[XFS_DEFER_OPS_TYPE_ATTR] = &xfs_attr_defer_type,
};
-static bool
+/*
+ * Ensure there's a log intent item associated with this deferred work item if
+ * the operation must be restarted on crash. Returns 1 if there's a log item;
+ * 0 if there isn't; or a negative errno.
+ */
+static int
xfs_defer_create_intent(
struct xfs_trans *tp,
struct xfs_defer_pending *dfp,
bool sort)
{
const struct xfs_defer_op_type *ops = defer_op_types[dfp->dfp_type];
+ struct xfs_log_item *lip;
- if (!dfp->dfp_intent)
- dfp->dfp_intent = ops->create_intent(tp, &dfp->dfp_work,
- dfp->dfp_count, sort);
- return dfp->dfp_intent != NULL;
+ if (dfp->dfp_intent)
+ return 1;
+
+ lip = ops->create_intent(tp, &dfp->dfp_work, dfp->dfp_count, sort);
+ if (!lip)
+ return 0;
+ if (IS_ERR(lip))
+ return PTR_ERR(lip);
+
+ dfp->dfp_intent = lip;
+ return 1;
}
/*
* For each pending item in the intake list, log its intent item and the
* associated extents, then add the entire intake list to the end of
* the pending list.
+ *
+ * Returns 1 if at least one log item was associated with the deferred work;
+ * 0 if there are no log items; or a negative errno.
*/
-static bool
+static int
xfs_defer_create_intents(
struct xfs_trans *tp)
{
struct xfs_defer_pending *dfp;
- bool ret = false;
+ int ret = 0;
list_for_each_entry(dfp, &tp->t_dfops, dfp_list) {
+ int ret2;
+
trace_xfs_defer_create_intent(tp->t_mountp, dfp);
- ret |= xfs_defer_create_intent(tp, dfp, true);
+ ret2 = xfs_defer_create_intent(tp, dfp, true);
+ if (ret2 < 0)
+ return ret2;
+ ret |= ret2;
}
return ret;
}
dfp->dfp_count--;
error = ops->finish_item(tp, dfp->dfp_done, li, &state);
if (error == -EAGAIN) {
+ int ret;
+
/*
* Caller wants a fresh transaction; put the work item
* back on the list and log a new log intent item to
dfp->dfp_count++;
dfp->dfp_done = NULL;
dfp->dfp_intent = NULL;
- xfs_defer_create_intent(tp, dfp, false);
+ ret = xfs_defer_create_intent(tp, dfp, false);
+ if (ret < 0)
+ error = ret;
}
if (error)
* of time that any one intent item can stick around in memory,
* pinning the log tail.
*/
- bool has_intents = xfs_defer_create_intents(*tp);
+ int has_intents = xfs_defer_create_intents(*tp);
list_splice_init(&(*tp)->t_dfops, &dop_pending);
+ if (has_intents < 0) {
+ error = has_intents;
+ goto out_shutdown;
+ }
if (has_intents || dfp) {
error = xfs_defer_trans_roll(tp);
if (error)
if (list_empty(&tp->t_dfops))
return NULL;
+ error = xfs_defer_create_intents(tp);
+ if (error < 0)
+ return ERR_PTR(error);
+
/* Create an object to capture the defer ops. */
dfc = kmem_zalloc(sizeof(*dfc), KM_NOFS);
INIT_LIST_HEAD(&dfc->dfc_list);
INIT_LIST_HEAD(&dfc->dfc_dfops);
- xfs_defer_create_intents(tp);
-
/* Move the dfops chain and transaction state to the capture struct. */
list_splice_init(&tp->t_dfops, &dfc->dfc_dfops);
dfc->dfc_tpflags = tp->t_flags & XFS_TRANS_LOWMODE;
/* If we don't capture anything, commit transaction and exit. */
dfc = xfs_defer_ops_capture(tp);
+ if (IS_ERR(dfc)) {
+ xfs_trans_cancel(tp);
+ return PTR_ERR(dfc);
+ }
if (!dfc)
return xfs_trans_commit(tp);
return container_of(lip, struct xfs_attri_log_item, attri_item);
}
+/*
+ * Shared xattr name/value buffers for logged extended attribute operations
+ *
+ * When logging updates to extended attributes, we can create quite a few
+ * attribute log intent items for a single xattr update. To avoid cycling the
+ * memory allocator and memcpy overhead, the name (and value, for setxattr)
+ * are kept in a refcounted object that is shared across all related log items
+ * and the upper-level deferred work state structure. The shared buffer has
+ * a control structure, followed by the name, and then the value.
