commit bb88e8da00 ("erofs: use meta buffers for xattr operations")
converted xattr operations to use on-stack erofs_buf instances.
erofs_init_inode_xattrs() uses such a metabuf while reading the inline
xattr header and shared xattr id array.
Some error paths after erofs_read_metabuf() leave through out_unlock
without dropping the metabuf, so the folio reference can leak.
Consolidate the cleanup at out_unlock. erofs_put_metabuf() is a
no-op if no folio has been acquired, and this keeps all paths after
taking EROFS_I_BL_XATTR_BIT covered by a single cleanup site.
Fixes: bb88e8da00 ("erofs: use meta buffers for xattr operations")
Signed-off-by: Jia Zhu <zhujia.zj@bytedance.com>
Reviewed-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Fixes: bb88e8da00 ("erofs: use meta buffers for xattr operations")
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
After unaligned compressed extents were introduced, the following race
could occur:
[Thread 1] [Thread 2]
(z_erofs_fill_bio_vec)
<handle a Z_EROFS_PREALLOCATED_FOLIO folio>
...
filemap_add_folio (1)
(z_erofs_bind_cache)
<the same folio is found..>
..
..
folio_attach_private (2)
filemap_add_folio (3) again
Since (1) is executed but (2) hasn't been executed yet, it's possible
that another thread finds the same managed folio in z_erofs_bind_cache()
for a different pcluster and calls filemap_add_folio() again since
folio->private is still Z_EROFS_PREALLOCATED_FOLIO.
Fix this by explicitly clearing folio->private before making the folio
visible in the managed cache so that another pcluster can simply wait
on the locked managed folio as what we did for other shared cases [1].
This only impacts unaligned data compression (`-E48bit` with zstd,
for example).
[1] Commit 9e2f9d34dd ("erofs: handle overlapped pclusters out of
crafted images properly") was originally introduced to handle crafted
overlapped extents, but it addresses unaligned extents as well.
Fixes: 7361d1e376 ("erofs: support unaligned encoded data")
Reported-by: Arseniy Krasnov <avkrasnov@salutedevices.com>
Closes: https://lore.kernel.org/r/4a2f3801-fac1-42fe-ae75-da315822e088@salutedevices.com
Tested-by: Arseniy Krasnov <avkrasnov@salutedevices.com>
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Pull smb server fixes from Steve French:
- Fix two null pointer dereferences and a memory leak
* tag 'v7.1-rc4-ksmbd-server-fixes' of git://git.samba.org/ksmbd:
ksmbd: fix null pointer dereference in compare_guid_key()
ksmbd: fix null pointer dereference in proc_show_files()
ksmbd: fix SID memory leak in set_posix_acl_entries_dacl() on overflow
Pull ntfs fixes from Namjae Jeon:
- Check the index depth limit via ntfs_icx_parent_inc(), avoiding
context corruption from excessively deep child chains
- Switch security descriptor allocation to kzalloc() to avoid leaking
uninitialized memory
- Prevent an inconsistent state where vol->volume_label becomes NULL on
allocation failure
- Validate MFT records by verifying that attrs_offset sits within
bytes_in_use
- Fix an off-by-one boundary comparison, correctly catching the
out-of-range MFT record number
- Validate the attribute name offset and length bounds prior to
AT_UNUSED enumeration
- Check for a valid left neighbor before runlist merges to prevent an
8byte out-of-bounds write on crafted volumes
- Add the missing record comparison against $MFTMirr during mount
- Fix wrong inode lookup when writing extent MFT records
- Redirty folio on memory allocation failure in ntfs_write_mft_block()
- Capture and propagate $MFTMirr sync errors during writeback
- Ensure MFT mirror and synchronous writes wait for I/O completion
- Fix buffer overflow/heap over-read in ntfs_bdev_write() when cluster
size is smaller than PAGE_SIZE
- Fix use-after-free in ntfs_inode_sync_filename() when parent index
inode is evicted while still holding its mrec_lock
- Update resident attribute length validation to match $AttrDef
- Fix refcount underflow and UAF of the global upcase table
- Fix two smatch warnings
* tag 'ntfs-for-7.1-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/linkinjeon/ntfs:
ntfs: restore $MFT mirror contents check
ntfs: fix empty_buf and ra lifetime bugs in ntfs_empty_logfile()
ntfs: validate attribute name bounds before returning it
ntfs: fix MFT bitmap scan 2^32 boundary check
ntfs: validate MFT attrs_offset against bytes_in_use
ntfs: fix missing kstrdup() error check in ntfs_write_volume_label()
ntfs: avoid leaking uninitialised bytes in new security descriptors
ntfs: fix out-of-bounds write in ntfs_index_walk_down()
ntfs: fix out-of-bounds write in ntfs_rl_collapse_range() merge path
ntfs: fix variable dereferenced before check ni in ntfs_attr_open()
ntfs: fix default_upcase refcount underflow and UAF on fs_context teardown
ntfs: match ntfs_resident_attr_min_value_length with $AttrDef
ntfs: avoid use-after-free of index inode in ntfs_inode_sync_filename()
ntfs: fix copy length in ntfs_bdev_write() for non-page-aligned start
ntfs: wait for sync mft writes to complete
ntfs: capture mft mirror sync errors in ntfs_write_mft_block()
ntfs: redirty folio when ntfs_write_mft_block() runs out of memory
ntfs: use base mft_no when looking up base inode for extent record
ntfs: fix variable dereferenced before check ni and attr in ntfs_attrlist_entry_add()
Pull vfs fixes from Christian Brauner:
"This contains a fixes for the current development cycle. Note that AI
related review sometimes delays fixes a bit because we find more fixes
for the fixes. I might try and send smaller but more fixes PRs if this
trend keeps up.
- Fix various netfslib bugs
- Fix an out-of-bounds write when listing idmappings
- Fix the return values in jfs_mkdir() and orangefs_mkdir()
- Fix a writeback writeback array overflow in fuse
- Fix a forced iversion increment on lazytime timestamp updates
- Reject a negative timeval component in kern_select()
- Fix error return when vfs_mkdir() fails in the cachefiles code
- Fix wrong error code returned for pidns ioctls"
* tag 'vfs-7.1-rc5.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs: (31 commits)
cachefiles: Fix error return when vfs_mkdir() fails
afs: Fix the locking used by afs_get_link()
netfs, afs: Fix write skipping in dir/link writepages
netfs: Fix netfs_read_folio() to wait on writeback
netfs: Fix folio->private handling in netfs_perform_write()
netfs: Fix partial invalidation of streaming-write folio
netfs: Fix potential UAF in netfs_unlock_abandoned_read_pages()
netfs: Fix leak of request in netfs_write_begin() error handling
netfs: Fix early put of sink folio in netfs_read_gaps()
netfs: Fix write streaming disablement if fd open O_RDWR
netfs: Fix read-gaps to remove netfs_folio from filled folio
netfs: Fix potential deadlock in write-through mode
netfs: Fix streaming write being overwritten
netfs: Defer the emission of trace_netfs_folio()
netfs: Fix netfs_invalidate_folio() to clear dirty bit if all changes gone
netfs: Fix overrun check in netfs_extract_user_iter()
netfs: fix error handling in netfs_extract_user_iter()
netfs: Fix potential uninitialised var in netfs_extract_user_iter()
netfs: fix VM_BUG_ON_FOLIO() issue in netfs_write_begin() call
netfs: Fix zeropoint update where i_size > remote_i_size
...
Pull ceph fixes from Ilya Dryomov:
"An important patch from Hristo that squashes a folio reference leak
that could lead to OOM kills in CephFS and a number of miscellaneous
fixes from Raphael and Slava.
All but two are marked for stable"
* tag 'ceph-for-7.1-rc4' of https://github.com/ceph/ceph-client:
libceph: Fix potential null-ptr-deref in decode_choose_args()
libceph: handle rbtree insertion error in decode_choose_args()
libceph: Fix potential out-of-bounds access in osdmap_decode()
ceph: put folios not suitable for writeback
ceph: add ceph_has_realms_with_quotas() check to ceph_quota_update_statfs()
libceph: Fix potential out-of-bounds access in __ceph_x_decrypt()
ceph: fix BUG_ON in __ceph_build_xattrs_blob() due to stale blob size
ceph: fix a buffer leak in __ceph_setxattr()
libceph: Fix unnecessarily high ceph_decode_need() for uniform bucket
libceph: Fix potential out-of-bounds access in crush_decode()
Pull btrfs fixes from David Sterba:
- fixup warning when allocating memory for readahead, __GFP_NOWARN was
accidentally dropped when setting mapping constraints
- in tracepoint of file sync, fix sleeping in atomic context when
handling dentries
- harden initial loading of block group on crafted/fuzzed images,
iterate all chunk mapping entries unconditionally
- fix freeing pages of submitted io after checking for errors
- fix incorrect inode size after remount when using fallocate KEEP_SIZE
mode (also requires disabled 'no-holes' feature)
* tag 'for-7.1-rc3-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: fix incorrect i_size after remount caused by KEEP_SIZE prealloc gap
btrfs: only release the dirty pages io tree after successful writes
btrfs: tracepoints: fix sleep while in atomic context in btrfs_sync_file()
btrfs: always pass __GFP_NOWARN from add_ra_bio_pages()
btrfs: fix check_chunk_block_group_mappings() to iterate all chunk maps
Pull xfs fixes from Carlos Maiolino:
"A few bug fixes, nothing really special stands out"
* tag 'xfs-fixes-7.1-rc4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
xfs: Fix typo in comment
xfs: fix the "limiting open zones" message
xfs: flush delalloc blocks on ENOSPC in xfs_trans_alloc_icreate
xfs: check da node block pad field during scrub
xfs: fix memory leak for data allocated by xfs_zone_gc_data_alloc()
xfs: fix memory leak on error in xfs_alloc_zone_info()
xfs: check directory data block header padding in scrub
xfs: zero directory data block padding on write verification
xfs: zero entire directory data block header region at init
xfs: remove the meaningless XFS_ALLOC_FLAG_FREEING
Pull nfsd fixes from Chuck Lever:
"Fixes for this release:
- Correctness fix for the new sunrpc cache netlink protocol
Marked for stable:
- Correctness fixes for delegated attributes
- Prevent an infinite loop when revoking layouts"
* tag 'nfsd-7.1-1' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
NFSD: Fix infinite loop in layout state revocation
sunrpc: start cache request seqno at 1 to fix netlink GET_REQS
nfsd: update mtime/ctime on COPY in presence of delegated attributes
nfsd: update mtime/ctime on CLONE in presense of delegated attributes
nfsd: fix file change detection in CB_GETATTR
nfsd: fix GET_DIR_DELEGATION when VFS leases are disabled
Pull block fixes from Jens Axboe:
- NVMe merge request via Keith:
- Fix memory leak on a passthrough integrity mapping failure (Keith)
- Hide secrets behind debug option (Hannes)
- Fix pci use-after-free for host memory buffer (Chia-Lin Kao)
- Fix tcp taregt use-after-free for data digest (Sagi)
- Revert a mistaken quirk (Alan Cui)
- Fix uevent and controller state race condition (Maurizio)
- Fix apple submission queue re-initialization (Nick Chan)
- Three fixes for blk-integrity, fixing an issue with the user data
mapping and two problems with recomputing number of segments
- Two fixes for the iov_iter bounce buffering
- Fix for the handling of dead zoned write plugs
- ublk max_sectors validation fix, with associated selftest addition
* tag 'block-7.1-20260515' of git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux:
nvme-apple: Reset q->sq_tail during queue init
block: align down bounces bios
block: pass a minsize argument to bio_iov_iter_bounce
selftests: ublk: cap nthreads to kernel's actual nr_hw_queues
block: fix handling of dead zone write plugs
block: bio-integrity: Fix null-ptr-deref in bio_integrity_map_user()
block: recompute nr_integrity_segments in blk_insert_cloned_request
block: don't overwrite bip_vcnt in bio_integrity_copy_user()
nvme: fix race condition between connected uevent and STARTED_ONCE flag
Revert "nvme: add quirk NVME_QUIRK_IGNORE_DEV_SUBNQN for 144d:a808"
nvmet-tcp: Fix potential UAF when ddgst mismatch
nvme-pci: fix use-after-free in nvme_free_host_mem()
nvmet-auth: Do not print DH-HMAC-CHAP secrets
nvme: fix bio leak on mapping failure
nvme: make prp passthrough usage less scary
ublk: reject max_sectors smaller than PAGE_SECTORS in parameter validation
When vfs_mkdir() fails, the error code is not extracted from the
returned error pointer. This causes mkdir_error to be reached with
ret=0, which leads to returning ERR_PTR(0) (NULL) instead of a
proper error pointer.
Fix this by extracting the error code from the error pointer when
vfs_mkdir() fails.
Fixes: 406fad7698 ("cachefiles: Fix oops in vfs_mkdir from cachefiles_get_directory")
Signed-off-by: Hongling Zeng <zenghongling@kylinos.cn>
Link: https://patch.msgid.link/20260513103406.202320-1-zenghongling@kylinos.cn
Signed-off-by: Christian Brauner <brauner@kernel.org>
SMB2 READ response validation in cifs_readv_receive() and
handle_read_data() checks data_offset + data_len against the received
buffer length. Both values are attacker-controlled fields from the
server response and are stored as unsigned int, so the addition can
wrap before the bounds check:
fs/smb/client/transport.c:1259
if (!use_rdma_mr && (data_offset + data_len > buflen))
fs/smb/client/smb2ops.c:4839
else if (buf_len >= data_offset + data_len)
A malicious SMB server can use this to bypass validation. In the
non-encrypted receive path the client attempts an oversized socket
read and stalls for the SMB response timeout (180 seconds) before
reconnecting. In the SMB3 encrypted path, runtime testing shows the
malformed length can reach copy_to_iter() in handle_read_data() with
attacker-controlled size, where usercopy hardening stops the oversized
copy before bytes reach userspace.
Guard both call sites with check_add_overflow(), which is already
used elsewhere in this subsystem (smb2pdu.c). On overflow, treat the
response as malformed and reject with -EIO.
Signed-off-by: Jeremy Erazo <mendozayt13@gmail.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
On 32-bit architectures, the infinite loop is as follows:
len = p->ErrorDataLength == 0xfffffff8
u8 *next = p->ErrorContextData + len
next == p
On 32-bit architectures, the out-of-bounds read is as follows:
len = p->ErrorDataLength == 0xfffffff0
u8 *next = p->ErrorContextData + len
next == (u8 *)p - 8
Reported-by: ChenXiaoSong <chenxiaosong@kylinos.cn>
Fixes: 76894f3e2f ("cifs: improve symlink handling for smb2+")
Cc: stable@vger.kernel.org
Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn>
Signed-off-by: Steve French <stfrench@microsoft.com>
session_fd_check() walks the per-inode m_op_list during durable-handle
session teardown and sets op->conn = NULL for every opinfo whose conn
matched the closing session's connection. The matching opinfo, however,
stays linked in its per-ClientGuid lease_table_list entry's lb->lease_list
because destroy_lease_table() only runs on full TCP-connection teardown,
not on SESSION_LOGOFF.
If the same TCP connection then negotiates a fresh session with the
same ClientGuid (ClientGuid is bound to NEGOTIATE, not the session, and
is unchanged across LOGOFF + SETUP) and issues a SMB2 CREATE with a
lease context on a different inode, find_same_lease_key() walks
lb->lease_list, reaches the stale opinfo, and calls compare_guid_key(),
which unconditionally dereferences opinfo->conn->ClientGUID. The conn
pointer is NULL and the kernel panics.
Reproducer requires only a successful SMB2 SESSION_SETUP and a share
configured with 'durable handles = yes'. KASAN report on mainline
70390501d1:
general protection fault, probably for non-canonical address
0xdffffc0000000069: 0000 [#1] SMP KASAN PTI
KASAN: null-ptr-deref in range [0x0000000000000348-0x000000000000034f]
Workqueue: ksmbd-io handle_ksmbd_work
RIP: 0010:bcmp+0x5b/0x230
Call Trace:
compare_guid_key+0x4b/0xd0
find_same_lease_key+0x324/0x690
smb2_open+0x6aea/0x8e60
handle_ksmbd_work+0x796/0xee0
...
Faulting address 0x348 is the offset of ClientGUID within struct
ksmbd_conn, confirming opinfo->conn was NULL.
Read opinfo->conn once and bail out if it has been cleared by a
concurrent session_fd_check(). A half-detached opinfo cannot be the
owner of an active lease, so returning 0 is the correct match result.
Fixes: c8efcc7861 ("ksmbd: add support for durable handles v1/v2")
Cc: stable@vger.kernel.org
Signed-off-by: Jeremy Laratro <research@aradex.io>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
When a SMB2 client opens a file with a durable v2 handle and then issues
SMB2 SESSION_LOGOFF, session_fd_check() clears fp->tcon = NULL on the
reconnectable file pointer but leaves the fp registered in global_ft.idr
until the durable scavenger fires (up to fp->durable_timeout seconds
later).
During that window any read of /proc/fs/ksmbd/files (mode 0400) panics
the kernel because proc_show_files() walks global_ft.idr and
unconditionally dereferences fp->tcon->id with no NULL guard.
Reproducer requires only a successful SMB2 SESSION_SETUP and a share
configured with 'durable handles = yes'. KASAN report on mainline
70390501d1:
general protection fault, probably for non-canonical address
0xdffffc0000000000: 0000 [#1] SMP KASAN PTI
KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
RIP: 0010:proc_show_files+0x118/0x740
Call Trace:
proc_show_files+0x118/0x740
seq_read_iter+0x4ef/0xe10
proc_reg_read_iter+0x1b7/0x280
...
Guard the dereference. A durable-disconnected fp legitimately has no
tcon; report its tree id as 0 rather than oopsing.
Fixes: b38f99c121 ("ksmbd: add procfs interface for runtime monitoring and statistics")
Cc: stable@vger.kernel.org
Signed-off-by: Jeremy Laratro <research@aradex.io>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Commit 299f962c0b ("ksmbd: use check_add_overflow() to prevent u16
DACL size overflow") added check_add_overflow() guards that break out
of the ACE-building loops in set_posix_acl_entries_dacl() when the
accumulated DACL size would wrap past 65535.
However, each iteration allocates a struct smb_sid via kmalloc_obj()
at the top of the loop and relies on the kfree(sid) call at the end
of the loop body (the 'pass_same_sid' label in the first loop, and
the explicit kfree at the tail of the second loop) to release it.
The newly introduced 'break' statements bypass those kfree() calls,
leaking the sid buffer every time an overflow is detected.
A malicious or malformed file with enough POSIX ACL entries to trip
the overflow check will leak one or more struct smb_sid allocations
on every request that touches the file's DACL, providing a trivial
kernel memory exhaustion vector.
Free sid before breaking out of the loops to plug the leak.
Fixes: 299f962c0b ("ksmbd: use check_add_overflow() to prevent u16 DACL size overflow")
Cc: stable@vger.kernel.org
Signed-off-by: Ferry Meng <mengferry@linux.alibaba.com>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
When bouncing for block size > PAGE_SIZE file systems that require
file system block size alignment (e.g. zoned XFS), the bio needs to
be big enough to fit an entire block.
Fixes: 8dd5e7c75d ("block: add helpers to bounce buffer an iov_iter into bios")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@kernel.org>
Link: https://patch.msgid.link/20260507050153.1298375-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
New Infolevels for QUERY_DIR (and QUERY_INFO) levels 78 through 81 are
now being used by Windows clients and were added to the documentation.
Add defines for them (and correct some typos in documentation). See
MS-SMB2 2.2.33 and MS-FSCC 2.4
Signed-off-by: Steve French <stfrench@microsoft.com>
The afs filesystem in the kernel doesn't do locking correctly for symbolic
links. There are a number of problems:
(1) It doesn't do any locking around afs_read_single() to prevent races
between multiple ->get_link() calls, thereby allowing the possibility
of leaks.
(2) It doesn't use RCU barriering when accessing the buffer pointers
during RCU pathwalk.
(3) It can race with another thread updating the contents of the symlink
if a third party updated it on the server.
Fix this by the following means:
(0) Move symlink handling into its own file as this makes it more
complicated.
(1) Take the validate_lock around afs_read_single() to prevent races
between multiple ->get_link() calls.
(2) Keep a separate copy of the symlink contents with an rcu_head. This
is always going to be a lot smaller than a page, so it can be
kmalloc'd and save quite a bit of memory. It also needs a refcount
for non-RCU pathwalk.
(3) Split the symlink read and write-to-cache routines in afs from those
for directories.
(4) Discard the I/O buffer as soon as the write-to-cache completes as this
is a full page (plus a folio_queue).
(5) If there's no cache, discard the I/O buffer immediately after reading
and copying if there is no cache.
Fixes: eae9e78951 ("afs: Use netfslib for symlinks, allowing them to be cached")
Fixes: 6698c02d64 ("afs: Locally initialise the contents of a new symlink on creation")
Closes: https://sashiko.dev/#/patchset/20260326104544.509518-1-dhowells%40redhat.com
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-25-dhowells@redhat.com
cc: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Fix netfs_write_single() and afs_single_writepages() to better handle a
write that would be skipped due to lock contention and WB_SYNC_NONE by
returning 1 from netfs_write_single() if it skipped and making
afs_single_writepages() skip also. If a skip occurs, the inode must be
re-marked as the VFS may have cleared the mark.
This is really only theoretical for directories in netfs_write_single() as
the only path to that is through afs_single_writepages() that takes the
->validate_lock around it, thereby serialising it.
Fixes: 6dd8093661 ("afs: Use netfslib for directories")
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-24-dhowells@redhat.com
cc: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Under some circumstances, netfs_perform_write() doesn't correctly
manipulate folio->private between NULL, NETFS_FOLIO_COPY_TO_CACHE, pointing
to a group and pointing to a netfs_folio struct, leading to potential
multiple attachments of private data with associated folio ref leaks and
also leaks of netfs_folio structs or netfs_group refs.
Fix this by consolidating the place at which a folio is marked uptodate in
one place and having that look at what's attached to folio->private and
decide how to clean it up and then set the new group. Also, the content
shouldn't be flushed if group is NULL, even if a group is specified in the
netfs_group parameter, as that would be the case for a new folio. A
filesystem should always specify netfs_group or never specify netfs_group.
The Sashiko auto-review tool noted that it was theoretically possible that
the fpos >= ctx->zero_point section might leak if it modified a streaming
write folio. This is unlikely, but with a network filesystem, third party
changes can happen. It also pointed out that __netfs_set_group() would
leak if called multiple times on the same folio from the "whole folio
modify section".
Fixes: 8f52de0077 ("netfs: Reduce number of conditional branches in netfs_perform_write()")
Closes: https://sashiko.dev/#/patchset/20260414082004.3756080-1-dhowells%40redhat.com
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-22-dhowells@redhat.com
cc: Paulo Alcantara <pc@manguebit.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Fix netfs_read_gaps() to release the sink page it uses after waiting for
the request to complete. The way the sink page is used is that an
ITER_BVEC-class iterator is created that has the gaps from the target folio
at either end, but has the sink page tiled over the middle so that a single
read op can fill in both gaps.
The bug was found by KASAN detecting a UAF on the generic/075 xfstest in
the cifsd kernel thread that handles reception of data from the TCP socket:
BUG: KASAN: use-after-free in _copy_to_iter+0x48a/0xa20
Write of size 885 at addr ffff888107f92000 by task cifsd/1285
CPU: 2 UID: 0 PID: 1285 Comm: cifsd Not tainted 7.0.0 #6 PREEMPT(lazy)
Call Trace:
dump_stack_lvl+0x5d/0x80
print_report+0x17f/0x4f1
kasan_report+0x100/0x1e0
kasan_check_range+0x10f/0x1e0
__asan_memcpy+0x3c/0x60
_copy_to_iter+0x48a/0xa20
__skb_datagram_iter+0x2c9/0x430
skb_copy_datagram_iter+0x6e/0x160
tcp_recvmsg_locked+0xce0/0x1130
tcp_recvmsg+0xeb/0x300
inet_recvmsg+0xcf/0x3a0
sock_recvmsg+0xea/0x100
cifs_readv_from_socket+0x3a6/0x4d0 [cifs]
cifs_read_iter_from_socket+0xdd/0x130 [cifs]
cifs_readv_receive+0xaad/0xb10 [cifs]
cifs_demultiplex_thread+0x1148/0x1740 [cifs]
kthread+0x1cf/0x210
Fixes: ee4cdf7ba8 ("netfs: Speed up buffered reading")
Reported-by: Steve French <sfrench@samba.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-18-dhowells@redhat.com
Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.org>
cc: Paulo Alcantara <pc@manguebit.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
In netfs_perform_write(), "write streaming" (the caching of dirty data in
dirty but !uptodate folios) is performed to avoid the need to read data
that is just going to get immediately overwritten. However, this is/will
be disabled in three circumstances: if the fd is open O_RDWR, if fscache is
in use (as we need to round out the blocks for DIO) or if content
encryption is enabled (again for rounding out purposes).
The idea behind disabling it if the fd is open O_RDWR is that we'd need to
flush the write-streaming page before we could read the data, particularly
through mmap. But netfs now fills in the gaps if ->read_folio() is called
on the page, so that is unnecessary. Further, this doesn't actually work
if a separate fd is open for reading.
Fix this by removing the check for O_RDWR, thereby allowing streaming
writes even when we might read.
This caused a number of problems with the generic/522 xfstest, but those
are now fixed.
Fixes: c38f4e96e6 ("netfs: Provide func to copy data to pagecache for buffered write")
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-17-dhowells@redhat.com
cc: Paulo Alcantara <pc@manguebit.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Fix netfs_read_gaps() to remove the netfs_folio record from the folio
record before marking the folio uptodate if it successfully fills the gaps
around the dirty data in a streaming write folio (dirty, but not uptodate).
Found with:
fsx -q -N 1000000 -p 10000 -o 128000 -l 600000 \
/xfstest.test/junk --replay-ops=junk.fsxops
using the following as junk.fsxops:
truncate 0x0 0x138b1 0x8b15d *
write 0x507ee 0x10df7 0x927c0
write 0x19993 0x10e04 0x927c0 *
mapwrite 0x66214 0x1a253 0x927c0
copy_range 0xb704 0x89b9 0x24429 0x79380
write 0x2402b 0x144a2 0x90660 *
mapwrite 0x204d5 0x140a0 0x927c0 *
copy_range 0x1f72c 0x137d0 0x7a906 0x927c0 *
read 0 0x9157c 0x9157c
on cifs with the default cache option.
It shows folio 0x24 misbehaving if the FMODE_READ check is commented out in
netfs_perform_write():
if (//(file->f_mode & FMODE_READ) ||
netfs_is_cache_enabled(ctx)) {
and no fscache. This was initially found with the generic/522 xfstest.
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-16-dhowells@redhat.com
Fixes: ee4cdf7ba8 ("netfs: Speed up buffered reading")
cc: Paulo Alcantara <pc@manguebit.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Fix netfs_advance_writethrough() to always unlock the supplied folio and to
mark it dirty if it isn't yet written to the end. Unfortunately, it can't
be marked for writeback until the folio is done with as that may cause a
deadlock against mmapped reads and writes.
Even though it has been marked dirty, premature writeback can't occur as
the caller is holding both inode->i_rwsem (which will prevent concurrent
truncation, fallocation, DIO and other writes) and ictx->wb_lock (which
will cause flushing to wait and writeback to skip or wait).
Note that this may be easier to deal with once the queuing of folios is
split from the generation of subrequests.
Fixes: 288ace2f57 ("netfs: New writeback implementation")
Closes: https://sashiko.dev/#/patchset/20260427154639.180684-1-dhowells%40redhat.com
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-15-dhowells@redhat.com
cc: Paulo Alcantara <pc@manguebit.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
In order to avoid reading whilst writing, netfslib will allow "streaming
writes" in which dirty data is stored directly into folios without reading
them first. Such folios are marked dirty but may not be marked uptodate.
If a folio is entirely written by a streaming write, uptodate will be set,
otherwise it will have a netfs_folio struct attached to ->private recording
the dirty region.
In the event that a partially written streaming write page is to be
overwritten entirely by a single write(), netfs_perform_write() will try to
copy over it, but doesn't discard the netfs_folio if it succeeds; further,
it doesn't correctly handle a partial copy that overwrites some of the
dirty data.
Fix this by the following:
(1) If the folio is successfully overwritten, free the netfs_folio struct
before marking the page uptodate.
(2) If the copy to the folio partially fails, but short of the dirty data,
just ignore the copy.
(3) If the copy partially fails and overwrites some of the dirty data,
accept the copy, update the netfs_folio struct to record the new data.
If the folio is now filled, free the netfs_folio and set uptodate,
otherwise return a partial write.
Found with:
fsx -q -N 1000000 -p 10000 -o 128000 -l 600000 \
/xfstest.test/junk --replay-ops=junk.fsxops
using the following as junk.fsxops:
truncate 0x0 0 0x927c0
write 0x63fb8 0x53c8 0
copy_range 0xb704 0x19b9 0x24429 0x79380
write 0x2402b 0x144a2 0x90660 *
write 0x204d5 0x140a0 0x927c0 *
copy_range 0x1f72c 0x137d0 0x7a906 0x927c0 *
read 0x00000 0x20000 0x9157c
read 0x20000 0x20000 0x9157c
read 0x40000 0x20000 0x9157c
read 0x60000 0x20000 0x9157c
read 0x7e1a0 0xcfb9 0x9157c
on cifs with the default cache option.
It shows folio 0x24 misbehaving if the FMODE_READ check is commented out in
netfs_perform_write():
if (//(file->f_mode & FMODE_READ) ||
netfs_is_cache_enabled(ctx)) {
and no fscache. This was initially found with the generic/522 xfstest.
Fixes: 8f52de0077 ("netfs: Reduce number of conditional branches in netfs_perform_write()")
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-14-dhowells@redhat.com
cc: Paulo Alcantara <pc@manguebit.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
If a streaming write is made, this will leave the relevant modified folio
in a not-uptodate, but dirty state with a netfs_folio struct hung off of
folio->private indicating the dirty range. Subsequently truncating the
file such that the dirty data in the folio is removed, but the first part
of the folio theoretically remains will cause the netfs_folio struct to be
discarded... but will leave the dirty flag set.
If the folio is then read via mmap(), netfs_read_folio() will see that the
page is dirty and jump to netfs_read_gaps() to fill in the missing bits.
netfs_read_gaps(), however, expects there to be a netfs_folio struct
present and can oops because truncate removed it.
Fix this by calling folio_cancel_dirty() in netfs_invalidate_folio() in the
event that all the dirty data in the folio is erased (as nfs does).
