From 84ebd795613488992b273220c2937d575d27d2a9 Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Wed, 31 Aug 2011 11:56:51 -0400 Subject: ext4: fake direct I/O mode for data=journal Currently attempts to open a file with O_DIRECT in data=journal mode causes the open to fail with -EINVAL. This makes it very hard to test data=journal mode. So we will let the open succeed, but then always fall back to O_DSYNC buffered writes. Signed-off-by: "Theodore Ts'o" --- fs/ext4/inode.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 18d2558b762..b84f127c085 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -2854,6 +2854,12 @@ static ssize_t ext4_direct_IO(int rw, struct kiocb *iocb, struct inode *inode = file->f_mapping->host; ssize_t ret; + /* + * If we are doing data journalling we don't support O_DIRECT + */ + if (ext4_should_journal_data(inode)) + return 0; + trace_ext4_direct_IO_enter(inode, offset, iov_length(iov, nr_segs), rw); if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) ret = ext4_ext_direct_IO(rw, iocb, iov, offset, nr_segs); @@ -2923,6 +2929,7 @@ static const struct address_space_operations ext4_journalled_aops = { .bmap = ext4_bmap, .invalidatepage = ext4_invalidatepage, .releasepage = ext4_releasepage, + .direct_IO = ext4_direct_IO, .is_partially_uptodate = block_is_partially_uptodate, .error_remove_page = generic_error_remove_page, }; -- cgit v1.2.3 From 4e96b2dbbf1d7e81f22047a50f862555a6cb87cb Mon Sep 17 00:00:00 2001 From: Allison Henderson Date: Sat, 3 Sep 2011 11:51:09 -0400 Subject: ext4: Add new ext4_discard_partial_page_buffers routines This patch adds two new routines: ext4_discard_partial_page_buffers and ext4_discard_partial_page_buffers_no_lock. The ext4_discard_partial_page_buffers routine is a wrapper function to ext4_discard_partial_page_buffers_no_lock. The wrapper function locks the page and passes it to ext4_discard_partial_page_buffers_no_lock. Calling functions that already have the page locked can call ext4_discard_partial_page_buffers_no_lock directly. The ext4_discard_partial_page_buffers_no_lock function zeros a specified range in a page, and unmaps the corresponding buffer heads. Only block aligned regions of the page will have their buffer heads unmapped. Unblock aligned regions will be mapped if needed so that they can be updated with the partial zero out. This function is meant to be used to update a page and its buffer heads to be zeroed and unmapped when the corresponding blocks have been released or will be released. This routine is used in the following scenarios: * A hole is punched and the non page aligned regions of the head and tail of the hole need to be discarded * The file is truncated and the partial page beyond EOF needs to be discarded * The end of a hole is in the same page as EOF. After the page is flushed, the partial page beyond EOF needs to be discarded. * A write operation begins or ends inside a hole and the partial page appearing before or after the write needs to be discarded * A write operation extends EOF and the partial page beyond EOF needs to be discarded This function takes a flag EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED which is used when a write operation begins or ends in a hole. When the EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED flag is used, only buffer heads that are already unmapped will have the corresponding regions of the page zeroed. Signed-off-by: Allison Henderson Signed-off-by: "Theodore Ts'o" --- fs/ext4/ext4.h | 11 +++ fs/ext4/inode.c | 224 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 235 insertions(+) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 5c38120c389..ccfa81f33bb 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -528,6 +528,11 @@ struct ext4_new_group_data { #define EXT4_FREE_BLOCKS_VALIDATED 0x0004 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008 +/* + * Flags used by ext4_discard_partial_page_buffers + */ +#define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001 + /* * ioctl commands */ @@ -1838,6 +1843,12 @@ extern int ext4_block_truncate_page(handle_t *handle, struct address_space *mapping, loff_t from); extern int ext4_block_zero_page_range(handle_t *handle, struct address_space *mapping, loff_t from, loff_t length); +extern int ext4_discard_partial_page_buffers(handle_t *handle, + struct address_space *mapping, loff_t from, + loff_t length, int flags); +extern int ext4_discard_partial_page_buffers_no_lock(handle_t *handle, + struct inode *inode, struct page *page, loff_t from, + loff_t length, int flags); extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); extern qsize_t *ext4_get_reserved_space(struct inode *inode); extern void ext4_da_update_reserve_space(struct inode *inode, diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index b84f127c085..d1b1ef71e5b 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -2966,6 +2966,230 @@ void ext4_set_aops(struct inode *inode) inode->i_mapping->a_ops = &ext4_journalled_aops; } + +/* + * ext4_discard_partial_page_buffers() + * Wrapper function for ext4_discard_partial_page_buffers_no_lock. + * This function finds and locks the page containing the offset + * "from" and passes it to ext4_discard_partial_page_buffers_no_lock. + * Calling functions that already have the page locked should call + * ext4_discard_partial_page_buffers_no_lock directly. + */ +int ext4_discard_partial_page_buffers(handle_t *handle, + struct address_space *mapping, loff_t from, + loff_t length, int flags) +{ + struct inode *inode = mapping->host; + struct page *page; + int err = 0; + + page = find_or_create_page(mapping, from >> PAGE_CACHE_SHIFT, + mapping_gfp_mask(mapping) & ~__GFP_FS); + if (!page) + return -EINVAL; + + err = ext4_discard_partial_page_buffers_no_lock(handle, inode, page, + from, length, flags); + + unlock_page(page); + page_cache_release(page); + return err; +} + +/* + * ext4_discard_partial_page_buffers_no_lock() + * Zeros a page range of length 'length' starting from offset 'from'. + * Buffer heads that correspond to the block aligned regions of the + * zeroed range will be unmapped. Unblock aligned regions + * will have the corresponding buffer head mapped if needed so that + * that region of the page can be updated with the partial zero out. + * + * This function assumes that the page has already been locked. The + * The range to be discarded must be contained with in the given page. + * If the specified range exceeds the end of the page it will be shortened + * to the end of the page that corresponds to 'from'. This function is + * appropriate for updating a page and it buffer heads to be unmapped and + * zeroed for blocks that have been either released, or are going to be + * released. + * + * handle: The journal handle + * inode: The files inode + * page: A locked page that contains the offset "from" + * from: The starting byte offset (from the begining of the file) + * to begin discarding + * len: The length of bytes to discard + * flags: Optional flags that may be used: + * + * EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED + * Only zero the regions of the page whose buffer heads + * have already been unmapped. This flag is appropriate + * for updateing the contents of a page whose blocks may + * have already been released, and we only want to zero + * out the regions that correspond to those released blocks. + * + * Returns zero on sucess or negative on failure. + */ +int ext4_discard_partial_page_buffers_no_lock(handle_t *handle, + struct inode *inode, struct page *page, loff_t from, + loff_t length, int flags) +{ + ext4_fsblk_t index = from >> PAGE_CACHE_SHIFT; + unsigned int offset = from & (PAGE_CACHE_SIZE-1); + unsigned int blocksize, max, pos; + unsigned int end_of_block, range_to_discard; + ext4_lblk_t iblock; + struct buffer_head *bh; + int err = 0; + + blocksize = inode->i_sb->s_blocksize; + max = PAGE_CACHE_SIZE - offset; + + if (index != page->index) + return -EINVAL; + + /* + * correct length if it does not fall between + * 'from' and the end of the page + */ + if (length > max || length < 0) + length = max; + + iblock = index << (PAGE_CACHE_SHIFT - inode->i_sb->s_blocksize_bits); + + if (!page_has_buffers(page)) { + /* + * If the range to be discarded covers a partial block + * we need to get the page buffers. This is because + * partial blocks cannot be released and the page needs + * to be updated with the contents of the block before + * we write the zeros on top of it. + */ + if (!(from & (blocksize - 1)) || + !((from + length) & (blocksize - 1))) { + create_empty_buffers(page, blocksize, 0); + } else { + /* + * If there are no partial blocks, + * there is nothing to update, + * so we can return now + */ + return 0; + } + } + + /* Find the buffer that contains "offset" */ + bh = page_buffers(page); + pos = blocksize; + while (offset >= pos) { + bh = bh->b_this_page; + iblock++; + pos += blocksize; + } + + pos = offset; + while (pos < offset + length) { + err = 0; + + /* The length of space left to zero and unmap */ + range_to_discard = offset + length - pos; + + /* The length of space until the end of the block */ + end_of_block = blocksize - (pos & (blocksize-1)); + + /* + * Do not unmap or zero past end of block + * for this buffer head + */ + if (range_to_discard > end_of_block) + range_to_discard = end_of_block; + + + /* + * Skip this buffer head if we are only zeroing unampped + * regions of the page + */ + if (flags & EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED && + buffer_mapped(bh)) + goto next; + + /* If the range is block aligned, unmap */ + if (range_to_discard == blocksize) { + clear_buffer_dirty(bh); + bh->b_bdev = NULL; + clear_buffer_mapped(bh); + clear_buffer_req(bh); + clear_buffer_new(bh); + clear_buffer_delay(bh); + clear_buffer_unwritten(bh); + clear_buffer_uptodate(bh); + zero_user(page, pos, range_to_discard); + BUFFER_TRACE(bh, "Buffer discarded"); + goto next; + } + + /* + * If this block is not completely contained in the range + * to be discarded, then it is not going to be released. Because + * we need to keep this block, we need to make sure this part + * of the page is uptodate before we modify it by writeing + * partial zeros on it. + */ + if (!buffer_mapped(bh)) { + /* + * Buffer head must be mapped before we can read + * from the block + */ + BUFFER_TRACE(bh, "unmapped"); + ext4_get_block(inode, iblock, bh, 0); + /* unmapped? It's a hole - nothing to do */ + if (!buffer_mapped(bh)) { + BUFFER_TRACE(bh, "still unmapped"); + goto next; + } + } + + /* Ok, it's mapped. Make sure it's up-to-date */ + if (PageUptodate(page)) + set_buffer_uptodate(bh); + + if (!buffer_uptodate(bh)) { + err = -EIO; + ll_rw_block(READ, 1, &bh); + wait_on_buffer(bh); + /* Uhhuh. Read error. Complain and punt.*/ + if (!buffer_uptodate(bh)) + goto next; + } + + if (ext4_should_journal_data(inode)) { + BUFFER_TRACE(bh, "get write access"); + err = ext4_journal_get_write_access(handle, bh); + if (err) + goto next; + } + + zero_user(page, pos, range_to_discard); + + err = 0; + if (ext4_should_journal_data(inode)) { + err = ext4_handle_dirty_metadata(handle, inode, bh); + } else { + if (ext4_should_order_data(inode) && + EXT4_I(inode)->jinode) + err = ext4_jbd2_file_inode(handle, inode); + mark_buffer_dirty(bh); + } + + BUFFER_TRACE(bh, "Partial buffer zeroed"); +next: + bh = bh->b_this_page; + iblock++; + pos += range_to_discard; + } + + return err; +} + /* * ext4_block_truncate_page() zeroes out a mapping from file offset `from' * up to the end of the block which corresponds to `from'. -- cgit v1.2.3 From decbd919f4bb9cb698486880c026c4104b13d3c3 Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Tue, 6 Sep 2011 02:37:06 -0400 Subject: ext4: only call ext4_jbd2_file_inode when an inode has been extended In delayed allocation mode, it's important to only call ext4_jbd2_file_inode when the file has been extended. This is necessary to avoid a race which first got introduced in commit 678aaf481, but which was made much more common with the introduction of the "punch hole" functionality. (Especially when dioread_nolock was enabled; when I could reliably reproduce this problem with xfstests #74.) The race is this: If while trying to writeback a delayed allocation inode, there is a need to map delalloc blocks, and we run out of space in the journal, *and* at the same time the inode is already on the committing transaction's t_inode_list (because for example while doing the punch hole operation, ext4_jbd2_file_inode() is called), then the commit operation will wait for the inode to finish all of its pending writebacks by calling filemap_fdatawait(), but since that inode has one or more pages with the PageWriteback flag set, the commit operation will wait forever, and the so the writeback of the inode can never take place, and the kjournald thread and the writeback thread end up waiting for each other --- forever. It's important at this point to recall why an inode is placed on the t_inode_list; it is to provide the data=ordered guarantees that we don't end up exposing stale data. In the case where we are truncating or punching a hole in the inode, there is no possibility that stale data could be exposed in the first place, so we don't need to put the inode on the t_inode_list! The right long-term fix is to get rid of data=ordered mode altogether, and only update the extent tree or indirect blocks after the data has been written. Until then, this change will also avoid some unnecessary waiting in the commit operation. Signed-off-by: "Theodore Ts'o" Cc: Allison Henderson Cc: Jan Kara --- fs/ext4/inode.c | 23 ++++++++--------------- 1 file changed, 8 insertions(+), 15 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index d1b1ef71e5b..f86b149fb8b 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -1471,13 +1471,13 @@ static void mpage_da_map_and_submit(struct mpage_da_data *mpd) for (i = 0; i < map.m_len; i++) unmap_underlying_metadata(bdev, map.m_pblk + i); - } - if (ext4_should_order_data(mpd->inode)) { - err = ext4_jbd2_file_inode(handle, mpd->inode); - if (err) - /* This only happens if the journal is aborted */ - return; + if (ext4_should_order_data(mpd->inode)) { + err = ext4_jbd2_file_inode(handle, mpd->inode); + if (err) + /* Only if the journal is aborted */ + return; + } } /* @@ -3173,12 +3173,8 @@ int ext4_discard_partial_page_buffers_no_lock(handle_t *handle, err = 0; if (ext4_should_journal_data(inode)) { err = ext4_handle_dirty_metadata(handle, inode, bh); - } else { - if (ext4_should_order_data(inode) && - EXT4_I(inode)->jinode) - err = ext4_jbd2_file_inode(handle, inode); + } else mark_buffer_dirty(bh); - } BUFFER_TRACE(bh, "Partial buffer zeroed"); next: @@ -3301,11 +3297,8 @@ int ext4_block_zero_page_range(handle_t *handle, err = 0; if (ext4_should_journal_data(inode)) { err = ext4_handle_dirty_metadata(handle, inode, bh); - } else { - if (ext4_should_order_data(inode) && EXT4_I(inode)->jinode) - err = ext4_jbd2_file_inode(handle, inode); + } else mark_buffer_dirty(bh); - } unlock: unlock_page(page); -- cgit v1.2.3 From 02fac1297eb3f471a27368271aadd285548297b0 Mon Sep 17 00:00:00 2001 From: Allison Henderson Date: Tue, 6 Sep 2011 21:53:01 -0400 Subject: ext4: fix partial page writes While running extended fsx tests to verify the preceeding patches, a similar bug was also found in the write operation When ever a write operation begins or ends in a hole, or extends EOF, the partial page contained in the hole or beyond EOF needs to be zeroed out. To correct this the new ext4_discard_partial_page_buffers_no_lock routine is used to zero out the partial page, but only for buffer heads that are already unmapped. Signed-off-by: Allison Henderson Signed-off-by: "Theodore Ts'o" --- fs/ext4/inode.c | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index f86b149fb8b..6ecc93979e4 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -2255,6 +2255,7 @@ static int ext4_da_write_begin(struct file *file, struct address_space *mapping, pgoff_t index; struct inode *inode = mapping->host; handle_t *handle; + loff_t page_len; index = pos >> PAGE_CACHE_SHIFT; @@ -2301,6 +2302,13 @@ retry: */ if (pos + len > inode->i_size) ext4_truncate_failed_write(inode); + } else { + page_len = pos & (PAGE_CACHE_SIZE - 1); + if (page_len > 0) { + ret = ext4_discard_partial_page_buffers_no_lock(handle, + inode, page, pos - page_len, page_len, + EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED); + } } if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries)) @@ -2343,6 +2351,7 @@ static int ext4_da_write_end(struct file *file, loff_t new_i_size; unsigned long start, end; int write_mode = (int)(unsigned long)fsdata; + loff_t page_len; if (write_mode == FALL_BACK_TO_NONDELALLOC) { if (ext4_should_order_data(inode)) { @@ -2391,6 +2400,16 @@ static int ext4_da_write_end(struct file *file, } ret2 = generic_write_end(file, mapping, pos, len, copied, page, fsdata); + + page_len = PAGE_CACHE_SIZE - + ((pos + copied - 1) & (PAGE_CACHE_SIZE - 1)); + + if (page_len > 0) { + ret = ext4_discard_partial_page_buffers_no_lock(handle, + inode, page, pos + copied - 1, page_len, + EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED); + } + copied = ret2; if (ret2 < 0) ret = ret2; -- cgit v1.2.3 From bab08ab9646288f1b0b72a7aaeecdff94bd62c18 Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Fri, 9 Sep 2011 18:36:51 -0400 Subject: ext4: enforce bigalloc restrictions (e.g., no online resizing, etc.) At least initially if the bigalloc feature is enabled, we will not support non-extent mapped inodes, online resizing, online defrag, or the FITRIM ioctl. This simplifies the initial implementation. Signed-off-by: "Theodore Ts'o" --- fs/ext4/indirect.c | 7 +++++++ fs/ext4/inode.c | 5 +++++ fs/ext4/ioctl.c | 33 +++++++++++++++++++++++++++++---- fs/ext4/super.c | 7 +++++++ 4 files changed, 48 insertions(+), 4 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/indirect.c b/fs/ext4/indirect.c index ea704dff866..3cfc73fbca8 100644 --- a/fs/ext4/indirect.c +++ b/fs/ext4/indirect.c @@ -699,6 +699,13 @@ int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, /* * Okay, we need to do block allocation. */ + if (EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb, + EXT4_FEATURE_RO_COMPAT_BIGALLOC)) { + EXT4_ERROR_INODE(inode, "Can't allocate blocks for " + "non-extent mapped inodes with bigalloc"); + return -ENOSPC; + } + goal = ext4_find_goal(inode, map->m_lblk, partial); /* the number of blocks need to allocate for [d,t]indirect blocks */ diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 6ecc93979e4..904a9a623da 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -3358,6 +3358,11 @@ int ext4_punch_hole(struct file *file, loff_t offset, loff_t length) return -ENOTSUPP; } + if (EXT4_SB(inode->i_sb)->s_cluster_ratio > 1) { + /* TODO: Add support for bigalloc file systems */ + return -ENOTSUPP; + } + return ext4_ext_punch_hole(file, offset, length); } diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c index f18bfe37aff..2046d699b4a 100644 --- a/fs/ext4/ioctl.c +++ b/fs/ext4/ioctl.c @@ -21,6 +21,7 @@ long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { struct inode *inode = filp->f_dentry->d_inode; + struct super_block *sb = inode->i_sb; struct ext4_inode_info *ei = EXT4_I(inode); unsigned int flags; @@ -183,7 +184,6 @@ setversion_out: * Returns 1 if it slept, else zero. */ { - struct super_block *sb = inode->i_sb; DECLARE_WAITQUEUE(wait, current); int ret = 0; @@ -199,7 +199,6 @@ setversion_out: #endif case EXT4_IOC_GROUP_EXTEND: { ext4_fsblk_t n_blocks_count; - struct super_block *sb = inode->i_sb; int err, err2=0; err = ext4_resize_begin(sb); @@ -209,6 +208,13 @@ setversion_out: if (get_user(n_blocks_count, (__u32 __user *)arg)) return -EFAULT; + if (EXT4_HAS_RO_COMPAT_FEATURE(sb, + EXT4_FEATURE_RO_COMPAT_BIGALLOC)) { + ext4_msg(sb, KERN_ERR, + "Online resizing not supported with bigalloc"); + return -EOPNOTSUPP; + } + err = mnt_want_write(filp->f_path.mnt); if (err) return err; @@ -250,6 +256,13 @@ setversion_out: goto mext_out; } + if (EXT4_HAS_RO_COMPAT_FEATURE(sb, + EXT4_FEATURE_RO_COMPAT_BIGALLOC)) { + ext4_msg(sb, KERN_ERR, + "Online defrag not supported with bigalloc"); + return -EOPNOTSUPP; + } + err = mnt_want_write(filp->f_path.mnt); if (err) goto mext_out; @@ -270,7 +283,6 @@ mext_out: case EXT4_IOC_GROUP_ADD: { struct ext4_new_group_data input; - struct super_block *sb = inode->i_sb; int err, err2=0; err = ext4_resize_begin(sb); @@ -281,6 +293,13 @@ mext_out: sizeof(input))) return -EFAULT; + if (EXT4_HAS_RO_COMPAT_FEATURE(sb, + EXT4_FEATURE_RO_COMPAT_BIGALLOC)) { + ext4_msg(sb, KERN_ERR, + "Online resizing not supported with bigalloc"); + return -EOPNOTSUPP; + } + err = mnt_want_write(filp->f_path.mnt); if (err) return err; @@ -337,7 +356,6 @@ mext_out: case FITRIM: { - struct super_block *sb = inode->i_sb; struct request_queue *q = bdev_get_queue(sb->s_bdev); struct fstrim_range range; int ret = 0; @@ -348,6 +366,13 @@ mext_out: if (!blk_queue_discard(q)) return -EOPNOTSUPP; + if (EXT4_HAS_RO_COMPAT_FEATURE(sb, + EXT4_FEATURE_RO_COMPAT_BIGALLOC)) { + ext4_msg(sb, KERN_ERR, + "FITRIM not supported with bigalloc"); + return -EOPNOTSUPP; + } + if (copy_from_user(&range, (struct fstrim_range *)arg, sizeof(range))) return -EFAULT; diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 823e7d9deee..25a4bfe3f39 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -2701,6 +2701,13 @@ static int ext4_feature_set_ok(struct super_block *sb, int readonly) return 0; } } + if (EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_BIGALLOC) && + !EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_EXTENTS)) { + ext4_msg(sb, KERN_ERR, + "Can't support bigalloc feature without " + "extents feature\n"); + return 0; + } return 1; } -- cgit v1.2.3 From 5704265188ffe4290ed73b3cb685206c3ed8209d Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Fri, 9 Sep 2011 18:56:51 -0400 Subject: ext4: convert s_{dirty,free}blocks_counter to s_{dirty,free}clusters_counter Convert the percpu counters s_dirtyblocks_counter and s_freeblocks_counter in struct ext4_super_info to be s_dirtyclusters_counter and s_freeclusters_counter. Signed-off-by: "Theodore Ts'o" --- fs/ext4/balloc.c | 10 +++++----- fs/ext4/ext4.h | 5 +++-- fs/ext4/ialloc.c | 3 ++- fs/ext4/inode.c | 18 ++++++++++-------- fs/ext4/mballoc.c | 12 +++++++----- fs/ext4/resize.c | 4 ++-- fs/ext4/super.c | 29 +++++++++++++++-------------- 7 files changed, 44 insertions(+), 37 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c index 89abf1f7b25..9080a857cda 100644 --- a/fs/ext4/balloc.c +++ b/fs/ext4/balloc.c @@ -414,16 +414,16 @@ static int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks, unsigned int flags) { s64 free_blocks, dirty_blocks, root_blocks; - struct percpu_counter *fbc = &sbi->s_freeblocks_counter; - struct percpu_counter *dbc = &sbi->s_dirtyblocks_counter; + struct percpu_counter *fcc = &sbi->s_freeclusters_counter; + struct percpu_counter *dbc = &sbi->s_dirtyclusters_counter; - free_blocks = percpu_counter_read_positive(fbc); + free_blocks = percpu_counter_read_positive(fcc); dirty_blocks = percpu_counter_read_positive(dbc); root_blocks = ext4_r_blocks_count(sbi->s_es); if (free_blocks - (nblocks + root_blocks + dirty_blocks) < EXT4_FREEBLOCKS_WATERMARK) { - free_blocks = percpu_counter_sum_positive(fbc); + free_blocks = EXT4_C2B(sbi, percpu_counter_sum_positive(fcc)); dirty_blocks = percpu_counter_sum_positive(dbc); } /* Check whether we have space after @@ -449,7 +449,7 @@ int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks, unsigned int flags) { if (ext4_has_free_blocks(sbi, nblocks, flags)) { - percpu_counter_add(&sbi->s_dirtyblocks_counter, nblocks); + percpu_counter_add(&sbi->s_dirtyclusters_counter, nblocks); return 0; } else return -ENOSPC; diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 030bfc1cb59..c7588366471 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -855,6 +855,7 @@ struct ext4_inode_info { ext4_group_t i_last_alloc_group; /* allocation reservation info for delalloc */ + /* In case of bigalloc, these refer to clusters rather than blocks */ unsigned int i_reserved_data_blocks; unsigned int i_reserved_meta_blocks; unsigned int i_allocated_meta_blocks; @@ -1144,10 +1145,10 @@ struct ext4_sb_info { u32 s_hash_seed[4]; int s_def_hash_version; int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */ - struct percpu_counter s_freeblocks_counter; + struct percpu_counter s_freeclusters_counter; struct percpu_counter s_freeinodes_counter; struct percpu_counter s_dirs_counter; - struct percpu_counter s_dirtyblocks_counter; + struct percpu_counter s_dirtyclusters_counter; struct blockgroup_lock *s_blockgroup_lock; struct proc_dir_entry *s_proc; struct kobject s_kobj; diff --git a/fs/ext4/ialloc.c b/fs/ext4/ialloc.c index b7a8130d0af..58115bad163 100644 --- a/fs/ext4/ialloc.c +++ b/fs/ext4/ialloc.c @@ -490,7 +490,8 @@ static int find_group_orlov(struct super_block *sb, struct inode *parent, freei = percpu_counter_read_positive(&sbi->s_freeinodes_counter); avefreei = freei / ngroups; - freeb = percpu_counter_read_positive(&sbi->s_freeblocks_counter); + freeb = EXT4_C2B(sbi, + percpu_counter_read_positive(&sbi->s_freeclusters_counter)); avefreeb = freeb; do_div(avefreeb, ngroups); ndirs = percpu_counter_read_positive(&sbi->s_dirs_counter); diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 904a9a623da..40f51aae42f 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -281,7 +281,7 @@ void ext4_da_update_reserve_space(struct inode *inode, /* Update per-inode reservations */ ei->i_reserved_data_blocks -= used; ei->i_reserved_meta_blocks -= ei->i_allocated_meta_blocks; - percpu_counter_sub(&sbi->s_dirtyblocks_counter, + percpu_counter_sub(&sbi->s_dirtyclusters_counter, used + ei->i_allocated_meta_blocks); ei->i_allocated_meta_blocks = 0; @@ -291,7 +291,7 @@ void ext4_da_update_reserve_space(struct inode *inode, * only when we have written all of the delayed * allocation blocks. */ - percpu_counter_sub(&sbi->s_dirtyblocks_counter, + percpu_counter_sub(&sbi->s_dirtyclusters_counter, ei->i_reserved_meta_blocks); ei->i_reserved_meta_blocks = 0; ei->i_da_metadata_calc_len = 0; @@ -1119,14 +1119,14 @@ static void ext4_da_release_space(struct inode *inode, int to_free) * only when we have written all of the delayed * allocation blocks. */ - percpu_counter_sub(&sbi->s_dirtyblocks_counter, + percpu_counter_sub(&sbi->s_dirtyclusters_counter, ei->i_reserved_meta_blocks); ei->i_reserved_meta_blocks = 0; ei->i_da_metadata_calc_len = 0; } /* update fs dirty data blocks counter */ - percpu_counter_sub(&sbi->s_dirtyblocks_counter, to_free); + percpu_counter_sub(&sbi->s_dirtyclusters_counter, to_free); spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); @@ -1349,9 +1349,10 @@ static void ext4_print_free_blocks(struct inode *inode) ext4_count_free_blocks(inode->i_sb)); printk(KERN_CRIT "Free/Dirty block details\n"); printk(KERN_CRIT "free_blocks=%lld\n", - (long long) percpu_counter_sum(&sbi->s_freeblocks_counter)); + (long long) EXT4_C2B(EXT4_SB(inode->i_sb), + percpu_counter_sum(&sbi->s_freeclusters_counter))); printk(KERN_CRIT "dirty_blocks=%lld\n", - (long long) percpu_counter_sum(&sbi->s_dirtyblocks_counter)); + (long long) percpu_counter_sum(&sbi->s_dirtyclusters_counter)); printk(KERN_CRIT "Block reservation details\n"); printk(KERN_CRIT "i_reserved_data_blocks=%u\n", EXT4_I(inode)->i_reserved_data_blocks); @@ -2226,8 +2227,9 @@ static int ext4_nonda_switch(struct super_block *sb) * Delalloc need an accurate free block accounting. So switch * to non delalloc when we are near to error range. */ - free_blocks = percpu_counter_read_positive(&sbi->s_freeblocks_counter); - dirty_blocks = percpu_counter_read_positive(&sbi->s_dirtyblocks_counter); + free_blocks = EXT4_C2B(sbi, + percpu_counter_read_positive(&sbi->s_freeclusters_counter)); + dirty_blocks = percpu_counter_read_positive(&sbi->s_dirtyclusters_counter); if (2 * free_blocks < 3 * dirty_blocks || free_blocks < (dirty_blocks + EXT4_FREEBLOCKS_WATERMARK)) { /* diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index 57ce6960e94..4a38b65bd56 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -2834,13 +2834,14 @@ ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac, gdp->bg_checksum = ext4_group_desc_csum(sbi, ac->ac_b_ex.fe_group, gdp); ext4_unlock_group(sb, ac->ac_b_ex.fe_group); - percpu_counter_sub(&sbi->s_freeblocks_counter, ac->ac_b_ex.fe_len); + percpu_counter_sub(&sbi->s_freeclusters_counter, ac->ac_b_ex.fe_len); /* * Now reduce the dirty block count also. Should not go negative */ if (!(ac->ac_flags & EXT4_MB_DELALLOC_RESERVED)) /* release all the reserved blocks if non delalloc */ - percpu_counter_sub(&sbi->s_dirtyblocks_counter, reserv_clstrs); + percpu_counter_sub(&sbi->s_dirtyclusters_counter, + reserv_clstrs); if (sbi->s_log_groups_per_flex) { ext4_group_t flex_group = ext4_flex_group(sbi, @@ -4384,7 +4385,7 @@ out: if (!ext4_test_inode_state(ar->inode, EXT4_STATE_DELALLOC_RESERVED)) /* release all the reserved blocks if non delalloc */ - percpu_counter_sub(&sbi->s_dirtyblocks_counter, + percpu_counter_sub(&sbi->s_dirtyclusters_counter, reserv_clstrs); } @@ -4691,7 +4692,7 @@ do_more: ext4_free_blks_set(sb, gdp, ret); gdp->bg_checksum = ext4_group_desc_csum(sbi, block_group, gdp); ext4_unlock_group(sb, block_group); - percpu_counter_add(&sbi->s_freeblocks_counter, count); + percpu_counter_add(&sbi->s_freeclusters_counter, count_clusters); if (sbi->s_log_groups_per_flex) { ext4_group_t flex_group = ext4_flex_group(sbi, block_group); @@ -4833,7 +4834,8 @@ int ext4_group_add_blocks(handle_t *handle, struct super_block *sb, ext4_free_blks_set(sb, desc, blk_free_count); desc->bg_checksum = ext4_group_desc_csum(sbi, block_group, desc); ext4_unlock_group(sb, block_group); - percpu_counter_add(&sbi->s_freeblocks_counter, blocks_freed); + percpu_counter_add(&sbi->s_freeclusters_counter, + EXT4_B2C(sbi, blocks_freed)); if (sbi->s_log_groups_per_flex) { ext4_group_t flex_group = ext4_flex_group(sbi, block_group); diff --git a/fs/ext4/resize.c b/fs/ext4/resize.c index 707d3f16f7c..a324a537f2d 100644 --- a/fs/ext4/resize.c +++ b/fs/ext4/resize.c @@ -937,8 +937,8 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input) input->reserved_blocks); /* Update the free space counts */ - percpu_counter_add(&sbi->s_freeblocks_counter, - input->free_blocks_count); + percpu_counter_add(&sbi->s_freeclusters_counter, + EXT4_B2C(sbi, input->free_blocks_count)); percpu_counter_add(&sbi->s_freeinodes_counter, EXT4_INODES_PER_GROUP(sb)); diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 25a4bfe3f39..f81e7e79165 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -837,10 +837,10 @@ static void ext4_put_super(struct super_block *sb) brelse(sbi->s_group_desc[i]); ext4_kvfree(sbi->s_group_desc); ext4_kvfree(sbi->s_flex_groups); - percpu_counter_destroy(&sbi->s_freeblocks_counter); + percpu_counter_destroy(&sbi->s_freeclusters_counter); percpu_counter_destroy(&sbi->s_freeinodes_counter); percpu_counter_destroy(&sbi->s_dirs_counter); - percpu_counter_destroy(&sbi->s_dirtyblocks_counter); + percpu_counter_destroy(&sbi->s_dirtyclusters_counter); brelse(sbi->s_sbh); #ifdef CONFIG_QUOTA for (i = 0; i < MAXQUOTAS; i++) @@ -2473,7 +2473,7 @@ static ssize_t delayed_allocation_blocks_show(struct ext4_attr *a, char *buf) { return snprintf(buf, PAGE_SIZE, "%llu\n", - (s64) percpu_counter_sum(&sbi->s_dirtyblocks_counter)); + (s64) percpu_counter_sum(&sbi->s_dirtyclusters_counter)); } static ssize_t session_write_kbytes_show(struct ext4_attr *a, @@ -3575,7 +3575,7 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) sbi->s_err_report.function = print_daily_error_info; sbi->s_err_report.data = (unsigned long) sb; - err = percpu_counter_init(&sbi->s_freeblocks_counter, + err = percpu_counter_init(&sbi->s_freeclusters_counter, ext4_count_free_blocks(sb)); if (!err) { err = percpu_counter_init(&sbi->s_freeinodes_counter, @@ -3586,7 +3586,7 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) ext4_count_dirs(sb)); } if (!err) { - err = percpu_counter_init(&sbi->s_dirtyblocks_counter, 0); + err = percpu_counter_init(&sbi->s_dirtyclusters_counter, 0); } if (err) { ext4_msg(sb, KERN_ERR, "insufficient memory"); @@ -3701,13 +3701,13 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) * The journal may have updated the bg summary counts, so we * need to update the global counters. */ - percpu_counter_set(&sbi->s_freeblocks_counter, + percpu_counter_set(&sbi->s_freeclusters_counter, ext4_count_free_blocks(sb)); percpu_counter_set(&sbi->s_freeinodes_counter, ext4_count_free_inodes(sb)); percpu_counter_set(&sbi->s_dirs_counter, ext4_count_dirs(sb)); - percpu_counter_set(&sbi->s_dirtyblocks_counter, 0); + percpu_counter_set(&sbi->s_dirtyclusters_counter, 0); no_journal: /* @@ -3847,10 +3847,10 @@ failed_mount3: del_timer(&sbi->s_err_report); if (sbi->s_flex_groups) ext4_kvfree(sbi->s_flex_groups); - percpu_counter_destroy(&sbi->s_freeblocks_counter); + percpu_counter_destroy(&sbi->s_freeclusters_counter); percpu_counter_destroy(&sbi->s_freeinodes_counter); percpu_counter_destroy(&sbi->s_dirs_counter); - percpu_counter_destroy(&sbi->s_dirtyblocks_counter); + percpu_counter_destroy(&sbi->s_dirtyclusters_counter); if (sbi->s_mmp_tsk) kthread_stop(sbi->s_mmp_tsk); failed_mount2: @@ -4173,8 +4173,9 @@ static int ext4_commit_super(struct super_block *sb, int sync) else es->s_kbytes_written = cpu_to_le64(EXT4_SB(sb)->s_kbytes_written); - ext4_free_blocks_count_set(es, percpu_counter_sum_positive( - &EXT4_SB(sb)->s_freeblocks_counter)); + ext4_free_blocks_count_set(es, + EXT4_C2B(EXT4_SB(sb), percpu_counter_sum_positive( + &EXT4_SB(sb)->s_freeclusters_counter))); es->s_free_inodes_count = cpu_to_le32(percpu_counter_sum_positive( &EXT4_SB(sb)->s_freeinodes_counter)); @@ -4629,10 +4630,10 @@ static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf) buf->f_type = EXT4_SUPER_MAGIC; buf->f_bsize = sb->s_blocksize; buf->f_blocks = ext4_blocks_count(es) - sbi->s_overhead_last; - bfree = percpu_counter_sum_positive(&sbi->s_freeblocks_counter) - - percpu_counter_sum_positive(&sbi->s_dirtyblocks_counter); + bfree = percpu_counter_sum_positive(&sbi->s_freeclusters_counter) - + percpu_counter_sum_positive(&sbi->s_dirtyclusters_counter); /* prevent underflow in case that few free space is available */ - buf->f_bfree = max_t(s64, bfree, 0); + buf->f_bfree = EXT4_C2B(sbi, max_t(s64, bfree, 0)); buf->f_bavail = buf->f_bfree - ext4_r_blocks_count(es); if (buf->f_bfree < ext4_r_blocks_count(es)) buf->f_bavail = 0; -- cgit v1.2.3 From 7b415bf60f6afb0499fd3dc0ee33444f54e28567 Mon Sep 17 00:00:00 2001 From: Aditya Kali Date: Fri, 9 Sep 2011 19:04:51 -0400 Subject: ext4: Fix bigalloc quota accounting and i_blocks value With bigalloc changes, the i_blocks value was not correctly set (it was still set to number of blocks being used, but in case of bigalloc, we want i_blocks to represent the number of clusters being used). Since the quota subsystem sets the i_blocks value, this patch fixes the quota accounting and makes sure that the i_blocks value is set correctly. Signed-off-by: Aditya Kali Signed-off-by: "Theodore Ts'o" --- fs/ext4/balloc.c | 5 +- fs/ext4/ext4.h | 16 ++- fs/ext4/ext4_extents.h | 2 + fs/ext4/extents.c | 306 ++++++++++++++++++++++++++++++++++++++++++++++++- fs/ext4/inode.c | 54 ++++++--- fs/ext4/mballoc.c | 5 +- fs/ext4/super.c | 3 +- 7 files changed, 366 insertions(+), 25 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c index 9080a857cda..bf42b3219e3 100644 --- a/fs/ext4/balloc.c +++ b/fs/ext4/balloc.c @@ -485,7 +485,7 @@ int ext4_should_retry_alloc(struct super_block *sb, int *retries) * @handle: handle to this transaction * @inode: file inode * @goal: given target block(filesystem wide) - * @count: pointer to total number of blocks needed + * @count: pointer to total number of clusters needed * @errp: error code * * Return 1st allocated block number on success, *count stores total account @@ -517,7 +517,8 @@ ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, spin_lock(&EXT4_I(inode)->i_block_reservation_lock); EXT4_I(inode)->i_allocated_meta_blocks += ar.len; spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); - dquot_alloc_block_nofail(inode, ar.len); + dquot_alloc_block_nofail(inode, + EXT4_C2B(EXT4_SB(inode->i_sb), ar.len)); } return ret; } diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index d2584224c89..a6307f7c980 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -144,9 +144,17 @@ struct ext4_allocation_request { #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten) #define EXT4_MAP_BOUNDARY (1 << BH_Boundary) #define EXT4_MAP_UNINIT (1 << BH_Uninit) +/* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of + * ext4_map_blocks wants to know whether or not the underlying cluster has + * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that + * the requested mapping was from previously mapped (or delayed allocated) + * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster + * should never appear on buffer_head's state flags. + */ +#define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster) #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\ EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\ - EXT4_MAP_UNINIT) + EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER) struct ext4_map_blocks { ext4_fsblk_t m_pblk; @@ -1884,6 +1892,7 @@ extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); extern qsize_t *ext4_get_reserved_space(struct inode *inode); extern void ext4_da_update_reserve_space(struct inode *inode, int used, int quota_claim); +extern int ext4_da_reserve_space(struct inode *inode, ext4_lblk_t lblock); /* indirect.c */ extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, @@ -2284,6 +2293,11 @@ extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t); enum ext4_state_bits { BH_Uninit /* blocks are allocated but uninitialized on disk */ = BH_JBDPrivateStart, + BH_AllocFromCluster, /* allocated blocks were part of already + * allocated cluster. Note that this flag will + * never, ever appear in a buffer_head's state + * flag. See EXT4_MAP_FROM_CLUSTER to see where + * this is used. */ }; BUFFER_FNS(Uninit, uninit) diff --git a/fs/ext4/ext4_extents.h b/fs/ext4/ext4_extents.h index 095c36f3b61..a52db3a69a3 100644 --- a/fs/ext4/ext4_extents.h +++ b/fs/ext4/ext4_extents.h @@ -290,5 +290,7 @@ extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t, struct ext4_ext_path *); extern void ext4_ext_drop_refs(struct ext4_ext_path *); extern int ext4_ext_check_inode(struct inode *inode); +extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk, + int search_hint_reverse); #endif /* _EXT4_EXTENTS */ diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index cd4479c0803..c4e00586453 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -2686,6 +2686,21 @@ again: } } + /* If we still have something in the partial cluster and we have removed + * even the first extent, then we should free the blocks in the partial + * cluster as well. */ + if (partial_cluster && path->p_hdr->eh_entries == 0) { + int flags = EXT4_FREE_BLOCKS_FORGET; + + if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode)) + flags |= EXT4_FREE_BLOCKS_METADATA; + + ext4_free_blocks(handle, inode, NULL, + EXT4_C2B(EXT4_SB(sb), partial_cluster), + EXT4_SB(sb)->s_cluster_ratio, flags); + partial_cluster = 0; + } + /* TODO: flexible tree reduction should be here */ if (path->p_hdr->eh_entries == 0) { /* @@ -3233,6 +3248,195 @@ static int check_eofblocks_fl(handle_t *handle, struct inode *inode, return ext4_mark_inode_dirty(handle, inode); } +/** + * ext4_find_delalloc_range: find delayed allocated block in the given range. + * + * Goes through the buffer heads in the range [lblk_start, lblk_end] and returns + * whether there are any buffers marked for delayed allocation. It returns '1' + * on the first delalloc'ed buffer head found. If no buffer head in the given + * range is marked for delalloc, it returns 0. + * lblk_start should always be <= lblk_end. + * search_hint_reverse is to indicate that searching in reverse from lblk_end to + * lblk_start might be more efficient (i.e., we will likely hit the delalloc'ed + * block sooner). This is useful when blocks are truncated sequentially from + * lblk_start towards lblk_end. + */ +static int ext4_find_delalloc_range(struct inode *inode, + ext4_lblk_t lblk_start, + ext4_lblk_t lblk_end, + int search_hint_reverse) +{ + struct address_space *mapping = inode->i_mapping; + struct buffer_head *head, *bh = NULL; + struct page *page; + ext4_lblk_t i, pg_lblk; + pgoff_t index; + + /* reverse search wont work if fs block size is less than page size */ + if (inode->i_blkbits < PAGE_CACHE_SHIFT) + search_hint_reverse = 0; + + if (search_hint_reverse) + i = lblk_end; + else + i = lblk_start; + + index = i >> (PAGE_CACHE_SHIFT - inode->i_blkbits); + + while ((i >= lblk_start) && (i <= lblk_end)) { + page = find_get_page(mapping, index); + if (!page || !PageDirty(page)) + goto nextpage; + + if (PageWriteback(page)) { + /* + * This might be a race with allocation and writeout. In + * this case we just assume that the rest of the range + * will eventually be written and there wont be any + * delalloc blocks left. + * TODO: the above assumption is troublesome, but might + * work better in practice. other option could be note + * somewhere that the cluster is getting written out and + * detect that here. + */ + page_cache_release(page); + return 0; + } + + if (!page_has_buffers(page)) + goto nextpage; + + head = page_buffers(page); + if (!head) + goto nextpage; + + bh = head; + pg_lblk = index << (PAGE_CACHE_SHIFT - + inode->i_blkbits); + do { + if (unlikely(pg_lblk < lblk_start)) { + /* + * This is possible when fs block size is less + * than page size and our cluster starts/ends in + * middle of the page. So we need to skip the + * initial few blocks till we reach the 'lblk' + */ + pg_lblk++; + continue; + } + + if (buffer_delay(bh)) { + page_cache_release(page); + return 1; + } + if (search_hint_reverse) + i--; + else + i++; + } while ((i >= lblk_start) && (i <= lblk_end) && + ((bh = bh->b_this_page) != head)); +nextpage: + if (page) + page_cache_release(page); + /* + * Move to next page. 'i' will be the first lblk in the next + * page. + */ + if (search_hint_reverse) + index--; + else + index++; + i = index << (PAGE_CACHE_SHIFT - inode->i_blkbits); + } + + return 0; +} + +int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk, + int search_hint_reverse) +{ + struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); + ext4_lblk_t lblk_start, lblk_end; + lblk_start = lblk & (~(sbi->s_cluster_ratio - 1)); + lblk_end = lblk_start + sbi->s_cluster_ratio - 1; + + return ext4_find_delalloc_range(inode, lblk_start, lblk_end, + search_hint_reverse); +} + +/** + * Determines how many complete clusters (out of those specified by the 'map') + * are under delalloc and were reserved quota for. + * This function is called when we are writing out the blocks that were + * originally written with their allocation delayed, but then the space was + * allocated using fallocate() before the delayed allocation could be resolved. + * The cases to look for are: + * ('=' indicated delayed allocated blocks + * '-' indicates non-delayed allocated blocks) + * (a) partial clusters towards beginning and/or end outside of allocated range + * are not delalloc'ed. + * Ex: + * |----c---=|====c====|====c====|===-c----| + * |++++++ allocated ++++++| + * ==> 4 complete clusters in above example + * + * (b) partial cluster (outside of allocated range) towards either end is + * marked for delayed allocation. In this case, we will exclude that + * cluster. + * Ex: + * |----====c========|========c========| + * |++++++ allocated ++++++| + * ==> 1 complete clusters in above example + * + * Ex: + * |================c================| + * |++++++ allocated ++++++| + * ==> 0 complete clusters in above example + * + * The ext4_da_update_reserve_space will be called only if we + * determine here that there were some "entire" clusters that span + * this 'allocated' range. + * In the non-bigalloc case, this function will just end up returning num_blks + * without ever calling ext4_find_delalloc_range. + */ +static unsigned int +get_reserved_cluster_alloc(struct inode *inode, ext4_lblk_t lblk_start, + unsigned int num_blks) +{ + struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); + ext4_lblk_t alloc_cluster_start, alloc_cluster_end; + ext4_lblk_t lblk_from, lblk_to, c_offset; + unsigned int allocated_clusters = 0; + + alloc_cluster_start = EXT4_B2C(sbi, lblk_start); + alloc_cluster_end = EXT4_B2C(sbi, lblk_start + num_blks - 1); + + /* max possible clusters for this allocation */ + allocated_clusters = alloc_cluster_end - alloc_cluster_start + 1; + + /* Check towards left side */ + c_offset = lblk_start & (sbi->s_cluster_ratio - 1); + if (c_offset) { + lblk_from = lblk_start & (~(sbi->s_cluster_ratio - 1)); + lblk_to = lblk_from + c_offset - 1; + + if (ext4_find_delalloc_range(inode, lblk_from, lblk_to, 0)) + allocated_clusters--; + } + + /* Now check towards right. */ + c_offset = (lblk_start + num_blks) & (sbi->s_cluster_ratio - 1); + if (allocated_clusters && c_offset) { + lblk_from = lblk_start + num_blks; + lblk_to = lblk_from + (sbi->s_cluster_ratio - c_offset) - 1; + + if (ext4_find_delalloc_range(inode, lblk_from, lblk_to, 0)) + allocated_clusters--; + } + + return allocated_clusters; +} + static int ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode, struct ext4_map_blocks *map, @@ -3338,8 +3542,15 @@ out: * But fallocate would have already updated quota and block * count for this offset. So cancel these reservation */ - if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) - ext4_da_update_reserve_space(inode, allocated, 0); + if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) { + unsigned int reserved_clusters; + reserved_clusters = get_reserved_cluster_alloc(inode, + map->m_lblk, map->m_len); + if (reserved_clusters) + ext4_da_update_reserve_space(inode, + reserved_clusters, + 0); + } map_out: map->m_flags |= EXT4_MAP_MAPPED; @@ -3484,6 +3695,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, ext4_fsblk_t newblock = 0; int free_on_err = 0, err = 0, depth, ret; unsigned int allocated = 0, offset = 0; + unsigned int allocated_clusters = 0, reserved_clusters = 0; unsigned int punched_out = 0; unsigned int result = 0; struct ext4_allocation_request ar; @@ -3499,6 +3711,10 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, if (!(flags & EXT4_GET_BLOCKS_PUNCH_OUT_EXT) && ext4_ext_in_cache(inode, map->m_lblk, &newex)) { if (!newex.ee_start_lo && !newex.ee_start_hi) { + if ((sbi->s_cluster_ratio > 1) && + ext4_find_delalloc_cluster(inode, map->m_lblk, 0)) + map->m_flags |= EXT4_MAP_FROM_CLUSTER; + if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) { /* * block isn't allocated yet and @@ -3509,6 +3725,8 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, /* we should allocate requested block */ } else { /* block is already allocated */ + if (sbi->s_cluster_ratio > 1) + map->m_flags |= EXT4_MAP_FROM_CLUSTER; newblock = map->m_lblk - le32_to_cpu(newex.ee_block) + ext4_ext_pblock(&newex); @@ -3665,6 +3883,10 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, } } + if ((sbi->s_cluster_ratio > 1) && + ext4_find_delalloc_cluster(inode, map->m_lblk, 0)) + map->m_flags |= EXT4_MAP_FROM_CLUSTER; + /* * requested block isn't allocated yet; * we couldn't try to create block if create flag is zero @@ -3681,6 +3903,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, /* * Okay, we need to do block allocation. */ + map->m_flags &= ~EXT4_MAP_FROM_CLUSTER; newex.ee_block = cpu_to_le32(map->m_lblk); cluster_offset = map->m_lblk & (sbi->s_cluster_ratio-1); @@ -3692,6 +3915,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, get_implied_cluster_alloc(sbi, map, ex, path)) { ar.len = allocated = map->m_len; newblock = map->m_pblk; + map->m_flags |= EXT4_MAP_FROM_CLUSTER; goto got_allocated_blocks; } @@ -3712,6 +3936,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, get_implied_cluster_alloc(sbi, map, ex2, path)) { ar.len = allocated = map->m_len; newblock = map->m_pblk; + map->m_flags |= EXT4_MAP_FROM_CLUSTER; goto got_allocated_blocks; } @@ -3765,6 +3990,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, ext_debug("allocate new block: goal %llu, found %llu/%u\n", ar.goal, newblock, allocated); free_on_err = 1; + allocated_clusters = ar.len; ar.len = EXT4_C2B(sbi, ar.len) - offset; if (ar.len > allocated) ar.len = allocated; @@ -3822,8 +4048,80 @@ got_allocated_blocks: * Update reserved blocks/metadata blocks after successful * block allocation which had been deferred till now. */ - if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) - ext4_da_update_reserve_space(inode, allocated, 1); + if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) { + /* + * Check how many clusters we had reserved this allocted range. + */ + reserved_clusters = get_reserved_cluster_alloc(inode, + map->m_lblk, allocated); + if (map->m_flags & EXT4_MAP_FROM_CLUSTER) { + if (reserved_clusters) { + /* + * We have clusters reserved for this range. + * But since we are not doing actual allocation + * and are simply using blocks from previously + * allocated cluster, we should release the + * reservation and not claim quota. + */ + ext4_da_update_reserve_space(inode, + reserved_clusters, 0); + } + } else { + BUG_ON(allocated_clusters < reserved_clusters); + /* We will claim quota for all newly allocated blocks.*/ + ext4_da_update_reserve_space(inode, allocated_clusters, + 1); + if (reserved_clusters < allocated_clusters) { + int reservation = allocated_clusters - + reserved_clusters; + /* + * It seems we claimed few clusters outside of + * the range of this allocation. We should give + * it back to the reservation pool. This can + * happen in the following case: + * + * * Suppose s_cluster_ratio is 4 (i.e., each + * cluster has 4 blocks. Thus, the clusters + * are [0-3],[4-7],[8-11]... + * * First comes delayed allocation write for + * logical blocks 10 & 11. Since there were no + * previous delayed allocated blocks in the + * range [8-11], we would reserve 1 cluster + * for this write. + * * Next comes write for logical blocks 3 to 8. + * In this case, we will reserve 2 clusters + * (for [0-3] and [4-7]; and not for [8-11] as + * that range has a delayed allocated blocks. + * Thus total reserved clusters now becomes 3. + * * Now, during the delayed allocation writeout + * time, we will first write blocks [3-8] and + * allocate 3 clusters for writing these + * blocks. Also, we would claim all these + * three clusters above. + * * Now when we come here to writeout the + * blocks [10-11], we would expect to claim + * the reservation of 1 cluster we had made + * (and we would claim it since there are no + * more delayed allocated blocks in the range + * [8-11]. But our reserved cluster count had + * already gone to 0. + * + * Thus, at the step 4 above when we determine + * that there are still some unwritten delayed + * allocated blocks outside of our current + * block range, we should increment the + * reserved clusters count so that when the + * remaining blocks finally gets written, we + * could claim them. + */ + while (reservation) { + ext4_da_reserve_space(inode, + map->m_lblk); + reservation--; + } + } + } + } /* * Cache the extent and update transaction to commit on fdatasync only diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 40f51aae42f..d1c17e47c1c 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -300,14 +300,14 @@ void ext4_da_update_reserve_space(struct inode *inode, /* Update quota subsystem for data blocks */ if (quota_claim) - dquot_claim_block(inode, used); + dquot_claim_block(inode, EXT4_C2B(sbi, used)); else { /* * We did fallocate with an offset that is already delayed * allocated. So on delayed allocated writeback we should * not re-claim the quota for fallocated blocks. */ - dquot_release_reservation_block(inode, used); + dquot_release_reservation_block(inode, EXT4_C2B(sbi, used)); } /* @@ -1037,14 +1037,14 @@ static int ext4_journalled_write_end(struct file *file, } /* - * Reserve a single block located at lblock + * Reserve a single cluster located at lblock */ -static int ext4_da_reserve_space(struct inode *inode, ext4_lblk_t lblock) +int ext4_da_reserve_space(struct inode *inode, ext4_lblk_t lblock) { int retries = 0; struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); struct ext4_inode_info *ei = EXT4_I(inode); - unsigned long md_needed; + unsigned int md_needed; int ret; /* @@ -1054,7 +1054,8 @@ static int ext4_da_reserve_space(struct inode *inode, ext4_lblk_t lblock) */ repeat: spin_lock(&ei->i_block_reservation_lock); - md_needed = ext4_calc_metadata_amount(inode, lblock); + md_needed = EXT4_NUM_B2C(sbi, + ext4_calc_metadata_amount(inode, lblock)); trace_ext4_da_reserve_space(inode, md_needed); spin_unlock(&ei->i_block_reservation_lock); @@ -1063,7 +1064,7 @@ repeat: * us from metadata over-estimation, though we may go over by * a small amount in the end. Here we just reserve for data. */ - ret = dquot_reserve_block(inode, 1); + ret = dquot_reserve_block(inode, EXT4_C2B(sbi, 1)); if (ret) return ret; /* @@ -1071,7 +1072,7 @@ repeat: * we cannot afford to run out of free blocks. */ if (ext4_claim_free_blocks(sbi, md_needed + 1, 0)) { - dquot_release_reservation_block(inode, 1); + dquot_release_reservation_block(inode, EXT4_C2B(sbi, 1)); if (ext4_should_retry_alloc(inode->i_sb, &retries)) { yield(); goto repeat; @@ -1118,6 +1119,8 @@ static void ext4_da_release_space(struct inode *inode, int to_free) * We can release all of the reserved metadata blocks * only when we have written all of the delayed * allocation blocks. + * Note that in case of bigalloc, i_reserved_meta_blocks, + * i_reserved_data_blocks, etc. refer to number of clusters. */ percpu_counter_sub(&sbi->s_dirtyclusters_counter, ei->i_reserved_meta_blocks); @@ -1130,7 +1133,7 @@ static void ext4_da_release_space(struct inode *inode, int to_free) spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); - dquot_release_reservation_block(inode, to_free); + dquot_release_reservation_block(inode, EXT4_C2B(sbi, to_free)); } static void ext4_da_page_release_reservation(struct page *page, @@ -1139,6 +1142,9 @@ static void ext4_da_page_release_reservation(struct page *page, int to_release = 0; struct buffer_head *head, *bh; unsigned int curr_off = 0; + struct inode *inode = page->mapping->host; + struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); + int num_clusters; head = page_buffers(page); bh = head; @@ -1151,7 +1157,20 @@ static void ext4_da_page_release_reservation(struct page *page, } curr_off = next_off; } while ((bh = bh->b_this_page) != head); - ext4_da_release_space(page->mapping->host, to_release); + + /* If we have released all the blocks belonging to a cluster, then we + * need to release the reserved space for that cluster. */ + num_clusters = EXT4_NUM_B2C(sbi, to_release); + while (num_clusters > 0) { + ext4_fsblk_t lblk; + lblk = (page->index << (PAGE_CACHE_SHIFT - inode->i_blkbits)) + + ((num_clusters - 1) << sbi->s_cluster_bits); + if (sbi->s_cluster_ratio == 1 || + !ext4_find_delalloc_cluster(inode, lblk, 1)) + ext4_da_release_space(inode, 1); + + num_clusters--; + } } /* @@ -1352,7 +1371,8 @@ static void ext4_print_free_blocks(struct inode *inode) (long long) EXT4_C2B(EXT4_SB(inode->i_sb), percpu_counter_sum(&sbi->s_freeclusters_counter))); printk(KERN_CRIT "dirty_blocks=%lld\n", - (long long) percpu_counter_sum(&sbi->s_dirtyclusters_counter)); + (long long) EXT4_C2B(EXT4_SB(inode->i_sb), + percpu_counter_sum(&sbi->s_dirtyclusters_counter))); printk(KERN_CRIT "Block reservation details\n"); printk(KERN_CRIT "i_reserved_data_blocks=%u\n", EXT4_I(inode)->i_reserved_data_blocks); @@ -1626,10 +1646,14 @@ static int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, /* * XXX: __block_write_begin() unmaps passed block, is it OK? */ - ret = ext4_da_reserve_space(inode, iblock); - if (ret) - /* not enough space to reserve */ - return ret; + /* If the block was allocated from previously allocated cluster, + * then we dont need to reserve it again. */ + if (!(map.m_flags & EXT4_MAP_FROM_CLUSTER)) { + ret = ext4_da_reserve_space(inode, iblock); + if (ret) + /* not enough space to reserve */ + return ret; + } map_bh(bh, inode->i_sb, invalid_block); set_buffer_new(bh); diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index 63dd5670334..5e1215d3833 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -4718,6 +4718,9 @@ do_more: freed += count; + if (!(flags & EXT4_FREE_BLOCKS_NO_QUOT_UPDATE)) + dquot_free_block(inode, EXT4_C2B(sbi, count_clusters)); + /* We dirtied the bitmap block */ BUFFER_TRACE(bitmap_bh, "dirtied bitmap block"); err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh); @@ -4736,8 +4739,6 @@ do_more: } ext4_mark_super_dirty(sb); error_return: - if (freed && !(flags & EXT4_FREE_BLOCKS_NO_QUOT_UPDATE)) - dquot_free_block(inode, freed); brelse(bitmap_bh); ext4_std_error(sb, err); return; diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 6810957e0ac..66b8cfa1563 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -2473,7 +2473,8 @@ static ssize_t delayed_allocation_blocks_show(struct ext4_attr *a, char *buf) { return snprintf(buf, PAGE_SIZE, "%llu\n", - (s64) percpu_counter_sum(&sbi->s_dirtyclusters_counter)); + (s64) EXT4_C2B(sbi, + percpu_counter_sum(&sbi->s_dirtyclusters_counter))); } static ssize_t session_write_kbytes_show(struct ext4_attr *a, -- cgit v1.2.3 From 5dee54372c1ea15ab482b959634cda8c01b042bd Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Fri, 9 Sep 2011 19:10:51 -0400 Subject: ext4: rename ext4_count_free_blocks() to ext4_count_free_clusters() This function really counts the free clusters reported in the block group descriptors, so rename it to reduce confusion. Signed-off-by: "Theodore Ts'o" --- fs/ext4/balloc.c | 11 ++++++----- fs/ext4/ext4.h | 2 +- fs/ext4/inode.c | 6 +++--- fs/ext4/super.c | 7 ++++--- 4 files changed, 14 insertions(+), 12 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c index 84dc3d62b40..4f303809e07 100644 --- a/fs/ext4/balloc.c +++ b/fs/ext4/balloc.c @@ -524,12 +524,12 @@ ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, } /** - * ext4_count_free_blocks() -- count filesystem free blocks + * ext4_count_free_clusters() -- count filesystem free clusters * @sb: superblock * - * Adds up the number of free blocks from each block group. + * Adds up the number of free clusters from each block group. */ -ext4_fsblk_t ext4_count_free_blocks(struct super_block *sb) +ext4_fsblk_t ext4_count_free_clusters(struct super_block *sb) { ext4_fsblk_t desc_count; struct ext4_group_desc *gdp; @@ -562,8 +562,9 @@ ext4_fsblk_t ext4_count_free_blocks(struct super_block *sb) bitmap_count += x; } brelse(bitmap_bh); - printk(KERN_DEBUG "ext4_count_free_blocks: stored = %llu" - ", computed = %llu, %llu\n", ext4_free_blocks_count(es), + printk(KERN_DEBUG "ext4_count_free_clusters: stored = %llu" + ", computed = %llu, %llu\n", + EXT4_B2C(sbi, ext4_free_blocks_count(es)), desc_count, bitmap_count); return bitmap_count; #else diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index db279c36201..f6963cd7f70 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -1775,7 +1775,7 @@ extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, int *errp); extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks, unsigned int flags); -extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *); +extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); extern void ext4_check_blocks_bitmap(struct super_block *); extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, ext4_group_t block_group, diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index d1c17e47c1c..f9198d7875c 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -1365,7 +1365,8 @@ static void ext4_print_free_blocks(struct inode *inode) { struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); printk(KERN_CRIT "Total free blocks count %lld\n", - ext4_count_free_blocks(inode->i_sb)); + EXT4_C2B(EXT4_SB(inode->i_sb), + ext4_count_free_clusters(inode->i_sb))); printk(KERN_CRIT "Free/Dirty block details\n"); printk(KERN_CRIT "free_blocks=%lld\n", (long long) EXT4_C2B(EXT4_SB(inode->i_sb), @@ -1451,8 +1452,7 @@ static void mpage_da_map_and_submit(struct mpage_da_data *mpd) if (err == -EAGAIN) goto submit_io; - if (err == -ENOSPC && - ext4_count_free_blocks(sb)) { + if (err == -ENOSPC && ext4_count_free_clusters(sb)) { mpd->retval = err; goto submit_io; } diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 30cf0e1bd89..07f3de34145 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -2153,7 +2153,8 @@ static int ext4_check_descriptors(struct super_block *sb, if (NULL != first_not_zeroed) *first_not_zeroed = grp; - ext4_free_blocks_count_set(sbi->s_es, ext4_count_free_blocks(sb)); + ext4_free_blocks_count_set(sbi->s_es, + EXT4_C2B(sbi, ext4_count_free_clusters(sb))); sbi->s_es->s_free_inodes_count =cpu_to_le32(ext4_count_free_inodes(sb)); return 1; } @@ -3577,7 +3578,7 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) sbi->s_err_report.data = (unsigned long) sb; err = percpu_counter_init(&sbi->s_freeclusters_counter, - ext4_count_free_blocks(sb)); + ext4_count_free_clusters(sb)); if (!err) { err = percpu_counter_init(&sbi->s_freeinodes_counter, ext4_count_free_inodes(sb)); @@ -3703,7 +3704,7 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) * need to update the global counters. */ percpu_counter_set(&sbi->s_freeclusters_counter, - ext4_count_free_blocks(sb)); + ext4_count_free_clusters(sb)); percpu_counter_set(&sbi->s_freeinodes_counter, ext4_count_free_inodes(sb)); percpu_counter_set(&sbi->s_dirs_counter, -- cgit v1.2.3 From e7d5f3156e6827970f75ab27ad7eb0155826eb0b Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Fri, 9 Sep 2011 19:14:51 -0400 Subject: ext4: rename ext4_claim_free_blocks() to ext4_claim_free_clusters() This function really claims a number of free clusters, not blocks, so rename it so it's clearer what's going on. Signed-off-by: "Theodore Ts'o" --- fs/ext4/balloc.c | 8 ++++---- fs/ext4/ext4.h | 4 ++-- fs/ext4/inode.c | 2 +- fs/ext4/mballoc.c | 2 +- 4 files changed, 8 insertions(+), 8 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c index 3d07c35c708..af0c5357e35 100644 --- a/fs/ext4/balloc.c +++ b/fs/ext4/balloc.c @@ -445,11 +445,11 @@ static int ext4_has_free_blocks(struct ext4_sb_info *sbi, return 0; } -int ext4_claim_free_blocks(struct ext4_sb_info *sbi, - s64 nblocks, unsigned int flags) +int ext4_claim_free_clusters(struct ext4_sb_info *sbi, + s64 nclusters, unsigned int flags) { - if (ext4_has_free_blocks(sbi, nblocks, flags)) { - percpu_counter_add(&sbi->s_dirtyclusters_counter, nblocks); + if (ext4_has_free_blocks(sbi, nclusters, flags)) { + percpu_counter_add(&sbi->s_dirtyclusters_counter, nclusters); return 0; } else return -ENOSPC; diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 1473c06e4c9..f97638634e3 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -1773,8 +1773,8 @@ extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode, unsigned int flags, unsigned long *count, int *errp); -extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, - s64 nblocks, unsigned int flags); +extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi, + s64 nclusters, unsigned int flags); extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *); extern void ext4_check_blocks_bitmap(struct super_block *); extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb, diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index f9198d7875c..557c34ffa05 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -1071,7 +1071,7 @@ repeat: * We do still charge estimated metadata to the sb though; * we cannot afford to run out of free blocks. */ - if (ext4_claim_free_blocks(sbi, md_needed + 1, 0)) { + if (ext4_claim_free_clusters(sbi, md_needed + 1, 0)) { dquot_release_reservation_block(inode, EXT4_C2B(sbi, 1)); if (ext4_should_retry_alloc(inode->i_sb, &retries)) { yield(); diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index 90a3ed7c565..1c83161090f 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -4297,7 +4297,7 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle, * and verify allocation doesn't exceed the quota limits. */ while (ar->len && - ext4_claim_free_blocks(sbi, ar->len, ar->flags)) { + ext4_claim_free_clusters(sbi, ar->len, ar->flags)) { /* let others to free the space */ yield(); -- cgit v1.2.3 From df55c99dc8ee4c3c886a5edc8a4aa6b131c95afc Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Fri, 9 Sep 2011 19:16:51 -0400 Subject: ext4: rename ext4_has_free_blocks() to ext4_has_free_clusters() Rename the function so it is more clear what is going on. Also rename the various variables so it's clearer what's happening. Also fix a missing blocks to cluster conversion when reading the number of reserved blocks for root. Signed-off-by: "Theodore Ts'o" --- fs/ext4/balloc.c | 43 ++++++++++++++++++++++--------------------- fs/ext4/ext4.h | 8 ++++---- fs/ext4/inode.c | 2 +- 3 files changed, 27 insertions(+), 26 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c index af0c5357e35..f6dba4505f1 100644 --- a/fs/ext4/balloc.c +++ b/fs/ext4/balloc.c @@ -403,42 +403,43 @@ ext4_read_block_bitmap(struct super_block *sb, ext4_group_t block_group) } /** - * ext4_has_free_blocks() + * ext4_has_free_clusters() * @sbi: in-core super block structure. - * @nblocks: number of needed blocks + * @nclusters: number of needed blocks + * @flags: flags from ext4_mb_new_blocks() * - * Check if filesystem has nblocks free & available for allocation. + * Check if filesystem has nclusters free & available for allocation. * On success return 1, return 0 on failure. */ -static int ext4_has_free_blocks(struct ext4_sb_info *sbi, - s64 nblocks, unsigned int flags) +static int ext4_has_free_clusters(struct ext4_sb_info *sbi, + s64 nclusters, unsigned int flags) { - s64 free_blocks, dirty_blocks, root_blocks; + s64 free_clusters, dirty_clusters, root_clusters; struct percpu_counter *fcc = &sbi->s_freeclusters_counter; - struct percpu_counter *dbc = &sbi->s_dirtyclusters_counter; + struct percpu_counter *dcc = &sbi->s_dirtyclusters_counter; - free_blocks = percpu_counter_read_positive(fcc); - dirty_blocks = percpu_counter_read_positive(dbc); - root_blocks = ext4_r_blocks_count(sbi->s_es); + free_clusters = percpu_counter_read_positive(fcc); + dirty_clusters = percpu_counter_read_positive(dcc); + root_clusters = EXT4_B2C(sbi, ext4_r_blocks_count(sbi->s_es)); - if (free_blocks - (nblocks + root_blocks + dirty_blocks) < - EXT4_FREEBLOCKS_WATERMARK) { - free_blocks = EXT4_C2B(sbi, percpu_counter_sum_positive(fcc)); - dirty_blocks = percpu_counter_sum_positive(dbc); + if (free_clusters - (nclusters + root_clusters + dirty_clusters) < + EXT4_FREECLUSTERS_WATERMARK) { + free_clusters = EXT4_C2B(sbi, percpu_counter_sum_positive(fcc)); + dirty_clusters = percpu_counter_sum_positive(dcc); } - /* Check whether we have space after - * accounting for current dirty blocks & root reserved blocks. + /* Check whether we have space after accounting for current + * dirty clusters & root reserved clusters. */ - if (free_blocks >= ((root_blocks + nblocks) + dirty_blocks)) + if (free_clusters >= ((root_clusters + nclusters) + dirty_clusters)) return 1; - /* Hm, nope. Are (enough) root reserved blocks available? */ + /* Hm, nope. Are (enough) root reserved clusters available? */ if (sbi->s_resuid == current_fsuid() || ((sbi->s_resgid != 0) && in_group_p(sbi->s_resgid)) || capable(CAP_SYS_RESOURCE) || (flags & EXT4_MB_USE_ROOT_BLOCKS)) { - if (free_blocks >= (nblocks + dirty_blocks)) + if (free_clusters >= (nclusters + dirty_clusters)) return 1; } @@ -448,7 +449,7 @@ static int ext4_has_free_blocks(struct ext4_sb_info *sbi, int ext4_claim_free_clusters(struct ext4_sb_info *sbi, s64 nclusters, unsigned int flags) { - if (ext4_has_free_blocks(sbi, nclusters, flags)) { + if (ext4_has_free_clusters(sbi, nclusters, flags)) { percpu_counter_add(&sbi->s_dirtyclusters_counter, nclusters); return 0; } else @@ -469,7 +470,7 @@ int ext4_claim_free_clusters(struct ext4_sb_info *sbi, */ int ext4_should_retry_alloc(struct super_block *sb, int *retries) { - if (!ext4_has_free_blocks(EXT4_SB(sb), 1, 0) || + if (!ext4_has_free_clusters(EXT4_SB(sb), 1, 0) || (*retries)++ > 3 || !EXT4_SB(sb)->s_journal) return 0; diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index f97638634e3..21ea65d8bd4 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -2103,13 +2103,13 @@ do { \ } while (0) #ifdef CONFIG_SMP -/* Each CPU can accumulate percpu_counter_batch blocks in their local - * counters. So we need to make sure we have free blocks more +/* Each CPU can accumulate percpu_counter_batch clusters in their local + * counters. So we need to make sure we have free clusters more * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times. */ -#define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) +#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids)) #else -#define EXT4_FREEBLOCKS_WATERMARK 0 +#define EXT4_FREECLUSTERS_WATERMARK 0 #endif static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize) diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 557c34ffa05..88dc63a0175 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -2255,7 +2255,7 @@ static int ext4_nonda_switch(struct super_block *sb) percpu_counter_read_positive(&sbi->s_freeclusters_counter)); dirty_blocks = percpu_counter_read_positive(&sbi->s_dirtyclusters_counter); if (2 * free_blocks < 3 * dirty_blocks || - free_blocks < (dirty_blocks + EXT4_FREEBLOCKS_WATERMARK)) { + free_blocks < (dirty_blocks + EXT4_FREECLUSTERS_WATERMARK)) { /* * free block count is less than 150% of dirty blocks * or free blocks is less than watermark -- cgit v1.2.3 From d8990240d8c911064447f8aa5a440f9345a6d692 Mon Sep 17 00:00:00 2001 From: Aditya Kali Date: Fri, 9 Sep 2011 19:18:51 -0400 Subject: ext4: add some tracepoints in ext4/extents.c This patch adds some tracepoints in ext4/extents.c and updates a tracepoint in ext4/inode.c. Tested: Built and ran the kernel and verified that these tracepoints work. Also ran xfstests. Signed-off-by: Aditya Kali Signed-off-by: "Theodore Ts'o" --- fs/ext4/ext4.h | 2 + fs/ext4/extents.c | 44 ++++- fs/ext4/inode.c | 3 +- fs/ext4/migrate.c | 1 - fs/ext4/move_extent.c | 1 - include/trace/events/ext4.h | 398 +++++++++++++++++++++++++++++++++++++++++++- 6 files changed, 433 insertions(+), 16 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 21ea65d8bd4..751277a4890 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -2339,4 +2339,6 @@ extern void ext4_resize_end(struct super_block *sb); #endif /* __KERNEL__ */ +#include "ext4_extents.h" + #endif /* _EXT4_H */ diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index c4e00586453..9b119308dae 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -42,7 +42,6 @@ #include #include #include "ext4_jbd2.h" -#include "ext4_extents.h" #include @@ -1969,6 +1968,7 @@ ext4_ext_put_in_cache(struct inode *inode, ext4_lblk_t block, struct ext4_ext_cache *cex; BUG_ON(len == 0); spin_lock(&EXT4_I(inode)->i_block_reservation_lock); + trace_ext4_ext_put_in_cache(inode, block, len, start); cex = &EXT4_I(inode)->i_cached_extent; cex->ec_block = block; cex->ec_len = len; @@ -2070,6 +2070,7 @@ errout: sbi->extent_cache_misses++; else sbi->extent_cache_hits++; + trace_ext4_ext_in_cache(inode, block, ret); spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); return ret; } @@ -2137,6 +2138,8 @@ static int ext4_ext_rm_idx(handle_t *handle, struct inode *inode, if (err) return err; ext_debug("index is empty, remove it, free block %llu\n", leaf); + trace_ext4_ext_rm_idx(inode, leaf); + ext4_free_blocks(handle, inode, NULL, leaf, 1, EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET); return err; @@ -2222,6 +2225,7 @@ static int ext4_remove_blocks(handle_t *handle, struct inode *inode, */ flags |= EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER; + trace_ext4_remove_blocks(inode, ex, from, to, *partial_cluster); /* * If we have a partial cluster, and it's different from the * cluster of the last block, we need to explicitly free the @@ -2336,6 +2340,8 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode, ex_ee_block = le32_to_cpu(ex->ee_block); ex_ee_len = ext4_ext_get_actual_len(ex); + trace_ext4_ext_rm_leaf(inode, start, ex, *partial_cluster); + while (ex >= EXT_FIRST_EXTENT(eh) && ex_ee_block + ex_ee_len > start) { @@ -2591,6 +2597,8 @@ static int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start) again: ext4_ext_invalidate_cache(inode); + trace_ext4_ext_remove_space(inode, start, depth); + /* * We start scanning from right side, freeing all the blocks * after i_size and walking into the tree depth-wise. @@ -2686,6 +2694,9 @@ again: } } + trace_ext4_ext_remove_space_done(inode, start, depth, partial_cluster, + path->p_hdr->eh_entries); + /* If we still have something in the partial cluster and we have removed * even the first extent, then we should free the blocks in the partial * cluster as well. */ @@ -3300,6 +3311,10 @@ static int ext4_find_delalloc_range(struct inode *inode, * detect that here. */ page_cache_release(page); + trace_ext4_find_delalloc_range(inode, + lblk_start, lblk_end, + search_hint_reverse, + 0, i); return 0; } @@ -3327,6 +3342,10 @@ static int ext4_find_delalloc_range(struct inode *inode, if (buffer_delay(bh)) { page_cache_release(page); + trace_ext4_find_delalloc_range(inode, + lblk_start, lblk_end, + search_hint_reverse, + 1, i); return 1; } if (search_hint_reverse) @@ -3349,6 +3368,8 @@ nextpage: i = index << (PAGE_CACHE_SHIFT - inode->i_blkbits); } + trace_ext4_find_delalloc_range(inode, lblk_start, lblk_end, + search_hint_reverse, 0, 0); return 0; } @@ -3414,6 +3435,8 @@ get_reserved_cluster_alloc(struct inode *inode, ext4_lblk_t lblk_start, /* max possible clusters for this allocation */ allocated_clusters = alloc_cluster_end - alloc_cluster_start + 1; + trace_ext4_get_reserved_cluster_alloc(inode, lblk_start, num_blks); + /* Check towards left side */ c_offset = lblk_start & (sbi->s_cluster_ratio - 1); if (c_offset) { @@ -3453,6 +3476,9 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode, flags, allocated); ext4_ext_show_leaf(inode, path); + trace_ext4_ext_handle_uninitialized_extents(inode, map, allocated, + newblock); + /* get_block() before submit the IO, split the extent */ if ((flags & EXT4_GET_BLOCKS_PRE_IO)) { ret = ext4_split_unwritten_extents(handle, inode, map, @@ -3572,7 +3598,7 @@ out2: * get_implied_cluster_alloc - check to see if the requested * allocation (in the map structure) overlaps with a cluster already * allocated in an extent. - * @sbi The ext4-specific superblock structure + * @sb The filesystem superblock structure * @map The requested lblk->pblk mapping * @ex The extent structure which might contain an implied * cluster allocation @@ -3609,11 +3635,12 @@ out2: * ext4_ext_map_blocks() will then allocate one or more new clusters * by calling ext4_mb_new_blocks(). */ -static int get_implied_cluster_alloc(struct ext4_sb_info *sbi, +static int get_implied_cluster_alloc(struct super_block *sb, struct ext4_map_blocks *map, struct ext4_extent *ex, struct ext4_ext_path *path) { + struct ext4_sb_info *sbi = EXT4_SB(sb); ext4_lblk_t c_offset = map->m_lblk & (sbi->s_cluster_ratio-1); ext4_lblk_t ex_cluster_start, ex_cluster_end; ext4_lblk_t rr_cluster_start, rr_cluster_end; @@ -3662,8 +3689,12 @@ static int get_implied_cluster_alloc(struct ext4_sb_info *sbi, ext4_lblk_t next = ext4_ext_next_allocated_block(path); map->m_len = min(map->m_len, next - map->m_lblk); } + + trace_ext4_get_implied_cluster_alloc_exit(sb, map, 1); return 1; } + + trace_ext4_get_implied_cluster_alloc_exit(sb, map, 0); return 0; } @@ -3772,6 +3803,9 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, * we split out initialized portions during a write. */ ee_len = ext4_ext_get_actual_len(ex); + + trace_ext4_ext_show_extent(inode, ee_block, ee_start, ee_len); + /* if found extent covers block, simply return it */ if (in_range(map->m_lblk, ee_block, ee_len)) { ext4_fsblk_t partial_cluster = 0; @@ -3912,7 +3946,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, * by ext4_ext_find_extent() implies a cluster we can use. */ if (cluster_offset && ex && - get_implied_cluster_alloc(sbi, map, ex, path)) { + get_implied_cluster_alloc(inode->i_sb, map, ex, path)) { ar.len = allocated = map->m_len; newblock = map->m_pblk; map->m_flags |= EXT4_MAP_FROM_CLUSTER; @@ -3933,7 +3967,7 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, /* Check if the extent after searching to the right implies a * cluster we can use. */ if ((sbi->s_cluster_ratio > 1) && ex2 && - get_implied_cluster_alloc(sbi, map, ex2, path)) { + get_implied_cluster_alloc(inode->i_sb, map, ex2, path)) { ar.len = allocated = map->m_len; newblock = map->m_pblk; map->m_flags |= EXT4_MAP_FROM_CLUSTER; diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 88dc63a0175..2dcd4fed96e 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -42,7 +42,6 @@ #include "ext4_jbd2.h" #include "xattr.h" #include "acl.h" -#include "ext4_extents.h" #include "truncate.h" #include @@ -268,7 +267,7 @@ void ext4_da_update_reserve_space(struct inode *inode, struct ext4_inode_info *ei = EXT4_I(inode); spin_lock(&ei->i_block_reservation_lock); - trace_ext4_da_update_reserve_space(inode, used); + trace_ext4_da_update_reserve_space(inode, used, quota_claim); if (unlikely(used > ei->i_reserved_data_blocks)) { ext4_msg(inode->i_sb, KERN_NOTICE, "%s: ino %lu, used %d " "with only %d reserved data blocks\n", diff --git a/fs/ext4/migrate.c b/fs/ext4/migrate.c index b57b98fb44d..6f07a06f243 100644 --- a/fs/ext4/migrate.c +++ b/fs/ext4/migrate.c @@ -15,7 +15,6 @@ #include #include #include "ext4_jbd2.h" -#include "ext4_extents.h" /* * The contiguous blocks details which can be diff --git a/fs/ext4/move_extent.c b/fs/ext4/move_extent.c index f57455a1b1b..c5826c623e7 100644 --- a/fs/ext4/move_extent.c +++ b/fs/ext4/move_extent.c @@ -17,7 +17,6 @@ #include #include #include "ext4_jbd2.h" -#include "ext4_extents.h" #include "ext4.h" /** diff --git a/include/trace/events/ext4.h b/include/trace/events/ext4.h index b50a5473624..c9a341e385a 100644 --- a/include/trace/events/ext4.h +++ b/include/trace/events/ext4.h @@ -12,6 +12,8 @@ struct ext4_allocation_request; struct ext4_prealloc_space; struct ext4_inode_info; struct mpage_da_data; +struct ext4_map_blocks; +struct ext4_extent; #define EXT4_I(inode) (container_of(inode, struct ext4_inode_info, vfs_inode)) @@ -1032,9 +1034,9 @@ TRACE_EVENT(ext4_forget, ); TRACE_EVENT(ext4_da_update_reserve_space, - TP_PROTO(struct inode *inode, int used_blocks), + TP_PROTO(struct inode *inode, int used_blocks, int quota_claim), - TP_ARGS(inode, used_blocks), + TP_ARGS(inode, used_blocks, quota_claim), TP_STRUCT__entry( __field( dev_t, dev ) @@ -1045,6 +1047,7 @@ TRACE_EVENT(ext4_da_update_reserve_space, __field( int, reserved_data_blocks ) __field( int, reserved_meta_blocks ) __field( int, allocated_meta_blocks ) + __field( int, quota_claim ) ), TP_fast_assign( @@ -1053,19 +1056,24 @@ TRACE_EVENT(ext4_da_update_reserve_space, __entry->mode = inode->i_mode; __entry->i_blocks = inode->i_blocks; __entry->used_blocks = used_blocks; - __entry->reserved_data_blocks = EXT4_I(inode)->i_reserved_data_blocks; - __entry->reserved_meta_blocks = EXT4_I(inode)->i_reserved_meta_blocks; - __entry->allocated_meta_blocks = EXT4_I(inode)->i_allocated_meta_blocks; + __entry->reserved_data_blocks = + EXT4_I(inode)->i_reserved_data_blocks; + __entry->reserved_meta_blocks = + EXT4_I(inode)->i_reserved_meta_blocks; + __entry->allocated_meta_blocks = + EXT4_I(inode)->i_allocated_meta_blocks; + __entry->quota_claim = quota_claim; ), TP_printk("dev %d,%d ino %lu mode 0%o i_blocks %llu used_blocks %d " "reserved_data_blocks %d reserved_meta_blocks %d " - "allocated_meta_blocks %d", + "allocated_meta_blocks %d quota_claim %d", MAJOR(__entry->dev), MINOR(__entry->dev), (unsigned long) __entry->ino, __entry->mode, __entry->i_blocks, __entry->used_blocks, __entry->reserved_data_blocks, - __entry->reserved_meta_blocks, __entry->allocated_meta_blocks) + __entry->reserved_meta_blocks, __entry->allocated_meta_blocks, + __entry->quota_claim) ); TRACE_EVENT(ext4_da_reserve_space, @@ -1589,6 +1597,382 @@ DEFINE_EVENT(ext4__trim, ext4_trim_all_free, TP_ARGS(sb, group, start, len) ); +TRACE_EVENT(ext4_ext_handle_uninitialized_extents, + TP_PROTO(struct inode *inode, struct ext4_map_blocks *map, + unsigned int allocated, ext4_fsblk_t newblock), + + TP_ARGS(inode, map, allocated, newblock), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, lblk ) + __field( ext4_fsblk_t, pblk ) + __field( unsigned int, len ) + __field( int, flags ) + __field( unsigned int, allocated ) + __field( ext4_fsblk_t, newblk ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->lblk = map->m_lblk; + __entry->pblk = map->m_pblk; + __entry->len = map->m_len; + __entry->flags = map->m_flags; + __entry->allocated = allocated; + __entry->newblk = newblock; + ), + + TP_printk("dev %d,%d ino %lu m_lblk %u m_pblk %llu m_len %u flags %d" + "allocated %d newblock %llu", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->lblk, (unsigned long long) __entry->pblk, + __entry->len, __entry->flags, + (unsigned int) __entry->allocated, + (unsigned long long) __entry->newblk) +); + +TRACE_EVENT(ext4_get_implied_cluster_alloc_exit, + TP_PROTO(struct super_block *sb, struct ext4_map_blocks *map, int ret), + + TP_ARGS(sb, map, ret), + + TP_STRUCT__entry( + __field( dev_t, dev ) + __field( ext4_lblk_t, lblk ) + __field( ext4_fsblk_t, pblk ) + __field( unsigned int, len ) + __field( unsigned int, flags ) + __field( int, ret ) + ), + + TP_fast_assign( + __entry->dev = sb->s_dev; + __entry->lblk = map->m_lblk; + __entry->pblk = map->m_pblk; + __entry->len = map->m_len; + __entry->flags = map->m_flags; + __entry->ret = ret; + ), + + TP_printk("dev %d,%d m_lblk %u m_pblk %llu m_len %u m_flags %u ret %d", + MAJOR(__entry->dev), MINOR(__entry->dev), + __entry->lblk, (unsigned long long) __entry->pblk, + __entry->len, __entry->flags, __entry->ret) +); + +TRACE_EVENT(ext4_ext_put_in_cache, + TP_PROTO(struct inode *inode, ext4_lblk_t lblk, unsigned int len, + ext4_fsblk_t start), + + TP_ARGS(inode, lblk, len, start), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, lblk ) + __field( unsigned int, len ) + __field( ext4_fsblk_t, start ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->lblk = lblk; + __entry->len = len; + __entry->start = start; + ), + + TP_printk("dev %d,%d ino %lu lblk %u len %u start %llu", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->lblk, + __entry->len, + (unsigned long long) __entry->start) +); + +TRACE_EVENT(ext4_ext_in_cache, + TP_PROTO(struct inode *inode, ext4_lblk_t lblk, int ret), + + TP_ARGS(inode, lblk, ret), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, lblk ) + __field( int, ret ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->lblk = lblk; + __entry->ret = ret; + ), + + TP_printk("dev %d,%d ino %lu lblk %u ret %d", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->lblk, + __entry->ret) + +); + +TRACE_EVENT(ext4_find_delalloc_range, + TP_PROTO(struct inode *inode, ext4_lblk_t from, ext4_lblk_t to, + int reverse, int found, ext4_lblk_t found_blk), + + TP_ARGS(inode, from, to, reverse, found, found_blk), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, from ) + __field( ext4_lblk_t, to ) + __field( int, reverse ) + __field( int, found ) + __field( ext4_lblk_t, found_blk ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->from = from; + __entry->to = to; + __entry->reverse = reverse; + __entry->found = found; + __entry->found_blk = found_blk; + ), + + TP_printk("dev %d,%d ino %lu from %u to %u reverse %d found %d " + "(blk = %u)", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->from, (unsigned) __entry->to, + __entry->reverse, __entry->found, + (unsigned) __entry->found_blk) +); + +TRACE_EVENT(ext4_get_reserved_cluster_alloc, + TP_PROTO(struct inode *inode, ext4_lblk_t lblk, unsigned int len), + + TP_ARGS(inode, lblk, len), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, lblk ) + __field( unsigned int, len ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->lblk = lblk; + __entry->len = len; + ), + + TP_printk("dev %d,%d ino %lu lblk %u len %u", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->lblk, + __entry->len) +); + +TRACE_EVENT(ext4_ext_show_extent, + TP_PROTO(struct inode *inode, ext4_lblk_t lblk, ext4_fsblk_t pblk, + unsigned short len), + + TP_ARGS(inode, lblk, pblk, len), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, lblk ) + __field( ext4_fsblk_t, pblk ) + __field( unsigned short, len ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->lblk = lblk; + __entry->pblk = pblk; + __entry->len = len; + ), + + TP_printk("dev %d,%d ino %lu lblk %u pblk %llu len %u", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->lblk, + (unsigned long long) __entry->pblk, + (unsigned short) __entry->len) +); + +TRACE_EVENT(ext4_remove_blocks, + TP_PROTO(struct inode *inode, struct ext4_extent *ex, + ext4_lblk_t from, ext4_fsblk_t to, + ext4_fsblk_t partial_cluster), + + TP_ARGS(inode, ex, from, to, partial_cluster), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, ee_lblk ) + __field( ext4_fsblk_t, ee_pblk ) + __field( unsigned short, ee_len ) + __field( ext4_lblk_t, from ) + __field( ext4_lblk_t, to ) + __field( ext4_fsblk_t, partial ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->ee_lblk = cpu_to_le32(ex->ee_block); + __entry->ee_pblk = ext4_ext_pblock(ex); + __entry->ee_len = ext4_ext_get_actual_len(ex); + __entry->from = from; + __entry->to = to; + __entry->partial = partial_cluster; + ), + + TP_printk("dev %d,%d ino %lu extent [%u(%llu), %u]" + "from %u to %u partial_cluster %u", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->ee_lblk, + (unsigned long long) __entry->ee_pblk, + (unsigned short) __entry->ee_len, + (unsigned) __entry->from, + (unsigned) __entry->to, + (unsigned) __entry->partial) +); + +TRACE_EVENT(ext4_ext_rm_leaf, + TP_PROTO(struct inode *inode, ext4_lblk_t start, + struct ext4_extent *ex, ext4_fsblk_t partial_cluster), + + TP_ARGS(inode, start, ex, partial_cluster), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, start ) + __field( ext4_lblk_t, ee_lblk ) + __field( ext4_fsblk_t, ee_pblk ) + __field( short, ee_len ) + __field( ext4_fsblk_t, partial ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->start = start; + __entry->ee_lblk = le32_to_cpu(ex->ee_block); + __entry->ee_pblk = ext4_ext_pblock(ex); + __entry->ee_len = ext4_ext_get_actual_len(ex); + __entry->partial = partial_cluster; + ), + + TP_printk("dev %d,%d ino %lu start_lblk %u last_extent [%u(%llu), %u]" + "partial_cluster %u", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->start, + (unsigned) __entry->ee_lblk, + (unsigned long long) __entry->ee_pblk, + (unsigned short) __entry->ee_len, + (unsigned) __entry->partial) +); + +TRACE_EVENT(ext4_ext_rm_idx, + TP_PROTO(struct inode *inode, ext4_fsblk_t pblk), + + TP_ARGS(inode, pblk), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_fsblk_t, pblk ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->pblk = pblk; + ), + + TP_printk("dev %d,%d ino %lu index_pblk %llu", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned long long) __entry->pblk) +); + +TRACE_EVENT(ext4_ext_remove_space, + TP_PROTO(struct inode *inode, ext4_lblk_t start, int depth), + + TP_ARGS(inode, start, depth), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, start ) + __field( int, depth ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->start = start; + __entry->depth = depth; + ), + + TP_printk("dev %d,%d ino %lu since %u depth %d", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->start, + __entry->depth) +); + +TRACE_EVENT(ext4_ext_remove_space_done, + TP_PROTO(struct inode *inode, ext4_lblk_t start, int depth, + ext4_lblk_t partial, unsigned short eh_entries), + + TP_ARGS(inode, start, depth, partial, eh_entries), + + TP_STRUCT__entry( + __field( ino_t, ino ) + __field( dev_t, dev ) + __field( ext4_lblk_t, start ) + __field( int, depth ) + __field( ext4_lblk_t, partial ) + __field( unsigned short, eh_entries ) + ), + + TP_fast_assign( + __entry->ino = inode->i_ino; + __entry->dev = inode->i_sb->s_dev; + __entry->start = start; + __entry->depth = depth; + __entry->partial = partial; + __entry->eh_entries = eh_entries; + ), + + TP_printk("dev %d,%d ino %lu since %u depth %d partial %u " + "remaining_entries %u", + MAJOR(__entry->dev), MINOR(__entry->dev), + (unsigned long) __entry->ino, + (unsigned) __entry->start, + __entry->depth, + (unsigned) __entry->partial, + (unsigned short) __entry->eh_entries) +); + #endif /* _TRACE_EXT4_H */ /* This part must be outside protection */ -- cgit v1.2.3 From 5356f2615cd558c57a1f7d7528d1ad4de3640d96 Mon Sep 17 00:00:00 2001 From: Aditya Kali Date: Fri, 9 Sep 2011 19:20:51 -0400 Subject: ext4: attempt to fix race in bigalloc code path Currently, there exists a race between delayed allocated writes and the writeback when bigalloc feature is in use. The race was because we wanted to determine what blocks in a cluster are under delayed allocation and we were using buffer_delayed(bh) check for it. But, the writeback codepath clears this bit without any synchronization which resulted in a race and an ext4 warning similar to: EXT4-fs (ram1): ext4_da_update_reserve_space: ino 13, used 1 with only 0 reserved data blocks The race existed in two places. (1) between ext4_find_delalloc_range() and ext4_map_blocks() when called from writeback code path. (2) between ext4_find_delalloc_range() and ext4_da_get_block_prep() (where buffer_delayed(bh) is set. To fix (1), this patch introduces a new buffer_head state bit - BH_Da_Mapped. This bit is set under the protection of EXT4_I(inode)->i_data_sem when we have actually mapped the delayed allocated blocks during the writeout time. We can now reliably check for this bit inside ext4_find_delalloc_range() to determine whether the reservation for the blocks have already been claimed or not. To fix (2), it was necessary to set buffer_delay(bh) under the protection of i_data_sem. So, I extracted the very beginning of ext4_map_blocks into a new function - ext4_da_map_blocks() - and performed the required setting of bh_delay bit and the quota reservation under the protection of i_data_sem. These two fixes makes the checking of buffer_delay(bh) and buffer_da_mapped(bh) consistent, thus removing the race. Tested: I was able to reproduce the problem by running 'dd' and 'fsync' in parallel. Also, xfstests sometimes used to reproduce this race. After the fix both my test and xfstests were successful and no race (warning message) was observed. Google-Bug-Id: 4997027 Signed-off-by: Aditya Kali Signed-off-by: "Theodore Ts'o" --- fs/ext4/ext4.h | 5 +- fs/ext4/extents.c | 38 +++++--------- fs/ext4/inode.c | 146 +++++++++++++++++++++++++++++++++++++++++++----------- 3 files changed, 134 insertions(+), 55 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 751277a4890..1bbd2caebe7 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -1893,7 +1893,6 @@ extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); extern qsize_t *ext4_get_reserved_space(struct inode *inode); extern void ext4_da_update_reserve_space(struct inode *inode, int used, int quota_claim); -extern int ext4_da_reserve_space(struct inode *inode, ext4_lblk_t lblock); /* indirect.c */ extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, @@ -2300,10 +2299,14 @@ enum ext4_state_bits { * never, ever appear in a buffer_head's state * flag. See EXT4_MAP_FROM_CLUSTER to see where * this is used. */ + BH_Da_Mapped, /* Delayed allocated block that now has a mapping. This + * flag is set when ext4_map_blocks is called on a + * delayed allocated block to get its real mapping. */ }; BUFFER_FNS(Uninit, uninit) TAS_BUFFER_FNS(Uninit, uninit) +BUFFER_FNS(Da_Mapped, da_mapped) /* * Add new method to test wether block and inode bitmaps are properly diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index 9b119308dae..ad39627c1fb 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -3296,28 +3296,9 @@ static int ext4_find_delalloc_range(struct inode *inode, while ((i >= lblk_start) && (i <= lblk_end)) { page = find_get_page(mapping, index); - if (!page || !PageDirty(page)) + if (!page) goto nextpage; - if (PageWriteback(page)) { - /* - * This might be a race with allocation and writeout. In - * this case we just assume that the rest of the range - * will eventually be written and there wont be any - * delalloc blocks left. - * TODO: the above assumption is troublesome, but might - * work better in practice. other option could be note - * somewhere that the cluster is getting written out and - * detect that here. - */ - page_cache_release(page); - trace_ext4_find_delalloc_range(inode, - lblk_start, lblk_end, - search_hint_reverse, - 0, i); - return 0; - } - if (!page_has_buffers(page)) goto nextpage; @@ -3340,7 +3321,11 @@ static int ext4_find_delalloc_range(struct inode *inode, continue; } - if (buffer_delay(bh)) { + /* Check if the buffer is delayed allocated and that it + * is not yet mapped. (when da-buffers are mapped during + * their writeout, their da_mapped bit is set.) + */ + if (buffer_delay(bh) && !buffer_da_mapped(bh)) { page_cache_release(page); trace_ext4_find_delalloc_range(inode, lblk_start, lblk_end, @@ -4106,6 +4091,7 @@ got_allocated_blocks: ext4_da_update_reserve_space(inode, allocated_clusters, 1); if (reserved_clusters < allocated_clusters) { + struct ext4_inode_info *ei = EXT4_I(inode); int reservation = allocated_clusters - reserved_clusters; /* @@ -4148,11 +4134,11 @@ got_allocated_blocks: * remaining blocks finally gets written, we * could claim them. */ - while (reservation) { - ext4_da_reserve_space(inode, - map->m_lblk); - reservation--; - } + dquot_reserve_block(inode, + EXT4_C2B(sbi, reservation)); + spin_lock(&ei->i_block_reservation_lock); + ei->i_reserved_data_blocks += reservation; + spin_unlock(&ei->i_block_reservation_lock); } } } diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 2dcd4fed96e..1380cd29c31 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -397,6 +397,49 @@ static pgoff_t ext4_num_dirty_pages(struct inode *inode, pgoff_t idx, return num; } +/* + * Sets the BH_Da_Mapped bit on the buffer heads corresponding to the given map. + */ +static void set_buffers_da_mapped(struct inode *inode, + struct ext4_map_blocks *map) +{ + struct address_space *mapping = inode->i_mapping; + struct pagevec pvec; + int i, nr_pages; + pgoff_t index, end; + + index = map->m_lblk >> (PAGE_CACHE_SHIFT - inode->i_blkbits); + end = (map->m_lblk + map->m_len - 1) >> + (PAGE_CACHE_SHIFT - inode->i_blkbits); + + pagevec_init(&pvec, 0); + while (index <= end) { + nr_pages = pagevec_lookup(&pvec, mapping, index, + min(end - index + 1, + (pgoff_t)PAGEVEC_SIZE)); + if (nr_pages == 0) + break; + for (i = 0; i < nr_pages; i++) { + struct page *page = pvec.pages[i]; + struct buffer_head *bh, *head; + + if (unlikely(page->mapping != mapping) || + !PageDirty(page)) + break; + + if (page_has_buffers(page)) { + bh = head = page_buffers(page); + do { + set_buffer_da_mapped(bh); + bh = bh->b_this_page; + } while (bh != head); + } + index++; + } + pagevec_release(&pvec); + } +} + /* * The ext4_map_blocks() function tries to look up the requested blocks, * and returns if the blocks are already mapped. @@ -516,9 +559,17 @@ int ext4_map_blocks(handle_t *handle, struct inode *inode, (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)) ext4_da_update_reserve_space(inode, retval, 1); } - if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) + if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) { ext4_clear_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED); + /* If we have successfully mapped the delayed allocated blocks, + * set the BH_Da_Mapped bit on them. Its important to do this + * under the protection of i_data_sem. + */ + if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) + set_buffers_da_mapped(inode, map); + } + up_write((&EXT4_I(inode)->i_data_sem)); if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) { int ret = check_block_validity(inode, map); @@ -1038,7 +1089,7 @@ static int ext4_journalled_write_end(struct file *file, /* * Reserve a single cluster located at lblock */ -int ext4_da_reserve_space(struct inode *inode, ext4_lblk_t lblock) +static int ext4_da_reserve_space(struct inode *inode, ext4_lblk_t lblock) { int retries = 0; struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); @@ -1153,6 +1204,7 @@ static void ext4_da_page_release_reservation(struct page *page, if ((offset <= curr_off) && (buffer_delay(bh))) { to_release++; clear_buffer_delay(bh); + clear_buffer_da_mapped(bh); } curr_off = next_off; } while ((bh = bh->b_this_page) != head); @@ -1271,6 +1323,8 @@ static int mpage_da_submit_io(struct mpage_da_data *mpd, clear_buffer_delay(bh); bh->b_blocknr = pblock; } + if (buffer_da_mapped(bh)) + clear_buffer_da_mapped(bh); if (buffer_unwritten(bh) || buffer_mapped(bh)) BUG_ON(bh->b_blocknr != pblock); @@ -1603,6 +1657,66 @@ static int ext4_bh_delay_or_unwritten(handle_t *handle, struct buffer_head *bh) return (buffer_delay(bh) || buffer_unwritten(bh)) && buffer_dirty(bh); } +/* + * This function is grabs code from the very beginning of + * ext4_map_blocks, but assumes that the caller is from delayed write + * time. This function looks up the requested blocks and sets the + * buffer delay bit under the protection of i_data_sem. + */ +static int ext4_da_map_blocks(struct inode *inode, sector_t iblock, + struct ext4_map_blocks *map, + struct buffer_head *bh) +{ + int retval; + sector_t invalid_block = ~((sector_t) 0xffff); + + if (invalid_block < ext4_blocks_count(EXT4_SB(inode->i_sb)->s_es)) + invalid_block = ~0; + + map->m_flags = 0; + ext_debug("ext4_da_map_blocks(): inode %lu, max_blocks %u," + "logical block %lu\n", inode->i_ino, map->m_len, + (unsigned long) map->m_lblk); + /* + * Try to see if we can get the block without requesting a new + * file system block. + */ + down_read((&EXT4_I(inode)->i_data_sem)); + if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) + retval = ext4_ext_map_blocks(NULL, inode, map, 0); + else + retval = ext4_ind_map_blocks(NULL, inode, map, 0); + + if (retval == 0) { + /* + * XXX: __block_prepare_write() unmaps passed block, + * is it OK? + */ + /* If the block was allocated from previously allocated cluster, + * then we dont need to reserve it again. */ + if (!(map->m_flags & EXT4_MAP_FROM_CLUSTER)) { + retval = ext4_da_reserve_space(inode, iblock); + if (retval) + /* not enough space to reserve */ + goto out_unlock; + } + + /* Clear EXT4_MAP_FROM_CLUSTER flag since its purpose is served + * and it should not appear on the bh->b_state. + */ + map->m_flags &= ~EXT4_MAP_FROM_CLUSTER; + + map_bh(bh, inode->i_sb, invalid_block); + set_buffer_new(bh); + set_buffer_delay(bh); + } + +out_unlock: + up_read((&EXT4_I(inode)->i_data_sem)); + + return retval; +} + /* * This is a special get_blocks_t callback which is used by * ext4_da_write_begin(). It will either return mapped block or @@ -1620,10 +1734,6 @@ static int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, { struct ext4_map_blocks map; int ret = 0; - sector_t invalid_block = ~((sector_t) 0xffff); - - if (invalid_block < ext4_blocks_count(EXT4_SB(inode->i_sb)->s_es)) - invalid_block = ~0; BUG_ON(create == 0); BUG_ON(bh->b_size != inode->i_sb->s_blocksize); @@ -1636,29 +1746,9 @@ static int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, * preallocated blocks are unmapped but should treated * the same as allocated blocks. */ - ret = ext4_map_blocks(NULL, inode, &map, 0); - if (ret < 0) + ret = ext4_da_map_blocks(inode, iblock, &map, bh); + if (ret <= 0) return ret; - if (ret == 0) { - if (buffer_delay(bh)) - return 0; /* Not sure this could or should happen */ - /* - * XXX: __block_write_begin() unmaps passed block, is it OK? - */ - /* If the block was allocated from previously allocated cluster, - * then we dont need to reserve it again. */ - if (!(map.m_flags & EXT4_MAP_FROM_CLUSTER)) { - ret = ext4_da_reserve_space(inode, iblock); - if (ret) - /* not enough space to reserve */ - return ret; - } - - map_bh(bh, inode->i_sb, invalid_block); - set_buffer_new(bh); - set_buffer_delay(bh); - return 0; - } map_bh(bh, inode->i_sb, map.m_pblk); bh->b_state = (bh->b_state & ~EXT4_MAP_FLAGS) | map.m_flags; -- cgit v1.2.3 From df3ab17072c31fbd394614711772682f0a956a2c Mon Sep 17 00:00:00 2001 From: Tao Ma Date: Sat, 8 Oct 2011 15:53:49 -0400 Subject: ext4: fix the comment describing ext4_ext_search_right() The comment describing what ext4_ext_search_right() does is incorrect. We return 0 in *phys when *logical is the 'largest' allocated block, not smallest. Fix a few other typos while we're at it. Cc: "Theodore Ts'o" Signed-off-by: Tao Ma --- fs/ext4/extents.c | 4 ++-- fs/ext4/inode.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index ad39627c1fb..142e3443476 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -1263,7 +1263,7 @@ static int ext4_ext_search_left(struct inode *inode, /* * search the closest allocated block to the right for *logical * and returns it at @logical + it's physical address at @phys - * if *logical is the smallest allocated block, the function + * if *logical is the largest allocated block, the function * returns 0 at @phys * return value contains 0 (success) or error code */ @@ -2168,7 +2168,7 @@ int ext4_ext_calc_credits_for_single_extent(struct inode *inode, int nrblocks, * need to account for leaf block credit * * bitmaps and block group descriptor blocks - * and other metadat blocks still need to be + * and other metadata blocks still need to be * accounted. */ /* 1 bitmap, 1 block group descriptor */ diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 1380cd29c31..4863238c375 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -458,7 +458,7 @@ static void set_buffers_da_mapped(struct inode *inode, * the buffer head is mapped. * * It returns 0 if plain look up failed (blocks have not been allocated), in - * that casem, buffer head is unmapped + * that case, buffer head is unmapped * * It returns the error in case of allocation failure. */ @@ -497,7 +497,7 @@ int ext4_map_blocks(handle_t *handle, struct inode *inode, * Returns if the blocks have already allocated * * Note that if blocks have been preallocated - * ext4_ext_get_block() returns th create = 0 + * ext4_ext_get_block() returns the create = 0 * with buffer head unmapped. */ if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) -- cgit v1.2.3 From 1bce63d1a2a2c8929442b79acd4eab2e3db10a0b Mon Sep 17 00:00:00 2001 From: Shaohua Li Date: Tue, 18 Oct 2011 10:55:51 -0400 Subject: ext4: add block plug for .writepages Add block plug for ext4 .writepages. Though ext4 .writepages already handles request merge very well, block plug is still helpful to reduce block lock contention. Signed-off-by: Shaohua Li Signed-off-by: "Theodore Ts'o" --- fs/ext4/inode.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 4863238c375..081bb25a9ad 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -2160,6 +2160,7 @@ static int ext4_da_writepages(struct address_space *mapping, struct ext4_sb_info *sbi = EXT4_SB(mapping->host->i_sb); pgoff_t done_index = 0; pgoff_t end; + struct blk_plug plug; trace_ext4_da_writepages(inode, wbc); @@ -2238,6 +2239,7 @@ retry: if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages) tag_pages_for_writeback(mapping, index, end); + blk_start_plug(&plug); while (!ret && wbc->nr_to_write > 0) { /* @@ -2302,6 +2304,7 @@ retry: */ break; } + blk_finish_plug(&plug); if (!io_done && !cycled) { cycled = 1; index = 0; -- cgit v1.2.3 From 09e0834fb0ce1ea2a63885177015bd5d7d2bc22d Mon Sep 17 00:00:00 2001 From: Akira Fujita Date: Thu, 20 Oct 2011 18:56:10 -0400 Subject: ext4: fix deadlock in ext4_ordered_write_end() If ext4_jbd2_file_inode() in ext4_ordered_write_end() fails for some reasons, this function returns to caller without unlocking the page. It leads to the deadlock, and the patch fixes this issue. Signed-off-by: Akira Fujita Signed-off-by: "Theodore Ts'o" --- fs/ext4/inode.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 081bb25a9ad..ecb57259192 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -959,7 +959,11 @@ static int ext4_ordered_write_end(struct file *file, ext4_orphan_add(handle, inode); if (ret2 < 0) ret = ret2; + } else { + unlock_page(page); + page_cache_release(page); } + ret2 = ext4_journal_stop(handle); if (!ret) ret = ret2; -- cgit v1.2.3 From 8de49e674a1133ab8998914a6e933ceb4b5f4b07 Mon Sep 17 00:00:00 2001 From: Kazuya Mio Date: Thu, 20 Oct 2011 19:23:08 -0400 Subject: ext4: fix the deadlock in mpage_da_map_and_submit() If ext4_jbd2_file_inode() in mpage_da_map_and_submit() fails due to journal abort, this function returns to caller without unlocking the page. It leads to the deadlock, and the patch fixes this issue by calling mpage_da_submit_io(). Signed-off-by: Kazuya Mio Signed-off-by: "Theodore Ts'o" --- fs/ext4/inode.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index ecb57259192..ff6aace0bb3 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -1552,9 +1552,11 @@ static void mpage_da_map_and_submit(struct mpage_da_data *mpd) if (ext4_should_order_data(mpd->inode)) { err = ext4_jbd2_file_inode(handle, mpd->inode); - if (err) + if (err) { /* Only if the journal is aborted */ - return; + mpd->retval = err; + goto submit_io; + } } } @@ -2294,11 +2296,12 @@ retry: ret = 0; } else if (ret == MPAGE_DA_EXTENT_TAIL) { /* - * got one extent now try with - * rest of the pages + * Got one extent now try with rest of the pages. + * If mpd.retval is set -EIO, journal is aborted. + * So we don't need to write any more. */ pages_written += mpd.pages_written; - ret = 0; + ret = mpd.retval; io_done = 1; } else if (wbc->nr_to_write) /* -- cgit v1.2.3 From a4e5d88b1b24827f4f6a3cba43228936a67d81ba Mon Sep 17 00:00:00 2001 From: Dmitry Monakhov Date: Tue, 25 Oct 2011 08:15:12 -0400 Subject: ext4: update EOFBLOCKS flag on fallocate properly EOFBLOCK_FL should be updated if called w/o FALLOCATE_FL_KEEP_SIZE Currently it happens only if new extent was allocated. TESTCASE: fallocate test_file -n -l4096 fallocate test_file -l4096 Last fallocate cmd has updated size, but keept EOFBLOCK_FL set. And fsck will complain about that. Also remove ping pong in ext4_fallocate() in case of new extents, where ext4_ext_map_blocks() clear EOFBLOCKS bit, and later ext4_falloc_update_inode() restore it again. Signed-off-by: Dmitry Monakhov Signed-off-by: "Theodore Ts'o" --- fs/ext4/ext4.h | 2 ++ fs/ext4/extents.c | 28 +++++++++++++++++----------- fs/ext4/inode.c | 6 ++++-- 3 files changed, 23 insertions(+), 13 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 4546da4f26c..3647ae0b21a 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -539,6 +539,8 @@ struct ext4_new_group_data { #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020 /* Don't normalize allocation size (used for fallocate) */ #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 + /* Request will not result in inode size update (user for fallocate) */ +#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080 /* * Flags used by ext4_free_blocks diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index d0e3f333d3f..8686eb756b3 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -3432,14 +3432,8 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode, /* buffered write, writepage time, convert*/ ret = ext4_ext_convert_to_initialized(handle, inode, map, path); - if (ret >= 0) { + if (ret >= 0) ext4_update_inode_fsync_trans(handle, inode, 1); - err = check_eofblocks_fl(handle, inode, map->m_lblk, path, - map->m_len); - if (err < 0) - goto out2; - } - out: if (ret <= 0) { err = ret; @@ -3480,6 +3474,12 @@ out: map_out: map->m_flags |= EXT4_MAP_MAPPED; + if ((flags & EXT4_GET_BLOCKS_KEEP_SIZE) == 0) { + err = check_eofblocks_fl(handle, inode, map->m_lblk, path, + map->m_len); + if (err < 0) + goto out2; + } out1: if (allocated > map->m_len) allocated = map->m_len; @@ -3955,7 +3955,10 @@ got_allocated_blocks: map->m_flags |= EXT4_MAP_UNINIT; } - err = check_eofblocks_fl(handle, inode, map->m_lblk, path, ar.len); + err = 0; + if ((flags & EXT4_GET_BLOCKS_KEEP_SIZE) == 0) + err = check_eofblocks_fl(handle, inode, map->m_lblk, + path, ar.len); if (!err) err = ext4_ext_insert_extent(handle, inode, path, &newex, flags); @@ -4214,6 +4217,7 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len) int ret = 0; int ret2 = 0; int retries = 0; + int flags; struct ext4_map_blocks map; unsigned int credits, blkbits = inode->i_blkbits; @@ -4250,6 +4254,10 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len) trace_ext4_fallocate_exit(inode, offset, max_blocks, ret); return ret; } + flags = EXT4_GET_BLOCKS_CREATE_UNINIT_EXT | + EXT4_GET_BLOCKS_NO_NORMALIZE; + if (mode & FALLOC_FL_KEEP_SIZE) + flags |= EXT4_GET_BLOCKS_KEEP_SIZE; retry: while (ret >= 0 && ret < max_blocks) { map.m_lblk = map.m_lblk + ret; @@ -4259,9 +4267,7 @@ retry: ret = PTR_ERR(handle); break; } - ret = ext4_map_blocks(handle, inode, &map, - EXT4_GET_BLOCKS_CREATE_UNINIT_EXT | - EXT4_GET_BLOCKS_NO_NORMALIZE); + ret = ext4_map_blocks(handle, inode, &map, flags); if (ret <= 0) { #ifdef EXT4FS_DEBUG WARN_ON(ret <= 0); diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index ff6aace0bb3..87ec615f0fd 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -477,9 +477,11 @@ int ext4_map_blocks(handle_t *handle, struct inode *inode, */ down_read((&EXT4_I(inode)->i_data_sem)); if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) { - retval = ext4_ext_map_blocks(handle, inode, map, 0); + retval = ext4_ext_map_blocks(handle, inode, map, flags & + EXT4_GET_BLOCKS_KEEP_SIZE); } else { - retval = ext4_ind_map_blocks(handle, inode, map, 0); + retval = ext4_ind_map_blocks(handle, inode, map, flags & + EXT4_GET_BLOCKS_KEEP_SIZE); } up_read((&EXT4_I(inode)->i_data_sem)); -- cgit v1.2.3 From fcbb5515825f1bb20b7a6f75ec48bee61416f879 Mon Sep 17 00:00:00 2001 From: Yongqiang Yang Date: Wed, 26 Oct 2011 05:00:19 -0400 Subject: ext4: let ext4_page_mkwrite stop started handle in failure The started journal handle should be stopped in failure case. Signed-off-by: Yongqiang Yang Signed-off-by: "Theodore Ts'o" Acked-by: Jan Kara Cc: stable@kernel.org --- fs/ext4/inode.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 87ec615f0fd..e4b26faac5f 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -4791,6 +4791,7 @@ retry_alloc: PAGE_CACHE_SIZE, NULL, do_journal_get_write_access)) { unlock_page(page); ret = VM_FAULT_SIGBUS; + ext4_journal_stop(handle); goto out; } ext4_set_inode_state(inode, EXT4_STATE_JDATA); -- cgit v1.2.3 From 0edeb71dc9133bfb505d3bf59642e07cd936613e Mon Sep 17 00:00:00 2001 From: Tao Ma Date: Mon, 31 Oct 2011 17:30:44 -0400 Subject: ext4: Create helper function for EXT4_IO_END_UNWRITTEN and i_aiodio_unwritten EXT4_IO_END_UNWRITTEN flag set and the increase of i_aiodio_unwritten should be done simultaneously since ext4_end_io_nolock always clear the flag and decrease the counter in the same time. We have found some bugs that the flag is set while leaving i_aiodio_unwritten unchanged(commit 32c80b32c053d). So this patch just tries to create a helper function to wrap them to avoid any future bug. The idea is inspired by Eric. Cc: Eric Sandeen Signed-off-by: Tao Ma Signed-off-by: "Theodore Ts'o" --- fs/ext4/ext4.h | 9 +++++++++ fs/ext4/extents.c | 18 ++++++------------ fs/ext4/inode.c | 5 +---- fs/ext4/page-io.c | 6 ++---- 4 files changed, 18 insertions(+), 20 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 657e82649fa..604c200216c 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -1275,6 +1275,15 @@ static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino) ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)); } +static inline void ext4_set_io_unwritten_flag(struct inode *inode, + struct ext4_io_end *io_end) +{ + if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { + io_end->flag |= EXT4_IO_END_UNWRITTEN; + atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten); + } +} + /* * Inode dynamic state flags */ diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index 36a0f177eca..a5c8caaaa09 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -3472,12 +3472,9 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode, * that this IO needs to conversion to written when IO is * completed */ - if (io) { - if (!(io->flag & EXT4_IO_END_UNWRITTEN)) { - io->flag = EXT4_IO_END_UNWRITTEN; - atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten); - } - } else + if (io) + ext4_set_io_unwritten_flag(inode, io); + else ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN); if (ext4_should_dioread_nolock(inode)) map->m_flags |= EXT4_MAP_UNINIT; @@ -4030,12 +4027,9 @@ got_allocated_blocks: * that we need to perform conversion when IO is done. */ if ((flags & EXT4_GET_BLOCKS_PRE_IO)) { - if (io) { - if (!(io->flag & EXT4_IO_END_UNWRITTEN)) { - io->flag = EXT4_IO_END_UNWRITTEN; - atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten); - } - } else + if (io) + ext4_set_io_unwritten_flag(inode, io); + else ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN); } diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index e4b26faac5f..60af5126eb0 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -2831,10 +2831,7 @@ static void ext4_end_io_buffer_write(struct buffer_head *bh, int uptodate) * but being more careful is always safe for the future change. */ inode = io_end->inode; - if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) { - io_end->flag |= EXT4_IO_END_UNWRITTEN; - atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten); - } + ext4_set_io_unwritten_flag(inode, io_end); /* Add the io_end to per-inode completed io list*/ spin_lock_irqsave(&EXT4_I(inode)->i_completed_io_lock, flags); diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c index 9eebf44646f..7ce1d0b19c9 100644 --- a/fs/ext4/page-io.c +++ b/fs/ext4/page-io.c @@ -336,10 +336,8 @@ submit_and_retry: if ((io_end->num_io_pages >= MAX_IO_PAGES) && (io_end->pages[io_end->num_io_pages-1] != io_page)) goto submit_and_retry; - if (buffer_uninit(bh) && !(io_end->flag & EXT4_IO_END_UNWRITTEN)) { - io_end->flag |= EXT4_IO_END_UNWRITTEN; - atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten); - } + if (buffer_uninit(bh)) + ext4_set_io_unwritten_flag(inode, io_end); io->io_end->size += bh->b_size; io->io_next_block++; ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh)); -- cgit v1.2.3 From e260daf27902b2189a9198f5b64fa4567939bb5b Mon Sep 17 00:00:00 2001 From: Yongqiang Yang Date: Mon, 31 Oct 2011 17:54:36 -0400 Subject: ext4: move vars to local scope in ext4_discard_partial_page_buffers_no_lock() Signed-off-by: Yongqiang Yang Signed-off-by: "Theodore Ts'o" --- fs/ext4/inode.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 60af5126eb0..f60b459b27d 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -3179,7 +3179,6 @@ int ext4_discard_partial_page_buffers_no_lock(handle_t *handle, ext4_fsblk_t index = from >> PAGE_CACHE_SHIFT; unsigned int offset = from & (PAGE_CACHE_SIZE-1); unsigned int blocksize, max, pos; - unsigned int end_of_block, range_to_discard; ext4_lblk_t iblock; struct buffer_head *bh; int err = 0; @@ -3231,6 +3230,8 @@ int ext4_discard_partial_page_buffers_no_lock(handle_t *handle, pos = offset; while (pos < offset + length) { + unsigned int end_of_block, range_to_discard; + err = 0; /* The length of space left to zero and unmap */ -- cgit v1.2.3 From 5129d05fda57be13f434dbe8536de39a6c25496d Mon Sep 17 00:00:00 2001 From: Yongqiang Yang Date: Mon, 31 Oct 2011 17:56:10 -0400 Subject: ext4: return ENOMEM if find_or_create_pages fails Signed-off-by: Yongqiang Yang Signed-off-by: "Theodore Ts'o" --- fs/ext4/inode.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index f60b459b27d..5fcef98f98e 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -3129,7 +3129,7 @@ int ext4_discard_partial_page_buffers(handle_t *handle, page = find_or_create_page(mapping, from >> PAGE_CACHE_SHIFT, mapping_gfp_mask(mapping) & ~__GFP_FS); if (!page) - return -EINVAL; + return -ENOMEM; err = ext4_discard_partial_page_buffers_no_lock(handle, inode, page, from, length, flags); @@ -3372,7 +3372,7 @@ int ext4_block_zero_page_range(handle_t *handle, page = find_or_create_page(mapping, from >> PAGE_CACHE_SHIFT, mapping_gfp_mask(mapping) & ~__GFP_FS); if (!page) - return -EINVAL; + return -ENOMEM; blocksize = inode->i_sb->s_blocksize; max = blocksize - (offset & (blocksize - 1)); -- cgit v1.2.3 From edb5ac8993e25143f6af1ab143843a65c52e2a15 Mon Sep 17 00:00:00 2001 From: Yongqiang Yang Date: Mon, 31 Oct 2011 18:04:38 -0400 Subject: ext4: let ext4_discard_partial_buffers handle unaligned range correctly As comment says, we should handle unaligned range rather than aligned one. This fixes a bug found by running xfstests #91. Signed-off-by: Yongqiang Yang --- fs/ext4/inode.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/ext4/inode.c') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 5fcef98f98e..de05e86f199 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -3206,8 +3206,8 @@ int ext4_discard_partial_page_buffers_no_lock(handle_t *handle, * to be updated with the contents of the block before * we write the zeros on top of it. */ - if (!(from & (blocksize - 1)) || - !((from + length) & (blocksize - 1))) { + if ((from & (blocksize - 1)) || + ((from + length) & (blocksize - 1))) { create_empty_buffers(page, blocksize, 0); } else { /* -- cgit v1.2.3