| Age | Commit message (Collapse) | Author |
|---|
|
git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
"A few more fixes that have been in the works during last twp weeks.
All have a user visible effect and are stable material:
- scrub: properly update progress after calling cancel ioctl, calling
'resume' would start from the beginning otherwise
- fix subvolume reference removal, after moving out of the original
path the reference is not recognized and will lead to transaction
abort
- fix reloc root lifetime checks, could lead to crashes when there's
subvolume cleaning running in parallel
- fix memory leak when quotas get disabled in the middle of extent
accounting
- fix transaction abort in case of balance being started on degraded
mount on eg. RAID1"
* tag 'for-5.5-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: check rw_devices, not num_devices for balance
Btrfs: always copy scrub arguments back to user space
btrfs: relocation: fix reloc_root lifespan and access
btrfs: fix memory leak in qgroup accounting
btrfs: do not delete mismatched root refs
btrfs: fix invalid removal of root ref
btrfs: rework arguments of btrfs_unlink_subvol
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse
Pull fuse fix from Miklos Szeredi:
"Fix a regression in the last release affecting the ftp module of the
gvfs filesystem"
* tag 'fuse-fixes-5.5-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse:
fuse: fix fuse_send_readpages() in the syncronous read case
|
|
The fstest btrfs/154 reports
[ 8675.381709] BTRFS: Transaction aborted (error -28)
[ 8675.383302] WARNING: CPU: 1 PID: 31900 at fs/btrfs/block-group.c:2038 btrfs_create_pending_block_groups+0x1e0/0x1f0 [btrfs]
[ 8675.390925] CPU: 1 PID: 31900 Comm: btrfs Not tainted 5.5.0-rc6-default+ #935
[ 8675.392780] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.12.0-59-gc9ba527-rebuilt.opensuse.org 04/01/2014
[ 8675.395452] RIP: 0010:btrfs_create_pending_block_groups+0x1e0/0x1f0 [btrfs]
[ 8675.402672] RSP: 0018:ffffb2090888fb00 EFLAGS: 00010286
[ 8675.404413] RAX: 0000000000000000 RBX: ffff92026dfa91c8 RCX: 0000000000000001
[ 8675.406609] RDX: 0000000000000000 RSI: ffffffff8e100899 RDI: ffffffff8e100971
[ 8675.408775] RBP: ffff920247c61660 R08: 0000000000000000 R09: 0000000000000000
[ 8675.410978] R10: 0000000000000000 R11: 0000000000000000 R12: 00000000ffffffe4
[ 8675.412647] R13: ffff92026db74000 R14: ffff920247c616b8 R15: ffff92026dfbc000
[ 8675.413994] FS: 00007fd5e57248c0(0000) GS:ffff92027d800000(0000) knlGS:0000000000000000
[ 8675.416146] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 8675.417833] CR2: 0000564aa51682d8 CR3: 000000006dcbc004 CR4: 0000000000160ee0
[ 8675.419801] Call Trace:
[ 8675.420742] btrfs_start_dirty_block_groups+0x355/0x480 [btrfs]
[ 8675.422600] btrfs_commit_transaction+0xc8/0xaf0 [btrfs]
[ 8675.424335] reset_balance_state+0x14a/0x190 [btrfs]
[ 8675.425824] btrfs_balance.cold+0xe7/0x154 [btrfs]
[ 8675.427313] ? kmem_cache_alloc_trace+0x235/0x2c0
[ 8675.428663] btrfs_ioctl_balance+0x298/0x350 [btrfs]
[ 8675.430285] btrfs_ioctl+0x466/0x2550 [btrfs]
[ 8675.431788] ? mem_cgroup_charge_statistics+0x51/0xf0
[ 8675.433487] ? mem_cgroup_commit_charge+0x56/0x400
[ 8675.435122] ? do_raw_spin_unlock+0x4b/0xc0
[ 8675.436618] ? _raw_spin_unlock+0x1f/0x30
[ 8675.438093] ? __handle_mm_fault+0x499/0x740
[ 8675.439619] ? do_vfs_ioctl+0x56e/0x770
[ 8675.441034] do_vfs_ioctl+0x56e/0x770
[ 8675.442411] ksys_ioctl+0x3a/0x70
[ 8675.443718] ? trace_hardirqs_off_thunk+0x1a/0x1c
[ 8675.445333] __x64_sys_ioctl+0x16/0x20
[ 8675.446705] do_syscall_64+0x50/0x210
[ 8675.448059] entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ 8675.479187] BTRFS: error (device vdb) in btrfs_create_pending_block_groups:2038: errno=-28 No space left
We now use btrfs_can_overcommit() to see if we can flip a block group
read only. Before this would fail because we weren't taking into
account the usable un-allocated space for allocating chunks. With my
patches we were allowed to do the balance, which is technically correct.
