#include <asm/byteorder.h>
#include <linux/swap.h>
#include <linux/pipe_fs_i.h>
+#include <linux/mpage.h>
+#include <linux/quotaops.h>
#define MLOG_MASK_PREFIX ML_FILE_IO
#include <cluster/masklog.h>
#include "suballoc.h"
#include "super.h"
#include "symlink.h"
+#include "refcounttree.h"
#include "buffer_head_io.h"
goto bail;
}
- status = ocfs2_read_block(OCFS2_SB(inode->i_sb),
- OCFS2_I(inode)->ip_blkno,
- &bh, OCFS2_BH_CACHED, inode);
+ status = ocfs2_read_inode_block(inode, &bh);
if (status < 0) {
mlog_errno(status);
goto bail;
}
fe = (struct ocfs2_dinode *) bh->b_data;
- if (!OCFS2_IS_VALID_DINODE(fe)) {
- mlog(ML_ERROR, "Invalid dinode #%llu: signature = %.*s\n",
- (unsigned long long)le64_to_cpu(fe->i_blkno), 7,
- fe->i_signature);
- goto bail;
- }
-
if ((u64)iblock >= ocfs2_clusters_to_blocks(inode->i_sb,
le32_to_cpu(fe->i_clusters))) {
mlog(ML_ERROR, "block offset is outside the allocated size: "
err = 0;
bail:
- if (bh)
- brelse(bh);
+ brelse(bh);
mlog_exit(err);
return err;
}
-static int ocfs2_get_block(struct inode *inode, sector_t iblock,
- struct buffer_head *bh_result, int create)
+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 p_blkno, past_eof;
+ 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,
goto bail;
}
- err = ocfs2_extent_map_get_blocks(inode, iblock, &p_blkno, NULL,
+ 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, "
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.
*/
- mlog_bug_on_msg(create && p_blkno == 0 && ocfs2_sparse_alloc(osb),
- "ino %lu, iblock %llu\n", inode->i_ino,
- (unsigned long long)iblock);
+ 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;
(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();
+ goto bail;
}
past_eof = ocfs2_blocks_for_bytes(inode->i_sb, i_size_read(inode));
return err;
}
+int ocfs2_read_inline_data(struct inode *inode, struct page *page,
+ struct buffer_head *di_bh)
+{
+ void *kaddr;
+ loff_t size;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+
+ if (!(le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_DATA_FL)) {
+ ocfs2_error(inode->i_sb, "Inode %llu lost inline data flag",
+ (unsigned long long)OCFS2_I(inode)->ip_blkno);
+ return -EROFS;
+ }
+
+ size = i_size_read(inode);
+
+ if (size > PAGE_CACHE_SIZE ||
+ 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",
+ (unsigned long long)OCFS2_I(inode)->ip_blkno,
+ (unsigned long long)size);
+ return -EROFS;
+ }
+
+ kaddr = kmap_atomic(page, KM_USER0);
+ if (size)
+ memcpy(kaddr, di->id2.i_data.id_data, size);
+ /* Clear the remaining part of the page */
+ memset(kaddr + size, 0, PAGE_CACHE_SIZE - size);
+ flush_dcache_page(page);
+ kunmap_atomic(kaddr, KM_USER0);
+
+ SetPageUptodate(page);
+
+ return 0;
+}
+
+static int ocfs2_readpage_inline(struct inode *inode, struct page *page)
+{
+ int ret;
+ struct buffer_head *di_bh = NULL;
+
+ BUG_ON(!PageLocked(page));
+ BUG_ON(!(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;
+ }
+
+ ret = ocfs2_read_inline_data(inode, page, di_bh);
+out:
+ unlock_page(page);
+
+ brelse(di_bh);
+ return ret;
+}
+
static int ocfs2_readpage(struct file *file, struct page *page)
{
struct inode *inode = page->mapping->host;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
loff_t start = (loff_t)page->index << PAGE_CACHE_SHIFT;
int ret, unlock = 1;
mlog_entry("(0x%p, %lu)\n", file, (page ? page->index : 0));
- ret = ocfs2_meta_lock_with_page(inode, NULL, 0, page);
+ ret = ocfs2_inode_lock_with_page(inode, NULL, 0, page);
if (ret != 0) {
if (ret == AOP_TRUNCATED_PAGE)
unlock = 0;
goto out;
}
- if (down_read_trylock(&OCFS2_I(inode)->ip_alloc_sem) == 0) {
+ if (down_read_trylock(&oi->ip_alloc_sem) == 0) {
ret = AOP_TRUNCATED_PAGE;
- goto out_meta_unlock;
+ goto out_inode_unlock;
}
/*
* XXX sys_readahead() seems to get that wrong?
*/
if (start >= i_size_read(inode)) {
- zero_user_page(page, 0, PAGE_SIZE, KM_USER0);
+ zero_user(page, 0, PAGE_SIZE);
SetPageUptodate(page);
ret = 0;
goto out_alloc;
}
- ret = ocfs2_data_lock_with_page(inode, 0, page);
- if (ret != 0) {
- if (ret == AOP_TRUNCATED_PAGE)
- unlock = 0;
- mlog_errno(ret);
- goto out_alloc;
- }
-
- ret = block_read_full_page(page, ocfs2_get_block);
+ if (oi->ip_dyn_features & OCFS2_INLINE_DATA_FL)
+ ret = ocfs2_readpage_inline(inode, page);
+ else
+ ret = block_read_full_page(page, ocfs2_get_block);
unlock = 0;
- ocfs2_data_unlock(inode, 0);
out_alloc:
up_read(&OCFS2_I(inode)->ip_alloc_sem);
-out_meta_unlock:
- ocfs2_meta_unlock(inode, 0);
+out_inode_unlock:
+ ocfs2_inode_unlock(inode, 0);
out:
if (unlock)
unlock_page(page);
return ret;
}
+/*
+ * 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().
+ */
+static int ocfs2_readpages(struct file *filp, struct address_space *mapping,
+ struct list_head *pages, unsigned nr_pages)
+{
+ int ret, err = -EIO;
+ struct inode *inode = mapping->host;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ loff_t start;
+ struct page *last;
+
+ /*
+ * 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 err;
+
+ if (down_read_trylock(&oi->ip_alloc_sem) == 0) {
+ ocfs2_inode_unlock(inode, 0);
+ return err;
+ }
+
+ /*
+ * 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_unlock;
+
+ /*
+ * Check whether a remote node truncated this file - we just
+ * drop out in that case as it's not worth handling here.
+ */
+ last = list_entry(pages->prev, struct page, lru);
+ start = (loff_t)last->index << PAGE_CACHE_SHIFT;
+ if (start >= i_size_read(inode))
+ goto out_unlock;
+
+ err = mpage_readpages(mapping, pages, nr_pages, ocfs2_get_block);
+
+out_unlock:
+ up_read(&oi->ip_alloc_sem);
+ ocfs2_inode_unlock(inode, 0);
+
+ return err;
+}
+
/* 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
{
int ret;
- down_read(&OCFS2_I(inode)->ip_alloc_sem);
-
ret = block_prepare_write(page, from, to, ocfs2_get_block);
- up_read(&OCFS2_I(inode)->ip_alloc_sem);
-
return ret;
}
unsigned to)
{
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
- handle_t *handle = NULL;
+ handle_t *handle;
int ret = 0;
handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
- if (!handle) {
+ if (IS_ERR(handle)) {
ret = -ENOMEM;
mlog_errno(ret);
goto out;
}
if (ocfs2_should_order_data(inode)) {
- ret = walk_page_buffers(handle,
- page_buffers(page),
- from, to, NULL,
- ocfs2_journal_dirty_data);
- if (ret < 0)
+ ret = ocfs2_jbd2_file_inode(handle, inode);
+ if (ret < 0)
mlog_errno(ret);
}
out:
if (ret) {
- if (handle)
+ if (!IS_ERR(handle))
ocfs2_commit_trans(osb, handle);
handle = ERR_PTR(ret);
}
* accessed concurrently from multiple nodes.
