No description
Find a file
Filipe Manana a6703c7115 btrfs: fix crash after non-aligned direct IO write with O_DSYNC
Whenever we attempt to do a non-aligned direct IO write with O_DSYNC, we
end up triggering an assertion and crashing. Example reproducer:

  $ cat test.sh
  #!/bin/bash

  DEV=/dev/sdj
  MNT=/mnt/sdj

  mkfs.btrfs -f $DEV > /dev/null
  mount $DEV $MNT

  # Do a direct IO write with O_DSYNC into a non-aligned range...
  xfs_io -f -d -s -c "pwrite -S 0xab -b 64K 1111 64K" $MNT/foobar

  umount $MNT

When running the reproducer an assertion fails and produces the following
trace:

  [ 2418.403134] assertion failed: !current->journal_info || flush != BTRFS_RESERVE_FLUSH_DATA, in fs/btrfs/space-info.c:1467
  [ 2418.403745] ------------[ cut here ]------------
  [ 2418.404306] kernel BUG at fs/btrfs/ctree.h:3286!
  [ 2418.404862] invalid opcode: 0000 [#2] PREEMPT SMP DEBUG_PAGEALLOC PTI
  [ 2418.405451] CPU: 1 PID: 64705 Comm: xfs_io Tainted: G      D           5.10.15-btrfs-next-87 #1
  [ 2418.406026] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
  [ 2418.407228] RIP: 0010:assertfail.constprop.0+0x18/0x26 [btrfs]
  [ 2418.407835] Code: e6 48 c7 (...)
  [ 2418.409078] RSP: 0018:ffffb06080d13c98 EFLAGS: 00010246
  [ 2418.409696] RAX: 000000000000006c RBX: ffff994c1debbf08 RCX: 0000000000000000
  [ 2418.410302] RDX: 0000000000000000 RSI: 0000000000000027 RDI: 00000000ffffffff
  [ 2418.410904] RBP: ffff994c21770000 R08: 0000000000000000 R09: 0000000000000000
  [ 2418.411504] R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000010000
  [ 2418.412111] R13: ffff994c22198400 R14: ffff994c21770000 R15: 0000000000000000
  [ 2418.412713] FS:  00007f54fd7aff00(0000) GS:ffff994d35200000(0000) knlGS:0000000000000000
  [ 2418.413326] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  [ 2418.413933] CR2: 000056549596d000 CR3: 000000010b928003 CR4: 0000000000370ee0
  [ 2418.414528] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  [ 2418.415109] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
  [ 2418.415669] Call Trace:
  [ 2418.416254]  btrfs_reserve_data_bytes.cold+0x22/0x22 [btrfs]
  [ 2418.416812]  btrfs_check_data_free_space+0x4c/0xa0 [btrfs]
  [ 2418.417380]  btrfs_buffered_write+0x1b0/0x7f0 [btrfs]
  [ 2418.418315]  btrfs_file_write_iter+0x2a9/0x770 [btrfs]
  [ 2418.418920]  new_sync_write+0x11f/0x1c0
  [ 2418.419430]  vfs_write+0x2bb/0x3b0
  [ 2418.419972]  __x64_sys_pwrite64+0x90/0xc0
  [ 2418.420486]  do_syscall_64+0x33/0x80
  [ 2418.420979]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
  [ 2418.421486] RIP: 0033:0x7f54fda0b986
  [ 2418.421981] Code: 48 c7 c0 (...)
  [ 2418.423019] RSP: 002b:00007ffc40569c38 EFLAGS: 00000246 ORIG_RAX: 0000000000000012
  [ 2418.423547] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f54fda0b986
  [ 2418.424075] RDX: 0000000000010000 RSI: 000056549595e000 RDI: 0000000000000003
  [ 2418.424596] RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000400
  [ 2418.425119] R10: 0000000000000400 R11: 0000000000000246 R12: 00000000ffffffff
  [ 2418.425644] R13: 0000000000000400 R14: 0000000000010000 R15: 0000000000000000
  [ 2418.426148] Modules linked in: btrfs blake2b_generic (...)
  [ 2418.429540] ---[ end trace ef2aeb44dc0afa34 ]---

1) At btrfs_file_write_iter() we set current->journal_info to
   BTRFS_DIO_SYNC_STUB;

2) We then call __btrfs_direct_write(), which calls btrfs_direct_IO();

3) We can't do the direct IO write because it starts at a non-aligned
   offset (1111). So at btrfs_direct_IO() we return -EINVAL (coming from
   check_direct_IO() which does the alignment check), but we leave
   current->journal_info set to BTRFS_DIO_SYNC_STUB - we only clear it
   at btrfs_dio_iomap_begin(), because we assume we always get there;

4) Then at __btrfs_direct_write() we see that the attempt to do the
   direct IO write was not successful, 0 bytes written, so we fallback
   to a buffered write by calling btrfs_buffered_write();

5) There we call btrfs_check_data_free_space() which in turn calls
   btrfs_alloc_data_chunk_ondemand() and that calls
   btrfs_reserve_data_bytes() with flush == BTRFS_RESERVE_FLUSH_DATA;

6) Then at btrfs_reserve_data_bytes() we have current->journal_info set to
   BTRFS_DIO_SYNC_STUB, therefore not NULL, and flush has the value
   BTRFS_RESERVE_FLUSH_DATA, triggering the second assertion:

  int btrfs_reserve_data_bytes(struct btrfs_fs_info *fs_info, u64 bytes,
                               enum btrfs_reserve_flush_enum flush)
  {
      struct btrfs_space_info *data_sinfo = fs_info->data_sinfo;
      int ret;

      ASSERT(flush == BTRFS_RESERVE_FLUSH_DATA ||
             flush == BTRFS_RESERVE_FLUSH_FREE_SPACE_INODE);
      ASSERT(!current->journal_info || flush != BTRFS_RESERVE_FLUSH_DATA);
  (...)

So fix that by setting the journal to NULL whenever check_direct_IO()
returns a failure.

This bug only affects 5.10 kernels, and the regression was introduced in
5.10-rc1 by commit 0eb79294db ("btrfs: dio iomap DSYNC workaround").
The bug does not exist in 5.11 kernels due to commit ecfdc08b8c
("btrfs: remove dio iomap DSYNC workaround"), which depends on a large
patchset that went into the merge window for 5.11. So this is a fix only
for 5.10.x stable kernels, as there are people hitting this bug.

Fixes: 0eb79294db ("btrfs: dio iomap DSYNC workaround")
CC: stable@vger.kernel.org # 5.10 (and only 5.10)
Acked-by: David Sterba <dsterba@suse.com>
Bugzilla: https://bugzilla.suse.com/show_bug.cgi?id=1181605
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-02-23 15:53:25 +01:00
arch xen/arm: don't ignore return errors from set_phys_to_machine 2021-02-23 15:53:24 +01:00
block bfq-iosched: Revert "bfq: Fix computation of shallow depth" 2021-02-17 11:02:24 +01:00
certs .gitignore: add SPDX License Identifier 2020-03-25 11:50:48 +01:00
crypto crypto: xor - Fix divide error in do_xor_speed() 2021-01-27 11:54:52 +01:00
Documentation ovl: implement volatile-specific fsync error behaviour 2021-02-10 09:29:16 +01:00
drivers Bluetooth: btusb: Always fallback to alt 1 for WBS 2021-02-23 15:53:25 +01:00
fs btrfs: fix crash after non-aligned direct IO write with O_DSYNC 2021-02-23 15:53:25 +01:00
include Xen/gntdev: correct error checking in gntdev_map_grant_pages() 2021-02-23 15:53:24 +01:00
init fgraph: Initialize tracing_graph_pause at task creation 2021-02-10 09:29:16 +01:00
ipc ipc: adjust proc_ipc_sem_dointvec definition to match prototype 2020-09-05 12:14:29 -07:00
kernel bpf: Check for integer overflow when using roundup_pow_of_two() 2021-02-17 11:02:25 +01:00
lib udp: fix skb_copy_and_csum_datagram with odd segment sizes 2021-02-17 11:02:28 +01:00
LICENSES LICENSES/deprecated: add Zlib license text 2020-09-16 14:33:49 +02:00
mm Revert "mm: memcontrol: avoid workload stalls when lowering memory.high" 2021-02-13 13:55:17 +01:00
net net: fix proc_fs init handling in af_packet and tls 2021-02-23 15:53:23 +01:00
samples samples/bpf: Fix possible hang in xdpsock with multiple threads 2020-12-30 11:53:49 +01:00
scripts scripts: set proper OpenSSL include dir also for sign-file 2021-02-17 11:02:28 +01:00
security cap: fix conversions on getxattr 2021-02-17 11:02:22 +01:00
sound ASoC: Intel: sof_sdw: set proper flags for Dell TGL-H SKU 0A5E 2021-02-13 13:55:07 +01:00
tools selftests: txtimestamp: fix compilation issue 2021-02-17 11:02:26 +01:00
usr Merge branch 'work.fdpic' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2020-08-07 13:29:39 -07:00
virt KVM: Forbid the use of tagged userspace addresses for memslots 2021-02-03 23:28:41 +01:00
.clang-format RDMA 5.10 pull request 2020-10-17 11:18:18 -07:00
.cocciconfig
.get_maintainer.ignore Opt out of scripts/get_maintainer.pl 2019-05-16 10:53:40 -07:00
.gitattributes .gitattributes: use 'dts' diff driver for dts files 2019-12-04 19:44:11 -08:00
.gitignore .gitignore: docs: ignore sphinx_*/ directories 2020-09-10 10:44:31 -06:00
.mailmap mailmap: add two more addresses of Uwe Kleine-König 2020-12-06 10:19:07 -08:00
COPYING COPYING: state that all contributions really are covered by this file 2020-02-10 13:32:20 -08:00
CREDITS MAINTAINERS: Move Jason Cooper to CREDITS 2020-11-30 10:20:34 +01:00
Kbuild kbuild: rename hostprogs-y/always to hostprogs/always-y 2020-02-04 01:53:07 +09:00
Kconfig kbuild: ensure full rebuild when the compiler is updated 2020-05-12 13:28:33 +09:00
MAINTAINERS Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net 2020-12-10 15:30:13 -08:00
Makefile Linux 5.10.17 2021-02-17 11:02:30 +01:00
README Drop all 00-INDEX files from Documentation/ 2018-09-09 15:08:58 -06:00

Linux kernel
============

There are several guides for kernel developers and users. These guides can
be rendered in a number of formats, like HTML and PDF. Please read
Documentation/admin-guide/README.rst first.

In order to build the documentation, use ``make htmldocs`` or
``make pdfdocs``.  The formatted documentation can also be read online at:

    https://www.kernel.org/doc/html/latest/

There are various text files in the Documentation/ subdirectory,
several of them using the Restructured Text markup notation.

Please read the Documentation/process/changes.rst file, as it contains the
requirements for building and running the kernel, and information about
the problems which may result by upgrading your kernel.