From: Genjian Zhang <zhanggenjian@xxxxxxxxxx>
Hello!
We found that 13b2856037a6 ("loop: Check for overflow while configuring loop") lost a unlock loop_ctl_mutex in loop_get_status(...).
which caused syzbot to report a UAF issue. However, the upstream patch does not have this issue.
So, we revert this patch and directly apply the unmodified upstream patch.
This series of patches was also sent to 4.19.y.
https://lore.kernel.org/all/20240301013028.2293831-1-zhanggenjian@xxxxxxx/
Risk use-after-free as reported by syzbot：
[ 84.669496] ==================================================================
[ 84.670021] BUG: KASAN: use-after-free in __mutex_lock.isra.9+0xc13/0xcb0
[ 84.670433] Read of size 4 at addr ffff88808dba43b8 by task syz-executor.22/14230
[ 84.670885]
[ 84.670987] CPU: 1 PID: 14230 Comm: syz-executor.22 Not tainted 5.4.270 #4
[ 84.671397] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-1kylin1 04/01/2014
[ 84.671927] Call Trace:
[ 84.672085] dump_stack+0x94/0xc7
[ 84.672293] ? __mutex_lock.isra.9+0xc13/0xcb0
[ 84.672569] print_address_description.constprop.6+0x16/0x220
[ 84.672915] ? __mutex_lock.isra.9+0xc13/0xcb0
[ 84.673187] ? __mutex_lock.isra.9+0xc13/0xcb0
[ 84.673462] __kasan_report.cold.9+0x1a/0x32
[ 84.673723] ? __mutex_lock.isra.9+0xc13/0xcb0
[ 84.673993] kasan_report+0x10/0x20
[ 84.674208] __mutex_lock.isra.9+0xc13/0xcb0
[ 84.674468] ? __mutex_lock.isra.9+0x617/0xcb0
[ 84.674739] ? ww_mutex_lock_interruptible+0x100/0x100
[ 84.675055] ? ww_mutex_lock_interruptible+0x100/0x100
[ 84.675369] ? kobject_get_unless_zero+0x144/0x190
[ 84.675668] ? kobject_del+0x60/0x60
[ 84.675893] ? __module_get+0x120/0x120
[ 84.676128] ? __mutex_lock_slowpath+0x10/0x10
[ 84.676399] mutex_lock_killable+0xde/0xf0
[ 84.676652] ? __mutex_lock_killable_slowpath+0x10/0x10
[ 84.676967] ? __mutex_lock_slowpath+0x10/0x10
[ 84.677243] ? disk_block_events+0x1d/0x120
[ 84.677509] lo_open+0x16/0xc0
[ 84.677701] ? lo_compat_ioctl+0x160/0x160
[ 84.677954] __blkdev_get+0xb0f/0x1160
[ 84.678185] ? bd_may_claim+0xd0/0xd0
[ 84.678410] ? bdev_disk_changed+0x190/0x190
[ 84.678674] ? _raw_spin_lock+0x7c/0xd0
[ 84.678915] ? _raw_write_lock_bh+0xd0/0xd0
[ 84.679172] blkdev_get+0x9b/0x290
[ 84.679381] ? ihold+0x1a/0x40
[ 84.679574] blkdev_open+0x1bd/0x240
[ 84.679794] do_dentry_open+0x439/0x1000
[ 84.680035] ? blkdev_get_by_dev+0x60/0x60
[ 84.680286] ? __x64_sys_fchdir+0x1a0/0x1a0
[ 84.680557] ? inode_permission+0x86/0x320
[ 84.680814] path_openat+0x998/0x4120
[ 84.681044] ? stack_trace_consume_entry+0x160/0x160
[ 84.681348] ? do_futex+0x136/0x1880
[ 84.681568] ? path_mountpoint+0xb50/0xb50
[ 84.681823] ? save_stack+0x4d/0x80
[ 84.682038] ? save_stack+0x19/0x80
[ 84.682253] ? __kasan_kmalloc.constprop.6+0xc1/0xd0
[ 84.682553] ? kmem_cache_alloc+0xc7/0x210
[ 84.682804] ? getname_flags+0xc4/0x560
[ 84.683045] ? do_sys_open+0x1ce/0x450
[ 84.683272] ? do_syscall_64+0x9a/0x330
[ 84.683509] ? entry_SYSCALL_64_after_hwframe+0x5c/0xc1
[ 84.683826] ? _raw_spin_lock+0x7c/0xd0
[ 84.684063] ? _raw_write_lock_bh+0xd0/0xd0
[ 84.684319] ? futex_exit_release+0x60/0x60
[ 84.684574] ? kasan_unpoison_shadow+0x30/0x40
[ 84.684844] ? __kasan_kmalloc.constprop.6+0xc1/0xd0
[ 84.685149] ? get_partial_node.isra.83.part.84+0x1e5/0x340
