2 * Back-end of the driver for virtual network devices. This portion of the
3 * driver exports a 'unified' network-device interface that can be accessed
4 * by any operating system that implements a compatible front end. A
5 * reference front-end implementation can be found in:
6 * drivers/net/xen-netfront.c
8 * Copyright (c) 2002-2005, K A Fraser
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public License version 2
12 * as published by the Free Software Foundation; or, when distributed
13 * separately from the Linux kernel or incorporated into other
14 * software packages, subject to the following license:
16 * Permission is hereby granted, free of charge, to any person obtaining a copy
17 * of this source file (the "Software"), to deal in the Software without
18 * restriction, including without limitation the rights to use, copy, modify,
19 * merge, publish, distribute, sublicense, and/or sell copies of the Software,
20 * and to permit persons to whom the Software is furnished to do so, subject to
21 * the following conditions:
23 * The above copyright notice and this permission notice shall be included in
24 * all copies or substantial portions of the Software.
26 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
27 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
28 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
29 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
30 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
31 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
37 #include <linux/kthread.h>
38 #include <linux/if_vlan.h>
39 #include <linux/udp.h>
40 #include <linux/highmem.h>
45 #include <xen/events.h>
46 #include <xen/interface/memory.h>
49 #include <asm/xen/hypercall.h>
51 /* Provide an option to disable split event channels at load time as
52 * event channels are limited resource. Split event channels are
55 bool separate_tx_rx_irq = true;
56 module_param(separate_tx_rx_irq, bool, 0644);
58 /* The time that packets can stay on the guest Rx internal queue
59 * before they are dropped.
61 unsigned int rx_drain_timeout_msecs = 10000;
62 module_param(rx_drain_timeout_msecs, uint, 0444);
64 /* The length of time before the frontend is considered unresponsive
65 * because it isn't providing Rx slots.
67 unsigned int rx_stall_timeout_msecs = 60000;
68 module_param(rx_stall_timeout_msecs, uint, 0444);
70 unsigned int xenvif_max_queues;
71 module_param_named(max_queues, xenvif_max_queues, uint, 0644);
72 MODULE_PARM_DESC(max_queues,
73 "Maximum number of queues per virtual interface");
76 * This is the maximum slots a skb can have. If a guest sends a skb
77 * which exceeds this limit it is considered malicious.
79 #define FATAL_SKB_SLOTS_DEFAULT 20
80 static unsigned int fatal_skb_slots = FATAL_SKB_SLOTS_DEFAULT;
81 module_param(fatal_skb_slots, uint, 0444);
83 /* The amount to copy out of the first guest Tx slot into the skb's
84 * linear area. If the first slot has more data, it will be mapped
85 * and put into the first frag.
87 * This is sized to avoid pulling headers from the frags for most
90 #define XEN_NETBACK_TX_COPY_LEN 128
93 static void xenvif_idx_release(struct xenvif_queue *queue, u16 pending_idx,
96 static void make_tx_response(struct xenvif_queue *queue,
97 struct xen_netif_tx_request *txp,
99 static void push_tx_responses(struct xenvif_queue *queue);
101 static inline int tx_work_todo(struct xenvif_queue *queue);
103 static struct xen_netif_rx_response *make_rx_response(struct xenvif_queue *queue,
110 static inline unsigned long idx_to_pfn(struct xenvif_queue *queue,
113 return page_to_pfn(queue->mmap_pages[idx]);
116 static inline unsigned long idx_to_kaddr(struct xenvif_queue *queue,
119 return (unsigned long)pfn_to_kaddr(idx_to_pfn(queue, idx));
122 #define callback_param(vif, pending_idx) \
123 (vif->pending_tx_info[pending_idx].callback_struct)
125 /* Find the containing VIF's structure from a pointer in pending_tx_info array
127 static inline struct xenvif_queue *ubuf_to_queue(const struct ubuf_info *ubuf)
129 u16 pending_idx = ubuf->desc;
130 struct pending_tx_info *temp =
131 container_of(ubuf, struct pending_tx_info, callback_struct);
132 return container_of(temp - pending_idx,
137 static u16 frag_get_pending_idx(skb_frag_t *frag)
139 return (u16)frag->page_offset;
142 static void frag_set_pending_idx(skb_frag_t *frag, u16 pending_idx)
144 frag->page_offset = pending_idx;
147 static inline pending_ring_idx_t pending_index(unsigned i)
149 return i & (MAX_PENDING_REQS-1);
152 bool xenvif_rx_ring_slots_available(struct xenvif_queue *queue, int needed)
157 prod = queue->rx.sring->req_prod;
158 cons = queue->rx.req_cons;
160 if (prod - cons >= needed)
163 queue->rx.sring->req_event = prod + 1;
165 /* Make sure event is visible before we check prod
169 } while (queue->rx.sring->req_prod != prod);
174 void xenvif_rx_queue_tail(struct xenvif_queue *queue, struct sk_buff *skb)
178 spin_lock_irqsave(&queue->rx_queue.lock, flags);
180 __skb_queue_tail(&queue->rx_queue, skb);
182 queue->rx_queue_len += skb->len;
183 if (queue->rx_queue_len > queue->rx_queue_max)
184 netif_tx_stop_queue(netdev_get_tx_queue(queue->vif->dev, queue->id));
186 spin_unlock_irqrestore(&queue->rx_queue.lock, flags);
189 static struct sk_buff *xenvif_rx_dequeue(struct xenvif_queue *queue)
193 spin_lock_irq(&queue->rx_queue.lock);
195 skb = __skb_dequeue(&queue->rx_queue);
197 queue->rx_queue_len -= skb->len;
199 spin_unlock_irq(&queue->rx_queue.lock);
204 static void xenvif_rx_queue_maybe_wake(struct xenvif_queue *queue)
206 spin_lock_irq(&queue->rx_queue.lock);
208 if (queue->rx_queue_len < queue->rx_queue_max)
209 netif_tx_wake_queue(netdev_get_tx_queue(queue->vif->dev, queue->id));
211 spin_unlock_irq(&queue->rx_queue.lock);
215 static void xenvif_rx_queue_purge(struct xenvif_queue *queue)
218 while ((skb = xenvif_rx_dequeue(queue)) != NULL)
222 static void xenvif_rx_queue_drop_expired(struct xenvif_queue *queue)
227 skb = skb_peek(&queue->rx_queue);
230 if (time_before(jiffies, XENVIF_RX_CB(skb)->expires))
232 xenvif_rx_dequeue(queue);
237 struct netrx_pending_operations {
238 unsigned copy_prod, copy_cons;
239 unsigned meta_prod, meta_cons;
240 struct gnttab_copy *copy;
241 struct xenvif_rx_meta *meta;
243 grant_ref_t copy_gref;
246 static struct xenvif_rx_meta *get_next_rx_buffer(struct xenvif_queue *queue,
247 struct netrx_pending_operations *npo)
249 struct xenvif_rx_meta *meta;
250 struct xen_netif_rx_request *req;
252 req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
254 meta = npo->meta + npo->meta_prod++;
255 meta->gso_type = XEN_NETIF_GSO_TYPE_NONE;
261 npo->copy_gref = req->gref;
267 * Set up the grant operations for this fragment. If it's a flipping
268 * interface, we also set up the unmap request from here.
270 static void xenvif_gop_frag_copy(struct xenvif_queue *queue, struct sk_buff *skb,
271 struct netrx_pending_operations *npo,
272 struct page *page, unsigned long size,
273 unsigned long offset, int *head)
275 struct gnttab_copy *copy_gop;
276 struct xenvif_rx_meta *meta;
278 int gso_type = XEN_NETIF_GSO_TYPE_NONE;
280 /* Data must not cross a page boundary. */
281 BUG_ON(size + offset > PAGE_SIZE<<compound_order(page));
283 meta = npo->meta + npo->meta_prod - 1;
285 /* Skip unused frames from start of page */
286 page += offset >> PAGE_SHIFT;
287 offset &= ~PAGE_MASK;
290 struct xen_page_foreign *foreign;
292 BUG_ON(offset >= PAGE_SIZE);
293 BUG_ON(npo->copy_off > MAX_BUFFER_OFFSET);
295 if (npo->copy_off == MAX_BUFFER_OFFSET)
296 meta = get_next_rx_buffer(queue, npo);
298 bytes = PAGE_SIZE - offset;
302 if (npo->copy_off + bytes > MAX_BUFFER_OFFSET)
303 bytes = MAX_BUFFER_OFFSET - npo->copy_off;
305 copy_gop = npo->copy + npo->copy_prod++;
306 copy_gop->flags = GNTCOPY_dest_gref;
307 copy_gop->len = bytes;
309 foreign = xen_page_foreign(page);
311 copy_gop->source.domid = foreign->domid;
312 copy_gop->source.u.ref = foreign->gref;
313 copy_gop->flags |= GNTCOPY_source_gref;
315 copy_gop->source.domid = DOMID_SELF;
316 copy_gop->source.u.gmfn =
317 virt_to_mfn(page_address(page));
319 copy_gop->source.offset = offset;
321 copy_gop->dest.domid = queue->vif->domid;
322 copy_gop->dest.offset = npo->copy_off;
323 copy_gop->dest.u.ref = npo->copy_gref;
325 npo->copy_off += bytes;
332 if (offset == PAGE_SIZE && size) {
333 BUG_ON(!PageCompound(page));
338 /* Leave a gap for the GSO descriptor. */
339 if (skb_is_gso(skb)) {
340 if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
341 gso_type = XEN_NETIF_GSO_TYPE_TCPV4;
342 else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
343 gso_type = XEN_NETIF_GSO_TYPE_TCPV6;
346 if (*head && ((1 << gso_type) & queue->vif->gso_mask))
347 queue->rx.req_cons++;
349 *head = 0; /* There must be something in this buffer now. */
355 * Prepare an SKB to be transmitted to the frontend.
357 * This function is responsible for allocating grant operations, meta
360 * It returns the number of meta structures consumed. The number of
361 * ring slots used is always equal to the number of meta slots used
362 * plus the number of GSO descriptors used. Currently, we use either
363 * zero GSO descriptors (for non-GSO packets) or one descriptor (for
364 * frontend-side LRO).
366 static int xenvif_gop_skb(struct sk_buff *skb,
367 struct netrx_pending_operations *npo,
368 struct xenvif_queue *queue)
370 struct xenvif *vif = netdev_priv(skb->dev);
371 int nr_frags = skb_shinfo(skb)->nr_frags;
373 struct xen_netif_rx_request *req;
374 struct xenvif_rx_meta *meta;
380 old_meta_prod = npo->meta_prod;
382 gso_type = XEN_NETIF_GSO_TYPE_NONE;
383 if (skb_is_gso(skb)) {
384 if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
385 gso_type = XEN_NETIF_GSO_TYPE_TCPV4;
386 else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
387 gso_type = XEN_NETIF_GSO_TYPE_TCPV6;
390 /* Set up a GSO prefix descriptor, if necessary */
391 if ((1 << gso_type) & vif->gso_prefix_mask) {
392 req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
393 meta = npo->meta + npo->meta_prod++;
394 meta->gso_type = gso_type;
395 meta->gso_size = skb_shinfo(skb)->gso_size;
400 req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
401 meta = npo->meta + npo->meta_prod++;
403 if ((1 << gso_type) & vif->gso_mask) {
404 meta->gso_type = gso_type;
405 meta->gso_size = skb_shinfo(skb)->gso_size;
407 meta->gso_type = XEN_NETIF_GSO_TYPE_NONE;
414 npo->copy_gref = req->gref;
417 while (data < skb_tail_pointer(skb)) {
418 unsigned int offset = offset_in_page(data);
419 unsigned int len = PAGE_SIZE - offset;
421 if (data + len > skb_tail_pointer(skb))
422 len = skb_tail_pointer(skb) - data;
424 xenvif_gop_frag_copy(queue, skb, npo,
425 virt_to_page(data), len, offset, &head);
429 for (i = 0; i < nr_frags; i++) {
430 xenvif_gop_frag_copy(queue, skb, npo,
431 skb_frag_page(&skb_shinfo(skb)->frags[i]),
432 skb_frag_size(&skb_shinfo(skb)->frags[i]),
433 skb_shinfo(skb)->frags[i].page_offset,
437 return npo->meta_prod - old_meta_prod;
441 * This is a twin to xenvif_gop_skb. Assume that xenvif_gop_skb was
442 * used to set up the operations on the top of
443 * netrx_pending_operations, which have since been done. Check that
444 * they didn't give any errors and advance over them.
446 static int xenvif_check_gop(struct xenvif *vif, int nr_meta_slots,
447 struct netrx_pending_operations *npo)
449 struct gnttab_copy *copy_op;
450 int status = XEN_NETIF_RSP_OKAY;
453 for (i = 0; i < nr_meta_slots; i++) {
454 copy_op = npo->copy + npo->copy_cons++;
455 if (copy_op->status != GNTST_okay) {
457 "Bad status %d from copy to DOM%d.\n",
458 copy_op->status, vif->domid);
459 status = XEN_NETIF_RSP_ERROR;
466 static void xenvif_add_frag_responses(struct xenvif_queue *queue, int status,
467 struct xenvif_rx_meta *meta,
471 unsigned long offset;
473 /* No fragments used */
474 if (nr_meta_slots <= 1)
479 for (i = 0; i < nr_meta_slots; i++) {
481 if (i == nr_meta_slots - 1)
484 flags = XEN_NETRXF_more_data;
487 make_rx_response(queue, meta[i].id, status, offset,
488 meta[i].size, flags);
492 void xenvif_kick_thread(struct xenvif_queue *queue)
497 static void xenvif_rx_action(struct xenvif_queue *queue)
501 struct xen_netif_rx_response *resp;
502 struct sk_buff_head rxq;
506 unsigned long offset;
507 bool need_to_notify = false;
509 struct netrx_pending_operations npo = {
510 .copy = queue->grant_copy_op,
514 skb_queue_head_init(&rxq);
516 while (xenvif_rx_ring_slots_available(queue, XEN_NETBK_RX_SLOTS_MAX)
517 && (skb = xenvif_rx_dequeue(queue)) != NULL) {
518 queue->last_rx_time = jiffies;
520 XENVIF_RX_CB(skb)->meta_slots_used = xenvif_gop_skb(skb, &npo, queue);
522 __skb_queue_tail(&rxq, skb);
525 BUG_ON(npo.meta_prod > ARRAY_SIZE(queue->meta));
530 BUG_ON(npo.copy_prod > MAX_GRANT_COPY_OPS);
531 gnttab_batch_copy(queue->grant_copy_op, npo.copy_prod);
533 while ((skb = __skb_dequeue(&rxq)) != NULL) {
535 if ((1 << queue->meta[npo.meta_cons].gso_type) &
536 queue->vif->gso_prefix_mask) {
537 resp = RING_GET_RESPONSE(&queue->rx,
538 queue->rx.rsp_prod_pvt++);
540 resp->flags = XEN_NETRXF_gso_prefix | XEN_NETRXF_more_data;
542 resp->offset = queue->meta[npo.meta_cons].gso_size;
543 resp->id = queue->meta[npo.meta_cons].id;
544 resp->status = XENVIF_RX_CB(skb)->meta_slots_used;
547 XENVIF_RX_CB(skb)->meta_slots_used--;
551 queue->stats.tx_bytes += skb->len;
552 queue->stats.tx_packets++;
554 status = xenvif_check_gop(queue->vif,
555 XENVIF_RX_CB(skb)->meta_slots_used,
558 if (XENVIF_RX_CB(skb)->meta_slots_used == 1)
561 flags = XEN_NETRXF_more_data;
563 if (skb->ip_summed == CHECKSUM_PARTIAL) /* local packet? */
564 flags |= XEN_NETRXF_csum_blank | XEN_NETRXF_data_validated;
565 else if (skb->ip_summed == CHECKSUM_UNNECESSARY)
566 /* remote but checksummed. */
567 flags |= XEN_NETRXF_data_validated;
570 resp = make_rx_response(queue, queue->meta[npo.meta_cons].id,
572 queue->meta[npo.meta_cons].size,
575 if ((1 << queue->meta[npo.meta_cons].gso_type) &
576 queue->vif->gso_mask) {
577 struct xen_netif_extra_info *gso =
578 (struct xen_netif_extra_info *)
579 RING_GET_RESPONSE(&queue->rx,
580 queue->rx.rsp_prod_pvt++);
582 resp->flags |= XEN_NETRXF_extra_info;
584 gso->u.gso.type = queue->meta[npo.meta_cons].gso_type;
585 gso->u.gso.size = queue->meta[npo.meta_cons].gso_size;
587 gso->u.gso.features = 0;
589 gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
593 xenvif_add_frag_responses(queue, status,
594 queue->meta + npo.meta_cons + 1,
595 XENVIF_RX_CB(skb)->meta_slots_used);
597 RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&queue->rx, ret);
599 need_to_notify |= !!ret;
601 npo.meta_cons += XENVIF_RX_CB(skb)->meta_slots_used;
607 notify_remote_via_irq(queue->rx_irq);
610 void xenvif_napi_schedule_or_enable_events(struct xenvif_queue *queue)
614 RING_FINAL_CHECK_FOR_REQUESTS(&queue->tx, more_to_do);
617 napi_schedule(&queue->napi);
620 static void tx_add_credit(struct xenvif_queue *queue)
622 unsigned long max_burst, max_credit;
625 * Allow a burst big enough to transmit a jumbo packet of up to 128kB.
