blob: 2596401308a86e24210efd65d531d91d6be60746 [file] [log] [blame]
Ian Campbellf942dc22011-03-15 00:06:18 +00001/*
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
7 *
8 * Copyright (c) 2002-2005, K A Fraser
9 *
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:
15 *
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:
22 *
23 * The above copyright notice and this permission notice shall be included in
24 * all copies or substantial portions of the Software.
25 *
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
32 * IN THE SOFTWARE.
33 */
34
35#include "common.h"
36
37#include <linux/kthread.h>
38#include <linux/if_vlan.h>
39#include <linux/udp.h>
40
41#include <net/tcp.h>
42
43#include <xen/events.h>
44#include <xen/interface/memory.h>
45
46#include <asm/xen/hypercall.h>
47#include <asm/xen/page.h>
48
49struct pending_tx_info {
50 struct xen_netif_tx_request req;
51 struct xenvif *vif;
52};
53typedef unsigned int pending_ring_idx_t;
54
55struct netbk_rx_meta {
56 int id;
57 int size;
58 int gso_size;
59};
60
61#define MAX_PENDING_REQS 256
62
Ian Campbellea066ad2011-10-05 00:28:46 +000063/* Discriminate from any valid pending_idx value. */
64#define INVALID_PENDING_IDX 0xFFFF
65
Ian Campbellf942dc22011-03-15 00:06:18 +000066#define MAX_BUFFER_OFFSET PAGE_SIZE
67
68/* extra field used in struct page */
69union page_ext {
70 struct {
71#if BITS_PER_LONG < 64
72#define IDX_WIDTH 8
73#define GROUP_WIDTH (BITS_PER_LONG - IDX_WIDTH)
74 unsigned int group:GROUP_WIDTH;
75 unsigned int idx:IDX_WIDTH;
76#else
77 unsigned int group, idx;
78#endif
79 } e;
80 void *mapping;
81};
82
83struct xen_netbk {
84 wait_queue_head_t wq;
85 struct task_struct *task;
86
87 struct sk_buff_head rx_queue;
88 struct sk_buff_head tx_queue;
89
90 struct timer_list net_timer;
91
92 struct page *mmap_pages[MAX_PENDING_REQS];
93
94 pending_ring_idx_t pending_prod;
95 pending_ring_idx_t pending_cons;
96 struct list_head net_schedule_list;
97
98 /* Protect the net_schedule_list in netif. */
99 spinlock_t net_schedule_list_lock;
100
101 atomic_t netfront_count;
102
103 struct pending_tx_info pending_tx_info[MAX_PENDING_REQS];
104 struct gnttab_copy tx_copy_ops[MAX_PENDING_REQS];
105
106 u16 pending_ring[MAX_PENDING_REQS];
107
108 /*
109 * Given MAX_BUFFER_OFFSET of 4096 the worst case is that each
110 * head/fragment page uses 2 copy operations because it
111 * straddles two buffers in the frontend.
112 */
113 struct gnttab_copy grant_copy_op[2*XEN_NETIF_RX_RING_SIZE];
114 struct netbk_rx_meta meta[2*XEN_NETIF_RX_RING_SIZE];
115};
116
117static struct xen_netbk *xen_netbk;
118static int xen_netbk_group_nr;
119
120void xen_netbk_add_xenvif(struct xenvif *vif)
121{
122 int i;
123 int min_netfront_count;
124 int min_group = 0;
125 struct xen_netbk *netbk;
126
127 min_netfront_count = atomic_read(&xen_netbk[0].netfront_count);
128 for (i = 0; i < xen_netbk_group_nr; i++) {
129 int netfront_count = atomic_read(&xen_netbk[i].netfront_count);
130 if (netfront_count < min_netfront_count) {
131 min_group = i;
132 min_netfront_count = netfront_count;
133 }
134 }
135
136 netbk = &xen_netbk[min_group];
137
138 vif->netbk = netbk;
139 atomic_inc(&netbk->netfront_count);
140}
141
142void xen_netbk_remove_xenvif(struct xenvif *vif)
143{
144 struct xen_netbk *netbk = vif->netbk;
145 vif->netbk = NULL;
146 atomic_dec(&netbk->netfront_count);
147}
148
149static void xen_netbk_idx_release(struct xen_netbk *netbk, u16 pending_idx);
150static void make_tx_response(struct xenvif *vif,
151 struct xen_netif_tx_request *txp,
152 s8 st);
153static struct xen_netif_rx_response *make_rx_response(struct xenvif *vif,
154 u16 id,
155 s8 st,
156 u16 offset,
157 u16 size,
158 u16 flags);
159
160static inline unsigned long idx_to_pfn(struct xen_netbk *netbk,
Ian Campbellea066ad2011-10-05 00:28:46 +0000161 u16 idx)
Ian Campbellf942dc22011-03-15 00:06:18 +0000162{
163 return page_to_pfn(netbk->mmap_pages[idx]);
164}
165
166static inline unsigned long idx_to_kaddr(struct xen_netbk *netbk,
Ian Campbellea066ad2011-10-05 00:28:46 +0000167 u16 idx)
Ian Campbellf942dc22011-03-15 00:06:18 +0000168{
169 return (unsigned long)pfn_to_kaddr(idx_to_pfn(netbk, idx));
170}
171
172/* extra field used in struct page */
173static inline void set_page_ext(struct page *pg, struct xen_netbk *netbk,
174 unsigned int idx)
175{
176 unsigned int group = netbk - xen_netbk;
177 union page_ext ext = { .e = { .group = group + 1, .idx = idx } };
178
179 BUILD_BUG_ON(sizeof(ext) > sizeof(ext.mapping));
180 pg->mapping = ext.mapping;
181}
182
183static int get_page_ext(struct page *pg,
184 unsigned int *pgroup, unsigned int *pidx)
185{
186 union page_ext ext = { .mapping = pg->mapping };
187 struct xen_netbk *netbk;
188 unsigned int group, idx;
189
190 group = ext.e.group - 1;
191
192 if (group < 0 || group >= xen_netbk_group_nr)
193 return 0;
194
195 netbk = &xen_netbk[group];
196
197 idx = ext.e.idx;
198
199 if ((idx < 0) || (idx >= MAX_PENDING_REQS))
200 return 0;
201
202 if (netbk->mmap_pages[idx] != pg)
203 return 0;
204
205 *pgroup = group;
206 *pidx = idx;
207
208 return 1;
209}
210
211/*
212 * This is the amount of packet we copy rather than map, so that the
213 * guest can't fiddle with the contents of the headers while we do
214 * packet processing on them (netfilter, routing, etc).
215 */
216#define PKT_PROT_LEN (ETH_HLEN + \
217 VLAN_HLEN + \
218 sizeof(struct iphdr) + MAX_IPOPTLEN + \
219 sizeof(struct tcphdr) + MAX_TCP_OPTION_SPACE)
220
Ian Campbellea066ad2011-10-05 00:28:46 +0000221static u16 frag_get_pending_idx(skb_frag_t *frag)
222{
223 return (u16)frag->page_offset;
224}
225
226static void frag_set_pending_idx(skb_frag_t *frag, u16 pending_idx)
227{
228 frag->page_offset = pending_idx;
229}
230
Ian Campbellf942dc22011-03-15 00:06:18 +0000231static inline pending_ring_idx_t pending_index(unsigned i)
232{
233 return i & (MAX_PENDING_REQS-1);
234}
235
236static inline pending_ring_idx_t nr_pending_reqs(struct xen_netbk *netbk)
237{
238 return MAX_PENDING_REQS -
239 netbk->pending_prod + netbk->pending_cons;
240}
241
242static void xen_netbk_kick_thread(struct xen_netbk *netbk)
243{
244 wake_up(&netbk->wq);
245}
246
247static int max_required_rx_slots(struct xenvif *vif)
248{
249 int max = DIV_ROUND_UP(vif->dev->mtu, PAGE_SIZE);
250
251 if (vif->can_sg || vif->gso || vif->gso_prefix)
252 max += MAX_SKB_FRAGS + 1; /* extra_info + frags */
253
254 return max;
255}
256
257int xen_netbk_rx_ring_full(struct xenvif *vif)
258{
259 RING_IDX peek = vif->rx_req_cons_peek;
260 RING_IDX needed = max_required_rx_slots(vif);
261
262 return ((vif->rx.sring->req_prod - peek) < needed) ||
263 ((vif->rx.rsp_prod_pvt + XEN_NETIF_RX_RING_SIZE - peek) < needed);
264}
265
266int xen_netbk_must_stop_queue(struct xenvif *vif)
267{
268 if (!xen_netbk_rx_ring_full(vif))
269 return 0;
270
271 vif->rx.sring->req_event = vif->rx_req_cons_peek +
272 max_required_rx_slots(vif);
273 mb(); /* request notification /then/ check the queue */
274
275 return xen_netbk_rx_ring_full(vif);
276}
277
278/*
279 * Returns true if we should start a new receive buffer instead of
280 * adding 'size' bytes to a buffer which currently contains 'offset'
281 * bytes.
282 */
283static bool start_new_rx_buffer(int offset, unsigned long size, int head)
284{
285 /* simple case: we have completely filled the current buffer. */
286 if (offset == MAX_BUFFER_OFFSET)
287 return true;
288
289 /*
290 * complex case: start a fresh buffer if the current frag
291 * would overflow the current buffer but only if:
292 * (i) this frag would fit completely in the next buffer
293 * and (ii) there is already some data in the current buffer
294 * and (iii) this is not the head buffer.
295 *
296 * Where:
297 * - (i) stops us splitting a frag into two copies
298 * unless the frag is too large for a single buffer.
299 * - (ii) stops us from leaving a buffer pointlessly empty.
300 * - (iii) stops us leaving the first buffer
301 * empty. Strictly speaking this is already covered
302 * by (ii) but is explicitly checked because
303 * netfront relies on the first buffer being
304 * non-empty and can crash otherwise.
305 *
306 * This means we will effectively linearise small
307 * frags but do not needlessly split large buffers
308 * into multiple copies tend to give large frags their
309 * own buffers as before.
310 */
311 if ((offset + size > MAX_BUFFER_OFFSET) &&
312 (size <= MAX_BUFFER_OFFSET) && offset && !head)
313 return true;
314
315 return false;
316}
317
318/*
319 * Figure out how many ring slots we're going to need to send @skb to
320 * the guest. This function is essentially a dry run of
321 * netbk_gop_frag_copy.
322 */
323unsigned int xen_netbk_count_skb_slots(struct xenvif *vif, struct sk_buff *skb)
324{
325 unsigned int count;
326 int i, copy_off;
327
328 count = DIV_ROUND_UP(
329 offset_in_page(skb->data)+skb_headlen(skb), PAGE_SIZE);
330
331 copy_off = skb_headlen(skb) % PAGE_SIZE;
332
333 if (skb_shinfo(skb)->gso_size)
334 count++;
335
336 for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
Eric Dumazet9e903e02011-10-18 21:00:24 +0000337 unsigned long size = skb_frag_size(&skb_shinfo(skb)->frags[i]);
Ian Campbellf942dc22011-03-15 00:06:18 +0000338 unsigned long bytes;
339 while (size > 0) {
340 BUG_ON(copy_off > MAX_BUFFER_OFFSET);
341
342 if (start_new_rx_buffer(copy_off, size, 0)) {
343 count++;
344 copy_off = 0;
345 }
346
347 bytes = size;
348 if (copy_off + bytes > MAX_BUFFER_OFFSET)
349 bytes = MAX_BUFFER_OFFSET - copy_off;
350
351 copy_off += bytes;
352 size -= bytes;
353 }
354 }
355 return count;
356}
357
358struct netrx_pending_operations {
359 unsigned copy_prod, copy_cons;
360 unsigned meta_prod, meta_cons;
361 struct gnttab_copy *copy;
362 struct netbk_rx_meta *meta;
363 int copy_off;
364 grant_ref_t copy_gref;
365};
366
367static struct netbk_rx_meta *get_next_rx_buffer(struct xenvif *vif,
368 struct netrx_pending_operations *npo)
369{
370 struct netbk_rx_meta *meta;
371 struct xen_netif_rx_request *req;
372
373 req = RING_GET_REQUEST(&vif->rx, vif->rx.req_cons++);
374
375 meta = npo->meta + npo->meta_prod++;
376 meta->gso_size = 0;
377 meta->size = 0;
378 meta->id = req->id;
379
380 npo->copy_off = 0;
381 npo->copy_gref = req->gref;
382
383 return meta;
384}
385
386/*
387 * Set up the grant operations for this fragment. If it's a flipping
388 * interface, we also set up the unmap request from here.
389 */
390static void netbk_gop_frag_copy(struct xenvif *vif, struct sk_buff *skb,
391 struct netrx_pending_operations *npo,
392 struct page *page, unsigned long size,
393 unsigned long offset, int *head)
394{
395 struct gnttab_copy *copy_gop;
396 struct netbk_rx_meta *meta;
397 /*
Wei Liue34c0242011-12-06 02:04:50 +0000398 * These variables are used iff get_page_ext returns true,
Ian Campbellf942dc22011-03-15 00:06:18 +0000399 * in which case they are guaranteed to be initialized.
400 */
401 unsigned int uninitialized_var(group), uninitialized_var(idx);
402 int foreign = get_page_ext(page, &group, &idx);
403 unsigned long bytes;
404
405 /* Data must not cross a page boundary. */
406 BUG_ON(size + offset > PAGE_SIZE);
407
408 meta = npo->meta + npo->meta_prod - 1;
409
410 while (size > 0) {
411 BUG_ON(npo->copy_off > MAX_BUFFER_OFFSET);
412
413 if (start_new_rx_buffer(npo->copy_off, size, *head)) {
414 /*
415 * Netfront requires there to be some data in the head
416 * buffer.
417 */
418 BUG_ON(*head);
419
420 meta = get_next_rx_buffer(vif, npo);
421 }
422
423 bytes = size;
424 if (npo->copy_off + bytes > MAX_BUFFER_OFFSET)
425 bytes = MAX_BUFFER_OFFSET - npo->copy_off;
426
427 copy_gop = npo->copy + npo->copy_prod++;
428 copy_gop->flags = GNTCOPY_dest_gref;
429 if (foreign) {
430 struct xen_netbk *netbk = &xen_netbk[group];
431 struct pending_tx_info *src_pend;
432
433 src_pend = &netbk->pending_tx_info[idx];
434
435 copy_gop->source.domid = src_pend->vif->domid;
436 copy_gop->source.u.ref = src_pend->req.gref;
437 copy_gop->flags |= GNTCOPY_source_gref;
438 } else {
439 void *vaddr = page_address(page);
440 copy_gop->source.domid = DOMID_SELF;
441 copy_gop->source.u.gmfn = virt_to_mfn(vaddr);
442 }
443 copy_gop->source.offset = offset;
444 copy_gop->dest.domid = vif->domid;
445
446 copy_gop->dest.offset = npo->copy_off;
447 copy_gop->dest.u.ref = npo->copy_gref;
448 copy_gop->len = bytes;
449
450 npo->copy_off += bytes;
451 meta->size += bytes;
452
453 offset += bytes;
454 size -= bytes;
455
456 /* Leave a gap for the GSO descriptor. */
457 if (*head && skb_shinfo(skb)->gso_size && !vif->gso_prefix)
458 vif->rx.req_cons++;
459
460 *head = 0; /* There must be something in this buffer now. */
461
462 }
463}
464
465/*
466 * Prepare an SKB to be transmitted to the frontend.
467 *
468 * This function is responsible for allocating grant operations, meta
469 * structures, etc.
470 *
471 * It returns the number of meta structures consumed. The number of
472 * ring slots used is always equal to the number of meta slots used
473 * plus the number of GSO descriptors used. Currently, we use either
474 * zero GSO descriptors (for non-GSO packets) or one descriptor (for
475 * frontend-side LRO).