The test is trying to start balance on degraded mount. So now we're
trying to allocate a chunk and cannot because we want to allocate a
RAID1 chunk, but there's only 1 device that's available for usage. This
results in an ENOSPC.
But we shouldn't even be making it this far, we don't have enough
devices to restripe. The problem is we're using btrfs_num_devices(),
that also includes missing devices. That's not actually what we want, we
need to use rw_devices.
The chunk_mutex is not needed here, rw_devices changes only in device
add, remove or replace, all are excluded by EXCL_OP mechanism.
Fixes: e4d8ec0f65b9 ("Btrfs: implement online profile changing")
CC: stable@vger.kernel.org # 4.4+
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ add stacktrace, update changelog, drop chunk_mutex ]
Signed-off-by: David Sterba <dsterba@suse.com>
|
|
If scrub returns an error we are not copying back the scrub arguments
structure to user space. This prevents user space to know how much
progress scrub has done if an error happened - this includes -ECANCELED
which is returned when users ask for scrub to stop. A particular use
case, which is used in btrfs-progs, is to resume scrub after it is
canceled, in that case it relies on checking the progress from the scrub
arguments structure and then use that progress in a call to resume
scrub.
So fix this by always copying the scrub arguments structure to user
space, overwriting the value returned to user space with -EFAULT only if
copying the structure failed to let user space know that either that
copying did not happen, and therefore the structure is stale, or it
happened partially and the structure is probably not valid and corrupt
due to the partial copy.
Reported-by: Graham Cobb <g.btrfs@cobb.uk.net>
Link: https://lore.kernel.org/linux-btrfs/d0a97688-78be-08de-ca7d-bcb4c7fb397e@cobb.uk.net/
Fixes: 06fe39ab15a6a4 ("Btrfs: do not overwrite scrub error with fault error in scrub ioctl")
CC: stable@vger.kernel.org # 5.1+
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Tested-by: Graham Cobb <g.btrfs@cobb.uk.net>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
|
|
If the credentials or the mm doesn't match, don't allow the task to
submit anything on behalf of this ring. The task that owns the ring can
pass the file descriptor to another task, but we don't want to allow
that task to submit an SQE that then assumes the ring mm and creds if
it needs to go async.
Cc: stable@vger.kernel.org
Suggested-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
On alpha and s390x:
fs/ubifs/debug.h:158:11: warning: format ‘%lu’ expects argument of type ‘long unsigned int’, but argument 4 has type ‘ino_t {aka unsigned int}’ [-Wformat=]
...
fs/ubifs/orphan.c:132:3: note: in expansion of macro ‘dbg_gen’
dbg_gen("deleted twice ino %lu", orph->inum);
...
fs/ubifs/orphan.c:140:3: note: in expansion of macro ‘dbg_gen’
dbg_gen("delete later ino %lu", orph->inum);
__kernel_ino_t is "unsigned long" on most architectures, but not on
alpha and s390x, where it is "unsigned int". Hence when printing an
ino_t, it should always be cast to "unsigned long" first.
Fix this by re-adding the recently removed casts.
Fixes: 8009ce956c3d2802 ("ubifs: Don't leak orphans on memory during commit")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Richard Weinberger <richard@nod.at>
|
|
In ubifs, concurrent execution of writepage and bulk read on the same file
may cause ABBA deadlock, for example (Reproduce method see Link):
Process A(Bulk-read starts from page4) Process B(write page4 back)
vfs_read wb_workfn or fsync
... ...
generic_file_buffered_read write_cache_pages
ubifs_readpage LOCK(page4)
ubifs_bulk_read ubifs_writepage
LOCK(ui->ui_mutex) ubifs_write_inode
ubifs_do_bulk_read LOCK(ui->ui_mutex)
find_or_create_page(alloc page4) ↑
LOCK(page4) <-- ABBA deadlock occurs!
In order to ensure the serialization execution of bulk read, we can't
remove the big lock 'ui->ui_mutex' in ubifs_bulk_read(). Instead, we
allow ubifs_do_bulk_read() to lock page failed by replacing
find_or_create_page(FGP_LOCK) with
pagecache_get_page(FGP_LOCK | FGP_NOWAIT).
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Suggested-by: zhangyi (F) <yi.zhang@huawei.com>
Cc: <Stable@vger.kernel.org>
Fixes: 4793e7c5e1c ("UBIFS: add bulk-read facility")
Link: https://bugzilla.kernel.org/show_bug.cgi?id=206153
Signed-off-by: Richard Weinberger <richard@nod.at>
|
|
In create_default_filesystem() when we allocate the idx node we must use
the idx_node_size we calculated just one line before, not tmp, which
contains completely other data.
Fixes: c4de6d7e4319 ("ubifs: Refactor create_default_filesystem()")
Cc: stable@vger.kernel.org # v4.20+
Reported-by: Naga Sureshkumar Relli <nagasure@xilinx.com>
Tested-by: Naga Sureshkumar Relli <nagasure@xilinx.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Richard Weinberger <richard@nod.at>
|
|
Make the FS_IOC_GETFLAGS ioctl on ubifs return the FS_ENCRYPT_FL flag on
encrypted files, like ext4 and f2fs do.
Also make this flag be ignored by FS_IOC_SETFLAGS, like ext4 and f2fs
do, since it's a recognized flag but is not directly settable.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
|
|
UBIFS's implementation of FS_IOC_SETFLAGS fails to preserve existing
inode flags that aren't settable by FS_IOC_SETFLAGS, namely the encrypt
flag. This causes the encrypt flag to be unexpectedly cleared.
Fix it by preserving existing unsettable flags, like ext4 and f2fs do.
Test case with kvm-xfstests shell:
FSTYP=ubifs KEYCTL_PROG=keyctl
. fs/ubifs/config
. ~/xfstests/common/encrypt
dev=$(__blkdev_to_ubi_volume /dev/vdc)
ubiupdatevol -t $dev
mount $dev /mnt -t ubifs
k=$(_generate_session_encryption_key)
mkdir /mnt/edir
xfs_io -c "set_encpolicy $k" /mnt/edir
echo contents > /mnt/edir/file
chattr +i /mnt/edir/file
chattr -i /mnt/edir/file
With the bug, the following errors occur on the last command:
[ 18.081559] fscrypt (ubifs, inode 67): Inconsistent encryption context (parent directory: 65)
chattr: Operation not permitted while reading flags on /mnt/edir/file
Fixes: d475a507457b ("ubifs: Add skeleton for fscrypto")
Cc: <stable@vger.kernel.org> # v4.10+
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
|
|
When log recovery is processing buffer log items, we should check that
the incoming iovec actually describes a region of memory large enough to
contain the log format and the dirty map.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
Increase XFS_BLF_DATAMAP_SIZE by 1 to fill in the implied padding at the
end of struct xfs_buf_log_format. This makes the size consistent so
that we can check it in xfs_ondisk.h, and will be needed once we start
logging attribute values.
On amd64 we get the following pahole:
struct xfs_buf_log_format {
short unsigned int blf_type; /* 0 2 */
short unsigned int blf_size; /* 2 2 */
short unsigned int blf_flags; /* 4 2 */
short unsigned int blf_len; /* 6 2 */
long long int blf_blkno; /* 8 8 */
unsigned int blf_map_size; /* 16 4 */
unsigned int blf_data_map[16]; /* 20 64 */
/* --- cacheline 1 boundary (64 bytes) was 20 bytes ago --- */
/* size: 88, cachelines: 2, members: 7 */
/* padding: 4 */
/* last cacheline: 24 bytes */
};
But on i386 we get the following:
struct xfs_buf_log_format {
short unsigned int blf_type; /* 0 2 */
short unsigned int blf_size; /* 2 2 */
short unsigned int blf_flags; /* 4 2 */
short unsigned int blf_len; /* 6 2 */
long long int blf_blkno; /* 8 8 */
unsigned int blf_map_size; /* 16 4 */
unsigned int blf_data_map[16]; /* 20 64 */
/* --- cacheline 1 boundary (64 bytes) was 20 bytes ago --- */
/* size: 84, cachelines: 2, members: 7 */
/* last cacheline: 20 bytes */
};
Notice how the amd64 compiler inserts 4 bytes of padding to the end of
the structure to ensure 8-byte alignment. Prior to "xfs: fix memory
corruption during remote attr value buffer invalidation" we would try to
write to blf_data_map[17], which is harmless on amd64 but really bad on
i386.
