On Tue, Feb 18, 2025 at 02:42:25PM -0800, Keith Busch wrote:
> From: Keith Busch <kbusch@xxxxxxxxxx>
>
> Similar to the fixed file path, requests may depend on a previous one
> to set up an index, so we need to allow linking them. The prep callback
> happens too soon for linked commands, so the lookup needs to be deferred
> to the issue path. Change the prep callbacks to just set the buf_index
> and let generic io_uring code handle the fixed buffer node setup, just
> like it already does for fixed files.
>
> Signed-off-by: Keith Busch <kbusch@xxxxxxxxxx>
> ---
...
> diff --git a/io_uring/net.c b/io_uring/net.c
> index 000dc70d08d0d..39838e8575b53 100644
> --- a/io_uring/net.c
> +++ b/io_uring/net.c
> @@ -1373,6 +1373,10 @@ int io_send_zc_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
> #endif
> if (unlikely(!io_msg_alloc_async(req)))
> return -ENOMEM;
> + if (zc->flags & IORING_RECVSEND_FIXED_BUF) {
> + req->buf_index = zc->buf_index;
> + req->flags |= REQ_F_FIXED_BUFFER;
> + }
> if (req->opcode != IORING_OP_SENDMSG_ZC)
> return io_send_setup(req, sqe);
> return io_sendmsg_setup(req, sqe);
> @@ -1434,25 +1438,10 @@ static int io_send_zc_import(struct io_kiocb *req, unsigned int issue_flags)
> struct io_async_msghdr *kmsg = req->async_data;
> int ret;
>
> - if (sr->flags & IORING_RECVSEND_FIXED_BUF) {
> - struct io_ring_ctx *ctx = req->ctx;
> - struct io_rsrc_node *node;
> -
> - ret = -EFAULT;
> - io_ring_submit_lock(ctx, issue_flags);
> - node = io_rsrc_node_lookup(&ctx->buf_table, sr->buf_index);
> - if (node) {
> - io_req_assign_buf_node(sr->notif, node);
Here the node buffer is assigned to ->notif req, instead of the current
request, and you may have to deal with this case here.
Otherwise, this patch looks fine:
Reviewed-by: Ming Lei <ming.lei@xxxxxxxxxx>
Thanks,
Ming
