* Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2011, 2016, Intel Corporation.
+ * Copyright (c) 2011, 2017, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
*/
#define DEBUG_SUBSYSTEM S_OSC
+#include <linux/workqueue.h>
#include <libcfs/libcfs.h>
-
+#include <linux/falloc.h>
#include <lprocfs_status.h>
-#include <lustre_debug.h>
#include <lustre_dlm.h>
#include <lustre_fid.h>
#include <lustre_ha.h>
#include <uapi/linux/lustre/lustre_ioctl.h>
#include <lustre_net.h>
#include <lustre_obdo.h>
-#include <uapi/linux/lustre/lustre_param.h>
#include <obd.h>
#include <obd_cksum.h>
#include <obd_class.h>
#include <lustre_osc.h>
+#include <linux/falloc.h>
#include "osc_internal.h"
+#include <lnet/lnet_rdma.h>
atomic_t osc_pool_req_count;
unsigned int osc_reqpool_maxreqcount;
static unsigned int osc_reqpool_mem_max = 5;
module_param(osc_reqpool_mem_max, uint, 0444);
+static int osc_idle_timeout = 20;
+module_param(osc_idle_timeout, uint, 0644);
+
#define osc_grant_args osc_brw_async_args
struct osc_setattr_args {
}
static int osc_setattr_interpret(const struct lu_env *env,
- struct ptlrpc_request *req,
- struct osc_setattr_args *sa, int rc)
+ struct ptlrpc_request *req, void *args, int rc)
{
- struct ost_body *body;
- ENTRY;
+ struct osc_setattr_args *sa = args;
+ struct ost_body *body;
- if (rc != 0)
- GOTO(out, rc);
+ ENTRY;
- body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
- if (body == NULL)
- GOTO(out, rc = -EPROTO);
+ if (rc != 0)
+ GOTO(out, rc);
+
+ body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
+ if (body == NULL)
+ GOTO(out, rc = -EPROTO);
lustre_get_wire_obdo(&req->rq_import->imp_connect_data, sa->sa_oa,
&body->oa);
out:
- rc = sa->sa_upcall(sa->sa_cookie, rc);
- RETURN(rc);
+ rc = sa->sa_upcall(sa->sa_cookie, rc);
+ RETURN(rc);
}
int osc_setattr_async(struct obd_export *exp, struct obdo *oa,
/* Do not wait for response. */
ptlrpcd_add_req(req);
} else {
- req->rq_interpret_reply =
- (ptlrpc_interpterer_t)osc_setattr_interpret;
+ req->rq_interpret_reply = osc_setattr_interpret;
- CLASSERT(sizeof(*sa) <= sizeof(req->rq_async_args));
- sa = ptlrpc_req_async_args(req);
+ sa = ptlrpc_req_async_args(sa, req);
sa->sa_oa = oa;
sa->sa_upcall = upcall;
sa->sa_cookie = cookie;
- if (rqset == PTLRPCD_SET)
- ptlrpcd_add_req(req);
- else
- ptlrpc_set_add_req(rqset, req);
+ ptlrpc_set_add_req(rqset, req);
}
RETURN(0);
}
req->rq_interpret_reply = osc_ladvise_interpret;
- CLASSERT(sizeof(*la) <= sizeof(req->rq_async_args));
- la = ptlrpc_req_async_args(req);
+ la = ptlrpc_req_async_args(la, req);
la->la_oa = oa;
la->la_upcall = upcall;
la->la_cookie = cookie;
- if (rqset == PTLRPCD_SET)
- ptlrpcd_add_req(req);
- else
- ptlrpc_set_add_req(rqset, req);
+ ptlrpc_set_add_req(rqset, req);
RETURN(0);
}
ptlrpc_request_set_replen(req);
- req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_setattr_interpret;
- CLASSERT(sizeof(*sa) <= sizeof(req->rq_async_args));
- sa = ptlrpc_req_async_args(req);
+ req->rq_interpret_reply = osc_setattr_interpret;
+ sa = ptlrpc_req_async_args(sa, req);
sa->sa_oa = oa;
sa->sa_upcall = upcall;
sa->sa_cookie = cookie;
}
EXPORT_SYMBOL(osc_punch_send);
+/**
+ * osc_fallocate_base() - Handles fallocate request.
+ *
+ * @exp: Export structure
+ * @oa: Attributes passed to OSS from client (obdo structure)
+ * @upcall: Primary & supplementary group information
+ * @cookie: Exclusive identifier
+ * @rqset: Request list.
+ * @mode: Operation done on given range.
+ *
+ * osc_fallocate_base() - Handles fallocate requests only. Only block
+ * allocation or standard preallocate operation is supported currently.
+ * Other mode flags is not supported yet. ftruncate(2) or truncate(2)
+ * is supported via SETATTR request.
+ *
+ * Return: Non-zero on failure and O on success.
+ */
+int osc_fallocate_base(struct obd_export *exp, struct obdo *oa,
+ obd_enqueue_update_f upcall, void *cookie, int mode)
+{
+ struct ptlrpc_request *req;
+ struct osc_setattr_args *sa;
+ struct ost_body *body;
+ struct obd_import *imp = class_exp2cliimp(exp);
+ int rc;
+ ENTRY;
+
+ oa->o_falloc_mode = mode;
+ req = ptlrpc_request_alloc(class_exp2cliimp(exp),
+ &RQF_OST_FALLOCATE);
+ if (req == NULL)
+ RETURN(-ENOMEM);
+
+ rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_FALLOCATE);
+ if (rc != 0) {
+ ptlrpc_request_free(req);
+ RETURN(rc);
+ }
+
+ body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
+ LASSERT(body);
+
+ lustre_set_wire_obdo(&imp->imp_connect_data, &body->oa, oa);
+
+ ptlrpc_request_set_replen(req);
+
+ req->rq_interpret_reply = osc_setattr_interpret;
+ BUILD_BUG_ON(sizeof(*sa) > sizeof(req->rq_async_args));
+ sa = ptlrpc_req_async_args(sa, req);
+ sa->sa_oa = oa;
+ sa->sa_upcall = upcall;
+ sa->sa_cookie = cookie;
+
+ ptlrpcd_add_req(req);
+
+ RETURN(0);
+}
+EXPORT_SYMBOL(osc_fallocate_base);
+
static int osc_sync_interpret(const struct lu_env *env,
- struct ptlrpc_request *req,
- void *arg, int rc)
+ struct ptlrpc_request *req, void *args, int rc)
{
- struct osc_fsync_args *fa = arg;
- struct ost_body *body;
- struct cl_attr *attr = &osc_env_info(env)->oti_attr;
- unsigned long valid = 0;
- struct cl_object *obj;
+ struct osc_fsync_args *fa = args;
+ struct ost_body *body;
+ struct cl_attr *attr = &osc_env_info(env)->oti_attr;
+ unsigned long valid = 0;
+ struct cl_object *obj;
ENTRY;
if (rc != 0)
ptlrpc_request_set_replen(req);
req->rq_interpret_reply = osc_sync_interpret;
- CLASSERT(sizeof(*fa) <= sizeof(req->rq_async_args));
- fa = ptlrpc_req_async_args(req);
+ fa = ptlrpc_req_async_args(fa, req);
fa->fa_obj = obj;
fa->fa_oa = oa;
fa->fa_upcall = upcall;
fa->fa_cookie = cookie;
- if (rqset == PTLRPCD_SET)
- ptlrpcd_add_req(req);
- else
- ptlrpc_set_add_req(rqset, req);
+ ptlrpc_set_add_req(rqset, req);
RETURN (0);
}
}
static int osc_destroy_interpret(const struct lu_env *env,
- struct ptlrpc_request *req, void *data,
- int rc)
+ struct ptlrpc_request *req, void *args, int rc)
{
struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
atomic_dec(&cli->cl_destroy_in_flight);
wake_up(&cli->cl_destroy_waitq);
+
return 0;
}
struct client_obd *cli = &exp->exp_obd->u.cli;
struct ptlrpc_request *req;
struct ost_body *body;
- struct list_head cancels = LIST_HEAD_INIT(cancels);
+ LIST_HEAD(cancels);
int rc, count;
ENTRY;
req->rq_interpret_reply = osc_destroy_interpret;
if (!osc_can_send_destroy(cli)) {
- struct l_wait_info lwi = LWI_INTR(LWI_ON_SIGNAL_NOOP, NULL);
-
/*
* Wait until the number of on-going destroy RPCs drops
* under max_rpc_in_flight
*/
- rc = l_wait_event_exclusive(cli->cl_destroy_waitq,
- osc_can_send_destroy(cli), &lwi);
+ rc = l_wait_event_abortable_exclusive(
+ cli->cl_destroy_waitq,
+ osc_can_send_destroy(cli));
if (rc) {
ptlrpc_req_finished(req);
- RETURN(rc);
+ RETURN(-EINTR);
}
}
oa->o_valid |= bits;
spin_lock(&cli->cl_loi_list_lock);
- if (OCD_HAS_FLAG(&cli->cl_import->imp_connect_data, GRANT_PARAM))
+ if (cli->cl_ocd_grant_param)
oa->o_dirty = cli->cl_dirty_grant;
else
oa->o_dirty = cli->cl_dirty_pages << PAGE_SHIFT;
- if (unlikely(cli->cl_dirty_pages - cli->cl_dirty_transit >
- cli->cl_dirty_max_pages)) {
- CERROR("dirty %lu - %lu > dirty_max %lu\n",
- cli->cl_dirty_pages, cli->cl_dirty_transit,
+ if (unlikely(cli->cl_dirty_pages > cli->cl_dirty_max_pages)) {
+ CERROR("dirty %lu > dirty_max %lu\n",
+ cli->cl_dirty_pages,
cli->cl_dirty_max_pages);
oa->o_undirty = 0;
- } else if (unlikely(atomic_long_read(&obd_dirty_pages) -
- atomic_long_read(&obd_dirty_transit_pages) >
+ } else if (unlikely(atomic_long_read(&obd_dirty_pages) >
(long)(obd_max_dirty_pages + 1))) {
/* The atomic_read() allowing the atomic_inc() are
* not covered by a lock thus they may safely race and trip
* this CERROR() unless we add in a small fudge factor (+1). */
- CERROR("%s: dirty %ld - %ld > system dirty_max %ld\n",
+ CERROR("%s: dirty %ld > system dirty_max %ld\n",
cli_name(cli), atomic_long_read(&obd_dirty_pages),
- atomic_long_read(&obd_dirty_transit_pages),
obd_max_dirty_pages);
oa->o_undirty = 0;
} else if (unlikely(cli->cl_dirty_max_pages - cli->cl_dirty_pages >
oa->o_undirty = 0;
} else {
unsigned long nrpages;
+ unsigned long undirty;
nrpages = cli->cl_max_pages_per_rpc;
nrpages *= cli->cl_max_rpcs_in_flight + 1;
nrpages = max(nrpages, cli->cl_dirty_max_pages);
- oa->o_undirty = nrpages << PAGE_SHIFT;
- if (OCD_HAS_FLAG(&cli->cl_import->imp_connect_data,
- GRANT_PARAM)) {
+ undirty = nrpages << PAGE_SHIFT;
+ if (cli->cl_ocd_grant_param) {
int nrextents;
/* take extent tax into account when asking for more
* grant space */
- nrextents = (nrpages + cli->cl_max_extent_pages - 1) /
+ nrextents = (nrpages + cli->cl_max_extent_pages - 1) /
cli->cl_max_extent_pages;
- oa->o_undirty += nrextents * cli->cl_grant_extent_tax;
+ undirty += nrextents * cli->cl_grant_extent_tax;
}
+ /* Do not ask for more than OBD_MAX_GRANT - a margin for server
+ * to add extent tax, etc.
+ */
+ oa->o_undirty = min(undirty, OBD_MAX_GRANT &
+ ~(PTLRPC_MAX_BRW_SIZE * 4UL));
}
oa->o_grant = cli->cl_avail_grant + cli->cl_reserved_grant;
- oa->o_dropped = cli->cl_lost_grant;
- cli->cl_lost_grant = 0;
+ /* o_dropped AKA o_misc is 32 bits, but cl_lost_grant is 64 bits */
+ if (cli->cl_lost_grant > INT_MAX) {
+ CDEBUG(D_CACHE,
+ "%s: avoided o_dropped overflow: cl_lost_grant %lu\n",
+ cli_name(cli), cli->cl_lost_grant);
+ oa->o_dropped = INT_MAX;
+ } else {
+ oa->o_dropped = cli->cl_lost_grant;
+ }
+ cli->cl_lost_grant -= oa->o_dropped;
spin_unlock(&cli->cl_loi_list_lock);
- CDEBUG(D_CACHE, "dirty: %llu undirty: %u dropped %u grant: %llu\n",
- oa->o_dirty, oa->o_undirty, oa->o_dropped, oa->o_grant);
+ CDEBUG(D_CACHE, "%s: dirty: %llu undirty: %u dropped %u grant: %llu"
+ " cl_lost_grant %lu\n", cli_name(cli), oa->o_dirty,
+ oa->o_undirty, oa->o_dropped, oa->o_grant, cli->cl_lost_grant);
}
void osc_update_next_shrink(struct client_obd *cli)
CDEBUG(D_CACHE, "next time %lld to shrink grant\n",
cli->cl_next_shrink_grant);
}
+EXPORT_SYMBOL(osc_update_next_shrink);
static void __osc_update_grant(struct client_obd *cli, u64 grant)
{
}
}
+/**
+ * grant thread data for shrinking space.
+ */
+struct grant_thread_data {
+ struct list_head gtd_clients;
+ struct mutex gtd_mutex;
+ unsigned long gtd_stopped:1;
+};
+static struct grant_thread_data client_gtd;
+
static int osc_shrink_grant_interpret(const struct lu_env *env,
- struct ptlrpc_request *req,
- void *aa, int rc)
+ struct ptlrpc_request *req,
+ void *args, int rc)
{
- struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
- struct obdo *oa = ((struct osc_grant_args *)aa)->aa_oa;
- struct ost_body *body;
+ struct osc_grant_args *aa = args;
+ struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
+ struct ost_body *body;
- if (rc != 0) {
- __osc_update_grant(cli, oa->o_grant);
- GOTO(out, rc);
- }
+ if (rc != 0) {
+ __osc_update_grant(cli, aa->aa_oa->o_grant);
+ GOTO(out, rc);
+ }
- body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
- LASSERT(body);
- osc_update_grant(cli, body);
+ body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
+ LASSERT(body);
+ osc_update_grant(cli, body);
out:
- OBDO_FREE(oa);
- return rc;
+ OBD_SLAB_FREE_PTR(aa->aa_oa, osc_obdo_kmem);
+ aa->aa_oa = NULL;
+
+ return rc;
}
static void osc_shrink_grant_local(struct client_obd *cli, struct obdo *oa)
osc_announce_cached(cli, &body->oa, 0);
spin_lock(&cli->cl_loi_list_lock);
+ if (target_bytes >= cli->cl_avail_grant) {
+ /* available grant has changed since target calculation */
+ spin_unlock(&cli->cl_loi_list_lock);
+ GOTO(out_free, rc = 0);
+ }
body->oa.o_grant = cli->cl_avail_grant - target_bytes;
cli->cl_avail_grant = target_bytes;
spin_unlock(&cli->cl_loi_list_lock);
sizeof(*body), body, NULL);
if (rc != 0)
__osc_update_grant(cli, body->oa.o_grant);
+out_free:
OBD_FREE_PTR(body);
RETURN(rc);
}
{
time64_t next_shrink = client->cl_next_shrink_grant;
- if ((client->cl_import->imp_connect_data.ocd_connect_flags &
- OBD_CONNECT_GRANT_SHRINK) == 0)
- return 0;
+ if (client->cl_import == NULL)
+ return 0;
+
+ if (!OCD_HAS_FLAG(&client->cl_import->imp_connect_data, GRANT_SHRINK) ||
+ client->cl_import->imp_grant_shrink_disabled) {
+ osc_update_next_shrink(client);
+ return 0;
+ }
if (ktime_get_seconds() >= next_shrink - 5) {
/* Get the current RPC size directly, instead of going via:
return 0;
}
-static int osc_grant_shrink_grant_cb(struct timeout_item *item, void *data)
+#define GRANT_SHRINK_RPC_BATCH 100
+
+static struct delayed_work work;
+
+static void osc_grant_work_handler(struct work_struct *data)
{
- struct client_obd *client;
+ struct client_obd *cli;
+ int rpc_sent;
+ bool init_next_shrink = true;
+ time64_t next_shrink = ktime_get_seconds() + GRANT_SHRINK_INTERVAL;
- list_for_each_entry(client, &item->ti_obd_list, cl_grant_shrink_list) {
- if (osc_should_shrink_grant(client))
- osc_shrink_grant(client);
+ rpc_sent = 0;
+ mutex_lock(&client_gtd.gtd_mutex);
+ list_for_each_entry(cli, &client_gtd.gtd_clients,
+ cl_grant_chain) {
+ if (rpc_sent < GRANT_SHRINK_RPC_BATCH &&
+ osc_should_shrink_grant(cli)) {
+ osc_shrink_grant(cli);
+ rpc_sent++;
+ }
+
+ if (!init_next_shrink) {
+ if (cli->cl_next_shrink_grant < next_shrink &&
+ cli->cl_next_shrink_grant > ktime_get_seconds())
+ next_shrink = cli->cl_next_shrink_grant;
+ } else {
+ init_next_shrink = false;
+ next_shrink = cli->cl_next_shrink_grant;
+ }
+ }
+ mutex_unlock(&client_gtd.gtd_mutex);
+
+ if (client_gtd.gtd_stopped == 1)
+ return;
+
+ if (next_shrink > ktime_get_seconds()) {
+ time64_t delay = next_shrink - ktime_get_seconds();
+
+ schedule_delayed_work(&work, cfs_time_seconds(delay));
+ } else {
+ schedule_work(&work.work);
}
- return 0;
}
-static int osc_add_shrink_grant(struct client_obd *client)
+void osc_schedule_grant_work(void)
{
- int rc;
+ cancel_delayed_work_sync(&work);
+ schedule_work(&work.work);
+}
+EXPORT_SYMBOL(osc_schedule_grant_work);
+
+/**
+ * Start grant thread for returing grant to server for idle clients.
+ */
+static int osc_start_grant_work(void)
+{
+ client_gtd.gtd_stopped = 0;
+ mutex_init(&client_gtd.gtd_mutex);
+ INIT_LIST_HEAD(&client_gtd.gtd_clients);
+
+ INIT_DELAYED_WORK(&work, osc_grant_work_handler);
+ schedule_work(&work.work);
- rc = ptlrpc_add_timeout_client(client->cl_grant_shrink_interval,
- TIMEOUT_GRANT,
- osc_grant_shrink_grant_cb, NULL,
- &client->cl_grant_shrink_list);
- if (rc) {
- CERROR("add grant client %s error %d\n", cli_name(client), rc);
- return rc;
- }
- CDEBUG(D_CACHE, "add grant client %s\n", cli_name(client));
- osc_update_next_shrink(client);
return 0;
}
-static int osc_del_shrink_grant(struct client_obd *client)
+static void osc_stop_grant_work(void)
+{
+ client_gtd.gtd_stopped = 1;
+ cancel_delayed_work_sync(&work);
+}
+
+static void osc_add_grant_list(struct client_obd *client)
{
- return ptlrpc_del_timeout_client(&client->cl_grant_shrink_list,
- TIMEOUT_GRANT);
+ mutex_lock(&client_gtd.gtd_mutex);
+ list_add(&client->cl_grant_chain, &client_gtd.gtd_clients);
+ mutex_unlock(&client_gtd.gtd_mutex);
+}
+
+static void osc_del_grant_list(struct client_obd *client)
+{
+ if (list_empty(&client->cl_grant_chain))
+ return;
+
+ mutex_lock(&client_gtd.gtd_mutex);
+ list_del_init(&client->cl_grant_chain);
+ mutex_unlock(&client_gtd.gtd_mutex);
}
void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
spin_lock(&cli->cl_loi_list_lock);
cli->cl_avail_grant = ocd->ocd_grant;
if (cli->cl_import->imp_state != LUSTRE_IMP_EVICTED) {
- cli->cl_avail_grant -= cli->cl_reserved_grant;
+ unsigned long consumed = cli->cl_reserved_grant;
+
if (OCD_HAS_FLAG(ocd, GRANT_PARAM))
- cli->cl_avail_grant -= cli->cl_dirty_grant;
+ consumed += cli->cl_dirty_grant;
else
- cli->cl_avail_grant -=
- cli->cl_dirty_pages << PAGE_SHIFT;
+ consumed += cli->cl_dirty_pages << PAGE_SHIFT;
+ if (cli->cl_avail_grant < consumed) {
+ CERROR("%s: granted %ld but already consumed %ld\n",
+ cli_name(cli), cli->cl_avail_grant, consumed);
+ cli->cl_avail_grant = 0;
+ } else {
+ cli->cl_avail_grant -= consumed;
+ }
}
if (OCD_HAS_FLAG(ocd, GRANT_PARAM)) {
~chunk_mask) & chunk_mask;
/* determine maximum extent size, in #pages */
size = (u64)ocd->ocd_grant_max_blks << ocd->ocd_grant_blkbits;
- cli->cl_max_extent_pages = size >> PAGE_SHIFT;
- if (cli->cl_max_extent_pages == 0)
- cli->cl_max_extent_pages = 1;
+ cli->cl_max_extent_pages = (size >> PAGE_SHIFT) ?: 1;
+ cli->cl_ocd_grant_param = 1;
} else {
+ cli->cl_ocd_grant_param = 0;
cli->cl_grant_extent_tax = 0;
cli->cl_chunkbits = PAGE_SHIFT;
cli->cl_max_extent_pages = DT_MAX_BRW_PAGES;
}
spin_unlock(&cli->cl_loi_list_lock);
- CDEBUG(D_CACHE, "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld."
