kernel.git/drivers/md/dm-thin.c, branch linux-5.4.y

dm thin: make get_first_thin use rcu-safe list first function

2025-02-01T17:18:45Z

commit 80f130bfad1dab93b95683fc39b87235682b8f72 upstream. The documentation in rculist.h explains the absence of list_empty_rcu() and cautions programmers against relying on a list_empty() -> list_first() sequence in RCU safe code. This is because each of these functions performs its own READ_ONCE() of the list head. This can lead to a situation where the list_empty() sees a valid list entry, but the subsequent list_first() sees a different view of list head state after a modification. In the case of dm-thin, this author had a production box crash from a GP fault in the process_deferred_bios path. This function saw a valid list head in get_first_thin() but when it subsequently dereferenced that and turned it into a thin_c, it got the inside of the struct pool, since the list was now empty and referring to itself. The kernel on which this occurred printed both a warning about a refcount_t being saturated, and a UBSAN error for an out-of-bounds cpuid access in the queued spinlock, prior to the fault itself. When the resulting kdump was examined, it was possible to see another thread patiently waiting in thin_dtr's synchronize_rcu. The thin_dtr call managed to pull the thin_c out of the active thins list (and have it be the last entry in the active_thins list) at just the wrong moment which lead to this crash. Fortunately, the fix here is straight forward. Switch get_first_thin() function to use list_first_or_null_rcu() which performs just a single READ_ONCE() and returns NULL if the list is already empty. This was run against the devicemapper test suite's thin-provisioning suites for delete and suspend and no regressions were observed. Signed-off-by: Krister Johansen Fixes: b10ebd34ccca ("dm thin: fix rcu_read_lock being held in code that can sleep") Cc: stable@vger.kernel.org Acked-by: Ming-Hung Tsai Signed-off-by: Mikulas Patocka Signed-off-by: Greg Kroah-Hartman

dm thin: Add missing destroy_work_on_stack()

2024-12-14T18:44:44Z

commit e74fa2447bf9ed03d085b6d91f0256cc1b53f1a8 upstream. This commit add missed destroy_work_on_stack() operations for pw->worker in pool_work_wait(). Fixes: e7a3e871d895 ("dm thin: cleanup noflush_work to use a proper completion") Cc: stable@vger.kernel.org Signed-off-by: Yuan Can Signed-off-by: Mikulas Patocka Signed-off-by: Greg Kroah-Hartman

dm thin: fix deadlock when swapping to thin device

2023-04-05T09:16:40Z

commit 9bbf5feecc7eab2c370496c1c161bbfe62084028 upstream. This is an already known issue that dm-thin volume cannot be used as swap, otherwise a deadlock may happen when dm-thin internal memory demand triggers swap I/O on the dm-thin volume itself. But thanks to commit a666e5c05e7c ("dm: fix deadlock when swapping to encrypted device"), the limit_swap_bios target flag can also be used for dm-thin to avoid the recursive I/O when it is used as swap. Fix is to simply set ti->limit_swap_bios to true in both pool_ctr() and thin_ctr(). In my test, I create a dm-thin volume /dev/vg/swap and use it as swap device. Then I run fio on another dm-thin volume /dev/vg/main and use large --blocksize to trigger swap I/O onto /dev/vg/swap. The following fio command line is used in my test, fio --name recursive-swap-io --lockmem 1 --iodepth 128 \ --ioengine libaio --filename /dev/vg/main --rw randrw \ --blocksize 1M --numjobs 32 --time_based --runtime=12h Without this fix, the whole system can be locked up within 15 seconds. With this fix, there is no any deadlock or hung task observed after 2 hours of running fio. Furthermore, if blocksize is changed from 1M to 128M, after around 30 seconds fio has no visible I/O, and the out-of-memory killer message shows up in kernel message. After around 20 minutes all fio processes are killed and the whole system is back to being alive. This is exactly what is expected when recursive I/O happens on dm-thin volume when it is used as swap. Depends-on: a666e5c05e7c ("dm: fix deadlock when swapping to encrypted device") Cc: stable@vger.kernel.org Signed-off-by: Coly Li Acked-by: Mikulas Patocka Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman

dm thin: add cond_resched() to various workqueue loops

2023-03-11T15:43:58Z

[ Upstream commit e4f80303c2353952e6e980b23914e4214487f2a6 ] Otherwise on resource constrained systems these workqueues may be too greedy. Signed-off-by: Mike Snitzer Signed-off-by: Sasha Levin

dm thin: resume even if in FAIL mode

2023-01-18T10:41:53Z

[ Upstream commit 19eb1650afeb1aa86151f61900e9e5f1de5d8d02 ] If a thinpool set fail_io while suspending, resume will fail with: device-mapper: resume ioctl on vg-thinpool failed: Invalid argument The thin-pool also can't be removed if an in-flight bio is in the deferred list. This can be easily reproduced using: echo "offline" > /sys/block/sda/device/state dd if=/dev/zero of=/dev/mapper/thin bs=4K count=1 dmsetup suspend /dev/mapper/pool mkfs.ext4 /dev/mapper/thin dmsetup resume /dev/mapper/pool The root cause is maybe_resize_data_dev() will check fail_io and return error before called dm_resume. Fix this by adding FAIL mode check at the end of pool_preresume(). Cc: stable@vger.kernel.org Fixes: da105ed5fd7e ("dm thin metadata: introduce dm_pool_abort_metadata") Signed-off-by: Luo Meng Signed-off-by: Mike Snitzer Signed-off-by: Sasha Levin

dm thin: Fix UAF in run_timer_softirq()

2023-01-18T10:41:47Z

commit 88430ebcbc0ec637b710b947738839848c20feff upstream. When dm_resume() and dm_destroy() are concurrent, it will lead to UAF, as follows: BUG: KASAN: use-after-free in __run_timers+0x173/0x710 Write of size 8 at addr ffff88816d9490f0 by task swapper/0/0 Call Trace: dump_stack_lvl+0x73/0x9f print_report.cold+0x132/0xaa2 _raw_spin_lock_irqsave+0xcd/0x160 __run_timers+0x173/0x710 kasan_report+0xad/0x110 __run_timers+0x173/0x710 __asan_store8+0x9c/0x140 __run_timers+0x173/0x710 call_timer_fn+0x310/0x310 pvclock_clocksource_read+0xfa/0x250 kvm_clock_read+0x2c/0x70 kvm_clock_get_cycles+0xd/0x20 ktime_get+0x5c/0x110 lapic_next_event+0x38/0x50 clockevents_program_event+0xf1/0x1e0 run_timer_softirq+0x49/0x90 __do_softirq+0x16e/0x62c __irq_exit_rcu+0x1fa/0x270 irq_exit_rcu+0x12/0x20 sysvec_apic_timer_interrupt+0x8e/0xc0 One of the concurrency UAF can be shown as below: use free do_resume | __find_device_hash_cell | dm_get | atomic_inc(&md->holders) | | dm_destroy | __dm_destroy | if (!dm_suspended_md(md)) | atomic_read(&md->holders) | msleep(1) dm_resume | __dm_resume | dm_table_resume_targets | pool_resume | do_waker #add delay work | dm_put | atomic_dec(&md->holders) | | dm_table_destroy | pool_dtr | __pool_dec | __pool_destroy | destroy_workqueue | kfree(pool) # free pool time out __do_softirq run_timer_softirq # pool has already been freed This can be easily reproduced using: 1. create thin-pool 2. dmsetup suspend pool 3. dmsetup resume pool 4. dmsetup remove_all # Concurrent with 3 The root cause of this UAF bug is that dm_resume() adds timer after dm_destroy() skips cancelling the timer because of suspend status. After timeout, it will call run_timer_softirq(), however pool has already been freed. The concurrency UAF bug will happen. Therefore, cancelling timer again in __pool_destroy(). Cc: stable@vger.kernel.org Fixes: 991d9fa02da0d ("dm: add thin provisioning target") Signed-off-by: Luo Meng Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman

dm thin: fix use-after-free crash in dm_sm_register_threshold_callback

2022-08-25T09:18:14Z

commit 3534e5a5ed2997ca1b00f44a0378a075bd05e8a3 upstream. Fault inject on pool metadata device reports: BUG: KASAN: use-after-free in dm_pool_register_metadata_threshold+0x40/0x80 Read of size 8 at addr ffff8881b9d50068 by task dmsetup/950 CPU: 7 PID: 950 Comm: dmsetup Tainted: G W 5.19.0-rc6 #1 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-1.fc33 04/01/2014 Call Trace: dump_stack_lvl+0x34/0x44 print_address_description.constprop.0.cold+0xeb/0x3f4 kasan_report.cold+0xe6/0x147 dm_pool_register_metadata_threshold+0x40/0x80 pool_ctr+0xa0a/0x1150 dm_table_add_target+0x2c8/0x640 table_load+0x1fd/0x430 ctl_ioctl+0x2c4/0x5a0 dm_ctl_ioctl+0xa/0x10 __x64_sys_ioctl+0xb3/0xd0 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x46/0xb0 This can be easily reproduced using: echo offline > /sys/block/sda/device/state dd if=/dev/zero of=/dev/mapper/thin bs=4k count=10 dmsetup load pool --table "0 20971520 thin-pool /dev/sda /dev/sdb 128 0 0" If a metadata commit fails, the transaction will be aborted and the metadata space maps will be destroyed. If a DM table reload then happens for this failed thin-pool, a use-after-free will occur in dm_sm_register_threshold_callback (called from dm_pool_register_metadata_threshold). Fix this by in dm_pool_register_metadata_threshold() by returning the -EINVAL error if the thin-pool is in fail mode. Also fail pool_ctr() with a new error message: "Error registering metadata threshold". Fixes: ac8c3f3df65e4 ("dm thin: generate event when metadata threshold passed") Cc: stable@vger.kernel.org Reported-by: Hulk Robot Signed-off-by: Luo Meng Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman

dm thin: don't allow changing data device during thin-pool reload

2020-02-24T07:36:49Z

[ Upstream commit 873937e75f9a8ea231a502c3d29d9cb6ad91b3ef ] The existing code allows changing the data device when the thin-pool target is reloaded. This capability is not required and only complicates device lifetime guarantees. This can cause crashes like the one reported here: https://bugzilla.redhat.com/show_bug.cgi?id=1788596 where the kernel tries to issue a flush bio located in a structure that was already freed. Take the first step to simplifying the thin-pool's data device lifetime by disallowing changing it. Like the thin-pool's metadata device, the data device is now set in pool_create() and it cannot be changed for a given thin-pool. Signed-off-by: Mikulas Patocka Signed-off-by: Mike Snitzer Signed-off-by: Sasha Levin

dm thin: fix use-after-free in metadata_pre_commit_callback

2020-02-05T21:22:53Z

[ Upstream commit a4a8d286586d4b28c8517a51db8d86954aadc74b ] dm-thin uses struct pool to hold the state of the pool. There may be multiple pool_c's pointing to a given pool, each pool_c represents a loaded target. pool_c's may be created and destroyed arbitrarily and the pool contains a reference count of pool_c's pointing to it. Since commit 694cfe7f31db3 ("dm thin: Flush data device before committing metadata") a pointer to pool_c is passed to dm_pool_register_pre_commit_callback and this function stores it in pmd->pre_commit_context. If this pool_c is freed, but pool is not (because there is another pool_c referencing it), we end up in a situation where pmd->pre_commit_context structure points to freed pool_c. It causes a crash in metadata_pre_commit_callback. Fix this by moving the dm_pool_register_pre_commit_callback() from pool_ctr() to pool_preresume(). This way the in-core thin-pool metadata is only ever armed with callback data whose lifetime matches the active thin-pool target. In should be noted that this fix preserves the ability to load a thin-pool table that uses a different data block device (that contains the same data) -- though it is unclear if that capability is still useful and/or needed. Fixes: 694cfe7f31db3 ("dm thin: Flush data device before committing metadata") Cc: stable@vger.kernel.org Reported-by: Zdenek Kabelac Reported-by: Mikulas Patocka Signed-off-by: Mike Snitzer Signed-off-by: Sasha Levin

dm thin: Flush data device before committing metadata

2019-12-21T10:05:03Z

commit 694cfe7f31db36912725e63a38a5179c8628a496 upstream. The thin provisioning target maintains per thin device mappings that map virtual blocks to data blocks in the data device. When we write to a shared block, in case of internal snapshots, or provision a new block, in case of external snapshots, we copy the shared block to a new data block (COW), update the mapping for the relevant virtual block and then issue the write to the new data block. Suppose the data device has a volatile write-back cache and the following sequence of events occur: 1. We write to a shared block 2. A new data block is allocated 3. We copy the shared block to the new data block using kcopyd (COW) 4. We insert the new mapping for the virtual block in the btree for that thin device. 5. The commit timeout expires and we commit the metadata, that now includes the new mapping from step (4). 6. The system crashes and the data device's cache has not been flushed, meaning that the COWed data are lost. The next time we read that virtual block of the thin device we read it from the data block allocated in step (2), since the metadata have been successfully committed. The data are lost due to the crash, so we read garbage instead of the old, shared data. This has the following implications: 1. In case of writes to shared blocks, with size smaller than the pool's block size (which means we first copy the whole block and then issue the smaller write), we corrupt data that the user never touched. 2. In case of writes to shared blocks, with size equal to the device's logical block size, we fail to provide atomic sector writes. When the system recovers the user will read garbage from that sector instead of the old data or the new data. 3. Even for writes to shared blocks, with size equal to the pool's block size (overwrites), after the system recovers, the written sectors will contain garbage instead of a random mix of sectors containing either old data or new data, thus we fail again to provide atomic sectors writes. 4. Even when the user flushes the thin device, because we first commit the metadata and then pass down the flush, the same risk for corruption exists (if the system crashes after the metadata have been committed but before the flush is passed down to the data device.) The only case which is unaffected is that of writes with size equal to the pool's block size and with the FUA flag set. But, because FUA writes trigger metadata commits, this case can trigger the corruption indirectly. Moreover, apart from internal and external snapshots, the same issue exists for newly provisioned blocks, when block zeroing is enabled. After the system recovers the provisioned blocks might contain garbage instead of zeroes. To solve this and avoid the potential data corruption we flush the pool's data device **before** committing its metadata. This ensures that the data blocks of any newly inserted mappings are properly written to non-volatile storage and won't be lost in case of a crash. Cc: stable@vger.kernel.org Signed-off-by: Nikos Tsironis Acked-by: Joe Thornber Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman