KASAN: use-after-free Read in raw_notifier_call_chain

[Wei Chen]

Dear Linux Developer,

Recently when using our tool to fuzz kernel, the following crash was triggered:

HEAD commit: c5eb0a61238d Linux 5.18-rc6
git tree: upstream
compiler: clang 12.0.0
console output:
https://drive.google.com/file/d/18zXJUa02xIJTl7FgYkXLMtCmDVz1SGYt/view?usp=sharing
kernel config: https://drive.google.com/file/d/12fNP5UArsFqTi2jjGomWuCk5evtgU0Gu/view?usp=sharing

Unfortunately, I don't have any reproducer for this crash yet.

BUG: KASAN: use-after-free in raw_notifier_call_chain+0x3e/0x170
Read of size 8 at addr ffff8880243a01f8 by task syz-executor/7401

CPU: 0 PID: 7401 Comm: syz-executor Not tainted 5.18.0-rc6 #2
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
1.13.0-1ubuntu1.1 04/01/2014
Call Trace:
 <TASK>
 dump_stack_lvl+0x1b1/0x28e
 print_address_description+0x65/0x4b0
 print_report+0xf4/0x1e0
 kasan_report+0xc3/0xf0
 raw_notifier_call_chain+0x3e/0x170
 netdev_run_todo+0x97e/0x1120
 tun_chr_close+0x11f/0x130
 __fput+0x3f3/0x860
 task_work_run+0x146/0x1c0
 do_exit+0x55d/0x1ef0
 do_group_exit+0x23b/0x2f0
 get_signal+0x2310/0x2360
 arch_do_signal_or_restart+0x7b/0x710
 exit_to_user_mode_prepare+0x154/0x210
 syscall_exit_to_user_mode+0x26/0x60
 do_syscall_64+0x4c/0x90
 entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7f5556695c4d
Code: Unable to access opcode bytes at RIP 0x7f5556695c23.
RSP: 002b:00007f5557863cd8 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca
RAX: fffffffffffffe00 RBX: 00007f55567bc0a0 RCX: 00007f5556695c4d
RDX: 0000000000000000 RSI: 0000000000000080 RDI: 00007f55567bc0a8
RBP: 00007f55567bc0a8 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 00007f55567bc0ac
R13: 00007ffcf6f9e50f R14: 00007ffcf6f9e6b0 R15: 00007f5557863dc0
 </TASK>

Allocated by task 6553:
 __kasan_slab_alloc+0xa3/0xd0
 kmem_cache_alloc+0x1c0/0x300
 copy_net_ns+0x13c/0x5c0
 create_new_namespaces+0x4ce/0xa00
 unshare_nsproxy_namespaces+0x11e/0x180
 ksys_unshare+0x573/0xb60
 __x64_sys_unshare+0x34/0x40
 do_syscall_64+0x3d/0x90
 entry_SYSCALL_64_after_hwframe+0x44/0xae

Freed by task 81:
 kasan_set_track+0x3d/0x60
 kasan_set_free_info+0x1f/0x40
 ____kasan_slab_free+0x130/0x1d0
 slab_free_freelist_hook+0x12e/0x1a0
 kmem_cache_free+0xc6/0x2c0
 cleanup_net+0xb81/0xc40
 process_one_work+0x83c/0x11a0
 worker_thread+0xa6c/0x1290
 kthread+0x266/0x300
 ret_from_fork+0x1f/0x30

Last potentially related work creation:
 kasan_save_stack+0x2b/0x50
 __kasan_record_aux_stack+0xac/0xc0
 insert_work+0x54/0x3e0
 __queue_work+0x979/0xcc0
 call_timer_fn+0xf5/0x210
 __run_timers+0x792/0x970
 __do_softirq+0x372/0x783

The buggy address belongs to the object at ffff8880243a0000
 which belongs to the cache net_namespace of size 6656
The buggy address is located 504 bytes inside of
 6656-byte region [ffff8880243a0000, ffff8880243a1a00)

The buggy address belongs to the physical page:
page:ffffea000090e800 refcount:1 mapcount:0 mapping:0000000000000000
index:0x0 pfn:0x243a0
head:ffffea000090e800 order:3 compound_mapcount:0 compound_pincount:0
flags: 0xfff00000010200(slab|head|node=0|zone=1|lastcpupid=0x7ff)
raw: 00fff00000010200 ffffea00008af200 dead000000000002 ffff8880115c73c0
raw: 0000000000000000 0000000080040004 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected
page_owner tracks the page as allocated
page last allocated via order 3, migratetype Unmovable, gfp_mask
0xd20c0(__GFP_IO|__GFP_FS|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP|__GFP_NOMEMALLOC),
pid 6553, tgid 6553 (syz-executor), ts 38572126085, free_ts
38569018784
 get_page_from_freelist+0x708/0xab0
 __alloc_pages+0x26c/0x5f0
 alloc_slab_page+0x70/0xf0
 allocate_slab+0x5d/0x380
 ___slab_alloc+0x3de/0xc30
 kmem_cache_alloc+0x26d/0x300
 copy_net_ns+0x13c/0x5c0
 create_new_namespaces+0x4ce/0xa00
 unshare_nsproxy_namespaces+0x11e/0x180
 ksys_unshare+0x573/0xb60
 __x64_sys_unshare+0x34/0x40
 do_syscall_64+0x3d/0x90
 entry_SYSCALL_64_after_hwframe+0x44/0xae
page last free stack trace:
 free_pcp_prepare+0xd20/0xec0
 free_unref_page+0x7d/0x560
 __unfreeze_partials+0x1ab/0x200
 put_cpu_partial+0x106/0x170
 ___cache_free+0x10f/0x1b0
 kasan_quarantine_reduce+0x127/0x190
 __kasan_slab_alloc+0x2f/0xd0
 kmem_cache_alloc+0x1c0/0x300
 getname_flags+0xba/0x650
 do_sys_openat2+0xba/0x4e0
 __x64_sys_open+0x221/0x270
 do_syscall_64+0x3d/0x90
 entry_SYSCALL_64_after_hwframe+0x44/0xae

Memory state around the buggy address:
 ffff8880243a0080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
 ffff8880243a0100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>ffff8880243a0180: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
                                                                ^
 ffff8880243a0200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
 ffff8880243a0280: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================

Best,
Wei