From: Chao Yu <chao@kernel.org>
To: Yangtao Li <frank.li@vivo.com>, Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Qi Han <hanqi@vivo.com>,
linux-kernel@vger.kernel.org,
linux-f2fs-devel@lists.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH] f2fs: compress: don't force buffered io when in COMPR_MODE_USER mode
Date: Mon, 19 Jun 2023 12:04:52 +0800 [thread overview]
Message-ID: <65ce42d6-889a-5e1d-8f04-af8f66fd0afa@kernel.org> (raw)
In-Reply-To: <fe223231-5445-61ef-1ba8-0d46f4f4ed5f@vivo.com>
On 2023/6/19 11:11, Yangtao Li wrote:
> On 2023/6/19 8:54, Chao Yu wrote:
>
>> On 2023/6/13 12:14, Yangtao Li wrote:
>>>
>>> On 2023/6/12 22:38, Chao Yu wrote:
>>>> On 2023/6/9 21:15, Yangtao Li wrote:
>>>>> It is observed that when in user compression mode
>>>>> (compress_extension=*),
>>>>> even though the file is not compressed, the file is still forced to use
>>>>> buffer io, which makes the AndroBench sequential read and write drop
>>>>> significantly. In fact, when the file is not compressed, we don't need
>>>>> to force it to buffer io.
>>>>>
>>>>> | w/o patch | w/ patch |
>>>>> seq read (MB/s) | 1320.068 | 3696.154 |
>>>>> seq write (MB/s) | 617.996 | 2978.478 |
>>>>>
>>>>> Fixes: 4c8ff7095bef ("f2fs: support data compression")
>>>>> Signed-off-by: Qi Han <hanqi@vivo.com>
>>>>> Signed-off-by: Yangtao Li <frank.li@vivo.com>
>>>>> ---
>>>>> fs/f2fs/f2fs.h | 14 ++++++++++++++
>>>>> fs/f2fs/file.c | 2 +-
>>>>> 2 files changed, 15 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
>>>>> index 1efcfd9e5a99..7f5472525310 100644
>>>>> --- a/fs/f2fs/f2fs.h
>>>>> +++ b/fs/f2fs/f2fs.h
>>>>> @@ -3168,6 +3168,20 @@ static inline int f2fs_compressed_file(struct
>>>>> inode *inode)
>>>>> is_inode_flag_set(inode, FI_COMPRESSED_FILE);
>>>>> }
>>>>> +static inline bool f2fs_is_compressed_file(struct inode *inode)
>>>>> +{
>>>>> + int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
>>>>> +
>>>>> + if (compress_mode == COMPR_MODE_FS)
>>>>> + return f2fs_compressed_file(inode);
>>>>> + else if (atomic_read(&F2FS_I(inode)->i_compr_blocks) ||
>>>>
>>>> Should check dirty page as well? i_compr_blocks may increase after
>>>> data writeback.
>>>>
>>> IIUC, in COMPR_MODE_USER mode, i_compr_blocks will only be updated when
>>> FI_ENABLE_COMPRESS is enabled.
>>>
>>> If FI_ENABLE_COMPRESS is not enabled, i_compr_blocks will never be
>>> updated after data writeback.
>>>
>>> So there is no need to additionally judge whether there is a dirty page?
>>
>> Oh, user mode, that's correct.
>>
>> If we allow dio/aio on compress file, it needs to consider race case in
>> between aio and ioc_compress_file.
>
>
> The inode_lock is already held in f2fs_file_write_iter and f2fs_ioc_compress_file, I guess this is enough?
>
> What else?
aio may complete outside inode lock, so it needs to call inode_dio_wait()
in f2fs_ioc_compress_file() to avoid the race case?
