diff options
Diffstat (limited to 'fs/ext3/inode.c')
| -rw-r--r-- | fs/ext3/inode.c | 510 |
1 files changed, 292 insertions, 218 deletions
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c index a9580617edd..2c6ccc49ba2 100644 --- a/fs/ext3/inode.c +++ b/fs/ext3/inode.c @@ -22,26 +22,18 @@ * Assorted race fixes, rewrite of ext3_get_block() by Al Viro, 2000 */ -#include <linux/module.h> -#include <linux/fs.h> -#include <linux/time.h> -#include <linux/ext3_jbd.h> -#include <linux/jbd.h> #include <linux/highuid.h> -#include <linux/pagemap.h> #include <linux/quotaops.h> -#include <linux/string.h> -#include <linux/buffer_head.h> #include <linux/writeback.h> #include <linux/mpage.h> -#include <linux/uio.h> -#include <linux/bio.h> -#include <linux/fiemap.h> #include <linux/namei.h> +#include <linux/aio.h> +#include "ext3.h" #include "xattr.h" #include "acl.h" static int ext3_writepage_trans_blocks(struct inode *inode); +static int ext3_block_truncate_page(struct inode *inode, loff_t from); /* * Test whether an inode is a fast symlink. @@ -70,6 +62,7 @@ int ext3_forget(handle_t *handle, int is_metadata, struct inode *inode, might_sleep(); + trace_ext3_forget(inode, is_metadata, blocknr); BUFFER_TRACE(bh, "enter"); jbd_debug(4, "forgetting bh %p: is_metadata = %d, mode %o, " @@ -194,20 +187,52 @@ static int truncate_restart_transaction(handle_t *handle, struct inode *inode) */ void ext3_evict_inode (struct inode *inode) { + struct ext3_inode_info *ei = EXT3_I(inode); struct ext3_block_alloc_info *rsv; handle_t *handle; int want_delete = 0; + trace_ext3_evict_inode(inode); if (!inode->i_nlink && !is_bad_inode(inode)) { dquot_initialize(inode); want_delete = 1; } - truncate_inode_pages(&inode->i_data, 0); + /* + * When journalling data dirty buffers are tracked only in the journal. + * So although mm thinks everything is clean and ready for reaping the + * inode might still have some pages to write in the running + * transaction or waiting to be checkpointed. Thus calling + * journal_invalidatepage() (via truncate_inode_pages()) to discard + * these buffers can cause data loss. Also even if we did not discard + * these buffers, we would have no way to find them after the inode + * is reaped and thus user could see stale data if he tries to read + * them before the transaction is checkpointed. So be careful and + * force everything to disk here... We use ei->i_datasync_tid to + * store the newest transaction containing inode's data. + * + * Note that directories do not have this problem because they don't + * use page cache. + * + * The s_journal check handles the case when ext3_get_journal() fails + * and puts the journal inode. + */ + if (inode->i_nlink && ext3_should_journal_data(inode) && + EXT3_SB(inode->i_sb)->s_journal && + (S_ISLNK(inode->i_mode) || S_ISREG(inode->i_mode)) && + inode->i_ino != EXT3_JOURNAL_INO) { + tid_t commit_tid = atomic_read(&ei->i_datasync_tid); + journal_t *journal = EXT3_SB(inode->i_sb)->s_journal; + + log_start_commit(journal, commit_tid); + log_wait_commit(journal, commit_tid); + filemap_write_and_wait(&inode->i_data); + } + truncate_inode_pages_final(&inode->i_data); ext3_discard_reservation(inode); - rsv = EXT3_I(inode)->i_block_alloc_info; - EXT3_I(inode)->i_block_alloc_info = NULL; + rsv = ei->i_block_alloc_info; + ei->i_block_alloc_info = NULL; if (unlikely(rsv)) kfree(rsv); @@ -231,15 +256,13 @@ void ext3_evict_inode (struct inode *inode) if (inode->i_blocks) ext3_truncate(inode); /* - * Kill off the orphan record which ext3_truncate created. - * AKPM: I think this can be inside the above `if'. - * Note that ext3_orphan_del() has to be able to cope with the - * deletion of a non-existent orphan - this is because we don't - * know if ext3_truncate() actually created an orphan record. - * (Well, we could do this if we need to, but heck - it works) + * Kill off the orphan record created when the inode lost the last + * link. Note that ext3_orphan_del() has to be able to cope with the + * deletion of a non-existent orphan - ext3_truncate() could + * have removed the record. */ ext3_orphan_del(handle, inode); - EXT3_I(inode)->i_dtime = get_seconds(); + ei->i_dtime = get_seconds(); /* * One subtle ordering requirement: if anything has gone wrong @@ -251,18 +274,18 @@ void ext3_evict_inode (struct inode *inode) if (ext3_mark_inode_dirty(handle, inode)) { /* If that failed, just dquot_drop() and be done with that */ dquot_drop(inode); - end_writeback(inode); + clear_inode(inode); } else { ext3_xattr_delete_inode(handle, inode); dquot_free_inode(inode); dquot_drop(inode); - end_writeback(inode); + clear_inode(inode); ext3_free_inode(handle, inode); } ext3_journal_stop(handle); return; no_delete: - end_writeback(inode); + clear_inode(inode); dquot_drop(inode); } @@ -655,6 +678,10 @@ static int ext3_alloc_branch(handle_t *handle, struct inode *inode, * parent to disk. */ bh = sb_getblk(inode->i_sb, new_blocks[n-1]); + if (unlikely(!bh)) { + err = -ENOMEM; + goto failed; + } branch[n].bh = bh; lock_buffer(bh); BUFFER_TRACE(bh, "call get_create_access"); @@ -696,7 +723,7 @@ failed: BUFFER_TRACE(branch[i].bh, "call journal_forget"); ext3_journal_forget(handle, branch[i].bh); } - for (i = 0; i <indirect_blks; i++) + for (i = 0; i < indirect_blks; i++) ext3_free_blocks(handle, inode, new_blocks[i], 1); ext3_free_blocks(handle, inode, new_blocks[i], num); @@ -725,6 +752,7 @@ static int ext3_splice_branch(handle_t *handle, struct inode *inode, struct ext3_block_alloc_info *block_i; ext3_fsblk_t current_block; struct ext3_inode_info *ei = EXT3_I(inode); + struct timespec now; block_i = ei->i_block_alloc_info; /* @@ -764,9 +792,11 @@ static int ext3_splice_branch(handle_t *handle, struct inode *inode, } /* We are done with atomic stuff, now do the rest of housekeeping */ - - inode->i_ctime = CURRENT_TIME_SEC; - ext3_mark_inode_dirty(handle, inode); + now = CURRENT_TIME_SEC; + if (!timespec_equal(&inode->i_ctime, &now) || !where->bh) { + inode->i_ctime = now; + ext3_mark_inode_dirty(handle, inode); + } /* ext3_mark_inode_dirty already updated i_sync_tid */ atomic_set(&ei->i_datasync_tid, handle->h_transaction->t_tid); @@ -842,6 +872,7 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode, ext3_fsblk_t first_block = 0; + trace_ext3_get_blocks_enter(inode, iblock, maxblocks, create); J_ASSERT(handle != NULL || create == 0); depth = ext3_block_to_path(inode,iblock,offsets,&blocks_to_boundary); @@ -886,6 +917,9 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode, if (!create || err == -EIO) goto cleanup; + /* + * Block out ext3_truncate while we alter the tree + */ mutex_lock(&ei->truncate_mutex); /* @@ -934,9 +968,6 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode, */ count = ext3_blks_to_allocate(partial, indirect_blks, maxblocks, blocks_to_boundary); - /* - * Block out ext3_truncate while we alter the tree - */ err = ext3_alloc_branch(handle, inode, indirect_blks, &count, goal, offsets + (partial - chain), partial); @@ -970,6 +1001,9 @@ cleanup: } BUFFER_TRACE(bh_result, "returned"); out: + trace_ext3_get_blocks_exit(inode, iblock, + depth ? le32_to_cpu(chain[depth-1].key) : 0, + count, err); return err; } @@ -1043,16 +1077,15 @@ struct buffer_head *ext3_getblk(handle_t *handle, struct inode *inode, * mapped. 0 in case of a HOLE. */ if (err > 0) { - if (err > 1) - WARN_ON(1); + WARN_ON(err > 1); err = 0; } *errp = err; if (!err && buffer_mapped(&dummy)) { struct buffer_head *bh; bh = sb_getblk(inode->i_sb, dummy.b_blocknr); - if (!bh) { - *errp = -EIO; + if (unlikely(!bh)) { + *errp = -ENOMEM; goto err; } if (buffer_new(&dummy)) { @@ -1100,9 +1133,11 @@ struct buffer_head *ext3_bread(handle_t *handle, struct inode *inode, bh = ext3_getblk(handle, inode, block, create, err); if (!bh) return bh; - if (buffer_uptodate(bh)) + if (bh_uptodate_or_lock(bh)) return bh; - ll_rw_block(READ_META, 1, &bh); + get_bh(bh); + bh->b_end_io = end_buffer_read_sync; + submit_bh(READ | REQ_META | REQ_PRIO, bh); wait_on_buffer(bh); if (buffer_uptodate(bh)) return bh; @@ -1202,6 +1237,16 @@ static void ext3_truncate_failed_write(struct inode *inode) ext3_truncate(inode); } +/* + * Truncate blocks that were not used by direct IO write. We have to zero out + * the last file block as well because direct IO might have written to it. + */ +static void ext3_truncate_failed_direct_write(struct inode *inode) +{ + ext3_block_truncate_page(inode, inode->i_size); + ext3_truncate(inode); +} + static int ext3_write_begin(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned flags, struct page **pagep, void **fsdata) @@ -1217,6 +1262,8 @@ static int ext3_write_begin(struct file *file, struct address_space *mapping, * we allocate blocks but write fails for some reason */ int needed_blocks = ext3_writepage_trans_blocks(inode) + 1; + trace_ext3_write_begin(inode, pos, len, flags); + index = pos >> PAGE_CACHE_SHIFT; from = pos & (PAGE_CACHE_SIZE - 1); to = from + len; @@ -1332,6 +1379,7 @@ static int ext3_ordered_write_end(struct file *file, unsigned from, to; int ret = 0, ret2; + trace_ext3_ordered_write_end(inode, pos, len, copied); copied = block_write_end(file, mapping, pos, len, copied, page, fsdata); from = pos & (PAGE_CACHE_SIZE - 1); @@ -1367,6 +1415,7 @@ static int ext3_writeback_write_end(struct file *file, struct inode *inode = file->f_mapping->host; int ret; + trace_ext3_writeback_write_end(inode, pos, len, copied); copied = block_write_end(file, mapping, pos, len, copied, page, fsdata); update_file_sizes(inode, pos, copied); /* @@ -1391,10 +1440,12 @@ static int ext3_journalled_write_end(struct file *file, { handle_t *handle = ext3_journal_current_handle(); struct inode *inode = mapping->host; + struct ext3_inode_info *ei = EXT3_I(inode); int ret = 0, ret2; int partial = 0; unsigned from, to; + trace_ext3_journalled_write_end(inode, pos, len, copied); from = pos & (PAGE_CACHE_SIZE - 1); to = from + len; @@ -1419,8 +1470,9 @@ static int ext3_journalled_write_end(struct file *file, if (pos + len > inode->i_size && ext3_can_truncate(inode)) ext3_orphan_add(handle, inode); ext3_set_inode_state(inode, EXT3_STATE_JDATA); - if (inode->i_size > EXT3_I(inode)->i_disksize) { - EXT3_I(inode)->i_disksize = inode->i_size; + atomic_set(&ei->i_datasync_tid, handle->h_transaction->t_tid); + if (inode->i_size > ei->i_disksize) { + ei->i_disksize = inode->i_size; ret2 = ext3_mark_inode_dirty(handle, inode); if (!ret) ret = ret2; @@ -1507,56 +1559,17 @@ static int buffer_unmapped(handle_t *handle, struct buffer_head *bh) } /* - * Note that we always start a transaction even if we're not journalling - * data. This is to preserve ordering: any hole instantiation within - * __block_write_full_page -> ext3_get_block() should be journalled - * along with the data so we don't crash and then get metadata which + * Note that whenever we need to map blocks we start a transaction even if + * we're not journalling data. This is to preserve ordering: any hole + * instantiation within __block_write_full_page -> ext3_get_block() should be + * journalled along with the data so we don't crash and then get metadata which * refers to old data. * * In all journalling modes block_write_full_page() will start the I/O. * - * Problem: - * - * ext3_writepage() -> kmalloc() -> __alloc_pages() -> page_launder() -> - * ext3_writepage() - * - * Similar for: - * - * ext3_file_write() -> generic_file_write() -> __alloc_pages() -> ... - * - * Same applies to ext3_get_block(). We will deadlock on various things like - * lock_journal and i_truncate_mutex. - * - * Setting PF_MEMALLOC here doesn't work - too many internal memory - * allocations fail. - * - * 16May01: If we're reentered then journal_current_handle() will be - * non-zero. We simply *return*. - * - * 1 July 2001: @@@ FIXME: - * In journalled data mode, a data buffer may be metadata against the - * current transaction. But the same file is part of a shared mapping - * and someone does a writepage() on it. - * - * We will move the buffer onto the async_data list, but *after* it has - * been dirtied. So there's a small window where we have dirty data on - * BJ_Metadata. - * - * Note that this only applies to the last partial page in the file. The - * bit which block_write_full_page() uses prepare/commit for. (That's - * broken code anyway: it's wrong for msync()). - * - * It's a rare case: affects the final partial page, for journalled data - * where the file is subject to bith write() and writepage() in the same - * transction. To fix it we'll need a custom block_write_full_page(). - * We'll probably need that anyway for journalling writepage() output. - * * We don't honour synchronous mounts for writepage(). That would be * disastrous. Any write() or metadata operation will sync the fs for * us. - * - * AKPM2: if all the page's buffers are mapped to disk and !data=journal, - * we don't need to open a transaction here. */ static int ext3_ordered_writepage(struct page *page, struct writeback_control *wbc) @@ -1568,7 +1581,13 @@ static int ext3_ordered_writepage(struct page *page, int err; J_ASSERT(PageLocked(page)); - WARN_ON_ONCE(IS_RDONLY(inode)); + /* + * We don't want to warn for emergency remount. The condition is + * ordered to avoid dereferencing inode->i_sb in non-error case to + * avoid slow-downs. + */ + WARN_ON_ONCE(IS_RDONLY(inode) && + !(EXT3_SB(inode->i_sb)->s_mount_state & EXT3_ERROR_FS)); /* * We give up here if we're reentered, because it might be for a @@ -1577,6 +1596,7 @@ static int ext3_ordered_writepage(struct page *page, if (ext3_journal_current_handle()) goto out_fail; + trace_ext3_ordered_writepage(page); if (!page_has_buffers(page)) { create_empty_buffers(page, inode->i_sb->s_blocksize, (1 << BH_Dirty)|(1 << BH_Uptodate)); @@ -1614,12 +1634,9 @@ static int ext3_ordered_writepage(struct page *page, * block_write_full_page() succeeded. Otherwise they are unmapped, * and generally junk. */ - if (ret == 0) { - err = walk_page_buffers(handle, page_bufs, 0, PAGE_CACHE_SIZE, + if (ret == 0) + ret = walk_page_buffers(handle, page_bufs, 0, PAGE_CACHE_SIZE, NULL, journal_dirty_data_fn); - if (!ret) - ret = err; - } walk_page_buffers(handle, page_bufs, 0, PAGE_CACHE_SIZE, NULL, bput_one); err = ext3_journal_stop(handle); @@ -1642,11 +1659,18 @@ static int ext3_writeback_writepage(struct page *page, int err; J_ASSERT(PageLocked(page)); - WARN_ON_ONCE(IS_RDONLY(inode)); + /* + * We don't want to warn for emergency remount. The condition is + * ordered to avoid dereferencing inode->i_sb in non-error case to + * avoid slow-downs. + */ + WARN_ON_ONCE(IS_RDONLY(inode) && + !(EXT3_SB(inode->i_sb)->s_mount_state & EXT3_ERROR_FS)); if (ext3_journal_current_handle()) goto out_fail; + trace_ext3_writeback_writepage(page); if (page_has_buffers(page)) { if (!walk_page_buffers(NULL, page_buffers(page), 0, PAGE_CACHE_SIZE, NULL, buffer_unmapped)) { @@ -1684,18 +1708,25 @@ static int ext3_journalled_writepage(struct page *page, int err; J_ASSERT(PageLocked(page)); - WARN_ON_ONCE(IS_RDONLY(inode)); - - if (ext3_journal_current_handle()) - goto no_write; - - handle = ext3_journal_start(inode, ext3_writepage_trans_blocks(inode)); - if (IS_ERR(handle)) { - ret = PTR_ERR(handle); - goto no_write; - } + /* + * We don't want to warn for emergency remount. The condition is + * ordered to avoid dereferencing inode->i_sb in non-error case to + * avoid slow-downs. + */ + WARN_ON_ONCE(IS_RDONLY(inode) && + !(EXT3_SB(inode->i_sb)->s_mount_state & EXT3_ERROR_FS)); + trace_ext3_journalled_writepage(page); if (!page_has_buffers(page) || PageChecked(page)) { + if (ext3_journal_current_handle()) + goto no_write; + + handle = ext3_journal_start(inode, + ext3_writepage_trans_blocks(inode)); + if (IS_ERR(handle)) { + ret = PTR_ERR(handle); + goto no_write; + } /* * It's mmapped pagecache. Add buffers and journal it. There * doesn't seem much point in redirtying the page here. @@ -1715,18 +1746,21 @@ static int ext3_journalled_writepage(struct page *page, if (ret == 0) ret = err; ext3_set_inode_state(inode, EXT3_STATE_JDATA); + atomic_set(&EXT3_I(inode)->i_datasync_tid, + handle->h_transaction->t_tid); unlock_page(page); + err = ext3_journal_stop(handle); + if (!ret) + ret = err; } else { /* - * It may be a page full of checkpoint-mode buffers. We don't - * really know unless we go poke around in the buffer_heads. - * But block_write_full_page will do the right thing. + * It is a page full of checkpoint-mode buffers. Go and write + * them. They should have been already mapped when they went + * to the journal so provide NULL get_block function to catch + * errors. */ - ret = block_write_full_page(page, ext3_get_block, wbc); + ret = block_write_full_page(page, NULL, wbc); } - err = ext3_journal_stop(handle); - if (!ret) - ret = err; out: return ret; @@ -1739,6 +1773,7 @@ out_unlock: static int ext3_readpage(struct file *file, struct page *page) { + trace_ext3_readpage(page); return mpage_readpage(page, ext3_get_block); } @@ -1749,23 +1784,27 @@ ext3_readpages(struct file *file, struct address_space *mapping, return mpage_readpages(mapping, pages, nr_pages, ext3_get_block); } -static void ext3_invalidatepage(struct page *page, unsigned long offset) +static void ext3_invalidatepage(struct page *page, unsigned int offset, + unsigned int length) { journal_t *journal = EXT3_JOURNAL(page->mapping->host); + trace_ext3_invalidatepage(page, offset, length); + /* * If it's a full truncate we just forget about the pending dirtying */ - if (offset == 0) + if (offset == 0 && length == PAGE_CACHE_SIZE) ClearPageChecked(page); - journal_invalidatepage(journal, page, offset); + journal_invalidatepage(journal, page, offset, length); } static int ext3_releasepage(struct page *page, gfp_t wait) { journal_t *journal = EXT3_JOURNAL(page->mapping->host); + trace_ext3_releasepage(page); WARN_ON(PageChecked(page)); if (!page_has_buffers(page)) return 0; @@ -1782,8 +1821,7 @@ static int ext3_releasepage(struct page *page, gfp_t wait) * VFS code falls back into buffered path in that case so we are safe. */ static ssize_t ext3_direct_IO(int rw, struct kiocb *iocb, - const struct iovec *iov, loff_t offset, - unsigned long nr_segs) + struct iov_iter *iter, loff_t offset) { struct file *file = iocb->ki_filp; struct inode *inode = file->f_mapping->host; @@ -1791,9 +1829,11 @@ static ssize_t ext3_direct_IO(int rw, struct kiocb *iocb, handle_t *handle; ssize_t ret; int orphan = 0; - size_t count = iov_length(iov, nr_segs); + size_t count = iov_iter_count(iter); int retries = 0; + trace_ext3_direct_IO_enter(inode, offset, count, rw); + if (rw == WRITE) { loff_t final_size = offset + count; @@ -1816,19 +1856,17 @@ static ssize_t ext3_direct_IO(int rw, struct kiocb *iocb, } retry: - ret = blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov, - offset, nr_segs, - ext3_get_block, NULL); + ret = blockdev_direct_IO(rw, iocb, inode, iter, offset, ext3_get_block); /* * In case of error extending write may have instantiated a few * blocks outside i_size. Trim these off again. */ if (unlikely((rw & WRITE) && ret < 0)) { loff_t isize = i_size_read(inode); - loff_t end = offset + iov_length(iov, nr_segs); + loff_t end = offset + count; if (end > isize) - vmtruncate(inode, isize); + ext3_truncate_failed_direct_write(inode); } if (ret == -ENOSPC && ext3_should_retry_alloc(inode->i_sb, &retries)) goto retry; @@ -1842,8 +1880,10 @@ retry: /* This is really bad luck. We've written the data * but cannot extend i_size. Truncate allocated blocks * and pretend the write failed... */ - ext3_truncate(inode); + ext3_truncate_failed_direct_write(inode); ret = PTR_ERR(handle); + if (inode->i_nlink) + ext3_orphan_del(NULL, inode); goto out; } if (inode->i_nlink) @@ -1868,6 +1908,7 @@ retry: ret = err; } out: + trace_ext3_direct_IO_exit(inode, offset, count, rw, ret); return ret; } @@ -1894,7 +1935,6 @@ static const struct address_space_operations ext3_ordered_aops = { .readpage = ext3_readpage, .readpages = ext3_readpages, .writepage = ext3_ordered_writepage, - .sync_page = block_sync_page, .write_begin = ext3_write_begin, .write_end = ext3_ordered_write_end, .bmap = ext3_bmap, @@ -1903,6 +1943,7 @@ static const struct address_space_operations ext3_ordered_aops = { .direct_IO = ext3_direct_IO, .migratepage = buffer_migrate_page, .is_partially_uptodate = block_is_partially_uptodate, + .is_dirty_writeback = buffer_check_dirty_writeback, .error_remove_page = generic_error_remove_page, }; @@ -1910,7 +1951,6 @@ static const struct address_space_operations ext3_writeback_aops = { .readpage = ext3_readpage, .readpages = ext3_readpages, .writepage = ext3_writeback_writepage, - .sync_page = block_sync_page, .write_begin = ext3_write_begin, .write_end = ext3_writeback_write_end, .bmap = ext3_bmap, @@ -1926,7 +1966,6 @@ static const struct address_space_operations ext3_journalled_aops = { .readpage = ext3_readpage, .readpages = ext3_readpages, .writepage = ext3_journalled_writepage, - .sync_page = block_sync_page, .write_begin = ext3_write_begin, .write_end = ext3_journalled_write_end, .set_page_dirty = ext3_journalled_set_page_dirty, @@ -1953,17 +1992,24 @@ void ext3_set_aops(struct inode *inode) * This required during truncate. We need to physically zero the tail end * of that block so it doesn't yield old data if the file is later grown. */ -static int ext3_block_truncate_page(handle_t *handle, struct page *page, - struct address_space *mapping, loff_t from) +static int ext3_block_truncate_page(struct inode *inode, loff_t from) { ext3_fsblk_t index = from >> PAGE_CACHE_SHIFT; - unsigned offset = from & (PAGE_CACHE_SIZE-1); + unsigned offset = from & (PAGE_CACHE_SIZE - 1); unsigned blocksize, iblock, length, pos; - struct inode *inode = mapping->host; + struct page *page; + handle_t *handle = NULL; struct buffer_head *bh; int err = 0; + /* Truncated on block boundary - nothing to do */ blocksize = inode->i_sb->s_blocksize; + if ((from & (blocksize - 1)) == 0) + return 0; + + page = grab_cache_page(inode->i_mapping, index); + if (!page) + return -ENOMEM; length = blocksize - (offset & (blocksize - 1)); iblock = index << (PAGE_CACHE_SHIFT - inode->i_sb->s_blocksize_bits); @@ -1999,20 +2045,30 @@ static int ext3_block_truncate_page(handle_t *handle, struct page *page, if (PageUptodate(page)) set_buffer_uptodate(bh); - if (!buffer_uptodate(bh)) { - err = -EIO; - ll_rw_block(READ, 1, &bh); - wait_on_buffer(bh); + if (!bh_uptodate_or_lock(bh)) { + err = bh_submit_read(bh); /* Uhhuh. Read error. Complain and punt. */ - if (!buffer_uptodate(bh)) + if (err) goto unlock; } + /* data=writeback mode doesn't need transaction to zero-out data */ + if (!ext3_should_writeback_data(inode)) { + /* We journal at most one block */ + handle = ext3_journal_start(inode, 1); + if (IS_ERR(handle)) { + clear_highpage(page); + flush_dcache_page(page); + err = PTR_ERR(handle); + goto unlock; + } + } + if (ext3_should_journal_data(inode)) { BUFFER_TRACE(bh, "get write access"); err = ext3_journal_get_write_access(handle, bh); if (err) - goto unlock; + goto stop; } zero_user(page, offset, length); @@ -2026,6 +2082,9 @@ static int ext3_block_truncate_page(handle_t *handle, struct page *page, err = ext3_journal_dirty_data(handle, bh); mark_buffer_dirty(bh); } +stop: + if (handle) + ext3_journal_stop(handle); unlock: unlock_page(page); @@ -2058,7 +2117,7 @@ static inline int all_zeroes(__le32 *p, __le32 *q) * * When we do truncate() we may have to clean the ends of several * indirect blocks but leave the blocks themselves alive. Block is - * partially truncated if some data below the new i_size is refered + * partially truncated if some data below the new i_size is referred * from it (and it is on the path to the first completely truncated * data block, indeed). We have to free the top of that path along * with everything to the right of the path. Since no allocation @@ -2145,13 +2204,15 @@ static void ext3_clear_blocks(handle_t *handle, struct inode *inode, if (try_to_extend_transaction(handle, inode)) { if (bh) { BUFFER_TRACE(bh, "call ext3_journal_dirty_metadata"); - ext3_journal_dirty_metadata(handle, bh); + if (ext3_journal_dirty_metadata(handle, bh)) + return; } ext3_mark_inode_dirty(handle, inode); truncate_restart_transaction(handle, inode); if (bh) { BUFFER_TRACE(bh, "retaking write access"); - ext3_journal_get_write_access(handle, bh); + if (ext3_journal_get_write_access(handle, bh)) + return; } } @@ -2185,7 +2246,7 @@ static void ext3_clear_blocks(handle_t *handle, struct inode *inode, * @first: array of block numbers * @last: points immediately past the end of array * - * We are freeing all blocks refered from that array (numbers are stored as + * We are freeing all blocks referred from that array (numbers are stored as * little-endian 32-bit) and updating @inode->i_blocks appropriately. * * We accumulate contiguous runs of blocks to free. Conveniently, if these @@ -2273,7 +2334,7 @@ static void ext3_free_data(handle_t *handle, struct inode *inode, * @last: pointer immediately past the end of array * @depth: depth of the branches to free * - * We are freeing all blocks refered from these branches (numbers are + * We are freeing all blocks referred from these branches (numbers are * stored as little-endian 32-bit) and updating @inode->i_blocks * appropriately. */ @@ -2392,8 +2453,6 @@ static void ext3_free_branches(handle_t *handle, struct inode *inode, int ext3_can_truncate(struct inode *inode) { - if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) - return 0; if (S_ISREG(inode->i_mode)) return 1; if (S_ISDIR(inode->i_mode)) @@ -2410,7 +2469,7 @@ int ext3_can_truncate(struct inode *inode) * transaction, and VFS/VM ensures that ext3_truncate() cannot run * simultaneously on behalf of the same inode. * - * As we work through the truncate and commmit bits of it to the journal there + * As we work through the truncate and commit bits of it to the journal there * is one core, guiding principle: the file's tree must always be consistent on * disk. We must be able to restart the truncate after a crash. * @@ -2437,7 +2496,6 @@ void ext3_truncate(struct inode *inode) struct ext3_inode_info *ei = EXT3_I(inode); __le32 *i_data = ei->i_data; int addr_per_block = EXT3_ADDR_PER_BLOCK(inode->i_sb); - struct address_space *mapping = inode->i_mapping; int offsets[4]; Indirect chain[4]; Indirect *partial; @@ -2445,7 +2503,8 @@ void ext3_truncate(struct inode *inode) int n; long last_block; unsigned blocksize = inode->i_sb->s_blocksize; - struct page *page; + + trace_ext3_truncate_enter(inode); if (!ext3_can_truncate(inode)) goto out_notrans; @@ -2453,37 +2512,12 @@ void ext3_truncate(struct inode *inode) if (inode->i_size == 0 && ext3_should_writeback_data(inode)) ext3_set_inode_state(inode, EXT3_STATE_FLUSH_ON_CLOSE); - /* - * We have to lock the EOF page here, because lock_page() nests - * outside journal_start(). - */ - if ((inode->i_size & (blocksize - 1)) == 0) { - /* Block boundary? Nothing to do */ - page = NULL; - } else { - page = grab_cache_page(mapping, - inode->i_size >> PAGE_CACHE_SHIFT); - if (!page) - goto out_notrans; - } - handle = start_transaction(inode); - if (IS_ERR(handle)) { - if (page) { - clear_highpage(page); - flush_dcache_page(page); - unlock_page(page); - page_cache_release(page); - } + if (IS_ERR(handle)) goto out_notrans; - } last_block = (inode->i_size + blocksize-1) >> EXT3_BLOCK_SIZE_BITS(inode->i_sb); - - if (page) - ext3_block_truncate_page(handle, page, mapping, inode->i_size); - n = ext3_block_to_path(inode, last_block, offsets, NULL); if (n == 0) goto out_stop; /* error */ @@ -2598,6 +2632,7 @@ out_stop: ext3_orphan_del(handle, inode); ext3_journal_stop(handle); + trace_ext3_truncate_exit(inode); return; out_notrans: /* @@ -2606,6 +2641,7 @@ out_notrans: */ if (inode->i_nlink) ext3_orphan_del(NULL, inode); + trace_ext3_truncate_exit(inode); } static ext3_fsblk_t ext3_get_inode_block(struct super_block *sb, @@ -2659,12 +2695,12 @@ static int __ext3_get_inode_loc(struct inode *inode, return -EIO; bh = sb_getblk(inode->i_sb, block); - if (!bh) { + if (unlikely(!bh)) { ext3_error (inode->i_sb, "ext3_get_inode_loc", "unable to read inode block - " "inode=%lu, block="E3FSBLK, inode->i_ino, block); - return -EIO; + return -ENOMEM; } if (!buffer_uptodate(bh)) { lock_buffer(bh); @@ -2713,7 +2749,7 @@ static int __ext3_get_inode_loc(struct inode *inode, bitmap_bh = sb_getblk(inode->i_sb, le32_to_cpu(desc->bg_inode_bitmap)); - if (!bitmap_bh) + if (unlikely(!bitmap_bh)) goto make_io; /* @@ -2747,9 +2783,10 @@ make_io: * has in-inode xattrs, or we don't have this inode in memory. * Read the block from disk. */ + trace_ext3_load_inode(inode); get_bh(bh); bh->b_end_io = end_buffer_read_sync; - submit_bh(READ_META, bh); + submit_bh(READ | REQ_META | REQ_PRIO, bh); wait_on_buffer(bh); if (!buffer_uptodate(bh)) { ext3_error(inode->i_sb, "ext3_get_inode_loc", @@ -2819,6 +2856,8 @@ struct inode *ext3_iget(struct super_block *sb, unsigned long ino) transaction_t *transaction; long ret; int block; + uid_t i_uid; + gid_t i_gid; inode = iget_locked(sb, ino); if (!inode) @@ -2835,13 +2874,15 @@ struct inode *ext3_iget(struct super_block *sb, unsigned long ino) bh = iloc.bh; raw_inode = ext3_raw_inode(&iloc); inode->i_mode = le16_to_cpu(raw_inode->i_mode); - inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low); - inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low); + i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low); + i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low); if(!(test_opt (inode->i_sb, NO_UID32))) { - inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16; - inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16; + i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16; + i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16; } - inode->i_nlink = le16_to_cpu(raw_inode->i_links_count); + i_uid_write(inode, i_uid); + i_gid_write(inode, i_gid); + set_nlink(inode, le16_to_cpu(raw_inode->i_links_count)); inode->i_size = le32_to_cpu(raw_inode->i_size); inode->i_atime.tv_sec = (signed)le32_to_cpu(raw_inode->i_atime); inode->i_ctime.tv_sec = (signed)le32_to_cpu(raw_inode->i_ctime); @@ -2996,6 +3037,10 @@ static int ext3_do_update_inode(handle_t *handle, struct ext3_inode_info *ei = EXT3_I(inode); struct buffer_head *bh = iloc->bh; int err = 0, rc, block; + int need_datasync = 0; + __le32 disksize; + uid_t i_uid; + gid_t i_gid; again: /* we can't allow multiple procs in here at once, its a bit racey */ @@ -3008,32 +3053,38 @@ again: ext3_get_inode_flags(ei); raw_inode->i_mode = cpu_to_le16(inode->i_mode); + i_uid = i_uid_read(inode); + i_gid = i_gid_read(inode); if(!(test_opt(inode->i_sb, NO_UID32))) { - raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid)); - raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid)); + raw_inode->i_uid_low = cpu_to_le16(low_16_bits(i_uid)); + raw_inode->i_gid_low = cpu_to_le16(low_16_bits(i_gid)); /* * Fix up interoperability with old kernels. Otherwise, old inodes get * re-used with the upper 16 bits of the uid/gid intact */ if(!ei->i_dtime) { raw_inode->i_uid_high = - cpu_to_le16(high_16_bits(inode->i_uid)); + cpu_to_le16(high_16_bits(i_uid)); raw_inode->i_gid_high = - cpu_to_le16(high_16_bits(inode->i_gid)); + cpu_to_le16(high_16_bits(i_gid)); } else { raw_inode->i_uid_high = 0; raw_inode->i_gid_high = 0; } } else { raw_inode->i_uid_low = - cpu_to_le16(fs_high2lowuid(inode->i_uid)); + cpu_to_le16(fs_high2lowuid(i_uid)); raw_inode->i_gid_low = - cpu_to_le16(fs_high2lowgid(inode->i_gid)); + cpu_to_le16(fs_high2lowgid(i_gid)); raw_inode->i_uid_high = 0; raw_inode->i_gid_high = 0; } raw_inode->i_links_count = cpu_to_le16(inode->i_nlink); - raw_inode->i_size = cpu_to_le32(ei->i_disksize); + disksize = cpu_to_le32(ei->i_disksize); + if (disksize != raw_inode->i_size) { + need_datasync = 1; + raw_inode->i_size = disksize; + } raw_inode->i_atime = cpu_to_le32(inode->i_atime.tv_sec); raw_inode->i_ctime = cpu_to_le32(inode->i_ctime.tv_sec); raw_inode->i_mtime = cpu_to_le32(inode->i_mtime.tv_sec); @@ -3049,8 +3100,11 @@ again: if (!S_ISREG(inode->i_mode)) { raw_inode->i_dir_acl = cpu_to_le32(ei->i_dir_acl); } else { - raw_inode->i_size_high = - cpu_to_le32(ei->i_disksize >> 32); + disksize = cpu_to_le32(ei->i_disksize >> 32); + if (disksize != raw_inode->i_size_high) { + raw_inode->i_size_high = disksize; + need_datasync = 1; + } if (ei->i_disksize > 0x7fffffffULL) { struct super_block *sb = inode->i_sb; if (!EXT3_HAS_RO_COMPAT_FEATURE(sb, @@ -3103,6 +3157,8 @@ again: ext3_clear_inode_state(inode, EXT3_STATE_NEW); atomic_set(&ei->i_sync_tid, handle->h_transaction->t_tid); + if (need_datasync) + atomic_set(&ei->i_datasync_tid, handle->h_transaction->t_tid); out_brelse: brelse (bh); ext3_std_error(inode->i_sb, err); @@ -3114,21 +3170,20 @@ out_brelse: * * We are called from a few places: * - * - Within generic_file_write() for O_SYNC files. + * - Within generic_file_aio_write() -> generic_write_sync() for O_SYNC files. * Here, there will be no transaction running. We wait for any running - * trasnaction to commit. + * transaction to commit. * - * - Within sys_sync(), kupdate and such. - * We wait on commit, if tol to. + * - Within flush work (for sys_sync(), kupdate and such). + * We wait on commit, if told to. * - * - Within prune_icache() (PF_MEMALLOC == true) - * Here we simply return. We can't afford to block kswapd on the - * journal commit. + * - Within iput_final() -> write_inode_now() + * We wait on commit, if told to. * * In all cases it is actually safe for us to return without doing anything, * because the inode has been copied into a raw inode buffer in - * ext3_mark_inode_dirty(). This is a correctness thing for O_SYNC and for - * knfsd. + * ext3_mark_inode_dirty(). This is a correctness thing for WB_SYNC_ALL + * writeback. * * Note that we are absolutely dependent upon all inode dirtiers doing the * right thing: they *must* call mark_inode_dirty() after dirtying info in @@ -3140,13 +3195,13 @@ out_brelse: * stuff(); * inode->i_size = expr; * - * is in error because a kswapd-driven write_inode() could occur while - * `stuff()' is running, and the new i_size will be lost. Plus the inode - * will no longer be on the superblock's dirty inode list. + * is in error because write_inode() could occur while `stuff()' is running, + * and the new i_size will be lost. Plus the inode will no longer be on the + * superblock's dirty inode list. */ int ext3_write_inode(struct inode *inode, struct writeback_control *wbc) { - if (current->flags & PF_MEMALLOC) + if (WARN_ON_ONCE(current->flags & PF_MEMALLOC)) return 0; if (ext3_journal_current_handle()) { @@ -3155,7 +3210,12 @@ int ext3_write_inode(struct inode *inode, struct writeback_control *wbc) return -EIO; } - if (wbc->sync_mode != WB_SYNC_ALL) + /* + * No need to force transaction in WB_SYNC_NONE mode. Also + * ext3_sync_fs() will force the commit after everything is + * written. + */ + if (wbc->sync_mode != WB_SYNC_ALL || wbc->for_sync) return 0; return ext3_force_commit(inode->i_sb); @@ -3190,8 +3250,8 @@ int ext3_setattr(struct dentry *dentry, struct iattr *attr) if (is_quota_modification(inode, attr)) dquot_initialize(inode); - if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) || - (ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) { + if ((ia_valid & ATTR_UID && !uid_eq(attr->ia_uid, inode->i_uid)) || + (ia_valid & ATTR_GID && !gid_eq(attr->ia_gid, inode->i_gid))) { handle_t *handle; /* (user+group)*(old+new) structure, inode write (sb, @@ -3217,6 +3277,9 @@ int ext3_setattr(struct dentry *dentry, struct iattr *attr) ext3_journal_stop(handle); } + if (attr->ia_valid & ATTR_SIZE) + inode_dio_wait(inode); + if (S_ISREG(inode->i_mode) && attr->ia_valid & ATTR_SIZE && attr->ia_size < inode->i_size) { handle_t *handle; @@ -3228,25 +3291,43 @@ int ext3_setattr(struct dentry *dentry, struct iattr *attr) } error = ext3_orphan_add(handle, inode); + if (error) { + ext3_journal_stop(handle); + goto err_out; + } EXT3_I(inode)->i_disksize = attr->ia_size; - rc = ext3_mark_inode_dirty(handle, inode); - if (!error) - error = rc; + error = ext3_mark_inode_dirty(handle, inode); ext3_journal_stop(handle); + if (error) { + /* Some hard fs error must have happened. Bail out. */ + ext3_orphan_del(NULL, inode); + goto err_out; + } + rc = ext3_block_truncate_page(inode, attr->ia_size); + if (rc) { + /* Cleanup orphan list and exit */ + handle = ext3_journal_start(inode, 3); + if (IS_ERR(handle)) { + ext3_orphan_del(NULL, inode); + goto err_out; + } + ext3_orphan_del(handle, inode); + ext3_journal_stop(handle); + goto err_out; + } } if ((attr->ia_valid & ATTR_SIZE) && attr->ia_size != i_size_read(inode)) { - rc = vmtruncate(inode, attr->ia_size); - if (rc) - goto err_out; + truncate_setsize(inode, attr->ia_size); + ext3_truncate(inode); } setattr_copy(inode, attr); mark_inode_dirty(inode); if (ia_valid & ATTR_MODE) - rc = ext3_acl_chmod(inode); + rc = posix_acl_chmod(inode, inode->i_mode); err_out: ext3_std_error(inode->i_sb, error); @@ -3292,7 +3373,7 @@ static int ext3_writepage_trans_blocks(struct inode *inode) if (ext3_should_journal_data(inode)) ret = 3 * (bpp + indirects) + 2; else - ret = 2 * (bpp + indirects) + 2; + ret = 2 * (bpp + indirects) + indirects + 2; #ifdef CONFIG_QUOTA /* We know that structure was already allocated during dquot_initialize so @@ -3358,14 +3439,6 @@ ext3_reserve_inode_write(handle_t *handle, struct inode *inode, * inode out, but prune_icache isn't a user-visible syncing function. * Whenever the user wants stuff synced (sys_sync, sys_msync, sys_fsync) * we start and wait on commits. - * - * Is this efficient/effective? Well, we're being nice to the system - * by cleaning up our inodes proactively so they can be reaped - * without I/O. But we are potentially leaving up to five seconds' - * worth of inodes floating about which prune_icache wants us to - * write out. One way to fix that would be to get prune_icache() - * to do a write_super() to free up some memory. It has the desired - * effect. */ int ext3_mark_inode_dirty(handle_t *handle, struct inode *inode) { @@ -3373,6 +3446,7 @@ int ext3_mark_inode_dirty(handle_t *handle, struct inode *inode) int err; might_sleep(); + trace_ext3_mark_inode_dirty(inode, _RET_IP_); err = ext3_reserve_inode_write(handle, inode, &iloc); if (!err) err = ext3_mark_iloc_dirty(handle, inode, &iloc); @@ -3393,7 +3467,7 @@ int ext3_mark_inode_dirty(handle_t *handle, struct inode *inode) * so would cause a commit on atime updates, which we don't bother doing. * We handle synchronous inodes at the highest possible level. */ -void ext3_dirty_inode(struct inode *inode) +void ext3_dirty_inode(struct inode *inode, int flags) { handle_t *current_handle = ext3_journal_current_handle(); handle_t *handle; |