Re: f2fs: Possible use-after-free when umount filesystem

[Gu Zheng <guz.fnst@cn.fujitsu.com>]

Hi Andrey,

On 07/21/2014 06:56 PM, Andrey Tsyvarev wrote:

> Hello,
> 
> Using memory error detector reveals the following use-after-free error in 3.15.0:
> 
> AddressSanitizer: heap-use-after-free in f2fs_evict_inode
> Read of size 8 by thread T22279:
>  [<ffffffffa02d8702>] f2fs_evict_inode+0x102/0x2e0 [f2fs] /home/tester/linux-sources/linux-kasan/fs/f2fs/f2fs.h:584
>  [<ffffffff812359af>] evict+0x15f/0x290 /home/tester/linux-sources/linux-kasan/fs/inode.c:550
>  [<     inlined    >] iput+0x196/0x280 iput_final /home/tester/linux-sources/linux-kasan/fs/inode.c:1418
>  [<ffffffff812369a6>] iput+0x196/0x280 /home/tester/linux-sources/linux-kasan/fs/inode.c:1436
>  [<ffffffffa02dc416>] f2fs_put_super+0xd6/0x170 [f2fs] /home/tester/linux-sources/linux-kasan/fs/f2fs/super.c:434
>  [<ffffffff81210095>] generic_shutdown_super+0xc5/0x1b0 /home/tester/linux-sources/linux-kasan/fs/super.c:406
>  [<ffffffff812105fd>] kill_block_super+0x4d/0xb0 /home/tester/linux-sources/linux-kasan/fs/super.c:1019
>  [<ffffffff81210a86>] deactivate_locked_super+0x66/0x80 /home/tester/linux-sources/linux-kasan/fs/super.c:284
>  [<ffffffff81211c98>] deactivate_super+0x68/0x80 /home/tester/linux-sources/linux-kasan/fs/super.c:307
>  [<ffffffff8123cc88>] mntput_no_expire+0x198/0x250 /home/tester/linux-sources/linux-kasan/fs/namespace.c:986 (discriminator 3)
>  [<     inlined    >] SyS_umount+0xe9/0x1a0 SYSC_umount /home/tester/linux-sources/linux-kasan/fs/namespace.c:1424
>  [<ffffffff8123f1c9>] SyS_umount+0xe9/0x1a0 /home/tester/linux-sources/linux-kasan/fs/namespace.c:1392
>  [<ffffffff81cc8df9>] system_call_fastpath+0x16/0x1b /home/tester/linux-sources/linux-kasan/arch/x86/kernel/entry_64.S:426
> 
> Freed by thread T3:
>  [<ffffffffa02dc337>] f2fs_i_callback+0x27/0x30 [f2fs] /home/tester/linux-sources/linux-kasan/fs/f2fs/super.c:408
>  [<     inlined    >] rcu_process_callbacks+0x2d6/0x930 __rcu_reclaim /home/tester/linux-sources/linux-kasan/kernel/rcu/rcu.h:114
>  [<     inlined    >] rcu_process_callbacks+0x2d6/0x930 rcu_do_batch /home/tester/linux-sources/linux-kasan/kernel/rcu/tree.c:2242
>  [<     inlined    >] rcu_process_callbacks+0x2d6/0x930 invoke_rcu_callbacks /home/tester/linux-sources/linux-kasan/kernel/rcu/tree.c:2499
>  [<     inlined    >] rcu_process_callbacks+0x2d6/0x930 __rcu_process_callbacks /home/tester/linux-sources/linux-kasan/kernel/rcu/tree.c:2466
>  [<ffffffff810fd266>] rcu_process_callbacks+0x2d6/0x930 /home/tester/linux-sources/linux-kasan/kernel/rcu/tree.c:2483
>  [<ffffffff8107cce2>] __do_softirq+0x142/0x380 /home/tester/linux-sources/linux-kasan/kernel/softirq.c:269
>  [<ffffffff8107cf50>] run_ksoftirqd+0x30/0x50 /home/tester/linux-sources/linux-kasan/kernel/softirq.c:658
>  [<ffffffff810b2a87>] smpboot_thread_fn+0x197/0x280 /home/tester/linux-sources/linux-kasan/kernel/smpboot.c:160
>  [<ffffffff810a8238>] kthread+0x148/0x160 /home/tester/linux-sources/linux-kasan/kernel/kthread.c:207
>  [<ffffffff81cc8d4c>] ret_from_fork+0x7c/0xb0 /home/tester/linux-sources/linux-kasan/arch/x86/kernel/entry_64.S:351
> 
> Allocated by thread T22276:
>  [<ffffffffa02dc7dd>] f2fs_alloc_inode+0x2d/0x170 [f2fs] /home/tester/linux-sources/linux-kasan/fs/f2fs/super.c:356
>  [<ffffffff8123471d>] alloc_inode+0x2d/0xe0 /home/tester/linux-sources/linux-kasan/fs/inode.c:208
>  [<ffffffff81235e2a>] iget_locked+0x10a/0x230 /home/tester/linux-sources/linux-kasan/fs/inode.c:1085
>  [<ffffffffa02d7495>] f2fs_iget+0x35/0xa80 [f2fs] /home/tester/linux-sources/linux-kasan/fs/f2fs/inode.c:129