+ */
+
+static inline struct xfs_attri_log_nameval *
+xfs_attri_log_nameval_get(
+ struct xfs_attri_log_nameval *nv)
+{
+ if (!refcount_inc_not_zero(&nv->refcount))
+ return NULL;
+ return nv;
+}
+
+static inline void
+xfs_attri_log_nameval_put(
+ struct xfs_attri_log_nameval *nv)
+{
+ if (!nv)
+ return;
+ if (refcount_dec_and_test(&nv->refcount))
+ kvfree(nv);
+}
+
+static inline struct xfs_attri_log_nameval *
+xfs_attri_log_nameval_alloc(
+ const void *name,
+ unsigned int name_len,
+ const void *value,
+ unsigned int value_len)
+{
+ struct xfs_attri_log_nameval *nv;
+
+ /*
+ * This could be over 64kB in length, so we have to use kvmalloc() for
+ * this. But kvmalloc() utterly sucks, so we use our own version.
+ */
+ nv = xlog_kvmalloc(sizeof(struct xfs_attri_log_nameval) +
+ name_len + value_len);
+ if (!nv)
+ return nv;
+
+ nv->name.i_addr = nv + 1;
+ nv->name.i_len = name_len;
+ nv->name.i_type = XLOG_REG_TYPE_ATTR_NAME;
+ memcpy(nv->name.i_addr, name, name_len);
+
+ if (value_len) {
+ nv->value.i_addr = nv->name.i_addr + name_len;
+ nv->value.i_len = value_len;
+ memcpy(nv->value.i_addr, value, value_len);
+ } else {
+ nv->value.i_addr = NULL;
+ nv->value.i_len = 0;
+ }
+ nv->value.i_type = XLOG_REG_TYPE_ATTR_VALUE;
+
+ refcount_set(&nv->refcount, 1);
+ return nv;
+}
+
STATIC void
xfs_attri_item_free(
struct xfs_attri_log_item *attrip)
{
kmem_free(attrip->attri_item.li_lv_shadow);
- kvfree(attrip);
+ xfs_attri_log_nameval_put(attrip->attri_nameval);
+ kmem_cache_free(xfs_attri_cache, attrip);
}
/*
int *nbytes)
{
struct xfs_attri_log_item *attrip = ATTRI_ITEM(lip);
+ struct xfs_attri_log_nameval *nv = attrip->attri_nameval;
*nvecs += 2;
*nbytes += sizeof(struct xfs_attri_log_format) +
- xlog_calc_iovec_len(attrip->attri_name_len);
+ xlog_calc_iovec_len(nv->name.i_len);
- if (!attrip->attri_value_len)
+ if (!nv->value.i_len)
return;
*nvecs += 1;
- *nbytes += xlog_calc_iovec_len(attrip->attri_value_len);
+ *nbytes += xlog_calc_iovec_len(nv->value.i_len);
}
/*
{
struct xfs_attri_log_item *attrip = ATTRI_ITEM(lip);
struct xfs_log_iovec *vecp = NULL;
+ struct xfs_attri_log_nameval *nv = attrip->attri_nameval;
attrip->attri_format.alfi_type = XFS_LI_ATTRI;
attrip->attri_format.alfi_size = 1;
* the log recovery.
