leandrof/linux
1
0
Fork 0
forked from Minki/linux
Code Issues Pull Requests Packages Projects Releases Wiki Activity
4ccc6db090
linux/io_uring/net.c

1048 lines
25 KiB
C
Raw Normal View History

io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
// SPDX-License-Identifier: GPL-2.0
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/file.h>
#include <linux/slab.h>
#include <linux/net.h>
#include <linux/compat.h>
#include <net/compat.h>
#include <linux/io_uring.h>
#include <uapi/linux/io_uring.h>
#include "io_uring.h"
io_uring: split provided buffers handling into its own file Move both the opcodes related to it, and the internals code dealing with it. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-13 13:07:23 +00:00
#include "kbuf.h"
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
#include "alloc_cache.h"
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
#include "net.h"
#if defined(CONFIG_NET)
struct io_shutdown {
struct file *file;
int how;
};
struct io_accept {
struct file *file;
struct sockaddr __user *addr;
int __user *addr_len;
int flags;
u32 file_slot;
unsigned long nofile;
};
struct io_socket {
struct file *file;
int domain;
int type;
int protocol;
int flags;
u32 file_slot;
unsigned long nofile;
};
struct io_connect {
struct file *file;
struct sockaddr __user *addr;
int addr_len;
};
struct io_sr_msg {
struct file *file;
union {
struct compat_msghdr __user *umsg_compat;
struct user_msghdr __user *umsg;
void __user *buf;
};
int msg_flags;
size_t len;
size_t done_io;
unsigned int flags;
};
#define IO_APOLL_MULTI_POLLED (REQ_F_APOLL_MULTISHOT | REQ_F_POLLED)
int io_shutdown_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
struct io_shutdown *shutdown = io_kiocb_to_cmd(req);
if (unlikely(sqe->off || sqe->addr || sqe->rw_flags ||
sqe->buf_index || sqe->splice_fd_in))
return -EINVAL;
shutdown->how = READ_ONCE(sqe->len);
return 0;
}
int io_shutdown(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_shutdown *shutdown = io_kiocb_to_cmd(req);
struct socket *sock;
int ret;
if (issue_flags & IO_URING_F_NONBLOCK)
return -EAGAIN;
sock = sock_from_file(req->file);
if (unlikely(!sock))
return -ENOTSOCK;
ret = __sys_shutdown_sock(sock, shutdown->how);
io_req_set_res(req, ret, 0);
return IOU_OK;
}
static bool io_net_retry(struct socket *sock, int flags)
{
if (!(flags & MSG_WAITALL))
return false;
return sock->type == SOCK_STREAM || sock->type == SOCK_SEQPACKET;
}
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
static void io_netmsg_recycle(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_async_msghdr *hdr = req->async_data;
if (!hdr || issue_flags & IO_URING_F_UNLOCKED)
return;
/* Let normal cleanup path reap it if we fail adding to the cache */
if (io_alloc_cache_put(&req->ctx->netmsg_cache, &hdr->cache)) {
req->async_data = NULL;
req->flags &= ~REQ_F_ASYNC_DATA;
}
}
static struct io_async_msghdr *io_recvmsg_alloc_async(struct io_kiocb *req,
unsigned int issue_flags)
{
struct io_ring_ctx *ctx = req->ctx;
struct io_cache_entry *entry;
if (!(issue_flags & IO_URING_F_UNLOCKED) &&
(entry = io_alloc_cache_get(&ctx->netmsg_cache)) != NULL) {
struct io_async_msghdr *hdr;
hdr = container_of(entry, struct io_async_msghdr, cache);
req->flags |= REQ_F_ASYNC_DATA;
req->async_data = hdr;
return hdr;
}
if (!io_alloc_async_data(req))
return req->async_data;
return NULL;
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
static int io_setup_async_msg(struct io_kiocb *req,
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
struct io_async_msghdr *kmsg,
unsigned int issue_flags)
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
{
struct io_async_msghdr *async_msg = req->async_data;
if (async_msg)
return -EAGAIN;
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
async_msg = io_recvmsg_alloc_async(req, issue_flags);
if (!async_msg) {
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
kfree(kmsg->free_iov);
return -ENOMEM;
}
req->flags |= REQ_F_NEED_CLEANUP;
memcpy(async_msg, kmsg, sizeof(*kmsg));
async_msg->msg.msg_name = &async_msg->addr;
/* if were using fast_iov, set it to the new one */
if (!async_msg->free_iov)
async_msg->msg.msg_iter.iov = async_msg->fast_iov;
return -EAGAIN;
}
static int io_sendmsg_copy_hdr(struct io_kiocb *req,
struct io_async_msghdr *iomsg)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
iomsg->msg.msg_name = &iomsg->addr;
iomsg->free_iov = iomsg->fast_iov;
return sendmsg_copy_msghdr(&iomsg->msg, sr->umsg, sr->msg_flags,
&iomsg->free_iov);
}
int io_sendmsg_prep_async(struct io_kiocb *req)
{
int ret;
ret = io_sendmsg_copy_hdr(req, req->async_data);
