Bug 204135 - F2FS: BUG: KASAN: null-ptr-deref in mempool_alloc+0xd8/0x2e0
Summary: F2FS: BUG: KASAN: null-ptr-deref in mempool_alloc+0xd8/0x2e0
Status: RESOLVED CODE_FIX
Alias: None
Product: File System
Classification: Unclassified
Component: f2fs (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: Default virtual assignee for f2fs
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-07-11 07:18 UTC by midwinter1993
Modified: 2019-07-30 06:25 UTC (History)
1 user (show)

See Also:
Kernel Version: 5.1.3
Subsystem:
Regression: No
Bisected commit-id:


Attachments

Description midwinter1993 2019-07-11 07:18:06 UTC
The following script triggers a KASAN null pointer dereference in f2fs file system.

I tested under kernel 5.1.3 using Qemu; this bug may not happen every time but according to my tries, it almost happens deterministically.

```
#!/bin/bash

MOUNT_DIR=/root/mnt
DISK=xx.img

dd if=/dev/zero of=$DISK bs=1M count=192
mkfs.f2fs -a 0 -o 19 -t 1 -z 3 -f -q $DISK


mkdir -pv $MOUNT_DIR

mount $DISK $MOUNT_DIR -o "noextent_cache"

mkdir $MOUNT_DIR/Umfoo_28

mount $DISK $MOUNT_DIR -o "remount,noextent_cache,mode=lfs,io_bits=7" # BUG OCCURS

umount $MOUNT_DIR
```