- "chunk bits: %d cl_max_extent_pages: %d\n",
- cli_name(cli),
- cli->cl_avail_grant, cli->cl_lost_grant, cli->cl_chunkbits,
- cli->cl_max_extent_pages);
+ CDEBUG(D_CACHE,
+ "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld. chunk bits: %d cl_max_extent_pages: %d\n",
+ cli_name(cli),
+ cli->cl_avail_grant, cli->cl_lost_grant, cli->cl_chunkbits,
+ cli->cl_max_extent_pages);
- if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT_SHRINK &&
- list_empty(&cli->cl_grant_shrink_list))
- osc_add_shrink_grant(cli);
+ if (OCD_HAS_FLAG(ocd, GRANT_SHRINK) && list_empty(&cli->cl_grant_chain))
+ osc_add_grant_list(cli);
}
EXPORT_SYMBOL(osc_init_grant);
/* return error if any niobuf was in error */
for (i = 0; i < niocount; i++) {
- if ((int)remote_rcs[i] < 0)
- return(remote_rcs[i]);
+ if ((int)remote_rcs[i] < 0) {
+ CDEBUG(D_INFO, "rc[%d]: %d req %p\n",
+ i, remote_rcs[i], req);
+ return remote_rcs[i];
+ }
if (remote_rcs[i] != 0) {
CDEBUG(D_INFO, "rc[%d] invalid (%d) req %p\n",
if (p1->flag != p2->flag) {
unsigned mask = ~(OBD_BRW_FROM_GRANT | OBD_BRW_NOCACHE |
OBD_BRW_SYNC | OBD_BRW_ASYNC |
- OBD_BRW_NOQUOTA | OBD_BRW_SOFT_SYNC);
+ OBD_BRW_NOQUOTA | OBD_BRW_SOFT_SYNC |
+ OBD_BRW_SYS_RESOURCE);
/* warn if we try to combine flags that we don't know to be
* safe to combine */
if (unlikely((p1->flag & mask) != (p2->flag & mask))) {
CWARN("Saw flags 0x%x and 0x%x in the same brw, please "
- "report this at https://jira.hpdd.intel.com/\n",
+ "report this at https://jira.whamcloud.com/\n",
p1->flag, p2->flag);
}
return 0;
return (p1->off + p1->count == p2->off);
}
-static u32 osc_checksum_bulk(int nob, size_t pg_count,
+#if IS_ENABLED(CONFIG_CRC_T10DIF)
+static int osc_checksum_bulk_t10pi(const char *obd_name, int nob,
+ size_t pg_count, struct brw_page **pga,
+ int opc, obd_dif_csum_fn *fn,
+ int sector_size,
+ u32 *check_sum, bool resend)
+{
+ struct ahash_request *req;
+ /* Used Adler as the default checksum type on top of DIF tags */
+ unsigned char cfs_alg = cksum_obd2cfs(OBD_CKSUM_T10_TOP);
+ struct page *__page;
+ unsigned char *buffer;
+ __u16 *guard_start;
+ unsigned int bufsize;
+ int guard_number;
+ int used_number = 0;
+ int used;
+ u32 cksum;
+ int rc = 0;
+ int i = 0;
+
+ LASSERT(pg_count > 0);
+
+ __page = alloc_page(GFP_KERNEL);
+ if (__page == NULL)
+ return -ENOMEM;
+
+ req = cfs_crypto_hash_init(cfs_alg, NULL, 0);
+ if (IS_ERR(req)) {
+ rc = PTR_ERR(req);
+ CERROR("%s: unable to initialize checksum hash %s: rc = %d\n",
+ obd_name, cfs_crypto_hash_name(cfs_alg), rc);
+ GOTO(out, rc);
+ }
+
+ buffer = kmap(__page);
+ guard_start = (__u16 *)buffer;
+ guard_number = PAGE_SIZE / sizeof(*guard_start);
+ CDEBUG(D_PAGE | (resend ? D_HA : 0),
+ "GRD tags per page=%u, resend=%u, bytes=%u, pages=%zu\n",
+ guard_number, resend, nob, pg_count);
+
+ while (nob > 0 && pg_count > 0) {
+ unsigned int count = pga[i]->count > nob ? nob : pga[i]->count;
+
+ /* corrupt the data before we compute the checksum, to
+ * simulate an OST->client data error */
+ if (unlikely(i == 0 && opc == OST_READ &&
+ OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE))) {
+ unsigned char *ptr = kmap(pga[i]->pg);
+ int off = pga[i]->off & ~PAGE_MASK;
+
+ memcpy(ptr + off, "bad1", min_t(typeof(nob), 4, nob));
+ kunmap(pga[i]->pg);
+ }
+
+ /*
+ * The left guard number should be able to hold checksums of a
+ * whole page
+ */
+ rc = obd_page_dif_generate_buffer(obd_name, pga[i]->pg,
+ pga[i]->off & ~PAGE_MASK,
+ count,
+ guard_start + used_number,
+ guard_number - used_number,
+ &used, sector_size,
+ fn);
+ if (unlikely(resend))
+ CDEBUG(D_PAGE | D_HA,
+ "pga[%u]: used %u off %llu+%u gen checksum: %*phN\n",
+ i, used, pga[i]->off & ~PAGE_MASK, count,
+ (int)(used * sizeof(*guard_start)),
+ guard_start + used_number);
+ if (rc)
+ break;
+
+ used_number += used;
+ if (used_number == guard_number) {
+ cfs_crypto_hash_update_page(req, __page, 0,
+ used_number * sizeof(*guard_start));
+ used_number = 0;
+ }
+
+ nob -= pga[i]->count;
+ pg_count--;
+ i++;
+ }
+ kunmap(__page);
+ if (rc)
+ GOTO(out, rc);
+
+ if (used_number != 0)
+ cfs_crypto_hash_update_page(req, __page, 0,
+ used_number * sizeof(*guard_start));
+
+ bufsize = sizeof(cksum);
+ cfs_crypto_hash_final(req, (unsigned char *)&cksum, &bufsize);
+
+ /* For sending we only compute the wrong checksum instead
+ * of corrupting the data so it is still correct on a redo */
+ if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
+ cksum++;
+
+ *check_sum = cksum;
+out:
+ __free_page(__page);
+ return rc;
+}
+#else /* !CONFIG_CRC_T10DIF */
+#define obd_dif_ip_fn NULL
+#define obd_dif_crc_fn NULL
+#define osc_checksum_bulk_t10pi(name, nob, pgc, pga, opc, fn, ssize, csum, re) \
+ -EOPNOTSUPP
+#endif /* CONFIG_CRC_T10DIF */
+
+static int osc_checksum_bulk(int nob, size_t pg_count,
struct brw_page **pga, int opc,
- enum cksum_types cksum_type)
+ enum cksum_types cksum_type,
+ u32 *cksum)
{
- u32 cksum;
int i = 0;
- struct cfs_crypto_hash_desc *hdesc;
+ struct ahash_request *req;
unsigned int bufsize;
unsigned char cfs_alg = cksum_obd2cfs(cksum_type);
LASSERT(pg_count > 0);
- hdesc = cfs_crypto_hash_init(cfs_alg, NULL, 0);
- if (IS_ERR(hdesc)) {
+ req = cfs_crypto_hash_init(cfs_alg, NULL, 0);
+ if (IS_ERR(req)) {
CERROR("Unable to initialize checksum hash %s\n",
cfs_crypto_hash_name(cfs_alg));
- return PTR_ERR(hdesc);
+ return PTR_ERR(req);
}
while (nob > 0 && pg_count > 0) {
memcpy(ptr + off, "bad1", min_t(typeof(nob), 4, nob));
kunmap(pga[i]->pg);
}
- cfs_crypto_hash_update_page(hdesc, pga[i]->pg,
+ cfs_crypto_hash_update_page(req, pga[i]->pg,
pga[i]->off & ~PAGE_MASK,
count);
LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d\n",
i++;
}
- bufsize = sizeof(cksum);
- cfs_crypto_hash_final(hdesc, (unsigned char *)&cksum, &bufsize);
+ bufsize = sizeof(*cksum);
+ cfs_crypto_hash_final(req, (unsigned char *)cksum, &bufsize);
/* For sending we only compute the wrong checksum instead
* of corrupting the data so it is still correct on a redo */
if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
- cksum++;
+ (*cksum)++;
- return cksum;
+ return 0;
+}
+
+static int osc_checksum_bulk_rw(const char *obd_name,
+ enum cksum_types cksum_type,
+ int nob, size_t pg_count,
+ struct brw_page **pga, int opc,
+ u32 *check_sum, bool resend)
+{
+ obd_dif_csum_fn *fn = NULL;
+ int sector_size = 0;
+ int rc;
+
+ ENTRY;
+ obd_t10_cksum2dif(cksum_type, &fn, §or_size);
+
+ if (fn)
+ rc = osc_checksum_bulk_t10pi(obd_name, nob, pg_count, pga,
+ opc, fn, sector_size, check_sum,
+ resend);
+ else
+ rc = osc_checksum_bulk(nob, pg_count, pga, opc, cksum_type,
+ check_sum);
+
+ RETURN(rc);
+}
+
+static inline void osc_release_bounce_pages(struct brw_page **pga,
+ u32 page_count)
+{
+#ifdef HAVE_LUSTRE_CRYPTO
+ int i;
+
+ for (i = 0; i < page_count; i++) {
+ /* Bounce pages allocated by a call to
+ * llcrypt_encrypt_pagecache_blocks() in osc_brw_prep_request()
+ * are identified thanks to the PageChecked flag.
+ */
+ if (PageChecked(pga[i]->pg))
+ llcrypt_finalize_bounce_page(&pga[i]->pg);
+ pga[i]->count -= pga[i]->bp_count_diff;
+ pga[i]->off += pga[i]->bp_off_diff;
+ }
+#endif
}
static int
u32 page_count, struct brw_page **pga,
struct ptlrpc_request **reqp, int resend)
{
- struct ptlrpc_request *req;
- struct ptlrpc_bulk_desc *desc;
- struct ost_body *body;
- struct obd_ioobj *ioobj;
- struct niobuf_remote *niobuf;
- int niocount, i, requested_nob, opc, rc, short_io_size;
- struct osc_brw_async_args *aa;
- struct req_capsule *pill;
- struct brw_page *pg_prev;
+ struct ptlrpc_request *req;
+ struct ptlrpc_bulk_desc *desc;
+ struct ost_body *body;
+ struct obd_ioobj *ioobj;
+ struct niobuf_remote *niobuf;
+ int niocount, i, requested_nob, opc, rc, short_io_size = 0;
+ struct osc_brw_async_args *aa;
+ struct req_capsule *pill;
+ struct brw_page *pg_prev;
void *short_io_buf;
+ const char *obd_name = cli->cl_import->imp_obd->obd_name;
+ struct inode *inode = NULL;
+ bool directio = false;
+ bool enable_checksum = true;
- ENTRY;
- if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
- RETURN(-ENOMEM); /* Recoverable */
- if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
- RETURN(-EINVAL); /* Fatal */
+ ENTRY;
+ if (pga[0]->pg) {
+ inode = page2inode(pga[0]->pg);
+ if (inode == NULL) {
+ /* Try to get reference to inode from cl_page if we are
+ * dealing with direct IO, as handled pages are not
+ * actual page cache pages.
+ */
+ struct osc_async_page *oap = brw_page2oap(pga[0]);
+ struct cl_page *clpage = oap2cl_page(oap);
+
+ inode = clpage->cp_inode;
+ if (inode)
+ directio = true;
+ }
+ }
+ if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
+ RETURN(-ENOMEM); /* Recoverable */
+ if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
+ RETURN(-EINVAL); /* Fatal */
if ((cmd & OBD_BRW_WRITE) != 0) {
opc = OST_WRITE;
if (req == NULL)
RETURN(-ENOMEM);
+ if (opc == OST_WRITE && inode && IS_ENCRYPTED(inode) &&
+ llcrypt_has_encryption_key(inode)) {
+ for (i = 0; i < page_count; i++) {
+ struct brw_page *pg = pga[i];
+ struct page *data_page = NULL;
+ bool retried = false;
+ bool lockedbymyself;
+ u32 nunits = (pg->off & ~PAGE_MASK) + pg->count;
+ struct address_space *map_orig = NULL;
+ pgoff_t index_orig;
+
+retry_encrypt:
+ nunits = round_up(nunits, LUSTRE_ENCRYPTION_UNIT_SIZE);
+ /* The page can already be locked when we arrive here.
+ * This is possible when cl_page_assume/vvp_page_assume
+ * is stuck on wait_on_page_writeback with page lock
+ * held. In this case there is no risk for the lock to
+ * be released while we are doing our encryption
+ * processing, because writeback against that page will
+ * end in vvp_page_completion_write/cl_page_completion,
+ * which means only once the page is fully processed.
+ */
+ lockedbymyself = trylock_page(pg->pg);
+ if (directio) {
+ map_orig = pg->pg->mapping;
+ pg->pg->mapping = inode->i_mapping;
+ index_orig = pg->pg->index;
+ pg->pg->index = pg->off >> PAGE_SHIFT;
+ }
+ data_page =
+ llcrypt_encrypt_pagecache_blocks(pg->pg,
+ nunits, 0,
+ GFP_NOFS);
+ if (directio) {
+ pg->pg->mapping = map_orig;
+ pg->pg->index = index_orig;
+ }
+ if (lockedbymyself)
+ unlock_page(pg->pg);
+ if (IS_ERR(data_page)) {
+ rc = PTR_ERR(data_page);
+ if (rc == -ENOMEM && !retried) {
+ retried = true;
+ rc = 0;
+ goto retry_encrypt;
+ }
+ ptlrpc_request_free(req);
+ RETURN(rc);
+ }
+ /* Set PageChecked flag on bounce page for
+ * disambiguation in osc_release_bounce_pages().
+ */
+ SetPageChecked(data_page);
+ pg->pg = data_page;
+ /* there should be no gap in the middle of page array */
+ if (i == page_count - 1) {
+ struct osc_async_page *oap = brw_page2oap(pg);
+
+ oa->o_size = oap->oap_count +
+ oap->oap_obj_off + oap->oap_page_off;
+ }
+ /* len is forced to nunits, and relative offset to 0
+ * so store the old, clear text info
+ */
+ pg->bp_count_diff = nunits - pg->count;
+ pg->count = nunits;
+ pg->bp_off_diff = pg->off & ~PAGE_MASK;
+ pg->off = pg->off & PAGE_MASK;
+ }
+ } else if (opc == OST_WRITE && inode && IS_ENCRYPTED(inode)) {
+ struct osc_async_page *oap = brw_page2oap(pga[0]);
+ struct cl_page *clpage = oap2cl_page(oap);
+ struct cl_object *clobj = clpage->cp_obj;
+ struct cl_attr attr = { 0 };
+ struct lu_env *env;
+ __u16 refcheck;
+
+ env = cl_env_get(&refcheck);
+ if (IS_ERR(env)) {
+ rc = PTR_ERR(env);
+ ptlrpc_request_free(req);
+ RETURN(rc);
+ }
+
+ cl_object_attr_lock(clobj);
+ rc = cl_object_attr_get(env, clobj, &attr);
+ cl_object_attr_unlock(clobj);
+ cl_env_put(env, &refcheck);
+ if (rc != 0) {
+ ptlrpc_request_free(req);
+ RETURN(rc);
+ }
+ if (attr.cat_size)
+ oa->o_size = attr.cat_size;
+ } else if (opc == OST_READ && inode && IS_ENCRYPTED(inode) &&
+ llcrypt_has_encryption_key(inode)) {
+ for (i = 0; i < page_count; i++) {
+ struct brw_page *pg = pga[i];
+ u32 nunits = (pg->off & ~PAGE_MASK) + pg->count;
+
+ nunits = round_up(nunits, LUSTRE_ENCRYPTION_UNIT_SIZE);
+ /* count/off are forced to cover the whole encryption
+ * unit size so that all encrypted data is stored on the
+ * OST, so adjust bp_{count,off}_diff for the size of
+ * the clear text.
+ */
+ pg->bp_count_diff = nunits - pg->count;
+ pg->count = nunits;
+ pg->bp_off_diff = pg->off & ~PAGE_MASK;
+ pg->off = pg->off & PAGE_MASK;
+ }
+ }
+
for (niocount = i = 1; i < page_count; i++) {
if (!can_merge_pages(pga[i - 1], pga[i]))
niocount++;
req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_CLIENT,
niocount * sizeof(*niobuf));
- for (i = 0; i < page_count; i++)
+ for (i = 0; i < page_count; i++) {
short_io_size += pga[i]->count;
+ if (!inode || !IS_ENCRYPTED(inode) ||
+ !llcrypt_has_encryption_key(inode)) {
+ pga[i]->bp_count_diff = 0;
+ pga[i]->bp_off_diff = 0;
+ }
+ }
+
+ if (brw_page2oap(pga[0])->oap_brw_flags & OBD_BRW_RDMA_ONLY) {
+ enable_checksum = false;
+ short_io_size = 0;
+ }
- /* Check if we can do a short io. */
- if (!(short_io_size <= cli->cl_short_io_bytes && niocount == 1 &&
- imp_connect_shortio(cli->cl_import)))
+ /* Check if read/write is small enough to be a short io. */
+ if (short_io_size > cli->cl_max_short_io_bytes || niocount > 1 ||
+ !imp_connect_shortio(cli->cl_import))
short_io_size = 0;
+ /* If this is an empty RPC to old server, just ignore it */
+ if (!short_io_size && !pga[0]->pg) {
+ ptlrpc_request_free(req);
+ RETURN(-ENODATA);
+ }
+
req_capsule_set_size(pill, &RMF_SHORT_IO, RCL_CLIENT,
opc == OST_READ ? 0 : short_io_size);
if (opc == OST_READ)
desc = ptlrpc_prep_bulk_imp(req, page_count,
cli->cl_import->imp_connect_data.ocd_brw_size >> LNET_MTU_BITS,
(opc == OST_WRITE ? PTLRPC_BULK_GET_SOURCE :
- PTLRPC_BULK_PUT_SINK) |
- PTLRPC_BULK_BUF_KIOV,
+ PTLRPC_BULK_PUT_SINK),
OST_BULK_PORTAL,
&ptlrpc_bulk_kiov_pin_ops);
lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
+ /* For READ and WRITE, we can't fill o_uid and o_gid using from_kuid()
+ * and from_kgid(), because they are asynchronous. Fortunately, variable
+ * oa contains valid o_uid and o_gid in these two operations.
+ * Besides, filling o_uid and o_gid is enough for nrs-tbf, see LU-9658.
+ * OBD_MD_FLUID and OBD_MD_FLUID is not set in order to avoid breaking
+ * other process logic */
+ body->oa.o_uid = oa->o_uid;
+ body->oa.o_gid = oa->o_gid;
+
obdo_to_ioobj(oa, ioobj);
ioobj->ioo_bufcnt = niocount;
/* The high bits of ioo_max_brw tells server _maximum_ number of bulks
LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
(pg->flag & OBD_BRW_SRVLOCK));
if (short_io_size != 0 && opc == OST_WRITE) {
- unsigned char *ptr = ll_kmap_atomic(pg->pg, KM_USER0);
+ unsigned char *ptr = kmap_atomic(pg->pg);
LASSERT(short_io_size >= requested_nob + pg->count);
memcpy(short_io_buf + requested_nob,
ptr + poff,
pg->count);
- ll_kunmap_atomic(ptr, KM_USER0);
+ kunmap_atomic(ptr);
} else if (short_io_size == 0) {
desc->bd_frag_ops->add_kiov_frag(desc, pg->pg, poff,
pg->count);
if (osc_should_shrink_grant(cli))
osc_shrink_grant_local(cli, &body->oa);
+ if (!cli->cl_checksum || sptlrpc_flavor_has_bulk(&req->rq_flvr))
+ enable_checksum = false;
+
/* size[REQ_REC_OFF] still sizeof (*body) */
if (opc == OST_WRITE) {
- if (cli->cl_checksum &&
- !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
+ if (enable_checksum) {
/* store cl_cksum_type in a local variable since
* it can be changed via lprocfs */
enum cksum_types cksum_type = cli->cl_cksum_type;
if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
body->oa.o_flags = 0;
- body->oa.o_flags |= cksum_type_pack(cksum_type);
+ body->oa.o_flags |= obd_cksum_type_pack(obd_name,
+ cksum_type);
body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
- body->oa.o_cksum = osc_checksum_bulk(requested_nob,
- page_count, pga,
- OST_WRITE,
- cksum_type);
- CDEBUG(D_PAGE, "checksum at write origin: %x\n",
- body->oa.o_cksum);
- /* save this in 'oa', too, for later checking */
- oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
- oa->o_flags |= cksum_type_pack(cksum_type);
+
+ rc = osc_checksum_bulk_rw(obd_name, cksum_type,
+ requested_nob, page_count,
+ pga, OST_WRITE,
+ &body->oa.o_cksum, resend);
+ if (rc < 0) {
+ CDEBUG(D_PAGE, "failed to checksum: rc = %d\n",
+ rc);
+ GOTO(out, rc);
+ }
+ CDEBUG(D_PAGE | (resend ? D_HA : 0),
+ "checksum at write origin: %x (%x)\n",
+ body->oa.o_cksum, cksum_type);
+
+ /* save this in 'oa', too, for later checking */
+ oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
+ oa->o_flags |= obd_cksum_type_pack(obd_name,
+ cksum_type);
} else {
/* clear out the checksum flag, in case this is a
* resend but cl_checksum is no longer set. b=11238 */
req_capsule_set_size(pill, &RMF_RCS, RCL_SERVER,
sizeof(__u32) * niocount);
} else {
- if (cli->cl_checksum &&
- !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
+ if (enable_checksum) {
if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
body->oa.o_flags = 0;
- body->oa.o_flags |= cksum_type_pack(cli->cl_cksum_type);
+ body->oa.o_flags |= obd_cksum_type_pack(obd_name,
+ cli->cl_cksum_type);
body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
- }
+ }
/* Client cksum has been already copied to wire obdo in previous
* lustre_set_wire_obdo(), and in the case a bulk-read is being
}
ptlrpc_request_set_replen(req);
- CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
- aa = ptlrpc_req_async_args(req);
+ aa = ptlrpc_req_async_args(aa, req);
aa->aa_oa = oa;
aa->aa_requested_nob = requested_nob;
aa->aa_nio_count = niocount;
int rc, i;
unsigned int len;
char *buf;
- mm_segment_t oldfs;
/* will only keep dump of pages on first error for the same range in
* file/fid, not during the resends/retries. */
snprintf(dbgcksum_file_name, sizeof(dbgcksum_file_name),
"%s-checksum_dump-osc-"DFID":[%llu-%llu]-%x-%x",
- (strncmp(libcfs_debug_file_path_arr, "NONE", 4) != 0 ?
- libcfs_debug_file_path_arr :
- LIBCFS_DEBUG_FILE_PATH_DEFAULT),
+ (strncmp(libcfs_debug_file_path, "NONE", 4) != 0 ?
+ libcfs_debug_file_path : LIBCFS_DEBUG_FILE_PATH_DEFAULT),
oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : 0ULL,
oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
pga[0]->off,
pga[page_count-1]->off + pga[page_count-1]->count - 1,
client_cksum, server_cksum);
+ CWARN("dumping checksum data to %s\n", dbgcksum_file_name);
filp = filp_open(dbgcksum_file_name,
O_CREAT | O_EXCL | O_WRONLY | O_LARGEFILE, 0600);
if (IS_ERR(filp)) {
return;
}
- oldfs = get_fs();
- set_fs(KERNEL_DS);
for (i = 0; i < page_count; i++) {
len = pga[i]->count;
buf = kmap(pga[i]->pg);
while (len != 0) {
- rc = vfs_write(filp, (__force const char __user *)buf,
- len, &filp->f_pos);
+ rc = cfs_kernel_write(filp, buf, len, &filp->f_pos);
if (rc < 0) {
CERROR("%s: wanted to write %u but got %d "
"error\n", dbgcksum_file_name, len, rc);
}
len -= rc;
buf += rc;
- CDEBUG(D_INFO, "%s: wrote %d bytes\n",
- dbgcksum_file_name, rc);
}
kunmap(pga[i]->pg);
}
- set_fs(oldfs);
- rc = ll_vfs_fsync_range(filp, 0, LLONG_MAX, 1);
+ rc = vfs_fsync_range(filp, 0, LLONG_MAX, 1);
if (rc)
CERROR("%s: sync returns %d\n", dbgcksum_file_name, rc);
filp_close(filp, NULL);
- return;
+
+ libcfs_debug_dumplog();
}
static int
-check_write_checksum(struct obdo *oa, const lnet_process_id_t *peer,
- __u32 client_cksum, __u32 server_cksum,
- struct osc_brw_async_args *aa)
+check_write_checksum(struct obdo *oa, const struct lnet_process_id *peer,
+ __u32 client_cksum, __u32 server_cksum,
+ struct osc_brw_async_args *aa)
{
- __u32 new_cksum;
- char *msg;
+ const char *obd_name = aa->aa_cli->cl_import->imp_obd->obd_name;
enum cksum_types cksum_type;
+ obd_dif_csum_fn *fn = NULL;
+ int sector_size = 0;
+ __u32 new_cksum;
+ char *msg;
+ int rc;
if (server_cksum == client_cksum) {
CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
dump_all_bulk_pages(oa, aa->aa_page_count, aa->aa_ppga,
server_cksum, client_cksum);
- cksum_type = cksum_type_unpack(oa->o_valid & OBD_MD_FLFLAGS ?
- oa->o_flags : 0);
- new_cksum = osc_checksum_bulk(aa->aa_requested_nob, aa->aa_page_count,
- aa->aa_ppga, OST_WRITE, cksum_type);
+ cksum_type = obd_cksum_type_unpack(oa->o_valid & OBD_MD_FLFLAGS ?
