-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
- *
+/*
* GPL HEADER START
*
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
* GPL HEADER END
*/
/*
- * Copyright 2008 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
+ *
+ * Copyright (c) 2011, 2014, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
* Implementation of cl_object for LOV layer.
*
* Author: Nikita Danilov <nikita.danilov@sun.com>
+ * Author: Jinshan Xiong <jinshan.xiong@whamcloud.com>
*/
#define DEBUG_SUBSYSTEM S_LOV
struct lov_object *lov,
const struct cl_object_conf *conf,
union lov_layout_state *state);
- void (*llo_delete)(const struct lu_env *env, struct lov_object *lov,
+ int (*llo_delete)(const struct lu_env *env, struct lov_object *lov,
union lov_layout_state *state);
void (*llo_fini)(const struct lu_env *env, struct lov_object *lov,
union lov_layout_state *state);
union lov_layout_state *state);
int (*llo_print)(const struct lu_env *env, void *cookie,
lu_printer_t p, const struct lu_object *o);
- struct cl_page *(*llo_page_init)(const struct lu_env *env,
- struct cl_object *obj,
- struct cl_page *page,
- cfs_page_t *vmpage);
+ int (*llo_page_init)(const struct lu_env *env, struct cl_object *obj,
+ struct cl_page *page, pgoff_t index);
int (*llo_lock_init)(const struct lu_env *env,
struct cl_object *obj, struct cl_lock *lock,
const struct cl_io *io);
struct cl_object *obj, struct cl_io *io);
int (*llo_getattr)(const struct lu_env *env, struct cl_object *obj,
struct cl_attr *attr);
+ int (*llo_find_cbdata)(const struct lu_env *env, struct cl_object *obj,
+ ldlm_iterator_t iter, void *data);
};
+static int lov_layout_wait(const struct lu_env *env, struct lov_object *lov);
+
+static void lov_lsm_put(struct lov_stripe_md *lsm)
+{
+ if (lsm != NULL)
+ lov_free_memmd(&lsm);
+}
+
/*****************************************************************************
*
* Lov object layout operations.
struct lov_object *lov,
union lov_layout_state *state)
{
- lov->u = *state;
}
static struct cl_object *lov_sub_find(const struct lu_env *env,
}
static int lov_init_sub(const struct lu_env *env, struct lov_object *lov,
- struct cl_object *stripe,
- struct lov_layout_raid0 *r0, int idx)
+ struct cl_object *stripe, struct lov_layout_raid0 *r0,
+ int idx)
{
- struct cl_object_header *hdr;
- struct cl_object_header *subhdr;
- struct cl_object_header *parent;
- struct lov_oinfo *oinfo;
- int result;
+ struct cl_object_header *hdr;
+ struct cl_object_header *subhdr;
+ struct cl_object_header *parent;
+ struct lov_oinfo *oinfo;
+ int result;
+
+ if (OBD_FAIL_CHECK(OBD_FAIL_LOV_INIT)) {
+ /* For sanity:test_206.
+ * Do not leave the object in cache to avoid accessing
+ * freed memory. This is because osc_object is referring to
+ * lov_oinfo of lsm_stripe_data which will be freed due to
+ * this failure. */
+ cl_object_kill(env, stripe);
+ cl_object_put(env, stripe);
+ return -EIO;
+ }
+
+ hdr = cl_object_header(lov2cl(lov));
+ subhdr = cl_object_header(stripe);
+
+ oinfo = lov->lo_lsm->lsm_oinfo[idx];
+ CDEBUG(D_INODE, DFID"@%p[%d] -> "DFID"@%p: ostid: "DOSTID
+ " idx: %d gen: %d\n",
+ PFID(&subhdr->coh_lu.loh_fid), subhdr, idx,
+ PFID(&hdr->coh_lu.loh_fid), hdr, POSTID(&oinfo->loi_oi),
+ oinfo->loi_ost_idx, oinfo->loi_ost_gen);
+
+ /* reuse ->coh_attr_guard to protect coh_parent change */
+ spin_lock(&subhdr->coh_attr_guard);
+ parent = subhdr->coh_parent;
+ if (parent == NULL) {
+ subhdr->coh_parent = hdr;
+ spin_unlock(&subhdr->coh_attr_guard);
+ subhdr->coh_nesting = hdr->coh_nesting + 1;
+ lu_object_ref_add(&stripe->co_lu, "lov-parent", lov);
+ r0->lo_sub[idx] = cl2lovsub(stripe);
+ r0->lo_sub[idx]->lso_super = lov;
+ r0->lo_sub[idx]->lso_index = idx;
+ result = 0;
+ } else {
+ struct lu_object *old_obj;
+ struct lov_object *old_lov;
+ unsigned int mask = D_INODE;
+
+ spin_unlock(&subhdr->coh_attr_guard);
+ old_obj = lu_object_locate(&parent->coh_lu, &lov_device_type);
+ LASSERT(old_obj != NULL);
+ old_lov = cl2lov(lu2cl(old_obj));
+ if (old_lov->lo_layout_invalid) {
+ /* the object's layout has already changed but isn't
+ * refreshed */
+ lu_object_unhash(env, &stripe->co_lu);
+ result = -EAGAIN;
+ } else {
+ mask = D_ERROR;
+ result = -EIO;
+ }
+
+ LU_OBJECT_DEBUG(mask, env, &stripe->co_lu,
+ "stripe %d is already owned.\n", idx);
+ LU_OBJECT_DEBUG(mask, env, old_obj, "owned.\n");
+ LU_OBJECT_HEADER(mask, env, lov2lu(lov), "try to own.\n");
+ cl_object_put(env, stripe);
+ }
+ return result;
+}
- hdr = cl_object_header(lov2cl(lov));
- subhdr = cl_object_header(stripe);
- parent = subhdr->coh_parent;
-
- oinfo = r0->lo_lsm->lsm_oinfo[idx];
- CDEBUG(D_INODE, DFID"@%p[%d] -> "DFID"@%p: id: "LPU64" seq: "LPU64
- " idx: %d gen: %d\n",
- PFID(&subhdr->coh_lu.loh_fid), subhdr, idx,
- PFID(&hdr->coh_lu.loh_fid), hdr,
- oinfo->loi_id, oinfo->loi_seq,
- oinfo->loi_ost_idx, oinfo->loi_ost_gen);
-
- if (parent == NULL) {
- subhdr->coh_parent = hdr;
- subhdr->coh_nesting = hdr->coh_nesting + 1;
- lu_object_ref_add(&stripe->co_lu, "lov-parent", lov);
- r0->lo_sub[idx] = cl2lovsub(stripe);
- r0->lo_sub[idx]->lso_super = lov;
- r0->lo_sub[idx]->lso_index = idx;
- result = 0;
- } else {
- CERROR("Stripe is already owned by other file (%i).\n", idx);
- LU_OBJECT_DEBUG(D_ERROR, env, &stripe->co_lu, "\n");
- LU_OBJECT_DEBUG(D_ERROR, env, lu_object_top(&parent->coh_lu),
- "old\n");
- LU_OBJECT_HEADER(D_ERROR, env, lov2lu(lov), "new\n");
- cl_object_put(env, stripe);
- result = -EIO;
- }
- return result;
+static int lov_page_slice_fixup(struct lov_object *lov,
+ struct cl_object *stripe)
+{
+ struct cl_object_header *hdr = cl_object_header(&lov->lo_cl);
+ struct cl_object *o;
+
+ if (stripe == NULL)
+ return hdr->coh_page_bufsize - lov->lo_cl.co_slice_off -
+ cfs_size_round(sizeof(struct lov_page));
+
+ cl_object_for_each(o, stripe)
+ o->co_slice_off += hdr->coh_page_bufsize;
+
+ return cl_object_header(stripe)->coh_page_bufsize;
}
static int lov_init_raid0(const struct lu_env *env,
struct lov_layout_raid0 *r0 = &state->raid0;
ENTRY;
- r0->lo_nr = conf->u.coc_md->lsm->lsm_stripe_count;
- r0->lo_lsm = conf->u.coc_md->lsm;
- LASSERT(r0->lo_nr <= lov_targets_nr(dev));
- OBD_ALLOC(r0->lo_sub, r0->lo_nr * sizeof r0->lo_sub[0]);
- if (r0->lo_sub != NULL) {
- result = 0;
- subconf->coc_inode = conf->coc_inode;
- cfs_spin_lock_init(&r0->lo_sub_lock);
+ if (lsm->lsm_magic != LOV_MAGIC_V1 && lsm->lsm_magic != LOV_MAGIC_V3) {
+ dump_lsm(D_ERROR, lsm);
+ LASSERTF(0, "magic mismatch, expected %d/%d, actual %d.\n",
+ LOV_MAGIC_V1, LOV_MAGIC_V3, lsm->lsm_magic);
+ }
+
+ LASSERT(lov->lo_lsm == NULL);
+ lov->lo_lsm = lsm_addref(lsm);
+ r0->lo_nr = lsm->lsm_stripe_count;
+ LASSERT(r0->lo_nr <= lov_targets_nr(dev));
+
+ lov->lo_layout_invalid = true;
+
+ OBD_ALLOC_LARGE(r0->lo_sub, r0->lo_nr * sizeof r0->lo_sub[0]);
+ if (r0->lo_sub != NULL) {
+ int psz = 0;
+
+ result = 0;
+ subconf->coc_inode = conf->coc_inode;
+ spin_lock_init(&r0->lo_sub_lock);
/*
* Create stripe cl_objects.
