This patch attempts to close a hole leading to a BUG seen with hot
removals during writes [1].

A block device (NVME namespace in this test case) is formatted to EXT4
without partitions. It's mounted and write I/O is run to a file, then
the device is hot removed from the slot. The superblock attempts to be
written to the drive which is no longer present.

The typical chain of events leading to the BUG:
ext4_commit_super()
  __sync_dirty_buffer()
    submit_bh()
      submit_bh_wbc()
        BUG_ON(!buffer_mapped(bh));

This fix checks for the superblock's buffer head being mapped prior to
syncing.

[1] https://www.spinics.net/lists/linux-ext4/msg56527.html



Signed-off-by: Jon Derrick <jonathan.derrick@intel.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

The kernel's ext4 mount-time checks were more permissive than
e2fsprogs's libext2fs checks when opening a file system.  The
superblock is considered too insane for debugfs or e2fsck to operate
on it, the kernel has no business trying to mount it.

This will make file system fuzzing tools work harder, but the failure
cases that they find will be more useful and be easier to evaluate.

Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

If there is a directory entry pointing to a system inode (such as a
journal inode), complain and declare the file system to be corrupted.

Also, if the superblock's first inode number field is too small,
refuse to mount the file system.

This addresses CVE-2018-10882.

https://bugzilla.kernel.org/show_bug.cgi?id=200069



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

Use a separate journal transaction if it turns out that we need to
convert an inline file to use an data block.  Otherwise we could end
up failing due to not having journal credits.

This addresses CVE-2018-10883.

https://bugzilla.kernel.org/show_bug.cgi?id=200071



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

Do not set the b_modified flag in block's journal head should not
until after we're sure that jbd2_journal_dirty_metadat() will not
abort with an error due to there not being enough space reserved in
the jbd2 handle.

Otherwise, future attempts to modify the buffer may lead a large
number of spurious errors and warnings.

This addresses CVE-2018-10883.

https://bugzilla.kernel.org/show_bug.cgi?id=200071



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

When expanding the extra isize space, we must never move the
system.data xattr out of the inode body.  For performance reasons, it
doesn't make any sense, and the inline data implementation assumes
that system.data xattr is never in the external xattr block.

This addresses CVE-2018-10880

https://bugzilla.kernel.org/show_bug.cgi?id=200005



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

When converting from an inode from storing the data in-line to a data
block, ext4_destroy_inline_data_nolock() was only clearing the on-disk
copy of the i_blocks[] array.  It was not clearing copy of the
i_blocks[] in ext4_inode_info, in i_data[], which is the copy actually
used by ext4_map_blocks().

This didn't matter much if we are using extents, since the extents
header would be invalid and thus the extents could would re-initialize
the extents tree.  But if we are using indirect blocks, the previous
contents of the i_blocks array will be treated as block numbers, with
potentially catastrophic results to the file system integrity and/or
user data.

This gets worse if the file system is using a 1k block size and
s_first_data is zero, but even without this, the file system can get
quite badly corrupted.

This addresses CVE-2018-10881.

https://bugzilla.kernel.org/show_bug.cgi?id=200015



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

If there is a corupted file system where the claimed depth of the
extent tree is -1, this can cause a massive buffer overrun leading to
sadness.

This addresses CVE-2018-10877.

https://bugzilla.kernel.org/show_bug.cgi?id=199417



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

The bg_flags field in the block group descripts is only valid if the
uninit_bg or metadata_csum feature is enabled.  We were not
consistently looking at this field; fix this.

Also block group #0 must never have uninitialized allocation bitmaps,
or need to be zeroed, since that's where the root inode, and other
special inodes are set up.  Check for these conditions and mark the
file system as corrupted if they are detected.

This addresses CVE-2018-10876.

https://bugzilla.kernel.org/show_bug.cgi?id=199403



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

It's really bad when the allocation bitmaps and the inode table
overlap with the block group descriptors, since it causes random
corruption of the bg descriptors.  So we really want to head those off
at the pass.

https://bugzilla.kernel.org/show_bug.cgi?id=199865



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

Regardless of whether the flex_bg feature is set, we should always
check to make sure the bits we are setting in the block bitmap are
within the block group bounds.

https://bugzilla.kernel.org/show_bug.cgi?id=199865



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org

If there an inode points to a block which is also some other type of
metadata block (such as a block allocation bitmap), the
buffer_verified flag can be set when it was validated as that other
metadata block type; however, it would make a really terrible external
attribute block.  The reason why we use the verified flag is to avoid
constantly reverifying the block.  However, it doesn't take much
overhead to make sure the magic number of the xattr block is correct,
and this will avoid potential crashes.

This addresses CVE-2018-10879.

https://bugzilla.kernel.org/show_bug.cgi?id=200001



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Cc: stable@kernel.org

In theory this should have been caught earlier when the xattr list was
verified, but in case it got missed, it's simple enough to add check
to make sure we don't overrun the xattr buffer.

This addresses CVE-2018-10879.

https://bugzilla.kernel.org/show_bug.cgi?id=200001



Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Cc: stable@kernel.org

This is very handy when debugging bugs handling maliciously corrupted
file systems.

Signed-off-by: Theodore Ts'o <tytso@mit.edu>

ext4_resize_fs() has an off-by-one bug when checking whether growing of
a filesystem will not overflow inode count. As a result it allows a
filesystem with 8192 inodes per group to grow to 64TB which overflows
inode count to 0 and makes filesystem unusable. Fix it.

Cc: stable@vger.kernel.org
Fixes: 3f8a6411


Reported-by: Jaco Kroon <jaco@uls.co.za>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>

Ext4 will always create ext4 extended attributes which do not have a
value (where e_value_size is zero) with e_value_offs set to zero.  In
most places e_value_offs will not be used in a substantive way if
e_value_size is zero.

There was one exception to this, which is in ext4_xattr_set_entry(),
where if there is a maliciously crafted file system where there is an
extended attribute with e_value_offs is non-zero and e_value_size is
0, the attempt to remove this xattr will result in a negative value
getting passed to memmove, leading to the following sadness:

[   41.225365] EXT4-fs (loop0): mounted filesystem with ordered data mode. Opts: (null)
[   44.538641] BUG: unable to handle kernel paging request at ffff9ec9a3000000
[   44.538733] IP: __memmove+0x81/0x1a0
[   44.538755] PGD 1249bd067 P4D 1249bd067 PUD 1249c1067 PMD 80000001230000e1
[   44.538793] Oops: 0003 [#1] SMP PTI
[   44.539074] CPU: 0 PID: 1470 Comm: poc Not tainted 4.16.0-rc1+ #1
    ...
[   44.539475] Call Trace:
[   44.539832]  ext4_xattr_set_entry+0x9e7/0xf80
    ...
[   44.539972]  ext4_xattr_block_set+0x212/0xea0
    ...
[   44.540041]  ext4_xattr_set_handle+0x514/0x610
[   44.540065]  ext4_xattr_set+0x7f/0x120
[   44.540090]  __vfs_removexattr+0x4d/0x60
[   44.540112]  vfs_removexattr+0x75/0xe0
[   44.540132]  removexattr+0x4d/0x80
    ...
[   44.540279]  path_removexattr+0x91/0xb0
[   44.540300]  SyS_removexattr+0xf/0x20
[   44.540322]  do_syscall_64+0x71/0x120
[   44.540344]  entry_SYSCALL_64_after_hwframe+0x21/0x86

https://bugzilla.kernel.org/show_bug.cgi?id=199347



This addresses CVE-2018-10840.

Reported-by: "Xu, Wen" <wen.xu@gatech.edu>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Cc: stable@kernel.org
Fixes: dec214d0 ("ext4: xattr inode deduplication")

If ext4_find_inline_data_nolock() returns an error it needs to get
reflected up to ext4_iget().  In order to fix this,
ext4_iget_extra_inode() needs to return an error (and not return
void).

This is related to "ext4: do not allow external inodes for inline
data" (which fixes CVE-2018-11412) in that in the errors=continue
case, it would be useful to for userspace to receive an error
indicating that file system is corrupted.

Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Cc: stable@kernel.org

The inline data feature was implemented before we added support for
external inodes for xattrs.  It makes no sense to support that
combination, but the problem is that there are a number of extended
attribute checks that are skipped if e_value_inum is non-zero.

Unfortunately, the inline data code is completely e_value_inum
unaware, and attempts to interpret the xattr fields as if it were an
inline xattr --- at which point, Hilarty Ensues.

This addresses CVE-2018-11412.

https://bugzilla.kernel.org/show_bug.cgi?id=199803



Reported-by: Jann Horn <jannh@google.com>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Fixes: e50e5129 ("ext4: xattr-in-inode support")
Cc: stable@kernel.org

This reserved space isn't committed yet but cannot be used for allocations.
For userspace it has no difference from used space. XFS already does this.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>
Fixes: 689c958c ("ext4: add project quota support")

Signed-off-by: Sean Fu <fxinrong@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

The kmem_cache_destroy() function already checks for null pointers, so
we can remove the check at the call site.

This patch also sets jbd2_handle_cache and jbd2_inode_cache to be NULL
after freeing them in jbd2_journal_destroy_handle_cache().

Signed-off-by: Wang Long <wanglong19@meituan.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>

See following dmesg output with jbd2 debug enabled:

...(start_this_handle, 313): New handle 00000000c88d6ceb going live.

...(start_this_handle, 383): Handle 00000000c88d6ceb given 53 credits (total 53, free 32681)

...(do_get_write_access, 838): journal_head 0000000002856fc0, force_copy 0

...(jbd2_journal_cancel_revoke, 421): journal_head 0000000002856fc0, cancelling revoke

We have an extra line with every messages, this is a waste of buffer,
we can fix it by removing "\n" in the caller or remove it in
the __jbd2_debug(), i checked every jbd2_debug() passed '\n' explicitly.

To avoid more lines, let's remove it inside __jbd2_debug().

Signed-off-by: Wang Shilong <wshilong@ddn.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>

When remounting ext4 from ro to rw, currently it allows its transition,
even if ext4_commit_super() returns EIO. Even worse thing is, after that,
fs/buffer complains buffer dirty bits like:

 Call trace:
 [<ffffff9750c259dc>] mark_buffer_dirty+0x184/0x1a4
 [<ffffff9750cb398c>] __ext4_handle_dirty_super+0x4c/0xfc
 [<ffffff9750c7a9fc>] ext4_file_open+0x154/0x1c0
 [<ffffff9750bea51c>] do_dentry_open+0x114/0x2d0
 [<ffffff9750bea75c>] vfs_open+0x5c/0x94
 [<ffffff9750bf879c>] path_openat+0x668/0xfe8
 [<ffffff9750bf8088>] do_filp_open+0x74/0x120
 [<ffffff9750beac98>] do_sys_open+0x148/0x254
 [<ffffff9750beade0>] SyS_openat+0x10/0x18
 [<ffffff9750a83ab0>] el0_svc_naked+0x24/0x28
 EXT4-fs (dm-1): previous I/O error to superblock detected
 Buffer I/O error on dev dm-1, logical block 0, lost sync page write
 EXT4-fs (dm-1): re-mounted. Opts: (null)
 Buffer I/O error on dev dm-1, logical block 80, lost async page write

Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

If fs is frozen after mount and before the first file open, the
update of s_last_mounted bypasses freeze protection and prints out
a WARNING splat:

$ mount /vdf
$ fsfreeze -f /vdf
$ cat /vdf/foo

[   31.578555] WARNING: CPU: 1 PID: 1415 at
fs/ext4/ext4_jbd2.c:53 ext4_journal_check_start+0x48/0x82

[   31.614016] Call Trace:
[   31.614997]  __ext4_journal_start_sb+0xe4/0x1a4
[   31.616771]  ? ext4_file_open+0xb6/0x189
[   31.618094]  ext4_file_open+0xb6/0x189

If fs is frozen, skip s_last_mounted update.

[backport hint: to apply to stable tree, need to apply also patches
 vfs: add the sb_start_intwrite_trylock() helper
 ext4: factor out helper ext4_sample_last_mounted()]

Cc: stable@vger.kernel.org
Fixes: bc0b0d6d

 ("ext4: update the s_last_mounted field in the superblock")
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>

Needed by ext4 to test frozen fs before updating s_last_mounted.

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>

Currently in ext4_punch_hole we're going to skip the mtime update if
there are no actual blocks to release. However we've actually modified
the file by zeroing the partial block so the mtime should be updated.

Moreover the sync and datasync handling is skipped as well, which is
also wrong. Fix it.

Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reported-by: Joe Habermann <joe.habermann@quantum.com>
Cc: <stable@vger.kernel.org>

The Linux VFS does not allow a way to set append/immuttable
attributes to symlinks, this is just not possible. If this is
detected inform the user as the filesystem must be corrupted.

Signed-off-by: Luis R. Rodriguez <mcgrof@kernel.org>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>

Use new return type vm_fault_t for fault handler. For now,
this is just documenting that the function returns a
VM_FAULT value rather than an errno. Once all instances are
converted, vm_fault_t will become a distinct type.

commit 1c8f4220

 ("mm: change return type to vm_fault_t")

Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Matthew Wilcox <mawilcox@microsoft.com>

When ext4_ind_map_blocks() computes a length of a hole, it doesn't count
with the fact that mapped offset may be somewhere in the middle of the
completely empty subtree. In such case it will return too large length
of the hole which then results in lseek(SEEK_DATA) to end up returning
an incorrect offset beyond the end of the hole.

Fix the problem by correctly taking offset within a subtree into account
when computing a length of a hole.

Fixes: facab4d9


CC: stable@vger.kernel.org
Reported-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

There are still some cases that we missed to set
block bitmaps corrupted bit properly:

1) block bitmap number is wrong.
2) failed to read block bitmap due to disk errors.
3) double free block bitmaps..
4) some mismatch check with bitmaps vs buddy information.

Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Signed-off-by: Wang Shilong <wshilong@ddn.com>
Reviewed-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>

There are still some cases that we missed to set
block bitmaps corrupted bit properly:

1)inode bitmap number is wrong.
2)failed to read block bitmap due to disk errors.
3)double allocations from bitmap

Also remove a duplicated call ext4_error() afer
ext4_read_inode_bitmap(), as ext4_error() have been
called inside ext4_read_inode_bitmap() properly.

Signed-off-by: Wang Shilong <wshilong@ddn.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>

Since there are many places to set inode/block bitmap
corrupt bit, add a new helper for it, which will make
codes more clear.

Signed-off-by: Wang Shilong <wshilong@ddn.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>

The only reason that sb_getblk() could fail is out of memory,
ext4 codes have returned -ENOMME for all other places except this
one, let's fix it here too.

Signed-off-by: Wang Shilong <wshilong@ddn.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Currently, creating large xattr (e.g. 2k) in ea_inode would cause
ea_inode refcount corruption, e.g.

  Pass 4: Checking reference counts
  Extended attribute inode 13 ref count is 0, should be 1. Fix? no

This is because that we save the lower 32bit of refcount in
inode->i_version and store it in raw_inode->i_disk_version on disk.
But since commit ee73f9a5 ("ext4: convert to new i_version
API"), we load/store modified i_disk_version from/to disk instead of
raw value, which causes on-disk ea_inode refcount corruption.

Fix it by loading/storing raw i_version/i_disk_version, because it's
a self-managed value in this case.

Fixes: ee73f9a5

 ("ext4: convert to new i_version API")
Cc: Tahsin Erdogan <tahsin@google.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

I hit ENOSPC error when creating new file in a newly created ext4
with ea_inode feature enabled, if selinux is enabled and ext4 is
mounted without any selinux context. e.g.

  mkfs -t ext4 -O ea_inode -F /dev/sda5
  mount /dev/sda5 /mnt/ext4
  touch /mnt/ext4/testfile  # got ENOSPC here

It turns out that we run out of journal credits in
ext4_xattr_set_handle() when creating new selinux label for the
newly created inode.

This is because that in __ext4_new_inode() we use
__ext4_xattr_set_credits() to calculate the reserved credits for new
xattr, with the 'is_create' argument being true, which implies less
credits in the ea_inode case. But we calculate the required credits
in ext4_xattr_set_handle() with 'is_create' being false, which means
we need more credits if ea_inode feature is enabled. So we don't
have enough credits and error out with ENOSPC.

Fix it by simply calling ext4_xattr_set_handle() with XATTR_CREATE
flag in ext4_initxattrs(), so we end up with requiring less credits
than reserved. The semantic of XATTR_CREATE is "Perform a pure
create, which fails if the named attribute exists already." (from
setxattr(2)), which is fine in this case, because we only call
ext4_initxattrs() on newly created inode.

Fixes: af65207c

 ("ext4: fix __ext4_new_inode() journal credits calculation")
Cc: Tahsin Erdogan <tahsin@google.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Since function ‘ext4_getfsmap_find_fixed_metadata’ can be made static,
make it so. Remove the following gcc warning (W=1):

  fs/ext4/fsmap.c:405:5: warning: no previous prototype for ‘ext4_getfsmap_find_fixed_metadata’ [-Wmissing-prototypes]

Signed-off-by: Mathieu Malaterre <malat@debian.org>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Pll KVM fixes from Radim Krčmář:
 "ARM:
   - Fix proxying of GICv2 CPU interface accesses
   - Fix crash when switching to BE
   - Track source vcpu git GICv2 SGIs
   - Fix an outdated bit of documentation

  x86:
   - Speed up injection of expired timers (for stable)"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: x86: remove APIC Timer periodic/oneshot spikes
  arm64: vgic-v2: Fix proxying of cpuif access
  KVM: arm/arm64: vgic_init: Cleanup reference to process_maintenance
  KVM: arm64: Fix order of vcpu_write_sys_reg() arguments
  KVM: arm/arm64: vgic: Fix source vcpu issues for GICv2 SGI