/*
* Glimpse callback handler for all quota locks. This function extracts
* information from the glimpse request.
*
* \param lock - is the lock targeted by the glimpse
* \param data - is a pointer to the glimpse ptlrpc request
* \param req - is the glimpse request
* \param desc - is the glimpse descriptor describing the purpose of the glimpse
* request.
* \param lvb - is the pointer to the lvb in the reply buffer
*
* \retval 0 on success and \desc, \lvb & \arg point to a valid structures,
* appropriate error on failure
*/
static int qsd_common_glimpse_ast(struct ptlrpc_request *req,
struct ldlm_gl_lquota_desc **desc, void **lvb)
{
int rc;
ENTRY;
LASSERT(lustre_msg_get_opc(req->rq_reqmsg) == LDLM_GL_CALLBACK);
/* glimpse on quota locks always packs a glimpse descriptor */
req_capsule_extend(&req->rq_pill, &RQF_LDLM_GL_DESC_CALLBACK);
/* extract glimpse descriptor */
*desc = req_capsule_client_get(&req->rq_pill, &RMF_DLM_GL_DESC);
if (*desc == NULL)
RETURN(-EFAULT);
/* prepare reply */
req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
sizeof(struct lquota_lvb));
rc = req_capsule_server_pack(&req->rq_pill);
if (rc != 0) {
CERROR("Can't pack response, rc %d\n", rc);
RETURN(rc);
}
/* extract lvb */
*lvb = req_capsule_server_get(&req->rq_pill, &RMF_DLM_LVB);
RETURN(0);
}
/**
* Send request reply from request \a req reply buffer.
* \a flags defines reply types
* Returns 0 on success or error code
*/
int ptlrpc_send_reply(struct ptlrpc_request *req, int flags)
{
struct ptlrpc_reply_state *rs = req->rq_reply_state;
struct ptlrpc_connection *conn;
int rc;
/* We must already have a reply buffer (only ptlrpc_error() may be
* called without one). The reply generated by sptlrpc layer (e.g.
* error notify, etc.) might have NULL rq->reqmsg; Otherwise we must
* have a request buffer which is either the actual (swabbed) incoming
* request, or a saved copy if this is a req saved in
* target_queue_final_reply().
*/
LASSERT(req->rq_no_reply == 0);
LASSERT(req->rq_reqbuf != NULL);
LASSERT(rs != NULL);
LASSERT((flags & PTLRPC_REPLY_MAYBE_DIFFICULT) || !rs->rs_difficult);
LASSERT(req->rq_repmsg != NULL);
LASSERT(req->rq_repmsg == rs->rs_msg);
LASSERT(rs->rs_cb_id.cbid_fn == reply_out_callback);
LASSERT(rs->rs_cb_id.cbid_arg == rs);
/* There may be no rq_export during failover */
if (unlikely(req->rq_export && req->rq_export->exp_obd &&
req->rq_export->exp_obd->obd_fail)) {
/* Failed obd's only send ENODEV */
req->rq_type = PTL_RPC_MSG_ERR;
req->rq_status = -ENODEV;
CDEBUG(D_HA, "sending ENODEV from failed obd %d\n",
req->rq_export->exp_obd->obd_minor);
}
/* In order to keep interoperability with the client (< 2.3) which
* doesn't have pb_jobid in ptlrpc_body, We have to shrink the
* ptlrpc_body in reply buffer to ptlrpc_body_v2, otherwise, the
* reply buffer on client will be overflow.
*
* XXX Remove this whenever we drop the interoperability with
* such client.
*/
req->rq_replen = lustre_shrink_msg(req->rq_repmsg, 0,
sizeof(struct ptlrpc_body_v2), 1);
if (req->rq_type != PTL_RPC_MSG_ERR)
req->rq_type = PTL_RPC_MSG_REPLY;
lustre_msg_set_type(req->rq_repmsg, req->rq_type);
lustre_msg_set_status(req->rq_repmsg,
ptlrpc_status_hton(req->rq_status));
lustre_msg_set_opc(req->rq_repmsg,
req->rq_reqmsg ? lustre_msg_get_opc(req->rq_reqmsg) : 0);
target_pack_pool_reply(req);
ptlrpc_at_set_reply(req, flags);
if (req->rq_export == NULL || req->rq_export->exp_connection == NULL)
conn = ptlrpc_connection_get(req->rq_peer, req->rq_self, NULL);
else
conn = ptlrpc_connection_addref(req->rq_export->exp_connection);
if (unlikely(conn == NULL)) {
CERROR("not replying on NULL connection\n"); /* bug 9635 */
return -ENOTCONN;
}
ptlrpc_rs_addref(rs); /* +1 ref for the network */
rc = sptlrpc_svc_wrap_reply(req);
if (unlikely(rc))
goto out;
req->rq_sent = ktime_get_real_seconds();
rc = ptl_send_buf(&rs->rs_md_h, rs->rs_repbuf, rs->rs_repdata_len,
(rs->rs_difficult && !rs->rs_no_ack) ?
LNET_ACK_REQ : LNET_NOACK_REQ,
&rs->rs_cb_id, conn,
ptlrpc_req2svc(req)->srv_rep_portal,
req->rq_xid, req->rq_reply_off);
out:
if (unlikely(rc != 0))
ptlrpc_req_drop_rs(req);
ptlrpc_connection_put(conn);
return rc;
}
/* TODO: handle requests in a similar way as MDT: see mdt_handle_common() */
static int ldlm_callback_handler(struct ptlrpc_request *req)
{
struct ldlm_namespace *ns;
struct ldlm_request *dlm_req;
struct ldlm_lock *lock;
int rc;
/* Requests arrive in sender's byte order. The ptlrpc service
* handler has already checked and, if necessary, byte-swapped the
* incoming request message body, but I am responsible for the
* message buffers. */
/* do nothing for sec context finalize */
if (lustre_msg_get_opc(req->rq_reqmsg) == SEC_CTX_FINI)
return 0;
req_capsule_init(&req->rq_pill, req, RCL_SERVER);
if (req->rq_export == NULL) {
rc = ldlm_callback_reply(req, -ENOTCONN);
ldlm_callback_errmsg(req, "Operate on unconnected server",
rc, NULL);
return 0;
}
LASSERT(req->rq_export != NULL);
LASSERT(req->rq_export->exp_obd != NULL);
switch (lustre_msg_get_opc(req->rq_reqmsg)) {
case LDLM_BL_CALLBACK:
if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_BL_CALLBACK_NET))
return 0;
break;
case LDLM_CP_CALLBACK:
if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_CP_CALLBACK_NET))
return 0;
break;
case LDLM_GL_CALLBACK:
if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_GL_CALLBACK_NET))
return 0;
break;
case LDLM_SET_INFO:
rc = ldlm_handle_setinfo(req);
ldlm_callback_reply(req, rc);
return 0;
case OBD_LOG_CANCEL: /* remove this eventually - for 1.4.0 compat */
CERROR("shouldn't be handling OBD_LOG_CANCEL on DLM thread\n");
req_capsule_set(&req->rq_pill, &RQF_LOG_CANCEL);
if (OBD_FAIL_CHECK(OBD_FAIL_OBD_LOG_CANCEL_NET))
return 0;
rc = llog_origin_handle_cancel(req);
if (OBD_FAIL_CHECK(OBD_FAIL_OBD_LOG_CANCEL_REP))
return 0;
ldlm_callback_reply(req, rc);
return 0;
case LLOG_ORIGIN_HANDLE_CREATE:
req_capsule_set(&req->rq_pill, &RQF_LLOG_ORIGIN_HANDLE_CREATE);
if (OBD_FAIL_CHECK(OBD_FAIL_OBD_LOGD_NET))
return 0;
rc = llog_origin_handle_open(req);
ldlm_callback_reply(req, rc);
return 0;
case LLOG_ORIGIN_HANDLE_NEXT_BLOCK:
req_capsule_set(&req->rq_pill,
&RQF_LLOG_ORIGIN_HANDLE_NEXT_BLOCK);
if (OBD_FAIL_CHECK(OBD_FAIL_OBD_LOGD_NET))
return 0;
rc = llog_origin_handle_next_block(req);
ldlm_callback_reply(req, rc);
return 0;
case LLOG_ORIGIN_HANDLE_READ_HEADER:
req_capsule_set(&req->rq_pill,
&RQF_LLOG_ORIGIN_HANDLE_READ_HEADER);
if (OBD_FAIL_CHECK(OBD_FAIL_OBD_LOGD_NET))
return 0;
rc = llog_origin_handle_read_header(req);
ldlm_callback_reply(req, rc);
return 0;
case LLOG_ORIGIN_HANDLE_CLOSE:
if (OBD_FAIL_CHECK(OBD_FAIL_OBD_LOGD_NET))
return 0;