--- Core dump ---
[   35.256703] ==================================================================
[   35.257772] BUG: KASAN: null-ptr-deref in mempool_alloc+0xd8/0x2e0
[   35.258646] Read of size 8 at addr 0000000000000020 by task umount/2161
[   35.259591] 
[   35.259833] CPU: 0 PID: 2161 Comm: umount Not tainted 5.1.3 #8
[   35.260646] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
[   35.261953] Call Trace:
[   35.262315]  dump_stack+0x8a/0xce
[   35.262787]  ? mempool_alloc+0xd8/0x2e0
[   35.263330]  ? mempool_alloc+0xd8/0x2e0
[   35.263819]  kasan_report.cold+0x5/0x32
[   35.264199]  ? mempool_alloc+0xd8/0x2e0
[   35.264678]  mempool_alloc+0xd8/0x2e0
[   35.265197]  ? mempool_resize+0x410/0x410
[   35.265760]  ? blk_init_request_from_bio+0xd0/0xd0
[   35.266439]  ? __write_data_page+0x4c9/0x1140
[   35.267063]  __submit_merged_bio+0x519/0x8c0
[   35.267671]  __submit_merged_write_cond+0x41a/0x520
[   35.268373]  f2fs_write_cache_pages+0xb08/0xb40
[   35.269021]  ? __write_data_page+0x1140/0x1140
[   35.269658]  ? update_group_capacity+0x340/0x340
[   35.270320]  f2fs_write_data_pages+0x796/0x8b0
[   35.270962]  ? f2fs_write_cache_pages+0xb40/0xb40
[   35.271631]  ? deref_stack_reg+0x84/0xd0
[   35.272195]  ? __read_once_size_nocheck.constprop.0+0x10/0x10
[   35.273005]  ? __orc_find+0x63/0xc0
[   35.273499]  ? unwind_next_frame+0x8a8/0x9b0
[   35.274104]  ? f2fs_write_cache_pages+0xb40/0xb40
[   35.274767]  do_writepages+0xbb/0x1e0
[   35.275287]  ? page_writeback_cpu_online+0x10/0x10
[   35.275970]  ? __orc_find+0x63/0xc0
[   35.276464]  ? stack_access_ok+0x35/0x90
[   35.277039]  ? unwind_next_frame+0x8a8/0x9b0
[   35.277644]  ? __update_load_avg_cfs_rq+0x2f/0x4a0
[   35.278340]  ? rb_next+0x33/0x80
[   35.278800]  __filemap_fdatawrite_range+0x142/0x1d0
[   35.279488]  ? delete_from_page_cache_batch+0x650/0x650
[   35.280239]  ? _raw_spin_lock_irq+0xd0/0xd0
[   35.280833]  f2fs_sync_dirty_inodes+0x167/0x490
[   35.281475]  f2fs_write_checkpoint+0x462/0x2f70
[   35.282115]  ? __schedule+0x481/0xc80
[   35.282633]  ? update_cfs_group+0x1d/0x1a0
[   35.283213]  ? __enqueue_entity+0xae/0xe0
[   35.283786]  ? pci_mmcfg_check_reserved+0x130/0x130
[   35.284478]  ? _raw_spin_lock_irq+0x76/0xd0
[   35.285073]  ? f2fs_wait_on_all_pages_writeback+0x1a0/0x1a0
[   35.285859]  ? try_to_wake_up+0xa4/0x750
[   35.286416]  ? bit_wait_io_timeout+0xc0/0xc0
[   35.287021]  ? __migrate_task.isra.0+0xa0/0xa0
[   35.287650]  ? pin_kill+0x121/0x260
[   35.288152]  ? pin_insert+0x30/0x30
[   35.288648]  ? kthread_stop+0x11b/0x260
[   35.289191]  kill_f2fs_super+0x1e7/0x210
[   35.289746]  ? __f2fs_commit_super+0xc0/0xc0
[   35.290347]  ? xas_start+0xbf/0x1c0
[   35.290846]  ? unregister_shrinker+0x9e/0xc0
[   35.291446]  ? kfree+0x9b/0x1c0
[   35.291899]  deactivate_locked_super+0x69/0xc0
[   35.292525]  deactivate_super+0x14e/0x160
[   35.293094]  ? iterate_supers_type+0x1c0/0x1c0
[   35.293723]  cleanup_mnt+0x71/0xc0
[   35.294207]  task_work_run+0x100/0x120
[   35.294663]  exit_to_usermode_loop+0x175/0x180
[   35.295149]  do_syscall_64+0x1f1/0x230
[   35.295560]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[   35.296114] RIP: 0033:0x7fe8e01f6d77
[   35.296509] Code: 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 31 f6 e9 09 00 00 00 66 0f 1f 84 00 00 00 00 00 b8 a6 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d f1 00 2b 00 f7 d8 64 89 01 48
[   35.298601] RSP: 002b:00007ffff47cec68 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6
[   35.299401] RAX: 0000000000000000 RBX: 000055f947ec7080 RCX: 00007fe8e01f6d77
[   35.300151] RDX: 0000000000000001 RSI: 0000000000000000 RDI: 000055f947ec7260
[   35.300950] RBP: 000055f947ec7260 R08: 000055f947ec7280 R09: 0000000000000014
[   35.301707] R10: 00000000000006b4 R11: 0000000000000246 R12: 00007fe8e06f8e64
[   35.302455] R13: 0000000000000000 R14: 0000000000000000 R15: 00007ffff47ceef0
[   35.303258] ==================================================================
[   35.304025] Disabling lock debugging due to kernel taint
[   35.305229] BUG: unable to handle kernel NULL pointer dereference at 0000000000000020
[   35.306356] #PF error: [normal kernel read fault]
[   35.307004] PGD 0 P4D 0 
[   35.307364] Oops: 0000 [#1] SMP KASAN PTI
[   35.307914] CPU: 1 PID: 2161 Comm: umount Tainted: G    B             5.1.3 #8
[   35.308879] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
[   35.310142] RIP: 0010:mempool_alloc+0xd8/0x2e0