*/
if (!INODE_JOURNAL(inode)) {
- err = ocfs2_meta_lock(inode, NULL, 0);
+ err = ocfs2_inode_lock(inode, NULL, 0);
if (err) {
if (err != -ENOENT)
mlog_errno(err);
down_read(&OCFS2_I(inode)->ip_alloc_sem);
}
- err = ocfs2_extent_map_get_blocks(inode, block, &p_blkno, NULL, NULL);
+ if (!(OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL))
+ err = ocfs2_extent_map_get_blocks(inode, block, &p_blkno, NULL,
+ NULL);
if (!INODE_JOURNAL(inode)) {
up_read(&OCFS2_I(inode)->ip_alloc_sem);
- ocfs2_meta_unlock(inode, 0);
+ ocfs2_inode_unlock(inode, 0);
}
if (err) {
goto bail;
}
-
bail:
status = err ? 0 : p_blkno;
*
* called like this: dio->get_blocks(dio->inode, fs_startblk,
* fs_count, map_bh, dio->rw == WRITE);
+ *
+ * Note that we never bother to allocate blocks here, and thus ignore the
+ * create argument.
*/
static int ocfs2_direct_IO_get_blocks(struct inode *inode, sector_t iblock,
struct buffer_head *bh_result, int create)
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,
goto bail;
}
- if (!ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb)) && !p_blkno) {
- 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;
- }
+ /* We should already CoW the refcounted extent in case of create. */
+ BUG_ON(create && (ext_flags & OCFS2_EXT_REFCOUNTED));
/*
* get_more_blocks() expects us to describe a hole by clearing
*/
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;
- }
-
+ else
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. */
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. Like the core uses
* i_alloc_sem, we use the rw_lock DLM lock to protect io on one node from
{
journal_t *journal = OCFS2_SB(page->mapping->host->i_sb)->journal->j_journal;
- journal_invalidatepage(journal, page, offset);
+ jbd2_journal_invalidatepage(journal, page, offset);
}
static int ocfs2_releasepage(struct page *page, gfp_t wait)
if (!page_has_buffers(page))
return 0;
- return journal_try_to_free_buffers(journal, page, wait);
+ return jbd2_journal_try_to_free_buffers(journal, page, wait);
}
static ssize_t ocfs2_direct_IO(int rw,
mlog_entry_void();
- if (!ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb))) {
- /*
- * We get PR data locks even for O_DIRECT. This
- * allows concurrent O_DIRECT I/O but doesn't let
- * O_DIRECT with extending and buffered zeroing writes
- * race. If they did race then the buffered zeroing
- * could be written back after the O_DIRECT I/O. It's
- * one thing to tell people not to mix buffered and
- * O_DIRECT writes, but expecting them to understand
- * that file extension is also an implicit buffered
- * write is too much. By getting the PR we force
- * writeback of the buffered zeroing before
- * proceeding.
- */
- ret = ocfs2_data_lock(inode, 0);
- if (ret < 0) {
- mlog_errno(ret);
- goto out;
- }
- ocfs2_data_unlock(inode, 0);
- }
+ /*
+ * 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 are appending. */
+ if (i_size_read(inode) <= offset)
+ return 0;
ret = blockdev_direct_IO_no_locking(rw, iocb, inode,
inode->i_sb->s_bdev, iov, offset,
- nr_segs,
+ nr_segs,
ocfs2_direct_IO_get_blocks,
ocfs2_dio_end_io);
-out:
+
mlog_exit(ret);
return ret;
}
}
/*
+ * 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_page_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.
+ */
+static int ocfs2_should_read_blk(struct inode *inode, struct page *page,
+ unsigned int block_start)
+{
+ u64 offset = page_offset(page) + 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_prepare_write(). 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.
set_buffer_uptodate(bh);
} else if (!buffer_uptodate(bh) && !buffer_delay(bh) &&
!buffer_new(bh) &&
+ ocfs2_should_read_blk(inode, page, block_start) &&
(block_start < from || block_end > to)) {
ll_rw_block(READ, 1, &bh);
*wait_bh++=bh;
if (block_start >= to)
break;
- zero_user_page(page, block_start, bh->b_size, KM_USER0);
+ zero_user(page, block_start, bh->b_size);
set_buffer_uptodate(bh);
mark_buffer_dirty(bh);
*/
unsigned c_new;
unsigned c_unwritten;
+ unsigned c_needs_zero;
};
-static inline int ocfs2_should_zero_cluster(struct ocfs2_write_cluster_desc *d)
-{
- return d->c_new || d->c_unwritten;
-}
-
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;
+
struct ocfs2_write_cluster_desc w_desc[OCFS2_MAX_CLUSTERS_PER_PAGE];
/*
struct ocfs2_cached_dealloc_ctxt w_dealloc;
};
-static void ocfs2_free_write_ctxt(struct ocfs2_write_ctxt *wc)
+void ocfs2_unlock_and_free_pages(struct page **pages, int num_pages)
{
int i;
- for(i = 0; i < wc->w_num_pages; i++) {
- if (wc->w_pages[i] == NULL)
- continue;
-
- unlock_page(wc->w_pages[i]);
- mark_page_accessed(wc->w_pages[i]);
- page_cache_release(wc->w_pages[i]);
+ for(i = 0; i < num_pages; i++) {
+ if (pages[i]) {
+ unlock_page(pages[i]);
+ mark_page_accessed(pages[i]);
+ page_cache_release(pages[i]);
+ }
}
+}
+
+static void ocfs2_free_write_ctxt(struct ocfs2_write_ctxt *wc)
+{
+ ocfs2_unlock_and_free_pages(wc->w_pages, wc->w_num_pages);
brelse(wc->w_di_bh);
kfree(wc);
return -ENOMEM;
wc->w_cpos = pos >> osb->s_clustersize_bits;
+ wc->w_first_new_cpos = UINT_MAX;
cend = (pos + len - 1) >> osb->s_clustersize_bits;
wc->w_clen = cend - wc->w_cpos + 1;
get_bh(di_bh);
start = max(from, block_start);
end = min(to, block_end);
- zero_user_page(page, start, end - start, KM_USER0);
+ zero_user_segment(page, start, end);
set_buffer_uptodate(bh);
}
loff_t user_pos, unsigned user_len)
{
int i;
- unsigned from, to;
+ unsigned from = user_pos & (PAGE_CACHE_SIZE - 1),
+ to = user_pos + user_len;
struct page *tmppage;
- ocfs2_zero_new_buffers(wc->w_target_page, user_pos, user_len);
-
- if (wc->w_large_pages) {
- from = wc->w_target_from;
- to = wc->w_target_to;
- } else {
- from = 0;
- to = PAGE_CACHE_SIZE;
- }
+ ocfs2_zero_new_buffers(wc->w_target_page, from, to);
for(i = 0; i < wc->w_num_pages; i++) {
tmppage = wc->w_pages[i];
- if (ocfs2_should_order_data(inode))
- walk_page_buffers(wc->w_handle, page_buffers(tmppage),
- from, to, NULL,
- ocfs2_journal_dirty_data);
+ if (page_has_buffers(tmppage)) {
+ if (ocfs2_should_order_data(inode))
+ ocfs2_jbd2_file_inode(wc->w_handle, inode);
- block_commit_write(tmppage, from, to);
+ block_commit_write(tmppage, from, to);
+ }
}
}
map_from = cluster_start;
map_to = cluster_end;
}
-
- wc->w_target_from = map_from;
- wc->w_target_to = map_to;
} else {
/*
* If we haven't allocated the new page yet, we
*/
static int ocfs2_write_cluster(struct address_space *mapping,
u32 phys, unsigned int unwritten,
+ unsigned int 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, new, should_zero = 0;
+ int ret, i, new;
u64 v_blkno, p_blkno;
struct inode *inode = mapping->host;
+ struct ocfs2_extent_tree et;
new = phys == 0 ? 1 : 0;
- if (new || unwritten)
- should_zero = 1;
-
if (new) {
u32 tmp_pos;
* any additional semaphores or cluster locks.
*/
tmp_pos = cpos;
- ret = ocfs2_do_extend_allocation(OCFS2_SB(inode->i_sb), inode,
- &tmp_pos, 1, 0, wc->w_di_bh,
- wc->w_handle, data_ac,
- meta_ac, NULL);
+ ret = ocfs2_add_inode_data(OCFS2_SB(inode->i_sb), inode,
+ &tmp_pos, 1, 0, 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
goto out;
}
} else if (unwritten) {
- ret = ocfs2_mark_extent_written(inode, wc->w_di_bh,
+ 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) {
if (tmpret) {
mlog_errno(tmpret);
if (ret == 0)
- tmpret = ret;
+ ret = tmpret;
}
}
loff_t pos, unsigned len)
{
int ret, i;
+ loff_t cluster_off;
+ unsigned int local_len = len;
struct ocfs2_write_cluster_desc *desc;
+ struct ocfs2_super *osb = OCFS2_SB(mapping->host->i_sb);
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_unwritten, data_ac, meta_ac,
- wc, desc->c_cpos, pos, len);
+ desc->c_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;
* newly allocated cluster.
*/
desc = &wc->w_desc[0];
- if (ocfs2_should_zero_cluster(desc))
+ if (desc->c_needs_zero)
ocfs2_figure_cluster_boundaries(osb,
desc->c_cpos,
&wc->w_target_from,
NULL);
desc = &wc->w_desc[wc->w_clen - 1];
- if (ocfs2_should_zero_cluster(desc))
+ if (desc->c_needs_zero)
ocfs2_figure_cluster_boundaries(osb,
desc->c_cpos,
NULL,
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.
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;
+ }
+
desc->c_phys = phys;
if (phys == 0) {
desc->c_new = 1;
+ desc->c_needs_zero = 1;
*clusters_to_alloc = *clusters_to_alloc + 1;
}
- if (ext_flags & OCFS2_EXT_UNWRITTEN)
+
+ if (ext_flags & OCFS2_EXT_UNWRITTEN) {
desc->c_unwritten = 1;
+ desc->c_needs_zero = 1;
+ }
num_clusters--;
}
return ret;
}
+static int ocfs2_write_begin_inline(struct address_space *mapping,
+ struct inode *inode,
+ struct ocfs2_write_ctxt *wc)
+{
+ int ret;
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ struct page *page;
+ handle_t *handle;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)wc->w_di_bh->b_data;
+
+ page = find_or_create_page(mapping, 0, GFP_NOFS);
+ if (!page) {
+ ret = -ENOMEM;
+ mlog_errno(ret);
+ goto out;
+ }
+ /*
+ * If we don't set w_num_pages then this page won't get unlocked
+ * and freed on cleanup of the write context.
+ */
+ wc->w_pages[0] = wc->w_target_page = page;
+ wc->w_num_pages = 1;
+
+ 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), 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 (!PageUptodate(page)) {
+ ret = ocfs2_read_inline_data(inode, page, wc->w_di_bh);
+ if (ret) {
+ ocfs2_commit_trans(osb, handle);
+
+ goto out;
+ }
+ }
+
+ wc->w_handle = handle;
+out:
+ return ret;
+}
+
+int ocfs2_size_fits_inline_data(struct buffer_head *di_bh, u64 new_size)
+{
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+
+ if (new_size <= le16_to_cpu(di->id2.i_data.id_count))
+ return 1;
+ return 0;
+}
+
+static int ocfs2_try_to_write_inline_data(struct address_space *mapping,
+ struct inode *inode, loff_t pos,
+ unsigned len, struct page *mmap_page,
+ struct ocfs2_write_ctxt *wc)
+{
+ int ret, written = 0;
+ loff_t end = pos + len;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct ocfs2_dinode *di = NULL;
+
+ mlog(0, "Inode %llu, write of %u bytes at off %llu. features: 0x%x\n",
+ (unsigned long long)oi->ip_blkno, len, (unsigned long long)pos,
+ oi->ip_dyn_features);
+
+ /*
+ * Handle inodes which already have inline data 1st.
+ */
+ if (oi->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
+ if (mmap_page == 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_page ||
+ 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.
+ */
+static int ocfs2_expand_nonsparse_inode(struct inode *inode, loff_t pos,
+ unsigned len,
+ struct ocfs2_write_ctxt *wc)
+{
+ int ret;
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ loff_t newsize = pos + len;
+
+ if (ocfs2_sparse_alloc(osb))
+ return 0;
+
+ if (newsize <= i_size_read(inode))
+ return 0;
+
+ ret = ocfs2_extend_no_holes(inode, newsize, pos);
+ if (ret)
+ mlog_errno(ret);
+
+ wc->w_first_new_cpos =
+ ocfs2_clusters_for_bytes(inode->i_sb, i_size_read(inode));
+
+ return ret;
+}
+
int ocfs2_write_begin_nolock(struct address_space *mapping,
loff_t pos, unsigned len, unsigned flags,
struct page **pagep, void **fsdata,
struct buffer_head *di_bh, struct page *mmap_page)
{
- int ret, credits = OCFS2_INODE_UPDATE_CREDITS;
+ int ret, cluster_of_pages, credits = OCFS2_INODE_UPDATE_CREDITS;
unsigned int clusters_to_alloc, extents_to_split;
struct ocfs2_write_ctxt *wc;
struct inode *inode = mapping->host;
struct ocfs2_alloc_context *data_ac = NULL;
struct ocfs2_alloc_context *meta_ac = NULL;
handle_t *handle;
+ struct ocfs2_extent_tree et;
ret = ocfs2_alloc_write_ctxt(&wc, osb, pos, len, di_bh);
if (ret) {
return ret;
}
+ if (ocfs2_supports_inline_data(osb)) {
+ ret = ocfs2_try_to_write_inline_data(mapping, inode, pos, len,
+ mmap_page, wc);
+ if (ret == 1) {
+ ret = 0;
+ goto success;
+ }
+ if (ret < 0) {
+ mlog_errno(ret);
+ goto out;
+ }
+ }
+
+ ret = ocfs2_expand_nonsparse_inode(inode, 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;
+ } else if (ret == 1) {
+ 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) {
* ocfs2_lock_allocators(). It greatly over-estimates
* the work to be done.
*/
- ret = ocfs2_lock_allocators(inode, di, clusters_to_alloc,
- extents_to_split, &data_ac, &meta_ac);
+ mlog(0, "extend inode %llu, i_size = %lld, di->i_clusters = %u,"
+ " clusters_to_add = %u, extents_to_split = %u\n",
+ (unsigned long long)OCFS2_I(inode)->ip_blkno,
+ (long long)i_size_read(inode), le32_to_cpu(di->i_clusters),
+ clusters_to_alloc, extents_to_split);
+
+ 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;
}
- credits = ocfs2_calc_extend_credits(inode->i_sb, di,
+ credits = ocfs2_calc_extend_credits(inode->i_sb,
+ &di->id2.i_list,
clusters_to_alloc);
}
- ocfs2_set_target_boundaries(osb, wc, pos, len,
- clusters_to_alloc + extents_to_split);
+ /*
+ * 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 (clusters_to_alloc || extents_to_split ||
+ (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;
+
+ ocfs2_set_target_boundaries(osb, wc, pos, len, cluster_of_pages);
handle = ocfs2_start_trans(osb, credits);
if (IS_ERR(handle)) {
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;
+ }
/*
* We don't want this to fail in ocfs2_write_end(), so do it
* here.
*/
- ret = ocfs2_journal_access(handle, inode, wc->w_di_bh,
- OCFS2_JOURNAL_ACCESS_WRITE);
+ ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), wc->w_di_bh,
+ OCFS2_JOURNAL_ACCESS_WRITE);
if (ret) {
mlog_errno(ret);
- goto out_commit;
+ goto out_quota;
}
/*
* extent.
*/
ret = ocfs2_grab_pages_for_write(mapping, wc, wc->w_cpos, pos,
- clusters_to_alloc + extents_to_split,
- mmap_page);
+ cluster_of_pages, mmap_page);
if (ret) {
mlog_errno(ret);
- goto out_commit;
+ goto out_quota;
}
ret = ocfs2_write_cluster_by_desc(mapping, data_ac, meta_ac, wc, pos,
len);
if (ret) {
mlog_errno(ret);
- goto out_commit;
+ goto out_quota;
}
if (data_ac)
if (meta_ac)
ocfs2_free_alloc_context(meta_ac);
+success:
*pagep = wc->w_target_page;
*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);
return ret;
}
-int ocfs2_write_begin(struct file *file, struct address_space *mapping,
- loff_t pos, unsigned len, unsigned flags,
- struct page **pagep, void **fsdata)
+static int ocfs2_write_begin(struct file *file, struct address_space *mapping,
+ loff_t pos, unsigned len, unsigned flags,
+ struct page **pagep, void **fsdata)
{
int ret;
struct buffer_head *di_bh = NULL;
struct inode *inode = mapping->host;
- ret = ocfs2_meta_lock(inode, &di_bh, 1);
+ ret = ocfs2_inode_lock(inode, &di_bh, 1);
if (ret) {
mlog_errno(ret);
return ret;
*/
down_write(&OCFS2_I(inode)->ip_alloc_sem);
- ret = ocfs2_data_lock(inode, 1);
- if (ret) {
- mlog_errno(ret);
- goto out_fail;
- }
-
ret = ocfs2_write_begin_nolock(mapping, pos, len, flags, pagep,
fsdata, di_bh, NULL);
if (ret) {
mlog_errno(ret);
- goto out_fail_data;
+ goto out_fail;
}
brelse(di_bh);
return 0;
-out_fail_data:
- ocfs2_data_unlock(inode, 1);
out_fail:
up_write(&OCFS2_I(inode)->ip_alloc_sem);
brelse(di_bh);
- ocfs2_meta_unlock(inode, 1);
+ ocfs2_inode_unlock(inode, 1);
return ret;
}
+static void ocfs2_write_end_inline(struct inode *inode, loff_t pos,
+ unsigned len, unsigned *copied,
+ struct ocfs2_dinode *di,
+ struct ocfs2_write_ctxt *wc)
+{
+ void *kaddr;
+
+ if (unlikely(*copied < len)) {
+ if (!PageUptodate(wc->w_target_page)) {
+ *copied = 0;
+ return;
+ }
+ }
+
+ kaddr = kmap_atomic(wc->w_target_page, KM_USER0);
+ memcpy(di->id2.i_data.id_data + pos, kaddr + pos, *copied);
+ kunmap_atomic(kaddr, KM_USER0);
+
+ mlog(0, "Data written to inode at offset %llu. "
+ "id_count = %u, copied = %u, i_dyn_features = 0x%x\n",
+ (unsigned long long)pos, *copied,
+ le16_to_cpu(di->id2.i_data.id_count),
+ le16_to_cpu(di->i_dyn_features));
+}
+
int ocfs2_write_end_nolock(struct address_space *mapping,
loff_t pos, unsigned len, unsigned copied,
struct page *page, void *fsdata)
handle_t *handle = wc->w_handle;
struct page *tmppage;
+ if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
+ ocfs2_write_end_inline(inode, pos, len, &copied, di, wc);
+ goto out_write_size;
+ }
+
if (unlikely(copied < len)) {
if (!PageUptodate(wc->w_target_page))
copied = 0;
to = PAGE_CACHE_SIZE;
}
- if (ocfs2_should_order_data(inode))
- walk_page_buffers(wc->w_handle, page_buffers(tmppage),
- from, to, NULL,
- ocfs2_journal_dirty_data);
-
- block_commit_write(tmppage, from, to);
+ if (page_has_buffers(tmppage)) {
+ if (ocfs2_should_order_data(inode))
+ ocfs2_jbd2_file_inode(wc->w_handle, inode);
+ block_commit_write(tmppage, from, to);
+ }
}
+out_write_size:
pos += copied;
if (pos > inode->i_size) {
i_size_write(inode, pos);
return copied;
}
-int ocfs2_write_end(struct file *file, struct address_space *mapping,
- loff_t pos, unsigned len, unsigned copied,
- struct page *page, void *fsdata)
+static int ocfs2_write_end(struct file *file, struct address_space *mapping,
+ loff_t pos, unsigned len, unsigned copied,
+ struct page *page, void *fsdata)
{
int ret;
struct inode *inode = mapping->host;
ret = ocfs2_write_end_nolock(mapping, pos, len, copied, page, fsdata);
- ocfs2_data_unlock(inode, 1);
up_write(&OCFS2_I(inode)->ip_alloc_sem);
- ocfs2_meta_unlock(inode, 1);
+ ocfs2_inode_unlock(inode, 1);
return ret;
}
const struct address_space_operations ocfs2_aops = {
- .readpage = ocfs2_readpage,
- .writepage = ocfs2_writepage,
- .bmap = ocfs2_bmap,
- .sync_page = block_sync_page,
- .direct_IO = ocfs2_direct_IO,
- .invalidatepage = ocfs2_invalidatepage,
- .releasepage = ocfs2_releasepage,
- .migratepage = buffer_migrate_page,
+ .readpage = ocfs2_readpage,
+ .readpages = ocfs2_readpages,
+ .writepage = ocfs2_writepage,
+ .write_begin = ocfs2_write_begin,
+ .write_end = ocfs2_write_end,
+ .bmap = ocfs2_bmap,
+ .sync_page = block_sync_page,
+ .direct_IO = ocfs2_direct_IO,
+ .invalidatepage = ocfs2_invalidatepage,
+ .releasepage = ocfs2_releasepage,
+ .migratepage = buffer_migrate_page,
+ .is_partially_uptodate = block_is_partially_uptodate,
+ .error_remove_page = generic_error_remove_page,
};
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff