leandrof/linux
1
0
Fork 0
forked from Minki/linux
Code Issues Pull Requests Packages Projects Releases Wiki Activity

xfs: share xattr name and value buffers when logging xattr updates

Browse Source 
While running xfs/297 and generic/642, I noticed a crash in
xfs_attri_item_relog when it tries to copy the attr name to the new
xattri log item.  I think what happened here was that we called
->iop_commit on the old attri item (which nulls out the pointers) as
part of a log force at the same time that a chained attr operation was
ongoing.  The system was busy enough that at some later point, the defer
ops operation decided it was necessary to relog the attri log item, but
as we've detached the name buffer from the old attri log item, we can't
copy it to the new one, and kaboom.

I think there's a broader refcounting problem with LARP mode -- the
setxattr code can return to userspace before the CIL actually formats
and commits the log item, which results in a UAF bug.  Therefore, the
xattr log item needs to be able to retain a reference to the name and
value buffers until the log items have completely cleared the log.
Furthermore, each time we create an intent log item, we allocate new
memory and (re)copy the contents; sharing here would be very useful.

Solve the UAF and the unnecessary memory allocations by having the log
code create a single refcounted buffer to contain the name and value
contents.  This buffer can be passed from old to new during a relog
operation, and the logging code can (optionally) attach it to the
xfs_attr_item for reuse when LARP mode is enabled.

This also fixes a problem where the xfs_attri_log_item objects weren't
being freed back to the same cache where they came from.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
This commit is contained in:
Darrick J. Wong 2022-05-23 08:43:46 +10:00 committed by Dave Chinner
parent 22a68ba724
commit 4183e4f27f
5 changed files with 222 additions and 133 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
fs/xfs/libxfs/xfs_attr.h
View File

@ -502,6 +502,8 @@ enum xfs_delattr_state {
{ XFS_DAS_NODE_REMOVE_ATTR, "XFS_DAS_NODE_REMOVE_ATTR" }, \ { XFS_DAS_NODE_REMOVE_ATTR, "XFS_DAS_NODE_REMOVE_ATTR" }, \
{ XFS_DAS_DONE, "XFS_DAS_DONE" } { XFS_DAS_DONE, "XFS_DAS_DONE" }
struct xfs_attri_log_nameval;
/* /*
* Context used for keeping track of delayed attribute operations * Context used for keeping track of delayed attribute operations
*/ */
@ -517,6 +519,12 @@ struct xfs_attr_intent {
struct xfs_da_args *xattri_da_args; struct xfs_da_args *xattri_da_args;
/*
* Shared buffer containing the attr name and value so that the logging
* code can share large memory buffers between log items.
*/
struct xfs_attri_log_nameval *xattri_nameval;
/* /*
* Used by xfs_attr_set to hold a leaf buffer across a transaction roll * Used by xfs_attr_set to hold a leaf buffer across a transaction roll
*/ */

59
fs/xfs/libxfs/xfs_defer.c
View File

@ -191,35 +191,56 @@ static const struct xfs_defer_op_type *defer_op_types[] = {
[XFS_DEFER_OPS_TYPE_ATTR] = &xfs_attr_defer_type, [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( xfs_defer_create_intent(
struct xfs_trans *tp, struct xfs_trans *tp,
struct xfs_defer_pending *dfp, struct xfs_defer_pending *dfp,
bool sort) bool sort)
{ {
const struct xfs_defer_op_type *ops = defer_op_types[dfp->dfp_type]; const struct xfs_defer_op_type *ops = defer_op_types[dfp->dfp_type];
struct xfs_log_item *lip;
if (!dfp->dfp_intent) if (dfp->dfp_intent)
dfp->dfp_intent = ops->create_intent(tp, &dfp->dfp_work, return 1;
dfp->dfp_count, sort);
return dfp->dfp_intent != NULL; 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 * 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 * associated extents, then add the entire intake list to the end of
* the pending list. * 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( xfs_defer_create_intents(
struct xfs_trans *tp) struct xfs_trans *tp)
{ {
struct xfs_defer_pending *dfp; struct xfs_defer_pending *dfp;
bool ret = false; int ret = 0;
list_for_each_entry(dfp, &tp->t_dfops, dfp_list) { list_for_each_entry(dfp, &tp->t_dfops, dfp_list) {
int ret2;
trace_xfs_defer_create_intent(tp->t_mountp, dfp); 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; return ret;
} }
@ -457,6 +478,8 @@ xfs_defer_finish_one(
dfp->dfp_count--; dfp->dfp_count--;
error = ops->finish_item(tp, dfp->dfp_done, li, &state); error = ops->finish_item(tp, dfp->dfp_done, li, &state);
if (error == -EAGAIN) { if (error == -EAGAIN) {
int ret;
/* /*
* Caller wants a fresh transaction; put the work item * Caller wants a fresh transaction; put the work item
* back on the list and log a new log intent item to * back on the list and log a new log intent item to
@ -467,7 +490,9 @@ xfs_defer_finish_one(
dfp->dfp_count++; dfp->dfp_count++;
dfp->dfp_done = NULL; dfp->dfp_done = NULL;
dfp->dfp_intent = 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) if (error)
@ -514,10 +539,14 @@ xfs_defer_finish_noroll(
* of time that any one intent item can stick around in memory, * of time that any one intent item can stick around in memory,
* pinning the log tail. * 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); list_splice_init(&(*tp)->t_dfops, &dop_pending);
if (has_intents < 0) {
error = has_intents;
goto out_shutdown;
}
if (has_intents || dfp) { if (has_intents || dfp) {
error = xfs_defer_trans_roll(tp); error = xfs_defer_trans_roll(tp);
if (error) if (error)
@ -676,13 +705,15 @@ xfs_defer_ops_capture(
if (list_empty(&tp->t_dfops)) if (list_empty(&tp->t_dfops))
return NULL; return NULL;
error = xfs_defer_create_intents(tp);
if (error < 0)
return ERR_PTR(error);
/* Create an object to capture the defer ops. */ /* Create an object to capture the defer ops. */
dfc = kmem_zalloc(sizeof(*dfc), KM_NOFS); dfc = kmem_zalloc(sizeof(*dfc), KM_NOFS);
INIT_LIST_HEAD(&dfc->dfc_list); INIT_LIST_HEAD(&dfc->dfc_list);
INIT_LIST_HEAD(&dfc->dfc_dfops); INIT_LIST_HEAD(&dfc->dfc_dfops);
xfs_defer_create_intents(tp);
/* Move the dfops chain and transaction state to the capture struct. */ /* Move the dfops chain and transaction state to the capture struct. */
list_splice_init(&tp->t_dfops, &dfc->dfc_dfops); list_splice_init(&tp->t_dfops, &dfc->dfc_dfops);
dfc->dfc_tpflags = tp->t_flags & XFS_TRANS_LOWMODE; dfc->dfc_tpflags = tp->t_flags & XFS_TRANS_LOWMODE;
@ -759,6 +790,10 @@ xfs_defer_ops_capture_and_commit(
/* If we don't capture anything, commit transaction and exit. */ /* If we don't capture anything, commit transaction and exit. */
dfc = xfs_defer_ops_capture(tp); dfc = xfs_defer_ops_capture(tp);
if (IS_ERR(dfc)) {
xfs_trans_cancel(tp);
return PTR_ERR(dfc);
}
if (!dfc) if (!dfc)
return xfs_trans_commit(tp); return xfs_trans_commit(tp);

268
fs/xfs/xfs_attr_item.c
View File

@ -42,12 +42,80 @@ static inline struct xfs_attri_log_item *ATTRI_ITEM(struct xfs_log_item *lip)
return container_of(lip, struct xfs_attri_log_item, attri_item); 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 STATIC void
xfs_attri_item_free( xfs_attri_item_free(
struct xfs_attri_log_item *attrip) struct xfs_attri_log_item *attrip)
{ {
kmem_free(attrip->attri_item.li_lv_shadow); 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);
} }
/* /*
@ -76,16 +144,17 @@ xfs_attri_item_size(
int *nbytes) int *nbytes)
{ {
struct xfs_attri_log_item *attrip = ATTRI_ITEM(lip); struct xfs_attri_log_item *attrip = ATTRI_ITEM(lip);
struct xfs_attri_log_nameval *nv = attrip->attri_nameval;
*nvecs += 2; *nvecs += 2;
*nbytes += sizeof(struct xfs_attri_log_format) + *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; return;
*nvecs += 1; *nvecs += 1;
*nbytes += xlog_calc_iovec_len(attrip->attri_value_len); *nbytes += xlog_calc_iovec_len(nv->value.i_len);
} }
/* /*
@ -100,6 +169,7 @@ xfs_attri_item_format(
{ {
struct xfs_attri_log_item *attrip = ATTRI_ITEM(lip); struct xfs_attri_log_item *attrip = ATTRI_ITEM(lip);
struct xfs_log_iovec *vecp = NULL; 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_type = XFS_LI_ATTRI;
attrip->attri_format.alfi_size = 1; attrip->attri_format.alfi_size = 1;
@ -111,22 +181,18 @@ xfs_attri_item_format(
* the log recovery. * the log recovery.
*/ */
ASSERT(attrip->attri_name_len > 0); ASSERT(nv->name.i_len > 0);
attrip->attri_format.alfi_size++; attrip->attri_format.alfi_size++;
if (attrip->attri_value_len > 0) if (nv->value.i_len > 0)
attrip->attri_format.alfi_size++; attrip->attri_format.alfi_size++;
xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_ATTRI_FORMAT, xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_ATTRI_FORMAT,
&attrip->attri_format, &attrip->attri_format,
sizeof(struct xfs_attri_log_format)); sizeof(struct xfs_attri_log_format));
xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_ATTR_NAME, xlog_copy_from_iovec(lv, &vecp, &nv->name);
attrip->attri_name, if (nv->value.i_len > 0)
attrip->attri_name_len); xlog_copy_from_iovec(lv, &vecp, &nv->value);
if (attrip->attri_value_len > 0)
xlog_copy_iovec(lv, &vecp, XLOG_REG_TYPE_ATTR_VALUE,
attrip->attri_value,
attrip->attri_value_len);
} }
/* /*
@ -161,41 +227,18 @@ xfs_attri_item_release(
STATIC struct xfs_attri_log_item * STATIC struct xfs_attri_log_item *
xfs_attri_init( xfs_attri_init(
struct xfs_mount *mp, struct xfs_mount *mp,
uint32_t name_len, struct xfs_attri_log_nameval *nv)
uint32_t value_len)
{ {
struct xfs_attri_log_item *attrip; struct xfs_attri_log_item *attrip;
uint32_t buffer_size = name_len + value_len;
if (buffer_size) { attrip = kmem_cache_zalloc(xfs_attri_cache, GFP_NOFS | __GFP_NOFAIL);
/*
* 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) * Grab an extra reference to the name/value buffer for this log item.
attrip->attri_name = ((char *)attrip) + * The caller retains its own reference!
sizeof(struct xfs_attri_log_item); */
else attrip->attri_nameval = xfs_attri_log_nameval_get(nv);
attrip->attri_name = NULL; ASSERT(attrip->attri_nameval);
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;
xfs_log_item_init(mp, &attrip->attri_item, XFS_LI_ATTRI, xfs_log_item_init(mp, &attrip->attri_item, XFS_LI_ATTRI,
&xfs_attri_item_ops); &xfs_attri_item_ops);
@ -354,17 +397,10 @@ xfs_attr_log_item(
attrp->alfi_ino = attr->xattri_da_args->dp->i_ino; attrp->alfi_ino = attr->xattri_da_args->dp->i_ino;
ASSERT(!(attr->xattri_op_flags & ~XFS_ATTRI_OP_FLAGS_TYPE_MASK)); ASSERT(!(attr->xattri_op_flags & ~XFS_ATTRI_OP_FLAGS_TYPE_MASK));
attrp->alfi_op_flags = attr->xattri_op_flags; attrp->alfi_op_flags = attr->xattri_op_flags;
attrp->alfi_value_len = attr->xattri_da_args->valuelen; attrp->alfi_value_len = attr->xattri_nameval->value.i_len;
attrp->alfi_name_len = attr->xattri_da_args->namelen; attrp->alfi_name_len = attr->xattri_nameval->name.i_len;
ASSERT(!(attr->xattri_da_args->attr_filter & ~XFS_ATTRI_FILTER_MASK)); ASSERT(!(attr->xattri_da_args->attr_filter & ~XFS_ATTRI_FILTER_MASK));
attrp->alfi_attr_filter = attr->xattri_da_args->attr_filter; 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. */ /* Get an ATTRI. */
@ -388,15 +424,30 @@ xfs_attr_create_intent(
* Each attr item only performs one attribute operation at a time, so * Each attr item only performs one attribute operation at a time, so
* this is a list of one * this is a list of one
*/ */
list_for_each_entry(attr, items, xattri_list) { attr = list_first_entry_or_null(items, struct xfs_attr_intent,
attrip = xfs_attri_init(mp, attr->xattri_da_args->namelen, xattri_list);
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); * 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; return &attrip->attri_item;
} }
@ -407,6 +458,7 @@ xfs_attr_free_item(
{ {
if (attr->xattri_da_state) if (attr->xattri_da_state)
xfs_da_state_free(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) if (attr->xattri_da_args->op_flags & XFS_DA_OP_RECOVERY)
kmem_free(attr); kmem_free(attr);
else else
@ -461,29 +513,6 @@ xfs_attr_cancel_item(
xfs_attr_free_item(attr); 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 STATIC bool
xfs_attri_item_match( xfs_attri_item_match(
struct xfs_log_item *lip, struct xfs_log_item *lip,
@ -547,6 +576,7 @@ xfs_attri_item_recover(
struct xfs_trans *tp; struct xfs_trans *tp;
struct xfs_trans_res tres; struct xfs_trans_res tres;
struct xfs_attri_log_format *attrp; struct xfs_attri_log_format *attrp;
struct xfs_attri_log_nameval *nv = attrip->attri_nameval;
int error, ret = 0; int error, ret = 0;
int total; int total;
int local; int local;
@ -558,7 +588,7 @@ xfs_attri_item_recover(
*/ */
attrp = &attrip->attri_format; attrp = &attrip->attri_format;
if (!xfs_attri_validate(mp, attrp) || 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; return -EFSCORRUPTED;
error = xlog_recover_iget(mp, attrp->alfi_ino, &ip); error = xlog_recover_iget(mp, attrp->alfi_ino, &ip);
@ -573,11 +603,19 @@ xfs_attri_item_recover(
attr->xattri_op_flags = attrp->alfi_op_flags & attr->xattri_op_flags = attrp->alfi_op_flags &
XFS_ATTRI_OP_FLAGS_TYPE_MASK; 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->dp = ip;
args->geo = mp->m_attr_geo; args->geo = mp->m_attr_geo;
args->whichfork = XFS_ATTR_FORK; args->whichfork = XFS_ATTR_FORK;
args->name = attrip->attri_name; args->name = nv->name.i_addr;
args->namelen = attrp->alfi_name_len; args->namelen = nv->name.i_len;
args->hashval = xfs_da_hashname(args->name, args->namelen); args->hashval = xfs_da_hashname(args->name, args->namelen);
args->attr_filter = attrp->alfi_attr_filter & XFS_ATTRI_FILTER_MASK; args->attr_filter = attrp->alfi_attr_filter & XFS_ATTRI_FILTER_MASK;
args->op_flags = XFS_DA_OP_RECOVERY | XFS_DA_OP_OKNOENT; args->op_flags = XFS_DA_OP_RECOVERY | XFS_DA_OP_OKNOENT;
@ -585,8 +623,8 @@ xfs_attri_item_recover(
switch (attr->xattri_op_flags) { switch (attr->xattri_op_flags) {
case XFS_ATTRI_OP_FLAGS_SET: case XFS_ATTRI_OP_FLAGS_SET:
case XFS_ATTRI_OP_FLAGS_REPLACE: case XFS_ATTRI_OP_FLAGS_REPLACE:
args->value = attrip->attri_value; args->value = nv->value.i_addr;
args->valuelen = attrp->alfi_value_len; args->valuelen = nv->value.i_len;
args->total = xfs_attr_calc_size(args, &local); args->total = xfs_attr_calc_size(args, &local);
if (xfs_inode_hasattr(args->dp)) if (xfs_inode_hasattr(args->dp))
attr->xattri_dela_state = xfs_attr_init_replace_state(args); attr->xattri_dela_state = xfs_attr_init_replace_state(args);
@ -660,8 +698,11 @@ xfs_attri_item_relog(
attrdp = xfs_trans_get_attrd(tp, old_attrip); attrdp = xfs_trans_get_attrd(tp, old_attrip);
set_bit(XFS_LI_DIRTY, &attrdp->attrd_item.li_flags); 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 = &new_attrip->attri_format;
new_attrp->alfi_ino = old_attrp->alfi_ino; new_attrp->alfi_ino = old_attrp->alfi_ino;
@ -670,13 +711,6 @@ xfs_attri_item_relog(
new_attrp->alfi_name_len = old_attrp->alfi_name_len; new_attrp->alfi_name_len = old_attrp->alfi_name_len;
new_attrp->alfi_attr_filter = old_attrp->alfi_attr_filter; 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); xfs_trans_add_item(tp, &new_attrip->attri_item);
set_bit(XFS_LI_DIRTY, &new_attrip->attri_item.li_flags); set_bit(XFS_LI_DIRTY, &new_attrip->attri_item.li_flags);
@ -690,14 +724,15 @@ xlog_recover_attri_commit_pass2(
struct xlog_recover_item *item, struct xlog_recover_item *item,
xfs_lsn_t lsn) xfs_lsn_t lsn)
{ {
int error;
struct xfs_mount *mp = log->l_mp; struct xfs_mount *mp = log->l_mp;
struct xfs_attri_log_item *attrip; struct xfs_attri_log_item *attrip;
struct xfs_attri_log_format *attri_formatp; struct xfs_attri_log_format *attri_formatp;
struct xfs_attri_log_nameval *nv;
const void *attr_value = NULL;
const void *attr_name; 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; attr_name = item->ri_buf[1].i_addr;
/* Validate xfs_attri_log_format before the large memory allocation */ /* Validate xfs_attri_log_format before the large memory allocation */
@ -711,27 +746,25 @@ xlog_recover_attri_commit_pass2(
return -EFSCORRUPTED; return -EFSCORRUPTED;
} }
/* memory alloc failure will cause replay to abort */ if (attri_formatp->alfi_value_len)
attrip = xfs_attri_init(mp, attri_formatp->alfi_name_len, attr_value = item->ri_buf[2].i_addr;
attri_formatp->alfi_value_len);
if (attrip == NULL) /*
* 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; return -ENOMEM;
error = xfs_attri_copy_format(&item->ri_buf[region], attrip = xfs_attri_init(mp, nv);
&attrip->attri_format); error = xfs_attri_copy_format(&item->ri_buf[0], &attrip->attri_format);
if (error) if (error)
goto out; 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 * 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 * ensure it makes it into the AIL. Insert the ATTRI into the AIL
@ -740,9 +773,11 @@ xlog_recover_attri_commit_pass2(
*/ */
xfs_trans_ail_insert(log->l_ailp, &attrip->attri_item, lsn); xfs_trans_ail_insert(log->l_ailp, &attrip->attri_item, lsn);
xfs_attri_release(attrip); xfs_attri_release(attrip);
xfs_attri_log_nameval_put(nv);
return 0; return 0;
out: out:
xfs_attri_item_free(attrip); xfs_attri_item_free(attrip);
xfs_attri_log_nameval_put(nv);
return error; return error;
} }
@ -822,7 +857,6 @@ static const struct xfs_item_ops xfs_attri_item_ops = {
.iop_size = xfs_attri_item_size, .iop_size = xfs_attri_item_size,
.iop_format = xfs_attri_item_format, .iop_format = xfs_attri_item_format,
.iop_unpin = xfs_attri_item_unpin, .iop_unpin = xfs_attri_item_unpin,
.iop_committed = xfs_attri_item_committed,
.iop_release = xfs_attri_item_release, .iop_release = xfs_attri_item_release,
.iop_recover = xfs_attri_item_recover, .iop_recover = xfs_attri_item_recover,
.iop_match = xfs_attri_item_match, .iop_match = xfs_attri_item_match,

13
fs/xfs/xfs_attr_item.h
View File

@ -11,6 +11,14 @@
struct xfs_mount; struct xfs_mount;
struct kmem_zone; struct kmem_zone;
struct xfs_attri_log_nameval {
struct xfs_log_iovec name;
struct xfs_log_iovec value;
refcount_t refcount;
/* name and value follow the end of this struct */
};
/* /*
* This is the "attr intention" log item. It is used to log the fact that some * This is the "attr intention" log item. It is used to log the fact that some
* extended attribute operations need to be processed. An operation is * extended attribute operations need to be processed. An operation is
@ -26,10 +34,7 @@ struct kmem_zone;
struct xfs_attri_log_item { struct xfs_attri_log_item {
struct xfs_log_item attri_item; struct xfs_log_item attri_item;
atomic_t attri_refcount; atomic_t attri_refcount;
int attri_name_len; struct xfs_attri_log_nameval *attri_nameval;
int attri_value_len;
void *attri_name;
void *attri_value;
struct xfs_attri_log_format attri_format; struct xfs_attri_log_format attri_format;
}; };

7
fs/xfs/xfs_log.h
View File

@ -86,6 +86,13 @@ xlog_copy_iovec(struct xfs_log_vec *lv, struct xfs_log_iovec **vecp,
return buf; return buf;
} }
static inline void *
xlog_copy_from_iovec(struct xfs_log_vec *lv, struct xfs_log_iovec **vecp,
const struct xfs_log_iovec *src)
{
return xlog_copy_iovec(lv, vecp, src->i_type, src->i_addr, src->i_len);
}
/* /*
* By comparing each component, we don't have to worry about extra * By comparing each component, we don't have to worry about extra
* endian issues in treating two 32 bit numbers as one 64 bit number * endian issues in treating two 32 bit numbers as one 64 bit number