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>
48 #include <asm/xen/hypercall.h>
49 #include <asm/xen/page.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 = 1;
56 module_param(separate_tx_rx_irq, bool, 0644);
58 /* When guest ring is filled up, qdisc queues the packets for us, but we have
59 * to timeout them, otherwise other guests' packets can get stuck there
61 unsigned int rx_drain_timeout_msecs = 10000;
62 module_param(rx_drain_timeout_msecs, uint, 0444);
63 unsigned int rx_drain_timeout_jiffies;
66 * This is the maximum slots a skb can have. If a guest sends a skb
67 * which exceeds this limit it is considered malicious.
69 #define FATAL_SKB_SLOTS_DEFAULT 20
70 static unsigned int fatal_skb_slots = FATAL_SKB_SLOTS_DEFAULT;
71 module_param(fatal_skb_slots, uint, 0444);
73 static void xenvif_idx_release(struct xenvif *vif, u16 pending_idx,
76 static void make_tx_response(struct xenvif *vif,
77 struct xen_netif_tx_request *txp,
80 static inline int tx_work_todo(struct xenvif *vif);
81 static inline int rx_work_todo(struct xenvif *vif);
83 static struct xen_netif_rx_response *make_rx_response(struct xenvif *vif,
90 static inline unsigned long idx_to_pfn(struct xenvif *vif,
93 return page_to_pfn(vif->mmap_pages[idx]);
96 static inline unsigned long idx_to_kaddr(struct xenvif *vif,
99 return (unsigned long)pfn_to_kaddr(idx_to_pfn(vif, idx));
102 #define callback_param(vif, pending_idx) \
103 (vif->pending_tx_info[pending_idx].callback_struct)
105 /* Find the containing VIF's structure from a pointer in pending_tx_info array
107 static inline struct xenvif* ubuf_to_vif(struct ubuf_info *ubuf)
109 u16 pending_idx = ubuf->desc;
110 struct pending_tx_info *temp =
111 container_of(ubuf, struct pending_tx_info, callback_struct);
112 return container_of(temp - pending_idx,
117 /* This is a miniumum size for the linear area to avoid lots of
118 * calls to __pskb_pull_tail() as we set up checksum offsets. The
119 * value 128 was chosen as it covers all IPv4 and most likely
122 #define PKT_PROT_LEN 128
124 static u16 frag_get_pending_idx(skb_frag_t *frag)
126 return (u16)frag->page_offset;
129 static void frag_set_pending_idx(skb_frag_t *frag, u16 pending_idx)
131 frag->page_offset = pending_idx;
134 static inline pending_ring_idx_t pending_index(unsigned i)
136 return i & (MAX_PENDING_REQS-1);
139 bool xenvif_rx_ring_slots_available(struct xenvif *vif, int needed)
144 prod = vif->rx.sring->req_prod;
145 cons = vif->rx.req_cons;
147 if (prod - cons >= needed)
150 vif->rx.sring->req_event = prod + 1;
152 /* Make sure event is visible before we check prod
156 } while (vif->rx.sring->req_prod != prod);
162 * Returns true if we should start a new receive buffer instead of
163 * adding 'size' bytes to a buffer which currently contains 'offset'
166 static bool start_new_rx_buffer(int offset, unsigned long size, int head)
168 /* simple case: we have completely filled the current buffer. */
169 if (offset == MAX_BUFFER_OFFSET)
173 * complex case: start a fresh buffer if the current frag
174 * would overflow the current buffer but only if:
175 * (i) this frag would fit completely in the next buffer
176 * and (ii) there is already some data in the current buffer
177 * and (iii) this is not the head buffer.
180 * - (i) stops us splitting a frag into two copies
181 * unless the frag is too large for a single buffer.
182 * - (ii) stops us from leaving a buffer pointlessly empty.
183 * - (iii) stops us leaving the first buffer
184 * empty. Strictly speaking this is already covered
185 * by (ii) but is explicitly checked because
186 * netfront relies on the first buffer being
187 * non-empty and can crash otherwise.
189 * This means we will effectively linearise small
190 * frags but do not needlessly split large buffers
191 * into multiple copies tend to give large frags their
192 * own buffers as before.
194 if ((offset + size > MAX_BUFFER_OFFSET) &&
195 (size <= MAX_BUFFER_OFFSET) && offset && !head)
201 struct netrx_pending_operations {
202 unsigned copy_prod, copy_cons;
203 unsigned meta_prod, meta_cons;
204 struct gnttab_copy *copy;
205 struct xenvif_rx_meta *meta;
207 grant_ref_t copy_gref;
210 static struct xenvif_rx_meta *get_next_rx_buffer(struct xenvif *vif,
211 struct netrx_pending_operations *npo)
213 struct xenvif_rx_meta *meta;
214 struct xen_netif_rx_request *req;
216 req = RING_GET_REQUEST(&vif->rx, vif->rx.req_cons++);
218 meta = npo->meta + npo->meta_prod++;
219 meta->gso_type = XEN_NETIF_GSO_TYPE_NONE;
225 npo->copy_gref = req->gref;
231 * Set up the grant operations for this fragment. If it's a flipping
232 * interface, we also set up the unmap request from here.
234 static void xenvif_gop_frag_copy(struct xenvif *vif, struct sk_buff *skb,
235 struct netrx_pending_operations *npo,
236 struct page *page, unsigned long size,
237 unsigned long offset, int *head,
238 struct xenvif *foreign_vif,
239 grant_ref_t foreign_gref)
241 struct gnttab_copy *copy_gop;
242 struct xenvif_rx_meta *meta;
244 int gso_type = XEN_NETIF_GSO_TYPE_NONE;
246 /* Data must not cross a page boundary. */
247 BUG_ON(size + offset > PAGE_SIZE<<compound_order(page));
249 meta = npo->meta + npo->meta_prod - 1;
251 /* Skip unused frames from start of page */
252 page += offset >> PAGE_SHIFT;
253 offset &= ~PAGE_MASK;
256 BUG_ON(offset >= PAGE_SIZE);
257 BUG_ON(npo->copy_off > MAX_BUFFER_OFFSET);
259 bytes = PAGE_SIZE - offset;
264 if (start_new_rx_buffer(npo->copy_off, bytes, *head)) {
266 * Netfront requires there to be some data in the head
271 meta = get_next_rx_buffer(vif, npo);
274 if (npo->copy_off + bytes > MAX_BUFFER_OFFSET)
275 bytes = MAX_BUFFER_OFFSET - npo->copy_off;
277 copy_gop = npo->copy + npo->copy_prod++;
278 copy_gop->flags = GNTCOPY_dest_gref;
279 copy_gop->len = bytes;
282 copy_gop->source.domid = foreign_vif->domid;
283 copy_gop->source.u.ref = foreign_gref;
284 copy_gop->flags |= GNTCOPY_source_gref;
286 copy_gop->source.domid = DOMID_SELF;
287 copy_gop->source.u.gmfn =
288 virt_to_mfn(page_address(page));
290 copy_gop->source.offset = offset;
292 copy_gop->dest.domid = vif->domid;
293 copy_gop->dest.offset = npo->copy_off;
294 copy_gop->dest.u.ref = npo->copy_gref;
296 npo->copy_off += bytes;
303 if (offset == PAGE_SIZE && size) {
304 BUG_ON(!PageCompound(page));
309 /* Leave a gap for the GSO descriptor. */
310 if (skb_is_gso(skb)) {
311 if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
312 gso_type = XEN_NETIF_GSO_TYPE_TCPV4;
313 else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
314 gso_type = XEN_NETIF_GSO_TYPE_TCPV6;
317 if (*head && ((1 << gso_type) & vif->gso_mask))
320 *head = 0; /* There must be something in this buffer now. */
326 * Prepare an SKB to be transmitted to the frontend.
328 * This function is responsible for allocating grant operations, meta
331 * It returns the number of meta structures consumed. The number of
332 * ring slots used is always equal to the number of meta slots used
333 * plus the number of GSO descriptors used. Currently, we use either
334 * zero GSO descriptors (for non-GSO packets) or one descriptor (for
335 * frontend-side LRO).
337 static int xenvif_gop_skb(struct sk_buff *skb,
338 struct netrx_pending_operations *npo)
340 struct xenvif *vif = netdev_priv(skb->dev);
341 int nr_frags = skb_shinfo(skb)->nr_frags;
343 struct xen_netif_rx_request *req;
344 struct xenvif_rx_meta *meta;
349 struct ubuf_info *ubuf = skb_shinfo(skb)->destructor_arg;
350 grant_ref_t foreign_grefs[MAX_SKB_FRAGS];
351 struct xenvif *foreign_vif = NULL;
353 old_meta_prod = npo->meta_prod;
355 gso_type = XEN_NETIF_GSO_TYPE_NONE;
356 if (skb_is_gso(skb)) {
357 if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
358 gso_type = XEN_NETIF_GSO_TYPE_TCPV4;
359 else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
360 gso_type = XEN_NETIF_GSO_TYPE_TCPV6;
363 /* Set up a GSO prefix descriptor, if necessary */
364 if ((1 << gso_type) & vif->gso_prefix_mask) {
365 req = RING_GET_REQUEST(&vif->rx, vif->rx.req_cons++);
366 meta = npo->meta + npo->meta_prod++;
367 meta->gso_type = gso_type;
368 meta->gso_size = skb_shinfo(skb)->gso_size;
373 req = RING_GET_REQUEST(&vif->rx, vif->rx.req_cons++);
374 meta = npo->meta + npo->meta_prod++;
376 if ((1 << gso_type) & vif->gso_mask) {
377 meta->gso_type = gso_type;
378 meta->gso_size = skb_shinfo(skb)->gso_size;
380 meta->gso_type = XEN_NETIF_GSO_TYPE_NONE;
387 npo->copy_gref = req->gref;
389 if ((skb_shinfo(skb)->tx_flags & SKBTX_DEV_ZEROCOPY) &&
390 (ubuf->callback == &xenvif_zerocopy_callback)) {
392 foreign_vif = ubuf_to_vif(ubuf);
395 u16 pending_idx = ubuf->desc;
397 foreign_vif->pending_tx_info[pending_idx].req.gref;
398 ubuf = (struct ubuf_info *) ubuf->ctx;
403 while (data < skb_tail_pointer(skb)) {
404 unsigned int offset = offset_in_page(data);
405 unsigned int len = PAGE_SIZE - offset;
407 if (data + len > skb_tail_pointer(skb))
408 len = skb_tail_pointer(skb) - data;
410 xenvif_gop_frag_copy(vif, skb, npo,
411 virt_to_page(data), len, offset, &head,
417 for (i = 0; i < nr_frags; i++) {
418 xenvif_gop_frag_copy(vif, skb, npo,
419 skb_frag_page(&skb_shinfo(skb)->frags[i]),
420 skb_frag_size(&skb_shinfo(skb)->frags[i]),
421 skb_shinfo(skb)->frags[i].page_offset,
427 return npo->meta_prod - old_meta_prod;
431 * This is a twin to xenvif_gop_skb. Assume that xenvif_gop_skb was
432 * used to set up the operations on the top of
433 * netrx_pending_operations, which have since been done. Check that
434 * they didn't give any errors and advance over them.
436 static int xenvif_check_gop(struct xenvif *vif, int nr_meta_slots,
437 struct netrx_pending_operations *npo)
439 struct gnttab_copy *copy_op;
440 int status = XEN_NETIF_RSP_OKAY;
443 for (i = 0; i < nr_meta_slots; i++) {
444 copy_op = npo->copy + npo->copy_cons++;
445 if (copy_op->status != GNTST_okay) {
447 "Bad status %d from copy to DOM%d.\n",
448 copy_op->status, vif->domid);
449 status = XEN_NETIF_RSP_ERROR;
456 static void xenvif_add_frag_responses(struct xenvif *vif, int status,
457 struct xenvif_rx_meta *meta,
461 unsigned long offset;
463 /* No fragments used */
464 if (nr_meta_slots <= 1)
469 for (i = 0; i < nr_meta_slots; i++) {
471 if (i == nr_meta_slots - 1)
474 flags = XEN_NETRXF_more_data;
477 make_rx_response(vif, meta[i].id, status, offset,
478 meta[i].size, flags);
482 struct xenvif_rx_cb {
486 #define XENVIF_RX_CB(skb) ((struct xenvif_rx_cb *)(skb)->cb)
488 void xenvif_kick_thread(struct xenvif *vif)
493 static void xenvif_rx_action(struct xenvif *vif)
497 struct xen_netif_rx_response *resp;
498 struct sk_buff_head rxq;
502 unsigned long offset;
503 bool need_to_notify = false;
505 struct netrx_pending_operations npo = {
506 .copy = vif->grant_copy_op,
510 skb_queue_head_init(&rxq);
512 while ((skb = skb_dequeue(&vif->rx_queue)) != NULL) {
513 RING_IDX max_slots_needed;
516 /* We need a cheap worse case estimate for the number of
520 max_slots_needed = DIV_ROUND_UP(offset_in_page(skb->data) +
523 for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
525 size = skb_frag_size(&skb_shinfo(skb)->frags[i]);
526 max_slots_needed += DIV_ROUND_UP(size, PAGE_SIZE);
528 if (skb_is_gso(skb) &&
529 (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4 ||
530 skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6))
533 /* If the skb may not fit then bail out now */
534 if (!xenvif_rx_ring_slots_available(vif, max_slots_needed)) {
535 skb_queue_head(&vif->rx_queue, skb);
536 need_to_notify = true;
537 vif->rx_last_skb_slots = max_slots_needed;
540 vif->rx_last_skb_slots = 0;
542 XENVIF_RX_CB(skb)->meta_slots_used = xenvif_gop_skb(skb, &npo);