476 */
477static int netbk_gop_skb(struct sk_buff *skb,
478 struct netrx_pending_operations *npo)
479{
480 struct xenvif *vif = netdev_priv(skb->dev);
481 int nr_frags = skb_shinfo(skb)->nr_frags;
482 int i;
483 struct xen_netif_rx_request *req;
484 struct netbk_rx_meta *meta;
485 unsigned char *data;
486 int head = 1;
487 int old_meta_prod;
488
489 old_meta_prod = npo->meta_prod;
490
491 /* Set up a GSO prefix descriptor, if necessary */
492 if (skb_shinfo(skb)->gso_size && vif->gso_prefix) {
493 req = RING_GET_REQUEST(&vif->rx, vif->rx.req_cons++);
494 meta = npo->meta + npo->meta_prod++;
495 meta->gso_size = skb_shinfo(skb)->gso_size;
496 meta->size = 0;
497 meta->id = req->id;
498 }
499
500 req = RING_GET_REQUEST(&vif->rx, vif->rx.req_cons++);
501 meta = npo->meta + npo->meta_prod++;
502
503 if (!vif->gso_prefix)
504 meta->gso_size = skb_shinfo(skb)->gso_size;
505 else
506 meta->gso_size = 0;
507
508 meta->size = 0;
509 meta->id = req->id;
510 npo->copy_off = 0;
511 npo->copy_gref = req->gref;
512
513 data = skb->data;
514 while (data < skb_tail_pointer(skb)) {
515 unsigned int offset = offset_in_page(data);
516 unsigned int len = PAGE_SIZE - offset;
517
518 if (data + len > skb_tail_pointer(skb))
519 len = skb_tail_pointer(skb) - data;
520
521 netbk_gop_frag_copy(vif, skb, npo,
522 virt_to_page(data), len, offset, &head);
523 data += len;
524 }
525
526 for (i = 0; i < nr_frags; i++) {
527 netbk_gop_frag_copy(vif, skb, npo,
Ian Campbellea066ad2011-10-05 00:28:46 +0000528 skb_frag_page(&skb_shinfo(skb)->frags[i]),
Eric Dumazet9e903e02011-10-18 21:00:24 +0000529 skb_frag_size(&skb_shinfo(skb)->frags[i]),
Ian Campbellf942dc22011-03-15 00:06:18 +0000530 skb_shinfo(skb)->frags[i].page_offset,
531 &head);
532 }
533
534 return npo->meta_prod - old_meta_prod;
535}
536
537/*
538 * This is a twin to netbk_gop_skb. Assume that netbk_gop_skb was
539 * used to set up the operations on the top of
540 * netrx_pending_operations, which have since been done. Check that
541 * they didn't give any errors and advance over them.
542 */
543static int netbk_check_gop(struct xenvif *vif, int nr_meta_slots,
544 struct netrx_pending_operations *npo)
545{
546 struct gnttab_copy *copy_op;
547 int status = XEN_NETIF_RSP_OKAY;
548 int i;
549
550 for (i = 0; i < nr_meta_slots; i++) {
551 copy_op = npo->copy + npo->copy_cons++;
552 if (copy_op->status != GNTST_okay) {
553 netdev_dbg(vif->dev,
554 "Bad status %d from copy to DOM%d.\n",
555 copy_op->status, vif->domid);
556 status = XEN_NETIF_RSP_ERROR;
557 }
558 }
559
560 return status;
561}
562
563static void netbk_add_frag_responses(struct xenvif *vif, int status,
564 struct netbk_rx_meta *meta,
565 int nr_meta_slots)
566{
567 int i;
568 unsigned long offset;
569
570 /* No fragments used */
571 if (nr_meta_slots <= 1)
572 return;
573
574 nr_meta_slots--;
575
576 for (i = 0; i < nr_meta_slots; i++) {
577 int flags;
578 if (i == nr_meta_slots - 1)
579 flags = 0;
580 else
581 flags = XEN_NETRXF_more_data;
582
583 offset = 0;
584 make_rx_response(vif, meta[i].id, status, offset,
585 meta[i].size, flags);
586 }
587}
588
589struct skb_cb_overlay {
590 int meta_slots_used;
591};
592
593static void xen_netbk_rx_action(struct xen_netbk *netbk)
594{
595 struct xenvif *vif = NULL, *tmp;
596 s8 status;
597 u16 irq, flags;
598 struct xen_netif_rx_response *resp;
599 struct sk_buff_head rxq;
600 struct sk_buff *skb;
601 LIST_HEAD(notify);
602 int ret;
603 int nr_frags;
604 int count;
605 unsigned long offset;
606 struct skb_cb_overlay *sco;
607
608 struct netrx_pending_operations npo = {
609 .copy = netbk->grant_copy_op,
610 .meta = netbk->meta,
611 };
612
613 skb_queue_head_init(&rxq);
614
615 count = 0;
616
617 while ((skb = skb_dequeue(&netbk->rx_queue)) != NULL) {
618 vif = netdev_priv(skb->dev);
619 nr_frags = skb_shinfo(skb)->nr_frags;
620
621 sco = (struct skb_cb_overlay *)skb->cb;
622 sco->meta_slots_used = netbk_gop_skb(skb, &npo);
623
624 count += nr_frags + 1;
625
626 __skb_queue_tail(&rxq, skb);
627
628 /* Filled the batch queue? */
629 if (count + MAX_SKB_FRAGS >= XEN_NETIF_RX_RING_SIZE)
630 break;
631 }
632
633 BUG_ON(npo.meta_prod > ARRAY_SIZE(netbk->meta));
634
635 if (!npo.copy_prod)
636 return;
637
638 BUG_ON(npo.copy_prod > ARRAY_SIZE(netbk->grant_copy_op));
639 ret = HYPERVISOR_grant_table_op(GNTTABOP_copy, &netbk->grant_copy_op,
640 npo.copy_prod);
641 BUG_ON(ret != 0);
642
643 while ((skb = __skb_dequeue(&rxq)) != NULL) {
644 sco = (struct skb_cb_overlay *)skb->cb;
645
646 vif = netdev_priv(skb->dev);
647
648 if (netbk->meta[npo.meta_cons].gso_size && vif->gso_prefix) {
649 resp = RING_GET_RESPONSE(&vif->rx,
650 vif->rx.rsp_prod_pvt++);
651
652 resp->flags = XEN_NETRXF_gso_prefix | XEN_NETRXF_more_data;
653
654 resp->offset = netbk->meta[npo.meta_cons].gso_size;
655 resp->id = netbk->meta[npo.meta_cons].id;
656 resp->status = sco->meta_slots_used;
657
658 npo.meta_cons++;
659 sco->meta_slots_used--;
660 }
661
662
663 vif->dev->stats.tx_bytes += skb->len;
664 vif->dev->stats.tx_packets++;
665
666 status = netbk_check_gop(vif, sco->meta_slots_used, &npo);
667
668 if (sco->meta_slots_used == 1)
669 flags = 0;
670 else
671 flags = XEN_NETRXF_more_data;
672
673 if (skb->ip_summed == CHECKSUM_PARTIAL) /* local packet? */
674 flags |= XEN_NETRXF_csum_blank | XEN_NETRXF_data_validated;
675 else if (skb->ip_summed == CHECKSUM_UNNECESSARY)
676 /* remote but checksummed. */
677 flags |= XEN_NETRXF_data_validated;
678
679 offset = 0;
680 resp = make_rx_response(vif, netbk->meta[npo.meta_cons].id,
681 status, offset,
682 netbk->meta[npo.meta_cons].size,
683 flags);
684
685 if (netbk->meta[npo.meta_cons].gso_size && !vif->gso_prefix) {
686 struct xen_netif_extra_info *gso =
687 (struct xen_netif_extra_info *)
688 RING_GET_RESPONSE(&vif->rx,
689 vif->rx.rsp_prod_pvt++);
690
691 resp->flags |= XEN_NETRXF_extra_info;
692
693 gso->u.gso.size = netbk->meta[npo.meta_cons].gso_size;
694 gso->u.gso.type = XEN_NETIF_GSO_TYPE_TCPV4;
695 gso->u.gso.pad = 0;
696 gso->u.gso.features = 0;
697
698 gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
699 gso->flags = 0;
700 }
701
702 netbk_add_frag_responses(vif, status,
703 netbk->meta + npo.meta_cons + 1,
704 sco->meta_slots_used);
705
706 RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&vif->rx, ret);
707 irq = vif->irq;
708 if (ret && list_empty(&vif->notify_list))
709 list_add_tail(&vif->notify_list, &notify);
710
711 xenvif_notify_tx_completion(vif);
712
713 xenvif_put(vif);
714 npo.meta_cons += sco->meta_slots_used;
