Re: [Linux Kernel Bug] UBSAN: shift-out-of-bounds in fault_around_bytes_set

[Kefeng Wang]

On 2024/3/1 15:42, Sam Sun wrote:
> Dear developers and maintainers,
>
> We found a shift-out-of-bounds bug in mm/memory.c. Kernel commit is b401b621758.
> Kernel config and C repro are attached to this email.
> UBSAN report is listed below.
> ```
> UBSAN: shift-out-of-bounds in /home/sy/linux-original/include/linux/log2.h:67:13
> shift exponent 4294967295 is too large for 64-bit type 'long unsigned int'
> CPU: 0 PID: 8091 Comm: syz-executor371 Not tainted 6.7.0-rc7 #1
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014
> Call Trace:
>   <TASK>
>   __dump_stack lib/dump_stack.c:88 [inline]
>   dump_stack_lvl+0x136/0x150 lib/dump_stack.c:106
>   ubsan_epilogue lib/ubsan.c:217 [inline]
>   __ubsan_handle_shift_out_of_bounds+0x24b/0x430 lib/ubsan.c:387
>   __rounddown_pow_of_two include/linux/log2.h:67 [inline]
>   fault_around_bytes_set.cold+0x19/0x1e mm/memory.c:4527
>   simple_attr_write_xsigned.constprop.0.isra.0+0x1ed/0x2d0 fs/libfs.c:1301
>   debugfs_attr_write_xsigned fs/debugfs/file.c:485 [inline]
>   debugfs_attr_write+0x74/0xa0 fs/debugfs/file.c:493
>   vfs_write+0x2a9/0xd80 fs/read_write.c:582
>   ksys_write+0x122/0x250 fs/read_write.c:637
>   do_syscall_x64 arch/x86/entry/common.c:52 [inline]
>   do_syscall_64+0x40/0x110 arch/x86/entry/common.c:83
>   entry_SYSCALL_64_after_hwframe+0x63/0x6b
> RIP: 0033:0x7fa30d5d7fcd
> Code: 28 c3 e8 46 1e 00 00 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48
> 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 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:00007ffc8b7ee1b8 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
> RAX: ffffffffffffffda RBX: 00007ffc8b7ee3b8 RCX: 00007fa30d5d7fcd
> RDX: 0000000000000002 RSI: 0000000020000040 RDI: 0000000000000003
> RBP: 0000000000000001 R08: 0000000000000000 R09: 00007ffc8b7ee3b8
> R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000001
> R13: 00007ffc8b7ee3a8 R14: 00007fa30d655530 R15: 0000000000000001
>   </TASK>
> ================================================================================
> ```
> In function simple_attr_write_xsigned, a user controlled string "buf"
> is copied and
> turned to long type by function "kstrtoll". If buf is "0", val passed
> to function
> fault_around_bytes_set is 0, which would trigger shift-out-of-bound bug.

Look like commit 53d36a56d8c4 ("mm: prefer fault_around_pages to 
fault_around_bytes") introduces the issue, please try the following change,

diff --git a/mm/memory.c b/mm/memory.c
index abd4f33d62c9..e17669d4f72f 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -4776,7 +4776,8 @@ static int fault_around_bytes_set(void *data, u64 val)
         * The minimum value is 1 page, however this results in no 
fault-around
         * at all. See should_fault_around().
         */
-       fault_around_pages = max(rounddown_pow_of_two(val) >> 
PAGE_SHIFT, 1UL);
+       val = max(val, PAGE_SIZE);
+       fault_around_pages = rounddown_pow_of_two(val) >> PAGE_SHIFT;

        return 0;
 }



> If you have any questions, please contact us.
> Reported by Yue Sun <samsun1006219@xxxxxxxxx>
>
> Best Regards,
> Yue