*/
- ASSERT(attrip->attri_name_len > 0);
+ ASSERT(nv->name.i_len > 0);
attrip->attri_format.alfi_size++;
- if (attrip->attri_value_len > 0)
+ if (nv->value.i_len > 0)
attrip->attri_format.alfi_size++;
xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_ATTRI_FORMAT,
&attrip->attri_format,
sizeof(struct xfs_attri_log_format));
- xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_ATTR_NAME,
- attrip->attri_name,
- attrip->attri_name_len);
- if (attrip->attri_value_len > 0)
- xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_ATTR_VALUE,
- attrip->attri_value,
- attrip->attri_value_len);
+ xlog_copy_from_iovec(lv, &vecp, &nv->name);
+ if (nv->value.i_len > 0)
+ xlog_copy_from_iovec(lv, &vecp, &nv->value);
}
/*
STATIC struct xfs_attri_log_item *
xfs_attri_init(
struct xfs_mount *mp,
- uint32_t name_len,
- uint32_t value_len)
-
+ struct xfs_attri_log_nameval *nv)
{
struct xfs_attri_log_item *attrip;
- uint32_t buffer_size = name_len + value_len;
-
- if (buffer_size) {
- /*
- * This could be over 64kB in length, so we have to use
- * kvmalloc() for this. But kvmalloc() utterly sucks, so we
- * use own version.
- */
- attrip = xlog_kvmalloc(sizeof(struct xfs_attri_log_item) +
- buffer_size);
- } else {
- attrip = kmem_cache_alloc(xfs_attri_cache,
- GFP_NOFS | __GFP_NOFAIL);
- }
- memset(attrip, 0, sizeof(struct xfs_attri_log_item));
- attrip->attri_name_len = name_len;
- if (name_len)
- attrip->attri_name = ((char *)attrip) +
- sizeof(struct xfs_attri_log_item);
- else
- attrip->attri_name = NULL;
+ attrip = kmem_cache_zalloc(xfs_attri_cache, GFP_NOFS | __GFP_NOFAIL);
- attrip->attri_value_len = value_len;
- if (value_len)
- attrip->attri_value = ((char *)attrip) +
- sizeof(struct xfs_attri_log_item) +
- name_len;
- else
- attrip->attri_value = NULL;
+ /*
+ * Grab an extra reference to the name/value buffer for this log item.
+ * The caller retains its own reference!
+ */
+ attrip->attri_nameval = xfs_attri_log_nameval_get(nv);
+ ASSERT(attrip->attri_nameval);
xfs_log_item_init(mp, &attrip->attri_item, XFS_LI_ATTRI,
&xfs_attri_item_ops);
attrp->alfi_ino = attr->xattri_da_args->dp->i_ino;
ASSERT(!(attr->xattri_op_flags & ~XFS_ATTRI_OP_FLAGS_TYPE_MASK));
attrp->alfi_op_flags = attr->xattri_op_flags;
- attrp->alfi_value_len = attr->xattri_da_args->valuelen;
- attrp->alfi_name_len = attr->xattri_da_args->namelen;
+ attrp->alfi_value_len = attr->xattri_nameval->value.i_len;
+ attrp->alfi_name_len = attr->xattri_nameval->name.i_len;
ASSERT(!(attr->xattri_da_args->attr_filter & ~XFS_ATTRI_FILTER_MASK));
attrp->alfi_attr_filter = attr->xattri_da_args->attr_filter;
-
- memcpy(attrip->attri_name, attr->xattri_da_args->name,
- attr->xattri_da_args->namelen);
- memcpy(attrip->attri_value, attr->xattri_da_args->value,
- attr->xattri_da_args->valuelen);
- attrip->attri_name_len = attr->xattri_da_args->namelen;
- attrip->attri_value_len = attr->xattri_da_args->valuelen;
}
/* Get an ATTRI. */
* Each attr item only performs one attribute operation at a time, so
* this is a list of one
*/
- list_for_each_entry(attr, items, xattri_list) {
- attrip = xfs_attri_init(mp, attr->xattri_da_args->namelen,
- attr->xattri_da_args->valuelen);
- if (attrip == NULL)
- return NULL;
-
- xfs_trans_add_item(tp, &attrip->attri_item);
- xfs_attr_log_item(tp, attrip, attr);
+ attr = list_first_entry_or_null(items, struct xfs_attr_intent,
+ xattri_list);
+
+ /*
+ * Create a buffer to store the attribute name and value. This buffer
+ * will be shared between the higher level deferred xattr work state
+ * and the lower level xattr log items.
