Re: [RFC PATCH] jbd2: make '0' an invalid transaction sequence

["wangjianjian (C)" <wangjianjian3@huawei.com>]

On 2024/7/12 18:04, wangjianjian (C) wrote:
> On 2024/7/12 17:53, Luis Henriques wrote:
>> Since there's code (in fast-commit) that already handles a '0' tid as a
>> special case, it's better to ensure that jbd2 never sets it to that value
>> when journal->j_transaction_sequence increment wraps.
>>
>> Suggested-by: Andreas Dilger <adilger@dilger.ca>
>> Signed-off-by: Luis Henriques (SUSE) <luis.henriques@linux.dev>
>> ---
>>   fs/jbd2/transaction.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
>> index 66513c18ca29..4dbdd37349c3 100644
>> --- a/fs/jbd2/transaction.c
>> +++ b/fs/jbd2/transaction.c
>> @@ -84,6 +84,8 @@ static void jbd2_get_transaction(journal_t *journal,
>>       transaction->t_state = T_RUNNING;
>>       transaction->t_start_time = ktime_get();
>>       transaction->t_tid = journal->j_transaction_sequence++;
>> +    if (unlikely(transaction->t_tid == 0))
>> +        transaction->t_tid = journal->j_transaction_sequence++;
> Do we need add j_transaction_sequence again here? if tansanction->t_tid 
> == 0, then journal->j_trnasaction_sequence must be 1
still need add 1. Sorry for confuse.
> 
>>       transaction->t_expires = jiffies + journal->j_commit_interval;
>>       atomic_set(&transaction->t_updates, 0);
>>       atomic_set(&transaction->t_outstanding_credits,
-- 
Regards