dax_supported() is defined whenever CONFIG_DAX is enabled. So dummy
implementation should be defined only in !CONFIG_DAX case, not in
!CONFIG_FS_DAX case.
Fixes: e2ec512825 ("dm: Call proper helper to determine dax support")
Cc: <[email protected]>
Reported-by: Geert Uytterhoeven <[email protected]>
Reported-by: Naresh Kamboju <[email protected]>
Reported-by: kernel test robot <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Dan Williams <[email protected]>
DM was calling generic_fsdax_supported() to determine whether a device
referenced in the DM table supports DAX. However this is a helper for "leaf" device drivers so that
they don't have to duplicate common generic checks. High level code
should call dax_supported() helper which that calls into appropriate
helper for the particular device. This problem manifested itself as
kernel messages:
dm-3: error: dax access failed (-95)
when lvm2-testsuite run in cases where a DM device was stacked on top of
another DM device.
Fixes: 7bf7eac8d6 ("dax: Arrange for dax_supported check to span multiple devices")
Cc: <[email protected]>
Tested-by: Adrian Huang <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Acked-by: Mike Snitzer <[email protected]>
Reported-by: kernel test robot <[email protected]>
Link: https://lore.kernel.org/r/160061715195.13131.5503173247632041975.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <[email protected]>
ext4_search_dir() and ext4_generic_delete_entry() can be called both for
standard director blocks and for inline directories stored inside inode
or inline xattr space. For the second case we didn't call
ext4_check_dir_entry() with proper constraints that could result in
accepting corrupted directory entry as well as false positive filesystem
errors like:
EXT4-fs error (device dm-0): ext4_search_dir:1395: inode #28320400:
block 113246792: comm dockerd: bad entry in directory: directory entry too
close to block end - offset=0, inode=28320403, rec_len=32, name_len=8,
size=4096
Fix the arguments passed to ext4_check_dir_entry().
Fixes: 109ba779d6 ("ext4: check for directory entries too close to block end")
CC: [email protected]
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
When remounting filesystem fails late during remount handling and
block_validity mount option is also changed during the remount, we fail
to restore system zone information to a state matching the mount option.
This is mostly harmless, just the block validity checking will not match
the situation described by the mount option. Make sure these two are always
consistent.
Reported-by: Lukas Czerner <[email protected]>
Reviewed-by: Lukas Czerner <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Currently, system zones just track ranges of block, that are "important"
fs metadata (bitmaps, group descriptors, journal blocks, etc.). This
however complicates how extent tree (or indirect blocks) can be checked
for inodes that actually track such metadata - currently the journal
inode but arguably we should be treating quota files or resize inode
similarly. We cannot run __ext4_ext_check() on such metadata inodes when
loading their extents as that would immediately trigger the validity
checks and so we just hack around that and special-case the journal
inode. This however leads to a situation that a journal inode which has
extent tree of depth at least one can have invalid extent tree that gets
unnoticed until ext4_cache_extents() crashes.
To overcome this limitation, track inode number each system zone belongs
to (0 is used for zones not belonging to any inode). We can then verify
inode number matches the expected one when verifying extent tree and
thus avoid the false errors. With this there's no need to to
special-case journal inode during extent tree checking anymore so remove
it.
Fixes: 0a944e8a6c ("ext4: don't perform block validity checks on the journal inode")
Reported-by: Wolfgang Frisch <[email protected]>
Reviewed-by: Lukas Czerner <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Currently, add_system_zone() just silently merges two added system zones
that overlap. However the overlap should not happen and it generally
suggests that some unrelated metadata overlap which indicates the fs is
corrupted. We should have caught such problems earlier (e.g. in
ext4_check_descriptors()) but add this check as another line of defense.
In later patch we also use this for stricter checking of journal inode
extent tree.
Reviewed-by: Lukas Czerner <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
A customer has reported a BUG_ON in ext4_clear_journal_err() hitting
during an LTP testing. Either this has been caused by a test setup
issue where the filesystem was being overwritten while LTP was mounting
it or the journal replay has overwritten the superblock with invalid
data. In either case it is preferable we don't take the machine down
with a BUG_ON. So handle the situation of unexpectedly missing
has_journal feature more gracefully. We issue warning and fail the mount
in the cases where the race window is narrow and the failed check is
most likely a programming error. In cases where fs corruption is more
likely, we do full ext4_error() handling before failing mount / remount.
Reviewed-by: Lukas Czerner <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Since commit 378f32bab3 ("ext4: introduce direct I/O write using iomap
infrastructure") we don't properly bail out of RWF_NOWAIT direct IO
write if underlying blocks are not allocated. Also
ext4_dio_write_checks() does not honor RWF_NOWAIT when re-acquiring
i_rwsem. Fix both issues.
Fixes: 378f32bab3 ("ext4: introduce direct I/O write using iomap infrastructure")
Cc: [email protected]
Reported-by: Filipe Manana <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Reviewed-by: Ritesh Harjani <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
When merging name events, fsids of the two involved events have to
match. Otherwise we could merge events from two different filesystems
and thus effectively loose the second event.
Backporting note: Although the commit cacfb956d4 introducing this bug
was merged for 5.7, the relevant code didn't get used in the end until
7e8283af6e ("fanotify: report parent fid + name + child fid") which
will be merged with this patch. So there's no need for backporting this.
Fixes: cacfb956d4 ("fanotify: record name info for FAN_DIR_MODIFY event")
Reported-by: Amir Goldstein <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
When user provides large buffer for events and there are lots of events
available, we can try to copy them all to userspace without scheduling
which can softlockup the kernel (furthermore exacerbated by the
contention on notification_lock). Add a scheduling point after copying
each event.
Note that usually the real underlying problem is the cost of fanotify
event merging and the resulting contention on notification_lock but this
is a cheap way to somewhat reduce the problem until we can properly
address that.
Reported-by: Francesco Ruggeri <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Reviewed-by: Amir Goldstein <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Switch link to Sourceforge in quota documentation to https and replace
link for libnl documentation with a working one from infradead.
Signed-off-by: Jan Kara <[email protected]>
Currently blk-mq does not report any event when two requests get merged
in the elevator. This then results in difficult to understand sequence
of events like:
...
8,0 34 1579 0.608765271 2718 I WS 215023504 + 40 [dbench]
8,0 34 1584 0.609184613 2719 A WS 215023544 + 56 <- (8,4) 2160568
8,0 34 1585 0.609184850 2719 Q WS 215023544 + 56 [dbench]
8,0 34 1586 0.609188524 2719 G WS 215023544 + 56 [dbench]
8,0 3 602 0.609684162 773 D WS 215023504 + 96 [kworker/3:1H]
8,0 34 1591 0.609843593 0 C WS 215023504 + 96 [0]
and you can only guess (after quite some headscratching since the above
excerpt is intermixed with a lot of other IO) that request 215023544+56
got merged to request 215023504+40. Provide proper event for request
merging like we used to do in the legacy block layer.
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Mostly for historical reasons, q->blk_trace is assigned through xchg()
and cmpxchg() atomic operations. Although this is correct, sparse
complains about this because it violates rcu annotations since commit
c780e86dd4 ("blktrace: Protect q->blk_trace with RCU") which started
to use rcu for accessing q->blk_trace. Furthermore there's no real need
for atomic operations anymore since all changes to q->blk_trace happen
under q->blk_trace_mutex and since it also makes more sense to check if
q->blk_trace is set with the mutex held earlier.
So let's just replace xchg() with rcu_replace_pointer() and cmpxchg()
with explicit check and rcu_assign_pointer(). This makes the code more
efficient and sparse happy.
Reported-by: kbuild test robot <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
The only use of I_DIRTY_TIME_EXPIRE is to detect in
__writeback_single_inode() that inode got there because flush worker
decided it's time to writeback the dirty inode time stamps (either
because we are syncing or because of age). However we can detect this
directly in __writeback_single_inode() and there's no need for the
strange propagation with I_DIRTY_TIME_EXPIRE flag.
Reviewed-by: Christoph Hellwig <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
When we are processing writeback for sync(2), move_expired_inodes()
didn't set any inode expiry value (older_than_this). This can result in
writeback never completing if there's steady stream of inodes added to
b_dirty_time list as writeback rechecks dirty lists after each writeback
round whether there's more work to be done. Fix the problem by using
sync(2) start time is inode expiry value when processing b_dirty_time
list similarly as for ordinarily dirtied inodes. This requires some
refactoring of older_than_this handling which simplifies the code
noticeably as a bonus.
Fixes: 0ae45f63d4 ("vfs: add support for a lazytime mount option")
CC: [email protected]
Reviewed-by: Christoph Hellwig <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Inode's i_io_list list head is used to attach inode to several different
lists - wb->{b_dirty, b_dirty_time, b_io, b_more_io}. When flush worker
prepares a list of inodes to writeback e.g. for sync(2), it moves inodes
to b_io list. Thus it is critical for sync(2) data integrity guarantees
that inode is not requeued to any other writeback list when inode is
queued for processing by flush worker. That's the reason why
writeback_single_inode() does not touch i_io_list (unless the inode is
completely clean) and why __mark_inode_dirty() does not touch i_io_list
if I_SYNC flag is set.
However there are two flaws in the current logic:
1) When inode has only I_DIRTY_TIME set but it is already queued in b_io
list due to sync(2), concurrent __mark_inode_dirty(inode, I_DIRTY_SYNC)
can still move inode back to b_dirty list resulting in skipping
writeback of inode time stamps during sync(2).
2) When inode is on b_dirty_time list and writeback_single_inode() races
with __mark_inode_dirty() like:
writeback_single_inode() __mark_inode_dirty(inode, I_DIRTY_PAGES)
inode->i_state |= I_SYNC
__writeback_single_inode()
inode->i_state |= I_DIRTY_PAGES;
if (inode->i_state & I_SYNC)
bail
if (!(inode->i_state & I_DIRTY_ALL))
- not true so nothing done
We end up with I_DIRTY_PAGES inode on b_dirty_time list and thus
standard background writeback will not writeback this inode leading to
possible dirty throttling stalls etc. (thanks to Martijn Coenen for this
analysis).
Fix these problems by tracking whether inode is queued in b_io or
b_more_io lists in a new I_SYNC_QUEUED flag. When this flag is set, we
know flush worker has queued inode and we should not touch i_io_list.
