| Age | Commit message (Collapse) | Author |
|
A client can commit a hole in a LAYOUTCOMMIT, because nfsd does not
check the extents against its layouts. xfs_iomap_write_unwritten() then
allocates written blocks there. A large hole overruns the block
reservation and shuts XFS down. A small one exposes stale data.
Refuse such a range with -EIO, using XFS_BMAPI_CONVERT_ONLY.
Link: https://lore.kernel.org/r/20261014130834.GB15773@lst.de
Fixes: 527851124d10 ("xfs: implement pNFS export operations")
Cc: stable@vger.kernel.org
Signed-off-by: Daejun Park <daejun7.park@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Carlos Maiolino <cem@kernel.org>
|
|
xfs_bmapi_write() with XFS_BMAPI_CONVERT also allocates blocks for
holes and delalloc extents in the range. Add XFS_BMAPI_CONVERT_ONLY for
callers that may only convert already allocated blocks. It stops at a
hole or delalloc extent, and returns -ENOSR if it stops before mapping
anything.
Marked for stable because "xfs: refuse a pNFS LAYOUTCOMMIT of a range
without blocks" needs it.
Suggested-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20261013120359.GB2944@lst.de
Cc: stable@vger.kernel.org
Signed-off-by: Daejun Park <daejun7.park@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Pankaj Raghav <p.raghav@samsung.com>
Signed-off-by: Carlos Maiolino <cem@kernel.org>
|
|
xfs_fs_map_blocks() marks the inode XFS_DIFLAG_PREALLOC only when it
allocates blocks for a write layout. A write layout past EOF can also
hand out allocated blocks, such as speculative preallocation. blockgc
then frees them under the layout, and another file may get them.
Mark the inode whenever a write mapping ends past EOF.
Fixes: 527851124d10 ("xfs: implement pNFS export operations")
Cc: stable@vger.kernel.org
Reported-by: Chuck Lever <cel@kernel.org>
Link: https://lore.kernel.org/r/4de5356f-bc4d-4f1a-9c81-fce09b1b46fd@slotpi15m67
Signed-off-by: Daejun Park <daejun7.park@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Pankaj Raghav <p.raghav@samsung.com>
Signed-off-by: Carlos Maiolino <cem@kernel.org>
|
|
bio_may_need_split() reads bi_io_vec, which for dma-buf bios aliases
bi_dmabuf_map, and interprets the map as a bvec. Always run the full
split check for dma-buf bios.
Fixes: d88e44f43e93 ("block: introduce dma map backed bio type")
Signed-off-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://patch.msgid.link/20261029105330.1427462-1-anuj1072538@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
A flush sequence with a DATA phase can complete its original request
twice. Both completions invoke ioc_rqos_done(), causing the request to
contribute two latency samples and adding the same queue wait time
twice. This gives such requests disproportionate weight compared to
ordinary writes, distorting the device-wide latency and queue wait
feedback used to adjust iocost's virtual time rate.
Skip requests marked RQF_FLUSH_SEQ in ioc_rqos_done(). The flag is set
when the request enters the flush machinery and is still set when the
data phase completes; blk_flush_restore_request() clears it before the
final completion, so accounting happens only once, after the whole
sequence finishes. The remaining sample covers the request's lifetime
through any postflush, and its queue wait time is accumulated only
once.
Fixes: 7caa47151ab2e ("blkcg: implement blk-iocost")
Cc: stable@vger.kernel.org
Signed-off-by: Zhe Liu <liuzhe1@kylinos.cn>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://patch.msgid.link/20261102061802.408173-1-liuzhe1@kylinos.cn
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
In virtblk_add_req(), each request consumes 2 extra descriptors (out_hdr
and in_hdr) in addition to the data scatter-gather segments.
When a hypervisor (e.g. QNX Hypervisor) advertises VIRTIO_BLK_F_SEG_MAX
with seg_max = 1024 alongside a 1024-entry split ring (vring.num = 1024)
and VIRTIO_RING_F_INDIRECT_DESC disabled, lim->max_segments is set to
1024. When the block layer submits requests with 1023 or 1024 data
segments, total_sg reaches 1025 or 1026. This exceeds vring.num (1024),
causing virtqueue_add_split() to return -ENOSPC and permanently wedge
the blk-mq queue.
Furthermore, the Virtio specification (2.7.5.3.1) requires that a
descriptor chain never exceed the Queue Size, and virtqueue_add_split()
falls back to direct descriptors if indirect table allocation fails.
Fix this by:
1. Rejecting queues with ring_size < 3 at probe, or
ring_size < (queue_max_segments + 2) during resume/reset recovery in
init_vq().
2. Unconditionally clamping sg_elems to (ring_size - 2) across all
virtqueues in virtblk_read_limits().
Fixes: 0864b79a1533 ("virtio: block: dynamic maximum segments")
Signed-off-by: Sergii Ushakov <sergiiushakov@google.com>
Link: https://patch.msgid.link/20260918135120.1261150-1-sergiiushakov@google.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
null_set_zone_cond() takes a zone condition, even though read-only and
offline are the only zone conditions that it can set. Because of this,
it needs a WARN_ON_ONCE() to reject any other zone condition, and its
name suggests a generic helper that can set any zone condition.
Rename null_set_zone_cond() to null_set_zone_readonly_or_offline(), and
replace its zone condition argument with a bool, set_offline, which
selects between the read-only and offline conditions. Also remove the
comment above the function, as it only repeats the new function name.
zone_cond_store() keeps taking a zone condition, as it, unlike
null_set_zone_readonly_or_offline(), contains generic code, and might
be extended to support additional zone conditions in the future.
No functional change, as zone_cond_store() is only called with the
read-only or offline condition.
Suggested-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Niklas Cassel <cassel@kernel.org>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://patch.msgid.link/20261022092158.2248831-8-cassel@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
null_set_zone_cond() calls null_finish_zone() to release the zone
resources held by a zone before setting it read-only or offline.
For an empty or closed zone, null_finish_zone() first runs the zone
resource checks, which can close an unrelated implicitly open zone, or
fail. As the error is ignored, a failure for a closed zone leaks
nr_zones_closed, permanently lowering the effective active zone limit.