626 * Otherwise the interface can seize up due to insufficient credit.
628 max_burst = RING_GET_REQUEST(&queue->tx, queue->tx.req_cons)->size;
629 max_burst = min(max_burst, 131072UL);
630 max_burst = max(max_burst, queue->credit_bytes);
632 /* Take care that adding a new chunk of credit doesn't wrap to zero. */
633 max_credit = queue->remaining_credit + queue->credit_bytes;
634 if (max_credit < queue->remaining_credit)
635 max_credit = ULONG_MAX; /* wrapped: clamp to ULONG_MAX */
637 queue->remaining_credit = min(max_credit, max_burst);
640 void xenvif_tx_credit_callback(unsigned long data)
642 struct xenvif_queue *queue = (struct xenvif_queue *)data;
643 tx_add_credit(queue);
644 xenvif_napi_schedule_or_enable_events(queue);
647 static void xenvif_tx_err(struct xenvif_queue *queue,
648 struct xen_netif_tx_request *txp, RING_IDX end)
650 RING_IDX cons = queue->tx.req_cons;
654 spin_lock_irqsave(&queue->response_lock, flags);
655 make_tx_response(queue, txp, XEN_NETIF_RSP_ERROR);
656 push_tx_responses(queue);
657 spin_unlock_irqrestore(&queue->response_lock, flags);
660 txp = RING_GET_REQUEST(&queue->tx, cons++);
662 queue->tx.req_cons = cons;
665 static void xenvif_fatal_tx_err(struct xenvif *vif)
667 netdev_err(vif->dev, "fatal error; disabling device\n");
668 vif->disabled = true;
669 /* Disable the vif from queue 0's kthread */
671 xenvif_kick_thread(&vif->queues[0]);
674 static int xenvif_count_requests(struct xenvif_queue *queue,
675 struct xen_netif_tx_request *first,
676 struct xen_netif_tx_request *txp,
679 RING_IDX cons = queue->tx.req_cons;
684 if (!(first->flags & XEN_NETTXF_more_data))
688 struct xen_netif_tx_request dropped_tx = { 0 };
690 if (slots >= work_to_do) {
691 netdev_err(queue->vif->dev,
692 "Asked for %d slots but exceeds this limit\n",
694 xenvif_fatal_tx_err(queue->vif);
698 /* This guest is really using too many slots and
699 * considered malicious.
701 if (unlikely(slots >= fatal_skb_slots)) {
702 netdev_err(queue->vif->dev,
703 "Malicious frontend using %d slots, threshold %u\n",
704 slots, fatal_skb_slots);
705 xenvif_fatal_tx_err(queue->vif);
709 /* Xen network protocol had implicit dependency on
710 * MAX_SKB_FRAGS. XEN_NETBK_LEGACY_SLOTS_MAX is set to
711 * the historical MAX_SKB_FRAGS value 18 to honor the
712 * same behavior as before. Any packet using more than
713 * 18 slots but less than fatal_skb_slots slots is
716 if (!drop_err && slots >= XEN_NETBK_LEGACY_SLOTS_MAX) {
718 netdev_dbg(queue->vif->dev,
719 "Too many slots (%d) exceeding limit (%d), dropping packet\n",
720 slots, XEN_NETBK_LEGACY_SLOTS_MAX);
727 memcpy(txp, RING_GET_REQUEST(&queue->tx, cons + slots),
730 /* If the guest submitted a frame >= 64 KiB then
731 * first->size overflowed and following slots will
732 * appear to be larger than the frame.
734 * This cannot be fatal error as there are buggy
735 * frontends that do this.
737 * Consume all slots and drop the packet.
739 if (!drop_err && txp->size > first->size) {
741 netdev_dbg(queue->vif->dev,
742 "Invalid tx request, slot size %u > remaining size %u\n",
743 txp->size, first->size);
747 first->size -= txp->size;
750 if (unlikely((txp->offset + txp->size) > PAGE_SIZE)) {
751 netdev_err(queue->vif->dev, "Cross page boundary, txp->offset: %u, size: %u\n",
752 txp->offset, txp->size);
753 xenvif_fatal_tx_err(queue->vif);
757 more_data = txp->flags & XEN_NETTXF_more_data;
765 xenvif_tx_err(queue, first, cons + slots);
773 struct xenvif_tx_cb {
777 #define XENVIF_TX_CB(skb) ((struct xenvif_tx_cb *)(skb)->cb)
779 static inline void xenvif_tx_create_map_op(struct xenvif_queue *queue,
781 struct xen_netif_tx_request *txp,
782 struct gnttab_map_grant_ref *mop)
784 queue->pages_to_map[mop-queue->tx_map_ops] = queue->mmap_pages[pending_idx];
785 gnttab_set_map_op(mop, idx_to_kaddr(queue, pending_idx),
786 GNTMAP_host_map | GNTMAP_readonly,
787 txp->gref, queue->vif->domid);
789 memcpy(&queue->pending_tx_info[pending_idx].req, txp,
793 static inline struct sk_buff *xenvif_alloc_skb(unsigned int size)
795 struct sk_buff *skb =
796 alloc_skb(size + NET_SKB_PAD + NET_IP_ALIGN,
797 GFP_ATOMIC | __GFP_NOWARN);
798 if (unlikely(skb == NULL))
801 /* Packets passed to netif_rx() must have some headroom. */
802 skb_reserve(skb, NET_SKB_PAD + NET_IP_ALIGN);
804 /* Initialize it here to avoid later surprises */
805 skb_shinfo(skb)->destructor_arg = NULL;
810 static struct gnttab_map_grant_ref *xenvif_get_requests(struct xenvif_queue *queue,
812 struct xen_netif_tx_request *txp,
813 struct gnttab_map_grant_ref *gop,
814 unsigned int frag_overflow,
815 struct sk_buff *nskb)
817 struct skb_shared_info *shinfo = skb_shinfo(skb);
818 skb_frag_t *frags = shinfo->frags;
819 u16 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
821 pending_ring_idx_t index;
822 unsigned int nr_slots;
824 nr_slots = shinfo->nr_frags;
826 /* Skip first skb fragment if it is on same page as header fragment. */
827 start = (frag_get_pending_idx(&shinfo->frags[0]) == pending_idx);
829 for (shinfo->nr_frags = start; shinfo->nr_frags < nr_slots;
830 shinfo->nr_frags++, txp++, gop++) {
831 index = pending_index(queue->pending_cons++);
832 pending_idx = queue->pending_ring[index];
833 xenvif_tx_create_map_op(queue, pending_idx, txp, gop);
834 frag_set_pending_idx(&frags[shinfo->nr_frags], pending_idx);
839 shinfo = skb_shinfo(nskb);
840 frags = shinfo->frags;
842 for (shinfo->nr_frags = 0; shinfo->nr_frags < frag_overflow;
843 shinfo->nr_frags++, txp++, gop++) {
844 index = pending_index(queue->pending_cons++);
845 pending_idx = queue->pending_ring[index];
846 xenvif_tx_create_map_op(queue, pending_idx, txp, gop);
847 frag_set_pending_idx(&frags[shinfo->nr_frags],
851 skb_shinfo(skb)->frag_list = nskb;
857 static inline void xenvif_grant_handle_set(struct xenvif_queue *queue,
859 grant_handle_t handle)
861 if (unlikely(queue->grant_tx_handle[pending_idx] !=
862 NETBACK_INVALID_HANDLE)) {
863 netdev_err(queue->vif->dev,
864 "Trying to overwrite active handle! pending_idx: 0x%x\n",
868 queue->grant_tx_handle[pending_idx] = handle;
871 static inline void xenvif_grant_handle_reset(struct xenvif_queue *queue,
874 if (unlikely(queue->grant_tx_handle[pending_idx] ==
875 NETBACK_INVALID_HANDLE)) {
876 netdev_err(queue->vif->dev,
877 "Trying to unmap invalid handle! pending_idx: 0x%x\n",
881 queue->grant_tx_handle[pending_idx] = NETBACK_INVALID_HANDLE;
884 static int xenvif_tx_check_gop(struct xenvif_queue *queue,
886 struct gnttab_map_grant_ref **gopp_map,
887 struct gnttab_copy **gopp_copy)
889 struct gnttab_map_grant_ref *gop_map = *gopp_map;
890 u16 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
891 /* This always points to the shinfo of the skb being checked, which
892 * could be either the first or the one on the frag_list
894 struct skb_shared_info *shinfo = skb_shinfo(skb);
895 /* If this is non-NULL, we are currently checking the frag_list skb, and
896 * this points to the shinfo of the first one
898 struct skb_shared_info *first_shinfo = NULL;
899 int nr_frags = shinfo->nr_frags;
900 const bool sharedslot = nr_frags &&
901 frag_get_pending_idx(&shinfo->frags[0]) == pending_idx;
904 /* Check status of header. */
905 err = (*gopp_copy)->status;
908 netdev_dbg(queue->vif->dev,
909 "Grant copy of header failed! status: %d pending_idx: %u ref: %u\n",
910 (*gopp_copy)->status,
912 (*gopp_copy)->source.u.ref);
913 /* The first frag might still have this slot mapped */
915 xenvif_idx_release(queue, pending_idx,
916 XEN_NETIF_RSP_ERROR);
921 for (i = 0; i < nr_frags; i++, gop_map++) {
924 pending_idx = frag_get_pending_idx(&shinfo->frags[i]);
926 /* Check error status: if okay then remember grant handle. */
927 newerr = gop_map->status;
929 if (likely(!newerr)) {
930 xenvif_grant_handle_set(queue,
933 /* Had a previous error? Invalidate this fragment. */
935 xenvif_idx_unmap(queue, pending_idx);
936 /* If the mapping of the first frag was OK, but
937 * the header's copy failed, and they are
938 * sharing a slot, send an error
940 if (i == 0 && sharedslot)
941 xenvif_idx_release(queue, pending_idx,
942 XEN_NETIF_RSP_ERROR);
944 xenvif_idx_release(queue, pending_idx,
950 /* Error on this fragment: respond to client with an error. */
952 netdev_dbg(queue->vif->dev,
953 "Grant map of %d. frag failed! status: %d pending_idx: %u ref: %u\n",
959 xenvif_idx_release(queue, pending_idx, XEN_NETIF_RSP_ERROR);
961 /* Not the first error? Preceding frags already invalidated. */
965 /* First error: if the header haven't shared a slot with the
966 * first frag, release it as well.
969 xenvif_idx_release(queue,
970 XENVIF_TX_CB(skb)->pending_idx,
973 /* Invalidate preceding fragments of this skb. */
974 for (j = 0; j < i; j++) {
975 pending_idx = frag_get_pending_idx(&shinfo->frags[j]);
976 xenvif_idx_unmap(queue, pending_idx);
977 xenvif_idx_release(queue, pending_idx,
981 /* And if we found the error while checking the frag_list, unmap
982 * the first skb's frags
985 for (j = 0; j < first_shinfo->nr_frags; j++) {
986 pending_idx = frag_get_pending_idx(&first_shinfo->frags[j]);
987 xenvif_idx_unmap(queue, pending_idx);
988 xenvif_idx_release(queue, pending_idx,
993 /* Remember the error: invalidate all subsequent fragments. */
997 if (skb_has_frag_list(skb) && !first_shinfo) {
998 first_shinfo = skb_shinfo(skb);
999 shinfo = skb_shinfo(skb_shinfo(skb)->frag_list);
1000 nr_frags = shinfo->nr_frags;
1005 *gopp_map = gop_map;
1009 static void xenvif_fill_frags(struct xenvif_queue *queue, struct sk_buff *skb)
1011 struct skb_shared_info *shinfo = skb_shinfo(skb);
1012 int nr_frags = shinfo->nr_frags;
1014 u16 prev_pending_idx = INVALID_PENDING_IDX;
1016 for (i = 0; i < nr_frags; i++) {
1017 skb_frag_t *frag = shinfo->frags + i;
1018 struct xen_netif_tx_request *txp;
1022 pending_idx = frag_get_pending_idx(frag);
1024 /* If this is not the first frag, chain it to the previous*/
1025 if (prev_pending_idx == INVALID_PENDING_IDX)
1026 skb_shinfo(skb)->destructor_arg =
1027 &callback_param(queue, pending_idx);
1029 callback_param(queue, prev_pending_idx).ctx =
1030 &callback_param(queue, pending_idx);
1032 callback_param(queue, pending_idx).ctx = NULL;
1033 prev_pending_idx = pending_idx;
1035 txp = &queue->pending_tx_info[pending_idx].req;
1036 page = virt_to_page(idx_to_kaddr(queue, pending_idx));
1037 __skb_fill_page_desc(skb, i, page, txp->offset, txp->size);
1038 skb->len += txp->size;
1039 skb->data_len += txp->size;
1040 skb->truesize += txp->size;
1042 /* Take an extra reference to offset network stack's put_page */
1043 get_page(queue->mmap_pages[pending_idx]);
1047 static int xenvif_get_extras(struct xenvif_queue *queue,
1048 struct xen_netif_extra_info *extras,
1051 struct xen_netif_extra_info extra;
1052 RING_IDX cons = queue->tx.req_cons;
1055 if (unlikely(work_to_do-- <= 0)) {
1056 netdev_err(queue->vif->dev, "Missing extra info\n");
1057 xenvif_fatal_tx_err(queue->vif);
1061 memcpy(&extra, RING_GET_REQUEST(&queue->tx, cons),
1063 if (unlikely(!extra.type ||
1064 extra.type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
1065 queue->tx.req_cons = ++cons;
1066 netdev_err(queue->vif->dev,
1067 "Invalid extra type: %d\n", extra.type);
1068 xenvif_fatal_tx_err(queue->vif);
1072 memcpy(&extras[extra.type - 1], &extra, sizeof(extra));
1073 queue->tx.req_cons = ++cons;
1074 } while (extra.flags & XEN_NETIF_EXTRA_FLAG_MORE);
1079 static int xenvif_set_skb_gso(struct xenvif *vif,
1080 struct sk_buff *skb,
1081 struct xen_netif_extra_info *gso)
1083 if (!gso->u.gso.size) {
1084 netdev_err(vif->dev, "GSO size must not be zero.\n");
1085 xenvif_fatal_tx_err(vif);
1089 switch (gso->u.gso.type) {
1090 case XEN_NETIF_GSO_TYPE_TCPV4:
1091 skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
1093 case XEN_NETIF_GSO_TYPE_TCPV6:
1094 skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
1097 netdev_err(vif->dev, "Bad GSO type %d.\n", gso->u.gso.type);
1098 xenvif_fatal_tx_err(vif);
1102 skb_shinfo(skb)->gso_size = gso->u.gso.size;
1103 /* gso_segs will be calculated later */
1108 static int checksum_setup(struct xenvif_queue *queue, struct sk_buff *skb)
1110 bool recalculate_partial_csum = false;
1112 /* A GSO SKB must be CHECKSUM_PARTIAL. However some buggy
1113 * peers can fail to set NETRXF_csum_blank when sending a GSO
1114 * frame. In this case force the SKB to CHECKSUM_PARTIAL and
1115 * recalculate the partial checksum.
1117 if (skb->ip_summed != CHECKSUM_PARTIAL && skb_is_gso(skb)) {
1118 queue->stats.rx_gso_checksum_fixup++;
1119 skb->ip_summed = CHECKSUM_PARTIAL;
1120 recalculate_partial_csum = true;
1123 /* A non-CHECKSUM_PARTIAL SKB does not require setup. */
1124 if (skb->ip_summed != CHECKSUM_PARTIAL)
1127 return skb_checksum_setup(skb, recalculate_partial_csum);
1130 static bool tx_credit_exceeded(struct xenvif_queue *queue, unsigned size)
1132 u64 now = get_jiffies_64();
1133 u64 next_credit = queue->credit_window_start +
1134 msecs_to_jiffies(queue->credit_usec / 1000);
1136 /* Timer could already be pending in rare cases. */
1137 if (timer_pending(&queue->credit_timeout))
1140 /* Passed the point where we can replenish credit? */
1141 if (time_after_eq64(now, next_credit)) {
1142 queue->credit_window_start = now;
1143 tx_add_credit(queue);
1146 /* Still too big to send right now? Set a callback. */
1147 if (size > queue->remaining_credit) {
1148 queue->credit_timeout.data =
1149 (unsigned long)queue;
1150 mod_timer(&queue->credit_timeout,
1152 queue->credit_window_start = next_credit;
1160 static void xenvif_tx_build_gops(struct xenvif_queue *queue,
1165 struct gnttab_map_grant_ref *gop = queue->tx_map_ops;
1166 struct sk_buff *skb, *nskb;
1168 unsigned int frag_overflow;
1170 while (skb_queue_len(&queue->tx_queue) < budget) {
1171 struct xen_netif_tx_request txreq;
1172 struct xen_netif_tx_request txfrags[XEN_NETBK_LEGACY_SLOTS_MAX];
1173 struct xen_netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX-1];
1177 unsigned int data_len;
1178 pending_ring_idx_t index;
1180 if (queue->tx.sring->req_prod - queue->tx.req_cons >
1181 XEN_NETIF_TX_RING_SIZE) {
1182 netdev_err(queue->vif->dev,
1183 "Impossible number of requests. "
1184 "req_prod %d, req_cons %d, size %ld\n",
1185 queue->tx.sring->req_prod, queue->tx.req_cons,
1186 XEN_NETIF_TX_RING_SIZE);
1187 xenvif_fatal_tx_err(queue->vif);
1191 work_to_do = RING_HAS_UNCONSUMED_REQUESTS(&queue->tx);
1195 idx = queue->tx.req_cons;
1196 rmb(); /* Ensure that we see the request before we copy it. */
1197 memcpy(&txreq, RING_GET_REQUEST(&queue->tx, idx), sizeof(txreq));
1199 /* Credit-based scheduling. */
1200 if (txreq.size > queue->remaining_credit &&
1201 tx_credit_exceeded(queue, txreq.size))
1204 queue->remaining_credit -= txreq.size;
1207 queue->tx.req_cons = ++idx;
1209 memset(extras, 0, sizeof(extras));
1210 if (txreq.flags & XEN_NETTXF_extra_info) {
1211 work_to_do = xenvif_get_extras(queue, extras,
1213 idx = queue->tx.req_cons;
1214 if (unlikely(work_to_do < 0))
1218 ret = xenvif_count_requests(queue, &txreq, txfrags, work_to_do);
1219 if (unlikely(ret < 0))
1224 if (unlikely(txreq.size < ETH_HLEN)) {
1225 netdev_dbg(queue->vif->dev,
1226 "Bad packet size: %d\n", txreq.size);
1227 xenvif_tx_err(queue, &txreq, idx);
1231 /* No crossing a page as the payload mustn't fragment. */
1232 if (unlikely((txreq.offset + txreq.size) > PAGE_SIZE)) {
1233 netdev_err(queue->vif->dev,
1234 "txreq.offset: %u, size: %u, end: %lu\n",
1235 txreq.offset, txreq.size,
1236 (unsigned long)(txreq.offset&~PAGE_MASK) + txreq.size);
1237 xenvif_fatal_tx_err(queue->vif);
1241 index = pending_index(queue->pending_cons);
1242 pending_idx = queue->pending_ring[index];
1244 data_len = (txreq.size > XEN_NETBACK_TX_COPY_LEN &&
1245 ret < XEN_NETBK_LEGACY_SLOTS_MAX) ?
1246 XEN_NETBACK_TX_COPY_LEN : txreq.size;
1248 skb = xenvif_alloc_skb(data_len);
1249 if (unlikely(skb == NULL)) {
1250 netdev_dbg(queue->vif->dev,
1251 "Can't allocate a skb in start_xmit.\n");
1252 xenvif_tx_err(queue, &txreq, idx);
1256 skb_shinfo(skb)->nr_frags = ret;
1257 if (data_len < txreq.size)
1258 skb_shinfo(skb)->nr_frags++;
1259 /* At this point shinfo->nr_frags is in fact the number of
1260 * slots, which can be as large as XEN_NETBK_LEGACY_SLOTS_MAX.
1264 if (skb_shinfo(skb)->nr_frags > MAX_SKB_FRAGS) {
1265 frag_overflow = skb_shinfo(skb)->nr_frags - MAX_SKB_FRAGS;
1266 BUG_ON(frag_overflow > MAX_SKB_FRAGS);
1267 skb_shinfo(skb)->nr_frags = MAX_SKB_FRAGS;
1268 nskb = xenvif_alloc_skb(0);
1269 if (unlikely(nskb == NULL)) {
1271 xenvif_tx_err(queue, &txreq, idx);
1272 if (net_ratelimit())
1273 netdev_err(queue->vif->dev,
1274 "Can't allocate the frag_list skb.\n");
1279 if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
1280 struct xen_netif_extra_info *gso;
1281 gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];
1283 if (xenvif_set_skb_gso(queue->vif, skb, gso)) {
1284 /* Failure in xenvif_set_skb_gso is fatal. */
1291 XENVIF_TX_CB(skb)->pending_idx = pending_idx;
1293 __skb_put(skb, data_len);
1294 queue->tx_copy_ops[*copy_ops].source.u.ref = txreq.gref;
1295 queue->tx_copy_ops[*copy_ops].source.domid = queue->vif->domid;
1296 queue->tx_copy_ops[*copy_ops].source.offset = txreq.offset;
1298 queue->tx_copy_ops[*copy_ops].dest.u.gmfn =
1299 virt_to_mfn(skb->data);
1300 queue->tx_copy_ops[*copy_ops].dest.domid = DOMID_SELF;
1301 queue->tx_copy_ops[*copy_ops].dest.offset =
1302 offset_in_page(skb->data);
1304 queue->tx_copy_ops[*copy_ops].len = data_len;
1305 queue->tx_copy_ops[*copy_ops].flags = GNTCOPY_source_gref;
1309 if (data_len < txreq.size) {
1310 frag_set_pending_idx(&skb_shinfo(skb)->frags[0],
1312 xenvif_tx_create_map_op(queue, pending_idx, &txreq, gop);
1315 frag_set_pending_idx(&skb_shinfo(skb)->frags[0],
1316 INVALID_PENDING_IDX);
1317 memcpy(&queue->pending_tx_info[pending_idx].req, &txreq,
1321 queue->pending_cons++;
1323 gop = xenvif_get_requests(queue, skb, txfrags, gop,
1324 frag_overflow, nskb);
1326 __skb_queue_tail(&queue->tx_queue, skb);
1328 queue->tx.req_cons = idx;
1330 if (((gop-queue->tx_map_ops) >= ARRAY_SIZE(queue->tx_map_ops)) ||
1331 (*copy_ops >= ARRAY_SIZE(queue->tx_copy_ops)))
1335 (*map_ops) = gop - queue->tx_map_ops;
1339 /* Consolidate skb with a frag_list into a brand new one with local pages on
1340 * frags. Returns 0 or -ENOMEM if can't allocate new pages.
1342 static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *skb)
1344 unsigned int offset = skb_headlen(skb);
1345 skb_frag_t frags[MAX_SKB_FRAGS];
1347 struct ubuf_info *uarg;
1348 struct sk_buff *nskb = skb_shinfo(skb)->frag_list;
1350 queue->stats.tx_zerocopy_sent += 2;
1351 queue->stats.tx_frag_overflow++;
1353 xenvif_fill_frags(queue, nskb);
1354 /* Subtract frags size, we will correct it later */
1355 skb->truesize -= skb->data_len;
1356 skb->len += nskb->len;
1357 skb->data_len += nskb->len;
1359 /* create a brand new frags array and coalesce there */
1360 for (i = 0; offset < skb->len; i++) {
1364 BUG_ON(i >= MAX_SKB_FRAGS);
1365 page = alloc_page(GFP_ATOMIC);
1368 skb->truesize += skb->data_len;
1369 for (j = 0; j < i; j++)
1370 put_page(frags[j].page.p);
1374 if (offset + PAGE_SIZE < skb->len)
1377 len = skb->len - offset;
1378 if (skb_copy_bits(skb, offset, page_address(page), len))
1382 frags[i].page.p = page;
1383 frags[i].page_offset = 0;
1384 skb_frag_size_set(&frags[i], len);
1387 /* Copied all the bits from the frag list -- free it. */
1388 skb_frag_list_init(skb);
1389 xenvif_skb_zerocopy_prepare(queue, nskb);
1392 /* Release all the original (foreign) frags. */
1393 for (f = 0; f < skb_shinfo(skb)->nr_frags; f++)
1394 skb_frag_unref(skb, f);
1395 uarg = skb_shinfo(skb)->destructor_arg;
1396 /* increase inflight counter to offset decrement in callback */
1397 atomic_inc(&queue->inflight_packets);
1398 uarg->callback(uarg, true);
1399 skb_shinfo(skb)->destructor_arg = NULL;
1401 /* Fill the skb with the new (local) frags. */
1402 memcpy(skb_shinfo(skb)->frags, frags, i * sizeof(skb_frag_t));
1403 skb_shinfo(skb)->nr_frags = i;
1404 skb->truesize += i * PAGE_SIZE;
1409 static int xenvif_tx_submit(struct xenvif_queue *queue)
1411 struct gnttab_map_grant_ref *gop_map = queue->tx_map_ops;
1412 struct gnttab_copy *gop_copy = queue->tx_copy_ops;
1413 struct sk_buff *skb;
1416 while ((skb = __skb_dequeue(&queue->tx_queue)) != NULL) {
1417 struct xen_netif_tx_request *txp;
1421 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
1422 txp = &queue->pending_tx_info[pending_idx].req;
1424 /* Check the remap error code. */
1425 if (unlikely(xenvif_tx_check_gop(queue, skb, &gop_map, &gop_copy))) {
1426 /* If there was an error, xenvif_tx_check_gop is
1427 * expected to release all the frags which were mapped,
1428 * so kfree_skb shouldn't do it again
1430 skb_shinfo(skb)->nr_frags = 0;
1431 if (skb_has_frag_list(skb)) {
1432 struct sk_buff *nskb =
1433 skb_shinfo(skb)->frag_list;
1434 skb_shinfo(nskb)->nr_frags = 0;
1440 data_len = skb->len;
1441 callback_param(queue, pending_idx).ctx = NULL;
1442 if (data_len < txp->size) {
1443 /* Append the packet payload as a fragment. */
1444 txp->offset += data_len;
1445 txp->size -= data_len;
1447 /* Schedule a response immediately. */
1448 xenvif_idx_release(queue, pending_idx,
1449 XEN_NETIF_RSP_OKAY);
1452 if (txp->flags & XEN_NETTXF_csum_blank)
1453 skb->ip_summed = CHECKSUM_PARTIAL;
1454 else if (txp->flags & XEN_NETTXF_data_validated)
1455 skb->ip_summed = CHECKSUM_UNNECESSARY;
1457 xenvif_fill_frags(queue, skb);
1459 if (unlikely(skb_has_frag_list(skb))) {
1460 if (xenvif_handle_frag_list(queue, skb)) {
1461 if (net_ratelimit())
1462 netdev_err(queue->vif->dev,
1463 "Not enough memory to consolidate frag_list!\n");
1464 xenvif_skb_zerocopy_prepare(queue, skb);
1470 skb->dev = queue->vif->dev;
1471 skb->protocol = eth_type_trans(skb, skb->dev);
1472 skb_reset_network_header(skb);
1474 if (checksum_setup(queue, skb)) {
1475 netdev_dbg(queue->vif->dev,
1476 "Can't setup checksum in net_tx_action\n");
1477 /* We have to set this flag to trigger the callback */
1478 if (skb_shinfo(skb)->destructor_arg)
1479 xenvif_skb_zerocopy_prepare(queue, skb);
1484 skb_probe_transport_header(skb, 0);
1486 /* If the packet is GSO then we will have just set up the
1487 * transport header offset in checksum_setup so it's now
1488 * straightforward to calculate gso_segs.
1490 if (skb_is_gso(skb)) {
1491 int mss = skb_shinfo(skb)->gso_size;
1492 int hdrlen = skb_transport_header(skb) -
1493 skb_mac_header(skb) +
1496 skb_shinfo(skb)->gso_segs =
1497 DIV_ROUND_UP(skb->len - hdrlen, mss);
1500 queue->stats.rx_bytes += skb->len;
1501 queue->stats.rx_packets++;
1505 /* Set this flag right before netif_receive_skb, otherwise
1506 * someone might think this packet already left netback, and
1507 * do a skb_copy_ubufs while we are still in control of the
1508 * skb. E.g. the __pskb_pull_tail earlier can do such thing.
1510 if (skb_shinfo(skb)->destructor_arg) {
1511 xenvif_skb_zerocopy_prepare(queue, skb);
1512 queue->stats.tx_zerocopy_sent++;
1515 netif_receive_skb(skb);
1521 void xenvif_zerocopy_callback(struct ubuf_info *ubuf, bool zerocopy_success)
1523 unsigned long flags;
1524 pending_ring_idx_t index;
1525 struct xenvif_queue *queue = ubuf_to_queue(ubuf);
1527 /* This is the only place where we grab this lock, to protect callbacks
1530 spin_lock_irqsave(&queue->callback_lock, flags);
1532 u16 pending_idx = ubuf->desc;
1533 ubuf = (struct ubuf_info *) ubuf->ctx;
1534 BUG_ON(queue->dealloc_prod - queue->dealloc_cons >=
1536 index = pending_index(queue->dealloc_prod);
1537 queue->dealloc_ring[index] = pending_idx;
1538 /* Sync with xenvif_tx_dealloc_action:
1539 * insert idx then incr producer.
1542 queue->dealloc_prod++;
1544 spin_unlock_irqrestore(&queue->callback_lock, flags);
1546 if (likely(zerocopy_success))
1547 queue->stats.tx_zerocopy_success++;
1549 queue->stats.tx_zerocopy_fail++;
1550 xenvif_skb_zerocopy_complete(queue);
1553 static inline void xenvif_tx_dealloc_action(struct xenvif_queue *queue)
1555 struct gnttab_unmap_grant_ref *gop;
1556 pending_ring_idx_t dc, dp;
1557 u16 pending_idx, pending_idx_release[MAX_PENDING_REQS];
1560 dc = queue->dealloc_cons;
1561 gop = queue->tx_unmap_ops;
1563 /* Free up any grants we have finished using */
1565 dp = queue->dealloc_prod;
1567 /* Ensure we see all indices enqueued by all
1568 * xenvif_zerocopy_callback().
1573 BUG_ON(gop - queue->tx_unmap_ops >= MAX_PENDING_REQS);
1575 queue->dealloc_ring[pending_index(dc++)];
1577 pending_idx_release[gop - queue->tx_unmap_ops] =
1579 queue->pages_to_unmap[gop - queue->tx_unmap_ops] =
1580 queue->mmap_pages[pending_idx];
1581 gnttab_set_unmap_op(gop,
1582 idx_to_kaddr(queue, pending_idx),
1584 queue->grant_tx_handle[pending_idx]);
1585 xenvif_grant_handle_reset(queue, pending_idx);
1589 } while (dp != queue->dealloc_prod);
1591 queue->dealloc_cons = dc;
1593 if (gop - queue->tx_unmap_ops > 0) {
1595 ret = gnttab_unmap_refs(queue->tx_unmap_ops,
1597 queue->pages_to_unmap,
1598 gop - queue->tx_unmap_ops);
1600 netdev_err(queue->vif->dev, "Unmap fail: nr_ops %tu ret %d\n",
1601 gop - queue->tx_unmap_ops, ret);
1602 for (i = 0; i < gop - queue->tx_unmap_ops; ++i) {
1603 if (gop[i].status != GNTST_okay)
1604 netdev_err(queue->vif->dev,
1605 " host_addr: 0x%llx handle: 0x%x status: %d\n",
1614 for (i = 0; i < gop - queue->tx_unmap_ops; ++i)
1615 xenvif_idx_release(queue, pending_idx_release[i],
1616 XEN_NETIF_RSP_OKAY);
1620 /* Called after netfront has transmitted */
1621 int xenvif_tx_action(struct xenvif_queue *queue, int budget)
1623 unsigned nr_mops, nr_cops = 0;
1626 if (unlikely(!tx_work_todo(queue)))
1629 xenvif_tx_build_gops(queue, budget, &nr_cops, &nr_mops);
1634 gnttab_batch_copy(queue->tx_copy_ops, nr_cops);
1636 ret = gnttab_map_refs(queue->tx_map_ops,
1638 queue->pages_to_map,
1643 work_done = xenvif_tx_submit(queue);
1648 static void xenvif_idx_release(struct xenvif_queue *queue, u16 pending_idx,
1651 struct pending_tx_info *pending_tx_info;
1652 pending_ring_idx_t index;
1653 unsigned long flags;
1655 pending_tx_info = &queue->pending_tx_info[pending_idx];
1657 spin_lock_irqsave(&queue->response_lock, flags);
1659 make_tx_response(queue, &pending_tx_info->req, status);
1661 /* Release the pending index before pusing the Tx response so
1662 * its available before a new Tx request is pushed by the
1665 index = pending_index(queue->pending_prod++);
1666 queue->pending_ring[index] = pending_idx;
1668 push_tx_responses(queue);
1670 spin_unlock_irqrestore(&queue->response_lock, flags);
1674 static void make_tx_response(struct xenvif_queue *queue,
1675 struct xen_netif_tx_request *txp,
1678 RING_IDX i = queue->tx.rsp_prod_pvt;
1679 struct xen_netif_tx_response *resp;
1681 resp = RING_GET_RESPONSE(&queue->tx, i);
1685 if (txp->flags & XEN_NETTXF_extra_info)
1686 RING_GET_RESPONSE(&queue->tx, ++i)->status = XEN_NETIF_RSP_NULL;
1688 queue->tx.rsp_prod_pvt = ++i;
1691 static void push_tx_responses(struct xenvif_queue *queue)
1695 RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&queue->tx, notify);
1697 notify_remote_via_irq(queue->tx_irq);
1700 static struct xen_netif_rx_response *make_rx_response(struct xenvif_queue *queue,
1707 RING_IDX i = queue->rx.rsp_prod_pvt;
1708 struct xen_netif_rx_response *resp;
1710 resp = RING_GET_RESPONSE(&queue->rx, i);
1711 resp->offset = offset;
1712 resp->flags = flags;
1714 resp->status = (s16)size;
1716 resp->status = (s16)st;
1718 queue->rx.rsp_prod_pvt = ++i;
1723 void xenvif_idx_unmap(struct xenvif_queue *queue, u16 pending_idx)
1726 struct gnttab_unmap_grant_ref tx_unmap_op;
1728 gnttab_set_unmap_op(&tx_unmap_op,
1729 idx_to_kaddr(queue, pending_idx),
1731 queue->grant_tx_handle[pending_idx]);
1732 xenvif_grant_handle_reset(queue, pending_idx);
1734 ret = gnttab_unmap_refs(&tx_unmap_op, NULL,
1735 &queue->mmap_pages[pending_idx], 1);
1737 netdev_err(queue->vif->dev,
1738 "Unmap fail: ret: %d pending_idx: %d host_addr: %llx handle: 0x%x status: %d\n",
1741 tx_unmap_op.host_addr,
1743 tx_unmap_op.status);
1748 static inline int tx_work_todo(struct xenvif_queue *queue)
1750 if (likely(RING_HAS_UNCONSUMED_REQUESTS(&queue->tx)))
1756 static inline bool tx_dealloc_work_todo(struct xenvif_queue *queue)
1758 return queue->dealloc_cons != queue->dealloc_prod;
1761 void xenvif_unmap_frontend_rings(struct xenvif_queue *queue)
1763 if (queue->tx.sring)
1764 xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(queue->vif),
1766 if (queue->rx.sring)
1767 xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(queue->vif),
1771 int xenvif_map_frontend_rings(struct xenvif_queue *queue,
1772 grant_ref_t tx_ring_ref,
1773 grant_ref_t rx_ring_ref)
1776 struct xen_netif_tx_sring *txs;
1777 struct xen_netif_rx_sring *rxs;
1781 err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(queue->vif),
1782 &tx_ring_ref, 1, &addr);
1786 txs = (struct xen_netif_tx_sring *)addr;
1787 BACK_RING_INIT(&queue->tx, txs, PAGE_SIZE);