[   35.310749] Code: 81 cb 00 20 09 00 4d 8d 74 24 18 89 44 24 08 65 48 8b 04 25 40 dd 01 00 48 89 44 24 10 e8 20 d9 f5 ff 4c 89 ff e8 d8 14 0c 00 <49> 8b 44 24 20 4c 89 f7 48 89 04 24 e8 c7 14 0c 00 49 8b 74 24 18
[   35.313330] RSP: 0018:ffff8881129af3a0 EFLAGS: 00010286
[   35.314034] RAX: 0000000000000000 RBX: 000000000009a800 RCX: ffffffff9694c53d
[   35.314959] RDX: 1ffffffff30c1557 RSI: 0000000000000246 RDI: ffffffff9860aab8
[   35.315914] RBP: 1ffff11022535e79 R08: 000000000000002c R09: ffffed1022e03c9b
[   35.316865] R10: ffffed1022e03c9a R11: ffff88811701e4d7 R12: 0000000000000000
[   35.317811] R13: 0000000000000400 R14: 0000000000000018 R15: 0000000000000020
[   35.318759] FS:  00007fe8e0914e40(0000) GS:ffff888117100000(0000) knlGS:0000000000000000
[   35.319845] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   35.320620] CR2: 00007f2f50033048 CR3: 00000001137d6000 CR4: 00000000000006e0
[   35.321568] Call Trace:
[   35.321914]  ? mempool_resize+0x410/0x410
[   35.322460]  ? blk_init_request_from_bio+0xd0/0xd0
[   35.323108]  ? __write_data_page+0x4c9/0x1140
[   35.323701]  __submit_merged_bio+0x519/0x8c0
[   35.324287]  __submit_merged_write_cond+0x41a/0x520
[   35.325004]  f2fs_write_cache_pages+0xb08/0xb40
[   35.325621]  ? __write_data_page+0x1140/0x1140
[   35.326226]  ? update_group_capacity+0x340/0x340
[   35.326853]  f2fs_write_data_pages+0x796/0x8b0
[   35.327459]  ? f2fs_write_cache_pages+0xb40/0xb40
[   35.328101]  ? deref_stack_reg+0x84/0xd0
[   35.328635]  ? __read_once_size_nocheck.constprop.0+0x10/0x10
[   35.329404]  ? __orc_find+0x63/0xc0
[   35.329879]  ? unwind_next_frame+0x8a8/0x9b0
[   35.330458]  ? f2fs_write_cache_pages+0xb40/0xb40
[   35.331092]  do_writepages+0xbb/0x1e0
[   35.331592]  ? page_writeback_cpu_online+0x10/0x10
[   35.332249]  ? __orc_find+0x63/0xc0
[   35.332730]  ? stack_access_ok+0x35/0x90
[   35.333270]  ? unwind_next_frame+0x8a8/0x9b0
[   35.333855]  ? __update_load_avg_cfs_rq+0x2f/0x4a0
[   35.334508]  ? rb_next+0x33/0x80
[   35.334958]  __filemap_fdatawrite_range+0x142/0x1d0
[   35.335622]  ? delete_from_page_cache_batch+0x650/0x650
[   35.336330]  ? _raw_spin_lock_irq+0xd0/0xd0
[   35.336898]  f2fs_sync_dirty_inodes+0x167/0x490
[   35.337509]  f2fs_write_checkpoint+0x462/0x2f70
[   35.338123]  ? __schedule+0x481/0xc80
[   35.338622]  ? update_cfs_group+0x1d/0x1a0
[   35.339176]  ? __enqueue_entity+0xae/0xe0
[   35.339721]  ? pci_mmcfg_check_reserved+0x130/0x130
[   35.340390]  ? _raw_spin_lock_irq+0x76/0xd0
[   35.340967]  ? f2fs_wait_on_all_pages_writeback+0x1a0/0x1a0
[   35.341726]  ? try_to_wake_up+0xa4/0x750
[   35.342267]  ? bit_wait_io_timeout+0xc0/0xc0
[   35.342854]  ? __migrate_task.isra.0+0xa0/0xa0
[   35.343465]  ? pin_kill+0x121/0x260
[   35.343953]  ? pin_insert+0x30/0x30
[   35.344438]  ? kthread_stop+0x11b/0x260
[   35.344967]  kill_f2fs_super+0x1e7/0x210
[   35.345509]  ? __f2fs_commit_super+0xc0/0xc0
[   35.346095]  ? xas_start+0xbf/0x1c0
[   35.346579]  ? unregister_shrinker+0x9e/0xc0
[   35.347165]  ? kfree+0x9b/0x1c0
[   35.347601]  deactivate_locked_super+0x69/0xc0
[   35.348189]  deactivate_super+0x14e/0x160
[   35.348740]  ? iterate_supers_type+0x1c0/0x1c0
[   35.349351]  cleanup_mnt+0x71/0xc0
[   35.349824]  task_work_run+0x100/0x120
[   35.350342]  exit_to_usermode_loop+0x175/0x180
[   35.350954]  do_syscall_64+0x1f1/0x230
[   35.351472]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[   35.352165] RIP: 0033:0x7fe8e01f6d77
[   35.352659] Code: 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 31 f6 e9 09 00 00 00 66 0f 1f 84 00 00 00 00 00 b8 a6 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d f1 00 2b 00 f7 d8 64 89 01 48
[   35.355180] RSP: 002b:00007ffff47cec68 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6
[   35.356208] RAX: 0000000000000000 RBX: 000055f947ec7080 RCX: 00007fe8e01f6d77
[   35.357171] RDX: 0000000000000001 RSI: 0000000000000000 RDI: 000055f947ec7260
[   35.358134] RBP: 000055f947ec7260 R08: 000055f947ec7280 R09: 0000000000000014
[   35.359097] R10: 00000000000006b4 R11: 0000000000000246 R12: 00007fe8e06f8e64
[   35.360062] R13: 0000000000000000 R14: 0000000000000000 R15: 00007ffff47ceef0
[   35.361021] Modules linked in:
[   35.361446] Dumping ftrace buffer:
[   35.361914]    (ftrace buffer empty)
[   35.362406] CR2: 0000000000000020
[   35.363006] ---[ end trace d88b774adc80a9fc ]---
Comment 1 Chao Yu 2019-07-16 09:09:15 UTC
I've sent one patch to fix this issue, just simply refusing remount w/ io_bits change, could you have a try?