Setting a zone read-only or offline requires no zone resources, so
release the zone resources held by the zone directly instead, using a
new helper, null_release_zone_resources().
Fixes: d3a5738849e0 ("null_blk: support read-only and offline zone conditions")
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Niklas Cassel <cassel@kernel.org>
Link: https://patch.msgid.link/20261022092158.2248831-7-cassel@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
When the open zone limit is reached, null_check_open() closes an
implicitly open zone only if the active zone limit is not reached.
Closing an implicitly open zone does not change the number of active
zones, so this check is wrong. For an empty zone, the active zone limit
was already checked by null_check_zone_resources(). For a closed zone,
the check makes writing to, opening or finishing the zone fail with
BLK_STS_ZONE_OPEN_RESOURCE at the active zone limit, whereas ZBC-3
(4.5.3.2.7.2) and NVMe ZNS (2.1.1.4.1) require closing an implicitly
open zone instead.
Remove the check, and update the comment above
null_check_zone_resources() accordingly.
Fixes: dc4d137ee3b7 ("null_blk: add support for max open/active zone limit for zoned devices")
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Niklas Cassel <cassel@kernel.org>
Link: https://patch.msgid.link/20261022092158.2248831-6-cassel@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
A KASAN slab-out-of-bounds read can occur in blk_mq_free_rqs() when there
is a mismatch between the number of hardware queues allocated for the IO
scheduler (et->nr_hw_queues) and the number of hardware queues in the block
tag set (set->nr_hw_queues).
This mismatch can happen if blk_mq_update_nr_hw_queues() fails halfway
through (e.g., due to memory pressure during
blk_mq_prealloc_tag_set_tags()). In this error path, the elevator is
restored with a larger number of hardware queues than the block tag set
actually has.
When the elevator is later freed, blk_mq_free_sched_tags() iterates up to
the new et->nr_hw_queues and calls blk_mq_free_rqs(). In blk_mq_free_rqs(),
it attempts to access set->tags[hctx_idx] to get the driver tags. Because
set->nr_hw_queues was not updated due to the earlier failure, set->tags
still has the old (smaller) size, leading to an out-of-bounds read.
BUG: KASAN: slab-out-of-bounds in blk_mq_free_rqs+0xde/0x680
Read of size 8 at addr ffff88818d67fc28 by task syz-executor117/5839
Call Trace:
blk_mq_free_rqs+0xde/0x680
blk_mq_free_map_and_rqs+0x40/0xf0
blk_mq_free_sched_tags
blk_mq_free_sched_res+0xeb/0x280
elevator_change_done+0x1d2/0x5c0
elevator_change+0x34f/0x480
elv_iosched_store+0x504/0x630
queue_attr_store+0x207/0x2b0
To fix this, explicitly check if hctx_idx < set->nr_hw_queues before
accessing set->tags[hctx_idx]. If hctx_idx >= set->nr_hw_queues, the
hardware queue doesn't exist in the tag set, meaning there are no driver
tags to clear mappings from. In this case, safely set drv_tags to NULL. The
subsequent call to blk_mq_clear_rq_mapping() already handles a NULL
drv_tags pointer and will safely return.
Fixes: 04225d13aef1 ("block: fix potential deadlock while running nr_hw_queue update")
Assisted-by: Gemini:gemini-3.5-flash Gemini:gemini-3.1-pro-preview syzbot
Reported-by: syzbot+e90526cab23b9efcd03c@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=e90526cab23b9efcd03c
Link: https://syzkaller.appspot.com/ai_job?id=827b5760-86a0-4f44-9c69-430b572eac12
Reviewed-by: Nilay Shorff <nilay@linux.ibm.com>
Cc: stable@vger.kernel.org # 6.17+
Signed-off-by: Krystian Kaniewski <krystianmkaniewski@gmail.com>
Tested-by: syzbot@syzkaller.appspotmail.com
Link: https://patch.msgid.link/20261014170538.1089448-1-krystianmkaniewski@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The timeval validation checks tv_usec < USEC_PER_SEC but not
tv_usec >= 0, so a negative tv_usec makes the timeout computation
wrap into a bogus value, later tripping "schedule_timeout: wrong
timeout value" instead of being rejected.
Fixes: d021c344051a ("VSOCK: Introduce VM Sockets")
Signed-off-by: Nguyen Dinh Phi <phind.uet@gmail.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Reviewed-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20261030064428.73930-1-phind.uet@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
init_vq() reads the writable num_request_queues and poll_queues module
parameters both at probe and when recreating virtqueues after suspend
or reset. However, virtio-blk does not update the blk-mq hardware queue
count or queue maps on restore.
Increasing poll_queues can therefore remove the completion callback
from a virtqueue still used for interrupt-driven I/O, leaving requests
uncompleted. Reducing the effective request queue count through
num_request_queues can instead leave existing hardware contexts
indexing beyond the reallocated vblk->vqs array.
Determine the virtqueue count and the split between interrupt-driven
and polling queues once at probe, and reuse this layout when
recreating the virtqueues. Changes to these module parameters then
affect only subsequently probed devices.
Fixes: 0989c41bed96 ("virtio-blk: add num_request_queues module parameter")
Signed-off-by: Yunye Zhao <yunye.zhao@linux.alibaba.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Link: https://patch.msgid.link/20261102133740.88919-1-yunye.zhao@linux.alibaba.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The block size obviously needs to be queried when it is not set, not when
it already is set.
Reported: Damien Le Moal <dlemoal@kernel.org>
Fixes: 5e9929e2f03c ("zloop: share more code for opening a zone backing file")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://patch.msgid.link/20261102055558.3212814-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
When restoring a zloop device, zloop_init_zone() checks that the size of
each conventional zone file matches the configured zone size. The check
prints an error and then jumps to out_fput, which releases the zone file
and returns the value left in "ret" by the preceding vfs_getattr() call,
that is, zero. zloop_ctl_add() therefore treats the zone as initialized
and continues with a zone->file pointer that was already released; any
I/O to that zone then dereferences it.
Return -EINVAL, like the sequential zone path does for the same kind of
inconsistency.
While at it, print the zone size that is actually compared instead of the
zone capacity, which may be smaller and only made the message confusing.
Fixes: eb0570c7df23 ("block: new zoned loop block device driver")
Signed-off-by: Shao Mingyin <shao.mingyin@zte.com.cn>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://patch.msgid.link/20261031221610676B5VoVcwMgkoDNVZH2bVDO@zte.com.cn
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Since commit 4ff58d6bc9dd ("block: serialize elevator changes for the
same queue using a writer lock"), a write to queue/scheduler fails with
-EBUSY whenever something else holds set->update_nr_hwq_lock. That
commit takes the lock for write, and it has to be a trylock because of
commit ce8ee8583ed8 ("block: use trylock to avoid lockdep circular
dependency in sysfs").
The lock covers the whole tag set, so it is easy to hit. A scheduler
write fails while another queue of the set is switching, or while a
sibling disk of the same SCSI host or NVMe controller is added or
removed. This matches the udev failures reported on 7.2.
Go back to taking the lock for read. The write lock was there to keep
two switches of one queue from overlapping in elevator_change_done(),
which touches the elevator sysfs and debugfs entries outside of
q->elevator_lock. Use QUEUE_FLAG_NO_ELV_SWITCH for that instead: a
write sets it before elevator_change() and clears it afterwards, and a
write that finds it set returns -EBUSY. Disk removal already sets the
flag under the write lock, so no switch can start once it has run.
Writes to the same queue, and writes that race with an nr_hw_queues
update, an nr_requests write or disk removal, can still get -EBUSY, as
with 4ff58d6bc9dd. A write to a disk that is being removed now gets
-EBUSY instead of -ENOENT.
Fixes: 4ff58d6bc9dd ("block: serialize elevator changes for the same queue using a writer lock")
Cc: stable@vger.kernel.org # v7.2+
Reported-by: Reindl Harald <h.reindl@thelounge.net>
Closes: https://lore.kernel.org/r/077dd42e-cbc9-4c23-a5ef-3d20e4fc7757@thelounge.net
Suggested-by: Nilay Shroff <nilay@linux.ibm.com>
Assisted-by: LLM
Signed-off-by: Laveesh Bansal <laveeshb@laveeshbansal.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Link: https://patch.msgid.link/20261031044141.5569-1-laveeshb@laveeshbansal.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
When blk_mq_alloc_disk() fails, "ret" still holds the zero returned by the
preceding blk_mq_alloc_tag_set() call, so the error message always reports
"err=0". Save PTR_ERR(zlo->disk) in "ret" before printing it, like the
blk_mq_alloc_tag_set() and add_disk() failure paths do.
Fixes: eb0570c7df23 ("block: new zoned loop block device driver")
Signed-off-by: Shao Mingyin <shao.mingyin@zte.com.cn>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://patch.msgid.link/20261031221737569WQ3bg8y80Ht7JdTCWwlkL@zte.com.cn
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
disk_discard_alignment_show() reports the physical I/O alignment offset
instead of the discard alignment. Devices can specify these offsets
independently, so the discard_alignment sysfs attribute can report the
wrong value.
Use bdev_discard_alignment() for the whole-disk attribute, as is already
done for partitions.
Fixes: 4e1462ffe899 ("block: remove queue_discard_alignment")
Signed-off-by: Avi Weiss <thnkslprpt@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://patch.msgid.link/20261031102023.3644494-2-thnkslprpt@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
All dma-buf (P2P) I/O to a device with integrity data is rejected
today, at registration and in blk_mq_submit_bio(). Only a bio for which
the block layer generates or verifies PI needs that. The block layer
then walks the bio_vecs, and a dma-buf bio has none.
Reject only those bios, and remove the check at registration. Then a
submitter that attaches its own PI, for example with
IORING_RW_ATTR_FLAG_PI, can do dma-buf I/O to such a device.
Link: https://lore.kernel.org/all/bbb87d6b-b7a3-487a-8be4-27f3f9ce4a1b@samsung.com/
Assisted-by: LLM
Reviewed-by: Anuj Gupta <anuj20.g@samsung.com>
Signed-off-by: Kyle Bader <kbader@ibm.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://patch.msgid.link/20261029155821.2983229-1-kbader@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
A GPT entry whose starting_lba is greater than its ending_lba is accepted,
and the kernel creates a partition with a bogus size. The bogus sector
count defeats the end-of-device check, so I/O through that partition is not
confined to it and can access sectors outside the partition on the same
disk.
is_pte_valid() only rejects an entry when its type is NULL or when
starting_lba or ending_lba is greater than lastlba; it never checks that
starting_lba <= ending_lba. For a reversed entry both field checks still
pass, but efi_partition() computes size = ending_lba - starting_lba + 1 in
u64, which underflows to a huge value. That value also wraps in the
end-of-device check in blk_add_partition() (from + size) and does not fit
the signed i_size, which turns negative.
Reject entries with starting_lba > ending_lba in is_pte_valid(), matching
the head-level check against last_usable_lba < first_usable_lba.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Zizhi Wo <wozizhi@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://patch.msgid.link/20261102080442.3775061-1-wozizhi@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
For zone append, the block layer copies req->__sector back into the
bio in blk_zone_append_update_request_bio(), so the driver must update
req->__sector with the sector reported by the device.
virtblk_request_done() does that, but the batched completion path used
by virtblk_poll() only unmaps and cleans up the requests before calling
blk_mq_end_request_batch().
A zone append completed through polling therefore keeps the sector that
was assigned at submission time, and the file system or application sees
the wrong LBA for the written data.
Move the sector update into a helper and call it from both completion
paths.
Fixes: 95bfec41bd3d ("virtio-blk: add support for zoned block devices")
Signed-off-by: Shao Mingyin <shao.mingyin@zte.com.cn>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://patch.msgid.link/202610261955196384aSLGuy3_0MiTvlYXVdiC@zte.com.cn
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
We observed I/O timeouts with requests stalled in per-CPU software
queues.
The affected queue had no elevator, but QUEUE_FLAG_SQ_SCHED remained
set. A failed elevator switch can leave this state when falling back to
none without clearing the flag. With empty hctx dispatch lists, a
queue-wide restart on CPU4 would skip requests pending in ctx5/ctx7:
CPU4: blk_mq_run_hw_queues(q)
sq_hctx = hctx4
hctx4: no pending requests -> selected
hctx5: 9 reads in ctx5 -> skipped
hctx7: 3 writes in ctx7 -> skipped
The pending requests may remain undispatched until another event runs
their hardware queues.