>  [<ffffffffa02e2393>] f2fs_fill_super+0xb53/0xff0 [f2fs] /home/tester/linux-sources/linux-kasan/fs/f2fs/super.c:1021
>  [<ffffffff81211bce>] mount_bdev+0x1de/0x240 /home/tester/linux-sources/linux-kasan/fs/super.c:992
>  [<ffffffffa02dbce0>] f2fs_mount+0x10/0x20 [f2fs] /home/tester/linux-sources/linux-kasan/fs/f2fs/super.c:1127
>  [<ffffffff81212a85>] mount_fs+0x55/0x220 /home/tester/linux-sources/linux-kasan/fs/super.c:1095
>  [<ffffffff8123c026>] vfs_kern_mount+0x66/0x200 /home/tester/linux-sources/linux-kasan/fs/namespace.c:851
>  [<     inlined    >] do_mount+0x2b4/0x1120 do_new_mount /home/tester/linux-sources/linux-kasan/fs/namespace.c:2129
>  [<ffffffff812400d4>] do_mount+0x2b4/0x1120 /home/tester/linux-sources/linux-kasan/fs/namespace.c:2453
>  [<     inlined    >] SyS_mount+0xb2/0x110 SYSC_mount /home/tester/linux-sources/linux-kasan/fs/namespace.c:2647
>  [<ffffffff812414a2>] SyS_mount+0xb2/0x110 /home/tester/linux-sources/linux-kasan/fs/namespace.c:2620
>  [<ffffffff81cc8df9>] system_call_fastpath+0x16/0x1b /home/tester/linux-sources/linux-kasan/arch/x86/kernel/entry_64.S:426
> 
> The buggy address ffff8800587866c8 is located 48 bytes inside
>  of 680-byte region [ffff880058786698, ffff880058786940)
> 
> Memory state around the buggy address:
>  ffff880058786100: ffffffff ffffffff ffffffff ffffffff
>  ffff880058786200: ffffffff ffffffff ffffffrr rrrrrrrr
>  ffff880058786300: rrrrrrrr rrffffff ffffffff ffffffff
>  ffff880058786400: ffffffff ffffffff ffffffff ffffffff
>  ffff880058786500: ffffffff ffffffff ffffffff fffffffr
>>ffff880058786600: rrrrrrrr rrrrrrrr rrrfffff ffffffff
>                                                ^
>  ffff880058786700: ffffffff ffffffff ffffffff ffffffff
>  ffff880058786800: ffffffff ffffffff ffffffff ffffffff
>  ffff880058786900: ffffffff rrrrrrrr rrrrrrrr rrrr....
>  ffff880058786a00: ........ ........ ........ ........
>  ffff880058786b00: ........ ........ ........ ........
> Legend:
>  f - 8 freed bytes
>  r - 8 redzone bytes
>  . - 8 allocated bytes
>  x=1..7 - x allocated bytes + (8-x) redzone bytes
> 
> 
> Investigation shows, that f2fs_evict_inode, when called for 'meta_inode', uses invalidate_mapping_pages() for 'node_inode'.
> But 'node_inode' is deleted before 'meta_inode' in f2fs_put_super via iput().
> 
> It seems that in common usage scenario this use-after-free is benign, because 'node_inode' remains partially valid data even after kmem_cache_free().
> But things may change if, while 'meta_inode' is evicted in one f2fs filesystem, another (mounted) f2fs filesystem requests inode from cache, and formely
> 'node_inode' of the first filesystem is returned.

The analysis seems reasonable. Have you tried to swap the reclaim order of node_inde
and meta_inode?

diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
index 870fe19..e114418 100644
--- a/fs/f2fs/super.c
+++ b/fs/f2fs/super.c
@@ -430,8 +430,8 @@ static void f2fs_put_super(struct super_block *sb)
        if (sbi->s_dirty && get_pages(sbi, F2FS_DIRTY_NODES))
                write_checkpoint(sbi, true);
 
-       iput(sbi->node_inode);
        iput(sbi->meta_inode);
+       iput(sbi->node_inode);
 
        /* destroy f2fs internal modules */
        destroy_node_manager(sbi);

Thanks,
Gu

> 
> 
> Found by Linux File System Verification project (linuxtesting.org).
> 
> 



------------------------------------------------------------------------------
Want fast and easy access to all the code in your enterprise? Index and
search up to 200,000 lines of code with a free copy of Black Duck
Code Sight - the same software that powers the world's largest code
search on Ohloh, the Black Duck Open Hub! Try it now.
http://p.sf.net/sfu/bds