/*
* fills in the vector of log iovecs for the given dquot log item.
*/
STATIC void
xfs_qm_dquot_logitem_format(
xfs_dq_logitem_t *logitem,
xfs_log_iovec_t *logvec)
{
ASSERT(logitem);
ASSERT(logitem->qli_dquot);
logvec->i_addr = (xfs_caddr_t)&logitem->qli_format;
logvec->i_len = sizeof(xfs_dq_logformat_t);
XLOG_VEC_SET_TYPE(logvec, XLOG_REG_TYPE_QFORMAT);
logvec++;
logvec->i_addr = (xfs_caddr_t)&logitem->qli_dquot->q_core;
logvec->i_len = sizeof(xfs_disk_dquot_t);
XLOG_VEC_SET_TYPE(logvec, XLOG_REG_TYPE_DQUOT);
ASSERT(2 == logitem->qli_item.li_desc->lid_size);
logitem->qli_format.qlf_size = 2;
}
/*
* This is called to fill in the vector of log iovecs for the
* given quotaoff log item. We use only 1 iovec, and we point that
* at the quotaoff_log_format structure embedded in the quotaoff item.
* It is at this point that we assert that all of the extent
* slots in the quotaoff item have been filled.
*/
STATIC void
xfs_qm_qoff_logitem_format(xfs_qoff_logitem_t *qf,
xfs_log_iovec_t *log_vector)
{
ASSERT(qf->qql_format.qf_type == XFS_LI_QUOTAOFF);
log_vector->i_addr = (xfs_caddr_t)&(qf->qql_format);
log_vector->i_len = sizeof(xfs_qoff_logitem_t);
XLOG_VEC_SET_TYPE(log_vector, XLOG_REG_TYPE_QUOTAOFF);
qf->qql_format.qf_size = 1;
}
/*
* Fill in the vector with pointers to data to be logged
* by this transaction. The transaction header takes
* the first vector, and then each dirty item takes the
* number of vectors it indicated it needed in xfs_trans_count_vecs().
*
* As each item fills in the entries it needs, also pin the item
* so that it cannot be flushed out until the log write completes.
*/
STATIC void
xfs_trans_fill_vecs(
xfs_trans_t *tp,
xfs_log_iovec_t *log_vector)
{
xfs_log_item_desc_t *lidp;
xfs_log_iovec_t *vecp;
uint nitems;
/*
* Skip over the entry for the transaction header, we'll
* fill that in at the end.
*/
vecp = log_vector + 1; /* pointer arithmetic */
nitems = 0;
lidp = xfs_trans_first_item(tp);
ASSERT(lidp != NULL);
while (lidp != NULL) {
/*
* Skip items which aren't dirty in this transaction.
*/
if (!(lidp->lid_flags & XFS_LID_DIRTY)) {
lidp = xfs_trans_next_item(tp, lidp);
continue;
}
/*
* The item may be marked dirty but not log anything.
* This can be used to get called when a transaction
* is committed.
*/
if (lidp->lid_size) {
nitems++;
}
IOP_FORMAT(lidp->lid_item, vecp);
vecp += lidp->lid_size; /* pointer arithmetic */
IOP_PIN(lidp->lid_item);
lidp = xfs_trans_next_item(tp, lidp);
}
/*
* Now that we've counted the number of items in this
* transaction, fill in the transaction header.
*/
tp->t_header.th_magic = XFS_TRANS_HEADER_MAGIC;
tp->t_header.th_type = tp->t_type;
tp->t_header.th_num_items = nitems;
log_vector->i_addr = (xfs_caddr_t)&tp->t_header;
log_vector->i_len = sizeof(xfs_trans_header_t);
XLOG_VEC_SET_TYPE(log_vector, XLOG_REG_TYPE_TRANSHDR);
}
/*
* This is called to fill in the vector of log iovecs for the
* given efi log item. We use only 1 iovec, and we point that
* at the efi_log_format structure embedded in the efi item.
* It is at this point that we assert that all of the extent
* slots in the efi item have been filled.
*/
STATIC void
xfs_efi_item_format(xfs_efi_log_item_t *efip,
xfs_log_iovec_t *log_vector)
{
uint size;
ASSERT(efip->efi_next_extent == efip->efi_format.efi_nextents);
efip->efi_format.efi_type = XFS_LI_EFI;
size = sizeof(xfs_efi_log_format_t);
size += (efip->efi_format.efi_nextents - 1) * sizeof(xfs_extent_t);
efip->efi_format.efi_size = 1;
log_vector->i_addr = (xfs_caddr_t)&(efip->efi_format);
log_vector->i_len = size;
XLOG_VEC_SET_TYPE(log_vector, XLOG_REG_TYPE_EFI_FORMAT);
ASSERT(size >= sizeof(xfs_efi_log_format_t));
}
/*
* This is called to fill in the vector of log iovecs for the
* given log buf item. It fills the first entry with a buf log
* format structure, and the rest point to contiguous chunks
* within the buffer.
*/
STATIC void
xfs_buf_item_format(
xfs_buf_log_item_t *bip,
xfs_log_iovec_t *log_vector)
{
uint base_size;
uint nvecs;
xfs_log_iovec_t *vecp;
xfs_buf_t *bp;
int first_bit;
int last_bit;
int next_bit;
uint nbits;
uint buffer_offset;
ASSERT(atomic_read(&bip->bli_refcount) > 0);
ASSERT((bip->bli_flags & XFS_BLI_LOGGED) ||
(bip->bli_flags & XFS_BLI_STALE));
bp = bip->bli_buf;
vecp = log_vector;
/*
* The size of the base structure is the size of the
* declared structure plus the space for the extra words
* of the bitmap. We subtract one from the map size, because
* the first element of the bitmap is accounted for in the
* size of the base structure.
*/
base_size =
(uint)(sizeof(xfs_buf_log_format_t) +
((bip->bli_format.blf_map_size - 1) * sizeof(uint)));
vecp->i_addr = (xfs_caddr_t)&bip->bli_format;
vecp->i_len = base_size;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_BFORMAT);
vecp++;
nvecs = 1;
if (bip->bli_flags & XFS_BLI_STALE) {
/*
* The buffer is stale, so all we need to log
* is the buf log format structure with the
* cancel flag in it.
*/
xfs_buf_item_trace("FORMAT STALE", bip);
ASSERT(bip->bli_format.blf_flags & XFS_BLI_CANCEL);
bip->bli_format.blf_size = nvecs;
return;
}
/*
* Fill in an iovec for each set of contiguous chunks.
*/
first_bit = xfs_next_bit(bip->bli_format.blf_data_map,
bip->bli_format.blf_map_size, 0);
ASSERT(first_bit != -1);
last_bit = first_bit;
nbits = 1;
for (;;) {
/*
* This takes the bit number to start looking from and
* returns the next set bit from there. It returns -1
* if there are no more bits set or the start bit is
* beyond the end of the bitmap.
*/
next_bit = xfs_next_bit(bip->bli_format.blf_data_map,
bip->bli_format.blf_map_size,
(uint)last_bit + 1);
/*
* If we run out of bits fill in the last iovec and get
* out of the loop.
* Else if we start a new set of bits then fill in the
* iovec for the series we were looking at and start
* counting the bits in the new one.
* Else we're still in the same set of bits so just
* keep counting and scanning.
*/
if (next_bit == -1) {
buffer_offset = first_bit * XFS_BLI_CHUNK;
vecp->i_addr = xfs_buf_offset(bp, buffer_offset);
vecp->i_len = nbits * XFS_BLI_CHUNK;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_BCHUNK);
nvecs++;
break;
} else if (next_bit != last_bit + 1) {
buffer_offset = first_bit * XFS_BLI_CHUNK;
vecp->i_addr = xfs_buf_offset(bp, buffer_offset);
vecp->i_len = nbits * XFS_BLI_CHUNK;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_BCHUNK);
nvecs++;
vecp++;
first_bit = next_bit;
last_bit = next_bit;
nbits = 1;
} else if (xfs_buf_offset(bp, next_bit << XFS_BLI_SHIFT) !=
(xfs_buf_offset(bp, last_bit << XFS_BLI_SHIFT) +
XFS_BLI_CHUNK)) {
buffer_offset = first_bit * XFS_BLI_CHUNK;
vecp->i_addr = xfs_buf_offset(bp, buffer_offset);
vecp->i_len = nbits * XFS_BLI_CHUNK;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_BCHUNK);
/* You would think we need to bump the nvecs here too, but we do not
* this number is used by recovery, and it gets confused by the boundary
* split here
* nvecs++;
*/
vecp++;
first_bit = next_bit;
last_bit = next_bit;
nbits = 1;
} else {
last_bit++;
nbits++;
}
}
bip->bli_format.blf_size = nvecs;
/*
* Check to make sure everything is consistent.
*/
xfs_buf_item_trace("FORMAT NORM", bip);
xfs_buf_item_log_check(bip);
}
/*
* This is called to fill in the vector of log iovecs for the
* given inode log item. It fills the first item with an inode
* log format structure, the second with the on-disk inode structure,
* and a possible third and/or fourth with the inode data/extents/b-tree
* root and inode attributes data/extents/b-tree root.
*/
STATIC void
xfs_inode_item_format(
xfs_inode_log_item_t *iip,
xfs_log_iovec_t *log_vector)
{
uint nvecs;
xfs_log_iovec_t *vecp;
xfs_inode_t *ip;
size_t data_bytes;
xfs_bmbt_rec_t *ext_buffer;
int nrecs;
xfs_mount_t *mp;
ip = iip->ili_inode;
vecp = log_vector;
vecp->i_addr = (xfs_caddr_t)&iip->ili_format;
vecp->i_len = sizeof(xfs_inode_log_format_t);
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_IFORMAT);
vecp++;
nvecs = 1;
/*
* Clear i_update_core if the timestamps (or any other
* non-transactional modification) need flushing/logging
* and we're about to log them with the rest of the core.
*
* This is the same logic as xfs_iflush() but this code can't
* run at the same time as xfs_iflush because we're in commit
* processing here and so we have the inode lock held in
* exclusive mode. Although it doesn't really matter
* for the timestamps if both routines were to grab the
* timestamps or not. That would be ok.
*
* We clear i_update_core before copying out the data.
* This is for coordination with our timestamp updates
* that don't hold the inode lock. They will always
* update the timestamps BEFORE setting i_update_core,
* so if we clear i_update_core after they set it we
* are guaranteed to see their updates to the timestamps
* either here. Likewise, if they set it after we clear it
* here, we'll see it either on the next commit of this
* inode or the next time the inode gets flushed via
* xfs_iflush(). This depends on strongly ordered memory
* semantics, but we have that. We use the SYNCHRONIZE
* macro to make sure that the compiler does not reorder
* the i_update_core access below the data copy below.
*/
if (ip->i_update_core) {
ip->i_update_core = 0;
SYNCHRONIZE();
}
/*
* We don't have to worry about re-ordering here because
* the update_size field is protected by the inode lock
* and we have that held in exclusive mode.
*/
if (ip->i_update_size)
ip->i_update_size = 0;
/*
* Make sure to get the latest atime from the Linux inode.
*/
xfs_synchronize_atime(ip);
/*
* make sure the linux inode is dirty
*/
xfs_mark_inode_dirty_sync(ip);
vecp->i_addr = (xfs_caddr_t)&ip->i_d;
vecp->i_len = sizeof(struct xfs_icdinode);
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_ICORE);
vecp++;
nvecs++;
iip->ili_format.ilf_fields |= XFS_ILOG_CORE;
/*
* If this is really an old format inode, then we need to
* log it as such. This means that we have to copy the link
* count from the new field to the old. We don't have to worry
* about the new fields, because nothing trusts them as long as
* the old inode version number is there. If the superblock already
* has a new version number, then we don't bother converting back.
*/
mp = ip->i_mount;
ASSERT(ip->i_d.di_version == 1 || xfs_sb_version_hasnlink(&mp->m_sb));
if (ip->i_d.di_version == 1) {
if (!xfs_sb_version_hasnlink(&mp->m_sb)) {
/*
* Convert it back.
*/
ASSERT(ip->i_d.di_nlink <= XFS_MAXLINK_1);
ip->i_d.di_onlink = ip->i_d.di_nlink;
} else {
/*
* The superblock version has already been bumped,
* so just make the conversion to the new inode
* format permanent.
*/
ip->i_d.di_version = 2;
ip->i_d.di_onlink = 0;
memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
}
}
switch (ip->i_d.di_format) {
case XFS_DINODE_FMT_EXTENTS:
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
XFS_ILOG_DEV | XFS_ILOG_UUID)));
if (iip->ili_format.ilf_fields & XFS_ILOG_DEXT) {
ASSERT(ip->i_df.if_bytes > 0);
ASSERT(ip->i_df.if_u1.if_extents != NULL);
ASSERT(ip->i_d.di_nextents > 0);
ASSERT(iip->ili_extents_buf == NULL);
nrecs = ip->i_df.if_bytes /
(uint)sizeof(xfs_bmbt_rec_t);
ASSERT(nrecs > 0);
#ifdef XFS_NATIVE_HOST
if (nrecs == ip->i_d.di_nextents) {
/*
* There are no delayed allocation
* extents, so just point to the
* real extents array.
*/
vecp->i_addr =
(char *)(ip->i_df.if_u1.if_extents);
vecp->i_len = ip->i_df.if_bytes;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_IEXT);
} else
#endif
{
/*
* There are delayed allocation extents
* in the inode, or we need to convert
* the extents to on disk format.
* Use xfs_iextents_copy()
* to copy only the real extents into
* a separate buffer. We'll free the
* buffer in the unlock routine.
*/
ext_buffer = kmem_alloc(ip->i_df.if_bytes,
KM_SLEEP);
iip->ili_extents_buf = ext_buffer;
vecp->i_addr = (xfs_caddr_t)ext_buffer;
vecp->i_len = xfs_iextents_copy(ip, ext_buffer,
XFS_DATA_FORK);
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_IEXT);
}
ASSERT(vecp->i_len <= ip->i_df.if_bytes);
iip->ili_format.ilf_dsize = vecp->i_len;
vecp++;
nvecs++;
}
break;
case XFS_DINODE_FMT_BTREE:
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_DDATA | XFS_ILOG_DEXT |
XFS_ILOG_DEV | XFS_ILOG_UUID)));
if (iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) {
ASSERT(ip->i_df.if_broot_bytes > 0);
ASSERT(ip->i_df.if_broot != NULL);
vecp->i_addr = (xfs_caddr_t)ip->i_df.if_broot;
vecp->i_len = ip->i_df.if_broot_bytes;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_IBROOT);
vecp++;
nvecs++;
iip->ili_format.ilf_dsize = ip->i_df.if_broot_bytes;
}
break;
case XFS_DINODE_FMT_LOCAL:
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
XFS_ILOG_DEV | XFS_ILOG_UUID)));
if (iip->ili_format.ilf_fields & XFS_ILOG_DDATA) {
ASSERT(ip->i_df.if_bytes > 0);
ASSERT(ip->i_df.if_u1.if_data != NULL);
ASSERT(ip->i_d.di_size > 0);
vecp->i_addr = (xfs_caddr_t)ip->i_df.if_u1.if_data;
/*
* Round i_bytes up to a word boundary.
* The underlying memory is guaranteed to
* to be there by xfs_idata_realloc().
*/
data_bytes = roundup(ip->i_df.if_bytes, 4);
ASSERT((ip->i_df.if_real_bytes == 0) ||
(ip->i_df.if_real_bytes == data_bytes));
vecp->i_len = (int)data_bytes;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_ILOCAL);
vecp++;
nvecs++;
iip->ili_format.ilf_dsize = (unsigned)data_bytes;
}
break;
case XFS_DINODE_FMT_DEV:
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
XFS_ILOG_DDATA | XFS_ILOG_UUID)));
if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
iip->ili_format.ilf_u.ilfu_rdev =
ip->i_df.if_u2.if_rdev;
}
break;
case XFS_DINODE_FMT_UUID:
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
XFS_ILOG_DDATA | XFS_ILOG_DEV)));
if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
iip->ili_format.ilf_u.ilfu_uuid =
ip->i_df.if_u2.if_uuid;
}
break;
default:
ASSERT(0);
break;
}
/*
* If there are no attributes associated with the file,
* then we're done.
