}
static void ext4_invalidatepage(struct page *page, unsigned long offset);
-static int noalloc_get_block_write(struct inode *inode, sector_t iblock,
- struct buffer_head *bh_result, int create);
-static int ext4_set_bh_endio(struct buffer_head *bh, struct inode *inode);
-static void ext4_end_io_buffer_write(struct buffer_head *bh, int uptodate);
static int __ext4_journalled_writepage(struct page *page, unsigned int len);
static int ext4_bh_delay_or_unwritten(handle_t *handle, struct buffer_head *bh);
static int ext4_discard_partial_page_buffers_no_lock(handle_t *handle,
return NULL;
bh = sb_getblk(inode->i_sb, map.m_pblk);
- if (!bh) {
- *errp = -EIO;
+ if (unlikely(!bh)) {
+ *errp = -ENOMEM;
return NULL;
}
if (map.m_flags & EXT4_MAP_NEW) {
* and the commit_write(). So doing the jbd2_journal_start at the start of
* prepare_write() is the right place.
*
- * Also, this function can nest inside ext4_writepage() ->
- * block_write_full_page(). In that case, we *know* that ext4_writepage()
- * has generated enough buffer credits to do the whole page. So we won't
- * block on the journal in that case, which is good, because the caller may
- * be PF_MEMALLOC.
+ * Also, this function can nest inside ext4_writepage(). In that case, we
+ * *know* that ext4_writepage() has generated enough buffer credits to do the
+ * whole page. So we won't block on the journal in that case, which is good,
+ * because the caller may be PF_MEMALLOC.
*
* By accident, ext4 can be reentered when a transaction is open via
* quota file writes. If we were to commit the transaction while thus
loff_t size = i_size_read(inode);
unsigned int len, block_start;
struct buffer_head *bh, *page_bufs = NULL;
- int journal_data = ext4_should_journal_data(inode);
sector_t pblock = 0, cur_logical = 0;
struct ext4_io_submit io_submit;
if (nr_pages == 0)
break;
for (i = 0; i < nr_pages; i++) {
- int commit_write = 0, skip_page = 0;
+ int skip_page = 0;
struct page *page = pvec.pages[i];
index = page->index;
BUG_ON(!PageLocked(page));
BUG_ON(PageWriteback(page));
- /*
- * If the page does not have buffers (for
- * whatever reason), try to create them using
- * __block_write_begin. If this fails,
- * skip the page and move on.
- */
- if (!page_has_buffers(page)) {
- if (__block_write_begin(page, 0, len,
- noalloc_get_block_write)) {
- skip_page:
- unlock_page(page);
- continue;
- }
- commit_write = 1;
- }
-
bh = page_bufs = page_buffers(page);
block_start = 0;
do {
- if (!bh)
- goto skip_page;
if (map && (cur_logical >= map->m_lblk) &&
(cur_logical <= (map->m_lblk +
(map->m_len - 1)))) {
pblock++;
} while (bh != page_bufs);
- if (skip_page)
- goto skip_page;
-
- if (commit_write)
- /* mark the buffer_heads as dirty & uptodate */
- block_commit_write(page, 0, len);
+ if (skip_page) {
+ unlock_page(page);
+ continue;
+ }
clear_page_dirty_for_io(page);
- /*
- * Delalloc doesn't support data journalling,
- * but eventually maybe we'll lift this
- * restriction.
