From: Jens Axboe <[email protected]>
To: "Alexander V. Buev" <[email protected]>,
[email protected], [email protected],
Christoph Hellwig <[email protected]>,
"Martin K . Petersen" <[email protected]>,
Pavel Begunkov <[email protected]>,
Chaitanya Kulkarni <[email protected]>,
Mikhail Malygin <[email protected]>,
[email protected]
Subject: Re: [PATCH v2 2/3] block: io_uring: add READV_PI/WRITEV_PI operations
Date: Thu, 10 Feb 2022 12:07:13 -0700 [thread overview]
Message-ID: <[email protected]> (raw)
In-Reply-To: <[email protected]>
On 2/10/22 12:03 PM, Alexander V. Buev wrote:
>> On 2/10/22 6:08 AM, Alexander V. Buev wrote:
>>> Added new READV_PI/WRITEV_PI operations to io_uring.
>>> Added new pi_addr & pi_len fields to SQE struct.
>>> Added new pi_iter field and IOCB_USE_PI flag to kiocb struct.
>>> Make corresponding corrections to io uring trace event.
>>>
>>> +struct io_rw_pi_state {
>>> + struct iov_iter iter;
>>> + struct iov_iter_state iter_state;
>>> + struct iovec fast_iov[UIO_FASTIOV_PI];
>>> +};
>>> +
>>> +struct io_rw_pi {
>>> + struct io_rw rw;
>>> + struct iovec *pi_iov;
>>> + u32 nr_pi_segs;
>>> + struct io_rw_pi_state *s;
>>> +};
>>
>> One immediate issue I see here is that io_rw_pi is big, and we try very
>> hard to keep the per-command payload to 64-bytes. This would be 88 bytes
>> by my count :-/
>>
>> Do you need everything from io_rw? If not, I'd just make io_rw_pi
>> contain the bits you need and see if you can squeeze it into the
>> existing cacheline.
>
> In short - Yes. Current patch code call existing io_read/io_write functions.
> This functions use io_rw struct information and process this data.
> I wanted to use existing functions but may be this is wrong way in this
> case.
>
> The second problem with request size is that the patch adds pi_iter
> pointer to kiocb struct. This also increase whole request union
> length.
>
> So I can see some (may be possible) solution for this:
>
> 1) do not store whole kiocb struct in request
> and write fully separated io_read/write_pi functions
>
> 2) make special CONFIG_XXX variable and simplify hide this code
> as default
Option 2 really sucks, because then obviously everyone wants their
feature enabled, and then we are back to square one. So never rely on a
config option, if it can be avoided.
I'd like to see what option 1 looks like, that sounds like a far better
solution.
--
Jens Axboe
next prev parent reply other threads:[~2022-02-10 19:07 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-02-10 13:08 [PATCH v2 0/3] implement direct IO with integrity Alexander V. Buev
2022-02-10 13:08 ` [PATCH v2 1/3] block: bio-integrity: add PI iovec to bio Alexander V. Buev
2022-02-10 17:49 ` Chaitanya Kulkarni
2022-02-10 13:08 ` [PATCH v2 2/3] block: io_uring: add READV_PI/WRITEV_PI operations Alexander V. Buev
2022-02-10 15:39 ` Jens Axboe
2022-02-10 19:03 ` Alexander V. Buev
2022-02-10 19:07 ` Jens Axboe [this message]
2022-02-11 9:39 ` Alexander V. Buev
2022-02-10 13:08 ` [PATCH v2 3/3] block: fops: handle IOCB_USE_PI in direct IO Alexander V. Buev
2022-02-10 17:53 ` Chaitanya Kulkarni
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 \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
/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