Also add some tracepoints to log modifications to a dirty page.
This can be reproduced with something like:
dd if=/dev/zero of=/xfstest.test/foo bs=1M count=1
umount /xfstest.test
mount /xfstest.test
xfs_io -c "w 0xbbbf 0xf96c" \
-c "truncate 0xbbbf" \
-c "mmap -r 0xb000 0x11000" \
-c "mr 0xb000 0x11000" \
/xfstest.test/foo
with fscaching disabled (otherwise streaming writes are suppressed) and a
change to netfs_perform_write() to disallow streaming writes if the fd is
open O_RDWR:
if (//(file->f_mode & FMODE_READ) || <--- comment this out
netfs_is_cache_enabled(ctx)) {
It should be reproducible even without this change, but if prevents the
above trivial xfs_io command from reproducing it.
Note that the initial dd is important: the file must start out sufficiently
large that the zero-point logic doesn't just clear the gaps because it
knows there's nothing in the file to read yet. Unmounting and mounting is
needed to clear the pagecache (there are other ways to do that that may
also work).
This was initially reproduced with the generic/522 xfstest on some patches
that remove the FMODE_READ restriction.
Fixes: 9ebff83e64 ("netfs: Prep to use folio->private for write grouping and streaming write")
Reported-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-12-dhowells@redhat.com
cc: Paulo Alcantara <pc@manguebit.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Fix the update of the zero point[*] by netfs_release_folio() when there is
uncommitted data in the pagecache beyond the folio being released but the
on-server EOF is in this folio (ie. i_size > remote_i_size). The update
needs to limit zero_point to remote_i_size, not i_size as i_size is a local
phenomenon reflecting updates made locally to the pagecache, not stuff
written to the server. remote_i_size tracks the server's i_size.
[*] The zero point is the file position from which we can assume that the
server will just return zeros, so we can avoid generating reads.
Note that netfs_invalidate_folio() probably doesn't need fixing as
zero_point should be updated by setattr after truncation or fallocate.
Found with:
fsx -q -N 1000000 -p 10000 -o 128000 -l 600000 \
/xfstest.test/junk --replay-ops=junk.fsxops
using the following as junk.fsxops:
truncate 0x0 0x1bbae 0x82864
write 0x3ef2e 0xf9c8 0x1bbae
write 0x67e05 0xcb5a 0x4e8f6
mapread 0x57781 0x85b6 0x7495f
copy_range 0x5d3d 0x10329 0x54fac 0x7495f
write 0x64710 0x1c2b 0x7495f
mapread 0x64000 0x1000 0x7495f
on cifs with the default cache option.
It shows read-gaps on folio 0x64 failing with a short read (ie. it hits
EOF) if the FMODE_READ check is commented out in netfs_perform_write():
if (//(file->f_mode & FMODE_READ) ||
netfs_is_cache_enabled(ctx)) {
and no fscache. This was initially found with the generic/522 xfstest.
Fixes: cce6bfa6ca ("netfs: Fix trimming of streaming-write folios in netfs_inval_folio()")
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-7-dhowells@redhat.com
cc: Paulo Alcantara <pc@manguebit.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
The list of subrequests attached to stream->subrequests is accessed without
locks by netfs_collect_read_results() and netfs_collect_write_results(),
and then they access subreq->flags without taking a barrier after getting
the subreq pointer from the list. Relatedly, the functions that build the
list don't use any sort of write barrier when constructing the list to make
sure that the NETFS_SREQ_IN_PROGRESS flag is perceived to be set first if
no lock is taken.
Fix this by:
(1) Add a new list_add_tail_release() function that uses a release barrier
to set the pointer to the new member of the list.
(2) Add a new list_first_entry_or_null_acquire() function that uses an
acquire barrier to read the pointer to the first member in a list (or
return NULL).
(3) Use list_add_tail_release() when adding a subreq to ->subrequests.
(4) Use list_first_entry_or_null_acquire() when initially accessing the
front of the list (when an item is removed, the pointer to the new
front iterm is obtained under the same lock).
Fixes: e2d46f2ec3 ("netfs: Change the read result collector to only use one work item")
Fixes: 288ace2f57 ("netfs: New writeback implementation")
Link: https://sashiko.dev/#/patchset/20260326104544.509518-1-dhowells%40redhat.com
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-4-dhowells@redhat.com
cc: Paulo Alcantara <pc@manguebit.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
When the preparation of a new subrequest for a read fails, if the
subrequest has already been added to the stream->subrequests list, it can't
simply be put and abandoned as the collector may see it. Also, if it
hasn't been queued yet, it has two outstanding refs that both need to be
put. Both DIO read and single-read dispatch fail at this; further, both
differ in the order they do things to the way buffered read works.
Fix cancellation of both DIO-read and single-read subrequests that failed
preparation by the following steps:
(1) Harmonise all three reads (buffered, dio, single) to queue the subreq
before prepping it.
(2) Make all three call netfs_queue_read() to do the queuing.
(3) Set NETFS_RREQ_ALL_QUEUED independently of the queuing as we don't
know the length of the subreq at this point.
(4) In all cases, set the error and NETFS_SREQ_FAILED flag on the subreq
and then call netfs_read_subreq_terminated() to deal with it. This
will pass responsibility off to the collector for dealing with it.
Fixes: e2d46f2ec3 ("netfs: Change the read result collector to only use one work item")
Closes: https://sashiko.dev/#/patchset/20260425125426.3855807-1-dhowells%40redhat.com
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://patch.msgid.link/20260512123404.719402-2-dhowells@redhat.com
cc: Paulo Alcantara <pc@manguebit.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
kern_select() normalises the user-supplied struct __kernel_old_timeval
with
tv.tv_sec + (tv.tv_usec / USEC_PER_SEC)
(tv.tv_usec % USEC_PER_SEC) * NSEC_PER_USEC
before calling poll_select_set_timeout() -> timespec64_valid(). Both
operands of the seconds sum are unbounded user-controlled signed long.
A crafted pair where tv_usec is a negative multiple of USEC_PER_SEC
drives the sum across the wrap boundary - e.g.
{ .tv_sec = LONG_MIN, .tv_usec = -1000000 }
yields sec = LONG_MAX, nsec = 0, which passes timespec64_valid() and
then flows through timespec64_add_safe(), which saturates the absolute
deadline to TIME64_MAX (clamped further to KTIME_MAX downstream).
select(2) therefore blocks effectively forever instead of returning
-EINVAL as POSIX requires for a negative timeout.
Only the legacy __NR_select syscall takes this path. pselect6, ppoll,
poll and epoll_pwait2 all hand the user's two fields directly to
poll_select_set_timeout(), which validates *before* doing any
arithmetic:
/* fs/select.c:271 -- the validator */
int poll_select_set_timeout(struct timespec64 *to, time64_t sec, long nsec)
{
struct timespec64 ts = {.tv_sec = sec, .tv_nsec = nsec};
if (!timespec64_valid(&ts))
return -EINVAL;
...
}
/* include/linux/time64.h:97 -- timespec64_valid */
if (ts->tv_sec < 0) return false;
if ((unsigned long)ts->tv_nsec >= NSEC_PER_SEC) return false;
/* fs/select.c:744 do_pselect() (pselect6, pselect6_time32) */
if (get_timespec64(&ts, tsp)) return -EFAULT;
if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec)) return -EINVAL;
/* fs/select.c:1097 ppoll */
if (get_timespec64(&ts, tsp)) return -EFAULT;
if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec)) return -EINVAL;
/* fs/select.c:1065 poll -- timeout_msecs is int; >= 0 gates the math */
if (timeout_msecs >= 0)
poll_select_set_timeout(to, timeout_msecs / MSEC_PER_SEC,
NSEC_PER_MSEC * (timeout_msecs % MSEC_PER_SEC));
/* fs/eventpoll.c:2512 epoll_pwait2 */
if (get_timespec64(&ts, timeout)) return -EFAULT;
if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec)) return -EINVAL;
In every one of these the wrap-prone arithmetic from kern_select()
simply does not exist; the user fields reach timespec64_valid()
unmodified. glibc routes the C-library select() through pselect6,
so the bug is reachable only via a direct syscall(__NR_select, ...).
The pre-validation negative check that used to live here was lost
when the syscall was switched to the poll_select_set_timeout() helper.
Restore it: reject tv_sec < 0 || tv_usec < 0 up front, mirroring what
glibc does in userspace. do_compat_select() has the same arithmetic
pattern but is only reachable on 32-bit compat and from a different
syscall entry; left for a follow-up so this change stays minimal.
Reproducer (returns -1/EINVAL on a fixed kernel; blocks indefinitely
on an unfixed one):
struct timeval tv = { .tv_sec = LONG_MIN, .tv_usec = -1000000 };
fd_set r;
int pfd[2];
pipe(pfd);
FD_ZERO(&r);
FD_SET(pfd[0], &r);
syscall(__NR_select, pfd[0] + 1, &r, NULL, NULL, &tv);
Fixes: 4d36a9e65d ("select: deal with math overflow from borderline valid userland data")
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260429-timeval-v1-1-4448e2588bbf@debian.org
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner <brauner@kernel.org>
check_mft_mirror() still computes the number of bytes to validate in each
mirrored MFT record, but the actual comparison against $MFTMirr was dropped
when the superblock code was updated.
As a result, mount misses a stale or inconsistent $MFTMirr as long as both
records pass the structural baad-record checks. Restore the comparison and
log an error when the primary $MFT record differs from its mirror copy.
Returning false lets the existing mount error handling mark the volume as
having NTFS errors and, with on_errors=remount-ro, continue read-only. The
default on_errors=continue mount policy still allows the mount to proceed.
Fixes: 6251f0b0de ("ntfs: update super block operations")
Signed-off-by: DaeMyung Kang <charsyam@gmail.com>
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
When executing NS_GET_PID_FROM_PIDNS (or similar pidns ioctls), if the
target task cannot be found in the corresponding pid_ns, the error code
should be ESRCH instead of ENOTTY.
This bug was introduced when the extensible ioctl handling was added.
Without proper return, ret would be overwritten by the default case in
the extensible ioctl switch statement.
Fixes: a1d220d9da ("nsfs: iterate through mount namespaces")
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Link: https://patch.msgid.link/20260507112301.1042757-1-chengzhihao1@huawei.com
Reviewed-by: Yang Erkun <yangerkun@huawei.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