715 dev_kfree_skb(skb);
716 }
717
718 list_for_each_entry_safe(vif, tmp, &notify, notify_list) {
719 notify_remote_via_irq(vif->irq);
720 list_del_init(&vif->notify_list);
721 }
722
723 /* More work to do? */
724 if (!skb_queue_empty(&netbk->rx_queue) &&
725 !timer_pending(&netbk->net_timer))
726 xen_netbk_kick_thread(netbk);
727}
728
729void xen_netbk_queue_tx_skb(struct xenvif *vif, struct sk_buff *skb)
730{
731 struct xen_netbk *netbk = vif->netbk;
732
733 skb_queue_tail(&netbk->rx_queue, skb);
734
735 xen_netbk_kick_thread(netbk);
736}
737
738static void xen_netbk_alarm(unsigned long data)
739{
740 struct xen_netbk *netbk = (struct xen_netbk *)data;
741 xen_netbk_kick_thread(netbk);
742}
743
744static int __on_net_schedule_list(struct xenvif *vif)
745{
746 return !list_empty(&vif->schedule_list);
747}
748
749/* Must be called with net_schedule_list_lock held */
750static void remove_from_net_schedule_list(struct xenvif *vif)
751{
752 if (likely(__on_net_schedule_list(vif))) {
753 list_del_init(&vif->schedule_list);
754 xenvif_put(vif);
755 }
756}
757
758static struct xenvif *poll_net_schedule_list(struct xen_netbk *netbk)
759{
760 struct xenvif *vif = NULL;
761
762 spin_lock_irq(&netbk->net_schedule_list_lock);
763 if (list_empty(&netbk->net_schedule_list))
764 goto out;
765
766 vif = list_first_entry(&netbk->net_schedule_list,
767 struct xenvif, schedule_list);
768 if (!vif)
769 goto out;
770
771 xenvif_get(vif);
772
773 remove_from_net_schedule_list(vif);
774out:
775 spin_unlock_irq(&netbk->net_schedule_list_lock);
776 return vif;
777}
778
779void xen_netbk_schedule_xenvif(struct xenvif *vif)
780{
781 unsigned long flags;
782 struct xen_netbk *netbk = vif->netbk;
783
784 if (__on_net_schedule_list(vif))
785 goto kick;
786
787 spin_lock_irqsave(&netbk->net_schedule_list_lock, flags);
788 if (!__on_net_schedule_list(vif) &&
789 likely(xenvif_schedulable(vif))) {
790 list_add_tail(&vif->schedule_list, &netbk->net_schedule_list);
791 xenvif_get(vif);
792 }
793 spin_unlock_irqrestore(&netbk->net_schedule_list_lock, flags);
794
795kick:
796 smp_mb();
797 if ((nr_pending_reqs(netbk) < (MAX_PENDING_REQS/2)) &&
798 !list_empty(&netbk->net_schedule_list))
799 xen_netbk_kick_thread(netbk);
800}
801
802void xen_netbk_deschedule_xenvif(struct xenvif *vif)
803{
804 struct xen_netbk *netbk = vif->netbk;
805 spin_lock_irq(&netbk->net_schedule_list_lock);
806 remove_from_net_schedule_list(vif);
807 spin_unlock_irq(&netbk->net_schedule_list_lock);
808}
809
810void xen_netbk_check_rx_xenvif(struct xenvif *vif)
811{
812 int more_to_do;
813
814 RING_FINAL_CHECK_FOR_REQUESTS(&vif->tx, more_to_do);
815
816 if (more_to_do)
817 xen_netbk_schedule_xenvif(vif);
818}
819
820static void tx_add_credit(struct xenvif *vif)
821{
822 unsigned long max_burst, max_credit;
823
824 /*
825 * Allow a burst big enough to transmit a jumbo packet of up to 128kB.
826 * Otherwise the interface can seize up due to insufficient credit.
827 */
828 max_burst = RING_GET_REQUEST(&vif->tx, vif->tx.req_cons)->size;
829 max_burst = min(max_burst, 131072UL);
830 max_burst = max(max_burst, vif->credit_bytes);
831
832 /* Take care that adding a new chunk of credit doesn't wrap to zero. */
833 max_credit = vif->remaining_credit + vif->credit_bytes;
834 if (max_credit < vif->remaining_credit)
835 max_credit = ULONG_MAX; /* wrapped: clamp to ULONG_MAX */
836
837 vif->remaining_credit = min(max_credit, max_burst);
838}
839
840static void tx_credit_callback(unsigned long data)
841{
842 struct xenvif *vif = (struct xenvif *)data;
843 tx_add_credit(vif);
844 xen_netbk_check_rx_xenvif(vif);
845}
846
847static void netbk_tx_err(struct xenvif *vif,
848 struct xen_netif_tx_request *txp, RING_IDX end)
849{
850 RING_IDX cons = vif->tx.req_cons;
851
852 do {
853 make_tx_response(vif, txp, XEN_NETIF_RSP_ERROR);
854 if (cons >= end)
855 break;
856 txp = RING_GET_REQUEST(&vif->tx, cons++);
857 } while (1);
858 vif->tx.req_cons = cons;
859 xen_netbk_check_rx_xenvif(vif);
860 xenvif_put(vif);
861}
862
863static int netbk_count_requests(struct xenvif *vif,
864 struct xen_netif_tx_request *first,
865 struct xen_netif_tx_request *txp,
866 int work_to_do)
867{
868 RING_IDX cons = vif->tx.req_cons;
869 int frags = 0;
870
871 if (!(first->flags & XEN_NETTXF_more_data))
872 return 0;
873
874 do {
875 if (frags >= work_to_do) {
876 netdev_dbg(vif->dev, "Need more frags\n");
877 return -frags;
878 }
879
880 if (unlikely(frags >= MAX_SKB_FRAGS)) {
881 netdev_dbg(vif->dev, "Too many frags\n");
882 return -frags;
883 }
884
885 memcpy(txp, RING_GET_REQUEST(&vif->tx, cons + frags),
886 sizeof(*txp));
887 if (txp->size > first->size) {
888 netdev_dbg(vif->dev, "Frags galore\n");
889 return -frags;
890 }
891
892 first->size -= txp->size;
893 frags++;
894
895 if (unlikely((txp->offset + txp->size) > PAGE_SIZE)) {
896 netdev_dbg(vif->dev, "txp->offset: %x, size: %u\n",
897 txp->offset, txp->size);
898 return -frags;
899 }
900 } while ((txp++)->flags & XEN_NETTXF_more_data);
901 return frags;
902}
903
904static struct page *xen_netbk_alloc_page(struct xen_netbk *netbk,
905 struct sk_buff *skb,
Ian Campbellea066ad2011-10-05 00:28:46 +0000906 u16 pending_idx)
Ian Campbellf942dc22011-03-15 00:06:18 +0000907{
908 struct page *page;
909 page = alloc_page(GFP_KERNEL|__GFP_COLD);
910 if (!page)
911 return NULL;
912 set_page_ext(page, netbk, pending_idx);
913 netbk->mmap_pages[pending_idx] = page;
914 return page;
915}
916
917static struct gnttab_copy *xen_netbk_get_requests(struct xen_netbk *netbk,
918 struct xenvif *vif,
919 struct sk_buff *skb,
920 struct xen_netif_tx_request *txp,
921 struct gnttab_copy *gop)
922{
923 struct skb_shared_info *shinfo = skb_shinfo(skb);
924 skb_frag_t *frags = shinfo->frags;
Ian Campbellea066ad2011-10-05 00:28:46 +0000925 u16 pending_idx = *((u16 *)skb->data);
Ian Campbellf942dc22011-03-15 00:06:18 +0000926 int i, start;
927
928 /* Skip first skb fragment if it is on same page as header fragment. */
Ian Campbellea066ad2011-10-05 00:28:46 +0000929 start = (frag_get_pending_idx(&shinfo->frags[0]) == pending_idx);
Ian Campbellf942dc22011-03-15 00:06:18 +0000930
931 for (i = start; i < shinfo->nr_frags; i++, txp++) {
932 struct page *page;
933 pending_ring_idx_t index;
934 struct pending_tx_info *pending_tx_info =
935 netbk->pending_tx_info;
936
937 index = pending_index(netbk->pending_cons++);
938 pending_idx = netbk->pending_ring[index];
939 page = xen_netbk_alloc_page(netbk, skb, pending_idx);
940 if (!page)
