/*
* For a given allocation, determine which allocators will need to be
* accessed, and lock them, reserving the appropriate number of bits.
*
* Called from ocfs2_extend_allocation() for file systems which don't
* support holes, and from ocfs2_write() for file systems which
* understand sparse inodes.
*/
int ocfs2_lock_allocators(struct inode *inode, struct ocfs2_dinode *di,
u32 clusters_to_add,
struct ocfs2_alloc_context **data_ac,
struct ocfs2_alloc_context **meta_ac)
{
int ret, num_free_extents;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
*meta_ac = NULL;
*data_ac = NULL;
mlog(0, "extend inode %llu, i_size = %lld, di->i_clusters = %u, "
"clusters_to_add = %u\n",
(unsigned long long)OCFS2_I(inode)->ip_blkno, i_size_read(inode),
le32_to_cpu(di->i_clusters), clusters_to_add);
num_free_extents = ocfs2_num_free_extents(osb, inode, di);
if (num_free_extents < 0) {
ret = num_free_extents;
mlog_errno(ret);
goto out;
}
/*
* Sparse allocation file systems need to be more conservative
* with reserving room for expansion - the actual allocation
* happens while we've got a journal handle open so re-taking
* a cluster lock (because we ran out of room for another
* extent) will violate ordering rules.
*
* Most of the time we'll only be seeing this 1 cluster at a time
* anyway.
*/
if (!num_free_extents ||
(ocfs2_sparse_alloc(osb) && num_free_extents < clusters_to_add)) {
ret = ocfs2_reserve_new_metadata(osb, di, meta_ac);
if (ret < 0) {
if (ret != -ENOSPC)
mlog_errno(ret);
goto out;
}
}
ret = ocfs2_reserve_clusters(osb, clusters_to_add, data_ac);
if (ret < 0) {
if (ret != -ENOSPC)
mlog_errno(ret);
goto out;
}
out:
if (ret) {
if (*meta_ac) {
ocfs2_free_alloc_context(*meta_ac);
*meta_ac = NULL;
}
/*
* We cannot have an error and a non null *data_ac.
*/
}
return ret;
}
/*
* TODO: Make this into a generic get_blocks function.
*
* From do_direct_io in direct-io.c:
* "So what we do is to permit the ->get_blocks function to populate
* bh.b_size with the size of IO which is permitted at this offset and
* this i_blkbits."
*
* This function is called directly from get_more_blocks in direct-io.c.
*
* called like this: dio->get_blocks(dio->inode, fs_startblk,
* fs_count, map_bh, dio->rw == WRITE);
*/
static int ocfs2_direct_IO_get_blocks(struct inode *inode, sector_t iblock,
struct buffer_head *bh_result, int create)
{
int ret;
u64 p_blkno, inode_blocks, contig_blocks;
unsigned int ext_flags;
unsigned char blocksize_bits = inode->i_sb->s_blocksize_bits;
unsigned long max_blocks = bh_result->b_size >> inode->i_blkbits;
/* This function won't even be called if the request isn't all
* nicely aligned and of the right size, so there's no need
* for us to check any of that. */
inode_blocks = ocfs2_blocks_for_bytes(inode->i_sb, i_size_read(inode));
/*
* Any write past EOF is not allowed because we'd be extending.
*/
if (create && (iblock + max_blocks) > inode_blocks) {
ret = -EIO;
goto bail;
}
/* This figures out the size of the next contiguous block, and
* our logical offset */
ret = ocfs2_extent_map_get_blocks(inode, iblock, &p_blkno,
&contig_blocks, &ext_flags);
if (ret) {
mlog(ML_ERROR, "get_blocks() failed iblock=%llu\n",
(unsigned long long)iblock);
ret = -EIO;
goto bail;
}
if (!ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb)) && !p_blkno && create) {
ocfs2_error(inode->i_sb,
"Inode %llu has a hole at block %llu\n",
(unsigned long long)OCFS2_I(inode)->ip_blkno,
(unsigned long long)iblock);
ret = -EROFS;
goto bail;
}
/*
* get_more_blocks() expects us to describe a hole by clearing
* the mapped bit on bh_result().
*
* Consider an unwritten extent as a hole.
*/
if (p_blkno && !(ext_flags & OCFS2_EXT_UNWRITTEN))
map_bh(bh_result, inode->i_sb, p_blkno);
else {
/*
* ocfs2_prepare_inode_for_write() should have caught
* the case where we'd be filling a hole and triggered
* a buffered write instead.
*/
if (create) {
ret = -EIO;
mlog_errno(ret);
goto bail;
}
clear_buffer_mapped(bh_result);
}
/* make sure we don't map more than max_blocks blocks here as
that's all the kernel will handle at this point. */
if (max_blocks < contig_blocks)
contig_blocks = max_blocks;
bh_result->b_size = contig_blocks << blocksize_bits;
bail:
return ret;
}
static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
loff_t *ppos,
size_t count,
int appending,
int *direct_io)
{
int ret = 0, meta_level = appending;
struct inode *inode = dentry->d_inode;
u32 clusters;
loff_t newsize, saved_pos;
/*
* We sample i_size under a read level meta lock to see if our write
* is extending the file, if it is we back off and get a write level
* meta lock.
*/
for(;;) {
ret = ocfs2_meta_lock(inode, NULL, meta_level);
if (ret < 0) {
meta_level = -1;
mlog_errno(ret);
goto out;
}
/* 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
* proceeed, 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 (should_remove_suid(dentry)) {
if (meta_level == 0) {
ocfs2_meta_unlock(inode, meta_level);
meta_level = 1;
continue;
}
ret = ocfs2_write_remove_suid(inode);
if (ret < 0) {
mlog_errno(ret);
goto out_unlock;
}
}
/* work on a copy of ppos until we're sure that we won't have
* to recalculate it due to relocking. */
if (appending) {
saved_pos = i_size_read(inode);
mlog(0, "O_APPEND: inode->i_size=%llu\n", saved_pos);
} else {
saved_pos = *ppos;
}
if (ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb))) {
loff_t end = saved_pos + count;
/*
* Skip the O_DIRECT checks if we don't need
* them.
*/
if (!direct_io || !(*direct_io))
break;
/*
* Allowing concurrent direct writes means
* i_size changes wouldn't be synchronized, so
* one node could wind up truncating another
* nodes writes.
*/
if (end > i_size_read(inode)) {
*direct_io = 0;
break;
}
/*
* We don't fill holes during direct io, so
* check for them here. If any are found, the
* caller will have to retake some cluster
* locks and initiate the io as buffered.
*/
ret = ocfs2_check_range_for_holes(inode, saved_pos,
count);
if (ret == 1) {
*direct_io = 0;
ret = 0;
} else if (ret < 0)
mlog_errno(ret);
break;
}
/*
* The rest of this loop is concerned with legacy file
* systems which don't support sparse files.
*/
newsize = count + saved_pos;
mlog(0, "pos=%lld newsize=%lld cursize=%lld\n",
(long long) saved_pos, (long long) newsize,
(long long) i_size_read(inode));
/* No need for a higher level metadata lock if we're
* never going past i_size. */
if (newsize <= i_size_read(inode))
break;
if (meta_level == 0) {
ocfs2_meta_unlock(inode, meta_level);
meta_level = 1;
continue;
}
spin_lock(&OCFS2_I(inode)->ip_lock);
clusters = ocfs2_clusters_for_bytes(inode->i_sb, newsize) -
OCFS2_I(inode)->ip_clusters;
spin_unlock(&OCFS2_I(inode)->ip_lock);
mlog(0, "Writing at EOF, may need more allocation: "
"i_size = %lld, newsize = %lld, need %u clusters\n",
(long long) i_size_read(inode), (long long) newsize,
clusters);
/* We only want to continue the rest of this loop if
* our extend will actually require more
* allocation. */
if (!clusters)
break;
ret = ocfs2_extend_file(inode, NULL, newsize, count);
if (ret < 0) {
if (ret != -ENOSPC)
mlog_errno(ret);
goto out_unlock;
}
break;
}
if (appending)
*ppos = saved_pos;
out_unlock:
ocfs2_meta_unlock(inode, meta_level);
out:
return ret;
}
static int ocfs2_get_block(struct inode *inode, sector_t iblock,
struct buffer_head *bh_result, int create)
{
int err = 0;
unsigned int ext_flags;
u64 max_blocks = bh_result->b_size >> inode->i_blkbits;
u64 p_blkno, count, past_eof;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
mlog_entry("(0x%p, %llu, 0x%p, %d)\n", inode,
(unsigned long long)iblock, bh_result, create);
if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_SYSTEM_FILE)
mlog(ML_NOTICE, "get_block on system inode 0x%p (%lu)\n",
inode, inode->i_ino);
if (S_ISLNK(inode->i_mode)) {
/* this always does I/O for some reason. */
err = ocfs2_symlink_get_block(inode, iblock, bh_result, create);
goto bail;
}
err = ocfs2_extent_map_get_blocks(inode, iblock, &p_blkno, &count,
&ext_flags);
if (err) {
mlog(ML_ERROR, "Error %d from get_blocks(0x%p, %llu, 1, "
"%llu, NULL)\n", err, inode, (unsigned long long)iblock,
(unsigned long long)p_blkno);
goto bail;
}
if (max_blocks < count)
count = max_blocks;
/*
* ocfs2 never allocates in this function - the only time we
* need to use BH_New is when we're extending i_size on a file
* system which doesn't support holes, in which case BH_New
* allows block_prepare_write() to zero.
*
* If we see this on a sparse file system, then a truncate has
* raced us and removed the cluster. In this case, we clear
* the buffers dirty and uptodate bits and let the buffer code
* ignore it as a hole.
*/
if (create && p_blkno == 0 && ocfs2_sparse_alloc(osb)) {
clear_buffer_dirty(bh_result);
clear_buffer_uptodate(bh_result);
goto bail;
}
/* Treat the unwritten extent as a hole for zeroing purposes. */
if (p_blkno && !(ext_flags & OCFS2_EXT_UNWRITTEN))
map_bh(bh_result, inode->i_sb, p_blkno);
bh_result->b_size = count << inode->i_blkbits;
if (!ocfs2_sparse_alloc(osb)) {
if (p_blkno == 0) {
err = -EIO;
mlog(ML_ERROR,
"iblock = %llu p_blkno = %llu blkno=(%llu)\n",
(unsigned long long)iblock,
(unsigned long long)p_blkno,
(unsigned long long)OCFS2_I(inode)->ip_blkno);
mlog(ML_ERROR, "Size %llu, clusters %u\n", (unsigned long long)i_size_read(inode), OCFS2_I(inode)->ip_clusters);
dump_stack();
}
past_eof = ocfs2_blocks_for_bytes(inode->i_sb, i_size_read(inode));
mlog(0, "Inode %lu, past_eof = %llu\n", inode->i_ino,
(unsigned long long)past_eof);
if (create && (iblock >= past_eof))
set_buffer_new(bh_result);
}
bail:
if (err < 0)
err = -EIO;
mlog_exit(err);
return err;
}
