user/sven/linux.git/fs/block_dev.c, branch v5.4.60

block: Fix use-after-free in blkdev_get()

2020-06-24T15:50:47Z

[ Upstream commit 2d3a8e2deddea6c89961c422ec0c5b851e648c14 ] In blkdev_get() we call __blkdev_get() to do some internal jobs and if there is some errors in __blkdev_get(), the bdput() is called which means we have released the refcount of the bdev (actually the refcount of the bdev inode). This means we cannot access bdev after that point. But acctually bdev is still accessed in blkdev_get() after calling __blkdev_get(). This results in use-after-free if the refcount is the last one we released in __blkdev_get(). Let's take a look at the following scenerio: CPU0 CPU1 CPU2 blkdev_open blkdev_open Remove disk bd_acquire blkdev_get __blkdev_get del_gendisk bdev_unhash_inode bd_acquire bdev_get_gendisk bd_forget failed because of unhashed bdput bdput (the last one) bdev_evict_inode access bdev => use after free [ 459.350216] BUG: KASAN: use-after-free in __lock_acquire+0x24c1/0x31b0 [ 459.351190] Read of size 8 at addr ffff88806c815a80 by task syz-executor.0/20132 [ 459.352347] [ 459.352594] CPU: 0 PID: 20132 Comm: syz-executor.0 Not tainted 4.19.90 #2 [ 459.353628] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 [ 459.354947] Call Trace: [ 459.355337] dump_stack+0x111/0x19e [ 459.355879] ? __lock_acquire+0x24c1/0x31b0 [ 459.356523] print_address_description+0x60/0x223 [ 459.357248] ? __lock_acquire+0x24c1/0x31b0 [ 459.357887] kasan_report.cold+0xae/0x2d8 [ 459.358503] __lock_acquire+0x24c1/0x31b0 [ 459.359120] ? _raw_spin_unlock_irq+0x24/0x40 [ 459.359784] ? lockdep_hardirqs_on+0x37b/0x580 [ 459.360465] ? _raw_spin_unlock_irq+0x24/0x40 [ 459.361123] ? finish_task_switch+0x125/0x600 [ 459.361812] ? finish_task_switch+0xee/0x600 [ 459.362471] ? mark_held_locks+0xf0/0xf0 [ 459.363108] ? __schedule+0x96f/0x21d0 [ 459.363716] lock_acquire+0x111/0x320 [ 459.364285] ? blkdev_get+0xce/0xbe0 [ 459.364846] ? blkdev_get+0xce/0xbe0 [ 459.365390] __mutex_lock+0xf9/0x12a0 [ 459.365948] ? blkdev_get+0xce/0xbe0 [ 459.366493] ? bdev_evict_inode+0x1f0/0x1f0 [ 459.367130] ? blkdev_get+0xce/0xbe0 [ 459.367678] ? destroy_inode+0xbc/0x110 [ 459.368261] ? mutex_trylock+0x1a0/0x1a0 [ 459.368867] ? __blkdev_get+0x3e6/0x1280 [ 459.369463] ? bdev_disk_changed+0x1d0/0x1d0 [ 459.370114] ? blkdev_get+0xce/0xbe0 [ 459.370656] blkdev_get+0xce/0xbe0 [ 459.371178] ? find_held_lock+0x2c/0x110 [ 459.371774] ? __blkdev_get+0x1280/0x1280 [ 459.372383] ? lock_downgrade+0x680/0x680 [ 459.373002] ? lock_acquire+0x111/0x320 [ 459.373587] ? bd_acquire+0x21/0x2c0 [ 459.374134] ? do_raw_spin_unlock+0x4f/0x250 [ 459.374780] blkdev_open+0x202/0x290 [ 459.375325] do_dentry_open+0x49e/0x1050 [ 459.375924] ? blkdev_get_by_dev+0x70/0x70 [ 459.376543] ? __x64_sys_fchdir+0x1f0/0x1f0 [ 459.377192] ? inode_permission+0xbe/0x3a0 [ 459.377818] path_openat+0x148c/0x3f50 [ 459.378392] ? kmem_cache_alloc+0xd5/0x280 [ 459.379016] ? entry_SYSCALL_64_after_hwframe+0x49/0xbe [ 459.379802] ? path_lookupat.isra.0+0x900/0x900 [ 459.380489] ? __lock_is_held+0xad/0x140 [ 459.381093] do_filp_open+0x1a1/0x280 [ 459.381654] ? may_open_dev+0xf0/0xf0 [ 459.382214] ? find_held_lock+0x2c/0x110 [ 459.382816] ? lock_downgrade+0x680/0x680 [ 459.383425] ? __lock_is_held+0xad/0x140 [ 459.384024] ? do_raw_spin_unlock+0x4f/0x250 [ 459.384668] ? _raw_spin_unlock+0x1f/0x30 [ 459.385280] ? __alloc_fd+0x448/0x560 [ 459.385841] do_sys_open+0x3c3/0x500 [ 459.386386] ? filp_open+0x70/0x70 [ 459.386911] ? trace_hardirqs_on_thunk+0x1a/0x1c [ 459.387610] ? trace_hardirqs_off_caller+0x55/0x1c0 [ 459.388342] ? do_syscall_64+0x1a/0x520 [ 459.388930] do_syscall_64+0xc3/0x520 [ 459.389490] entry_SYSCALL_64_after_hwframe+0x49/0xbe [ 459.390248] RIP: 0033:0x416211 [ 459.390720] Code: 75 14 b8 02 00 00 00 0f 05 48 3d 01 f0 ff ff 0f 83 04 19 00 00 c3 48 83 ec 08 e8 0a fa ff ff 48 89 04 24 b8 02 00 00 00 0f 05 <48> 8b 3c 24 48 89 c2 e8 53 fa ff ff 48 89 d0 48 83 c4 08 48 3d 01 [ 459.393483] RSP: 002b:00007fe45dfe9a60 EFLAGS: 00000293 ORIG_RAX: 0000000000000002 [ 459.394610] RAX: ffffffffffffffda RBX: 00007fe45dfea6d4 RCX: 0000000000416211 [ 459.395678] RDX: 00007fe45dfe9b0a RSI: 0000000000000002 RDI: 00007fe45dfe9b00 [ 459.396758] RBP: 000000000076bf20 R08: 0000000000000000 R09: 000000000000000a [ 459.397930] R10: 0000000000000075 R11: 0000000000000293 R12: 00000000ffffffff [ 459.399022] R13: 0000000000000bd9 R14: 00000000004cdb80 R15: 000000000076bf2c [ 459.400168] [ 459.400430] Allocated by task 20132: [ 459.401038] kasan_kmalloc+0xbf/0xe0 [ 459.401652] kmem_cache_alloc+0xd5/0x280 [ 459.402330] bdev_alloc_inode+0x18/0x40 [ 459.402970] alloc_inode+0x5f/0x180 [ 459.403510] iget5_locked+0x57/0xd0 [ 459.404095] bdget+0x94/0x4e0 [ 459.404607] bd_acquire+0xfa/0x2c0 [ 459.405113] blkdev_open+0x110/0x290 [ 459.405702] do_dentry_open+0x49e/0x1050 [ 459.406340] path_openat+0x148c/0x3f50 [ 459.406926] do_filp_open+0x1a1/0x280 [ 459.407471] do_sys_open+0x3c3/0x500 [ 459.408010] do_syscall_64+0xc3/0x520 [ 459.408572] entry_SYSCALL_64_after_hwframe+0x49/0xbe [ 459.409415] [ 459.409679] Freed by task 1262: [ 459.410212] __kasan_slab_free+0x129/0x170 [ 459.410919] kmem_cache_free+0xb2/0x2a0 [ 459.411564] rcu_process_callbacks+0xbb2/0x2320 [ 459.412318] __do_softirq+0x225/0x8ac Fix this by delaying bdput() to the end of blkdev_get() which means we have finished accessing bdev. Fixes: 77ea887e433a ("implement in-kernel gendisk events handling") Reported-by: Hulk Robot Signed-off-by: Jason Yan Tested-by: Sedat Dilek Reviewed-by: Jan Kara Reviewed-by: Christoph Hellwig Reviewed-by: Dan Carpenter Cc: Christoph Hellwig Cc: Jens Axboe Cc: Ming Lei Cc: Jan Kara Cc: Dan Carpenter Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

