Re: [PATCH v3 05/10] ext4: restart handle if credits are insufficient during allocating blocks

[Zhang Yi <yi.zhang@huaweicloud.com>]

On 2025/7/4 16:18, Jan Kara wrote:
> On Fri 04-07-25 09:40:23, Zhang Yi wrote:
>> On 2025/7/4 0:27, Jan Kara wrote:
>>> On Thu 03-07-25 10:13:07, Zhang Yi wrote:
>>>> On 2025/7/2 22:18, Jan Kara wrote:
>>>>> On Tue 01-07-25 21:06:30, Zhang Yi wrote:
>>>>>> From: Zhang Yi <yi.zhang@huawei.com>
>>>>>>
>>>>>> After large folios are supported on ext4, writing back a sufficiently
>>>>>> large and discontinuous folio may consume a significant number of
>>>>>> journal credits, placing considerable strain on the journal. For
>>>>>> example, in a 20GB filesystem with 1K block size and 1MB journal size,
>>>>>> writing back a 2MB folio could require thousands of credits in the
>>>>>> worst-case scenario (when each block is discontinuous and distributed
>>>>>> across different block groups), potentially exceeding the journal size.
>>>>>> This issue can also occur in ext4_write_begin() and ext4_page_mkwrite()
>>>>>> when delalloc is not enabled.
>>>>>>
>>>>>> Fix this by ensuring that there are sufficient journal credits before
>>>>>> allocating an extent in mpage_map_one_extent() and
>>>>>> ext4_block_write_begin(). If there are not enough credits, return
>>>>>> -EAGAIN, exit the current mapping loop, restart a new handle and a new
>>>>>> transaction, and allocating blocks on this folio again in the next
>>>>>> iteration.
>>>>>>
>>>>>> Suggested-by: Jan Kara <jack@suse.cz>
>>>>>> Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
>>>>>
>>>>> Very nice. Feel free to add:
>>>>>
>>>>> Reviewed-by: Jan Kara <jack@suse.cz>
>>>>>
>>>>> One small comment below:
>>>>>
>>>>>> +/*
>>>>>> + * Make sure that the current journal transaction has enough credits to map
>>>>>> + * one extent. Return -EAGAIN if it cannot extend the current running
>>>>>> + * transaction.
>>>>>> + */
>>>>>> +static inline int ext4_journal_ensure_extent_credits(handle_t *handle,
>>>>>> +						     struct inode *inode)
>>>>>> +{
>>>>>> +	int credits;
>>>>>> +	int ret;
>>>>>> +
>>>>>> +	if (!handle)
>>>>>
>>>>> Shouldn't this rather be ext4_handle_valid(handle) to catch nojournal mode
>>>>> properly?
>>>>>
>>>> __ext4_journal_ensure_credits() already calls ext4_handle_valid() to handle
>>>> nojournal mode, and the '!handle' check here is to handle the case where
>>>> ext4_block_write_begin() passes in a NULL 'handle'.
>>>
>>> Ah, right. But then you don't need the test at all, do you? Anyway,
>>> whatever you decide to do with this (or nothing) is fine by me.
>>>
>>
>> Yeah, remove this test is fine with me. I added this one is because the
>> comments in ext4_handle_valid() said "Do not use this for NULL handles."
>> I think it is best to follow this rule. :)
> 
> Right, I didn't read that comment :) So maybe the best fix will be just
> adding a comment before the test like:
> 
> 	/* Called from ext4_da_write_begin() which has no handle started? */
> 	if (!handle)
> 		return 0;
> 

Sure, it looks good, will do.

Thanks,
Yi.