if (!ret)
req->flags |= REQ_F_NEED_CLEANUP;
return ret;
}
void io_sendmsg_recvmsg_cleanup(struct io_kiocb *req)
{
struct io_async_msghdr *io = req->async_data;
kfree(io->free_iov);
}
int io_sendmsg_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
if (unlikely(sqe->file_index || sqe->addr2))
return -EINVAL;
sr->umsg = u64_to_user_ptr(READ_ONCE(sqe->addr));
sr->len = READ_ONCE(sqe->len);
sr->flags = READ_ONCE(sqe->ioprio);
if (sr->flags & ~IORING_RECVSEND_POLL_FIRST)
return -EINVAL;
sr->msg_flags = READ_ONCE(sqe->msg_flags) | MSG_NOSIGNAL;
if (sr->msg_flags & MSG_DONTWAIT)
req->flags |= REQ_F_NOWAIT;
#ifdef CONFIG_COMPAT
if (req->ctx->compat)
sr->msg_flags |= MSG_CMSG_COMPAT;
#endif
sr->done_io = 0;
return 0;
}
int io_sendmsg(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
struct io_async_msghdr iomsg, *kmsg;
struct socket *sock;
unsigned flags;
int min_ret = 0;
int ret;
sock = sock_from_file(req->file);
if (unlikely(!sock))
return -ENOTSOCK;
if (req_has_async_data(req)) {
kmsg = req->async_data;
} else {
ret = io_sendmsg_copy_hdr(req, &iomsg);
if (ret)
return ret;
kmsg = &iomsg;
}
if (!(req->flags & REQ_F_POLLED) &&
(sr->flags & IORING_RECVSEND_POLL_FIRST))
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
return io_setup_async_msg(req, kmsg, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
flags = sr->msg_flags;
if (issue_flags & IO_URING_F_NONBLOCK)
flags |= MSG_DONTWAIT;
if (flags & MSG_WAITALL)
min_ret = iov_iter_count(&kmsg->msg.msg_iter);
ret = __sys_sendmsg_sock(sock, &kmsg->msg, flags);
if (ret < min_ret) {
if (ret == -EAGAIN && (issue_flags & IO_URING_F_NONBLOCK))
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
return io_setup_async_msg(req, kmsg, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (ret == -ERESTARTSYS)
ret = -EINTR;
if (ret > 0 && io_net_retry(sock, flags)) {
sr->done_io += ret;
req->flags |= REQ_F_PARTIAL_IO;
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
return io_setup_async_msg(req, kmsg, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
}
req_set_fail(req);
}
/* fast path, check for non-NULL to avoid function call */
if (kmsg->free_iov)
kfree(kmsg->free_iov);
req->flags &= ~REQ_F_NEED_CLEANUP;
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
io_netmsg_recycle(req, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (ret >= 0)
ret += sr->done_io;
else if (sr->done_io)
ret = sr->done_io;
io_req_set_res(req, ret, 0);
return IOU_OK;
}
int io_send(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
struct msghdr msg;
struct iovec iov;
struct socket *sock;
unsigned flags;
int min_ret = 0;
int ret;
if (!(req->flags & REQ_F_POLLED) &&
(sr->flags & IORING_RECVSEND_POLL_FIRST))
return -EAGAIN;
sock = sock_from_file(req->file);
if (unlikely(!sock))
return -ENOTSOCK;
ret = import_single_range(WRITE, sr->buf, sr->len, &iov, &msg.msg_iter);
if (unlikely(ret))
return ret;
msg.msg_name = NULL;
msg.msg_control = NULL;
msg.msg_controllen = 0;
msg.msg_namelen = 0;
flags = sr->msg_flags;
if (issue_flags & IO_URING_F_NONBLOCK)
flags |= MSG_DONTWAIT;
if (flags & MSG_WAITALL)
min_ret = iov_iter_count(&msg.msg_iter);
msg.msg_flags = flags;
ret = sock_sendmsg(sock, &msg);
if (ret < min_ret) {
if (ret == -EAGAIN && (issue_flags & IO_URING_F_NONBLOCK))
return -EAGAIN;
if (ret == -ERESTARTSYS)
ret = -EINTR;
if (ret > 0 && io_net_retry(sock, flags)) {
sr->len -= ret;
sr->buf += ret;
sr->done_io += ret;
req->flags |= REQ_F_PARTIAL_IO;
return -EAGAIN;
}
req_set_fail(req);
}
if (ret >= 0)
ret += sr->done_io;
else if (sr->done_io)
ret = sr->done_io;
io_req_set_res(req, ret, 0);
return IOU_OK;
}
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
static bool io_recvmsg_multishot_overflow(struct io_async_msghdr *iomsg)
{
unsigned long hdr;
if (check_add_overflow(sizeof(struct io_uring_recvmsg_out),
(unsigned long)iomsg->namelen, &hdr))
return true;
if (check_add_overflow(hdr, iomsg->controllen, &hdr))
return true;
if (hdr > INT_MAX)
return true;
return false;
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
static int __io_recvmsg_copy_hdr(struct io_kiocb *req,
struct io_async_msghdr *iomsg)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
net: copy from user before calling __copy_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-2-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:56 +00:00
struct user_msghdr msg;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
int ret;
net: copy from user before calling __copy_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-2-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:56 +00:00
if (copy_from_user(&msg, sr->umsg, sizeof(*sr->umsg)))
return -EFAULT;
ret = __copy_msghdr(&iomsg->msg, &msg, &iomsg->uaddr);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (ret)
return ret;
if (req->flags & REQ_F_BUFFER_SELECT) {
net: copy from user before calling __copy_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-2-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:56 +00:00
if (msg.msg_iovlen == 0) {
io_uring: allow iov_len = 0 for recvmsg and buffer select When using BUFFER_SELECT there is no technical requirement that the user actually provides iov, and this removes one copy_from_user call. So allow iov_len to be 0. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-4-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:22 +00:00
sr->len = iomsg->fast_iov[0].iov_len = 0;
iomsg->fast_iov[0].iov_base = NULL;
iomsg->free_iov = NULL;
net: copy from user before calling __copy_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-2-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:56 +00:00
} else if (msg.msg_iovlen > 1) {
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
return -EINVAL;
io_uring: allow iov_len = 0 for recvmsg and buffer select When using BUFFER_SELECT there is no technical requirement that the user actually provides iov, and this removes one copy_from_user call. So allow iov_len to be 0. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-4-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:22 +00:00
} else {
net: copy from user before calling __copy_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-2-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:56 +00:00
if (copy_from_user(iomsg->fast_iov, msg.msg_iov, sizeof(*msg.msg_iov)))
io_uring: allow iov_len = 0 for recvmsg and buffer select When using BUFFER_SELECT there is no technical requirement that the user actually provides iov, and this removes one copy_from_user call. So allow iov_len to be 0. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-4-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:22 +00:00
return -EFAULT;
sr->len = iomsg->fast_iov[0].iov_len;
iomsg->free_iov = NULL;
}
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
if (req->flags & REQ_F_APOLL_MULTISHOT) {
iomsg->namelen = msg.msg_namelen;
iomsg->controllen = msg.msg_controllen;
if (io_recvmsg_multishot_overflow(iomsg))
return -EOVERFLOW;
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
} else {
iomsg->free_iov = iomsg->fast_iov;
net: copy from user before calling __copy_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-2-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:56 +00:00
ret = __import_iovec(READ, msg.msg_iov, msg.msg_iovlen, UIO_FASTIOV,
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
&iomsg->free_iov, &iomsg->msg.msg_iter,
false);
if (ret > 0)
ret = 0;
}
return ret;
}
#ifdef CONFIG_COMPAT
static int __io_compat_recvmsg_copy_hdr(struct io_kiocb *req,
struct io_async_msghdr *iomsg)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
net: copy from user before calling __get_compat_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-3-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:57 +00:00
struct compat_msghdr msg;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
struct compat_iovec __user *uiov;
int ret;
net: copy from user before calling __get_compat_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-3-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:57 +00:00
if (copy_from_user(&msg, sr->umsg_compat, sizeof(msg)))
return -EFAULT;
ret = __get_compat_msghdr(&iomsg->msg, sr->umsg_compat, &iomsg->uaddr);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (ret)
return ret;
net: copy from user before calling __get_compat_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-3-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:57 +00:00
uiov = compat_ptr(msg.msg_iov);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (req->flags & REQ_F_BUFFER_SELECT) {
compat_ssize_t clen;
net: copy from user before calling __get_compat_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-3-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:57 +00:00
if (msg.msg_iovlen == 0) {
io_uring: support 0 length iov in buffer select in compat Match up work done in "io_uring: allow iov_len = 0 for recvmsg and buffer select", but for compat code path. Fixes: a68caad69ce5 ("io_uring: allow iov_len = 0 for recvmsg and buffer select") Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220708181838.1495428-3-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-08 18:18:36 +00:00
sr->len = 0;
iomsg->free_iov = NULL;
net: copy from user before calling __get_compat_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-3-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:57 +00:00
} else if (msg.msg_iovlen > 1) {
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
return -EINVAL;
io_uring: support 0 length iov in buffer select in compat Match up work done in "io_uring: allow iov_len = 0 for recvmsg and buffer select", but for compat code path. Fixes: a68caad69ce5 ("io_uring: allow iov_len = 0 for recvmsg and buffer select") Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220708181838.1495428-3-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-08 18:18:36 +00:00
} else {
if (!access_ok(uiov, sizeof(*uiov)))
return -EFAULT;
if (__get_user(clen, &uiov->iov_len))
return -EFAULT;
if (clen < 0)
return -EINVAL;
sr->len = clen;
iomsg->free_iov = NULL;
}
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
if (req->flags & REQ_F_APOLL_MULTISHOT) {
iomsg->namelen = msg.msg_namelen;
iomsg->controllen = msg.msg_controllen;
if (io_recvmsg_multishot_overflow(iomsg))
return -EOVERFLOW;
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
} else {
iomsg->free_iov = iomsg->fast_iov;
net: copy from user before calling __get_compat_msghdr this is in preparation for multishot receive from io_uring, where it needs to have access to the original struct user_msghdr. functionally this should be a no-op. Acked-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-3-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:57 +00:00
ret = __import_iovec(READ, (struct iovec __user *)uiov, msg.msg_iovlen,
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
UIO_FASTIOV, &iomsg->free_iov,
&iomsg->msg.msg_iter, true);
if (ret < 0)
return ret;
}
return 0;
}
#endif
static int io_recvmsg_copy_hdr(struct io_kiocb *req,
struct io_async_msghdr *iomsg)
{
iomsg->msg.msg_name = &iomsg->addr;
#ifdef CONFIG_COMPAT
if (req->ctx->compat)
return __io_compat_recvmsg_copy_hdr(req, iomsg);
#endif
return __io_recvmsg_copy_hdr(req, iomsg);
}
int io_recvmsg_prep_async(struct io_kiocb *req)
{
int ret;
ret = io_recvmsg_copy_hdr(req, req->async_data);
if (!ret)
req->flags |= REQ_F_NEED_CLEANUP;
return ret;
}
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
#define RECVMSG_FLAGS (IORING_RECVSEND_POLL_FIRST | IORING_RECV_MULTISHOT)
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
int io_recvmsg_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
if (unlikely(sqe->file_index || sqe->addr2))
return -EINVAL;
sr->umsg = u64_to_user_ptr(READ_ONCE(sqe->addr));
sr->len = READ_ONCE(sqe->len);
sr->flags = READ_ONCE(sqe->ioprio);
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
if (sr->flags & ~(RECVMSG_FLAGS))
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
return -EINVAL;
sr->msg_flags = READ_ONCE(sqe->msg_flags) | MSG_NOSIGNAL;
if (sr->msg_flags & MSG_DONTWAIT)
req->flags |= REQ_F_NOWAIT;
if (sr->msg_flags & MSG_ERRQUEUE)
req->flags |= REQ_F_CLEAR_POLLIN;
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
if (sr->flags & IORING_RECV_MULTISHOT) {
if (!(req->flags & REQ_F_BUFFER_SELECT))
return -EINVAL;
if (sr->msg_flags & MSG_WAITALL)
return -EINVAL;
if (req->opcode == IORING_OP_RECV && sr->len)
return -EINVAL;
req->flags |= REQ_F_APOLL_MULTISHOT;
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
#ifdef CONFIG_COMPAT
if (req->ctx->compat)
sr->msg_flags |= MSG_CMSG_COMPAT;
#endif
sr->done_io = 0;
return 0;
}
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
static inline void io_recv_prep_retry(struct io_kiocb *req)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
sr->done_io = 0;
sr->len = 0; /* get from the provided buffer */
}
/*
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
* Finishes io_recv and io_recvmsg.
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
*
* Returns true if it is actually finished, or false if it should run
* again (for multishot).
*/
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
static inline bool io_recv_finish(struct io_kiocb *req, int *ret,
unsigned int cflags, bool mshot_finished)
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
{
if (!(req->flags & REQ_F_APOLL_MULTISHOT)) {
io_req_set_res(req, *ret, cflags);
*ret = IOU_OK;
return true;
}
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
if (!mshot_finished) {
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
if (io_post_aux_cqe(req->ctx, req->cqe.user_data, *ret,
cflags | IORING_CQE_F_MORE, false)) {
io_recv_prep_retry(req);
return false;
}
/*
* Otherwise stop multishot but use the current result.