+ */
+ if (!attr->xattri_nameval) {
+ struct xfs_da_args *args = attr->xattri_da_args;
+
+ /*
+ * Transfer our reference to the name/value buffer to the
+ * deferred work state structure.
+ */
+ attr->xattri_nameval = xfs_attri_log_nameval_alloc(args->name,
+ args->namelen, args->value, args->valuelen);
}
+ if (!attr->xattri_nameval)
+ return ERR_PTR(-ENOMEM);
+
+ attrip = xfs_attri_init(mp, attr->xattri_nameval);
+ xfs_trans_add_item(tp, &attrip->attri_item);
+ xfs_attr_log_item(tp, attrip, attr);
return &attrip->attri_item;
}
{
if (attr->xattri_da_state)
xfs_da_state_free(attr->xattri_da_state);
+ xfs_attri_log_nameval_put(attr->xattri_nameval);
if (attr->xattri_da_args->op_flags & XFS_DA_OP_RECOVERY)
kmem_free(attr);
else
xfs_attr_free_item(attr);
}
-STATIC xfs_lsn_t
-xfs_attri_item_committed(
- struct xfs_log_item *lip,
- xfs_lsn_t lsn)
-{
- struct xfs_attri_log_item *attrip = ATTRI_ITEM(lip);
-
- /*
- * The attrip refers to xfs_attr_item memory to log the name and value
- * with the intent item. This already occurred when the intent was
- * committed so these fields are no longer accessed. Clear them out of
- * caution since we're about to free the xfs_attr_item.
- */
- attrip->attri_name = NULL;
- attrip->attri_value = NULL;
-
- /*
- * The ATTRI is logged only once and cannot be moved in the log, so
- * simply return the lsn at which it's been logged.
- */
- return lsn;
-}
-
STATIC bool
xfs_attri_item_match(
struct xfs_log_item *lip,
struct xfs_trans *tp;
struct xfs_trans_res tres;
struct xfs_attri_log_format *attrp;
+ struct xfs_attri_log_nameval *nv = attrip->attri_nameval;
int error, ret = 0;
int total;
int local;
*/
attrp = &attrip->attri_format;
if (!xfs_attri_validate(mp, attrp) ||
- !xfs_attr_namecheck(attrip->attri_name, attrip->attri_name_len))
+ !xfs_attr_namecheck(nv->name.i_addr, nv->name.i_len))
return -EFSCORRUPTED;
error = xlog_recover_iget(mp, attrp->alfi_ino, &ip);
attr->xattri_op_flags = attrp->alfi_op_flags &
XFS_ATTRI_OP_FLAGS_TYPE_MASK;
+ /*
+ * We're reconstructing the deferred work state structure from the
+ * recovered log item. Grab a reference to the name/value buffer and
+ * attach it to the new work state.
+ */
+ attr->xattri_nameval = xfs_attri_log_nameval_get(nv);
+ ASSERT(attr->xattri_nameval);
+
args->dp = ip;
args->geo = mp->m_attr_geo;
args->whichfork = XFS_ATTR_FORK;
- args->name = attrip->attri_name;
- args->namelen = attrp->alfi_name_len;
+ args->name = nv->name.i_addr;
+ args->namelen = nv->name.i_len;
args->hashval = xfs_da_hashname(args->name, args->namelen);
args->attr_filter = attrp->alfi_attr_filter & XFS_ATTRI_FILTER_MASK;
args->op_flags = XFS_DA_OP_RECOVERY | XFS_DA_OP_OKNOENT;
switch (attr->xattri_op_flags) {
case XFS_ATTRI_OP_FLAGS_SET:
case XFS_ATTRI_OP_FLAGS_REPLACE:
- args->value = attrip->attri_value;
- args->valuelen = attrp->alfi_value_len;
+ args->value = nv->value.i_addr;
+ args->valuelen = nv->value.i_len;
args->total = xfs_attr_calc_size(args, &local);
if (xfs_inode_hasattr(args->dp))
attr->xattri_dela_state = xfs_attr_init_replace_state(args);
attrdp = xfs_trans_get_attrd(tp, old_attrip);
set_bit(XFS_LI_DIRTY, &attrdp->attrd_item.li_flags);
- new_attrip = xfs_attri_init(tp->t_mountp, old_attrp->alfi_name_len,
- old_attrp->alfi_value_len);
+ /*
+ * Create a new log item that shares the same name/value buffer as the
+ * old log item.