*/
struct lov_oinfo *oinfo = lsm->lsm_oinfo[i];
int ost_idx = oinfo->loi_ost_idx;
- fid_ostid_unpack(ofid, &oinfo->loi_oi,
- oinfo->loi_ost_idx);
- subdev = lovsub2cl_dev(dev->ld_target[ost_idx]);
- subconf->u.coc_oinfo = oinfo;
- LASSERTF(subdev != NULL, "not init ost %d\n", ost_idx);
- stripe = lov_sub_find(env, subdev, ofid, subconf);
- if (!IS_ERR(stripe))
- result = lov_init_sub(env, lov, stripe, r0, i);
- else
- result = PTR_ERR(stripe);
+ if (lov_oinfo_is_dummy(oinfo))
+ continue;
+
+ result = ostid_to_fid(ofid, &oinfo->loi_oi,
+ oinfo->loi_ost_idx);
+ if (result != 0)
+ GOTO(out, result);
+
+ subdev = lovsub2cl_dev(dev->ld_target[ost_idx]);
+ subconf->u.coc_oinfo = oinfo;
+ LASSERTF(subdev != NULL, "not init ost %d\n", ost_idx);
+ /* In the function below, .hs_keycmp resolves to
+ * lu_obj_hop_keycmp() */
+ /* coverity[overrun-buffer-val] */
+ stripe = lov_sub_find(env, subdev, ofid, subconf);
+ if (!IS_ERR(stripe)) {
+ result = lov_init_sub(env, lov, stripe, r0, i);
+ if (result == -EAGAIN) { /* try again */
+ --i;
+ result = 0;
+ continue;
+ }
+ } else {
+ result = PTR_ERR(stripe);
+ }
+
+ if (result == 0) {
+ int sz = lov_page_slice_fixup(lov, stripe);
+ LASSERT(ergo(psz > 0, psz == sz));
+ psz = sz;
+ }
}
- } else
- result = -ENOMEM;
- RETURN(result);
+ if (result == 0)
+ cl_object_header(&lov->lo_cl)->coh_page_bufsize += psz;
+ } else
+ result = -ENOMEM;
+out:
+ RETURN(result);
+}
+
+static int lov_init_released(const struct lu_env *env,
+ struct lov_device *dev, struct lov_object *lov,
+ const struct cl_object_conf *conf,
+ union lov_layout_state *state)
+{
+ struct lov_stripe_md *lsm = conf->u.coc_md->lsm;
+
+ LASSERT(lsm != NULL);
+ LASSERT(lsm_is_released(lsm));
+ LASSERT(lov->lo_lsm == NULL);
+
+ lov->lo_lsm = lsm_addref(lsm);
+ return 0;
}
-static void lov_delete_empty(const struct lu_env *env, struct lov_object *lov,
- union lov_layout_state *state)
+static struct cl_object *lov_find_subobj(const struct lu_env *env,
+ struct lov_object *lov,
+ struct lov_stripe_md *lsm,
+ int stripe_idx)
{
- LASSERT(lov->lo_type == LLT_EMPTY);
+ struct lov_device *dev = lu2lov_dev(lov2lu(lov)->lo_dev);
+ struct lov_oinfo *oinfo = lsm->lsm_oinfo[stripe_idx];
+ struct lov_thread_info *lti = lov_env_info(env);
+ struct lu_fid *ofid = <i->lti_fid;
+ struct cl_device *subdev;
+ int ost_idx;
+ int rc;
+ struct cl_object *result;
+
+ if (lov->lo_type != LLT_RAID0)
+ GOTO(out, result = NULL);
+
+ ost_idx = oinfo->loi_ost_idx;
+ rc = ostid_to_fid(ofid, &oinfo->loi_oi, ost_idx);
+ if (rc != 0)
+ GOTO(out, result = NULL);
+
+ subdev = lovsub2cl_dev(dev->ld_target[ost_idx]);
+ result = lov_sub_find(env, subdev, ofid, NULL);
+out:
+ if (result == NULL)
+ result = ERR_PTR(-EINVAL);
+ return result;
+}
+
+static int lov_delete_empty(const struct lu_env *env, struct lov_object *lov,
+ union lov_layout_state *state)
+{
+ LASSERT(lov->lo_type == LLT_EMPTY || lov->lo_type == LLT_RELEASED);
+
+ lov_layout_wait(env, lov);
+ return 0;
}
static void lov_subobject_kill(const struct lu_env *env, struct lov_object *lov,
- struct lovsub_object *los, int idx)
+ struct lovsub_object *los, int idx)
{
- struct cl_object *sub;
- struct lov_layout_raid0 *r0;
- struct lu_site *site;
- cfs_waitlink_t *waiter;
+ struct cl_object *sub;
+ struct lov_layout_raid0 *r0;
+ struct lu_site *site;
+ struct lu_site_bkt_data *bkt;
+ wait_queue_t *waiter;
r0 = &lov->u.raid0;
LASSERT(r0->lo_sub[idx] == los);
sub = lovsub2cl(los);
site = sub->co_lu.lo_dev->ld_site;
+ bkt = lu_site_bkt_from_fid(site, &sub->co_lu.lo_header->loh_fid);
cl_object_kill(env, sub);
/* release a reference to the sub-object and ... */
/* ... wait until it is actually destroyed---sub-object clears its
* ->lo_sub[] slot in lovsub_object_fini() */
- if (r0->lo_sub[idx] == los) {
- waiter = &lov_env_info(env)->lti_waiter;
- cfs_waitlink_init(waiter);
- cfs_waitq_add(&site->ls_marche_funebre, waiter);
- cfs_set_current_state(CFS_TASK_UNINT);
- while (1) {
- /* this wait-queue is signaled at the end of
- * lu_object_free(). */
- cfs_set_current_state(CFS_TASK_UNINT);
- cfs_spin_lock(&r0->lo_sub_lock);
- if (r0->lo_sub[idx] == los) {
- cfs_spin_unlock(&r0->lo_sub_lock);
- cfs_waitq_wait(waiter, CFS_TASK_UNINT);
- } else {
- cfs_spin_unlock(&r0->lo_sub_lock);
- cfs_set_current_state(CFS_TASK_RUNNING);
- break;
- }
- }
- cfs_waitq_del(&site->ls_marche_funebre, waiter);
- }
- LASSERT(r0->lo_sub[idx] == NULL);
+ if (r0->lo_sub[idx] == los) {
+ waiter = &lov_env_info(env)->lti_waiter;
+ init_waitqueue_entry(waiter, current);
+ add_wait_queue(&bkt->lsb_marche_funebre, waiter);
+ set_current_state(TASK_UNINTERRUPTIBLE);
+ while (1) {
+ /* this wait-queue is signaled at the end of
+ * lu_object_free(). */
+ set_current_state(TASK_UNINTERRUPTIBLE);
+ spin_lock(&r0->lo_sub_lock);
+ if (r0->lo_sub[idx] == los) {
+ spin_unlock(&r0->lo_sub_lock);
+ schedule();
+ } else {
+ spin_unlock(&r0->lo_sub_lock);
+ set_current_state(TASK_RUNNING);
+ break;
+ }
+ }
+ remove_wait_queue(&bkt->lsb_marche_funebre, waiter);
+ }
+ LASSERT(r0->lo_sub[idx] == NULL);
}
-static void lov_delete_raid0(const struct lu_env *env, struct lov_object *lov,
- union lov_layout_state *state)
+static int lov_delete_raid0(const struct lu_env *env, struct lov_object *lov,
+ union lov_layout_state *state)
{
- struct lov_layout_raid0 *r0 = &state->raid0;
- int i;
+ struct lov_layout_raid0 *r0 = &state->raid0;
+ struct lov_stripe_md *lsm = lov->lo_lsm;
+ int i;
- ENTRY;
+ ENTRY;
+
+ dump_lsm(D_INODE, lsm);
+
+ lov_layout_wait(env, lov);
if (r0->lo_sub != NULL) {
for (i = 0; i < r0->lo_nr; ++i) {
struct lovsub_object *los = r0->lo_sub[i];
- if (los != NULL)
+ if (los != NULL) {
+ cl_object_prune(env, &los->lso_cl);
/*
* If top-level object is to be evicted from
* the cache, so are its sub-objects.