941 return NULL;
942
Ian Campbellf942dc22011-03-15 00:06:18 +0000943 gop->source.u.ref = txp->gref;
944 gop->source.domid = vif->domid;
945 gop->source.offset = txp->offset;
946
947 gop->dest.u.gmfn = virt_to_mfn(page_address(page));
948 gop->dest.domid = DOMID_SELF;
949 gop->dest.offset = txp->offset;
950
951 gop->len = txp->size;
952 gop->flags = GNTCOPY_source_gref;
953
954 gop++;
955
956 memcpy(&pending_tx_info[pending_idx].req, txp, sizeof(*txp));
957 xenvif_get(vif);
958 pending_tx_info[pending_idx].vif = vif;
Ian Campbellea066ad2011-10-05 00:28:46 +0000959 frag_set_pending_idx(&frags[i], pending_idx);
Ian Campbellf942dc22011-03-15 00:06:18 +0000960 }
961
962 return gop;
963}
964
965static int xen_netbk_tx_check_gop(struct xen_netbk *netbk,
966 struct sk_buff *skb,
967 struct gnttab_copy **gopp)
968{
969 struct gnttab_copy *gop = *gopp;
Ian Campbellea066ad2011-10-05 00:28:46 +0000970 u16 pending_idx = *((u16 *)skb->data);
Ian Campbellf942dc22011-03-15 00:06:18 +0000971 struct pending_tx_info *pending_tx_info = netbk->pending_tx_info;
972 struct xenvif *vif = pending_tx_info[pending_idx].vif;
973 struct xen_netif_tx_request *txp;
974 struct skb_shared_info *shinfo = skb_shinfo(skb);
975 int nr_frags = shinfo->nr_frags;
976 int i, err, start;
977
978 /* Check status of header. */
979 err = gop->status;
980 if (unlikely(err)) {
981 pending_ring_idx_t index;
982 index = pending_index(netbk->pending_prod++);
983 txp = &pending_tx_info[pending_idx].req;
984 make_tx_response(vif, txp, XEN_NETIF_RSP_ERROR);
985 netbk->pending_ring[index] = pending_idx;
986 xenvif_put(vif);
987 }
988
989 /* Skip first skb fragment if it is on same page as header fragment. */
Ian Campbellea066ad2011-10-05 00:28:46 +0000990 start = (frag_get_pending_idx(&shinfo->frags[0]) == pending_idx);
Ian Campbellf942dc22011-03-15 00:06:18 +0000991
992 for (i = start; i < nr_frags; i++) {
993 int j, newerr;
994 pending_ring_idx_t index;
995
Ian Campbellea066ad2011-10-05 00:28:46 +0000996 pending_idx = frag_get_pending_idx(&shinfo->frags[i]);
Ian Campbellf942dc22011-03-15 00:06:18 +0000997
998 /* Check error status: if okay then remember grant handle. */
999 newerr = (++gop)->status;
1000 if (likely(!newerr)) {
1001 /* Had a previous error? Invalidate this fragment. */
1002 if (unlikely(err))
1003 xen_netbk_idx_release(netbk, pending_idx);
1004 continue;
1005 }
1006
1007 /* Error on this fragment: respond to client with an error. */
1008 txp = &netbk->pending_tx_info[pending_idx].req;
1009 make_tx_response(vif, txp, XEN_NETIF_RSP_ERROR);
1010 index = pending_index(netbk->pending_prod++);
1011 netbk->pending_ring[index] = pending_idx;
1012 xenvif_put(vif);
1013
1014 /* Not the first error? Preceding frags already invalidated. */
1015 if (err)
1016 continue;
1017
1018 /* First error: invalidate header and preceding fragments. */
1019 pending_idx = *((u16 *)skb->data);
1020 xen_netbk_idx_release(netbk, pending_idx);
1021 for (j = start; j < i; j++) {
Jan Beulich5ccb3ea2011-11-18 05:42:05 +00001022 pending_idx = frag_get_pending_idx(&shinfo->frags[j]);
Ian Campbellf942dc22011-03-15 00:06:18 +00001023 xen_netbk_idx_release(netbk, pending_idx);
1024 }
1025
1026 /* Remember the error: invalidate all subsequent fragments. */
1027 err = newerr;
1028 }
1029
1030 *gopp = gop + 1;
1031 return err;
1032}
1033
1034static void xen_netbk_fill_frags(struct xen_netbk *netbk, struct sk_buff *skb)
1035{
1036 struct skb_shared_info *shinfo = skb_shinfo(skb);
1037 int nr_frags = shinfo->nr_frags;
1038 int i;
1039
1040 for (i = 0; i < nr_frags; i++) {
1041 skb_frag_t *frag = shinfo->frags + i;
1042 struct xen_netif_tx_request *txp;
Ian Campbellea066ad2011-10-05 00:28:46 +00001043 struct page *page;
1044 u16 pending_idx;
Ian Campbellf942dc22011-03-15 00:06:18 +00001045
Ian Campbellea066ad2011-10-05 00:28:46 +00001046 pending_idx = frag_get_pending_idx(frag);
Ian Campbellf942dc22011-03-15 00:06:18 +00001047
1048 txp = &netbk->pending_tx_info[pending_idx].req;
Ian Campbellea066ad2011-10-05 00:28:46 +00001049 page = virt_to_page(idx_to_kaddr(netbk, pending_idx));
1050 __skb_fill_page_desc(skb, i, page, txp->offset, txp->size);
Ian Campbellf942dc22011-03-15 00:06:18 +00001051 skb->len += txp->size;
1052 skb->data_len += txp->size;
1053 skb->truesize += txp->size;
1054
1055 /* Take an extra reference to offset xen_netbk_idx_release */
1056 get_page(netbk->mmap_pages[pending_idx]);
1057 xen_netbk_idx_release(netbk, pending_idx);
1058 }
1059}
1060
1061static int xen_netbk_get_extras(struct xenvif *vif,
1062 struct xen_netif_extra_info *extras,
1063 int work_to_do)
1064{
1065 struct xen_netif_extra_info extra;
1066 RING_IDX cons = vif->tx.req_cons;
1067
1068 do {
1069 if (unlikely(work_to_do-- <= 0)) {
1070 netdev_dbg(vif->dev, "Missing extra info\n");
1071 return -EBADR;
1072 }
1073
1074 memcpy(&extra, RING_GET_REQUEST(&vif->tx, cons),
1075 sizeof(extra));
1076 if (unlikely(!extra.type ||
1077 extra.type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
1078 vif->tx.req_cons = ++cons;
1079 netdev_dbg(vif->dev,
1080 "Invalid extra type: %d\n", extra.type);
1081 return -EINVAL;
1082 }
1083
1084 memcpy(&extras[extra.type - 1], &extra, sizeof(extra));
1085 vif->tx.req_cons = ++cons;
1086 } while (extra.flags & XEN_NETIF_EXTRA_FLAG_MORE);
1087
1088 return work_to_do;
1089}
1090
1091static int netbk_set_skb_gso(struct xenvif *vif,
1092 struct sk_buff *skb,
1093 struct xen_netif_extra_info *gso)
1094{
1095 if (!gso->u.gso.size) {
1096 netdev_dbg(vif->dev, "GSO size must not be zero.\n");
1097 return -EINVAL;
1098 }
1099
1100 /* Currently only TCPv4 S.O. is supported. */
1101 if (gso->u.gso.type != XEN_NETIF_GSO_TYPE_TCPV4) {
1102 netdev_dbg(vif->dev, "Bad GSO type %d.\n", gso->u.gso.type);
1103 return -EINVAL;
1104 }
1105
1106 skb_shinfo(skb)->gso_size = gso->u.gso.size;
1107 skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
1108
1109 /* Header must be checked, and gso_segs computed. */
1110 skb_shinfo(skb)->gso_type |= SKB_GSO_DODGY;
1111 skb_shinfo(skb)->gso_segs = 0;
1112
1113 return 0;
1114}
1115
1116static int checksum_setup(struct xenvif *vif, struct sk_buff *skb)
1117{
1118 struct iphdr *iph;
1119 unsigned char *th;
1120 int err = -EPROTO;
1121 int recalculate_partial_csum = 0;
1122
1123 /*
1124 * A GSO SKB must be CHECKSUM_PARTIAL. However some buggy
1125 * peers can fail to set NETRXF_csum_blank when sending a GSO
1126 * frame. In this case force the SKB to CHECKSUM_PARTIAL and
1127 * recalculate the partial checksum.
1128 */
1129 if (skb->ip_summed != CHECKSUM_PARTIAL && skb_is_gso(skb)) {
1130 vif->rx_gso_checksum_fixup++;
1131 skb->ip_summed = CHECKSUM_PARTIAL;
1132 recalculate_partial_csum = 1;
1133 }
1134
1135 /* A non-CHECKSUM_PARTIAL SKB does not require setup. */
1136 if (skb->ip_summed != CHECKSUM_PARTIAL)
1137 return 0;
1138
1139 if (skb->protocol != htons(ETH_P_IP))
1140 goto out;
1141
1142 iph = (void *)skb->data;
1143 th = skb->data + 4 * iph->ihl;
1144 if (th >= skb_tail_pointer(skb))
1145 goto out;
1146
1147 skb->csum_start = th - skb->head;
1148 switch (iph->protocol) {
1149 case IPPROTO_TCP:
1150 skb->csum_offset = offsetof(struct tcphdr, check);
1151
1152 if (recalculate_partial_csum) {
1153 struct tcphdr *tcph = (struct tcphdr *)th;
1154 tcph->check = ~csum_tcpudp_magic(iph->saddr, iph->daddr,
1155 skb->len - iph->ihl*4,
1156 IPPROTO_TCP, 0);
1157 }
1158 break;
1159 case IPPROTO_UDP:
1160 skb->csum_offset = offsetof(struct udphdr, check);
1161
1162 if (recalculate_partial_csum) {
1163 struct udphdr *udph = (struct udphdr *)th;
1164 udph->check = ~csum_tcpudp_magic(iph->saddr, iph->daddr,
1165 skb->len - iph->ihl*4,
1166 IPPROTO_UDP, 0);
1167 }
1168 break;
1169 default:
1170 if (net_ratelimit())
1171 netdev_err(vif->dev,
1172 "Attempting to checksum a non-TCP/UDP packet, dropping a protocol %d packet\n",
1173 iph->protocol);
1174 goto out;
1175 }
1176
1177 if ((th + skb->csum_offset + 2) > skb_tail_pointer(skb))
1178 goto out;
1179
1180 err = 0;
1181
1182out:
1183 return err;
1184}
1185
1186static bool tx_credit_exceeded(struct xenvif *vif, unsigned size)
1187{
1188 unsigned long now = jiffies;
1189 unsigned long next_credit =
1190 vif->credit_timeout.expires +
1191 msecs_to_jiffies(vif->credit_usec / 1000);
1192
1193 /* Timer could already be pending in rare cases. */
1194 if (timer_pending(&vif->credit_timeout))
1195 return true;
1196
1197 /* Passed the point where we can replenish credit? */
1198 if (time_after_eq(now, next_credit)) {
1199 vif->credit_timeout.expires = now;
1200 tx_add_credit(vif);
1201 }
1202
1203 /* Still too big to send right now? Set a callback. */
1204 if (size > vif->remaining_credit) {
1205 vif->credit_timeout.data =
1206 (unsigned long)vif;
1207 vif->credit_timeout.function =
1208 tx_credit_callback;
1209 mod_timer(&vif->credit_timeout,
1210 next_credit);
1211
1212 return true;
1213 }
1214
1215 return false;
1216}
1217
1218static unsigned xen_netbk_tx_build_gops(struct xen_netbk *netbk)
1219{
1220 struct gnttab_copy *gop = netbk->tx_copy_ops, *request_gop;
1221 struct sk_buff *skb;
1222 int ret;
1223
1224 while (((nr_pending_reqs(netbk) + MAX_SKB_FRAGS) < MAX_PENDING_REQS) &&
1225 !list_empty(&netbk->net_schedule_list)) {
1226 struct xenvif *vif;
1227 struct xen_netif_tx_request txreq;
1228 struct xen_netif_tx_request txfrags[MAX_SKB_FRAGS];
1229 struct page *page;
1230 struct xen_netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX-1];
1231 u16 pending_idx;
1232 RING_IDX idx;
1233 int work_to_do;
1234 unsigned int data_len;
1235 pending_ring_idx_t index;
1236
1237 /* Get a netif from the list with work to do. */
1238 vif = poll_net_schedule_list(netbk);
1239 if (!vif)
1240 continue;
1241
1242 RING_FINAL_CHECK_FOR_REQUESTS(&vif->tx, work_to_do);
1243 if (!work_to_do) {
1244 xenvif_put(vif);
1245 continue;
1246 }
1247
1248 idx = vif->tx.req_cons;
1249 rmb(); /* Ensure that we see the request before we copy it. */
1250 memcpy(&txreq, RING_GET_REQUEST(&vif->tx, idx), sizeof(txreq));
1251
1252 /* Credit-based scheduling. */
1253 if (txreq.size > vif->remaining_credit &&
1254 tx_credit_exceeded(vif, txreq.size)) {
1255 xenvif_put(vif);
1256 continue;
1257 }
1258
1259 vif->remaining_credit -= txreq.size;
1260
1261 work_to_do--;
1262 vif->tx.req_cons = ++idx;
1263
1264 memset(extras, 0, sizeof(extras));
1265 if (txreq.flags & XEN_NETTXF_extra_info) {
1266 work_to_do = xen_netbk_get_extras(vif, extras,
1267 work_to_do);
1268 idx = vif->tx.req_cons;
1269 if (unlikely(work_to_do < 0)) {
1270 netbk_tx_err(vif, &txreq, idx);
1271 continue;
1272 }
1273 }
1274
1275 ret = netbk_count_requests(vif, &txreq, txfrags, work_to_do);
1276 if (unlikely(ret < 0)) {
1277 netbk_tx_err(vif, &txreq, idx - ret);
1278 continue;
1279 }
1280 idx += ret;
1281
1282 if (unlikely(txreq.size < ETH_HLEN)) {
1283 netdev_dbg(vif->dev,
1284 "Bad packet size: %d\n", txreq.size);
1285 netbk_tx_err(vif, &txreq, idx);
1286 continue;
1287 }
1288
1289 /* No crossing a page as the payload mustn't fragment. */
1290 if (unlikely((txreq.offset + txreq.size) > PAGE_SIZE)) {
1291 netdev_dbg(vif->dev,
1292 "txreq.offset: %x, size: %u, end: %lu\n",
1293 txreq.offset, txreq.size,
1294 (txreq.offset&~PAGE_MASK) + txreq.size);
1295 netbk_tx_err(vif, &txreq, idx);
1296 continue;
1297 }
1298
1299 index = pending_index(netbk->pending_cons);
1300 pending_idx = netbk->pending_ring[index];
1301
1302 data_len = (txreq.size > PKT_PROT_LEN &&
1303 ret < MAX_SKB_FRAGS) ?