rc = llog_origin_handle_close(req);
ldlm_callback_reply(req, rc);
return 0;
case OBD_QC_CALLBACK:
req_capsule_set(&req->rq_pill, &RQF_QC_CALLBACK);
if (OBD_FAIL_CHECK(OBD_FAIL_OBD_QC_CALLBACK_NET))
return 0;
rc = ldlm_handle_qc_callback(req);
ldlm_callback_reply(req, rc);
return 0;
default:
CERROR("unknown opcode %u\n",
lustre_msg_get_opc(req->rq_reqmsg));
ldlm_callback_reply(req, -EPROTO);
return 0;
}
ns = req->rq_export->exp_obd->obd_namespace;
LASSERT(ns != NULL);
req_capsule_set(&req->rq_pill, &RQF_LDLM_CALLBACK);
dlm_req = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
if (dlm_req == NULL) {
rc = ldlm_callback_reply(req, -EPROTO);
ldlm_callback_errmsg(req, "Operate without parameter", rc,
NULL);
return 0;
}
/* Force a known safe race, send a cancel to the server for a lock
* which the server has already started a blocking callback on. */
if (OBD_FAIL_CHECK(OBD_FAIL_LDLM_CANCEL_BL_CB_RACE) &&
lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK) {
rc = ldlm_cli_cancel(&dlm_req->lock_handle[0], 0);
if (rc < 0)
CERROR("ldlm_cli_cancel: %d\n", rc);
}
lock = ldlm_handle2lock_long(&dlm_req->lock_handle[0], 0);
if (!lock) {
CDEBUG(D_DLMTRACE, "callback on lock "LPX64" - lock "
"disappeared\n", dlm_req->lock_handle[0].cookie);
rc = ldlm_callback_reply(req, -EINVAL);
ldlm_callback_errmsg(req, "Operate with invalid parameter", rc,
&dlm_req->lock_handle[0]);
return 0;
}
if ((lock->l_flags & LDLM_FL_FAIL_LOC) &&
lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK)
OBD_RACE(OBD_FAIL_LDLM_CP_BL_RACE);
/* Copy hints/flags (e.g. LDLM_FL_DISCARD_DATA) from AST. */
lock_res_and_lock(lock);
lock->l_flags |= ldlm_flags_from_wire(dlm_req->lock_flags &
LDLM_AST_FLAGS);
if (lustre_msg_get_opc(req->rq_reqmsg) == LDLM_BL_CALLBACK) {
/* If somebody cancels lock and cache is already dropped,
* or lock is failed before cp_ast received on client,
* we can tell the server we have no lock. Otherwise, we
* should send cancel after dropping the cache. */
if (((lock->l_flags & LDLM_FL_CANCELING) &&
(lock->l_flags & LDLM_FL_BL_DONE)) ||
(lock->l_flags & LDLM_FL_FAILED)) {
LDLM_DEBUG(lock, "callback on lock "
LPX64" - lock disappeared\n",
dlm_req->lock_handle[0].cookie);
unlock_res_and_lock(lock);
LDLM_LOCK_RELEASE(lock);
rc = ldlm_callback_reply(req, -EINVAL);
ldlm_callback_errmsg(req, "Operate on stale lock", rc,
&dlm_req->lock_handle[0]);
return 0;
}
/* BL_AST locks are not needed in LRU.
* Let ldlm_cancel_lru() be fast. */
ldlm_lock_remove_from_lru(lock);
lock->l_flags |= LDLM_FL_BL_AST;
}
unlock_res_and_lock(lock);
/* We want the ost thread to get this reply so that it can respond
* to ost requests (write cache writeback) that might be triggered
* in the callback.
