1 // SPDX-License-Identifier: GPL-2.0
2 /* Copyright (c) 2017 - 2018 Covalent IO, Inc. http://covalent.io */
4 #include <linux/skmsg.h>
5 #include <linux/skbuff.h>
6 #include <linux/scatterlist.h>
11 #include <trace/events/sock.h>
13 static bool sk_msg_try_coalesce_ok(struct sk_msg *msg, int elem_first_coalesce)
15 if (msg->sg.end > msg->sg.start &&
16 elem_first_coalesce < msg->sg.end)
19 if (msg->sg.end < msg->sg.start &&
20 (elem_first_coalesce > msg->sg.start ||
21 elem_first_coalesce < msg->sg.end))
27 int sk_msg_alloc(struct sock *sk, struct sk_msg *msg, int len,
28 int elem_first_coalesce)
30 struct page_frag *pfrag = sk_page_frag(sk);
31 u32 osize = msg->sg.size;
36 struct scatterlist *sge;
40 if (!sk_page_frag_refill(sk, pfrag)) {
45 orig_offset = pfrag->offset;
46 use = min_t(int, len, pfrag->size - orig_offset);
47 if (!sk_wmem_schedule(sk, use)) {
53 sk_msg_iter_var_prev(i);
54 sge = &msg->sg.data[i];
56 if (sk_msg_try_coalesce_ok(msg, elem_first_coalesce) &&
57 sg_page(sge) == pfrag->page &&
58 sge->offset + sge->length == orig_offset) {
61 if (sk_msg_full(msg)) {
66 sge = &msg->sg.data[msg->sg.end];
68 sg_set_page(sge, pfrag->page, use, orig_offset);
69 get_page(pfrag->page);
70 sk_msg_iter_next(msg, end);
73 sk_mem_charge(sk, use);
82 sk_msg_trim(sk, msg, osize);
85 EXPORT_SYMBOL_GPL(sk_msg_alloc);
87 int sk_msg_clone(struct sock *sk, struct sk_msg *dst, struct sk_msg *src,
90 int i = src->sg.start;
91 struct scatterlist *sge = sk_msg_elem(src, i);
92 struct scatterlist *sgd = NULL;
96 if (sge->length > off)
99 sk_msg_iter_var_next(i);
100 if (i == src->sg.end && off)
102 sge = sk_msg_elem(src, i);
106 sge_len = sge->length - off;
111 sgd = sk_msg_elem(dst, dst->sg.end - 1);
114 (sg_page(sge) == sg_page(sgd)) &&
115 (sg_virt(sge) + off == sg_virt(sgd) + sgd->length)) {
116 sgd->length += sge_len;
117 dst->sg.size += sge_len;
118 } else if (!sk_msg_full(dst)) {
119 sge_off = sge->offset + off;
120 sk_msg_page_add(dst, sg_page(sge), sge_len, sge_off);
127 sk_mem_charge(sk, sge_len);
128 sk_msg_iter_var_next(i);
129 if (i == src->sg.end && len)
131 sge = sk_msg_elem(src, i);
136 EXPORT_SYMBOL_GPL(sk_msg_clone);
138 void sk_msg_return_zero(struct sock *sk, struct sk_msg *msg, int bytes)
140 int i = msg->sg.start;
143 struct scatterlist *sge = sk_msg_elem(msg, i);
145 if (bytes < sge->length) {
146 sge->length -= bytes;
147 sge->offset += bytes;
148 sk_mem_uncharge(sk, bytes);
152 sk_mem_uncharge(sk, sge->length);
153 bytes -= sge->length;
156 sk_msg_iter_var_next(i);
157 } while (bytes && i != msg->sg.end);
160 EXPORT_SYMBOL_GPL(sk_msg_return_zero);
162 void sk_msg_return(struct sock *sk, struct sk_msg *msg, int bytes)
164 int i = msg->sg.start;
167 struct scatterlist *sge = &msg->sg.data[i];
168 int uncharge = (bytes < sge->length) ? bytes : sge->length;
170 sk_mem_uncharge(sk, uncharge);
172 sk_msg_iter_var_next(i);
173 } while (i != msg->sg.end);
175 EXPORT_SYMBOL_GPL(sk_msg_return);
177 static int sk_msg_free_elem(struct sock *sk, struct sk_msg *msg, u32 i,
180 struct scatterlist *sge = sk_msg_elem(msg, i);
181 u32 len = sge->length;
183 /* When the skb owns the memory we free it from consume_skb path. */
186 sk_mem_uncharge(sk, len);
187 put_page(sg_page(sge));
189 memset(sge, 0, sizeof(*sge));
193 static int __sk_msg_free(struct sock *sk, struct sk_msg *msg, u32 i,
196 struct scatterlist *sge = sk_msg_elem(msg, i);
199 while (msg->sg.size) {
200 msg->sg.size -= sge->length;
201 freed += sk_msg_free_elem(sk, msg, i, charge);
202 sk_msg_iter_var_next(i);
203 sk_msg_check_to_free(msg, i, msg->sg.size);
204 sge = sk_msg_elem(msg, i);
206 consume_skb(msg->skb);
211 int sk_msg_free_nocharge(struct sock *sk, struct sk_msg *msg)
213 return __sk_msg_free(sk, msg, msg->sg.start, false);
215 EXPORT_SYMBOL_GPL(sk_msg_free_nocharge);
217 int sk_msg_free(struct sock *sk, struct sk_msg *msg)
219 return __sk_msg_free(sk, msg, msg->sg.start, true);
221 EXPORT_SYMBOL_GPL(sk_msg_free);
223 static void __sk_msg_free_partial(struct sock *sk, struct sk_msg *msg,
224 u32 bytes, bool charge)
226 struct scatterlist *sge;
227 u32 i = msg->sg.start;
230 sge = sk_msg_elem(msg, i);
233 if (bytes < sge->length) {
235 sk_mem_uncharge(sk, bytes);
236 sge->length -= bytes;
237 sge->offset += bytes;
238 msg->sg.size -= bytes;
242 msg->sg.size -= sge->length;
243 bytes -= sge->length;
244 sk_msg_free_elem(sk, msg, i, charge);
245 sk_msg_iter_var_next(i);
246 sk_msg_check_to_free(msg, i, bytes);
251 void sk_msg_free_partial(struct sock *sk, struct sk_msg *msg, u32 bytes)
253 __sk_msg_free_partial(sk, msg, bytes, true);
255 EXPORT_SYMBOL_GPL(sk_msg_free_partial);
257 void sk_msg_free_partial_nocharge(struct sock *sk, struct sk_msg *msg,
260 __sk_msg_free_partial(sk, msg, bytes, false);
263 void sk_msg_trim(struct sock *sk, struct sk_msg *msg, int len)
265 int trim = msg->sg.size - len;
273 sk_msg_iter_var_prev(i);
275 while (msg->sg.data[i].length &&
276 trim >= msg->sg.data[i].length) {
277 trim -= msg->sg.data[i].length;
278 sk_msg_free_elem(sk, msg, i, true);
279 sk_msg_iter_var_prev(i);
284 msg->sg.data[i].length -= trim;
285 sk_mem_uncharge(sk, trim);
286 /* Adjust copybreak if it falls into the trimmed part of last buf */
287 if (msg->sg.curr == i && msg->sg.copybreak > msg->sg.data[i].length)
288 msg->sg.copybreak = msg->sg.data[i].length;
290 sk_msg_iter_var_next(i);
293 /* If we trim data a full sg elem before curr pointer update
294 * copybreak and current so that any future copy operations
295 * start at new copy location.
296 * However trimed data that has not yet been used in a copy op
297 * does not require an update.
300 msg->sg.curr = msg->sg.start;
301 msg->sg.copybreak = 0;
302 } else if (sk_msg_iter_dist(msg->sg.start, msg->sg.curr) >=
303 sk_msg_iter_dist(msg->sg.start, msg->sg.end)) {
304 sk_msg_iter_var_prev(i);
306 msg->sg.copybreak = msg->sg.data[i].length;
309 EXPORT_SYMBOL_GPL(sk_msg_trim);
311 int sk_msg_zerocopy_from_iter(struct sock *sk, struct iov_iter *from,
312 struct sk_msg *msg, u32 bytes)
314 int i, maxpages, ret = 0, num_elems = sk_msg_elem_used(msg);
315 const int to_max_pages = MAX_MSG_FRAGS;
316 struct page *pages[MAX_MSG_FRAGS];
317 ssize_t orig, copied, use, offset;
322 maxpages = to_max_pages - num_elems;
328 copied = iov_iter_get_pages2(from, pages, bytes, maxpages,
336 msg->sg.size += copied;
339 use = min_t(int, copied, PAGE_SIZE - offset);
340 sg_set_page(&msg->sg.data[msg->sg.end],
341 pages[i], use, offset);
342 sg_unmark_end(&msg->sg.data[msg->sg.end]);
343 sk_mem_charge(sk, use);
347 sk_msg_iter_next(msg, end);
351 /* When zerocopy is mixed with sk_msg_*copy* operations we
352 * may have a copybreak set in this case clear and prefer
353 * zerocopy remainder when possible.
355 msg->sg.copybreak = 0;
356 msg->sg.curr = msg->sg.end;
359 /* Revert iov_iter updates, msg will need to use 'trim' later if it
360 * also needs to be cleared.
363 iov_iter_revert(from, msg->sg.size - orig);
366 EXPORT_SYMBOL_GPL(sk_msg_zerocopy_from_iter);
368 int sk_msg_memcopy_from_iter(struct sock *sk, struct iov_iter *from,
369 struct sk_msg *msg, u32 bytes)
371 int ret = -ENOSPC, i = msg->sg.curr;
372 struct scatterlist *sge;
377 sge = sk_msg_elem(msg, i);
378 /* This is possible if a trim operation shrunk the buffer */
379 if (msg->sg.copybreak >= sge->length) {
380 msg->sg.copybreak = 0;
381 sk_msg_iter_var_next(i);
382 if (i == msg->sg.end)
384 sge = sk_msg_elem(msg, i);
387 buf_size = sge->length - msg->sg.copybreak;
388 copy = (buf_size > bytes) ? bytes : buf_size;
389 to = sg_virt(sge) + msg->sg.copybreak;
390 msg->sg.copybreak += copy;
391 if (sk->sk_route_caps & NETIF_F_NOCACHE_COPY)
392 ret = copy_from_iter_nocache(to, copy, from);
394 ret = copy_from_iter(to, copy, from);
402 msg->sg.copybreak = 0;
403 sk_msg_iter_var_next(i);
404 } while (i != msg->sg.end);
409 EXPORT_SYMBOL_GPL(sk_msg_memcopy_from_iter);
411 /* Receive sk_msg from psock->ingress_msg to @msg. */
412 int sk_msg_recvmsg(struct sock *sk, struct sk_psock *psock, struct msghdr *msg,
415 struct iov_iter *iter = &msg->msg_iter;
416 int peek = flags & MSG_PEEK;
417 struct sk_msg *msg_rx;
420 msg_rx = sk_psock_peek_msg(psock);
421 while (copied != len) {
422 struct scatterlist *sge;
424 if (unlikely(!msg_rx))
427 i = msg_rx->sg.start;
432 sge = sk_msg_elem(msg_rx, i);
435 if (copied + copy > len)
437 copy = copy_page_to_iter(page, sge->offset, copy, iter);
439 copied = copied ? copied : -EFAULT;
448 sk_mem_uncharge(sk, copy);
449 msg_rx->sg.size -= copy;
452 sk_msg_iter_var_next(i);
457 /* Lets not optimize peek case if copy_page_to_iter
458 * didn't copy the entire length lets just break.
460 if (copy != sge->length)
462 sk_msg_iter_var_next(i);
467 } while ((i != msg_rx->sg.end) && !sg_is_last(sge));
469 if (unlikely(peek)) {
470 msg_rx = sk_psock_next_msg(psock, msg_rx);
476 msg_rx->sg.start = i;
477 if (!sge->length && (i == msg_rx->sg.end || sg_is_last(sge))) {
478 msg_rx = sk_psock_dequeue_msg(psock);
479 kfree_sk_msg(msg_rx);
481 msg_rx = sk_psock_peek_msg(psock);
486 EXPORT_SYMBOL_GPL(sk_msg_recvmsg);
488 bool sk_msg_is_readable(struct sock *sk)
490 struct sk_psock *psock;
494 psock = sk_psock(sk);
496 empty = list_empty(&psock->ingress_msg);
500 EXPORT_SYMBOL_GPL(sk_msg_is_readable);
502 static struct sk_msg *alloc_sk_msg(gfp_t gfp)
506 msg = kzalloc(sizeof(*msg), gfp | __GFP_NOWARN);
509 sg_init_marker(msg->sg.data, NR_MSG_FRAG_IDS);
513 static struct sk_msg *sk_psock_create_ingress_msg(struct sock *sk,
516 if (atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf)
519 if (!sk_rmem_schedule(sk, skb, skb->truesize))
522 return alloc_sk_msg(GFP_KERNEL);
525 static int sk_psock_skb_ingress_enqueue(struct sk_buff *skb,
527 struct sk_psock *psock,
533 num_sge = skb_to_sgvec(skb, msg->sg.data, off, len);
535 /* skb linearize may fail with ENOMEM, but lets simply try again
536 * later if this happens. Under memory pressure we don't want to
537 * drop the skb. We need to linearize the skb so that the mapping
538 * in skb_to_sgvec can not error.
540 if (skb_linearize(skb))
543 num_sge = skb_to_sgvec(skb, msg->sg.data, off, len);
544 if (unlikely(num_sge < 0))
550 msg->sg.size = copied;
551 msg->sg.end = num_sge;
554 sk_psock_queue_msg(psock, msg);
555 sk_psock_data_ready(sk, psock);
559 static int sk_psock_skb_ingress_self(struct sk_psock *psock, struct sk_buff *skb,
562 static int sk_psock_skb_ingress(struct sk_psock *psock, struct sk_buff *skb,
565 struct sock *sk = psock->sk;
569 /* If we are receiving on the same sock skb->sk is already assigned,
570 * skip memory accounting and owner transition seeing it already set
573 if (unlikely(skb->sk == sk))
574 return sk_psock_skb_ingress_self(psock, skb, off, len);
575 msg = sk_psock_create_ingress_msg(sk, skb);
579 /* This will transition ownership of the data from the socket where
580 * the BPF program was run initiating the redirect to the socket
581 * we will eventually receive this data on. The data will be released
582 * from skb_consume found in __tcp_bpf_recvmsg() after its been copied
585 skb_set_owner_r(skb, sk);
586 err = sk_psock_skb_ingress_enqueue(skb, off, len, psock, sk, msg);
592 /* Puts an skb on the ingress queue of the socket already assigned to the
593 * skb. In this case we do not need to check memory limits or skb_set_owner_r
594 * because the skb is already accounted for here.
596 static int sk_psock_skb_ingress_self(struct sk_psock *psock, struct sk_buff *skb,
599 struct sk_msg *msg = alloc_sk_msg(GFP_ATOMIC);
600 struct sock *sk = psock->sk;
605 skb_set_owner_r(skb, sk);
606 err = sk_psock_skb_ingress_enqueue(skb, off, len, psock, sk, msg);
612 static int sk_psock_handle_skb(struct sk_psock *psock, struct sk_buff *skb,
613 u32 off, u32 len, bool ingress)
618 if (!sock_writeable(psock->sk))
620 return skb_send_sock(psock->sk, skb, off, len);
623 err = sk_psock_skb_ingress(psock, skb, off, len);
629 static void sk_psock_skb_state(struct sk_psock *psock,
630 struct sk_psock_work_state *state,
633 spin_lock_bh(&psock->ingress_lock);
634 if (sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED)) {
638 spin_unlock_bh(&psock->ingress_lock);
641 static void sk_psock_backlog(struct work_struct *work)
643 struct delayed_work *dwork = to_delayed_work(work);
644 struct sk_psock *psock = container_of(dwork, struct sk_psock, work);
645 struct sk_psock_work_state *state = &psock->work_state;
646 struct sk_buff *skb = NULL;
647 u32 len = 0, off = 0;
651 mutex_lock(&psock->work_mutex);
652 if (unlikely(state->len)) {
657 while ((skb = skb_peek(&psock->ingress_skb))) {
660 if (skb_bpf_strparser(skb)) {
661 struct strp_msg *stm = strp_msg(skb);
666 ingress = skb_bpf_ingress(skb);
667 skb_bpf_redirect_clear(skb);
670 if (!sock_flag(psock->sk, SOCK_DEAD))
671 ret = sk_psock_handle_skb(psock, skb, off,
674 if (ret == -EAGAIN) {
675 sk_psock_skb_state(psock, state, len, off);
677 /* Delay slightly to prioritize any
678 * other work that might be here.
680 if (sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED))
681 schedule_delayed_work(&psock->work, 1);
684 /* Hard errors break pipe and stop xmit. */
685 sk_psock_report_error(psock, ret ? -ret : EPIPE);
686 sk_psock_clear_state(psock, SK_PSOCK_TX_ENABLED);
693 skb = skb_dequeue(&psock->ingress_skb);
697 mutex_unlock(&psock->work_mutex);
700 struct sk_psock *sk_psock_init(struct sock *sk, int node)
702 struct sk_psock *psock;
705 write_lock_bh(&sk->sk_callback_lock);
707 if (sk_is_inet(sk) && inet_csk_has_ulp(sk)) {
708 psock = ERR_PTR(-EINVAL);
712 if (sk->sk_user_data) {
713 psock = ERR_PTR(-EBUSY);
717 psock = kzalloc_node(sizeof(*psock), GFP_ATOMIC | __GFP_NOWARN, node);
719 psock = ERR_PTR(-ENOMEM);
723 prot = READ_ONCE(sk->sk_prot);
725 psock->eval = __SK_NONE;
726 psock->sk_proto = prot;
727 psock->saved_unhash = prot->unhash;
728 psock->saved_destroy = prot->destroy;
729 psock->saved_close = prot->close;
730 psock->saved_write_space = sk->sk_write_space;
732 INIT_LIST_HEAD(&psock->link);
733 spin_lock_init(&psock->link_lock);
735 INIT_DELAYED_WORK(&psock->work, sk_psock_backlog);
736 mutex_init(&psock->work_mutex);
737 INIT_LIST_HEAD(&psock->ingress_msg);
738 spin_lock_init(&psock->ingress_lock);
739 skb_queue_head_init(&psock->ingress_skb);
741 sk_psock_set_state(psock, SK_PSOCK_TX_ENABLED);
742 refcount_set(&psock->refcnt, 1);
744 __rcu_assign_sk_user_data_with_flags(sk, psock,
745 SK_USER_DATA_NOCOPY |
750 write_unlock_bh(&sk->sk_callback_lock);
753 EXPORT_SYMBOL_GPL(sk_psock_init);
755 struct sk_psock_link *sk_psock_link_pop(struct sk_psock *psock)
757 struct sk_psock_link *link;
759 spin_lock_bh(&psock->link_lock);
760 link = list_first_entry_or_null(&psock->link, struct sk_psock_link,
763 list_del(&link->list);
764 spin_unlock_bh(&psock->link_lock);
768 static void __sk_psock_purge_ingress_msg(struct sk_psock *psock)
770 struct sk_msg *msg, *tmp;
772 list_for_each_entry_safe(msg, tmp, &psock->ingress_msg, list) {
773 list_del(&msg->list);
774 sk_msg_free(psock->sk, msg);
779 static void __sk_psock_zap_ingress(struct sk_psock *psock)
783 while ((skb = skb_dequeue(&psock->ingress_skb)) != NULL) {
784 skb_bpf_redirect_clear(skb);
785 sock_drop(psock->sk, skb);
787 __sk_psock_purge_ingress_msg(psock);
790 static void sk_psock_link_destroy(struct sk_psock *psock)
792 struct sk_psock_link *link, *tmp;
794 list_for_each_entry_safe(link, tmp, &psock->link, list) {
795 list_del(&link->list);
796 sk_psock_free_link(link);
800 void sk_psock_stop(struct sk_psock *psock)
802 spin_lock_bh(&psock->ingress_lock);
803 sk_psock_clear_state(psock, SK_PSOCK_TX_ENABLED);
804 sk_psock_cork_free(psock);
805 spin_unlock_bh(&psock->ingress_lock);
808 static void sk_psock_done_strp(struct sk_psock *psock);
810 static void sk_psock_destroy(struct work_struct *work)
812 struct sk_psock *psock = container_of(to_rcu_work(work),
813 struct sk_psock, rwork);
814 /* No sk_callback_lock since already detached. */
816 sk_psock_done_strp(psock);
818 cancel_delayed_work_sync(&psock->work);
819 __sk_psock_zap_ingress(psock);
820 mutex_destroy(&psock->work_mutex);
822 psock_progs_drop(&psock->progs);
824 sk_psock_link_destroy(psock);
825 sk_psock_cork_free(psock);
828 sock_put(psock->sk_redir);
833 void sk_psock_drop(struct sock *sk, struct sk_psock *psock)
835 write_lock_bh(&sk->sk_callback_lock);
836 sk_psock_restore_proto(sk, psock);
837 rcu_assign_sk_user_data(sk, NULL);
838 if (psock->progs.stream_parser)
839 sk_psock_stop_strp(sk, psock);
840 else if (psock->progs.stream_verdict || psock->progs.skb_verdict)
841 sk_psock_stop_verdict(sk, psock);
842 write_unlock_bh(&sk->sk_callback_lock);
844 sk_psock_stop(psock);
846 INIT_RCU_WORK(&psock->rwork, sk_psock_destroy);
847 queue_rcu_work(system_wq, &psock->rwork);
849 EXPORT_SYMBOL_GPL(sk_psock_drop);
851 static int sk_psock_map_verd(int verdict, bool redir)
855 return redir ? __SK_REDIRECT : __SK_PASS;
864 int sk_psock_msg_verdict(struct sock *sk, struct sk_psock *psock,
867 struct bpf_prog *prog;
871 prog = READ_ONCE(psock->progs.msg_parser);
872 if (unlikely(!prog)) {
877 sk_msg_compute_data_pointers(msg);
879 ret = bpf_prog_run_pin_on_cpu(prog, msg);
880 ret = sk_psock_map_verd(ret, msg->sk_redir);
881 psock->apply_bytes = msg->apply_bytes;
882 if (ret == __SK_REDIRECT) {
883 if (psock->sk_redir) {
884 sock_put(psock->sk_redir);
885 psock->sk_redir = NULL;
887 if (!msg->sk_redir) {
891 psock->redir_ingress = sk_msg_to_ingress(msg);
892 psock->sk_redir = msg->sk_redir;
893 sock_hold(psock->sk_redir);
899 EXPORT_SYMBOL_GPL(sk_psock_msg_verdict);
901 static int sk_psock_skb_redirect(struct sk_psock *from, struct sk_buff *skb)
903 struct sk_psock *psock_other;
904 struct sock *sk_other;
906 sk_other = skb_bpf_redirect_fetch(skb);
907 /* This error is a buggy BPF program, it returned a redirect
908 * return code, but then didn't set a redirect interface.
910 if (unlikely(!sk_other)) {
911 skb_bpf_redirect_clear(skb);
912 sock_drop(from->sk, skb);
915 psock_other = sk_psock(sk_other);
916 /* This error indicates the socket is being torn down or had another
917 * error that caused the pipe to break. We can't send a packet on
918 * a socket that is in this state so we drop the skb.
920 if (!psock_other || sock_flag(sk_other, SOCK_DEAD)) {
921 skb_bpf_redirect_clear(skb);
922 sock_drop(from->sk, skb);
925 spin_lock_bh(&psock_other->ingress_lock);
926 if (!sk_psock_test_state(psock_other, SK_PSOCK_TX_ENABLED)) {
927 spin_unlock_bh(&psock_other->ingress_lock);
928 skb_bpf_redirect_clear(skb);
929 sock_drop(from->sk, skb);
933 skb_queue_tail(&psock_other->ingress_skb, skb);
934 schedule_delayed_work(&psock_other->work, 0);
935 spin_unlock_bh(&psock_other->ingress_lock);
939 static void sk_psock_tls_verdict_apply(struct sk_buff *skb,
940 struct sk_psock *from, int verdict)
944 sk_psock_skb_redirect(from, skb);
953 int sk_psock_tls_strp_read(struct sk_psock *psock, struct sk_buff *skb)
955 struct bpf_prog *prog;
959 prog = READ_ONCE(psock->progs.stream_verdict);
963 skb_bpf_redirect_clear(skb);
964 ret = bpf_prog_run_pin_on_cpu(prog, skb);
965 ret = sk_psock_map_verd(ret, skb_bpf_redirect_fetch(skb));
968 sk_psock_tls_verdict_apply(skb, psock, ret);
972 EXPORT_SYMBOL_GPL(sk_psock_tls_strp_read);
974 static int sk_psock_verdict_apply(struct sk_psock *psock, struct sk_buff *skb,
977 struct sock *sk_other;
984 sk_other = psock->sk;
985 if (sock_flag(sk_other, SOCK_DEAD) ||
986 !sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED))
989 skb_bpf_set_ingress(skb);
991 /* If the queue is empty then we can submit directly
992 * into the msg queue. If its not empty we have to
993 * queue work otherwise we may get OOO data. Otherwise,
994 * if sk_psock_skb_ingress errors will be handled by
995 * retrying later from workqueue.
997 if (skb_queue_empty(&psock->ingress_skb)) {
1000 if (skb_bpf_strparser(skb)) {
1001 struct strp_msg *stm = strp_msg(skb);
1004 len = stm->full_len;
1006 err = sk_psock_skb_ingress_self(psock, skb, off, len);
1009 spin_lock_bh(&psock->ingress_lock);
1010 if (sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED)) {
1011 skb_queue_tail(&psock->ingress_skb, skb);
1012 schedule_delayed_work(&psock->work, 0);
1015 spin_unlock_bh(&psock->ingress_lock);
1021 tcp_eat_skb(psock->sk, skb);
1022 err = sk_psock_skb_redirect(psock, skb);
1027 skb_bpf_redirect_clear(skb);
1028 tcp_eat_skb(psock->sk, skb);
1029 sock_drop(psock->sk, skb);
1035 static void sk_psock_write_space(struct sock *sk)
1037 struct sk_psock *psock;
1038 void (*write_space)(struct sock *sk) = NULL;
1041 psock = sk_psock(sk);
1042 if (likely(psock)) {
1043 if (sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED))
1044 schedule_delayed_work(&psock->work, 0);
1045 write_space = psock->saved_write_space;
1052 #if IS_ENABLED(CONFIG_BPF_STREAM_PARSER)
1053 static void sk_psock_strp_read(struct strparser *strp, struct sk_buff *skb)
1055 struct sk_psock *psock;
1056 struct bpf_prog *prog;
1057 int ret = __SK_DROP;
1062 psock = sk_psock(sk);
1063 if (unlikely(!psock)) {
1067 prog = READ_ONCE(psock->progs.stream_verdict);
1071 skb_bpf_redirect_clear(skb);
1072 ret = bpf_prog_run_pin_on_cpu(prog, skb);
1073 skb_bpf_set_strparser(skb);
1074 ret = sk_psock_map_verd(ret, skb_bpf_redirect_fetch(skb));
1077 sk_psock_verdict_apply(psock, skb, ret);
1082 static int sk_psock_strp_read_done(struct strparser *strp, int err)
1087 static int sk_psock_strp_parse(struct strparser *strp, struct sk_buff *skb)
1089 struct sk_psock *psock = container_of(strp, struct sk_psock, strp);
1090 struct bpf_prog *prog;
1094 prog = READ_ONCE(psock->progs.stream_parser);
1096 skb->sk = psock->sk;
1097 ret = bpf_prog_run_pin_on_cpu(prog, skb);
1104 /* Called with socket lock held. */
1105 static void sk_psock_strp_data_ready(struct sock *sk)
1107 struct sk_psock *psock;
1109 trace_sk_data_ready(sk);
1112 psock = sk_psock(sk);
1113 if (likely(psock)) {
1114 if (tls_sw_has_ctx_rx(sk)) {
1115 psock->saved_data_ready(sk);
1117 write_lock_bh(&sk->sk_callback_lock);
1118 strp_data_ready(&psock->strp);
1119 write_unlock_bh(&sk->sk_callback_lock);
1125 int sk_psock_init_strp(struct sock *sk, struct sk_psock *psock)
1129 static const struct strp_callbacks cb = {
1130 .rcv_msg = sk_psock_strp_read,
1131 .read_sock_done = sk_psock_strp_read_done,
1132 .parse_msg = sk_psock_strp_parse,
1135 ret = strp_init(&psock->strp, sk, &cb);
1137 sk_psock_set_state(psock, SK_PSOCK_RX_STRP_ENABLED);
1142 void sk_psock_start_strp(struct sock *sk, struct sk_psock *psock)
1144 if (psock->saved_data_ready)
1147 psock->saved_data_ready = sk->sk_data_ready;
1148 sk->sk_data_ready = sk_psock_strp_data_ready;
1149 sk->sk_write_space = sk_psock_write_space;
1152 void sk_psock_stop_strp(struct sock *sk, struct sk_psock *psock)
1154 psock_set_prog(&psock->progs.stream_parser, NULL);
1156 if (!psock->saved_data_ready)
1159 sk->sk_data_ready = psock->saved_data_ready;
1160 psock->saved_data_ready = NULL;
1161 strp_stop(&psock->strp);
1164 static void sk_psock_done_strp(struct sk_psock *psock)
1166 /* Parser has been stopped */
1167 if (sk_psock_test_state(psock, SK_PSOCK_RX_STRP_ENABLED))
1168 strp_done(&psock->strp);
1171 static void sk_psock_done_strp(struct sk_psock *psock)
1174 #endif /* CONFIG_BPF_STREAM_PARSER */
1176 static int sk_psock_verdict_recv(struct sock *sk, struct sk_buff *skb)
1178 struct sk_psock *psock;
1179 struct bpf_prog *prog;
1180 int ret = __SK_DROP;
1184 psock = sk_psock(sk);
1185 if (unlikely(!psock)) {
1187 tcp_eat_skb(sk, skb);
1191 prog = READ_ONCE(psock->progs.stream_verdict);
1193 prog = READ_ONCE(psock->progs.skb_verdict);
1196 skb_bpf_redirect_clear(skb);
1197 ret = bpf_prog_run_pin_on_cpu(prog, skb);
1198 ret = sk_psock_map_verd(ret, skb_bpf_redirect_fetch(skb));
1200 ret = sk_psock_verdict_apply(psock, skb, ret);
1208 static void sk_psock_verdict_data_ready(struct sock *sk)
1210 struct socket *sock = sk->sk_socket;
1211 const struct proto_ops *ops;
1214 trace_sk_data_ready(sk);
1216 if (unlikely(!sock))
1218 ops = READ_ONCE(sock->ops);
1219 if (!ops || !ops->read_skb)
1221 copied = ops->read_skb(sk, sk_psock_verdict_recv);
1223 struct sk_psock *psock;
1226 psock = sk_psock(sk);
1228 psock->saved_data_ready(sk);
1233 void sk_psock_start_verdict(struct sock *sk, struct sk_psock *psock)
1235 if (psock->saved_data_ready)
1238 psock->saved_data_ready = sk->sk_data_ready;
1239 sk->sk_data_ready = sk_psock_verdict_data_ready;
1240 sk->sk_write_space = sk_psock_write_space;
1243 void sk_psock_stop_verdict(struct sock *sk, struct sk_psock *psock)
1245 psock_set_prog(&psock->progs.stream_verdict, NULL);
1246 psock_set_prog(&psock->progs.skb_verdict, NULL);
1248 if (!psock->saved_data_ready)
1251 sk->sk_data_ready = psock->saved_data_ready;
1252 psock->saved_data_ready = NULL;