1789 err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(queue->vif),
1790 &rx_ring_ref, 1, &addr);
1794 rxs = (struct xen_netif_rx_sring *)addr;
1795 BACK_RING_INIT(&queue->rx, rxs, PAGE_SIZE);
1800 xenvif_unmap_frontend_rings(queue);
1804 static void xenvif_queue_carrier_off(struct xenvif_queue *queue)
1806 struct xenvif *vif = queue->vif;
1808 queue->stalled = true;
1810 /* At least one queue has stalled? Disable the carrier. */
1811 spin_lock(&vif->lock);
1812 if (vif->stalled_queues++ == 0) {
1813 netdev_info(vif->dev, "Guest Rx stalled");
1814 netif_carrier_off(vif->dev);
1816 spin_unlock(&vif->lock);
1819 static void xenvif_queue_carrier_on(struct xenvif_queue *queue)
1821 struct xenvif *vif = queue->vif;
1823 queue->last_rx_time = jiffies; /* Reset Rx stall detection. */
1824 queue->stalled = false;
1826 /* All queues are ready? Enable the carrier. */
1827 spin_lock(&vif->lock);
1828 if (--vif->stalled_queues == 0) {
1829 netdev_info(vif->dev, "Guest Rx ready");
1830 netif_carrier_on(vif->dev);
1832 spin_unlock(&vif->lock);
1835 static bool xenvif_rx_queue_stalled(struct xenvif_queue *queue)
1837 RING_IDX prod, cons;
1839 prod = queue->rx.sring->req_prod;
1840 cons = queue->rx.req_cons;
1842 return !queue->stalled
1843 && prod - cons < XEN_NETBK_RX_SLOTS_MAX
1844 && time_after(jiffies,
1845 queue->last_rx_time + queue->vif->stall_timeout);
1848 static bool xenvif_rx_queue_ready(struct xenvif_queue *queue)
1850 RING_IDX prod, cons;
1852 prod = queue->rx.sring->req_prod;
1853 cons = queue->rx.req_cons;
1855 return queue->stalled
1856 && prod - cons >= XEN_NETBK_RX_SLOTS_MAX;
1859 static bool xenvif_have_rx_work(struct xenvif_queue *queue)
1861 return (!skb_queue_empty(&queue->rx_queue)
1862 && xenvif_rx_ring_slots_available(queue, XEN_NETBK_RX_SLOTS_MAX))
1863 || (queue->vif->stall_timeout &&
1864 (xenvif_rx_queue_stalled(queue)
1865 || xenvif_rx_queue_ready(queue)))
1866 || kthread_should_stop()
1867 || queue->vif->disabled;
1870 static long xenvif_rx_queue_timeout(struct xenvif_queue *queue)
1872 struct sk_buff *skb;
1875 skb = skb_peek(&queue->rx_queue);
1877 return MAX_SCHEDULE_TIMEOUT;
1879 timeout = XENVIF_RX_CB(skb)->expires - jiffies;
1880 return timeout < 0 ? 0 : timeout;
1883 /* Wait until the guest Rx thread has work.
1885 * The timeout needs to be adjusted based on the current head of the
1886 * queue (and not just the head at the beginning). In particular, if
1887 * the queue is initially empty an infinite timeout is used and this
1888 * needs to be reduced when a skb is queued.
1890 * This cannot be done with wait_event_timeout() because it only
1891 * calculates the timeout once.
1893 static void xenvif_wait_for_rx_work(struct xenvif_queue *queue)
1897 if (xenvif_have_rx_work(queue))
1903 prepare_to_wait(&queue->wq, &wait, TASK_INTERRUPTIBLE);
1904 if (xenvif_have_rx_work(queue))
1906 ret = schedule_timeout(xenvif_rx_queue_timeout(queue));
1910 finish_wait(&queue->wq, &wait);
1913 int xenvif_kthread_guest_rx(void *data)
1915 struct xenvif_queue *queue = data;
1916 struct xenvif *vif = queue->vif;
1918 if (!vif->stall_timeout)
1919 xenvif_queue_carrier_on(queue);
1922 xenvif_wait_for_rx_work(queue);
1924 if (kthread_should_stop())
1927 /* This frontend is found to be rogue, disable it in
1928 * kthread context. Currently this is only set when
1929 * netback finds out frontend sends malformed packet,
1930 * but we cannot disable the interface in softirq
1931 * context so we defer it here, if this thread is
1932 * associated with queue 0.
1934 if (unlikely(vif->disabled && queue->id == 0)) {
1935 xenvif_carrier_off(vif);
1939 if (!skb_queue_empty(&queue->rx_queue))
1940 xenvif_rx_action(queue);
1942 /* If the guest hasn't provided any Rx slots for a
1943 * while it's probably not responsive, drop the
1944 * carrier so packets are dropped earlier.
1946 if (vif->stall_timeout) {
1947 if (xenvif_rx_queue_stalled(queue))
1948 xenvif_queue_carrier_off(queue);
1949 else if (xenvif_rx_queue_ready(queue))
1950 xenvif_queue_carrier_on(queue);
1953 /* Queued packets may have foreign pages from other
1954 * domains. These cannot be queued indefinitely as
1955 * this would starve guests of grant refs and transmit
1958 xenvif_rx_queue_drop_expired(queue);
1960 xenvif_rx_queue_maybe_wake(queue);
1965 /* Bin any remaining skbs */
1966 xenvif_rx_queue_purge(queue);
1971 static bool xenvif_dealloc_kthread_should_stop(struct xenvif_queue *queue)
1973 /* Dealloc thread must remain running until all inflight
1976 return kthread_should_stop() &&
1977 !atomic_read(&queue->inflight_packets);
1980 int xenvif_dealloc_kthread(void *data)
1982 struct xenvif_queue *queue = data;
1985 wait_event_interruptible(queue->dealloc_wq,
1986 tx_dealloc_work_todo(queue) ||
1987 xenvif_dealloc_kthread_should_stop(queue));
1988 if (xenvif_dealloc_kthread_should_stop(queue))
1991 xenvif_tx_dealloc_action(queue);
1995 /* Unmap anything remaining*/
1996 if (tx_dealloc_work_todo(queue))
1997 xenvif_tx_dealloc_action(queue);
2002 static int __init netback_init(void)
2009 /* Allow as many queues as there are CPUs, by default */
2010 xenvif_max_queues = num_online_cpus();
2012 if (fatal_skb_slots < XEN_NETBK_LEGACY_SLOTS_MAX) {
2013 pr_info("fatal_skb_slots too small (%d), bump it to XEN_NETBK_LEGACY_SLOTS_MAX (%d)\n",
2014 fatal_skb_slots, XEN_NETBK_LEGACY_SLOTS_MAX);
2015 fatal_skb_slots = XEN_NETBK_LEGACY_SLOTS_MAX;
2018 rc = xenvif_xenbus_init();
2022 #ifdef CONFIG_DEBUG_FS
2023 xen_netback_dbg_root = debugfs_create_dir("xen-netback", NULL);
2024 if (IS_ERR_OR_NULL(xen_netback_dbg_root))
2025 pr_warn("Init of debugfs returned %ld!\n",
2026 PTR_ERR(xen_netback_dbg_root));
2027 #endif /* CONFIG_DEBUG_FS */
2035 module_init(netback_init);
2037 static void __exit netback_fini(void)
2039 #ifdef CONFIG_DEBUG_FS
2040 if (!IS_ERR_OR_NULL(xen_netback_dbg_root))
2041 debugfs_remove_recursive(xen_netback_dbg_root);
2042 #endif /* CONFIG_DEBUG_FS */
2043 xenvif_xenbus_fini();
2045 module_exit(netback_fini);
2047 MODULE_LICENSE("Dual BSD/GPL");
2048 MODULE_ALIAS("xen-backend:vif");