- */
- if (unlikely(journal_data && PageChecked(page)))
- err = __ext4_journalled_writepage(page, len);
- else if (test_opt(inode->i_sb, MBLK_IO_SUBMIT))
- err = ext4_bio_write_page(&io_submit, page,
- len, mpd->wbc);
- else if (buffer_uninit(page_bufs)) {
- ext4_set_bh_endio(page_bufs, inode);
- err = block_write_full_page_endio(page,
- noalloc_get_block_write,
- mpd->wbc, ext4_end_io_buffer_write);
- } else
- err = block_write_full_page(page,
- noalloc_get_block_write, mpd->wbc);
-
+ err = ext4_bio_write_page(&io_submit, page, len,
+ mpd->wbc);
if (!err)
mpd->pages_written++;
/*
*
* @mpd->lbh - extent of blocks
* @logical - logical number of the block in the file
- * @bh - bh of the block (used to access block's state)
+ * @b_state - b_state of the buffer head added
*
* the function is used to collect contig. blocks in same state
*/
-static void mpage_add_bh_to_extent(struct mpage_da_data *mpd,
- sector_t logical, size_t b_size,
+static void mpage_add_bh_to_extent(struct mpage_da_data *mpd, sector_t logical,
unsigned long b_state)
{
sector_t next;
- int nrblocks = mpd->b_size >> mpd->inode->i_blkbits;
+ int blkbits = mpd->inode->i_blkbits;
+ int nrblocks = mpd->b_size >> blkbits;
/*
* XXX Don't go larger than mballoc is willing to allocate
* mpage_da_submit_io() into this function and then call
* ext4_map_blocks() multiple times in a loop
*/
- if (nrblocks >= 8*1024*1024/mpd->inode->i_sb->s_blocksize)
+ if (nrblocks >= (8*1024*1024 >> blkbits))
goto flush_it;
- /* check if thereserved journal credits might overflow */
- if (!(ext4_test_inode_flag(mpd->inode, EXT4_INODE_EXTENTS))) {
+ /* check if the reserved journal credits might overflow */
+ if (!ext4_test_inode_flag(mpd->inode, EXT4_INODE_EXTENTS)) {
if (nrblocks >= EXT4_MAX_TRANS_DATA) {
/*
* With non-extent format we are limited by the journal
* nrblocks. So limit nrblocks.
*/
goto flush_it;
- } else if ((nrblocks + (b_size >> mpd->inode->i_blkbits)) >
- EXT4_MAX_TRANS_DATA) {
- /*
- * Adding the new buffer_head would make it cross the
- * allowed limit for which we have journal credit
- * reserved. So limit the new bh->b_size
- */
- b_size = (EXT4_MAX_TRANS_DATA - nrblocks) <<
- mpd->inode->i_blkbits;
- /* we will do mpage_da_submit_io in the next loop */
}
}
/*
*/
if (mpd->b_size == 0) {
mpd->b_blocknr = logical;
- mpd->b_size = b_size;
+ mpd->b_size = 1 << blkbits;
mpd->b_state = b_state & BH_FLAGS;
return;
}
* Can we merge the block to our big extent?
*/
if (logical == next && (b_state & BH_FLAGS) == mpd->b_state) {
- mpd->b_size += b_size;
+ mpd->b_size += 1 << blkbits;
return;
}
return 0;
}
-/*
- * This function is used as a standard get_block_t calback function
- * when there is no desire to allocate any blocks. It is used as a
- * callback function for block_write_begin() and block_write_full_page().
- * These functions should only try to map a single block at a time.
- *
- * Since this function doesn't do block allocations even if the caller
- * requests it by passing in create=1, it is critically important that
- * any caller checks to make sure that any buffer heads are returned
- * by this function are either all already mapped or marked for
- * delayed allocation before calling block_write_full_page(). Otherwise,
- * b_blocknr could be left unitialized, and the page write functions will
- * be taken by surprise.
- */
-static int noalloc_get_block_write(struct inode *inode, sector_t iblock,
- struct buffer_head *bh_result, int create)
-{
- BUG_ON(bh_result->b_size != inode->i_sb->s_blocksize);
- return _ext4_get_block(inode, iblock, bh_result, 0);
-}
-
static int bget_one(handle_t *handle, struct buffer_head *bh)
{
get_bh(bh);
static int ext4_writepage(struct page *page,
struct writeback_control *wbc)
{
- int ret = 0, commit_write = 0;
+ int ret = 0;
loff_t size;
unsigned int len;
struct buffer_head *page_bufs = NULL;
struct inode *inode = page->mapping->host;
+ struct ext4_io_submit io_submit;
trace_ext4_writepage(page);
size = i_size_read(inode);
else
len = PAGE_CACHE_SIZE;
+ page_bufs = page_buffers(page);
/*
- * If the page does not have buffers (for whatever reason),
- * try to create them using __block_write_begin. If this
- * fails, redirty the page and move on.
+ * We cannot do block allocation or other extent handling in this
+ * function. If there are buffers needing that, we have to redirty
+ * the page. But we may reach here when we do a journal commit via
+ * journal_submit_inode_data_buffers() and in that case we must write
+ * allocated buffers to achieve data=ordered mode guarantees.
*/
- if (!page_has_buffers(page)) {
- if (__block_write_begin(page, 0, len,
- noalloc_get_block_write)) {
- redirty_page:
- redirty_page_for_writepage(wbc, page);
+ if (ext4_walk_page_buffers(NULL, page_bufs, 0, len, NULL,
+ ext4_bh_delay_or_unwritten)) {
+ redirty_page_for_writepage(wbc, page);
+ if (current->flags & PF_MEMALLOC) {
+ /*
+ * For memory cleaning there's no point in writing only
+ * some buffers. So just bail out. Warn if we came here
+ * from direct reclaim.
+ */
+ WARN_ON_ONCE((current->flags & (PF_MEMALLOC|PF_KSWAPD))
+ == PF_MEMALLOC);
unlock_page(page);
return 0;
}
- commit_write = 1;
}
- page_bufs = page_buffers(page);
- if (ext4_walk_page_buffers(NULL, page_bufs, 0, len, NULL,
- ext4_bh_delay_or_unwritten)) {
- /*
- * We don't want to do block allocation, so redirty
- * the page and return. We may reach here when we do
- * a journal commit via journal_submit_inode_data_buffers.
- * We can also reach here via shrink_page_list but it
- * should never be for direct reclaim so warn if that
- * happens
- */
- WARN_ON_ONCE((current->flags & (PF_MEMALLOC|PF_KSWAPD)) ==
- PF_MEMALLOC);
- goto redirty_page;
- }
- if (commit_write)
- /* now mark the buffer_heads as dirty and uptodate */
- block_commit_write(page, 0, len);
if (PageChecked(page) && ext4_should_journal_data(inode))
/*
*/
return __ext4_journalled_writepage(page, len);
- if (buffer_uninit(page_bufs)) {
- ext4_set_bh_endio(page_bufs, inode);
- ret = block_write_full_page_endio(page, noalloc_get_block_write,
- wbc, ext4_end_io_buffer_write);
- } else
- ret = block_write_full_page(page, noalloc_get_block_write,
- wbc);
-
+ memset(&io_submit, 0, sizeof(io_submit));
+ ret = ext4_bio_write_page(&io_submit, page, len, wbc);
+ ext4_io_submit(&io_submit);
return ret;
}
logical = (sector_t) page->index <<
(PAGE_CACHE_SHIFT - inode->i_blkbits);
- if (!page_has_buffers(page)) {
- mpage_add_bh_to_extent(mpd, logical,
- PAGE_CACHE_SIZE,
- (1 << BH_Dirty) | (1 << BH_Uptodate));
- if (mpd->io_done)
- goto ret_extent_tail;
- } else {
+ /* Add all dirty buffers to mpd */
+ head = page_buffers(page);
+ bh = head;
+ do {
+ BUG_ON(buffer_locked(bh));
/*
- * Page with regular buffer heads,
- * just add all dirty ones
+ * We need to try to allocate unmapped blocks
+ * in the same page. Otherwise we won't make
+ * progress with the page in ext4_writepage
*/
- head = page_buffers(page);
- bh = head;
- do {
- BUG_ON(buffer_locked(bh));
+ if (ext4_bh_delay_or_unwritten(NULL, bh)) {
+ mpage_add_bh_to_extent(mpd, logical,
+ bh->b_state);
+ if (mpd->io_done)
+ goto ret_extent_tail;
+ } else if (buffer_dirty(bh) &&
+ buffer_mapped(bh)) {
/*
- * We need to try to allocate
- * unmapped blocks in the same page.
- * Otherwise we won't make progress
- * with the page in ext4_writepage
+ * mapped dirty buffer. We need to
+ * update the b_state because we look
+ * at b_state in mpage_da_map_blocks.
+ * We don't update b_size because if we
+ * find an unmapped buffer_head later
+ * we need to use the b_state flag of
+ * that buffer_head.
*/
- if (ext4_bh_delay_or_unwritten(NULL, bh)) {
- mpage_add_bh_to_extent(mpd, logical,
- bh->b_size,
- bh->b_state);
- if (mpd->io_done)
- goto ret_extent_tail;
- } else if (buffer_dirty(bh) && (buffer_mapped(bh))) {
- /*
- * mapped dirty buffer. We need
- * to update the b_state
- * because we look at b_state
- * in mpage_da_map_blocks. We
- * don't update b_size because
- * if we find an unmapped
- * buffer_head later we need to
- * use the b_state flag of that
- * buffer_head.
- */
- if (mpd->b_size == 0)
- mpd->b_state = bh->b_state & BH_FLAGS;
- }
- logical++;
- } while ((bh = bh->b_this_page) != head);
- }
+ if (mpd->b_size == 0)
+ mpd->b_state =
+ bh->b_state & BH_FLAGS;
+ }
+ logical++;
+ } while ((bh = bh->b_this_page) != head);
if (nr_to_write > 0) {
nr_to_write--;
return mpage_readpages(mapping, pages, nr_pages, ext4_get_block);
}
-static void ext4_invalidatepage_free_endio(struct page *page, unsigned long offset)
+static void ext4_invalidatepage(struct page *page, unsigned long offset)
{
- struct buffer_head *head, *bh;
- unsigned int curr_off = 0;
+ trace_ext4_invalidatepage(page, offset);
- if (!page_has_buffers(page))
- return;
- head = bh = page_buffers(page);
- do {
- if (offset <= curr_off && test_clear_buffer_uninit(bh)
- && bh->b_private) {
- ext4_free_io_end(bh->b_private);
- bh->b_private = NULL;
- bh->b_end_io = NULL;
- }
- curr_off = curr_off + bh->b_size;
- bh = bh->b_this_page;
- } while (bh != head);
+ /* No journalling happens on data buffers when this function is used */
+ WARN_ON(page_has_buffers(page) && buffer_jbd(page_buffers(page)));
+
+ block_invalidatepage(page, offset);
}
-static void ext4_invalidatepage(struct page *page, unsigned long offset)
+static int __ext4_journalled_invalidatepage(struct page *page,
+ unsigned long offset)
{
journal_t *journal = EXT4_JOURNAL(page->mapping->host);
- trace_ext4_invalidatepage(page, offset);
+ trace_ext4_journalled_invalidatepage(page, offset);
- /*
- * free any io_end structure allocated for buffers to be discarded
- */
- if (ext4_should_dioread_nolock(page->mapping->host))
- ext4_invalidatepage_free_endio(page, offset);
/*
* If it's a full truncate we just forget about the pending dirtying
*/
if (offset == 0)
ClearPageChecked(page);
- if (journal)
- jbd2_journal_invalidatepage(journal, page, offset);
- else
- block_invalidatepage(page, offset);
+ return jbd2_journal_invalidatepage(journal, page, offset);
+}
+
+/* Wrapper for aops... */
+static void ext4_journalled_invalidatepage(struct page *page,
+ unsigned long offset)
+{
+ WARN_ON(__ext4_journalled_invalidatepage(page, offset) < 0);
}
static int ext4_releasepage(struct page *page, gfp_t wait)
ext4_add_complete_io(io_end);
}
-static void ext4_end_io_buffer_write(struct buffer_head *bh, int uptodate)
-{
- ext4_io_end_t *io_end = bh->b_private;
- struct inode *inode;
-
- if (!test_clear_buffer_uninit(bh) || !io_end)
- goto out;
-
- if (!(io_end->inode->i_sb->s_flags & MS_ACTIVE)) {
- ext4_msg(io_end->inode->i_sb, KERN_INFO,
- "sb umounted, discard end_io request for inode %lu",
- io_end->inode->i_ino);
- ext4_free_io_end(io_end);
- goto out;
- }
-
- /*
- * It may be over-defensive here to check EXT4_IO_END_UNWRITTEN now,
- * but being more careful is always safe for the future change.
- */
- inode = io_end->inode;
- ext4_set_io_unwritten_flag(inode, io_end);
- ext4_add_complete_io(io_end);
-out:
- bh->b_private = NULL;
- bh->b_end_io = NULL;
- clear_buffer_uninit(bh);
- end_buffer_async_write(bh, uptodate);
-}
-
-static int ext4_set_bh_endio(struct buffer_head *bh, struct inode *inode)
-{
- ext4_io_end_t *io_end;
- struct page *page = bh->b_page;
- loff_t offset = (sector_t)page->index << PAGE_CACHE_SHIFT;
- size_t size = bh->b_size;
-
-retry:
- io_end = ext4_init_io_end(inode, GFP_ATOMIC);
- if (!io_end) {
- pr_warn_ratelimited("%s: allocation fail\n", __func__);
- schedule();
- goto retry;
- }
- io_end->offset = offset;
- io_end->size = size;
- /*
- * We need to hold a reference to the page to make sure it
- * doesn't get evicted before ext4_end_io_work() has a chance
- * to convert the extent from written to unwritten.
- */
- io_end->page = page;
- get_page(io_end->page);
-
- bh->b_private = io_end;
- bh->b_end_io = ext4_end_io_buffer_write;
- return 0;
-}
-
/*
* For ext4 extent files, ext4 will do direct-io write to holes,
* preallocated extents, and those write extend the file, no need to
.write_end = ext4_journalled_write_end,
.set_page_dirty = ext4_journalled_set_page_dirty,
.bmap = ext4_bmap,
- .invalidatepage = ext4_invalidatepage,
+ .invalidatepage = ext4_journalled_invalidatepage,
.releasepage = ext4_releasepage,
.direct_IO = ext4_direct_IO,
.is_partially_uptodate = block_is_partially_uptodate,
if (!S_ISREG(inode->i_mode))
return -EOPNOTSUPP;
- if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
- /* TODO: Add support for non extent hole punching */
- return -EOPNOTSUPP;
- }
+ if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
+ return ext4_ind_punch_hole(file, offset, length);
if (EXT4_SB(inode->i_sb)->s_cluster_ratio > 1) {
/* TODO: Add support for bigalloc file systems */
return -EOPNOTSUPP;
}
+ trace_ext4_punch_hole(inode, offset, length);
+
return ext4_ext_punch_hole(file, offset, length);
}
iloc->offset = (inode_offset % inodes_per_block) * EXT4_INODE_SIZE(sb);
bh = sb_getblk(sb, block);
- if (!bh) {
- EXT4_ERROR_INODE_BLOCK(inode, block,
- "unable to read itable block");
- return -EIO;
- }
+ if (unlikely(!bh))
+ return -ENOMEM;
if (!buffer_uptodate(bh)) {
lock_buffer(bh);
/* Is the inode bitmap in cache? */
bitmap_bh = sb_getblk(sb, ext4_inode_bitmap(sb, gdp));
- if (!bitmap_bh)
+ if (unlikely(!bitmap_bh))
goto make_io;
/*
return err;
}
+/*
+ * In data=journal mode ext4_journalled_invalidatepage() may fail to invalidate
+ * buffers that are attached to a page stradding i_size and are undergoing
+ * commit. In that case we have to wait for commit to finish and try again.
+ */
+static void ext4_wait_for_tail_page_commit(struct inode *inode)
+{
+ struct page *page;
+ unsigned offset;
+ journal_t *journal = EXT4_SB(inode->i_sb)->s_journal;
+ tid_t commit_tid = 0;
+ int ret;
+
+ offset = inode->i_size & (PAGE_CACHE_SIZE - 1);
+ /*
+ * All buffers in the last page remain valid? Then there's nothing to
+ * do. We do the check mainly to optimize the common PAGE_CACHE_SIZE ==
+ * blocksize case
+ */
+ if (offset > PAGE_CACHE_SIZE - (1 << inode->i_blkbits))
+ return;
+ while (1) {
+ page = find_lock_page(inode->i_mapping,
+ inode->i_size >> PAGE_CACHE_SHIFT);
+ if (!page)
+ return;
+ ret = __ext4_journalled_invalidatepage(page, offset);
+ unlock_page(page);
+ page_cache_release(page);
+ if (ret != -EBUSY)
+ return;
+ commit_tid = 0;
+ read_lock(&journal->j_state_lock);
+ if (journal->j_committing_transaction)
+ commit_tid = journal->j_committing_transaction->t_tid;
+ read_unlock(&journal->j_state_lock);
+ if (commit_tid)
+ jbd2_log_wait_commit(journal, commit_tid);
+ }
+}
+
/*
* ext4_setattr()
*
}
if (attr->ia_valid & ATTR_SIZE) {
- if (attr->ia_size != i_size_read(inode)) {
- truncate_setsize(inode, attr->ia_size);
- /* Inode size will be reduced, wait for dio in flight.
- * Temporarily disable dioread_nolock to prevent
- * livelock. */
+ if (attr->ia_size != inode->i_size) {
+ loff_t oldsize = inode->i_size;
+
+ i_size_write(inode, attr->ia_size);
+ /*
+ * Blocks are going to be removed from the inode. Wait
+ * for dio in flight. Temporarily disable
+ * dioread_nolock to prevent livelock.
+ */
if (orphan) {
- ext4_inode_block_unlocked_dio(inode);
- inode_dio_wait(inode);
- ext4_inode_resume_unlocked_dio(inode);
+ if (!ext4_should_journal_data(inode)) {
+ ext4_inode_block_unlocked_dio(inode);
+ inode_dio_wait(inode);
+ ext4_inode_resume_unlocked_dio(inode);
+ } else
+ ext4_wait_for_tail_page_commit(inode);
}
+ /*
+ * Truncate pagecache after we've waited for commit
+ * in data=journal mode to make pages freeable.
+ */
+ truncate_pagecache(inode, oldsize, inode->i_size);
}
ext4_truncate(inode);
}