All of lore.kernel.org
 help / color / mirror / Atom feed
From: Zhao Lei <zhaolei@cn.fujitsu.com>
To: <linux-btrfs@vger.kernel.org>
Cc: Zhao Lei <zhaolei@cn.fujitsu.com>
Subject: [PATCH] btrfs: Fix no space bug caused by removing bg
Date: Mon, 21 Sep 2015 20:59:13 +0800	[thread overview]
Message-ID: <15fc8f8d002e4ffcdb46e769736f240ae7ace20b.1442839332.git.zhaolei@cn.fujitsu.com> (raw)

btrfs in v4.3-rc1 failed many xfstests items with '-o nospace_cache'
mount option.

Failed cases are:
 btrfs/008,016,019,020,026,027,028,029,031,041,046,048,050,051,053,054,
 077,083,084,087,092,094
 generic/004,010,014,023,024,074,075,080,086,087,089,091,092,100,112,123,
 124,125,126,127,131,133,192,193,198,207,208,209,213,214,215,228,239,240,
 246,247,248,255,263,285,306,313,316,323

We can reproduce this bug with following simple command:
 TEST_DEV=/dev/vdh
 TEST_DIR=/mnt/tmp

 umount "$TEST_DEV" >/dev/null
 mkfs.btrfs -f "$TEST_DEV"
 mount "$TEST_DEV" "$TEST_DIR"

 umount "$TEST_DEV"
 mount "$TEST_DEV" "$TEST_DIR"

 cp /bin/bash $TEST_DIR

Result is:
 (omit previous commands)
 # cp /bin/bash $TEST_DIR
 cp: writing `/mnt/tmp/bash': No space left on device

By bisect, we can see it is triggered by patch titled:
 commit e44163e17796
 ("btrfs: explictly delete unused block groups in close_ctree and ro-remount")

But the wrong code is not in above patch, btrfs delete all chunks
if no data in filesystem, and above patch just make it obviously.

Detail reason:
 1: mkfs a blank filesystem, or delete everything in filesystem
 2: umount fs
    (current code will delete all data chunks)
 3: mount fs
    Because no any data chunks, data's space_cache have no chance
    to init, it means: space_info->total_bytes == 0, and
    space_info->full == 1.
 4: do some write
    Current code will ignore chunk allocate because space_info->full,
    and return -ENOSPC.

Fix:
 Don't auto-delete last blockgroup for a raid type.
 If we delete all blockgroup for a raidtype, it not only cause above bug,
 but also may change filesystem to all-single in some case.

Test:
 Test by above script, and confirmed the logic by debug output.

Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
---
 fs/btrfs/extent-tree.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 5411f0a..35cf7eb 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -10012,7 +10012,8 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 					       bg_list);
 		space_info = block_group->space_info;
 		list_del_init(&block_group->bg_list);
-		if (ret || btrfs_mixed_space_info(space_info)) {
+		if (ret || btrfs_mixed_space_info(space_info) ||
+		    block_group->list.next == block_group->list.prev) {
 			btrfs_put_block_group(block_group);
 			continue;
 		}
-- 
1.8.5.1


             reply	other threads:[~2015-09-21 13:00 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-09-21 12:59 Zhao Lei [this message]
2015-09-21 13:27 ` [PATCH] btrfs: Fix no space bug caused by removing bg Filipe David Manana
2015-09-21 13:37   ` Filipe David Manana
2015-09-22 10:06   ` Zhao Lei
2015-09-22 10:22     ` Filipe David Manana
2015-09-22 11:24       ` Zhao Lei
2015-09-22 12:45         ` Filipe David Manana
2015-09-23  1:59           ` Zhao Lei
2015-09-22 10:22     ` Zhao Lei
2015-09-22 12:59 ` Jeff Mahoney
2015-09-22 13:28   ` Hugo Mills
2015-09-22 13:36   ` Holger Hoffstätte
2015-09-22 13:41     ` Hugo Mills
2015-09-22 14:23       ` David Sterba
2015-09-22 14:36         ` Hugo Mills
2015-09-22 14:54           ` Austin S Hemmelgarn
2015-09-22 15:39             ` Hugo Mills
2015-09-22 17:32               ` Austin S Hemmelgarn
2015-09-22 17:37                 ` Austin S Hemmelgarn
2015-09-23  4:49                 ` Duncan
2015-09-23 13:28               ` David Sterba
2015-09-23 13:57                 ` Austin S Hemmelgarn
2015-09-23 14:05                 ` Hugo Mills
2015-09-23 13:12           ` David Sterba
2015-09-23 13:19             ` Qu Wenruo
2015-09-23 13:32               ` Austin S Hemmelgarn
2015-09-23 14:00                 ` Qu Wenruo
2015-09-23 17:28                   ` David Sterba
2015-09-23 13:37               ` David Sterba
2015-09-23 13:45               ` Hugo Mills
2015-09-23 13:28             ` Hugo Mills
2015-09-22 16:23     ` Jeff Mahoney
2015-09-23  2:14   ` Zhao Lei

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=15fc8f8d002e4ffcdb46e769736f240ae7ace20b.1442839332.git.zhaolei@cn.fujitsu.com \
    --to=zhaolei@cn.fujitsu.com \
    --cc=linux-btrfs@vger.kernel.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.