+ oa->o_flags : 0);
+
+ switch (cksum_type) {
+ case OBD_CKSUM_T10IP512:
+ fn = obd_dif_ip_fn;
+ sector_size = 512;
+ break;
+ case OBD_CKSUM_T10IP4K:
+ fn = obd_dif_ip_fn;
+ sector_size = 4096;
+ break;
+ case OBD_CKSUM_T10CRC512:
+ fn = obd_dif_crc_fn;
+ sector_size = 512;
+ break;
+ case OBD_CKSUM_T10CRC4K:
+ fn = obd_dif_crc_fn;
+ sector_size = 4096;
+ break;
+ default:
+ break;
+ }
+
+ if (fn)
+ rc = osc_checksum_bulk_t10pi(obd_name, aa->aa_requested_nob,
+ aa->aa_page_count, aa->aa_ppga,
+ OST_WRITE, fn, sector_size,
+ &new_cksum, true);
+ else
+ rc = osc_checksum_bulk(aa->aa_requested_nob, aa->aa_page_count,
+ aa->aa_ppga, OST_WRITE, cksum_type,
+ &new_cksum);
- if (cksum_type != cksum_type_unpack(aa->aa_oa->o_flags))
+ if (rc < 0)
+ msg = "failed to calculate the client write checksum";
+ else if (cksum_type != obd_cksum_type_unpack(aa->aa_oa->o_flags))
msg = "the server did not use the checksum type specified in "
"the original request - likely a protocol problem";
else if (new_cksum == server_cksum)
DFID " object "DOSTID" extent [%llu-%llu], original "
"client csum %x (type %x), server csum %x (type %x),"
" client csum now %x\n",
- aa->aa_cli->cl_import->imp_obd->obd_name,
- msg, libcfs_nid2str(peer->nid),
+ obd_name, msg, libcfs_nid2str(peer->nid),
oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : (__u64)0,
oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
POSTID(&oa->o_oi), aa->aa_ppga[0]->off,
aa->aa_ppga[aa->aa_page_count - 1]->off +
aa->aa_ppga[aa->aa_page_count-1]->count - 1,
- client_cksum, cksum_type_unpack(aa->aa_oa->o_flags),
+ client_cksum,
+ obd_cksum_type_unpack(aa->aa_oa->o_flags),
server_cksum, cksum_type, new_cksum);
return 1;
}
/* Note rc enters this function as number of bytes transferred */
static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
{
- struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
+ struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
+ struct client_obd *cli = aa->aa_cli;
+ const char *obd_name = cli->cl_import->imp_obd->obd_name;
const struct lnet_process_id *peer =
- &req->rq_import->imp_connection->c_peer;
- struct client_obd *cli = aa->aa_cli;
- struct ost_body *body;
+ &req->rq_import->imp_connection->c_peer;
+ struct ost_body *body;
u32 client_cksum = 0;
- ENTRY;
+ struct inode *inode;
+ unsigned int blockbits = 0, blocksize = 0;
- if (rc < 0 && rc != -EDQUOT) {
- DEBUG_REQ(D_INFO, req, "Failed request with rc = %d\n", rc);
- RETURN(rc);
- }
+ ENTRY;
- LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
- body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
- if (body == NULL) {
- DEBUG_REQ(D_INFO, req, "Can't unpack body\n");
- RETURN(-EPROTO);
- }
+ if (rc < 0 && rc != -EDQUOT) {
+ DEBUG_REQ(D_INFO, req, "Failed request: rc = %d", rc);
+ RETURN(rc);
+ }
+
+ LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
+ body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
+ if (body == NULL) {
+ DEBUG_REQ(D_INFO, req, "cannot unpack body");
+ RETURN(-EPROTO);
+ }
/* set/clear over quota flag for a uid/gid/projid */
if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
unsigned qid[LL_MAXQUOTAS] = {
body->oa.o_uid, body->oa.o_gid,
body->oa.o_projid };
- CDEBUG(D_QUOTA, "setdq for [%u %u %u] with valid %#llx, flags %x\n",
+ CDEBUG(D_QUOTA,
+ "setdq for [%u %u %u] with valid %#llx, flags %x\n",
body->oa.o_uid, body->oa.o_gid, body->oa.o_projid,
body->oa.o_valid, body->oa.o_flags);
- osc_quota_setdq(cli, qid, body->oa.o_valid,
- body->oa.o_flags);
- }
+ osc_quota_setdq(cli, req->rq_xid, qid, body->oa.o_valid,
+ body->oa.o_flags);
+ }
- osc_update_grant(cli, body);
+ osc_update_grant(cli, body);
- if (rc < 0)
- RETURN(rc);
+ if (rc < 0)
+ RETURN(rc);
- if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
- client_cksum = aa->aa_oa->o_cksum; /* save for later */
+ if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
+ client_cksum = aa->aa_oa->o_cksum; /* save for later */
- if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
- if (rc > 0) {
- CERROR("Unexpected +ve rc %d\n", rc);
- RETURN(-EPROTO);
- }
+ if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
+ if (rc > 0) {
+ CERROR("%s: unexpected positive size %d\n",
+ obd_name, rc);
+ RETURN(-EPROTO);
+ }
if (req->rq_bulk != NULL &&
sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
- RETURN(-EAGAIN);
+ RETURN(-EAGAIN);
- if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
- check_write_checksum(&body->oa, peer, client_cksum,
+ if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
+ check_write_checksum(&body->oa, peer, client_cksum,
body->oa.o_cksum, aa))
- RETURN(-EAGAIN);
+ RETURN(-EAGAIN);
- rc = check_write_rcs(req, aa->aa_requested_nob,aa->aa_nio_count,
- aa->aa_page_count, aa->aa_ppga);
- GOTO(out, rc);
- }
+ rc = check_write_rcs(req, aa->aa_requested_nob,
+ aa->aa_nio_count, aa->aa_page_count,
+ aa->aa_ppga);
+ GOTO(out, rc);
+ }
- /* The rest of this function executes only for OST_READs */
+ /* The rest of this function executes only for OST_READs */
if (req->rq_bulk == NULL) {
rc = req_capsule_get_size(&req->rq_pill, &RMF_SHORT_IO,
/* if unwrap_bulk failed, return -EAGAIN to retry */
rc = sptlrpc_cli_unwrap_bulk_read(req, req->rq_bulk, rc);
}
- if (rc < 0)
- GOTO(out, rc = -EAGAIN);
+ if (rc < 0)
+ GOTO(out, rc = -EAGAIN);
- if (rc > aa->aa_requested_nob) {
- CERROR("Unexpected rc %d (%d requested)\n", rc,
- aa->aa_requested_nob);
- RETURN(-EPROTO);
- }
+ if (rc > aa->aa_requested_nob) {
+ CERROR("%s: unexpected size %d, requested %d\n", obd_name,
+ rc, aa->aa_requested_nob);
+ RETURN(-EPROTO);
+ }
if (req->rq_bulk != NULL && rc != req->rq_bulk->bd_nob_transferred) {
- CERROR ("Unexpected rc %d (%d transferred)\n",
- rc, req->rq_bulk->bd_nob_transferred);
- return (-EPROTO);
- }
+ CERROR("%s: unexpected size %d, transferred %d\n", obd_name,
+ rc, req->rq_bulk->bd_nob_transferred);
+ RETURN(-EPROTO);
+ }
if (req->rq_bulk == NULL) {
/* short io */
CDEBUG(D_CACHE, "page %p count %d\n",
aa->aa_ppga[i]->pg, count);
- ptr = ll_kmap_atomic(aa->aa_ppga[i]->pg, KM_USER0);
+ ptr = kmap_atomic(aa->aa_ppga[i]->pg);
memcpy(ptr + (aa->aa_ppga[i]->off & ~PAGE_MASK), buf,
count);
- ll_kunmap_atomic((void *) ptr, KM_USER0);
+ kunmap_atomic((void *) ptr);
buf += count;
nob -= count;
}
}
- if (rc < aa->aa_requested_nob)
- handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
+ if (rc < aa->aa_requested_nob)
+ handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
- if (body->oa.o_valid & OBD_MD_FLCKSUM) {
- static int cksum_counter;
- u32 server_cksum = body->oa.o_cksum;
- char *via = "";
- char *router = "";
+ if (body->oa.o_valid & OBD_MD_FLCKSUM) {
+ static int cksum_counter;
+ u32 server_cksum = body->oa.o_cksum;
+ int nob = rc;
+ char *via = "";
+ char *router = "";
enum cksum_types cksum_type;
+ u32 o_flags = body->oa.o_valid & OBD_MD_FLFLAGS ?
+ body->oa.o_flags : 0;
- cksum_type = cksum_type_unpack(body->oa.o_valid &OBD_MD_FLFLAGS?
- body->oa.o_flags : 0);
- client_cksum = osc_checksum_bulk(rc, aa->aa_page_count,
- aa->aa_ppga, OST_READ,
- cksum_type);
+ cksum_type = obd_cksum_type_unpack(o_flags);
+ rc = osc_checksum_bulk_rw(obd_name, cksum_type, nob,
+ aa->aa_page_count, aa->aa_ppga,
+ OST_READ, &client_cksum, false);
+ if (rc < 0)
+ GOTO(out, rc);
if (req->rq_bulk != NULL &&
peer->nid != req->rq_bulk->bd_sender) {
if (server_cksum != client_cksum) {
struct ost_body *clbody;
+ __u32 client_cksum2;
u32 page_count = aa->aa_page_count;
+ osc_checksum_bulk_rw(obd_name, cksum_type, nob,
+ page_count, aa->aa_ppga,
+ OST_READ, &client_cksum2, true);
clbody = req_capsule_client_get(&req->rq_pill,
&RMF_OST_BODY);
if (cli->cl_checksum_dump)
LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
"%s%s%s inode "DFID" object "DOSTID
- " extent [%llu-%llu], client %x, "
+ " extent [%llu-%llu], client %x/%x, "
"server %x, cksum_type %x\n",
- req->rq_import->imp_obd->obd_name,
+ obd_name,
libcfs_nid2str(peer->nid),
via, router,
clbody->oa.o_valid & OBD_MD_FLFID ?
aa->aa_ppga[0]->off,
aa->aa_ppga[page_count-1]->off +
aa->aa_ppga[page_count-1]->count - 1,
- client_cksum, server_cksum,
- cksum_type);
+ client_cksum, client_cksum2,
+ server_cksum, cksum_type);
cksum_counter = 0;
aa->aa_oa->o_cksum = client_cksum;
rc = -EAGAIN;
CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
rc = 0;
}
- } else if (unlikely(client_cksum)) {
- static int cksum_missed;
+ } else if (unlikely(client_cksum)) {
+ static int cksum_missed;
+
+ cksum_missed++;
+ if ((cksum_missed & (-cksum_missed)) == cksum_missed)
+ CERROR("%s: checksum %u requested from %s but not sent\n",
+ obd_name, cksum_missed,
+ libcfs_nid2str(peer->nid));
+ } else {
+ rc = 0;
+ }
+
+ inode = page2inode(aa->aa_ppga[0]->pg);
+ if (inode == NULL) {
+ /* Try to get reference to inode from cl_page if we are
+ * dealing with direct IO, as handled pages are not
+ * actual page cache pages.
+ */
+ struct osc_async_page *oap = brw_page2oap(aa->aa_ppga[0]);
+
+ inode = oap2cl_page(oap)->cp_inode;
+ if (inode) {
+ blockbits = inode->i_blkbits;
+ blocksize = 1 << blockbits;
+ }
+ }
+ if (inode && IS_ENCRYPTED(inode)) {
+ int idx;
+
+ if (!llcrypt_has_encryption_key(inode)) {
+ CDEBUG(D_SEC, "no enc key for ino %lu\n", inode->i_ino);
+ GOTO(out, rc);
+ }
+ for (idx = 0; idx < aa->aa_page_count; idx++) {
+ struct brw_page *pg = aa->aa_ppga[idx];
+ unsigned int offs = 0;
+
+ while (offs < PAGE_SIZE) {
+ /* do not decrypt if page is all 0s */
+ if (memchr_inv(page_address(pg->pg) + offs, 0,
+ LUSTRE_ENCRYPTION_UNIT_SIZE) == NULL) {
+ /* if page is empty forward info to
+ * upper layers (ll_io_zero_page) by
+ * clearing PagePrivate2
+ */
+ if (!offs)
+ ClearPagePrivate2(pg->pg);
+ break;
+ }
+
+ if (blockbits) {
+ /* This is direct IO case. Directly call
+ * decrypt function that takes inode as
+ * input parameter. Page does not need
+ * to be locked.
+ */
+ u64 lblk_num =
+ ((u64)(pg->off >> PAGE_SHIFT) <<
+ (PAGE_SHIFT - blockbits)) +
+ (offs >> blockbits);
+ unsigned int i;
+
+ for (i = offs;
+ i < offs +
+ LUSTRE_ENCRYPTION_UNIT_SIZE;
+ i += blocksize, lblk_num++) {
+ rc =
+ llcrypt_decrypt_block_inplace(
+ inode, pg->pg,
+ blocksize, i,
+ lblk_num);
+ if (rc)
+ break;
+ }
+ } else {
+ rc = llcrypt_decrypt_pagecache_blocks(
+ pg->pg,
+ LUSTRE_ENCRYPTION_UNIT_SIZE,
+ offs);
+ }
+ if (rc)
+ GOTO(out, rc);
+
+ offs += LUSTRE_ENCRYPTION_UNIT_SIZE;
+ }
+ }
+ }
- cksum_missed++;
- if ((cksum_missed & (-cksum_missed)) == cksum_missed)
- CERROR("Checksum %u requested from %s but not sent\n",
- cksum_missed, libcfs_nid2str(peer->nid));
- } else {
- rc = 0;
- }
out:
if (rc >= 0)
lustre_get_wire_obdo(&req->rq_import->imp_connect_data,
aa->aa_oa, &body->oa);
- RETURN(rc);
+ RETURN(rc);
}
static int osc_brw_redo_request(struct ptlrpc_request *request,
struct osc_brw_async_args *aa, int rc)
{
- struct ptlrpc_request *new_req;
- struct osc_brw_async_args *new_aa;
- struct osc_async_page *oap;
- ENTRY;
+ struct ptlrpc_request *new_req;
+ struct osc_brw_async_args *new_aa;
+ struct osc_async_page *oap;
+ ENTRY;
+ /* The below message is checked in replay-ost-single.sh test_8ae*/
DEBUG_REQ(rc == -EINPROGRESS ? D_RPCTRACE : D_ERROR, request,
"redo for recoverable error %d", rc);
RETURN(rc);
list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
- if (oap->oap_request != NULL) {
- LASSERTF(request == oap->oap_request,
- "request %p != oap_request %p\n",
- request, oap->oap_request);
- if (oap->oap_interrupted) {
- ptlrpc_req_finished(new_req);
- RETURN(-EINTR);
- }
- }
- }
- /* New request takes over pga and oaps from old request.
- * Note that copying a list_head doesn't work, need to move it... */
- aa->aa_resends++;
- new_req->rq_interpret_reply = request->rq_interpret_reply;
- new_req->rq_async_args = request->rq_async_args;
+ if (oap->oap_request != NULL) {
+ LASSERTF(request == oap->oap_request,
+ "request %p != oap_request %p\n",
+ request, oap->oap_request);
+ }
+ }
+ /*
+ * New request takes over pga and oaps from old request.
