status = ocfs2_read_inode_block(inode, &bh); if (status < 0) {
mlog_errno(status); goto bail;
}
fe = (struct ocfs2_dinode *) bh->b_data;
if ((u64)iblock >= ocfs2_clusters_to_blocks(inode->i_sb,
le32_to_cpu(fe->i_clusters))) {
err = -ENOMEM;
mlog(ML_ERROR, "block offset is outside the allocated size: " "%llu\n", (unsignedlonglong)iblock); goto bail;
}
/* We don't use the page cache to create symlink data, so if
* need be, copy it over from the buffer cache. */ if (!buffer_uptodate(bh_result) && ocfs2_inode_is_new(inode)) {
u64 blkno = le64_to_cpu(fe->id2.i_list.l_recs[0].e_blkno) +
iblock;
buffer_cache_bh = sb_getblk(osb->sb, blkno); if (!buffer_cache_bh) {
err = -ENOMEM;
mlog(ML_ERROR, "couldn't getblock for symlink!\n"); goto bail;
}
/* we haven't locked out transactions, so a commit * could've happened. Since we've got a reference on * the bh, even if it commits while we're doing the
* copy, the data is still good. */ if (buffer_jbd(buffer_cache_bh) && ocfs2_inode_is_new(inode)) {
memcpy_to_folio(bh_result->b_folio,
bh_result->b_size * iblock,
buffer_cache_bh->b_data,
bh_result->b_size);
set_buffer_uptodate(bh_result);
}
brelse(buffer_cache_bh);
}
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;
}
/* * 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_write_begin() 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);
if (!(le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_DATA_FL)) {
ocfs2_error(inode->i_sb, "Inode %llu lost inline data flag\n",
(unsignedlonglong)OCFS2_I(inode)->ip_blkno); return -EROFS;
}
size = i_size_read(inode);
if (size > folio_size(folio) ||
size > ocfs2_max_inline_data_with_xattr(inode->i_sb, di)) {
ocfs2_error(inode->i_sb, "Inode %llu has with inline data has bad size: %Lu\n",
(unsignedlonglong)OCFS2_I(inode)->ip_blkno,
(unsignedlonglong)size); return -EROFS;
}
ret = ocfs2_inode_lock_with_folio(inode, NULL, 0, folio); if (ret != 0) { if (ret == AOP_TRUNCATED_PAGE)
unlock = 0;
mlog_errno(ret); goto out;
}
if (down_read_trylock(&oi->ip_alloc_sem) == 0) { /* * Unlock the folio and cycle ip_alloc_sem so that we don't * busyloop waiting for ip_alloc_sem to unlock
*/
ret = AOP_TRUNCATED_PAGE;
folio_unlock(folio);
unlock = 0;
down_read(&oi->ip_alloc_sem);
up_read(&oi->ip_alloc_sem); goto out_inode_unlock;
}
/* * i_size might have just been updated as we grabbed the meta lock. We * might now be discovering a truncate that hit on another node. * block_read_full_folio->get_block freaks out if it is asked to read * beyond the end of a file, so we check here. Callers * (generic_file_read, vm_ops->fault) are clever enough to check i_size * and notice that the folio they just read isn't needed. * * XXX sys_readahead() seems to get that wrong?
*/ if (start >= i_size_read(inode)) {
folio_zero_segment(folio, 0, folio_size(folio));
folio_mark_uptodate(folio);
ret = 0; goto out_alloc;
}
if (oi->ip_dyn_features & OCFS2_INLINE_DATA_FL)
ret = ocfs2_readpage_inline(inode, folio); else
ret = block_read_full_folio(folio, ocfs2_get_block);
unlock = 0;
/* * This is used only for read-ahead. Failures or difficult to handle * situations are safe to ignore. * * Right now, we don't bother with BH_Boundary - in-inode extent lists * are quite large (243 extents on 4k blocks), so most inodes don't * grow out to a tree. If need be, detecting boundary extents could * trivially be added in a future version of ocfs2_get_block().
*/ staticvoid ocfs2_readahead(struct readahead_control *rac)
{ int ret; struct inode *inode = rac->mapping->host; struct ocfs2_inode_info *oi = OCFS2_I(inode);
/* * Use the nonblocking flag for the dlm code to avoid page * lock inversion, but don't bother with retrying.
*/
ret = ocfs2_inode_lock_full(inode, NULL, 0, OCFS2_LOCK_NONBLOCK); if (ret) return;
if (down_read_trylock(&oi->ip_alloc_sem) == 0) goto out_unlock;
/* * Don't bother with inline-data. There isn't anything * to read-ahead in that case anyway...
*/ if (oi->ip_dyn_features & OCFS2_INLINE_DATA_FL) goto out_up;
/* * Check whether a remote node truncated this file - we just * drop out in that case as it's not worth handling here.
*/ if (readahead_pos(rac) >= i_size_read(inode)) goto out_up;
/* Note: Because we don't support holes, our allocation has * already happened (allocation writes zeros to the file data) * so we don't have to worry about ordered writes in * ocfs2_writepages. * * ->writepages is called during the process of invalidating the page cache * during blocked lock processing. It can't block on any cluster locks * to during block mapping. It's relying on the fact that the block * mapping can't have disappeared under the dirty pages that it is * being asked to write back.
