1

net: extend ubuf_info callback to ops structure

We'll need to associate additional callbacks with ubuf_info, introduce
a structure holding ubuf_info callbacks. Apart from a more smarter
io_uring notification management introduced in next patches, it can be
used to generalise msg_zerocopy_put_abort() and also store
->sg_from_iter, which is currently passed in struct msghdr.

Reviewed-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://lore.kernel.org/all/a62015541de49c0e2a8a0377a1d5d0a5aeb07016.1713369317.git.asml.silence@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This commit is contained in:
Pavel Begunkov 2024-04-19 12:08:39 +01:00 committed by Jakub Kicinski
parent 0bbac3facb
commit 7ab4f16f9e
9 changed files with 53 additions and 30 deletions

View File

@ -754,7 +754,7 @@ static ssize_t tap_get_user(struct tap_queue *q, void *msg_control,
skb_zcopy_init(skb, msg_control); skb_zcopy_init(skb, msg_control);
} else if (msg_control) { } else if (msg_control) {
struct ubuf_info *uarg = msg_control; struct ubuf_info *uarg = msg_control;
uarg->callback(NULL, uarg, false); uarg->ops->complete(NULL, uarg, false);
} }
dev_queue_xmit(skb); dev_queue_xmit(skb);

View File

@ -1906,7 +1906,7 @@ static ssize_t tun_get_user(struct tun_struct *tun, struct tun_file *tfile,
skb_zcopy_init(skb, msg_control); skb_zcopy_init(skb, msg_control);
} else if (msg_control) { } else if (msg_control) {
struct ubuf_info *uarg = msg_control; struct ubuf_info *uarg = msg_control;
uarg->callback(NULL, uarg, false); uarg->ops->complete(NULL, uarg, false);
} }
skb_reset_network_header(skb); skb_reset_network_header(skb);

View File

@ -390,9 +390,8 @@ bool xenvif_rx_queue_tail(struct xenvif_queue *queue, struct sk_buff *skb);
void xenvif_carrier_on(struct xenvif *vif); void xenvif_carrier_on(struct xenvif *vif);
/* Callback from stack when TX packet can be released */ /* Callbacks from stack when TX packet can be released */
void xenvif_zerocopy_callback(struct sk_buff *skb, struct ubuf_info *ubuf, extern const struct ubuf_info_ops xenvif_ubuf_ops;
bool zerocopy_success);
static inline pending_ring_idx_t nr_pending_reqs(struct xenvif_queue *queue) static inline pending_ring_idx_t nr_pending_reqs(struct xenvif_queue *queue)
{ {

View File

@ -593,7 +593,7 @@ int xenvif_init_queue(struct xenvif_queue *queue)
for (i = 0; i < MAX_PENDING_REQS; i++) { for (i = 0; i < MAX_PENDING_REQS; i++) {
queue->pending_tx_info[i].callback_struct = (struct ubuf_info_msgzc) queue->pending_tx_info[i].callback_struct = (struct ubuf_info_msgzc)
{ { .callback = xenvif_zerocopy_callback }, { { .ops = &xenvif_ubuf_ops },
{ { .ctx = NULL, { { .ctx = NULL,
.desc = i } } }; .desc = i } } };
queue->grant_tx_handle[i] = NETBACK_INVALID_HANDLE; queue->grant_tx_handle[i] = NETBACK_INVALID_HANDLE;

View File

@ -1156,7 +1156,7 @@ static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *s
uarg = skb_shinfo(skb)->destructor_arg; uarg = skb_shinfo(skb)->destructor_arg;
/* increase inflight counter to offset decrement in callback */ /* increase inflight counter to offset decrement in callback */
atomic_inc(&queue->inflight_packets); atomic_inc(&queue->inflight_packets);
uarg->callback(NULL, uarg, true); uarg->ops->complete(NULL, uarg, true);
skb_shinfo(skb)->destructor_arg = NULL; skb_shinfo(skb)->destructor_arg = NULL;
/* Fill the skb with the new (local) frags. */ /* Fill the skb with the new (local) frags. */
@ -1278,7 +1278,8 @@ static int xenvif_tx_submit(struct xenvif_queue *queue)
return work_done; return work_done;
} }
void xenvif_zerocopy_callback(struct sk_buff *skb, struct ubuf_info *ubuf_base, static void xenvif_zerocopy_callback(struct sk_buff *skb,
struct ubuf_info *ubuf_base,
bool zerocopy_success) bool zerocopy_success)
{ {
unsigned long flags; unsigned long flags;
@ -1312,6 +1313,10 @@ void xenvif_zerocopy_callback(struct sk_buff *skb, struct ubuf_info *ubuf_base,
xenvif_skb_zerocopy_complete(queue); xenvif_skb_zerocopy_complete(queue);
} }
const struct ubuf_info_ops xenvif_ubuf_ops = {
.complete = xenvif_zerocopy_callback,
};
static inline void xenvif_tx_dealloc_action(struct xenvif_queue *queue) static inline void xenvif_tx_dealloc_action(struct xenvif_queue *queue)
{ {
struct gnttab_unmap_grant_ref *gop; struct gnttab_unmap_grant_ref *gop;

View File

@ -380,7 +380,7 @@ static void vhost_zerocopy_signal_used(struct vhost_net *net,
} }
} }
static void vhost_zerocopy_callback(struct sk_buff *skb, static void vhost_zerocopy_complete(struct sk_buff *skb,
struct ubuf_info *ubuf_base, bool success) struct ubuf_info *ubuf_base, bool success)
{ {
struct ubuf_info_msgzc *ubuf = uarg_to_msgzc(ubuf_base); struct ubuf_info_msgzc *ubuf = uarg_to_msgzc(ubuf_base);
@ -408,6 +408,10 @@ static void vhost_zerocopy_callback(struct sk_buff *skb,
rcu_read_unlock_bh(); rcu_read_unlock_bh();
} }
static const struct ubuf_info_ops vhost_ubuf_ops = {
.complete = vhost_zerocopy_complete,
};
static inline unsigned long busy_clock(void) static inline unsigned long busy_clock(void)
{ {
return local_clock() >> 10; return local_clock() >> 10;
@ -879,7 +883,7 @@ static void handle_tx_zerocopy(struct vhost_net *net, struct socket *sock)
vq->heads[nvq->upend_idx].len = VHOST_DMA_IN_PROGRESS; vq->heads[nvq->upend_idx].len = VHOST_DMA_IN_PROGRESS;
ubuf->ctx = nvq->ubufs; ubuf->ctx = nvq->ubufs;
ubuf->desc = nvq->upend_idx; ubuf->desc = nvq->upend_idx;
ubuf->ubuf.callback = vhost_zerocopy_callback; ubuf->ubuf.ops = &vhost_ubuf_ops;
ubuf->ubuf.flags = SKBFL_ZEROCOPY_FRAG; ubuf->ubuf.flags = SKBFL_ZEROCOPY_FRAG;
refcount_set(&ubuf->ubuf.refcnt, 1); refcount_set(&ubuf->ubuf.refcnt, 1);
msg.msg_control = &ctl; msg.msg_control = &ctl;

View File

@ -527,6 +527,11 @@ enum {
#define SKBFL_ALL_ZEROCOPY (SKBFL_ZEROCOPY_FRAG | SKBFL_PURE_ZEROCOPY | \ #define SKBFL_ALL_ZEROCOPY (SKBFL_ZEROCOPY_FRAG | SKBFL_PURE_ZEROCOPY | \
SKBFL_DONT_ORPHAN | SKBFL_MANAGED_FRAG_REFS) SKBFL_DONT_ORPHAN | SKBFL_MANAGED_FRAG_REFS)
struct ubuf_info_ops {
void (*complete)(struct sk_buff *, struct ubuf_info *,
bool zerocopy_success);
};
/* /*
* The callback notifies userspace to release buffers when skb DMA is done in * The callback notifies userspace to release buffers when skb DMA is done in
* lower device, the skb last reference should be 0 when calling this. * lower device, the skb last reference should be 0 when calling this.
@ -536,8 +541,7 @@ enum {
* The desc field is used to track userspace buffer index. * The desc field is used to track userspace buffer index.
*/ */
struct ubuf_info { struct ubuf_info {
void (*callback)(struct sk_buff *, struct ubuf_info *, const struct ubuf_info_ops *ops;
bool zerocopy_success);
refcount_t refcnt; refcount_t refcnt;
u8 flags; u8 flags;
}; };
@ -1662,14 +1666,13 @@ static inline void skb_set_end_offset(struct sk_buff *skb, unsigned int offset)
} }
#endif #endif
extern const struct ubuf_info_ops msg_zerocopy_ubuf_ops;
struct ubuf_info *msg_zerocopy_realloc(struct sock *sk, size_t size, struct ubuf_info *msg_zerocopy_realloc(struct sock *sk, size_t size,
struct ubuf_info *uarg); struct ubuf_info *uarg);
void msg_zerocopy_put_abort(struct ubuf_info *uarg, bool have_uref); void msg_zerocopy_put_abort(struct ubuf_info *uarg, bool have_uref);
void msg_zerocopy_callback(struct sk_buff *skb, struct ubuf_info *uarg,
bool success);
int __zerocopy_sg_from_iter(struct msghdr *msg, struct sock *sk, int __zerocopy_sg_from_iter(struct msghdr *msg, struct sock *sk,
struct sk_buff *skb, struct iov_iter *from, struct sk_buff *skb, struct iov_iter *from,
size_t length); size_t length);
@ -1757,13 +1760,13 @@ static inline void *skb_zcopy_get_nouarg(struct sk_buff *skb)
static inline void net_zcopy_put(struct ubuf_info *uarg) static inline void net_zcopy_put(struct ubuf_info *uarg)
{ {
if (uarg) if (uarg)
uarg->callback(NULL, uarg, true); uarg->ops->complete(NULL, uarg, true);
} }
static inline void net_zcopy_put_abort(struct ubuf_info *uarg, bool have_uref) static inline void net_zcopy_put_abort(struct ubuf_info *uarg, bool have_uref)
{ {
if (uarg) { if (uarg) {
if (uarg->callback == msg_zerocopy_callback) if (uarg->ops == &msg_zerocopy_ubuf_ops)
msg_zerocopy_put_abort(uarg, have_uref); msg_zerocopy_put_abort(uarg, have_uref);
else if (have_uref) else if (have_uref)
net_zcopy_put(uarg); net_zcopy_put(uarg);
@ -1777,7 +1780,7 @@ static inline void skb_zcopy_clear(struct sk_buff *skb, bool zerocopy_success)
if (uarg) { if (uarg) {
if (!skb_zcopy_is_nouarg(skb)) if (!skb_zcopy_is_nouarg(skb))
uarg->callback(skb, uarg, zerocopy_success); uarg->ops->complete(skb, uarg, zerocopy_success);
skb_shinfo(skb)->flags &= ~SKBFL_ALL_ZEROCOPY; skb_shinfo(skb)->flags &= ~SKBFL_ALL_ZEROCOPY;
} }

View File

@ -24,7 +24,7 @@ static void io_notif_complete_tw_ext(struct io_kiocb *notif, struct io_tw_state
io_req_task_complete(notif, ts); io_req_task_complete(notif, ts);
} }
static void io_tx_ubuf_callback(struct sk_buff *skb, struct ubuf_info *uarg, static void io_tx_ubuf_complete(struct sk_buff *skb, struct ubuf_info *uarg,
bool success) bool success)
{ {
struct io_notif_data *nd = container_of(uarg, struct io_notif_data, uarg); struct io_notif_data *nd = container_of(uarg, struct io_notif_data, uarg);
@ -45,19 +45,27 @@ static void io_tx_ubuf_callback_ext(struct sk_buff *skb, struct ubuf_info *uarg,
else if (!success && !nd->zc_copied) else if (!success && !nd->zc_copied)
WRITE_ONCE(nd->zc_copied, true); WRITE_ONCE(nd->zc_copied, true);
} }
io_tx_ubuf_callback(skb, uarg, success); io_tx_ubuf_complete(skb, uarg, success);
} }
static const struct ubuf_info_ops io_ubuf_ops = {
.complete = io_tx_ubuf_complete,
};
static const struct ubuf_info_ops io_ubuf_ops_ext = {
.complete = io_tx_ubuf_callback_ext,
};
void io_notif_set_extended(struct io_kiocb *notif) void io_notif_set_extended(struct io_kiocb *notif)
{ {
struct io_notif_data *nd = io_notif_to_data(notif); struct io_notif_data *nd = io_notif_to_data(notif);
if (nd->uarg.callback != io_tx_ubuf_callback_ext) { if (nd->uarg.ops != &io_ubuf_ops_ext) {
nd->account_pages = 0; nd->account_pages = 0;
nd->zc_report = false; nd->zc_report = false;
nd->zc_used = false; nd->zc_used = false;
nd->zc_copied = false; nd->zc_copied = false;
nd->uarg.callback = io_tx_ubuf_callback_ext; nd->uarg.ops = &io_ubuf_ops_ext;
notif->io_task_work.func = io_notif_complete_tw_ext; notif->io_task_work.func = io_notif_complete_tw_ext;
} }
} }
@ -80,7 +88,7 @@ struct io_kiocb *io_alloc_notif(struct io_ring_ctx *ctx)
nd = io_notif_to_data(notif); nd = io_notif_to_data(notif);
nd->uarg.flags = IO_NOTIF_UBUF_FLAGS; nd->uarg.flags = IO_NOTIF_UBUF_FLAGS;
nd->uarg.callback = io_tx_ubuf_callback; nd->uarg.ops = &io_ubuf_ops;
refcount_set(&nd->uarg.refcnt, 1); refcount_set(&nd->uarg.refcnt, 1);
return notif; return notif;
} }

View File

@ -1708,7 +1708,7 @@ static struct ubuf_info *msg_zerocopy_alloc(struct sock *sk, size_t size)
return NULL; return NULL;
} }
uarg->ubuf.callback = msg_zerocopy_callback; uarg->ubuf.ops = &msg_zerocopy_ubuf_ops;
uarg->id = ((u32)atomic_inc_return(&sk->sk_zckey)) - 1; uarg->id = ((u32)atomic_inc_return(&sk->sk_zckey)) - 1;
uarg->len = 1; uarg->len = 1;
uarg->bytelen = size; uarg->bytelen = size;
@ -1734,7 +1734,7 @@ struct ubuf_info *msg_zerocopy_realloc(struct sock *sk, size_t size,
u32 bytelen, next; u32 bytelen, next;
/* there might be non MSG_ZEROCOPY users */ /* there might be non MSG_ZEROCOPY users */
if (uarg->callback != msg_zerocopy_callback) if (uarg->ops != &msg_zerocopy_ubuf_ops)
return NULL; return NULL;
/* realloc only when socket is locked (TCP, UDP cork), /* realloc only when socket is locked (TCP, UDP cork),
@ -1845,7 +1845,7 @@ release:
sock_put(sk); sock_put(sk);
} }
void msg_zerocopy_callback(struct sk_buff *skb, struct ubuf_info *uarg, static void msg_zerocopy_complete(struct sk_buff *skb, struct ubuf_info *uarg,
bool success) bool success)
{ {
struct ubuf_info_msgzc *uarg_zc = uarg_to_msgzc(uarg); struct ubuf_info_msgzc *uarg_zc = uarg_to_msgzc(uarg);
@ -1855,7 +1855,6 @@ void msg_zerocopy_callback(struct sk_buff *skb, struct ubuf_info *uarg,
if (refcount_dec_and_test(&uarg->refcnt)) if (refcount_dec_and_test(&uarg->refcnt))
__msg_zerocopy_callback(uarg_zc); __msg_zerocopy_callback(uarg_zc);
} }
EXPORT_SYMBOL_GPL(msg_zerocopy_callback);
void msg_zerocopy_put_abort(struct ubuf_info *uarg, bool have_uref) void msg_zerocopy_put_abort(struct ubuf_info *uarg, bool have_uref)
{ {
@ -1865,10 +1864,15 @@ void msg_zerocopy_put_abort(struct ubuf_info *uarg, bool have_uref)
uarg_to_msgzc(uarg)->len--; uarg_to_msgzc(uarg)->len--;
if (have_uref) if (have_uref)
msg_zerocopy_callback(NULL, uarg, true); msg_zerocopy_complete(NULL, uarg, true);
} }
EXPORT_SYMBOL_GPL(msg_zerocopy_put_abort); EXPORT_SYMBOL_GPL(msg_zerocopy_put_abort);
const struct ubuf_info_ops msg_zerocopy_ubuf_ops = {
.complete = msg_zerocopy_complete,
};
EXPORT_SYMBOL_GPL(msg_zerocopy_ubuf_ops);
int skb_zerocopy_iter_stream(struct sock *sk, struct sk_buff *skb, int skb_zerocopy_iter_stream(struct sock *sk, struct sk_buff *skb,
struct msghdr *msg, int len, struct msghdr *msg, int len,
struct ubuf_info *uarg) struct ubuf_info *uarg)