KMSAN: uninit-value in f2fs_lookup

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit: c5a13b33 kmsan: clang-format core
git tree: https://github.com/google/kmsan.git master
console output: https://syzkaller.appspot.com/x/log.txt?x=14f5b19b900000
kernel config: https://syzkaller.appspot.com/x/.config?x=20f149ad694ba4be
dashboard link: https://syzkaller.appspot.com/bug?extid=0eac6f0bbd558fd866d7
compiler: clang version 10.0.0 (https://github.com/llvm/llvm-project/ c2443155a0fb245c8f17f2c1c72b6ea391e86e81)
userspace arch: i386

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

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+0eac6f...@syzkaller.appspotmail.com

=====================================================
BUG: KMSAN: uninit-value in f2fs_lookup+0xe05/0x1a80 fs/f2fs/namei.c:503
CPU: 0 PID: 20216 Comm: syz-executor.5 Not tainted 5.9.0-rc4-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0x21c/0x280 lib/dump_stack.c:118
kmsan_report+0xf7/0x1e0 mm/kmsan/kmsan_report.c:122
__msan_warning+0x58/0xa0 mm/kmsan/kmsan_instr.c:219
f2fs_lookup+0xe05/0x1a80 fs/f2fs/namei.c:503
lookup_open fs/namei.c:3082 [inline]
open_last_lookups fs/namei.c:3177 [inline]
path_openat+0x2729/0x6a90 fs/namei.c:3365
do_filp_open+0x2b8/0x710 fs/namei.c:3395
do_sys_openat2+0xa88/0x1140 fs/open.c:1168
do_sys_open fs/open.c:1184 [inline]
__do_compat_sys_openat fs/open.c:1242 [inline]
__se_compat_sys_openat+0x2a4/0x310 fs/open.c:1240
__ia32_compat_sys_openat+0x56/0x70 fs/open.c:1240
do_syscall_32_irqs_on arch/x86/entry/common.c:80 [inline]
__do_fast_syscall_32+0x129/0x180 arch/x86/entry/common.c:139
do_fast_syscall_32+0x6a/0xc0 arch/x86/entry/common.c:162
do_SYSENTER_32+0x73/0x90 arch/x86/entry/common.c:205
entry_SYSENTER_compat_after_hwframe+0x4d/0x5c
RIP: 0023:0xf7f73549
Code: b8 01 10 06 03 74 b4 01 10 07 03 74 b0 01 10 08 03 74 d8 01 00 00 00 00 00 00 00 00 00 00 00 00 00 51 52 55 89 e5 0f 34 cd 80 <5d> 5a 59 c3 90 90 90 90 eb 0d 90 90 90 90 90 90 90 90 90 90 90 90
RSP: 002b:00000000f554c0cc EFLAGS: 00000296 ORIG_RAX: 0000000000000127
RAX: ffffffffffffffda RBX: 00000000ffffff9c RCX: 0000000020000980
RDX: 000000000002f042 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000

Local variable ----page@f2fs_lookup created at:
f2fs_lookup+0x8f/0x1a80 fs/f2fs/namei.c:477
f2fs_lookup+0x8f/0x1a80 fs/f2fs/namei.c:477
=====================================================


---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

[Chao Yu]

Hi,

I don't see any problem here, thanks for your report. :)

Thanks,

> _______________________________________________
> Linux-f2fs-devel mailing list
> Linux-f2...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
> .
>

[Dan Carpenter]

On Fri, Sep 25, 2020 at 05:06:33PM +0800, Chao Yu wrote:
> Hi,
>
> I don't see any problem here, thanks for your report. :)
>

I bet the uninitialize value is because "max_depth" is zero.


352 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
353 const struct f2fs_filename *fname,
354 struct page **res_page)
^^^^^^^^
The stack trace says this isn't initialized.

355 {
356 unsigned long npages = dir_blocks(dir);
357 struct f2fs_dir_entry *de = NULL;
358 unsigned int max_depth;
359 unsigned int level;
360
361 if (f2fs_has_inline_dentry(dir)) {
362 *res_page = NULL;
363 de = f2fs_find_in_inline_dir(dir, fname, res_page);
364 goto out;
365 }
366
367 if (npages == 0) {
368 *res_page = NULL;
369 goto out;
370 }
371
372 max_depth = F2FS_I(dir)->i_current_depth;
373 if (unlikely(max_depth > MAX_DIR_HASH_DEPTH)) {
374 f2fs_warn(F2FS_I_SB(dir), "Corrupted max_depth of %lu: %u",
375 dir->i_ino, max_depth);
376 max_depth = MAX_DIR_HASH_DEPTH;
377 f2fs_i_depth_write(dir, max_depth);
378 }
379
380 for (level = 0; level < max_depth; level++) {
^^^^^^^^^^^^^^^^^
If "max_depth" is zero, then we never enter this loop.

381 *res_page = NULL;
382 de = find_in_level(dir, level, fname, res_page);
383 if (de || IS_ERR(*res_page))
384 break;
385 }
386 out:
387 /* This is to increase the speed of f2fs_create */
388 if (!de)
389 F2FS_I(dir)->task = current;
390 return de;

Which means that we return a NULL "de" and "*res_page" is uninitialized
and that matches what syzbot found throug runtime testing.

391 }

regards,
dan carpenter

[Chao Yu]

Hi Dan,

On 2020-9-25 18:57, Dan Carpenter wrote:
> On Fri, Sep 25, 2020 at 05:06:33PM +0800, Chao Yu wrote:
>> Hi,
>>
>> I don't see any problem here, thanks for your report. :)
>>
>
> I bet the uninitialize value is because "max_depth" is zero.

I agree with you, thanks for the hint. :)

Thanks,

[Eric Biggers]

On Fri, Sep 25, 2020 at 05:06:33PM +0800, Chao Yu wrote:

> Hi,
>
> I don't see any problem here, thanks for your report. :)
>
> Thanks,

What about if max_depth == 0 in __f2fs_find_entry()? Then __f2fs_find_entry()
would return NULL without initializing *res_page.

A fix could be:

diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
index 069f498af1e3..ceb4431b5669 100644
--- a/fs/f2fs/dir.c
+++ b/fs/f2fs/dir.c
@@ -357,16 +357,15 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
unsigned int max_depth;
unsigned int level;

+ *res_page = NULL;
+
if (f2fs_has_inline_dentry(dir)) {
- *res_page = NULL;

de = f2fs_find_in_inline_dir(dir, fname, res_page);

goto out;
}

- if (npages == 0) {
- *res_page = NULL;
+ if (npages == 0)
goto out;
- }

max_depth = F2FS_I(dir)->i_current_depth;
if (unlikely(max_depth > MAX_DIR_HASH_DEPTH)) {
@@ -377,7 +376,6 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,

}

for (level = 0; level < max_depth; level++) {

- *res_page = NULL;

de = find_in_level(dir, level, fname, res_page);

if (de || IS_ERR(*res_page))
break;

[Eric Biggers]

On Fri, Sep 25, 2020 at 09:38:19AM -0700, Eric Biggers wrote:
> On Fri, Sep 25, 2020 at 05:06:33PM +0800, Chao Yu wrote:
> > Hi,
> >
> > I don't see any problem here, thanks for your report. :)
> >
> > Thanks,
>
> What about if max_depth == 0 in __f2fs_find_entry()? Then __f2fs_find_entry()
> would return NULL without initializing *res_page.

... and I now see Dan Carpenter already pointed this out. I was a bit late!

- Eric

[Chao Yu]

Thanks for your check as well. :)

Thanks,

>
> - Eric
>