https://lore.kernel.org/linux-f2fs-devel/20190712085700.4239-1-yuchao0@huawei.com/T/#u
Comment 2 midwinter1993 2019-07-17 01:51:22 UTC
(In reply to Chao Yu from comment #1)
> I've sent one patch to fix this issue, just simply refusing remount w/
> io_bits change, could you have a try?
> 
> https://lore.kernel.org/linux-f2fs-devel/20190712085700.4239-1-
> yuchao0@huawei.com/T/#u

I have tested this patch but have compiling errors:

> /linux-5.1.3/fs/f2fs/super.c:1538:22: error: implicit declaration of function F2FS_IO_ALIGNED; did you mean F2FS_BLK_ALIGN? [-Werror=implicit-function-declaration]

> /linux-5.1.3/fs/f2fs/super.c:1621:3: error: implicit declaration of function f2fs_warn [-Werror=implicit-function-declaration]

I "grep" the whole file system directory but not found the symbol "F2FS_IO_ALIGNED" and "f2fs_warn".

Should I change the kernel version to test?
Comment 3 Chao Yu 2019-07-17 01:58:40 UTC
F2FS_IO_ALIGNED() was defined in last patch for your previous issue... so, should apply them before this patch.

Could you change to last code in dev branch in Jaegeuk's git repo.

https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git/log/?h=dev
Comment 4 midwinter1993 2019-07-22 03:38:43 UTC
(In reply to Chao Yu from comment #3)
> F2FS_IO_ALIGNED() was defined in last patch for your previous issue... so,
> should apply them before this patch.
> 
> Could you change to last code in dev branch in Jaegeuk's git repo.
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git/log/?h=dev

Hi! I tested the patch under commit:
> Hash: e3110f4; f2fs: fix to read source block before invalidating it

But the F2FS_IO_ALIGNED was still missing!? (Or something wrong I did?)
Comment 5 Chao Yu 2019-07-22 03:49:23 UTC
Sorry, I meant I made the two fix patches based Jaegeuk's git repo, however Jaegeuk hasn't merge the two patches in his branch.

So you can do the merge by yourself, or you can use below codes in my repo, I've update the all fixes on it.

https://git.kernel.org/pub/scm/linux/kernel/git/chao/linux.git/log/?h=dev
Comment 6 midwinter1993 2019-07-30 03:56:29 UTC
(In reply to Chao Yu from comment #5)
> Sorry, I meant I made the two fix patches based Jaegeuk's git repo, however
> Jaegeuk hasn't merge the two patches in his branch.
> 
> So you can do the merge by yourself, or you can use below codes in my repo,
> I've update the all fixes on it.
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/chao/linux.git/log/?h=dev

I have tested several times in your repo and under dev branch; the above-mentioned bug did not happen anymore. :-P
Comment 7 Chao Yu 2019-07-30 06:25:55 UTC
Thanks for the test, let me close this track. :)

Note You need to log in before you can comment on or make changes to this bug.