This shouldn't cause any changes in the ondisk logging formats because
the log code writes out the log vectors with the appropriate size for
the log item's map_size, and log recovery treats the data_map array as a
VLA.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
Complain if someone calls xfs_buf_item_init on a buffer that is larger
than the dirty bitmap can handle, or tries to log a region that's past
the end of the dirty bitmap.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
The only thing that can cause a nonzero return from
xfs_buf_item_get_format is if the kmem_alloc fails, which it can't.
Get rid of all the unnecessary error handling.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
Now that we know we don't have to take a transaction to stale the incore
buffers for a remote value, get rid of the unnecessary memory allocation
in the leaf walker and call the rmt_stale function directly. Flatten
the loop while we're at it.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
While running generic/103, I observed what looks like memory corruption
and (with slub debugging turned on) a slub redzone warning on i386 when
inactivating an inode with a 64k remote attr value.
On a v5 filesystem, maximally sized remote attr values require one block
more than 64k worth of space to hold both the remote attribute value
header (64 bytes). On a 4k block filesystem this results in a 68k
buffer; on a 64k block filesystem, this would be a 128k buffer. Note
that even though we'll never use more than 65,600 bytes of this buffer,
XFS_MAX_BLOCKSIZE is 64k.
This is a problem because the definition of struct xfs_buf_log_format
allows for XFS_MAX_BLOCKSIZE worth of dirty bitmap (64k). On i386 when we
invalidate a remote attribute, xfs_trans_binval zeroes all 68k worth of
the dirty map, writing right off the end of the log item and corrupting
memory. We've gotten away with this on x86_64 for years because the
compiler inserts a u32 padding on the end of struct xfs_buf_log_format.
Fortunately for us, remote attribute values are written to disk with
xfs_bwrite(), which is to say that they are not logged. Fix the problem
by removing all places where we could end up creating a buffer log item
for a remote attribute value and leave a note explaining why. Next,
replace the open-coded buffer invalidation with a call to the helper we
created in the previous patch that does better checking for bad metadata
before marking the buffer stale.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
Hoist the code that invalidates remote extended attribute value buffers
into a separate helper function. This prepares us for a memory
corruption fix in the next patch.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
Commit 60e4cf67a58 (reiserfs: fix extended attributes on the root
directory) introduced a regression open_xa_root started returning
-EOPNOTSUPP but it was not handled properly in reiserfs_for_each_xattr.
When the reiserfs module is built without CONFIG_REISERFS_FS_XATTR,
deleting an inode would result in a warning and chowning an inode
would also result in a warning and then fail to complete.
With CONFIG_REISERFS_FS_XATTR enabled, the xattr root would always be
present for read-write operations.
This commit handles -EOPNOSUPP in the same way -ENODATA is handled.
Fixes: 60e4cf67a582 ("reiserfs: fix extended attributes on the root directory")
CC: stable@vger.kernel.org # Commit 60e4cf67a58 was picked up by stable
Link: https://lore.kernel.org/r/20200115180059.6935-1-jeffm@suse.com
Reported-by: Michael Brunnbauer <brunni@netestate.de>
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Jan Kara <jack@suse.cz>
|
|
Buffered read in fuse normally goes via:
-> generic_file_buffered_read()
-> fuse_readpages()
-> fuse_send_readpages()
->fuse_simple_request() [called since v5.4]
In the case of a read request, fuse_simple_request() will return a
non-negative bytecount on success or a negative error value. A positive
bytecount was taken to be an error and the PG_error flag set on the page.
This resulted in generic_file_buffered_read() falling back to ->readpage(),
which would repeat the read request and succeed. Because of the repeated
read succeeding the bug was not detected with regression tests or other use
cases.
The FTP module in GVFS however fails the second read due to the
non-seekable nature of FTP downloads.
Fix by checking and ignoring positive return value from
fuse_simple_request().
Reported-by: Ondrej Holy <oholy@redhat.com>
Link: https://gitlab.gnome.org/GNOME/gvfs/issues/441
Fixes: 134831e36bbd ("fuse: convert readpages to simple api")
Cc: <stable@vger.kernel.org> # v5.4
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
|
|
Direct I/O reads can also be used with RWF_NOWAIT & co. Fix the inode
locking in xfs_file_dio_aio_read to take IOCB_NOWAIT into account.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
A previous commit moved the locking for the async sqthread, but didn't
take into account that the io-wq workers still need it. We can't use
req->in_async for this anymore as both the sqthread and io-wq workers
set it, gate the need for locking on io_wq_current_is_worker() instead.
Fixes: 8a4955ff1cca ("io_uring: sqthread should grab ctx->uring_lock for submissions")
Reported-by: Bijan Mottahedeh <bijan.mottahedeh@oracle.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
req->result is cleared when io_issue_sqe() calls io_read/write_pre()
routines. Those routines however are not called when the sqe
argument is NULL, which is the case when io_issue_sqe() is called from
io_wq_submit_work(). io_issue_sqe() may then examine a stale result if
a polled request had previously failed with -EAGAIN:
if (ctx->flags & IORING_SETUP_IOPOLL) {
if (req->result == -EAGAIN)
return -EAGAIN;
io_iopoll_req_issued(req);
}
and in turn cause a subsequently completed request to be re-issued in
io_wq_submit_work().
Signed-off-by: Bijan Mottahedeh <bijan.mottahedeh@oracle.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Detected kmemleak.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
1.
f2fs_quota_sync
-> down_read(&sbi->quota_sem)
-> dquot_writeback_dquots
-> f2fs_dquot_commit
-> down_read(&sbi->quota_sem)
2.
f2fs_quota_sync
-> down_read(&sbi->quota_sem)
-> f2fs_write_data_pages
-> f2fs_write_single_data_page
-> down_write(&F2FS_I(inode)->i_sem)
f2fs_mkdir
-> f2fs_do_add_link
-> down_write(&F2FS_I(inode)->i_sem)
-> f2fs_init_inode_metadata
-> f2fs_new_node_page
-> dquot_alloc_inode
-> f2fs_dquot_mark_dquot_dirty
-> down_read(&sbi->quota_sem)
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
In f2fs_rename(), new_page is gone after f2fs_set_link(), but it tries
to put again when whiteout is failed and jumped to put_out_dir.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
This patch moves setting I_LINKABLE early in rename2(whiteout) to avoid the
below warning.
[ 3189.163385] WARNING: CPU: 3 PID: 59523 at fs/inode.c:358 inc_nlink+0x32/0x40
[ 3189.246979] Call Trace:
[ 3189.248707] f2fs_init_inode_metadata+0x2d6/0x440 [f2fs]
[ 3189.251399] f2fs_add_inline_entry+0x162/0x8c0 [f2fs]
[ 3189.254010] f2fs_add_dentry+0x69/0xe0 [f2fs]
[ 3189.256353] f2fs_do_add_link+0xc5/0x100 [f2fs]
[ 3189.258774] f2fs_rename2+0xabf/0x1010 [f2fs]
[ 3189.261079] vfs_rename+0x3f8/0xaa0
[ 3189.263056] ? tomoyo_path_rename+0x44/0x60
[ 3189.265283] ? do_renameat2+0x49b/0x550
[ 3189.267324] do_renameat2+0x49b/0x550
[ 3189.269316] __x64_sys_renameat2+0x20/0x30
[ 3189.271441] do_syscall_64+0x5a/0x230
[ 3189.273410] entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ 3189.275848] RIP: 0033:0x7f270b4d9a49
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
META_MAPPING is used to move blocks for both encrypted and verity files.
So the META_MAPPING invalidation condition in do_checkpoint() should
consider verity too, not just encrypt.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
In low memory scenario, we can allocate multiple bios without
submitting any of them.
- f2fs_write_checkpoint()
- block_operations()
- f2fs_sync_node_pages()
step 1) flush cold nodes, allocate new bio from mempool
- bio_alloc()
- mempool_alloc()
step 2) flush hot nodes, allocate a bio from mempool
- bio_alloc()
- mempool_alloc()
step 3) flush warm nodes, be stuck in below call path
- bio_alloc()
- mempool_alloc()
- loop to wait mempool element release, as we only
reserved memory for two bio allocation, however above
allocated two bios may never be submitted.
So we need avoid using default bioset, in this patch we introduce a
private bioset, in where we enlarg mempool element count to total
number of log header, so that we can make sure we have enough
backuped memory pool in scenario of allocating/holding multiple
bios.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
Remove duplicate sbi->aw_cnt stats counter that tracks
the number of atomic files currently opened (it also shows
incorrect value sometimes). Use more relit lable sbi->atomic_files
to show in the stats.
Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
To catch f2fs bugs in write pointer handling code for zoned block
devices, check write pointers of non-open zones that current segments do
not point to. Do this check at mount time, after the fsync data recovery
and current segments' write pointer consistency fix. Or when fsync data
recovery is disabled by mount option, do the check when there is no fsync
data.
Check two items comparing write pointers with valid block maps in SIT.
The first item is check for zones with no valid blocks. When there is no
valid blocks in a zone, the write pointer should be at the start of the
zone. If not, next write operation to the zone will cause unaligned write
error. If write pointer is not at the zone start, reset the write pointer
to place at the zone start.
The second item is check between the write pointer position and the last
valid block in the zone. It is unexpected that the last valid block
position is beyond the write pointer. In such a case, report as a bug.
Fix is not required for such zone, because the zone is not selected for
next write operation until the zone get discarded.
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
On sudden f2fs shutdown, write pointers of zoned block devices can go
further but f2fs meta data keeps current segments at positions before the
write operations. After remounting the f2fs, this inconsistency causes
write operations not at write pointers and "Unaligned write command"
error is reported.
To avoid the error, compare current segments with write pointers of open
zones the current segments point to, during mount operation. If the write
pointer position is not aligned with the current segment position, assign
a new zone to the current segment. Also check the newly assigned zone has
write pointer at zone start. If not, reset write pointer of the zone.
Perform the consistency check during fsync recovery. Not to lose the
fsync data, do the check after fsync data gets restored and before
checkpoint commit which flushes data at current segment positions. Not to
cause conflict with kworker's dirfy data/node flush, do the fix within
SBI_POR_DOING protection.
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
Pull vfs fixes from Al Viro:
"Fixes for mountpoint_last() bugs (by converting to use of
lookup_last()) and an autofs regression fix from this cycle (caused by
follow_managed() breakage introduced in barrier fixes series)"
* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
fix autofs regression caused by follow_managed() changes
reimplement path_mountpoint() with less magic
|
|
xfs_check_ondisk_structs() verifies that the sizes of the data types
used by xfs are correct via the XFS_CHECK_STRUCT_SIZE() macro.
Since the structures padding can vary depending on the ABI (e.g. on
ARM OABI structures are padded to multiple of 32 bits), it may happen
that xfs_dir2_sf_entry_t size check breaks the compilation with the
assertion below:
In file included from linux/include/linux/string.h:6,
from linux/include/linux/uuid.h:12,
from linux/fs/xfs/xfs_linux.h:10,
from linux/fs/xfs/xfs.h:22,
from linux/fs/xfs/xfs_super.c:7:
In function ‘xfs_check_ondisk_structs’,
inlined from ‘init_xfs_fs’ at linux/fs/xfs/xfs_super.c:2025:2:
linux/include/linux/compiler.h:350:38:
error: call to ‘__compiletime_assert_107’ declared with attribute
error: XFS: sizeof(xfs_dir2_sf_entry_t) is wrong, expected 3
_compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
Restore the correct behavior adding __packed to the structure definition.
Cc: Darrick J. Wong <darrick.wong@oracle.com>
Suggested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
In gfs2_inode_lookup, we initialize inode->i_atime to the lowest
possibly value after gfs2_inode_refresh may already have been called.
This should be the other way around, but we didn't notice because
usually the inode type is known from the directory entry and so
gfs2_inode_lookup won't call gfs2_inode_refresh.
In addition, only initialize ip->i_no_formal_ino from no_formal_ino when
actually needed.
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
|
|
we need to reload ->d_flags after the call of ->d_manage() - the thing
might've been called with dentry still negative and have the damn thing
turned positive while we'd waited.
Fixes: d41efb522e90 "fs/namei.c: pull positivity check into follow_managed()"
Reported-by: Ian Kent <raven@themaw.net>
Tested-by: Ian Kent <raven@themaw.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
|
|
... and get rid of a bunch of bugs in it. Background:
the reason for path_mountpoint() is that umount() really doesn't
want attempts to revalidate the root of what it's trying to umount.
The thing we want to avoid actually happen from complete_walk();
solution was to do something parallel to normal path_lookupat()
and it both went overboard and got the boilerplate subtly
(and not so subtly) wrong.
A better solution is to do pretty much what the normal path_lookupat()
does, but instead of complete_walk() do unlazy_walk(). All it takes
to avoid that ->d_weak_revalidate() call... mountpoint_last() goes
away, along with everything it got wrong, and so does the magic around
LOOKUP_NO_REVAL.
Another source of bugs is that when we traverse mounts at the final
location (and we need to do that - umount . expects to get whatever's
overmounting ., if any, out of the lookup) we really ought to take
care of ->d_manage() - as it is, manual umount of autofs automount
in progress can lead to unpleasant surprises for the daemon. Easily
solved by using handle_lookup_down() instead of follow_mount().
Tested-by: Ian Kent <raven@themaw.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
|
|
If we pass back dependent work in case of links, we need to always
ensure that we call the link setup and work prep handler. If not, we
might be missing some setup for the next work item.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
If we require mm and user context, mark the request for cancellation
if we fail to acquire the desired mm.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
<linux/kernel.h> already provides a macro u64_to_user_ptr().
Use it instead of open-coding the two casts.
No change in behavior.
Link: https://lore.kernel.org/r/20191231175408.20524-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
When initializing an fs-verity hash algorithm, also initialize a mempool
that contains a single preallocated hash request object. Then replace
the direct calls to ahash_request_alloc() and ahash_request_free() with
allocating and freeing from this mempool.
This eliminates the possibility of the allocation failing, which is
desirable for the I/O path.
This doesn't cause deadlocks because there's no case where multiple hash
requests are needed at a time to make forward progress.
Link: https://lore.kernel.org/r/20191231175545.20709-1-ebiggers@kernel.org
Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
When fs-verity verifies data pages, currently it reads each Merkle tree
page synchronously using read_mapping_page().
Therefore, when the Merkle tree pages aren't already cached, fs-verity
causes an extra 4 KiB I/O request for every 512 KiB of data (assuming
that the Merkle tree uses SHA-256 and 4 KiB blocks). This results in
more I/O requests and performance loss than is strictly necessary.
Therefore, implement readahead of the Merkle tree pages.
For simplicity, we take advantage of the fact that the kernel already
does readahead of the file's *data*, just like it does for any other
file. Due to this, we don't really need a separate readahead state
(struct file_ra_state) just for the Merkle tree, but rather we just need
to piggy-back on the existing data readahead requests.
We also only really need to bother with the first level of the Merkle
tree, since the usual fan-out factor is 128, so normally over 99% of
Merkle tree I/O requests are for the first level.
Therefore, make fsverity_verify_bio() enable readahead of the first
Merkle tree level, for up to 1/4 the number of pages in the bio, when it
sees that the REQ_RAHEAD flag is set on the bio. The readahead size is
then passed down to ->read_merkle_tree_page() for the filesystem to
(optionally) implement if it sees that the requested page is uncached.
While we're at it, also make build_merkle_tree_level() set the Merkle
tree readahead size, since it's easy to do there.
However, for now don't set the readahead size in fsverity_verify_page(),
since currently it's only used to verify holes on ext4 and f2fs, and it
would need parameters added to know how much to read ahead.
This patch significantly improves fs-verity sequential read performance.
Some quick benchmarks with 'cat'-ing a 250MB file after dropping caches:
On an ARM64 phone (using sha256-ce):
Before: 217 MB/s
After: 263 MB/s
(compare to sha256sum of non-verity file: 357 MB/s)
In an x86_64 VM (using sha256-avx2):
Before: 173 MB/s
After: 215 MB/s
(compare to sha256sum of non-verity file: 223 MB/s)
Link: https://lore.kernel.org/r/20200106205533.137005-1-ebiggers@kernel.org
Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
When it builds the first level of the Merkle tree, FS_IOC_ENABLE_VERITY
sequentially reads each page of the file using read_mapping_page().
This works fine if the file's data is already in pagecache, which should
normally be the case, since this ioctl is normally used immediately
after writing out the file.
But in any other case this implementation performs very poorly, since
only one page is read at a time.
Fix this by implementing readahead using the functions from
mm/readahead.c.
This improves performance in the uncached case by about 20x, as seen in
the following benchmarks done on a 250MB file (on x86_64 with SHA-NI):
FS_IOC_ENABLE_VERITY uncached (before) 3.299s
FS_IOC_ENABLE_VERITY uncached (after) 0.160s
FS_IOC_ENABLE_VERITY cached 0.147s
sha256sum uncached 0.191s
sha256sum cached 0.145s
Note: we could instead switch to kernel_read(). But that would mean
we'd no longer be hashing the data directly from the pagecache, which is
a nice optimization of its own. And using kernel_read() would require
allocating another temporary buffer, hashing the data and tree pages
separately, and explicitly zero-padding the last page -- so it wouldn't
really be any simpler than direct pagecache access, at least for now.
Link: https://lore.kernel.org/r/20200106205410.136707-1-ebiggers@kernel.org
Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Document that fscrypt_encrypt_pagecache_blocks() allocates the bounce
page from a mempool, and document what this means for the @gfp_flags
argument.
Link: https://lore.kernel.org/r/20191231181026.47400-1-ebiggers@kernel.org
Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Currently fscrypt_zeroout_range() issues and waits on a bio for each
block it writes, which makes it very slow.
Optimize it to write up to 16 pages at a time instead.
Also add a function comment, and improve reliability by allowing the
allocations of the bio and the first ciphertext page to wait on the
corresponding mempools.
Link: https://lore.kernel.org/r/20191226160813.53182-1-ebiggers@kernel.org
Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Merge misc fixes from David Howells.
Two afs fixes and a key refcounting fix.
* dhowells:
afs: Fix afs_lookup() to not clobber the version on a new dentry
afs: Fix use-after-loss-of-ref
keys: Fix request_key() cache
|
|
Fix afs_lookup() to not clobber the version set on a new dentry by
afs_do_lookup() - especially as it's using the wrong version of the
version (we need to use the one given to us by whatever op the dir
contents correspond to rather than what's in the afs_vnode).
Fixes: 9dd0b82ef530 ("afs: Fix missing dentry data version updating")
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
afs_lookup() has a tracepoint to indicate the outcome of
d_splice_alias(), passing it the inode to retrieve the fid from.
However, the function gave up its ref on that inode when it called
d_splice_alias(), which may have failed and dropped the inode.
Fix this by caching the fid.
Fixes: 80548b03991f ("afs: Add more tracepoints")
Reported-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
I observed a hang in generic/308 while running fstests on a i686 kernel.
The hang occurred when trying to purge the pagecache on a large sparse
file that had a page created past MAX_LFS_FILESIZE, which caused an
integer overflow in the pagecache xarray and resulted in an infinite
loop.
I then noticed that Linus changed the definition of MAX_LFS_FILESIZE in
commit 0cc3b0ec23ce ("Clarify (and fix) MAX_LFS_FILESIZE macros") so
that it is now one page short of the maximum page index on 32-bit
kernels. Because the XFS function to compute max offset open-codes the
2005-era MAX_LFS_FILESIZE computation and neither the vfs nor mm perform
any sanity checking of s_maxbytes, the code in generic/308 can create a
page above the pagecache's limit and kaboom.
Fix all this by setting s_maxbytes to MAX_LFS_FILESIZE directly and
aborting the mount with a warning if our assumptions ever break. I have
no answer for why this seems to have been broken for years and nobody
noticed.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
xfs_itruncate_extents_flags() is supposed to unmap every block in a file
from EOF onwards. Oddly, it uses s_maxbytes as the upper limit to the
bunmapi range, even though s_maxbytes reflects the highest offset the
pagecache can support, not the highest offset that XFS supports.
The result of this confusion is that if you create a 20T file on a
64-bit machine, mount the filesystem on a 32-bit machine, and remove the
file, we leak everything above 16T. Fix this by capping the bunmapi
request at the maximum possible block offset, not s_maxbytes.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
Introduce a new #define for the maximum supported file block offset.
We'll use this in the next patch to make it more obvious that we're
doing some operation for all possible inode fork mappings after a given
offset. We can't use ULLONG_MAX here because bunmapi uses that to
detect when it's done.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