On the other hand we also know that once flush worker is done with the
inode it will requeue the inode to appropriate dirty list. When
I_SYNC_QUEUED is not set, __mark_inode_dirty() can (and must) move inode
to appropriate dirty list.
Reported-by: Martijn Coenen <[email protected]>
Reviewed-by: Martijn Coenen <[email protected]>
Tested-by: Martijn Coenen <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
Fixes: 0ae45f63d4 ("vfs: add support for a lazytime mount option")
CC: [email protected]
Signed-off-by: Jan Kara <[email protected]>
Currently, operations on inode->i_io_list are protected by
wb->list_lock. In the following patches we'll need to maintain
consistency between inode->i_state and inode->i_io_list so change the
code so that inode->i_lock protects also all inode's i_io_list handling.
Reviewed-by: Martijn Coenen <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
CC: [email protected] # Prerequisite for "writeback: Avoid skipping inode writeback"
Signed-off-by: Jan Kara <[email protected]>
ext4_writepages() currently works in a loop like:
start a transaction
scan inode for pages to write
map and submit these pages
stop the transaction
This loop results in starting transaction once more than is needed
because in the last iteration we start a transaction only to scan the
inode and find there are no pages to write. This can be significant
increase in number of transaction starts for single-extent files or
files that have all blocks already mapped. Furthermore we already know
from previous iteration whether there are more pages to write or not. So
propagate the information from mpage_prepare_extent_to_map() and avoid
unnecessary looping in case there are no more pages to write.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
When reserved transaction handle is unused, we subtract its reserved
credits in __jbd2_journal_unreserve_handle() called from
jbd2_journal_stop(). However this function forgets to remove reserved
credits from transaction->t_outstanding_credits and thus the transaction
space that was reserved remains effectively leaked. The leaked
transaction space can be quite significant in some cases and leads to
unnecessarily small transactions and thus reducing throughput of the
journalling machinery. E.g. fsmark workload creating lots of 4k files
was observed to have about 20% lower throughput due to this when ext4 is
mounted with dioread_nolock mount option.
Subtract reserved credits from t_outstanding_credits as well.
CC: [email protected]
Fixes: 8f7d89f368 ("jbd2: transaction reservation support")
Reviewed-by: Andreas Dilger <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
When we are evicting inode with journalled data, we may race with
transaction commit in the following way:
CPU0 CPU1
jbd2_journal_commit_transaction() evict(inode)
inode_io_list_del()
inode_wait_for_writeback()
process BJ_Forget list
__jbd2_journal_insert_checkpoint()
__jbd2_journal_refile_buffer()
__jbd2_journal_unfile_buffer()
if (test_clear_buffer_jbddirty(bh))
mark_buffer_dirty(bh)
__mark_inode_dirty(inode)
ext4_evict_inode(inode)
frees the inode
This results in use-after-free issues in the writeback code (or
the assertion added in the previous commit triggering).
Fix the problem by removing inode from writeback lists once all the page
cache is evicted and so inode cannot be added to writeback lists again.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Ext4 needs to remove inode from writeback lists after it is out of
visibility of its journalling machinery (which can still dirty the
inode). Export inode_io_list_del() for it.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Currently informational messages within block trace do not have PID
information of the process reporting the message included. With BFQ it
is sometimes useful to have the information and there's no good reason
to omit the information from the trace. So just fill in pid information
when generating note message.
Signed-off-by: Jan Kara <[email protected]>
Reviewed-by: Chaitanya Kulkarni <[email protected]>
Acked-by: Paolo Valente <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
We do not want to create initialized extents beyond end of file because
for e2fsck it is impossible to distinguish them from a case of corrupted
file size / extent tree and so it complains like:
Inode 12, i_size is 147456, should be 163840. Fix? no
Code in ext4_ext_convert_to_initialized() and
ext4_split_convert_extents() try to make sure it does not create
initialized extents beyond inode size however they check against
inode->i_size which is wrong. They should instead check against
EXT4_I(inode)->i_disksize which is the current inode size on disk.
That's what e2fsck is going to see in case of crash before all dirty
data is written. This bug manifests as generic/456 test failure (with
recent enough fstests where fsx got fixed to properly pass
FALLOC_KEEP_SIZE_FL flags to the kernel) when run with dioread_lock
mount option.
CC: [email protected]
Fixes: 21ca087a38 ("ext4: Do not zero out uninitialized extents beyond i_size")
Reviewed-by: Lukas Czerner <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Theodore Ts'o <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Commit 769071ac9f "ns: Introduce Time Namespace" broke reporting of
inotify ucounts (max_inotify_instances, max_inotify_watches) in
/proc/sys/user because it has added UCOUNT_TIME_NAMESPACES into enum
ucount_type but didn't properly update reporting in
kernel/ucount.c:setup_userns_sysctls(). This problem got fixed in commit
eeec26d5da "time/namespace: Add max_time_namespaces ucount".
Add BUILD_BUG_ON to catch a similar problem in the future.
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Acked-by: Andrei Vagin <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
When handling a page fault, we drop mmap_sem to start async readahead so
that we don't block on IO submission with mmap_sem held. However there's
no point to drop mmap_sem in case readahead is disabled. Handle that case
to avoid pointless dropping of mmap_sem and retrying the fault. This was
actually reported to block mlockall(MCL_CURRENT) indefinitely.
Fixes: 6b4c9f4469 ("filemap: drop the mmap_sem for all blocking operations")
Reported-by: Minchan Kim <[email protected]>
Reported-by: Robert Stupp <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Reviewed-by: Josef Bacik <[email protected]>
Reviewed-by: Minchan Kim <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Linus Torvalds <[email protected]>
When ext4 is running on a filesystem without a journal, it tries not to
reuse recently deleted inodes to provide better chances for filesystem
recovery in case of crash. However this logic forbids reuse of freed
inodes for up to 5 minutes and especially for filesystems with smaller
number of inodes can lead to ENOSPC errors returned when allocating new
inodes.
Fix the problem by allowing to reuse recently deleted inode if there's
no other inode free in the scanned range.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
When some events have directory id and some object id,
fanotify_event_has_fid() becomes mostly useless and confusing because we
usually need to know which type of file handle the event has. So just
drop the function and use fanotify_event_object_fh() instead.
Signed-off-by: Jan Kara <[email protected]>
Breakup the union and make them both inherit from abstract fanotify_event.
fanotify_path_event, fanotify_fid_event and fanotify_perm_event inherit
from fanotify_event.
type field in abstract fanotify_event determines the concrete event type.
fanotify_path_event, fanotify_fid_event and fanotify_perm_event are
allocated from separate memcache pools.
Rename fanotify_perm_event casting macro to FANOTIFY_PERM(), so that
FANOTIFY_PE() and FANOTIFY_FE() can be used as casting macros to
fanotify_path_event and fanotify_fid_event.
[JK: Cleanup FANOTIFY_PE() and FANOTIFY_FE() to be proper inline
functions and remove requirement that fanotify_event is the first in
event structures]
Link: https://lore.kernel.org/r/[email protected]
Suggested-by: Jan Kara <[email protected]>
Signed-off-by: Amir Goldstein <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Currently, struct fanotify_fid groups fsid and file handle and is
unioned together with struct path to save space. Also there is fh_type
and fh_len directly in struct fanotify_event to avoid padding overhead.
In the follwing patches, we will be adding more event types and this
packing makes code difficult to follow. So unpack everything and create
struct fanotify_fh which groups members logically related to file handle
to make code easier to follow. In the following patch we will pack
things again differently to make events smaller.
Signed-off-by: Jan Kara <[email protected]>
create_fd() is never used with invalid path. Also the only thing it
needs to know from fanotify_event is the path. Simplify the function to
take path directly and assume it is correct.
Signed-off-by: Jan Kara <[email protected]>
Fix a debug-only build error in ext2/xattr.c:
When building without extra debugging, (and with another patch that uses
no_printk() instead of <empty> for the ext2-xattr debug-print macros,
this build error happens:
../fs/ext2/xattr.c: In function ‘ext2_xattr_cache_insert’:
../fs/ext2/xattr.c:869:18: error: ‘ext2_xattr_cache’ undeclared (first use in
this function); did you mean ‘ext2_xattr_list’?
atomic_read(&ext2_xattr_cache->c_entry_count));
Fix the problem by removing cached entry count from the debug message
since otherwise we'd have to export the mbcache structure just for that.
Fixes: be0726d33c ("ext2: convert to mbcache2")
Reported-by: Randy Dunlap <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Lockdep complains about a chain:
sb_internal#2 --> &ei->xattr_sem#2 --> fs_reclaim
and shrink_dentry_list -> ext2_evict_inode -> ext2_xattr_delete_inode ->
down_write(ei->xattr_sem) creating a locking cycle in the reclaim path.
This is however a false positive because when we are in
ext2_evict_inode() we are the only holder of the inode reference and
nobody else should touch xattr_sem of that inode. So we cannot ever
block on acquiring the xattr_sem in the reclaim path.
Silence the lockdep warning by using down_write_trylock() in
ext2_xattr_delete_inode() to not create false locking dependency.
Reported-by: "J. R. Okajima" <[email protected]>
Reviewed-by: Ritesh Harjani <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
KASAN is reporting that __blk_add_trace() has a use-after-free issue
when accessing q->blk_trace. Indeed the switching of block tracing (and
thus eventual freeing of q->blk_trace) is completely unsynchronized with
the currently running tracing and thus it can happen that the blk_trace
structure is being freed just while __blk_add_trace() works on it.
Protect accesses to q->blk_trace by RCU during tracing and make sure we
wait for the end of RCU grace period when shutting down tracing. Luckily
that is rare enough event that we can afford that. Note that postponing
the freeing of blk_trace to an RCU callback should better be avoided as
it could have unexpected user visible side-effects as debugfs files
would be still existing for a short while block tracing has been shut
down.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=205711
CC: [email protected]
Reviewed-by: Chaitanya Kulkarni <[email protected]>
Reviewed-by: Ming Lei <[email protected]>
Tested-by: Ming Lei <[email protected]>
Reviewed-by: Bart Van Assche <[email protected]>
Reported-by: Tristan Madani <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
When CONFIG_QFMT_V2 is configured as a module, the test in
ext4_feature_set_ok() fails and so mount of filesystems with quota or
project features fails. Fix the test to use IS_ENABLED macro which
works properly even for modules.
Link: https://lore.kernel.org/r/[email protected]
Fixes: d65d87a074 ("ext4: improve explanation of a mount failure caused by a misconfigured kernel")
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Theodore Ts'o <[email protected]>
Cc: [email protected]
When a kernel is configured without CONFIG_DEV_DAX_PMEM_COMPAT, the
compilation of tools/testing/nvdimm fails with:
Building modules, stage 2.
MODPOST 11 modules
ERROR: "dax_pmem_compat_test" [tools/testing/nvdimm/test/nfit_test.ko] undefined!
Fix the problem by calling dax_pmem_compat_test() only if the kernel has
the required functionality.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Dan Williams <[email protected]>
DIR_INDEX has been introduced as a compat ext4 feature. That means that
even kernels / tools that don't understand the feature may modify the
filesystem. This works because for kernels not understanding indexed dir
format, internal htree nodes appear just as empty directory entries.
Index dir aware kernels then check the htree structure is still
consistent before using the data. This all worked reasonably well until
metadata checksums were introduced. The problem is that these
effectively made DIR_INDEX only ro-compatible because internal htree
nodes store checksums in a different place than normal directory blocks.
Thus any modification ignorant to DIR_INDEX (or just clearing
EXT4_INDEX_FL from the inode) will effectively cause checksum mismatch
and trigger kernel errors. So we have to be more careful when dealing
with indexed directories on filesystems with checksumming enabled.
1) We just disallow loading any directory inodes with EXT4_INDEX_FL when
DIR_INDEX is not enabled. This is harsh but it should be very rare (it
means someone disabled DIR_INDEX on existing filesystem and didn't run
e2fsck), e2fsck can fix the problem, and we don't want to answer the
difficult question: "Should we rather corrupt the directory more or
should we ignore that DIR_INDEX feature is not set?"
2) When we find out htree structure is corrupted (but the filesystem and
the directory should in support htrees), we continue just ignoring htree
information for reading but we refuse to add new entries to the
directory to avoid corrupting it more.
Link: https://lore.kernel.org/r/[email protected]
Fixes: dbe8944404 ("ext4: Calculate and verify checksums for htree nodes")
Reviewed-by: Andreas Dilger <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Theodore Ts'o <[email protected]>
Cc: [email protected]
UDF does not have separate preallocated table of inodes. So similarly to
XFS we pretend that every free block is also a free inode in statfs(2)
output. Clarify this in a comment.
Signed-off-by: Jan Kara <[email protected]>
UDF 2.60 standard states in section 2.2.14.2:
A partition with Access Type 3 (rewritable) shall define a Freed
Space Bitmap or a Freed Space Table, see 2.3.3. All other partitions
shall not define a Freed Space Bitmap or a Freed Space Table.
Rewritable partitions are used on media that require some form of
preprocessing before re-writing data (for example legacy MO). Such
partitions shall use Access Type 3.
Overwritable partitions are used on media that do not require
preprocessing before overwriting data (for example: CD-RW, DVD-RW,
DVD+RW, DVD-RAM, BD-RE, HD DVD-Rewritable). Such partitions shall
use Access Type 4.
however older versions of the standard didn't have this wording and
there are tools out there that create UDF filesystems with rewritable
partitions but that don't contain a Freed Space Bitmap or a Freed Space
Table on media that does not require pre-processing before overwriting a
block. So instead of forcing media with rewritable partition read-only,
base this decision on presence of a Freed Space Bitmap or a Freed Space
Table.
Reported-by: Pali Rohár <[email protected]>
Reviewed-by: Pali Rohár <[email protected]>
Fixes: b085fbe2ef ("udf: Fix crash during mount")
Link: https://lore.kernel.org/linux-fsdevel/20200112144735.hj2emsoy4uwsouxz@pali
Signed-off-by: Jan Kara <[email protected]>
Free space on filesystems with metadata or virtual partition maps
currently gets misreported. This is because these partitions are just
remapped onto underlying real partitions from which keep track of free
blocks. Take this remapping into account when counting free blocks as
well.
Reviewed-by: Pali Rohár <[email protected]>
Reported-by: Pali Rohár <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Currently we start transaction for mapping every extent for writing
using direct IO. This is unnecessary when we know we are overwriting
already allocated blocks and the overhead of starting a transaction can
be significant especially for multithreaded workloads doing small writes.
Use iomap operations that avoid starting a transaction for direct IO
overwrites.
This improves throughput of 4k random writes - fio jobfile:
[global]
rw=randrw
norandommap=1
invalidate=0
bs=4k
numjobs=16
time_based=1
ramp_time=30
runtime=120
group_reporting=1
ioengine=psync
direct=1
size=16G
filename=file1.0.0:file1.0.1:file1.0.2:file1.0.3:file1.0.4:file1.0.5:file1.0.6:file1.0.7:file1.0.8:file1.0.9:file1.0.10:file1.0.11:file1.0.12:file1.0.13:file1.0.14:file1.0.15:file1.0.16:file1.0.17:file1.0.18:file1.0.19:file1.0.20:file1.0.21:file1.0.22:file1.0.23:file1.0.24:file1.0.25:file1.0.26:file1.0.27:file1.0.28:file1.0.29:file1.0.30:file1.0.31
file_service_type=random
nrfiles=32
from 3018MB/s to 4059MB/s in my test VM running test against simulated
pmem device (note that before iomap conversion, this workload was able
to achieve 3708MB/s because old direct IO path avoided transaction start
for overwrites as well). For dax, the win is even larger improving
throughput from 3042MB/s to 4311MB/s.