* Assert that no attribute-related log flags are set.
*/
if (!XFS_IFORK_Q(ip)) {
ASSERT(nvecs == iip->ili_item.li_desc->lid_size);
iip->ili_format.ilf_size = nvecs;
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT)));
return;
}
switch (ip->i_d.di_aformat) {
case XFS_DINODE_FMT_EXTENTS:
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_ADATA | XFS_ILOG_ABROOT)));
if (iip->ili_format.ilf_fields & XFS_ILOG_AEXT) {
ASSERT(ip->i_afp->if_bytes > 0);
ASSERT(ip->i_afp->if_u1.if_extents != NULL);
ASSERT(ip->i_d.di_anextents > 0);
#ifdef DEBUG
nrecs = ip->i_afp->if_bytes /
(uint)sizeof(xfs_bmbt_rec_t);
#endif
ASSERT(nrecs > 0);
ASSERT(nrecs == ip->i_d.di_anextents);
#ifdef XFS_NATIVE_HOST
/*
* There are not delayed allocation extents
* for attributes, so just point at the array.
*/
vecp->i_addr = (char *)(ip->i_afp->if_u1.if_extents);
vecp->i_len = ip->i_afp->if_bytes;
#else
ASSERT(iip->ili_aextents_buf == NULL);
/*
* Need to endian flip before logging
*/
ext_buffer = kmem_alloc(ip->i_afp->if_bytes,
KM_SLEEP);
iip->ili_aextents_buf = ext_buffer;
vecp->i_addr = (xfs_caddr_t)ext_buffer;
vecp->i_len = xfs_iextents_copy(ip, ext_buffer,
XFS_ATTR_FORK);
#endif
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_IATTR_EXT);
iip->ili_format.ilf_asize = vecp->i_len;
vecp++;
nvecs++;
}
break;
case XFS_DINODE_FMT_BTREE:
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_ADATA | XFS_ILOG_AEXT)));
if (iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) {
ASSERT(ip->i_afp->if_broot_bytes > 0);
ASSERT(ip->i_afp->if_broot != NULL);
vecp->i_addr = (xfs_caddr_t)ip->i_afp->if_broot;
vecp->i_len = ip->i_afp->if_broot_bytes;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_IATTR_BROOT);
vecp++;
nvecs++;
iip->ili_format.ilf_asize = ip->i_afp->if_broot_bytes;
}
break;
case XFS_DINODE_FMT_LOCAL:
ASSERT(!(iip->ili_format.ilf_fields &
(XFS_ILOG_ABROOT | XFS_ILOG_AEXT)));
if (iip->ili_format.ilf_fields & XFS_ILOG_ADATA) {
ASSERT(ip->i_afp->if_bytes > 0);
ASSERT(ip->i_afp->if_u1.if_data != NULL);
vecp->i_addr = (xfs_caddr_t)ip->i_afp->if_u1.if_data;
/*
* Round i_bytes up to a word boundary.
* The underlying memory is guaranteed to
* to be there by xfs_idata_realloc().
*/
data_bytes = roundup(ip->i_afp->if_bytes, 4);
ASSERT((ip->i_afp->if_real_bytes == 0) ||
(ip->i_afp->if_real_bytes == data_bytes));
vecp->i_len = (int)data_bytes;
XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_IATTR_LOCAL);
vecp++;
nvecs++;
iip->ili_format.ilf_asize = (unsigned)data_bytes;
}
break;
default:
ASSERT(0);
break;
}
ASSERT(nvecs == iip->ili_item.li_desc->lid_size);
iip->ili_format.ilf_size = nvecs;
}