Clear the flag and restore nr_requests and async_depth to the driver
queue depth in the common scheduler teardown path, covering both
scheduler exit and failed initialization.
Fixes: 4d337cebcb1c ("blk-mq: avoid to touch q->elevator without any protection")
Assisted-by: LLM
Signed-off-by: Jianan Huang <jnhuang95@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://patch.msgid.link/20261023092611.116795-1-jnhuang95@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
bs_blocks and bs_xflags were both read from bstat32->bs_size.
Read each one from its own field instead.
Fixes: e94fc4a43e5c39 ("[XFS] Add compat handlers for swapext ioctl")
Cc: stable@vger.kernel.org # v2.6.29
Signed-off-by: Reid Downing <skibletz@gmail.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Signed-off-by: Carlos Maiolino <cem@kernel.org>
|
|
BLKBSZGET returns the block device's soft block size, while BLKSSZGET
returns its logical sector size. The compat BLKBSZGET_32 handler uses
bdev_logical_block_size(), so it returns the wrong value when these
sizes differ.
For example, on an x86_64 kernel a native query of a RAM or loop device
with 512-byte logical sectors returns the initial soft block size of
4096 bytes, but an i386 process gets 512 bytes. This makes 32-bit
blockdev --getbsz and the BSZ column of --report disagree with native
queries, even without a preceding BLKBSZSET.
Use block_size() as the native handler and the pre-regression compat
handler do, and correct the case comment to describe the soft size.
Fixes: 9b81648cb5e3 ("compat_ioctl: simplify up block/ioctl.c")
Cc: stable@vger.kernel.org
Signed-off-by: Hungyu Lin <dennylin0707@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20261023153645.37546-1-dennylin0707@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
blk_mq_run_work_fn() can call blk_mq_sched_dispatch_requests() through
blk_mq_run_dispatch_ops(). The dispatcher checks QUEUE_FLAG_QUIESCED
while the flag is clear. It can then be preempted before it gets a
request. At that time, it has no queue usage reference, so
blk_mq_freeze_queue() does not wait for it.
virtblk_freeze_priv() then calls blk_mq_quiesce_queue_nowait() and
blk_mq_unfreeze_queue(). A new request can enter the queue. The old
dispatcher can run again, use its old flag check, get the new request,
and call virtio_queue_rq(). At the same time, virtblk_freeze_priv() can
reset the device, delete the virtqueues, and free vblk->vqs. The old
dispatcher can then use a deleted virtqueue or a NULL vblk->vqs, and the
kernel can crash.
Fix this by using blk_mq_quiesce_queue(), which waits for dispatch code
that was already running. Keep the existing freeze and unfreeze order.
The queue remains quiesced until virtblk_restore_priv() calls
blk_mq_unquiesce_queue().
Fixes: 7678abee0867 ("virtio-blk: don't keep queue frozen during system suspend")
Cc: stable@vger.kernel.org
Acked-by: Jason Wang <jasowangio@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Cong Zhang <cong.zhang@oss.qualcomm.com>
Link: https://lore.kernel.org/20260911-virtblk_sync_quiesce-v1-1-a883f8f31258@oss.qualcomm.com
Link: https://patch.msgid.link/20261103-virtblk_sync_quiesce-v3-1-f40233f9bdd9@oss.qualcomm.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The CACHE flag belongs to nullb_device, which outlives each powered-on
nullb instance. Power-off discards the runtime cache but leaves the flag
set. If cache_size changes from 1 to 0 before the next power-on, the
queue advertises write-through while new page writes still go into the
cache tree. A completed write can then disappear at the next power-off.
Clear CACHE before creating the new instance. The existing cache_size
setup sets it again when the new instance needs a cache.
Signed-off-by: Baek Kyu, Kim <pocj8ur4in@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://patch.msgid.link/20261025161133.59344-1-pocj8ur4in@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
When blk_crypto_rq_get_keyslot() fails, blk_mq_submit_bio() ends the bio
by hand and frees the request. By then blk_mq_bio_to_request() has done
the IO accounting, which blk_mq_free_request() doesn't undo, so the
in_flight counter never drops and iostat shows a device that is 100%
busy. If the bio went through zone write plugging the zone stays plugged
as well, and every later write to it waits forever.
Use blk_mq_end_request() like the dispatch path does for a request the
driver rejected, after marking it for zone write plugging.
Move the call to blk_zone_write_plug_init_request() before
blk_crypto_rq_get_keyslot() so that it doesn't need to be duplicated.
It doesn't depend on the keyslot.
Fixes: 99dc422335d8 ("block: support to account io_ticks precisely")
Fixes: dd291d77cc90 ("block: Introduce zone write plugging")
Cc: stable@vger.kernel.org
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Eric Biggers <ebiggers@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://patch.msgid.link/20261029200110.1328103-1-ebiggers@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
io_sq_thread_stop() carries a WARN_ON_ONCE() asserting that
IO_SQ_THREAD_SHOULD_STOP isn't set yet, i.e. that stopping the SQPOLL
thread is never attempted twice. Since commit 22d9db9fd64a
("io_uring/sqpoll: don't attach to a thread that is canceling"), the
SQPOLL thread itself sets that bit on its way out, so that
io_sq_offload_create() won't attach a new ring to a thread that is
canceling.
When the thread exits on a fatal signal, the bit is thus set before the
ring teardown gets to io_sq_thread_stop(), which now triggers the
warning on every such exit:
WARNING: io_uring/sqpoll.c:88 at io_sq_thread_stop
Workqueue: iou_exit io_ring_exit_work
Call Trace:
io_put_sq_data
io_sq_thread_finish
io_ring_ctx_free
io_ring_exit_work
Introduce a separate IO_SQ_THREAD_EXITING bit for the thread marking
itself as on its way out, and check it in io_sq_offload_create() along
with IO_SQ_THREAD_SHOULD_STOP. IO_SQ_THREAD_SHOULD_STOP regains its
original meaning, "a stop has been requested", and the WARN_ON_ONCE()
holds again.
Reproduced with an SQPOLL ring and an immediate fatal signal: the
warning fires on v7.3.0+ (22d9db9fd64a present) and is gone with this
change.
Reported-by: syzbot+675af112519ca34f5502@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=675af112519ca34f5502
Fixes: 22d9db9fd64a ("io_uring/sqpoll: don't attach to a thread that is canceling")
Signed-off-by: Shuai Liu <847807867@qq.com>
Link: https://patch.msgid.link/tencent_BB08960D79E0261B5C1EB1E589787734E907@qq.com
[axboe: trim excessive comment]
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
unregister_console_locked() is called from within
hlist_for_each_entry_safe(console_list) loops. This would be safe if
unregister_console_locked() only unregistered the specified console. After
all, that is what the _safe variant is for.
However, unregister_console_locked() calls printk_kthreads_check_locked(),
which can unregister _multiple_ consoles if kthreads fail to start. This
could lead to hlist_for_each_entry_safe() use-after-free scenarios because
not only was the list entry removed, but possibly the following entry as
well (the @n temporary storage of the next entry).
Move the call to printk_kthreads_check_locked() outside of
unregister_console_locked(). Add comments to both functions to clarify
if they are safe to be used in hlist_for_each_entry_safe(console_list)
loops.
Reported-by: Sashiko <sashiko-bot@kernel.org>
Closes: https://lore.kernel.org/lkml/87ik31iz4x.fsf@jogness.linutronix.de
Fixes: 76f258bf3f2a ("printk: nbcon: Introduce printer kthreads")
Signed-off-by: John Ogness <john.ogness@linutronix.de>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Tested-by: Derek Barbosa <debarbos@redhat.com>
Link: https://patch.msgid.link/20261030135802.663893-1-john.ogness@linutronix.de
Signed-off-by: Petr Mladek <pmladek@suse.com>
|
|
Firmware reports udma_shift as the log2 of the number of queues in
a queue group. A value of zero means the queues are not striped
across two UDMAs, so the QID space is linear and the device has one UDMA.
The interleaved bitid/qid transform treats bit 0 as the UDMA selector
when the group shift is zero. QID 1 then becomes bitid
1 << half_qid_shift, and freeing that QID releases the wrong allocator
index. Skip the transform and select one UDMA so QP and CQ allocation
uses the QID directly.
Fixes: e8521822c733 ("RDMA/ionic: Register device ops for control path")
Reported-by: Keith Lucas <keith.lucas@oracle.com>
Link: https://lore.kernel.org/r/20261021161651.4020222-1-haakon.bugge@oracle.com
Cc: stable@vger.kernel.org # 6.18
Signed-off-by: Abhijit Gangurde <abhijit.gangurde@amd.com>
Link: https://patch.msgid.link/20261030082402.841751-1-abhijit.gangurde@amd.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
|
|
Previously, IB_MR_REREG_PD would cause the driver to set
iwmr->ibmr.pd to the new PD before reregistering with HW because
irdma_hwreg_mr() retrieves the PD from there.
ib_uverbs_rereg_mr() only moves the PD usecnt from the old PD to
the new one and updates mr->pd itself after the driver's
rereg_user_mr() call succeeds.
So, if the rereg_user_mr() call fails, the MR is left pointing at
a PD that does not account for it while the old PD keeps the reference,
which means the old PD can never be deallocated, while the new PD can
be deallocated even though the MR still points at it.
Fix this by removing the PD update from the driver and just letting
the core handle it. This required plumbing the PD as an argument
through a few functions. This works because the core passes the
PD unconditionally and provides the current PD if IB_MR_REREG_PD is
not set.
Fixes: 5ac388db27c4 ("RDMA/irdma: Add support to re-register a memory region")
Signed-off-by: Jacob Moroni <jmoroni@google.com>
Link: https://patch.msgid.link/20261029135306.1372393-1-jmoroni@google.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
|
|
An MR can become "defunct" if a previous rereg_mr happened
to fail in between deregistering the old MR in HW and
reregistering it (e.g., due to ib_umem_get_va() failing
in the IB_MR_REREG_TRANS case).
In this case, the MR is left in a state where the umem may
have already been released.
If rereg_mr then runs again, but without IB_MR_REREG_TRANS,
it will end up reregistering the MR with stale translation
info.
Fix this by rejecting any attempt to reregister an MR that
has previously failed during rereg_mr after the point where
the MR has been deregistered in HW.
Fixes: 5ac388db27c4 ("RDMA/irdma: Add support to re-register a memory region")
Signed-off-by: Jacob Moroni <jmoroni@google.com>
Link: https://patch.msgid.link/20261029133121.1204636-1-jmoroni@google.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
|
|
During rereg_mr, the umem is released in various failure cases
but mmr->umem was left pointing to a freed umem, so it could end
up getting double-freed when the MR is finally deregistered, or
cause a UAF in ib_umem_check_rereg() if another rereg_mr attempt
is made.
Fix this by setting mmr->umem to NULL after ib_umem_release().
Additionally, when mlx4_mr_rereg_mem_write() fails, mtt.order is
left set but with an invalid offset, so during dereg,
mlx4_mtt_cleanup() ends up freeing an MTT range that was never
allocated. Fix this by resetting mtt.order to -1 in that path.
Fixes: 9376932d0c26 ("IB/mlx4_ib: Add support for user MR re-registration")
Signed-off-by: Jacob Moroni <jmoroni@google.com>
Link: https://patch.msgid.link/20261029132740.1169950-1-jmoroni@google.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
|
|
inotify compares the mask, watch descriptor and name when coalescing
adjacent unread events, but ignores their rename cookies. Distinct
IN_MOVED_FROM events can therefore be coalesced, losing the cookie needed
to match a later IN_MOVED_TO event.
For example, watch two directories with IN_MOVE. Move an entry from the
first directory into an unwatched directory, recreate it under the same
name, then move it from the first directory into the second watched
directory before reading the events. The second IN_MOVED_FROM is merged
with the first, while the IN_MOVED_TO retains the second rename's cookie.
Userspace can no longer pair the second rename's events.
The original inotify implementation compared cookies before coalescing.