543 BUG_ON(XENVIF_RX_CB(skb)->meta_slots_used > max_slots_needed);
545 __skb_queue_tail(&rxq, skb);
548 BUG_ON(npo.meta_prod > ARRAY_SIZE(vif->meta));
553 BUG_ON(npo.copy_prod > MAX_GRANT_COPY_OPS);
554 gnttab_batch_copy(vif->grant_copy_op, npo.copy_prod);
556 while ((skb = __skb_dequeue(&rxq)) != NULL) {
558 if ((1 << vif->meta[npo.meta_cons].gso_type) &
559 vif->gso_prefix_mask) {
560 resp = RING_GET_RESPONSE(&vif->rx,
561 vif->rx.rsp_prod_pvt++);
563 resp->flags = XEN_NETRXF_gso_prefix | XEN_NETRXF_more_data;
565 resp->offset = vif->meta[npo.meta_cons].gso_size;
566 resp->id = vif->meta[npo.meta_cons].id;
567 resp->status = XENVIF_RX_CB(skb)->meta_slots_used;
570 XENVIF_RX_CB(skb)->meta_slots_used--;
574 vif->dev->stats.tx_bytes += skb->len;
575 vif->dev->stats.tx_packets++;
577 status = xenvif_check_gop(vif,
578 XENVIF_RX_CB(skb)->meta_slots_used,
581 if (XENVIF_RX_CB(skb)->meta_slots_used == 1)
584 flags = XEN_NETRXF_more_data;
586 if (skb->ip_summed == CHECKSUM_PARTIAL) /* local packet? */
587 flags |= XEN_NETRXF_csum_blank | XEN_NETRXF_data_validated;
588 else if (skb->ip_summed == CHECKSUM_UNNECESSARY)
589 /* remote but checksummed. */
590 flags |= XEN_NETRXF_data_validated;
593 resp = make_rx_response(vif, vif->meta[npo.meta_cons].id,
595 vif->meta[npo.meta_cons].size,
598 if ((1 << vif->meta[npo.meta_cons].gso_type) &
600 struct xen_netif_extra_info *gso =
601 (struct xen_netif_extra_info *)
602 RING_GET_RESPONSE(&vif->rx,
603 vif->rx.rsp_prod_pvt++);
605 resp->flags |= XEN_NETRXF_extra_info;
607 gso->u.gso.type = vif->meta[npo.meta_cons].gso_type;
608 gso->u.gso.size = vif->meta[npo.meta_cons].gso_size;
610 gso->u.gso.features = 0;
612 gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
616 xenvif_add_frag_responses(vif, status,
617 vif->meta + npo.meta_cons + 1,
618 XENVIF_RX_CB(skb)->meta_slots_used);
620 RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&vif->rx, ret);
622 need_to_notify |= !!ret;
624 npo.meta_cons += XENVIF_RX_CB(skb)->meta_slots_used;
630 notify_remote_via_irq(vif->rx_irq);
633 void xenvif_check_rx_xenvif(struct xenvif *vif)
637 RING_FINAL_CHECK_FOR_REQUESTS(&vif->tx, more_to_do);
640 napi_schedule(&vif->napi);
643 static void tx_add_credit(struct xenvif *vif)
645 unsigned long max_burst, max_credit;
648 * Allow a burst big enough to transmit a jumbo packet of up to 128kB.
649 * Otherwise the interface can seize up due to insufficient credit.
651 max_burst = RING_GET_REQUEST(&vif->tx, vif->tx.req_cons)->size;
652 max_burst = min(max_burst, 131072UL);
653 max_burst = max(max_burst, vif->credit_bytes);
655 /* Take care that adding a new chunk of credit doesn't wrap to zero. */
656 max_credit = vif->remaining_credit + vif->credit_bytes;
657 if (max_credit < vif->remaining_credit)
658 max_credit = ULONG_MAX; /* wrapped: clamp to ULONG_MAX */
660 vif->remaining_credit = min(max_credit, max_burst);
663 static void tx_credit_callback(unsigned long data)
665 struct xenvif *vif = (struct xenvif *)data;
667 xenvif_check_rx_xenvif(vif);
670 static void xenvif_tx_err(struct xenvif *vif,
671 struct xen_netif_tx_request *txp, RING_IDX end)
673 RING_IDX cons = vif->tx.req_cons;
677 spin_lock_irqsave(&vif->response_lock, flags);
678 make_tx_response(vif, txp, XEN_NETIF_RSP_ERROR);
679 spin_unlock_irqrestore(&vif->response_lock, flags);
682 txp = RING_GET_REQUEST(&vif->tx, cons++);
684 vif->tx.req_cons = cons;
687 static void xenvif_fatal_tx_err(struct xenvif *vif)
689 netdev_err(vif->dev, "fatal error; disabling device\n");
690 xenvif_carrier_off(vif);
693 static int xenvif_count_requests(struct xenvif *vif,
694 struct xen_netif_tx_request *first,
695 struct xen_netif_tx_request *txp,
698 RING_IDX cons = vif->tx.req_cons;
703 if (!(first->flags & XEN_NETTXF_more_data))
707 struct xen_netif_tx_request dropped_tx = { 0 };
709 if (slots >= work_to_do) {
711 "Asked for %d slots but exceeds this limit\n",
713 xenvif_fatal_tx_err(vif);
717 /* This guest is really using too many slots and
718 * considered malicious.
720 if (unlikely(slots >= fatal_skb_slots)) {
722 "Malicious frontend using %d slots, threshold %u\n",
723 slots, fatal_skb_slots);
724 xenvif_fatal_tx_err(vif);
728 /* Xen network protocol had implicit dependency on
729 * MAX_SKB_FRAGS. XEN_NETBK_LEGACY_SLOTS_MAX is set to
730 * the historical MAX_SKB_FRAGS value 18 to honor the
731 * same behavior as before. Any packet using more than
732 * 18 slots but less than fatal_skb_slots slots is
735 if (!drop_err && slots >= XEN_NETBK_LEGACY_SLOTS_MAX) {
738 "Too many slots (%d) exceeding limit (%d), dropping packet\n",
739 slots, XEN_NETBK_LEGACY_SLOTS_MAX);
746 memcpy(txp, RING_GET_REQUEST(&vif->tx, cons + slots),
749 /* If the guest submitted a frame >= 64 KiB then
750 * first->size overflowed and following slots will
751 * appear to be larger than the frame.
753 * This cannot be fatal error as there are buggy
754 * frontends that do this.
756 * Consume all slots and drop the packet.
758 if (!drop_err && txp->size > first->size) {
761 "Invalid tx request, slot size %u > remaining size %u\n",
762 txp->size, first->size);
766 first->size -= txp->size;
769 if (unlikely((txp->offset + txp->size) > PAGE_SIZE)) {
770 netdev_err(vif->dev, "Cross page boundary, txp->offset: %x, size: %u\n",
771 txp->offset, txp->size);
772 xenvif_fatal_tx_err(vif);
776 more_data = txp->flags & XEN_NETTXF_more_data;
784 xenvif_tx_err(vif, first, cons + slots);
792 struct xenvif_tx_cb {
796 #define XENVIF_TX_CB(skb) ((struct xenvif_tx_cb *)(skb)->cb)
798 static inline void xenvif_tx_create_gop(struct xenvif *vif,
800 struct xen_netif_tx_request *txp,
801 struct gnttab_map_grant_ref *gop)
803 vif->pages_to_map[gop-vif->tx_map_ops] = vif->mmap_pages[pending_idx];
804 gnttab_set_map_op(gop, idx_to_kaddr(vif, pending_idx),
805 GNTMAP_host_map | GNTMAP_readonly,
806 txp->gref, vif->domid);
808 memcpy(&vif->pending_tx_info[pending_idx].req, txp,
812 static inline struct sk_buff *xenvif_alloc_skb(unsigned int size)
814 struct sk_buff *skb =
815 alloc_skb(size + NET_SKB_PAD + NET_IP_ALIGN,
816 GFP_ATOMIC | __GFP_NOWARN);
817 if (unlikely(skb == NULL))
820 /* Packets passed to netif_rx() must have some headroom. */
821 skb_reserve(skb, NET_SKB_PAD + NET_IP_ALIGN);
823 /* Initialize it here to avoid later surprises */
824 skb_shinfo(skb)->destructor_arg = NULL;
829 static struct gnttab_map_grant_ref *xenvif_get_requests(struct xenvif *vif,
831 struct xen_netif_tx_request *txp,
832 struct gnttab_map_grant_ref *gop)
834 struct skb_shared_info *shinfo = skb_shinfo(skb);
835 skb_frag_t *frags = shinfo->frags;
836 u16 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
838 pending_ring_idx_t index;
839 unsigned int nr_slots, frag_overflow = 0;
841 /* At this point shinfo->nr_frags is in fact the number of
842 * slots, which can be as large as XEN_NETBK_LEGACY_SLOTS_MAX.
844 if (shinfo->nr_frags > MAX_SKB_FRAGS) {
845 frag_overflow = shinfo->nr_frags - MAX_SKB_FRAGS;
846 BUG_ON(frag_overflow > MAX_SKB_FRAGS);
847 shinfo->nr_frags = MAX_SKB_FRAGS;
849 nr_slots = shinfo->nr_frags;
851 /* Skip first skb fragment if it is on same page as header fragment. */
852 start = (frag_get_pending_idx(&shinfo->frags[0]) == pending_idx);
854 for (shinfo->nr_frags = start; shinfo->nr_frags < nr_slots;
855 shinfo->nr_frags++, txp++, gop++) {
856 index = pending_index(vif->pending_cons++);
857 pending_idx = vif->pending_ring[index];
858 xenvif_tx_create_gop(vif, pending_idx, txp, gop);
859 frag_set_pending_idx(&frags[shinfo->nr_frags], pending_idx);
863 struct sk_buff *nskb = xenvif_alloc_skb(0);
864 if (unlikely(nskb == NULL)) {
867 "Can't allocate the frag_list skb.\n");
871 shinfo = skb_shinfo(nskb);
872 frags = shinfo->frags;
874 for (shinfo->nr_frags = 0; shinfo->nr_frags < frag_overflow;
875 shinfo->nr_frags++, txp++, gop++) {
876 index = pending_index(vif->pending_cons++);
877 pending_idx = vif->pending_ring[index];
878 xenvif_tx_create_gop(vif, pending_idx, txp, gop);
879 frag_set_pending_idx(&frags[shinfo->nr_frags],
883 skb_shinfo(skb)->frag_list = nskb;
889 static inline void xenvif_grant_handle_set(struct xenvif *vif,
891 grant_handle_t handle)
893 if (unlikely(vif->grant_tx_handle[pending_idx] !=
894 NETBACK_INVALID_HANDLE)) {
896 "Trying to overwrite active handle! pending_idx: %x\n",
900 vif->grant_tx_handle[pending_idx] = handle;
903 static inline void xenvif_grant_handle_reset(struct xenvif *vif,
906 if (unlikely(vif->grant_tx_handle[pending_idx] ==
907 NETBACK_INVALID_HANDLE)) {
909 "Trying to unmap invalid handle! pending_idx: %x\n",
913 vif->grant_tx_handle[pending_idx] = NETBACK_INVALID_HANDLE;
916 static int xenvif_tx_check_gop(struct xenvif *vif,
918 struct gnttab_map_grant_ref **gopp)
920 struct gnttab_map_grant_ref *gop = *gopp;
921 u16 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
922 struct skb_shared_info *shinfo = skb_shinfo(skb);
923 struct pending_tx_info *tx_info;
924 int nr_frags = shinfo->nr_frags;
926 struct sk_buff *first_skb = NULL;
928 /* Check status of header. */
931 xenvif_idx_release(vif, pending_idx, XEN_NETIF_RSP_ERROR);
933 xenvif_grant_handle_set(vif, pending_idx , gop->handle);
935 /* Skip first skb fragment if it is on same page as header fragment. */
936 start = (frag_get_pending_idx(&shinfo->frags[0]) == pending_idx);
939 for (i = start; i < nr_frags; i++) {
942 pending_idx = frag_get_pending_idx(&shinfo->frags[i]);
943 tx_info = &vif->pending_tx_info[pending_idx];
945 /* Check error status: if okay then remember grant handle. */
946 newerr = (++gop)->status;
948 if (likely(!newerr)) {
949 xenvif_grant_handle_set(vif, pending_idx , gop->handle);
950 /* Had a previous error? Invalidate this fragment. */
952 xenvif_idx_unmap(vif, pending_idx);
956 /* Error on this fragment: respond to client with an error. */
957 xenvif_idx_release(vif, pending_idx, XEN_NETIF_RSP_ERROR);
959 /* Not the first error? Preceding frags already invalidated. */
962 /* First error: invalidate header and preceding fragments. */
964 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
966 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
967 xenvif_idx_unmap(vif, pending_idx);
968 for (j = start; j < i; j++) {
969 pending_idx = frag_get_pending_idx(&shinfo->frags[j]);
970 xenvif_idx_unmap(vif, pending_idx);
973 /* Remember the error: invalidate all subsequent fragments. */
977 if (skb_has_frag_list(skb)) {
979 skb = shinfo->frag_list;
980 shinfo = skb_shinfo(skb);
981 nr_frags = shinfo->nr_frags;
987 /* There was a mapping error in the frag_list skb. We have to unmap
988 * the first skb's frags
990 if (first_skb && err) {
992 shinfo = skb_shinfo(first_skb);
993 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
994 start = (frag_get_pending_idx(&shinfo->frags[0]) == pending_idx);
995 for (j = start; j < shinfo->nr_frags; j++) {
996 pending_idx = frag_get_pending_idx(&shinfo->frags[j]);
997 xenvif_idx_unmap(vif, pending_idx);
1005 static void xenvif_fill_frags(struct xenvif *vif, struct sk_buff *skb)
1007 struct skb_shared_info *shinfo = skb_shinfo(skb);
1008 int nr_frags = shinfo->nr_frags;
1010 u16 prev_pending_idx = INVALID_PENDING_IDX;
1012 if (skb_shinfo(skb)->destructor_arg)
1013 prev_pending_idx = XENVIF_TX_CB(skb)->pending_idx;
1015 for (i = 0; i < nr_frags; i++) {
1016 skb_frag_t *frag = shinfo->frags + i;
1017 struct xen_netif_tx_request *txp;
1021 pending_idx = frag_get_pending_idx(frag);
1023 /* If this is not the first frag, chain it to the previous*/
1024 if (unlikely(prev_pending_idx == INVALID_PENDING_IDX))
1025 skb_shinfo(skb)->destructor_arg =
1026 &callback_param(vif, pending_idx);
1027 else if (likely(pending_idx != prev_pending_idx))
1028 callback_param(vif, prev_pending_idx).ctx =
1029 &callback_param(vif, pending_idx);
1031 callback_param(vif, pending_idx).ctx = NULL;
1032 prev_pending_idx = pending_idx;
1034 txp = &vif->pending_tx_info[pending_idx].req;
1035 page = virt_to_page(idx_to_kaddr(vif, pending_idx));
1036 __skb_fill_page_desc(skb, i, page, txp->offset, txp->size);
1037 skb->len += txp->size;
1038 skb->data_len += txp->size;
1039 skb->truesize += txp->size;
1041 /* Take an extra reference to offset network stack's put_page */
1042 get_page(vif->mmap_pages[pending_idx]);
1044 /* FIXME: __skb_fill_page_desc set this to true because page->pfmemalloc
1045 * overlaps with "index", and "mapping" is not set. I think mapping
1046 * should be set. If delivered to local stack, it would drop this
1047 * skb in sk_filter unless the socket has the right to use it.
1049 skb->pfmemalloc = false;
1052 static int xenvif_get_extras(struct xenvif *vif,
1053 struct xen_netif_extra_info *extras,
1056 struct xen_netif_extra_info extra;
1057 RING_IDX cons = vif->tx.req_cons;
1060 if (unlikely(work_to_do-- <= 0)) {
1061 netdev_err(vif->dev, "Missing extra info\n");
1062 xenvif_fatal_tx_err(vif);
1066 memcpy(&extra, RING_GET_REQUEST(&vif->tx, cons),
1068 if (unlikely(!extra.type ||
1069 extra.type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
1070 vif->tx.req_cons = ++cons;
1071 netdev_err(vif->dev,
1072 "Invalid extra type: %d\n", extra.type);
1073 xenvif_fatal_tx_err(vif);
1077 memcpy(&extras[extra.type - 1], &extra, sizeof(extra));
1078 vif->tx.req_cons = ++cons;
1079 } while (extra.flags & XEN_NETIF_EXTRA_FLAG_MORE);
1084 static int xenvif_set_skb_gso(struct xenvif *vif,
1085 struct sk_buff *skb,
1086 struct xen_netif_extra_info *gso)
1088 if (!gso->u.gso.size) {
1089 netdev_err(vif->dev, "GSO size must not be zero.\n");
1090 xenvif_fatal_tx_err(vif);
1094 switch (gso->u.gso.type) {
1095 case XEN_NETIF_GSO_TYPE_TCPV4:
1096 skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
1098 case XEN_NETIF_GSO_TYPE_TCPV6:
1099 skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
1102 netdev_err(vif->dev, "Bad GSO type %d.\n", gso->u.gso.type);
1103 xenvif_fatal_tx_err(vif);
1107 skb_shinfo(skb)->gso_size = gso->u.gso.size;
1108 /* gso_segs will be calculated later */
1113 static int checksum_setup(struct xenvif *vif, struct sk_buff *skb)
1115 bool recalculate_partial_csum = false;
1117 /* A GSO SKB must be CHECKSUM_PARTIAL. However some buggy
1118 * peers can fail to set NETRXF_csum_blank when sending a GSO
1119 * frame. In this case force the SKB to CHECKSUM_PARTIAL and
1120 * recalculate the partial checksum.