1304 PKT_PROT_LEN : txreq.size;
1305
1306 skb = alloc_skb(data_len + NET_SKB_PAD + NET_IP_ALIGN,
1307 GFP_ATOMIC | __GFP_NOWARN);
1308 if (unlikely(skb == NULL)) {
1309 netdev_dbg(vif->dev,
1310 "Can't allocate a skb in start_xmit.\n");
1311 netbk_tx_err(vif, &txreq, idx);
1312 break;
1313 }
1314
1315 /* Packets passed to netif_rx() must have some headroom. */
1316 skb_reserve(skb, NET_SKB_PAD + NET_IP_ALIGN);
1317
1318 if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
1319 struct xen_netif_extra_info *gso;
1320 gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];
1321
1322 if (netbk_set_skb_gso(vif, skb, gso)) {
1323 kfree_skb(skb);
1324 netbk_tx_err(vif, &txreq, idx);
1325 continue;
1326 }
1327 }
1328
1329 /* XXX could copy straight to head */
1330 page = xen_netbk_alloc_page(netbk, skb, pending_idx);
1331 if (!page) {
1332 kfree_skb(skb);
1333 netbk_tx_err(vif, &txreq, idx);
1334 continue;
1335 }
1336
Ian Campbellf942dc22011-03-15 00:06:18 +00001337 gop->source.u.ref = txreq.gref;
1338 gop->source.domid = vif->domid;
1339 gop->source.offset = txreq.offset;
1340
1341 gop->dest.u.gmfn = virt_to_mfn(page_address(page));
1342 gop->dest.domid = DOMID_SELF;
1343 gop->dest.offset = txreq.offset;
1344
1345 gop->len = txreq.size;
1346 gop->flags = GNTCOPY_source_gref;
1347
1348 gop++;
1349
1350 memcpy(&netbk->pending_tx_info[pending_idx].req,
1351 &txreq, sizeof(txreq));
1352 netbk->pending_tx_info[pending_idx].vif = vif;
1353 *((u16 *)skb->data) = pending_idx;
1354
1355 __skb_put(skb, data_len);
1356
1357 skb_shinfo(skb)->nr_frags = ret;
1358 if (data_len < txreq.size) {
1359 skb_shinfo(skb)->nr_frags++;
Ian Campbellea066ad2011-10-05 00:28:46 +00001360 frag_set_pending_idx(&skb_shinfo(skb)->frags[0],
1361 pending_idx);
Ian Campbellf942dc22011-03-15 00:06:18 +00001362 } else {
Ian Campbellea066ad2011-10-05 00:28:46 +00001363 frag_set_pending_idx(&skb_shinfo(skb)->frags[0],
1364 INVALID_PENDING_IDX);
Ian Campbellf942dc22011-03-15 00:06:18 +00001365 }
1366
1367 __skb_queue_tail(&netbk->tx_queue, skb);
1368
1369 netbk->pending_cons++;
1370
1371 request_gop = xen_netbk_get_requests(netbk, vif,
1372 skb, txfrags, gop);
1373 if (request_gop == NULL) {
1374 kfree_skb(skb);
1375 netbk_tx_err(vif, &txreq, idx);
1376 continue;
1377 }
1378 gop = request_gop;
1379
1380 vif->tx.req_cons = idx;
1381 xen_netbk_check_rx_xenvif(vif);
1382
1383 if ((gop-netbk->tx_copy_ops) >= ARRAY_SIZE(netbk->tx_copy_ops))
1384 break;
1385 }
1386
1387 return gop - netbk->tx_copy_ops;
1388}
1389
1390static void xen_netbk_tx_submit(struct xen_netbk *netbk)
1391{
1392 struct gnttab_copy *gop = netbk->tx_copy_ops;
1393 struct sk_buff *skb;
1394
1395 while ((skb = __skb_dequeue(&netbk->tx_queue)) != NULL) {
1396 struct xen_netif_tx_request *txp;
1397 struct xenvif *vif;
1398 u16 pending_idx;
1399 unsigned data_len;
1400
1401 pending_idx = *((u16 *)skb->data);
1402 vif = netbk->pending_tx_info[pending_idx].vif;
1403 txp = &netbk->pending_tx_info[pending_idx].req;
1404
1405 /* Check the remap error code. */
1406 if (unlikely(xen_netbk_tx_check_gop(netbk, skb, &gop))) {
1407 netdev_dbg(vif->dev, "netback grant failed.\n");
1408 skb_shinfo(skb)->nr_frags = 0;
1409 kfree_skb(skb);
1410 continue;
1411 }
1412
1413 data_len = skb->len;
1414 memcpy(skb->data,
1415 (void *)(idx_to_kaddr(netbk, pending_idx)|txp->offset),
1416 data_len);
1417 if (data_len < txp->size) {
1418 /* Append the packet payload as a fragment. */
1419 txp->offset += data_len;
1420 txp->size -= data_len;
1421 } else {
1422 /* Schedule a response immediately. */
1423 xen_netbk_idx_release(netbk, pending_idx);
1424 }
1425
1426 if (txp->flags & XEN_NETTXF_csum_blank)
1427 skb->ip_summed = CHECKSUM_PARTIAL;
1428 else if (txp->flags & XEN_NETTXF_data_validated)
1429 skb->ip_summed = CHECKSUM_UNNECESSARY;
1430
1431 xen_netbk_fill_frags(netbk, skb);
1432
1433 /*
1434 * If the initial fragment was < PKT_PROT_LEN then
1435 * pull through some bytes from the other fragments to
1436 * increase the linear region to PKT_PROT_LEN bytes.
1437 */
1438 if (skb_headlen(skb) < PKT_PROT_LEN && skb_is_nonlinear(skb)) {
1439 int target = min_t(int, skb->len, PKT_PROT_LEN);
1440 __pskb_pull_tail(skb, target - skb_headlen(skb));
1441 }
1442
1443 skb->dev = vif->dev;
1444 skb->protocol = eth_type_trans(skb, skb->dev);
1445
1446 if (checksum_setup(vif, skb)) {
1447 netdev_dbg(vif->dev,
1448 "Can't setup checksum in net_tx_action\n");
1449 kfree_skb(skb);
1450 continue;
1451 }
1452
1453 vif->dev->stats.rx_bytes += skb->len;
1454 vif->dev->stats.rx_packets++;
1455
1456 xenvif_receive_skb(vif, skb);
1457 }
1458}
1459
1460/* Called after netfront has transmitted */
1461static void xen_netbk_tx_action(struct xen_netbk *netbk)
1462{
1463 unsigned nr_gops;
1464 int ret;
1465
1466 nr_gops = xen_netbk_tx_build_gops(netbk);
1467
1468 if (nr_gops == 0)
1469 return;
1470 ret = HYPERVISOR_grant_table_op(GNTTABOP_copy,
1471 netbk->tx_copy_ops, nr_gops);
1472 BUG_ON(ret);
1473
1474 xen_netbk_tx_submit(netbk);
1475
1476}
1477
1478static void xen_netbk_idx_release(struct xen_netbk *netbk, u16 pending_idx)
1479{
1480 struct xenvif *vif;
1481 struct pending_tx_info *pending_tx_info;
1482 pending_ring_idx_t index;
1483
1484 /* Already complete? */
1485 if (netbk->mmap_pages[pending_idx] == NULL)
1486 return;
1487
1488 pending_tx_info = &netbk->pending_tx_info[pending_idx];
1489
1490 vif = pending_tx_info->vif;
1491
1492 make_tx_response(vif, &pending_tx_info->req, XEN_NETIF_RSP_OKAY);
1493
1494 index = pending_index(netbk->pending_prod++);
1495 netbk->pending_ring[index] = pending_idx;
1496
1497 xenvif_put(vif);
1498
1499 netbk->mmap_pages[pending_idx]->mapping = 0;
1500 put_page(netbk->mmap_pages[pending_idx]);
1501 netbk->mmap_pages[pending_idx] = NULL;
1502}
1503
1504static void make_tx_response(struct xenvif *vif,
1505 struct xen_netif_tx_request *txp,
1506 s8 st)
1507{
1508 RING_IDX i = vif->tx.rsp_prod_pvt;
1509 struct xen_netif_tx_response *resp;
1510 int notify;
1511
1512 resp = RING_GET_RESPONSE(&vif->tx, i);
1513 resp->id = txp->id;
1514 resp->status = st;
1515
1516 if (txp->flags & XEN_NETTXF_extra_info)
1517 RING_GET_RESPONSE(&vif->tx, ++i)->status = XEN_NETIF_RSP_NULL;
1518
1519 vif->tx.rsp_prod_pvt = ++i;
1520 RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&vif->tx, notify);
1521 if (notify)
1522 notify_remote_via_irq(vif->irq);
1523}
1524
1525static struct xen_netif_rx_response *make_rx_response(struct xenvif *vif,
1526 u16 id,
1527 s8 st,
1528 u16 offset,
1529 u16 size,
1530 u16 flags)
1531{
1532 RING_IDX i = vif->rx.rsp_prod_pvt;
1533 struct xen_netif_rx_response *resp;
1534
1535 resp = RING_GET_RESPONSE(&vif->rx, i);
1536 resp->offset = offset;
1537 resp->flags = flags;
1538 resp->id = id;
1539 resp->status = (s16)size;
1540 if (st < 0)
1541 resp->status = (s16)st;
1542
1543 vif->rx.rsp_prod_pvt = ++i;
1544
1545 return resp;
1546}
1547
1548static inline int rx_work_todo(struct xen_netbk *netbk)
1549{
1550 return !skb_queue_empty(&netbk->rx_queue);
1551}
1552
1553static inline int tx_work_todo(struct xen_netbk *netbk)
1554{
1555
1556 if (((nr_pending_reqs(netbk) + MAX_SKB_FRAGS) < MAX_PENDING_REQS) &&
1557 !list_empty(&netbk->net_schedule_list))
1558 return 1;
1559
1560 return 0;
1561}
1562
1563static int xen_netbk_kthread(void *data)
1564{
1565 struct xen_netbk *netbk = data;
1566 while (!kthread_should_stop()) {
1567 wait_event_interruptible(netbk->wq,
1568 rx_work_todo(netbk) ||
1569 tx_work_todo(netbk) ||
1570 kthread_should_stop());
1571 cond_resched();
1572
1573 if (kthread_should_stop())
1574 break;
1575
1576 if (rx_work_todo(netbk))
1577 xen_netbk_rx_action(netbk);
1578
1579 if (tx_work_todo(netbk))
1580 xen_netbk_tx_action(netbk);
1581 }
1582
1583 return 0;
1584}
1585
1586void xen_netbk_unmap_frontend_rings(struct xenvif *vif)
1587{
David Vrabelc9d63692011-09-29 16:53:31 +01001588 if (vif->tx.sring)
1589 xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(vif),
1590 vif->tx.sring);
1591 if (vif->rx.sring)
1592 xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(vif),
1593 vif->rx.sring);
Ian Campbellf942dc22011-03-15 00:06:18 +00001594}
1595
1596int xen_netbk_map_frontend_rings(struct xenvif *vif,
1597 grant_ref_t tx_ring_ref,
1598 grant_ref_t rx_ring_ref)
1599{
David Vrabelc9d63692011-09-29 16:53:31 +01001600 void *addr;
Ian Campbellf942dc22011-03-15 00:06:18 +00001601 struct xen_netif_tx_sring *txs;
1602 struct xen_netif_rx_sring *rxs;
1603
1604 int err = -ENOMEM;
1605
David Vrabelc9d63692011-09-29 16:53:31 +01001606 err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(vif),
1607 tx_ring_ref, &addr);
1608 if (err)
Ian Campbellf942dc22011-03-15 00:06:18 +00001609 goto err;
1610
David Vrabelc9d63692011-09-29 16:53:31 +01001611 txs = (struct xen_netif_tx_sring *)addr;
Ian Campbellf942dc22011-03-15 00:06:18 +00001612 BACK_RING_INIT(&vif->tx, txs, PAGE_SIZE);
1613
David Vrabelc9d63692011-09-29 16:53:31 +01001614 err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(vif),
1615 rx_ring_ref, &addr);
1616 if (err)
Ian Campbellf942dc22011-03-15 00:06:18 +00001617 goto err;
Ian Campbellf942dc22011-03-15 00:06:18 +00001618
David Vrabelc9d63692011-09-29 16:53:31 +01001619 rxs = (struct xen_netif_rx_sring *)addr;
Ian Campbellf942dc22011-03-15 00:06:18 +00001620 BACK_RING_INIT(&vif->rx, rxs, PAGE_SIZE);
1621
David Vrabelc9d63692011-09-29 16:53:31 +01001622 vif->rx_req_cons_peek = 0;
1623
Ian Campbellf942dc22011-03-15 00:06:18 +00001624 return 0;
1625
1626err:
1627 xen_netbk_unmap_frontend_rings(vif);
1628 return err;
1629}
1630
1631static int __init netback_init(void)
1632{
1633 int i;
1634 int rc = 0;
1635 int group;
1636
Daniel De Graaf2a14b2442011-12-14 15:12:13 -05001637 if (!xen_domain())
Ian Campbellf942dc22011-03-15 00:06:18 +00001638 return -ENODEV;
1639
1640 xen_netbk_group_nr = num_online_cpus();
1641 xen_netbk = vzalloc(sizeof(struct xen_netbk) * xen_netbk_group_nr);
Joe Perchese404dec2012-01-29 12:56:23 +00001642 if (!xen_netbk)
Ian Campbellf942dc22011-03-15 00:06:18 +00001643 return -ENOMEM;
Ian Campbellf942dc22011-03-15 00:06:18 +00001644
1645 for (group = 0; group < xen_netbk_group_nr; group++) {
1646 struct xen_netbk *netbk = &xen_netbk[group];
1647 skb_queue_head_init(&netbk->rx_queue);
1648 skb_queue_head_init(&netbk->tx_queue);
1649
1650 init_timer(&netbk->net_timer);
1651 netbk->net_timer.data = (unsigned long)netbk;
1652 netbk->net_timer.function = xen_netbk_alarm;
1653
1654 netbk->pending_cons = 0;
1655 netbk->pending_prod = MAX_PENDING_REQS;
1656 for (i = 0; i < MAX_PENDING_REQS; i++)
1657 netbk->pending_ring[i] = i;
1658
1659 init_waitqueue_head(&netbk->wq);
1660 netbk->task = kthread_create(xen_netbk_kthread,
1661 (void *)netbk,
1662 "netback/%u", group);
1663
1664 if (IS_ERR(netbk->task)) {
Wei Liu6b84bd12011-12-05 06:57:44 +00001665 printk(KERN_ALERT "kthread_create() fails at netback\n");
Ian Campbellf942dc22011-03-15 00:06:18 +00001666 del_timer(&netbk->net_timer);
1667 rc = PTR_ERR(netbk->task);
1668 goto failed_init;
1669 }
1670
1671 kthread_bind(netbk->task, group);
1672
1673 INIT_LIST_HEAD(&netbk->net_schedule_list);
1674
1675 spin_lock_init(&netbk->net_schedule_list_lock);
1676
1677 atomic_set(&netbk->netfront_count, 0);
1678
1679 wake_up_process(netbk->task);
1680 }
1681
1682 rc = xenvif_xenbus_init();
1683 if (rc)
1684 goto failed_init;
1685
1686 return 0;
1687
1688failed_init:
1689 while (--group >= 0) {
1690 struct xen_netbk *netbk = &xen_netbk[group];
1691 for (i = 0; i < MAX_PENDING_REQS; i++) {
1692 if (netbk->mmap_pages[i])
1693 __free_page(netbk->mmap_pages[i]);
1694 }
1695 del_timer(&netbk->net_timer);
1696 kthread_stop(netbk->task);
1697 }
1698 vfree(xen_netbk);
1699 return rc;
1700
1701}
1702
1703module_init(netback_init);
1704
1705MODULE_LICENSE("Dual BSD/GPL");
Bastian Blankf984cec2011-06-30 11:19:09 -07001706MODULE_ALIAS("xen-backend:vif");