*/ staticint ocfs2_writepages(struct address_space *mapping, struct writeback_control *wbc)
{ return mpage_writepages(mapping, wbc, ocfs2_get_block);
}
/* Taken from ext3. We don't necessarily need the full blown * functionality yet, but IMHO it's better to cut and paste the whole * thing so we can avoid introducing our own bugs (and easily pick up
* their fixes when they happen) --Mark */ int walk_page_buffers( handle_t *handle, struct buffer_head *head, unsigned from, unsigned to, int *partial, int (*fn)( handle_t *handle, struct buffer_head *bh))
{ struct buffer_head *bh; unsigned block_start, block_end; unsigned blocksize = head->b_size; int err, ret = 0; struct buffer_head *next;
for ( bh = head, block_start = 0;
ret == 0 && (bh != head || !block_start);
block_start = block_end, bh = next)
{
next = bh->b_this_page;
block_end = block_start + blocksize; if (block_end <= from || block_start >= to) { if (partial && !buffer_uptodate(bh))
*partial = 1; continue;
}
err = (*fn)(handle, bh); if (!ret)
ret = err;
} return ret;
}
/* * The swap code (ab-)uses ->bmap to get a block mapping and then * bypasseѕ the file system for actual I/O. We really can't allow * that on refcounted inodes, so we have to skip out here. And yes, * 0 is the magic code for a bmap error..
*/ if (ocfs2_is_refcount_inode(inode)) return 0;
/* We don't need to lock journal system files, since they aren't * accessed concurrently from multiple nodes.
*/ if (!INODE_JOURNAL(inode)) {
err = ocfs2_inode_lock(inode, NULL, 0); if (err) { if (err != -ENOENT)
mlog_errno(err); goto bail;
}
down_read(&OCFS2_I(inode)->ip_alloc_sem);
}
if (start)
*start = cluster_start; if (end)
*end = cluster_end;
}
/* * 'from' and 'to' are the region in the page to avoid zeroing. * * If pagesize > clustersize, this function will avoid zeroing outside * of the cluster boundary. * * from == to == 0 is code for "zero the entire cluster region"
*/ staticvoid ocfs2_clear_folio_regions(struct folio *folio, struct ocfs2_super *osb, u32 cpos, unsigned from, unsigned to)
{ void *kaddr; unsignedint cluster_start, cluster_end;
if (from || to) { if (from > cluster_start)
memset(kaddr + cluster_start, 0, from - cluster_start); if (to < cluster_end)
memset(kaddr + to, 0, cluster_end - to);
} else {
memset(kaddr + cluster_start, 0, cluster_end - cluster_start);
}
kunmap_local(kaddr);
}
/* * Nonsparse file systems fully allocate before we get to the write * code. This prevents ocfs2_write() from tagging the write as an * allocating one, which means ocfs2_map_folio_blocks() might try to * read-in the blocks at the tail of our file. Avoid reading them by * testing i_size against each block offset.
*/ staticint ocfs2_should_read_blk(struct inode *inode, struct folio *folio, unsignedint block_start)
{
u64 offset = folio_pos(folio) + block_start;
if (ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb))) return 1;
if (i_size_read(inode) > offset) return 1;
return 0;
}
/* * Some of this taken from __block_write_begin(). We already have our * mapping by now though, and the entire write will be allocating or * it won't, so not much need to use BH_New. * * This will also skip zeroing, which is handled externally.
*/ int ocfs2_map_folio_blocks(struct folio *folio, u64 *p_blkno, struct inode *inode, unsignedint from, unsignedint to, intnew)
{ int ret = 0; struct buffer_head *head, *bh, *wait[2], **wait_bh = wait; unsignedint block_end, block_start; unsignedint bsize = i_blocksize(inode);
head = folio_buffers(folio); if (!head)
head = create_empty_buffers(folio, bsize, 0);
/* * Ignore blocks outside of our i/o range - * they may belong to unallocated clusters.
*/ if (block_start >= to || block_end <= from) { if (folio_test_uptodate(folio))
set_buffer_uptodate(bh); continue;
}
/* * For an allocating write with cluster size >= page * size, we always write the entire page.
*/ if (new)
set_buffer_new(bh);
if (!buffer_mapped(bh)) {
map_bh(bh, inode->i_sb, *p_blkno);
clean_bdev_bh_alias(bh);
}
/* * If we issued read requests - let them complete.
*/ while(wait_bh > wait) {
wait_on_buffer(*--wait_bh); if (!buffer_uptodate(*wait_bh))
ret = -EIO;
}
if (ret == 0 || !new) return ret;
/* * If we get -EIO above, zero out any newly allocated blocks * to avoid exposing stale data.
*/
bh = head;
block_start = 0; do {
block_end = block_start + bsize; if (block_end <= from) goto next_bh; if (block_start >= to) break;
/* * Describe the state of a single cluster to be written to.
*/ struct ocfs2_write_cluster_desc {
u32 c_cpos;
u32 c_phys; /* * Give this a unique field because c_phys eventually gets * filled.
*/ unsigned c_new; unsigned c_clear_unwritten; unsigned c_needs_zero;
};
struct ocfs2_write_ctxt { /* Logical cluster position / len of write */
u32 w_cpos;
u32 w_clen;
/* First cluster allocated in a nonsparse extend */
u32 w_first_new_cpos;
/* Type of caller. Must be one of buffer, mmap, direct. */
ocfs2_write_type_t w_type;
/* * This is true if page_size > cluster_size. * * It triggers a set of special cases during write which might * have to deal with allocating writes to partial pages.
*/ unsignedint w_large_pages;
/* * Folios involved in this write. * * w_target_folio is the folio being written to by the user. * * w_folios is an array of folios which always contains * w_target_folio, and in the case of an allocating write with * page_size < cluster size, it will contain zero'd and mapped * pages adjacent to w_target_folio which need to be written * out in so that future reads from that region will get * zero's.