1122 if (skb->ip_summed != CHECKSUM_PARTIAL && skb_is_gso(skb)) {
1123 vif->rx_gso_checksum_fixup++;
1124 skb->ip_summed = CHECKSUM_PARTIAL;
1125 recalculate_partial_csum = true;
1128 /* A non-CHECKSUM_PARTIAL SKB does not require setup. */
1129 if (skb->ip_summed != CHECKSUM_PARTIAL)
1132 return skb_checksum_setup(skb, recalculate_partial_csum);
1135 static bool tx_credit_exceeded(struct xenvif *vif, unsigned size)
1137 u64 now = get_jiffies_64();
1138 u64 next_credit = vif->credit_window_start +
1139 msecs_to_jiffies(vif->credit_usec / 1000);
1141 /* Timer could already be pending in rare cases. */
1142 if (timer_pending(&vif->credit_timeout))
1145 /* Passed the point where we can replenish credit? */
1146 if (time_after_eq64(now, next_credit)) {
1147 vif->credit_window_start = now;
1151 /* Still too big to send right now? Set a callback. */
1152 if (size > vif->remaining_credit) {
1153 vif->credit_timeout.data =
1155 vif->credit_timeout.function =
1157 mod_timer(&vif->credit_timeout,
1159 vif->credit_window_start = next_credit;
1167 static unsigned xenvif_tx_build_gops(struct xenvif *vif, int budget)
1169 struct gnttab_map_grant_ref *gop = vif->tx_map_ops, *request_gop;
1170 struct sk_buff *skb;
1173 while (skb_queue_len(&vif->tx_queue) < budget) {
1174 struct xen_netif_tx_request txreq;
1175 struct xen_netif_tx_request txfrags[XEN_NETBK_LEGACY_SLOTS_MAX];
1176 struct xen_netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX-1];
1180 unsigned int data_len;
1181 pending_ring_idx_t index;
1183 if (vif->tx.sring->req_prod - vif->tx.req_cons >
1184 XEN_NETIF_TX_RING_SIZE) {
1185 netdev_err(vif->dev,
1186 "Impossible number of requests. "
1187 "req_prod %d, req_cons %d, size %ld\n",
1188 vif->tx.sring->req_prod, vif->tx.req_cons,
1189 XEN_NETIF_TX_RING_SIZE);
1190 xenvif_fatal_tx_err(vif);
1194 work_to_do = RING_HAS_UNCONSUMED_REQUESTS(&vif->tx);
1198 idx = vif->tx.req_cons;
1199 rmb(); /* Ensure that we see the request before we copy it. */
1200 memcpy(&txreq, RING_GET_REQUEST(&vif->tx, idx), sizeof(txreq));
1202 /* Credit-based scheduling. */
1203 if (txreq.size > vif->remaining_credit &&
1204 tx_credit_exceeded(vif, txreq.size))
1207 vif->remaining_credit -= txreq.size;
1210 vif->tx.req_cons = ++idx;
1212 memset(extras, 0, sizeof(extras));
1213 if (txreq.flags & XEN_NETTXF_extra_info) {
1214 work_to_do = xenvif_get_extras(vif, extras,
1216 idx = vif->tx.req_cons;
1217 if (unlikely(work_to_do < 0))
1221 ret = xenvif_count_requests(vif, &txreq, txfrags, work_to_do);
1222 if (unlikely(ret < 0))
1227 if (unlikely(txreq.size < ETH_HLEN)) {
1228 netdev_dbg(vif->dev,
1229 "Bad packet size: %d\n", txreq.size);
1230 xenvif_tx_err(vif, &txreq, idx);
1234 /* No crossing a page as the payload mustn't fragment. */
1235 if (unlikely((txreq.offset + txreq.size) > PAGE_SIZE)) {
1236 netdev_err(vif->dev,
1237 "txreq.offset: %x, size: %u, end: %lu\n",
1238 txreq.offset, txreq.size,
1239 (txreq.offset&~PAGE_MASK) + txreq.size);
1240 xenvif_fatal_tx_err(vif);
1244 index = pending_index(vif->pending_cons);
1245 pending_idx = vif->pending_ring[index];
1247 data_len = (txreq.size > PKT_PROT_LEN &&
1248 ret < XEN_NETBK_LEGACY_SLOTS_MAX) ?
1249 PKT_PROT_LEN : txreq.size;
1251 skb = xenvif_alloc_skb(data_len);
1252 if (unlikely(skb == NULL)) {
1253 netdev_dbg(vif->dev,
1254 "Can't allocate a skb in start_xmit.\n");
1255 xenvif_tx_err(vif, &txreq, idx);
1259 if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
1260 struct xen_netif_extra_info *gso;
1261 gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];
1263 if (xenvif_set_skb_gso(vif, skb, gso)) {
1264 /* Failure in xenvif_set_skb_gso is fatal. */
1270 xenvif_tx_create_gop(vif, pending_idx, &txreq, gop);
1274 XENVIF_TX_CB(skb)->pending_idx = pending_idx;
1276 __skb_put(skb, data_len);
1278 skb_shinfo(skb)->nr_frags = ret;
1279 if (data_len < txreq.size) {
1280 skb_shinfo(skb)->nr_frags++;
1281 frag_set_pending_idx(&skb_shinfo(skb)->frags[0],
1284 frag_set_pending_idx(&skb_shinfo(skb)->frags[0],
1285 INVALID_PENDING_IDX);
1288 vif->pending_cons++;
1290 request_gop = xenvif_get_requests(vif, skb, txfrags, gop);
1291 if (request_gop == NULL) {
1293 xenvif_tx_err(vif, &txreq, idx);
1298 __skb_queue_tail(&vif->tx_queue, skb);
1300 vif->tx.req_cons = idx;
1302 if ((gop-vif->tx_map_ops) >= ARRAY_SIZE(vif->tx_map_ops))
1306 return gop - vif->tx_map_ops;
1309 /* Consolidate skb with a frag_list into a brand new one with local pages on
1310 * frags. Returns 0 or -ENOMEM if can't allocate new pages.
1312 static int xenvif_handle_frag_list(struct xenvif *vif, struct sk_buff *skb)
1314 unsigned int offset = skb_headlen(skb);
1315 skb_frag_t frags[MAX_SKB_FRAGS];
1317 struct ubuf_info *uarg;
1318 struct sk_buff *nskb = skb_shinfo(skb)->frag_list;
1320 vif->tx_zerocopy_sent += 2;
1321 vif->tx_frag_overflow++;
1323 xenvif_fill_frags(vif, nskb);
1324 /* Subtract frags size, we will correct it later */
1325 skb->truesize -= skb->data_len;
1326 skb->len += nskb->len;
1327 skb->data_len += nskb->len;
1329 /* create a brand new frags array and coalesce there */
1330 for (i = 0; offset < skb->len; i++) {
1334 BUG_ON(i >= MAX_SKB_FRAGS);
1335 page = alloc_page(GFP_ATOMIC|__GFP_COLD);
1338 skb->truesize += skb->data_len;
1339 for (j = 0; j < i; j++)
1340 put_page(frags[j].page.p);
1344 if (offset + PAGE_SIZE < skb->len)
1347 len = skb->len - offset;
1348 if (skb_copy_bits(skb, offset, page_address(page), len))
1352 frags[i].page.p = page;
1353 frags[i].page_offset = 0;
1354 skb_frag_size_set(&frags[i], len);
1356 /* swap out with old one */