hibernate: Allow uswsusp to write to swap

2020-04-23T08:36:34Z

[ Upstream commit 56939e014a6c212b317414faa307029e2e80c3b9 ] It turns out that there is one use case for programs being able to write to swap devices, and that is the userspace hibernation code. Quick fix: disable the S_SWAPFILE check if hibernation is configured. Fixes: dc617f29dbe5 ("vfs: don't allow writes to swap files") Reported-by: Domenico Andreoli Reported-by: Marian Klein Signed-off-by: Domenico Andreoli Reviewed-by: Darrick J. Wong Signed-off-by: Darrick J. Wong Signed-off-by: Sasha Levin

bdev: Refresh bdev size for disks without partitioning

2019-12-13T07:43:19Z

commit cba22d86e0a10b7070d2e6a7379dbea51aa0883c upstream. Currently, block device size in not updated on second and further open for block devices where partition scan is disabled. This is particularly annoying for example for DVD drives as that means block device size does not get updated once the media is inserted into a drive if the device is already open when inserting the media. This is actually always the case for example when pktcdvd is in use. Fix the problem by revalidating block device size on every open even for devices with partition scan disabled. Signed-off-by: Jan Kara Signed-off-by: Jens Axboe Cc: Laura Abbott Signed-off-by: Greg Kroah-Hartman

bdev: Factor out bdev revalidation into a common helper

2019-12-13T07:43:18Z

commit 731dc4868311ee097757b8746eaa1b4f8b2b4f1c upstream. Factor out code handling revalidation of bdev on disk change into a common helper. Signed-off-by: Jan Kara Signed-off-by: Jens Axboe Cc: Laura Abbott Signed-off-by: Greg Kroah-Hartman

Merge tag 'vfs-5.4-merge-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux

2019-09-19T00:35:20Z

Pull swap access updates from Darrick Wong: "Prohibit writing to active swap files and swap partitions. There's no non-malicious use case for allowing userspace to scribble on storage that the kernel thinks it owns" * tag 'vfs-5.4-merge-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux: vfs: don't allow writes to swap files mm: set S_SWAPFILE on blockdev swap devices

vfs: don't allow writes to swap files

2019-08-20T14:55:16Z

Don't let userspace write to an active swap file because the kernel effectively has a long term lease on the storage and things could get seriously corrupted if we let this happen. Signed-off-by: Darrick J. Wong Reviewed-by: Christoph Hellwig

block: remove REQ_NOWAIT_INLINE

2019-08-15T17:09:16Z

We had a few issues with this code, and there's still a problem around how we deal with error handling for chained/split bios. For now, just revert the code and we'll try again with a thoroug solution. This reverts commits: e15c2ffa1091 ("block: fix O_DIRECT error handling for bio fragments") 0eb6ddfb865c ("block: Fix __blkdev_direct_IO() for bio fragments") 6a43074e2f46 ("block: properly handle IOCB_NOWAIT for async O_DIRECT IO") 893a1c97205a ("blk-mq: allow REQ_NOWAIT to return an error inline") Signed-off-by: Jens Axboe

bdev: Fixup error handling in blkdev_get()

2019-08-08T13:37:03Z

Commit 89e524c04fa9 ("loop: Fix mount(2) failure due to race with LOOP_SET_FD") converted blkdev_get() to use the new helpers for finishing claiming of a block device. However the conversion botched the error handling in blkdev_get() and thus the bdev has been marked as held even in case __blkdev_get() returned error. This led to occasional warnings with block/001 test from blktests like: kernel: WARNING: CPU: 5 PID: 907 at fs/block_dev.c:1899 __blkdev_put+0x396/0x3a0 Correct the error handling. CC: stable@vger.kernel.org Fixes: 89e524c04fa9 ("loop: Fix mount(2) failure due to race with LOOP_SET_FD") Signed-off-by: Jan Kara Signed-off-by: Jens Axboe

block: fix O_DIRECT error handling for bio fragments

2019-08-07T18:19:43Z

0eb6ddfb865c tried to fix this up, but introduced a use-after-free of dio. Additionally, we still had an issue with error handling, as reported by Darrick: "I noticed a regression in xfs/747 (an unreleased xfstest for the xfs_scrub media scanning feature) on 5.3-rc3. I'll condense that down to a simpler reproducer: error-test: 0 209 linear 8:48 0 error-test: 209 1 error error-test: 210 6446894 linear 8:48 210 Basically we have a ~3G /dev/sdd and we set up device mapper to fail IO for sector 209 and to pass the io to the scsi device everywhere else. On 5.3-rc3, performing a directio pread of this range with a < 1M buffer (in other words, a request for fewer than MAX_BIO_PAGES bytes) yields EIO like you'd expect: pread64(3, 0x7f880e1c7000, 1048576, 0) = -1 EIO (Input/output error) pread: Input/output error +++ exited with 0 +++ But doing it with a larger buffer succeeds(!): pread64(3, "XFSB\0\0\20\0\0\0\0\0\0\fL\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 1146880, 0) = 1146880 read 1146880/1146880 bytes at offset 0 1 MiB, 1 ops; 0.0009 sec (1.124 GiB/sec and 1052.6316 ops/sec) +++ exited with 0 +++ (Note that the part of the buffer corresponding to the dm-error area is uninitialized) On 5.3-rc2, both commands would fail with EIO like you'd expect. The only change between rc2 and rc3 is commit 0eb6ddfb865c ("block: Fix __blkdev_direct_IO() for bio fragments"). AFAICT we end up in __blkdev_direct_IO with a 1120K buffer, which gets split into two bios: one for the first BIO_MAX_PAGES worth of data (1MB) and a second one for the 96k after that." Fix this by noting that it's always safe to dereference dio if we get BLK_QC_T_EAGAIN returned, as end_io hasn't been run for that case. So we can safely increment the dio size before calling submit_bio(), and then decrement it on failure (not that it really matters, as the bio and dio are going away). For error handling, return to the original method of just using 'ret' for tracking the error, and the size tracking in dio->size. Fixes: 0eb6ddfb865c ("block: Fix __blkdev_direct_IO() for bio fragments") Fixes: 6a43074e2f46 ("block: properly handle IOCB_NOWAIT for async O_DIRECT IO") Reported-by: Darrick J. Wong Signed-off-by: Jens Axboe

block: Fix __blkdev_direct_IO() for bio fragments

2019-08-01T19:51:18Z

The recent fix to properly handle IOCB_NOWAIT for async O_DIRECT IO (patch 6a43074e2f46) introduced two problems with BIO fragment handling for direct IOs: 1) The dio size processed is calculated by incrementing the ret variable by the size of the bio fragment issued for the dio. However, this size is obtained directly from bio->bi_iter.bi_size AFTER the bio submission which may result in referencing the bi_size value after the bio completed, resulting in an incorrect value use. 2) The ret variable is not incremented by the size of the last bio fragment issued for the bio, leading to an invalid IO size being returned to the user. Fix both problem by using dio->size (which is incremented before the bio submission) to update the value of ret after bio submissions, including for the last bio fragment issued. Fixes: 6a43074e2f46 ("block: properly handle IOCB_NOWAIT for async O_DIRECT IO") Reported-by: Masato Suzuki Signed-off-by: Damien Le Moal Signed-off-by: Jens Axboe