*/
lov_subobject_kill(env, lov, los, i);
- }
- }
- EXIT;
+ }
+ }
+ }
+ RETURN(0);
}
static void lov_fini_empty(const struct lu_env *env, struct lov_object *lov,
union lov_layout_state *state)
{
- LASSERT(lov->lo_type == LLT_EMPTY);
+ LASSERT(lov->lo_type == LLT_EMPTY || lov->lo_type == LLT_RELEASED);
}
static void lov_fini_raid0(const struct lu_env *env, struct lov_object *lov,
union lov_layout_state *state)
{
- struct lov_layout_raid0 *r0 = &state->raid0;
+ struct lov_layout_raid0 *r0 = &state->raid0;
+ ENTRY;
- ENTRY;
- if (r0->lo_sub != NULL) {
- OBD_FREE(r0->lo_sub, r0->lo_nr * sizeof r0->lo_sub[0]);
- r0->lo_sub = NULL;
- }
- EXIT;
+ if (r0->lo_sub != NULL) {
+ OBD_FREE_LARGE(r0->lo_sub, r0->lo_nr * sizeof r0->lo_sub[0]);
+ r0->lo_sub = NULL;
+ }
+
+ dump_lsm(D_INODE, lov->lo_lsm);
+ lov_free_memmd(&lov->lo_lsm);
+
+ EXIT;
+}
+
+static void lov_fini_released(const struct lu_env *env, struct lov_object *lov,
+ union lov_layout_state *state)
+{
+ ENTRY;
+ dump_lsm(D_INODE, lov->lo_lsm);
+ lov_free_memmd(&lov->lo_lsm);
+ EXIT;
}
static int lov_print_empty(const struct lu_env *env, void *cookie,
lu_printer_t p, const struct lu_object *o)
{
- (*p)(env, cookie, "empty\n");
+ (*p)(env, cookie, "empty %d\n", lu2lov(o)->lo_layout_invalid);
return 0;
}
static int lov_print_raid0(const struct lu_env *env, void *cookie,
- lu_printer_t p, const struct lu_object *o)
+ lu_printer_t p, const struct lu_object *o)
{
- struct lov_object *lov = lu2lov(o);
- struct lov_layout_raid0 *r0 = lov_r0(lov);
- int i;
-
- (*p)(env, cookie, "stripes: %d:\n", r0->lo_nr);
- for (i = 0; i < r0->lo_nr; ++i) {
- struct lu_object *sub;
+ struct lov_object *lov = lu2lov(o);
+ struct lov_layout_raid0 *r0 = lov_r0(lov);
+ struct lov_stripe_md *lsm = lov->lo_lsm;
+ int i;
+
+ (*p)(env, cookie, "stripes: %d, %s, lsm{%p 0x%08X %d %u %u}:\n",
+ r0->lo_nr, lov->lo_layout_invalid ? "invalid" : "valid", lsm,
+ lsm->lsm_magic, atomic_read(&lsm->lsm_refc),
+ lsm->lsm_stripe_count, lsm->lsm_layout_gen);
+ for (i = 0; i < r0->lo_nr; ++i) {
+ struct lu_object *sub;
+
+ if (r0->lo_sub[i] != NULL) {
+ sub = lovsub2lu(r0->lo_sub[i]);
+ lu_object_print(env, cookie, p, sub);
+ } else {
+ (*p)(env, cookie, "sub %d absent\n", i);
+ }
+ }
+ return 0;
+}
- if (r0->lo_sub[i] != NULL) {
- sub = lovsub2lu(r0->lo_sub[i]);
- lu_object_print(env, cookie, p, sub);
- } else
- (*p)(env, cookie, "sub %d absent\n", i);
- }
- return 0;
+static int lov_print_released(const struct lu_env *env, void *cookie,
+ lu_printer_t p, const struct lu_object *o)
+{
+ struct lov_object *lov = lu2lov(o);
+ struct lov_stripe_md *lsm = lov->lo_lsm;
+
+ (*p)(env, cookie,
+ "released: %s, lsm{%p 0x%08X %d %u %u}:\n",
+ lov->lo_layout_invalid ? "invalid" : "valid", lsm,
+ lsm->lsm_magic, atomic_read(&lsm->lsm_refc),
+ lsm->lsm_stripe_count, lsm->lsm_layout_gen);
+ return 0;
}
/**
static int lov_attr_get_raid0(const struct lu_env *env, struct cl_object *obj,
struct cl_attr *attr)
{
- struct lov_object *lov = cl2lov(obj);
- struct lov_layout_raid0 *r0 = lov_r0(lov);
- struct lov_stripe_md *lsm = lov->u.raid0.lo_lsm;
- struct ost_lvb *lvb = &lov_env_info(env)->lti_lvb;
- __u64 kms;
- int result = 0;
+ struct lov_object *lov = cl2lov(obj);
+ struct lov_layout_raid0 *r0 = lov_r0(lov);
+ struct cl_attr *lov_attr = &r0->lo_attr;
+ int result = 0;
ENTRY;
- if (!r0->lo_attr_valid) {
- /*
- * Fill LVB with attributes already initialized by the upper
- * layer.
- */
- cl_attr2lvb(lvb, attr);
- kms = attr->cat_kms;
- /*
- * XXX that should be replaced with a loop over sub-objects,
- * doing cl_object_attr_get() on them. But for now, let's
- * reuse old lov code.
- */
+ /* this is called w/o holding type guard mutex, so it must be inside
+ * an on going IO otherwise lsm may be replaced.
+ * LU-2117: it turns out there exists one exception. For mmaped files,
+ * the lock of those files may be requested in the other file's IO
+ * context, and this function is called in ccc_lock_state(), it will
+ * hit this assertion.
+ * Anyway, it's still okay to call attr_get w/o type guard as layout
+ * can't go if locks exist. */
+ /* LASSERT(atomic_read(&lsm->lsm_refc) > 1); */
+
+ if (!r0->lo_attr_valid) {
+ struct lov_stripe_md *lsm = lov->lo_lsm;
+ struct ost_lvb *lvb = &lov_env_info(env)->lti_lvb;
+ __u64 kms = 0;
+
+ memset(lvb, 0, sizeof(*lvb));
+ /* XXX: timestamps can be negative by sanity:test_39m,
+ * how can it be? */
+ lvb->lvb_atime = LLONG_MIN;
+ lvb->lvb_ctime = LLONG_MIN;
+ lvb->lvb_mtime = LLONG_MIN;
+
+ /*
+ * XXX that should be replaced with a loop over sub-objects,
+ * doing cl_object_attr_get() on them. But for now, let's
+ * reuse old lov code.
+ */
+
+ /*
+ * XXX take lsm spin-lock to keep lov_merge_lvb_kms()
+ * happy. It's not needed, because new code uses
+ * ->coh_attr_guard spin-lock to protect consistency of
+ * sub-object attributes.
+ */
+ lov_stripe_lock(lsm);
+ result = lov_merge_lvb_kms(lsm, lvb, &kms);
+ lov_stripe_unlock(lsm);
+ if (result == 0) {
+ cl_lvb2attr(lov_attr, lvb);
+ lov_attr->cat_kms = kms;
+ r0->lo_attr_valid = 1;
+ }
+ }
+ if (result == 0) { /* merge results */
+ attr->cat_blocks = lov_attr->cat_blocks;
+ attr->cat_size = lov_attr->cat_size;
+ attr->cat_kms = lov_attr->cat_kms;
+ if (attr->cat_atime < lov_attr->cat_atime)
+ attr->cat_atime = lov_attr->cat_atime;
+ if (attr->cat_ctime < lov_attr->cat_ctime)
+ attr->cat_ctime = lov_attr->cat_ctime;
+ if (attr->cat_mtime < lov_attr->cat_mtime)
+ attr->cat_mtime = lov_attr->cat_mtime;
+ }
+ RETURN(result);
+}
- /*
- * XXX take lsm spin-lock to keep lov_merge_lvb_kms()
- * happy. It's not needed, because new code uses
- * ->coh_attr_guard spin-lock to protect consistency of
- * sub-object attributes.