1357 memcpy(skb_shinfo(skb)->frags,
1359 i * sizeof(skb_frag_t));
1360 skb_shinfo(skb)->nr_frags = i;
1361 skb->truesize += i * PAGE_SIZE;
1363 /* remove traces of mapped pages and frag_list */
1364 skb_frag_list_init(skb);
1365 uarg = skb_shinfo(skb)->destructor_arg;
1366 uarg->callback(uarg, true);
1367 skb_shinfo(skb)->destructor_arg = NULL;
1369 skb_shinfo(nskb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
1375 static int xenvif_tx_submit(struct xenvif *vif)
1377 struct gnttab_map_grant_ref *gop = vif->tx_map_ops;
1378 struct sk_buff *skb;
1381 while ((skb = __skb_dequeue(&vif->tx_queue)) != NULL) {
1382 struct xen_netif_tx_request *txp;
1386 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
1387 txp = &vif->pending_tx_info[pending_idx].req;
1389 /* Check the remap error code. */
1390 if (unlikely(xenvif_tx_check_gop(vif, skb, &gop))) {
1391 netdev_dbg(vif->dev, "netback grant failed.\n");
1392 skb_shinfo(skb)->nr_frags = 0;
1397 data_len = skb->len;
1399 (void *)(idx_to_kaddr(vif, pending_idx)|txp->offset),
1401 callback_param(vif, pending_idx).ctx = NULL;
1402 if (data_len < txp->size) {
1403 /* Append the packet payload as a fragment. */
1404 txp->offset += data_len;
1405 txp->size -= data_len;
1406 skb_shinfo(skb)->destructor_arg =
1407 &callback_param(vif, pending_idx);
1409 /* Schedule a response immediately. */
1410 xenvif_idx_unmap(vif, pending_idx);
1413 if (txp->flags & XEN_NETTXF_csum_blank)
1414 skb->ip_summed = CHECKSUM_PARTIAL;
1415 else if (txp->flags & XEN_NETTXF_data_validated)
1416 skb->ip_summed = CHECKSUM_UNNECESSARY;
1418 xenvif_fill_frags(vif, skb);
1420 if (unlikely(skb_has_frag_list(skb))) {
1421 if (xenvif_handle_frag_list(vif, skb)) {
1422 if (net_ratelimit())
1423 netdev_err(vif->dev,
1424 "Not enough memory to consolidate frag_list!\n");
1425 skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
1431 if (skb_is_nonlinear(skb) && skb_headlen(skb) < PKT_PROT_LEN) {
1432 int target = min_t(int, skb->len, PKT_PROT_LEN);
1433 __pskb_pull_tail(skb, target - skb_headlen(skb));
1436 skb->dev = vif->dev;
1437 skb->protocol = eth_type_trans(skb, skb->dev);
1438 skb_reset_network_header(skb);
1440 if (checksum_setup(vif, skb)) {
1441 netdev_dbg(vif->dev,
1442 "Can't setup checksum in net_tx_action\n");
1443 /* We have to set this flag to trigger the callback */
1444 if (skb_shinfo(skb)->destructor_arg)
1445 skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
1450 skb_probe_transport_header(skb, 0);
1452 /* If the packet is GSO then we will have just set up the
1453 * transport header offset in checksum_setup so it's now
1454 * straightforward to calculate gso_segs.
1456 if (skb_is_gso(skb)) {
1457 int mss = skb_shinfo(skb)->gso_size;
1458 int hdrlen = skb_transport_header(skb) -
1459 skb_mac_header(skb) +
1462 skb_shinfo(skb)->gso_segs =
1463 DIV_ROUND_UP(skb->len - hdrlen, mss);
1466 vif->dev->stats.rx_bytes += skb->len;
1467 vif->dev->stats.rx_packets++;
1471 /* Set this flag right before netif_receive_skb, otherwise
1472 * someone might think this packet already left netback, and
1473 * do a skb_copy_ubufs while we are still in control of the
1474 * skb. E.g. the __pskb_pull_tail earlier can do such thing.
1476 if (skb_shinfo(skb)->destructor_arg) {
1477 skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
1478 vif->tx_zerocopy_sent++;
1481 netif_receive_skb(skb);
1487 void xenvif_zerocopy_callback(struct ubuf_info *ubuf, bool zerocopy_success)
1489 unsigned long flags;
1490 pending_ring_idx_t index;
1491 struct xenvif *vif = ubuf_to_vif(ubuf);
1493 /* This is the only place where we grab this lock, to protect callbacks
1496 spin_lock_irqsave(&vif->callback_lock, flags);
1498 u16 pending_idx = ubuf->desc;
1499 ubuf = (struct ubuf_info *) ubuf->ctx;
1500 BUG_ON(vif->dealloc_prod - vif->dealloc_cons >=
1502 index = pending_index(vif->dealloc_prod);
1503 vif->dealloc_ring[index] = pending_idx;
1504 /* Sync with xenvif_tx_dealloc_action:
1505 * insert idx then incr producer.
1508 vif->dealloc_prod++;
1510 wake_up(&vif->dealloc_wq);
1511 spin_unlock_irqrestore(&vif->callback_lock, flags);
1513 if (likely(zerocopy_success))
1514 vif->tx_zerocopy_success++;
1516 vif->tx_zerocopy_fail++;
1519 static inline void xenvif_tx_dealloc_action(struct xenvif *vif)
1521 struct gnttab_unmap_grant_ref *gop;
1522 pending_ring_idx_t dc, dp;
1523 u16 pending_idx, pending_idx_release[MAX_PENDING_REQS];
1526 dc = vif->dealloc_cons;
1527 gop = vif->tx_unmap_ops;
1529 /* Free up any grants we have finished using */
1531 dp = vif->dealloc_prod;
1533 /* Ensure we see all indices enqueued by all
1534 * xenvif_zerocopy_callback().
1539 BUG_ON(gop - vif->tx_unmap_ops > MAX_PENDING_REQS);
1541 vif->dealloc_ring[pending_index(dc++)];
1543 pending_idx_release[gop-vif->tx_unmap_ops] =
1545 vif->pages_to_unmap[gop-vif->tx_unmap_ops] =
1546 vif->mmap_pages[pending_idx];
1547 gnttab_set_unmap_op(gop,
1548 idx_to_kaddr(vif, pending_idx),
1550 vif->grant_tx_handle[pending_idx]);
1551 xenvif_grant_handle_reset(vif, pending_idx);
1555 } while (dp != vif->dealloc_prod);
1557 vif->dealloc_cons = dc;
1559 if (gop - vif->tx_unmap_ops > 0) {
1561 ret = gnttab_unmap_refs(vif->tx_unmap_ops,
1563 vif->pages_to_unmap,
1564 gop - vif->tx_unmap_ops);
1566 netdev_err(vif->dev, "Unmap fail: nr_ops %tx ret %d\n",
1567 gop - vif->tx_unmap_ops, ret);
1568 for (i = 0; i < gop - vif->tx_unmap_ops; ++i) {
1569 if (gop[i].status != GNTST_okay)
1570 netdev_err(vif->dev,
1571 " host_addr: %llx handle: %x status: %d\n",
1580 for (i = 0; i < gop - vif->tx_unmap_ops; ++i)
1581 xenvif_idx_release(vif, pending_idx_release[i],
1582 XEN_NETIF_RSP_OKAY);
1586 /* Called after netfront has transmitted */
1587 int xenvif_tx_action(struct xenvif *vif, int budget)
1592 if (unlikely(!tx_work_todo(vif)))
1595 nr_gops = xenvif_tx_build_gops(vif, budget);
1600 ret = gnttab_map_refs(vif->tx_map_ops,
1606 work_done = xenvif_tx_submit(vif);
1611 static void xenvif_idx_release(struct xenvif *vif, u16 pending_idx,
1614 struct pending_tx_info *pending_tx_info;
1615 pending_ring_idx_t index;
1616 unsigned long flags;
1618 pending_tx_info = &vif->pending_tx_info[pending_idx];
1619 spin_lock_irqsave(&vif->response_lock, flags);
1620 make_tx_response(vif, &pending_tx_info->req, status);
1621 index = pending_index(vif->pending_prod);
1622 vif->pending_ring[index] = pending_idx;
1623 /* TX shouldn't use the index before we give it back here */
1625 vif->pending_prod++;
1626 spin_unlock_irqrestore(&vif->response_lock, flags);
1630 static void make_tx_response(struct xenvif *vif,
1631 struct xen_netif_tx_request *txp,
1634 RING_IDX i = vif->tx.rsp_prod_pvt;
1635 struct xen_netif_tx_response *resp;
1638 resp = RING_GET_RESPONSE(&vif->tx, i);
1642 if (txp->flags & XEN_NETTXF_extra_info)
1643 RING_GET_RESPONSE(&vif->tx, ++i)->status = XEN_NETIF_RSP_NULL;
1645 vif->tx.rsp_prod_pvt = ++i;
1646 RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&vif->tx, notify);
1648 notify_remote_via_irq(vif->tx_irq);
1651 static struct xen_netif_rx_response *make_rx_response(struct xenvif *vif,
1658 RING_IDX i = vif->rx.rsp_prod_pvt;
1659 struct xen_netif_rx_response *resp;
1661 resp = RING_GET_RESPONSE(&vif->rx, i);
1662 resp->offset = offset;
1663 resp->flags = flags;
1665 resp->status = (s16)size;
1667 resp->status = (s16)st;
1669 vif->rx.rsp_prod_pvt = ++i;
1674 void xenvif_idx_unmap(struct xenvif *vif, u16 pending_idx)
1677 struct gnttab_unmap_grant_ref tx_unmap_op;
1679 gnttab_set_unmap_op(&tx_unmap_op,
1680 idx_to_kaddr(vif, pending_idx),
1682 vif->grant_tx_handle[pending_idx]);
1683 xenvif_grant_handle_reset(vif, pending_idx);
1685 ret = gnttab_unmap_refs(&tx_unmap_op, NULL,
1686 &vif->mmap_pages[pending_idx], 1);
1688 netdev_err(vif->dev,
1689 "Unmap fail: ret: %d pending_idx: %d host_addr: %llx handle: %x status: %d\n",
1692 tx_unmap_op.host_addr,
1694 tx_unmap_op.status);
1698 xenvif_idx_release(vif, pending_idx, XEN_NETIF_RSP_OKAY);
1701 static inline int rx_work_todo(struct xenvif *vif)
1703 return (!skb_queue_empty(&vif->rx_queue) &&
1704 xenvif_rx_ring_slots_available(vif, vif->rx_last_skb_slots)) ||
1705 vif->rx_queue_purge;
1708 static inline int tx_work_todo(struct xenvif *vif)
1711 if (likely(RING_HAS_UNCONSUMED_REQUESTS(&vif->tx)))
1717 static inline bool tx_dealloc_work_todo(struct xenvif *vif)
1719 return vif->dealloc_cons != vif->dealloc_prod;
1722 void xenvif_unmap_frontend_rings(struct xenvif *vif)
1725 xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(vif),
1728 xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(vif),
1732 int xenvif_map_frontend_rings(struct xenvif *vif,
1733 grant_ref_t tx_ring_ref,
1734 grant_ref_t rx_ring_ref)
1737 struct xen_netif_tx_sring *txs;
1738 struct xen_netif_rx_sring *rxs;
1742 err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(vif),
1743 tx_ring_ref, &addr);
1747 txs = (struct xen_netif_tx_sring *)addr;
1748 BACK_RING_INIT(&vif->tx, txs, PAGE_SIZE);
1750 err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(vif),
1751 rx_ring_ref, &addr);
1755 rxs = (struct xen_netif_rx_sring *)addr;
1756 BACK_RING_INIT(&vif->rx, rxs, PAGE_SIZE);
1761 xenvif_unmap_frontend_rings(vif);
1765 void xenvif_stop_queue(struct xenvif *vif)
1767 if (!vif->can_queue)
1770 netif_stop_queue(vif->dev);
1773 static void xenvif_start_queue(struct xenvif *vif)
1775 if (xenvif_schedulable(vif))
1776 netif_wake_queue(vif->dev);
1779 int xenvif_kthread_guest_rx(void *data)
1781 struct xenvif *vif = data;
1782 struct sk_buff *skb;
1784 while (!kthread_should_stop()) {
1785 wait_event_interruptible(vif->wq,
1786 rx_work_todo(vif) ||
1787 kthread_should_stop());
1788 if (kthread_should_stop())
1791 if (vif->rx_queue_purge) {
1792 skb_queue_purge(&vif->rx_queue);
1793 vif->rx_queue_purge = false;
1796 if (!skb_queue_empty(&vif->rx_queue))
1797 xenvif_rx_action(vif);
1799 if (skb_queue_empty(&vif->rx_queue) &&
1800 netif_queue_stopped(vif->dev)) {
1801 del_timer_sync(&vif->wake_queue);
1802 xenvif_start_queue(vif);
1808 /* Bin any remaining skbs */
1809 while ((skb = skb_dequeue(&vif->rx_queue)) != NULL)
1815 int xenvif_dealloc_kthread(void *data)
1817 struct xenvif *vif = data;
1819 while (!kthread_should_stop()) {
1820 wait_event_interruptible(vif->dealloc_wq,
1821 tx_dealloc_work_todo(vif) ||
1822 kthread_should_stop());
1823 if (kthread_should_stop())
1826 xenvif_tx_dealloc_action(vif);
1830 /* Unmap anything remaining*/
1831 if (tx_dealloc_work_todo(vif))
1832 xenvif_tx_dealloc_action(vif);
1837 static int __init netback_init(void)
1844 if (fatal_skb_slots < XEN_NETBK_LEGACY_SLOTS_MAX) {
1845 pr_info("fatal_skb_slots too small (%d), bump it to XEN_NETBK_LEGACY_SLOTS_MAX (%d)\n",
1846 fatal_skb_slots, XEN_NETBK_LEGACY_SLOTS_MAX);
1847 fatal_skb_slots = XEN_NETBK_LEGACY_SLOTS_MAX;
1850 rc = xenvif_xenbus_init();
1854 rx_drain_timeout_jiffies = msecs_to_jiffies(rx_drain_timeout_msecs);
1862 module_init(netback_init);
1864 static void __exit netback_fini(void)
1866 xenvif_xenbus_fini();
1868 module_exit(netback_fini);
1870 MODULE_LICENSE("Dual BSD/GPL");
1871 MODULE_ALIAS("xen-backend:vif");