*/ unsignedint w_num_folios; struct folio *w_folios[OCFS2_MAX_CTXT_PAGES]; struct folio *w_target_folio;
/* * w_target_locked is used for page_mkwrite path indicating no unlocking * against w_target_folio in ocfs2_write_end_nolock.
*/ unsignedint w_target_locked:1;
/* * ocfs2_write_end() uses this to know what the real range to * write in the target should be.
*/ unsignedint w_target_from; unsignedint w_target_to;
/* * We could use journal_current_handle() but this is cleaner, * IMHO -Mark
*/
handle_t *w_handle;
void ocfs2_unlock_and_free_folios(struct folio **folios, int num_folios)
{ int i;
for(i = 0; i < num_folios; i++) { if (!folios[i]) continue;
folio_unlock(folios[i]);
folio_mark_accessed(folios[i]);
folio_put(folios[i]);
}
}
staticvoid ocfs2_unlock_folios(struct ocfs2_write_ctxt *wc)
{ int i;
/* * w_target_locked is only set to true in the page_mkwrite() case. * The intent is to allow us to lock the target page from write_begin() * to write_end(). The caller must hold a ref on w_target_folio.
*/ if (wc->w_target_locked) {
BUG_ON(!wc->w_target_folio); for (i = 0; i < wc->w_num_folios; i++) { if (wc->w_target_folio == wc->w_folios[i]) {
wc->w_folios[i] = NULL; break;
}
}
folio_mark_accessed(wc->w_target_folio);
folio_put(wc->w_target_folio);
}
ocfs2_unlock_and_free_folios(wc->w_folios, wc->w_num_folios);
}
/* * If a page has any new buffers, zero them out here, and mark them uptodate * and dirty so they'll be written out (in order to prevent uninitialised * block data from leaking). And clear the new bit.
*/ staticvoid ocfs2_zero_new_buffers(struct folio *folio, size_t from, size_t to)
{ unsignedint block_start, block_end; struct buffer_head *head, *bh;
BUG_ON(!folio_test_locked(folio));
head = folio_buffers(folio); if (!head) return;
/* * Only called when we have a failure during allocating write to write * zero's to the newly allocated region.
*/ staticvoid ocfs2_write_failure(struct inode *inode, struct ocfs2_write_ctxt *wc,
loff_t user_pos, unsigned user_len)
{ int i; unsigned from = user_pos & (PAGE_SIZE - 1),
to = user_pos + user_len;
if (wc->w_target_folio)
ocfs2_zero_new_buffers(wc->w_target_folio, from, to);
for (i = 0; i < wc->w_num_folios; i++) { struct folio *folio = wc->w_folios[i];
if (folio && folio_buffers(folio)) { if (ocfs2_should_order_data(inode))
ocfs2_jbd2_inode_add_write(wc->w_handle, inode,
user_pos, user_len);
/* treat the write as new if the a hole/lseek spanned across * the page boundary.
*/ new = new | ((i_size_read(inode) <= folio_pos(folio)) &&
(folio_pos(folio) <= user_pos));
if (new)
ret = ocfs2_map_folio_blocks(folio, p_blkno, inode,
cluster_start, cluster_end, new); else
ret = ocfs2_map_folio_blocks(folio, p_blkno, inode,
map_from, map_to, new); if (ret) {
mlog_errno(ret); goto out;
}
user_data_from = map_from;
user_data_to = map_to; if (new) {
map_from = cluster_start;
map_to = cluster_end;
}
} else { /* * If we haven't allocated the new folio yet, we * shouldn't be writing it out without copying user * data. This is likely a math error from the caller.
*/
BUG_ON(!new);
map_from = cluster_start;
map_to = cluster_end;
ret = ocfs2_map_folio_blocks(folio, p_blkno, inode,
cluster_start, cluster_end, new); if (ret) {
mlog_errno(ret); goto out;
}
}
/* * Parts of newly allocated folios need to be zero'd. * * Above, we have also rewritten 'to' and 'from' - as far as * the rest of the function is concerned, the entire cluster * range inside of a folio needs to be written. * * We can skip this if the folio is uptodate - it's already * been zero'd from being read in as a hole.
*/ if (new && !folio_test_uptodate(folio))
ocfs2_clear_folio_regions(folio, OCFS2_SB(inode->i_sb),
cpos, user_data_from, user_data_to);
flush_dcache_folio(folio);
out: return ret;
}
/* * This function will only grab one clusters worth of pages.
*/ staticint ocfs2_grab_folios_for_write(struct address_space *mapping, struct ocfs2_write_ctxt *wc, u32 cpos, loff_t user_pos, unsigned user_len, intnew, struct folio *mmap_folio)
{ int ret = 0, i; unsignedlong start, target_index, end_index, index; struct inode *inode = mapping->host;
loff_t last_byte;
target_index = user_pos >> PAGE_SHIFT;
/* * Figure out how many pages we'll be manipulating here. For * non allocating write, we just change the one * page. Otherwise, we'll need a whole clusters worth. If we're * writing past i_size, we only need enough pages to cover the * last page of the write.
*/ if (new) {
wc->w_num_folios = ocfs2_pages_per_cluster(inode->i_sb);
start = ocfs2_align_clusters_to_page_index(inode->i_sb, cpos); /* * We need the index *past* the last page we could possibly * touch. This is the page past the end of the write or * i_size, whichever is greater.
