if (file->f_mode & FMODE_WRITE) {
status = dquot_initialize(inode); if (status) goto leave;
}
spin_lock(&oi->ip_lock);
/* Check that the inode hasn't been wiped from disk by another * node. If it hasn't then we're safe as long as we hold the
* spin lock until our increment of open count. */ if (oi->ip_flags & OCFS2_INODE_DELETED) {
spin_unlock(&oi->ip_lock);
status = -ENOENT; goto leave;
}
if (mode & O_DIRECT)
oi->ip_flags |= OCFS2_INODE_OPEN_DIRECT;
oi->ip_open_count++;
spin_unlock(&oi->ip_lock);
status = ocfs2_init_file_private(inode, file); if (status) { /* * We want to set open count back if we're failing the * open.
*/
spin_lock(&oi->ip_lock);
oi->ip_open_count--;
spin_unlock(&oi->ip_lock);
}
/* * We can be called with no vfsmnt structure - NFSD will * sometimes do this. * * Note that our action here is different than touch_atime() - * if we can't tell whether this is a noatime mount, then we * don't know whether to trust the value of s_atime_quantum.
*/ if (vfsmnt == NULL) return 0;
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) {
ret = PTR_ERR(handle);
mlog_errno(ret); goto out;
}
ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret) {
mlog_errno(ret); goto out_commit;
}
/* * Don't use ocfs2_mark_inode_dirty() here as we don't always * have i_rwsem to guard against concurrent changes to other * inode fields.
*/
inode_set_atime_to_ts(inode, current_time(inode));
di->i_atime = cpu_to_le64(inode_get_atime_sec(inode));
di->i_atime_nsec = cpu_to_le32(inode_get_atime_nsec(inode));
ocfs2_update_inode_fsync_trans(handle, inode, 0);
ocfs2_journal_dirty(handle, bh);
/* * If the new offset is aligned to the range of the cluster, there is * no space for ocfs2_zero_range_for_truncate to fill, so no need to * CoW either.
*/ if ((offset & (OCFS2_SB(inode->i_sb)->s_clustersize - 1)) == 0) return 0;
status = ocfs2_get_clusters(inode, cpos, &phys,
&num_clusters, &ext_flags); if (status) {
mlog_errno(status); goto out;
}
if (!(ext_flags & OCFS2_EXT_REFCOUNTED)) goto out;
/* * We need to CoW the cluster contains the offset if it is reflinked * since we will call ocfs2_zero_range_for_truncate later which will * write "0" from offset to the end of the cluster.
*/
status = ocfs2_cow_file_pos(inode, fe_bh, new_i_size); if (status) {
mlog_errno(status); return status;
}
/* TODO: This needs to actually orphan the inode in this
* transaction. */
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) {
status = PTR_ERR(handle);
mlog_errno(status); goto out;
}
status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), fe_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (status < 0) {
mlog_errno(status); goto out_commit;
}
/* * Do this before setting i_size.
*/
cluster_bytes = ocfs2_align_bytes_to_clusters(inode->i_sb, new_i_size);
status = ocfs2_zero_range_for_truncate(inode, handle, new_i_size,
cluster_bytes); if (status) {
mlog_errno(status); goto out_commit;
}
/* * The inode lock forced other nodes to sync and drop their * pages, which (correctly) happens even if we have a truncate * without allocation change - ocfs2 cluster sizes can be much * greater than page size, so we have to truncate them * anyway.
*/
if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
unmap_mapping_range(inode->i_mapping,
new_i_size + PAGE_SIZE - 1, 0, 1);
truncate_inode_pages(inode->i_mapping, new_i_size);
status = ocfs2_truncate_inline(inode, di_bh, new_i_size,
i_size_read(inode), 1); if (status)
mlog_errno(status);
goto bail_unlock_sem;
}
/* alright, we're going to need to do a full blown alloc size * change. Orphan the inode so that recovery can complete the * truncate if necessary. This does the task of marking
* i_size. */
status = ocfs2_orphan_for_truncate(osb, inode, di_bh, new_i_size); if (status < 0) {
mlog_errno(status); goto bail_unlock_sem;
}
status = ocfs2_commit_truncate(osb, inode, di_bh); if (status < 0) {
mlog_errno(status); goto bail_unlock_sem;
}
/* TODO: orphan dir cleanup here. */
bail_unlock_sem:
up_write(&OCFS2_I(inode)->ip_alloc_sem);
bail: if (!status && OCFS2_I(inode)->ip_clusters == 0)
status = ocfs2_try_remove_refcount_tree(inode, di_bh);
return status;
}
/* * extend file allocation only here. * we'll update all the disk stuff, and oip->alloc_size * * expect stuff to be locked, a transaction started and enough data / * metadata reservations in the contexts. * * Will return -EAGAIN, and a reason if a restart is needed. * If passed in, *reason will always be set, even in error.
*/ int ocfs2_add_inode_data(struct ocfs2_super *osb, struct inode *inode,
u32 *logical_offset,
u32 clusters_to_add, int mark_unwritten, struct buffer_head *fe_bh,
handle_t *handle, struct ocfs2_alloc_context *data_ac, struct ocfs2_alloc_context *meta_ac, enum ocfs2_alloc_restarted *reason_ret)
{ struct ocfs2_extent_tree et;
status = dquot_alloc_space_nodirty(inode,
ocfs2_clusters_to_bytes(osb->sb, clusters_to_add)); if (status) goto leave;
did_quota = 1;
/* reserve a write to the file entry early on - that we if we * run out of credits in the allocation path, we can still
* update i_size. */
status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (status < 0) {
mlog_errno(status); goto leave;
}
if (why != RESTART_NONE && clusters_to_add) { if (why == RESTART_META) {
restart_func = 1;
status = 0;
} else {
BUG_ON(why != RESTART_TRANS);
status = ocfs2_allocate_extend_trans(handle, 1); if (status < 0) { /* handle still has to be committed at
* this point. */
status = -ENOMEM;
mlog_errno(status); goto leave;
} goto restarted_transaction;
}
}
leave: if (status < 0 && did_quota)
dquot_free_space(inode,
ocfs2_clusters_to_bytes(osb->sb, clusters_to_add)); if (handle) {
ocfs2_commit_trans(osb, handle);
handle = NULL;
} if (data_ac) {
ocfs2_free_alloc_context(data_ac);
data_ac = NULL;
} if (meta_ac) {
ocfs2_free_alloc_context(meta_ac);
meta_ac = NULL;
} if ((!status) && restart_func) {
restart_func = 0; goto restart_all;
}
brelse(bh);
bh = NULL;
return status;
}
/* * While a write will already be ordering the data, a truncate will not. * Thus, we need to explicitly order the zeroed pages.
*/ static handle_t *ocfs2_zero_start_ordered_transaction(struct inode *inode, struct buffer_head *di_bh,
loff_t start_byte,
loff_t length)
{ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
handle_t *handle = NULL; int ret = 0;
if (!ocfs2_should_order_data(inode)) goto out;
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) {
ret = -ENOMEM;
mlog_errno(ret); goto out;
}
ret = ocfs2_jbd2_inode_add_write(handle, inode, start_byte, length); if (ret < 0) {
mlog_errno(ret); goto out;
}
ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret)
mlog_errno(ret);
ocfs2_update_inode_fsync_trans(handle, inode, 1);
out: if (ret) { if (!IS_ERR(handle))
ocfs2_commit_trans(osb, handle);
handle = ERR_PTR(ret);
} return handle;
}
/* Some parts of this taken from generic_cont_expand, which turned out * to be too fragile to do exactly what we need without us having to
* worry about recursive locking in ->write_begin() and ->write_end(). */ staticint ocfs2_write_zero_page(struct inode *inode, u64 abs_from,
u64 abs_to, struct buffer_head *di_bh)
{ struct address_space *mapping = inode->i_mapping; struct folio *folio; unsignedlong index = abs_from >> PAGE_SHIFT;
handle_t *handle; int ret = 0; unsigned zero_from, zero_to, block_start, block_end; struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
handle = ocfs2_zero_start_ordered_transaction(inode, di_bh,
abs_from,
abs_to - abs_from); if (IS_ERR(handle)) {
ret = PTR_ERR(handle); goto out;
}
folio = __filemap_get_folio(mapping, index,
FGP_LOCK | FGP_ACCESSED | FGP_CREAT, GFP_NOFS); if (IS_ERR(folio)) {
ret = PTR_ERR(folio);
mlog_errno(ret); goto out_commit_trans;
}
/* Get the offsets within the folio that we want to zero */
zero_from = offset_in_folio(folio, abs_from);
zero_to = offset_in_folio(folio, abs_to); if (!zero_to)
zero_to = folio_size(folio);
/* We know that zero_from is block aligned */ for (block_start = zero_from; block_start < zero_to;
block_start = block_end) {
block_end = block_start + i_blocksize(inode);
/* * block_start is block-aligned. Bump it by one to force * __block_write_begin and block_commit_write to zero the * whole block.
*/
ret = __block_write_begin(folio, block_start + 1, 0,
ocfs2_get_block); if (ret < 0) {
mlog_errno(ret); goto out_unlock;
}
/* must not update i_size! */
block_commit_write(folio, block_start + 1, block_start + 1);
}
/* * fs-writeback will release the dirty pages without page lock * whose offset are over inode size, the release happens at * block_write_full_folio().
*/
i_size_write(inode, abs_to);
inode->i_blocks = ocfs2_inode_sector_count(inode);
di->i_size = cpu_to_le64((u64)i_size_read(inode));
inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode));
di->i_mtime = di->i_ctime = cpu_to_le64(inode_get_mtime_sec(inode));
di->i_ctime_nsec = cpu_to_le32(inode_get_mtime_nsec(inode));
di->i_mtime_nsec = di->i_ctime_nsec; if (handle) {
ocfs2_journal_dirty(handle, di_bh);
ocfs2_update_inode_fsync_trans(handle, inode, 1);
}
/* * Find the next range to zero. We do this in terms of bytes because * that's what ocfs2_zero_extend() wants, and it is dealing with the * pagecache. We may return multiple extents. * * zero_start and zero_end are ocfs2_zero_extend()s current idea of what * needs to be zeroed. range_start and range_end return the next zeroing * range. A subsequent call should pass the previous range_end as its * zero_start. If range_end is 0, there's nothing to do. * * Unwritten extents are skipped over. Refcounted extents are CoWd.
*/ staticint ocfs2_zero_extend_get_range(struct inode *inode, struct buffer_head *di_bh,
u64 zero_start, u64 zero_end,
u64 *range_start, u64 *range_end)
{ int rc = 0, needs_cow = 0;
u32 p_cpos, zero_clusters = 0;
u32 zero_cpos =
zero_start >> OCFS2_SB(inode->i_sb)->s_clustersize_bits;
u32 last_cpos = ocfs2_clusters_for_bytes(inode->i_sb, zero_end); unsignedint num_clusters = 0; unsignedint ext_flags = 0;
while (zero_cpos < last_cpos) {
rc = ocfs2_get_clusters(inode, zero_cpos, &p_cpos,
&num_clusters, &ext_flags); if (rc) {
mlog_errno(rc); goto out;
}
if (p_cpos && !(ext_flags & OCFS2_EXT_UNWRITTEN)) {
zero_clusters = num_clusters; if (ext_flags & OCFS2_EXT_REFCOUNTED)
needs_cow = 1; break;
}
/* * Zero one range returned from ocfs2_zero_extend_get_range(). The caller * has made sure that the entire range needs zeroing.
*/ staticint ocfs2_zero_extend_range(struct inode *inode, u64 range_start,
u64 range_end, struct buffer_head *di_bh)
{ int rc = 0;
u64 next_pos;
u64 zero_pos = range_start;
/* * Only quota files call this without a bh, and they can't be * refcounted.
*/
BUG_ON(!di_bh && ocfs2_is_refcount_inode(inode));
BUG_ON(!di_bh && !(oi->ip_flags & OCFS2_INODE_SYSTEM_FILE));
if (clusters_to_add) {
ret = ocfs2_extend_allocation(inode, oi->ip_clusters,
clusters_to_add, 0); if (ret) {
mlog_errno(ret); goto out;
}
}
/* * Call this even if we don't add any clusters to the tree. We * still need to zero the area between the old i_size and the * new i_size.
*/
ret = ocfs2_zero_extend(inode, di_bh, zero_to); if (ret < 0)
mlog_errno(ret);
out: return ret;
}
staticint ocfs2_extend_file(struct inode *inode, struct buffer_head *di_bh,
u64 new_i_size)
{ int ret = 0; struct ocfs2_inode_info *oi = OCFS2_I(inode);
BUG_ON(!di_bh);
/* setattr sometimes calls us like this. */ if (new_i_size == 0) goto out;
if (i_size_read(inode) == new_i_size) goto out;
BUG_ON(new_i_size < i_size_read(inode));
/* * The alloc sem blocks people in read/write from reading our * allocation until we're done changing it. We depend on * i_rwsem to block other extend/truncate calls while we're * here. We even have to hold it for sparse files because there * might be some tail zeroing.
*/
down_write(&oi->ip_alloc_sem);
if (oi->ip_dyn_features & OCFS2_INLINE_DATA_FL) { /* * We can optimize small extends by keeping the inodes * inline data.
*/ if (ocfs2_size_fits_inline_data(di_bh, new_i_size)) {
up_write(&oi->ip_alloc_sem); goto out_update_size;
}
ret = ocfs2_convert_inline_data_to_extents(inode, di_bh); if (ret) {
up_write(&oi->ip_alloc_sem);
mlog_errno(ret); goto out;
}
}
if (ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb)))
ret = ocfs2_zero_extend(inode, di_bh, new_i_size); else
ret = ocfs2_extend_no_holes(inode, di_bh, new_i_size,
new_i_size);
up_write(&oi->ip_alloc_sem);
if (ret < 0) {
mlog_errno(ret); goto out;
}
out_update_size:
ret = ocfs2_simple_size_update(inode, di_bh, new_i_size); if (ret < 0)
mlog_errno(ret);
out: return ret;
}
int ocfs2_setattr(struct mnt_idmap *idmap, struct dentry *dentry, struct iattr *attr)
{ int status = 0, size_change; int inode_locked = 0; struct inode *inode = d_inode(dentry); struct super_block *sb = inode->i_sb; struct ocfs2_super *osb = OCFS2_SB(sb); struct buffer_head *bh = NULL;
handle_t *handle = NULL; struct dquot *transfer_to[MAXQUOTAS] = { }; int qtype; int had_lock; struct ocfs2_lock_holder oh;
status = setattr_prepare(&nop_mnt_idmap, dentry, attr); if (status) return status;
if (is_quota_modification(&nop_mnt_idmap, inode, attr)) {
status = dquot_initialize(inode); if (status) return status;
}
size_change = S_ISREG(inode->i_mode) && attr->ia_valid & ATTR_SIZE; if (size_change) { /* * Here we should wait dio to finish before inode lock * to avoid a deadlock between ocfs2_setattr() and * ocfs2_dio_end_io_write()
*/
inode_dio_wait(inode);
status = ocfs2_rw_lock(inode, 1); if (status < 0) {
mlog_errno(status); goto bail;
}
}
had_lock = ocfs2_inode_lock_tracker(inode, &bh, 1, &oh); if (had_lock < 0) {
status = had_lock; goto bail_unlock_rw;
} elseif (had_lock) { /* * As far as we know, ocfs2_setattr() could only be the first * VFS entry point in the call chain of recursive cluster * locking issue. * * For instance: * chmod_common() * notify_change() * ocfs2_setattr() * posix_acl_chmod() * ocfs2_iop_get_acl() * * But, we're not 100% sure if it's always true, because the * ordering of the VFS entry points in the call chain is out * of our control. So, we'd better dump the stack here to * catch the other cases of recursive locking.
*/
mlog(ML_ERROR, "Another case of recursive locking:\n");
dump_stack();
}
inode_locked = 1;
if (size_change) {
status = inode_newsize_ok(inode, attr->ia_size); if (status) goto bail_unlock;
if (i_size_read(inode) >= attr->ia_size) { if (ocfs2_should_order_data(inode)) {
status = ocfs2_begin_ordered_truncate(inode,
attr->ia_size); if (status) goto bail_unlock;
}
status = ocfs2_truncate_file(inode, bh, attr->ia_size);
} else
status = ocfs2_extend_file(inode, bh, attr->ia_size); if (status < 0) { if (status != -ENOSPC)
mlog_errno(status);
status = -ENOSPC; goto bail_unlock;
}
}
if ((attr->ia_valid & ATTR_UID && !uid_eq(attr->ia_uid, inode->i_uid)) ||
(attr->ia_valid & ATTR_GID && !gid_eq(attr->ia_gid, inode->i_gid))) { /* * Gather pointers to quota structures so that allocation / * freeing of quota structures happens here and not inside * dquot_transfer() where we have problems with lock ordering
*/ if (attr->ia_valid & ATTR_UID && !uid_eq(attr->ia_uid, inode->i_uid)
&& OCFS2_HAS_RO_COMPAT_FEATURE(sb,
OCFS2_FEATURE_RO_COMPAT_USRQUOTA)) {
transfer_to[USRQUOTA] = dqget(sb, make_kqid_uid(attr->ia_uid)); if (IS_ERR(transfer_to[USRQUOTA])) {
status = PTR_ERR(transfer_to[USRQUOTA]);
transfer_to[USRQUOTA] = NULL; goto bail_unlock;
}
} if (attr->ia_valid & ATTR_GID && !gid_eq(attr->ia_gid, inode->i_gid)
&& OCFS2_HAS_RO_COMPAT_FEATURE(sb,
OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)) {
transfer_to[GRPQUOTA] = dqget(sb, make_kqid_gid(attr->ia_gid)); if (IS_ERR(transfer_to[GRPQUOTA])) {
status = PTR_ERR(transfer_to[GRPQUOTA]);
transfer_to[GRPQUOTA] = NULL; goto bail_unlock;
}
}
down_write(&OCFS2_I(inode)->ip_alloc_sem);
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS +
2 * ocfs2_quota_trans_credits(sb)); if (IS_ERR(handle)) {
status = PTR_ERR(handle);
mlog_errno(status); goto bail_unlock_alloc;
}
status = __dquot_transfer(inode, transfer_to); if (status < 0) goto bail_commit;
} else {
down_write(&OCFS2_I(inode)->ip_alloc_sem);
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) {
status = PTR_ERR(handle);
mlog_errno(status); goto bail_unlock_alloc;
}
}
err = ocfs2_inode_revalidate(path->dentry); if (err) { if (err != -ENOENT)
mlog_errno(err); goto bail;
}
generic_fillattr(&nop_mnt_idmap, request_mask, inode, stat); /* * If there is inline data in the inode, the inode will normally not * have data blocks allocated (it may have an external xattr block). * Report at least one sector for such files, so tools like tar, rsync, * others don't incorrectly think the file is completely sparse.
*/ if (unlikely(OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL))
stat->blocks += (stat->size + 511)>>9;
/* We set the blksize from the cluster size for performance */
stat->blksize = osb->s_clustersize;
bail: return err;
}
int ocfs2_permission(struct mnt_idmap *idmap, struct inode *inode, int mask)
{ int ret, had_lock; struct ocfs2_lock_holder oh;
if (mask & MAY_NOT_BLOCK) return -ECHILD;
had_lock = ocfs2_inode_lock_tracker(inode, NULL, 0, &oh); if (had_lock < 0) {
ret = had_lock; goto out;
} elseif (had_lock) { /* See comments in ocfs2_setattr() for details. * The call chain of this case could be: * do_sys_open() * may_open() * inode_permission() * ocfs2_permission() * ocfs2_iop_get_acl()
*/
mlog(ML_ERROR, "Another case of recursive locking:\n");
dump_stack();
}
ret = generic_permission(&nop_mnt_idmap, inode, mask);
ret = ocfs2_read_inode_block(inode, &bh); if (ret < 0) {
mlog_errno(ret); goto out;
}
ret = __ocfs2_write_remove_suid(inode, bh);
out:
brelse(bh); return ret;
}
/* * Allocate enough extents to cover the region starting at byte offset * start for len bytes. Existing extents are skipped, any extents * added are marked as "unwritten".
*/ staticint ocfs2_allocate_unwritten_extents(struct inode *inode,
u64 start, u64 len)
{ int ret;
u32 cpos, phys_cpos, clusters, alloc_size;
u64 end = start + len; struct buffer_head *di_bh = NULL;
if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
ret = ocfs2_read_inode_block(inode, &di_bh); if (ret) {
mlog_errno(ret); goto out;
}
/* * Nothing to do if the requested reservation range * fits within the inode.
*/ if (ocfs2_size_fits_inline_data(di_bh, end)) goto out;
ret = ocfs2_convert_inline_data_to_extents(inode, di_bh); if (ret) {
mlog_errno(ret); goto out;
}
}
/* * We consider both start and len to be inclusive.
*/
cpos = start >> OCFS2_SB(inode->i_sb)->s_clustersize_bits;
clusters = ocfs2_clusters_for_bytes(inode->i_sb, start + len);
clusters -= cpos;
while (clusters) {
ret = ocfs2_get_clusters(inode, cpos, &phys_cpos,
&alloc_size, NULL); if (ret) {
mlog_errno(ret); goto out;
}
/* * Hole or existing extent len can be arbitrary, so * cap it to our own allocation request.
*/ if (alloc_size > clusters)
alloc_size = clusters;
if (phys_cpos) { /* * We already have an allocation at this * region so we can safely skip it.
*/ goto next;
}
ret = ocfs2_extend_allocation(inode, cpos, alloc_size, 1); if (ret) { if (ret != -ENOSPC)
mlog_errno(ret); goto out;
}
/* * Truncate a byte range, avoiding pages within partial clusters. This * preserves those pages for the zeroing code to write to.
*/ staticvoid ocfs2_truncate_cluster_pages(struct inode *inode, u64 byte_start,
u64 byte_len)
{ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
loff_t start, end; struct address_space *mapping = inode->i_mapping;
start = (loff_t)ocfs2_align_bytes_to_clusters(inode->i_sb, byte_start);
end = byte_start + byte_len;
end = end & ~(osb->s_clustersize - 1);
if (start < end) {
unmap_mapping_range(mapping, start, end - start, 0);
truncate_inode_pages_range(mapping, start, end - 1);
}
}
/* * zero out partial blocks of one cluster. * * start: file offset where zero starts, will be made upper block aligned. * len: it will be trimmed to the end of current cluster if "start + len" * is bigger than it.
*/ staticint ocfs2_zeroout_partial_cluster(struct inode *inode,
u64 start, u64 len)
{ int ret;
u64 start_block, end_block, nr_blocks;
u64 p_block, offset;
u32 cluster, p_cluster, nr_clusters; struct super_block *sb = inode->i_sb;
u64 end = ocfs2_align_bytes_to_clusters(sb, start);
/* * The "start" and "end" values are NOT necessarily part of * the range whose allocation is being deleted. Rather, this * is what the user passed in with the request. We must zero * partial clusters here. There's no need to worry about * physical allocation - the zeroing code knows to skip holes.
*/
trace_ocfs2_zero_partial_clusters(
(unsignedlonglong)OCFS2_I(inode)->ip_blkno,
(unsignedlonglong)start, (unsignedlonglong)end);
/* * If both edges are on a cluster boundary then there's no * zeroing required as the region is part of the allocation to * be truncated.
*/ if ((start & (csize - 1)) == 0 && (end & (csize - 1)) == 0) goto out;
/* No page cache for EOF blocks, issue zero out to disk. */ if (end > isize) { /* * zeroout eof blocks in last cluster starting from * "isize" even "start" > "isize" because it is * complicated to zeroout just at "start" as "start" * may be not aligned with block size, buffer write * would be required to do that, but out of eof buffer * write is not supported.
*/
ret = ocfs2_zeroout_partial_cluster(inode, isize,
end - isize); if (ret) {
mlog_errno(ret); goto out;
} if (start >= isize) goto out;
end = isize;
}
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) {
ret = PTR_ERR(handle);
mlog_errno(ret); goto out;
}
/* * If start is on a cluster boundary and end is somewhere in another * cluster, we have not COWed the cluster starting at start, unless * end is also within the same cluster. So, in this case, we skip this * first call to ocfs2_zero_range_for_truncate() truncate and move on * to the next one.
*/ if ((start & (csize - 1)) != 0) { /* * We want to get the byte offset of the end of the 1st * cluster.
*/
tmpend = (u64)osb->s_clustersize +
(start & ~(osb->s_clustersize - 1)); if (tmpend > end)
tmpend = end;
ret = ocfs2_zero_range_for_truncate(inode, handle, start,
tmpend); if (ret)
mlog_errno(ret);
}
if (tmpend < end) { /* * This may make start and end equal, but the zeroing * code will skip any work in that case so there's no * need to catch it up here.
*/
start = end & ~(osb->s_clustersize - 1);
for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
rec = &el->l_recs[i];
if (le32_to_cpu(rec->e_cpos) < pos) break;
}
return i;
}
/* * Helper to calculate the punching pos and length in one run, we handle the * following three cases in order: * * - remove the entire record * - remove a partial record * - no record needs to be removed (hole-punching completed)
*/ staticvoid ocfs2_calc_trunc_pos(struct inode *inode, struct ocfs2_extent_list *el, struct ocfs2_extent_rec *rec,
u32 trunc_start, u32 *trunc_cpos,
u32 *trunc_len, u32 *trunc_end,
u64 *blkno, int *done)
{ int ret = 0;
u32 coff, range;
range = le32_to_cpu(rec->e_cpos) + ocfs2_rec_clusters(el, rec);
if (le32_to_cpu(rec->e_cpos) >= trunc_start) { /* * remove an entire extent record.
*/
*trunc_cpos = le32_to_cpu(rec->e_cpos); /* * Skip holes if any.
*/ if (range < *trunc_end)
*trunc_end = range;
*trunc_len = *trunc_end - le32_to_cpu(rec->e_cpos);
*blkno = le64_to_cpu(rec->e_blkno);
*trunc_end = le32_to_cpu(rec->e_cpos);
} elseif (range > trunc_start) { /* * remove a partial extent record, which means we're * removing the last extent record.
*/
*trunc_cpos = trunc_start; /* * skip hole if any.
*/ if (range < *trunc_end)
*trunc_end = range;
*trunc_len = *trunc_end - trunc_start;
coff = trunc_start - le32_to_cpu(rec->e_cpos);
*blkno = le64_to_cpu(rec->e_blkno) +
ocfs2_clusters_to_blocks(inode->i_sb, coff);
*trunc_end = trunc_start;
} else { /* * It may have two following possibilities: * * - last record has been removed * - trunc_start was within a hole * * both two cases mean the completion of hole punching.
*/
ret = 1;
}
if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) { int id_count = ocfs2_max_inline_data_with_xattr(inode->i_sb, di);
if (byte_start > id_count || byte_start + byte_len > id_count) {
ret = -EINVAL;
mlog_errno(ret); goto out;
}
ret = ocfs2_truncate_inline(inode, di_bh, byte_start,
byte_start + byte_len, 0); if (ret) {
mlog_errno(ret); goto out;
} /* * There's no need to get fancy with the page cache * truncate of an inline-data inode. We're talking * about less than a page here, which will be cached * in the dinode buffer anyway.
*/
unmap_mapping_range(mapping, 0, 0, 0);
truncate_inode_pages(mapping, 0); goto out;
}
/* * For reflinks, we may need to CoW 2 clusters which might be * partially zero'd later, if hole's start and end offset were * within one cluster(means is not exactly aligned to clustersize).
*/
if (ocfs2_is_refcount_inode(inode)) {
ret = ocfs2_cow_file_pos(inode, di_bh, byte_start); if (ret) {
mlog_errno(ret); goto out;
}
ret = ocfs2_cow_file_pos(inode, di_bh, byte_start + byte_len); if (ret) {
mlog_errno(ret); goto out;
}
}
/* * Parts of this function taken from xfs_change_file_space()
*/ staticint __ocfs2_change_file_space(struct file *file, struct inode *inode,
loff_t f_pos, unsignedint cmd, struct ocfs2_space_resv *sr, int change_size)
{ int ret;
s64 llen;
loff_t size, orig_isize; struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); struct buffer_head *di_bh = NULL;
handle_t *handle; unsignedlonglong max_off = inode->i_sb->s_maxbytes;
if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb)) return -EROFS;
inode_lock(inode);
/* Wait all existing dio workers, newcomers will block on i_rwsem */
inode_dio_wait(inode); /* * This prevents concurrent writes on other nodes
*/
ret = ocfs2_rw_lock(inode, 1); if (ret) {
mlog_errno(ret); goto out;
}
ret = ocfs2_inode_lock(inode, &di_bh, 1); if (ret) {
mlog_errno(ret); goto out_rw_unlock;
}
if (inode->i_flags & (S_IMMUTABLE|S_APPEND)) {
ret = -EPERM; goto out_inode_unlock;
}
switch (sr->l_whence) { case 0: /*SEEK_SET*/ break; case 1: /*SEEK_CUR*/
sr->l_start += f_pos; break; case 2: /*SEEK_END*/
sr->l_start += i_size_read(inode); break; default:
ret = -EINVAL; goto out_inode_unlock;
}
sr->l_whence = 0;
if (cmd == OCFS2_IOC_RESVSP || cmd == OCFS2_IOC_RESVSP64 ||
cmd == OCFS2_IOC_UNRESVSP || cmd == OCFS2_IOC_UNRESVSP64) { if (sr->l_len <= 0) {
ret = -EINVAL; goto out_inode_unlock;
}
}
if (file && setattr_should_drop_suidgid(&nop_mnt_idmap, file_inode(file))) {
ret = __ocfs2_write_remove_suid(inode, di_bh); if (ret) {
mlog_errno(ret); goto out_inode_unlock;
}
}
down_write(&OCFS2_I(inode)->ip_alloc_sem); switch (cmd) { case OCFS2_IOC_RESVSP: case OCFS2_IOC_RESVSP64: /* * This takes unsigned offsets, but the signed ones we * pass have been checked against overflow above.
*/
ret = ocfs2_allocate_unwritten_extents(inode, sr->l_start,
sr->l_len); break; case OCFS2_IOC_UNRESVSP: case OCFS2_IOC_UNRESVSP64:
ret = ocfs2_remove_inode_range(inode, di_bh, sr->l_start,
sr->l_len); break; default:
ret = -EINVAL;
}
orig_isize = i_size_read(inode); /* zeroout eof blocks in the cluster. */ if (!ret && change_size && orig_isize < size) {
ret = ocfs2_zeroout_partial_cluster(inode, orig_isize,
size - orig_isize); if (!ret)
i_size_write(inode, size);
}
up_write(&OCFS2_I(inode)->ip_alloc_sem); if (ret) {
mlog_errno(ret); goto out_inode_unlock;
}
/* * We update c/mtime for these changes
*/
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) {
ret = PTR_ERR(handle);
mlog_errno(ret); goto out_inode_unlock;
}
inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode));
ret = ocfs2_mark_inode_dirty(handle, inode, di_bh); if (ret < 0)
mlog_errno(ret);
if (file && (file->f_flags & O_SYNC))
handle->h_sync = 1;
staticint ocfs2_inode_lock_for_extent_tree(struct inode *inode, struct buffer_head **di_bh, int meta_level, int write_sem, int wait)
{ int ret = 0;
if (wait)
ret = ocfs2_inode_lock(inode, di_bh, meta_level); else
ret = ocfs2_try_inode_lock(inode, di_bh, meta_level); if (ret < 0) goto out;
if (wait) { if (write_sem)
down_write(&OCFS2_I(inode)->ip_alloc_sem); else
down_read(&OCFS2_I(inode)->ip_alloc_sem);
} else { if (write_sem)
ret = down_write_trylock(&OCFS2_I(inode)->ip_alloc_sem); else
ret = down_read_trylock(&OCFS2_I(inode)->ip_alloc_sem);
/* * We start with a read level meta lock and only jump to an ex * if we need to make modifications here.
*/ for(;;) {
ret = ocfs2_inode_lock_for_extent_tree(inode,
&di_bh,
meta_level,
write_sem,
wait); if (ret < 0) { if (ret != -EAGAIN)
mlog_errno(ret); goto out;
}
/* * Check if IO will overwrite allocated blocks in case * IOCB_NOWAIT flag is set.
*/ if (!wait && !overwrite_io) {
overwrite_io = 1;
ret = ocfs2_overwrite_io(inode, di_bh, pos, count); if (ret < 0) { if (ret != -EAGAIN)
mlog_errno(ret); goto out_unlock;
}
}
/* Clear suid / sgid if necessary. We do this here * instead of later in the write path because * remove_suid() calls ->setattr without any hint that * we may have already done our cluster locking. Since * ocfs2_setattr() *must* take cluster locks to * proceed, this will lead us to recursively lock the * inode. There's also the dinode i_size state which * can be lost via setattr during extending writes (we
* set inode->i_size at the end of a write. */ if (setattr_should_drop_suidgid(&nop_mnt_idmap, inode)) { if (meta_level == 0) {
ocfs2_inode_unlock_for_extent_tree(inode,
&di_bh,
meta_level,
write_sem);
meta_level = 1; continue;
}
ret = ocfs2_write_remove_suid(inode); if (ret < 0) {
mlog_errno(ret); goto out_unlock;
}
}
ret = ocfs2_check_range_for_refcount(inode, pos, count); if (ret == 1) {
ocfs2_inode_unlock_for_extent_tree(inode,
&di_bh,
meta_level,
write_sem);
meta_level = 1;
write_sem = 1;
ret = ocfs2_inode_lock_for_extent_tree(inode,
&di_bh,
meta_level,
write_sem,
wait); if (ret < 0) { if (ret != -EAGAIN)
mlog_errno(ret); goto out;
}
if (nowait) { if (!inode_trylock(inode)) return -EAGAIN;
} else
inode_lock(inode);
ocfs2_iocb_init_rw_locked(iocb);
/* * Concurrent O_DIRECT writes are allowed with * mount_option "coherency=buffered". * For append write, we must take rw EX.
*/
rw_level = (!direct_io || full_coherency || append_write);
if (nowait)
ret = ocfs2_try_rw_lock(inode, rw_level); else
ret = ocfs2_rw_lock(inode, rw_level); if (ret < 0) { if (ret != -EAGAIN)
mlog_errno(ret); goto out_mutex;
}
/* * O_DIRECT writes with "coherency=full" need to take EX cluster * inode_lock to guarantee coherency.
*/ if (direct_io && full_coherency) { /* * We need to take and drop the inode lock to force * other nodes to drop their caches. Buffered I/O * already does this in write_begin().
*/ if (nowait)
ret = ocfs2_try_inode_lock(inode, NULL, 1); else
ret = ocfs2_inode_lock(inode, NULL, 1); if (ret < 0) { if (ret != -EAGAIN)
mlog_errno(ret); goto out;
}
ocfs2_inode_unlock(inode, 1);
}
ret = generic_write_checks(iocb, from); if (ret <= 0) { if (ret)
mlog_errno(ret); goto out;
}
count = ret;
ret = ocfs2_prepare_inode_for_write(file, iocb->ki_pos, count, !nowait); if (ret < 0) { if (ret != -EAGAIN)
mlog_errno(ret); goto out;
}
if (direct_io && !is_sync_kiocb(iocb) &&
ocfs2_is_io_unaligned(inode, count, iocb->ki_pos)) { /* * Make it a sync io if it's an unaligned aio.
*/
saved_ki_complete = xchg(&iocb->ki_complete, NULL);
}
/* communicate with ocfs2_dio_end_io */
ocfs2_iocb_set_rw_locked(iocb, rw_level);
written = __generic_file_write_iter(iocb, from); /* buffered aio wouldn't have proper lock coverage today */
BUG_ON(written == -EIOCBQUEUED && !direct_io);
/* * deep in g_f_a_w_n()->ocfs2_direct_IO we pass in a ocfs2_dio_end_io * function pointer which is called when o_direct io completes so that * it can unlock our rw lock. * Unfortunately there are error cases which call end_io and others * that don't. so we don't have to unlock the rw_lock if either an * async dio is going to do it in the future or an end_io after an * error has already done it.
*/ if ((written == -EIOCBQUEUED) || (!ocfs2_iocb_is_rw_locked(iocb))) {
rw_level = -1;
}
if (unlikely(written <= 0)) goto out;
if (((file->f_flags & O_DSYNC) && !direct_io) ||
IS_SYNC(inode)) {
ret = filemap_fdatawrite_range(file->f_mapping,
iocb->ki_pos - written,
iocb->ki_pos - 1); if (ret < 0)
written = ret;
if (!ret) {
ret = jbd2_journal_force_commit(osb->journal->j_journal); if (ret < 0)
written = ret;
}
if (!ret)
ret = filemap_fdatawait_range(file->f_mapping,
iocb->ki_pos - written,
iocb->ki_pos - 1);
}
out: if (saved_ki_complete)
xchg(&iocb->ki_complete, saved_ki_complete);
if (rw_level != -1)
ocfs2_rw_unlock(inode, rw_level);
if (!inode) {
ret = -EINVAL;
mlog_errno(ret); goto bail;
}
if (!direct_io && nowait) return -EOPNOTSUPP;
ocfs2_iocb_init_rw_locked(iocb);
/* * buffered reads protect themselves in ->read_folio(). O_DIRECT reads * need locks to protect pending reads from racing with truncate.
*/ if (direct_io) { if (nowait)
ret = ocfs2_try_rw_lock(inode, 0); else
ret = ocfs2_rw_lock(inode, 0);
if (ret < 0) { if (ret != -EAGAIN)
mlog_errno(ret); goto bail;
}
rw_level = 0; /* communicate with ocfs2_dio_end_io */
ocfs2_iocb_set_rw_locked(iocb, rw_level);
}
/* * We're fine letting folks race truncates and extending * writes with read across the cluster, just like they can * locally. Hence no rw_lock during read. * * Take and drop the meta data lock to update inode fields * like i_size. This allows the checks down below * copy_splice_read() a chance of actually working.
*/
ret = ocfs2_inode_lock_atime(inode, filp->f_path.mnt, &lock_level,
!nowait); if (ret < 0) { if (ret != -EAGAIN)
mlog_errno(ret); goto bail;
}
ocfs2_inode_unlock(inode, lock_level);
ret = generic_file_read_iter(iocb, to);
trace_generic_file_read_iter_ret(ret);
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.