Re: KASAN: use-after-free Read in inode_cgwb_move_to_attached

[Jan Kara]

On Tue 21-02-23 21:09:23, Palash Oswal wrote:
> Hello,
> I found the following issue using syzkaller on:
> HEAD commit : e60276b8c11ab4a8be23807bc67b04
> 8cfb937dfa (v6.0.8)
> git tree: stable
> 
> C Reproducer : https://gist.github.com/oswalpalash/bed0eba75def3cdd34a285428e9bcdc4
> Kernel .config :
> https://gist.github.com/oswalpalash/0962c70d774e5ec736a047bba917cecb
> 
> Console log :
> 
> ==================================================================
> BUG: KASAN: use-after-free in __list_del_entry_valid+0xf2/0x110
> Read of size 8 at addr ffff8880273c4358 by task syz-executor.1/6475

OK, so FAT inode was on writeback list (through inode->i_io_list) when
being freed. This should be fixed by commit 4e3c51f4e805 ("fs: do not
update freeing inode i_io_list"). Can you check please?

								Honza

> 
> CPU: 0 PID: 6475 Comm: syz-executor.1 Not tainted 6.0.8-pasta #2
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
> 1.13.0-1ubuntu1.1 04/01/2014
> Call Trace:
>  <TASK>
>  dump_stack_lvl+0xcd/0x134
>  print_report.cold+0xe5/0x63a
>  kasan_report+0x8a/0x1b0
>  __list_del_entry_valid+0xf2/0x110
>  inode_cgwb_move_to_attached+0x2ee/0x4e0
>  writeback_single_inode+0x3fa/0x510
>  write_inode_now+0x16a/0x1e0
>  blkdev_flush_mapping+0x168/0x220
>  blkdev_put_whole+0xd1/0xf0
>  blkdev_put+0x29b/0x700
>  deactivate_locked_super+0x8c/0xf0
>  deactivate_super+0xad/0xd0
>  cleanup_mnt+0x347/0x4b0
>  task_work_run+0xe0/0x1a0
>  exit_to_user_mode_prepare+0x25d/0x270
>  syscall_exit_to_user_mode+0x19/0x50
>  do_syscall_64+0x42/0xb0
>  entry_SYSCALL_64_after_hwframe+0x63/0xcd
> RIP: 0033:0x7f22bd29143b
> Code: ff ff ff f7 d8 64 89 01 48 83 c8 ff c3 66 90 f3 0f 1e fa 31 f6
> e9 05 00 00 00 0f 1f 44 00 00 f3 0f 1e fa b8 a6 00 00 00 0f 05 <48> 3d
> 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
> RSP: 002b:00007ffe505103b8 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6
> RAX: 0000000000000000 RBX: 0000000000000000 RCX: 00007f22bd29143b
> RDX: 00007f22bd228a90 RSI: 000000000000000a RDI: 00007ffe50510480
> RBP: 00007ffe50510480 R08: 00007f22bd2fba1f R09: 00007ffe50510240
> R10: 00000000fffffffb R11: 0000000000000246 R12: 00007f22bd2fb9f8
> R13: 00007ffe50511520 R14: 0000555556f4bd90 R15: 0000000000000032
>  </TASK>
> 
> Allocated by task 7810:
>  kasan_save_stack+0x1e/0x40
>  __kasan_slab_alloc+0x85/0xb0
>  kmem_cache_alloc_lru+0x25b/0xfb0
>  fat_alloc_inode+0x23/0x1e0
>  alloc_inode+0x61/0x1e0
>  new_inode_pseudo+0x13/0x80
>  new_inode+0x1b/0x40
>  fat_build_inode+0x146/0x2d0
>  vfat_create+0x249/0x390
>  lookup_open+0x10bc/0x1640
>  path_openat+0xa42/0x2840
>  do_filp_open+0x1ca/0x2a0
>  do_sys_openat2+0x61b/0x990
>  do_sys_open+0xc3/0x140
>  do_syscall_64+0x35/0xb0
>  entry_SYSCALL_64_after_hwframe+0x63/0xcd
> 
> Freed by task 16:
>  kasan_save_stack+0x1e/0x40
>  kasan_set_track+0x21/0x30
>  kasan_set_free_info+0x20/0x30
>  __kasan_slab_free+0xf5/0x180
>  kmem_cache_free.part.0+0xfc/0x4a0
>  i_callback+0x3f/0x70
>  rcu_core+0x785/0x1720
>  __do_softirq+0x1d0/0x908
> 
> Last potentially related work creation:
>  kasan_save_stack+0x1e/0x40
>  __kasan_record_aux_stack+0x7e/0x90
>  call_rcu+0x99/0x740
>  destroy_inode+0x129/0x1b0
>  iput.part.0+0x5cd/0x800
>  iput+0x58/0x70
>  dentry_unlink_inode+0x2e2/0x4a0
>  __dentry_kill+0x374/0x5e0
>  dput+0x656/0xbe0
>  __fput+0x3cc/0xa90
>  task_work_run+0xe0/0x1a0
>  exit_to_user_mode_prepare+0x25d/0x270
>  syscall_exit_to_user_mode+0x19/0x50
>  do_syscall_64+0x42/0xb0
>  entry_SYSCALL_64_after_hwframe+0x63/0xcd
> 
> The buggy address belongs to the object at ffff8880273c4080
>  which belongs to the cache fat_inode_cache of size 1488
> The buggy address is located 728 bytes inside of
>  1488-byte region [ffff8880273c4080, ffff8880273c4650)
> 
> The buggy address belongs to the physical page:
> page:ffffea00009cf100 refcount:1 mapcount:0 mapping:0000000000000000
> index:0xffff8880273c4ffe pfn:0x273c4
> flags: 0xfff00000000200(slab|node=0|zone=1|lastcpupid=0x7ff)
> raw: 00fff00000000200 ffffea00009cf0c8 ffff88801820e450 ffff888103e00e00
> raw: ffff8880273c4ffe ffff8880273c4080 0000000100000002 0000000000000000
> page dumped because: kasan: bad access detected
> page_owner tracks the page as allocated
> page last allocated via order 0, migratetype Reclaimable, gfp_mask
> 0x242050(__GFP_IO|__GFP_NOWARN|__GFP_COMP|__GFP_THISNODE|__GFP_RECLAIMABLE),
> pid 7810, tgid 7808 (syz-executor.1), ts 50543388569, free_ts
> 21285403579
>  prep_new_page+0x2c6/0x350
>  get_page_from_freelist+0xae9/0x3a80
>  __alloc_pages+0x321/0x710
>  cache_grow_begin+0x75/0x360
>  kmem_cache_alloc_lru+0xe72/0xfb0
>  fat_alloc_inode+0x23/0x1e0
>  alloc_inode+0x61/0x1e0
>  new_inode_pseudo+0x13/0x80
>  new_inode+0x1b/0x40
>  fat_fill_super+0x1c37/0x3710
>  mount_bdev+0x34d/0x410
>  legacy_get_tree+0x105/0x220
>  vfs_get_tree+0x89/0x2f0
>  path_mount+0x121b/0x1cb0
>  do_mount+0xf3/0x110
>  __x64_sys_mount+0x18f/0x230
> page last free stack trace:
>  free_pcp_prepare+0x5ab/0xd00
>  free_unref_page+0x19/0x410
>  slab_destroy+0x14/0x50
>  slabs_destroy+0x6a/0x90
>  ___cache_free+0x1e3/0x3b0
>  qlist_free_all+0x51/0x1c0
>  kasan_quarantine_reduce+0x13d/0x180
>  __kasan_slab_alloc+0x97/0xb0
>  kmem_cache_alloc+0x204/0xcc0
>  getname_flags+0xd2/0x5b0
>  vfs_fstatat+0x73/0xb0
>  __do_sys_newlstat+0x8b/0x110
>  do_syscall_64+0x35/0xb0
>  entry_SYSCALL_64_after_hwframe+0x63/0xcd
> 
> Memory state around the buggy address:
>  ffff8880273c4200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>  ffff8880273c4280: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> >ffff8880273c4300: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>                                                     ^
>  ffff8880273c4380: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>  ffff8880273c4400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> ==================================================================
-- 
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR