+ goto out_1;
+ }
+
+ rc = -EFAULT; /* If I SEGV... */
+
+ for (i = 0; i < n_ids; i++) {
+ tmpid.pid = info->pi_pid;
+ tmpid.nid = info->pi_ni[i].ns_nid;
+ if (copy_to_user(&ids[i], &tmpid, sizeof(tmpid)))
+ goto out_1;
+ }
+ rc = info->pi_nnis;