Thanks,
>
>
> 4691 static ssize_t f2fs_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
> 4692 {
> 4693 struct inode *inode = file_inode(iocb->ki_filp);
> 4694 const loff_t orig_pos = iocb->ki_pos;
> 4695 const size_t orig_count = iov_iter_count(from);
> 4696 loff_t target_size;
> 4697 bool dio;
> 4698 bool may_need_sync = true;
> 4699 int preallocated;
> 4700 ssize_t ret;
> 4701
> 4702 if (unlikely(f2fs_cp_error(F2FS_I_SB(inode)))) {
> 4703 ret = -EIO;
> 4704 goto out;
> 4705 }
> 4706
> 4707 if (!f2fs_is_compress_backend_ready(inode)) {
> 4708 ret = -EOPNOTSUPP;
> 4709 goto out;
> 4710 }
> 4711
> 4712 if (iocb->ki_flags & IOCB_NOWAIT) {
> 4713 if (!inode_trylock(inode)) {
> 4714 ret = -EAGAIN;
> 4715 goto out;
> 4716 }
> 4717 } else {
> 4718 inode_lock(inode);
> 4719 }
>
>
> 4115 static int f2fs_ioc_compress_file(struct file *filp)
> 4116 {
> 4117 struct inode *inode = file_inode(filp);
> 4118 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
> 4119 pgoff_t page_idx = 0, last_idx;
> 4120 unsigned int blk_per_seg = sbi->blocks_per_seg;
> 4121 int cluster_size = F2FS_I(inode)->i_cluster_size;
> 4122 int count, ret;
> 4123
> 4124 if (!f2fs_sb_has_compression(sbi) ||
> 4125 F2FS_OPTION(sbi).compress_mode != COMPR_MODE_USER)
> 4126 return -EOPNOTSUPP;
> 4127
> 4128 if (!(filp->f_mode & FMODE_WRITE))
> 4129 return -EBADF;
> 4130
> 4131 if (!f2fs_compressed_file(inode))
> 4132 return -EINVAL;
> 4133
> 4134 f2fs_balance_fs(sbi, true);
> 4135
> 4136 file_start_write(filp);
> 4137 inode_lock(inode);
>
>
> Thanks,
>
>>
>> Thanks,
>>
>>>
>>>
>>> Thanks,
>>>
>>>> Thanks,
>>>>
>>>>> + is_inode_flag_set(inode, FI_COMPRESS_RELEASED) ||
>>>>> + is_inode_flag_set(inode, FI_ENABLE_COMPRESS))
>>>>> + return true;
>>>>> +
>>>>> + return false;
>>>>> +}
>>>>> +
>>>>> static inline bool f2fs_need_compress_data(struct inode *inode)
>>>>> {
>>>>> int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode;
>>>>> diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
>>>>> index 74ecc9e20619..0698129b2165 100644
>>>>> --- a/fs/f2fs/file.c
>>>>> +++ b/fs/f2fs/file.c
>>>>> @@ -821,7 +821,7 @@ static bool f2fs_force_buffered_io(struct inode
>>>>> *inode, int rw)
>>>>> return true;
>>>>> if (fsverity_active(inode))
>>>>> return true;
>>>>> - if (f2fs_compressed_file(inode))
>>>>> + if (f2fs_is_compressed_file(inode))
>>>>> return true;
>>>>> /* disallow direct IO if any of devices has unaligned blksize */
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
next prev parent reply other threads:[~2023-06-19 4:05 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-09 13:15 [f2fs-dev] [PATCH] f2fs: compress: don't force buffered io when in COMPR_MODE_USER mode Yangtao Li via Linux-f2fs-devel
2023-06-12 14:38 ` Chao Yu
2023-06-13 4:14 ` Yangtao Li via Linux-f2fs-devel
2023-06-19 0:54 ` Chao Yu
2023-06-19 3:11 ` Yangtao Li via Linux-f2fs-devel
2023-06-19 4:04 ` Chao Yu [this message]
2023-06-19 9:31 ` Yangtao Li via Linux-f2fs-devel
2023-06-19 14:13 ` Chao Yu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=65ce42d6-889a-5e1d-8f04-af8f66fd0afa@kernel.org \
--to=chao@kernel.org \
--cc=frank.li@vivo.com \
--cc=hanqi@vivo.com \
--cc=jaegeuk@kernel.org \
--cc=linux-f2fs-devel@lists.sourceforge.net \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).