Reported-by: Dan Williams <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
We cannot look at 'i->pipe' unless we know the iter is a pipe. Move the
ring_size load to a branch in iov_iter_alignment() where we've already
checked the iter is a pipe to avoid bogus dereference.
Reported-by: [email protected]
Fixes: 8cefc107ca ("pipe: Use head and tail pointers for the ring, not cursor and length")
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Al Viro <[email protected]>
When we fail to allocate string for journal device name we jump to
'error' label which tries to unlock reiserfs write lock which is not
held. Jump to 'error_unlocked' instead.
Fixes: f32485be83 ("reiserfs: delay reiserfs lock until journal initialization")
Signed-off-by: Jan Kara <[email protected]>
When a filesystem is mounted with jdev mount option, we store the
journal device name in an allocated string in superblock. However we
fail to ever free that string. Fix it.
Reported-by: [email protected]
Fixes: c3aa077648 ("reiserfs: Properly display mount options in /proc/mounts")
CC: [email protected]
Signed-off-by: Jan Kara <[email protected]>
ext4_check_dir_entry() currently does not catch a case when a directory
entry ends so close to the block end that the header of the next
directory entry would not fit in the remaining space. This can lead to
directory iteration code trying to access address beyond end of current
buffer head leading to oops.
CC: [email protected]
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Function ext4_empty_dir() doesn't correctly handle directories with
holes and crashes on bh->b_data dereference when bh is NULL. Reorganize
the loop to use 'offset' variable all the times instead of comparing
pointers to current direntry with bh->b_data pointer. Also add more
strict checking of '.' and '..' directory entries to avoid entering loop
in possibly invalid state on corrupted filesystems.
References: CVE-2019-19037
CC: [email protected]
Fixes: 4e19d6b65f ("ext4: allow directory holes")
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
iomap_dio_bio_actor() copies iter to a local variable and then limits it
to a file extent we have mapped. When IO is submitted,
iomap_dio_bio_actor() advances the original iter while the copied iter
is advanced inside bio_iov_iter_get_pages(). This logic is non-obvious
especially because both iters still point to same shared structures
(such as pipe info) so if iov_iter_advance() changes anything in the
shared structure, this scheme breaks. Let's just truncate and reexpand
the original iter as needed instead of playing games with copying iters
and keeping them in sync.
Signed-off-by: Jan Kara <[email protected]>
Reviewed-by: Darrick J. Wong <[email protected]>
Signed-off-by: Darrick J. Wong <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
When splicing using iomap_dio_rw() to a pipe, we may leak pipe pages
because bio_iov_iter_get_pages() records that the pipe will have full
extent worth of data however if file size is not block size aligned
iomap_dio_rw() returns less than what bio_iov_iter_get_pages() set up
and splice code gets confused leaking a pipe page with the file tail.
Handle the situation similarly to the old direct IO implementation and
revert iter to actually returned read amount which makes iter consistent
with value returned from iomap_dio_rw() and thus the splice code is
happy.
Fixes: ff6a9292e6 ("iomap: implement direct I/O")
CC: [email protected]
Reported-by: [email protected]
Signed-off-by: Jan Kara <[email protected]>
Reviewed-by: Darrick J. Wong <[email protected]>
Signed-off-by: Darrick J. Wong <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
The helper jbd2_handle_buffer_credits() doesn't correctly handle reserved
handles which can lead to crashes. Fix it getting of journal pointer to
work for reserved handles as well.
Fixes: a9a8344ee1 ("ext4, jbd2: Provide accessor function for handle credits")
Reported-by: Eric Biggers <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Commit 8fcc3a5806 ("ext4: rework reserved cluster accounting when
invalidating pages") moved freeing of delayed allocation reservations
from dirty page invalidation time to time when we evict corresponding
status extent from extent status tree. For inodes which don't have any
blocks allocated this may actually happen only in ext4_clear_blocks()
which is after we've dropped references to quota structures from the
inode. Thus reservation of quota leaked. Fix the problem by clearing
quota information from the inode only after evicting extent status tree
in ext4_clear_inode().
Link: https://lore.kernel.org/r/[email protected]
Reported-by: Konstantin Khlebnikov <[email protected]>
Fixes: 8fcc3a5806 ("ext4: rework reserved cluster accounting when invalidating pages")
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Theodore Ts'o <[email protected]>
Somewhat simplify the logic initializing search start and end in
ext2_try_to_allocate(). No functional change.
Signed-off-by: Jan Kara <[email protected]>
With transition to blk-mq, the elevator= kernel argument was removed as
it makes less and less sense with the current variety of devices. Since
this may surprise some users and there are advices on the Internet that
still suggest to use it, let's at least warn if the parameter is used.
Reviewed-by: Jeff Moyer <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Currently we reserve j_max_transaction_buffers / 32 for transaction
descriptor blocks. Now that revoke descriptors are accounted for
separately this estimate is unnecessarily high and we can actually
compute much tighter estimate. In the common case of 32k journal blocks
and 4k blocksize this actually reduces the amount of reserved descriptor
blocks from 256 to ~25 which allows us to fit more real data into a
transaction.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
So far we have reserved only relatively high fixed amount of revoke
credits for each transaction. We over-reserved by large amount for most
cases but when freeing large directories or files with data journalling,
the fixed amount is not enough. In fact the worst case estimate is
inconveniently large (maximum extent size) for freeing of one extent.
We fix this by doing proper estimate of the amount of blocks that need
to be revoked when removing blocks from the inode due to truncate or
hole punching and otherwise reserve just a small amount of revoke
credits for each transaction to accommodate freeing of xattrs block or
so.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Make checking of available credits in jbd2_journal_dirty_metadata() more
strict. There should be always enough credits in the handle to write all
potential revoke descriptors. Also we warn in case there are not enough
credits since this is a bug in the filesystem.
Reviewed-by: Theodore Ts'o <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Extend functions for starting, extending, and restarting transaction
handles to take number of revoke records handle must be able to
accommodate. These functions then make sure transaction has enough
credits to be able to store resulting revoke descriptor blocks. Also
revoke code tracks number of revoke records created by a handle to catch
situation where some place didn't reserve enough space for revoke
records. Similarly to standard transaction credits, space for unused
reserved revoke records is released when the handle is stopped.
On the ext4 side we currently take a simplistic approach of reserving
space for 1024 revoke records for any transaction. This grows amount of
credits reserved for each handle only by a few and is enough for any
normal workload so that we don't hit warnings in jbd2. We will refine
the logic in following commits.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Currently, journal descriptor blocks were not accounted in
transaction->t_outstanding_credits and we were just leaving some slack
space in the journal for them (in jbd2_log_space_left() and
jbd2_space_needed()). This is making proper accounting (and reservation
we want to add) of descriptor blocks difficult so switch to accounting
descriptor blocks in transaction->t_outstanding_credits and just reserve
the same amount of credits in t_outstanding credits for journal
descriptor blocks when creating transaction.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
jbd2__journal_restart() has quite some code that is common with
jbd2_journal_stop(). Factor this functionality into stop_this_handle()
helper and use it from both functions. Note that this also drops
t_handle_lock protection from jbd2__journal_restart() as
jbd2_journal_stop() does the same thing without it.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
When we drop last handle from a transaction and journal->j_barrier_count
> 0, jbd2_journal_stop() wakes up journal->j_wait_transaction_locked
wait queue. This looks pointless - wait for outstanding handles always
happens on journal->j_wait_updates waitqueue.
journal->j_wait_transaction_locked is used to wait for transaction state
changes and by start_this_handle() for waiting until
journal->j_barrier_count drops to 0. The first case is clearly
irrelevant here since only jbd2 thread changes transaction state. The
second case looks related but jbd2_journal_unlock_updates() is
responsible for the wakeup in this case. So just drop the wakeup.
Reviewed-by: Theodore Ts'o <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
If a transaction is larger than journal->j_max_transaction_buffers, that
is a bug and not a trigger for transaction commit. Also the very next
attempt to start new handle will start transaction commit anyway. So
just remove the pointless check. Arguably, we could start transaction
commit whenever the transaction size is *close* to
journal->j_max_transaction_buffers. This has a potential to reduce
latency of the next jbd2_journal_start() at the cost of somewhat smaller
transactions. However for this to have any effect, it would mean that
there isn't someone already waiting in jbd2_journal_start() which means
metadata load for the fs is pretty light anyway so probably this
optimization is not worth it.
Reviewed-by: Theodore Ts'o <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
jbd2 statistics counting number of blocks logged in a transaction was
wrong. It didn't count the commit block and more importantly it didn't
count revoke descriptor blocks. Make sure these get properly counted.
Reviewed-by: Theodore Ts'o <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Provide ext4_journal_ensure_credits_fn() function to ensure transaction
has given amount of credits and call helper function to prepare for
restarting a transaction. This allows to remove some boilerplate code
from various places, add proper error handling for the case where
transaction extension or restart fails, and reduces following changes
needed for proper revoke record reservation tracking.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Error cleanup path in ext4_alloc_branch() calls ext4_forget() on freshly
allocated indirect blocks with 'metadata' set to 1. This results in
generating revoke records for these blocks. However this is unnecessary
as the freed blocks are only allocated in the current transaction and
thus they will never be journalled. Make this cleanup path similar to
e.g. cleanup in ext4_splice_branch() and use ext4_free_blocks() to
handle block forgetting by passing EXT4_FREE_BLOCKS_FORGET and not
EXT4_FREE_BLOCKS_METADATA to ext4_free_blocks(). This also allows
allocating transaction not to reserve any credits for revoke records.
Reviewed-by: Theodore Ts'o <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Similarly to directories, EA inodes do only journalled modifications to
their data. Change ext4_should_journal_data() to return true for them so
that we don't have to special-case them during truncate.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Estimate for the number of credits needed for final freeing of inode in
ext4_evict_inode() was to small. We may modify 4 blocks (inode & sb for
orphan deletion, bitmap & group descriptor for inode freeing) and not
just 3.
[ Fixed minor whitespace nit. -- TYT ]
Fixes: e50e5129f3 ("ext4: xattr-in-inode support")
CC: [email protected]
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
When ext4_mkdir(), ext4_symlink(), ext4_create(), or ext4_mknod() fail
to add entry into directory, it ends up dropping freshly created inode
under the running transaction and thus inode truncation happens under
that transaction. That breaks assumptions that evict() does not get
called from a transaction context and at least in ext4_symlink() case it
can result in inode eviction deadlocking in inode_wait_for_writeback()
when flush worker finds symlink inode, starts to write it back and
blocks on starting a transaction. So change the code in ext4_mkdir() and
ext4_add_nondir() to drop inode reference only after the transaction is
stopped. We also have to add inode to the orphan list in that case as
otherwise the inode would get leaked in case we crash before inode
deletion is committed.
CC: [email protected]
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
With 32-bit block numbers, we don't allocate the array for journal
buffer heads large enough for corresponding descriptor tags to fill the
descriptor block. Thus we end up writing out half-full descriptor blocks
to the journal unnecessarily growing the transaction. Fix the logic to
allocate the array large enough.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
When number of free space in the journal is very low, the arithmetic in
jbd2_log_space_left() could underflow resulting in very high number of
free blocks and thus triggering assertion failure in transaction commit
code complaining there's not enough space in the journal:
J_ASSERT(journal->j_free > 1);
Properly check for the low number of free blocks.
CC: [email protected]
Reviewed-by: Theodore Ts'o <[email protected]>
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Make dquot_get_state() gracefully handle a situation when there are no
quota files present even though quotas are enabled.
Signed-off-by: Jan Kara <[email protected]>
Quota on and quota off are protected by s_umount semaphore held in
exclusive mode since commit 7d6cd73d33 "quota: Hold s_umount in
exclusive mode when enabling / disabling quotas". This makes it
impossible for dquot_disable() to race with other enabling or disabling
of quotas. Simplify the cleanup done by dquot_disable() based on this
fact and also remove some stale comments. As a bonus this cleanup makes
dquot_disable() properly handle a case when there are no quota inodes.
Signed-off-by: Jan Kara <[email protected]>
Now dquot_enable() has only two internal callers and both of them just
need to update quota flags and don't need most of checks. Just drop
dquot_enable() and fold necessary functionality into the two calling
places.
Signed-off-by: Jan Kara <[email protected]>
Use dquot_load_quota_inode from filesystems instead of dquot_enable().
In all three cases we want to load quota inode and never use the
function to update quota flags.
Signed-off-by: Jan Kara <[email protected]>
Rename vfs_load_quota_inode() to dquot_load_quota_inode() to be
consistent with naming of other functions used for enabling quota
accounting from filesystems. Also export the function and add some
sanity checks to assure filesystems are calling the function properly.
Signed-off-by: Jan Kara <[email protected]>
We already have quota inode loaded when resuming quotas. Use
vfs_load_quota() to avoid some pointless churn with the quota inode.
Signed-off-by: Jan Kara <[email protected]>
Factor out setting up of quota inode and eventual error cleanup from
vfs_load_quota_inode(). This will simplify situation for filesystems
that don't have any quota inodes.
Signed-off-by: Jan Kara <[email protected]>
Currently, block device size in not updated on second and further open
for block devices where partition scan is disabled. This is particularly
annoying for example for DVD drives as that means block device size does
not get updated once the media is inserted into a drive if the device is
already open when inserting the media. This is actually always the case
for example when pktcdvd is in use.
Fix the problem by revalidating block device size on every open even for
devices with partition scan disabled.
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
Factor out code handling revalidation of bdev on disk change into a
common helper.
Signed-off-by: Jan Kara <[email protected]>
Signed-off-by: Jens Axboe <[email protected]>
jbd2_journal_forget() jumps to 'not_jbd' branch which calls __bforget()
in cases where the buffer is clean which is pointless. In case of failed
assertion, it can be even argued that it is safer not to touch buffer's
dirty bits. Also logically it makes more sense to just jump to 'drop'
and that will make logic also simpler when we switch bh_state_lock to a
spinlock.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
__jbd2_journal_unfile_buffer() and __jbd2_journal_refile_buffer() drop
transaction's jh reference when they remove jh from a transaction. This
will be however inconvenient once we move state lock into journal_head
itself as we still need to unlock it and we'd need to grab jh reference
just for that. Move dropping of jh reference out of these functions into
the few callers.
Signed-off-by: Jan Kara <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Theodore Ts'o <[email protected]>
Use iomap_dio_rw() to wait for unaligned direct IO instead of opencoding
the wait.
Signed-off-by: Jan Kara <[email protected]>
Reviewed-by: Darrick J. Wong <[email protected]>
Signed-off-by: Darrick J. Wong <[email protected]>
Filesystems do not support doing IO as asynchronous in some cases. For
example in case of unaligned writes or in case file size needs to be
extended (e.g. for ext4). Instead of forcing filesystem to wait for AIO
in such cases, add argument to iomap_dio_rw() which makes the function
wait for IO completion. This also results in executing
iomap_dio_complete() inline in iomap_dio_rw() providing its return value
to the caller as for ordinary sync IO.
Signed-off-by: Jan Kara <[email protected]>
Reviewed-by: Darrick J. Wong <[email protected]>
Signed-off-by: Darrick J. Wong <[email protected]>
Jan Kara