Viewing: mdt_mds.c
// SPDX-License-Identifier: GPL-2.0
/*
* Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
* Copyright (c) 2013, 2017, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
*
* Lustre Metadata Service Layer
*
* Author: Di Wang <di.wang@whamcloud.com>
*/
#define DEBUG_SUBSYSTEM S_MDS
#include <linux/module.h>
#include <obd_support.h>
/* struct ptlrpc_request */
#include <lustre_net.h>
/* struct obd_export */
#include <lustre_export.h>
/* struct obd_device */
#include <obd.h>
/* lu2dt_dev() */
#include <dt_object.h>
#include <lustre_mds.h>
#include "mdt_internal.h"
#include <lustre_quota.h>
#include <lustre_acl.h>
#include <lustre_nodemap.h>
#include <uapi/linux/lustre/lustre_param.h>
struct mds_device {
/* super-class */
struct md_device mds_md_dev;
struct ptlrpc_service *mds_regular_service;
struct ptlrpc_service *mds_readpage_service;
struct ptlrpc_service *mds_out_service;
struct ptlrpc_service *mds_mdsc_service;
struct ptlrpc_service *mds_mdss_service;
struct ptlrpc_service *mds_fld_service;
struct ptlrpc_service *mds_io_service;
struct mutex mds_health_mutex;
};
/*
* * Initialized in mds_mod_init().
* */
static unsigned long mds_num_threads;
module_param(mds_num_threads, ulong, 0444);
MODULE_PARM_DESC(mds_num_threads, "number of MDS service threads to start");
static unsigned int mds_cpu_bind = 1;
module_param(mds_cpu_bind, uint, 0444);
MODULE_PARM_DESC(mds_cpu_bind,
"bind MDS threads to particular CPU partitions");
static int mds_max_io_threads = 512;
module_param(mds_max_io_threads, int, 0444);
MODULE_PARM_DESC(mds_max_io_threads,
"maximum number of MDS IO service threads");
static unsigned int mds_io_cpu_bind = 1;
module_param(mds_io_cpu_bind, uint, 0444);
MODULE_PARM_DESC(mds_io_cpu_bind,
"bind MDS IO threads to particular CPU partitions");
static char *mds_io_num_cpts;
module_param(mds_io_num_cpts, charp, 0444);
MODULE_PARM_DESC(mds_io_num_cpts,
"CPU partitions MDS IO threads should run on");
static struct cfs_cpt_table *mdt_io_cptable;
static char *mds_num_cpts;
module_param(mds_num_cpts, charp, 0444);
MODULE_PARM_DESC(mds_num_cpts, "CPU partitions MDS threads should run on");
static unsigned long mds_rdpg_num_threads;
module_param(mds_rdpg_num_threads, ulong, 0444);
MODULE_PARM_DESC(mds_rdpg_num_threads,
"number of MDS readpage service threads to start");
static unsigned int mds_rdpg_cpu_bind = 1;
module_param(mds_rdpg_cpu_bind, uint, 0444);
MODULE_PARM_DESC(mds_rdpg_cpu_bind,
"bind MDS readpage threads to particular CPU partitions");
static char *mds_rdpg_num_cpts;
module_param(mds_rdpg_num_cpts, charp, 0444);
MODULE_PARM_DESC(mds_rdpg_num_cpts,
"CPU partitions MDS readpage threads should run on");
unsigned int mdt_enable_flr_ec;
module_param(mdt_enable_flr_ec, uint, 0644);
MODULE_PARM_DESC(mdt_enable_flr_ec,
"enable FLR EC connect flag, off by default");
/* device init/fini methods */
static void mds_stop_ptlrpc_service(struct mds_device *m)
{
ENTRY;
mutex_lock(&m->mds_health_mutex);
if (m->mds_regular_service != NULL) {
ptlrpc_unregister_service(m->mds_regular_service);
m->mds_regular_service = NULL;
}
if (m->mds_readpage_service != NULL) {
ptlrpc_unregister_service(m->mds_readpage_service);
m->mds_readpage_service = NULL;
}
if (m->mds_out_service != NULL) {
ptlrpc_unregister_service(m->mds_out_service);
m->mds_out_service = NULL;
}
if (m->mds_mdsc_service != NULL) {
ptlrpc_unregister_service(m->mds_mdsc_service);
m->mds_mdsc_service = NULL;
}
if (m->mds_mdss_service != NULL) {
ptlrpc_unregister_service(m->mds_mdss_service);
m->mds_mdss_service = NULL;
}
if (m->mds_fld_service != NULL) {
ptlrpc_unregister_service(m->mds_fld_service);
m->mds_fld_service = NULL;
}
if (m->mds_io_service != NULL) {
ptlrpc_unregister_service(m->mds_io_service);
m->mds_io_service = NULL;
}
mutex_unlock(&m->mds_health_mutex);
if (mdt_io_cptable != NULL) {
cfs_cpt_table_free(mdt_io_cptable);
mdt_io_cptable = NULL;
}
EXIT;
}
static int ldlm_enqueue_hpreq_check(struct ptlrpc_request *req)
{
struct ldlm_request *dlm_req;
int rc = 0;
ENTRY;
if ((lustre_msg_get_flags(req->rq_reqmsg) & (MSG_REPLAY|MSG_RESENT)) !=
MSG_RESENT)
RETURN(0);
req_capsule_init(&req->rq_pill, req, RCL_SERVER);
req_capsule_set(&req->rq_pill, &RQF_LDLM_ENQUEUE);
dlm_req = req_capsule_client_get(&req->rq_pill, &RMF_DLM_REQ);
if (dlm_req == NULL)
RETURN(-EFAULT);
if (dlm_req->lock_count > 0) {
struct ldlm_lock *lock;
lock = cfs_hash_lookup(req->rq_export->exp_lock_hash,
(void *)&dlm_req->lock_handle[0]);
DEBUG_REQ(D_RPCTRACE, req, "lock %p cookie 0x%llx",
lock, dlm_req->lock_handle[0].cookie);
if (lock != NULL) {
rc = lock->l_granted_mode == lock->l_req_mode;
if (rc)
LDLM_DEBUG(lock, "hpreq resend");
ldlm_lock_put(lock);
}
}
RETURN(rc);
}
static struct ptlrpc_hpreq_ops ldlm_enqueue_hpreq_ops = {
.hpreq_lock_match = NULL,
.hpreq_check = ldlm_enqueue_hpreq_check,
.hpreq_fini = NULL,
};
static int mds_hpreq_handler(struct ptlrpc_request *req)
{
if (lustre_msg_get_opc(req->rq_reqmsg) == LDLM_ENQUEUE)
req->rq_ops = &ldlm_enqueue_hpreq_ops;
else
ptlrpc_hpreq_handler(req);
return 0;
}
static int mds_start_ptlrpc_service(struct mds_device *m)
{
static struct ptlrpc_service_conf conf;
struct obd_device *obd = m->mds_md_dev.md_lu_dev.ld_obd;
nodemask_t *mask;
int rc = 0;
ENTRY;
conf = (typeof(conf)) {
.psc_name = LUSTRE_MDT_NAME,
.psc_watchdog_factor = MDT_SERVICE_WATCHDOG_FACTOR,
.psc_buf = {
.bc_nbufs = MDS_NBUFS,
.bc_buf_size = MDS_REG_BUFSIZE,
.bc_req_max_size = MDS_REG_MAXREQSIZE,
.bc_rep_max_size = MDS_REG_MAXREPSIZE,
.bc_req_portal = MDS_REQUEST_PORTAL,
.bc_rep_portal = MDC_REPLY_PORTAL,
},
/*
* We'd like to have a mechanism to set this on a per-device
* basis, but alas...
*/
.psc_thr = {
.tc_thr_name = LUSTRE_MDT_NAME,
.tc_thr_factor = MDS_THR_FACTOR,
.tc_nthrs_init = MDS_NTHRS_INIT,
.tc_nthrs_base = MDS_NTHRS_BASE,
.tc_nthrs_max = MDS_NTHRS_MAX,
.tc_nthrs_user = mds_num_threads,
.tc_cpu_bind = mds_cpu_bind,
/* LCT_DT_THREAD is required as MDT threads may scan
* all LDLM namespaces (including OFD-originated) to
* cancel LDLM locks */
.tc_ctx_tags = LCT_MD_THREAD | LCT_DT_THREAD,
},
.psc_cpt = {
.cc_pattern = mds_num_cpts,
.cc_affinity = true,
},
.psc_ops = {
.so_req_handler = tgt_request_handle,
.so_req_printer = target_print_req,
.so_hpreq_handler = mds_hpreq_handler,
},
};
m->mds_regular_service = ptlrpc_register_service(&conf, &obd->obd_kset,
obd->obd_debugfs_entry);
if (IS_ERR(m->mds_regular_service)) {
rc = PTR_ERR(m->mds_regular_service);
CERROR("failed to start regular mdt service: %d\n", rc);
m->mds_regular_service = NULL;
RETURN(rc);
}
/*
* readpage service configuration. Parameters have to be adjusted,
* ideally.
*/
memset(&conf, 0, sizeof(conf));
conf = (typeof(conf)) {
.psc_name = LUSTRE_MDT_NAME "_readpage",
.psc_watchdog_factor = MDT_SERVICE_WATCHDOG_FACTOR,
.psc_buf = {
.bc_nbufs = MDS_NBUFS,
.bc_buf_size = MDS_BUFSIZE,
.bc_req_max_size = MDS_MAXREQSIZE,
.bc_rep_max_size = MDS_MAXREPSIZE,
.bc_req_portal = MDS_READPAGE_PORTAL,
.bc_rep_portal = MDC_REPLY_PORTAL,
},
.psc_thr = {
.tc_thr_name = LUSTRE_MDT_NAME "_rdpg",
.tc_thr_factor = MDS_RDPG_THR_FACTOR,
.tc_nthrs_init = MDS_RDPG_NTHRS_INIT,
.tc_nthrs_base = MDS_RDPG_NTHRS_BASE,
.tc_nthrs_max = MDS_RDPG_NTHRS_MAX,
.tc_nthrs_user = mds_rdpg_num_threads,
.tc_cpu_bind = mds_rdpg_cpu_bind,
.tc_ctx_tags = LCT_MD_THREAD,
},
.psc_cpt = {
.cc_pattern = mds_rdpg_num_cpts,
.cc_affinity = true,
},
.psc_ops = {
.so_req_handler = tgt_request_handle,
.so_req_printer = target_print_req,
},
};
m->mds_readpage_service = ptlrpc_register_service(&conf, &obd->obd_kset,
obd->obd_debugfs_entry);
if (IS_ERR(m->mds_readpage_service)) {
rc = PTR_ERR(m->mds_readpage_service);
CERROR("failed to start readpage service: %d\n", rc);
m->mds_readpage_service = NULL;
GOTO(err_mds_svc, rc);
}
/* Object update service */
conf = (typeof(conf)) {
.psc_name = LUSTRE_MDT_NAME "_out",
.psc_watchdog_factor = MDT_SERVICE_WATCHDOG_FACTOR,
.psc_buf = {
.bc_nbufs = MDS_NBUFS,
.bc_buf_size = OUT_BUFSIZE,
.bc_req_max_size = OUT_MAXREQSIZE,
.bc_rep_max_size = OUT_MAXREPSIZE,
.bc_req_portal = OUT_PORTAL,
.bc_rep_portal = OSC_REPLY_PORTAL,
},
/*
* We'd like to have a mechanism to set this on a per-device
* basis, but alas...
*/
.psc_thr = {
.tc_thr_name = LUSTRE_MDT_NAME "_out",
.tc_thr_factor = MDS_THR_FACTOR,
.tc_nthrs_init = MDS_NTHRS_INIT,
.tc_nthrs_base = MDS_NTHRS_BASE,
.tc_nthrs_max = MDS_NTHRS_MAX,
.tc_nthrs_user = mds_num_threads,
.tc_cpu_bind = mds_cpu_bind,
.tc_ctx_tags = LCT_MD_THREAD |
LCT_DT_THREAD,
},
.psc_cpt = {
.cc_pattern = mds_num_cpts,
.cc_affinity = true,
},
.psc_ops = {
.so_req_handler = tgt_request_handle,
.so_req_printer = target_print_req,
.so_hpreq_handler = NULL,
},
};
m->mds_out_service = ptlrpc_register_service(&conf, &obd->obd_kset,
obd->obd_debugfs_entry);
if (IS_ERR(m->mds_out_service)) {
rc = PTR_ERR(m->mds_out_service);
CERROR("failed to start out service: %d\n", rc);
m->mds_out_service = NULL;
GOTO(err_mds_svc, rc);
}
/*
* sequence controller service configuration
*/
memset(&conf, 0, sizeof(conf));
conf = (typeof(conf)) {
.psc_name = LUSTRE_MDT_NAME "_seqs",
.psc_watchdog_factor = MDT_SERVICE_WATCHDOG_FACTOR,
.psc_buf = {
.bc_nbufs = MDS_NBUFS,
.bc_buf_size = SEQ_BUFSIZE,
.bc_req_max_size = SEQ_MAXREQSIZE,
.bc_rep_max_size = SEQ_MAXREPSIZE,
.bc_req_portal = SEQ_CONTROLLER_PORTAL,
.bc_rep_portal = MDC_REPLY_PORTAL,
},
.psc_thr = {
.tc_thr_name = LUSTRE_MDT_NAME "_seqs",
.tc_nthrs_init = MDS_OTHR_NTHRS_INIT,
.tc_nthrs_max = MDS_OTHR_NTHRS_MAX,
.tc_ctx_tags = LCT_MD_THREAD,
},
.psc_ops = {
.so_req_handler = tgt_request_handle,
.so_req_printer = target_print_req,
.so_hpreq_handler = NULL,
},
};
m->mds_mdsc_service = ptlrpc_register_service(&conf, &obd->obd_kset,
obd->obd_debugfs_entry);
if (IS_ERR(m->mds_mdsc_service)) {
rc = PTR_ERR(m->mds_mdsc_service);
CERROR("failed to start seq controller service: %d\n", rc);
m->mds_mdsc_service = NULL;
GOTO(err_mds_svc, rc);
}
/*
* metadata sequence server service configuration
*/
memset(&conf, 0, sizeof(conf));
conf = (typeof(conf)) {
.psc_name = LUSTRE_MDT_NAME "_seqm",
.psc_watchdog_factor = MDT_SERVICE_WATCHDOG_FACTOR,
.psc_buf = {
.bc_nbufs = MDS_NBUFS,
.bc_buf_size = SEQ_BUFSIZE,
.bc_req_max_size = SEQ_MAXREQSIZE,
.bc_rep_max_size = SEQ_MAXREPSIZE,
.bc_req_portal = SEQ_METADATA_PORTAL,
.bc_rep_portal = MDC_REPLY_PORTAL,
},
.psc_thr = {
.tc_thr_name = LUSTRE_MDT_NAME "_seqm",
.tc_nthrs_init = MDS_OTHR_NTHRS_INIT,
.tc_nthrs_max = MDS_OTHR_NTHRS_MAX,
.tc_ctx_tags = LCT_MD_THREAD | LCT_DT_THREAD
},
.psc_ops = {
.so_req_handler = tgt_request_handle,
.so_req_printer = target_print_req,
.so_hpreq_handler = NULL,
},
};
m->mds_mdss_service = ptlrpc_register_service(&conf, &obd->obd_kset,
obd->obd_debugfs_entry);
if (IS_ERR(m->mds_mdss_service)) {
rc = PTR_ERR(m->mds_mdss_service);
CERROR("failed to start metadata seq server service: %d\n", rc);
m->mds_mdss_service = NULL;
GOTO(err_mds_svc, rc);
}
/* FLD service start */
memset(&conf, 0, sizeof(conf));
conf = (typeof(conf)) {
.psc_name = LUSTRE_MDT_NAME "_fld",
.psc_watchdog_factor = MDT_SERVICE_WATCHDOG_FACTOR,
.psc_buf = {
.bc_nbufs = MDS_NBUFS,
.bc_buf_size = FLD_BUFSIZE,
.bc_req_max_size = FLD_MAXREQSIZE,
.bc_rep_max_size = FLD_MAXREPSIZE,