*/
last_byte = max(user_pos + user_len, i_size_read(inode));
BUG_ON(last_byte < 1);
end_index = ((last_byte - 1) >> PAGE_SHIFT) + 1; if ((start + wc->w_num_folios) > end_index)
wc->w_num_folios = end_index - start;
} else {
wc->w_num_folios = 1;
start = target_index;
}
end_index = (user_pos + user_len - 1) >> PAGE_SHIFT;
for(i = 0; i < wc->w_num_folios; i++) {
index = start + i;
if (index >= target_index && index <= end_index &&
wc->w_type == OCFS2_WRITE_MMAP) { /* * ocfs2_pagemkwrite() is a little different * and wants us to directly use the page * passed in.
*/
folio_lock(mmap_folio);
/* Exit and let the caller retry */ if (mmap_folio->mapping != mapping) {
WARN_ON(mmap_folio->mapping);
folio_unlock(mmap_folio);
ret = -EAGAIN; goto out;
}
if (index == target_index)
wc->w_target_folio = wc->w_folios[i];
}
out: if (ret)
wc->w_target_locked = false; return ret;
}
/* * Prepare a single cluster for write one cluster into the file.
*/ staticint ocfs2_write_cluster(struct address_space *mapping,
u32 *phys, unsignedintnew, unsignedint clear_unwritten, unsignedint should_zero, struct ocfs2_alloc_context *data_ac, struct ocfs2_alloc_context *meta_ac, struct ocfs2_write_ctxt *wc, u32 cpos,
loff_t user_pos, unsigned user_len)
{ int ret, i;
u64 p_blkno; struct inode *inode = mapping->host; struct ocfs2_extent_tree et; int bpc = ocfs2_clusters_to_blocks(inode->i_sb, 1);
if (new) {
u32 tmp_pos;
/* * This is safe to call with the page locks - it won't take * any additional semaphores or cluster locks.
*/
tmp_pos = cpos;
ret = ocfs2_add_inode_data(OCFS2_SB(inode->i_sb), inode,
&tmp_pos, 1, !clear_unwritten,
wc->w_di_bh, wc->w_handle,
data_ac, meta_ac, NULL); /* * This shouldn't happen because we must have already * calculated the correct meta data allocation required. The * internal tree allocation code should know how to increase * transaction credits itself. * * If need be, we could handle -EAGAIN for a * RESTART_TRANS here.
*/
mlog_bug_on_msg(ret == -EAGAIN, "Inode %llu: EAGAIN return during allocation.\n",
(unsignedlonglong)OCFS2_I(inode)->ip_blkno); if (ret < 0) {
mlog_errno(ret); goto out;
}
} elseif (clear_unwritten) {
ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode),
wc->w_di_bh);
ret = ocfs2_mark_extent_written(inode, &et,
wc->w_handle, cpos, 1, *phys,
meta_ac, &wc->w_dealloc); if (ret < 0) {
mlog_errno(ret); goto out;
}
}
/* * The only reason this should fail is due to an inability to * find the extent added.
*/
ret = ocfs2_get_clusters(inode, cpos, phys, NULL, NULL); if (ret < 0) {
mlog(ML_ERROR, "Get physical blkno failed for inode %llu, " "at logical cluster %u",
(unsignedlonglong)OCFS2_I(inode)->ip_blkno, cpos); goto out;
}
for (i = 0; i < wc->w_clen; i++) {
desc = &wc->w_desc[i];
/* * We have to make sure that the total write passed in * doesn't extend past a single cluster.
*/
local_len = len;
cluster_off = pos & (osb->s_clustersize - 1); if ((cluster_off + local_len) > osb->s_clustersize)
local_len = osb->s_clustersize - cluster_off;
ret = ocfs2_write_cluster(mapping, &desc->c_phys,
desc->c_new,
desc->c_clear_unwritten,
desc->c_needs_zero,
data_ac, meta_ac,
wc, desc->c_cpos, pos, local_len); if (ret) {
mlog_errno(ret); goto out;
}
len -= local_len;
pos += local_len;
}
ret = 0;
out: return ret;
}
/* * ocfs2_write_end() wants to know which parts of the target page it * should complete the write on. It's easiest to compute them ahead of * time when a more complete view of the write is available.
*/ staticvoid ocfs2_set_target_boundaries(struct ocfs2_super *osb, struct ocfs2_write_ctxt *wc,
loff_t pos, unsigned len, int alloc)
{ struct ocfs2_write_cluster_desc *desc;
/* * Allocating write - we may have different boundaries based * on page size and cluster size. * * NOTE: We can no longer compute one value from the other as * the actual write length and user provided length may be * different.
*/
if (wc->w_large_pages) { /* * We only care about the 1st and last cluster within * our range and whether they should be zero'd or not. Either * value may be extended out to the start/end of a * newly allocated cluster.
*/
desc = &wc->w_desc[0]; if (desc->c_needs_zero)
ocfs2_figure_cluster_boundaries(osb,
desc->c_cpos,
&wc->w_target_from,
NULL);
/* * Check if this extent is marked UNWRITTEN by direct io. If so, we need not to * do the zero work. And should not to clear UNWRITTEN since it will be cleared * by the direct io procedure. * If this is a new extent that allocated by direct io, we should mark it in * the ip_unwritten_list.
