From 61391d562229ed94899ed4b4973dc2f0c015292a Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Fri, 9 May 2014 17:17:40 +0100 Subject: [PATCH 01/80] Btrfs: fix hang on error (such as ENOSPC) when writing extent pages When running low on available disk space and having several processes doing buffered file IO, I got the following trace in dmesg: [ 4202.720152] INFO: task kworker/u8:1:5450 blocked for more than 120 seconds. [ 4202.720401] Not tainted 3.13.0-fdm-btrfs-next-26+ #1 [ 4202.720596] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 4202.720874] kworker/u8:1 D 0000000000000001 0 5450 2 0x00000000 [ 4202.720904] Workqueue: btrfs-flush_delalloc normal_work_helper [btrfs] [ 4202.720908] ffff8801f62ddc38 0000000000000082 ffff880203ac2490 00000000001d3f40 [ 4202.720913] ffff8801f62ddfd8 00000000001d3f40 ffff8800c4f0c920 ffff880203ac2490 [ 4202.720918] 00000000001d4a40 ffff88020fe85a40 ffff88020fe85ab8 0000000000000001 [ 4202.720922] Call Trace: [ 4202.720931] [] schedule+0x29/0x70 [ 4202.720950] [] btrfs_start_ordered_extent+0x6d/0x110 [btrfs] [ 4202.720956] [] ? bit_waitqueue+0xc0/0xc0 [ 4202.720972] [] btrfs_run_ordered_extent_work+0x29/0x40 [btrfs] [ 4202.720988] [] normal_work_helper+0x137/0x2c0 [btrfs] [ 4202.720994] [] process_one_work+0x1f5/0x530 (...) [ 4202.721027] 2 locks held by kworker/u8:1/5450: [ 4202.721028] #0: (%s-%s){++++..}, at: [] process_one_work+0x193/0x530 [ 4202.721037] #1: ((&work->normal_work)){+.+...}, at: [] process_one_work+0x193/0x530 [ 4202.721054] INFO: task btrfs:7891 blocked for more than 120 seconds. [ 4202.721258] Not tainted 3.13.0-fdm-btrfs-next-26+ #1 [ 4202.721444] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 4202.721699] btrfs D 0000000000000001 0 7891 7890 0x00000001 [ 4202.721704] ffff88018c2119e8 0000000000000086 ffff8800a33d2490 00000000001d3f40 [ 4202.721710] ffff88018c211fd8 00000000001d3f40 ffff8802144b0000 ffff8800a33d2490 [ 4202.721714] ffff8800d8576640 ffff88020fe85bc0 ffff88020fe85bc8 7fffffffffffffff [ 4202.721718] Call Trace: [ 4202.721723] [] schedule+0x29/0x70 [ 4202.721727] [] schedule_timeout+0x1dc/0x270 [ 4202.721732] [] ? mark_held_locks+0xb9/0x140 [ 4202.721736] [] ? _raw_spin_unlock_irq+0x30/0x40 [ 4202.721740] [] ? trace_hardirqs_on_caller+0x10d/0x1d0 [ 4202.721744] [] wait_for_completion+0xdf/0x120 [ 4202.721749] [] ? try_to_wake_up+0x310/0x310 [ 4202.721765] [] btrfs_wait_ordered_extents+0x1f4/0x280 [btrfs] [ 4202.721781] [] btrfs_mksubvol.isra.62+0x30e/0x5a0 [btrfs] [ 4202.721786] [] ? bit_waitqueue+0xc0/0xc0 [ 4202.721799] [] btrfs_ioctl_snap_create_transid+0x1a9/0x1b0 [btrfs] [ 4202.721813] [] btrfs_ioctl_snap_create_v2+0x10a/0x170 [btrfs] (...) It turns out that extent_io.c:__extent_writepage(), which ends up being called through filemap_fdatawrite_range() in btrfs_start_ordered_extent(), was getting -ENOSPC when calling the fill_delalloc callback. In this situation, it returned without the writepage_end_io_hook callback (inode.c:btrfs_writepage_end_io_hook) ever being called for the respective page, which prevents the ordered extent's bytes_left count from ever reaching 0, and therefore a finish_ordered_fn work is never queued into the endio_write_workers queue. This makes the task that called btrfs_start_ordered_extent() hang forever on the wait queue of the ordered extent. This is fairly easy to reproduce using a small filesystem and fsstress on a quad core vm: mkfs.btrfs -f -b `expr 2100 \* 1024 \* 1024` /dev/sdd mount /dev/sdd /mnt fsstress -p 6 -d /mnt -n 100000 -x \ "btrfs subvolume snapshot -r /mnt /mnt/mysnap" \ -f allocsp=0 \ -f bulkstat=0 \ -f bulkstat1=0 \ -f chown=0 \ -f creat=1 \ -f dread=0 \ -f dwrite=0 \ -f fallocate=1 \ -f fdatasync=0 \ -f fiemap=0 \ -f freesp=0 \ -f fsync=0 \ -f getattr=0 \ -f getdents=0 \ -f link=0 \ -f mkdir=0 \ -f mknod=0 \ -f punch=1 \ -f read=0 \ -f readlink=0 \ -f rename=0 \ -f resvsp=0 \ -f rmdir=0 \ -f setxattr=0 \ -f stat=0 \ -f symlink=0 \ -f sync=0 \ -f truncate=1 \ -f unlink=0 \ -f unresvsp=0 \ -f write=4 So just ensure that if an error happens while writing the extent page we call the writepage_end_io_hook callback. Also make it return the error code and ensure the caller (extent_write_cache_pages) processes all pages in the page vector even if an error happens only for some of them, so that ordered extents end up released. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-) diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 3955e475ceec..fa31c8d2c095 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -3278,6 +3278,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, end - cur + 1, 1); if (IS_ERR_OR_NULL(em)) { SetPageError(page); + ret = PTR_ERR_OR_ZERO(em); break; } @@ -3364,13 +3365,17 @@ done: set_page_writeback(page); end_page_writeback(page); } + if (PageError(page)) { + ret = ret < 0 ? ret : -EIO; + end_extent_writepage(page, ret, start, page_end); + } unlock_page(page); done_unlocked: /* drop our reference on any cached states */ free_extent_state(cached_state); - return 0; + return ret; } static int eb_wait(void *word) @@ -3690,6 +3695,7 @@ static int extent_write_cache_pages(struct extent_io_tree *tree, struct inode *inode = mapping->host; int ret = 0; int done = 0; + int err = 0; int nr_to_write_done = 0; struct pagevec pvec; int nr_pages; @@ -3776,8 +3782,8 @@ retry: unlock_page(page); ret = 0; } - if (ret) - done = 1; + if (!err && ret < 0) + err = ret; /* * the filesystem may choose to bump up nr_to_write. @@ -3789,7 +3795,7 @@ retry: pagevec_release(&pvec); cond_resched(); } - if (!scanned && !done) { + if (!scanned && !done && !err) { /* * We hit the last page and there is more work to be done: wrap * back to the start of the file @@ -3799,7 +3805,7 @@ retry: goto retry; } btrfs_add_delayed_iput(inode); - return ret; + return err; } static void flush_epd_write_bio(struct extent_page_data *epd) From 98806b446db8f15e83b59d065dae62901138e5cb Mon Sep 17 00:00:00 2001 From: Stefan Behrens Date: Fri, 9 May 2014 15:28:07 +0200 Subject: [PATCH 02/80] btrfs: check_int: propagate out-of-memory error upwards This issue was not causing any harm but IMO (and in the opinion of the static code checker) it is better to propagate this error status upwards. Signed-off-by: Stefan Behrens Reported-by: Dan Carpenter Signed-off-by: Chris Mason --- fs/btrfs/check-integrity.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/check-integrity.c b/fs/btrfs/check-integrity.c index 0e8388e72d8d..ce92ae30250f 100644 --- a/fs/btrfs/check-integrity.c +++ b/fs/btrfs/check-integrity.c @@ -1093,6 +1093,7 @@ leaf_item_out_of_bounce_error: next_stack = btrfsic_stack_frame_alloc(); if (NULL == next_stack) { + sf->error = -1; btrfsic_release_block_ctx( &sf-> next_block_ctx); @@ -1190,8 +1191,10 @@ continue_with_current_node_stack_frame: sf->next_block_ctx.datav[0]; next_stack = btrfsic_stack_frame_alloc(); - if (NULL == next_stack) + if (NULL == next_stack) { + sf->error = -1; goto one_stack_frame_backwards; + } next_stack->i = -1; next_stack->block = sf->next_block; From 60e1975acb48fc3d74a3422b21dde74c977ac3d5 Mon Sep 17 00:00:00 2001 From: Zach Brown Date: Fri, 9 May 2014 17:15:08 -0400 Subject: [PATCH 03/80] btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown Signed-off-by: Chris Mason --- fs/btrfs/lzo.c | 14 +++++++------- fs/btrfs/zlib.c | 26 +++++++++++++------------- 2 files changed, 20 insertions(+), 20 deletions(-) diff --git a/fs/btrfs/lzo.c b/fs/btrfs/lzo.c index b47f669aca75..dfad8514f0da 100644 --- a/fs/btrfs/lzo.c +++ b/fs/btrfs/lzo.c @@ -143,7 +143,7 @@ static int lzo_compress_pages(struct list_head *ws, if (ret != LZO_E_OK) { printk(KERN_DEBUG "BTRFS: deflate in loop returned %d\n", ret); - ret = -1; + ret = -EIO; goto out; } @@ -189,7 +189,7 @@ static int lzo_compress_pages(struct list_head *ws, kunmap(out_page); if (nr_pages == nr_dest_pages) { out_page = NULL; - ret = -1; + ret = -E2BIG; goto out; } @@ -208,7 +208,7 @@ static int lzo_compress_pages(struct list_head *ws, /* we're making it bigger, give up */ if (tot_in > 8192 && tot_in < tot_out) { - ret = -1; + ret = -E2BIG; goto out; } @@ -335,7 +335,7 @@ cont: break; if (page_in_index + 1 >= total_pages_in) { - ret = -1; + ret = -EIO; goto done; } @@ -358,7 +358,7 @@ cont: kunmap(pages_in[page_in_index - 1]); if (ret != LZO_E_OK) { printk(KERN_WARNING "BTRFS: decompress failed\n"); - ret = -1; + ret = -EIO; break; } @@ -402,12 +402,12 @@ static int lzo_decompress(struct list_head *ws, unsigned char *data_in, ret = lzo1x_decompress_safe(data_in, in_len, workspace->buf, &out_len); if (ret != LZO_E_OK) { printk(KERN_WARNING "BTRFS: decompress failed!\n"); - ret = -1; + ret = -EIO; goto out; } if (out_len < start_byte) { - ret = -1; + ret = -EIO; goto out; } diff --git a/fs/btrfs/zlib.c b/fs/btrfs/zlib.c index 8e57191950cb..4f196314c0c1 100644 --- a/fs/btrfs/zlib.c +++ b/fs/btrfs/zlib.c @@ -98,7 +98,7 @@ static int zlib_compress_pages(struct list_head *ws, if (Z_OK != zlib_deflateInit(&workspace->def_strm, 3)) { printk(KERN_WARNING "BTRFS: deflateInit failed\n"); - ret = -1; + ret = -EIO; goto out; } @@ -110,7 +110,7 @@ static int zlib_compress_pages(struct list_head *ws, out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM); if (out_page == NULL) { - ret = -1; + ret = -ENOMEM; goto out; } cpage_out = kmap(out_page); @@ -128,7 +128,7 @@ static int zlib_compress_pages(struct list_head *ws, printk(KERN_DEBUG "BTRFS: deflate in loop returned %d\n", ret); zlib_deflateEnd(&workspace->def_strm); - ret = -1; + ret = -EIO; goto out; } @@ -136,7 +136,7 @@ static int zlib_compress_pages(struct list_head *ws, if (workspace->def_strm.total_in > 8192 && workspace->def_strm.total_in < workspace->def_strm.total_out) { - ret = -1; + ret = -EIO; goto out; } /* we need another page for writing out. Test this @@ -147,12 +147,12 @@ static int zlib_compress_pages(struct list_head *ws, kunmap(out_page); if (nr_pages == nr_dest_pages) { out_page = NULL; - ret = -1; + ret = -E2BIG; goto out; } out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM); if (out_page == NULL) { - ret = -1; + ret = -ENOMEM; goto out; } cpage_out = kmap(out_page); @@ -188,12 +188,12 @@ static int zlib_compress_pages(struct list_head *ws, zlib_deflateEnd(&workspace->def_strm); if (ret != Z_STREAM_END) { - ret = -1; + ret = -EIO; goto out; } if (workspace->def_strm.total_out >= workspace->def_strm.total_in) { - ret = -1; + ret = -E2BIG; goto out; } @@ -253,7 +253,7 @@ static int zlib_decompress_biovec(struct list_head *ws, struct page **pages_in, if (Z_OK != zlib_inflateInit2(&workspace->inf_strm, wbits)) { printk(KERN_WARNING "BTRFS: inflateInit failed\n"); - return -1; + return -EIO; } while (workspace->inf_strm.total_in < srclen) { ret = zlib_inflate(&workspace->inf_strm, Z_NO_FLUSH); @@ -295,7 +295,7 @@ static int zlib_decompress_biovec(struct list_head *ws, struct page **pages_in, } } if (ret != Z_STREAM_END) - ret = -1; + ret = -EIO; else ret = 0; done: @@ -337,7 +337,7 @@ static int zlib_decompress(struct list_head *ws, unsigned char *data_in, if (Z_OK != zlib_inflateInit2(&workspace->inf_strm, wbits)) { printk(KERN_WARNING "BTRFS: inflateInit failed\n"); - return -1; + return -EIO; } while (bytes_left > 0) { @@ -354,7 +354,7 @@ static int zlib_decompress(struct list_head *ws, unsigned char *data_in, total_out = workspace->inf_strm.total_out; if (total_out == buf_start) { - ret = -1; + ret = -EIO; break; } @@ -382,7 +382,7 @@ next: } if (ret != Z_STREAM_END && bytes_left != 0) - ret = -1; + ret = -EIO; else ret = 0; From 774bcb35f0d9ccb9adfd3a391328cc25523659fc Mon Sep 17 00:00:00 2001 From: Zach Brown Date: Fri, 9 May 2014 17:15:09 -0400 Subject: [PATCH 04/80] btrfs: return ptr error from compression workspace The btrfs compression wrappers translated errors from workspace allocation to either -ENOMEM or -1. The compression type workspace allocators are already returning a ERR_PTR(-ENOMEM). Just return that and get rid of the magical -1. This helps a future patch return errors from the compression wrappers. Signed-off-by: Zach Brown Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/compression.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c index d43c544d3b68..92371c414228 100644 --- a/fs/btrfs/compression.c +++ b/fs/btrfs/compression.c @@ -887,7 +887,7 @@ int btrfs_compress_pages(int type, struct address_space *mapping, workspace = find_workspace(type); if (IS_ERR(workspace)) - return -1; + return PTR_ERR(workspace); ret = btrfs_compress_op[type-1]->compress_pages(workspace, mapping, start, len, pages, @@ -923,7 +923,7 @@ static int btrfs_decompress_biovec(int type, struct page **pages_in, workspace = find_workspace(type); if (IS_ERR(workspace)) - return -ENOMEM; + return PTR_ERR(workspace); ret = btrfs_compress_op[type-1]->decompress_biovec(workspace, pages_in, disk_start, @@ -945,7 +945,7 @@ int btrfs_decompress(int type, unsigned char *data_in, struct page *dest_page, workspace = find_workspace(type); if (IS_ERR(workspace)) - return -ENOMEM; + return PTR_ERR(workspace); ret = btrfs_compress_op[type-1]->decompress(workspace, data_in, dest_page, start_byte, From 166ae5a418756dacd2e8883c6f562c2d0ce2db2d Mon Sep 17 00:00:00 2001 From: Zach Brown Date: Fri, 9 May 2014 17:15:10 -0400 Subject: [PATCH 05/80] btrfs: fix inline compressed read err corruption uncompress_inline() is dropping the error from btrfs_decompress() after testing it and zeroing the page that was supposed to hold decompressed data. This can silently turn compressed inline data in to zeros if decompression fails due to corrupt compressed data or memory allocation failure. I verified this by manually forcing the error from btrfs_decompress() for a silly named copy of od: if (!strcmp(current->comm, "failod")) ret = -ENOMEM; # od -x /mnt/btrfs/dir/80 | head -1 0000000 3031 3038 310a 2d30 6f70 6e69 0a74 3031 # echo 3 > /proc/sys/vm/drop_caches # cp $(which od) /tmp/failod # /tmp/failod -x /mnt/btrfs/dir/80 | head -1 0000000 0000 0000 0000 0000 0000 0000 0000 0000 The fix is to pass the error to its caller. Which still has a BUG_ON(). So we fix that too. There seems to be no reason for the zeroing of the page on the error from btrfs_decompress() but not from the allocation error a few lines above. So the page zeroing is removed. Signed-off-by: Zach Brown Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 15 +++++---------- 1 file changed, 5 insertions(+), 10 deletions(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 5f805bc944fa..24dfa27a4f76 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6086,16 +6086,8 @@ static noinline int uncompress_inline(struct btrfs_path *path, max_size = min_t(unsigned long, PAGE_CACHE_SIZE, max_size); ret = btrfs_decompress(compress_type, tmp, page, extent_offset, inline_size, max_size); - if (ret) { - char *kaddr = kmap_atomic(page); - unsigned long copy_size = min_t(u64, - PAGE_CACHE_SIZE - pg_offset, - max_size - extent_offset); - memset(kaddr + pg_offset, 0, copy_size); - kunmap_atomic(kaddr); - } kfree(tmp); - return 0; + return ret; } /* @@ -6296,7 +6288,10 @@ next: ret = uncompress_inline(path, inode, page, pg_offset, extent_offset, item); - BUG_ON(ret); /* -ENOMEM */ + if (ret) { + err = ret; + goto out; + } } else { map = kmap(page); read_extent_buffer(leaf, map + pg_offset, ptr, From 7fb18a06644a33f0a2eba810d57d4b3b2c982f5b Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Fri, 25 Apr 2014 14:58:05 +0200 Subject: [PATCH 06/80] btrfs: Remove unnecessary check for NULL iput() already checks for the inode being NULL, thus it's unnecessary to check before calling. Signed-off-by: Tobias Klauser Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index 0be77993378e..bd850b599a99 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -717,8 +717,8 @@ static int scrub_fixup_readpage(u64 inum, u64 offset, u64 root, void *fixup_ctx) out: if (page) put_page(page); - if (inode) - iput(inode); + + iput(inode); if (ret < 0) return ret; From a1a50f60a6bf4f861eb94793420274bc1ccd409a Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Sat, 26 Apr 2014 01:35:31 +0100 Subject: [PATCH 07/80] Btrfs: read inode size after acquiring the mutex when punching a hole In a previous change, commit 12870f1c9b2de7d475d22e73fd7db1b418599725, I accidentally moved the roundup of inode->i_size to outside of the critical section delimited by the inode mutex, which is not atomic and not correct since the size can be changed by other task before we acquire the mutex. Therefore fix it. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/file.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index ae6af072b635..e4ff2d52ea23 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -2187,13 +2187,14 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) bool same_page = ((offset >> PAGE_CACHE_SHIFT) == ((offset + len - 1) >> PAGE_CACHE_SHIFT)); bool no_holes = btrfs_fs_incompat(root->fs_info, NO_HOLES); - u64 ino_size = round_up(inode->i_size, PAGE_CACHE_SIZE); + u64 ino_size; ret = btrfs_wait_ordered_range(inode, offset, len); if (ret) return ret; mutex_lock(&inode->i_mutex); + ino_size = round_up(inode->i_size, PAGE_CACHE_SIZE); /* * We needn't truncate any page which is beyond the end of the file * because we are sure there is no data there. From d2cbf2a260ab18c833f07fda66e30c4d4344162e Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Tue, 29 Apr 2014 13:07:58 +0800 Subject: [PATCH 08/80] Btrfs: do not increment on bio_index one by one 'bio_index' is just a index, it's really not necessary to do increment one by one. Signed-off-by: Liu Bo Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/file-item.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c index 127555b29f58..67751b776c7b 100644 --- a/fs/btrfs/file-item.c +++ b/fs/btrfs/file-item.c @@ -281,10 +281,10 @@ static int __btrfs_lookup_bio_sums(struct btrfs_root *root, found: csum += count * csum_size; nblocks -= count; + bio_index += count; while (count--) { disk_bytenr += bvec->bv_len; offset += bvec->bv_len; - bio_index++; bvec++; } } From fc19c5e73645f95d3eca12b4e91e7b56faf1e4a4 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Tue, 29 Apr 2014 13:18:40 +0100 Subject: [PATCH 09/80] Btrfs: fix leaf corruption caused by ENOSPC while hole punching While running a stress test with multiple threads writing to the same btrfs file system, I ended up with a situation where a leaf was corrupted in that it had 2 file extent item keys that had the same exact key. I was able to detect this quickly thanks to the following patch which triggers an assertion as soon as a leaf is marked dirty if there are duplicated keys or out of order keys: Btrfs: check if items are ordered when a leaf is marked dirty (https://patchwork.kernel.org/patch/3955431/) Basically while running the test, I got the following in dmesg: [28877.415877] WARNING: CPU: 2 PID: 10706 at fs/btrfs/file.c:553 btrfs_drop_extent_cache+0x435/0x440 [btrfs]() (...) [28877.415917] Call Trace: [28877.415922] [] dump_stack+0x4e/0x68 [28877.415926] [] warn_slowpath_common+0x8c/0xc0 [28877.415929] [] warn_slowpath_null+0x1a/0x20 [28877.415944] [] btrfs_drop_extent_cache+0x435/0x440 [btrfs] [28877.415949] [] ? kmem_cache_alloc+0xfe/0x1c0 [28877.415962] [] fill_holes+0x229/0x3e0 [btrfs] [28877.415972] [] ? block_rsv_add_bytes+0x55/0x80 [btrfs] [28877.415984] [] btrfs_fallocate+0xb6b/0xc20 [btrfs] (...) [29854.132560] BTRFS critical (device sdc): corrupt leaf, bad key order: block=955232256,root=1, slot=24 [29854.132565] BTRFS info (device sdc): leaf 955232256 total ptrs 40 free space 778 (...) [29854.132637] item 23 key (3486 108 667648) itemoff 2694 itemsize 53 [29854.132638] extent data disk bytenr 14574411776 nr 286720 [29854.132639] extent data offset 0 nr 286720 ram 286720 [29854.132640] item 24 key (3486 108 954368) itemoff 2641 itemsize 53 [29854.132641] extent data disk bytenr 0 nr 0 [29854.132643] extent data offset 0 nr 0 ram 0 [29854.132644] item 25 key (3486 108 954368) itemoff 2588 itemsize 53 [29854.132645] extent data disk bytenr 8699670528 nr 77824 [29854.132646] extent data offset 0 nr 77824 ram 77824 [29854.132647] item 26 key (3486 108 1146880) itemoff 2535 itemsize 53 [29854.132648] extent data disk bytenr 8699670528 nr 77824 [29854.132649] extent data offset 0 nr 77824 ram 77824 (...) [29854.132707] kernel BUG at fs/btrfs/ctree.h:3901! (...) [29854.132771] Call Trace: [29854.132779] [] setup_items_for_insert+0x2dc/0x400 [btrfs] [29854.132791] [] __btrfs_drop_extents+0xba7/0xdd0 [btrfs] [29854.132794] [] ? trace_hardirqs_on_caller+0x16/0x1d0 [29854.132797] [] ? trace_hardirqs_on+0xd/0x10 [29854.132800] [] ? kmem_cache_alloc+0xfe/0x1c0 [29854.132810] [] insert_reserved_file_extent.constprop.66+0xab/0x310 [btrfs] [29854.132820] [] __btrfs_prealloc_file_range+0x116/0x340 [btrfs] [29854.132830] [] btrfs_prealloc_file_range+0x23/0x30 [btrfs] (...) So this is caused by getting an -ENOSPC error while punching a file hole, more specifically, we get -ENOSPC error from __btrfs_drop_extents in the while loop of file.c:btrfs_punch_hole() when it's unable to modify the btree to delete one or more file extent items due to lack of enough free space. When this happens, in btrfs_punch_hole(), we attempt to reclaim free space by switching our transaction block reservation object to root->fs_info->trans_block_rsv, end our transaction and start a new transaction basically - and, we keep increasing our current offset (cur_offset) as long as it's smaller than the end of the target range (lockend) - this makes use leave the loop with cur_offset == drop_end which in turn makes us call fill_holes() for inserting a file extent item that represents a 0 bytes range hole (and this insertion succeeds, as in the meanwhile more space became available). This 0 bytes file hole extent item is a problem because any subsequent caller of __btrfs_drop_extents (regular file writes, or fallocate calls for e.g.), with a start file offset that is equal to the offset of the hole, will not remove this extent item due to the following conditional in the while loop of __btrfs_drop_extents: if (extent_end <= search_start) { path->slots[0]++; goto next_slot; } This later makes the call to setup_items_for_insert() (at the very end of __btrfs_drop_extents), insert a new file extent item with the same offset as the 0 bytes file hole extent item that follows it. Needless is to say that this causes chaos, either when reading the leaf from disk (btree_readpage_end_io_hook), where we perform leaf sanity checks or in subsequent operations that manipulate file extent items, as in the fallocate call as shown by the dmesg trace above. Without my other patch to perform the leaf sanity checks once a leaf is marked as dirty (if the integrity checker is enabled), it would have been much harder to debug this issue. This change might fix a few similar issues reported by users in the mailing list regarding assertion failures in btrfs_set_item_key_safe calls performed by __btrfs_drop_extents, such as the following report: http://comments.gmane.org/gmane.comp.file-systems.btrfs/32938 Asking fill_holes() to create a 0 bytes wide file hole item also produced the first warning in the trace above, as we passed a range to btrfs_drop_extent_cache that has an end smaller (by -1) than its start. On 3.14 kernels this issue manifests itself through leaf corruption, as we get duplicated file extent item keys in a leaf when calling setup_items_for_insert(), but on older kernels, setup_items_for_insert() isn't called by __btrfs_drop_extents(), instead we have callers of __btrfs_drop_extents(), namely the functions inode.c:insert_inline_extent() and inode.c:insert_reserved_file_extent(), calling btrfs_insert_empty_item() to insert the new file extent item, which would fail with error -EEXIST, instead of inserting a duplicated key - which is still a serious issue as it would make all similar file extent item replace operations keep failing if they target the same file range. Cc: stable@vger.kernel.org Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/file.c | 20 +++++++++++++++++++- 1 file changed, 19 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index e4ff2d52ea23..3029925e96d7 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -780,6 +780,18 @@ next_slot: extent_end = search_start; } + /* + * Don't skip extent items representing 0 byte lengths. They + * used to be created (bug) if while punching holes we hit + * -ENOSPC condition. So if we find one here, just ensure we + * delete it, otherwise we would insert a new file extent item + * with the same key (offset) as that 0 bytes length file + * extent item in the call to setup_items_for_insert() later + * in this function. + */ + if (extent_end == key.offset && extent_end >= search_start) + goto delete_extent_item; + if (extent_end <= search_start) { path->slots[0]++; goto next_slot; @@ -893,6 +905,7 @@ next_slot: * | ------ extent ------ | */ if (start <= key.offset && end >= extent_end) { +delete_extent_item: if (del_nr == 0) { del_slot = path->slots[0]; del_nr = 1; @@ -2348,7 +2361,12 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) } trans->block_rsv = &root->fs_info->trans_block_rsv; - if (cur_offset < ino_size) { + /* + * Don't insert file hole extent item if it's for a range beyond eof + * (because it's useless) or if it represents a 0 bytes range (when + * cur_offset == drop_end). + */ + if (cur_offset < ino_size && cur_offset < drop_end) { ret = fill_holes(trans, inode, path, cur_offset, drop_end); if (ret) { err = ret; From 7d824b6f9cf28917d8a05891ef423fb0e4e34c69 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Wed, 7 May 2014 17:37:51 +0200 Subject: [PATCH 10/80] btrfs: balance filter: add limit of processed chunks This started as debugging helper, to watch the effects of converting between raid levels on multiple devices, but could be useful standalone. In my case the usage filter was not finegrained enough and led to converting too many chunks at once. Another example use is in connection with drange+devid or vrange filters that allow to work with a specific chunk or even with a chunk on a given device. The limit filter applies last, the value of 0 means no limiting. CC: Ilya Dryomov CC: Hugo Mills Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 7 ++++++- fs/btrfs/volumes.c | 18 ++++++++++++++++++ fs/btrfs/volumes.h | 1 + include/uapi/linux/btrfs.h | 3 ++- 4 files changed, 27 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index ba6b88528dc7..e6f899dc5e47 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -840,7 +840,10 @@ struct btrfs_disk_balance_args { /* BTRFS_BALANCE_ARGS_* */ __le64 flags; - __le64 unused[8]; + /* BTRFS_BALANCE_ARGS_LIMIT value */ + __le64 limit; + + __le64 unused[7]; } __attribute__ ((__packed__)); /* @@ -2897,6 +2900,7 @@ btrfs_disk_balance_args_to_cpu(struct btrfs_balance_args *cpu, cpu->vend = le64_to_cpu(disk->vend); cpu->target = le64_to_cpu(disk->target); cpu->flags = le64_to_cpu(disk->flags); + cpu->limit = le64_to_cpu(disk->limit); } static inline void @@ -2914,6 +2918,7 @@ btrfs_cpu_balance_args_to_disk(struct btrfs_disk_balance_args *disk, disk->vend = cpu_to_le64(cpu->vend); disk->target = cpu_to_le64(cpu->target); disk->flags = cpu_to_le64(cpu->flags); + disk->limit = cpu_to_le64(cpu->limit); } /* struct btrfs_super_block */ diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 49d7fab73360..3b761a456acd 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -2922,6 +2922,16 @@ static int should_balance_chunk(struct btrfs_root *root, return 0; } + /* + * limited by count, must be the last filter + */ + if ((bargs->flags & BTRFS_BALANCE_ARGS_LIMIT)) { + if (bargs->limit == 0) + return 0; + else + bargs->limit--; + } + return 1; } @@ -2944,6 +2954,9 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info) int ret; int enospc_errors = 0; bool counting = true; + u64 limit_data = bctl->data.limit; + u64 limit_meta = bctl->meta.limit; + u64 limit_sys = bctl->sys.limit; /* step one make some room on all the devices */ devices = &fs_info->fs_devices->devices; @@ -2982,6 +2995,11 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info) memset(&bctl->stat, 0, sizeof(bctl->stat)); spin_unlock(&fs_info->balance_lock); again: + if (!counting) { + bctl->data.limit = limit_data; + bctl->meta.limit = limit_meta; + bctl->sys.limit = limit_sys; + } key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID; key.offset = (u64)-1; key.type = BTRFS_CHUNK_ITEM_KEY; diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h index 80754f9dd3df..1a15bbeb65e2 100644 --- a/fs/btrfs/volumes.h +++ b/fs/btrfs/volumes.h @@ -255,6 +255,7 @@ struct map_lookup { #define BTRFS_BALANCE_ARGS_DEVID (1ULL << 2) #define BTRFS_BALANCE_ARGS_DRANGE (1ULL << 3) #define BTRFS_BALANCE_ARGS_VRANGE (1ULL << 4) +#define BTRFS_BALANCE_ARGS_LIMIT (1ULL << 5) /* * Profile changing flags. When SOFT is set we won't relocate chunk if diff --git a/include/uapi/linux/btrfs.h b/include/uapi/linux/btrfs.h index b4d69092fbdb..901a3c563f60 100644 --- a/include/uapi/linux/btrfs.h +++ b/include/uapi/linux/btrfs.h @@ -211,7 +211,8 @@ struct btrfs_balance_args { __u64 flags; - __u64 unused[8]; + __u64 limit; /* limit number of processed chunks */ + __u64 unused[7]; } __attribute__ ((__packed__)); /* report balance progress to userspace */ From 80a773fbfc2d6b5b2478377e8ac271d495f55e73 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Wed, 7 May 2014 18:17:06 +0200 Subject: [PATCH 11/80] btrfs: retrieve more info from FS_INFO ioctl Provide the basic information about filesystem through the ioctl: * b-tree node size (same as leaf size) * sector size * expected alignment of CLONE_RANGE and EXTENT_SAME ioctl arguments Backward compatibility: if the values are 0, kernel does not provide this information, the applications should ignore them. Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 4 ++++ include/uapi/linux/btrfs.h | 6 +++++- 2 files changed, 9 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 2f6d7b13b5bd..c6c8e3560e73 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2574,6 +2574,10 @@ static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg) } mutex_unlock(&fs_devices->device_list_mutex); + fi_args->nodesize = root->fs_info->super_copy->nodesize; + fi_args->sectorsize = root->fs_info->super_copy->sectorsize; + fi_args->clone_alignment = root->fs_info->super_copy->sectorsize; + if (copy_to_user(arg, fi_args, sizeof(*fi_args))) ret = -EFAULT; diff --git a/include/uapi/linux/btrfs.h b/include/uapi/linux/btrfs.h index 901a3c563f60..7554fd381a56 100644 --- a/include/uapi/linux/btrfs.h +++ b/include/uapi/linux/btrfs.h @@ -181,7 +181,11 @@ struct btrfs_ioctl_fs_info_args { __u64 max_id; /* out */ __u64 num_devices; /* out */ __u8 fsid[BTRFS_FSID_SIZE]; /* out */ - __u64 reserved[124]; /* pad to 1k */ + __u32 nodesize; /* out */ + __u32 sectorsize; /* out */ + __u32 clone_alignment; /* out */ + __u32 reserved32; + __u64 reserved[122]; /* pad to 1k */ }; struct btrfs_ioctl_feature_flags { From df93589a17378277c0ac394858dd8a2a41e5fb5a Mon Sep 17 00:00:00 2001 From: David Sterba Date: Wed, 7 May 2014 18:17:16 +0200 Subject: [PATCH 12/80] btrfs: export more from FS_INFO to sysfs Similar to the FS_INFO updates, export the basic filesystem info through sysfs: node size, sector size and clone alignment. Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/sysfs.c | 40 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 40 insertions(+) diff --git a/fs/btrfs/sysfs.c b/fs/btrfs/sysfs.c index c5eb2143dc66..ba2a645dee07 100644 --- a/fs/btrfs/sysfs.c +++ b/fs/btrfs/sysfs.c @@ -396,8 +396,48 @@ static ssize_t btrfs_label_store(struct kobject *kobj, } BTRFS_ATTR_RW(label, 0644, btrfs_label_show, btrfs_label_store); +static ssize_t btrfs_no_store(struct kobject *kobj, + struct kobj_attribute *a, + const char *buf, size_t len) +{ + return -EPERM; +} + +static ssize_t btrfs_nodesize_show(struct kobject *kobj, + struct kobj_attribute *a, char *buf) +{ + struct btrfs_fs_info *fs_info = to_fs_info(kobj); + + return snprintf(buf, PAGE_SIZE, "%u\n", fs_info->super_copy->nodesize); +} + +BTRFS_ATTR_RW(nodesize, 0444, btrfs_nodesize_show, btrfs_no_store); + +static ssize_t btrfs_sectorsize_show(struct kobject *kobj, + struct kobj_attribute *a, char *buf) +{ + struct btrfs_fs_info *fs_info = to_fs_info(kobj); + + return snprintf(buf, PAGE_SIZE, "%u\n", fs_info->super_copy->sectorsize); +} + +BTRFS_ATTR_RW(sectorsize, 0444, btrfs_sectorsize_show, btrfs_no_store); + +static ssize_t btrfs_clone_alignment_show(struct kobject *kobj, + struct kobj_attribute *a, char *buf) +{ + struct btrfs_fs_info *fs_info = to_fs_info(kobj); + + return snprintf(buf, PAGE_SIZE, "%u\n", fs_info->super_copy->sectorsize); +} + +BTRFS_ATTR_RW(clone_alignment, 0444, btrfs_clone_alignment_show, btrfs_no_store); + static struct attribute *btrfs_attrs[] = { BTRFS_ATTR_PTR(label), + BTRFS_ATTR_PTR(nodesize), + BTRFS_ATTR_PTR(sectorsize), + BTRFS_ATTR_PTR(clone_alignment), NULL, }; From 7d6213c5a764d0591afeb9faa3f7ee18e8e74c25 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 24 Apr 2014 15:09:10 +0200 Subject: [PATCH 13/80] btrfs: make DEV_INFO ioctl available to anyone This ioctl provides basic info about the devices that can be obtained in other ways (eg. sysfs), there's no reason to restrict it to CAP_SYSADMIN. Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index c6c8e3560e73..92a5184ac696 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2593,9 +2593,6 @@ static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg) int ret = 0; char *s_uuid = NULL; - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - di_args = memdup_user(arg, sizeof(*di_args)); if (IS_ERR(di_args)) return PTR_ERR(di_args); From e4ef90ff6198ac58d4c7dbbeae8cc7b24867d1ca Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 24 Apr 2014 15:09:18 +0200 Subject: [PATCH 14/80] btrfs: make FS_INFO ioctl available to anyone This ioctl provides basic info about the filesystem that can be obtained in other ways (eg. sysfs), there's no reason to restrict it to CAP_SYSADMIN. Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 92a5184ac696..2db12fc3f748 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2557,9 +2557,6 @@ static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg) struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices; int ret = 0; - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL); if (!fi_args) return -ENOMEM; From ef3b9af50bfa6a1f02cd7b3f5124b712b1ba3e3c Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Sun, 27 Apr 2014 20:40:45 +0100 Subject: [PATCH 15/80] Btrfs: implement inode_operations callback tmpfile This implements the tmpfile callback of struct inode_operations, introduced in the linux kernel 3.11, and implemented already by some filesystems. This callback is invoked by the VFS when the flag O_TMPFILE is passed to the open system call. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason Reviewed-by: David Sterba --- fs/btrfs/inode.c | 118 +++++++++++++++++++++++++++++++++++++++-------- 1 file changed, 98 insertions(+), 20 deletions(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 24dfa27a4f76..5c0def8dd761 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -5553,6 +5553,7 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, struct btrfs_inode_ref *ref; struct btrfs_key key[2]; u32 sizes[2]; + int nitems = name ? 2 : 1; unsigned long ptr; int ret; @@ -5572,7 +5573,7 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, */ inode->i_ino = objectid; - if (dir) { + if (dir && name) { trace_btrfs_inode_request(dir); ret = btrfs_set_inode_index(dir, index); @@ -5581,6 +5582,8 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, iput(inode); return ERR_PTR(ret); } + } else if (dir) { + *index = 0; } /* * index_cnt is ignored for everything but a dir, @@ -5605,21 +5608,24 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY); key[0].offset = 0; - /* - * Start new inodes with an inode_ref. This is slightly more - * efficient for small numbers of hard links since they will - * be packed into one item. Extended refs will kick in if we - * add more hard links than can fit in the ref item. - */ - key[1].objectid = objectid; - btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY); - key[1].offset = ref_objectid; - sizes[0] = sizeof(struct btrfs_inode_item); - sizes[1] = name_len + sizeof(*ref); + + if (name) { + /* + * Start new inodes with an inode_ref. This is slightly more + * efficient for small numbers of hard links since they will + * be packed into one item. Extended refs will kick in if we + * add more hard links than can fit in the ref item. + */ + key[1].objectid = objectid; + btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY); + key[1].offset = ref_objectid; + + sizes[1] = name_len + sizeof(*ref); + } path->leave_spinning = 1; - ret = btrfs_insert_empty_items(trans, root, path, key, sizes, 2); + ret = btrfs_insert_empty_items(trans, root, path, key, sizes, nitems); if (ret != 0) goto fail; @@ -5632,12 +5638,14 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, sizeof(*inode_item)); fill_inode_item(trans, path->nodes[0], inode_item, inode); - ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1, - struct btrfs_inode_ref); - btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len); - btrfs_set_inode_ref_index(path->nodes[0], ref, *index); - ptr = (unsigned long)(ref + 1); - write_extent_buffer(path->nodes[0], name, ptr, name_len); + if (name) { + ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1, + struct btrfs_inode_ref); + btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len); + btrfs_set_inode_ref_index(path->nodes[0], ref, *index); + ptr = (unsigned long)(ref + 1); + write_extent_buffer(path->nodes[0], name, ptr, name_len); + } btrfs_mark_buffer_dirty(path->nodes[0]); btrfs_free_path(path); @@ -5673,7 +5681,7 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans, return inode; fail: - if (dir) + if (dir && name) BTRFS_I(dir)->index_cnt--; btrfs_free_path(path); iput(inode); @@ -5958,6 +5966,15 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir, err = btrfs_update_inode(trans, root, inode); if (err) goto fail; + if (inode->i_nlink == 1) { + /* + * If new hard link count is 1, it's a file created + * with open(2) O_TMPFILE flag. + */ + err = btrfs_orphan_del(trans, inode); + if (err) + goto fail; + } d_instantiate(dentry, inode); btrfs_log_new_name(trans, inode, NULL, parent); } @@ -8884,6 +8901,66 @@ static int btrfs_permission(struct inode *inode, int mask) return generic_permission(inode, mask); } +static int btrfs_tmpfile(struct inode *dir, struct dentry *dentry, umode_t mode) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(dir)->root; + struct inode *inode = NULL; + u64 objectid; + u64 index; + int ret = 0; + + /* + * 5 units required for adding orphan entry + */ + trans = btrfs_start_transaction(root, 5); + if (IS_ERR(trans)) + return PTR_ERR(trans); + + ret = btrfs_find_free_ino(root, &objectid); + if (ret) + goto out; + + inode = btrfs_new_inode(trans, root, dir, NULL, 0, + btrfs_ino(dir), objectid, mode, &index); + if (IS_ERR(inode)) { + ret = PTR_ERR(inode); + inode = NULL; + goto out; + } + + ret = btrfs_init_inode_security(trans, inode, dir, NULL); + if (ret) + goto out; + + ret = btrfs_update_inode(trans, root, inode); + if (ret) + goto out; + + inode->i_fop = &btrfs_file_operations; + inode->i_op = &btrfs_file_inode_operations; + + inode->i_mapping->a_ops = &btrfs_aops; + inode->i_mapping->backing_dev_info = &root->fs_info->bdi; + BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops; + + ret = btrfs_orphan_add(trans, inode); + if (ret) + goto out; + + d_tmpfile(dentry, inode); + mark_inode_dirty(inode); + +out: + btrfs_end_transaction(trans, root); + if (ret) + iput(inode); + btrfs_balance_delayed_items(root); + btrfs_btree_balance_dirty(root); + + return ret; +} + static const struct inode_operations btrfs_dir_inode_operations = { .getattr = btrfs_getattr, .lookup = btrfs_lookup, @@ -8904,6 +8981,7 @@ static const struct inode_operations btrfs_dir_inode_operations = { .get_acl = btrfs_get_acl, .set_acl = btrfs_set_acl, .update_time = btrfs_update_time, + .tmpfile = btrfs_tmpfile, }; static const struct inode_operations btrfs_dir_ro_inode_operations = { .lookup = btrfs_lookup, From 944a4515b27302ac42d5964b7400c72f4d692782 Mon Sep 17 00:00:00 2001 From: Daeseok Youn Date: Mon, 14 Apr 2014 15:37:02 +0900 Subject: [PATCH 16/80] btrfs: remove redundant null check in btrfs_dentry_release() It doesn't need to check NULL for kfree() Signed-off-by: Daeseok Youn Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 5c0def8dd761..36c13e391ae3 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -5168,8 +5168,7 @@ static int btrfs_dentry_delete(const struct dentry *dentry) static void btrfs_dentry_release(struct dentry *dentry) { - if (dentry->d_fsdata) - kfree(dentry->d_fsdata); + kfree(dentry->d_fsdata); } static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry, From 521e0546c970c3d845076f243828fa7abd71edfc Mon Sep 17 00:00:00 2001 From: David Sterba Date: Tue, 15 Apr 2014 16:41:44 +0200 Subject: [PATCH 17/80] btrfs: protect snapshots from deleting during send The patch "Btrfs: fix protection between send and root deletion" (18f687d538449373c37c) does not actually prevent to delete the snapshot and just takes care during background cleaning, but this seems rather user unfriendly, this patch implements the idea presented in http://www.spinics.net/lists/linux-btrfs/msg30813.html - add an internal root_item flag to denote a dead root - check if the send_in_progress is set and refuse to delete, otherwise set the flag and proceed - check the flag in send similar to the btrfs_root_readonly checks, for all involved roots The root lookup in send via btrfs_read_fs_root_no_name will check if the root is really dead or not. If it is, ENOENT, aborted send. If it's alive, it's protected by send_in_progress, send can continue. CC: Miao Xie CC: Wang Shilong Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 11 +++++++++++ fs/btrfs/ioctl.c | 30 ++++++++++++++++++++++++++++++ fs/btrfs/send.c | 14 ++++++++++++-- 3 files changed, 53 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index e6f899dc5e47..f4a439e1a43f 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -756,6 +756,12 @@ struct btrfs_dir_item { #define BTRFS_ROOT_SUBVOL_RDONLY (1ULL << 0) +/* + * Internal in-memory flag that a subvolume has been marked for deletion but + * still visible as a directory + */ +#define BTRFS_ROOT_SUBVOL_DEAD (1ULL << 48) + struct btrfs_root_item { struct btrfs_inode_item inode; __le64 generation; @@ -2791,6 +2797,11 @@ static inline bool btrfs_root_readonly(struct btrfs_root *root) return (root->root_item.flags & cpu_to_le64(BTRFS_ROOT_SUBVOL_RDONLY)) != 0; } +static inline bool btrfs_root_dead(struct btrfs_root *root) +{ + return (root->root_item.flags & cpu_to_le64(BTRFS_ROOT_SUBVOL_DEAD)) != 0; +} + /* struct btrfs_root_backup */ BTRFS_SETGET_STACK_FUNCS(backup_tree_root, struct btrfs_root_backup, tree_root, 64); diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 2db12fc3f748..2db3fe12f50d 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2219,6 +2219,7 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file, struct btrfs_ioctl_vol_args *vol_args; struct btrfs_trans_handle *trans; struct btrfs_block_rsv block_rsv; + u64 root_flags; u64 qgroup_reserved; int namelen; int ret; @@ -2240,6 +2241,7 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file, if (err) goto out; + err = mutex_lock_killable_nested(&dir->i_mutex, I_MUTEX_PARENT); if (err == -EINTR) goto out_drop_write; @@ -2301,6 +2303,27 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file, } mutex_lock(&inode->i_mutex); + + /* + * Don't allow to delete a subvolume with send in progress. This is + * inside the i_mutex so the error handling that has to drop the bit + * again is not run concurrently. + */ + spin_lock(&dest->root_item_lock); + root_flags = btrfs_root_flags(&root->root_item); + if (root->send_in_progress == 0) { + btrfs_set_root_flags(&root->root_item, + root_flags | BTRFS_ROOT_SUBVOL_DEAD); + spin_unlock(&dest->root_item_lock); + } else { + spin_unlock(&dest->root_item_lock); + btrfs_warn(root->fs_info, + "Attempt to delete subvolume %llu during send", + root->root_key.objectid); + err = -EPERM; + goto out_dput; + } + err = d_invalidate(dentry); if (err) goto out_unlock; @@ -2389,6 +2412,13 @@ out_release: out_up_write: up_write(&root->fs_info->subvol_sem); out_unlock: + if (err) { + spin_lock(&dest->root_item_lock); + root_flags = btrfs_root_flags(&root->root_item); + btrfs_set_root_flags(&root->root_item, + root_flags & ~BTRFS_ROOT_SUBVOL_DEAD); + spin_unlock(&dest->root_item_lock); + } mutex_unlock(&inode->i_mutex); if (!err) { shrink_dcache_sb(root->fs_info->sb); diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index 484aacac2c89..c76400dda4df 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -5518,7 +5518,7 @@ long btrfs_ioctl_send(struct file *mnt_file, void __user *arg_) /* * The subvolume must remain read-only during send, protect against - * making it RW. + * making it RW. This also protects against deletion. */ spin_lock(&send_root->root_item_lock); send_root->send_in_progress++; @@ -5578,6 +5578,15 @@ long btrfs_ioctl_send(struct file *mnt_file, void __user *arg_) } sctx->send_root = send_root; + /* + * Unlikely but possible, if the subvolume is marked for deletion but + * is slow to remove the directory entry, send can still be started + */ + if (btrfs_root_dead(sctx->send_root)) { + ret = -EPERM; + goto out; + } + sctx->clone_roots_cnt = arg->clone_sources_count; sctx->send_max_size = BTRFS_SEND_BUF_SIZE; @@ -5667,7 +5676,8 @@ long btrfs_ioctl_send(struct file *mnt_file, void __user *arg_) spin_lock(&sctx->parent_root->root_item_lock); sctx->parent_root->send_in_progress++; - if (!btrfs_root_readonly(sctx->parent_root)) { + if (!btrfs_root_readonly(sctx->parent_root) || + btrfs_root_dead(sctx->parent_root)) { spin_unlock(&sctx->parent_root->root_item_lock); srcu_read_unlock(&fs_info->subvol_srcu, index); ret = -EPERM; From 61155aa04ebcba6a33f2a281c46f351ca150d816 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Tue, 15 Apr 2014 16:42:03 +0200 Subject: [PATCH 18/80] btrfs: assert that send is not in progres before root deletion CC: Miao Xie CC: Wang Shilong Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 1 + fs/btrfs/transaction.c | 13 ------------- 2 files changed, 1 insertion(+), 13 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 2db3fe12f50d..ebac486f59af 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2424,6 +2424,7 @@ out_unlock: shrink_dcache_sb(root->fs_info->sb); btrfs_invalidate_inodes(dest); d_delete(dentry); + ASSERT(dest->send_in_progress == 0); /* the last ref */ if (dest->cache_inode) { diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 7579f6d0b854..c30815e79235 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -1984,19 +1984,6 @@ int btrfs_clean_one_deleted_snapshot(struct btrfs_root *root) } root = list_first_entry(&fs_info->dead_roots, struct btrfs_root, root_list); - /* - * Make sure root is not involved in send, - * if we fail with first root, we return - * directly rather than continue. - */ - spin_lock(&root->root_item_lock); - if (root->send_in_progress) { - spin_unlock(&fs_info->trans_lock); - spin_unlock(&root->root_item_lock); - return 0; - } - spin_unlock(&root->root_item_lock); - list_del_init(&root->root_list); spin_unlock(&fs_info->trans_lock); From 5a1972bd9fd4b2fb1bac8b7a0b636d633d8717e3 Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Wed, 16 Apr 2014 17:02:32 +0800 Subject: [PATCH 19/80] btrfs: Add ctime/mtime update for btrfs device add/remove. Btrfs will send uevent to udev inform the device change, but ctime/mtime for the block device inode is not udpated, which cause libblkid used by btrfs-progs unable to detect device change and use old cache, causing 'btrfs dev scan; btrfs dev rmove; btrfs dev scan' give an error message. Reported-by: Tsutomu Itoh Cc: Karel Zak Signed-off-by: Qu Wenruo Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 26 ++++++++++++++++++++++++-- 1 file changed, 24 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 3b761a456acd..ad8e342c4b78 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1452,6 +1452,22 @@ out: return ret; } +/* + * Function to update ctime/mtime for a given device path. + * Mainly used for ctime/mtime based probe like libblkid. + */ +static void update_dev_time(char *path_name) +{ + struct file *filp; + + filp = filp_open(path_name, O_RDWR, 0); + if (!filp) + return; + file_update_time(filp); + filp_close(filp, NULL); + return; +} + static int btrfs_rm_dev_item(struct btrfs_root *root, struct btrfs_device *device) { @@ -1704,10 +1720,14 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) ret = 0; - /* Notify udev that device has changed */ - if (bdev) + if (bdev) { + /* Notify udev that device has changed */ btrfs_kobject_uevent(bdev, KOBJ_CHANGE); + /* Update ctime/mtime for device path for libblkid */ + update_dev_time(device_path); + } + error_brelse: brelse(bh); if (bdev) @@ -2146,6 +2166,8 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path) ret = btrfs_commit_transaction(trans, root); } + /* Update ctime/mtime for libblkid */ + update_dev_time(device_path); return ret; error_trans: From 32d6b47fe6fc1714d5f1bba1b9f38e0ab0ad58a8 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 24 Apr 2014 13:31:55 +0800 Subject: [PATCH 20/80] Btrfs: output warning instead of error when loading free space cache failed If we fail to load a free space cache, we can rebuild it from the extent tree, so it is not a serious error, we should not output a error message that would make the users uncomfortable. This patch uses warning message instead of it. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/free-space-cache.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 73f3de7a083c..a6bd654dcd47 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -831,7 +831,7 @@ int load_free_space_cache(struct btrfs_fs_info *fs_info, if (!matched) { __btrfs_remove_free_space_cache(ctl); - btrfs_err(fs_info, "block group %llu has wrong amount of free space", + btrfs_warn(fs_info, "block group %llu has wrong amount of free space", block_group->key.objectid); ret = -1; } @@ -843,7 +843,7 @@ out: spin_unlock(&block_group->lock); ret = 0; - btrfs_err(fs_info, "failed to load free space cache for block group %llu", + btrfs_warn(fs_info, "failed to load free space cache for block group %llu, rebuild it now", block_group->key.objectid); } From 21c7e75654b77b53a4436bf28496aac11536b6ba Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 13 May 2014 17:29:04 -0700 Subject: [PATCH 21/80] Btrfs: reclaim the reserved metadata space at background Before applying this patch, the task had to reclaim the metadata space by itself if the metadata space was not enough. And When the task started the space reclamation, all the other tasks which wanted to reserve the metadata space were blocked. At some cases, they would be blocked for a long time, it made the performance fluctuate wildly. So we introduce the background metadata space reclamation, when the space is about to be exhausted, we insert a reclaim work into the workqueue, the worker of the workqueue helps us to reclaim the reserved space at the background. By this way, the tasks needn't reclaim the space by themselves at most cases, and even if the tasks have to reclaim the space or are blocked for the space reclamation, they will get enough space more quickly. Here is my test result(Tested by compilebench): Memory: 2GB CPU: 2Cores * 1CPU Partition: 40GB(SSD) Test command: # compilebench -D -m Without this patch: intial create total runs 30 avg 54.36 MB/s (user 0.52s sys 2.44s) compile total runs 30 avg 123.72 MB/s (user 0.13s sys 1.17s) read compiled tree total runs 3 avg 81.15 MB/s (user 0.74s sys 4.89s) delete compiled tree total runs 30 avg 5.32 seconds (user 0.35s sys 4.37s) With this patch: intial create total runs 30 avg 59.80 MB/s (user 0.52s sys 2.53s) compile total runs 30 avg 151.44 MB/s (user 0.13s sys 1.11s) read compiled tree total runs 3 avg 83.25 MB/s (user 0.76s sys 4.91s) delete compiled tree total runs 30 avg 5.29 seconds (user 0.34s sys 4.34s) Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 6 +++ fs/btrfs/disk-io.c | 3 ++ fs/btrfs/extent-tree.c | 105 ++++++++++++++++++++++++++++++++++++++++- fs/btrfs/super.c | 1 + 4 files changed, 114 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index f4a439e1a43f..0a805b8d61cc 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -33,6 +33,7 @@ #include #include #include +#include #include "extent_io.h" #include "extent_map.h" #include "async-thread.h" @@ -1322,6 +1323,8 @@ struct btrfs_stripe_hash_table { #define BTRFS_STRIPE_HASH_TABLE_BITS 11 +void btrfs_init_async_reclaim_work(struct work_struct *work); + /* fs_info */ struct reloc_control; struct btrfs_device; @@ -1697,6 +1700,9 @@ struct btrfs_fs_info { struct semaphore uuid_tree_rescan_sem; unsigned int update_uuid_tree_gen:1; + + /* Used to reclaim the metadata space in the background. */ + struct work_struct async_reclaim_work; }; struct btrfs_subvolume_writers { diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 983314932af3..4b1f16dd9ce3 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2291,6 +2291,7 @@ int open_ctree(struct super_block *sb, atomic_set(&fs_info->balance_cancel_req, 0); fs_info->balance_ctl = NULL; init_waitqueue_head(&fs_info->balance_wait_q); + btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work); sb->s_blocksize = 4096; sb->s_blocksize_bits = blksize_bits(4096); @@ -3603,6 +3604,8 @@ int close_ctree(struct btrfs_root *root) /* clear out the rbtree of defraggable inodes */ btrfs_cleanup_defrag_inodes(fs_info); + cancel_work_sync(&fs_info->async_reclaim_work); + if (!(fs_info->sb->s_flags & MS_RDONLY)) { ret = btrfs_commit_super(root); if (ret) diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 5590af92094b..15467e3f5876 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4204,6 +4204,104 @@ static int flush_space(struct btrfs_root *root, return ret; } + +static inline u64 +btrfs_calc_reclaim_metadata_size(struct btrfs_root *root, + struct btrfs_space_info *space_info) +{ + u64 used; + u64 expected; + u64 to_reclaim; + + to_reclaim = min_t(u64, num_online_cpus() * 1024 * 1024, + 16 * 1024 * 1024); + spin_lock(&space_info->lock); + if (can_overcommit(root, space_info, to_reclaim, + BTRFS_RESERVE_FLUSH_ALL)) { + to_reclaim = 0; + goto out; + } + + used = space_info->bytes_used + space_info->bytes_reserved + + space_info->bytes_pinned + space_info->bytes_readonly + + space_info->bytes_may_use; + if (can_overcommit(root, space_info, 1024 * 1024, + BTRFS_RESERVE_FLUSH_ALL)) + expected = div_factor_fine(space_info->total_bytes, 95); + else + expected = div_factor_fine(space_info->total_bytes, 90); + + if (used > expected) + to_reclaim = used - expected; + else + to_reclaim = 0; + to_reclaim = min(to_reclaim, space_info->bytes_may_use + + space_info->bytes_reserved); +out: + spin_unlock(&space_info->lock); + + return to_reclaim; +} + +static inline int need_do_async_reclaim(struct btrfs_space_info *space_info, + struct btrfs_fs_info *fs_info, u64 used) +{ + return (used >= div_factor_fine(space_info->total_bytes, 98) && + !btrfs_fs_closing(fs_info) && + !test_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state)); +} + +static int btrfs_need_do_async_reclaim(struct btrfs_space_info *space_info, + struct btrfs_fs_info *fs_info) +{ + u64 used; + + spin_lock(&space_info->lock); + used = space_info->bytes_used + space_info->bytes_reserved + + space_info->bytes_pinned + space_info->bytes_readonly + + space_info->bytes_may_use; + if (need_do_async_reclaim(space_info, fs_info, used)) { + spin_unlock(&space_info->lock); + return 1; + } + spin_unlock(&space_info->lock); + + return 0; +} + +static void btrfs_async_reclaim_metadata_space(struct work_struct *work) +{ + struct btrfs_fs_info *fs_info; + struct btrfs_space_info *space_info; + u64 to_reclaim; + int flush_state; + + fs_info = container_of(work, struct btrfs_fs_info, async_reclaim_work); + space_info = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA); + + to_reclaim = btrfs_calc_reclaim_metadata_size(fs_info->fs_root, + space_info); + if (!to_reclaim) + return; + + flush_state = FLUSH_DELAYED_ITEMS_NR; + do { + flush_space(fs_info->fs_root, space_info, to_reclaim, + to_reclaim, flush_state); + flush_state++; + if (!btrfs_need_do_async_reclaim(space_info, fs_info)) + return; + } while (flush_state <= COMMIT_TRANS); + + if (btrfs_need_do_async_reclaim(space_info, fs_info)) + queue_work(system_unbound_wq, work); +} + +void btrfs_init_async_reclaim_work(struct work_struct *work) +{ + INIT_WORK(work, btrfs_async_reclaim_metadata_space); +} + /** * reserve_metadata_bytes - try to reserve bytes from the block_rsv's space * @root - the root we're allocating for @@ -4311,8 +4409,13 @@ again: if (ret && flush != BTRFS_RESERVE_NO_FLUSH) { flushing = true; space_info->flush = 1; + } else if (!ret && space_info->flags & BTRFS_BLOCK_GROUP_METADATA) { + used += orig_bytes; + if (need_do_async_reclaim(space_info, root->fs_info, used) && + !work_busy(&root->fs_info->async_reclaim_work)) + queue_work(system_unbound_wq, + &root->fs_info->async_reclaim_work); } - spin_unlock(&space_info->lock); if (!ret || flush == BTRFS_RESERVE_NO_FLUSH) diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 9601d25a4607..8f7c03db1f8d 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -1413,6 +1413,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data) * this also happens on 'umount -rf' or on shutdown, when * the filesystem is busy. */ + cancel_work_sync(&fs_info->async_reclaim_work); /* wait for the uuid_scan task to finish */ down(&fs_info->uuid_tree_rescan_sem); From 65d33fd7a6e4aaf9cd6577555d0b3c18fbefe415 Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Tue, 22 Apr 2014 17:13:51 +0800 Subject: [PATCH 22/80] btrfs: Add check to avoid cleanup roots already in fs_info->dead_roots. Current btrfs_orphan_cleanup will also cleanup roots which is already in fs_info->dead_roots without protection. This will have conditional race with fs_info->cleaner_kthread. This patch will use refs in root->root_item to detect roots in dead_roots and avoid conflicts. Signed-off-by: Qu Wenruo Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 39 +++++++++++++++++++++++++++++++-------- 1 file changed, 31 insertions(+), 8 deletions(-) diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 4b1f16dd9ce3..85f514483bd1 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -3534,28 +3534,51 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info) { u64 root_objectid = 0; struct btrfs_root *gang[8]; - int i; - int ret; + int i = 0; + int err = 0; + unsigned int ret = 0; + int index; while (1) { + index = srcu_read_lock(&fs_info->subvol_srcu); ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix, (void **)gang, root_objectid, ARRAY_SIZE(gang)); - if (!ret) + if (!ret) { + srcu_read_unlock(&fs_info->subvol_srcu, index); break; - + } root_objectid = gang[ret - 1]->root_key.objectid + 1; - for (i = 0; i < ret; i++) { - int err; + for (i = 0; i < ret; i++) { + /* Avoid to grab roots in dead_roots */ + if (btrfs_root_refs(&gang[i]->root_item) == 0) { + gang[i] = NULL; + continue; + } + /* grab all the search result for later use */ + gang[i] = btrfs_grab_fs_root(gang[i]); + } + srcu_read_unlock(&fs_info->subvol_srcu, index); + + for (i = 0; i < ret; i++) { + if (!gang[i]) + continue; root_objectid = gang[i]->root_key.objectid; err = btrfs_orphan_cleanup(gang[i]); if (err) - return err; + break; + btrfs_put_fs_root(gang[i]); } root_objectid++; } - return 0; + + /* release the uncleaned roots due to error */ + for (; i < ret; i++) { + if (gang[i]) + btrfs_put_fs_root(gang[i]); + } + return err; } int btrfs_commit_super(struct btrfs_root *root) From 5f43f86e3f62734dfeeb36c4d940425a2c6789dc Mon Sep 17 00:00:00 2001 From: Gui Hecheng Date: Mon, 21 Apr 2014 20:13:11 +0800 Subject: [PATCH 23/80] btrfs: fix wrong max system array size check in kernel space For system chunk array, We copy a "disk_key" and an chunk item each time, so there should be enough space to hold both of them, not only the chunk item. Signed-off-by: Gui Hecheng Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index ad8e342c4b78..5864f05243c2 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -3921,7 +3921,8 @@ static int btrfs_add_system_chunk(struct btrfs_root *root, u8 *ptr; array_size = btrfs_super_sys_array_size(super_copy); - if (array_size + item_size > BTRFS_SYSTEM_CHUNK_ARRAY_SIZE) + if (array_size + item_size + sizeof(disk_key) + > BTRFS_SYSTEM_CHUNK_ARRAY_SIZE) return -EFBIG; ptr = super_copy->sys_chunk_array + array_size; From 23f8f9b7caca83d3397b1288a519c6579b7fda88 Mon Sep 17 00:00:00 2001 From: Gui Hecheng Date: Mon, 21 Apr 2014 20:13:12 +0800 Subject: [PATCH 24/80] btrfs: add dev maxs limit for __btrfs_alloc_chunk in kernel space For RAID0,5,6,10, For system chunk, there shouldn't be too many stripes to make a btrfs_chunk that exceeds BTRFS_SYSTEM_CHUNK_ARRAY_SIZE For data/meta chunk, there shouldn't be too many stripes to make a btrfs_chunk that exceeds a leaf. Signed-off-by: Gui Hecheng Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 5864f05243c2..8113f4567084 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -4027,6 +4027,16 @@ static void check_raid56_incompat_flag(struct btrfs_fs_info *info, u64 type) btrfs_set_fs_incompat(info, RAID56); } +#define BTRFS_MAX_DEVS(r) ((BTRFS_LEAF_DATA_SIZE(r) \ + - sizeof(struct btrfs_item) \ + - sizeof(struct btrfs_chunk)) \ + / sizeof(struct btrfs_stripe) + 1) + +#define BTRFS_MAX_DEVS_SYS_CHUNK ((BTRFS_SYSTEM_CHUNK_ARRAY_SIZE \ + - 2 * sizeof(struct btrfs_disk_key) \ + - 2 * sizeof(struct btrfs_chunk)) \ + / sizeof(struct btrfs_stripe) + 1) + static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, struct btrfs_root *extent_root, u64 start, u64 type) @@ -4076,6 +4086,8 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, if (type & BTRFS_BLOCK_GROUP_DATA) { max_stripe_size = 1024 * 1024 * 1024; max_chunk_size = 10 * max_stripe_size; + if (!devs_max) + devs_max = BTRFS_MAX_DEVS(info->chunk_root); } else if (type & BTRFS_BLOCK_GROUP_METADATA) { /* for larger filesystems, use larger metadata chunks */ if (fs_devices->total_rw_bytes > 50ULL * 1024 * 1024 * 1024) @@ -4083,9 +4095,13 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, else max_stripe_size = 256 * 1024 * 1024; max_chunk_size = max_stripe_size; + if (!devs_max) + devs_max = BTRFS_MAX_DEVS(info->chunk_root); } else if (type & BTRFS_BLOCK_GROUP_SYSTEM) { max_stripe_size = 32 * 1024 * 1024; max_chunk_size = 2 * max_stripe_size; + if (!devs_max) + devs_max = BTRFS_MAX_DEVS_SYS_CHUNK; } else { btrfs_err(info, "invalid chunk type 0x%llx requested\n", type); From b46ab97bcd4ae7954b3a150f642a82cdd1434f40 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Fri, 21 Mar 2014 12:46:54 +0000 Subject: [PATCH 25/80] Btrfs: send, avoid unnecessary inode item lookup in the btree Regardless of whether the caller is interested or not in knowing the inode's generation (dir_gen != NULL), get_first_ref always does a btree lookup to get the inode item. Avoid this useless lookup if dir_gen parameter is NULL (which is in some cases). Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/send.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index c76400dda4df..98e9e0cf7e8e 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -1693,10 +1693,12 @@ static int get_first_ref(struct btrfs_root *root, u64 ino, goto out; btrfs_release_path(path); - ret = get_inode_info(root, parent_dir, NULL, dir_gen, NULL, NULL, - NULL, NULL); - if (ret < 0) - goto out; + if (dir_gen) { + ret = get_inode_info(root, parent_dir, NULL, dir_gen, NULL, + NULL, NULL, NULL); + if (ret < 0) + goto out; + } *dir = parent_dir; @@ -1712,13 +1714,12 @@ static int is_first_ref(struct btrfs_root *root, int ret; struct fs_path *tmp_name; u64 tmp_dir; - u64 tmp_dir_gen; tmp_name = fs_path_alloc(); if (!tmp_name) return -ENOMEM; - ret = get_first_ref(root, ino, &tmp_dir, &tmp_dir_gen, tmp_name); + ret = get_first_ref(root, ino, &tmp_dir, NULL, tmp_name); if (ret < 0) goto out; From c992ec94f24c3e7135d6c23860615f269f0b1d87 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Sat, 22 Mar 2014 17:15:24 +0000 Subject: [PATCH 26/80] Btrfs: send, account for orphan directories when building path strings If we have directories with a pending move/rename operation, we must take into account any orphan directories that got created before executing the pending move/rename. Those orphan directories are directories with an inode number higher then the current send progress and that don't exist in the parent snapshot, they are created before current progress reaches their inode number, with a generated name of the form oN-M-I and at the root of the filesystem tree, and later when progress matches their inode number, moved/renamed to their final location. Reproducer: $ mkfs.btrfs -f /dev/sdd $ mount /dev/sdd /mnt $ mkdir -p /mnt/a/b/c/d $ mkdir /mnt/a/b/e $ mv /mnt/a/b/c /mnt/a/b/e/CC $ mkdir /mnt/a/b/e/CC/d/f $ mkdir /mnt/a/g $ btrfs subvolume snapshot -r /mnt /mnt/snap1 $ btrfs send /mnt/snap1 -f /tmp/base.send $ mkdir /mnt/a/g/h $ mv /mnt/a/b/e /mnt/a/g/h/EE $ mv /mnt/a/g/h/EE/CC/d /mnt/a/g/h/EE/DD $ btrfs subvolume snapshot -r /mnt /mnt/snap2 $ btrfs send -p /mnt/snap1 /mnt/snap2 -f /tmp/incremental.send The second receive command failed with the following error: ERROR: rename a/b/e/CC/d -> o264-7-0/EE/DD failed. No such file or directory A test case for xfstests follows soon. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/send.c | 35 ++++++++++------------------------- 1 file changed, 10 insertions(+), 25 deletions(-) diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index 98e9e0cf7e8e..c2bfd60ba245 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -3055,32 +3055,17 @@ static int apply_dir_move(struct send_ctx *sctx, struct pending_dir_move *pm) if (ret < 0) goto out; - if (parent_ino == sctx->cur_ino) { - /* child only renamed, not moved */ - ASSERT(parent_gen == sctx->cur_inode_gen); - ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, - from_path); - if (ret < 0) - goto out; - ret = fs_path_add_path(from_path, name); - if (ret < 0) - goto out; - } else { - /* child moved and maybe renamed too */ - sctx->send_progress = pm->ino; - ret = get_cur_path(sctx, pm->ino, pm->gen, from_path); - if (ret < 0) - goto out; - } - - fs_path_free(name); - name = NULL; - - to_path = fs_path_alloc(); - if (!to_path) { - ret = -ENOMEM; + ret = get_cur_path(sctx, parent_ino, parent_gen, + from_path); + if (ret < 0) goto out; - } + ret = fs_path_add_path(from_path, name); + if (ret < 0) + goto out; + + fs_path_reset(name); + to_path = name; + name = NULL; sctx->send_progress = sctx->cur_ino + 1; ret = get_cur_path(sctx, pm->ino, pm->gen, to_path); From a10c40766c30a002affc0c47dd515d048c3959b4 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Sat, 22 Mar 2014 17:16:30 +0000 Subject: [PATCH 27/80] Btrfs: send, remove dead code from __get_cur_name_and_parent Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/send.c | 6 ------ 1 file changed, 6 deletions(-) diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index c2bfd60ba245..fb6aeed9c223 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -2030,7 +2030,6 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx, { int ret; int nce_ret; - struct btrfs_path *path = NULL; struct name_cache_entry *nce = NULL; /* @@ -2056,10 +2055,6 @@ static int __get_cur_name_and_parent(struct send_ctx *sctx, } } - path = alloc_path_for_send(); - if (!path) - return -ENOMEM; - /* * If the inode is not existent yet, add the orphan name and return 1. * This should only happen for the parent dir that we determine in @@ -2135,7 +2130,6 @@ out_cache: name_cache_clean_unused(sctx); out: - btrfs_free_path(path); return ret; } From f959492fc15b60d874a9cbf55ae4760f2ef261ed Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Thu, 27 Mar 2014 20:14:01 +0000 Subject: [PATCH 28/80] Btrfs: send, fix more issues related to directory renames This is a continuation of the previous changes titled: Btrfs: fix incremental send's decision to delay a dir move/rename Btrfs: part 2, fix incremental send's decision to delay a dir move/rename There's a few more cases where a directory rename/move must be delayed which was previously overlooked. If our immediate ancestor has a lower inode number than ours and it doesn't have a delayed rename/move operation associated to it, it doesn't mean there isn't any non-direct ancestor of our current inode that needs to be renamed/moved before our current inode (i.e. with a higher inode number than ours). So we can't stop the search if our immediate ancestor has a lower inode number than ours, we need to navigate the directory hierarchy upwards until we hit the root or: 1) find an ancestor with an higher inode number that was renamed/moved in the send root too (or already has a pending rename/move registered); 2) find an ancestor that is a new directory (higher inode number than ours and exists only in the send root). Reproducer for case 1) $ mkfs.btrfs -f /dev/sdd $ mount /dev/sdd /mnt $ mkdir -p /mnt/a/b $ mkdir -p /mnt/a/c/d $ mkdir /mnt/a/b/e $ mkdir /mnt/a/c/d/f $ mv /mnt/a/b /mnt/a/c/d/2b $ mkdir /mnt/a/x $ mkdir /mnt/a/y $ btrfs subvolume snapshot -r /mnt /mnt/snap1 $ btrfs send /mnt/snap1 -f /tmp/base.send $ mv /mnt/a/x /mnt/a/y $ mv /mnt/a/c/d/2b/e /mnt/a/c/d/2b/2e $ mv /mnt/a/c/d /mnt/a/h/2d $ mv /mnt/a/c /mnt/a/h/2d/2b/2c $ btrfs subvolume snapshot -r /mnt /mnt/snap2 $ btrfs send -p /mnt/snap1 /mnt/snap2 -f /tmp/incremental.send Simple reproducer for case 2) $ mkfs.btrfs -f /dev/sdd $ mount /dev/sdd /mnt $ mkdir -p /mnt/a/b $ mkdir /mnt/a/c $ mv /mnt/a/b /mnt/a/c/b2 $ mkdir /mnt/a/e $ btrfs subvolume snapshot -r /mnt /mnt/snap1 $ btrfs send /mnt/snap1 -f /tmp/base.send $ mv /mnt/a/c/b2 /mnt/a/e/b3 $ mkdir /mnt/a/e/b3/f $ mkdir /mnt/a/h $ mv /mnt/a/c /mnt/a/e/b3/f/c2 $ mv /mnt/a/e /mnt/a/h/e2 $ btrfs subvolume snapshot -r /mnt /mnt/snap2 $ btrfs send -p /mnt/snap1 /mnt/snap2 -f /tmp/incremental.send Another simple reproducer for case 2) $ mkfs.btrfs -f /dev/sdd $ mount /dev/sdd /mnt $ mkdir -p /mnt/a/b $ mkdir /mnt/a/c $ mkdir /mnt/a/b/d $ mkdir /mnt/a/c/e $ btrfs subvolume snapshot -r /mnt /mnt/snap1 $ btrfs send /mnt/snap1 -f /tmp/base.send $ mkdir /mnt/a/b/d/f $ mkdir /mnt/a/b/g $ mv /mnt/a/c/e /mnt/a/b/g/e2 $ mv /mnt/a/c /mnt/a/b/d/f/c2 $ mv /mnt/a/b/d/f /mnt/a/b/g/e2/f2 $ btrfs subvolume snapshot -r /mnt /mnt/snap2 $ btrfs send -p /mnt/snap1 /mnt/snap2 -f /tmp/incremental.send More complex reproducer for case 2) $ mkfs.btrfs -f /dev/sdd $ mount /dev/sdd /mnt $ mkdir -p /mnt/a/b $ mkdir -p /mnt/a/c/d $ mkdir /mnt/a/b/e $ mkdir /mnt/a/c/d/f $ mv /mnt/a/b /mnt/a/c/d/2b $ mkdir /mnt/a/x $ mkdir /mnt/a/y $ btrfs subvolume snapshot -r /mnt /mnt/snap1 $ btrfs send /mnt/snap1 -f /tmp/base.send $ mv /mnt/a/x /mnt/a/y $ mv /mnt/a/c/d/2b/e /mnt/a/c/d/2b/2e $ mv /mnt/a/c/d /mnt/a/h/2d $ mv /mnt/a/c /mnt/a/h/2d/2b/2c $ btrfs subvolume snapshot -r /mnt /mnt/snap2 $ btrfs send -p /mnt/snap1 /mnt/snap2 -f /tmp/incremental.send For both cases the incremental send would enter an infinite loop when building path strings. While solving these cases, this change also re-implements the code to detect when directory moves/renames should be delayed. Instead of dealing with several specific cases separately, it's now more generic handling all cases with a simple detection algorithm and if when applying a delayed move/rename there's a path loop detected, it further delays the move/rename registering a new ancestor inode as the dependency inode (so our rename happens after that ancestor is renamed). Tests for these cases is being added to xfstests too. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/send.c | 190 ++++++++++++++++++++++++------------------------ 1 file changed, 96 insertions(+), 94 deletions(-) diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index fb6aeed9c223..3f14b31c0c96 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -2940,7 +2940,9 @@ static void free_waiting_dir_move(struct send_ctx *sctx, static int add_pending_dir_move(struct send_ctx *sctx, u64 ino, u64 ino_gen, - u64 parent_ino) + u64 parent_ino, + struct list_head *new_refs, + struct list_head *deleted_refs) { struct rb_node **p = &sctx->pending_dir_moves.rb_node; struct rb_node *parent = NULL; @@ -2972,12 +2974,12 @@ static int add_pending_dir_move(struct send_ctx *sctx, } } - list_for_each_entry(cur, &sctx->deleted_refs, list) { + list_for_each_entry(cur, deleted_refs, list) { ret = dup_ref(cur, &pm->update_refs); if (ret < 0) goto out; } - list_for_each_entry(cur, &sctx->new_refs, list) { + list_for_each_entry(cur, new_refs, list) { ret = dup_ref(cur, &pm->update_refs); if (ret < 0) goto out; @@ -3020,6 +3022,48 @@ static struct pending_dir_move *get_pending_dir_moves(struct send_ctx *sctx, return NULL; } +static int path_loop(struct send_ctx *sctx, struct fs_path *name, + u64 ino, u64 gen, u64 *ancestor_ino) +{ + int ret = 0; + u64 parent_inode = 0; + u64 parent_gen = 0; + u64 start_ino = ino; + + *ancestor_ino = 0; + while (ino != BTRFS_FIRST_FREE_OBJECTID) { + fs_path_reset(name); + + if (is_waiting_for_rm(sctx, ino)) + break; + if (is_waiting_for_move(sctx, ino)) { + if (*ancestor_ino == 0) + *ancestor_ino = ino; + ret = get_first_ref(sctx->parent_root, ino, + &parent_inode, &parent_gen, name); + } else { + ret = __get_cur_name_and_parent(sctx, ino, gen, + &parent_inode, + &parent_gen, name); + if (ret > 0) { + ret = 0; + break; + } + } + if (ret < 0) + break; + if (parent_inode == start_ino) { + ret = 1; + if (*ancestor_ino == 0) + *ancestor_ino = ino; + break; + } + ino = parent_inode; + gen = parent_gen; + } + return ret; +} + static int apply_dir_move(struct send_ctx *sctx, struct pending_dir_move *pm) { struct fs_path *from_path = NULL; @@ -3031,6 +3075,7 @@ static int apply_dir_move(struct send_ctx *sctx, struct pending_dir_move *pm) struct waiting_dir_move *dm = NULL; u64 rmdir_ino = 0; int ret; + u64 ancestor = 0; name = fs_path_alloc(); from_path = fs_path_alloc(); @@ -3057,11 +3102,25 @@ static int apply_dir_move(struct send_ctx *sctx, struct pending_dir_move *pm) if (ret < 0) goto out; + sctx->send_progress = sctx->cur_ino + 1; + ret = path_loop(sctx, name, pm->ino, pm->gen, &ancestor); + if (ret) { + LIST_HEAD(deleted_refs); + ASSERT(ancestor > BTRFS_FIRST_FREE_OBJECTID); + ret = add_pending_dir_move(sctx, pm->ino, pm->gen, ancestor, + &pm->update_refs, &deleted_refs); + if (ret < 0) + goto out; + if (rmdir_ino) { + dm = get_waiting_dir_move(sctx, pm->ino); + ASSERT(dm); + dm->rmdir_ino = rmdir_ino; + } + goto out; + } fs_path_reset(name); to_path = name; name = NULL; - - sctx->send_progress = sctx->cur_ino + 1; ret = get_cur_path(sctx, pm->ino, pm->gen, to_path); if (ret < 0) goto out; @@ -3185,127 +3244,74 @@ out: static int wait_for_parent_move(struct send_ctx *sctx, struct recorded_ref *parent_ref) { - int ret; + int ret = 0; u64 ino = parent_ref->dir; u64 parent_ino_before, parent_ino_after; - u64 old_gen; struct fs_path *path_before = NULL; struct fs_path *path_after = NULL; int len1, len2; - int register_upper_dirs; - u64 gen; - - if (is_waiting_for_move(sctx, ino)) - return 1; - - if (parent_ref->dir <= sctx->cur_ino) - return 0; - - ret = get_inode_info(sctx->parent_root, ino, NULL, &old_gen, - NULL, NULL, NULL, NULL); - if (ret == -ENOENT) - return 0; - else if (ret < 0) - return ret; - - if (parent_ref->dir_gen != old_gen) - return 0; - - path_before = fs_path_alloc(); - if (!path_before) - return -ENOMEM; - - ret = get_first_ref(sctx->parent_root, ino, &parent_ino_before, - NULL, path_before); - if (ret == -ENOENT) { - ret = 0; - goto out; - } else if (ret < 0) { - goto out; - } path_after = fs_path_alloc(); - if (!path_after) { + path_before = fs_path_alloc(); + if (!path_after || !path_before) { ret = -ENOMEM; goto out; } - ret = get_first_ref(sctx->send_root, ino, &parent_ino_after, - &gen, path_after); - if (ret == -ENOENT) { - ret = 0; - goto out; - } else if (ret < 0) { - goto out; - } - - len1 = fs_path_len(path_before); - len2 = fs_path_len(path_after); - if (parent_ino_before != parent_ino_after || len1 != len2 || - memcmp(path_before->start, path_after->start, len1)) { - ret = 1; - goto out; - } - ret = 0; - /* - * Ok, our new most direct ancestor has a higher inode number but - * wasn't moved/renamed. So maybe some of the new ancestors higher in - * the hierarchy have an higher inode number too *and* were renamed - * or moved - in this case we need to wait for the ancestor's rename - * or move operation before we can do the move/rename for the current - * inode. + * Our current directory inode may not yet be renamed/moved because some + * ancestor (immediate or not) has to be renamed/moved first. So find if + * such ancestor exists and make sure our own rename/move happens after + * that ancestor is processed. */ - register_upper_dirs = 0; - ino = parent_ino_after; -again: - while ((ret == 0 || register_upper_dirs) && ino > sctx->cur_ino) { - u64 parent_gen; + while (ino > BTRFS_FIRST_FREE_OBJECTID) { + if (is_waiting_for_move(sctx, ino)) { + ret = 1; + break; + } fs_path_reset(path_before); fs_path_reset(path_after); ret = get_first_ref(sctx->send_root, ino, &parent_ino_after, - &parent_gen, path_after); + NULL, path_after); if (ret < 0) goto out; ret = get_first_ref(sctx->parent_root, ino, &parent_ino_before, NULL, path_before); - if (ret == -ENOENT) { - ret = 0; - break; - } else if (ret < 0) { + if (ret < 0 && ret != -ENOENT) { goto out; + } else if (ret == -ENOENT) { + ret = 1; + break; } len1 = fs_path_len(path_before); len2 = fs_path_len(path_after); - if (parent_ino_before != parent_ino_after || len1 != len2 || - memcmp(path_before->start, path_after->start, len1)) { + if (ino > sctx->cur_ino && + (parent_ino_before != parent_ino_after || len1 != len2 || + memcmp(path_before->start, path_after->start, len1))) { ret = 1; - if (register_upper_dirs) { - break; - } else { - register_upper_dirs = 1; - ino = parent_ref->dir; - gen = parent_ref->dir_gen; - goto again; - } - } else if (register_upper_dirs) { - ret = add_pending_dir_move(sctx, ino, gen, - parent_ino_after); - if (ret < 0 && ret != -EEXIST) - goto out; + break; } - ino = parent_ino_after; - gen = parent_gen; } out: fs_path_free(path_before); fs_path_free(path_after); + if (ret == 1) { + ret = add_pending_dir_move(sctx, + sctx->cur_ino, + sctx->cur_inode_gen, + ino, + &sctx->new_refs, + &sctx->deleted_refs); + if (!ret) + ret = 1; + } + return ret; } @@ -3466,10 +3472,6 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino); if (ret < 0) goto out; if (ret) { - ret = add_pending_dir_move(sctx, - sctx->cur_ino, - sctx->cur_inode_gen, - cur->dir); *pending_move = 1; } else { ret = send_rename(sctx, valid_path, From 27cdeb7096b86f05ad018a24cdb63acdf0850a5d Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Wed, 2 Apr 2014 19:51:05 +0800 Subject: [PATCH 29/80] Btrfs: use bitfield instead of integer data type for the some variants in btrfs_root Signed-off-by: Miao Xie Signed-off-by: Wang Shilong Signed-off-by: Chris Mason --- fs/btrfs/ctree.c | 25 ++++++++++++++----------- fs/btrfs/ctree.h | 39 +++++++++++++++++++++------------------ fs/btrfs/disk-io.c | 33 +++++++++++++++------------------ fs/btrfs/extent-tree.c | 6 +++--- fs/btrfs/file.c | 4 +++- fs/btrfs/inode.c | 29 ++++++++++++++++++----------- fs/btrfs/ioctl.c | 4 ++-- fs/btrfs/relocation.c | 17 +++++++++-------- fs/btrfs/root-tree.c | 2 +- fs/btrfs/transaction.c | 33 +++++++++++++++++---------------- fs/btrfs/tree-defrag.c | 2 +- fs/btrfs/tree-log.c | 9 +++++---- 12 files changed, 109 insertions(+), 94 deletions(-) diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index 1bcfcdb23cf4..2f10e12ae94c 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -224,7 +224,8 @@ static struct extent_buffer *btrfs_read_lock_root_node(struct btrfs_root *root) static void add_root_to_dirty_list(struct btrfs_root *root) { spin_lock(&root->fs_info->trans_lock); - if (root->track_dirty && list_empty(&root->dirty_list)) { + if (test_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state) && + list_empty(&root->dirty_list)) { list_add(&root->dirty_list, &root->fs_info->dirty_cowonly_roots); } @@ -246,9 +247,10 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans, int level; struct btrfs_disk_key disk_key; - WARN_ON(root->ref_cows && trans->transid != - root->fs_info->running_transaction->transid); - WARN_ON(root->ref_cows && trans->transid != root->last_trans); + WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) && + trans->transid != root->fs_info->running_transaction->transid); + WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) && + trans->transid != root->last_trans); level = btrfs_header_level(buf); if (level == 0) @@ -997,14 +999,14 @@ int btrfs_block_can_be_shared(struct btrfs_root *root, * snapshot and the block was not allocated by tree relocation, * we know the block is not shared. */ - if (root->ref_cows && + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) && buf != root->node && buf != root->commit_root && (btrfs_header_generation(buf) <= btrfs_root_last_snapshot(&root->root_item) || btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC))) return 1; #ifdef BTRFS_COMPAT_EXTENT_TREE_V0 - if (root->ref_cows && + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) && btrfs_header_backref_rev(buf) < BTRFS_MIXED_BACKREF_REV) return 1; #endif @@ -1146,9 +1148,10 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans, btrfs_assert_tree_locked(buf); - WARN_ON(root->ref_cows && trans->transid != - root->fs_info->running_transaction->transid); - WARN_ON(root->ref_cows && trans->transid != root->last_trans); + WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) && + trans->transid != root->fs_info->running_transaction->transid); + WARN_ON(test_bit(BTRFS_ROOT_REF_COWS, &root->state) && + trans->transid != root->last_trans); level = btrfs_header_level(buf); @@ -1193,7 +1196,7 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans, return ret; } - if (root->ref_cows) { + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) { ret = btrfs_reloc_cow_block(trans, root, buf, cow); if (ret) return ret; @@ -1556,7 +1559,7 @@ static inline int should_cow_block(struct btrfs_trans_handle *trans, !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) && !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID && btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)) && - !root->force_cow) + !test_bit(BTRFS_ROOT_FORCE_COW, &root->state)) return 0; return 1; } diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 0a805b8d61cc..efd3bf61696d 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1710,6 +1710,26 @@ struct btrfs_subvolume_writers { wait_queue_head_t wait; }; +/* + * The state of btrfs root + */ +/* + * btrfs_record_root_in_trans is a multi-step process, + * and it can race with the balancing code. But the + * race is very small, and only the first time the root + * is added to each transaction. So IN_TRANS_SETUP + * is used to tell us when more checks are required + */ +#define BTRFS_ROOT_IN_TRANS_SETUP 0 +#define BTRFS_ROOT_REF_COWS 1 +#define BTRFS_ROOT_TRACK_DIRTY 2 +#define BTRFS_ROOT_IN_RADIX 3 +#define BTRFS_ROOT_DUMMY_ROOT 4 +#define BTRFS_ROOT_ORPHAN_ITEM_INSERTED 5 +#define BTRFS_ROOT_DEFRAG_RUNNING 6 +#define BTRFS_ROOT_FORCE_COW 7 +#define BTRFS_ROOT_MULTI_LOG_TASKS 8 + /* * in ram representation of the tree. extent_root is used for all allocations * and for the extent tree extent_root root. @@ -1721,6 +1741,7 @@ struct btrfs_root { struct btrfs_root *log_root; struct btrfs_root *reloc_root; + unsigned long state; struct btrfs_root_item root_item; struct btrfs_key root_key; struct btrfs_fs_info *fs_info; @@ -1755,7 +1776,6 @@ struct btrfs_root { /* Just be updated when the commit succeeds. */ int last_log_commit; pid_t log_start_pid; - bool log_multiple_pids; u64 objectid; u64 last_trans; @@ -1775,23 +1795,9 @@ struct btrfs_root { u64 highest_objectid; - /* btrfs_record_root_in_trans is a multi-step process, - * and it can race with the balancing code. But the - * race is very small, and only the first time the root - * is added to each transaction. So in_trans_setup - * is used to tell us when more checks are required - */ - unsigned long in_trans_setup; - int ref_cows; - int track_dirty; - int in_radix; -#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS - int dummy_root; -#endif u64 defrag_trans_start; struct btrfs_key defrag_progress; struct btrfs_key defrag_max; - int defrag_running; char *name; /* the dirty list is only used by non-reference counted roots */ @@ -1805,7 +1811,6 @@ struct btrfs_root { spinlock_t orphan_lock; atomic_t orphan_inodes; struct btrfs_block_rsv *orphan_block_rsv; - int orphan_item_inserted; int orphan_cleanup_state; spinlock_t inode_lock; @@ -1823,8 +1828,6 @@ struct btrfs_root { */ dev_t anon_dev; - int force_cow; - spinlock_t root_item_lock; atomic_t refs; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 85f514483bd1..cf8427a7a615 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1201,10 +1201,7 @@ static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, root->nodesize = nodesize; root->leafsize = leafsize; root->stripesize = stripesize; - root->ref_cows = 0; - root->track_dirty = 0; - root->in_radix = 0; - root->orphan_item_inserted = 0; + root->state = 0; root->orphan_cleanup_state = 0; root->objectid = objectid; @@ -1265,7 +1262,6 @@ static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, else root->defrag_trans_start = 0; init_completion(&root->kobj_unregister); - root->defrag_running = 0; root->root_key.objectid = objectid; root->anon_dev = 0; @@ -1290,7 +1286,7 @@ struct btrfs_root *btrfs_alloc_dummy_root(void) if (!root) return ERR_PTR(-ENOMEM); __setup_root(4096, 4096, 4096, 4096, root, NULL, 1); - root->dummy_root = 1; + set_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state); return root; } @@ -1341,8 +1337,7 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans, btrfs_mark_buffer_dirty(leaf); root->commit_root = btrfs_root_node(root); - root->track_dirty = 1; - + set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state); root->root_item.flags = 0; root->root_item.byte_limit = 0; @@ -1396,13 +1391,15 @@ static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans, root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID; root->root_key.type = BTRFS_ROOT_ITEM_KEY; root->root_key.offset = BTRFS_TREE_LOG_OBJECTID; + /* + * DON'T set REF_COWS for log trees + * * log trees do not get reference counted because they go away * before a real commit is actually done. They do store pointers * to file data extents, and those reference counts still get * updated (along with back refs to the log tree). */ - root->ref_cows = 0; leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0, BTRFS_TREE_LOG_OBJECTID, NULL, @@ -1536,7 +1533,7 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_root *tree_root, return root; if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) { - root->ref_cows = 1; + set_bit(BTRFS_ROOT_REF_COWS, &root->state); btrfs_check_and_init_root_item(&root->root_item); } @@ -1606,7 +1603,7 @@ int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info, (unsigned long)root->root_key.objectid, root); if (ret == 0) - root->in_radix = 1; + set_bit(BTRFS_ROOT_IN_RADIX, &root->state); spin_unlock(&fs_info->fs_roots_radix_lock); radix_tree_preload_end(); @@ -1662,7 +1659,7 @@ again: if (ret < 0) goto fail; if (ret == 0) - root->orphan_item_inserted = 1; + set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state); ret = btrfs_insert_fs_root(fs_info, root); if (ret) { @@ -2101,7 +2098,7 @@ static void del_fs_roots(struct btrfs_fs_info *fs_info) struct btrfs_root, root_list); list_del(&gang[0]->root_list); - if (gang[0]->in_radix) { + if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) { btrfs_drop_and_free_fs_root(fs_info, gang[0]); } else { free_extent_buffer(gang[0]->node); @@ -2694,7 +2691,7 @@ retry_root_backup: ret = PTR_ERR(extent_root); goto recovery_tree_root; } - extent_root->track_dirty = 1; + set_bit(BTRFS_ROOT_TRACK_DIRTY, &extent_root->state); fs_info->extent_root = extent_root; location.objectid = BTRFS_DEV_TREE_OBJECTID; @@ -2703,7 +2700,7 @@ retry_root_backup: ret = PTR_ERR(dev_root); goto recovery_tree_root; } - dev_root->track_dirty = 1; + set_bit(BTRFS_ROOT_TRACK_DIRTY, &dev_root->state); fs_info->dev_root = dev_root; btrfs_init_devices_late(fs_info); @@ -2713,13 +2710,13 @@ retry_root_backup: ret = PTR_ERR(csum_root); goto recovery_tree_root; } - csum_root->track_dirty = 1; + set_bit(BTRFS_ROOT_TRACK_DIRTY, &csum_root->state); fs_info->csum_root = csum_root; location.objectid = BTRFS_QUOTA_TREE_OBJECTID; quota_root = btrfs_read_tree_root(tree_root, &location); if (!IS_ERR(quota_root)) { - quota_root->track_dirty = 1; + set_bit(BTRFS_ROOT_TRACK_DIRTY, "a_root->state); fs_info->quota_enabled = 1; fs_info->pending_quota_state = 1; fs_info->quota_root = quota_root; @@ -2734,7 +2731,7 @@ retry_root_backup: create_uuid_tree = true; check_uuid_tree = false; } else { - uuid_root->track_dirty = 1; + set_bit(BTRFS_ROOT_TRACK_DIRTY, &uuid_root->state); fs_info->uuid_root = uuid_root; create_uuid_tree = false; check_uuid_tree = diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 15467e3f5876..7c18566fd2b4 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2983,7 +2983,7 @@ static int __btrfs_mod_ref(struct btrfs_trans_handle *trans, nritems = btrfs_header_nritems(buf); level = btrfs_header_level(buf); - if (!root->ref_cows && level == 0) + if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state) && level == 0) return 0; if (inc) @@ -4472,7 +4472,7 @@ static struct btrfs_block_rsv *get_block_rsv( { struct btrfs_block_rsv *block_rsv = NULL; - if (root->ref_cows) + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) block_rsv = trans->block_rsv; if (root == root->fs_info->csum_root && trans->adding_csums) @@ -7838,7 +7838,7 @@ int btrfs_drop_snapshot(struct btrfs_root *root, } } - if (root->in_radix) { + if (test_bit(BTRFS_ROOT_IN_RADIX, &root->state)) { btrfs_drop_and_free_fs_root(tree_root->fs_info, root); } else { free_extent_buffer(root->node); diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 3029925e96d7..5c6947dbc948 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -714,7 +714,7 @@ int __btrfs_drop_extents(struct btrfs_trans_handle *trans, int recow; int ret; int modify_tree = -1; - int update_refs = (root->ref_cows || root == root->fs_info->tree_root); + int update_refs; int found = 0; int leafs_visited = 0; @@ -724,6 +724,8 @@ int __btrfs_drop_extents(struct btrfs_trans_handle *trans, if (start >= BTRFS_I(inode)->disk_i_size && !replace_extent) modify_tree = 0; + update_refs = (test_bit(BTRFS_ROOT_REF_COWS, &root->state) || + root == root->fs_info->tree_root); while (1) { recow = 0; ret = btrfs_lookup_file_extent(trans, root, path, ino, diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 36c13e391ae3..ef3bd3deca12 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -2947,14 +2947,15 @@ void btrfs_orphan_commit_root(struct btrfs_trans_handle *trans, root->orphan_block_rsv = NULL; spin_unlock(&root->orphan_lock); - if (root->orphan_item_inserted && + if (test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state) && btrfs_root_refs(&root->root_item) > 0) { ret = btrfs_del_orphan_item(trans, root->fs_info->tree_root, root->root_key.objectid); if (ret) btrfs_abort_transaction(trans, root, ret); else - root->orphan_item_inserted = 0; + clear_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, + &root->state); } if (block_rsv) { @@ -3271,7 +3272,8 @@ int btrfs_orphan_cleanup(struct btrfs_root *root) btrfs_block_rsv_release(root, root->orphan_block_rsv, (u64)-1); - if (root->orphan_block_rsv || root->orphan_item_inserted) { + if (root->orphan_block_rsv || + test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state)) { trans = btrfs_join_transaction(root); if (!IS_ERR(trans)) btrfs_end_transaction(trans, root); @@ -3998,7 +4000,8 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, * not block aligned since we will be keeping the last block of the * extent just the way it is. */ - if (root->ref_cows || root == root->fs_info->tree_root) + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) || + root == root->fs_info->tree_root) btrfs_drop_extent_cache(inode, ALIGN(new_size, root->sectorsize), (u64)-1, 0); @@ -4091,7 +4094,9 @@ search_again: extent_num_bytes); num_dec = (orig_num_bytes - extent_num_bytes); - if (root->ref_cows && extent_start != 0) + if (test_bit(BTRFS_ROOT_REF_COWS, + &root->state) && + extent_start != 0) inode_sub_bytes(inode, num_dec); btrfs_mark_buffer_dirty(leaf); } else { @@ -4105,7 +4110,8 @@ search_again: num_dec = btrfs_file_extent_num_bytes(leaf, fi); if (extent_start != 0) { found_extent = 1; - if (root->ref_cows) + if (test_bit(BTRFS_ROOT_REF_COWS, + &root->state)) inode_sub_bytes(inode, num_dec); } } @@ -4120,10 +4126,9 @@ search_again: btrfs_file_extent_other_encoding(leaf, fi) == 0) { u32 size = new_size - found_key.offset; - if (root->ref_cows) { + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) inode_sub_bytes(inode, item_end + 1 - new_size); - } /* * update the ram bytes to properly reflect @@ -4133,7 +4138,8 @@ search_again: size = btrfs_file_extent_calc_inline_size(size); btrfs_truncate_item(root, path, size, 1); - } else if (root->ref_cows) { + } else if (test_bit(BTRFS_ROOT_REF_COWS, + &root->state)) { inode_sub_bytes(inode, item_end + 1 - found_key.offset); } @@ -4155,8 +4161,9 @@ delete: } else { break; } - if (found_extent && (root->ref_cows || - root == root->fs_info->tree_root)) { + if (found_extent && + (test_bit(BTRFS_ROOT_REF_COWS, &root->state) || + root == root->fs_info->tree_root)) { btrfs_set_path_blocking(path); ret = btrfs_free_extent(trans, root, extent_start, extent_num_bytes, 0, diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index ebac486f59af..242a37cd26b2 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -638,7 +638,7 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir, struct btrfs_trans_handle *trans; int ret; - if (!root->ref_cows) + if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state)) return -EINVAL; atomic_inc(&root->will_be_snapshoted); @@ -2369,7 +2369,7 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file, dest->root_item.drop_level = 0; btrfs_set_root_refs(&dest->root_item, 0); - if (!xchg(&dest->orphan_item_inserted, 1)) { + if (!test_and_set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &dest->state)) { ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root, dest->root_key.objectid); diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 7f92ab1daa87..b9cf0f522513 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -528,7 +528,7 @@ static int should_ignore_root(struct btrfs_root *root) { struct btrfs_root *reloc_root; - if (!root->ref_cows) + if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state)) return 0; reloc_root = root->reloc_root; @@ -610,7 +610,7 @@ struct btrfs_root *find_tree_root(struct reloc_control *rc, root = read_fs_root(rc->extent_root->fs_info, root_objectid); BUG_ON(IS_ERR(root)); - if (root->ref_cows && + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) && generation != btrfs_root_generation(&root->root_item)) return NULL; @@ -887,7 +887,7 @@ again: goto out; } - if (!root->ref_cows) + if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state)) cur->cowonly = 1; if (btrfs_root_level(&root->root_item) == cur->level) { @@ -954,7 +954,8 @@ again: upper->bytenr = eb->start; upper->owner = btrfs_header_owner(eb); upper->level = lower->level + 1; - if (!root->ref_cows) + if (!test_bit(BTRFS_ROOT_REF_COWS, + &root->state)) upper->cowonly = 1; /* @@ -2441,7 +2442,7 @@ struct btrfs_root *select_reloc_root(struct btrfs_trans_handle *trans, next = walk_up_backref(next, edges, &index); root = next->root; BUG_ON(!root); - BUG_ON(!root->ref_cows); + BUG_ON(!test_bit(BTRFS_ROOT_REF_COWS, &root->state)); if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) { record_reloc_root_in_trans(trans, root); @@ -2506,7 +2507,7 @@ struct btrfs_root *select_one_root(struct btrfs_trans_handle *trans, BUG_ON(!root); /* no other choice for non-references counted tree */ - if (!root->ref_cows) + if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state)) return root; if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) @@ -2893,14 +2894,14 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans, goto out; } - if (!root || root->ref_cows) { + if (!root || test_bit(BTRFS_ROOT_REF_COWS, &root->state)) { ret = reserve_metadata_space(trans, rc, node); if (ret) goto out; } if (root) { - if (root->ref_cows) { + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) { BUG_ON(node->new_bytenr); BUG_ON(!list_empty(&node->list)); btrfs_record_root_in_trans(trans, root); diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c index 38bb47e7d6b1..360a728a639f 100644 --- a/fs/btrfs/root-tree.c +++ b/fs/btrfs/root-tree.c @@ -306,7 +306,7 @@ int btrfs_find_orphan_roots(struct btrfs_root *tree_root) break; } - root->orphan_item_inserted = 1; + set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state); err = btrfs_insert_fs_root(root->fs_info, root); if (err) { diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index c30815e79235..7c4c049da871 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -241,18 +241,19 @@ loop: static int record_root_in_trans(struct btrfs_trans_handle *trans, struct btrfs_root *root) { - if (root->ref_cows && root->last_trans < trans->transid) { + if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) && + root->last_trans < trans->transid) { WARN_ON(root == root->fs_info->extent_root); WARN_ON(root->commit_root != root->node); /* - * see below for in_trans_setup usage rules + * see below for IN_TRANS_SETUP usage rules * we have the reloc mutex held now, so there * is only one writer in this function */ - root->in_trans_setup = 1; + set_bit(BTRFS_ROOT_IN_TRANS_SETUP, &root->state); - /* make sure readers find in_trans_setup before + /* make sure readers find IN_TRANS_SETUP before * they find our root->last_trans update */ smp_wmb(); @@ -279,7 +280,7 @@ static int record_root_in_trans(struct btrfs_trans_handle *trans, * But, we have to set root->last_trans before we * init the relocation root, otherwise, we trip over warnings * in ctree.c. The solution used here is to flag ourselves - * with root->in_trans_setup. When this is 1, we're still + * with root IN_TRANS_SETUP. When this is 1, we're still * fixing up the reloc trees and everyone must wait. * * When this is zero, they can trust root->last_trans and fly @@ -288,8 +289,8 @@ static int record_root_in_trans(struct btrfs_trans_handle *trans, * done before we pop in the zero below */ btrfs_init_reloc_root(trans, root); - smp_wmb(); - root->in_trans_setup = 0; + smp_mb__before_clear_bit(); + clear_bit(BTRFS_ROOT_IN_TRANS_SETUP, &root->state); } return 0; } @@ -298,16 +299,16 @@ static int record_root_in_trans(struct btrfs_trans_handle *trans, int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans, struct btrfs_root *root) { - if (!root->ref_cows) + if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state)) return 0; /* - * see record_root_in_trans for comments about in_trans_setup usage + * see record_root_in_trans for comments about IN_TRANS_SETUP usage * and barriers */ smp_rmb(); if (root->last_trans == trans->transid && - !root->in_trans_setup) + !test_bit(BTRFS_ROOT_IN_TRANS_SETUP, &root->state)) return 0; mutex_lock(&root->fs_info->reloc_mutex); @@ -365,7 +366,7 @@ static int may_wait_transaction(struct btrfs_root *root, int type) static inline bool need_reserve_reloc_root(struct btrfs_root *root) { if (!root->fs_info->reloc_ctl || - !root->ref_cows || + !test_bit(BTRFS_ROOT_REF_COWS, &root->state) || root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID || root->reloc_root) return false; @@ -1049,8 +1050,8 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans, btrfs_save_ino_cache(root, trans); /* see comments in should_cow_block() */ - root->force_cow = 0; - smp_wmb(); + clear_bit(BTRFS_ROOT_FORCE_COW, &root->state); + smp_mb__after_clear_bit(); if (root->commit_root != root->node) { list_add_tail(&root->dirty_list, @@ -1081,7 +1082,7 @@ int btrfs_defrag_root(struct btrfs_root *root) struct btrfs_trans_handle *trans; int ret; - if (xchg(&root->defrag_running, 1)) + if (test_and_set_bit(BTRFS_ROOT_DEFRAG_RUNNING, &root->state)) return 0; while (1) { @@ -1104,7 +1105,7 @@ int btrfs_defrag_root(struct btrfs_root *root) break; } } - root->defrag_running = 0; + clear_bit(BTRFS_ROOT_DEFRAG_RUNNING, &root->state); return ret; } @@ -1271,7 +1272,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans, } /* see comments in should_cow_block() */ - root->force_cow = 1; + set_bit(BTRFS_ROOT_FORCE_COW, &root->state); smp_wmb(); btrfs_set_root_node(new_root_item, tmp); diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c index 76928ca97741..a63719cc9578 100644 --- a/fs/btrfs/tree-defrag.c +++ b/fs/btrfs/tree-defrag.c @@ -49,7 +49,7 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans, goto out; } - if (root->ref_cows == 0) + if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state)) goto out; if (btrfs_test_opt(root, SSD)) diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index e2f45fc02610..ffee15856130 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -152,9 +152,9 @@ static int start_log_trans(struct btrfs_trans_handle *trans, if (!root->log_start_pid) { root->log_start_pid = current->pid; - root->log_multiple_pids = false; + clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state); } else if (root->log_start_pid != current->pid) { - root->log_multiple_pids = true; + set_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state); } atomic_inc(&root->log_batch); @@ -181,7 +181,7 @@ static int start_log_trans(struct btrfs_trans_handle *trans, if (ret) goto out; } - root->log_multiple_pids = false; + clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state); root->log_start_pid = current->pid; atomic_inc(&root->log_batch); atomic_inc(&root->log_writers); @@ -2500,7 +2500,8 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, while (1) { int batch = atomic_read(&root->log_batch); /* when we're on an ssd, just kick the log commit out */ - if (!btrfs_test_opt(root, SSD) && root->log_multiple_pids) { + if (!btrfs_test_opt(root, SSD) && + test_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state)) { mutex_unlock(&root->log_mutex); schedule_timeout_uninterruptible(1); mutex_lock(&root->log_mutex); From 4d90d28b1c5c5ab6292c0888dc1ebdfc3023f877 Mon Sep 17 00:00:00 2001 From: Anand Jain Date: Sun, 6 Apr 2014 12:59:07 +0800 Subject: [PATCH 30/80] btrfs: btrfs_rm_device() should zero mirror SB as well This fix will ensure all SB copies on the disk is zeroed when the disk is intentionally removed. This helps to better manage disks in the user land. This version of patch also merges the Zach patch as below. btrfs: don't double brelse on device rm Signed-off-by: Anand Jain Signed-off-by: Zach Brown Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 31 +++++++++++++++++++++++++++++++ 1 file changed, 31 insertions(+) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 8113f4567084..ee71622088f9 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1710,12 +1710,43 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) * remove it from the devices list and zero out the old super */ if (clear_super && disk_super) { + u64 bytenr; + int i; + /* make sure this device isn't detected as part of * the FS anymore */ memset(&disk_super->magic, 0, sizeof(disk_super->magic)); set_buffer_dirty(bh); sync_dirty_buffer(bh); + + /* clear the mirror copies of super block on the disk + * being removed, 0th copy is been taken care above and + * the below would take of the rest + */ + for (i = 1; i < BTRFS_SUPER_MIRROR_MAX; i++) { + bytenr = btrfs_sb_offset(i); + if (bytenr + BTRFS_SUPER_INFO_SIZE >= + i_size_read(bdev->bd_inode)) + break; + + brelse(bh); + bh = __bread(bdev, bytenr / 4096, + BTRFS_SUPER_INFO_SIZE); + if (!bh) + continue; + + disk_super = (struct btrfs_super_block *)bh->b_data; + + if (btrfs_super_bytenr(disk_super) != bytenr || + btrfs_super_magic(disk_super) != BTRFS_MAGIC) { + continue; + } + memset(&disk_super->magic, 0, + sizeof(disk_super->magic)); + set_buffer_dirty(bh); + sync_dirty_buffer(bh); + } } ret = 0; From 776e4aae5508848730a82b4695edfbd7bc802c33 Mon Sep 17 00:00:00 2001 From: ZhangZhen Date: Tue, 8 Apr 2014 19:30:19 +0800 Subject: [PATCH 31/80] btrfs: remove useless ACL check posix_acl_xattr_set() already does the check, and it's the only way to feed in an ACL from userspace. So the check here is useless, remove it. Signed-off-by: zhang zhen Signed-off-by: Chris Mason --- fs/btrfs/acl.c | 7 ------- 1 file changed, 7 deletions(-) diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c index ff9b3995d453..9a0124a95851 100644 --- a/fs/btrfs/acl.c +++ b/fs/btrfs/acl.c @@ -79,13 +79,6 @@ static int __btrfs_set_acl(struct btrfs_trans_handle *trans, const char *name; char *value = NULL; - if (acl) { - ret = posix_acl_valid(acl); - if (ret < 0) - return ret; - ret = 0; - } - switch (type) { case ACL_TYPE_ACCESS: name = POSIX_ACL_XATTR_ACCESS; From 59885b39303edc0b3284f324a1ca42cb6e7742d0 Mon Sep 17 00:00:00 2001 From: Tsutomu Itoh Date: Wed, 9 Apr 2014 09:18:04 +0900 Subject: [PATCH 32/80] Btrfs: fix possible memory leak in btrfs_create_tree() In btrfs_create_tree(), if btrfs_insert_root() fails, we should free root->commit_root. Reported-by: Alex Lyakas Signed-off-by: Tsutomu Itoh Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index cf8427a7a615..d3bd34e434f6 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1366,6 +1366,7 @@ struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans, fail: if (leaf) { btrfs_tree_unlock(leaf); + free_extent_buffer(root->commit_root); free_extent_buffer(leaf); } kfree(root); From de348ee022175401e77d7662b7ca6e231a94e3fd Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Wed, 9 Apr 2014 19:23:22 +0800 Subject: [PATCH 33/80] Btrfs: make sure there are not any read requests before stopping workers In close_ctree(), after we have stopped all workers,there maybe still some read requests(for example readahead) to submit and this *maybe* trigger an oops that user reported before: kernel BUG at fs/btrfs/async-thread.c:619! By hacking codes, i can reproduce this problem with one cpu available. We fix this potential problem by invalidating all btree inode pages before stopping all workers. Thanks to Miao for pointing out this problem. Signed-off-by: Wang Shilong Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index d3bd34e434f6..f4e708239c09 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -3657,6 +3657,11 @@ int close_ctree(struct btrfs_root *root) btrfs_free_block_groups(fs_info); + /* + * we must make sure there is not any read request to + * submit after we stopping all workers. + */ + invalidate_inode_pages2(fs_info->btree_inode->i_mapping); btrfs_stop_all_workers(fs_info); free_root_pointers(fs_info, 1); From 35045bf2fd7c030f2583dbd80a2015f427778bf1 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Wed, 9 Apr 2014 14:38:34 +0100 Subject: [PATCH 34/80] Btrfs: don't access non-existent key when csum tree is empty When the csum tree is empty, our leaf (path->nodes[0]) has a number of items equal to 0 and since btrfs_header_nritems() returns an unsigned integer (and so is our local nritems variable) the following comparison always evaluates to false: if (path->slots[0] >= nritems - 1) { As the casting rules lead to: if ((u32)0 >= (u32)4294967295) { This makes us access key at slot paths->slots[0] + 1 (1) of the empty leaf some lines below: btrfs_item_key_to_cpu(path->nodes[0], &found_key, slot); if (found_key.objectid != BTRFS_EXTENT_CSUM_OBJECTID || found_key.type != BTRFS_EXTENT_CSUM_KEY) { found_next = 1; goto insert; } So just don't access such non-existent slot and don't set found_next to 1 when the tree is empty. It's very unlikely we'll get a random key with the objectid and type values above, which is where we could go into trouble. If nritems is 0, just set found_next to 1 anyway as it will make us insert a csum item covering our whole extent (or the whole leaf) when the tree is empty. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/file-item.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c index 67751b776c7b..609d56b9fd8e 100644 --- a/fs/btrfs/file-item.c +++ b/fs/btrfs/file-item.c @@ -750,7 +750,7 @@ again: int slot = path->slots[0] + 1; /* we didn't find a csum item, insert one */ nritems = btrfs_header_nritems(path->nodes[0]); - if (path->slots[0] >= nritems - 1) { + if (!nritems || (path->slots[0] >= nritems - 1)) { ret = btrfs_next_leaf(root, path); if (ret == 1) found_next = 1; From 1f21ef0a34c37de0daf66fce7141b23b88630f29 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Wed, 9 Apr 2014 15:37:06 +0100 Subject: [PATCH 35/80] Btrfs: check if items are ordered when a leaf is marked dirty To ease finding bugs during development related to modifying btree leaves in such a way that it makes its items not sorted by key anymore. Since this is an expensive check, it's only enabled if CONFIG_BTRFS_FS_CHECK_INTEGRITY is set, which isn't meant to be enabled for regular users. Signed-off-by: Filipe David Borba Manana Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index f4e708239c09..e54f0cd5cdf6 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -3738,6 +3738,12 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf) __percpu_counter_add(&root->fs_info->dirty_metadata_bytes, buf->len, root->fs_info->dirty_metadata_batch); +#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY + if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) { + btrfs_print_leaf(root, buf); + ASSERT(0); + } +#endif } static void __btrfs_btree_balance_dirty(struct btrfs_root *root, From 995946dd2961776bb45e653810ae96643fbf4ce8 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Wed, 2 Apr 2014 19:51:06 +0800 Subject: [PATCH 36/80] Btrfs: use helpers for last_trans_log_full_commit instead of opencode Signed-off-by: Miao Xie Signed-off-by: Wang Shilong Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 5 ++--- fs/btrfs/inode.c | 2 +- fs/btrfs/tree-log.c | 40 +++++++++++++++++----------------------- fs/btrfs/tree-log.h | 16 ++++++++++++++++ 4 files changed, 36 insertions(+), 27 deletions(-) diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 7c18566fd2b4..203d9a5a4030 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -26,10 +26,9 @@ #include #include #include "hash.h" -#include "ctree.h" +#include "tree-log.h" #include "disk-io.h" #include "print-tree.h" -#include "transaction.h" #include "volumes.h" #include "raid56.h" #include "locking.h" @@ -8714,7 +8713,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, extent_root = root->fs_info->extent_root; - root->fs_info->last_trans_log_full_commit = trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); cache = btrfs_create_block_group_cache(root, chunk_offset, size); if (!cache) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index ef3bd3deca12..95ac6b4f2390 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -8329,7 +8329,7 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry, BTRFS_I(old_inode)->dir_index = 0ULL; if (unlikely(old_ino == BTRFS_FIRST_FREE_OBJECTID)) { /* force full log commit if subvolume involved. */ - root->fs_info->last_trans_log_full_commit = trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); } else { ret = btrfs_insert_inode_ref(trans, dest, new_dentry->d_name.name, diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index ffee15856130..9e1f2cd5e67a 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -20,13 +20,11 @@ #include #include #include -#include "ctree.h" -#include "transaction.h" +#include "tree-log.h" #include "disk-io.h" #include "locking.h" #include "print-tree.h" #include "backref.h" -#include "tree-log.h" #include "hash.h" /* magic values for the inode_only field in btrfs_log_inode: @@ -144,12 +142,10 @@ static int start_log_trans(struct btrfs_trans_handle *trans, mutex_lock(&root->log_mutex); if (root->log_root) { - if (ACCESS_ONCE(root->fs_info->last_trans_log_full_commit) == - trans->transid) { + if (btrfs_need_log_full_commit(root->fs_info, trans)) { ret = -EAGAIN; goto out; } - if (!root->log_start_pid) { root->log_start_pid = current->pid; clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state); @@ -2512,8 +2508,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, } /* bail out if we need to do a full commit */ - if (ACCESS_ONCE(root->fs_info->last_trans_log_full_commit) == - trans->transid) { + if (btrfs_need_log_full_commit(root->fs_info, trans)) { ret = -EAGAIN; btrfs_free_logged_extents(log, log_transid); mutex_unlock(&root->log_mutex); @@ -2534,8 +2529,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, blk_finish_plug(&plug); btrfs_abort_transaction(trans, root, ret); btrfs_free_logged_extents(log, log_transid); - ACCESS_ONCE(root->fs_info->last_trans_log_full_commit) = - trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); mutex_unlock(&root->log_mutex); goto out; } @@ -2578,8 +2572,8 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, list_del_init(&root_log_ctx.list); blk_finish_plug(&plug); - ACCESS_ONCE(root->fs_info->last_trans_log_full_commit) = - trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); + if (ret != -ENOSPC) { btrfs_abort_transaction(trans, root, ret); mutex_unlock(&log_root_tree->log_mutex); @@ -2623,8 +2617,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, * now that we've moved on to the tree of log tree roots, * check the full commit flag again */ - if (ACCESS_ONCE(root->fs_info->last_trans_log_full_commit) == - trans->transid) { + if (btrfs_need_log_full_commit(root->fs_info, trans)) { blk_finish_plug(&plug); btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); btrfs_free_logged_extents(log, log_transid); @@ -2638,8 +2631,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, EXTENT_DIRTY | EXTENT_NEW); blk_finish_plug(&plug); if (ret) { - ACCESS_ONCE(root->fs_info->last_trans_log_full_commit) = - trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); btrfs_abort_transaction(trans, root, ret); btrfs_free_logged_extents(log, log_transid); mutex_unlock(&log_root_tree->log_mutex); @@ -2668,8 +2660,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, */ ret = write_ctree_super(trans, root->fs_info->tree_root, 1); if (ret) { - ACCESS_ONCE(root->fs_info->last_trans_log_full_commit) = - trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); btrfs_abort_transaction(trans, root, ret); goto out_wake_log_root; } @@ -2887,7 +2878,7 @@ fail: out_unlock: mutex_unlock(&BTRFS_I(dir)->log_mutex); if (ret == -ENOSPC) { - root->fs_info->last_trans_log_full_commit = trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); ret = 0; } else if (ret < 0) btrfs_abort_transaction(trans, root, ret); @@ -2920,7 +2911,7 @@ int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans, dirid, &index); mutex_unlock(&BTRFS_I(inode)->log_mutex); if (ret == -ENOSPC) { - root->fs_info->last_trans_log_full_commit = trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); ret = 0; } else if (ret < 0 && ret != -ENOENT) btrfs_abort_transaction(trans, root, ret); @@ -4131,8 +4122,7 @@ static noinline int check_parent_dirs_for_sync(struct btrfs_trans_handle *trans, * make sure any commits to the log are forced * to be full commits */ - root->fs_info->last_trans_log_full_commit = - trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); ret = 1; break; } @@ -4178,6 +4168,10 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans, goto end_no_trans; } + /* + * The prev transaction commit doesn't complete, we need do + * full commit by ourselves. + */ if (root->fs_info->last_trans_log_full_commit > root->fs_info->last_trans_committed) { ret = 1; @@ -4247,7 +4241,7 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans, end_trans: dput(old_parent); if (ret < 0) { - root->fs_info->last_trans_log_full_commit = trans->transid; + btrfs_set_log_full_commit(root->fs_info, trans); ret = 1; } diff --git a/fs/btrfs/tree-log.h b/fs/btrfs/tree-log.h index 91b145fce333..7f5b41bd5373 100644 --- a/fs/btrfs/tree-log.h +++ b/fs/btrfs/tree-log.h @@ -19,6 +19,9 @@ #ifndef __TREE_LOG_ #define __TREE_LOG_ +#include "ctree.h" +#include "transaction.h" + /* return value for btrfs_log_dentry_safe that means we don't need to log it at all */ #define BTRFS_NO_LOG_SYNC 256 @@ -35,6 +38,19 @@ static inline void btrfs_init_log_ctx(struct btrfs_log_ctx *ctx) INIT_LIST_HEAD(&ctx->list); } +static inline void btrfs_set_log_full_commit(struct btrfs_fs_info *fs_info, + struct btrfs_trans_handle *trans) +{ + ACCESS_ONCE(fs_info->last_trans_log_full_commit) = trans->transid; +} + +static inline int btrfs_need_log_full_commit(struct btrfs_fs_info *fs_info, + struct btrfs_trans_handle *trans) +{ + return ACCESS_ONCE(fs_info->last_trans_log_full_commit) == + trans->transid; +} + int btrfs_sync_log(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_log_ctx *ctx); int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root); From f017f15f7c1893cb518ddae796f3a75c0c413a92 Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Thu, 13 Mar 2014 13:19:47 +0800 Subject: [PATCH 37/80] Btrfs: fix joining same transaction handle more than twice We hit something like the following function call flows: |->run_delalloc_range() |->btrfs_join_transaction() |->cow_file_range() |->btrfs_join_transaction() |->find_free_extent() |->btrfs_join_transaction() Trace infomation can be seen as: [ 7411.127040] ------------[ cut here ]------------ [ 7411.127060] WARNING: CPU: 0 PID: 11557 at fs/btrfs/transaction.c:383 start_transaction+0x561/0x580 [btrfs]() [ 7411.127079] CPU: 0 PID: 11557 Comm: kworker/u8:9 Tainted: G O 3.13.0+ #4 [ 7411.127080] Hardware name: LENOVO QiTianM4350/ , BIOS F1KT52AUS 05/24/2013 [ 7411.127085] Workqueue: writeback bdi_writeback_workfn (flush-btrfs-5) [ 7411.127092] Call Trace: [ 7411.127097] [] dump_stack+0x45/0x56 [ 7411.127101] [] warn_slowpath_common+0x7d/0xa0 [ 7411.127102] [] warn_slowpath_null+0x1a/0x20 [ 7411.127109] [] start_transaction+0x561/0x580 [btrfs] [ 7411.127115] [] btrfs_join_transaction+0x17/0x20 [btrfs] [ 7411.127120] [] find_free_extent+0xa21/0xb50 [btrfs] [ 7411.127126] [] btrfs_reserve_extent+0xa8/0x1a0 [btrfs] [ 7411.127131] [] btrfs_alloc_free_block+0xee/0x440 [btrfs] [ 7411.127137] [] ? btree_set_page_dirty+0xe/0x10 [btrfs] [ 7411.127142] [] __btrfs_cow_block+0x121/0x530 [btrfs] [ 7411.127146] [] btrfs_cow_block+0x11f/0x1c0 [btrfs] [ 7411.127151] [] btrfs_search_slot+0x1d4/0x9c0 [btrfs] [ 7411.127157] [] btrfs_lookup_file_extent+0x37/0x40 [btrfs] [ 7411.127163] [] __btrfs_drop_extents+0x16c/0xd90 [btrfs] [ 7411.127169] [] ? start_transaction+0x93/0x580 [btrfs] [ 7411.127171] [] ? kmem_cache_alloc+0x132/0x140 [ 7411.127176] [] ? btrfs_alloc_path+0x1a/0x20 [btrfs] [ 7411.127182] [] cow_file_range_inline+0x181/0x2e0 [btrfs] [ 7411.127187] [] cow_file_range+0x2ed/0x440 [btrfs] [ 7411.127194] [] ? free_extent_buffer+0x4f/0xb0 [btrfs] [ 7411.127200] [] run_delalloc_nocow+0x38f/0xa60 [btrfs] [ 7411.127207] [] ? test_range_bit+0x30/0x180 [btrfs] [ 7411.127212] [] run_delalloc_range+0x2e8/0x350 [btrfs] [ 7411.127219] [] ? find_lock_delalloc_range+0x1a9/0x1e0 [btrfs] [ 7411.127222] [] ? blk_queue_bio+0x2c1/0x330 [ 7411.127228] [] __extent_writepage+0x2f4/0x760 [btrfs] Here we fix it by avoiding joining transaction again if we have held a transaction handle when allocating chunk in find_free_extent(). Signed-off-by: Wang Shilong Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 203d9a5a4030..aff579df5f47 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6616,8 +6616,14 @@ loop: loop++; if (loop == LOOP_ALLOC_CHUNK) { struct btrfs_trans_handle *trans; + int exist = 0; + + trans = current->journal_info; + if (trans) + exist = 1; + else + trans = btrfs_join_transaction(root); - trans = btrfs_join_transaction(root); if (IS_ERR(trans)) { ret = PTR_ERR(trans); goto out; @@ -6634,7 +6640,8 @@ loop: root, ret); else ret = 0; - btrfs_end_transaction(trans, root); + if (!exist) + btrfs_end_transaction(trans, root); if (ret) goto out; } From 29cc83f69c8338ff8fd1383c9be263d4bdf52d73 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Sun, 11 May 2014 23:14:59 +0800 Subject: [PATCH 38/80] Btrfs: fix NULL pointer crash of deleting a seed device Same as normal devices, seed devices should be initialized with fs_info->dev_root as well, otherwise we'll get a NULL pointer crash. Cc: Chris Murphy Reported-by: Chris Murphy Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index ee71622088f9..018cde61cf9d 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -6146,10 +6146,14 @@ void btrfs_init_devices_late(struct btrfs_fs_info *fs_info) struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; struct btrfs_device *device; - mutex_lock(&fs_devices->device_list_mutex); - list_for_each_entry(device, &fs_devices->devices, dev_list) - device->dev_root = fs_info->dev_root; - mutex_unlock(&fs_devices->device_list_mutex); + while (fs_devices) { + mutex_lock(&fs_devices->device_list_mutex); + list_for_each_entry(device, &fs_devices->devices, dev_list) + device->dev_root = fs_info->dev_root; + mutex_unlock(&fs_devices->device_list_mutex); + + fs_devices = fs_devices->seed; + } } static void __btrfs_reset_dev_stats(struct btrfs_device *dev) From 5dca6eea91653e9949ce6eb9e9acab6277e2f2c4 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Mon, 12 May 2014 12:47:36 +0800 Subject: [PATCH 39/80] Btrfs: mark mapping with error flag to report errors to userspace According to commit 865ffef3797da2cac85b3354b5b6050dc9660978 (fs: fix fsync() error reporting), it's not stable to just check error pages because pages can be truncated or invalidated, we should also mark mapping with error flag so that a later fsync can catch the error. Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index fa31c8d2c095..a55580f4e611 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -2367,6 +2367,8 @@ int end_extent_writepage(struct page *page, int err, u64 start, u64 end) if (!uptodate) { ClearPageUptodate(page); SetPageError(page); + ret = ret < 0 ? ret : -EIO; + mapping_set_error(page->mapping, ret); } return 0; } From fcebe4562dec83b3f8d3088d77584727b09130b2 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 13 May 2014 17:30:47 -0700 Subject: [PATCH 40/80] Btrfs: rework qgroup accounting Currently qgroups account for space by intercepting delayed ref updates to fs trees. It does this by adding sequence numbers to delayed ref updates so that it can figure out how the tree looked before the update so we can adjust the counters properly. The problem with this is that it does not allow delayed refs to be merged, so if you say are defragging an extent with 5k snapshots pointing to it we will thrash the delayed ref lock because we need to go back and manually merge these things together. Instead we want to process quota changes when we know they are going to happen, like when we first allocate an extent, we free a reference for an extent, we add new references etc. This patch accomplishes this by only adding qgroup operations for real ref changes. We only modify the sequence number when we need to lookup roots for bytenrs, this reduces the amount of churn on the sequence number and allows us to merge delayed refs as we add them most of the time. This patch encompasses a bunch of architectural changes 1) qgroup ref operations: instead of tracking qgroup operations through the delayed refs we simply add new ref operations whenever we notice that we need to when we've modified the refs themselves. 2) tree mod seq: we no longer have this separation of major/minor counters. this makes the sequence number stuff much more sane and we can remove some locking that was needed to protect the counter. 3) delayed ref seq: we now read the tree mod seq number and use that as our sequence. This means each new delayed ref doesn't have it's own unique sequence number, rather whenever we go to lookup backrefs we inc the sequence number so we can make sure to keep any new operations from screwing up our world view at that given point. This allows us to merge delayed refs during runtime. With all of these changes the delayed ref stuff is a little saner and the qgroup accounting stuff no longer goes negative in some cases like it was before. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/backref.h | 4 +- fs/btrfs/ctree.c | 45 +- fs/btrfs/ctree.h | 59 +-- fs/btrfs/delayed-ref.c | 39 +- fs/btrfs/delayed-ref.h | 24 +- fs/btrfs/disk-io.c | 4 + fs/btrfs/extent-tree.c | 206 +++++---- fs/btrfs/file.c | 5 +- fs/btrfs/ioctl.c | 63 ++- fs/btrfs/qgroup.c | 937 ++++++++++++++++++++++++++++++----------- fs/btrfs/qgroup.h | 107 +++++ fs/btrfs/transaction.c | 52 +-- 12 files changed, 1055 insertions(+), 490 deletions(-) create mode 100644 fs/btrfs/qgroup.h diff --git a/fs/btrfs/backref.h b/fs/btrfs/backref.h index a910b27a8ad9..94e94429f3e9 100644 --- a/fs/btrfs/backref.h +++ b/fs/btrfs/backref.h @@ -55,8 +55,8 @@ int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info, int paths_from_inode(u64 inum, struct inode_fs_paths *ipath); int btrfs_find_all_roots(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, u64 bytenr, - u64 time_seq, struct ulist **roots); + struct btrfs_fs_info *fs_info, u64 bytenr, + u64 time_seq, struct ulist **roots); char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path, u32 name_len, unsigned long name_off, struct extent_buffer *eb_in, u64 parent, diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index 2f10e12ae94c..bbbe4f1c5086 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -356,43 +356,13 @@ static inline void tree_mod_log_write_unlock(struct btrfs_fs_info *fs_info) } /* - * Increment the upper half of tree_mod_seq, set lower half zero. - * - * Must be called with fs_info->tree_mod_seq_lock held. + * Pull a new tree mod seq number for our operation. */ -static inline u64 btrfs_inc_tree_mod_seq_major(struct btrfs_fs_info *fs_info) -{ - u64 seq = atomic64_read(&fs_info->tree_mod_seq); - seq &= 0xffffffff00000000ull; - seq += 1ull << 32; - atomic64_set(&fs_info->tree_mod_seq, seq); - return seq; -} - -/* - * Increment the lower half of tree_mod_seq. - * - * Must be called with fs_info->tree_mod_seq_lock held. The way major numbers - * are generated should not technically require a spin lock here. (Rationale: - * incrementing the minor while incrementing the major seq number is between its - * atomic64_read and atomic64_set calls doesn't duplicate sequence numbers, it - * just returns a unique sequence number as usual.) We have decided to leave - * that requirement in here and rethink it once we notice it really imposes a - * problem on some workload. - */ -static inline u64 btrfs_inc_tree_mod_seq_minor(struct btrfs_fs_info *fs_info) +static inline u64 btrfs_inc_tree_mod_seq(struct btrfs_fs_info *fs_info) { return atomic64_inc_return(&fs_info->tree_mod_seq); } -/* - * return the last minor in the previous major tree_mod_seq number - */ -u64 btrfs_tree_mod_seq_prev(u64 seq) -{ - return (seq & 0xffffffff00000000ull) - 1ull; -} - /* * This adds a new blocker to the tree mod log's blocker list if the @elem * passed does not already have a sequence number set. So when a caller expects @@ -404,19 +374,16 @@ u64 btrfs_tree_mod_seq_prev(u64 seq) u64 btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info, struct seq_list *elem) { - u64 seq; - tree_mod_log_write_lock(fs_info); spin_lock(&fs_info->tree_mod_seq_lock); if (!elem->seq) { - elem->seq = btrfs_inc_tree_mod_seq_major(fs_info); + elem->seq = btrfs_inc_tree_mod_seq(fs_info); list_add_tail(&elem->list, &fs_info->tree_mod_seq_list); } - seq = btrfs_inc_tree_mod_seq_minor(fs_info); spin_unlock(&fs_info->tree_mod_seq_lock); tree_mod_log_write_unlock(fs_info); - return seq; + return elem->seq; } void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info, @@ -489,9 +456,7 @@ __tree_mod_log_insert(struct btrfs_fs_info *fs_info, struct tree_mod_elem *tm) BUG_ON(!tm); - spin_lock(&fs_info->tree_mod_seq_lock); - tm->seq = btrfs_inc_tree_mod_seq_minor(fs_info); - spin_unlock(&fs_info->tree_mod_seq_lock); + tm->seq = btrfs_inc_tree_mod_seq(fs_info); tm_root = &fs_info->tree_mod_log; new = &tm_root->rb_node; diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index efd3bf61696d..06cc384933cc 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1648,7 +1648,10 @@ struct btrfs_fs_info { /* holds configuration and tracking. Protected by qgroup_lock */ struct rb_root qgroup_tree; + struct rb_root qgroup_op_tree; spinlock_t qgroup_lock; + spinlock_t qgroup_op_lock; + atomic_t qgroup_op_seq; /* * used to avoid frequently calling ulist_alloc()/ulist_free() @@ -3300,9 +3303,9 @@ int btrfs_reserve_extent(struct btrfs_root *root, u64 num_bytes, u64 min_alloc_size, u64 empty_size, u64 hint_byte, struct btrfs_key *ins, int is_data); int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct extent_buffer *buf, int full_backref, int for_cow); + struct extent_buffer *buf, int full_backref, int no_quota); int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct extent_buffer *buf, int full_backref, int for_cow); + struct extent_buffer *buf, int full_backref, int no_quota); int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, u64 flags, @@ -3310,7 +3313,7 @@ int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans, int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, u64 parent, u64 root_objectid, - u64 owner, u64 offset, int for_cow); + u64 owner, u64 offset, int no_quota); int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len); int btrfs_free_and_pin_reserved_extent(struct btrfs_root *root, @@ -3322,7 +3325,7 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans, int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, u64 parent, - u64 root_objectid, u64 owner, u64 offset, int for_cow); + u64 root_objectid, u64 owner, u64 offset, int no_quota); int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans, struct btrfs_root *root); @@ -3410,7 +3413,6 @@ int btrfs_init_space_info(struct btrfs_fs_info *fs_info); int btrfs_delayed_refs_qgroup_accounting(struct btrfs_trans_handle *trans, struct btrfs_fs_info *fs_info); int __get_raid_index(u64 flags); - int btrfs_start_nocow_write(struct btrfs_root *root); void btrfs_end_nocow_write(struct btrfs_root *root); /* ctree.c */ @@ -3586,7 +3588,6 @@ u64 btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info, struct seq_list *elem); void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info, struct seq_list *elem); -u64 btrfs_tree_mod_seq_prev(u64 seq); int btrfs_old_root_level(struct btrfs_root *root, u64 time_seq); /* root-item.c */ @@ -4094,52 +4095,6 @@ void btrfs_reada_detach(void *handle); int btree_readahead_hook(struct btrfs_root *root, struct extent_buffer *eb, u64 start, int err); -/* qgroup.c */ -struct qgroup_update { - struct list_head list; - struct btrfs_delayed_ref_node *node; - struct btrfs_delayed_extent_op *extent_op; -}; - -int btrfs_quota_enable(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info); -int btrfs_quota_disable(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info); -int btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info); -void btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info); -int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info); -int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, u64 src, u64 dst); -int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, u64 src, u64 dst); -int btrfs_create_qgroup(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, u64 qgroupid, - char *name); -int btrfs_remove_qgroup(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, u64 qgroupid); -int btrfs_limit_qgroup(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, u64 qgroupid, - struct btrfs_qgroup_limit *limit); -int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info); -void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info); -struct btrfs_delayed_extent_op; -int btrfs_qgroup_record_ref(struct btrfs_trans_handle *trans, - struct btrfs_delayed_ref_node *node, - struct btrfs_delayed_extent_op *extent_op); -int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, - struct btrfs_delayed_ref_node *node, - struct btrfs_delayed_extent_op *extent_op); -int btrfs_run_qgroups(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info); -int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, u64 srcid, u64 objectid, - struct btrfs_qgroup_inherit *inherit); -int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes); -void btrfs_qgroup_free(struct btrfs_root *root, u64 num_bytes); - -void assert_qgroups_uptodate(struct btrfs_trans_handle *trans); - static inline int is_fstree(u64 rootid) { if (rootid == BTRFS_FS_TREE_OBJECTID || diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c index 31299646024d..6d16bea94e1c 100644 --- a/fs/btrfs/delayed-ref.c +++ b/fs/btrfs/delayed-ref.c @@ -106,6 +106,10 @@ static int comp_entry(struct btrfs_delayed_ref_node *ref2, return -1; if (ref1->type > ref2->type) return 1; + if (ref1->no_quota > ref2->no_quota) + return 1; + if (ref1->no_quota < ref2->no_quota) + return -1; /* merging of sequenced refs is not allowed */ if (compare_seq) { if (ref1->seq < ref2->seq) @@ -635,7 +639,7 @@ add_delayed_tree_ref(struct btrfs_fs_info *fs_info, struct btrfs_delayed_ref_head *head_ref, struct btrfs_delayed_ref_node *ref, u64 bytenr, u64 num_bytes, u64 parent, u64 ref_root, int level, - int action, int for_cow) + int action, int no_quota) { struct btrfs_delayed_ref_node *existing; struct btrfs_delayed_tree_ref *full_ref; @@ -645,6 +649,8 @@ add_delayed_tree_ref(struct btrfs_fs_info *fs_info, if (action == BTRFS_ADD_DELAYED_EXTENT) action = BTRFS_ADD_DELAYED_REF; + if (is_fstree(ref_root)) + seq = atomic64_read(&fs_info->tree_mod_seq); delayed_refs = &trans->transaction->delayed_refs; /* first set the basic ref node struct up */ @@ -655,9 +661,7 @@ add_delayed_tree_ref(struct btrfs_fs_info *fs_info, ref->action = action; ref->is_head = 0; ref->in_tree = 1; - - if (need_ref_seq(for_cow, ref_root)) - seq = btrfs_get_tree_mod_seq(fs_info, &trans->delayed_ref_elem); + ref->no_quota = no_quota; ref->seq = seq; full_ref = btrfs_delayed_node_to_tree_ref(ref); @@ -697,7 +701,7 @@ add_delayed_data_ref(struct btrfs_fs_info *fs_info, struct btrfs_delayed_ref_head *head_ref, struct btrfs_delayed_ref_node *ref, u64 bytenr, u64 num_bytes, u64 parent, u64 ref_root, u64 owner, - u64 offset, int action, int for_cow) + u64 offset, int action, int no_quota) { struct btrfs_delayed_ref_node *existing; struct btrfs_delayed_data_ref *full_ref; @@ -709,6 +713,9 @@ add_delayed_data_ref(struct btrfs_fs_info *fs_info, delayed_refs = &trans->transaction->delayed_refs; + if (is_fstree(ref_root)) + seq = atomic64_read(&fs_info->tree_mod_seq); + /* first set the basic ref node struct up */ atomic_set(&ref->refs, 1); ref->bytenr = bytenr; @@ -717,9 +724,7 @@ add_delayed_data_ref(struct btrfs_fs_info *fs_info, ref->action = action; ref->is_head = 0; ref->in_tree = 1; - - if (need_ref_seq(for_cow, ref_root)) - seq = btrfs_get_tree_mod_seq(fs_info, &trans->delayed_ref_elem); + ref->no_quota = no_quota; ref->seq = seq; full_ref = btrfs_delayed_node_to_data_ref(ref); @@ -762,12 +767,15 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info, u64 bytenr, u64 num_bytes, u64 parent, u64 ref_root, int level, int action, struct btrfs_delayed_extent_op *extent_op, - int for_cow) + int no_quota) { struct btrfs_delayed_tree_ref *ref; struct btrfs_delayed_ref_head *head_ref; struct btrfs_delayed_ref_root *delayed_refs; + if (!is_fstree(ref_root) || !fs_info->quota_enabled) + no_quota = 0; + BUG_ON(extent_op && extent_op->is_data); ref = kmem_cache_alloc(btrfs_delayed_tree_ref_cachep, GFP_NOFS); if (!ref) @@ -793,10 +801,8 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info, add_delayed_tree_ref(fs_info, trans, head_ref, &ref->node, bytenr, num_bytes, parent, ref_root, level, action, - for_cow); + no_quota); spin_unlock(&delayed_refs->lock); - if (need_ref_seq(for_cow, ref_root)) - btrfs_qgroup_record_ref(trans, &ref->node, extent_op); return 0; } @@ -810,12 +816,15 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info, u64 parent, u64 ref_root, u64 owner, u64 offset, int action, struct btrfs_delayed_extent_op *extent_op, - int for_cow) + int no_quota) { struct btrfs_delayed_data_ref *ref; struct btrfs_delayed_ref_head *head_ref; struct btrfs_delayed_ref_root *delayed_refs; + if (!is_fstree(ref_root) || !fs_info->quota_enabled) + no_quota = 0; + BUG_ON(extent_op && !extent_op->is_data); ref = kmem_cache_alloc(btrfs_delayed_data_ref_cachep, GFP_NOFS); if (!ref) @@ -841,10 +850,8 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info, add_delayed_data_ref(fs_info, trans, head_ref, &ref->node, bytenr, num_bytes, parent, ref_root, owner, offset, - action, for_cow); + action, no_quota); spin_unlock(&delayed_refs->lock); - if (need_ref_seq(for_cow, ref_root)) - btrfs_qgroup_record_ref(trans, &ref->node, extent_op); return 0; } diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h index 4ba9b93022ff..a764e2340d48 100644 --- a/fs/btrfs/delayed-ref.h +++ b/fs/btrfs/delayed-ref.h @@ -52,6 +52,7 @@ struct btrfs_delayed_ref_node { unsigned int action:8; unsigned int type:8; + unsigned int no_quota:1; /* is this node still in the rbtree? */ unsigned int is_head:1; unsigned int in_tree:1; @@ -196,14 +197,14 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info, u64 bytenr, u64 num_bytes, u64 parent, u64 ref_root, int level, int action, struct btrfs_delayed_extent_op *extent_op, - int for_cow); + int no_quota); int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info, struct btrfs_trans_handle *trans, u64 bytenr, u64 num_bytes, u64 parent, u64 ref_root, u64 owner, u64 offset, int action, struct btrfs_delayed_extent_op *extent_op, - int for_cow); + int no_quota); int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info, struct btrfs_trans_handle *trans, u64 bytenr, u64 num_bytes, @@ -230,25 +231,6 @@ int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info, struct btrfs_delayed_ref_root *delayed_refs, u64 seq); -/* - * delayed refs with a ref_seq > 0 must be held back during backref walking. - * this only applies to items in one of the fs-trees. for_cow items never need - * to be held back, so they won't get a ref_seq number. - */ -static inline int need_ref_seq(int for_cow, u64 rootid) -{ - if (for_cow) - return 0; - - if (rootid == BTRFS_FS_TREE_OBJECTID) - return 1; - - if ((s64)rootid >= (s64)BTRFS_FIRST_FREE_OBJECTID) - return 1; - - return 0; -} - /* * a node might live in a head or a regular ref, this lets you * test for the proper type to use. diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index e54f0cd5cdf6..77f92a32e230 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -49,6 +49,7 @@ #include "dev-replace.h" #include "raid56.h" #include "sysfs.h" +#include "qgroup.h" #ifdef CONFIG_X86 #include @@ -2219,6 +2220,7 @@ int open_ctree(struct super_block *sb, spin_lock_init(&fs_info->free_chunk_lock); spin_lock_init(&fs_info->tree_mod_seq_lock); spin_lock_init(&fs_info->super_lock); + spin_lock_init(&fs_info->qgroup_op_lock); spin_lock_init(&fs_info->buffer_lock); rwlock_init(&fs_info->tree_mod_log_lock); mutex_init(&fs_info->reloc_mutex); @@ -2244,6 +2246,7 @@ int open_ctree(struct super_block *sb, atomic_set(&fs_info->async_submit_draining, 0); atomic_set(&fs_info->nr_async_bios, 0); atomic_set(&fs_info->defrag_running, 0); + atomic_set(&fs_info->qgroup_op_seq, 0); atomic64_set(&fs_info->tree_mod_seq, 0); fs_info->sb = sb; fs_info->max_inline = 8192 * 1024; @@ -2353,6 +2356,7 @@ int open_ctree(struct super_block *sb, spin_lock_init(&fs_info->qgroup_lock); mutex_init(&fs_info->qgroup_ioctl_lock); fs_info->qgroup_tree = RB_ROOT; + fs_info->qgroup_op_tree = RB_ROOT; INIT_LIST_HEAD(&fs_info->dirty_qgroups); fs_info->qgroup_seq = 1; fs_info->quota_enabled = 0; diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index aff579df5f47..343eb10230a1 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -35,6 +35,7 @@ #include "free-space-cache.h" #include "math.h" #include "sysfs.h" +#include "qgroup.h" #undef SCRAMBLE_DELAYED_REFS @@ -80,7 +81,8 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, u64 bytenr, u64 num_bytes, u64 parent, u64 root_objectid, u64 owner_objectid, u64 owner_offset, int refs_to_drop, - struct btrfs_delayed_extent_op *extra_op); + struct btrfs_delayed_extent_op *extra_op, + int no_quota); static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op, struct extent_buffer *leaf, struct btrfs_extent_item *ei); @@ -93,7 +95,8 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 parent, u64 root_objectid, u64 flags, struct btrfs_disk_key *key, - int level, struct btrfs_key *ins); + int level, struct btrfs_key *ins, + int no_quota); static int do_chunk_alloc(struct btrfs_trans_handle *trans, struct btrfs_root *extent_root, u64 flags, int force); @@ -1270,7 +1273,7 @@ fail: static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, - int refs_to_drop) + int refs_to_drop, int *last_ref) { struct btrfs_key key; struct btrfs_extent_data_ref *ref1 = NULL; @@ -1306,6 +1309,7 @@ static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans, if (num_refs == 0) { ret = btrfs_del_item(trans, root, path); + *last_ref = 1; } else { if (key.type == BTRFS_EXTENT_DATA_REF_KEY) btrfs_set_extent_data_ref_count(leaf, ref1, num_refs); @@ -1763,7 +1767,8 @@ void update_inline_extent_backref(struct btrfs_root *root, struct btrfs_path *path, struct btrfs_extent_inline_ref *iref, int refs_to_mod, - struct btrfs_delayed_extent_op *extent_op) + struct btrfs_delayed_extent_op *extent_op, + int *last_ref) { struct extent_buffer *leaf; struct btrfs_extent_item *ei; @@ -1807,6 +1812,7 @@ void update_inline_extent_backref(struct btrfs_root *root, else btrfs_set_shared_data_ref_count(leaf, sref, refs); } else { + *last_ref = 1; size = btrfs_extent_inline_ref_size(type); item_size = btrfs_item_size_nr(leaf, path->slots[0]); ptr = (unsigned long)iref; @@ -1838,7 +1844,7 @@ int insert_inline_extent_backref(struct btrfs_trans_handle *trans, if (ret == 0) { BUG_ON(owner < BTRFS_FIRST_FREE_OBJECTID); update_inline_extent_backref(root, path, iref, - refs_to_add, extent_op); + refs_to_add, extent_op, NULL); } else if (ret == -ENOENT) { setup_inline_extent_backref(root, path, iref, parent, root_objectid, owner, offset, @@ -1871,17 +1877,19 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct btrfs_path *path, struct btrfs_extent_inline_ref *iref, - int refs_to_drop, int is_data) + int refs_to_drop, int is_data, int *last_ref) { int ret = 0; BUG_ON(!is_data && refs_to_drop != 1); if (iref) { update_inline_extent_backref(root, path, iref, - -refs_to_drop, NULL); + -refs_to_drop, NULL, last_ref); } else if (is_data) { - ret = remove_extent_data_ref(trans, root, path, refs_to_drop); + ret = remove_extent_data_ref(trans, root, path, refs_to_drop, + last_ref); } else { + *last_ref = 1; ret = btrfs_del_item(trans, root, path); } return ret; @@ -1945,7 +1953,8 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, u64 parent, - u64 root_objectid, u64 owner, u64 offset, int for_cow) + u64 root_objectid, u64 owner, u64 offset, + int no_quota) { int ret; struct btrfs_fs_info *fs_info = root->fs_info; @@ -1957,12 +1966,12 @@ int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans, ret = btrfs_add_delayed_tree_ref(fs_info, trans, bytenr, num_bytes, parent, root_objectid, (int)owner, - BTRFS_ADD_DELAYED_REF, NULL, for_cow); + BTRFS_ADD_DELAYED_REF, NULL, no_quota); } else { ret = btrfs_add_delayed_data_ref(fs_info, trans, bytenr, num_bytes, parent, root_objectid, owner, offset, - BTRFS_ADD_DELAYED_REF, NULL, for_cow); + BTRFS_ADD_DELAYED_REF, NULL, no_quota); } return ret; } @@ -1972,31 +1981,64 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans, u64 bytenr, u64 num_bytes, u64 parent, u64 root_objectid, u64 owner, u64 offset, int refs_to_add, + int no_quota, struct btrfs_delayed_extent_op *extent_op) { + struct btrfs_fs_info *fs_info = root->fs_info; struct btrfs_path *path; struct extent_buffer *leaf; struct btrfs_extent_item *item; + struct btrfs_key key; u64 refs; int ret; + enum btrfs_qgroup_operation_type type = BTRFS_QGROUP_OPER_ADD_EXCL; path = btrfs_alloc_path(); if (!path) return -ENOMEM; + if (!is_fstree(root_objectid) || !root->fs_info->quota_enabled) + no_quota = 1; + path->reada = 1; path->leave_spinning = 1; /* this will setup the path even if it fails to insert the back ref */ - ret = insert_inline_extent_backref(trans, root->fs_info->extent_root, - path, bytenr, num_bytes, parent, + ret = insert_inline_extent_backref(trans, fs_info->extent_root, path, + bytenr, num_bytes, parent, root_objectid, owner, offset, refs_to_add, extent_op); - if (ret != -EAGAIN) + if ((ret < 0 && ret != -EAGAIN) || (!ret && no_quota)) goto out; + /* + * Ok we were able to insert an inline extent and it appears to be a new + * reference, deal with the qgroup accounting. + */ + if (!ret && !no_quota) { + ASSERT(root->fs_info->quota_enabled); + leaf = path->nodes[0]; + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); + item = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_extent_item); + if (btrfs_extent_refs(leaf, item) > (u64)refs_to_add) + type = BTRFS_QGROUP_OPER_ADD_SHARED; + btrfs_release_path(path); + ret = btrfs_qgroup_record_ref(trans, fs_info, root_objectid, + bytenr, num_bytes, type, 0); + goto out; + } + + /* + * Ok we had -EAGAIN which means we didn't have space to insert and + * inline extent ref, so just update the reference count and add a + * normal backref. + */ leaf = path->nodes[0]; + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item); refs = btrfs_extent_refs(leaf, item); + if (refs) + type = BTRFS_QGROUP_OPER_ADD_SHARED; btrfs_set_extent_refs(leaf, item, refs + refs_to_add); if (extent_op) __run_delayed_extent_op(extent_op, leaf, item); @@ -2004,9 +2046,15 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans, btrfs_mark_buffer_dirty(leaf); btrfs_release_path(path); + if (!no_quota) { + ret = btrfs_qgroup_record_ref(trans, fs_info, root_objectid, + bytenr, num_bytes, type, 0); + if (ret) + goto out; + } + path->reada = 1; path->leave_spinning = 1; - /* now insert the actual backref */ ret = insert_extent_backref(trans, root->fs_info->extent_root, path, bytenr, parent, root_objectid, @@ -2040,8 +2088,7 @@ static int run_delayed_data_ref(struct btrfs_trans_handle *trans, if (node->type == BTRFS_SHARED_DATA_REF_KEY) parent = ref->parent; - else - ref_root = ref->root; + ref_root = ref->root; if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) { if (extent_op) @@ -2055,13 +2102,13 @@ static int run_delayed_data_ref(struct btrfs_trans_handle *trans, node->num_bytes, parent, ref_root, ref->objectid, ref->offset, node->ref_mod, - extent_op); + node->no_quota, extent_op); } else if (node->action == BTRFS_DROP_DELAYED_REF) { ret = __btrfs_free_extent(trans, root, node->bytenr, node->num_bytes, parent, ref_root, ref->objectid, ref->offset, node->ref_mod, - extent_op); + extent_op, node->no_quota); } else { BUG(); } @@ -2198,8 +2245,7 @@ static int run_delayed_tree_ref(struct btrfs_trans_handle *trans, if (node->type == BTRFS_SHARED_BLOCK_REF_KEY) parent = ref->parent; - else - ref_root = ref->root; + ref_root = ref->root; ins.objectid = node->bytenr; if (skinny_metadata) { @@ -2217,15 +2263,18 @@ static int run_delayed_tree_ref(struct btrfs_trans_handle *trans, parent, ref_root, extent_op->flags_to_set, &extent_op->key, - ref->level, &ins); + ref->level, &ins, + node->no_quota); } else if (node->action == BTRFS_ADD_DELAYED_REF) { ret = __btrfs_inc_extent_ref(trans, root, node->bytenr, node->num_bytes, parent, ref_root, - ref->level, 0, 1, extent_op); + ref->level, 0, 1, node->no_quota, + extent_op); } else if (node->action == BTRFS_DROP_DELAYED_REF) { ret = __btrfs_free_extent(trans, root, node->bytenr, node->num_bytes, parent, ref_root, - ref->level, 0, 1, extent_op); + ref->level, 0, 1, extent_op, + node->no_quota); } else { BUG(); } @@ -2573,42 +2622,6 @@ static u64 find_middle(struct rb_root *root) } #endif -int btrfs_delayed_refs_qgroup_accounting(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info) -{ - struct qgroup_update *qgroup_update; - int ret = 0; - - if (list_empty(&trans->qgroup_ref_list) != - !trans->delayed_ref_elem.seq) { - /* list without seq or seq without list */ - btrfs_err(fs_info, - "qgroup accounting update error, list is%s empty, seq is %#x.%x", - list_empty(&trans->qgroup_ref_list) ? "" : " not", - (u32)(trans->delayed_ref_elem.seq >> 32), - (u32)trans->delayed_ref_elem.seq); - BUG(); - } - - if (!trans->delayed_ref_elem.seq) - return 0; - - while (!list_empty(&trans->qgroup_ref_list)) { - qgroup_update = list_first_entry(&trans->qgroup_ref_list, - struct qgroup_update, list); - list_del(&qgroup_update->list); - if (!ret) - ret = btrfs_qgroup_account_ref( - trans, fs_info, qgroup_update->node, - qgroup_update->extent_op); - kfree(qgroup_update); - } - - btrfs_put_tree_mod_seq(fs_info, &trans->delayed_ref_elem); - - return ret; -} - static inline u64 heads_to_leaves(struct btrfs_root *root, u64 heads) { u64 num_bytes; @@ -2697,8 +2710,6 @@ int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans, if (root == root->fs_info->extent_root) root = root->fs_info->tree_root; - btrfs_delayed_refs_qgroup_accounting(trans, root->fs_info); - delayed_refs = &trans->transaction->delayed_refs; if (count == 0) { count = atomic_read(&delayed_refs->num_entries) * 2; @@ -2757,6 +2768,9 @@ again: goto again; } out: + ret = btrfs_delayed_qgroup_accounting(trans, root->fs_info); + if (ret) + return ret; assert_qgroups_uptodate(trans); return 0; } @@ -2963,7 +2977,7 @@ out: static int __btrfs_mod_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf, - int full_backref, int inc, int for_cow) + int full_backref, int inc, int no_quota) { u64 bytenr; u64 num_bytes; @@ -3013,7 +3027,7 @@ static int __btrfs_mod_ref(struct btrfs_trans_handle *trans, key.offset -= btrfs_file_extent_offset(buf, fi); ret = process_func(trans, root, bytenr, num_bytes, parent, ref_root, key.objectid, - key.offset, for_cow); + key.offset, no_quota); if (ret) goto fail; } else { @@ -3021,7 +3035,7 @@ static int __btrfs_mod_ref(struct btrfs_trans_handle *trans, num_bytes = btrfs_level_size(root, level - 1); ret = process_func(trans, root, bytenr, num_bytes, parent, ref_root, level - 1, 0, - for_cow); + no_quota); if (ret) goto fail; } @@ -3032,15 +3046,15 @@ fail: } int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct extent_buffer *buf, int full_backref, int for_cow) + struct extent_buffer *buf, int full_backref, int no_quota) { - return __btrfs_mod_ref(trans, root, buf, full_backref, 1, for_cow); + return __btrfs_mod_ref(trans, root, buf, full_backref, 1, no_quota); } int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, - struct extent_buffer *buf, int full_backref, int for_cow) + struct extent_buffer *buf, int full_backref, int no_quota) { - return __btrfs_mod_ref(trans, root, buf, full_backref, 0, for_cow); + return __btrfs_mod_ref(trans, root, buf, full_backref, 0, no_quota); } static int write_one_cache_group(struct btrfs_trans_handle *trans, @@ -5723,7 +5737,8 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, u64 bytenr, u64 num_bytes, u64 parent, u64 root_objectid, u64 owner_objectid, u64 owner_offset, int refs_to_drop, - struct btrfs_delayed_extent_op *extent_op) + struct btrfs_delayed_extent_op *extent_op, + int no_quota) { struct btrfs_key key; struct btrfs_path *path; @@ -5739,9 +5754,14 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, int num_to_del = 1; u32 item_size; u64 refs; + int last_ref = 0; + enum btrfs_qgroup_operation_type type = BTRFS_QGROUP_OPER_SUB_EXCL; bool skinny_metadata = btrfs_fs_incompat(root->fs_info, SKINNY_METADATA); + if (!info->quota_enabled || !is_fstree(root_objectid)) + no_quota = 1; + path = btrfs_alloc_path(); if (!path) return -ENOMEM; @@ -5789,7 +5809,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, BUG_ON(iref); ret = remove_extent_backref(trans, extent_root, path, NULL, refs_to_drop, - is_data); + is_data, &last_ref); if (ret) { btrfs_abort_transaction(trans, extent_root, ret); goto out; @@ -5916,6 +5936,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, refs -= refs_to_drop; if (refs > 0) { + type = BTRFS_QGROUP_OPER_SUB_SHARED; if (extent_op) __run_delayed_extent_op(extent_op, leaf, ei); /* @@ -5931,7 +5952,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, if (found_extent) { ret = remove_extent_backref(trans, extent_root, path, iref, refs_to_drop, - is_data); + is_data, &last_ref); if (ret) { btrfs_abort_transaction(trans, extent_root, ret); goto out; @@ -5952,6 +5973,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, } } + last_ref = 1; ret = btrfs_del_items(trans, extent_root, path, path->slots[0], num_to_del); if (ret) { @@ -5974,6 +5996,20 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, goto out; } } + btrfs_release_path(path); + + /* Deal with the quota accounting */ + if (!ret && last_ref && !no_quota) { + int mod_seq = 0; + + if (owner_objectid >= BTRFS_FIRST_FREE_OBJECTID && + type == BTRFS_QGROUP_OPER_SUB_SHARED) + mod_seq = 1; + + ret = btrfs_qgroup_record_ref(trans, info, root_objectid, + bytenr, num_bytes, type, + mod_seq); + } out: btrfs_free_path(path); return ret; @@ -6110,7 +6146,7 @@ out: /* Can return -ENOMEM */ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, u64 parent, u64 root_objectid, - u64 owner, u64 offset, int for_cow) + u64 owner, u64 offset, int no_quota) { int ret; struct btrfs_fs_info *fs_info = root->fs_info; @@ -6130,13 +6166,13 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, ret = btrfs_add_delayed_tree_ref(fs_info, trans, bytenr, num_bytes, parent, root_objectid, (int)owner, - BTRFS_DROP_DELAYED_REF, NULL, for_cow); + BTRFS_DROP_DELAYED_REF, NULL, no_quota); } else { ret = btrfs_add_delayed_data_ref(fs_info, trans, bytenr, num_bytes, parent, root_objectid, owner, offset, BTRFS_DROP_DELAYED_REF, - NULL, for_cow); + NULL, no_quota); } return ret; } @@ -6842,6 +6878,13 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans, btrfs_mark_buffer_dirty(path->nodes[0]); btrfs_free_path(path); + /* Always set parent to 0 here since its exclusive anyway. */ + ret = btrfs_qgroup_record_ref(trans, fs_info, root_objectid, + ins->objectid, ins->offset, + BTRFS_QGROUP_OPER_ADD_EXCL, 0); + if (ret) + return ret; + ret = update_block_group(root, ins->objectid, ins->offset, 1); if (ret) { /* -ENOENT, logic error */ btrfs_err(fs_info, "update block group failed for %llu %llu", @@ -6856,7 +6899,8 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 parent, u64 root_objectid, u64 flags, struct btrfs_disk_key *key, - int level, struct btrfs_key *ins) + int level, struct btrfs_key *ins, + int no_quota) { int ret; struct btrfs_fs_info *fs_info = root->fs_info; @@ -6866,6 +6910,7 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans, struct btrfs_path *path; struct extent_buffer *leaf; u32 size = sizeof(*extent_item) + sizeof(*iref); + u64 num_bytes = ins->offset; bool skinny_metadata = btrfs_fs_incompat(root->fs_info, SKINNY_METADATA); @@ -6899,6 +6944,7 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans, if (skinny_metadata) { iref = (struct btrfs_extent_inline_ref *)(extent_item + 1); + num_bytes = root->leafsize; } else { block_info = (struct btrfs_tree_block_info *)(extent_item + 1); btrfs_set_tree_block_key(leaf, block_info, key); @@ -6920,6 +6966,14 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans, btrfs_mark_buffer_dirty(leaf); btrfs_free_path(path); + if (!no_quota) { + ret = btrfs_qgroup_record_ref(trans, fs_info, root_objectid, + ins->objectid, num_bytes, + BTRFS_QGROUP_OPER_ADD_EXCL, 0); + if (ret) + return ret; + } + ret = update_block_group(root, ins->objectid, root->leafsize, 1); if (ret) { /* -ENOENT, logic error */ btrfs_err(fs_info, "update block group failed for %llu %llu", diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 5c6947dbc948..8accf94ef220 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -40,6 +40,7 @@ #include "tree-log.h" #include "locking.h" #include "volumes.h" +#include "qgroup.h" static struct kmem_cache *btrfs_inode_defrag_cachep; /* @@ -849,7 +850,7 @@ next_slot: disk_bytenr, num_bytes, 0, root->root_key.objectid, new_key.objectid, - start - extent_offset, 0); + start - extent_offset, 1); BUG_ON(ret); /* -ENOMEM */ } key.offset = start; @@ -1206,7 +1207,7 @@ again: ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0, root->root_key.objectid, - ino, orig_offset, 0); + ino, orig_offset, 1); BUG_ON(ret); /* -ENOMEM */ if (split == start) { diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 242a37cd26b2..a21a4ac537b7 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -58,6 +58,7 @@ #include "dev-replace.h" #include "props.h" #include "sysfs.h" +#include "qgroup.h" #ifdef CONFIG_64BIT /* If we have a 32-bit userspace and 64-bit kernel, then the UAPI @@ -2941,6 +2942,41 @@ out: return ret; } +/* Helper to check and see if this root currently has a ref on the given disk + * bytenr. If it does then we need to update the quota for this root. This + * doesn't do anything if quotas aren't enabled. + */ +static int check_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, + u64 disko) +{ + struct seq_list tree_mod_seq_elem = {}; + struct ulist *roots; + struct ulist_iterator uiter; + struct ulist_node *root_node = NULL; + int ret; + + if (!root->fs_info->quota_enabled) + return 1; + + btrfs_get_tree_mod_seq(root->fs_info, &tree_mod_seq_elem); + ret = btrfs_find_all_roots(trans, root->fs_info, disko, + tree_mod_seq_elem.seq, &roots); + if (ret < 0) + goto out; + ret = 0; + ULIST_ITER_INIT(&uiter); + while ((root_node = ulist_next(roots, &uiter))) { + if (root_node->val == root->objectid) { + ret = 1; + break; + } + } + ulist_free(roots); +out: + btrfs_put_tree_mod_seq(root->fs_info, &tree_mod_seq_elem); + return ret; +} + /** * btrfs_clone() - clone a range from inode file to another * @@ -2964,7 +3000,9 @@ static int btrfs_clone(struct inode *src, struct inode *inode, u32 nritems; int slot; int ret; + int no_quota; u64 len = olen_aligned; + u64 last_disko = 0; ret = -ENOMEM; buf = vmalloc(btrfs_level_size(root, 0)); @@ -2996,6 +3034,7 @@ static int btrfs_clone(struct inode *src, struct inode *inode, nritems = btrfs_header_nritems(path->nodes[0]); process_slot: + no_quota = 1; if (path->slots[0] >= nritems) { ret = btrfs_next_leaf(BTRFS_I(src)->root, path); if (ret < 0) @@ -3128,6 +3167,28 @@ process_slot: datao); btrfs_set_file_extent_num_bytes(leaf, extent, datal); + + /* + * We need to look up the roots that point at + * this bytenr and see if the new root does. If + * it does not we need to make sure we update + * quotas appropriately. + */ + if (disko && root != BTRFS_I(src)->root && + disko != last_disko) { + no_quota = check_ref(trans, root, + disko); + if (no_quota < 0) { + btrfs_abort_transaction(trans, + root, + ret); + btrfs_end_transaction(trans, + root); + ret = no_quota; + goto out; + } + } + if (disko) { inode_add_bytes(inode, datal); ret = btrfs_inc_extent_ref(trans, root, @@ -3135,7 +3196,7 @@ process_slot: root->root_key.objectid, btrfs_ino(inode), new_key.offset - datao, - 0); + no_quota); if (ret) { btrfs_abort_transaction(trans, root, diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c index 2cf905877aaf..09b8cc83965c 100644 --- a/fs/btrfs/qgroup.c +++ b/fs/btrfs/qgroup.c @@ -32,6 +32,7 @@ #include "ulist.h" #include "backref.h" #include "extent_io.h" +#include "qgroup.h" /* TODO XXX FIXME * - subvol delete -> delete when ref goes to 0? delete limits also? @@ -84,8 +85,8 @@ struct btrfs_qgroup { /* * temp variables for accounting operations */ - u64 tag; - u64 refcnt; + u64 old_refcnt; + u64 new_refcnt; }; /* @@ -98,6 +99,9 @@ struct btrfs_qgroup_list { struct btrfs_qgroup *member; }; +#define ptr_to_u64(x) ((u64)(uintptr_t)x) +#define u64_to_ptr(x) ((struct btrfs_qgroup *)(uintptr_t)x) + static int qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid, int init_flags); @@ -1174,33 +1178,198 @@ out: mutex_unlock(&fs_info->qgroup_ioctl_lock); return ret; } +static int comp_oper(struct btrfs_qgroup_operation *oper1, + struct btrfs_qgroup_operation *oper2) +{ + if (oper1->bytenr < oper2->bytenr) + return -1; + if (oper1->bytenr > oper2->bytenr) + return 1; + if (oper1->seq < oper2->seq) + return -1; + if (oper1->seq > oper2->seq) + return -1; + if (oper1->ref_root < oper2->ref_root) + return -1; + if (oper1->ref_root > oper2->ref_root) + return 1; + if (oper1->type < oper2->type) + return -1; + if (oper1->type > oper2->type) + return 1; + return 0; +} + +static int insert_qgroup_oper(struct btrfs_fs_info *fs_info, + struct btrfs_qgroup_operation *oper) +{ + struct rb_node **p; + struct rb_node *parent = NULL; + struct btrfs_qgroup_operation *cur; + int cmp; + + spin_lock(&fs_info->qgroup_op_lock); + p = &fs_info->qgroup_op_tree.rb_node; + while (*p) { + parent = *p; + cur = rb_entry(parent, struct btrfs_qgroup_operation, n); + cmp = comp_oper(cur, oper); + if (cmp < 0) { + p = &(*p)->rb_right; + } else if (cmp) { + p = &(*p)->rb_left; + } else { + spin_unlock(&fs_info->qgroup_op_lock); + return -EEXIST; + } + } + rb_link_node(&oper->n, parent, p); + rb_insert_color(&oper->n, &fs_info->qgroup_op_tree); + spin_unlock(&fs_info->qgroup_op_lock); + return 0; +} /* - * btrfs_qgroup_record_ref is called when the ref is added or deleted. it puts - * the modification into a list that's later used by btrfs_end_transaction to - * pass the recorded modifications on to btrfs_qgroup_account_ref. + * Record a quota operation for processing later on. + * @trans: the transaction we are adding the delayed op to. + * @fs_info: the fs_info for this fs. + * @ref_root: the root of the reference we are acting on, + * @bytenr: the bytenr we are acting on. + * @num_bytes: the number of bytes in the reference. + * @type: the type of operation this is. + * @mod_seq: do we need to get a sequence number for looking up roots. + * + * We just add it to our trans qgroup_ref_list and carry on and process these + * operations in order at some later point. If the reference root isn't a fs + * root then we don't bother with doing anything. + * + * MUST BE HOLDING THE REF LOCK. */ int btrfs_qgroup_record_ref(struct btrfs_trans_handle *trans, - struct btrfs_delayed_ref_node *node, - struct btrfs_delayed_extent_op *extent_op) + struct btrfs_fs_info *fs_info, u64 ref_root, + u64 bytenr, u64 num_bytes, + enum btrfs_qgroup_operation_type type, int mod_seq) { - struct qgroup_update *u; + struct btrfs_qgroup_operation *oper; + int ret; - BUG_ON(!trans->delayed_ref_elem.seq); - u = kmalloc(sizeof(*u), GFP_NOFS); - if (!u) + if (!is_fstree(ref_root) || !fs_info->quota_enabled) + return 0; + + oper = kmalloc(sizeof(*oper), GFP_NOFS); + if (!oper) return -ENOMEM; - u->node = node; - u->extent_op = extent_op; - list_add_tail(&u->list, &trans->qgroup_ref_list); + oper->ref_root = ref_root; + oper->bytenr = bytenr; + oper->num_bytes = num_bytes; + oper->type = type; + oper->seq = atomic_inc_return(&fs_info->qgroup_op_seq); + INIT_LIST_HEAD(&oper->elem.list); + oper->elem.seq = 0; + ret = insert_qgroup_oper(fs_info, oper); + if (ret) { + /* Shouldn't happen so have an assert for developers */ + ASSERT(0); + kfree(oper); + return ret; + } + list_add_tail(&oper->list, &trans->qgroup_ref_list); + + if (mod_seq) + btrfs_get_tree_mod_seq(fs_info, &oper->elem); return 0; } -static int qgroup_account_ref_step1(struct btrfs_fs_info *fs_info, - struct ulist *roots, struct ulist *tmp, - u64 seq) +/* + * The easy accounting, if we are adding/removing the only ref for an extent + * then this qgroup and all of the parent qgroups get their refrence and + * exclusive counts adjusted. + */ +static int qgroup_excl_accounting(struct btrfs_fs_info *fs_info, + struct btrfs_qgroup_operation *oper) +{ + struct btrfs_qgroup *qgroup; + struct ulist *tmp; + struct btrfs_qgroup_list *glist; + struct ulist_node *unode; + struct ulist_iterator uiter; + int sign = 0; + int ret = 0; + + tmp = ulist_alloc(GFP_NOFS); + if (!tmp) + return -ENOMEM; + + spin_lock(&fs_info->qgroup_lock); + if (!fs_info->quota_root) + goto out; + qgroup = find_qgroup_rb(fs_info, oper->ref_root); + if (!qgroup) + goto out; + switch (oper->type) { + case BTRFS_QGROUP_OPER_ADD_EXCL: + sign = 1; + break; + case BTRFS_QGROUP_OPER_SUB_EXCL: + sign = -1; + break; + default: + ASSERT(0); + } + qgroup->rfer += sign * oper->num_bytes; + qgroup->rfer_cmpr += sign * oper->num_bytes; + + WARN_ON(sign < 0 && qgroup->excl < oper->num_bytes); + qgroup->excl += sign * oper->num_bytes; + qgroup->excl_cmpr += sign * oper->num_bytes; + + qgroup_dirty(fs_info, qgroup); + + /* Get all of the parent groups that contain this qgroup */ + list_for_each_entry(glist, &qgroup->groups, next_group) { + ret = ulist_add(tmp, glist->group->qgroupid, + ptr_to_u64(glist->group), GFP_ATOMIC); + if (ret < 0) + goto out; + } + + /* Iterate all of the parents and adjust their reference counts */ + ULIST_ITER_INIT(&uiter); + while ((unode = ulist_next(tmp, &uiter))) { + qgroup = u64_to_ptr(unode->aux); + qgroup->rfer += sign * oper->num_bytes; + qgroup->rfer_cmpr += sign * oper->num_bytes; + qgroup->excl += sign * oper->num_bytes; + if (sign < 0) + WARN_ON(qgroup->excl < oper->num_bytes); + qgroup->excl_cmpr += sign * oper->num_bytes; + qgroup_dirty(fs_info, qgroup); + + /* Add any parents of the parents */ + list_for_each_entry(glist, &qgroup->groups, next_group) { + ret = ulist_add(tmp, glist->group->qgroupid, + ptr_to_u64(glist->group), GFP_ATOMIC); + if (ret < 0) + goto out; + } + } + ret = 0; +out: + spin_unlock(&fs_info->qgroup_lock); + ulist_free(tmp); + return ret; +} + +/* + * Walk all of the roots that pointed to our bytenr and adjust their refcnts as + * properly. + */ +static int qgroup_calc_old_refcnt(struct btrfs_fs_info *fs_info, + u64 root_to_skip, struct ulist *tmp, + struct ulist *roots, struct ulist *qgroups, + u64 seq, int *old_roots, int rescan) { struct ulist_node *unode; struct ulist_iterator uiter; @@ -1211,129 +1380,482 @@ static int qgroup_account_ref_step1(struct btrfs_fs_info *fs_info, ULIST_ITER_INIT(&uiter); while ((unode = ulist_next(roots, &uiter))) { + /* We don't count our current root here */ + if (unode->val == root_to_skip) + continue; qg = find_qgroup_rb(fs_info, unode->val); if (!qg) continue; + /* + * We could have a pending removal of this same ref so we may + * not have actually found our ref root when doing + * btrfs_find_all_roots, so we need to keep track of how many + * old roots we find in case we removed ours and added a + * different one at the same time. I don't think this could + * happen in practice but that sort of thinking leads to pain + * and suffering and to the dark side. + */ + (*old_roots)++; ulist_reinit(tmp); - /* XXX id not needed */ - ret = ulist_add(tmp, qg->qgroupid, - (u64)(uintptr_t)qg, GFP_ATOMIC); + ret = ulist_add(qgroups, qg->qgroupid, ptr_to_u64(qg), + GFP_ATOMIC); + if (ret < 0) + return ret; + ret = ulist_add(tmp, qg->qgroupid, ptr_to_u64(qg), GFP_ATOMIC); if (ret < 0) return ret; ULIST_ITER_INIT(&tmp_uiter); while ((tmp_unode = ulist_next(tmp, &tmp_uiter))) { struct btrfs_qgroup_list *glist; - qg = (struct btrfs_qgroup *)(uintptr_t)tmp_unode->aux; - if (qg->refcnt < seq) - qg->refcnt = seq + 1; + qg = u64_to_ptr(tmp_unode->aux); + /* + * We use this sequence number to keep from having to + * run the whole list and 0 out the refcnt every time. + * We basically use sequnce as the known 0 count and + * then add 1 everytime we see a qgroup. This is how we + * get how many of the roots actually point up to the + * upper level qgroups in order to determine exclusive + * counts. + * + * For rescan we want to set old_refcnt to seq so our + * exclusive calculations end up correct. + */ + if (rescan) + qg->old_refcnt = seq; + else if (qg->old_refcnt < seq) + qg->old_refcnt = seq + 1; else - ++qg->refcnt; + qg->old_refcnt++; + if (qg->new_refcnt < seq) + qg->new_refcnt = seq + 1; + else + qg->new_refcnt++; list_for_each_entry(glist, &qg->groups, next_group) { + ret = ulist_add(qgroups, glist->group->qgroupid, + ptr_to_u64(glist->group), + GFP_ATOMIC); + if (ret < 0) + return ret; ret = ulist_add(tmp, glist->group->qgroupid, - (u64)(uintptr_t)glist->group, + ptr_to_u64(glist->group), GFP_ATOMIC); if (ret < 0) return ret; } } } - return 0; } -static int qgroup_account_ref_step2(struct btrfs_fs_info *fs_info, - struct ulist *roots, struct ulist *tmp, - u64 seq, int sgn, u64 num_bytes, - struct btrfs_qgroup *qgroup) +/* + * We need to walk forward in our operation tree and account for any roots that + * were deleted after we made this operation. + */ +static int qgroup_account_deleted_refs(struct btrfs_fs_info *fs_info, + struct btrfs_qgroup_operation *oper, + struct ulist *tmp, + struct ulist *qgroups, u64 seq, + int *old_roots) { struct ulist_node *unode; struct ulist_iterator uiter; struct btrfs_qgroup *qg; - struct btrfs_qgroup_list *glist; + struct btrfs_qgroup_operation *tmp_oper; + struct rb_node *n; int ret; ulist_reinit(tmp); - ret = ulist_add(tmp, qgroup->qgroupid, (uintptr_t)qgroup, GFP_ATOMIC); - if (ret < 0) - return ret; + /* + * We only walk forward in the tree since we're only interested in + * removals that happened _after_ our operation. + */ + spin_lock(&fs_info->qgroup_op_lock); + n = rb_next(&oper->n); + spin_unlock(&fs_info->qgroup_op_lock); + if (!n) + return 0; + tmp_oper = rb_entry(n, struct btrfs_qgroup_operation, n); + while (tmp_oper->bytenr == oper->bytenr) { + /* + * If it's not a removal we don't care, additions work out + * properly with our refcnt tracking. + */ + if (tmp_oper->type != BTRFS_QGROUP_OPER_SUB_SHARED && + tmp_oper->type != BTRFS_QGROUP_OPER_SUB_EXCL) + goto next; + qg = find_qgroup_rb(fs_info, tmp_oper->ref_root); + if (!qg) + goto next; + ret = ulist_add(qgroups, qg->qgroupid, ptr_to_u64(qg), + GFP_ATOMIC); + if (ret) { + if (ret < 0) + return ret; + /* + * We only want to increase old_roots if this qgroup is + * not already in the list of qgroups. If it is already + * there then that means it must have been re-added or + * the delete will be discarded because we had an + * existing ref that we haven't looked up yet. In this + * case we don't want to increase old_roots. So if ret + * == 1 then we know that this is the first time we've + * seen this qgroup and we can bump the old_roots. + */ + (*old_roots)++; + ret = ulist_add(tmp, qg->qgroupid, ptr_to_u64(qg), + GFP_ATOMIC); + if (ret < 0) + return ret; + } +next: + spin_lock(&fs_info->qgroup_op_lock); + n = rb_next(&tmp_oper->n); + spin_unlock(&fs_info->qgroup_op_lock); + if (!n) + break; + tmp_oper = rb_entry(n, struct btrfs_qgroup_operation, n); + } + + /* Ok now process the qgroups we found */ ULIST_ITER_INIT(&uiter); while ((unode = ulist_next(tmp, &uiter))) { - qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux; - if (qg->refcnt < seq) { - /* not visited by step 1 */ - qg->rfer += sgn * num_bytes; - qg->rfer_cmpr += sgn * num_bytes; - if (roots->nnodes == 0) { - qg->excl += sgn * num_bytes; - qg->excl_cmpr += sgn * num_bytes; - } - qgroup_dirty(fs_info, qg); - } - WARN_ON(qg->tag >= seq); - qg->tag = seq; + struct btrfs_qgroup_list *glist; + qg = u64_to_ptr(unode->aux); + if (qg->old_refcnt < seq) + qg->old_refcnt = seq + 1; + else + qg->old_refcnt++; + if (qg->new_refcnt < seq) + qg->new_refcnt = seq + 1; + else + qg->new_refcnt++; list_for_each_entry(glist, &qg->groups, next_group) { + ret = ulist_add(qgroups, glist->group->qgroupid, + ptr_to_u64(glist->group), GFP_ATOMIC); + if (ret < 0) + return ret; ret = ulist_add(tmp, glist->group->qgroupid, - (uintptr_t)glist->group, GFP_ATOMIC); + ptr_to_u64(glist->group), GFP_ATOMIC); if (ret < 0) return ret; } } - return 0; } -static int qgroup_account_ref_step3(struct btrfs_fs_info *fs_info, - struct ulist *roots, struct ulist *tmp, - u64 seq, int sgn, u64 num_bytes) +/* Add refcnt for the newly added reference. */ +static int qgroup_calc_new_refcnt(struct btrfs_fs_info *fs_info, + struct btrfs_qgroup_operation *oper, + struct btrfs_qgroup *qgroup, + struct ulist *tmp, struct ulist *qgroups, + u64 seq) { struct ulist_node *unode; struct ulist_iterator uiter; struct btrfs_qgroup *qg; - struct ulist_node *tmp_unode; - struct ulist_iterator tmp_uiter; int ret; + ulist_reinit(tmp); + ret = ulist_add(qgroups, qgroup->qgroupid, ptr_to_u64(qgroup), + GFP_ATOMIC); + if (ret < 0) + return ret; + ret = ulist_add(tmp, qgroup->qgroupid, ptr_to_u64(qgroup), + GFP_ATOMIC); + if (ret < 0) + return ret; + ULIST_ITER_INIT(&uiter); + while ((unode = ulist_next(tmp, &uiter))) { + struct btrfs_qgroup_list *glist; + + qg = u64_to_ptr(unode->aux); + if (oper->type == BTRFS_QGROUP_OPER_ADD_SHARED) { + if (qg->new_refcnt < seq) + qg->new_refcnt = seq + 1; + else + qg->new_refcnt++; + } else { + if (qg->old_refcnt < seq) + qg->old_refcnt = seq + 1; + else + qg->old_refcnt++; + } + list_for_each_entry(glist, &qg->groups, next_group) { + ret = ulist_add(tmp, glist->group->qgroupid, + ptr_to_u64(glist->group), GFP_ATOMIC); + if (ret < 0) + return ret; + ret = ulist_add(qgroups, glist->group->qgroupid, + ptr_to_u64(glist->group), GFP_ATOMIC); + if (ret < 0) + return ret; + } + } + return 0; +} + +/* + * This adjusts the counters for all referenced qgroups if need be. + */ +static int qgroup_adjust_counters(struct btrfs_fs_info *fs_info, + u64 root_to_skip, u64 num_bytes, + struct ulist *qgroups, u64 seq, + int old_roots, int new_roots, int rescan) +{ + struct ulist_node *unode; + struct ulist_iterator uiter; + struct btrfs_qgroup *qg; + u64 cur_new_count, cur_old_count; + + ULIST_ITER_INIT(&uiter); + while ((unode = ulist_next(qgroups, &uiter))) { + bool dirty = false; + + qg = u64_to_ptr(unode->aux); + /* + * Wasn't referenced before but is now, add to the reference + * counters. + */ + if (qg->old_refcnt <= seq && qg->new_refcnt > seq) { + qg->rfer += num_bytes; + qg->rfer_cmpr += num_bytes; + dirty = true; + } + + /* + * Was referenced before but isn't now, subtract from the + * reference counters. + */ + if (qg->old_refcnt > seq && qg->new_refcnt <= seq) { + qg->rfer -= num_bytes; + qg->rfer_cmpr -= num_bytes; + dirty = true; + } + + if (qg->old_refcnt < seq) + cur_old_count = 0; + else + cur_old_count = qg->old_refcnt - seq; + if (qg->new_refcnt < seq) + cur_new_count = 0; + else + cur_new_count = qg->new_refcnt - seq; + + /* + * If our refcount was the same as the roots previously but our + * new count isn't the same as the number of roots now then we + * went from having a exclusive reference on this range to not. + */ + if (old_roots && cur_old_count == old_roots && + (cur_new_count != new_roots || new_roots == 0)) { + WARN_ON(cur_new_count != new_roots && new_roots == 0); + qg->excl -= num_bytes; + qg->excl_cmpr -= num_bytes; + dirty = true; + } + + /* + * If we didn't reference all the roots before but now we do we + * have an exclusive reference to this range. + */ + if ((!old_roots || (old_roots && cur_old_count != old_roots)) + && cur_new_count == new_roots) { + qg->excl += num_bytes; + qg->excl_cmpr += num_bytes; + dirty = true; + } + + if (dirty) + qgroup_dirty(fs_info, qg); + } + return 0; +} + +/* + * If we removed a data extent and there were other references for that bytenr + * then we need to lookup all referenced roots to make sure we still don't + * reference this bytenr. If we do then we can just discard this operation. + */ +static int check_existing_refs(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, + struct btrfs_qgroup_operation *oper) +{ + struct ulist *roots = NULL; + struct ulist_node *unode; + struct ulist_iterator uiter; + int ret = 0; + + ret = btrfs_find_all_roots(trans, fs_info, oper->bytenr, + oper->elem.seq, &roots); + if (ret < 0) + return ret; + ret = 0; + ULIST_ITER_INIT(&uiter); while ((unode = ulist_next(roots, &uiter))) { - qg = find_qgroup_rb(fs_info, unode->val); - if (!qg) - continue; - - ulist_reinit(tmp); - ret = ulist_add(tmp, qg->qgroupid, (uintptr_t)qg, GFP_ATOMIC); - if (ret < 0) - return ret; - - ULIST_ITER_INIT(&tmp_uiter); - while ((tmp_unode = ulist_next(tmp, &tmp_uiter))) { - struct btrfs_qgroup_list *glist; - - qg = (struct btrfs_qgroup *)(uintptr_t)tmp_unode->aux; - if (qg->tag == seq) - continue; - - if (qg->refcnt - seq == roots->nnodes) { - qg->excl -= sgn * num_bytes; - qg->excl_cmpr -= sgn * num_bytes; - qgroup_dirty(fs_info, qg); - } - - list_for_each_entry(glist, &qg->groups, next_group) { - ret = ulist_add(tmp, glist->group->qgroupid, - (uintptr_t)glist->group, - GFP_ATOMIC); - if (ret < 0) - return ret; - } + if (unode->val == oper->ref_root) { + ret = 1; + break; } } + ulist_free(roots); + btrfs_put_tree_mod_seq(fs_info, &oper->elem); - return 0; + return ret; +} + +/* + * If we share a reference across multiple roots then we may need to adjust + * various qgroups referenced and exclusive counters. The basic premise is this + * + * 1) We have seq to represent a 0 count. Instead of looping through all of the + * qgroups and resetting their refcount to 0 we just constantly bump this + * sequence number to act as the base reference count. This means that if + * anybody is equal to or below this sequence they were never referenced. We + * jack this sequence up by the number of roots we found each time in order to + * make sure we don't have any overlap. + * + * 2) We first search all the roots that reference the area _except_ the root + * we're acting on currently. This makes up the old_refcnt of all the qgroups + * before. + * + * 3) We walk all of the qgroups referenced by the root we are currently acting + * on, and will either adjust old_refcnt in the case of a removal or the + * new_refcnt in the case of an addition. + * + * 4) Finally we walk all the qgroups that are referenced by this range + * including the root we are acting on currently. We will adjust the counters + * based on the number of roots we had and will have after this operation. + * + * Take this example as an illustration + * + * [qgroup 1/0] + * / | \ + * [qg 0/0] [qg 0/1] [qg 0/2] + * \ | / + * [ extent ] + * + * Say we are adding a reference that is covered by qg 0/0. The first step + * would give a refcnt of 1 to qg 0/1 and 0/2 and a refcnt of 2 to qg 1/0 with + * old_roots being 2. Because it is adding new_roots will be 1. We then go + * through qg 0/0 which will get the new_refcnt set to 1 and add 1 to qg 1/0's + * new_refcnt, bringing it to 3. We then walk through all of the qgroups, we + * notice that the old refcnt for qg 0/0 < the new refcnt, so we added a + * reference and thus must add the size to the referenced bytes. Everything + * else is the same so nothing else changes. + */ +static int qgroup_shared_accounting(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, + struct btrfs_qgroup_operation *oper) +{ + struct ulist *roots = NULL; + struct ulist *qgroups, *tmp; + struct btrfs_qgroup *qgroup; + struct seq_list elem = {}; + u64 seq; + int old_roots = 0; + int new_roots = 0; + int ret = 0; + + if (oper->elem.seq) { + ret = check_existing_refs(trans, fs_info, oper); + if (ret < 0) + return ret; + if (ret) + return 0; + } + + qgroups = ulist_alloc(GFP_NOFS); + if (!qgroups) + return -ENOMEM; + + tmp = ulist_alloc(GFP_NOFS); + if (!tmp) + return -ENOMEM; + + btrfs_get_tree_mod_seq(fs_info, &elem); + ret = btrfs_find_all_roots(trans, fs_info, oper->bytenr, elem.seq, + &roots); + btrfs_put_tree_mod_seq(fs_info, &elem); + if (ret < 0) { + ulist_free(qgroups); + ulist_free(tmp); + return ret; + } + spin_lock(&fs_info->qgroup_lock); + qgroup = find_qgroup_rb(fs_info, oper->ref_root); + if (!qgroup) + goto out; + seq = fs_info->qgroup_seq; + + /* + * So roots is the list of all the roots currently pointing at the + * bytenr, including the ref we are adding if we are adding, or not if + * we are removing a ref. So we pass in the ref_root to skip that root + * in our calculations. We set old_refnct and new_refcnt cause who the + * hell knows what everything looked like before, and it doesn't matter + * except... + */ + ret = qgroup_calc_old_refcnt(fs_info, oper->ref_root, tmp, roots, qgroups, + seq, &old_roots, 0); + if (ret < 0) + goto out; + + /* + * Now adjust the refcounts of the qgroups that care about this + * reference, either the old_count in the case of removal or new_count + * in the case of an addition. + */ + ret = qgroup_calc_new_refcnt(fs_info, oper, qgroup, tmp, qgroups, + seq); + if (ret < 0) + goto out; + + /* + * ...in the case of removals. If we had a removal before we got around + * to processing this operation then we need to find that guy and count + * his references as if they really existed so we don't end up screwing + * up the exclusive counts. Then whenever we go to process the delete + * everything will be grand and we can account for whatever exclusive + * changes need to be made there. We also have to pass in old_roots so + * we have an accurate count of the roots as it pertains to this + * operations view of the world. + */ + ret = qgroup_account_deleted_refs(fs_info, oper, tmp, qgroups, seq, + &old_roots); + if (ret < 0) + goto out; + + /* + * We are adding our root, need to adjust up the number of roots, + * otherwise old_roots is the number of roots we want. + */ + if (oper->type == BTRFS_QGROUP_OPER_ADD_SHARED) { + new_roots = old_roots + 1; + } else { + new_roots = old_roots; + old_roots++; + } + fs_info->qgroup_seq += old_roots + 1; + + + /* + * And now the magic happens, bless Arne for having a pretty elegant + * solution for this. + */ + qgroup_adjust_counters(fs_info, oper->ref_root, oper->num_bytes, + qgroups, seq, old_roots, new_roots, 0); +out: + spin_unlock(&fs_info->qgroup_lock); + ulist_free(qgroups); + ulist_free(roots); + ulist_free(tmp); + return ret; } /* @@ -1342,125 +1864,65 @@ static int qgroup_account_ref_step3(struct btrfs_fs_info *fs_info, * then the space is accounted accordingly to the different roots. The * accounting algorithm works in 3 steps documented inline. */ -int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans, - struct btrfs_fs_info *fs_info, - struct btrfs_delayed_ref_node *node, - struct btrfs_delayed_extent_op *extent_op) +static int btrfs_qgroup_account(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, + struct btrfs_qgroup_operation *oper) { - struct btrfs_root *quota_root; - u64 ref_root; - struct btrfs_qgroup *qgroup; - struct ulist *roots = NULL; - u64 seq; int ret = 0; - int sgn; if (!fs_info->quota_enabled) return 0; BUG_ON(!fs_info->quota_root); - if (node->type == BTRFS_TREE_BLOCK_REF_KEY || - node->type == BTRFS_SHARED_BLOCK_REF_KEY) { - struct btrfs_delayed_tree_ref *ref; - ref = btrfs_delayed_node_to_tree_ref(node); - ref_root = ref->root; - } else if (node->type == BTRFS_EXTENT_DATA_REF_KEY || - node->type == BTRFS_SHARED_DATA_REF_KEY) { - struct btrfs_delayed_data_ref *ref; - ref = btrfs_delayed_node_to_data_ref(node); - ref_root = ref->root; - } else { - BUG(); - } - - if (!is_fstree(ref_root)) { - /* - * non-fs-trees are not being accounted - */ - return 0; - } - - switch (node->action) { - case BTRFS_ADD_DELAYED_REF: - case BTRFS_ADD_DELAYED_EXTENT: - sgn = 1; - seq = btrfs_tree_mod_seq_prev(node->seq); - break; - case BTRFS_DROP_DELAYED_REF: - sgn = -1; - seq = node->seq; - break; - case BTRFS_UPDATE_DELAYED_HEAD: - return 0; - default: - BUG(); - } - mutex_lock(&fs_info->qgroup_rescan_lock); if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) { - if (fs_info->qgroup_rescan_progress.objectid <= node->bytenr) { + if (fs_info->qgroup_rescan_progress.objectid <= oper->bytenr) { mutex_unlock(&fs_info->qgroup_rescan_lock); return 0; } } mutex_unlock(&fs_info->qgroup_rescan_lock); - /* - * the delayed ref sequence number we pass depends on the direction of - * the operation. for add operations, we pass - * tree_mod_log_prev_seq(node->seq) to skip - * the delayed ref's current sequence number, because we need the state - * of the tree before the add operation. for delete operations, we pass - * (node->seq) to include the delayed ref's current sequence number, - * because we need the state of the tree after the delete operation. - */ - ret = btrfs_find_all_roots(trans, fs_info, node->bytenr, seq, &roots); - if (ret < 0) - return ret; + ASSERT(is_fstree(oper->ref_root)); - spin_lock(&fs_info->qgroup_lock); + switch (oper->type) { + case BTRFS_QGROUP_OPER_ADD_EXCL: + case BTRFS_QGROUP_OPER_SUB_EXCL: + ret = qgroup_excl_accounting(fs_info, oper); + break; + case BTRFS_QGROUP_OPER_ADD_SHARED: + case BTRFS_QGROUP_OPER_SUB_SHARED: + ret = qgroup_shared_accounting(trans, fs_info, oper); + break; + default: + ASSERT(0); + } + return ret; +} - quota_root = fs_info->quota_root; - if (!quota_root) - goto unlock; - - qgroup = find_qgroup_rb(fs_info, ref_root); - if (!qgroup) - goto unlock; - - /* - * step 1: for each old ref, visit all nodes once and inc refcnt - */ - ulist_reinit(fs_info->qgroup_ulist); - seq = fs_info->qgroup_seq; - fs_info->qgroup_seq += roots->nnodes + 1; /* max refcnt */ - - ret = qgroup_account_ref_step1(fs_info, roots, fs_info->qgroup_ulist, - seq); - if (ret) - goto unlock; - - /* - * step 2: walk from the new root - */ - ret = qgroup_account_ref_step2(fs_info, roots, fs_info->qgroup_ulist, - seq, sgn, node->num_bytes, qgroup); - if (ret) - goto unlock; - - /* - * step 3: walk again from old refs - */ - ret = qgroup_account_ref_step3(fs_info, roots, fs_info->qgroup_ulist, - seq, sgn, node->num_bytes); - if (ret) - goto unlock; - -unlock: - spin_unlock(&fs_info->qgroup_lock); - ulist_free(roots); +/* + * Needs to be called everytime we run delayed refs, even if there is an error + * in order to cleanup outstanding operations. + */ +int btrfs_delayed_qgroup_accounting(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info) +{ + struct btrfs_qgroup_operation *oper; + int ret = 0; + while (!list_empty(&trans->qgroup_ref_list)) { + oper = list_first_entry(&trans->qgroup_ref_list, + struct btrfs_qgroup_operation, list); + list_del_init(&oper->list); + if (!ret || !trans->aborted) + ret = btrfs_qgroup_account(trans, fs_info, oper); + spin_lock(&fs_info->qgroup_op_lock); + rb_erase(&oper->n, &fs_info->qgroup_op_tree); + spin_unlock(&fs_info->qgroup_op_lock); + btrfs_put_tree_mod_seq(fs_info, &oper->elem); + kfree(oper); + } return ret; } @@ -1629,8 +2091,16 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, srcgroup = find_qgroup_rb(fs_info, srcid); if (!srcgroup) goto unlock; - dstgroup->rfer = srcgroup->rfer - level_size; - dstgroup->rfer_cmpr = srcgroup->rfer_cmpr - level_size; + + /* + * We call inherit after we clone the root in order to make sure + * our counts don't go crazy, so at this point the only + * difference between the two roots should be the root node. + */ + dstgroup->rfer = srcgroup->rfer; + dstgroup->rfer_cmpr = srcgroup->rfer_cmpr; + dstgroup->excl = level_size; + dstgroup->excl_cmpr = level_size; srcgroup->excl = level_size; srcgroup->excl_cmpr = level_size; qgroup_dirty(fs_info, dstgroup); @@ -1734,7 +2204,7 @@ int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes) struct btrfs_qgroup *qg; struct btrfs_qgroup_list *glist; - qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux; + qg = u64_to_ptr(unode->aux); if ((qg->lim_flags & BTRFS_QGROUP_LIMIT_MAX_RFER) && qg->reserved + (s64)qg->rfer + num_bytes > @@ -1766,7 +2236,7 @@ int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes) while ((unode = ulist_next(fs_info->qgroup_ulist, &uiter))) { struct btrfs_qgroup *qg; - qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux; + qg = u64_to_ptr(unode->aux); qg->reserved += num_bytes; } @@ -1812,7 +2282,7 @@ void btrfs_qgroup_free(struct btrfs_root *root, u64 num_bytes) struct btrfs_qgroup *qg; struct btrfs_qgroup_list *glist; - qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux; + qg = u64_to_ptr(unode->aux); qg->reserved -= num_bytes; @@ -1848,15 +2318,15 @@ void assert_qgroups_uptodate(struct btrfs_trans_handle *trans) */ static int qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path, - struct btrfs_trans_handle *trans, struct ulist *tmp, - struct extent_buffer *scratch_leaf) + struct btrfs_trans_handle *trans, struct ulist *qgroups, + struct ulist *tmp, struct extent_buffer *scratch_leaf) { struct btrfs_key found; struct ulist *roots = NULL; - struct ulist_node *unode; - struct ulist_iterator uiter; struct seq_list tree_mod_seq_elem = {}; + u64 num_bytes; u64 seq; + int new_roots; int slot; int ret; @@ -1897,8 +2367,6 @@ qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path, mutex_unlock(&fs_info->qgroup_rescan_lock); for (; slot < btrfs_header_nritems(scratch_leaf); ++slot) { - u64 num_bytes; - btrfs_item_key_to_cpu(scratch_leaf, &found, slot); if (found.type != BTRFS_EXTENT_ITEM_KEY && found.type != BTRFS_METADATA_ITEM_KEY) @@ -1908,76 +2376,34 @@ qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path, else num_bytes = found.offset; - ret = btrfs_find_all_roots(trans, fs_info, found.objectid, - tree_mod_seq_elem.seq, &roots); + ulist_reinit(qgroups); + ret = btrfs_find_all_roots(NULL, fs_info, found.objectid, 0, + &roots); if (ret < 0) goto out; spin_lock(&fs_info->qgroup_lock); seq = fs_info->qgroup_seq; fs_info->qgroup_seq += roots->nnodes + 1; /* max refcnt */ - ret = qgroup_account_ref_step1(fs_info, roots, tmp, seq); - if (ret) { + new_roots = 0; + ret = qgroup_calc_old_refcnt(fs_info, 0, tmp, roots, qgroups, + seq, &new_roots, 1); + if (ret < 0) { spin_unlock(&fs_info->qgroup_lock); ulist_free(roots); goto out; } - /* - * step2 of btrfs_qgroup_account_ref works from a single root, - * we're doing all at once here. - */ - ulist_reinit(tmp); - ULIST_ITER_INIT(&uiter); - while ((unode = ulist_next(roots, &uiter))) { - struct btrfs_qgroup *qg; - - qg = find_qgroup_rb(fs_info, unode->val); - if (!qg) - continue; - - ret = ulist_add(tmp, qg->qgroupid, (uintptr_t)qg, - GFP_ATOMIC); - if (ret < 0) { - spin_unlock(&fs_info->qgroup_lock); - ulist_free(roots); - goto out; - } + ret = qgroup_adjust_counters(fs_info, 0, num_bytes, qgroups, + seq, 0, new_roots, 1); + if (ret < 0) { + spin_unlock(&fs_info->qgroup_lock); + ulist_free(roots); + goto out; } - - /* this loop is similar to step 2 of btrfs_qgroup_account_ref */ - ULIST_ITER_INIT(&uiter); - while ((unode = ulist_next(tmp, &uiter))) { - struct btrfs_qgroup *qg; - struct btrfs_qgroup_list *glist; - - qg = (struct btrfs_qgroup *)(uintptr_t) unode->aux; - qg->rfer += num_bytes; - qg->rfer_cmpr += num_bytes; - WARN_ON(qg->tag >= seq); - if (qg->refcnt - seq == roots->nnodes) { - qg->excl += num_bytes; - qg->excl_cmpr += num_bytes; - } - qgroup_dirty(fs_info, qg); - - list_for_each_entry(glist, &qg->groups, next_group) { - ret = ulist_add(tmp, glist->group->qgroupid, - (uintptr_t)glist->group, - GFP_ATOMIC); - if (ret < 0) { - spin_unlock(&fs_info->qgroup_lock); - ulist_free(roots); - goto out; - } - } - } - spin_unlock(&fs_info->qgroup_lock); ulist_free(roots); - ret = 0; } - out: btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem); @@ -1990,13 +2416,16 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work) qgroup_rescan_work); struct btrfs_path *path; struct btrfs_trans_handle *trans = NULL; - struct ulist *tmp = NULL; + struct ulist *tmp = NULL, *qgroups = NULL; struct extent_buffer *scratch_leaf = NULL; int err = -ENOMEM; path = btrfs_alloc_path(); if (!path) goto out; + qgroups = ulist_alloc(GFP_NOFS); + if (!qgroups) + goto out; tmp = ulist_alloc(GFP_NOFS); if (!tmp) goto out; @@ -2015,7 +2444,7 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work) err = -EINTR; } else { err = qgroup_rescan_leaf(fs_info, path, trans, - tmp, scratch_leaf); + qgroups, tmp, scratch_leaf); } if (err > 0) btrfs_commit_transaction(trans, fs_info->fs_root); @@ -2025,7 +2454,7 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work) out: kfree(scratch_leaf); - ulist_free(tmp); + ulist_free(qgroups); btrfs_free_path(path); mutex_lock(&fs_info->qgroup_rescan_lock); diff --git a/fs/btrfs/qgroup.h b/fs/btrfs/qgroup.h new file mode 100644 index 000000000000..5952ff1fbd7a --- /dev/null +++ b/fs/btrfs/qgroup.h @@ -0,0 +1,107 @@ +/* + * Copyright (C) 2014 Facebook. All rights reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public + * License v2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public + * License along with this program; if not, write to the + * Free Software Foundation, Inc., 59 Temple Place - Suite 330, + * Boston, MA 021110-1307, USA. + */ + +#ifndef __BTRFS_QGROUP__ +#define __BTRFS_QGROUP__ + +/* + * A description of the operations, all of these operations only happen when we + * are adding the 1st reference for that subvolume in the case of adding space + * or on the last reference delete in the case of subtraction. The only + * exception is the last one, which is added for confusion. + * + * BTRFS_QGROUP_OPER_ADD_EXCL: adding bytes where this subvolume is the only + * one pointing at the bytes we are adding. This is called on the first + * allocation. + * + * BTRFS_QGROUP_OPER_ADD_SHARED: adding bytes where this bytenr is going to be + * shared between subvols. This is called on the creation of a ref that already + * has refs from a different subvolume, so basically reflink. + * + * BTRFS_QGROUP_OPER_SUB_EXCL: removing bytes where this subvolume is the only + * one referencing the range. + * + * BTRFS_QGROUP_OPER_SUB_SHARED: removing bytes where this subvolume shares with + * refs with other subvolumes. + */ +enum btrfs_qgroup_operation_type { + BTRFS_QGROUP_OPER_ADD_EXCL, + BTRFS_QGROUP_OPER_ADD_SHARED, + BTRFS_QGROUP_OPER_SUB_EXCL, + BTRFS_QGROUP_OPER_SUB_SHARED, +}; + +struct btrfs_qgroup_operation { + u64 ref_root; + u64 bytenr; + u64 num_bytes; + u64 seq; + enum btrfs_qgroup_operation_type type; + struct seq_list elem; + struct rb_node n; + struct list_head list; +}; + +int btrfs_quota_enable(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info); +int btrfs_quota_disable(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info); +int btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info); +void btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info); +int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info); +int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 src, u64 dst); +int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 src, u64 dst); +int btrfs_create_qgroup(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 qgroupid, + char *name); +int btrfs_remove_qgroup(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 qgroupid); +int btrfs_limit_qgroup(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 qgroupid, + struct btrfs_qgroup_limit *limit); +int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info); +void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info); +struct btrfs_delayed_extent_op; +int btrfs_qgroup_record_ref(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 ref_root, + u64 bytenr, u64 num_bytes, + enum btrfs_qgroup_operation_type type, + int mod_seq); +int btrfs_delayed_qgroup_accounting(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info); +void btrfs_remove_qgroup_operation(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, + struct btrfs_qgroup_operation *oper); +int btrfs_run_qgroups(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info); +int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 srcid, u64 objectid, + struct btrfs_qgroup_inherit *inherit); +int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes); +void btrfs_qgroup_free(struct btrfs_root *root, u64 num_bytes); + +void assert_qgroups_uptodate(struct btrfs_trans_handle *trans); + +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS +int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid, + u64 rfer, u64 excl); +#endif + +#endif /* __BTRFS_QGROUP__ */ diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 7c4c049da871..3aafbde8b637 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -31,6 +31,7 @@ #include "inode-map.h" #include "volumes.h" #include "dev-replace.h" +#include "qgroup.h" #define BTRFS_ROOT_TRANS_TAG 0 @@ -703,23 +704,9 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans, return 0; } - /* - * do the qgroup accounting as early as possible - */ - err = btrfs_delayed_refs_qgroup_accounting(trans, info); - btrfs_trans_release_metadata(trans, root); trans->block_rsv = NULL; - if (trans->qgroup_reserved) { - /* - * the same root has to be passed here between start_transaction - * and end_transaction. Subvolume quota depends on this. - */ - btrfs_qgroup_free(trans->root, trans->qgroup_reserved); - trans->qgroup_reserved = 0; - } - if (!list_empty(&trans->new_bgs)) btrfs_create_pending_block_groups(trans, root); @@ -730,6 +717,15 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans, btrfs_run_delayed_refs(trans, root, cur); } + if (trans->qgroup_reserved) { + /* + * the same root has to be passed here between start_transaction + * and end_transaction. Subvolume quota depends on this. + */ + btrfs_qgroup_free(trans->root, trans->qgroup_reserved); + trans->qgroup_reserved = 0; + } + btrfs_trans_release_metadata(trans, root); trans->block_rsv = NULL; @@ -1169,12 +1165,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans, goto no_free_objectid; } - pending->error = btrfs_qgroup_inherit(trans, fs_info, - root->root_key.objectid, - objectid, pending->inherit); - if (pending->error) - goto no_free_objectid; - key.objectid = objectid; key.offset = (u64)-1; key.type = BTRFS_ROOT_ITEM_KEY; @@ -1271,6 +1261,22 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans, goto fail; } + /* + * We need to flush delayed refs in order to make sure all of our quota + * operations have been done before we call btrfs_qgroup_inherit. + */ + ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + goto fail; + } + + pending->error = btrfs_qgroup_inherit(trans, fs_info, + root->root_key.objectid, + objectid, pending->inherit); + if (pending->error) + goto no_free_objectid; + /* see comments in should_cow_block() */ set_bit(BTRFS_ROOT_FORCE_COW, &root->state); smp_wmb(); @@ -1599,12 +1605,6 @@ static int btrfs_flush_all_pending_stuffs(struct btrfs_trans_handle *trans, * them now so that they hinder processing of more delayed refs * as little as possible. */ - if (ret) { - btrfs_delayed_refs_qgroup_accounting(trans, root->fs_info); - return ret; - } - - ret = btrfs_delayed_refs_qgroup_accounting(trans, root->fs_info); if (ret) return ret; From faa2dbf004e89e8f7ccd28fbe6f07c308417b8ae Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 7 May 2014 17:06:09 -0400 Subject: [PATCH 41/80] Btrfs: add sanity tests for new qgroup accounting code This exercises the various parts of the new qgroup accounting code. We do some basic stuff and do some things with the shared refs to make sure all that code works. I had to add a bunch of infrastructure because I needed to be able to insert items into a fake tree without having to do all the hard work myself, hopefully this will be usefull in the future. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/Makefile | 2 +- fs/btrfs/backref.c | 4 + fs/btrfs/ctree.c | 4 + fs/btrfs/ctree.h | 6 + fs/btrfs/disk-io.c | 18 +- fs/btrfs/disk-io.h | 1 + fs/btrfs/extent-tree.c | 17 ++ fs/btrfs/extent_io.c | 47 ++++ fs/btrfs/extent_io.h | 2 + fs/btrfs/qgroup.c | 23 ++ fs/btrfs/super.c | 3 + fs/btrfs/tests/btrfs-tests.c | 97 +++++++ fs/btrfs/tests/btrfs-tests.h | 9 + fs/btrfs/tests/inode-tests.c | 35 +-- fs/btrfs/tests/qgroup-tests.c | 468 ++++++++++++++++++++++++++++++++++ fs/btrfs/transaction.h | 1 + 16 files changed, 700 insertions(+), 37 deletions(-) create mode 100644 fs/btrfs/tests/qgroup-tests.c diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile index f341a98031d2..6d1d0b93b1aa 100644 --- a/fs/btrfs/Makefile +++ b/fs/btrfs/Makefile @@ -16,4 +16,4 @@ btrfs-$(CONFIG_BTRFS_FS_CHECK_INTEGRITY) += check-integrity.o btrfs-$(CONFIG_BTRFS_FS_RUN_SANITY_TESTS) += tests/free-space-tests.o \ tests/extent-buffer-tests.o tests/btrfs-tests.o \ - tests/extent-io-tests.o tests/inode-tests.o + tests/extent-io-tests.o tests/inode-tests.o tests/qgroup-tests.o diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index 10db21fa0926..f09aa18ab710 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -900,7 +900,11 @@ again: goto out; BUG_ON(ret == 0); +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + if (trans && likely(trans->type != __TRANS_DUMMY)) { +#else if (trans) { +#endif /* * look if there are updates for this ref queued and lock the * head diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index bbbe4f1c5086..d99d9651dd58 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -1506,6 +1506,10 @@ static inline int should_cow_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf) { +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + if (unlikely(test_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state))) + return 0; +#endif /* ensure we can see the force_cow */ smp_rmb(); diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 06cc384933cc..4896d7a947eb 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1798,6 +1798,10 @@ struct btrfs_root { u64 highest_objectid; +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + u64 alloc_bytenr; +#endif + u64 defrag_trans_start; struct btrfs_key defrag_progress; struct btrfs_key defrag_max; @@ -4111,6 +4115,8 @@ static inline int btrfs_defrag_cancelled(struct btrfs_fs_info *fs_info) /* Sanity test specific functions */ #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS void btrfs_test_destroy_inode(struct inode *inode); +int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid, + u64 rfer, u64 excl); #endif #endif diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 77f92a32e230..0c0fa78ef452 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1110,6 +1110,11 @@ struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root, struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize) { +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + if (unlikely(test_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state))) + return alloc_test_extent_buffer(root->fs_info, bytenr, + blocksize); +#endif return alloc_extent_buffer(root->fs_info, bytenr, blocksize); } @@ -1288,6 +1293,7 @@ struct btrfs_root *btrfs_alloc_dummy_root(void) return ERR_PTR(-ENOMEM); __setup_root(4096, 4096, 4096, 4096, root, NULL, 1); set_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state); + root->alloc_bytenr = 0; return root; } @@ -2089,7 +2095,7 @@ static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root) free_root_extent_buffers(info->chunk_root); } -static void del_fs_roots(struct btrfs_fs_info *fs_info) +void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info) { int ret; struct btrfs_root *gang[8]; @@ -2969,7 +2975,7 @@ fail_qgroup: fail_trans_kthread: kthread_stop(fs_info->transaction_kthread); btrfs_cleanup_transaction(fs_info->tree_root); - del_fs_roots(fs_info); + btrfs_free_fs_roots(fs_info); fail_cleaner: kthread_stop(fs_info->cleaner_kthread); @@ -3504,8 +3510,10 @@ void btrfs_drop_and_free_fs_root(struct btrfs_fs_info *fs_info, if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) btrfs_free_log(NULL, root); - __btrfs_remove_free_space_cache(root->free_ino_pinned); - __btrfs_remove_free_space_cache(root->free_ino_ctl); + if (root->free_ino_pinned) + __btrfs_remove_free_space_cache(root->free_ino_pinned); + if (root->free_ino_ctl) + __btrfs_remove_free_space_cache(root->free_ino_ctl); free_fs_root(root); } @@ -3655,7 +3663,7 @@ int close_ctree(struct btrfs_root *root) btrfs_sysfs_remove_one(fs_info); - del_fs_roots(fs_info); + btrfs_free_fs_roots(fs_info); btrfs_put_block_group_cache(fs_info); diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h index 53059df350f8..23ce3ceba0a9 100644 --- a/fs/btrfs/disk-io.h +++ b/fs/btrfs/disk-io.h @@ -68,6 +68,7 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_root *tree_root, int btrfs_init_fs_root(struct btrfs_root *root); int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root); +void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info); struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_key *key, diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 343eb10230a1..ddf16bfb9270 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2992,6 +2992,10 @@ static int __btrfs_mod_ref(struct btrfs_trans_handle *trans, int (*process_func)(struct btrfs_trans_handle *, struct btrfs_root *, u64, u64, u64, u64, u64, u64, int); +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + if (unlikely(test_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state))) + return 0; +#endif ref_root = btrfs_header_owner(buf); nritems = btrfs_header_nritems(buf); level = btrfs_header_level(buf); @@ -6151,6 +6155,10 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, int ret; struct btrfs_fs_info *fs_info = root->fs_info; +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + if (unlikely(test_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state))) + return 0; +#endif add_pinned_bytes(root->fs_info, num_bytes, owner, root_objectid); /* @@ -7157,6 +7165,15 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans, bool skinny_metadata = btrfs_fs_incompat(root->fs_info, SKINNY_METADATA); +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + if (unlikely(test_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state))) { + buf = btrfs_init_new_buffer(trans, root, root->alloc_bytenr, + blocksize, level); + if (!IS_ERR(buf)) + root->alloc_bytenr += blocksize; + return buf; + } +#endif block_rsv = use_block_rsv(trans, root, blocksize); if (IS_ERR(block_rsv)) return ERR_CAST(block_rsv); diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index a55580f4e611..8285ed0464fa 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -4549,6 +4549,53 @@ struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info, return NULL; } +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS +struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info, + u64 start, unsigned long len) +{ + struct extent_buffer *eb, *exists = NULL; + int ret; + + eb = find_extent_buffer(fs_info, start); + if (eb) + return eb; + eb = alloc_dummy_extent_buffer(start, len); + if (!eb) + return NULL; + eb->fs_info = fs_info; +again: + ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM); + if (ret) + goto free_eb; + spin_lock(&fs_info->buffer_lock); + ret = radix_tree_insert(&fs_info->buffer_radix, + start >> PAGE_CACHE_SHIFT, eb); + spin_unlock(&fs_info->buffer_lock); + radix_tree_preload_end(); + if (ret == -EEXIST) { + exists = find_extent_buffer(fs_info, start); + if (exists) + goto free_eb; + else + goto again; + } + check_buffer_tree_ref(eb); + set_bit(EXTENT_BUFFER_IN_TREE, &eb->bflags); + + /* + * We will free dummy extent buffer's if they come into + * free_extent_buffer with a ref count of 2, but if we are using this we + * want the buffers to stay in memory until we're done with them, so + * bump the ref count again. + */ + atomic_inc(&eb->refs); + return eb; +free_eb: + btrfs_release_extent_buffer(eb); + return exists; +} +#endif + struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info, u64 start, unsigned long len) { diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h index c488b45237bf..8b63f2d46518 100644 --- a/fs/btrfs/extent_io.h +++ b/fs/btrfs/extent_io.h @@ -350,5 +350,7 @@ noinline u64 find_lock_delalloc_range(struct inode *inode, struct extent_io_tree *tree, struct page *locked_page, u64 *start, u64 *end, u64 max_bytes); +struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info, + u64 start, unsigned long len); #endif #endif diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c index 09b8cc83965c..73f706c9a342 100644 --- a/fs/btrfs/qgroup.c +++ b/fs/btrfs/qgroup.c @@ -246,6 +246,21 @@ static int del_relation_rb(struct btrfs_fs_info *fs_info, return -ENOENT; } +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS +int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid, + u64 rfer, u64 excl) +{ + struct btrfs_qgroup *qgroup; + + qgroup = find_qgroup_rb(fs_info, qgroupid); + if (!qgroup) + return -EINVAL; + if (qgroup->rfer != rfer || qgroup->excl != excl) + return -EINVAL; + return 0; +} +#endif + /* * The full config is read in one go, only called from open_ctree() * It doesn't use any locking, as at this point we're still single-threaded @@ -524,6 +539,10 @@ static int add_qgroup_item(struct btrfs_trans_handle *trans, struct extent_buffer *leaf; struct btrfs_key key; +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + if (unlikely(test_bit(BTRFS_ROOT_DUMMY_ROOT, "a_root->state))) + return 0; +#endif path = btrfs_alloc_path(); if (!path) return -ENOMEM; @@ -673,6 +692,10 @@ static int update_qgroup_info_item(struct btrfs_trans_handle *trans, int ret; int slot; +#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS + if (unlikely(test_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state))) + return 0; +#endif key.objectid = 0; key.type = BTRFS_QGROUP_INFO_KEY; key.offset = qgroup->qgroupid; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 8f7c03db1f8d..d80a2786f563 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -1895,6 +1895,9 @@ static int btrfs_run_sanity_tests(void) if (ret) goto out; ret = btrfs_test_inodes(); + if (ret) + goto out; + ret = btrfs_test_qgroups(); out: btrfs_destroy_test_fs(); return ret; diff --git a/fs/btrfs/tests/btrfs-tests.c b/fs/btrfs/tests/btrfs-tests.c index 757ef00a75a4..a5dcacb5df9c 100644 --- a/fs/btrfs/tests/btrfs-tests.c +++ b/fs/btrfs/tests/btrfs-tests.c @@ -21,6 +21,9 @@ #include #include "btrfs-tests.h" #include "../ctree.h" +#include "../volumes.h" +#include "../disk-io.h" +#include "../qgroup.h" static struct vfsmount *test_mnt = NULL; @@ -72,3 +75,97 @@ void btrfs_destroy_test_fs(void) kern_unmount(test_mnt); unregister_filesystem(&test_type); } + +struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(void) +{ + struct btrfs_fs_info *fs_info = kzalloc(sizeof(struct btrfs_fs_info), + GFP_NOFS); + + if (!fs_info) + return fs_info; + fs_info->fs_devices = kzalloc(sizeof(struct btrfs_fs_devices), + GFP_NOFS); + if (!fs_info->fs_devices) { + kfree(fs_info); + return NULL; + } + fs_info->super_copy = kzalloc(sizeof(struct btrfs_super_block), + GFP_NOFS); + if (!fs_info->super_copy) { + kfree(fs_info->fs_devices); + kfree(fs_info); + return NULL; + } + + if (init_srcu_struct(&fs_info->subvol_srcu)) { + kfree(fs_info->fs_devices); + kfree(fs_info->super_copy); + kfree(fs_info); + return NULL; + } + + spin_lock_init(&fs_info->buffer_lock); + spin_lock_init(&fs_info->qgroup_lock); + spin_lock_init(&fs_info->qgroup_op_lock); + spin_lock_init(&fs_info->super_lock); + spin_lock_init(&fs_info->fs_roots_radix_lock); + spin_lock_init(&fs_info->tree_mod_seq_lock); + mutex_init(&fs_info->qgroup_ioctl_lock); + mutex_init(&fs_info->qgroup_rescan_lock); + rwlock_init(&fs_info->tree_mod_log_lock); + fs_info->running_transaction = NULL; + fs_info->qgroup_tree = RB_ROOT; + fs_info->qgroup_ulist = NULL; + atomic64_set(&fs_info->tree_mod_seq, 0); + INIT_LIST_HEAD(&fs_info->dirty_qgroups); + INIT_LIST_HEAD(&fs_info->dead_roots); + INIT_LIST_HEAD(&fs_info->tree_mod_seq_list); + INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC); + INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC); + return fs_info; +} + +static void btrfs_free_dummy_fs_info(struct btrfs_fs_info *fs_info) +{ + struct radix_tree_iter iter; + void **slot; + + spin_lock(&fs_info->buffer_lock); +restart: + radix_tree_for_each_slot(slot, &fs_info->buffer_radix, &iter, 0) { + struct extent_buffer *eb; + + eb = radix_tree_deref_slot(slot); + if (!eb) + continue; + /* Shouldn't happen but that kind of thinking creates CVE's */ + if (radix_tree_exception(eb)) { + if (radix_tree_deref_retry(eb)) + goto restart; + continue; + } + spin_unlock(&fs_info->buffer_lock); + free_extent_buffer_stale(eb); + spin_lock(&fs_info->buffer_lock); + } + spin_unlock(&fs_info->buffer_lock); + + btrfs_free_qgroup_config(fs_info); + btrfs_free_fs_roots(fs_info); + cleanup_srcu_struct(&fs_info->subvol_srcu); + kfree(fs_info->super_copy); + kfree(fs_info->fs_devices); + kfree(fs_info); +} + +void btrfs_free_dummy_root(struct btrfs_root *root) +{ + if (!root) + return; + if (root->node) + free_extent_buffer(root->node); + if (root->fs_info) + btrfs_free_dummy_fs_info(root->fs_info); + kfree(root); +} + diff --git a/fs/btrfs/tests/btrfs-tests.h b/fs/btrfs/tests/btrfs-tests.h index 312560a9123d..fd3954224480 100644 --- a/fs/btrfs/tests/btrfs-tests.h +++ b/fs/btrfs/tests/btrfs-tests.h @@ -23,13 +23,18 @@ #define test_msg(fmt, ...) pr_info("BTRFS: selftest: " fmt, ##__VA_ARGS__) +struct btrfs_root; + int btrfs_test_free_space_cache(void); int btrfs_test_extent_buffer_operations(void); int btrfs_test_extent_io(void); int btrfs_test_inodes(void); +int btrfs_test_qgroups(void); int btrfs_init_test_fs(void); void btrfs_destroy_test_fs(void); struct inode *btrfs_new_test_inode(void); +struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(void); +void btrfs_free_dummy_root(struct btrfs_root *root); #else static inline int btrfs_test_free_space_cache(void) { @@ -54,6 +59,10 @@ static inline int btrfs_test_inodes(void) { return 0; } +static inline int btrfs_test_qgroups(void) +{ + return 0; +} #endif #endif diff --git a/fs/btrfs/tests/inode-tests.c b/fs/btrfs/tests/inode-tests.c index 397d1f99a8eb..3ae0f5b8bb80 100644 --- a/fs/btrfs/tests/inode-tests.c +++ b/fs/btrfs/tests/inode-tests.c @@ -23,33 +23,6 @@ #include "../extent_io.h" #include "../volumes.h" -static struct btrfs_fs_info *alloc_dummy_fs_info(void) -{ - struct btrfs_fs_info *fs_info = kzalloc(sizeof(struct btrfs_fs_info), - GFP_NOFS); - if (!fs_info) - return fs_info; - fs_info->fs_devices = kzalloc(sizeof(struct btrfs_fs_devices), - GFP_NOFS); - if (!fs_info->fs_devices) { - kfree(fs_info); - return NULL; - } - return fs_info; -} -static void free_dummy_root(struct btrfs_root *root) -{ - if (!root) - return; - if (root->fs_info) { - kfree(root->fs_info->fs_devices); - kfree(root->fs_info); - } - if (root->node) - free_extent_buffer(root->node); - kfree(root); -} - static void insert_extent(struct btrfs_root *root, u64 start, u64 len, u64 ram_bytes, u64 offset, u64 disk_bytenr, u64 disk_len, u32 type, u8 compression, int slot) @@ -276,7 +249,7 @@ static noinline int test_btrfs_get_extent(void) * We do this since btrfs_get_extent wants to assign em->bdev to * root->fs_info->fs_devices->latest_bdev. */ - root->fs_info = alloc_dummy_fs_info(); + root->fs_info = btrfs_alloc_dummy_fs_info(); if (!root->fs_info) { test_msg("Couldn't allocate dummy fs info\n"); goto out; @@ -837,7 +810,7 @@ out: if (!IS_ERR(em)) free_extent_map(em); iput(inode); - free_dummy_root(root); + btrfs_free_dummy_root(root); return ret; } @@ -864,7 +837,7 @@ static int test_hole_first(void) goto out; } - root->fs_info = alloc_dummy_fs_info(); + root->fs_info = btrfs_alloc_dummy_fs_info(); if (!root->fs_info) { test_msg("Couldn't allocate dummy fs info\n"); goto out; @@ -934,7 +907,7 @@ out: if (!IS_ERR(em)) free_extent_map(em); iput(inode); - free_dummy_root(root); + btrfs_free_dummy_root(root); return ret; } diff --git a/fs/btrfs/tests/qgroup-tests.c b/fs/btrfs/tests/qgroup-tests.c new file mode 100644 index 000000000000..fa691b754aaf --- /dev/null +++ b/fs/btrfs/tests/qgroup-tests.c @@ -0,0 +1,468 @@ +/* + * Copyright (C) 2013 Facebook. All rights reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public + * License v2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public + * License along with this program; if not, write to the + * Free Software Foundation, Inc., 59 Temple Place - Suite 330, + * Boston, MA 021110-1307, USA. + */ + +#include "btrfs-tests.h" +#include "../ctree.h" +#include "../transaction.h" +#include "../disk-io.h" +#include "../qgroup.h" + +static void init_dummy_trans(struct btrfs_trans_handle *trans) +{ + memset(trans, 0, sizeof(*trans)); + trans->transid = 1; + INIT_LIST_HEAD(&trans->qgroup_ref_list); + trans->type = __TRANS_DUMMY; +} + +static int insert_normal_tree_ref(struct btrfs_root *root, u64 bytenr, + u64 num_bytes, u64 parent, u64 root_objectid) +{ + struct btrfs_trans_handle trans; + struct btrfs_extent_item *item; + struct btrfs_extent_inline_ref *iref; + struct btrfs_tree_block_info *block_info; + struct btrfs_path *path; + struct extent_buffer *leaf; + struct btrfs_key ins; + u32 size = sizeof(*item) + sizeof(*iref) + sizeof(*block_info); + int ret; + + init_dummy_trans(&trans); + + ins.objectid = bytenr; + ins.type = BTRFS_EXTENT_ITEM_KEY; + ins.offset = num_bytes; + + path = btrfs_alloc_path(); + if (!path) { + test_msg("Couldn't allocate path\n"); + return -ENOMEM; + } + + path->leave_spinning = 1; + ret = btrfs_insert_empty_item(&trans, root, path, &ins, size); + if (ret) { + test_msg("Couldn't insert ref %d\n", ret); + btrfs_free_path(path); + return ret; + } + + leaf = path->nodes[0]; + item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item); + btrfs_set_extent_refs(leaf, item, 1); + btrfs_set_extent_generation(leaf, item, 1); + btrfs_set_extent_flags(leaf, item, BTRFS_EXTENT_FLAG_TREE_BLOCK); + block_info = (struct btrfs_tree_block_info *)(item + 1); + btrfs_set_tree_block_level(leaf, block_info, 1); + iref = (struct btrfs_extent_inline_ref *)(block_info + 1); + if (parent > 0) { + btrfs_set_extent_inline_ref_type(leaf, iref, + BTRFS_SHARED_BLOCK_REF_KEY); + btrfs_set_extent_inline_ref_offset(leaf, iref, parent); + } else { + btrfs_set_extent_inline_ref_type(leaf, iref, BTRFS_TREE_BLOCK_REF_KEY); + btrfs_set_extent_inline_ref_offset(leaf, iref, root_objectid); + } + btrfs_free_path(path); + return 0; +} + +static int add_tree_ref(struct btrfs_root *root, u64 bytenr, u64 num_bytes, + u64 parent, u64 root_objectid) +{ + struct btrfs_trans_handle trans; + struct btrfs_extent_item *item; + struct btrfs_path *path; + struct btrfs_key key; + u64 refs; + int ret; + + init_dummy_trans(&trans); + + key.objectid = bytenr; + key.type = BTRFS_EXTENT_ITEM_KEY; + key.offset = num_bytes; + + path = btrfs_alloc_path(); + if (!path) { + test_msg("Couldn't allocate path\n"); + return -ENOMEM; + } + + path->leave_spinning = 1; + ret = btrfs_search_slot(&trans, root, &key, path, 0, 1); + if (ret) { + test_msg("Couldn't find extent ref\n"); + btrfs_free_path(path); + return ret; + } + + item = btrfs_item_ptr(path->nodes[0], path->slots[0], + struct btrfs_extent_item); + refs = btrfs_extent_refs(path->nodes[0], item); + btrfs_set_extent_refs(path->nodes[0], item, refs + 1); + btrfs_release_path(path); + + key.objectid = bytenr; + if (parent) { + key.type = BTRFS_SHARED_BLOCK_REF_KEY; + key.offset = parent; + } else { + key.type = BTRFS_TREE_BLOCK_REF_KEY; + key.offset = root_objectid; + } + + ret = btrfs_insert_empty_item(&trans, root, path, &key, 0); + if (ret) + test_msg("Failed to insert backref\n"); + btrfs_free_path(path); + return ret; +} + +static int remove_extent_item(struct btrfs_root *root, u64 bytenr, + u64 num_bytes) +{ + struct btrfs_trans_handle trans; + struct btrfs_key key; + struct btrfs_path *path; + int ret; + + init_dummy_trans(&trans); + + key.objectid = bytenr; + key.type = BTRFS_EXTENT_ITEM_KEY; + key.offset = num_bytes; + + path = btrfs_alloc_path(); + if (!path) { + test_msg("Couldn't allocate path\n"); + return -ENOMEM; + } + path->leave_spinning = 1; + + ret = btrfs_search_slot(&trans, root, &key, path, -1, 1); + if (ret) { + test_msg("Didn't find our key %d\n", ret); + btrfs_free_path(path); + return ret; + } + btrfs_del_item(&trans, root, path); + btrfs_free_path(path); + return 0; +} + +static int remove_extent_ref(struct btrfs_root *root, u64 bytenr, + u64 num_bytes, u64 parent, u64 root_objectid) +{ + struct btrfs_trans_handle trans; + struct btrfs_extent_item *item; + struct btrfs_path *path; + struct btrfs_key key; + u64 refs; + int ret; + + init_dummy_trans(&trans); + + key.objectid = bytenr; + key.type = BTRFS_EXTENT_ITEM_KEY; + key.offset = num_bytes; + + path = btrfs_alloc_path(); + if (!path) { + test_msg("Couldn't allocate path\n"); + return -ENOMEM; + } + + path->leave_spinning = 1; + ret = btrfs_search_slot(&trans, root, &key, path, 0, 1); + if (ret) { + test_msg("Couldn't find extent ref\n"); + btrfs_free_path(path); + return ret; + } + + item = btrfs_item_ptr(path->nodes[0], path->slots[0], + struct btrfs_extent_item); + refs = btrfs_extent_refs(path->nodes[0], item); + btrfs_set_extent_refs(path->nodes[0], item, refs - 1); + btrfs_release_path(path); + + key.objectid = bytenr; + if (parent) { + key.type = BTRFS_SHARED_BLOCK_REF_KEY; + key.offset = parent; + } else { + key.type = BTRFS_TREE_BLOCK_REF_KEY; + key.offset = root_objectid; + } + + ret = btrfs_search_slot(&trans, root, &key, path, -1, 1); + if (ret) { + test_msg("Couldn't find backref %d\n", ret); + btrfs_free_path(path); + return ret; + } + btrfs_del_item(&trans, root, path); + btrfs_free_path(path); + return ret; +} + +static int test_no_shared_qgroup(struct btrfs_root *root) +{ + struct btrfs_trans_handle trans; + struct btrfs_fs_info *fs_info = root->fs_info; + int ret; + + init_dummy_trans(&trans); + + test_msg("Qgroup basic add\n"); + ret = btrfs_create_qgroup(NULL, fs_info, 5, NULL); + if (ret) { + test_msg("Couldn't create a qgroup %d\n", ret); + return ret; + } + + ret = btrfs_qgroup_record_ref(&trans, fs_info, 5, 4096, 4096, + BTRFS_QGROUP_OPER_ADD_EXCL, 0); + if (ret) { + test_msg("Couldn't add space to a qgroup %d\n", ret); + return ret; + } + + ret = insert_normal_tree_ref(root, 4096, 4096, 0, 5); + if (ret) + return ret; + + ret = btrfs_delayed_qgroup_accounting(&trans, fs_info); + if (ret) { + test_msg("Delayed qgroup accounting failed %d\n", ret); + return ret; + } + + if (btrfs_verify_qgroup_counts(fs_info, 5, 4096, 4096)) { + test_msg("Qgroup counts didn't match expected values\n"); + return -EINVAL; + } + + ret = remove_extent_item(root, 4096, 4096); + if (ret) + return -EINVAL; + + ret = btrfs_qgroup_record_ref(&trans, fs_info, 5, 4096, 4096, + BTRFS_QGROUP_OPER_SUB_EXCL, 0); + if (ret) { + test_msg("Couldn't remove space from the qgroup %d\n", ret); + return -EINVAL; + } + + ret = btrfs_delayed_qgroup_accounting(&trans, fs_info); + if (ret) { + test_msg("Qgroup accounting failed %d\n", ret); + return -EINVAL; + } + + if (btrfs_verify_qgroup_counts(fs_info, 5, 0, 0)) { + test_msg("Qgroup counts didn't match expected values\n"); + return -EINVAL; + } + + return 0; +} + +/* + * Add a ref for two different roots to make sure the shared value comes out + * right, also remove one of the roots and make sure the exclusive count is + * adjusted properly. + */ +static int test_multiple_refs(struct btrfs_root *root) +{ + struct btrfs_trans_handle trans; + struct btrfs_fs_info *fs_info = root->fs_info; + int ret; + + init_dummy_trans(&trans); + + test_msg("Qgroup multiple refs test\n"); + + /* We have 5 created already from the previous test */ + ret = btrfs_create_qgroup(NULL, fs_info, 256, NULL); + if (ret) { + test_msg("Couldn't create a qgroup %d\n", ret); + return ret; + } + + ret = insert_normal_tree_ref(root, 4096, 4096, 0, 5); + if (ret) + return ret; + + ret = btrfs_qgroup_record_ref(&trans, fs_info, 5, 4096, 4096, + BTRFS_QGROUP_OPER_ADD_EXCL, 0); + if (ret) { + test_msg("Couldn't add space to a qgroup %d\n", ret); + return ret; + } + + ret = btrfs_delayed_qgroup_accounting(&trans, fs_info); + if (ret) { + test_msg("Delayed qgroup accounting failed %d\n", ret); + return ret; + } + + if (btrfs_verify_qgroup_counts(fs_info, 5, 4096, 4096)) { + test_msg("Qgroup counts didn't match expected values\n"); + return -EINVAL; + } + + ret = add_tree_ref(root, 4096, 4096, 0, 256); + if (ret) + return ret; + + ret = btrfs_qgroup_record_ref(&trans, fs_info, 256, 4096, 4096, + BTRFS_QGROUP_OPER_ADD_SHARED, 0); + if (ret) { + test_msg("Qgroup record ref failed %d\n", ret); + return ret; + } + + ret = btrfs_delayed_qgroup_accounting(&trans, fs_info); + if (ret) { + test_msg("Qgroup accounting failed %d\n", ret); + return ret; + } + + if (btrfs_verify_qgroup_counts(fs_info, 5, 4096, 0)) { + test_msg("Qgroup counts didn't match expected values\n"); + return -EINVAL; + } + + if (btrfs_verify_qgroup_counts(fs_info, 256, 4096, 0)) { + test_msg("Qgroup counts didn't match expected values\n"); + return -EINVAL; + } + + ret = remove_extent_ref(root, 4096, 4096, 0, 256); + if (ret) + return ret; + + ret = btrfs_qgroup_record_ref(&trans, fs_info, 256, 4096, 4096, + BTRFS_QGROUP_OPER_SUB_SHARED, 0); + if (ret) { + test_msg("Qgroup record ref failed %d\n", ret); + return ret; + } + + ret = btrfs_delayed_qgroup_accounting(&trans, fs_info); + if (ret) { + test_msg("Qgroup accounting failed %d\n", ret); + return ret; + } + + if (btrfs_verify_qgroup_counts(fs_info, 256, 0, 0)) { + test_msg("Qgroup counts didn't match expected values\n"); + return -EINVAL; + } + + if (btrfs_verify_qgroup_counts(fs_info, 5, 4096, 4096)) { + test_msg("Qgroup counts didn't match expected values\n"); + return -EINVAL; + } + + return 0; +} + +int btrfs_test_qgroups(void) +{ + struct btrfs_root *root; + struct btrfs_root *tmp_root; + int ret = 0; + + root = btrfs_alloc_dummy_root(); + if (IS_ERR(root)) { + test_msg("Couldn't allocate root\n"); + return PTR_ERR(root); + } + + root->fs_info = btrfs_alloc_dummy_fs_info(); + if (!root->fs_info) { + test_msg("Couldn't allocate dummy fs info\n"); + ret = -ENOMEM; + goto out; + } + + /* + * Can't use bytenr 0, some things freak out + * *cough*backref walking code*cough* + */ + root->node = alloc_test_extent_buffer(root->fs_info, 4096, 4096); + if (!root->node) { + test_msg("Couldn't allocate dummy buffer\n"); + ret = -ENOMEM; + goto out; + } + root->alloc_bytenr += 8192; + + tmp_root = btrfs_alloc_dummy_root(); + if (IS_ERR(tmp_root)) { + test_msg("Couldn't allocate a fs root\n"); + ret = PTR_ERR(tmp_root); + goto out; + } + + tmp_root->root_key.objectid = 5; + root->fs_info->fs_root = tmp_root; + ret = btrfs_insert_fs_root(root->fs_info, tmp_root); + if (ret) { + test_msg("Couldn't insert fs root %d\n", ret); + goto out; + } + + tmp_root = btrfs_alloc_dummy_root(); + if (IS_ERR(tmp_root)) { + test_msg("Couldn't allocate a fs root\n"); + ret = PTR_ERR(tmp_root); + goto out; + } + + tmp_root->root_key.objectid = 256; + ret = btrfs_insert_fs_root(root->fs_info, tmp_root); + if (ret) { + test_msg("Couldn't insert fs root %d\n", ret); + goto out; + } + + /* We are using this root as our extent root */ + root->fs_info->extent_root = root; + + /* + * Some of the paths we test assume we have a filled out fs_info, so we + * just need to addt he root in there so we don't panic. + */ + root->fs_info->tree_root = root; + root->fs_info->quota_root = root; + root->fs_info->quota_enabled = 1; + + test_msg("Running qgroup tests\n"); + ret = test_no_shared_qgroup(root); + if (ret) + goto out; + ret = test_multiple_refs(root); +out: + btrfs_free_dummy_root(root); + return ret; +} diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h index b57b924e8e03..7dd558ed0716 100644 --- a/fs/btrfs/transaction.h +++ b/fs/btrfs/transaction.h @@ -69,6 +69,7 @@ struct btrfs_transaction { #define __TRANS_ATTACH (1U << 10) #define __TRANS_JOIN (1U << 11) #define __TRANS_JOIN_NOLOCK (1U << 12) +#define __TRANS_DUMMY (1U << 13) #define TRANS_USERSPACE (__TRANS_USERSPACE | __TRANS_FREEZABLE) #define TRANS_START (__TRANS_START | __TRANS_FREEZABLE) From 45ff35d6b9f301cbfb6b1703466ebe4431c7567f Mon Sep 17 00:00:00 2001 From: Guangliang Zhao Date: Mon, 12 May 2014 11:04:33 +0800 Subject: [PATCH 42/80] Btrfs: remove OPT_acl parse when acl disabled Even CONFIG_BTRFS_FS_POSIX_ACL is not defined, the acl still could been enabled using a mount option, and now fs/btrfs/acl.o is not built, so the mount options will appear to be supported but will be silently ignored. Signed-off-by: Guangliang Zhao Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/super.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index d80a2786f563..77fcb5706700 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -580,8 +580,15 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) } break; case Opt_acl: +#ifdef CONFIG_BTRFS_FS_POSIX_ACL root->fs_info->sb->s_flags |= MS_POSIXACL; break; +#else + btrfs_err(root->fs_info, + "support for ACL not compiled in!"); + ret = -EINVAL; + goto out; +#endif case Opt_noacl: root->fs_info->sb->s_flags &= ~MS_POSIXACL; break; From 298658414a2f0bea1f05a81876a45c1cd96aa2e0 Mon Sep 17 00:00:00 2001 From: Wang Shilong Date: Tue, 13 May 2014 17:05:06 +0800 Subject: [PATCH 43/80] Btrfs: set right total device count for seeding support Seeding device support allows us to create a new filesystem based on existed filesystem. However newly created filesystem's @total_devices should include seed devices. This patch fix the following problem: # mkfs.btrfs -f /dev/sdb # btrfstune -S 1 /dev/sdb # mount /dev/sdb /mnt # btrfs device add -f /dev/sdc /mnt --->fs_devices->total_devices = 1 # umount /mnt # mount /dev/sdc /mnt --->fs_devices->total_devices = 2 This is because we record right @total_devices in superblock, but @fs_devices->total_devices is reset to be 0 in btrfs_prepare_sprout(). Fix this problem by not resetting @fs_devices->total_devices. Signed-off-by: Wang Shilong Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 1 - 1 file changed, 1 deletion(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 018cde61cf9d..be8d6714494e 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1934,7 +1934,6 @@ static int btrfs_prepare_sprout(struct btrfs_root *root) fs_devices->seeding = 0; fs_devices->num_devices = 0; fs_devices->open_devices = 0; - fs_devices->total_devices = 0; fs_devices->seed = seed_devices; generate_random_uuid(fs_devices->fsid); From 58dfae6365666e26cb2d172dc437773120fc4e1a Mon Sep 17 00:00:00 2001 From: ZhangZhen Date: Tue, 13 May 2014 16:36:08 +0800 Subject: [PATCH 44/80] btrfs: replace simple_strtoull() with kstrtoull() use the newer and more pleasant kstrtoull() to replace simple_strtoull(), because simple_strtoull() is marked for obsoletion. Signed-off-by: Zhang Zhen Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index a21a4ac537b7..fba7a004e7e5 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1503,11 +1503,12 @@ static noinline int btrfs_ioctl_resize(struct file *file, sizestr = vol_args->name; devstr = strchr(sizestr, ':'); if (devstr) { - char *end; sizestr = devstr + 1; *devstr = '\0'; devstr = vol_args->name; - devid = simple_strtoull(devstr, &end, 10); + ret = kstrtoull(devstr, 10, &devid); + if (ret) + goto out_free; if (!devid) { ret = -EINVAL; goto out_free; From 7d78874273463a784759916fc3e0b4e2eb141c70 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Wed, 21 May 2014 05:49:54 -0700 Subject: [PATCH 45/80] Btrfs: fix double free in find_lock_delalloc_range We need to NULL the cached_state after freeing it, otherwise we might free it again if find_delalloc_range doesn't find anything. Signed-off-by: Chris Mason cc: stable@vger.kernel.org --- fs/btrfs/extent_io.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 8285ed0464fa..0f425dea4523 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -1693,6 +1693,7 @@ again: * shortening the size of the delalloc range we're searching */ free_extent_state(cached_state); + cached_state = NULL; if (!loops) { max_bytes = PAGE_CACHE_SIZE; loops = 1; From 351fd3532159441e810d458a5b681090ff8449fd Mon Sep 17 00:00:00 2001 From: David Sterba Date: Thu, 15 May 2014 16:48:20 +0200 Subject: [PATCH 46/80] btrfs: remove stale newlines from log messages I've noticed an extra line after "use no compression", but search revealed much more in messages of more critical levels and rare errors. Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 +- fs/btrfs/inode.c | 4 ++-- fs/btrfs/ordered-data.c | 2 +- fs/btrfs/relocation.c | 4 ++-- fs/btrfs/send.c | 4 ++-- fs/btrfs/super.c | 2 +- fs/btrfs/volumes.c | 10 +++++----- 7 files changed, 14 insertions(+), 14 deletions(-) diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index ddf16bfb9270..bb5b3067ddc3 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5932,7 +5932,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans, refs = btrfs_extent_refs(leaf, ei); if (refs < refs_to_drop) { btrfs_err(info, "trying to drop %d refs but we only have %Lu " - "for bytenr %Lu\n", refs_to_drop, refs, bytenr); + "for bytenr %Lu", refs_to_drop, refs, bytenr); ret = -EINVAL; btrfs_abort_transaction(trans, extent_root, ret); goto out; diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 95ac6b4f2390..0b8ce3002cfe 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -3475,7 +3475,7 @@ cache_acl: ret = btrfs_load_inode_props(inode, path); if (ret) btrfs_err(root->fs_info, - "error loading props for ino %llu (root %llu): %d\n", + "error loading props for ino %llu (root %llu): %d", btrfs_ino(inode), root->root_key.objectid, ret); } @@ -8010,7 +8010,7 @@ int btrfs_create_subvol_root(struct btrfs_trans_handle *trans, err = btrfs_subvol_inherit_props(trans, new_root, parent_root); if (err) btrfs_err(new_root->fs_info, - "error inheriting subvolume %llu properties: %d\n", + "error inheriting subvolume %llu properties: %d", new_root->root_key.objectid, err); err = btrfs_update_inode(trans, new_root, inode); diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c index a94b05f72869..e12441c7cf1d 100644 --- a/fs/btrfs/ordered-data.c +++ b/fs/btrfs/ordered-data.c @@ -67,7 +67,7 @@ static void ordered_data_tree_panic(struct inode *inode, int errno, { struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); btrfs_panic(fs_info, errno, "Inconsistency in ordered tree at offset " - "%llu\n", offset); + "%llu", offset); } /* diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index b9cf0f522513..65245a07275b 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -337,7 +337,7 @@ static void backref_tree_panic(struct rb_node *rb_node, int errno, u64 bytenr) if (bnode->root) fs_info = bnode->root->fs_info; btrfs_panic(fs_info, errno, "Inconsistency in backref cache " - "found at offset %llu\n", bytenr); + "found at offset %llu", bytenr); } /* @@ -1259,7 +1259,7 @@ static int __must_check __add_reloc_root(struct btrfs_root *root) if (rb_node) { btrfs_panic(root->fs_info, -EEXIST, "Duplicate root found " "for start=%llu while inserting into relocation " - "tree\n", node->bytenr); + "tree", node->bytenr); kfree(node); return -EEXIST; } diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index 3f14b31c0c96..a86c049932f9 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -1349,7 +1349,7 @@ static int find_extent_clone(struct send_ctx *sctx, ret = -EIO; btrfs_err(sctx->send_root->fs_info, "did not find backref in " "send_root. inode=%llu, offset=%llu, " - "disk_byte=%llu found extent=%llu\n", + "disk_byte=%llu found extent=%llu", ino, data_offset, disk_byte, found_key.objectid); goto out; } @@ -5472,7 +5472,7 @@ static void btrfs_root_dec_send_in_progress(struct btrfs_root* root) */ if (root->send_in_progress < 0) btrfs_err(root->fs_info, - "send_in_progres unbalanced %d root %llu\n", + "send_in_progres unbalanced %d root %llu", root->send_in_progress, root->root_key.objectid); spin_unlock(&root->root_item_lock); } diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 77fcb5706700..4662d92a4b73 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -511,7 +511,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) } else if (compress) { if (!btrfs_test_opt(root, COMPRESS)) btrfs_info(root->fs_info, - "btrfs: use %s compression\n", + "btrfs: use %s compression", compress_type); } break; diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index be8d6714494e..7d725a9ce670 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -4133,7 +4133,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, if (!devs_max) devs_max = BTRFS_MAX_DEVS_SYS_CHUNK; } else { - btrfs_err(info, "invalid chunk type 0x%llx requested\n", + btrfs_err(info, "invalid chunk type 0x%llx requested", type); BUG_ON(1); } @@ -4381,7 +4381,7 @@ int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans, if (em->start != chunk_offset || em->len != chunk_size) { btrfs_crit(extent_root->fs_info, "found a bad mapping, wanted" - " %Lu-%Lu, found %Lu-%Lu\n", chunk_offset, + " %Lu-%Lu, found %Lu-%Lu", chunk_offset, chunk_size, em->start, em->len); free_extent_map(em); return -EINVAL; @@ -4583,14 +4583,14 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len) * and exit, so return 1 so the callers don't try to use other copies. */ if (!em) { - btrfs_crit(fs_info, "No mapping for %Lu-%Lu\n", logical, + btrfs_crit(fs_info, "No mapping for %Lu-%Lu", logical, logical+len); return 1; } if (em->start > logical || em->start + em->len < logical) { btrfs_crit(fs_info, "Invalid mapping for %Lu-%Lu, got " - "%Lu-%Lu\n", logical, logical+len, em->start, + "%Lu-%Lu", logical, logical+len, em->start, em->start + em->len); free_extent_map(em); return 1; @@ -4771,7 +4771,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, if (em->start > logical || em->start + em->len < logical) { btrfs_crit(fs_info, "found a bad mapping, wanted %Lu, " - "found %Lu-%Lu\n", logical, em->start, + "found %Lu-%Lu", logical, em->start, em->start + em->len); free_extent_map(em); return -EINVAL; From 67a77eb147a36fdc346f1b9ff84649209d51f1da Mon Sep 17 00:00:00 2001 From: David Sterba Date: Fri, 16 May 2014 15:15:45 +0200 Subject: [PATCH 47/80] btrfs: remove newline from inode cache kthread name Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/inode-map.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c index 86935f5ae291..888fbe19079f 100644 --- a/fs/btrfs/inode-map.c +++ b/fs/btrfs/inode-map.c @@ -174,7 +174,7 @@ static void start_caching(struct btrfs_root *root) BTRFS_LAST_FREE_OBJECTID - objectid + 1); } - tsk = kthread_run(caching_kthread, root, "btrfs-ino-cache-%llu\n", + tsk = kthread_run(caching_kthread, root, "btrfs-ino-cache-%llu", root->root_key.objectid); if (IS_ERR(tsk)) { btrfs_warn(root->fs_info, "failed to start inode caching task"); From 402a0f475938f26b0c7b3175c8163b91a12a8a9a Mon Sep 17 00:00:00 2001 From: Anand Jain Date: Tue, 20 May 2014 14:38:11 +0800 Subject: [PATCH 48/80] btrfs: usage error should not be logged into system log I have an opinion that system logs /var/log/messages are valuable info to investigate the real system issues at the data center. People handling data center issues do spend a lot time and efforts analyzing messages files. Having usage error logged into /var/log/messages is something we should avoid. Signed-off-by: Anand Jain Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/sysfs.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/fs/btrfs/sysfs.c b/fs/btrfs/sysfs.c index ba2a645dee07..c0dfda5644f2 100644 --- a/fs/btrfs/sysfs.c +++ b/fs/btrfs/sysfs.c @@ -374,11 +374,8 @@ static ssize_t btrfs_label_store(struct kobject *kobj, struct btrfs_root *root = fs_info->fs_root; int ret; - if (len >= BTRFS_LABEL_SIZE) { - pr_err("BTRFS: unable to set label with more than %d bytes\n", - BTRFS_LABEL_SIZE - 1); + if (len >= BTRFS_LABEL_SIZE) return -EINVAL; - } trans = btrfs_start_transaction(root, 0); if (IS_ERR(trans)) From 2a10840945a86dfa2356bce526ae78cd4c1a356e Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 20 May 2014 09:23:31 -0400 Subject: [PATCH 49/80] Btrfs: free tmp ulist for qgroup rescan Memory leaks are bad mmkay? Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/qgroup.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c index 73f706c9a342..cf5aead95a7f 100644 --- a/fs/btrfs/qgroup.c +++ b/fs/btrfs/qgroup.c @@ -2478,6 +2478,7 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work) out: kfree(scratch_leaf); ulist_free(qgroups); + ulist_free(tmp); btrfs_free_path(path); mutex_lock(&fs_info->qgroup_rescan_lock); From d4452bc526c431a882cc6ba854619e166cb7dfe4 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 19 May 2014 20:47:56 -0700 Subject: [PATCH 50/80] Btrfs: break up __btrfs_write_out_cache to cut down stack usage __btrfs_write_out_cache was one of our stack pigs. This breaks it up into helper functions and slims it down to 194 bytes. Signed-off-by: Chris Mason --- fs/btrfs/free-space-cache.c | 314 ++++++++++++++++++++++-------------- 1 file changed, 194 insertions(+), 120 deletions(-) diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index a6bd654dcd47..372b05ff1943 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -851,90 +851,44 @@ out: return ret; } -/** - * __btrfs_write_out_cache - write out cached info to an inode - * @root - the root the inode belongs to - * @ctl - the free space cache we are going to write out - * @block_group - the block_group for this cache if it belongs to a block_group - * @trans - the trans handle - * @path - the path to use - * @offset - the offset for the key we'll insert - * - * This function writes out a free space cache struct to disk for quick recovery - * on mount. This will return 0 if it was successfull in writing the cache out, - * and -1 if it was not. - */ -static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, - struct btrfs_free_space_ctl *ctl, - struct btrfs_block_group_cache *block_group, - struct btrfs_trans_handle *trans, - struct btrfs_path *path, u64 offset) +static noinline_for_stack +int write_cache_extent_entries(struct io_ctl *io_ctl, + struct btrfs_free_space_ctl *ctl, + struct btrfs_block_group_cache *block_group, + int *entries, int *bitmaps, + struct list_head *bitmap_list) { - struct btrfs_free_space_header *header; - struct extent_buffer *leaf; - struct rb_node *node; - struct list_head *pos, *n; - struct extent_state *cached_state = NULL; - struct btrfs_free_cluster *cluster = NULL; - struct extent_io_tree *unpin = NULL; - struct io_ctl io_ctl; - struct list_head bitmap_list; - struct btrfs_key key; - u64 start, extent_start, extent_end, len; - int entries = 0; - int bitmaps = 0; int ret; - int err = -1; - - INIT_LIST_HEAD(&bitmap_list); - - if (!i_size_read(inode)) - return -1; - - ret = io_ctl_init(&io_ctl, inode, root); - if (ret) - return -1; + struct btrfs_free_cluster *cluster = NULL; + struct rb_node *node = rb_first(&ctl->free_space_offset); /* Get the cluster for this block_group if it exists */ - if (block_group && !list_empty(&block_group->cluster_list)) + if (block_group && !list_empty(&block_group->cluster_list)) { cluster = list_entry(block_group->cluster_list.next, struct btrfs_free_cluster, block_group_list); + } - /* Lock all pages first so we can lock the extent safely. */ - io_ctl_prepare_pages(&io_ctl, inode, 0); - - lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1, - 0, &cached_state); - - node = rb_first(&ctl->free_space_offset); if (!node && cluster) { node = rb_first(&cluster->root); cluster = NULL; } - /* Make sure we can fit our crcs into the first page */ - if (io_ctl.check_crcs && - (io_ctl.num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE) - goto out_nospc; - - io_ctl_set_generation(&io_ctl, trans->transid); - /* Write out the extent entries */ while (node) { struct btrfs_free_space *e; e = rb_entry(node, struct btrfs_free_space, offset_index); - entries++; + *entries += 1; - ret = io_ctl_add_entry(&io_ctl, e->offset, e->bytes, + ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes, e->bitmap); if (ret) - goto out_nospc; + goto fail; if (e->bitmap) { - list_add_tail(&e->list, &bitmap_list); - bitmaps++; + list_add_tail(&e->list, bitmap_list); + *bitmaps += 1; } node = rb_next(node); if (!node && cluster) { @@ -942,13 +896,84 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, cluster = NULL; } } + return 0; +fail: + return -ENOSPC; +} + +static noinline_for_stack int +update_cache_item(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct inode *inode, + struct btrfs_path *path, u64 offset, + int entries, int bitmaps) +{ + struct btrfs_key key; + struct btrfs_free_space_header *header; + struct extent_buffer *leaf; + int ret; + + key.objectid = BTRFS_FREE_SPACE_OBJECTID; + key.offset = offset; + key.type = 0; + + ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + if (ret < 0) { + clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1, + EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL, + GFP_NOFS); + goto fail; + } + leaf = path->nodes[0]; + if (ret > 0) { + struct btrfs_key found_key; + ASSERT(path->slots[0]); + path->slots[0]--; + btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); + if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID || + found_key.offset != offset) { + clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, + inode->i_size - 1, + EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, + NULL, GFP_NOFS); + btrfs_release_path(path); + goto fail; + } + } + + BTRFS_I(inode)->generation = trans->transid; + header = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_free_space_header); + btrfs_set_free_space_entries(leaf, header, entries); + btrfs_set_free_space_bitmaps(leaf, header, bitmaps); + btrfs_set_free_space_generation(leaf, header, trans->transid); + btrfs_mark_buffer_dirty(leaf); + btrfs_release_path(path); + + return 0; + +fail: + return -1; +} + +static noinline_for_stack int +add_ioctl_entries(struct btrfs_root *root, + struct inode *inode, + struct btrfs_block_group_cache *block_group, + struct io_ctl *io_ctl, + struct extent_state **cached_state, + struct list_head *bitmap_list, + int *entries) +{ + u64 start, extent_start, extent_end, len; + struct list_head *pos, *n; + struct extent_io_tree *unpin = NULL; + int ret; /* * We want to add any pinned extents to our free space cache * so we don't leak the space - */ - - /* + * * We shouldn't have switched the pinned extents yet so this is the * right one */ @@ -977,8 +1002,8 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, block_group->key.offset, extent_end + 1); len = extent_end - extent_start; - entries++; - ret = io_ctl_add_entry(&io_ctl, extent_start, len, NULL); + *entries += 1; + ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL); if (ret) goto out_nospc; @@ -986,74 +1011,129 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, } /* Write out the bitmaps */ - list_for_each_safe(pos, n, &bitmap_list) { + list_for_each_safe(pos, n, bitmap_list) { struct btrfs_free_space *entry = list_entry(pos, struct btrfs_free_space, list); - ret = io_ctl_add_bitmap(&io_ctl, entry->bitmap); + ret = io_ctl_add_bitmap(io_ctl, entry->bitmap); if (ret) goto out_nospc; list_del_init(&entry->list); } /* Zero out the rest of the pages just to make sure */ - io_ctl_zero_remaining_pages(&io_ctl); + io_ctl_zero_remaining_pages(io_ctl); - ret = btrfs_dirty_pages(root, inode, io_ctl.pages, io_ctl.num_pages, - 0, i_size_read(inode), &cached_state); - io_ctl_drop_pages(&io_ctl); + ret = btrfs_dirty_pages(root, inode, io_ctl->pages, io_ctl->num_pages, + 0, i_size_read(inode), cached_state); + io_ctl_drop_pages(io_ctl); unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0, - i_size_read(inode) - 1, &cached_state, GFP_NOFS); + i_size_read(inode) - 1, cached_state, GFP_NOFS); if (ret) - goto out; + goto fail; ret = btrfs_wait_ordered_range(inode, 0, (u64)-1); if (ret) { clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1, EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL, GFP_NOFS); + goto fail; + } + return 0; + +fail: + return -1; + +out_nospc: + return -ENOSPC; +} + +static void noinline_for_stack +cleanup_write_cache_enospc(struct inode *inode, + struct io_ctl *io_ctl, + struct extent_state **cached_state, + struct list_head *bitmap_list) +{ + struct list_head *pos, *n; + list_for_each_safe(pos, n, bitmap_list) { + struct btrfs_free_space *entry = + list_entry(pos, struct btrfs_free_space, list); + list_del_init(&entry->list); + } + io_ctl_drop_pages(io_ctl); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0, + i_size_read(inode) - 1, cached_state, + GFP_NOFS); +} + +/** + * __btrfs_write_out_cache - write out cached info to an inode + * @root - the root the inode belongs to + * @ctl - the free space cache we are going to write out + * @block_group - the block_group for this cache if it belongs to a block_group + * @trans - the trans handle + * @path - the path to use + * @offset - the offset for the key we'll insert + * + * This function writes out a free space cache struct to disk for quick recovery + * on mount. This will return 0 if it was successfull in writing the cache out, + * and -1 if it was not. + */ +static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode, + struct btrfs_free_space_ctl *ctl, + struct btrfs_block_group_cache *block_group, + struct btrfs_trans_handle *trans, + struct btrfs_path *path, u64 offset) +{ + struct extent_state *cached_state = NULL; + struct io_ctl io_ctl; + struct list_head bitmap_list; + int entries = 0; + int bitmaps = 0; + int ret; + int err = -1; + + INIT_LIST_HEAD(&bitmap_list); + + if (!i_size_read(inode)) + return -1; + + ret = io_ctl_init(&io_ctl, inode, root); + if (ret) + return -1; + + /* Lock all pages first so we can lock the extent safely. */ + io_ctl_prepare_pages(&io_ctl, inode, 0); + + lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1, + 0, &cached_state); + + + /* Make sure we can fit our crcs into the first page */ + if (io_ctl.check_crcs && + (io_ctl.num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE) + goto out_nospc; + + io_ctl_set_generation(&io_ctl, trans->transid); + + ret = write_cache_extent_entries(&io_ctl, ctl, + block_group, &entries, &bitmaps, + &bitmap_list); + if (ret) + goto out_nospc; + + ret = add_ioctl_entries(root, inode, block_group, &io_ctl, + &cached_state, &bitmap_list, &entries); + + if (ret == -ENOSPC) + goto out_nospc; + else if (ret) goto out; - } - key.objectid = BTRFS_FREE_SPACE_OBJECTID; - key.offset = offset; - key.type = 0; + err = update_cache_item(trans, root, inode, path, offset, + entries, bitmaps); - ret = btrfs_search_slot(trans, root, &key, path, 0, 1); - if (ret < 0) { - clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1, - EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL, - GFP_NOFS); - goto out; - } - leaf = path->nodes[0]; - if (ret > 0) { - struct btrfs_key found_key; - ASSERT(path->slots[0]); - path->slots[0]--; - btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); - if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID || - found_key.offset != offset) { - clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, - inode->i_size - 1, - EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, - NULL, GFP_NOFS); - btrfs_release_path(path); - goto out; - } - } - - BTRFS_I(inode)->generation = trans->transid; - header = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_free_space_header); - btrfs_set_free_space_entries(leaf, header, entries); - btrfs_set_free_space_bitmaps(leaf, header, bitmaps); - btrfs_set_free_space_generation(leaf, header, trans->transid); - btrfs_mark_buffer_dirty(leaf); - btrfs_release_path(path); - - err = 0; out: io_ctl_free(&io_ctl); if (err) { @@ -1064,14 +1144,8 @@ out: return err; out_nospc: - list_for_each_safe(pos, n, &bitmap_list) { - struct btrfs_free_space *entry = - list_entry(pos, struct btrfs_free_space, list); - list_del_init(&entry->list); - } - io_ctl_drop_pages(&io_ctl); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0, - i_size_read(inode) - 1, &cached_state, GFP_NOFS); + + cleanup_write_cache_enospc(inode, &io_ctl, &cached_state, &bitmap_list); goto out; } From 0e378df15cd87f540f1ba9503e4aa039e1c72741 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 19 May 2014 20:55:27 -0700 Subject: [PATCH 51/80] Btrfs: cut down stack usage in btree_write_cache_pages This adds noinline_for_stack to two helpers used by btree_write_cache_pages. It shaves us down from 424 bytes on the stack to 280. Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 0f425dea4523..51299c261d56 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -3393,9 +3393,10 @@ void wait_on_extent_buffer_writeback(struct extent_buffer *eb) TASK_UNINTERRUPTIBLE); } -static int lock_extent_buffer_for_io(struct extent_buffer *eb, - struct btrfs_fs_info *fs_info, - struct extent_page_data *epd) +static noinline_for_stack int +lock_extent_buffer_for_io(struct extent_buffer *eb, + struct btrfs_fs_info *fs_info, + struct extent_page_data *epd) { unsigned long i, num_pages; int flush = 0; @@ -3500,7 +3501,7 @@ static void end_bio_extent_buffer_writepage(struct bio *bio, int err) bio_put(bio); } -static int write_one_eb(struct extent_buffer *eb, +static noinline_for_stack int write_one_eb(struct extent_buffer *eb, struct btrfs_fs_info *fs_info, struct writeback_control *wbc, struct extent_page_data *epd) From fc4adbff823f76577ece26dcb88bf6f8392dbd43 Mon Sep 17 00:00:00 2001 From: Alex Gartrell Date: Tue, 20 May 2014 13:07:56 -0700 Subject: [PATCH 52/80] btrfs: Drop EXTENT_UPTODATE check in hole punching and direct locking In these instances, we are trying to determine if a page has been accessed since we began the operation for the sake of retry. This is easily accomplished by doing a gang lookup in the page mapping radix tree, and it saves us the dependency on the flag (so that we might eventually delete it). btrfs_page_exists_in_range borrows heavily from find_get_page, replacing the radix tree look up with a gang lookup of 1, so that we can find the next highest page >= index and see if it falls into our lock range. Signed-off-by: Chris Mason Signed-off-by: Alex Gartrell --- fs/btrfs/btrfs_inode.h | 2 ++ fs/btrfs/file.c | 4 +-- fs/btrfs/inode.c | 72 +++++++++++++++++++++++++++++++++++++++--- 3 files changed, 71 insertions(+), 7 deletions(-) diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h index c9a24444ec9a..a0cf3e56fe20 100644 --- a/fs/btrfs/btrfs_inode.h +++ b/fs/btrfs/btrfs_inode.h @@ -284,4 +284,6 @@ static inline void btrfs_inode_resume_unlocked_dio(struct inode *inode) &BTRFS_I(inode)->runtime_flags); } +bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end); + #endif diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 8accf94ef220..e46bfaf6cde2 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -2266,9 +2266,7 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) if ((!ordered || (ordered->file_offset + ordered->len <= lockstart || ordered->file_offset > lockend)) && - !test_range_bit(&BTRFS_I(inode)->io_tree, lockstart, - lockend, EXTENT_UPTODATE, 0, - cached_state)) { + !btrfs_page_exists_in_range(inode, lockstart, lockend)) { if (ordered) btrfs_put_ordered_extent(ordered); break; diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 0b8ce3002cfe..a462da1a3e6a 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6735,6 +6735,71 @@ out: return ret; } +bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end) +{ + struct radix_tree_root *root = &inode->i_mapping->page_tree; + int found = false; + void **pagep = NULL; + struct page *page = NULL; + int start_idx; + int end_idx; + + start_idx = start >> PAGE_CACHE_SHIFT; + + /* + * end is the last byte in the last page. end == start is legal + */ + end_idx = end >> PAGE_CACHE_SHIFT; + + rcu_read_lock(); + + /* Most of the code in this while loop is lifted from + * find_get_page. It's been modified to begin searching from a + * page and return just the first page found in that range. If the + * found idx is less than or equal to the end idx then we know that + * a page exists. If no pages are found or if those pages are + * outside of the range then we're fine (yay!) */ + while (page == NULL && + radix_tree_gang_lookup_slot(root, &pagep, NULL, start_idx, 1)) { + page = radix_tree_deref_slot(pagep); + if (unlikely(!page)) + break; + + if (radix_tree_exception(page)) { + if (radix_tree_deref_retry(page)) + continue; + /* + * Otherwise, shmem/tmpfs must be storing a swap entry + * here as an exceptional entry: so return it without + * attempting to raise page count. + */ + break; /* TODO: Is this relevant for this use case? */ + } + + if (!page_cache_get_speculative(page)) + continue; + + /* + * Has the page moved? + * This is part of the lockless pagecache protocol. See + * include/linux/pagemap.h for details. + */ + if (unlikely(page != *pagep)) { + page_cache_release(page); + page = NULL; + } + } + + if (page) { + if (page->index <= end_idx) + found = true; + page_cache_release(page); + } + + rcu_read_unlock(); + return found; +} + static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend, struct extent_state **cached_state, int writing) { @@ -6759,10 +6824,9 @@ static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend, * invalidate needs to happen so that reads after a write do not * get stale data. */ - if (!ordered && (!writing || - !test_range_bit(&BTRFS_I(inode)->io_tree, - lockstart, lockend, EXTENT_UPTODATE, 0, - *cached_state))) + if (!ordered && + (!writing || + !btrfs_page_exists_in_range(inode, lockstart, lockend))) break; unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, From 40f765805f082ed679c55bf6ab60212e55fb6fc1 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Wed, 21 May 2014 13:35:51 -0700 Subject: [PATCH 53/80] Btrfs: split up __extent_writepage to lower stack usage __extent_writepage has two unrelated parts. First it does the delayed allocation dance and second it does the mapping and IO for the page we're actually writing. This splits it up into those two parts so the stack from one doesn't impact the stack from the other. Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 330 +++++++++++++++++++++++++------------------ fs/btrfs/inode.c | 2 +- 2 files changed, 194 insertions(+), 138 deletions(-) diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 51299c261d56..0b5fa91d9a88 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -3101,143 +3101,130 @@ static noinline void update_nr_written(struct page *page, } /* - * the writepage semantics are similar to regular writepage. extent - * records are inserted to lock ranges in the tree, and as dirty areas - * are found, they are marked writeback. Then the lock bits are removed - * and the end_io handler clears the writeback ranges + * helper for __extent_writepage, doing all of the delayed allocation setup. + * + * This returns 1 if our fill_delalloc function did all the work required + * to write the page (copy into inline extent). In this case the IO has + * been started and the page is already unlocked. + * + * This returns 0 if all went well (page still locked) + * This returns < 0 if there were errors (page still locked) */ -static int __extent_writepage(struct page *page, struct writeback_control *wbc, - void *data) +static noinline_for_stack int writepage_delalloc(struct inode *inode, + struct page *page, struct writeback_control *wbc, + struct extent_page_data *epd, + u64 delalloc_start, + unsigned long *nr_written) +{ + struct extent_io_tree *tree = epd->tree; + u64 page_end = delalloc_start + PAGE_CACHE_SIZE - 1; + u64 nr_delalloc; + u64 delalloc_to_write = 0; + u64 delalloc_end = 0; + int ret; + int page_started = 0; + + if (epd->extent_locked || !tree->ops || !tree->ops->fill_delalloc) + return 0; + + while (delalloc_end < page_end) { + nr_delalloc = find_lock_delalloc_range(inode, tree, + page, + &delalloc_start, + &delalloc_end, + 128 * 1024 * 1024); + if (nr_delalloc == 0) { + delalloc_start = delalloc_end + 1; + continue; + } + ret = tree->ops->fill_delalloc(inode, page, + delalloc_start, + delalloc_end, + &page_started, + nr_written); + /* File system has been set read-only */ + if (ret) { + SetPageError(page); + /* fill_delalloc should be return < 0 for error + * but just in case, we use > 0 here meaning the + * IO is started, so we don't want to return > 0 + * unless things are going well. + */ + ret = ret < 0 ? ret : -EIO; + goto done; + } + /* + * delalloc_end is already one less than the total + * length, so we don't subtract one from + * PAGE_CACHE_SIZE + */ + delalloc_to_write += (delalloc_end - delalloc_start + + PAGE_CACHE_SIZE) >> + PAGE_CACHE_SHIFT; + delalloc_start = delalloc_end + 1; + } + if (wbc->nr_to_write < delalloc_to_write) { + int thresh = 8192; + + if (delalloc_to_write < thresh * 2) + thresh = delalloc_to_write; + wbc->nr_to_write = min_t(u64, delalloc_to_write, + thresh); + } + + /* did the fill delalloc function already unlock and start + * the IO? + */ + if (page_started) { + /* + * we've unlocked the page, so we can't update + * the mapping's writeback index, just update + * nr_to_write. + */ + wbc->nr_to_write -= *nr_written; + return 1; + } + + ret = 0; + +done: + return ret; +} + +/* + * helper for __extent_writepage. This calls the writepage start hooks, + * and does the loop to map the page into extents and bios. + * + * We return 1 if the IO is started and the page is unlocked, + * 0 if all went well (page still locked) + * < 0 if there were errors (page still locked) + */ +static noinline_for_stack int __extent_writepage_io(struct inode *inode, + struct page *page, + struct writeback_control *wbc, + struct extent_page_data *epd, + loff_t i_size, + unsigned long nr_written, + int write_flags, int *nr_ret) { - struct inode *inode = page->mapping->host; - struct extent_page_data *epd = data; struct extent_io_tree *tree = epd->tree; u64 start = page_offset(page); - u64 delalloc_start; u64 page_end = start + PAGE_CACHE_SIZE - 1; u64 end; u64 cur = start; u64 extent_offset; - u64 last_byte = i_size_read(inode); u64 block_start; u64 iosize; sector_t sector; struct extent_state *cached_state = NULL; struct extent_map *em; struct block_device *bdev; - int ret; - int nr = 0; size_t pg_offset = 0; size_t blocksize; - loff_t i_size = i_size_read(inode); - unsigned long end_index = i_size >> PAGE_CACHE_SHIFT; - u64 nr_delalloc; - u64 delalloc_end; - int page_started; - int compressed; - int write_flags; - unsigned long nr_written = 0; - bool fill_delalloc = true; + int ret = 0; + int nr = 0; + bool compressed; - if (wbc->sync_mode == WB_SYNC_ALL) - write_flags = WRITE_SYNC; - else - write_flags = WRITE; - - trace___extent_writepage(page, inode, wbc); - - WARN_ON(!PageLocked(page)); - - ClearPageError(page); - - pg_offset = i_size & (PAGE_CACHE_SIZE - 1); - if (page->index > end_index || - (page->index == end_index && !pg_offset)) { - page->mapping->a_ops->invalidatepage(page, 0, PAGE_CACHE_SIZE); - unlock_page(page); - return 0; - } - - if (page->index == end_index) { - char *userpage; - - userpage = kmap_atomic(page); - memset(userpage + pg_offset, 0, - PAGE_CACHE_SIZE - pg_offset); - kunmap_atomic(userpage); - flush_dcache_page(page); - } - pg_offset = 0; - - set_page_extent_mapped(page); - - if (!tree->ops || !tree->ops->fill_delalloc) - fill_delalloc = false; - - delalloc_start = start; - delalloc_end = 0; - page_started = 0; - if (!epd->extent_locked && fill_delalloc) { - u64 delalloc_to_write = 0; - /* - * make sure the wbc mapping index is at least updated - * to this page. - */ - update_nr_written(page, wbc, 0); - - while (delalloc_end < page_end) { - nr_delalloc = find_lock_delalloc_range(inode, tree, - page, - &delalloc_start, - &delalloc_end, - 128 * 1024 * 1024); - if (nr_delalloc == 0) { - delalloc_start = delalloc_end + 1; - continue; - } - ret = tree->ops->fill_delalloc(inode, page, - delalloc_start, - delalloc_end, - &page_started, - &nr_written); - /* File system has been set read-only */ - if (ret) { - SetPageError(page); - goto done; - } - /* - * delalloc_end is already one less than the total - * length, so we don't subtract one from - * PAGE_CACHE_SIZE - */ - delalloc_to_write += (delalloc_end - delalloc_start + - PAGE_CACHE_SIZE) >> - PAGE_CACHE_SHIFT; - delalloc_start = delalloc_end + 1; - } - if (wbc->nr_to_write < delalloc_to_write) { - int thresh = 8192; - - if (delalloc_to_write < thresh * 2) - thresh = delalloc_to_write; - wbc->nr_to_write = min_t(u64, delalloc_to_write, - thresh); - } - - /* did the fill delalloc function already unlock and start - * the IO? - */ - if (page_started) { - ret = 0; - /* - * we've unlocked the page, so we can't update - * the mapping's writeback index, just update - * nr_to_write. - */ - wbc->nr_to_write -= nr_written; - goto done_unlocked; - } - } if (tree->ops && tree->ops->writepage_start_hook) { ret = tree->ops->writepage_start_hook(page, start, page_end); @@ -3247,9 +3234,10 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, wbc->pages_skipped++; else redirty_page_for_writepage(wbc, page); + update_nr_written(page, wbc, nr_written); unlock_page(page); - ret = 0; + ret = 1; goto done_unlocked; } } @@ -3261,7 +3249,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, update_nr_written(page, wbc, nr_written + 1); end = page_end; - if (last_byte <= start) { + if (i_size <= start) { if (tree->ops && tree->ops->writepage_end_io_hook) tree->ops->writepage_end_io_hook(page, start, page_end, NULL, 1); @@ -3271,7 +3259,8 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, blocksize = inode->i_sb->s_blocksize; while (cur <= end) { - if (cur >= last_byte) { + u64 em_end; + if (cur >= i_size) { if (tree->ops && tree->ops->writepage_end_io_hook) tree->ops->writepage_end_io_hook(page, cur, page_end, NULL, 1); @@ -3286,9 +3275,10 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, } extent_offset = cur - em->start; - BUG_ON(extent_map_end(em) <= cur); + em_end = extent_map_end(em); + BUG_ON(em_end <= cur); BUG_ON(end < cur); - iosize = min(extent_map_end(em) - cur, end - cur + 1); + iosize = min(em_end - cur, end - cur + 1); iosize = ALIGN(iosize, blocksize); sector = (em->block_start + extent_offset) >> 9; bdev = em->bdev; @@ -3324,13 +3314,6 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, pg_offset += iosize; continue; } - /* leave this out until we have a page_mkwrite call */ - if (0 && !test_range_bit(tree, cur, cur + iosize - 1, - EXTENT_DIRTY, 0, NULL)) { - cur = cur + iosize; - pg_offset += iosize; - continue; - } if (tree->ops && tree->ops->writepage_io_hook) { ret = tree->ops->writepage_io_hook(page, cur, @@ -3341,7 +3324,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, if (ret) { SetPageError(page); } else { - unsigned long max_nr = end_index + 1; + unsigned long max_nr = (i_size >> PAGE_CACHE_SHIFT) + 1; set_range_writeback(tree, cur, cur + iosize - 1); if (!PageWriteback(page)) { @@ -3362,6 +3345,81 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc, pg_offset += iosize; nr++; } +done: + *nr_ret = nr; + +done_unlocked: + + /* drop our reference on any cached states */ + free_extent_state(cached_state); + return ret; +} + +/* + * the writepage semantics are similar to regular writepage. extent + * records are inserted to lock ranges in the tree, and as dirty areas + * are found, they are marked writeback. Then the lock bits are removed + * and the end_io handler clears the writeback ranges + */ +static int __extent_writepage(struct page *page, struct writeback_control *wbc, + void *data) +{ + struct inode *inode = page->mapping->host; + struct extent_page_data *epd = data; + u64 start = page_offset(page); + u64 page_end = start + PAGE_CACHE_SIZE - 1; + int ret; + int nr = 0; + size_t pg_offset = 0; + loff_t i_size = i_size_read(inode); + unsigned long end_index = i_size >> PAGE_CACHE_SHIFT; + int write_flags; + unsigned long nr_written = 0; + + if (wbc->sync_mode == WB_SYNC_ALL) + write_flags = WRITE_SYNC; + else + write_flags = WRITE; + + trace___extent_writepage(page, inode, wbc); + + WARN_ON(!PageLocked(page)); + + ClearPageError(page); + + pg_offset = i_size & (PAGE_CACHE_SIZE - 1); + if (page->index > end_index || + (page->index == end_index && !pg_offset)) { + page->mapping->a_ops->invalidatepage(page, 0, PAGE_CACHE_SIZE); + unlock_page(page); + return 0; + } + + if (page->index == end_index) { + char *userpage; + + userpage = kmap_atomic(page); + memset(userpage + pg_offset, 0, + PAGE_CACHE_SIZE - pg_offset); + kunmap_atomic(userpage); + flush_dcache_page(page); + } + + pg_offset = 0; + + set_page_extent_mapped(page); + + ret = writepage_delalloc(inode, page, wbc, epd, start, &nr_written); + if (ret == 1) + goto done_unlocked; + if (ret) + goto done; + + ret = __extent_writepage_io(inode, page, wbc, epd, + i_size, nr_written, write_flags, &nr); + if (ret == 1) + goto done_unlocked; + done: if (nr == 0) { /* make sure the mapping tag for page dirty gets cleared */ @@ -3373,12 +3431,10 @@ done: end_extent_writepage(page, ret, start, page_end); } unlock_page(page); + return ret; done_unlocked: - - /* drop our reference on any cached states */ - free_extent_state(cached_state); - return ret; + return 0; } static int eb_wait(void *word) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index a462da1a3e6a..992aae6c00b0 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -125,7 +125,7 @@ static int btrfs_init_inode_security(struct btrfs_trans_handle *trans, * the btree. The caller should have done a btrfs_drop_extents so that * no overlapping inline items exist in the btree */ -static noinline int insert_inline_extent(struct btrfs_trans_handle *trans, +static int insert_inline_extent(struct btrfs_trans_handle *trans, struct btrfs_path *path, int extent_inserted, struct btrfs_root *root, struct inode *inode, u64 start, size_t size, size_t compressed_size, From a79b7d4b3e8118f265dcb4bdf9a572c392f02708 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 22 May 2014 16:18:52 -0700 Subject: [PATCH 54/80] Btrfs: async delayed refs Delayed extent operations are triggered during transaction commits. The goal is to queue up a healthly batch of changes to the extent allocation tree and run through them in bulk. This farms them off to async helper threads. The goal is to have the bulk of the delayed operations being done in the background, but this is also important to limit our stack footprint. Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 5 +++ fs/btrfs/disk-io.c | 6 ++++ fs/btrfs/extent-tree.c | 79 ++++++++++++++++++++++++++++++++++++++++++ fs/btrfs/inode.c | 1 + fs/btrfs/transaction.c | 19 ++++++++-- 5 files changed, 107 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 4896d7a947eb..02895a126ab9 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1546,6 +1546,9 @@ struct btrfs_fs_info { */ struct btrfs_workqueue *fixup_workers; struct btrfs_workqueue *delayed_workers; + + /* the extent workers do delayed refs on the extent allocation tree */ + struct btrfs_workqueue *extent_workers; struct task_struct *transaction_kthread; struct task_struct *cleaner_kthread; int thread_pool_size; @@ -3268,6 +3271,8 @@ int btrfs_check_space_for_delayed_refs(struct btrfs_trans_handle *trans, void btrfs_put_block_group(struct btrfs_block_group_cache *cache); int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans, struct btrfs_root *root, unsigned long count); +int btrfs_async_run_delayed_refs(struct btrfs_root *root, + unsigned long count, int wait); int btrfs_lookup_extent(struct btrfs_root *root, u64 start, u64 len); int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 0c0fa78ef452..8bb4aa19898f 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2069,6 +2069,7 @@ static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info) btrfs_destroy_workqueue(fs_info->readahead_workers); btrfs_destroy_workqueue(fs_info->flush_workers); btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers); + btrfs_destroy_workqueue(fs_info->extent_workers); } static void free_root_extent_buffers(struct btrfs_root *root) @@ -2586,6 +2587,10 @@ int open_ctree(struct super_block *sb, btrfs_alloc_workqueue("readahead", flags, max_active, 2); fs_info->qgroup_rescan_workers = btrfs_alloc_workqueue("qgroup-rescan", flags, 1, 0); + fs_info->extent_workers = + btrfs_alloc_workqueue("extent-refs", flags, + min_t(u64, fs_devices->num_devices, + max_active), 8); if (!(fs_info->workers && fs_info->delalloc_workers && fs_info->submit_workers && fs_info->flush_workers && @@ -2595,6 +2600,7 @@ int open_ctree(struct super_block *sb, fs_info->endio_freespace_worker && fs_info->rmw_workers && fs_info->caching_workers && fs_info->readahead_workers && fs_info->fixup_workers && fs_info->delayed_workers && + fs_info->fixup_workers && fs_info->extent_workers && fs_info->qgroup_rescan_workers)) { err = -ENOMEM; goto fail_sb_buffer; diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index bb5b3067ddc3..6caddd5970e4 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -2674,15 +2674,94 @@ int btrfs_should_throttle_delayed_refs(struct btrfs_trans_handle *trans, u64 num_entries = atomic_read(&trans->transaction->delayed_refs.num_entries); u64 avg_runtime; + u64 val; smp_mb(); avg_runtime = fs_info->avg_delayed_ref_runtime; + val = num_entries * avg_runtime; if (num_entries * avg_runtime >= NSEC_PER_SEC) return 1; + if (val >= NSEC_PER_SEC / 2) + return 2; return btrfs_check_space_for_delayed_refs(trans, root); } +struct async_delayed_refs { + struct btrfs_root *root; + int count; + int error; + int sync; + struct completion wait; + struct btrfs_work work; +}; + +static void delayed_ref_async_start(struct btrfs_work *work) +{ + struct async_delayed_refs *async; + struct btrfs_trans_handle *trans; + int ret; + + async = container_of(work, struct async_delayed_refs, work); + + trans = btrfs_join_transaction(async->root); + if (IS_ERR(trans)) { + async->error = PTR_ERR(trans); + goto done; + } + + /* + * trans->sync means that when we call end_transaciton, we won't + * wait on delayed refs + */ + trans->sync = true; + ret = btrfs_run_delayed_refs(trans, async->root, async->count); + if (ret) + async->error = ret; + + ret = btrfs_end_transaction(trans, async->root); + if (ret && !async->error) + async->error = ret; +done: + if (async->sync) + complete(&async->wait); + else + kfree(async); +} + +int btrfs_async_run_delayed_refs(struct btrfs_root *root, + unsigned long count, int wait) +{ + struct async_delayed_refs *async; + int ret; + + async = kmalloc(sizeof(*async), GFP_NOFS); + if (!async) + return -ENOMEM; + + async->root = root->fs_info->tree_root; + async->count = count; + async->error = 0; + if (wait) + async->sync = 1; + else + async->sync = 0; + init_completion(&async->wait); + + btrfs_init_work(&async->work, delayed_ref_async_start, + NULL, NULL); + + btrfs_queue_work(root->fs_info->extent_workers, &async->work); + + if (wait) { + wait_for_completion(&async->wait); + ret = async->error; + kfree(async); + return ret; + } + return 0; +} + /* * this starts processing the delayed reference count updates and * extent insertions we have queued up so far. count can be diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 992aae6c00b0..38d1e7b976d8 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -2678,6 +2678,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent) trans = NULL; goto out_unlock; } + trans->block_rsv = &root->fs_info->delalloc_block_rsv; if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags)) diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 3aafbde8b637..1c54e2eb74ab 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -697,6 +697,7 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans, unsigned long cur = trans->delayed_ref_updates; int lock = (trans->type != TRANS_JOIN_NOLOCK); int err = 0; + int must_run_delayed_refs = 0; if (trans->use_count > 1) { trans->use_count--; @@ -711,10 +712,18 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans, btrfs_create_pending_block_groups(trans, root); trans->delayed_ref_updates = 0; - if (!trans->sync && btrfs_should_throttle_delayed_refs(trans, root)) { + if (!trans->sync) { + must_run_delayed_refs = + btrfs_should_throttle_delayed_refs(trans, root); cur = max_t(unsigned long, cur, 32); - trans->delayed_ref_updates = 0; - btrfs_run_delayed_refs(trans, root, cur); + + /* + * don't make the caller wait if they are from a NOLOCK + * or ATTACH transaction, it will deadlock with commit + */ + if (must_run_delayed_refs == 1 && + (trans->type & (__TRANS_JOIN_NOLOCK | __TRANS_ATTACH))) + must_run_delayed_refs = 2; } if (trans->qgroup_reserved) { @@ -775,6 +784,10 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans, assert_qgroups_uptodate(trans); kmem_cache_free(btrfs_trans_handle_cachep, trans); + if (must_run_delayed_refs) { + btrfs_async_run_delayed_refs(root, cur, + must_run_delayed_refs == 1); + } return err; } From 1af56070e3ef9477dbc7eba3b9ad7446979c7974 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Sun, 25 May 2014 04:49:24 +0100 Subject: [PATCH 55/80] Btrfs: send, don't error in the presence of subvols/snapshots If we are doing an incremental send and the base snapshot has a directory with name X that doesn't exist anymore in the second snapshot and a new subvolume/snapshot exists in the second snapshot that has the same name as the directory (name X), the incremental send would fail with -ENOENT error. This is because it attempts to lookup for an inode with a number matching the objectid of a root, which doesn't exist. Steps to reproduce: mkfs.btrfs -f /dev/sdd mount /dev/sdd /mnt mkdir /mnt/testdir btrfs subvolume snapshot -r /mnt /mnt/mysnap1 rmdir /mnt/testdir btrfs subvolume create /mnt/testdir btrfs subvolume snapshot -r /mnt /mnt/mysnap2 btrfs send -p /mnt/mysnap1 /mnt/mysnap2 -f /tmp/send.data A test case for xfstests follows. Reported-by: Robert White Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/send.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index a86c049932f9..15cdc67ce9ad 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -1628,6 +1628,10 @@ static int lookup_dir_item_inode(struct btrfs_root *root, goto out; } btrfs_dir_item_key_to_cpu(path->nodes[0], di, &key); + if (key.type == BTRFS_ROOT_ITEM_KEY) { + ret = -ENOENT; + goto out; + } *found_inode = key.objectid; *found_type = btrfs_dir_type(path->nodes[0], di); From 7e3ae33efad1490d01040f552ef50e58ed6376ca Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Fri, 23 May 2014 20:15:16 +0100 Subject: [PATCH 56/80] Btrfs: send, use the right limits for xattr names and values We were limiting the sum of the xattr name and value lengths to PATH_MAX, which is not correct, specially on filesystems created with btrfs-progs v3.12 or higher, where the default leaf size is max(16384, PAGE_SIZE), or systems with page sizes larger than 4096 bytes. Xattrs have their own specific maximum name and value lengths, which depend on the leaf size, therefore use these limits to be able to send xattrs with sizes larger than PATH_MAX. A test case for xfstests follows. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/send.c | 30 +++++++++++++++++++++++------- 1 file changed, 23 insertions(+), 7 deletions(-) diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index 15cdc67ce9ad..6528aa662181 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -975,7 +975,7 @@ static int iterate_dir_item(struct btrfs_root *root, struct btrfs_path *path, struct btrfs_dir_item *di; struct btrfs_key di_key; char *buf = NULL; - const int buf_len = PATH_MAX; + int buf_len; u32 name_len; u32 data_len; u32 cur; @@ -985,6 +985,11 @@ static int iterate_dir_item(struct btrfs_root *root, struct btrfs_path *path, int num; u8 type; + if (found_key->type == BTRFS_XATTR_ITEM_KEY) + buf_len = BTRFS_MAX_XATTR_SIZE(root); + else + buf_len = PATH_MAX; + buf = kmalloc(buf_len, GFP_NOFS); if (!buf) { ret = -ENOMEM; @@ -1006,12 +1011,23 @@ static int iterate_dir_item(struct btrfs_root *root, struct btrfs_path *path, type = btrfs_dir_type(eb, di); btrfs_dir_item_key_to_cpu(eb, di, &di_key); - /* - * Path too long - */ - if (name_len + data_len > buf_len) { - ret = -ENAMETOOLONG; - goto out; + if (type == BTRFS_FT_XATTR) { + if (name_len > XATTR_NAME_MAX) { + ret = -ENAMETOOLONG; + goto out; + } + if (name_len + data_len > buf_len) { + ret = -E2BIG; + goto out; + } + } else { + /* + * Path too long + */ + if (name_len + data_len > buf_len) { + ret = -ENAMETOOLONG; + goto out; + } } read_extent_buffer(eb, buf, (unsigned long)(di + 1), From c1895442be01c58449e3bf9272f22062a670e08f Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Tue, 27 May 2014 12:59:57 -0400 Subject: [PATCH 57/80] btrfs: allocate raid type kobjects dynamically We are currently allocating space_info objects in an array when we allocate space_info. When a user does something like: # btrfs balance start -mconvert=raid1 -dconvert=raid1 /mnt # btrfs balance start -mconvert=single -dconvert=single /mnt -f # btrfs balance start -mconvert=raid1 -dconvert=raid1 / We can end up with memory corruption since the kobject hasn't been reinitialized properly and the name pointer was left set. The rationale behind allocating them statically was to avoid creating a separate kobject container that just contained the raid type. It used the index in the array to determine the index. Ultimately, though, this wastes more memory than it saves in all but the most complex scenarios and introduces kobject lifetime questions. This patch allocates the kobjects dynamically instead. Note that we also remove the kobject_get/put of the parent kobject since kobject_add and kobject_del do that internally. Signed-off-by: Jeff Mahoney Reported-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 8 +++++++- fs/btrfs/extent-tree.c | 39 ++++++++++++++++++++++++++------------- fs/btrfs/sysfs.c | 5 +++-- 3 files changed, 36 insertions(+), 16 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 02895a126ab9..af523d695432 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1123,6 +1123,12 @@ struct btrfs_qgroup_limit_item { __le64 rsv_excl; } __attribute__ ((__packed__)); +/* For raid type sysfs entries */ +struct raid_kobject { + int raid_type; + struct kobject kobj; +}; + struct btrfs_space_info { spinlock_t lock; @@ -1173,7 +1179,7 @@ struct btrfs_space_info { wait_queue_head_t wait; struct kobject kobj; - struct kobject block_group_kobjs[BTRFS_NR_RAID_TYPES]; + struct kobject *block_group_kobjs[BTRFS_NR_RAID_TYPES]; }; #define BTRFS_BLOCK_RSV_GLOBAL 1 diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 6caddd5970e4..fafb3e53ecde 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3497,10 +3497,8 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags, return ret; } - for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) { + for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) INIT_LIST_HEAD(&found->block_groups[i]); - kobject_init(&found->block_group_kobjs[i], &btrfs_raid_ktype); - } init_rwsem(&found->groups_sem); spin_lock_init(&found->lock); found->flags = flags & BTRFS_BLOCK_GROUP_TYPE_MASK; @@ -8586,8 +8584,9 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info) list_del(&space_info->list); for (i = 0; i < BTRFS_NR_RAID_TYPES; i++) { struct kobject *kobj; - kobj = &space_info->block_group_kobjs[i]; - if (kobj->parent) { + kobj = space_info->block_group_kobjs[i]; + space_info->block_group_kobjs[i] = NULL; + if (kobj) { kobject_del(kobj); kobject_put(kobj); } @@ -8611,17 +8610,26 @@ static void __link_block_group(struct btrfs_space_info *space_info, up_write(&space_info->groups_sem); if (first) { - struct kobject *kobj = &space_info->block_group_kobjs[index]; + struct raid_kobject *rkobj; int ret; - kobject_get(&space_info->kobj); /* put in release */ - ret = kobject_add(kobj, &space_info->kobj, "%s", - get_raid_name(index)); + rkobj = kzalloc(sizeof(*rkobj), GFP_NOFS); + if (!rkobj) + goto out_err; + rkobj->raid_type = index; + kobject_init(&rkobj->kobj, &btrfs_raid_ktype); + ret = kobject_add(&rkobj->kobj, &space_info->kobj, + "%s", get_raid_name(index)); if (ret) { - pr_warn("BTRFS: failed to add kobject for block cache. ignoring.\n"); - kobject_put(&space_info->kobj); + kobject_put(&rkobj->kobj); + goto out_err; } + space_info->block_group_kobjs[index] = &rkobj->kobj; } + + return; +out_err: + pr_warn("BTRFS: failed to add kobject for block cache. ignoring.\n"); } static struct btrfs_block_group_cache * @@ -8956,6 +8964,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, struct btrfs_root *tree_root = root->fs_info->tree_root; struct btrfs_key key; struct inode *inode; + struct kobject *kobj = NULL; int ret; int index; int factor; @@ -9055,11 +9064,15 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, */ list_del_init(&block_group->list); if (list_empty(&block_group->space_info->block_groups[index])) { - kobject_del(&block_group->space_info->block_group_kobjs[index]); - kobject_put(&block_group->space_info->block_group_kobjs[index]); + kobj = block_group->space_info->block_group_kobjs[index]; + block_group->space_info->block_group_kobjs[index] = NULL; clear_avail_alloc_bits(root->fs_info, block_group->flags); } up_write(&block_group->space_info->groups_sem); + if (kobj) { + kobject_del(kobj); + kobject_put(kobj); + } if (block_group->cached == BTRFS_CACHE_STARTED) wait_block_group_cache_done(block_group); diff --git a/fs/btrfs/sysfs.c b/fs/btrfs/sysfs.c index c0dfda5644f2..df39458f1487 100644 --- a/fs/btrfs/sysfs.c +++ b/fs/btrfs/sysfs.c @@ -254,6 +254,7 @@ static ssize_t global_rsv_reserved_show(struct kobject *kobj, BTRFS_ATTR(global_rsv_reserved, 0444, global_rsv_reserved_show); #define to_space_info(_kobj) container_of(_kobj, struct btrfs_space_info, kobj) +#define to_raid_kobj(_kobj) container_of(_kobj, struct raid_kobject, kobj) static ssize_t raid_bytes_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf); @@ -266,7 +267,7 @@ static ssize_t raid_bytes_show(struct kobject *kobj, { struct btrfs_space_info *sinfo = to_space_info(kobj->parent); struct btrfs_block_group_cache *block_group; - int index = kobj - sinfo->block_group_kobjs; + int index = to_raid_kobj(kobj)->raid_type; u64 val = 0; down_read(&sinfo->groups_sem); @@ -288,7 +289,7 @@ static struct attribute *raid_attributes[] = { static void release_raid_kobj(struct kobject *kobj) { - kobject_put(kobj->parent); + kfree(to_raid_kobj(kobj)); } struct kobj_type btrfs_raid_ktype = { From 8321cf2596d283821acc466377c2b85bcd3422b7 Mon Sep 17 00:00:00 2001 From: Rickard Strandqvist Date: Thu, 22 May 2014 22:43:43 +0200 Subject: [PATCH 58/80] fs: btrfs: volumes.c: Fix for possible null pointer dereference There is otherwise a risk of a possible null pointer dereference. Was largely found by using a static code analysis program called cppcheck. Signed-off-by: Rickard Strandqvist Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 7d725a9ce670..ffeed6d6326f 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1690,11 +1690,12 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) struct btrfs_fs_devices *fs_devices; fs_devices = root->fs_info->fs_devices; while (fs_devices) { - if (fs_devices->seed == cur_devices) + if (fs_devices->seed == cur_devices) { + fs_devices->seed = cur_devices->seed; break; + } fs_devices = fs_devices->seed; } - fs_devices->seed = cur_devices->seed; cur_devices->seed = NULL; lock_chunks(root); __btrfs_close_devices(cur_devices); From c125b8bff1d9f6c8c91ce4eb8bd5616058c7d510 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Fri, 23 May 2014 05:03:34 +0100 Subject: [PATCH 59/80] Btrfs: ensure readers see new data after a clone operation We were cleaning the clone target file range from the page cache before we did replace the file extent items in the fs tree. This was racy, as right after cleaning the relevant range from the page cache and before replacing the file extent items, a read against that range could be performed by another task and populate again the page cache with stale data (stale after the cloning finishes). This would result in reads after the clone operation successfully finishes to get old data (and potentially for a very long time). Therefore evict the pages after replacing the file extent items, so that subsequent reads will always get the new data. Similarly, we were prone to races while cloning the file extent items because we weren't locking the target range and wait for any existing ordered extents against that range to complete. It was possible that after cloning the extent items, a write operation that was performed before the clone operation and overlaps the same range, would end up undoing all or part of the work the clone operation did (a worker task running inode.c:btrfs_finish_ordered_io). Therefore lock the target range in the io tree, wait for all pending ordered extents against that range to finish and then safely perform the cloning. The issue of reading stale data after the clone operation is easy to reproduce by running the following C program in a loop until it exits with return value 1. #include #include #include #include #include #include #include #include #include #include #include #include #include #define SRC_FILE "/mnt/sdd/foo" #define DST_FILE "/mnt/sdd/bar" #define FILE_SIZE (16 * 1024) #define PATTERN_SRC 'X' #define PATTERN_DST 'Y' struct btrfs_ioctl_clone_range_args { __s64 src_fd; __u64 src_offset, src_length; __u64 dest_offset; }; #define BTRFS_IOCTL_MAGIC 0x94 #define BTRFS_IOC_CLONE_RANGE _IOW(BTRFS_IOCTL_MAGIC, 13, \ struct btrfs_ioctl_clone_range_args) static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER; static int clone_done = 0; static int reader_ready = 0; static int stale_data = 0; static void *reader_loop(void *arg) { char buf[4096], want_buf[4096]; memset(want_buf, PATTERN_SRC, 4096); pthread_mutex_lock(&mutex); reader_ready = 1; pthread_mutex_unlock(&mutex); while (1) { int done, fd, ret; fd = open(DST_FILE, O_RDONLY); assert(fd != -1); pthread_mutex_lock(&mutex); done = clone_done; pthread_mutex_unlock(&mutex); ret = read(fd, buf, 4096); assert(ret == 4096); close(fd); if (done) { ret = memcmp(buf, want_buf, 4096); if (ret == 0) { printf("Found new content\n"); } else { printf("Found old content\n"); pthread_mutex_lock(&mutex); stale_data = 1; pthread_mutex_unlock(&mutex); } break; } } return NULL; } int main(int argc, char *argv[]) { pthread_t reader; int ret, i, fd; struct btrfs_ioctl_clone_range_args clone_args; int fd1, fd2; ret = remove(SRC_FILE); if (ret == -1 && errno != ENOENT) { fprintf(stderr, "Error deleting src file: %s\n", strerror(errno)); return 1; } ret = remove(DST_FILE); if (ret == -1 && errno != ENOENT) { fprintf(stderr, "Error deleting dst file: %s\n", strerror(errno)); return 1; } fd = open(SRC_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU); assert(fd != -1); for (i = 0; i < FILE_SIZE; i++) { char c = PATTERN_SRC; ret = write(fd, &c, 1); assert(ret == 1); } close(fd); fd = open(DST_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU); assert(fd != -1); for (i = 0; i < FILE_SIZE; i++) { char c = PATTERN_DST; ret = write(fd, &c, 1); assert(ret == 1); } close(fd); sync(); ret = pthread_create(&reader, NULL, reader_loop, NULL); assert(ret == 0); while (1) { int r; pthread_mutex_lock(&mutex); r = reader_ready; pthread_mutex_unlock(&mutex); if (r) break; } fd1 = open(SRC_FILE, O_RDONLY); if (fd1 < 0) { fprintf(stderr, "Error open src file: %s\n", strerror(errno)); return 1; } fd2 = open(DST_FILE, O_RDWR); if (fd2 < 0) { fprintf(stderr, "Error open dst file: %s\n", strerror(errno)); return 1; } clone_args.src_fd = fd1; clone_args.src_offset = 0; clone_args.src_length = 4096; clone_args.dest_offset = 0; ret = ioctl(fd2, BTRFS_IOC_CLONE_RANGE, &clone_args); assert(ret == 0); close(fd1); close(fd2); pthread_mutex_lock(&mutex); clone_done = 1; pthread_mutex_unlock(&mutex); ret = pthread_join(reader, NULL); assert(ret == 0); pthread_mutex_lock(&mutex); ret = stale_data ? 1 : 0; pthread_mutex_unlock(&mutex); return ret; } Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 36 +++++++++++++++++++++++++++++++----- 1 file changed, 31 insertions(+), 5 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index fba7a004e7e5..362720a3fea2 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -3410,15 +3410,41 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd, goto out_unlock; } - /* truncate page cache pages from target inode range */ - truncate_inode_pages_range(&inode->i_data, destoff, - PAGE_CACHE_ALIGN(destoff + len) - 1); + /* + * Lock the target range too. Right after we replace the file extent + * items in the fs tree (which now point to the cloned data), we might + * have a worker replace them with extent items relative to a write + * operation that was issued before this clone operation (i.e. confront + * with inode.c:btrfs_finish_ordered_io). + */ + if (same_inode) { + u64 lock_start = min_t(u64, off, destoff); + u64 lock_len = max_t(u64, off, destoff) + len - lock_start; - lock_extent_range(src, off, len); + lock_extent_range(src, lock_start, lock_len); + } else { + lock_extent_range(src, off, len); + lock_extent_range(inode, destoff, len); + } ret = btrfs_clone(src, inode, off, olen, len, destoff); - unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1); + if (same_inode) { + u64 lock_start = min_t(u64, off, destoff); + u64 lock_end = max_t(u64, off, destoff) + len - 1; + + unlock_extent(&BTRFS_I(src)->io_tree, lock_start, lock_end); + } else { + unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1); + unlock_extent(&BTRFS_I(inode)->io_tree, destoff, + destoff + len - 1); + } + /* + * Truncate page cache pages so that future reads will see the cloned + * data immediately and not the previous data. + */ + truncate_inode_pages_range(&inode->i_data, destoff, + PAGE_CACHE_ALIGN(destoff + len) - 1); out_unlock: if (!same_inode) { if (inode < src) { From c55bfa67e94e22ec0449fe7c55b3ef20fbe13348 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Sun, 25 May 2014 03:55:44 +0100 Subject: [PATCH 60/80] Btrfs: set dead flag on the right root when destroying snapshot We were setting the BTRFS_ROOT_SUBVOL_DEAD flag on the root of the parent of our target snapshot, instead of setting it in the target snapshot's root. This is easy to observe by running the following scenario: mkfs.btrfs -f /dev/sdd mount /dev/sdd /mnt btrfs subvolume create /mnt/first_subvol btrfs subvolume snapshot -r /mnt /mnt/mysnap1 btrfs subvolume delete /mnt/first_subvol btrfs subvolume snapshot -r /mnt /mnt/mysnap2 btrfs send -p /mnt/mysnap1 /mnt/mysnap2 -f /tmp/send.data The send command failed because the send ioctl returned -EPERM. A test case for xfstests follows. Signed-off-by: Filipe David Borba Manana Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 362720a3fea2..38f2169b73a4 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2312,16 +2312,16 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file, * again is not run concurrently. */ spin_lock(&dest->root_item_lock); - root_flags = btrfs_root_flags(&root->root_item); - if (root->send_in_progress == 0) { - btrfs_set_root_flags(&root->root_item, + root_flags = btrfs_root_flags(&dest->root_item); + if (dest->send_in_progress == 0) { + btrfs_set_root_flags(&dest->root_item, root_flags | BTRFS_ROOT_SUBVOL_DEAD); spin_unlock(&dest->root_item_lock); } else { spin_unlock(&dest->root_item_lock); btrfs_warn(root->fs_info, "Attempt to delete subvolume %llu during send", - root->root_key.objectid); + dest->root_key.objectid); err = -EPERM; goto out_dput; } @@ -2416,8 +2416,8 @@ out_up_write: out_unlock: if (err) { spin_lock(&dest->root_item_lock); - root_flags = btrfs_root_flags(&root->root_item); - btrfs_set_root_flags(&root->root_item, + root_flags = btrfs_root_flags(&dest->root_item); + btrfs_set_root_flags(&dest->root_item, root_flags & ~BTRFS_ROOT_SUBVOL_DEAD); spin_unlock(&dest->root_item_lock); } From 2c463823cb897a8e1758518c911646f5605cdd82 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Sat, 31 May 2014 02:31:05 +0100 Subject: [PATCH 61/80] Btrfs: avoid visiting all extent items when cloning a range When cloning a range of a file, we were visiting all the extent items in the btree that belong to our source inode. We don't need to visit those extent items that don't overlap the range we are cloning, as doing so only makes us waste time and do unnecessary btree navigations (btrfs_next_leaf) for inodes that have a large number of file extent items in the btree. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 26 ++++++++++++++++++++++---- 1 file changed, 22 insertions(+), 4 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 38f2169b73a4..f0b4237cedfc 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -3020,7 +3020,7 @@ static int btrfs_clone(struct inode *src, struct inode *inode, /* clone data */ key.objectid = btrfs_ino(src); key.type = BTRFS_EXTENT_DATA_KEY; - key.offset = 0; + key.offset = off; while (1) { /* @@ -3032,6 +3032,17 @@ static int btrfs_clone(struct inode *src, struct inode *inode, 0, 0); if (ret < 0) goto out; + /* + * First search, if no extent item that starts at offset off was + * found but the previous item is an extent item, it's possible + * it might overlap our target range, therefore process it. + */ + if (key.offset == off && ret > 0 && path->slots[0] > 0) { + btrfs_item_key_to_cpu(path->nodes[0], &key, + path->slots[0] - 1); + if (key.type == BTRFS_EXTENT_DATA_KEY) + path->slots[0]--; + } nritems = btrfs_header_nritems(path->nodes[0]); process_slot: @@ -3081,10 +3092,16 @@ process_slot: extent); } - if (key.offset + datal <= off || - key.offset >= off + len - 1) { + /* + * The first search might have left us at an extent + * item that ends before our target range's start, can + * happen if we have holes and NO_HOLES feature enabled. + */ + if (key.offset + datal <= off) { path->slots[0]++; goto process_slot; + } else if (key.offset >= off + len) { + break; } size = btrfs_item_size_nr(leaf, slot); @@ -3291,6 +3308,8 @@ process_slot: goto out; } ret = btrfs_end_transaction(trans, root); + if (new_key.offset + datal >= destoff + len) + break; } btrfs_release_path(path); key.offset++; @@ -3298,7 +3317,6 @@ process_slot: ret = 0; out: - btrfs_release_path(path); btrfs_free_path(path); vfree(buf); return ret; From ff5df9b884268d319d041934c6059c09733a4be2 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Fri, 30 May 2014 17:56:24 +0100 Subject: [PATCH 62/80] Btrfs: ioctl, don't re-lock extent range when not necessary In ioctl.c:lock_extent_range(), after locking our target range, the ordered extent that btrfs_lookup_first_ordered_extent() returns us may not overlap our target range at all. In this case we would just unlock our target range, wait for any new ordered extents that overlap the range to complete, lock again the range and repeat all these steps until we don't get any ordered extent and the delalloc flag isn't set in the io tree for our target range. Therefore just stop if we get an ordered extent that doesn't overlap our target range and the dealalloc flag isn't set for the range in the inode's io tree. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index f0b4237cedfc..04ece8fab921 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2700,10 +2700,15 @@ static inline void lock_extent_range(struct inode *inode, u64 off, u64 len) lock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1); ordered = btrfs_lookup_first_ordered_extent(inode, off + len - 1); - if (!ordered && + if ((!ordered || + ordered->file_offset + ordered->len <= off || + ordered->file_offset >= off + len) && !test_range_bit(&BTRFS_I(inode)->io_tree, off, - off + len - 1, EXTENT_DELALLOC, 0, NULL)) + off + len - 1, EXTENT_DELALLOC, 0, NULL)) { + if (ordered) + btrfs_put_ordered_extent(ordered); break; + } unlock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1); if (ordered) btrfs_put_ordered_extent(ordered); From 3821f348889e506efbd268cc8149e0ebfa47c4e5 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Tue, 3 Jun 2014 12:41:44 +0100 Subject: [PATCH 63/80] Btrfs: update commit root on snapshot creation after orphan cleanup On snapshot creation (either writable or read-only), we do orphan cleanup against the root of the snapshot. If the cleanup did remove any orphans, then the current root node will be different from the commit root node until the next transaction commit happens. A send operation always uses the commit root of a snapshot - this means it will see the orphans if it starts computing the send stream before the next transaction commit happens (triggered by a timer or sync() for .e.g), which is when the commit root gets assigned a reference to current root, where the orphans are not visible anymore. The consequence of send seeing the orphans is explained below. For example: mkfs.btrfs -f /dev/sdd mount -o commit=999 /dev/sdd /mnt # open a file with O_TMPFILE and leave it open # write some data to the file btrfs subvolume snapshot -r /mnt /mnt/snap1 btrfs send /mnt/snap1 -f /tmp/send.data The send operation will fail with the following error: ERROR: send ioctl failed with -116: Stale file handle What happens here is that our snapshot has an orphan inode still visible through the commit root, that corresponds to the tmpfile. However send will attempt to call inode.c:btrfs_iget(), with the goal of reading the file's data, which will return -ESTALE because it will use the current root (and not the commit root) of the snapshot. Of course, there are other cases where we can get orphans, but this example using a tmpfile makes it much easier to reproduce the issue. Therefore on snapshot creation, after calling btrfs_orphan_cleanup, if the commit root is different from the current root, just commit the transaction associated with the snapshot's root (if it exists), so that a send will not see any orphans that don't exist anymore. This also guarantees a send will always see the same content regardless of whether a transaction commit happened already before the send was requested and after the orphan cleanup (meaning the commit root and current roots are the same) or it hasn't happened yet (commit and current roots are different). Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 04ece8fab921..219e26fa9069 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -712,6 +712,35 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir, if (ret) goto fail; + /* + * If orphan cleanup did remove any orphans, it means the tree was + * modified and therefore the commit root is not the same as the + * current root anymore. This is a problem, because send uses the + * commit root and therefore can see inode items that don't exist + * in the current root anymore, and for example make calls to + * btrfs_iget, which will do tree lookups based on the current root + * and not on the commit root. Those lookups will fail, returning a + * -ESTALE error, and making send fail with that error. So make sure + * a send does not see any orphans we have just removed, and that it + * will see the same inodes regardless of whether a transaction + * commit happened before it started (meaning that the commit root + * will be the same as the current root) or not. + */ + if (readonly && pending_snapshot->snap->node != + pending_snapshot->snap->commit_root) { + trans = btrfs_join_transaction(pending_snapshot->snap); + if (IS_ERR(trans) && PTR_ERR(trans) != -ENOENT) { + ret = PTR_ERR(trans); + goto fail; + } + if (!IS_ERR(trans)) { + ret = btrfs_commit_transaction(trans, + pending_snapshot->snap); + if (ret) + goto fail; + } + } + inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry); if (IS_ERR(inode)) { ret = PTR_ERR(inode); From d77815461f047e561f77a07754ae923ade597d4e Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Fri, 30 May 2014 15:16:10 +0800 Subject: [PATCH 64/80] btrfs: Avoid trucating page or punching hole in a already existed hole. btrfs_punch_hole() will truncate unaligned pages or punch hole on a already existed hole. This will cause unneeded zero page or holes splitting the original huge hole. This patch will skip already existed holes before any page truncating or hole punching. Signed-off-by: Qu Wenruo Signed-off-by: Chris Mason --- fs/btrfs/file.c | 112 ++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 98 insertions(+), 14 deletions(-) diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index e46bfaf6cde2..eb3f2708a01d 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -2184,6 +2184,37 @@ out: return 0; } +/* + * Find a hole extent on given inode and change start/len to the end of hole + * extent.(hole/vacuum extent whose em->start <= start && + * em->start + em->len > start) + * When a hole extent is found, return 1 and modify start/len. + */ +static int find_first_non_hole(struct inode *inode, u64 *start, u64 *len) +{ + struct extent_map *em; + int ret = 0; + + em = btrfs_get_extent(inode, NULL, 0, *start, *len, 0); + if (IS_ERR_OR_NULL(em)) { + if (!em) + ret = -ENOMEM; + else + ret = PTR_ERR(em); + return ret; + } + + /* Hole or vacuum extent(only exists in no-hole mode) */ + if (em->block_start == EXTENT_MAP_HOLE) { + ret = 1; + *len = em->start + em->len > *start + *len ? + 0 : *start + *len - em->start - em->len; + *start = em->start + em->len; + } + free_extent_map(em); + return ret; +} + static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) { struct btrfs_root *root = BTRFS_I(inode)->root; @@ -2191,17 +2222,18 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) struct btrfs_path *path; struct btrfs_block_rsv *rsv; struct btrfs_trans_handle *trans; - u64 lockstart = round_up(offset, BTRFS_I(inode)->root->sectorsize); - u64 lockend = round_down(offset + len, - BTRFS_I(inode)->root->sectorsize) - 1; - u64 cur_offset = lockstart; + u64 lockstart; + u64 lockend; + u64 tail_start; + u64 tail_len; + u64 orig_start = offset; + u64 cur_offset; u64 min_size = btrfs_calc_trunc_metadata_size(root, 1); u64 drop_end; int ret = 0; int err = 0; int rsv_count; - bool same_page = ((offset >> PAGE_CACHE_SHIFT) == - ((offset + len - 1) >> PAGE_CACHE_SHIFT)); + bool same_page; bool no_holes = btrfs_fs_incompat(root->fs_info, NO_HOLES); u64 ino_size; @@ -2211,6 +2243,21 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) mutex_lock(&inode->i_mutex); ino_size = round_up(inode->i_size, PAGE_CACHE_SIZE); + ret = find_first_non_hole(inode, &offset, &len); + if (ret < 0) + goto out_only_mutex; + if (ret && !len) { + /* Already in a large hole */ + ret = 0; + goto out_only_mutex; + } + + lockstart = round_up(offset , BTRFS_I(inode)->root->sectorsize); + lockend = round_down(offset + len, + BTRFS_I(inode)->root->sectorsize) - 1; + same_page = ((offset >> PAGE_CACHE_SHIFT) == + ((offset + len - 1) >> PAGE_CACHE_SHIFT)); + /* * We needn't truncate any page which is beyond the end of the file * because we are sure there is no data there. @@ -2222,8 +2269,7 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) if (same_page && len < PAGE_CACHE_SIZE) { if (offset < ino_size) ret = btrfs_truncate_page(inode, offset, len, 0); - mutex_unlock(&inode->i_mutex); - return ret; + goto out_only_mutex; } /* zero back part of the first page */ @@ -2235,12 +2281,39 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) } } - /* zero the front end of the last page */ - if (offset + len < ino_size) { - ret = btrfs_truncate_page(inode, offset + len, 0, 1); - if (ret) { - mutex_unlock(&inode->i_mutex); - return ret; + /* Check the aligned pages after the first unaligned page, + * if offset != orig_start, which means the first unaligned page + * including serveral following pages are already in holes, + * the extra check can be skipped */ + if (offset == orig_start) { + /* after truncate page, check hole again */ + len = offset + len - lockstart; + offset = lockstart; + ret = find_first_non_hole(inode, &offset, &len); + if (ret < 0) + goto out_only_mutex; + if (ret && !len) { + ret = 0; + goto out_only_mutex; + } + lockstart = offset; + } + + /* Check the tail unaligned part is in a hole */ + tail_start = lockend + 1; + tail_len = offset + len - tail_start; + if (tail_len) { + ret = find_first_non_hole(inode, &tail_start, &tail_len); + if (unlikely(ret < 0)) + goto out_only_mutex; + if (!ret) { + /* zero the front end of the last page */ + if (tail_start + tail_len < ino_size) { + ret = btrfs_truncate_page(inode, + tail_start + tail_len, 0, 1); + if (ret) + goto out_only_mutex; + } } } @@ -2314,6 +2387,8 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) BUG_ON(ret); trans->block_rsv = rsv; + cur_offset = lockstart; + len = lockend - cur_offset; while (cur_offset < lockend) { ret = __btrfs_drop_extents(trans, root, inode, path, cur_offset, lockend + 1, @@ -2354,6 +2429,14 @@ static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) rsv, min_size); BUG_ON(ret); /* shouldn't happen */ trans->block_rsv = rsv; + + ret = find_first_non_hole(inode, &cur_offset, &len); + if (unlikely(ret < 0)) + break; + if (ret && !len) { + ret = 0; + break; + } } if (ret) { @@ -2392,6 +2475,7 @@ out_free: out: unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, &cached_state, GFP_NOFS); +out_only_mutex: mutex_unlock(&inode->i_mutex); if (ret && !err) err = ret; From b05fd8742f6291b67571ad0fdad4da6b6eb98025 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Thu, 29 May 2014 23:31:39 +0100 Subject: [PATCH 65/80] Btrfs: fix transaction leak during fsync call If btrfs_log_dentry_safe() returns an error, we set ret to 1 and fall through with the goal of committing the transaction. However, in the case where the inode doesn't need a full sync, we would call btrfs_wait_ordered_range() against the target range for our inode, and if it returned an error, we would return without commiting or ending the transaction. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/file.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index eb3f2708a01d..ad7c05909a49 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -2025,8 +2025,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) if (!full_sync) { ret = btrfs_wait_ordered_range(inode, start, end - start + 1); - if (ret) + if (ret) { + btrfs_end_transaction(trans, root); goto out; + } } ret = btrfs_commit_transaction(trans, root); } else { From 902c68a4da74442f0ab1c0b458f7723a68dfd3b1 Mon Sep 17 00:00:00 2001 From: Gui Hecheng Date: Thu, 29 May 2014 09:19:58 +0800 Subject: [PATCH 66/80] btrfs: replace EINVAL with ERANGE for resize when ULLONG_MAX To be accurate about the error case, if the new size is beyond ULLONG_MAX, return ERANGE instead of EINVAL. Signed-off-by: Gui Hecheng Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 219e26fa9069..55f4d717d541 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1593,7 +1593,7 @@ static noinline int btrfs_ioctl_resize(struct file *file, new_size = old_size - new_size; } else if (mod > 0) { if (new_size > ULLONG_MAX - old_size) { - ret = -EINVAL; + ret = -ERANGE; goto out_free; } new_size = old_size + new_size; From 964930312aec583809a690868119ce716f4ee926 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Tue, 27 May 2014 13:53:20 -0400 Subject: [PATCH 67/80] btrfs: free delayed node outside of root->inode_lock On heavy workloads, we're seeing soft lockup warnings on root->inode_lock in __btrfs_release_delayed_node. The low hanging fruit is to reduce the size of the critical section. Signed-off-by: Jeff Mahoney Reviewed-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/delayed-inode.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c index 33e561a84013..da775bfdebc9 100644 --- a/fs/btrfs/delayed-inode.c +++ b/fs/btrfs/delayed-inode.c @@ -149,8 +149,8 @@ again: spin_lock(&root->inode_lock); ret = radix_tree_insert(&root->delayed_nodes_tree, ino, node); if (ret == -EEXIST) { - kmem_cache_free(delayed_node_cache, node); spin_unlock(&root->inode_lock); + kmem_cache_free(delayed_node_cache, node); radix_tree_preload_end(); goto again; } @@ -267,14 +267,17 @@ static void __btrfs_release_delayed_node( mutex_unlock(&delayed_node->mutex); if (atomic_dec_and_test(&delayed_node->refs)) { + bool free = false; struct btrfs_root *root = delayed_node->root; spin_lock(&root->inode_lock); if (atomic_read(&delayed_node->refs) == 0) { radix_tree_delete(&root->delayed_nodes_tree, delayed_node->inode_id); - kmem_cache_free(delayed_node_cache, delayed_node); + free = true; } spin_unlock(&root->inode_lock); + if (free) + kmem_cache_free(delayed_node_cache, delayed_node); } } From f82a9901b0321feb4b4dea1583c51bd75707dd4e Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Sun, 1 Jun 2014 01:50:28 +0100 Subject: [PATCH 68/80] Btrfs: fix clone to deal with holes when NO_HOLES feature is enabled If the NO_HOLES feature is enabled holes don't have file extent items in the btree that represent them anymore. This made the clone operation ignore the gaps that exist between consecutive file extent items and therefore not create the holes at the destination. When not using the NO_HOLES feature, the holes were created at the destination. A test case for xfstests follows. Signed-off-by: Filipe David Borba Manana Reviewed-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 108 ++++++++++++++++++++++++++++++++++++----------- 1 file changed, 83 insertions(+), 25 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 55f4d717d541..44dcfd054ca6 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -3012,6 +3012,37 @@ out: return ret; } +static int clone_finish_inode_update(struct btrfs_trans_handle *trans, + struct inode *inode, + u64 endoff, + const u64 destoff, + const u64 olen) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + + inode_inc_iversion(inode); + inode->i_mtime = inode->i_ctime = CURRENT_TIME; + /* + * We round up to the block size at eof when determining which + * extents to clone above, but shouldn't round up the file size. + */ + if (endoff > destoff + olen) + endoff = destoff + olen; + if (endoff > inode->i_size) + btrfs_i_size_write(inode, endoff); + + ret = btrfs_update_inode(trans, root, inode); + if (ret) { + btrfs_abort_transaction(trans, root, ret); + btrfs_end_transaction(trans, root); + goto out; + } + ret = btrfs_end_transaction(trans, root); +out: + return ret; +} + /** * btrfs_clone() - clone a range from inode file to another * @@ -3024,7 +3055,8 @@ out: * @destoff: Offset within @inode to start clone */ static int btrfs_clone(struct inode *src, struct inode *inode, - u64 off, u64 olen, u64 olen_aligned, u64 destoff) + const u64 off, const u64 olen, const u64 olen_aligned, + const u64 destoff) { struct btrfs_root *root = BTRFS_I(inode)->root; struct btrfs_path *path = NULL; @@ -3036,8 +3068,9 @@ static int btrfs_clone(struct inode *src, struct inode *inode, int slot; int ret; int no_quota; - u64 len = olen_aligned; + const u64 len = olen_aligned; u64 last_disko = 0; + u64 last_dest_end = destoff; ret = -ENOMEM; buf = vmalloc(btrfs_level_size(root, 0)); @@ -3105,7 +3138,7 @@ process_slot: u64 disko = 0, diskl = 0; u64 datao = 0, datal = 0; u8 comp; - u64 endoff; + u64 drop_start; extent = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); @@ -3153,6 +3186,18 @@ process_slot: else new_key.offset = destoff; + /* + * Deal with a hole that doesn't have an extent item + * that represents it (NO_HOLES feature enabled). + * This hole is either in the middle of the cloning + * range or at the beginning (fully overlaps it or + * partially overlaps it). + */ + if (new_key.offset != last_dest_end) + drop_start = last_dest_end; + else + drop_start = new_key.offset; + /* * 1 - adjusting old extent (we may have to split it) * 1 - add new extent @@ -3182,7 +3227,7 @@ process_slot: } ret = btrfs_drop_extents(trans, root, inode, - new_key.offset, + drop_start, new_key.offset + datal, 1); if (ret) { @@ -3283,7 +3328,7 @@ process_slot: aligned_end = ALIGN(new_key.offset + datal, root->sectorsize); ret = btrfs_drop_extents(trans, root, inode, - new_key.offset, + drop_start, aligned_end, 1); if (ret) { @@ -3321,27 +3366,12 @@ process_slot: btrfs_mark_buffer_dirty(leaf); btrfs_release_path(path); - inode_inc_iversion(inode); - inode->i_mtime = inode->i_ctime = CURRENT_TIME; - - /* - * we round up to the block size at eof when - * determining which extents to clone above, - * but shouldn't round up the file size - */ - endoff = new_key.offset + datal; - if (endoff > destoff+olen) - endoff = destoff+olen; - if (endoff > inode->i_size) - btrfs_i_size_write(inode, endoff); - - ret = btrfs_update_inode(trans, root, inode); - if (ret) { - btrfs_abort_transaction(trans, root, ret); - btrfs_end_transaction(trans, root); + last_dest_end = new_key.offset + datal; + ret = clone_finish_inode_update(trans, inode, + last_dest_end, + destoff, olen); + if (ret) goto out; - } - ret = btrfs_end_transaction(trans, root); if (new_key.offset + datal >= destoff + len) break; } @@ -3350,6 +3380,34 @@ process_slot: } ret = 0; + if (last_dest_end < destoff + len) { + /* + * We have an implicit hole (NO_HOLES feature is enabled) that + * fully or partially overlaps our cloning range at its end. + */ + btrfs_release_path(path); + + /* + * 1 - remove extent(s) + * 1 - inode update + */ + trans = btrfs_start_transaction(root, 2); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + goto out; + } + ret = btrfs_drop_extents(trans, root, inode, + last_dest_end, destoff + len, 1); + if (ret) { + if (ret != -EOPNOTSUPP) + btrfs_abort_transaction(trans, root, ret); + btrfs_end_transaction(trans, root); + goto out; + } + ret = clone_finish_inode_update(trans, inode, destoff + len, + destoff, olen); + } + out: btrfs_free_path(path); vfree(buf); From 337c6f6830a5ceb650eaf10d38ed8e8f496c8f61 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Mon, 9 Jun 2014 13:22:13 +0100 Subject: [PATCH 69/80] Btrfs: ensure btrfs_prev_leaf doesn't miss 1 item We might have had an item with the previous key in the tree right before we released our path. And after we released our path, that item might have been pushed to the first slot (0) of the leaf we were holding due to a tree balance. Alternatively, an item with the previous key can exist as the only element of a leaf (big fat item). Therefore account for these 2 cases, so that our callers (like btrfs_previous_item) don't miss an existing item with a key matching the previous key we computed above. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/ctree.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index d99d9651dd58..4eada52f3969 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -5097,7 +5097,17 @@ int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path) return ret; btrfs_item_key(path->nodes[0], &found_key, 0); ret = comp_keys(&found_key, &key); - if (ret < 0) + /* + * We might have had an item with the previous key in the tree right + * before we released our path. And after we released our path, that + * item might have been pushed to the first slot (0) of the leaf we + * were holding due to a tree balance. Alternatively, an item with the + * previous key can exist as the only element of a leaf (big fat item). + * Therefore account for these 2 cases, so that our callers (like + * btrfs_previous_item) don't miss an existing item with a key matching + * the previous key we computed above. + */ + if (ret <= 0) return 0; return 1; } From 0b43e04f700bce8eecf3581612e9d3e15bf258bc Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Mon, 9 Jun 2014 11:04:49 +0800 Subject: [PATCH 70/80] Btrfs: fix leaf corruption after __btrfs_drop_extents Several reports about leaf corruption has been floating on the list, one of them points to __btrfs_drop_extents(), and we find that the leaf becomes corrupted after __btrfs_drop_extents(), it's really a rare case but it does exist. The problem turns out to be btrfs_next_leaf() called in __btrfs_drop_extents(). So in btrfs_next_leaf(), we release the current path to re-search the last key of the leaf for locating next leaf, and we've taken it into account that there might be balance operations between leafs during this 'unlock and re-lock' dance, so we check the path again and advance it if there are now more items available. But things are a bit different if that last key happens to be removed and balance gets a bigger key as the last one, and btrfs_search_slot will return it with ret > 0, IOW, nothing change in this leaf except the new last key, then we think we're okay because there is no more item balanced in, fine, we thinks we can go to the next leaf. However, we should return that bigger key, otherwise we deserve leaf corruption, for example, in endio, skipping that key means that __btrfs_drop_extents() thinks it has dropped all extent matched the required range and finish_ordered_io can safely insert a new extent, but it actually doesn't and ends up a leaf corruption. One may be asking that why our locking on extent io tree doesn't work as expected, ie. it should avoid this kind of race situation. But in __btrfs_drop_extents(), we don't always find extents which are included within our locking range, IOW, extents can start before our searching start, in this case locking on extent io tree doesn't protect us from the race. This takes the special case into account. Reviewed-by: Filipe Manana Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/ctree.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index 4eada52f3969..aeab453b8e24 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -5718,6 +5718,24 @@ again: ret = 0; goto done; } + /* + * So the above check misses one case: + * - after releasing the path above, someone has removed the item that + * used to be at the very end of the block, and balance between leafs + * gets another one with bigger key.offset to replace it. + * + * This one should be returned as well, or we can get leaf corruption + * later(esp. in __btrfs_drop_extents()). + * + * And a bit more explanation about this check, + * with ret > 0, the key isn't found, the path points to the slot + * where it should be inserted, so the path->slots[0] item must be the + * bigger one. + */ + if (nritems > 0 && ret > 0 && path->slots[0] == nritems - 1) { + ret = 0; + goto done; + } while (level < BTRFS_MAX_LEVEL) { if (!path->nodes[level]) { From 9391558411e89915f7f7c5c996d3091eced1a0ef Mon Sep 17 00:00:00 2001 From: Antonio Ospite Date: Wed, 4 Jun 2014 14:03:48 +0200 Subject: [PATCH 71/80] trivial: fs/btrfs/ioctl.c: fix typo s/substract/subtract/ Signed-off-by: Antonio Ospite Cc: Chris Mason Cc: Josef Bacik Cc: linux-btrfs@vger.kernel.org Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 44dcfd054ca6..dd9a02a53c7b 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -3216,11 +3216,11 @@ process_slot: * | ------------- extent ------------- | */ - /* substract range b */ + /* subtract range b */ if (key.offset + datal > off + len) datal = off + len - key.offset; - /* substract range a */ + /* subtract range a */ if (off > key.offset) { datao += off - key.offset; datal -= off - key.offset; From c81d57679ee4b87a5b50e9afe4e5d1866d08b9ad Mon Sep 17 00:00:00 2001 From: Gui Hecheng Date: Thu, 5 Jun 2014 10:02:51 +0800 Subject: [PATCH 72/80] btrfs: replace EINVAL with EOPNOTSUPP for dev_replace raid56 To return EOPNOTSUPP is more user friendly than to return EINVAL, and then user-space tool will show that the dev_replace operation for raid56 is not currently supported rather than showing that there is an invalid argument. Signed-off-by: Gui Hecheng Signed-off-by: Chris Mason --- fs/btrfs/dev-replace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/dev-replace.c b/fs/btrfs/dev-replace.c index 9f2290509aca..2af6e66fe788 100644 --- a/fs/btrfs/dev-replace.c +++ b/fs/btrfs/dev-replace.c @@ -313,7 +313,7 @@ int btrfs_dev_replace_start(struct btrfs_root *root, if (btrfs_fs_incompat(fs_info, RAID56)) { btrfs_warn(fs_info, "dev_replace cannot yet handle RAID5/RAID6"); - return -EINVAL; + return -EOPNOTSUPP; } switch (args->start.cont_reading_from_srcdev_mode) { From 91405151eb7005949072bee751343390757f8417 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Thu, 5 Jun 2014 13:22:24 +0100 Subject: [PATCH 73/80] Btrfs: make sure we retry if we couldn't get the page In inode.c:btrfs_page_exists_in_range(), if we can't get the page we need to retry. However we weren't retrying because we weren't setting page to NULL, which makes the while loop exit immediately and will make us call page_cache_release after exiting the loop which is incorrect because our page get didn't succeed. This could also make us return true when we shouldn't. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 38d1e7b976d8..cdbd20e62bb6 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6777,8 +6777,10 @@ bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end) break; /* TODO: Is this relevant for this use case? */ } - if (!page_cache_get_speculative(page)) + if (!page_cache_get_speculative(page)) { + page = NULL; continue; + } /* * Has the page moved? From 809f9016253dbfec00f71e98e11c4bcbf01b6c26 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Thu, 5 Jun 2014 13:22:25 +0100 Subject: [PATCH 74/80] Btrfs: make sure we retry if page is a retriable exception In inode.c:btrfs_page_exists_in_range(), if the page we get from the radix tree is an exception which should make us retry, set page to NULL in order to really retry, because otherwise we don't get another loop iteration executed (page != NULL makes the while loop exit). This also was making us call page_cache_release after exiting the loop, which isn't correct because page doesn't point to a valid page, and possibly return true from the function when we shouldn't. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index cdbd20e62bb6..f265f413acba 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6767,8 +6767,10 @@ bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end) break; if (radix_tree_exception(page)) { - if (radix_tree_deref_retry(page)) + if (radix_tree_deref_retry(page)) { + page = NULL; continue; + } /* * Otherwise, shmem/tmpfs must be storing a swap entry * here as an exceptional entry: so return it without From 6fdef6d43cb897bcbbed13786e8cc8cccabc442a Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Thu, 5 Jun 2014 13:22:26 +0100 Subject: [PATCH 75/80] Btrfs: don't release invalid page in btrfs_page_exists_in_range() In inode.c:btrfs_page_exists_in_range(), if the page we got from the radix tree is an exception entry, which can't be retried, we exit the loop with a non-NULL page and then call page_cache_release against it, which is not ok since it's not a valid page. This could also make us return true when we shouldn't. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 1 + 1 file changed, 1 insertion(+) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index f265f413acba..477e64a1146f 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6776,6 +6776,7 @@ bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end) * here as an exceptional entry: so return it without * attempting to raise page count. */ + page = NULL; break; /* TODO: Is this relevant for this use case? */ } From 8a56457f5f8fa7c2698ffae8545214c5b96a2cb5 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 5 Jun 2014 16:08:45 -0400 Subject: [PATCH 76/80] Btrfs: don't check nodes for extent items The backref code was looking at nodes as well as leaves when we tried to populate extent item entries. This is not good, and although we go away with it for the most part because we'd skip where disk_bytenr != random_memory, sometimes random_memory would match and suddenly boom. This fixes that problem. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/backref.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index f09aa18ab710..84d0912eecb2 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -988,11 +988,12 @@ again: goto out; } if (ref->count && ref->parent) { - if (extent_item_pos && !ref->inode_list) { + if (extent_item_pos && !ref->inode_list && + ref->level == 0) { u32 bsz; struct extent_buffer *eb; bsz = btrfs_level_size(fs_info->extent_root, - info_level); + ref->level); eb = read_tree_block(fs_info->extent_root, ref->parent, bsz, 0); if (!eb || !extent_buffer_uptodate(eb)) { From cd857dd6bc2ae9ecea14e75a34e8a8fdc158e307 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Sun, 8 Jun 2014 19:04:13 +0800 Subject: [PATCH 77/80] Btrfs: use right type to get real comparison We want to make sure the point is still within the extent item, not to verify the memory it's pointing to. Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/backref.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index 84d0912eecb2..00f9d9f911ca 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -1428,7 +1428,7 @@ static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb, *out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1); } *ptr = (unsigned long)*out_eiref; - if ((void *)*ptr >= (void *)ei + item_size) + if ((unsigned long)(*ptr) >= (unsigned long)ei + item_size) return -ENOENT; } From 7ffbb598a059b73487909619d73150f99b50337a Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Mon, 9 Jun 2014 03:48:05 +0100 Subject: [PATCH 78/80] Btrfs: make fsync work after cloning into a file When cloning into a file, we were correctly replacing the extent items in the target range and removing the extent maps. However we weren't replacing the extent maps with new ones that point to the new extents - as a consequence, an incremental fsync (when the inode doesn't have the full sync flag) was a NOOP, since it relies on the existence of extent maps in the modified list of the inode's extent map tree, which was empty. Therefore add new extent maps to reflect the target clone range. A test case for xfstests follows. Signed-off-by: Filipe David Borba Manana Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 6 ++++ fs/btrfs/file-item.c | 76 ++++++++++++++++++++++++++++++++++++++++++++ fs/btrfs/inode.c | 42 +++--------------------- fs/btrfs/ioctl.c | 69 ++++++++++++++++++++++++++++++++++++++++ 4 files changed, 155 insertions(+), 38 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index af523d695432..b7e2c1c1ef36 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -3749,6 +3749,12 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode, struct bio *bio, u64 file_start, int contig); int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end, struct list_head *list, int search_commit); +void btrfs_extent_item_to_extent_map(struct inode *inode, + const struct btrfs_path *path, + struct btrfs_file_extent_item *fi, + const bool new_inline, + struct extent_map *em); + /* inode.c */ struct btrfs_delalloc_work { struct inode *inode; diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c index 609d56b9fd8e..f46cfe45d686 100644 --- a/fs/btrfs/file-item.c +++ b/fs/btrfs/file-item.c @@ -885,3 +885,79 @@ out: fail_unlock: goto out; } + +void btrfs_extent_item_to_extent_map(struct inode *inode, + const struct btrfs_path *path, + struct btrfs_file_extent_item *fi, + const bool new_inline, + struct extent_map *em) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_buffer *leaf = path->nodes[0]; + const int slot = path->slots[0]; + struct btrfs_key key; + u64 extent_start, extent_end; + u64 bytenr; + u8 type = btrfs_file_extent_type(leaf, fi); + int compress_type = btrfs_file_extent_compression(leaf, fi); + + em->bdev = root->fs_info->fs_devices->latest_bdev; + btrfs_item_key_to_cpu(leaf, &key, slot); + extent_start = key.offset; + + if (type == BTRFS_FILE_EXTENT_REG || + type == BTRFS_FILE_EXTENT_PREALLOC) { + extent_end = extent_start + + btrfs_file_extent_num_bytes(leaf, fi); + } else if (type == BTRFS_FILE_EXTENT_INLINE) { + size_t size; + size = btrfs_file_extent_inline_len(leaf, slot, fi); + extent_end = ALIGN(extent_start + size, root->sectorsize); + } + + em->ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi); + if (type == BTRFS_FILE_EXTENT_REG || + type == BTRFS_FILE_EXTENT_PREALLOC) { + em->start = extent_start; + em->len = extent_end - extent_start; + em->orig_start = extent_start - + btrfs_file_extent_offset(leaf, fi); + em->orig_block_len = btrfs_file_extent_disk_num_bytes(leaf, fi); + bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); + if (bytenr == 0) { + em->block_start = EXTENT_MAP_HOLE; + return; + } + if (compress_type != BTRFS_COMPRESS_NONE) { + set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); + em->compress_type = compress_type; + em->block_start = bytenr; + em->block_len = em->orig_block_len; + } else { + bytenr += btrfs_file_extent_offset(leaf, fi); + em->block_start = bytenr; + em->block_len = em->len; + if (type == BTRFS_FILE_EXTENT_PREALLOC) + set_bit(EXTENT_FLAG_PREALLOC, &em->flags); + } + } else if (type == BTRFS_FILE_EXTENT_INLINE) { + em->block_start = EXTENT_MAP_INLINE; + em->start = extent_start; + em->len = extent_end - extent_start; + /* + * Initialize orig_start and block_len with the same values + * as in inode.c:btrfs_get_extent(). + */ + em->orig_start = EXTENT_MAP_HOLE; + em->block_len = (u64)-1; + if (!new_inline && compress_type != BTRFS_COMPRESS_NONE) { + set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); + em->compress_type = compress_type; + } + } else { + btrfs_err(root->fs_info, + "unknown file extent item type %d, inode %llu, offset %llu, root %llu", + type, btrfs_ino(inode), extent_start, + root->root_key.objectid); + } +} diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 477e64a1146f..372b2cb2b297 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6129,7 +6129,6 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, { int ret; int err = 0; - u64 bytenr; u64 extent_start = 0; u64 extent_end = 0; u64 objectid = btrfs_ino(inode); @@ -6143,7 +6142,7 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_trans_handle *trans = NULL; - int compress_type; + const bool new_inline = !page || create; again: read_lock(&em_tree->lock); @@ -6217,7 +6216,6 @@ again: found_type = btrfs_file_extent_type(leaf, item); extent_start = found_key.offset; - compress_type = btrfs_file_extent_compression(leaf, item); if (found_type == BTRFS_FILE_EXTENT_REG || found_type == BTRFS_FILE_EXTENT_PREALLOC) { extent_end = extent_start + @@ -6252,32 +6250,10 @@ next: goto not_found_em; } - em->ram_bytes = btrfs_file_extent_ram_bytes(leaf, item); + btrfs_extent_item_to_extent_map(inode, path, item, new_inline, em); + if (found_type == BTRFS_FILE_EXTENT_REG || found_type == BTRFS_FILE_EXTENT_PREALLOC) { - em->start = extent_start; - em->len = extent_end - extent_start; - em->orig_start = extent_start - - btrfs_file_extent_offset(leaf, item); - em->orig_block_len = btrfs_file_extent_disk_num_bytes(leaf, - item); - bytenr = btrfs_file_extent_disk_bytenr(leaf, item); - if (bytenr == 0) { - em->block_start = EXTENT_MAP_HOLE; - goto insert; - } - if (compress_type != BTRFS_COMPRESS_NONE) { - set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); - em->compress_type = compress_type; - em->block_start = bytenr; - em->block_len = em->orig_block_len; - } else { - bytenr += btrfs_file_extent_offset(leaf, item); - em->block_start = bytenr; - em->block_len = em->len; - if (found_type == BTRFS_FILE_EXTENT_PREALLOC) - set_bit(EXTENT_FLAG_PREALLOC, &em->flags); - } goto insert; } else if (found_type == BTRFS_FILE_EXTENT_INLINE) { unsigned long ptr; @@ -6286,12 +6262,8 @@ next: size_t extent_offset; size_t copy_size; - em->block_start = EXTENT_MAP_INLINE; - if (!page || create) { - em->start = extent_start; - em->len = extent_end - extent_start; + if (new_inline) goto out; - } size = btrfs_file_extent_inline_len(leaf, path->slots[0], item); extent_offset = page_offset(page) + pg_offset - extent_start; @@ -6301,10 +6273,6 @@ next: em->len = ALIGN(copy_size, root->sectorsize); em->orig_block_len = em->len; em->orig_start = em->start; - if (compress_type) { - set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); - em->compress_type = compress_type; - } ptr = btrfs_file_extent_inline_start(item) + extent_offset; if (create == 0 && !PageUptodate(page)) { if (btrfs_file_extent_compression(leaf, item) != @@ -6351,8 +6319,6 @@ next: set_extent_uptodate(io_tree, em->start, extent_map_end(em) - 1, NULL, GFP_NOFS); goto insert; - } else { - WARN(1, KERN_ERR "btrfs unknown found_type %d\n", found_type); } not_found: em->start = start; diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index dd9a02a53c7b..c2e796b664c1 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -3043,6 +3043,60 @@ out: return ret; } +static void clone_update_extent_map(struct inode *inode, + const struct btrfs_trans_handle *trans, + const struct btrfs_path *path, + struct btrfs_file_extent_item *fi, + const u64 hole_offset, + const u64 hole_len) +{ + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + struct extent_map *em; + int ret; + + em = alloc_extent_map(); + if (!em) { + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + return; + } + + if (fi) { + btrfs_extent_item_to_extent_map(inode, path, fi, false, em); + em->generation = -1; + if (btrfs_file_extent_type(path->nodes[0], fi) == + BTRFS_FILE_EXTENT_INLINE) + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + } else { + em->start = hole_offset; + em->len = hole_len; + em->ram_bytes = em->len; + em->orig_start = hole_offset; + em->block_start = EXTENT_MAP_HOLE; + em->block_len = 0; + em->orig_block_len = 0; + em->compress_type = BTRFS_COMPRESS_NONE; + em->generation = trans->transid; + } + + while (1) { + write_lock(&em_tree->lock); + ret = add_extent_mapping(em_tree, em, 1); + write_unlock(&em_tree->lock); + if (ret != -EEXIST) { + free_extent_map(em); + break; + } + btrfs_drop_extent_cache(inode, em->start, + em->start + em->len - 1, 0); + } + + if (unlikely(ret)) + set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); +} + /** * btrfs_clone() - clone a range from inode file to another * @@ -3361,8 +3415,19 @@ process_slot: btrfs_item_ptr_offset(leaf, slot), size); inode_add_bytes(inode, datal); + extent = btrfs_item_ptr(leaf, slot, + struct btrfs_file_extent_item); } + /* If we have an implicit hole (NO_HOLES feature). */ + if (drop_start < new_key.offset) + clone_update_extent_map(inode, trans, + path, NULL, drop_start, + new_key.offset - drop_start); + + clone_update_extent_map(inode, trans, path, + extent, 0, 0); + btrfs_mark_buffer_dirty(leaf); btrfs_release_path(path); @@ -3406,6 +3471,10 @@ process_slot: } ret = clone_finish_inode_update(trans, inode, destoff + len, destoff, olen); + if (ret) + goto out; + clone_update_extent_map(inode, trans, path, NULL, last_dest_end, + destoff + len - last_dest_end); } out: From 6eda71d0c030af0fc2f68aaa676e6d445600855b Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Mon, 9 Jun 2014 10:54:07 +0800 Subject: [PATCH 79/80] Btrfs: fix scrub_print_warning to handle skinny metadata extents The skinny extents are intepreted incorrectly in scrub_print_warning(), and end up hitting the BUG() in btrfs_extent_inline_ref_size. Reported-by: Konstantinos Skarlatos Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/backref.c | 30 +++++++++++++++++++----------- fs/btrfs/backref.h | 4 ++-- fs/btrfs/scrub.c | 5 +++-- 3 files changed, 24 insertions(+), 15 deletions(-) diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index 00f9d9f911ca..e25564bfcb46 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -1409,9 +1409,10 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical, * returns <0 on error */ static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb, - struct btrfs_extent_item *ei, u32 item_size, - struct btrfs_extent_inline_ref **out_eiref, - int *out_type) + struct btrfs_key *key, + struct btrfs_extent_item *ei, u32 item_size, + struct btrfs_extent_inline_ref **out_eiref, + int *out_type) { unsigned long end; u64 flags; @@ -1421,9 +1422,16 @@ static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb, /* first call */ flags = btrfs_extent_flags(eb, ei); if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) { - info = (struct btrfs_tree_block_info *)(ei + 1); - *out_eiref = - (struct btrfs_extent_inline_ref *)(info + 1); + if (key->type == BTRFS_METADATA_ITEM_KEY) { + /* a skinny metadata extent */ + *out_eiref = + (struct btrfs_extent_inline_ref *)(ei + 1); + } else { + WARN_ON(key->type != BTRFS_EXTENT_ITEM_KEY); + info = (struct btrfs_tree_block_info *)(ei + 1); + *out_eiref = + (struct btrfs_extent_inline_ref *)(info + 1); + } } else { *out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1); } @@ -1433,7 +1441,7 @@ static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb, } end = (unsigned long)ei + item_size; - *out_eiref = (struct btrfs_extent_inline_ref *)*ptr; + *out_eiref = (struct btrfs_extent_inline_ref *)(*ptr); *out_type = btrfs_extent_inline_ref_type(eb, *out_eiref); *ptr += btrfs_extent_inline_ref_size(*out_type); @@ -1452,8 +1460,8 @@ static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb, * <0 on error. */ int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb, - struct btrfs_extent_item *ei, u32 item_size, - u64 *out_root, u8 *out_level) + struct btrfs_key *key, struct btrfs_extent_item *ei, + u32 item_size, u64 *out_root, u8 *out_level) { int ret; int type; @@ -1464,8 +1472,8 @@ int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb, return 1; while (1) { - ret = __get_extent_inline_ref(ptr, eb, ei, item_size, - &eiref, &type); + ret = __get_extent_inline_ref(ptr, eb, key, ei, item_size, + &eiref, &type); if (ret < 0) return ret; diff --git a/fs/btrfs/backref.h b/fs/btrfs/backref.h index 94e94429f3e9..86fc20fec282 100644 --- a/fs/btrfs/backref.h +++ b/fs/btrfs/backref.h @@ -40,8 +40,8 @@ int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical, u64 *flags); int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb, - struct btrfs_extent_item *ei, u32 item_size, - u64 *out_root, u8 *out_level); + struct btrfs_key *key, struct btrfs_extent_item *ei, + u32 item_size, u64 *out_root, u8 *out_level); int iterate_extent_inodes(struct btrfs_fs_info *fs_info, u64 extent_item_objectid, diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index bd850b599a99..ac80188eec88 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -588,8 +588,9 @@ static void scrub_print_warning(const char *errstr, struct scrub_block *sblock) if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) { do { - ret = tree_backref_for_extent(&ptr, eb, ei, item_size, - &ref_root, &ref_level); + ret = tree_backref_for_extent(&ptr, eb, &found_key, ei, + item_size, &ref_root, + &ref_level); printk_in_rcu(KERN_WARNING "BTRFS: %s at logical %llu on dev %s, " "sector %llu: metadata %s (level %d) in tree " From c7548af69d9ef71512eb52d8009521eba3e768fd Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Tue, 10 Jun 2014 13:06:56 -0700 Subject: [PATCH 80/80] Btrfs: convert smp_mb__{before,after}_clear_bit The new call is smp_mb__{before,after}_atomic. The __ gives us extra protection from the atomic rays. Signed-off-by: Chris Mason --- fs/btrfs/transaction.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 1c54e2eb74ab..9630f10f8e1e 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -290,7 +290,7 @@ static int record_root_in_trans(struct btrfs_trans_handle *trans, * done before we pop in the zero below */ btrfs_init_reloc_root(trans, root); - smp_mb__before_clear_bit(); + smp_mb__before_atomic(); clear_bit(BTRFS_ROOT_IN_TRANS_SETUP, &root->state); } return 0; @@ -1060,7 +1060,7 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans, /* see comments in should_cow_block() */ clear_bit(BTRFS_ROOT_FORCE_COW, &root->state); - smp_mb__after_clear_bit(); + smp_mb__after_atomic(); if (root->commit_root != root->node) { list_add_tail(&root->dirty_list,