*
* But we'd also like to be able to indicate in the reply that we're
* cancelling right now, because it's unused, or have an intent result
* in the reply, so we might have to push the responsibility for sending
* the reply down into the AST handlers, alas. */
switch (lustre_msg_get_opc(req->rq_reqmsg)) {
case LDLM_BL_CALLBACK:
CDEBUG(D_INODE, "blocking ast\n");
req_capsule_extend(&req->rq_pill, &RQF_LDLM_BL_CALLBACK);
if (!(lock->l_flags & LDLM_FL_CANCEL_ON_BLOCK)) {
rc = ldlm_callback_reply(req, 0);
if (req->rq_no_reply || rc)
ldlm_callback_errmsg(req, "Normal process", rc,
&dlm_req->lock_handle[0]);
}
if (ldlm_bl_to_thread_lock(ns, &dlm_req->lock_desc, lock))
ldlm_handle_bl_callback(ns, &dlm_req->lock_desc, lock);
break;
case LDLM_CP_CALLBACK:
CDEBUG(D_INODE, "completion ast\n");
req_capsule_extend(&req->rq_pill, &RQF_LDLM_CP_CALLBACK);
ldlm_callback_reply(req, 0);
ldlm_handle_cp_callback(req, ns, dlm_req, lock);
break;
case LDLM_GL_CALLBACK:
CDEBUG(D_INODE, "glimpse ast\n");
req_capsule_extend(&req->rq_pill, &RQF_LDLM_GL_CALLBACK);
ldlm_handle_gl_callback(req, ns, dlm_req, lock);
break;
default:
LBUG(); /* checked above */
}
return 0;
}
/* TODO: handle requests in a similar way as MDT: see mdt_handle_common() */
int mgs_handle(struct ptlrpc_request *req)
{
int fail = OBD_FAIL_MGS_ALL_REPLY_NET;
int opc, rc = 0;
ENTRY;
req_capsule_init(&req->rq_pill, req, RCL_SERVER);
CFS_FAIL_TIMEOUT_MS(OBD_FAIL_MGS_PAUSE_REQ, cfs_fail_val);
if (CFS_FAIL_CHECK(OBD_FAIL_MGS_ALL_REQUEST_NET))
RETURN(0);
LASSERT(current->journal_info == NULL);
opc = lustre_msg_get_opc(req->rq_reqmsg);
if (opc == SEC_CTX_INIT ||
opc == SEC_CTX_INIT_CONT ||
opc == SEC_CTX_FINI)
GOTO(out, rc = 0);
if (opc != MGS_CONNECT) {
if (!class_connected_export(req->rq_export)) {
DEBUG_REQ(D_MGS, req, "operation on unconnected MGS\n");
req->rq_status = -ENOTCONN;
GOTO(out, rc = -ENOTCONN);
}
}
switch (opc) {
case MGS_CONNECT:
DEBUG_REQ(D_MGS, req, "connect");
/* MGS and MDS have same request format for connect */
req_capsule_set(&req->rq_pill, &RQF_MDS_CONNECT);
rc = target_handle_connect(req);
if (rc == 0)
rc = mgs_connect_check_sptlrpc(req);
if (!rc && (lustre_msg_get_conn_cnt(req->rq_reqmsg) > 1))
/* Make clients trying to reconnect after a MGS restart
happy; also requires obd_replayable */
lustre_msg_add_op_flags(req->rq_repmsg,
MSG_CONNECT_RECONNECT);
break;
case MGS_DISCONNECT:
DEBUG_REQ(D_MGS, req, "disconnect");
/* MGS and MDS have same request format for disconnect */
req_capsule_set(&req->rq_pill, &RQF_MDS_DISCONNECT);
rc = target_handle_disconnect(req);
req->rq_status = rc; /* superfluous? */
break;
case MGS_EXCEPTION:
DEBUG_REQ(D_MGS, req, "exception");
rc = mgs_handle_exception(req);
break;
case MGS_TARGET_REG:
DEBUG_REQ(D_MGS, req, "target add");
req_capsule_set(&req->rq_pill, &RQF_MGS_TARGET_REG);
rc = mgs_handle_target_reg(req);
break;
case MGS_TARGET_DEL:
DEBUG_REQ(D_MGS, req, "target del");
rc = mgs_handle_target_del(req);
break;
case MGS_SET_INFO:
DEBUG_REQ(D_MGS, req, "set_info");
req_capsule_set(&req->rq_pill, &RQF_MGS_SET_INFO);
rc = mgs_set_info_rpc(req);
break;
case MGS_CONFIG_READ:
DEBUG_REQ(D_MGS, req, "read config");
req_capsule_set(&req->rq_pill, &RQF_MGS_CONFIG_READ);
rc = mgs_config_read(req);
break;
case LDLM_ENQUEUE:
DEBUG_REQ(D_MGS, req, "enqueue");
req_capsule_set(&req->rq_pill, &RQF_LDLM_ENQUEUE);
rc = ldlm_handle_enqueue(req, ldlm_server_completion_ast,
ldlm_server_blocking_ast, NULL);
break;
case LDLM_BL_CALLBACK:
case LDLM_CP_CALLBACK:
DEBUG_REQ(D_MGS, req, "callback");
CERROR("callbacks should not happen on MGS\n");
LBUG();
break;
case OBD_PING:
DEBUG_REQ(D_INFO, req, "ping");
req_capsule_set(&req->rq_pill, &RQF_OBD_PING);
rc = target_handle_ping(req);
break;
case OBD_LOG_CANCEL:
DEBUG_REQ(D_MGS, req, "log cancel");
rc = -ENOTSUPP; /* la la la */
break;
case LLOG_ORIGIN_HANDLE_CREATE:
DEBUG_REQ(D_MGS, req, "llog_init");
req_capsule_set(&req->rq_pill, &RQF_LLOG_ORIGIN_HANDLE_CREATE);
rc = llog_origin_handle_create(req);
if (rc == 0)
(void)mgs_handle_fslog_hack(req);
break;
case LLOG_ORIGIN_HANDLE_NEXT_BLOCK:
DEBUG_REQ(D_MGS, req, "llog next block");
req_capsule_set(&req->rq_pill,
&RQF_LLOG_ORIGIN_HANDLE_NEXT_BLOCK);
rc = llog_origin_handle_next_block(req);
break;
case LLOG_ORIGIN_HANDLE_READ_HEADER:
DEBUG_REQ(D_MGS, req, "llog read header");
req_capsule_set(&req->rq_pill,
&RQF_LLOG_ORIGIN_HANDLE_READ_HEADER);
rc = llog_origin_handle_read_header(req);
break;
case LLOG_ORIGIN_HANDLE_CLOSE:
DEBUG_REQ(D_MGS, req, "llog close");
rc = llog_origin_handle_close(req);
break;
case LLOG_CATINFO:
DEBUG_REQ(D_MGS, req, "llog catinfo");
req_capsule_set(&req->rq_pill, &RQF_LLOG_CATINFO);
rc = llog_catinfo(req);
break;
default:
req->rq_status = -ENOTSUPP;
rc = ptlrpc_error(req);
RETURN(rc);
}
LASSERT(current->journal_info == NULL);
if (rc)
CERROR("MGS handle cmd=%d rc=%d\n", opc, rc);
out:
target_send_reply(req, rc, fail);
RETURN(0);
}
int mdt_handle_idmap(struct mdt_thread_info *info)
{
struct ptlrpc_request *req = mdt_info_req(info);
struct mdt_device *mdt = info->mti_mdt;
struct mdt_export_data *med;
struct ptlrpc_user_desc *pud = req->rq_user_desc;
struct md_identity *identity;
__u32 opc;
int rc = 0;
ENTRY;
if (!req->rq_export)
RETURN(0);
med = mdt_req2med(req);
if (!exp_connect_rmtclient(info->mti_exp))
RETURN(0);
opc = lustre_msg_get_opc(req->rq_reqmsg);
/* Bypass other opc */
if ((opc != SEC_CTX_INIT) && (opc != SEC_CTX_INIT_CONT) &&
(opc != SEC_CTX_FINI) && (opc != MDS_CONNECT))
RETURN(0);
LASSERT(med->med_idmap);
if (unlikely(!pud)) {
CDEBUG(D_SEC, "remote client must run with rq_user_desc "
"present\n");
RETURN(-EACCES);
}
if (req->rq_auth_mapped_uid == INVALID_UID) {
CDEBUG(D_SEC, "invalid authorized mapped uid, please check "
"/etc/lustre/idmap.conf!\n");
RETURN(-EACCES);
}
if (is_identity_get_disabled(mdt->mdt_identity_cache)) {
CDEBUG(D_SEC, "remote client must run with identity_get "
"enabled!\n");
RETURN(-EACCES);
}
identity = mdt_identity_get(mdt->mdt_identity_cache,
req->rq_auth_mapped_uid);
if (IS_ERR(identity)) {
CDEBUG(D_SEC, "can't get mdt identity(%u), no mapping added\n",
req->rq_auth_mapped_uid);
RETURN(-EACCES);
}
switch (opc) {
case SEC_CTX_INIT:
case SEC_CTX_INIT_CONT:
case MDS_CONNECT:
rc = lustre_idmap_add(med->med_idmap,
pud->pud_uid, identity->mi_uid,
pud->pud_gid, identity->mi_gid);
break;
case SEC_CTX_FINI:
rc = lustre_idmap_del(med->med_idmap,
pud->pud_uid, identity->mi_uid,
pud->pud_gid, identity->mi_gid);
break;
}
mdt_identity_put(mdt->mdt_identity_cache, identity);
if (rc)
RETURN(rc);
switch (opc) {
case SEC_CTX_INIT:
case SEC_CTX_INIT_CONT:
case SEC_CTX_FINI:
mdt_revoke_export_locks(req->rq_export);
break;
}
RETURN(0);
}