* Probably will end up going into overflow, but this means
* we cannot trust the ordering anymore
*/
}
io_req_set_res(req, *ret, cflags);
if (req->flags & REQ_F_POLLED)
*ret = IOU_STOP_MULTISHOT;
io_uring: fix multishot ending when not polled If multishot is not actually polling then return IOU_OK rather than the result. If the result was > 0 this will confuse things further up the callstack which expect a return <= 0. Fixes: 1300ebb20286 ("io_uring: multishot recv") Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220708181838.1495428-2-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-08 18:18:35 +00:00
else
*ret = IOU_OK;
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
return true;
}
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
static int io_recvmsg_prep_multishot(struct io_async_msghdr *kmsg,
struct io_sr_msg *sr, void __user **buf,
size_t *len)
{
unsigned long ubuf = (unsigned long) *buf;
unsigned long hdr;
hdr = sizeof(struct io_uring_recvmsg_out) + kmsg->namelen +
kmsg->controllen;
if (*len < hdr)
return -EFAULT;
if (kmsg->controllen) {
unsigned long control = ubuf + hdr - kmsg->controllen;
kmsg->msg.msg_control_user = (void *) control;
kmsg->msg.msg_controllen = kmsg->controllen;
}
sr->buf = *buf; /* stash for later copy */
*buf = (void *) (ubuf + hdr);
kmsg->payloadlen = *len = *len - hdr;
return 0;
}
struct io_recvmsg_multishot_hdr {
struct io_uring_recvmsg_out msg;
struct sockaddr_storage addr;
};
static int io_recvmsg_multishot(struct socket *sock, struct io_sr_msg *io,
struct io_async_msghdr *kmsg,
unsigned int flags, bool *finished)
{
int err;
int copy_len;
struct io_recvmsg_multishot_hdr hdr;
if (kmsg->namelen)
kmsg->msg.msg_name = &hdr.addr;
kmsg->msg.msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT);
kmsg->msg.msg_namelen = 0;
if (sock->file->f_flags & O_NONBLOCK)
flags |= MSG_DONTWAIT;
err = sock_recvmsg(sock, &kmsg->msg, flags);
*finished = err <= 0;
if (err < 0)
return err;
hdr.msg = (struct io_uring_recvmsg_out) {
.controllen = kmsg->controllen - kmsg->msg.msg_controllen,
.flags = kmsg->msg.msg_flags & ~MSG_CMSG_COMPAT
};
hdr.msg.payloadlen = err;
if (err > kmsg->payloadlen)
err = kmsg->payloadlen;
copy_len = sizeof(struct io_uring_recvmsg_out);
if (kmsg->msg.msg_namelen > kmsg->namelen)
copy_len += kmsg->namelen;
else
copy_len += kmsg->msg.msg_namelen;
/*
* "fromlen shall refer to the value before truncation.."
* 1003.1g
*/
hdr.msg.namelen = kmsg->msg.msg_namelen;
/* ensure that there is no gap between hdr and sockaddr_storage */
BUILD_BUG_ON(offsetof(struct io_recvmsg_multishot_hdr, addr) !=
sizeof(struct io_uring_recvmsg_out));
if (copy_to_user(io->buf, &hdr, copy_len)) {
*finished = true;
return -EFAULT;
}
return sizeof(struct io_uring_recvmsg_out) + kmsg->namelen +
kmsg->controllen + err;
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
int io_recvmsg(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
struct io_async_msghdr iomsg, *kmsg;
struct socket *sock;
unsigned int cflags;
unsigned flags;
int ret, min_ret = 0;
bool force_nonblock = issue_flags & IO_URING_F_NONBLOCK;
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
bool mshot_finished = true;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
sock = sock_from_file(req->file);
if (unlikely(!sock))
return -ENOTSOCK;
if (req_has_async_data(req)) {
kmsg = req->async_data;
} else {
ret = io_recvmsg_copy_hdr(req, &iomsg);
if (ret)
return ret;
kmsg = &iomsg;
}
if (!(req->flags & REQ_F_POLLED) &&
(sr->flags & IORING_RECVSEND_POLL_FIRST))
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
return io_setup_async_msg(req, kmsg, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
retry_multishot:
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (io_do_buffer_select(req)) {
void __user *buf;
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
size_t len = sr->len;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
buf = io_buffer_select(req, &len, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (!buf)
return -ENOBUFS;
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
if (req->flags & REQ_F_APOLL_MULTISHOT) {
ret = io_recvmsg_prep_multishot(kmsg, sr, &buf, &len);
if (ret) {
io_kbuf_recycle(req, issue_flags);
return ret;
}
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
kmsg->fast_iov[0].iov_base = buf;
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
kmsg->fast_iov[0].iov_len = len;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
iov_iter_init(&kmsg->msg.msg_iter, READ, kmsg->fast_iov, 1,
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
len);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
}
flags = sr->msg_flags;
if (force_nonblock)
flags |= MSG_DONTWAIT;
if (flags & MSG_WAITALL)
min_ret = iov_iter_count(&kmsg->msg.msg_iter);
kmsg->msg.msg_get_inq = 1;
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
if (req->flags & REQ_F_APOLL_MULTISHOT)
ret = io_recvmsg_multishot(sock, sr, kmsg, flags,
&mshot_finished);
else
ret = __sys_recvmsg_sock(sock, &kmsg->msg, sr->umsg,
kmsg->uaddr, flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (ret < min_ret) {
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
if (ret == -EAGAIN && force_nonblock) {
ret = io_setup_async_msg(req, kmsg, issue_flags);
if (ret == -EAGAIN && (req->flags & IO_APOLL_MULTI_POLLED) ==
IO_APOLL_MULTI_POLLED) {
io_kbuf_recycle(req, issue_flags);
return IOU_ISSUE_SKIP_COMPLETE;
}
return ret;
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (ret == -ERESTARTSYS)
ret = -EINTR;
if (ret > 0 && io_net_retry(sock, flags)) {
sr->done_io += ret;
req->flags |= REQ_F_PARTIAL_IO;
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
return io_setup_async_msg(req, kmsg, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
}
req_set_fail(req);
} else if ((flags & MSG_WAITALL) && (kmsg->msg.msg_flags & (MSG_TRUNC | MSG_CTRUNC))) {
req_set_fail(req);
}
io_uring: recycle buffers on error Rather than passing an error back to the user with a buffer attached, recycle the buffer immediately. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-5-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:23 +00:00
if (ret > 0)
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
ret += sr->done_io;
else if (sr->done_io)
ret = sr->done_io;
io_uring: recycle buffers on error Rather than passing an error back to the user with a buffer attached, recycle the buffer immediately. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-5-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:23 +00:00
else
io_kbuf_recycle(req, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
cflags = io_put_kbuf(req, issue_flags);
if (kmsg->msg.msg_inq)
cflags |= IORING_CQE_F_SOCK_NONEMPTY;
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
if (!io_recv_finish(req, &ret, cflags, mshot_finished))
goto retry_multishot;
if (mshot_finished) {
io_netmsg_recycle(req, issue_flags);
/* fast path, check for non-NULL to avoid function call */
if (kmsg->free_iov)
kfree(kmsg->free_iov);
req->flags &= ~REQ_F_NEED_CLEANUP;
}
return ret;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
}
int io_recv(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_sr_msg *sr = io_kiocb_to_cmd(req);
struct msghdr msg;
struct socket *sock;
struct iovec iov;
unsigned int cflags;
unsigned flags;
int ret, min_ret = 0;
bool force_nonblock = issue_flags & IO_URING_F_NONBLOCK;
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
size_t len = sr->len;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (!(req->flags & REQ_F_POLLED) &&
(sr->flags & IORING_RECVSEND_POLL_FIRST))
return -EAGAIN;
sock = sock_from_file(req->file);
if (unlikely(!sock))
return -ENOTSOCK;
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
retry_multishot:
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (io_do_buffer_select(req)) {
void __user *buf;
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
buf = io_buffer_select(req, &len, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (!buf)
return -ENOBUFS;
sr->buf = buf;
}
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
ret = import_single_range(READ, sr->buf, len, &iov, &msg.msg_iter);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (unlikely(ret))
goto out_free;
msg.msg_name = NULL;
msg.msg_namelen = 0;
msg.msg_control = NULL;
msg.msg_get_inq = 1;
msg.msg_flags = 0;
msg.msg_controllen = 0;
msg.msg_iocb = NULL;
flags = sr->msg_flags;
if (force_nonblock)
flags |= MSG_DONTWAIT;
if (flags & MSG_WAITALL)
min_ret = iov_iter_count(&msg.msg_iter);
ret = sock_recvmsg(sock, &msg, flags);
if (ret < min_ret) {
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
if (ret == -EAGAIN && force_nonblock) {
if ((req->flags & IO_APOLL_MULTI_POLLED) == IO_APOLL_MULTI_POLLED) {
io_kbuf_recycle(req, issue_flags);
return IOU_ISSUE_SKIP_COMPLETE;
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
return -EAGAIN;
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
if (ret == -ERESTARTSYS)
ret = -EINTR;
if (ret > 0 && io_net_retry(sock, flags)) {
sr->len -= ret;
sr->buf += ret;
sr->done_io += ret;
req->flags |= REQ_F_PARTIAL_IO;
return -EAGAIN;
}
req_set_fail(req);
} else if ((flags & MSG_WAITALL) && (msg.msg_flags & (MSG_TRUNC | MSG_CTRUNC))) {
out_free:
req_set_fail(req);
}
io_uring: recycle buffers on error Rather than passing an error back to the user with a buffer attached, recycle the buffer immediately. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-5-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:23 +00:00
if (ret > 0)
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
ret += sr->done_io;
else if (sr->done_io)
ret = sr->done_io;
io_uring: recycle buffers on error Rather than passing an error back to the user with a buffer attached, recycle the buffer immediately. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-5-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:23 +00:00
else
io_kbuf_recycle(req, issue_flags);
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
cflags = io_put_kbuf(req, issue_flags);
if (msg.msg_inq)
cflags |= IORING_CQE_F_SOCK_NONEMPTY;
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
io_uring: support multishot in recvmsg Similar to multishot recv, this will require provided buffers to be used. However recvmsg is much more complex than recv as it has multiple outputs. Specifically flags, name, and control messages. Support this by introducing a new struct io_uring_recvmsg_out with 4 fields. namelen, controllen and flags match the similar out fields in msghdr from standard recvmsg(2), payloadlen is the length of the payload following the header. This struct is placed at the start of the returned buffer. Based on what the user specifies in struct msghdr, the next bytes of the buffer will be name (the next msg_namelen bytes), and then control (the next msg_controllen bytes). The payload will come at the end. The return value in the CQE is the total used size of the provided buffer. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220714110258.1336200-4-dylany@fb.com [axboe: style fixups, see link] Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-14 11:02:58 +00:00
if (!io_recv_finish(req, &ret, cflags, ret <= 0))
io_uring: multishot recv Support multishot receive for io_uring. Typical server applications will run a loop where for each recv CQE it requeues another recv/recvmsg. This can be simplified by using the existing multishot functionality combined with io_uring's provided buffers. The API is to add the IORING_RECV_MULTISHOT flag to the SQE. CQEs will then be posted (with IORING_CQE_F_MORE flag set) when data is available and is read. Once an error occurs or the socket ends, the multishot will be removed and a completion without IORING_CQE_F_MORE will be posted. The benefit to this is that the recv is much more performant. * Subsequent receives are queued up straight away without requiring the application to finish a processing loop. * If there are more data in the socket (sat the provided buffer size is smaller than the socket buffer) then the data is immediately returned, improving batching. * Poll is only armed once and reused, saving CPU cycles Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-11-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:29 +00:00
goto retry_multishot;
return ret;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
}
int io_accept_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
struct io_accept *accept = io_kiocb_to_cmd(req);
unsigned flags;
if (sqe->len || sqe->buf_index)
return -EINVAL;
accept->addr = u64_to_user_ptr(READ_ONCE(sqe->addr));
accept->addr_len = u64_to_user_ptr(READ_ONCE(sqe->addr2));
accept->flags = READ_ONCE(sqe->accept_flags);
accept->nofile = rlimit(RLIMIT_NOFILE);
flags = READ_ONCE(sqe->ioprio);
if (flags & ~IORING_ACCEPT_MULTISHOT)
return -EINVAL;
accept->file_slot = READ_ONCE(sqe->file_index);
if (accept->file_slot) {
if (accept->flags & SOCK_CLOEXEC)
return -EINVAL;
if (flags & IORING_ACCEPT_MULTISHOT &&
accept->file_slot != IORING_FILE_INDEX_ALLOC)
return -EINVAL;
}
if (accept->flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK))
return -EINVAL;
if (SOCK_NONBLOCK != O_NONBLOCK && (accept->flags & SOCK_NONBLOCK))
accept->flags = (accept->flags & ~SOCK_NONBLOCK) | O_NONBLOCK;
if (flags & IORING_ACCEPT_MULTISHOT)
req->flags |= REQ_F_APOLL_MULTISHOT;
return 0;
}
int io_accept(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_ring_ctx *ctx = req->ctx;
struct io_accept *accept = io_kiocb_to_cmd(req);
bool force_nonblock = issue_flags & IO_URING_F_NONBLOCK;
unsigned int file_flags = force_nonblock ? O_NONBLOCK : 0;
bool fixed = !!accept->file_slot;
struct file *file;
int ret, fd;
retry:
if (!fixed) {
fd = __get_unused_fd_flags(accept->flags, accept->nofile);
if (unlikely(fd < 0))
return fd;
}
file = do_accept(req->file, file_flags, accept->addr, accept->addr_len,
accept->flags);
if (IS_ERR(file)) {
if (!fixed)
put_unused_fd(fd);
ret = PTR_ERR(file);
if (ret == -EAGAIN && force_nonblock) {
/*
* if it's multishot and polled, we don't need to
* return EAGAIN to arm the poll infra since it
* has already been done
*/
if ((req->flags & IO_APOLL_MULTI_POLLED) ==
IO_APOLL_MULTI_POLLED)
ret = IOU_ISSUE_SKIP_COMPLETE;
return ret;
}
if (ret == -ERESTARTSYS)
ret = -EINTR;
req_set_fail(req);
} else if (!fixed) {
fd_install(fd, file);
ret = fd;
} else {
ret = io_fixed_fd_install(req, issue_flags, file,
accept->file_slot);
}
if (!(req->flags & REQ_F_APOLL_MULTISHOT)) {
io_req_set_res(req, ret, 0);
return IOU_OK;
}
io_uring: fix multishot accept ordering Similar to multishot poll, drop multishot accept when CQE overflow occurs. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-10-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:28 +00:00
if (ret >= 0 &&
io_post_aux_cqe(ctx, req->cqe.user_data, ret, IORING_CQE_F_MORE, false))
io_uring: don't expose io_fill_cqe_aux() Deduplicate some code and add a helper for filling an aux CQE, locking and notification. Signed-off-by: Pavel Begunkov <asml.silence@gmail.com> Link: https://lore.kernel.org/r/b7c6557c8f9dc5c4cfb01292116c682a0ff61081.1655455613.git.asml.silence@gmail.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-17 08:48:00 +00:00
goto retry;
io_uring: fix multishot accept ordering Similar to multishot poll, drop multishot accept when CQE overflow occurs. Signed-off-by: Dylan Yudaken <dylany@fb.com> Link: https://lore.kernel.org/r/20220630091231.1456789-10-dylany@fb.com Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-06-30 09:12:28 +00:00
io_req_set_res(req, ret, 0);
if (req->flags & REQ_F_POLLED)
return IOU_STOP_MULTISHOT;
return IOU_OK;
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
}
int io_socket_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
struct io_socket *sock = io_kiocb_to_cmd(req);
if (sqe->addr || sqe->rw_flags || sqe->buf_index)
return -EINVAL;
sock->domain = READ_ONCE(sqe->fd);
sock->type = READ_ONCE(sqe->off);
sock->protocol = READ_ONCE(sqe->len);
sock->file_slot = READ_ONCE(sqe->file_index);
sock->nofile = rlimit(RLIMIT_NOFILE);
sock->flags = sock->type & ~SOCK_TYPE_MASK;
if (sock->file_slot && (sock->flags & SOCK_CLOEXEC))
return -EINVAL;
if (sock->flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK))
return -EINVAL;
return 0;
}
int io_socket(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_socket *sock = io_kiocb_to_cmd(req);
bool fixed = !!sock->file_slot;
struct file *file;
int ret, fd;
if (!fixed) {
fd = __get_unused_fd_flags(sock->flags, sock->nofile);
if (unlikely(fd < 0))
return fd;
}
file = __sys_socket_file(sock->domain, sock->type, sock->protocol);
if (IS_ERR(file)) {
if (!fixed)
put_unused_fd(fd);
ret = PTR_ERR(file);
if (ret == -EAGAIN && (issue_flags & IO_URING_F_NONBLOCK))
return -EAGAIN;
if (ret == -ERESTARTSYS)
ret = -EINTR;
req_set_fail(req);
} else if (!fixed) {
fd_install(fd, file);
ret = fd;
} else {
ret = io_fixed_fd_install(req, issue_flags, file,
sock->file_slot);
}
io_req_set_res(req, ret, 0);
return IOU_OK;
}
int io_connect_prep_async(struct io_kiocb *req)
{
struct io_async_connect *io = req->async_data;
struct io_connect *conn = io_kiocb_to_cmd(req);
return move_addr_to_kernel(conn->addr, conn->addr_len, &io->address);
}
int io_connect_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
struct io_connect *conn = io_kiocb_to_cmd(req);
if (sqe->len || sqe->buf_index || sqe->rw_flags || sqe->splice_fd_in)
return -EINVAL;
conn->addr = u64_to_user_ptr(READ_ONCE(sqe->addr));
conn->addr_len = READ_ONCE(sqe->addr2);
return 0;
}
int io_connect(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_connect *connect = io_kiocb_to_cmd(req);
struct io_async_connect __io, *io;
unsigned file_flags;
int ret;
bool force_nonblock = issue_flags & IO_URING_F_NONBLOCK;
if (req_has_async_data(req)) {
io = req->async_data;
} else {
ret = move_addr_to_kernel(connect->addr,
connect->addr_len,
&__io.address);
if (ret)
goto out;
io = &__io;
}
file_flags = force_nonblock ? O_NONBLOCK : 0;
ret = __sys_connect_file(req->file, &io->address,
connect->addr_len, file_flags);
if ((ret == -EAGAIN || ret == -EINPROGRESS) && force_nonblock) {
if (req_has_async_data(req))
return -EAGAIN;
if (io_alloc_async_data(req)) {
ret = -ENOMEM;
goto out;
}
memcpy(req->async_data, &__io, sizeof(__io));
return -EAGAIN;
}
if (ret == -ERESTARTSYS)
ret = -EINTR;
out:
if (ret < 0)
req_set_fail(req);
io_req_set_res(req, ret, 0);
return IOU_OK;
}
io_uring: add netmsg cache For recvmsg/sendmsg, if they don't complete inline, we currently need to allocate a struct io_async_msghdr for each request. This is a somewhat large struct. Hook up sendmsg/recvmsg to use the io_alloc_cache. This reduces the alloc + free overhead considerably, yielding 4-5% of extra performance running netbench. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-07-07 20:30:09 +00:00
void io_netmsg_cache_free(struct io_cache_entry *entry)
{
kfree(container_of(entry, struct io_async_msghdr, cache));
}
io_uring: split network related opcodes into its own file While at it, convert the handlers to just use io_eopnotsupp_prep() if CONFIG_NET isn't set. Signed-off-by: Jens Axboe <axboe@kernel.dk>
2022-05-25 12:25:13 +00:00
#endif
Reference in New Issue Copy Permalink