.bc_req_portal = FLD_REQUEST_PORTAL,
.bc_rep_portal = MDC_REPLY_PORTAL,
},
.psc_thr = {
.tc_thr_name = LUSTRE_MDT_NAME "_fld",
.tc_nthrs_init = MDS_OTHR_NTHRS_INIT,
.tc_nthrs_max = MDS_OTHR_NTHRS_MAX,
.tc_ctx_tags = LCT_DT_THREAD | LCT_MD_THREAD,
},
.psc_ops = {
.so_req_handler = tgt_request_handle,
.so_req_printer = target_print_req,
.so_hpreq_handler = NULL,
},
};
m->mds_fld_service = ptlrpc_register_service(&conf, &obd->obd_kset,
obd->obd_debugfs_entry);
if (IS_ERR(m->mds_fld_service)) {
rc = PTR_ERR(m->mds_fld_service);
CERROR("failed to start fld service: %d\n", rc);
m->mds_fld_service = NULL;
GOTO(err_mds_svc, rc);
}
mask = cfs_cpt_nodemask(cfs_cpt_tab, CFS_CPT_ANY);
/* event CPT feature is disabled in libcfs level by set partition
* number to 1, we still want to set node affinity for io service */
if (cfs_cpt_number(cfs_cpt_tab) == 1 && nodes_weight(*mask) > 1) {
int cpt = 0;
int i;
mdt_io_cptable = cfs_cpt_table_alloc(nodes_weight(*mask));
for_each_node_mask(i, *mask) {
if (mdt_io_cptable == NULL) {
CWARN("MDS failed to create CPT table\n");
break;
}
rc = cfs_cpt_set_node(mdt_io_cptable, cpt++, i);
if (!rc) {
CWARN("MDS Failed to set node %d for IO CPT table\n",
i);
cfs_cpt_table_free(mdt_io_cptable);
mdt_io_cptable = NULL;
break;
}
}
rc = 0;
}
memset(&conf, 0, sizeof(conf));
conf = (typeof(conf)) {
.psc_name = LUSTRE_MDT_NAME "_io",
.psc_watchdog_factor = MDT_SERVICE_WATCHDOG_FACTOR,
.psc_buf = {
.bc_nbufs = OST_NBUFS,
.bc_buf_size = OST_IO_BUFSIZE,
.bc_req_max_size = OST_IO_MAXREQSIZE,
.bc_rep_max_size = OST_IO_MAXREPSIZE,
.bc_req_portal = MDS_IO_PORTAL,
.bc_rep_portal = MDC_REPLY_PORTAL,
},
.psc_thr = {
.tc_thr_name = LUSTRE_MDT_NAME "_io",
.tc_thr_factor = OSS_THR_FACTOR,
.tc_nthrs_init = OSS_NTHRS_INIT,
.tc_nthrs_base = OSS_NTHRS_BASE,
.tc_nthrs_max = mds_max_io_threads,
.tc_nthrs_user = mds_num_threads,
.tc_cpu_bind = mds_io_cpu_bind,
.tc_ctx_tags = LCT_DT_THREAD | LCT_MD_THREAD,
},
.psc_cpt = {
.cc_cptable = mdt_io_cptable,
.cc_pattern = mdt_io_cptable == NULL ?
mds_io_num_cpts : NULL,
.cc_affinity = true,
},
.psc_ops = {
.so_thr_init = tgt_io_thread_init,
.so_thr_done = tgt_io_thread_done,
.so_req_handler = tgt_request_handle,
.so_req_printer = target_print_req,
.so_hpreq_handler = tgt_hpreq_handler,
},
};
m->mds_io_service = ptlrpc_register_service(&conf, &obd->obd_kset,
obd->obd_debugfs_entry);
if (IS_ERR(m->mds_io_service)) {
rc = PTR_ERR(m->mds_io_service);
CERROR("failed to start MDT I/O service: %d\n", rc);
m->mds_io_service = NULL;
GOTO(err_mds_svc, rc);
}
EXIT;
err_mds_svc:
if (rc)
mds_stop_ptlrpc_service(m);
return rc;
}
static inline struct mds_device *mds_dev(struct lu_device *d)
{
return container_of_safe(d, struct mds_device, mds_md_dev.md_lu_dev);
}
static struct lu_device *mds_device_fini(const struct lu_env *env,
struct lu_device *d)
{
struct mds_device *m = mds_dev(d);
struct obd_device *obd = d->ld_obd;
ENTRY;
mds_stop_ptlrpc_service(m);
lprocfs_obd_cleanup(obd);
RETURN(NULL);
}
static struct lu_device *mds_device_free(const struct lu_env *env,
