From: Pavel Begunkov <asml.silence@gmail.com>
To: Jens Axboe <axboe@kernel.dk>, io-uring@vger.kernel.org
Subject: Re: [PATCH 1/4] io_uring/uring_cmd: get rid of using req->imu
Date: Fri, 18 Oct 2024 20:34:25 +0100 [thread overview]
Message-ID: <19d5a8bd-60cd-496f-a7ba-ffde5dd2e906@gmail.com> (raw)
In-Reply-To: <20241018183948.464779-2-axboe@kernel.dk>
On 10/18/24 19:38, Jens Axboe wrote:
> It's pretty pointless to use io_kiocb as intermediate storage for this,
> so split the validity check and the actual usage.
The table is uring_lock protected, if we don't resolve in advance
we should take care of locking when importing.
Another concern is adding a gap b/w setting a rsrc node and looking
up a buffer. That should be fine, but worth mentioning that when
you grab a rsrc node it also prevent destruction of all objects that
are created after this point.
> Signed-off-by: Jens Axboe <axboe@kernel.dk>
> ---
> io_uring/uring_cmd.c | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/io_uring/uring_cmd.c b/io_uring/uring_cmd.c
> index 39c3c816ec78..cc8bb5550ff5 100644
> --- a/io_uring/uring_cmd.c
> +++ b/io_uring/uring_cmd.c
> @@ -211,11 +211,10 @@ int io_uring_cmd_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
> struct io_ring_ctx *ctx = req->ctx;
> u16 index;
>
> - req->buf_index = READ_ONCE(sqe->buf_index);
> + index = READ_ONCE(sqe->buf_index);
> + req->buf_index = array_index_nospec(index, ctx->nr_user_bufs);
> if (unlikely(req->buf_index >= ctx->nr_user_bufs))
> return -EFAULT;
The rsrc should hold the table destruction, but it feels like it
should better move where importing happens.
> - index = array_index_nospec(req->buf_index, ctx->nr_user_bufs);
> - req->imu = ctx->user_bufs[index];
> io_req_set_rsrc_node(req, ctx, 0);
> }
> ioucmd->cmd_op = READ_ONCE(sqe->cmd_op);
> @@ -272,8 +271,10 @@ int io_uring_cmd_import_fixed(u64 ubuf, unsigned long len, int rw,
> struct iov_iter *iter, void *ioucmd)
> {
> struct io_kiocb *req = cmd_to_io_kiocb(ioucmd);
> + struct io_mapped_ubuf *imu;
>
> - return io_import_fixed(rw, iter, req->imu, ubuf, len);
> + imu = req->ctx->user_bufs[req->buf_index];
> + return io_import_fixed(rw, iter, imu, ubuf, len);
> }
> EXPORT_SYMBOL_GPL(io_uring_cmd_import_fixed);
>
--
Pavel Begunkov
next prev parent reply other threads:[~2024-10-18 19:34 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-10-18 18:38 [PATCHSET for-next 0/4] Get rid of io_kiocb->imu Jens Axboe
2024-10-18 18:38 ` [PATCH 1/4] io_uring/uring_cmd: get rid of using req->imu Jens Axboe
2024-10-18 19:34 ` Pavel Begunkov [this message]
2024-10-19 21:56 ` Jens Axboe
2024-10-18 18:38 ` [PATCH 2/4] io_uring/rw: " Jens Axboe
2024-10-18 18:38 ` [PATCH 3/4] io_uring/net: move send zc fixed buffer import to issue path Jens Axboe
2024-10-18 18:38 ` [PATCH 4/4] io_uring: kill 'imu' from struct io_kiocb Jens Axboe
-- strict thread matches above, loose matches on Subject: below --
2024-10-22 2:03 [PATCHSET v2 for-next 0/4] Get rid of io_kiocb->imu Jens Axboe
2024-10-22 2:03 ` [PATCH 1/4] io_uring/uring_cmd: get rid of using req->imu Jens Axboe
2024-10-22 2:43 ` Ming Lei
2024-10-22 2:59 ` Jens Axboe
2024-10-22 8:34 ` Anuj Gupta
2024-10-22 13:18 ` Jens Axboe
2024-10-22 13:24 ` Jens Axboe
2024-10-22 13:32 [PATCHSET v3 for-next 0/4] Get rid of io_kiocb->imu Jens Axboe
2024-10-22 13:32 ` [PATCH 1/4] io_uring/uring_cmd: get rid of using req->imu Jens Axboe
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=19d5a8bd-60cd-496f-a7ba-ffde5dd2e906@gmail.com \
--to=asml.silence@gmail.com \
--cc=axboe@kernel.dk \
--cc=io-uring@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