- */
- lov_stripe_lock(lsm);
- result = lov_merge_lvb_kms(lsm, lvb, &kms);
- lov_stripe_unlock(lsm);
- if (result == 0) {
- cl_lvb2attr(attr, lvb);
- attr->cat_kms = kms;
- r0->lo_attr_valid = 1;
- r0->lo_attr = *attr;
- }
- } else
- *attr = r0->lo_attr;
- RETURN(result);
+static int lov_find_cbdata_empty(const struct lu_env *env,
+ struct cl_object *obj, ldlm_iterator_t iter,
+ void *data)
+{
+ return 0;
+}
+
+static int lov_find_cbdata_raid0(const struct lu_env *env,
+ struct cl_object *obj, ldlm_iterator_t iter,
+ void *data)
+{
+ struct lov_object *lov = cl2lov(obj);
+ struct lov_layout_raid0 *r0 = lov_r0(lov);
+ struct cl_object *subobj;
+ int i;
+ int rc = 0;
+
+ for (i = 0; i < r0->lo_nr; ++i) {
+ if (r0->lo_sub[i] == NULL)
+ continue;
+
+ subobj = lovsub2cl(r0->lo_sub[i]);
+
+ rc = cl_object_find_cbdata(env, subobj, iter, data);
+ if (rc != 0)
+ break;
+ }
+
+ return rc;
}
const static struct lov_layout_operations lov_dispatch[] = {
.llo_install = lov_install_empty,
.llo_print = lov_print_empty,
.llo_page_init = lov_page_init_empty,
- .llo_lock_init = NULL,
+ .llo_lock_init = lov_lock_init_empty,
.llo_io_init = lov_io_init_empty,
- .llo_getattr = lov_attr_get_empty
+ .llo_getattr = lov_attr_get_empty,
+ .llo_find_cbdata = lov_find_cbdata_empty
},
[LLT_RAID0] = {
.llo_init = lov_init_raid0,
.llo_page_init = lov_page_init_raid0,
.llo_lock_init = lov_lock_init_raid0,
.llo_io_init = lov_io_init_raid0,
- .llo_getattr = lov_attr_get_raid0
+ .llo_getattr = lov_attr_get_raid0,
+ .llo_find_cbdata = lov_find_cbdata_raid0
+ },
+ [LLT_RELEASED] = {
+ .llo_init = lov_init_released,
+ .llo_delete = lov_delete_empty,
+ .llo_fini = lov_fini_released,
+ .llo_install = lov_install_empty,
+ .llo_print = lov_print_released,
+ .llo_page_init = lov_page_init_empty,
+ .llo_lock_init = lov_lock_init_empty,
+ .llo_io_init = lov_io_init_released,
+ .llo_getattr = lov_attr_get_empty,
+ .llo_find_cbdata = lov_find_cbdata_empty
}
};
-
/**
* Performs a double-dispatch based on the layout type of an object.
*/
lov_dispatch[__llt].op(__VA_ARGS__); \
})
+/**
+ * Return lov_layout_type associated with a given lsm
+ */
+static enum lov_layout_type lov_type(struct lov_stripe_md *lsm)
+{
+ if (lsm == NULL)
+ return LLT_EMPTY;
+ if (lsm_is_released(lsm))
+ return LLT_RELEASED;
+ return LLT_RAID0;
+}
+
+static inline void lov_conf_freeze(struct lov_object *lov)
+{
+ if (lov->lo_owner != current)
+ down_read(&lov->lo_type_guard);
+}
+
+static inline void lov_conf_thaw(struct lov_object *lov)
+{
+ if (lov->lo_owner != current)
+ up_read(&lov->lo_type_guard);
+}
+
#define LOV_2DISPATCH_MAYLOCK(obj, op, lock, ...) \
({ \
struct lov_object *__obj = (obj); \
int __lock = !!(lock); \
typeof(lov_dispatch[0].op(__VA_ARGS__)) __result; \
\
- __lock &= __obj->lo_owner != cfs_current(); \
if (__lock) \
- cfs_down_read(&__obj->lo_type_guard); \
+ lov_conf_freeze(__obj); \
__result = LOV_2DISPATCH_NOLOCK(obj, op, __VA_ARGS__); \
if (__lock) \
- cfs_up_read(&__obj->lo_type_guard); \
+ lov_conf_thaw(__obj); \
__result; \
})
struct lov_object *__obj = (obj); \
enum lov_layout_type __llt; \
\
- if (__obj->lo_owner != cfs_current()) \
- cfs_down_read(&__obj->lo_type_guard); \
+ lov_conf_freeze(__obj); \
__llt = __obj->lo_type; \
LASSERT(0 <= __llt && __llt < ARRAY_SIZE(lov_dispatch)); \
lov_dispatch[__llt].op(__VA_ARGS__); \
- if (__obj->lo_owner != cfs_current()) \
- cfs_up_read(&__obj->lo_type_guard); \
+ lov_conf_thaw(__obj); \
} while (0)
-static int lov_layout_change(const struct lu_env *env,
- struct lov_object *obj, enum lov_layout_type llt,
- const struct cl_object_conf *conf)
+static void lov_conf_lock(struct lov_object *lov)
{
- int result;
- union lov_layout_state *state = &lov_env_info(env)->lti_state;
- const struct lov_layout_operations *old_ops;
- const struct lov_layout_operations *new_ops;
+ LASSERT(lov->lo_owner != current);
+ down_write(&lov->lo_type_guard);
+ LASSERT(lov->lo_owner == NULL);
+ lov->lo_owner = current;
+}
- LASSERT(0 <= obj->lo_type && obj->lo_type < ARRAY_SIZE(lov_dispatch));
- LASSERT(0 <= llt && llt < ARRAY_SIZE(lov_dispatch));
- ENTRY;
+static void lov_conf_unlock(struct lov_object *lov)
+{
+ lov->lo_owner = NULL;
+ up_write(&lov->lo_type_guard);
+}
- old_ops = &lov_dispatch[obj->lo_type];
- new_ops = &lov_dispatch[llt];
-
- result = new_ops->llo_init(env, lu2lov_dev(obj->lo_cl.co_lu.lo_dev),
- obj, conf, state);
- if (result == 0) {
- struct cl_object_header *hdr = cl_object_header(&obj->lo_cl);
- void *cookie;
- struct lu_env *nested;
- int refcheck;
-
- cookie = cl_env_reenter();
- nested = cl_env_get(&refcheck);
- if (!IS_ERR(nested))
- cl_object_prune(nested, &obj->lo_cl);
- else
- result = PTR_ERR(nested);
- cl_env_put(nested, &refcheck);
- cl_env_reexit(cookie);
-
- old_ops->llo_fini(env, obj, &obj->u);
- LASSERT(cfs_list_empty(&hdr->coh_locks));
- LASSERT(hdr->coh_tree.rnode == NULL);
- LASSERT(hdr->coh_pages == 0);
-
- new_ops->llo_install(env, obj, state);
- obj->lo_type = llt;
- } else
- new_ops->llo_fini(env, obj, state);
- RETURN(result);
+static int lov_layout_wait(const struct lu_env *env, struct lov_object *lov)
+{
+ struct l_wait_info lwi = { 0 };
+ ENTRY;
+
+ while (atomic_read(&lov->lo_active_ios) > 0) {
+ CDEBUG(D_INODE, "file:"DFID" wait for active IO, now: %d.\n",
+ PFID(lu_object_fid(lov2lu(lov))),
+ atomic_read(&lov->lo_active_ios));
+
+ l_wait_event(lov->lo_waitq,
+ atomic_read(&lov->lo_active_ios) == 0, &lwi);
+ }
+ RETURN(0);
+}
+
+static int lov_layout_change(const struct lu_env *unused,
+ struct lov_object *lov,
+ const struct cl_object_conf *conf)
+{
+ int result;
+ enum lov_layout_type llt = LLT_EMPTY;
+ union lov_layout_state *state = &lov->u;
+ const struct lov_layout_operations *old_ops;
+ const struct lov_layout_operations *new_ops;
+
+ void *cookie;
+ struct lu_env *env;
+ int refcheck;
+ ENTRY;
+
+ LASSERT(0 <= lov->lo_type && lov->lo_type < ARRAY_SIZE(lov_dispatch));
+
+ if (conf->u.coc_md != NULL)
+ llt = lov_type(conf->u.coc_md->lsm);
+ LASSERT(0 <= llt && llt < ARRAY_SIZE(lov_dispatch));
+
+ cookie = cl_env_reenter();
+ env = cl_env_get(&refcheck);
+ if (IS_ERR(env)) {
+ cl_env_reexit(cookie);
+ RETURN(PTR_ERR(env));
+ }
+
+ CDEBUG(D_INODE, DFID" from %s to %s\n",
+ PFID(lu_object_fid(lov2lu(lov))),
+ llt2str(lov->lo_type), llt2str(llt));
+
+ old_ops = &lov_dispatch[lov->lo_type];
+ new_ops = &lov_dispatch[llt];
+
+ result = cl_object_prune(env, &lov->lo_cl);
+ if (result != 0)
+ GOTO(out, result);
+
+ result = old_ops->llo_delete(env, lov, &lov->u);
+ if (result == 0) {
+ old_ops->llo_fini(env, lov, &lov->u);
+
+ LASSERT(atomic_read(&lov->lo_active_ios) == 0);
+
+ lov->lo_type = LLT_EMPTY;
+ /* page bufsize fixup */
+ cl_object_header(&lov->lo_cl)->coh_page_bufsize -=
+ lov_page_slice_fixup(lov, NULL);
+
+ result = new_ops->llo_init(env,
+ lu2lov_dev(lov->lo_cl.co_lu.lo_dev),
+ lov, conf, state);
+ if (result == 0) {
+ new_ops->llo_install(env, lov, state);
+ lov->lo_type = llt;
+ } else {
+ new_ops->llo_delete(env, lov, state);
+ new_ops->llo_fini(env, lov, state);
+ /* this file becomes an EMPTY file. */
+ }
+ }
+
+out:
+ cl_env_put(env, &refcheck);
+ cl_env_reexit(cookie);
+ RETURN(result);
}
/*****************************************************************************
* Lov object operations.