+ * Note that copying a list_head doesn't work, need to move it...
+ */
+ aa->aa_resends++;
+ new_req->rq_interpret_reply = request->rq_interpret_reply;
+ new_req->rq_async_args = request->rq_async_args;
new_req->rq_commit_cb = request->rq_commit_cb;
/* cap resend delay to the current request timeout, this is similar to
* what ptlrpc does (see after_reply()) */
new_req->rq_generation_set = 1;
new_req->rq_import_generation = request->rq_import_generation;
- new_aa = ptlrpc_req_async_args(new_req);
+ new_aa = ptlrpc_req_async_args(new_aa, new_req);
INIT_LIST_HEAD(&new_aa->aa_oaps);
list_splice_init(&aa->aa_oaps, &new_aa->aa_oaps);
static void osc_release_ppga(struct brw_page **ppga, size_t count)
{
- LASSERT(ppga != NULL);
- OBD_FREE(ppga, sizeof(*ppga) * count);
+ LASSERT(ppga != NULL);
+ OBD_FREE_PTR_ARRAY_LARGE(ppga, count);
}
static int brw_interpret(const struct lu_env *env,
- struct ptlrpc_request *req, void *data, int rc)
+ struct ptlrpc_request *req, void *args, int rc)
{
- struct osc_brw_async_args *aa = data;
+ struct osc_brw_async_args *aa = args;
struct osc_extent *ext;
struct osc_extent *tmp;
struct client_obd *cli = aa->aa_cli;
- unsigned long transferred = 0;
- ENTRY;
+ unsigned long transferred = 0;
+
+ ENTRY;
+
+ rc = osc_brw_fini_request(req, rc);
+ CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
- rc = osc_brw_fini_request(req, rc);
- CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
- /* When server return -EINPROGRESS, client should always retry
- * regardless of the number of times the bulk was resent already. */
+ /* restore clear text pages */
+ osc_release_bounce_pages(aa->aa_ppga, aa->aa_page_count);
+
+ /*
+ * When server returns -EINPROGRESS, client should always retry
+ * regardless of the number of times the bulk was resent already.
+ */
if (osc_recoverable_error(rc) && !req->rq_no_delay) {
if (req->rq_import_generation !=
req->rq_import->imp_generation) {
req->rq_import->imp_obd->obd_name,
POSTID(&aa->aa_oa->o_oi), rc);
} else if (rc == -EINPROGRESS ||
- client_should_resend(aa->aa_resends, aa->aa_cli)) {
+ client_should_resend(aa->aa_resends, aa->aa_cli)) {
rc = osc_brw_redo_request(req, aa, rc);
} else {
CERROR("%s: too many resent retries for object: "
cl_object_attr_update(env, obj, attr, valid);
cl_object_attr_unlock(obj);
}
- OBDO_FREE(aa->aa_oa);
+ OBD_SLAB_FREE_PTR(aa->aa_oa, osc_obdo_kmem);
+ aa->aa_oa = NULL;
if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE && rc == 0)
osc_inc_unstable_pages(req);
list_for_each_entry_safe(ext, tmp, &aa->aa_exts, oe_link) {
list_del_init(&ext->oe_link);
osc_extent_finish(env, ext, 1,
- rc && req->rq_no_delay ? -EWOULDBLOCK : rc);
+ rc && req->rq_no_delay ? -EAGAIN : rc);
}
LASSERT(list_empty(&aa->aa_exts));
LASSERT(list_empty(&aa->aa_oaps));
struct cl_req_attr *crattr = NULL;
loff_t starting_offset = OBD_OBJECT_EOF;
loff_t ending_offset = 0;
- int mpflag = 0;
+ /* '1' for consistency with code that checks !mpflag to restore */
+ int mpflag = 1;
int mem_tight = 0;
int page_count = 0;
bool soft_sync = false;
- bool interrupted = false;
bool ndelay = false;
int i;
int grant = 0;
int rc;
__u32 layout_version = 0;
- struct list_head rpc_list = LIST_HEAD_INIT(rpc_list);
+ LIST_HEAD(rpc_list);
struct ost_body *body;
ENTRY;
LASSERT(!list_empty(ext_list));
mem_tight |= ext->oe_memalloc;
grant += ext->oe_grants;
page_count += ext->oe_nr_pages;
- layout_version = MAX(layout_version, ext->oe_layout_version);
+ layout_version = max(layout_version, ext->oe_layout_version);
if (obj == NULL)
obj = ext->oe_obj;
}
soft_sync = osc_over_unstable_soft_limit(cli);
if (mem_tight)
- mpflag = cfs_memory_pressure_get_and_set();
+ mpflag = memalloc_noreclaim_save();
- OBD_ALLOC(pga, sizeof(*pga) * page_count);
+ OBD_ALLOC_PTR_ARRAY_LARGE(pga, page_count);
if (pga == NULL)
GOTO(out, rc = -ENOMEM);
- OBDO_ALLOC(oa);
+ OBD_SLAB_ALLOC_PTR_GFP(oa, osc_obdo_kmem, GFP_NOFS);
if (oa == NULL)
GOTO(out, rc = -ENOMEM);
else
LASSERT(oap->oap_page_off + oap->oap_count ==
PAGE_SIZE);
- if (oap->oap_interrupted)
- interrupted = true;
}
if (ext->oe_ndelay)
ndelay = true;
}
/* first page in the list */
- oap = list_entry(rpc_list.next, typeof(*oap), oap_rpc_item);
+ oap = list_first_entry(&rpc_list, typeof(*oap), oap_rpc_item);
crattr = &osc_env_info(env)->oti_req_attr;
memset(crattr, 0, sizeof(*crattr));
req->rq_interpret_reply = brw_interpret;
req->rq_memalloc = mem_tight != 0;
oap->oap_request = ptlrpc_request_addref(req);
- if (interrupted && !req->rq_intr)
- ptlrpc_mark_interrupted(req);
if (ndelay) {
req->rq_no_resend = req->rq_no_delay = 1;
/* probably set a shorter timeout value.
* way to do this in a single call. bug 10150 */
body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
crattr->cra_oa = &body->oa;
- crattr->cra_flags = OBD_MD_FLMTIME|OBD_MD_FLCTIME|OBD_MD_FLATIME;
+ crattr->cra_flags = OBD_MD_FLMTIME | OBD_MD_FLCTIME | OBD_MD_FLATIME;
cl_req_attr_set(env, osc2cl(obj), crattr);
lustre_msg_set_jobid(req->rq_reqmsg, crattr->cra_jobid);
- CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
- aa = ptlrpc_req_async_args(req);
+ aa = ptlrpc_req_async_args(aa, req);
INIT_LIST_HEAD(&aa->aa_oaps);
list_splice_init(&rpc_list, &aa->aa_oaps);
INIT_LIST_HEAD(&aa->aa_exts);
}
spin_unlock(&cli->cl_loi_list_lock);
- DEBUG_REQ(D_INODE, req, "%d pages, aa %p. now %ur/%uw in flight",
+ DEBUG_REQ(D_INODE, req, "%d pages, aa %p, now %ur/%uw in flight",
page_count, aa, cli->cl_r_in_flight,
cli->cl_w_in_flight);
OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_DELAY_IO, cfs_fail_val);
EXIT;
out:
- if (mem_tight != 0)
- cfs_memory_pressure_restore(mpflag);
+ if (mem_tight)
+ memalloc_noreclaim_restore(mpflag);
if (rc != 0) {
LASSERT(req == NULL);
if (oa)
- OBDO_FREE(oa);
- if (pga)
- OBD_FREE(pga, sizeof(*pga) * page_count);
+ OBD_SLAB_FREE_PTR(oa, osc_obdo_kmem);
+ if (pga) {
+ osc_release_bounce_pages(pga, page_count);
+ osc_release_ppga(pga, page_count);
+ }
/* this should happen rarely and is pretty bad, it makes the
- * pending list not follow the dirty order */
- while (!list_empty(ext_list)) {
- ext = list_entry(ext_list->next, struct osc_extent,
- oe_link);
+ * pending list not follow the dirty order
+ */
+ while ((ext = list_first_entry_or_null(ext_list,
+ struct osc_extent,
+ oe_link)) != NULL) {
list_del_init(&ext->oe_link);
osc_extent_finish(env, ext, 0, rc);
}
RETURN(rc);
}
+/* This is to refresh our lock in face of no RPCs. */
+void osc_send_empty_rpc(struct osc_object *osc, pgoff_t start)
+{
+ struct ptlrpc_request *req;
+ struct obdo oa;
+ struct brw_page bpg = { .off = start, .count = 1};
+ struct brw_page *pga = &bpg;
+ int rc;
+
+ memset(&oa, 0, sizeof(oa));
+ oa.o_oi = osc->oo_oinfo->loi_oi;
+ oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP | OBD_MD_FLFLAGS;
+ /* For updated servers - don't do a read */
+ oa.o_flags = OBD_FL_NORPC;
+
+ rc = osc_brw_prep_request(OBD_BRW_READ, osc_cli(osc), &oa, 1, &pga,
+ &req, 0);
+
+ /* If we succeeded we ship it off, if not there's no point in doing
+ * anything. Also no resends.
+ * No interpret callback, no commit callback.
+ */
+ if (!rc) {
+ req->rq_no_resend = 1;
+ ptlrpcd_add_req(req);
+ }
+}
+
static int osc_set_lock_data(struct ldlm_lock *lock, void *data)
{
int set = 0;
}
int osc_enqueue_interpret(const struct lu_env *env, struct ptlrpc_request *req,
- struct osc_enqueue_args *aa, int rc)
+ void *args, int rc)
{
+ struct osc_enqueue_args *aa = args;
struct ldlm_lock *lock;
struct lustre_handle *lockh = &aa->oa_lockh;
enum ldlm_mode mode = aa->oa_mode;
struct ost_lvb *lvb = aa->oa_lvb;
__u32 lvb_len = sizeof(*lvb);
__u64 flags = 0;
+ struct ldlm_enqueue_info einfo = {
+ .ei_type = aa->oa_type,
+ .ei_mode = mode,
+ };
ENTRY;
}
/* Complete obtaining the lock procedure. */
- rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, aa->oa_type, 1,
- aa->oa_mode, aa->oa_flags, lvb, lvb_len,
- lockh, rc);
+ rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, &einfo, 1, aa->oa_flags,
+ lvb, lvb_len, lockh, rc, false);
/* Complete osc stuff. */
rc = osc_enqueue_fini(req, aa->oa_upcall, aa->oa_cookie, lockh, mode,
aa->oa_flags, aa->oa_speculative, rc);
RETURN(rc);
}
-struct ptlrpc_request_set *PTLRPCD_SET = (void *)1;
-
/* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
* from the 2nd OSC before a lock from the 1st one. This does not deadlock with
* other synchronous requests, however keeping some locks and trying to obtain
* release locks just after they are obtained. */
int osc_enqueue_base(struct obd_export *exp, struct ldlm_res_id *res_id,
__u64 *flags, union ldlm_policy_data *policy,
- struct ost_lvb *lvb, int kms_valid,
- osc_enqueue_upcall_f upcall, void *cookie,
- struct ldlm_enqueue_info *einfo,
+ struct ost_lvb *lvb, osc_enqueue_upcall_f upcall,
+ void *cookie, struct ldlm_enqueue_info *einfo,
struct ptlrpc_request_set *rqset, int async,
bool speculative)
{
policy->l_extent.start -= policy->l_extent.start & ~PAGE_MASK;
policy->l_extent.end |= ~PAGE_MASK;
- /*
- * kms is not valid when either object is completely fresh (so that no
- * locks are cached), or object was evicted. In the latter case cached
- * lock cannot be used, because it would prime inode state with
- * potentially stale LVB.
- */
- if (!kms_valid)
- goto no_match;
-
/* Next, search for already existing extent locks that will cover us */
/* If we're trying to read, we also search for an existing PW lock. The
* VFS and page cache already protect us locally, so lots of readers/
if (intent != 0)
match_flags |= LDLM_FL_BLOCK_GRANTED;
mode = ldlm_lock_match(obd->obd_namespace, match_flags, res_id,
- einfo->ei_type, policy, mode, &lockh, 0);
+ einfo->ei_type, policy, mode, &lockh);
if (mode) {
struct ldlm_lock *matched;
}
}
-no_match:
if (*flags & (LDLM_FL_TEST_LOCK | LDLM_FL_MATCH_LOCK))
RETURN(-ENOLCK);
- if (intent) {
- req = ptlrpc_request_alloc(class_exp2cliimp(exp),
- &RQF_LDLM_ENQUEUE_LVB);
- if (req == NULL)
- RETURN(-ENOMEM);
-
- rc = ldlm_prep_enqueue_req(exp, req, NULL, 0);
- if (rc) {
- ptlrpc_request_free(req);
- RETURN(rc);
- }
-
- req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
- sizeof *lvb);
- ptlrpc_request_set_replen(req);
- }
-
/* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
*flags &= ~LDLM_FL_BLOCK_GRANTED;
if (async) {
if (!rc) {
struct osc_enqueue_args *aa;
- CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
- aa = ptlrpc_req_async_args(req);
+ aa = ptlrpc_req_async_args(aa, req);
aa->oa_exp = exp;
aa->oa_mode = einfo->ei_mode;
aa->oa_type = einfo->ei_type;
aa->oa_flags = NULL;
}
- req->rq_interpret_reply =
- (ptlrpc_interpterer_t)osc_enqueue_interpret;
- if (rqset == PTLRPCD_SET)
- ptlrpcd_add_req(req);
- else
- ptlrpc_set_add_req(rqset, req);
- } else if (intent) {
- ptlrpc_req_finished(req);
+ req->rq_interpret_reply = osc_enqueue_interpret;
+ ptlrpc_set_add_req(rqset, req);
}
RETURN(rc);
}
rc = osc_enqueue_fini(req, upcall, cookie, &lockh, einfo->ei_mode,
flags, speculative, rc);
- if (intent)
- ptlrpc_req_finished(req);
RETURN(rc);
}
-int osc_match_base(struct obd_export *exp, struct ldlm_res_id *res_id,
- enum ldlm_type type, union ldlm_policy_data *policy,
- enum ldlm_mode mode, __u64 *flags, void *data,
- struct lustre_handle *lockh, int unref)
+int osc_match_base(const struct lu_env *env, struct obd_export *exp,
+ struct ldlm_res_id *res_id, enum ldlm_type type,
+ union ldlm_policy_data *policy, enum ldlm_mode mode,
+ __u64 *flags, struct osc_object *obj,
+ struct lustre_handle *lockh, enum ldlm_match_flags match_flags)
{
struct obd_device *obd = exp->exp_obd;
__u64 lflags = *flags;
policy->l_extent.start -= policy->l_extent.start & ~PAGE_MASK;
policy->l_extent.end |= ~PAGE_MASK;
- /* Next, search for already existing extent locks that will cover us */
- /* If we're trying to read, we also search for an existing PW lock. The
- * VFS and page cache already protect us locally, so lots of readers/
- * writers can share a single PW lock. */
- rc = mode;
- if (mode == LCK_PR)
- rc |= LCK_PW;
- rc = ldlm_lock_match(obd->obd_namespace, lflags,
- res_id, type, policy, rc, lockh, unref);
+ /* Next, search for already existing extent locks that will cover us */
+ rc = ldlm_lock_match_with_skip(obd->obd_namespace, lflags, 0,
+ res_id, type, policy, mode, lockh,
+ match_flags);
if (rc == 0 || lflags & LDLM_FL_TEST_LOCK)
RETURN(rc);
- if (data != NULL) {
+ if (obj != NULL) {
struct ldlm_lock *lock = ldlm_handle2lock(lockh);
LASSERT(lock != NULL);
- if (!osc_set_lock_data(lock, data)) {
+ if (osc_set_lock_data(lock, obj)) {
+ lock_res_and_lock(lock);
+ if (!ldlm_is_lvb_cached(lock)) {
+ LASSERT(lock->l_ast_data == obj);
+ osc_lock_lvb_update(env, obj, lock, NULL);
+ ldlm_set_lvb_cached(lock);
+ }
+ unlock_res_and_lock(lock);
+ } else {
ldlm_lock_decref(lockh, rc);
rc = 0;
}
}
static int osc_statfs_interpret(const struct lu_env *env,
- struct ptlrpc_request *req,
- struct osc_async_args *aa, int rc)
+ struct ptlrpc_request *req, void *args, int rc)
{
- struct obd_statfs *msfs;
- ENTRY;
+ struct osc_async_args *aa = args;
+ struct obd_statfs *msfs;
- if (rc == -EBADR)
- /* The request has in fact never been sent
- * due to issues at a higher level (LOV).
- * Exit immediately since the caller is
- * aware of the problem and takes care
- * of the clean up */
- RETURN(rc);
+ ENTRY;
+ if (rc == -EBADR)
+ /*
+ * The request has in fact never been sent due to issues at
+ * a higher level (LOV). Exit immediately since the caller
+ * is aware of the problem and takes care of the clean up.
+ */
+ RETURN(rc);
- if ((rc == -ENOTCONN || rc == -EAGAIN) &&
- (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
- GOTO(out, rc = 0);
+ if ((rc == -ENOTCONN || rc == -EAGAIN) &&
+ (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
+ GOTO(out, rc = 0);
- if (rc != 0)
- GOTO(out, rc);
+ if (rc != 0)
+ GOTO(out, rc);
- msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
- if (msfs == NULL) {
+ msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
+ if (msfs == NULL)
GOTO(out, rc = -EPROTO);
- }
- *aa->aa_oi->oi_osfs = *msfs;
+ *aa->aa_oi->oi_osfs = *msfs;
out:
- rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
- RETURN(rc);
+ rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
+
+ RETURN(rc);
}
static int osc_statfs_async(struct obd_export *exp,
- struct obd_info *oinfo, __u64 max_age,
+ struct obd_info *oinfo, time64_t max_age,
struct ptlrpc_request_set *rqset)
{
struct obd_device *obd = class_exp2obd(exp);
struct ptlrpc_request *req;
struct osc_async_args *aa;
- int rc;
+ int rc;
ENTRY;
+ if (obd->obd_osfs_age >= max_age) {
+ CDEBUG(D_SUPER,
+ "%s: use %p cache blocks %llu/%llu objects %llu/%llu\n",
+ obd->obd_name, &obd->obd_osfs,
+ obd->obd_osfs.os_bavail, obd->obd_osfs.os_blocks,
+ obd->obd_osfs.os_ffree, obd->obd_osfs.os_files);
+ spin_lock(&obd->obd_osfs_lock);
+ memcpy(oinfo->oi_osfs, &obd->obd_osfs, sizeof(*oinfo->oi_osfs));
+ spin_unlock(&obd->obd_osfs_lock);
+ oinfo->oi_flags |= OBD_STATFS_FROM_CACHE;
+ if (oinfo->oi_cb_up)
+ oinfo->oi_cb_up(oinfo, 0);
+
+ RETURN(0);
+ }
+
/* We could possibly pass max_age in the request (as an absolute
* timestamp or a "seconds.usec ago") so the target can avoid doing
* extra calls into the filesystem if that isn't necessary (e.g.
ptlrpc_request_free(req);
RETURN(rc);
}
- ptlrpc_request_set_replen(req);
- req->rq_request_portal = OST_CREATE_PORTAL;
- ptlrpc_at_set_req_timeout(req);
+ ptlrpc_request_set_replen(req);
+ req->rq_request_portal = OST_CREATE_PORTAL;
+ ptlrpc_at_set_req_timeout(req);
- if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
- /* procfs requests not want stat in wait for avoid deadlock */
- req->rq_no_resend = 1;
- req->rq_no_delay = 1;
- }
+ if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
+ /* procfs requests not want stat in wait for avoid deadlock */
+ req->rq_no_resend = 1;
+ req->rq_no_delay = 1;
+ }
- req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_statfs_interpret;
- CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
- aa = ptlrpc_req_async_args(req);
+ req->rq_interpret_reply = osc_statfs_interpret;
+ aa = ptlrpc_req_async_args(aa, req);
aa->aa_oi = oinfo;
ptlrpc_set_add_req(rqset, req);
}
static int osc_statfs(const struct lu_env *env, struct obd_export *exp,
- struct obd_statfs *osfs, __u64 max_age, __u32 flags)
+ struct obd_statfs *osfs, time64_t max_age, __u32 flags)
{
- struct obd_device *obd = class_exp2obd(exp);
- struct obd_statfs *msfs;
- struct ptlrpc_request *req;
- struct obd_import *imp = NULL;
- int rc;
- ENTRY;
+ struct obd_device *obd = class_exp2obd(exp);
+ struct obd_statfs *msfs;
+ struct ptlrpc_request *req;
+ struct obd_import *imp, *imp0;
+ int rc;
+ ENTRY;
- /*Since the request might also come from lprocfs, so we need
- *sync this with client_disconnect_export Bug15684*/
- down_read(&obd->u.cli.cl_sem);
- if (obd->u.cli.cl_import)
- imp = class_import_get(obd->u.cli.cl_import);
- up_read(&obd->u.cli.cl_sem);
- if (!imp)
- RETURN(-ENODEV);
+ /*Since the request might also come from lprocfs, so we need
+ *sync this with client_disconnect_export Bug15684
+ */
+ with_imp_locked(obd, imp0, rc)
+ imp = class_import_get(imp0);
+ if (rc)
+ RETURN(rc);
- /* We could possibly pass max_age in the request (as an absolute
- * timestamp or a "seconds.usec ago") so the target can avoid doing
- * extra calls into the filesystem if that isn't necessary (e.g.
- * during mount that would help a bit). Having relative timestamps
- * is not so great if request processing is slow, while absolute
- * timestamps are not ideal because they need time synchronization. */
- req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
+ /* We could possibly pass max_age in the request (as an absolute
+ * timestamp or a "seconds.usec ago") so the target can avoid doing
+ * extra calls into the filesystem if that isn't necessary (e.g.
+ * during mount that would help a bit). Having relative timestamps
+ * is not so great if request processing is slow, while absolute
+ * timestamps are not ideal because they need time synchronization. */
+ req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
- class_import_put(imp);
+ class_import_put(imp);
- if (req == NULL)
- RETURN(-ENOMEM);
+ if (req == NULL)
+ RETURN(-ENOMEM);
- rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
- if (rc) {
- ptlrpc_request_free(req);
- RETURN(rc);
- }
- ptlrpc_request_set_replen(req);
- req->rq_request_portal = OST_CREATE_PORTAL;
- ptlrpc_at_set_req_timeout(req);
+ rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
+ if (rc) {
+ ptlrpc_request_free(req);
+ RETURN(rc);
+ }
+ ptlrpc_request_set_replen(req);
+ req->rq_request_portal = OST_CREATE_PORTAL;
+ ptlrpc_at_set_req_timeout(req);
- if (flags & OBD_STATFS_NODELAY) {
- /* procfs requests not want stat in wait for avoid deadlock */
- req->rq_no_resend = 1;
- req->rq_no_delay = 1;
- }
+ if (flags & OBD_STATFS_NODELAY) {
+ /* procfs requests not want stat in wait for avoid deadlock */
+ req->rq_no_resend = 1;
+ req->rq_no_delay = 1;
+ }
- rc = ptlrpc_queue_wait(req);
- if (rc)
- GOTO(out, rc);
+ rc = ptlrpc_queue_wait(req);
+ if (rc)
+ GOTO(out, rc);
- msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
- if (msfs == NULL) {
- GOTO(out, rc = -EPROTO);
- }
+ msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
+ if (msfs == NULL)
+ GOTO(out, rc = -EPROTO);
- *osfs = *msfs;
+ *osfs = *msfs;
- EXIT;
- out:
- ptlrpc_req_finished(req);
- return rc;
+ EXIT;
+out:
+ ptlrpc_req_finished(req);
+ return rc;
}
static int osc_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 = karg;
- int err = 0;
- ENTRY;
+ struct obd_device *obd = exp->exp_obd;
+ struct obd_ioctl_data *data = karg;
+ int rc = 0;
+ ENTRY;
if (!try_module_get(THIS_MODULE)) {
CERROR("%s: cannot get module '%s'\n", obd->obd_name,
module_name(THIS_MODULE));
return -EINVAL;
}
- switch (cmd) {
- case OBD_IOC_CLIENT_RECOVER:
- err = ptlrpc_recover_import(obd->u.cli.cl_import,
- data->ioc_inlbuf1, 0);
- if (err > 0)
- err = 0;
- GOTO(out, err);
- case IOC_OSC_SET_ACTIVE:
- err = ptlrpc_set_import_active(obd->u.cli.cl_import,
- data->ioc_offset);
- GOTO(out, err);
- case OBD_IOC_PING_TARGET:
- err = ptlrpc_obd_ping(obd);
- GOTO(out, err);
+ switch (cmd) {
+ case OBD_IOC_CLIENT_RECOVER:
+ rc = ptlrpc_recover_import(obd->u.cli.cl_import,
+ data->ioc_inlbuf1, 0);
+ if (rc > 0)
+ rc = 0;
+ break;
+ case IOC_OSC_SET_ACTIVE:
+ rc = ptlrpc_set_import_active(obd->u.cli.cl_import,
+ data->ioc_offset);
+ break;
default:
- CDEBUG(D_INODE, "unrecognised ioctl %#x by %s\n",
- cmd, current_comm());
- GOTO(out, err = -ENOTTY);
+ rc = -ENOTTY;
+ CDEBUG(D_INODE, "%s: unrecognised ioctl %#x by %s: rc = %d\n",
+ obd->obd_name, cmd, current->comm, rc);
+ break;
}
-out:
+
module_put(THIS_MODULE);
- return err;
+ return rc;
}
int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
RETURN(0);
}
- if (KEY_IS(KEY_CACHE_SET)) {
- struct client_obd *cli = &obd->u.cli;
-
- LASSERT(cli->cl_cache == NULL); /* only once */
- cli->cl_cache = (struct cl_client_cache *)val;
- cl_cache_incref(cli->cl_cache);
- cli->cl_lru_left = &cli->cl_cache->ccc_lru_left;
-
- /* add this osc into entity list */
- LASSERT(list_empty(&cli->cl_lru_osc));
- spin_lock(&cli->cl_cache->ccc_lru_lock);
- list_add(&cli->cl_lru_osc, &cli->cl_cache->ccc_lru);
- spin_unlock(&cli->cl_cache->ccc_lru_lock);
-
- RETURN(0);
- }
-
if (KEY_IS(KEY_CACHE_LRU_SHRINK)) {
struct client_obd *cli = &obd->u.cli;
long nr = atomic_long_read(&cli->cl_lru_in_list) >> 1;
struct osc_grant_args *aa;
struct obdo *oa;
- CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
- aa = ptlrpc_req_async_args(req);
- OBDO_ALLOC(oa);
+ aa = ptlrpc_req_async_args(aa, req);
+ OBD_SLAB_ALLOC_PTR_GFP(oa, osc_obdo_kmem, GFP_NOFS);
if (!oa) {
ptlrpc_req_finished(req);
RETURN(-ENOMEM);
spin_lock(&cli->cl_loi_list_lock);
grant = cli->cl_avail_grant + cli->cl_reserved_grant;
- if (data->ocd_connect_flags & OBD_CONNECT_GRANT_PARAM)
+ if (data->ocd_connect_flags & OBD_CONNECT_GRANT_PARAM) {
+ /* restore ocd_grant_blkbits as client page bits */
+ data->ocd_grant_blkbits = PAGE_SHIFT;
grant += cli->cl_dirty_grant;
- else
+ } else {
grant += cli->cl_dirty_pages << PAGE_SHIFT;
+ }
data->ocd_grant = grant ? : 2 * cli_brw_size(obd);
lost_grant = cli->cl_lost_grant;
cli->cl_lost_grant = 0;
struct obd_device *obd = class_exp2obd(exp);
int rc;
- rc = client_disconnect_export(exp);
- /**
- * Initially we put del_shrink_grant before disconnect_export, but it
- * causes the following problem if setup (connect) and cleanup
- * (disconnect) are tangled together.
- * connect p1 disconnect p2
- * ptlrpc_connect_import
- * ............... class_manual_cleanup
- * osc_disconnect
- * del_shrink_grant
- * ptlrpc_connect_interrupt
- * init_grant_shrink
- * add this client to shrink list
- * cleanup_osc
- * Bang! pinger trigger the shrink.
- * So the osc should be disconnected from the shrink list, after we
- * are sure the import has been destroyed. BUG18662
- */
- if (obd->u.cli.cl_import == NULL)
- osc_del_shrink_grant(&obd->u.cli);
- return rc;
+ rc = client_disconnect_export(exp);
+ /**
+ * Initially we put del_shrink_grant before disconnect_export, but it
+ * causes the following problem if setup (connect) and cleanup
+ * (disconnect) are tangled together.
+ * connect p1 disconnect p2
+ * ptlrpc_connect_import
+ * ............... class_manual_cleanup
+ * osc_disconnect
+ * del_shrink_grant
+ * ptlrpc_connect_interrupt
+ * osc_init_grant
+ * add this client to shrink list
+ * cleanup_osc
+ * Bang! grant shrink thread trigger the shrink. BUG18662
+ */
+ osc_del_grant_list(&obd->u.cli);
+ return rc;
}
EXPORT_SYMBOL(osc_disconnect);
* Cancel all unused and granted extent lock.
*/
if (lock->l_resource->lr_type == LDLM_EXTENT &&
- lock->l_granted_mode == lock->l_req_mode &&
+ ldlm_is_granted(lock) &&
osc_ldlm_weigh_ast(lock) == 0)
RETURN(1);
GOTO(out_ptlrpcd_work, rc);
cli->cl_grant_shrink_interval = GRANT_SHRINK_INTERVAL;
+ cli->cl_root_squash = 0;
+ osc_update_next_shrink(cli);
- INIT_LIST_HEAD(&cli->cl_grant_shrink_list);
RETURN(rc);
out_ptlrpcd_work:
int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
{
struct client_obd *cli = &obd->u.cli;
- struct obd_type *type;
int adding;
int added;
int req_count;
if (rc < 0)
RETURN(rc);
-#ifdef CONFIG_PROC_FS
- obd->obd_vars = lprocfs_osc_obd_vars;
-#endif
- /* If this is true then both client (osc) and server (osp) are on the
- * same node. The osp layer if loaded first will register the osc proc
- * directory. In that case this obd_device will be attached its proc
- * tree to type->typ_procsym instead of obd->obd_type->typ_procroot.
- */
- type = class_search_type(LUSTRE_OSP_NAME);
- if (type && type->typ_procsym) {
- obd->obd_proc_entry = lprocfs_register(obd->obd_name,
- type->typ_procsym,
- obd->obd_vars, obd);
- if (IS_ERR(obd->obd_proc_entry)) {
- rc = PTR_ERR(obd->obd_proc_entry);
- CERROR("error %d setting up lprocfs for %s\n", rc,
- obd->obd_name);
- obd->obd_proc_entry = NULL;
- }
- }
-
- rc = lprocfs_obd_setup(obd, false);
- if (!rc) {
- /* If the basic OSC proc tree construction succeeded then
- * lets do the rest.
- */
- lproc_osc_attach_seqstat(obd);
- sptlrpc_lprocfs_cliobd_attach(obd);
- ptlrpc_lprocfs_register_obd(obd);
- }
+ rc = osc_tunables_init(obd);
+ if (rc)
+ RETURN(rc);
/*
* We try to control the total number of requests with a upper limit
atomic_add(added, &osc_pool_req_count);
}
- INIT_LIST_HEAD(&cli->cl_grant_shrink_list);
ns_register_cancel(obd->obd_namespace, osc_cancel_weight);
spin_lock(&osc_shrink_lock);
list_add_tail(&cli->cl_shrink_list, &osc_shrink_list);
spin_unlock(&osc_shrink_lock);
+ cli->cl_import->imp_idle_timeout = osc_idle_timeout;
+ cli->cl_import->imp_idle_debug = D_HA;
RETURN(0);
}
osc_precleanup_common(obd);
ptlrpc_lprocfs_unregister_obd(obd);
- lprocfs_obd_cleanup(obd);
RETURN(0);
}
}
EXPORT_SYMBOL(osc_cleanup_common);
-int osc_process_config_base(struct obd_device *obd, struct lustre_cfg *lcfg)
-{
- int rc = class_process_proc_param(PARAM_OSC, obd->obd_vars, lcfg, obd);
- return rc > 0 ? 0: rc;
-}
-
-static int osc_process_config(struct obd_device *obd, size_t len, void *buf)
-{
- return osc_process_config_base(obd, buf);
-}
-
-static struct obd_ops osc_obd_ops = {
+static const struct obd_ops osc_obd_ops = {
.o_owner = THIS_MODULE,
.o_setup = osc_setup,
.o_precleanup = osc_precleanup,
.o_cleanup = osc_cleanup_common,
.o_add_conn = client_import_add_conn,
.o_del_conn = client_import_del_conn,
- .o_connect = client_connect_import,
+ .o_connect = client_connect_import,
.o_reconnect = osc_reconnect,
.o_disconnect = osc_disconnect,
.o_statfs = osc_statfs,
.o_iocontrol = osc_iocontrol,
.o_set_info_async = osc_set_info_async,
.o_import_event = osc_import_event,
- .o_process_config = osc_process_config,
.o_quotactl = osc_quotactl,
};
-static struct shrinker *osc_cache_shrinker;
-struct list_head osc_shrink_list = LIST_HEAD_INIT(osc_shrink_list);
+LIST_HEAD(osc_shrink_list);
DEFINE_SPINLOCK(osc_shrink_lock);
-#ifndef HAVE_SHRINKER_COUNT
-static int osc_cache_shrink(SHRINKER_ARGS(sc, nr_to_scan, gfp_mask))
+#ifdef HAVE_SHRINKER_COUNT
+static struct shrinker osc_cache_shrinker = {
+ .count_objects = osc_cache_shrink_count,
+ .scan_objects = osc_cache_shrink_scan,
+ .seeks = DEFAULT_SEEKS,
+};
+#else
+static int osc_cache_shrink(struct shrinker *shrinker,
+ struct shrink_control *sc)
{
- struct shrink_control scv = {
- .nr_to_scan = shrink_param(sc, nr_to_scan),
- .gfp_mask = shrink_param(sc, gfp_mask)
- };
-#if !defined(HAVE_SHRINKER_WANT_SHRINK_PTR) && !defined(HAVE_SHRINK_CONTROL)
- struct shrinker *shrinker = NULL;
-#endif
+ (void)osc_cache_shrink_scan(shrinker, sc);
- (void)osc_cache_shrink_scan(shrinker, &scv);
-
- return osc_cache_shrink_count(shrinker, &scv);
+ return osc_cache_shrink_count(shrinker, sc);
}
+
+static struct shrinker osc_cache_shrinker = {
+ .shrink = osc_cache_shrink,
+ .seeks = DEFAULT_SEEKS,
+};
#endif
static int __init osc_init(void)
{
- bool enable_proc = true;
- struct obd_type *type;
unsigned int reqpool_size;
unsigned int reqsize;
int rc;
- DEF_SHRINKER_VAR(osc_shvar, osc_cache_shrink,
- osc_cache_shrink_count, osc_cache_shrink_scan);
ENTRY;
/* print an address of _any_ initialized kernel symbol from this
if (rc)
RETURN(rc);
- type = class_search_type(LUSTRE_OSP_NAME);
- if (type != NULL && type->typ_procsym != NULL)
- enable_proc = false;
-
- rc = class_register_type(&osc_obd_ops, NULL, enable_proc, NULL,
+ rc = class_register_type(&osc_obd_ops, NULL, true,
LUSTRE_OSC_NAME, &osc_device_type);
if (rc)
GOTO(out_kmem, rc);
- osc_cache_shrinker = set_shrinker(DEFAULT_SEEKS, &osc_shvar);
+ rc = register_shrinker(&osc_cache_shrinker);
+ if (rc)
+ GOTO(out_type, rc);
/* This is obviously too much memory, only prevent overflow here */
if (osc_reqpool_mem_max >= 1 << 12 || osc_reqpool_mem_max == 0)
- GOTO(out_type, rc = -EINVAL);
+ GOTO(out_shrinker, rc = -EINVAL);
reqpool_size = osc_reqpool_mem_max << 20;
osc_rq_pool = ptlrpc_init_rq_pool(0, OST_IO_MAXREQSIZE,
ptlrpc_add_rqs_to_pool);
- if (osc_rq_pool != NULL)
- GOTO(out, rc);
- rc = -ENOMEM;
+ if (osc_rq_pool == NULL)
+ GOTO(out_shrinker, rc = -ENOMEM);
+
+ rc = osc_start_grant_work();
+ if (rc != 0)
+ GOTO(out_req_pool, rc);
+
+ RETURN(rc);
+
+out_req_pool:
+ ptlrpc_free_rq_pool(osc_rq_pool);
+out_shrinker:
+ unregister_shrinker(&osc_cache_shrinker);
out_type:
class_unregister_type(LUSTRE_OSC_NAME);
out_kmem:
lu_kmem_fini(osc_caches);
-out:
+
RETURN(rc);
}
static void __exit osc_exit(void)
{
- remove_shrinker(osc_cache_shrinker);
+ osc_stop_grant_work();
+ unregister_shrinker(&osc_cache_shrinker);
class_unregister_type(LUSTRE_OSC_NAME);
lu_kmem_fini(osc_caches);
ptlrpc_free_rq_pool(osc_rq_pool);
git://git.whamcloud.com / fs / lustre-release.git / blobdiff