linux-ext4 - Re: [PATCH v2 4/4] ext4: order mode should not take effect for DIO

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <0600ebcc-d7dd-3586-b855-a94985d3b3da@huawei.com>
Date: Fri, 7 Nov 2025 19:54:02 +0800
From: yangerkun <yangerkun@...wei.com>
To: Jan Kara <jack@...e.cz>
CC: <linux-ext4@...r.kernel.org>, <tytso@....edu>, <adilger.kernel@...ger.ca>,
	<yi.zhang@...wei.com>, <libaokun1@...wei.com>, <yangerkun@...weicloud.com>
Subject: Re: [PATCH v2 4/4] ext4: order mode should not take effect for DIO



在 2025/11/4 23:07, Jan Kara 写道:
> On Tue 04-11-25 21:17:50, Yang Erkun wrote:
>> Since the size will be updated after the DIO completes, the data
>> will not be shown to userspace before that.
>>
>> Reviewed-by: Zhang Yi <yi.zhang@...wei.com>
>> Signed-off-by: Yang Erkun <yangerkun@...wei.com>
> 
> Hum, is there some measurable performance benefit from this? If yes, it
> would be good to mention it in the changelog. If not, then why bother?

Hi,

I discovered this while debugging. Today, I also conducted some
performance tests on this patch, but it appears that there is no
noticeable improvement. Therefore, I agree with your suggestion and will
remove it in the next version!


Thanks,
Erkun.

> 
> 								Honza
> 
>> ---
>>   fs/ext4/ext4.h              | 2 ++
>>   fs/ext4/inode.c             | 5 +++--
>>   include/trace/events/ext4.h | 1 +
>>   3 files changed, 6 insertions(+), 2 deletions(-)
>>
>> diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
>> index 96d7d649ccb0..d0331697467d 100644
>> --- a/fs/ext4/ext4.h
>> +++ b/fs/ext4/ext4.h
>> @@ -715,6 +715,8 @@ enum {
>>   #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
>>   	/* Don't normalize allocation size (used for fallocate) */
>>   #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
>> +	/* Get blocks from DIO */
>> +#define EXT4_GET_BLOCKS_DIO			0x0080
>>   	/* Convert written extents to unwritten */
>>   #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
>>   	/* Write zeros to newly created written extents */
>> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
>> index 3d8ada26d5cd..168dbcc9e921 100644
>> --- a/fs/ext4/inode.c
>> +++ b/fs/ext4/inode.c
>> @@ -818,6 +818,7 @@ int ext4_map_blocks(handle_t *handle, struct inode *inode,
>>   		if (map->m_flags & EXT4_MAP_NEW &&
>>   		    !(map->m_flags & EXT4_MAP_UNWRITTEN) &&
>>   		    !(flags & EXT4_GET_BLOCKS_ZERO) &&
>> +		    !(flags & EXT4_GET_BLOCKS_DIO) &&
>>   		    !ext4_is_quota_file(inode) &&
>>   		    ext4_should_order_data(inode)) {
>>   			loff_t start_byte =
>> @@ -3729,9 +3730,9 @@ static int ext4_iomap_alloc(struct inode *inode, struct ext4_map_blocks *map,
>>   	 * happening and thus expose allocated blocks to direct I/O reads.
>>   	 */
>>   	else if (((loff_t)map->m_lblk << blkbits) >= i_size_read(inode))
>> -		m_flags = EXT4_GET_BLOCKS_CREATE;
>> +		m_flags = EXT4_GET_BLOCKS_CREATE | EXT4_GET_BLOCKS_DIO;
>>   	else if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
>> -		m_flags = EXT4_GET_BLOCKS_IO_CREATE_EXT;
>> +		m_flags = EXT4_GET_BLOCKS_IO_CREATE_EXT | EXT4_GET_BLOCKS_DIO;
>>   
>>   	if (flags & IOMAP_ATOMIC)
>>   		ret = ext4_map_blocks_atomic_write(handle, inode, map, m_flags,
>> diff --git a/include/trace/events/ext4.h b/include/trace/events/ext4.h
>> index ada2b9223df5..de6d848f2e37 100644
>> --- a/include/trace/events/ext4.h
>> +++ b/include/trace/events/ext4.h
>> @@ -43,6 +43,7 @@ struct partial_cluster;
>>   	{ EXT4_GET_BLOCKS_CONVERT,		"CONVERT" },		\
>>   	{ EXT4_GET_BLOCKS_METADATA_NOFAIL,	"METADATA_NOFAIL" },	\
>>   	{ EXT4_GET_BLOCKS_NO_NORMALIZE,		"NO_NORMALIZE" },	\
>> +	{ EXT4_GET_BLOCKS_DIO,			"DIO" },		\
>>   	{ EXT4_GET_BLOCKS_CONVERT_UNWRITTEN,	"CONVERT_UNWRITTEN" },  \
>>   	{ EXT4_GET_BLOCKS_ZERO,			"ZERO" },		\
>>   	{ EXT4_GET_BLOCKS_IO_SUBMIT,		"IO_SUBMIT" },		\
>> -- 
>> 2.39.2
>>