The fsnotify conversion lost this comparison. Do not coalesce
IN_MOVED_FROM or IN_MOVED_TO events, which carry the identity of their
respective renames. Leave coalescing of other event types unchanged.
Fixes: 63c882a05416 ("inotify: reimplement inotify using fsnotify")
Cc: stable@vger.kernel.org
Suggested-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Hungyu Lin <dennylin0707@gmail.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Link: https://patch.msgid.link/20261027192027.8617-1-dennylin0707@gmail.com
Signed-off-by: Jan Kara <jack@suse.cz>
|
|
Every change of an ext2 directory is written to disk before the system call
returns, whether the directory is DIRSYNC or not: its pages with
filemap_write_and_wait() and its inode with sync_inode_metadata().
Writes which reach the device for 2000 files, and 200 directories, in a
directory without DIRSYNC, with the flusher turned off:
create rename unlink mkdir
before 4000 8000 4000 800
after 0 0 0 0
In a directory with chattr +D the numbers are those of "before", before and
after.
commit bc943f4872a7 ("ext2: Don't flush page immediately for DIRSYNC
directories") moved the flush from ext2_commit_chunk() to the new
ext2_handle_dirsync(). The IS_DIRSYNC() check which was around it didn't
move along.
Fix by putting the check into ext2_handle_dirsync().
Fixes: bc943f4872a7 ("ext2: Don't flush page immediately for DIRSYNC directories")
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Link: https://patch.msgid.link/20261027-b4-ext2-dirsync-v1-1-3dcd012d518d@toxicpanda.com
Signed-off-by: Jan Kara <jack@suse.cz>
|
|
Lockdep reports the following circular locking dependency when a virtio-net
device is probed with lockdep enabled on a system whose devices sit behind
an IOMMU:
======================================================
WARNING: possible circular locking dependency detected
7.3.0-rc1+ #84 Not tainted
------------------------------------------------------
kworker/u34:0/57 is trying to acquire lock:
ffff800104f83ee8 (&md->mutex){+.+.}-{4:4}, at: msi_domain_get_virq+0xf0/0x118
but task is already holding lock:
ffffc00081edd930 (cpu_hotplug_lock){++++}-{0:0}, at: cpus_read_lock+0x10/0x1c
which lock already depends on the new lock.
Chain exists of:
&md->mutex --> &group->mutex --> cpu_hotplug_lock
The cycle is composed of three dependencies:
* cpu_hotplug_lock -> &md->mutex (which triggers the splat). virtio_net's
init_vqs() calls cpus_read_lock() around virtnet_set_affinity(), which
reaches vp_set_vq_affinity(). That calls pci_irq_vector(), which takes
the per-device MSI mutex (&dev->msi.data->mutex) via
msi_domain_get_virq().
* &md->mutex -> &domain->mutex -> &group->mutex. MSI-X allocation of any
device behind the IOMMU holds the MSI mutex while allocating interrupts
through the IRQ domain hierarchy. The MSI parent domain (GICv2m, or ITS
likewise) calls iommu_dma_prepare_msi() from its .alloc() to map the MSI
doorbell, which takes the IOMMU group mutex.
* &group->mutex -> cpu_hotplug_lock. bus_iommu_probe() calls
iommu_setup_dma_ops() under the group mutex; initializing the IOVA
rcaches registers a cpuhp instance, which takes cpu_hotplug_lock.
The cycle only became complete since commit 288683c92b1a ("iommu: Make
iommu_dma_prepare_msi() into a generic operation"), where
iommu_dma_prepare_msi() switched to use the group mutex.
pci_irq_vector() is a pure lookup of a vector that is bound to the
queue for the whole lifetime of the queue, so there is no reason to
take the MSI mutex for it from cpu hotplug locked contexts.
Cache the IRQ number in virtio_pci_vq_info {} at vector binding time and
make vp_set_vq_affinity() use the cached value.
This also covers the CPU hotplug callbacks
(virtnet_cpu_online/virtnet_cpu_dead), which call virtnet_set_affinity()
with cpu_hotplug_lock already held.
Signed-off-by: Zenghui Yu (Huawei) <zenghui.yu@linux.dev>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261029121648.22155-1-zenghui.yu@linux.dev>
|
|
The virtqueue_add_indirect_packed() function fills an indirect
descriptor table with addr, len, and flags for each entry, but does
not initialize the id field of struct vring_packed_desc. The table is
allocated via kmalloc() (not kzalloc) and then DMA-mapped with
DMA_TO_DEVICE, making the uninitialized 2-byte id field of each
descriptor readable by the device.
This leaks kernel heap memory to the device for every packed indirect
descriptor entry. A malicious or compromised device could extract heap
contents by reading the id fields from the DMA-mapped indirect
descriptor tables. This affects all virtio devices that use packed
virtqueues with indirect descriptors, which is the common configuration
for modern virtio devices.
Fix this by explicitly initializing desc[i].id to zero in the
descriptor filling loop. The id field is unused in indirect descriptor
table entries per the virtio specification (only the head descriptor in
the main ring uses id for buffer identification), so setting it to zero
is semantically correct.
This matches the split ring path where all four fields of struct
vring_desc (addr, len, flags, next) are explicitly written in
virtqueue_add_desc_split().
Signed-off-by: zhang jiao <zhangjiao2@cmss.chinamobile.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261029072511.851-1-zhangjiao2@cmss.chinamobile.com>
|
|
iov_iter_get_pages2() returns a byte count that excludes the offset into
the first page, but also returns a reference to that page in pages[].
On the misaligned-I/O fallback, vhost_scsi_revert_map_iov_to_sgl() puts
pages already added to the SG list. The cleanup loop starts at
pages[npages], but accounts for its first page as PAGE_SIZE bytes
rather than PAGE_SIZE - offset. If the remaining data starts at a
non-zero offset, it can finish one page early and leave a page reference
behind.
Use PAGE_SIZE - offset for the first remaining page and clear offset for
later pages, as in the mapping loop. Without this, each affected request
leaks one page reference.
Fixes: bca939d5bcd0 ("vhost-scsi: Dynamically allocate scatterlists")
Reviewed-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Jia Jia <physicalmtea@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261023104035.6686-1-physicalmtea@gmail.com>
|
|
vduse_domain_get_coherent_page() assumes that iotlb entries above
bounce_size were installed by vduse_domain_alloc_coherent(), so it
treats map->addr as a physical address. But vduse registers with
use_va = true, and when the device is bound to vhost_vdpa,
vhost_vdpa_va_map() stores a userspace virtual address in map->addr,
which vduse_domain_set_map() then copies into the same iotlb. A
process holding the domain file descriptor across a rebind can
therefore fault in an arbitrary physical page with read/write
access, ignoring the entry permissions.
Additionally, since the domain file mmap lacks VM_IO, the domain
file itself can be mapped and passed to vhost_vdpa_va_map(),
producing self-referential entries that bypass the coherent path
check.
To fix, only translate entries whose backing file is the domain
file itself in vduse_domain_get_coherent_page(), reject such
entries in vduse_domain_set_map() since only external mappings
should be imported, and set VM_IO on the domain file mmap so
vhost_vdpa_va_map() refuses to map it in the first place.
Fixes: 8c773d53fb7b ("vduse: Implement an MMU-based software IOTLB")
Cc: Xie Yongji <xieyongji@bytedance.com>
Cc: stable@vger.kernel.org
Reported-by: chenglong.ma <longpareto@gmail.com>
Signed-off-by: chenglong.ma <longpareto@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Co-Authored-By: Claude Code <noreply@anthropic.com>
|
|
remove_common() frees the driver data during freeze, but does not clear
vdev->priv, which points to freed memory. If probe_common() then fails
during restore, a later remove or freeze can dereference the dangling
pointer.
Clear vdev->priv on the probe_common() error cleanup path and before
freeing the driver data in remove_common(). Also reset the device and
return early from remove_common() if vdev->priv is already NULL for a
safe teardown after a failed restore.
Fixes: 08e53fbdb85c ("virtio-rng: support multiple virtio-rng devices")
Cc: stable@vger.kernel.org
Signed-off-by: Thorsten Blum <blum@kernel.org>
Reviewed-by: PrasannaKumar Muralidharan <prasannatsmkumar@gmail.com>.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261016103640.4989-2-blum@kernel.org>
|
|
Use virtqueue consistently in the notify function comments
across the MMIO and PCI transports.
Assisted-by: LLM
Signed-off-by: Wei-Shuo Chen <tavhadia@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261016072318.24015-1-tavhadia@gmail.com>
|
|
Since c578123e63da ("virtio_balloon: Treat stats requests as wakeup
events") the stats_request() virtqueue callback takes vb->wakeup_lock
via start_wakeup_event(). init_vqs() registers that callback and even
primes and kicks the stats queue, so the lock is reachable from irq
context as soon as init_vqs() returns, long before the driver sets
DRIVER_OK: vring_interrupt() dispatches to the callback without any
DRIVER_OK check unless CONFIG_VIRTIO_HARDEN_NOTIFICATION is set, and
that option depends on BROKEN. syzbot hit it from a threaded irq:
INFO: trying to register non-static key in stats_request
...
assign_lock_key
register_lock_class
lock_acquire
rt_spin_lock
start_wakeup_event drivers/virtio/virtio_balloon.c:150 [inline]
stats_request drivers/virtio/virtio_balloon.c:435
The lock is only initialized near the end of virtballoon_probe(),
after init_vqs() and several steps that can still fail and unwind
through del_vqs(), where a racing interrupt still finds the callback
registered. Until then it is whatever kzalloc() left in vb: lockdep
complains about the unregistered lock class, and wakeup_signal_mask
and processing_wakeup_event get updated from irq context while probe
is still running.
The rt_spin_lock() in the report is just spin_lock() on PREEMPT_RT;
any PROVE_LOCKING build takes the same path.
stop_update_lock(), which the same callback also acquires, is already
initialized before init_vqs(); do the same for wakeup_lock.
Fixes: c578123e63da ("virtio_balloon: Treat stats requests as wakeup events")
Reported-by: syzbot+13df72e0dc4be5c62215@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=13df72e0dc4be5c62215
Signed-off-by: Mingbing Zhai <zhaimingbing@cmss.chinamobile.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261016025429.901-1-zhaimingbing@cmss.chinamobile.com>
|
|
Several headers related to virtio, vDPA, VDUSE and vhost were not
covered by any specific MAINTAINERS entry and fell through to the
catch-all THE REST section:
- include/linux/device-id/virtio.h: not matched by the existing
"include/linux/virtio*.h" glob, which only matches files directly
under include/linux/ and does not recurse into subdirectories.
- include/uapi/linux/vdpa.h: the in-kernel include/linux/vdpa.h was
listed under VIRTIO CORE but its UAPI companion was missing; the
"include/uapi/linux/virtio_*.h" pattern does not match it because
the filename lacks the "virtio_" prefix.
- include/uapi/linux/vduse.h: there is no dedicated VDUSE section in
MAINTAINERS; the driver lives under drivers/vdpa/ which is already
covered by VIRTIO CORE, so the UAPI header belongs there too.
- include/uapi/linux/vhost_types.h: the VIRTIO HOST (VHOST) section
explicitly lists include/uapi/linux/vhost.h but omits its companion
types header.
Assisted-by: claude-sonnet 4.6
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261015094049.107999-1-eperezma@redhat.com>
|
|
Initialize the completion and as hash buckets before cdev_device_add()
makes the device accessible.
Fixes: 4c8cf31885f6 ("vhost: introduce vDPA-based backend")
Assisted-by: OpenAI-Codex:GPT-5.6
Signed-off-by: Linfeng Sun <linfeng.sun.dev@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261011-fix-null-pointer-in-vhost_vdpa_release-v2-1-66ba929971ce@gmail.com>
|
|
Enabling a backend without VHOST_SET_VRING_KICK dereferences a NULL
vq->kick in vhost_poll_start(). Reject the request with -EINVAL.
Fixes: 264ee5aa81ec ("tools/virtio: Add --reset")
Assisted-by: OpenAI-Codex:GPT-5.6
Signed-off-by: Linfeng Sun <linfeng.sun.dev@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261011-fix-null-pointer-in-vhost_test_set_backend-v1-1-60b2f12d64c6@gmail.com>
|
|
vhost_scsi_set_endpoint() allocates command pools only for I/O
virtqueues that are already set up, but publishes the backend for every
virtqueue. Userspace can complete a skipped queue after attaching the
endpoint, then kick it with scsi_tags uninitialized and trigger the
alloc_hint warning in sbitmap_get().
Delayed virtqueue setup is not supported. Leave skipped I/O virtqueues
inactive and reject changes to the vring size or addresses while an
endpoint is attached. Userspace must clear the endpoint before changing
the vring configuration.
I found this bug myself, though the patch was written with AI assistance.
Fixes: 25b98b64e284 ("vhost scsi: alloc cmds per vq instead of session")
Assisted-by: OpenAI-Codex:GPT-5.6
Signed-off-by: Linfeng Sun <linfeng.sun.dev@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261010-fix-warning-in-sbitmap_get-v3-1-33de562d1359@gmail.com>
|
|
Split virtqueues save descriptor flags and next indexes in desc_extra
before publishing descriptors to the device. The detach path uses the
shadow next index, but decides whether to continue by rereading the NEXT
flag from the shared descriptor.
Use the shadow flags for the continuation decision as well. This keeps
all detach metadata in the same driver-owned snapshot and avoids an
unnecessary shared-ring read.
Fixes: 72b5e8958738 ("virtio-ring: store DMA metadata in desc_extra for split virtqueue")
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Assisted-by: LLM
Signed-off-by: sungbyeongchan <tjdqudcks0424@naver.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20261008050423.32221-1-tjdqudcks0424@naver.com>
|
|
vhost_get_vq_desc() returns the number of device readable (output) and
device writable (input) iovec segments. A guest can submit an eventq
descriptor chain that fills vq->iov with output segments, setting out
to ARRAY_SIZE(vq->iov) and in to zero.
When the host later queues a vhost-scsi event, vhost_scsi_do_evt_work()
uses out as the index of the first input segment. This accesses one element
beyond vq->iov and triggers an array-index-out-of-bounds report. With
panic_on_warn enabled, the report panics the host.
The event queue accepts device writable buffers only. Reject chains that
contain output segments or lack an input segment before using out as an
index. The input check prevents the out-of-bounds access, while the output
check validates the eventq descriptor direction.
I found the descriptor direction issue myself, though the patch was
written with AI assistance. The out-of-bounds read with an
output-only chain was separately reported by Jia Jia.
Fixes: a6c9af87363c ("tcm_vhost: Add hotplug/hotunplug support")
Reported-by: Jia Jia <physicalmtea@gmail.com>
Assisted-by: OpenAI-Codex:GPT-5
Signed-off-by: Linfeng Sun <linfeng.sun.dev@gmail.com>
Message-ID: <20261009-fix-arrary-index-out-of-bounds-in-vhost_scsi_complete_events-v2-1-1f7689a54735@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
|
|
When the WCN7850 is powered up from the Bluetooth side first, i.e.
BT_EN is asserted while WLAN_EN is still low, the chip starts from a
cold state and is not ready to accept the first HCI command 50 ms after
BT_EN goes high. The first command (0xfc00, read QCA version) is lost,
hci_qca does not resend it and fails after the 2 s command timeout:
Bluetooth: hci0: Reading QCA version information failed (-110)
When WLAN is powered first, 50 ms is sufficient, so the failure only
shows up in the BT-first ordering. Measured on the affected board by
which enable pin rose first:
pwup_delay_ms BT-first boots with -110
50 9 9
75 23 0
100 21 0
300 20 0
Raise the power-up delay of WCN7850 to 100 ms. The delay is shared by
the Bluetooth and WLAN targets, so WLAN power-up also waits 50 ms
longer.
Fixes: 2f1630f437df ("power: pwrseq: add a driver for the PMU module on the QCom WCN chipsets")
Signed-off-by: Cheng Jiang <cheng.jiang@oss.qualcomm.com>
Signed-off-by: Shuai Zhang <shuai.zhang@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://patch.msgid.link/20261102094800.3225454-1-shuai.zhang@oss.qualcomm.com
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
|
|
s3c24xx_serial_getclk() returns the best baud clock with the reference
which clk_get() has taken. s3c24xx_serial_set_termios() stores it in
ourport->baudclk and disables the clock which was there before, but does
not put it.
clk_get() hands out a new handle on every call, so the new clock never
equals the old one, and every call of ->set_termios() leaks a handle: a
struct clk and a copy of the connection name. Whoever has the port open
can repeat that at will with tcsetattr().
Put the old clock when it is replaced, and the new one in case it is not
stored.
Fixes: 5f5a7a5578c5 ("serial: samsung: switch to clkdev based clock lookup")
Cc: stable <stable@kernel.org>
Cc: Krzysztof Kozlowski <krzk@kernel.org>
Cc: Peter Griffin <peter.griffin@linaro.org>
Cc: Alim Akhtar <alim.akhtar@samsung.com>
Cc: Jiri Slaby <jirislaby@kernel.org>
Cc: Kukjin Kim <kgene.kim@samsung.com>
Cc: Thomas Abraham <thomas.abraham@linaro.org>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://patch.msgid.link/20261027-tty-llm-v1-22-99b0202ae546@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
|
s3c24xx_serial_rx_chars_dma() starts by taking a reference to the tty.
If the interrupt is not a receive timeout, it switches the port from PIO
back to DMA and leaves through the "finish" label, which does not drop
the reference.
A port with RX DMA changes to PIO on every receive timeout, that is at
the end of every burst of data, and comes back this way when the next
burst starts. So each burst takes a reference which is never dropped,
and the tty structure is not freed when the port is closed for the last
time.
Drop the reference on both paths.
Fixes: b543c301c2c5 ("serial: samsung: add DMA support for RX")
Cc: stable <stable@kernel.org>
Cc: Krzysztof Kozlowski <krzk@kernel.org>
Cc: Peter Griffin <peter.griffin@linaro.org>
Cc: Alim Akhtar <alim.akhtar@samsung.com>
Cc: Jiri Slaby <jirislaby@kernel.org>
Cc: Robert Baldyga <r.baldyga@samsung.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://patch.msgid.link/20261027-tty-llm-v1-21-99b0202ae546@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|