linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Sasha Levin <sashal@kernel.org>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: Josef Bacik <josef@toxicpanda.com>, Qu Wenruo <wqu@suse.com>,
	David Sterba <dsterba@suse.com>, Sasha Levin <sashal@kernel.org>,
	linux-btrfs@vger.kernel.org
Subject: [PATCH AUTOSEL 5.5 54/56] btrfs: remove a BUG_ON() from merge_reloc_roots()
Date: Thu,  9 Apr 2020 23:47:58 -0400	[thread overview]
Message-ID: <20200410034800.8381-54-sashal@kernel.org> (raw)
In-Reply-To: <20200410034800.8381-1-sashal@kernel.org>

From: Josef Bacik <josef@toxicpanda.com>

[ Upstream commit 7b7b74315b24dc064bc1c683659061c3d48f8668 ]

This was pretty subtle, we default to reloc roots having 0 root refs, so
if we crash in the middle of the relocation they can just be deleted.
If we successfully complete the relocation operations we'll set our root
refs to 1 in prepare_to_merge() and then go on to merge_reloc_roots().

At prepare_to_merge() time if any of the reloc roots have a 0 reference
still, we will remove that reloc root from our reloc root rb tree, and
then clean it up later.

However this only happens if we successfully start a transaction.  If
we've aborted previously we will skip this step completely, and only
have reloc roots with a reference count of 0, but were never properly
removed from the reloc control's rb tree.

This isn't a problem per-se, our references are held by the list the
reloc roots are on, and by the original root the reloc root belongs to.
If we end up in this situation all the reloc roots will be added to the
dirty_reloc_list, and then properly dropped at that point.  The reloc
control will be free'd and the rb tree is no longer used.

There were two options when fixing this, one was to remove the BUG_ON(),
the other was to make prepare_to_merge() handle the case where we
couldn't start a trans handle.

IMO this is the cleaner solution.  I started with handling the error in
prepare_to_merge(), but it turned out super ugly.  And in the end this
BUG_ON() simply doesn't matter, the cleanup was happening properly, we
were just panicing because this BUG_ON() only matters in the success
case.  So I've opted to just remove it and add a comment where it was.

Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/btrfs/relocation.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 9546ce3c86dc4..e350706b483f1 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -2561,7 +2561,21 @@ void merge_reloc_roots(struct reloc_control *rc)
 			free_reloc_roots(&reloc_roots);
 	}
 
-	BUG_ON(!RB_EMPTY_ROOT(&rc->reloc_root_tree.rb_root));
+	/*
+	 * We used to have
+	 *
+	 * BUG_ON(!RB_EMPTY_ROOT(&rc->reloc_root_tree.rb_root));
+	 *
+	 * here, but it's wrong.  If we fail to start the transaction in
+	 * prepare_to_merge() we will have only 0 ref reloc roots, none of which
+	 * have actually been removed from the reloc_root_tree rb tree.  This is
+	 * fine because we're bailing here, and we hold a reference on the root
+	 * for the list that holds it, so these roots will be cleaned up when we
+	 * do the reloc_dirty_list afterwards.  Meanwhile the root->reloc_root
+	 * will be cleaned up on unmount.
+	 *
+	 * The remaining nodes will be cleaned up by free_reloc_control.
+	 */
 }
 
 static void free_block_list(struct rb_root *blocks)
-- 
2.20.1


  parent reply	other threads:[~2020-04-10  3:49 UTC|newest]

Thread overview: 56+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-04-10  3:47 [PATCH AUTOSEL 5.5 01/56] cpufreq: imx6q: Fixes unwanted cpu overclocking on i.MX6ULL Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 02/56] EDAC/mc: Report "unknown memory" on too many DIMM labels found Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 03/56] usb: ucsi: ccg: disable runtime pm during fw flashing Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 04/56] staging: wilc1000: avoid double unlocking of 'wilc->hif_cs' mutex Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 05/56] media: vimc: streamer: fix memory leak in vimc subdevs if kthread_run fails Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 06/56] media: venus: hfi_parser: Ignore HEVC encoding for V1 Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 07/56] firmware: arm_sdei: fix double-lock on hibernate with shared events Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 08/56] driver core: Reevaluate dev->links.need_for_probe as suppliers are added Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 09/56] sched/vtime: Prevent unstable evaluation of WARN(vtime->state) Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 10/56] null_blk: Fix the null_add_dev() error path Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 11/56] blk-mq: Fix a recently introduced regression in blk_mq_realloc_hw_ctxs() Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 12/56] null_blk: Handle null_add_dev() failures properly Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 13/56] null_blk: fix spurious IO errors after failed past-wp access Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 14/56] media: imx: imx7_mipi_csis: Power off the source when stopping streaming Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 15/56] media: imx: imx7-media-csi: Fix video field handling Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 16/56] xhci: bail out early if driver can't accress host in resume Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 17/56] ACPI: EC: Do not clear boot_ec_is_ecdt in acpi_ec_add() Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 18/56] x86: Don't let pgprot_modify() change the page encryption bit Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 19/56] dma-mapping: Fix dma_pgprot() for unencrypted coherent pages Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 20/56] block: keep bdi->io_pages in sync with max_sectors_kb for stacked devices Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 21/56] debugfs: Check module state before warning in {full/open}_proxy_open() Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 22/56] spi: spi-fsl-dspi: Avoid NULL pointer in dspi_slave_abort for non-DMA mode Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 23/56] irqchip/versatile-fpga: Handle chained IRQs properly Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 24/56] time/sched_clock: Expire timer in hardirq context Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 25/56] media: allegro: fix type of gop_length in channel_create message Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 26/56] sched: Avoid scale real weight down to zero Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 27/56] sched/fair: Fix condition of avg_load calculation Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 28/56] selftests/x86/ptrace_syscall_32: Fix no-vDSO segfault Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 29/56] PCI/switchtec: Fix init_completion race condition with poll_wait() Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 30/56] block, bfq: move forward the getting of an extra ref in bfq_bfqq_move Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 31/56] media: i2c: video-i2c: fix build errors due to 'imply hwmon' Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 32/56] libata: Remove extra scsi_host_put() in ata_scsi_add_hosts() Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 33/56] pstore/platform: fix potential mem leak if pstore_init_fs failed Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 34/56] gfs2: Do log_flush in gfs2_ail_empty_gl even if ail list is empty Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 35/56] gfs2: Don't demote a glock until its revokes are written Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 36/56] cpufreq: imx6q: fix error handling Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 37/56] x86/boot: Use unsigned comparison for addresses Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 38/56] efi/x86: Ignore the memory attributes table on i386 Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 39/56] genirq/irqdomain: Check pointer in irq_domain_alloc_irqs_hierarchy() Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 40/56] blk-mq: Keep set->nr_hw_queues and set->map[].nr_queues in sync Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 41/56] block: Fix use-after-free issue accessing struct io_cq Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 42/56] block, zoned: fix integer overflow with BLKRESETZONE et al Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 43/56] media: mtk-vpu: avoid unaligned access to DTCM buffer Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 44/56] media: i2c: ov5695: Fix power on and off sequences Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 45/56] usb: dwc3: core: add support for disabling SS instances in park mode Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 46/56] irqchip/gic-v4: Provide irq_retrigger to avoid circular locking dependency Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 47/56] md: check arrays is suspended in mddev_detach before call quiesce operations Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 48/56] firmware: fix a double abort case with fw_load_sysfs_fallback Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 49/56] spi: spi-fsl-dspi: Replace interruptible wait queue with a simple completion Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 50/56] locking/lockdep: Avoid recursion in lockdep_count_{for,back}ward_deps() Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 51/56] block, bfq: fix use-after-free in bfq_idle_slice_timer_body Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 52/56] btrfs: hold a ref on the root in btrfs_recover_relocation Sasha Levin
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 53/56] btrfs: qgroup: ensure qgroup_rescan_running is only set when the worker is at least queued Sasha Levin
2020-04-10  3:47 ` Sasha Levin [this message]
2020-04-10  3:47 ` [PATCH AUTOSEL 5.5 55/56] btrfs: restart relocate_tree_blocks properly Sasha Levin
2020-04-10  3:48 ` [PATCH AUTOSEL 5.5 56/56] btrfs: track reloc roots based on their commit root bytenr Sasha Levin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200410034800.8381-54-sashal@kernel.org \
    --to=sashal@kernel.org \
    --cc=dsterba@suse.com \
    --cc=josef@toxicpanda.com \
    --cc=linux-btrfs@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=stable@vger.kernel.org \
    --cc=wqu@suse.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).