在 2025/1/10 09:50, Boris Burkov 写道:
On Thu, Dec 12, 2024 at 04:43:58PM +1030, Qu Wenruo wrote:[BUG] When testing with COW fixup marked as BUG_ON() (this is involved with the new pin_user_pages*() change, which should not result new out-of-band dirty pages), I hit a crash triggered by the BUG_ON() from hitting COW fixup path. This BUG_ON() happens just after a failed btrfs_run_delalloc_range(): BTRFS error (device dm-2): failed to run delalloc range, root 348 ino 405 folio 65536 submit_bitmap 6-15 start 90112 len 106496: -28 ------------[ cut here ]------------ kernel BUG at fs/btrfs/extent_io.c:1444! Internal error: Oops - BUG: 00000000f2000800 [#1] SMP CPU: 0 UID: 0 PID: 434621 Comm: kworker/u24:8 Tainted: G OE 6.12.0-rc7-custom+ #86 Hardware name: QEMU KVM Virtual Machine, BIOS unknown 2/2/2022 Workqueue: events_unbound btrfs_async_reclaim_data_space [btrfs] pc : extent_writepage_io+0x2d4/0x308 [btrfs] lr : extent_writepage_io+0x2d4/0x308 [btrfs] Call trace: extent_writepage_io+0x2d4/0x308 [btrfs] extent_writepage+0x218/0x330 [btrfs] extent_write_cache_pages+0x1d4/0x4b0 [btrfs] btrfs_writepages+0x94/0x150 [btrfs] do_writepages+0x74/0x190 filemap_fdatawrite_wbc+0x88/0xc8 start_delalloc_inodes+0x180/0x3b0 [btrfs] btrfs_start_delalloc_roots+0x174/0x280 [btrfs] shrink_delalloc+0x114/0x280 [btrfs] flush_space+0x250/0x2f8 [btrfs] btrfs_async_reclaim_data_space+0x180/0x228 [btrfs] process_one_work+0x164/0x408 worker_thread+0x25c/0x388 kthread+0x100/0x118 ret_from_fork+0x10/0x20 Code: aa1403e1 9402f3ef aa1403e0 9402f36f (d4210000) ---[ end trace 0000000000000000 ]--- [CAUSE] That failure is mostly from cow_file_range(), where we can hit -ENOSPC. Although the -ENOSPC is already a bug related to our space reservation code, let's just focus on the error handling. For example, we have the following dirty range [0, 64K) of an inode, with 4K sector size and 4K page size: 0 16K 32K 48K 64K |///////////////////////////////////////| |#######################################| Where |///| means page are still dirty, and |###| means the extent io tree has EXTENT_DELALLOC flag. - Enter extent_writepage() for page 0 - Enter btrfs_run_delalloc_range() for range [0, 64K) - Enter cow_file_range() for range [0, 64K) - Function btrfs_reserve_extent() only reserved one 16K extent So we created extent map and ordered extent for range [0, 16K) 0 16K 32K 48K 64K |////////|//////////////////////////////| |<- OE ->|##############################| And range [0, 16K) has its delalloc flag cleared. But since we haven't yet submit any bio, involved 4 pages are still dirty. - Function btrfs_reserve_extent() return with -ENOSPC Now we have to run error cleanup, which will clear all EXTENT_DELALLOC* flags and clear the dirty flags for the remaining ranges: 0 16K 32K 48K 64K |////////| | | | | Note that range [0, 16K) still has their pages dirty. - Some time later, writeback are triggered again for the range [0, 16K) since the page range still have dirty flags. - btrfs_run_delalloc_range() will do nothing because there is no EXTENT_DELALLOC flag. - extent_writepage_io() find page 0 has no ordered flag Which falls into the COW fixup path, triggering the BUG_ON(). Unfortunately this error handling bug dates back to the introduction of btrfs. Thankfully with the abuse of cow fixup, at least it won't crash the kernel. [FIX] Instead of immediately unlock the extent and folios, we keep the extent and folios locked until either erroring out or the whole delalloc range finished. When the whole delalloc range finished without error, we just unlock the whole range with PAGE_SET_ORDERED (and PAGE_UNLOCK for !keep_locked cases), with EXTENT_DELALLOC and EXTENT_LOCKED cleared. And those involved folios will be properly submitted, with their dirty flags cleared during submission. For the error path, it will be a little more complex: - The range with ordered extent allocated (range (1)) We only clear the EXTENT_DELALLOC and EXTENT_LOCKED, as the remaining flags are cleaned up by btrfs_mark_ordered_io_finished()->btrfs_finish_one_ordered(). For folios we finish the IO (clear dirty, start writeback and immediately finish the writeback) and unlock the folios. - The range with reserved extent but no ordered extent (range(2)) - The range we never touched (range(3)) For both range (2) and range(3) the behavior is not changed. Now even if cow_file_range() failed halfway with some successfully reserved extents/ordered extents, we will keep all folios clean, so there will be no future writeback triggered on them.2 qs, to make sure I understand: This changes the happy path, in that IO can't start on the allocated ordered extents until the whole range is done allocating and unlocked or errors. But it shouldn't be a big deal unless we have this race a lot?
If we race a lot, it already means the fs is fragmented thus we need a lot of loops to allocate quite some small extents.
For normal cases, we should really get a large extent for the delalloc range in one go, and in that case, the lock holding period is not changed.
So if we really hit some races, it already means our fs is fragmented and no one can expect a quick run anyway.
What is the new behavior in your test case? The whole range correctly is not dirty, no IO happens, and the mapping has an error set on it? Have you managed to demonstrate something to that effect more explicitly than not hitting the BUG_ON in your new code?
Unfortunately I have no better way to verify the behavior, other than BUG_ON() in cow fixup path.
It may reduce the warning from space reservation code (because we properly free the reserved space), but it's less obvious than the avoided BUG_ON().
However, assuming I understood correctly, LGTM. Reviewed-by: Boris Burkov <boris@xxxxxx>
Thanks a lot for the review. Qu
Cc: stable@xxxxxxxxxxxxxxx Signed-off-by: Qu Wenruo <wqu@xxxxxxxx> --- fs/btrfs/inode.c | 65 ++++++++++++++++++++++++------------------------ 1 file changed, 32 insertions(+), 33 deletions(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 5ba8d044757b..19c88b7d0363 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -1364,6 +1364,17 @@ static noinline int cow_file_range(struct btrfs_inode *inode,alloc_hint = btrfs_get_extent_allocation_hint(inode, start, num_bytes); + /*+ * We're not doing compressed IO, don't unlock the first page + * (which the caller expects to stay locked), don't clear any + * dirty bits and don't set any writeback bits + * + * Do set the Ordered (Private2) bit so we know this page was + * properly setup for writepage. + */ + page_ops = (keep_locked ? 0 : PAGE_UNLOCK); + page_ops |= PAGE_SET_ORDERED; + /* * Relocation relies on the relocated extents to have exactly the same * size as the original extents. Normally writeback for relocation data @@ -1423,6 +1434,10 @@ static noinline int cow_file_range(struct btrfs_inode *inode, file_extent.offset = 0; file_extent.compression = BTRFS_COMPRESS_NONE;+ /*+ * Locked range will be released either during error clean up or + * after the whole range is finished. + */ lock_extent(&inode->io_tree, start, start + cur_alloc_size - 1, &cached);@@ -1468,21 +1483,6 @@ static noinline int cow_file_range(struct btrfs_inode *inode, btrfs_dec_block_group_reservations(fs_info, ins.objectid); - /*- * We're not doing compressed IO, don't unlock the first page - * (which the caller expects to stay locked), don't clear any - * dirty bits and don't set any writeback bits - * - * Do set the Ordered flag so we know this page was - * properly setup for writepage. - */ - page_ops = (keep_locked ? 0 : PAGE_UNLOCK); - page_ops |= PAGE_SET_ORDERED; - - extent_clear_unlock_delalloc(inode, start, start + cur_alloc_size - 1, - locked_folio, &cached, - EXTENT_LOCKED | EXTENT_DELALLOC, - page_ops); if (num_bytes < cur_alloc_size) num_bytes = 0; else @@ -1499,6 +1499,9 @@ static noinline int cow_file_range(struct btrfs_inode *inode, if (ret) goto out_unlock; } + extent_clear_unlock_delalloc(inode, orig_start, end, locked_folio, &cached, + EXTENT_LOCKED | EXTENT_DELALLOC, + page_ops); done: if (done_offset) *done_offset = end; @@ -1519,35 +1522,31 @@ static noinline int cow_file_range(struct btrfs_inode *inode, * We process each region below. */- clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |- EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV; - page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK; - /* * For the range (1). We have already instantiated the ordered extents * for this region. They are cleaned up by * btrfs_cleanup_ordered_extents() in e.g, - * btrfs_run_delalloc_range(). EXTENT_LOCKED | EXTENT_DELALLOC are - * already cleared in the above loop. And, EXTENT_DELALLOC_NEW | - * EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV are handled by the cleanup - * function. + * btrfs_run_delalloc_range(). + * EXTENT_DELALLOC_NEW | EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV + * are also handled by the cleanup function. * - * However, in case of @keep_locked, we still need to unlock the pages - * (except @locked_folio) to ensure all the pages are unlocked. + * So here we only clear EXTENT_LOCKED and EXTENT_DELALLOC flag, + * and finish the writeback of the involved folios, which will be + * never submitted. */ - if (keep_locked && orig_start < start) { + if (orig_start < start) { + clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC; + page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK; + if (!locked_folio) mapping_set_error(inode->vfs_inode.i_mapping, ret); extent_clear_unlock_delalloc(inode, orig_start, start - 1, - locked_folio, NULL, 0, page_ops); + locked_folio, NULL, clear_bits, page_ops); }- /*- * At this point we're unlocked, we want to make sure we're only - * clearing these flags under the extent lock, so lock the rest of the - * range and clear everything up. - */ - lock_extent(&inode->io_tree, start, end, NULL); + clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW | + EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV; + page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK;/** For the range (2). If we reserved an extent for our delalloc range -- 2.47.1