struct lu_device *d)
{
struct mds_device *m = mds_dev(d);
ENTRY;
md_device_fini(&m->mds_md_dev);
OBD_FREE_PTR(m);
RETURN(NULL);
}
static struct lu_device *mds_device_alloc(const struct lu_env *env,
struct lu_device_type *t,
struct lustre_cfg *cfg)
{
struct mds_device *m;
struct obd_device *obd;
struct lu_device *l;
int rc;
OBD_ALLOC_PTR(m);
if (m == NULL)
return ERR_PTR(-ENOMEM);
md_device_init(&m->mds_md_dev, t);
l = &m->mds_md_dev.md_lu_dev;
obd = class_name2obd(lustre_cfg_string(cfg, 0));
LASSERT(obd != NULL);
l->ld_obd = obd;
/* set this lu_device to obd, because error handling need it */
obd->obd_lu_dev = l;
rc = lprocfs_obd_setup(obd, true);
if (rc != 0) {
mds_device_free(env, l);
l = ERR_PTR(rc);
return l;
}
mutex_init(&m->mds_health_mutex);
rc = mds_start_ptlrpc_service(m);
if (rc != 0) {
lprocfs_obd_cleanup(obd);
mds_device_free(env, l);
l = ERR_PTR(rc);
return l;
}
return l;
}
/* type constructor/destructor: mdt_type_init, mdt_type_fini */
LU_TYPE_INIT_FINI(mds, &mdt_thread_key);
static const struct lu_device_type_operations mds_device_type_ops = {
.ldto_init = mds_type_init,
.ldto_fini = mds_type_fini,
.ldto_start = mds_type_start,
.ldto_stop = mds_type_stop,
.ldto_device_alloc = mds_device_alloc,
.ldto_device_free = mds_device_free,
.ldto_device_fini = mds_device_fini
};
static struct lu_device_type mds_device_type = {
.ldt_tags = LU_DEVICE_MD,
.ldt_name = LUSTRE_MDS_NAME,
.ldt_ops = &mds_device_type_ops,
.ldt_ctx_tags = LCT_MD_THREAD
};
static int mds_health_check(const struct lu_env *env, struct obd_device *obd)
{
struct mds_device *mds = mds_dev(obd->obd_lu_dev);
int rc = 0;
mutex_lock(&mds->mds_health_mutex);
rc |= ptlrpc_service_health_check(mds->mds_regular_service);
rc |= ptlrpc_service_health_check(mds->mds_readpage_service);
rc |= ptlrpc_service_health_check(mds->mds_out_service);
rc |= ptlrpc_service_health_check(mds->mds_mdsc_service);
rc |= ptlrpc_service_health_check(mds->mds_mdss_service);
rc |= ptlrpc_service_health_check(mds->mds_fld_service);
rc |= ptlrpc_service_health_check(mds->mds_io_service);
mutex_unlock(&mds->mds_health_mutex);
return rc != 0 ? 1 : 0;
}
/* ioctls on obd dev */
static int mds_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
void *karg, void __user *uarg)
{
struct obd_device *obd = exp->exp_obd;
struct obd_ioctl_data *data;
bool dynamic = true;
int rc = 0;
ENTRY;
CDEBUG(D_IOCTL, "%s: cmd=%x len=%u karg=%pK uarg=%pK\n",
obd->obd_name, cmd, len, karg, uarg);
data = karg;
/* we only support nodemap ioctls, for now */
if (cmd != OBD_IOC_NODEMAP)
GOTO(out, rc = -EINVAL);
rc = server_iocontrol_nodemap(obd, data, &dynamic, NULL, NULL);
if (rc)
GOTO(out, rc);
out:
RETURN(rc);
}
static const struct obd_ops mds_obd_device_ops = {
.o_owner = THIS_MODULE,
.o_health_check = mds_health_check,
.o_iocontrol = mds_iocontrol,
};
int mds_mod_init(void)
{
return class_register_type(&mds_obd_device_ops, NULL, false,
LUSTRE_MDS_NAME, &mds_device_type);
}
void mds_mod_exit(void)
{
class_unregister_type(LUSTRE_MDS_NAME);
}