[ 84.685485] ? __fget_light+0x1d1/0x550
[ 84.685721] do_filp_open+0x197/0x270
[ 84.685946] ? may_open_dev+0xd0/0xd0
[ 84.686172] ? kasan_unpoison_shadow+0x30/0x40
[ 84.686443] ? __kasan_kmalloc.constprop.6+0xc1/0xd0
[ 84.686743] ? __alloc_fd+0x1a3/0x580
[ 84.686973] do_sys_open+0x2c7/0x450
[ 84.687195] ? filp_open+0x60/0x60
[ 84.687406] ? __x64_sys_timer_settime32+0x280/0x280
[ 84.687707] do_syscall_64+0x9a/0x330
[ 84.687931] ? syscall_return_slowpath+0x17a/0x230
[ 84.688221] entry_SYSCALL_64_after_hwframe+0x5c/0xc1
[ 84.688524]
[ 84.688622] Allocated by task 14056:
[ 84.688842] save_stack+0x19/0x80
[ 84.689044] __kasan_kmalloc.constprop.6+0xc1/0xd0
[ 84.689333] kmem_cache_alloc_node+0xe2/0x230
[ 84.689600] copy_process+0x165c/0x72d0
[ 84.689833] _do_fork+0xf9/0x9a0
[ 84.690032] __x64_sys_clone+0x17a/0x200
[ 84.690271] do_syscall_64+0x9a/0x330
[ 84.690496] entry_SYSCALL_64_after_hwframe+0x5c/0xc1
[ 84.690800]
[ 84.690903] Freed by task 0:
[ 84.691081] save_stack+0x19/0x80
[ 84.691287] __kasan_slab_free+0x125/0x170
[ 84.691535] kmem_cache_free+0x7a/0x2a0
[ 84.691774] __put_task_struct+0x1ec/0x4a0
[ 84.692023] delayed_put_task_struct+0x178/0x1d0
[ 84.692303] rcu_core+0x538/0x16c0
[ 84.692512] __do_softirq+0x175/0x63d
[ 84.692741]
[ 84.692840] The buggy address belongs to the object at ffff88808dba4380
[ 84.692840] which belongs to the cache task_struct of size 3328
[ 84.693584] The buggy address is located 56 bytes inside of
[ 84.693584] 3328-byte region [ffff88808dba4380, ffff88808dba5080)
[ 84.694272] The buggy address belongs to the page:
[ 84.694563] page:ffffea000236e800 refcount:1 mapcount:0 mapping:ffff8881838acdc0 index:0x0 compound_mapcount: 0
[ 84.695166] flags: 0x100000000010200(slab|head)
[ 84.695457] raw: 0100000000010200 dead000000000100 dead000000000122 ffff8881838acdc0
[ 84.695919] raw: 0000000000000000 0000000000090009 00000001ffffffff 0000000000000000
[ 84.696375] page dumped because: kasan: bad access detected
[ 84.696705]
[ 84.696801] Memory state around the buggy address:
[ 84.697089] ffff88808dba4280: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 84.697519] ffff88808dba4300: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[ 84.697945] >ffff88808dba4380: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 84.698371] ^
[ 84.698674] ffff88808dba4400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 84.699111] ffff88808dba4480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[ 84.699537] ==================================================================
[ 84.699965] Disabling lock debugging due to kernel taint
Best regards
Genjian
Genjian Zhang (1):
Revert "loop: Check for overflow while configuring loop"
Martijn Coenen (5):
loop: Call loop_config_discard() only after new config is applied
loop: Remove sector_t truncation checks
loop: Factor out setting loop device size
loop: Refactor loop_set_status() size calculation
loop: Factor out configuring loop from status
Siddh Raman Pant (1):
loop: Check for overflow while configuring loop
Zhong Jinghua (1):
loop: loop_set_status_from_info() check before assignment
drivers/block/loop.c | 185 ++++++++++++++++++++++++-------------------
1 file changed, 104 insertions(+), 81 deletions(-)
--
2.25.1
