Re: [PATCH] ext4: fix uninitialized ratelimit_state->lock access in __ext4_fill_super()

[Baokun Li <libaokun@huaweicloud.com>]

Hi Ted,

Would you consider merging in this patch in the current merge
window? This patch has been merged in our internal branch for
a while now and so far it is safe.

Regards,
Baokun

On 2024/4/23 9:41, Baokun Li wrote:
> Gently ping again.
>
> On 2024/1/2 21:37, Baokun Li wrote:
>> In the following concurrency we will access the uninitialized rs->lock:
>>
>> ext4_fill_super
>>    ext4_register_sysfs
>>     // sysfs registered msg_ratelimit_interval_ms
>>                               // Other processes modify rs->interval to
>>                               // non-zero via msg_ratelimit_interval_ms
>>    ext4_orphan_cleanup
>>      ext4_msg(sb, KERN_INFO, "Errors on filesystem, "
>>        __ext4_msg
>> ___ratelimit(&(EXT4_SB(sb)->s_msg_ratelimit_state)
>>            if (!rs->interval)  // do nothing if interval is 0
>>              return 1;
>>            raw_spin_trylock_irqsave(&rs->lock, flags)
>>              raw_spin_trylock(lock)
>>                _raw_spin_trylock
>>                  __raw_spin_trylock
>>                    spin_acquire(&lock->dep_map, 0, 1, _RET_IP_)
>>                      lock_acquire
>>                        __lock_acquire
>>                          register_lock_class
>>                            assign_lock_key
>>                              dump_stack();
>>    ratelimit_state_init(&sbi->s_msg_ratelimit_state, 5 * HZ, 10);
>>      raw_spin_lock_init(&rs->lock);
>>      // init rs->lock here
>>
>> and get the following dump_stack:
>>
>> =========================================================
>> INFO: trying to register non-static key.
>> The code is fine but needs lockdep annotation, or maybe
>> you didn't initialize this object before use?
>> turning off the locking correctness validator.
>> CPU: 12 PID: 753 Comm: mount Tainted: G E 6.7.0-rc6-next-20231222 #504
>> [...]
>> Call Trace:
>>   dump_stack_lvl+0xc5/0x170
>>   dump_stack+0x18/0x30
>>   register_lock_class+0x740/0x7c0
>>   __lock_acquire+0x69/0x13a0
>>   lock_acquire+0x120/0x450
>>   _raw_spin_trylock+0x98/0xd0
>>   ___ratelimit+0xf6/0x220
>>   __ext4_msg+0x7f/0x160 [ext4]
>>   ext4_orphan_cleanup+0x665/0x740 [ext4]
>>   __ext4_fill_super+0x21ea/0x2b10 [ext4]
>>   ext4_fill_super+0x14d/0x360 [ext4]
>> [...]
>> =========================================================
>>
>> Normally interval is 0 until s_msg_ratelimit_state is initialized, so
>> ___ratelimit() does nothing. But registering sysfs precedes initializing
>> rs->lock, so it is possible to change rs->interval to a non-zero value
>> via the msg_ratelimit_interval_ms interface of sysfs while rs->lock is
>> uninitialized, and then a call to ext4_msg triggers the problem by
>> accessing an uninitialized rs->lock. Therefore register sysfs after all
>> initializations are complete to avoid such problems.
>>
>> Signed-off-by: Baokun Li <libaokun1@huawei.com>
>> ---
>>   fs/ext4/super.c | 22 ++++++++++------------
>>   1 file changed, 10 insertions(+), 12 deletions(-)
>>
>> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
>> index 0980845c8b8f..1db23b0e8a4f 100644
>> --- a/fs/ext4/super.c
>> +++ b/fs/ext4/super.c
>> @@ -5564,19 +5564,15 @@ static int __ext4_fill_super(struct 
>> fs_context *fc, struct super_block *sb)
>>       if (err)
>>           goto failed_mount6;
>>   -    err = ext4_register_sysfs(sb);
>> -    if (err)
>> -        goto failed_mount7;
>> -
>>       err = ext4_init_orphan_info(sb);
>>       if (err)
>> -        goto failed_mount8;
>> +        goto failed_mount7;
>>   #ifdef CONFIG_QUOTA
>>       /* Enable quota usage during mount. */
>>       if (ext4_has_feature_quota(sb) && !sb_rdonly(sb)) {
>>           err = ext4_enable_quotas(sb);
>>           if (err)
>> -            goto failed_mount9;
>> +            goto failed_mount8;
>>       }
>>   #endif  /* CONFIG_QUOTA */
>>   @@ -5602,7 +5598,7 @@ static int __ext4_fill_super(struct 
>> fs_context *fc, struct super_block *sb)
>>           ext4_msg(sb, KERN_INFO, "recovery complete");
>>           err = ext4_mark_recovery_complete(sb, es);
>>           if (err)
>> -            goto failed_mount10;
>> +            goto failed_mount9;
>>       }
>>         if (test_opt(sb, DISCARD) && 
>> !bdev_max_discard_sectors(sb->s_bdev))
>> @@ -5619,15 +5615,17 @@ static int __ext4_fill_super(struct 
>> fs_context *fc, struct super_block *sb)
>>       atomic_set(&sbi->s_warning_count, 0);
>>       atomic_set(&sbi->s_msg_count, 0);
>>   +    /* Register sysfs after all initializations are complete. */
>> +    err = ext4_register_sysfs(sb);
>> +    if (err)
>> +        goto failed_mount9;
>> +
>>       return 0;
>>   -failed_mount10:
>> +failed_mount9:
>>       ext4_quotas_off(sb, EXT4_MAXQUOTAS);
>> -failed_mount9: __maybe_unused
>> +failed_mount8: __maybe_unused
>>       ext4_release_orphan_info(sb);
>> -failed_mount8:
>> -    ext4_unregister_sysfs(sb);
>> -    kobject_put(&sbi->s_kobj);
>>   failed_mount7:
>>       ext4_unregister_li_request(sb);
>>   failed_mount6:
>