*
*/
-
int lov_object_init(const struct lu_env *env, struct lu_object *obj,
const struct lu_object_conf *conf)
{
struct lov_device *dev = lu2lov_dev(obj->lo_dev);
struct lov_object *lov = lu2lov(obj);
const struct cl_object_conf *cconf = lu2cl_conf(conf);
- union lov_layout_state *set = &lov_env_info(env)->lti_state;
+ union lov_layout_state *set = &lov->u;
const struct lov_layout_operations *ops;
int result;
ENTRY;
- cfs_init_rwsem(&lov->lo_type_guard);
+ init_rwsem(&lov->lo_type_guard);
+ atomic_set(&lov->lo_active_ios, 0);
+ init_waitqueue_head(&lov->lo_waitq);
+
+ cl_object_page_init(lu2cl(obj), sizeof(struct lov_page));
/* no locking is necessary, as object is being created */
- lov->lo_type = cconf->u.coc_md->lsm != NULL ? LLT_RAID0 : LLT_EMPTY;
+ lov->lo_type = lov_type(cconf->u.coc_md->lsm);
ops = &lov_dispatch[lov->lo_type];
result = ops->llo_init(env, dev, lov, cconf, set);
if (result == 0)
ops->llo_install(env, lov, set);
- else
- ops->llo_fini(env, lov, set);
RETURN(result);
}
static int lov_conf_set(const struct lu_env *env, struct cl_object *obj,
const struct cl_object_conf *conf)
{
- struct lov_object *lov = cl2lov(obj);
- int result;
-
- ENTRY;
- /*
- * Currently only LLT_EMPTY -> LLT_RAID0 transition is supported.
- */
- LASSERT(lov->lo_owner != cfs_current());
- cfs_down_write(&lov->lo_type_guard);
- LASSERT(lov->lo_owner == NULL);
- lov->lo_owner = cfs_current();
- if (lov->lo_type == LLT_EMPTY && conf->u.coc_md->lsm != NULL)
- result = lov_layout_change(env, lov, LLT_RAID0, conf);
- else
- result = -EOPNOTSUPP;
- lov->lo_owner = NULL;
- cfs_up_write(&lov->lo_type_guard);
- RETURN(result);
+ struct lov_stripe_md *lsm = NULL;
+ struct lov_object *lov = cl2lov(obj);
+ int result = 0;
+ ENTRY;
+
+ lov_conf_lock(lov);
+ if (conf->coc_opc == OBJECT_CONF_INVALIDATE) {
+ lov->lo_layout_invalid = true;
+ GOTO(out, result = 0);
+ }
+
+ if (conf->coc_opc == OBJECT_CONF_WAIT) {
+ if (lov->lo_layout_invalid &&
+ atomic_read(&lov->lo_active_ios) > 0) {
+ lov_conf_unlock(lov);
+ result = lov_layout_wait(env, lov);
+ lov_conf_lock(lov);
+ }
+ GOTO(out, result);
+ }
+
+ LASSERT(conf->coc_opc == OBJECT_CONF_SET);
+
+ if (conf->u.coc_md != NULL)
+ lsm = conf->u.coc_md->lsm;
+ if ((lsm == NULL && lov->lo_lsm == NULL) ||
+ ((lsm != NULL && lov->lo_lsm != NULL) &&
+ (lov->lo_lsm->lsm_layout_gen == lsm->lsm_layout_gen) &&
+ (lov->lo_lsm->lsm_pattern == lsm->lsm_pattern))) {
+ /* same version of layout */
+ lov->lo_layout_invalid = false;
+ GOTO(out, result = 0);
+ }
+
+ /* will change layout - check if there still exists active IO. */
+ if (atomic_read(&lov->lo_active_ios) > 0) {
+ lov->lo_layout_invalid = true;
+ GOTO(out, result = -EBUSY);
+ }
+
+ result = lov_layout_change(env, lov, conf);
+ lov->lo_layout_invalid = result != 0;
+ EXIT;
+
+out:
+ lov_conf_unlock(lov);
+ CDEBUG(D_INODE, DFID" lo_layout_invalid=%d\n",
+ PFID(lu_object_fid(lov2lu(lov))), lov->lo_layout_invalid);
+ RETURN(result);
}
static void lov_object_delete(const struct lu_env *env, struct lu_object *obj)
static int lov_object_print(const struct lu_env *env, void *cookie,
lu_printer_t p, const struct lu_object *o)
{
- return LOV_2DISPATCH(lu2lov(o), llo_print, env, cookie, p, o);
+ return LOV_2DISPATCH_NOLOCK(lu2lov(o), llo_print, env, cookie, p, o);
}
-struct cl_page *lov_page_init(const struct lu_env *env, struct cl_object *obj,
- struct cl_page *page, cfs_page_t *vmpage)
+int lov_page_init(const struct lu_env *env, struct cl_object *obj,
+ struct cl_page *page, pgoff_t index)
{
- return LOV_2DISPATCH(cl2lov(obj),
- llo_page_init, env, obj, page, vmpage);
+ return LOV_2DISPATCH_NOLOCK(cl2lov(obj), llo_page_init, env, obj, page,
+ index);
}
/**
* layer. Dispatches to the appropriate layout io initialization method.
*/
int lov_io_init(const struct lu_env *env, struct cl_object *obj,
- struct cl_io *io)
+ struct cl_io *io)
{
- CL_IO_SLICE_CLEAN(lov_env_io(env), lis_cl);
- /*
- * Do not take lock in case of CIT_MISC io, because
- *
- * - if this is an io for a glimpse, then we don't care;
- *
- * - if this not a glimpse (writepage or lock cancellation), then
- * layout change cannot happen because a page or a lock
- * already exist; and
- *
- * - lock ordering (lock mutex nests within layout rw-semaphore)
- * is obeyed in case of lock cancellation.
- */
- return LOV_2DISPATCH_MAYLOCK(cl2lov(obj), llo_io_init,
- io->ci_type != CIT_MISC, env, obj, io);
+ CL_IO_SLICE_CLEAN(lov_env_io(env), lis_cl);
+ return LOV_2DISPATCH_MAYLOCK(cl2lov(obj), llo_io_init,
+ !io->ci_ignore_layout, env, obj, io);
}
/**
return LOV_2DISPATCH_NOLOCK(cl2lov(obj), llo_getattr, env, obj, attr);
}
-static int lov_attr_set(const struct lu_env *env, struct cl_object *obj,
- const struct cl_attr *attr, unsigned valid)
+static int lov_attr_update(const struct lu_env *env, struct cl_object *obj,
+ const struct cl_attr *attr, unsigned valid)
{
- /*
- * No dispatch is required here, as no layout implements this.
- */
- return 0;
+ /*
+ * No dispatch is required here, as no layout implements this.
+ */
+ return 0;
}
int lov_lock_init(const struct lu_env *env, struct cl_object *obj,
- struct cl_lock *lock, const struct cl_io *io)
+ struct cl_lock *lock, const struct cl_io *io)
{
- return LOV_2DISPATCH(cl2lov(obj), llo_lock_init, env, obj, lock, io);
+ /* No need to lock because we've taken one refcount of layout. */
+ return LOV_2DISPATCH_NOLOCK(cl2lov(obj), llo_lock_init, env, obj, lock,
+ io);
+}
+
+/**
+ * We calculate on which OST the mapping will end. If the length of mapping
+ * is greater than (stripe_size * stripe_count) then the last_stripe will
+ * will be one just before start_stripe. Else we check if the mapping
+ * intersects each OST and find last_stripe.
+ * This function returns the last_stripe and also sets the stripe_count
+ * over which the mapping is spread
+ *
+ * \param lsm [in] striping information for the file
+ * \param fm_start [in] logical start of mapping
+ * \param fm_end [in] logical end of mapping
+ * \param start_stripe [in] starting stripe of the mapping
+ * \param stripe_count [out] the number of stripes across which to map is
+ * returned
+ *
+ * \retval last_stripe return the last stripe of the mapping
+ */
+static int fiemap_calc_last_stripe(struct lov_stripe_md *lsm,
+ loff_t fm_start, loff_t fm_end,
+ int start_stripe, int *stripe_count)
+{
+ int last_stripe;
+ loff_t obd_start;
+ loff_t obd_end;
+ int i, j;
+
+ if (fm_end - fm_start > lsm->lsm_stripe_size * lsm->lsm_stripe_count) {
+ last_stripe = (start_stripe < 1 ? lsm->lsm_stripe_count - 1 :
+ start_stripe - 1);
+ *stripe_count = lsm->lsm_stripe_count;
+ } else {
+ for (j = 0, i = start_stripe; j < lsm->lsm_stripe_count;
+ i = (i + 1) % lsm->lsm_stripe_count, j++) {
+ if ((lov_stripe_intersects(lsm, i, fm_start, fm_end,
+ &obd_start, &obd_end)) == 0)
+ break;
+ }
+ *stripe_count = j;
+ last_stripe = (start_stripe + j - 1) % lsm->lsm_stripe_count;
+ }
+
+ return last_stripe;
+}
+
+/**
+ * Set fe_device and copy extents from local buffer into main return buffer.
+ *
+ * \param fiemap [out] fiemap to hold all extents
+ * \param lcl_fm_ext [in] array of fiemap extents get from OSC layer
+ * \param ost_index [in] OST index to be written into the fm_device
+ * field for each extent
+ * \param ext_count [in] number of extents to be copied
+ * \param current_extent [in] where to start copying in the extent array
+ */
+static void fiemap_prepare_and_copy_exts(struct fiemap *fiemap,
+ struct fiemap_extent *lcl_fm_ext,
+ int ost_index, unsigned int ext_count,
+ int current_extent)
+{
+ char *to;
+ unsigned int ext;
+
+ for (ext = 0; ext < ext_count; ext++) {
+ lcl_fm_ext[ext].fe_device = ost_index;
+ lcl_fm_ext[ext].fe_flags |= FIEMAP_EXTENT_NET;
+ }
+
+ /* Copy fm_extent's from fm_local to return buffer */
+ to = (char *)fiemap + fiemap_count_to_size(current_extent);
+ memcpy(to, lcl_fm_ext, ext_count * sizeof(struct fiemap_extent));
+}
+
+#define FIEMAP_BUFFER_SIZE 4096
+
+/**
+ * Non-zero fe_logical indicates that this is a continuation FIEMAP
+ * call. The local end offset and the device are sent in the first
+ * fm_extent. This function calculates the stripe number from the index.
+ * This function returns a stripe_no on which mapping is to be restarted.
+ *
+ * This function returns fm_end_offset which is the in-OST offset at which
+ * mapping should be restarted. If fm_end_offset=0 is returned then caller
+ * will re-calculate proper offset in next stripe.
+ * Note that the first extent is passed to lov_get_info via the value field.
+ *
+ * \param fiemap [in] fiemap request header
+ * \param lsm [in] striping information for the file
+ * \param fm_start [in] logical start of mapping
+ * \param fm_end [in] logical end of mapping
+ * \param start_stripe [out] starting stripe will be returned in this
+ */
+static loff_t fiemap_calc_fm_end_offset(struct fiemap *fiemap,
+ struct lov_stripe_md *lsm,
+ loff_t fm_start, loff_t fm_end,
+ int *start_stripe)
+{
+ loff_t local_end = fiemap->fm_extents[0].fe_logical;
+ loff_t lun_start;
+ loff_t lun_end;
+ loff_t fm_end_offset;
+ int stripe_no = -1;
+ int i;
+
+ if (fiemap->fm_extent_count == 0 ||
+ fiemap->fm_extents[0].fe_logical == 0)
+ return 0;
+
+ /* Find out stripe_no from ost_index saved in the fe_device */
+ for (i = 0; i < lsm->lsm_stripe_count; i++) {
+ struct lov_oinfo *oinfo = lsm->lsm_oinfo[i];
+
+ if (lov_oinfo_is_dummy(oinfo))
+ continue;
+
+ if (oinfo->loi_ost_idx == fiemap->fm_extents[0].fe_device) {
+ stripe_no = i;
+ break;
+ }
+ }
+
+ if (stripe_no == -1)
+ return -EINVAL;
+
+ /* If we have finished mapping on previous device, shift logical
+ * offset to start of next device */
+ if (lov_stripe_intersects(lsm, stripe_no, fm_start, fm_end,
+ &lun_start, &lun_end) != 0 &&
+ local_end < lun_end) {
+ fm_end_offset = local_end;
+ *start_stripe = stripe_no;
+ } else {
+ /* This is a special value to indicate that caller should
+ * calculate offset in next stripe. */
+ fm_end_offset = 0;
+ *start_stripe = (stripe_no + 1) % lsm->lsm_stripe_count;
+ }
+
+ return fm_end_offset;
+}
+
+/**
+ * Break down the FIEMAP request and send appropriate calls to individual OSTs.
+ * This also handles the restarting of FIEMAP calls in case mapping overflows
+ * the available number of extents in single call.
+ *
+ * \param env [in] lustre environment
+ * \param obj [in] file object
+ * \param fmkey [in] fiemap request header and other info
+ * \param fiemap [out] fiemap buffer holding retrived map extents
+ * \param buflen [in/out] max buffer length of @fiemap, when iterate
+ * each OST, it is used to limit max map needed
+ * \retval 0 success
+ * \retval < 0 error
+ */
+static int lov_object_fiemap(const struct lu_env *env, struct cl_object *obj,
+ struct ll_fiemap_info_key *fmkey,
+ struct fiemap *fiemap, size_t *buflen)
+{
+ struct lov_stripe_md *lsm;
+ struct cl_object *subobj = NULL;
+ struct lov_obd *lov = lu2lov_dev(obj->co_lu.lo_dev)->ld_lov;
+ struct fiemap *fm_local = NULL;
+ struct fiemap_extent *lcl_fm_ext;
+ loff_t fm_start;
+ loff_t fm_end;
+ loff_t fm_length;
+ loff_t fm_end_offset;
+ int count_local;
+ int ost_index = 0;
+ int start_stripe;
+ int current_extent = 0;
+ int rc = 0;
+ int last_stripe;
+ int cur_stripe = 0;
+ int cur_stripe_wrap = 0;
+ int stripe_count;
+ unsigned int buffer_size = FIEMAP_BUFFER_SIZE;
+ /* Whether have we collected enough extents */
+ bool enough = false;
+ /* EOF for object */
+ bool ost_eof = false;
+ /* done with required mapping for this OST? */
+ bool ost_done = false;
+ ENTRY;
+
+ lsm = lov_lsm_addref(cl2lov(obj));
+ if (lsm == NULL)
+ RETURN(-ENODATA);
+
+ /**
+ * If the stripe_count > 1 and the application does not understand
+ * DEVICE_ORDER flag, it cannot interpret the extents correctly.
+ */
+ if (lsm->lsm_stripe_count > 1 && !(fiemap->fm_flags &
+ FIEMAP_FLAG_DEVICE_ORDER))
+ GOTO(out, rc = -ENOTSUPP);
+
+ if (lsm_is_released(lsm)) {
+ if (fiemap->fm_start < fmkey->lfik_oa.o_size) {
+ /**
+ * released file, return a minimal FIEMAP if
+ * request fits in file-size.
+ */
+ fiemap->fm_mapped_extents = 1;
+ fiemap->fm_extents[0].fe_logical = fiemap->fm_start;
+ if (fiemap->fm_start + fiemap->fm_length <
+ fmkey->lfik_oa.o_size)
+ fiemap->fm_extents[0].fe_length =
+ fiemap->fm_length;
+ else
+ fiemap->fm_extents[0].fe_length =
+ fmkey->lfik_oa.o_size -
+ fiemap->fm_start;
+ fiemap->fm_extents[0].fe_flags |=
+ FIEMAP_EXTENT_UNKNOWN | FIEMAP_EXTENT_LAST;
+ }
+ GOTO(out, rc = 0);
+ }
+
+ if (fiemap_count_to_size(fiemap->fm_extent_count) < buffer_size)
+ buffer_size = fiemap_count_to_size(fiemap->fm_extent_count);
+
+ OBD_ALLOC_LARGE(fm_local, buffer_size);
+ if (fm_local == NULL)
+ GOTO(out, rc = -ENOMEM);
+ lcl_fm_ext = &fm_local->fm_extents[0];
+ count_local = fiemap_size_to_count(buffer_size);
+
+ fm_start = fiemap->fm_start;
+ fm_length = fiemap->fm_length;
+ /* Calculate start stripe, last stripe and length of mapping */
+ start_stripe = lov_stripe_number(lsm, fm_start);
+ fm_end = (fm_length == ~0ULL) ? fmkey->lfik_oa.o_size :
+ fm_start + fm_length - 1;
+ /* If fm_length != ~0ULL but fm_start_fm_length-1 exceeds file size */
+ if (fm_end > fmkey->lfik_oa.o_size)
+ fm_end = fmkey->lfik_oa.o_size;
+
+ last_stripe = fiemap_calc_last_stripe(lsm, fm_start, fm_end,
+ start_stripe, &stripe_count);
+ fm_end_offset = fiemap_calc_fm_end_offset(fiemap, lsm, fm_start, fm_end,
+ &start_stripe);
+ if (fm_end_offset == -EINVAL)
+ GOTO(out, rc = -EINVAL);
+
+ /**
+ * Requested extent count exceeds the fiemap buffer size, shrink our
+ * ambition.
+ */
+ if (fiemap_count_to_size(fiemap->fm_extent_count) > *buflen)
+ fiemap->fm_extent_count = fiemap_size_to_count(*buflen);
+ if (fiemap->fm_extent_count == 0)
+ count_local = 0;
+
+ /* Check each stripe */
+ for (cur_stripe = start_stripe; stripe_count > 0;
+ --stripe_count,
+ cur_stripe = (cur_stripe + 1) % lsm->lsm_stripe_count) {
+ loff_t req_fm_len; /* Stores length of required mapping */
+ loff_t len_mapped_single_call;
+ loff_t lun_start;
+ loff_t lun_end;
+ loff_t obd_object_end;
+ unsigned int ext_count;
+
+ cur_stripe_wrap = cur_stripe;
+
+ /* Find out range of mapping on this stripe */
+ if ((lov_stripe_intersects(lsm, cur_stripe, fm_start, fm_end,
+ &lun_start, &obd_object_end)) == 0)
+ continue;
+
+ if (lov_oinfo_is_dummy(lsm->lsm_oinfo[cur_stripe]))
+ GOTO(out, rc = -EIO);
+
+ /* If this is a continuation FIEMAP call and we are on
+ * starting stripe then lun_start needs to be set to
+ * fm_end_offset */
+ if (fm_end_offset != 0 && cur_stripe == start_stripe)
+ lun_start = fm_end_offset;
+
+ if (fm_length != ~0ULL) {
+ /* Handle fm_start + fm_length overflow */
+ if (fm_start + fm_length < fm_start)
+ fm_length = ~0ULL - fm_start;
+ lun_end = lov_size_to_stripe(lsm, fm_start + fm_length,
+ cur_stripe);
+ } else {
+ lun_end = ~0ULL;
+ }
+
+ if (lun_start == lun_end)
+ continue;
+
+ req_fm_len = obd_object_end - lun_start;
+ fm_local->fm_length = 0;
+ len_mapped_single_call = 0;
+
+ /* find lobsub object */
+ subobj = lov_find_subobj(env, cl2lov(obj), lsm,
+ cur_stripe);
+ if (IS_ERR(subobj))
+ GOTO(out, rc = PTR_ERR(subobj));
+ /* If the output buffer is very large and the objects have many
+ * extents we may need to loop on a single OST repeatedly */
+ ost_eof = false;
+ ost_done = false;
+ do {
+ if (fiemap->fm_extent_count > 0) {
+ /* Don't get too many extents. */
+ if (current_extent + count_local >
+ fiemap->fm_extent_count)
+ count_local = fiemap->fm_extent_count -
+ current_extent;
+ }
+
+ lun_start += len_mapped_single_call;
+ fm_local->fm_length = req_fm_len -
+ len_mapped_single_call;
+ req_fm_len = fm_local->fm_length;
+ fm_local->fm_extent_count = enough ? 1 : count_local;
+ fm_local->fm_mapped_extents = 0;
+ fm_local->fm_flags = fiemap->fm_flags;
+
+ ost_index = lsm->lsm_oinfo[cur_stripe]->loi_ost_idx;
+
+ if (ost_index < 0 ||
+ ost_index >= lov->desc.ld_tgt_count)
+ GOTO(obj_put, rc = -EINVAL);
+ /* If OST is inactive, return extent with UNKNOWN
+ * flag. */
+ if (!lov->lov_tgts[ost_index]->ltd_active) {
+ fm_local->fm_flags |= FIEMAP_EXTENT_LAST;
+ fm_local->fm_mapped_extents = 1;
+
+ lcl_fm_ext[0].fe_logical = lun_start;
+ lcl_fm_ext[0].fe_length = obd_object_end -
+ lun_start;
+ lcl_fm_ext[0].fe_flags |= FIEMAP_EXTENT_UNKNOWN;
+
+ goto inactive_tgt;
+ }
+
+ fm_local->fm_start = lun_start;
+ fm_local->fm_flags &= ~FIEMAP_FLAG_DEVICE_ORDER;
+ memcpy(&fmkey->lfik_fiemap, fm_local,
+ sizeof(*fm_local));
+ *buflen = fiemap_count_to_size(
+ fm_local->fm_extent_count);
+
+ rc = cl_object_fiemap(env, subobj, fmkey, fm_local,
+ buflen);
+ if (rc != 0)
+ GOTO(obj_put, rc);
+inactive_tgt:
+ ext_count = fm_local->fm_mapped_extents;
+ if (ext_count == 0) {
+ ost_done = true;
+ /* If last stripe has hold at the end,
+ * we need to return */
+ if (cur_stripe_wrap == last_stripe) {
+ fiemap->fm_mapped_extents = 0;
+ goto finish;
+ }
+ break;
+ } else if (enough) {
+ /*
+ * We've collected enough extents and there are
+ * more extents after it.
+ */
+ goto finish;
+ }
+
+ /* If we just need num of extents, got to next device */
+ if (fiemap->fm_extent_count == 0) {
+ current_extent += ext_count;
+ break;
+ }
+
+ /* prepare to copy retrived map extents */
+ len_mapped_single_call =
+ lcl_fm_ext[ext_count - 1].fe_logical -
+ lun_start + lcl_fm_ext[ext_count - 1].fe_length;
+
+ /* Have we finished mapping on this device? */
+ if (req_fm_len <= len_mapped_single_call)
+ ost_done = true;
+
+ /* Clear the EXTENT_LAST flag which can be present on
+ * the last extent */
+ if (lcl_fm_ext[ext_count - 1].fe_flags &
+ FIEMAP_EXTENT_LAST)
+ lcl_fm_ext[ext_count - 1].fe_flags &=
+ ~FIEMAP_EXTENT_LAST;
+ if (lov_stripe_size(lsm,
+ lcl_fm_ext[ext_count - 1].fe_logical +
+ lcl_fm_ext[ext_count - 1].fe_length,
+ cur_stripe) >= fmkey->lfik_oa.o_size)
+ ost_eof = true;
+
+ fiemap_prepare_and_copy_exts(fiemap, lcl_fm_ext,
+ ost_index, ext_count,
+ current_extent);
+ current_extent += ext_count;
+
+ /* Ran out of available extents? */
+ if (current_extent >= fiemap->fm_extent_count)
+ enough = true;
+ } while (!ost_done && !ost_eof);
+
+ cl_object_put(env, subobj);
+ subobj = NULL;
+
+ if (cur_stripe_wrap == last_stripe)
+ goto finish;
+ } /* for each stripe */
+finish:
+ /* Indicate that we are returning device offsets unless file just has
+ * single stripe */
+ if (lsm->lsm_stripe_count > 1)
+ fiemap->fm_flags |= FIEMAP_FLAG_DEVICE_ORDER;
+
+ if (fiemap->fm_extent_count == 0)
+ goto skip_last_device_calc;
+
+ /* Check if we have reached the last stripe and whether mapping for that
+ * stripe is done. */
+ if ((cur_stripe_wrap == last_stripe) && (ost_done || ost_eof))
+ fiemap->fm_extents[current_extent - 1].fe_flags |=
+ FIEMAP_EXTENT_LAST;
+skip_last_device_calc:
+ fiemap->fm_mapped_extents = current_extent;
+obj_put:
+ if (subobj != NULL)
+ cl_object_put(env, subobj);
+out:
+ if (fm_local != NULL)
+ OBD_FREE_LARGE(fm_local, buffer_size);
+
+ lov_lsm_put(lsm);
+
+ return rc;
+}
+
+static int lov_object_getstripe(const struct lu_env *env, struct cl_object *obj,
+ struct lov_user_md __user *lum)
+{
+ struct lov_object *lov = cl2lov(obj);
+ struct lov_stripe_md *lsm;
+ int rc = 0;
+ ENTRY;
+
+ lsm = lov_lsm_addref(lov);
+ if (lsm == NULL)
+ RETURN(-ENODATA);
+
+ rc = lov_getstripe(cl2lov(obj), lsm, lum);
+ lov_lsm_put(lsm);
+ RETURN(rc);
+}
+
+static int lov_object_layout_get(const struct lu_env *env,
+ struct cl_object *obj,
+ struct cl_layout *cl)
+{
+ struct lov_object *lov = cl2lov(obj);
+ struct lov_stripe_md *lsm = lov_lsm_addref(lov);
+ struct lu_buf *buf = &cl->cl_buf;
+ ssize_t rc;
+ ENTRY;
+
+ if (lsm == NULL) {
+ cl->cl_size = 0;
+ cl->cl_layout_gen = CL_LAYOUT_GEN_EMPTY;
+ cl->cl_is_released = false;
+
+ RETURN(0);
+ }
+
+ cl->cl_size = lov_mds_md_size(lsm->lsm_stripe_count, lsm->lsm_magic);
+ cl->cl_layout_gen = lsm->lsm_layout_gen;
+ cl->cl_is_released = lsm_is_released(lsm);
+
+ rc = lov_lsm_pack(lsm, buf->lb_buf, buf->lb_len);
+ lov_lsm_put(lsm);
+
+ RETURN(rc < 0 ? rc : 0);
+}
+
+static loff_t lov_object_maxbytes(struct cl_object *obj)
+{
+ struct lov_object *lov = cl2lov(obj);
+ struct lov_stripe_md *lsm = lov_lsm_addref(lov);
+ loff_t maxbytes;
+
+ if (lsm == NULL)
+ return LLONG_MAX;
+
+ maxbytes = lsm->lsm_maxbytes;
+
+ lov_lsm_put(lsm);
+
+ return maxbytes;
+}
+
+static int lov_object_find_cbdata(const struct lu_env *env,
+ struct cl_object *obj, ldlm_iterator_t iter,
+ void *data)
+{
+ int rc;
+ ENTRY;
+
+ /* call cl_object_find_cbdata for sub obj */
+ rc = LOV_2DISPATCH_NOLOCK(cl2lov(obj), llo_find_cbdata,
+ env, obj, iter, data);
+ RETURN(rc);
}
static const struct cl_object_operations lov_ops = {
- .coo_page_init = lov_page_init,
- .coo_lock_init = lov_lock_init,
- .coo_io_init = lov_io_init,
- .coo_attr_get = lov_attr_get,
- .coo_attr_set = lov_attr_set,
- .coo_conf_set = lov_conf_set
+ .coo_page_init = lov_page_init,
+ .coo_lock_init = lov_lock_init,
+ .coo_io_init = lov_io_init,
+ .coo_attr_get = lov_attr_get,
+ .coo_attr_update = lov_attr_update,
+ .coo_conf_set = lov_conf_set,
+ .coo_getstripe = lov_object_getstripe,
+ .coo_layout_get = lov_object_layout_get,
+ .coo_maxbytes = lov_object_maxbytes,
+ .coo_find_cbdata = lov_object_find_cbdata,
+ .coo_fiemap = lov_object_fiemap,
};
static const struct lu_object_operations lov_lu_obj_ops = {
};
struct lu_object *lov_object_alloc(const struct lu_env *env,
- const struct lu_object_header *unused,
- struct lu_device *dev)
+ const struct lu_object_header *unused,
+ struct lu_device *dev)
{
- struct lov_object *lov;
- struct lu_object *obj;
+ struct lov_object *lov;
+ struct lu_object *obj;
+
+ ENTRY;
+ OBD_SLAB_ALLOC_PTR_GFP(lov, lov_object_kmem, GFP_NOFS);
+ if (lov != NULL) {
+ obj = lov2lu(lov);
+ lu_object_init(obj, NULL, dev);
+ lov->lo_cl.co_ops = &lov_ops;
+ lov->lo_type = -1; /* invalid, to catch uninitialized type */
+ /*
+ * object io operation vector (cl_object::co_iop) is installed
+ * later in lov_object_init(), as different vectors are used
+ * for object with different layouts.
+ */
+ obj->lo_ops = &lov_lu_obj_ops;
+ } else
+ obj = NULL;
+ RETURN(obj);
+}
- ENTRY;
- OBD_SLAB_ALLOC_PTR_GFP(lov, lov_object_kmem, CFS_ALLOC_IO);
- if (lov != NULL) {
- obj = lov2lu(lov);
- lu_object_init(obj, NULL, dev);
- lov->lo_cl.co_ops = &lov_ops;
- lov->lo_type = -1; /* invalid, to catch uninitialized type */
- /*
- * object io operation vector (cl_object::co_iop) is installed
- * later in lov_object_init(), as different vectors are used
- * for object with different layouts.
- */
- obj->lo_ops = &lov_lu_obj_ops;
- } else
- obj = NULL;
- RETURN(obj);
+struct lov_stripe_md *lov_lsm_addref(struct lov_object *lov)
+{
+ struct lov_stripe_md *lsm = NULL;
+
+ lov_conf_freeze(lov);
+ if (lov->lo_lsm != NULL) {
+ lsm = lsm_addref(lov->lo_lsm);
+ CDEBUG(D_INODE, "lsm %p addref %d/%d by %p.\n",
+ lsm, atomic_read(&lsm->lsm_refc),
+ lov->lo_layout_invalid, current);
+ }
+ lov_conf_thaw(lov);
+ return lsm;
+}
+
+int lov_read_and_clear_async_rc(struct cl_object *clob)
+{
+ struct lu_object *luobj;
+ int rc = 0;
+ ENTRY;
+
+ luobj = lu_object_locate(&cl_object_header(clob)->coh_lu,
+ &lov_device_type);
+ if (luobj != NULL) {
+ struct lov_object *lov = lu2lov(luobj);
+
+ lov_conf_freeze(lov);
+ switch (lov->lo_type) {
+ case LLT_RAID0: {
+ struct lov_stripe_md *lsm;
+ int i;
+
+ lsm = lov->lo_lsm;
+ LASSERT(lsm != NULL);
+ for (i = 0; i < lsm->lsm_stripe_count; i++) {
+ struct lov_oinfo *loi = lsm->lsm_oinfo[i];
+
+ if (lov_oinfo_is_dummy(loi))
+ continue;
+
+ if (loi->loi_ar.ar_rc && !rc)
+ rc = loi->loi_ar.ar_rc;
+ loi->loi_ar.ar_rc = 0;
+ }
+ }
+ case LLT_RELEASED:
+ case LLT_EMPTY:
+ break;
+ default:
+ LBUG();
+ }
+ lov_conf_thaw(lov);
+ }
+ RETURN(rc);
}
+EXPORT_SYMBOL(lov_read_and_clear_async_rc);
/** @} lov */
git://git.whamcloud.com / fs / lustre-release.git / blobdiff