+ */
+ new_attrip = xfs_attri_init(tp->t_mountp, old_attrip->attri_nameval);
new_attrp = &new_attrip->attri_format;
new_attrp->alfi_ino = old_attrp->alfi_ino;
new_attrp->alfi_name_len = old_attrp->alfi_name_len;
new_attrp->alfi_attr_filter = old_attrp->alfi_attr_filter;
- memcpy(new_attrip->attri_name, old_attrip->attri_name,
- new_attrip->attri_name_len);
-
- if (new_attrip->attri_value_len > 0)
- memcpy(new_attrip->attri_value, old_attrip->attri_value,
- new_attrip->attri_value_len);
-
xfs_trans_add_item(tp, &new_attrip->attri_item);
set_bit(XFS_LI_DIRTY, &new_attrip->attri_item.li_flags);
struct xlog_recover_item *item,
xfs_lsn_t lsn)
{
- int error;
struct xfs_mount *mp = log->l_mp;
struct xfs_attri_log_item *attrip;
struct xfs_attri_log_format *attri_formatp;
+ struct xfs_attri_log_nameval *nv;
+ const void *attr_value = NULL;
const void *attr_name;
- int region = 0;
+ int error;
- attri_formatp = item->ri_buf[region].i_addr;
+ attri_formatp = item->ri_buf[0].i_addr;
attr_name = item->ri_buf[1].i_addr;
/* Validate xfs_attri_log_format before the large memory allocation */
return -EFSCORRUPTED;
}
- /* memory alloc failure will cause replay to abort */
- attrip = xfs_attri_init(mp, attri_formatp->alfi_name_len,
- attri_formatp->alfi_value_len);
- if (attrip == NULL)
+ if (attri_formatp->alfi_value_len)
+ attr_value = item->ri_buf[2].i_addr;
+
+ /*
+ * Memory alloc failure will cause replay to abort. We attach the
+ * name/value buffer to the recovered incore log item and drop our
+ * reference.
+ */
+ nv = xfs_attri_log_nameval_alloc(attr_name,
+ attri_formatp->alfi_name_len, attr_value,
+ attri_formatp->alfi_value_len);
+ if (!nv)
return -ENOMEM;
- error = xfs_attri_copy_format(&item->ri_buf[region],
- &attrip->attri_format);
+ attrip = xfs_attri_init(mp, nv);
+ error = xfs_attri_copy_format(&item->ri_buf[0], &attrip->attri_format);
if (error)
goto out;
- region++;
- memcpy(attrip->attri_name, item->ri_buf[region].i_addr,
- attrip->attri_name_len);
-
- if (attrip->attri_value_len > 0) {
- region++;
- memcpy(attrip->attri_value, item->ri_buf[region].i_addr,
- attrip->attri_value_len);
- }
-
/*
* The ATTRI has two references. One for the ATTRD and one for ATTRI to
* ensure it makes it into the AIL. Insert the ATTRI into the AIL
*/
xfs_trans_ail_insert(log->l_ailp, &attrip->attri_item, lsn);
xfs_attri_release(attrip);
+ xfs_attri_log_nameval_put(nv);
return 0;
out:
xfs_attri_item_free(attrip);
+ xfs_attri_log_nameval_put(nv);
return error;
}
.iop_size = xfs_attri_item_size,
.iop_format = xfs_attri_item_format,
.iop_unpin = xfs_attri_item_unpin,
- .iop_committed = xfs_attri_item_committed,
.iop_release = xfs_attri_item_release,
.iop_recover = xfs_attri_item_recover,
.iop_match = xfs_attri_item_match,
projects / users / hch / xfs.git / commitdiff