Re: [PATCH v2 3/4] exfat: dirty all new pages when extending valid_size

[Chi Zhiling <chizhiling@163.com>]

On 10/5/26 5:08 PM, Namjae Jeon wrote:
>> @@ -664,90 +664,30 @@ int exfat_file_fsync(struct file *filp, loff_t start, loff_t end, int datasync)
>>   static int exfat_zero_new_range(struct inode *inode, loff_t start, loff_t end)
> Function name and comments no longer seem to match. What do you think
> about changing them like this?
> 
> /*
>   * exfat_prepare_valid_range - populate and dirty folios covering [start, end)
>   *
>   * Populate missing blocks via the read path, which zero-fills data
>   * beyond the current valid_size while preserving uptodate blocks.
>   * Mark entire folios dirty so the newly valid range is written back.
>   *
>   * Call before advancing valid_size.
>   */

Okay, I'll add it to v3.

> 
> And generic/538 test fails with this patch set. Please check if this
> test failure is reproducible for you as well.

I ran the tests, but I wasn't able to reproduce the failure. Could you 
tell me the block size and cluster size used to reproduce it?

There is also a new failure in generic/551. This failure can be fixed by 
adding inode_dio_wait() to exfat_extend_valid_size(). Jiale's patch 
includes this fix:

https://lore.kernel.org/exfat/20261003093035.532916-4-yaojiale02@163.com/T/#u


Thanks.