*/ staticint ocfs2_unwritten_check(struct inode *inode, struct ocfs2_write_ctxt *wc, struct ocfs2_write_cluster_desc *desc)
{ struct ocfs2_inode_info *oi = OCFS2_I(inode); struct ocfs2_unwritten_extent *ue = NULL, *new = NULL; int ret = 0;
if (!desc->c_needs_zero) return 0;
retry:
spin_lock(&oi->ip_lock); /* Needs not to zero no metter buffer or direct. The one who is zero * the cluster is doing zero. And he will clear unwritten after all
* cluster io finished. */
list_for_each_entry(ue, &oi->ip_unwritten_list, ue_ip_node) { if (desc->c_cpos == ue->ue_cpos) {
BUG_ON(desc->c_new);
desc->c_needs_zero = 0;
desc->c_clear_unwritten = 0; goto unlock;
}
}
if (wc->w_type != OCFS2_WRITE_DIRECT) goto unlock;
if (new == NULL) {
spin_unlock(&oi->ip_lock); new = kmalloc(sizeof(struct ocfs2_unwritten_extent),
GFP_NOFS); if (new == NULL) {
ret = -ENOMEM; goto out;
} goto retry;
} /* This direct write will doing zero. */
new->ue_cpos = desc->c_cpos;
new->ue_phys = desc->c_phys;
desc->c_clear_unwritten = 0;
list_add_tail(&new->ue_ip_node, &oi->ip_unwritten_list);
list_add_tail(&new->ue_node, &wc->w_unwritten_list);
wc->w_unwritten_count++; new = NULL;
unlock:
spin_unlock(&oi->ip_lock);
out:
kfree(new); return ret;
}
/* * Populate each single-cluster write descriptor in the write context * with information about the i/o to be done. * * Returns the number of clusters that will have to be allocated, as * well as a worst case estimate of the number of extent records that * would have to be created during a write to an unwritten region.
*/ staticint ocfs2_populate_write_desc(struct inode *inode, struct ocfs2_write_ctxt *wc, unsignedint *clusters_to_alloc, unsignedint *extents_to_split)
{ int ret; struct ocfs2_write_cluster_desc *desc; unsignedint num_clusters = 0; unsignedint ext_flags = 0;
u32 phys = 0; int i;
*clusters_to_alloc = 0;
*extents_to_split = 0;
for (i = 0; i < wc->w_clen; i++) {
desc = &wc->w_desc[i];
desc->c_cpos = wc->w_cpos + i;
if (num_clusters == 0) { /* * Need to look up the next extent record.
*/
ret = ocfs2_get_clusters(inode, desc->c_cpos, &phys,
&num_clusters, &ext_flags); if (ret) {
mlog_errno(ret); goto out;
}
/* We should already CoW the refcountd extent. */
BUG_ON(ext_flags & OCFS2_EXT_REFCOUNTED);
/* * Assume worst case - that we're writing in * the middle of the extent. * * We can assume that the write proceeds from * left to right, in which case the extent * insert code is smart enough to coalesce the * next splits into the previous records created.
*/ if (ext_flags & OCFS2_EXT_UNWRITTEN)
*extents_to_split = *extents_to_split + 2;
} elseif (phys) { /* * Only increment phys if it doesn't describe * a hole.
*/
phys++;
}
/* * If w_first_new_cpos is < UINT_MAX, we have a non-sparse * file that got extended. w_first_new_cpos tells us * where the newly allocated clusters are so we can * zero them.
*/ if (desc->c_cpos >= wc->w_first_new_cpos) {
BUG_ON(phys == 0);
desc->c_needs_zero = 1;
}
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) {
ret = PTR_ERR(handle);
mlog_errno(ret); goto out;
}
folio = __filemap_get_folio(mapping, 0,
FGP_LOCK | FGP_ACCESSED | FGP_CREAT, GFP_NOFS); if (IS_ERR(folio)) {
ocfs2_commit_trans(osb, handle);
ret = PTR_ERR(folio);
mlog_errno(ret); goto out;
} /* * If we don't set w_num_folios then this folio won't get unlocked * and freed on cleanup of the write context.
*/
wc->w_target_folio = folio;
wc->w_folios[0] = folio;
wc->w_num_folios = 1;
ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), wc->w_di_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret) {
ocfs2_commit_trans(osb, handle);
mlog_errno(ret); goto out;
}
if (!(OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL))
ocfs2_set_inode_data_inline(inode, di);
if (!folio_test_uptodate(folio)) {
ret = ocfs2_read_inline_data(inode, folio, wc->w_di_bh); if (ret) {
ocfs2_commit_trans(osb, handle);
/* * Handle inodes which already have inline data 1st.
*/ if (oi->ip_dyn_features & OCFS2_INLINE_DATA_FL) { if (mmap_folio == NULL &&
ocfs2_size_fits_inline_data(wc->w_di_bh, end)) goto do_inline_write;
/* * The write won't fit - we have to give this inode an * inline extent list now.
*/
ret = ocfs2_convert_inline_data_to_extents(inode, wc->w_di_bh); if (ret)
mlog_errno(ret); goto out;
}
/* * Check whether the inode can accept inline data.
*/ if (oi->ip_clusters != 0 || i_size_read(inode) != 0) return 0;
/* * Check whether the write can fit.
*/
di = (struct ocfs2_dinode *)wc->w_di_bh->b_data; if (mmap_folio ||
end > ocfs2_max_inline_data_with_xattr(inode->i_sb, di)) return 0;
do_inline_write:
ret = ocfs2_write_begin_inline(mapping, inode, wc); if (ret) {
mlog_errno(ret); goto out;
}
/* * This signals to the caller that the data can be written * inline.
*/
written = 1;
out: return written ? written : ret;
}
/* * This function only does anything for file systems which can't * handle sparse files. * * What we want to do here is fill in any hole between the current end * of allocation and the end of our write. That way the rest of the * write path can treat it as an non-allocating write, which has no * special case code for sparse/nonsparse files.
*/ staticint ocfs2_expand_nonsparse_inode(struct inode *inode, struct buffer_head *di_bh,
loff_t pos, unsigned len, struct ocfs2_write_ctxt *wc)
{ int ret;
loff_t newsize = pos + len;
try_again:
ret = ocfs2_alloc_write_ctxt(&wc, osb, pos, len, type, di_bh); if (ret) {
mlog_errno(ret); return ret;
}
if (ocfs2_supports_inline_data(osb)) {
ret = ocfs2_try_to_write_inline_data(mapping, inode, pos, len,
mmap_folio, wc); if (ret == 1) {
ret = 0; goto success;
} if (ret < 0) {
mlog_errno(ret); goto out;
}
}
/* Direct io change i_size late, should not zero tail here. */ if (type != OCFS2_WRITE_DIRECT) { if (ocfs2_sparse_alloc(osb))
ret = ocfs2_zero_tail(inode, di_bh, pos); else
ret = ocfs2_expand_nonsparse_inode(inode, di_bh, pos,
len, wc); if (ret) {
mlog_errno(ret); goto out;
}
}
ret = ocfs2_check_range_for_refcount(inode, pos, len); if (ret < 0) {
mlog_errno(ret); goto out;
} elseif (ret == 1) {
clusters_need = wc->w_clen;
ret = ocfs2_refcount_cow(inode, di_bh,
wc->w_cpos, wc->w_clen, UINT_MAX); if (ret) {
mlog_errno(ret); goto out;
}
}
ret = ocfs2_populate_write_desc(inode, wc, &clusters_to_alloc,
&extents_to_split); if (ret) {
mlog_errno(ret); goto out;
}
clusters_need += clusters_to_alloc;
/* * We set w_target_from, w_target_to here so that * ocfs2_write_end() knows which range in the target page to * write out. An allocation requires that we write the entire * cluster range.
*/ if (clusters_to_alloc || extents_to_split) { /* * XXX: We are stretching the limits of * ocfs2_lock_allocators(). It greatly over-estimates * the work to be done.
*/
ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode),
wc->w_di_bh);
ret = ocfs2_lock_allocators(inode, &et,
clusters_to_alloc, extents_to_split,
&data_ac, &meta_ac); if (ret) {
mlog_errno(ret); goto out;
}
if (data_ac)
data_ac->ac_resv = &OCFS2_I(inode)->ip_la_data_resv;
credits = ocfs2_calc_extend_credits(inode->i_sb,
&di->id2.i_list);
} elseif (type == OCFS2_WRITE_DIRECT) /* direct write needs not to start trans if no extents alloc. */ goto success;
/* * We have to zero sparse allocated clusters, unwritten extent clusters, * and non-sparse clusters we just extended. For non-sparse writes, * we know zeros will only be needed in the first and/or last cluster.
*/ if (wc->w_clen && (wc->w_desc[0].c_needs_zero ||
wc->w_desc[wc->w_clen - 1].c_needs_zero))
cluster_of_pages = 1; else
cluster_of_pages = 0;
handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) {
ret = PTR_ERR(handle);
mlog_errno(ret); goto out;
}
wc->w_handle = handle;
if (clusters_to_alloc) {
ret = dquot_alloc_space_nodirty(inode,
ocfs2_clusters_to_bytes(osb->sb, clusters_to_alloc)); if (ret) goto out_commit;
}
ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), wc->w_di_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret) {
mlog_errno(ret); goto out_quota;
}
/* * Fill our folio array first. That way we've grabbed enough so * that we can zero and flush if we error after adding the * extent.
*/
ret = ocfs2_grab_folios_for_write(mapping, wc, wc->w_cpos, pos, len,
cluster_of_pages, mmap_folio); if (ret) { /* * ocfs2_grab_folios_for_write() returns -EAGAIN if it * could not lock the target folio. In this case, we exit * with no error and no target folio. This will trigger * the caller, page_mkwrite(), to re-try the operation.
*/ if (type == OCFS2_WRITE_MMAP && ret == -EAGAIN) {
BUG_ON(wc->w_target_folio);
ret = 0; goto out_quota;
}
mlog_errno(ret); goto out_quota;
}
ret = ocfs2_write_cluster_by_desc(mapping, data_ac, meta_ac, wc, pos,
len); if (ret) {
mlog_errno(ret); goto out_quota;
}
if (data_ac)
ocfs2_free_alloc_context(data_ac); if (meta_ac)
ocfs2_free_alloc_context(meta_ac);
success: if (foliop)
*foliop = wc->w_target_folio;
*fsdata = wc; return 0;
out_quota: if (clusters_to_alloc)
dquot_free_space(inode,
ocfs2_clusters_to_bytes(osb->sb, clusters_to_alloc));
out_commit:
ocfs2_commit_trans(osb, handle);
out: /* * The mmapped page won't be unlocked in ocfs2_free_write_ctxt(), * even in case of error here like ENOSPC and ENOMEM. So, we need * to unlock the target page manually to prevent deadlocks when * retrying again on ENOSPC, or when returning non-VM_FAULT_LOCKED * to VM code.
*/ if (wc->w_target_locked)
folio_unlock(mmap_folio);
ocfs2_free_write_ctxt(inode, wc);
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 (ret == -ENOSPC && try_free) { /* * Try to free some truncate log so that we can have enough * clusters to allocate.
*/
try_free = 0;
ret1 = ocfs2_try_to_free_truncate_log(osb, clusters_need); if (ret1 == 1) goto try_again;
ret = ocfs2_inode_lock(inode, &di_bh, 1); if (ret) {
mlog_errno(ret); return ret;
}
/* * Take alloc sem here to prevent concurrent lookups. That way * the mapping, zeroing and tree manipulation within * ocfs2_write() will be safe against ->read_folio(). This * should also serve to lock out allocation from a shared * writeable region.
*/
down_write(&OCFS2_I(inode)->ip_alloc_sem);
ret = ocfs2_write_begin_nolock(mapping, pos, len, OCFS2_WRITE_BUFFER,
foliop, fsdata, di_bh, NULL); if (ret) {
mlog_errno(ret); goto out_fail;
}
if (unlikely(copied < len) && wc->w_target_folio) {
loff_t new_isize;
if (!folio_test_uptodate(wc->w_target_folio))
copied = 0;
new_isize = max_t(loff_t, i_size_read(inode), pos + copied); if (new_isize > folio_pos(wc->w_target_folio))
ocfs2_zero_new_buffers(wc->w_target_folio, start+copied,
start+len); else { /* * When folio is fully beyond new isize (data copy * failed), do not bother zeroing the folio. Invalidate * it instead so that writeback does not get confused * put page & buffer dirty bits into inconsistent * state.
*/
block_invalidate_folio(wc->w_target_folio, 0,
folio_size(wc->w_target_folio));
}
} if (wc->w_target_folio)
flush_dcache_folio(wc->w_target_folio);
for (i = 0; i < wc->w_num_folios; i++) { struct folio *folio = wc->w_folios[i];
/* This is the direct io target folio */ if (folio == NULL) continue;
if (folio == wc->w_target_folio) {
from = wc->w_target_from;
to = wc->w_target_to;
BUG_ON(from > folio_size(folio) ||
to > folio_size(folio) ||
to < from);
} else { /* * Pages adjacent to the target (if any) imply * a hole-filling write in which case we want * to flush their entire range.
*/
from = 0;
to = folio_size(folio);
}
if (folio_buffers(folio)) { if (handle && ocfs2_should_order_data(inode)) {
loff_t start_byte = folio_pos(folio) + from;
loff_t length = to - from;
ocfs2_jbd2_inode_add_write(handle, inode,
start_byte, length);
}
block_commit_write(folio, from, to);
}
}
out_write_size: /* Direct io do not update i_size here. */ if (wc->w_type != OCFS2_WRITE_DIRECT) {
pos += copied; if (pos > i_size_read(inode)) {
i_size_write(inode, pos);
mark_inode_dirty(inode);
}
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_mtime_nsec = di->i_ctime_nsec = cpu_to_le32(inode_get_mtime_nsec(inode)); if (handle)
ocfs2_update_inode_fsync_trans(handle, inode, 1);
} if (handle)
ocfs2_journal_dirty(handle, wc->w_di_bh);
out: /* unlock pages before dealloc since it needs acquiring j_trans_barrier * lock, or it will cause a deadlock since journal commit threads holds * this lock and will ask for the page lock when flushing the data. * put it here to preserve the unlock order.
*/
ocfs2_unlock_folios(wc);
/* * 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);
*/ staticint ocfs2_dio_wr_get_block(struct inode *inode, sector_t iblock, struct buffer_head *bh_result, int create)
{ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); struct ocfs2_inode_info *oi = OCFS2_I(inode); struct ocfs2_write_ctxt *wc; struct ocfs2_write_cluster_desc *desc = NULL; struct ocfs2_dio_write_ctxt *dwc = NULL; struct buffer_head *di_bh = NULL;
u64 p_blkno; unsignedint i_blkbits = inode->i_sb->s_blocksize_bits;
loff_t pos = iblock << i_blkbits;
sector_t endblk = (i_size_read(inode) - 1) >> i_blkbits; unsigned len, total_len = bh_result->b_size; int ret = 0, first_get_block = 0;
len = osb->s_clustersize - (pos & (osb->s_clustersize - 1));
len = min(total_len, len);
/* * bh_result->b_size is count in get_more_blocks according to write * "pos" and "end", we need map twice to return different buffer state: * 1. area in file size, not set NEW; * 2. area out file size, set NEW. * * iblock endblk * |--------|---------|---------|--------- * |<-------area in file------->|
*/
mlog(0, "get block of %lu at %llu:%u req %u\n",
inode->i_ino, pos, len, total_len);
/* * Because we need to change file size in ocfs2_dio_end_io_write(), or * we may need to add it to orphan dir. So can not fall to fast path * while file size will be changed.
*/ if (pos + total_len <= i_size_read(inode)) {
/* This is the fast path for re-write. */
ret = ocfs2_lock_get_block(inode, iblock, bh_result, create); if (buffer_mapped(bh_result) &&
!buffer_new(bh_result) &&
ret == 0) goto out;
/* Clear state set by ocfs2_get_block. */
bh_result->b_state = 0;
}
dwc = ocfs2_dio_alloc_write_ctx(bh_result, &first_get_block); if (unlikely(dwc == NULL)) {
ret = -ENOMEM;
mlog_errno(ret); goto out;
}
if (ocfs2_clusters_for_bytes(inode->i_sb, pos + total_len) >
ocfs2_clusters_for_bytes(inode->i_sb, i_size_read(inode)) &&
!dwc->dw_orphaned) { /* * when we are going to alloc extents beyond file size, add the * inode to orphan dir, so we can recall those spaces when * system crashed during write.
*/
ret = ocfs2_add_inode_to_orphan(osb, inode); if (ret < 0) {
mlog_errno(ret); goto out;
}
dwc->dw_orphaned = 1;
}
ret = ocfs2_inode_lock(inode, &di_bh, 1); if (ret) {
mlog_errno(ret); goto out;
}
down_write(&oi->ip_alloc_sem);
if (first_get_block) { if (ocfs2_sparse_alloc(osb))
ret = ocfs2_zero_tail(inode, di_bh, pos); else
ret = ocfs2_expand_nonsparse_inode(inode, di_bh, pos,
total_len, NULL); if (ret < 0) {
mlog_errno(ret); goto unlock;
}
}
ret = ocfs2_write_begin_nolock(inode->i_mapping, pos, len,
OCFS2_WRITE_DIRECT, NULL,
(void **)&wc, di_bh, NULL); if (ret) {
mlog_errno(ret); goto unlock;
}
map_bh(bh_result, inode->i_sb, p_blkno);
bh_result->b_size = len; if (desc->c_needs_zero)
set_buffer_new(bh_result);
if (iblock > endblk)
set_buffer_new(bh_result);
/* May sleep in end_io. It should not happen in a irq context. So defer
* it to dio work queue. */
set_buffer_defer_completion(bh_result);
if (!list_empty(&wc->w_unwritten_list)) { struct ocfs2_unwritten_extent *ue = NULL;
ue = list_first_entry(&wc->w_unwritten_list, struct ocfs2_unwritten_extent,
ue_node);
BUG_ON(ue->ue_cpos != desc->c_cpos); /* The physical address may be 0, fill it. */
ue->ue_phys = desc->c_phys;
/* We do clear unwritten, delete orphan, change i_size here. If neither
* of these happen, we can skip all this. */ if (list_empty(&dwc->dw_zero_list) &&
end <= i_size_read(inode) &&
!dwc->dw_orphaned) goto out;
ret = ocfs2_inode_lock(inode, &di_bh, 1); if (ret < 0) {
mlog_errno(ret); goto out;
}
down_write(&oi->ip_alloc_sem);
/* Delete orphan before acquire i_rwsem. */ if (dwc->dw_orphaned) {
BUG_ON(dwc->dw_writer_pid != task_pid_nr(current));
end = end > i_size_read(inode) ? end : 0;
ret = ocfs2_del_inode_from_orphan(osb, inode, di_bh,
!!end, end); if (ret < 0)
mlog_errno(ret);
}
/* Attach dealloc with extent tree in case that we may reuse extents * which are already unlinked from current extent tree due to extent * rotation and merging.
*/
et.et_dealloc = &dealloc;
ret = ocfs2_lock_allocators(inode, &et, 0, dwc->dw_zero_count*2,
&data_ac, &meta_ac); if (ret) {
mlog_errno(ret); goto unlock;
}
handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) {
ret = PTR_ERR(handle);
mlog_errno(ret); goto unlock;
}
ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
OCFS2_JOURNAL_ACCESS_WRITE); if (ret) {
mlog_errno(ret); goto commit;
}
list_for_each_entry(ue, &dwc->dw_zero_list, ue_node) {
ret = ocfs2_assure_trans_credits(handle, credits); if (ret < 0) {
mlog_errno(ret); break;
}
ret = ocfs2_mark_extent_written(inode, &et, handle,
ue->ue_cpos, 1,
ue->ue_phys,
meta_ac, &dealloc); if (ret < 0) {
mlog_errno(ret); break;
}
}
if (end > i_size_read(inode)) {
ret = ocfs2_set_inode_size(handle, inode, di_bh, end); if (ret < 0)
mlog_errno(ret);
}
commit:
ocfs2_commit_trans(osb, handle);
unlock:
up_write(&oi->ip_alloc_sem);
ocfs2_inode_unlock(inode, 1);
brelse(di_bh);
out: if (data_ac)
ocfs2_free_alloc_context(data_ac); if (meta_ac)
ocfs2_free_alloc_context(meta_ac);
ocfs2_run_deallocs(osb, &dealloc);
ocfs2_dio_free_write_ctx(inode, dwc);
return ret;
}
/* * ocfs2_dio_end_io is called by the dio core when a dio is finished. We're * particularly interested in the aio/dio case. We use the rw_lock DLM lock * to protect io on one node from truncation on another.
*/ staticint ocfs2_dio_end_io(struct kiocb *iocb,
loff_t offset,
ssize_t bytes, void *private)
{ struct inode *inode = file_inode(iocb->ki_filp); int level; int ret = 0;
/* this io's submitter should not have unlocked this before we could */
BUG_ON(!ocfs2_iocb_is_rw_locked(iocb));
if (bytes <= 0)
mlog_ratelimited(ML_ERROR, "Direct IO failed, bytes = %lld",
(longlong)bytes); if (private) { if (bytes > 0)
ret = ocfs2_dio_end_io_write(inode, private, offset,
bytes); else
ocfs2_dio_free_write_ctx(inode, private);
}
/* * Fallback to buffered I/O if we see an inode without * extents.
*/ if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) return 0;
/* Fallback to buffered I/O if we do not support append dio. */ if (iocb->ki_pos + iter->count > i_size_read(inode) &&
!ocfs2_supports_append_dio(osb)) return 0;
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.
