drivers/net/xen-netback/netback.c

   1 /*
   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 #include <linux/highmem.h>
  41
  42 #include <net/tcp.h>
  43
  44 #include <xen/xen.h>
  45 #include <xen/events.h>
  46 #include <xen/interface/memory.h>
  47 #include <xen/page.h>
  48
  49 #include <asm/xen/hypercall.h>
  50
  51 /* Provide an option to disable split event channels at load time as
  52  * event channels are limited resource. Split event channels are
  53  * enabled by default.
  54  */
  55 bool separate_tx_rx_irq = true;
  56 module_param(separate_tx_rx_irq, bool, 0644);
  57
  58 /* The time that packets can stay on the guest Rx internal queue
  59  * before they are dropped.
  60  */
  61 unsigned int rx_drain_timeout_msecs = 10000;
  62 module_param(rx_drain_timeout_msecs, uint, 0444);
  63
  64 /* The length of time before the frontend is considered unresponsive
  65  * because it isn't providing Rx slots.
  66  */
  67 unsigned int rx_stall_timeout_msecs = 60000;
  68 module_param(rx_stall_timeout_msecs, uint, 0444);
  69
  70 unsigned int xenvif_max_queues;
  71 module_param_named(max_queues, xenvif_max_queues, uint, 0644);
  72 MODULE_PARM_DESC(max_queues,
  73                  "Maximum number of queues per virtual interface");
  74
  75 /*
  76  * This is the maximum slots a skb can have. If a guest sends a skb
  77  * which exceeds this limit it is considered malicious.
  78  */
  79 #define FATAL_SKB_SLOTS_DEFAULT 20
  80 static unsigned int fatal_skb_slots = FATAL_SKB_SLOTS_DEFAULT;
  81 module_param(fatal_skb_slots, uint, 0444);
  82
  83 /* The amount to copy out of the first guest Tx slot into the skb's
  84  * linear area.  If the first slot has more data, it will be mapped
  85  * and put into the first frag.
  86  *
  87  * This is sized to avoid pulling headers from the frags for most
  88  * TCP/IP packets.
  89  */
  90 #define XEN_NETBACK_TX_COPY_LEN 128
  91
  92
  93 static void xenvif_idx_release(struct xenvif_queue *queue, u16 pending_idx,
  94                                u8 status);
  95
  96 static void make_tx_response(struct xenvif_queue *queue,
  97                              struct xen_netif_tx_request *txp,
  98                              s8       st);
  99 static void push_tx_responses(struct xenvif_queue *queue);
 100
 101 static inline int tx_work_todo(struct xenvif_queue *queue);
 102
 103 static struct xen_netif_rx_response *make_rx_response(struct xenvif_queue *queue,
 104                                              u16      id,
 105                                              s8       st,
 106                                              u16      offset,
 107                                              u16      size,
 108                                              u16      flags);
 109
 110 static inline unsigned long idx_to_pfn(struct xenvif_queue *queue,
 111                                        u16 idx)
 112 {
 113         return page_to_pfn(queue->mmap_pages[idx]);
 114 }
 115
 116 static inline unsigned long idx_to_kaddr(struct xenvif_queue *queue,
 117                                          u16 idx)
 118 {
 119         return (unsigned long)pfn_to_kaddr(idx_to_pfn(queue, idx));
 120 }
 121
 122 #define callback_param(vif, pending_idx) \
 123         (vif->pending_tx_info[pending_idx].callback_struct)
 124
 125 /* Find the containing VIF's structure from a pointer in pending_tx_info array
 126  */
 127 static inline struct xenvif_queue *ubuf_to_queue(const struct ubuf_info *ubuf)
 128 {
 129         u16 pending_idx = ubuf->desc;
 130         struct pending_tx_info *temp =
 131                 container_of(ubuf, struct pending_tx_info, callback_struct);
 132         return container_of(temp - pending_idx,
 133                             struct xenvif_queue,
 134                             pending_tx_info[0]);
 135 }
 136
 137 static u16 frag_get_pending_idx(skb_frag_t *frag)
 138 {
 139         return (u16)frag->page_offset;
 140 }
 141
 142 static void frag_set_pending_idx(skb_frag_t *frag, u16 pending_idx)
 143 {
 144         frag->page_offset = pending_idx;
 145 }
 146
 147 static inline pending_ring_idx_t pending_index(unsigned i)
 148 {
 149         return i & (MAX_PENDING_REQS-1);
 150 }
 151
 152 static int xenvif_rx_ring_slots_needed(struct xenvif *vif)
 153 {
 154         if (vif->gso_mask)
 155                 return DIV_ROUND_UP(vif->dev->gso_max_size, XEN_PAGE_SIZE) + 1;
 156         else
 157                 return DIV_ROUND_UP(vif->dev->mtu, XEN_PAGE_SIZE);
 158 }
 159
 160 static bool xenvif_rx_ring_slots_available(struct xenvif_queue *queue)
 161 {
 162         RING_IDX prod, cons;
 163         int needed;
 164
 165         needed = xenvif_rx_ring_slots_needed(queue->vif);
 166
 167         do {
 168                 prod = queue->rx.sring->req_prod;
 169                 cons = queue->rx.req_cons;
 170
 171                 if (prod - cons >= needed)
 172                         return true;
 173
 174                 queue->rx.sring->req_event = prod + 1;
 175
 176                 /* Make sure event is visible before we check prod
 177                  * again.
 178                  */
 179                 mb();
 180         } while (queue->rx.sring->req_prod != prod);
 181
 182         return false;
 183 }
 184
 185 void xenvif_rx_queue_tail(struct xenvif_queue *queue, struct sk_buff *skb)
 186 {
 187         unsigned long flags;
 188
 189         spin_lock_irqsave(&queue->rx_queue.lock, flags);
 190
 191         __skb_queue_tail(&queue->rx_queue, skb);
 192
 193         queue->rx_queue_len += skb->len;
 194         if (queue->rx_queue_len > queue->rx_queue_max)
 195                 netif_tx_stop_queue(netdev_get_tx_queue(queue->vif->dev, queue->id));
 196
 197         spin_unlock_irqrestore(&queue->rx_queue.lock, flags);
 198 }
 199
 200 static struct sk_buff *xenvif_rx_dequeue(struct xenvif_queue *queue)
 201 {
 202         struct sk_buff *skb;
 203
 204         spin_lock_irq(&queue->rx_queue.lock);
 205
 206         skb = __skb_dequeue(&queue->rx_queue);
 207         if (skb)
 208                 queue->rx_queue_len -= skb->len;
 209
 210         spin_unlock_irq(&queue->rx_queue.lock);
 211
 212         return skb;
 213 }
 214
 215 static void xenvif_rx_queue_maybe_wake(struct xenvif_queue *queue)
 216 {
 217         spin_lock_irq(&queue->rx_queue.lock);
 218
 219         if (queue->rx_queue_len < queue->rx_queue_max)
 220                 netif_tx_wake_queue(netdev_get_tx_queue(queue->vif->dev, queue->id));
 221
 222         spin_unlock_irq(&queue->rx_queue.lock);
 223 }
 224
 225
 226 static void xenvif_rx_queue_purge(struct xenvif_queue *queue)
 227 {
 228         struct sk_buff *skb;
 229         while ((skb = xenvif_rx_dequeue(queue)) != NULL)
 230                 kfree_skb(skb);
 231 }
 232
 233 static void xenvif_rx_queue_drop_expired(struct xenvif_queue *queue)
 234 {
 235         struct sk_buff *skb;
 236
 237         for(;;) {
 238                 skb = skb_peek(&queue->rx_queue);
 239                 if (!skb)
 240                         break;
 241                 if (time_before(jiffies, XENVIF_RX_CB(skb)->expires))
 242                         break;
 243                 xenvif_rx_dequeue(queue);
 244                 kfree_skb(skb);
 245         }
 246 }
 247
 248 struct netrx_pending_operations {
 249         unsigned copy_prod, copy_cons;
 250         unsigned meta_prod, meta_cons;
 251         struct gnttab_copy *copy;
 252         struct xenvif_rx_meta *meta;
 253         int copy_off;
 254         grant_ref_t copy_gref;
 255 };
 256
 257 static struct xenvif_rx_meta *get_next_rx_buffer(struct xenvif_queue *queue,
 258                                                  struct netrx_pending_operations *npo)
 259 {
 260         struct xenvif_rx_meta *meta;
 261         struct xen_netif_rx_request *req;
 262
 263         req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
 264
 265         meta = npo->meta + npo->meta_prod++;
 266         meta->gso_type = XEN_NETIF_GSO_TYPE_NONE;
 267         meta->gso_size = 0;
 268         meta->size = 0;
 269         meta->id = req->id;
 270
 271         npo->copy_off = 0;
 272         npo->copy_gref = req->gref;
 273
 274         return meta;
 275 }
 276
 277 struct gop_frag_copy {
 278         struct xenvif_queue *queue;
 279         struct netrx_pending_operations *npo;
 280         struct xenvif_rx_meta *meta;
 281         int head;
 282         int gso_type;
 283
 284         struct page *page;
 285 };
 286
 287 static void xenvif_setup_copy_gop(unsigned long gfn,
 288                                   unsigned int offset,
 289                                   unsigned int *len,
 290                                   struct gop_frag_copy *info)
 291 {
 292         struct gnttab_copy *copy_gop;
 293         struct xen_page_foreign *foreign;
 294         /* Convenient aliases */
 295         struct xenvif_queue *queue = info->queue;
 296         struct netrx_pending_operations *npo = info->npo;
 297         struct page *page = info->page;
 298
 299         BUG_ON(npo->copy_off > MAX_BUFFER_OFFSET);
 300
 301         if (npo->copy_off == MAX_BUFFER_OFFSET)
 302                 info->meta = get_next_rx_buffer(queue, npo);
 303
 304         if (npo->copy_off + *len > MAX_BUFFER_OFFSET)
 305                 *len = MAX_BUFFER_OFFSET - npo->copy_off;
 306
 307         copy_gop = npo->copy + npo->copy_prod++;
 308         copy_gop->flags = GNTCOPY_dest_gref;
 309         copy_gop->len = *len;
 310
 311         foreign = xen_page_foreign(page);
 312         if (foreign) {
 313                 copy_gop->source.domid = foreign->domid;
 314                 copy_gop->source.u.ref = foreign->gref;
 315                 copy_gop->flags |= GNTCOPY_source_gref;
 316         } else {
 317                 copy_gop->source.domid = DOMID_SELF;
 318                 copy_gop->source.u.gmfn = gfn;
 319         }
 320         copy_gop->source.offset = offset;
 321
 322         copy_gop->dest.domid = queue->vif->domid;
 323         copy_gop->dest.offset = npo->copy_off;
 324         copy_gop->dest.u.ref = npo->copy_gref;
 325
 326         npo->copy_off += *len;
 327         info->meta->size += *len;
 328
 329         /* Leave a gap for the GSO descriptor. */
 330         if (info->head && ((1 << info->gso_type) & queue->vif->gso_mask))
 331                 queue->rx.req_cons++;
 332
 333         info->head = 0; /* There must be something in this buffer now */
 334 }
 335
 336 static void xenvif_gop_frag_copy_grant(unsigned long gfn,
 337                                        unsigned offset,
 338                                        unsigned int len,
 339                                        void *data)
 340 {
 341         unsigned int bytes;
 342
 343         while (len) {
 344                 bytes = len;
 345                 xenvif_setup_copy_gop(gfn, offset, &bytes, data);
 346                 offset += bytes;
 347                 len -= bytes;
 348         }
 349 }
 350
 351 /*
 352  * Set up the grant operations for this fragment. If it's a flipping
 353  * interface, we also set up the unmap request from here.
 354  */
 355 static void xenvif_gop_frag_copy(struct xenvif_queue *queue, struct sk_buff *skb,
 356                                  struct netrx_pending_operations *npo,
 357                                  struct page *page, unsigned long size,
 358                                  unsigned long offset, int *head)
 359 {
 360         struct gop_frag_copy info = {
 361                 .queue = queue,
 362                 .npo = npo,
 363                 .head = *head,
 364                 .gso_type = XEN_NETIF_GSO_TYPE_NONE,
 365         };
 366         unsigned long bytes;
 367
 368         if (skb_is_gso(skb)) {
 369                 if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
 370                         info.gso_type = XEN_NETIF_GSO_TYPE_TCPV4;
 371                 else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
 372                         info.gso_type = XEN_NETIF_GSO_TYPE_TCPV6;
 373         }
 374
 375         /* Data must not cross a page boundary. */
 376         BUG_ON(size + offset > PAGE_SIZE<<compound_order(page));
 377
 378         info.meta = npo->meta + npo->meta_prod - 1;
 379
 380         /* Skip unused frames from start of page */
 381         page += offset >> PAGE_SHIFT;
 382         offset &= ~PAGE_MASK;
 383
 384         while (size > 0) {
 385                 BUG_ON(offset >= PAGE_SIZE);
 386
 387                 bytes = PAGE_SIZE - offset;
 388                 if (bytes > size)
 389                         bytes = size;
 390
 391                 info.page = page;
 392                 gnttab_foreach_grant_in_range(page, offset, bytes,
 393                                               xenvif_gop_frag_copy_grant,
 394                                               &info);
 395                 size -= bytes;
 396                 offset = 0;
 397
 398                 /* Next page */
 399                 if (size) {
 400                         BUG_ON(!PageCompound(page));
 401                         page++;
 402                 }
 403         }
 404
 405         *head = info.head;
 406 }
 407
 408 /*
 409  * Prepare an SKB to be transmitted to the frontend.
 410  *
 411  * This function is responsible for allocating grant operations, meta
 412  * structures, etc.
 413  *
 414  * It returns the number of meta structures consumed. The number of
 415  * ring slots used is always equal to the number of meta slots used
 416  * plus the number of GSO descriptors used. Currently, we use either
 417  * zero GSO descriptors (for non-GSO packets) or one descriptor (for
 418  * frontend-side LRO).
 419  */
 420 static int xenvif_gop_skb(struct sk_buff *skb,
 421                           struct netrx_pending_operations *npo,
 422                           struct xenvif_queue *queue)
 423 {
 424         struct xenvif *vif = netdev_priv(skb->dev);
 425         int nr_frags = skb_shinfo(skb)->nr_frags;
 426         int i;
 427         struct xen_netif_rx_request *req;
 428         struct xenvif_rx_meta *meta;
 429         unsigned char *data;
 430         int head = 1;
 431         int old_meta_prod;
 432         int gso_type;
 433
 434         old_meta_prod = npo->meta_prod;
 435
 436         gso_type = XEN_NETIF_GSO_TYPE_NONE;
 437         if (skb_is_gso(skb)) {
 438                 if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
 439                         gso_type = XEN_NETIF_GSO_TYPE_TCPV4;
 440                 else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
 441                         gso_type = XEN_NETIF_GSO_TYPE_TCPV6;
 442         }
 443
 444         /* Set up a GSO prefix descriptor, if necessary */
 445         if ((1 << gso_type) & vif->gso_prefix_mask) {
 446                 req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
 447                 meta = npo->meta + npo->meta_prod++;
 448                 meta->gso_type = gso_type;
 449                 meta->gso_size = skb_shinfo(skb)->gso_size;
 450                 meta->size = 0;
 451                 meta->id = req->id;
 452         }
 453
 454         req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
 455         meta = npo->meta + npo->meta_prod++;
 456
 457         if ((1 << gso_type) & vif->gso_mask) {
 458                 meta->gso_type = gso_type;
 459                 meta->gso_size = skb_shinfo(skb)->gso_size;
 460         } else {
 461                 meta->gso_type = XEN_NETIF_GSO_TYPE_NONE;
 462                 meta->gso_size = 0;
 463         }
 464
 465         meta->size = 0;
 466         meta->id = req->id;
 467         npo->copy_off = 0;
 468         npo->copy_gref = req->gref;
 469
 470         data = skb->data;
 471         while (data < skb_tail_pointer(skb)) {
 472                 unsigned int offset = offset_in_page(data);
 473                 unsigned int len = PAGE_SIZE - offset;
 474
 475                 if (data + len > skb_tail_pointer(skb))
 476                         len = skb_tail_pointer(skb) - data;
 477
 478                 xenvif_gop_frag_copy(queue, skb, npo,
 479                                      virt_to_page(data), len, offset, &head);
 480                 data += len;
 481         }
 482
 483         for (i = 0; i < nr_frags; i++) {
 484                 xenvif_gop_frag_copy(queue, skb, npo,
 485                                      skb_frag_page(&skb_shinfo(skb)->frags[i]),
 486                                      skb_frag_size(&skb_shinfo(skb)->frags[i]),
 487                                      skb_shinfo(skb)->frags[i].page_offset,
 488                                      &head);
 489         }
 490
 491         return npo->meta_prod - old_meta_prod;
 492 }
 493
 494 /*
 495  * This is a twin to xenvif_gop_skb.  Assume that xenvif_gop_skb was
 496  * used to set up the operations on the top of
 497  * netrx_pending_operations, which have since been done.  Check that
 498  * they didn't give any errors and advance over them.
 499  */
 500 static int xenvif_check_gop(struct xenvif *vif, int nr_meta_slots,
 501                             struct netrx_pending_operations *npo)
 502 {
 503         struct gnttab_copy     *copy_op;
 504         int status = XEN_NETIF_RSP_OKAY;
 505         int i;
 506
 507         for (i = 0; i < nr_meta_slots; i++) {
 508                 copy_op = npo->copy + npo->copy_cons++;
 509                 if (copy_op->status != GNTST_okay) {
 510                         netdev_dbg(vif->dev,
 511                                    "Bad status %d from copy to DOM%d.\n",
 512                                    copy_op->status, vif->domid);
 513                         status = XEN_NETIF_RSP_ERROR;
 514                 }
 515         }
 516
 517         return status;
 518 }
 519
 520 static void xenvif_add_frag_responses(struct xenvif_queue *queue, int status,
 521                                       struct xenvif_rx_meta *meta,
 522                                       int nr_meta_slots)
 523 {
 524         int i;
 525         unsigned long offset;
 526
 527         /* No fragments used */
 528         if (nr_meta_slots <= 1)
 529                 return;
 530
 531         nr_meta_slots--;
 532
 533         for (i = 0; i < nr_meta_slots; i++) {
 534                 int flags;
 535                 if (i == nr_meta_slots - 1)
 536                         flags = 0;
 537                 else
 538                         flags = XEN_NETRXF_more_data;
 539
 540                 offset = 0;
 541                 make_rx_response(queue, meta[i].id, status, offset,
 542                                  meta[i].size, flags);
 543         }
 544 }
 545
 546 void xenvif_kick_thread(struct xenvif_queue *queue)
 547 {
 548         wake_up(&queue->wq);
 549 }
 550
 551 static void xenvif_rx_action(struct xenvif_queue *queue)
 552 {
 553         s8 status;
 554         u16 flags;
 555         struct xen_netif_rx_response *resp;
 556         struct sk_buff_head rxq;
 557         struct sk_buff *skb;
 558         LIST_HEAD(notify);
 559         int ret;
 560         unsigned long offset;
 561         bool need_to_notify = false;
 562
 563         struct netrx_pending_operations npo = {
 564                 .copy  = queue->grant_copy_op,
 565                 .meta  = queue->meta,
 566         };
 567
 568         skb_queue_head_init(&rxq);
 569
 570         while (xenvif_rx_ring_slots_available(queue)
 571                && (skb = xenvif_rx_dequeue(queue)) != NULL) {
 572                 queue->last_rx_time = jiffies;
 573
 574                 XENVIF_RX_CB(skb)->meta_slots_used = xenvif_gop_skb(skb, &npo, queue);
 575
 576                 __skb_queue_tail(&rxq, skb);
 577         }
 578
 579         BUG_ON(npo.meta_prod > ARRAY_SIZE(queue->meta));
 580
 581         if (!npo.copy_prod)
 582                 goto done;
 583
 584         BUG_ON(npo.copy_prod > MAX_GRANT_COPY_OPS);
 585         gnttab_batch_copy(queue->grant_copy_op, npo.copy_prod);
 586
 587         while ((skb = __skb_dequeue(&rxq)) != NULL) {
 588
 589                 if ((1 << queue->meta[npo.meta_cons].gso_type) &
 590                     queue->vif->gso_prefix_mask) {
 591                         resp = RING_GET_RESPONSE(&queue->rx,
 592                                                  queue->rx.rsp_prod_pvt++);
 593
 594                         resp->flags = XEN_NETRXF_gso_prefix | XEN_NETRXF_more_data;
 595
 596                         resp->offset = queue->meta[npo.meta_cons].gso_size;
 597                         resp->id = queue->meta[npo.meta_cons].id;
 598                         resp->status = XENVIF_RX_CB(skb)->meta_slots_used;
 599
 600                         npo.meta_cons++;
 601                         XENVIF_RX_CB(skb)->meta_slots_used--;
 602                 }
 603
 604
 605                 queue->stats.tx_bytes += skb->len;
 606                 queue->stats.tx_packets++;
 607
 608                 status = xenvif_check_gop(queue->vif,
 609                                           XENVIF_RX_CB(skb)->meta_slots_used,
 610                                           &npo);
 611
 612                 if (XENVIF_RX_CB(skb)->meta_slots_used == 1)
 613                         flags = 0;
 614                 else
 615                         flags = XEN_NETRXF_more_data;
 616
 617                 if (skb->ip_summed == CHECKSUM_PARTIAL) /* local packet? */
 618                         flags |= XEN_NETRXF_csum_blank | XEN_NETRXF_data_validated;
 619                 else if (skb->ip_summed == CHECKSUM_UNNECESSARY)
 620                         /* remote but checksummed. */
 621                         flags |= XEN_NETRXF_data_validated;
 622
 623                 offset = 0;
 624                 resp = make_rx_response(queue, queue->meta[npo.meta_cons].id,
 625                                         status, offset,
 626                                         queue->meta[npo.meta_cons].size,
 627                                         flags);
 628
 629                 if ((1 << queue->meta[npo.meta_cons].gso_type) &
 630                     queue->vif->gso_mask) {
 631                         struct xen_netif_extra_info *gso =
 632                                 (struct xen_netif_extra_info *)
 633                                 RING_GET_RESPONSE(&queue->rx,
 634                                                   queue->rx.rsp_prod_pvt++);
 635
 636                         resp->flags |= XEN_NETRXF_extra_info;
 637
 638                         gso->u.gso.type = queue->meta[npo.meta_cons].gso_type;
 639                         gso->u.gso.size = queue->meta[npo.meta_cons].gso_size;
 640                         gso->u.gso.pad = 0;
 641                         gso->u.gso.features = 0;
 642
 643                         gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
 644                         gso->flags = 0;
 645                 }
 646
 647                 xenvif_add_frag_responses(queue, status,
 648                                           queue->meta + npo.meta_cons + 1,
 649                                           XENVIF_RX_CB(skb)->meta_slots_used);
 650
 651                 RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&queue->rx, ret);
 652
 653                 need_to_notify |= !!ret;
 654
 655                 npo.meta_cons += XENVIF_RX_CB(skb)->meta_slots_used;
 656                 dev_kfree_skb(skb);
 657         }
 658
 659 done:
 660         if (need_to_notify)
 661                 notify_remote_via_irq(queue->rx_irq);
 662 }
 663
 664 void xenvif_napi_schedule_or_enable_events(struct xenvif_queue *queue)
 665 {
 666         int more_to_do;
 667
 668         RING_FINAL_CHECK_FOR_REQUESTS(&queue->tx, more_to_do);
 669
 670         if (more_to_do)
 671                 napi_schedule(&queue->napi);
 672 }
 673
 674 static void tx_add_credit(struct xenvif_queue *queue)
 675 {
 676         unsigned long max_burst, max_credit;
 677
 678         /*
 679          * Allow a burst big enough to transmit a jumbo packet of up to 128kB.
 680          * Otherwise the interface can seize up due to insufficient credit.
 681          */
 682         max_burst = RING_GET_REQUEST(&queue->tx, queue->tx.req_cons)->size;
 683         max_burst = min(max_burst, 131072UL);
 684         max_burst = max(max_burst, queue->credit_bytes);
 685
 686         /* Take care that adding a new chunk of credit doesn't wrap to zero. */
 687         max_credit = queue->remaining_credit + queue->credit_bytes;
 688         if (max_credit < queue->remaining_credit)
 689                 max_credit = ULONG_MAX; /* wrapped: clamp to ULONG_MAX */
 690
 691         queue->remaining_credit = min(max_credit, max_burst);
 692 }
 693
 694 void xenvif_tx_credit_callback(unsigned long data)
 695 {
 696         struct xenvif_queue *queue = (struct xenvif_queue *)data;
 697         tx_add_credit(queue);
 698         xenvif_napi_schedule_or_enable_events(queue);
 699 }
 700
 701 static void xenvif_tx_err(struct xenvif_queue *queue,
 702                           struct xen_netif_tx_request *txp, RING_IDX end)
 703 {
 704         RING_IDX cons = queue->tx.req_cons;
 705         unsigned long flags;
 706
 707         do {
 708                 spin_lock_irqsave(&queue->response_lock, flags);
 709                 make_tx_response(queue, txp, XEN_NETIF_RSP_ERROR);
 710                 push_tx_responses(queue);
 711                 spin_unlock_irqrestore(&queue->response_lock, flags);
 712                 if (cons == end)
 713                         break;
 714                 txp = RING_GET_REQUEST(&queue->tx, cons++);
 715         } while (1);
 716         queue->tx.req_cons = cons;
 717 }
 718
 719 static void xenvif_fatal_tx_err(struct xenvif *vif)
 720 {
 721         netdev_err(vif->dev, "fatal error; disabling device\n");
 722         vif->disabled = true;
 723         /* Disable the vif from queue 0's kthread */
 724         if (vif->queues)
 725                 xenvif_kick_thread(&vif->queues[0]);
 726 }
 727
 728 static int xenvif_count_requests(struct xenvif_queue *queue,
 729                                  struct xen_netif_tx_request *first,
 730                                  struct xen_netif_tx_request *txp,
 731                                  int work_to_do)
 732 {
 733         RING_IDX cons = queue->tx.req_cons;
 734         int slots = 0;
 735         int drop_err = 0;
 736         int more_data;
 737
 738         if (!(first->flags & XEN_NETTXF_more_data))
 739                 return 0;
 740
 741         do {
 742                 struct xen_netif_tx_request dropped_tx = { 0 };
 743
 744                 if (slots >= work_to_do) {
 745                         netdev_err(queue->vif->dev,
 746                                    "Asked for %d slots but exceeds this limit\n",
 747                                    work_to_do);
 748                         xenvif_fatal_tx_err(queue->vif);
 749                         return -ENODATA;
 750                 }
 751
 752                 /* This guest is really using too many slots and
 753                  * considered malicious.
 754                  */
 755                 if (unlikely(slots >= fatal_skb_slots)) {
 756                         netdev_err(queue->vif->dev,
 757                                    "Malicious frontend using %d slots, threshold %u\n",
 758                                    slots, fatal_skb_slots);
 759                         xenvif_fatal_tx_err(queue->vif);
 760                         return -E2BIG;
 761                 }
 762
 763                 /* Xen network protocol had implicit dependency on
 764                  * MAX_SKB_FRAGS. XEN_NETBK_LEGACY_SLOTS_MAX is set to
 765                  * the historical MAX_SKB_FRAGS value 18 to honor the
 766                  * same behavior as before. Any packet using more than
 767                  * 18 slots but less than fatal_skb_slots slots is
 768                  * dropped
 769                  */
 770                 if (!drop_err && slots >= XEN_NETBK_LEGACY_SLOTS_MAX) {
 771                         if (net_ratelimit())
 772                                 netdev_dbg(queue->vif->dev,
 773                                            "Too many slots (%d) exceeding limit (%d), dropping packet\n",
 774                                            slots, XEN_NETBK_LEGACY_SLOTS_MAX);
 775                         drop_err = -E2BIG;
 776                 }
 777
 778                 if (drop_err)
 779                         txp = &dropped_tx;
 780
 781                 memcpy(txp, RING_GET_REQUEST(&queue->tx, cons + slots),
 782                        sizeof(*txp));
 783
 784                 /* If the guest submitted a frame >= 64 KiB then
 785                  * first->size overflowed and following slots will
 786                  * appear to be larger than the frame.
 787                  *
 788                  * This cannot be fatal error as there are buggy
 789                  * frontends that do this.
 790                  *
 791                  * Consume all slots and drop the packet.
 792                  */
 793                 if (!drop_err && txp->size > first->size) {
 794                         if (net_ratelimit())
 795                                 netdev_dbg(queue->vif->dev,
 796                                            "Invalid tx request, slot size %u > remaining size %u\n",
 797                                            txp->size, first->size);
 798                         drop_err = -EIO;
 799                 }
 800
 801                 first->size -= txp->size;
 802                 slots++;
 803
 804                 if (unlikely((txp->offset + txp->size) > XEN_PAGE_SIZE)) {
 805                         netdev_err(queue->vif->dev, "Cross page boundary, txp->offset: %u, size: %u\n",
 806                                  txp->offset, txp->size);
 807                         xenvif_fatal_tx_err(queue->vif);
 808                         return -EINVAL;
 809                 }
 810
 811                 more_data = txp->flags & XEN_NETTXF_more_data;
 812
 813                 if (!drop_err)
 814                         txp++;
 815
 816         } while (more_data);
 817
 818         if (drop_err) {
 819                 xenvif_tx_err(queue, first, cons + slots);
 820                 return drop_err;
 821         }
 822
 823         return slots;
 824 }
 825
 826
 827 struct xenvif_tx_cb {
 828         u16 pending_idx;
 829 };
 830
 831 #define XENVIF_TX_CB(skb) ((struct xenvif_tx_cb *)(skb)->cb)
 832
 833 static inline void xenvif_tx_create_map_op(struct xenvif_queue *queue,
 834                                           u16 pending_idx,
 835                                           struct xen_netif_tx_request *txp,
 836                                           struct gnttab_map_grant_ref *mop)
 837 {
 838         queue->pages_to_map[mop-queue->tx_map_ops] = queue->mmap_pages[pending_idx];
 839         gnttab_set_map_op(mop, idx_to_kaddr(queue, pending_idx),
 840                           GNTMAP_host_map | GNTMAP_readonly,
 841                           txp->gref, queue->vif->domid);
 842
 843         memcpy(&queue->pending_tx_info[pending_idx].req, txp,
 844                sizeof(*txp));
 845 }
 846
 847 static inline struct sk_buff *xenvif_alloc_skb(unsigned int size)
 848 {
 849         struct sk_buff *skb =
 850                 alloc_skb(size + NET_SKB_PAD + NET_IP_ALIGN,
 851                           GFP_ATOMIC | __GFP_NOWARN);
 852         if (unlikely(skb == NULL))
 853                 return NULL;
 854
 855         /* Packets passed to netif_rx() must have some headroom. */
 856         skb_reserve(skb, NET_SKB_PAD + NET_IP_ALIGN);
 857
 858         /* Initialize it here to avoid later surprises */
 859         skb_shinfo(skb)->destructor_arg = NULL;
 860
 861         return skb;
 862 }
 863
 864 static struct gnttab_map_grant_ref *xenvif_get_requests(struct xenvif_queue *queue,
 865                                                         struct sk_buff *skb,
 866                                                         struct xen_netif_tx_request *txp,
 867                                                         struct gnttab_map_grant_ref *gop,
 868                                                         unsigned int frag_overflow,
 869                                                         struct sk_buff *nskb)
 870 {
 871         struct skb_shared_info *shinfo = skb_shinfo(skb);
 872         skb_frag_t *frags = shinfo->frags;
 873         u16 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
 874         int start;
 875         pending_ring_idx_t index;
 876         unsigned int nr_slots;
 877
 878         nr_slots = shinfo->nr_frags;
 879
 880         /* Skip first skb fragment if it is on same page as header fragment. */
 881         start = (frag_get_pending_idx(&shinfo->frags[0]) == pending_idx);
 882
 883         for (shinfo->nr_frags = start; shinfo->nr_frags < nr_slots;
 884              shinfo->nr_frags++, txp++, gop++) {
 885                 index = pending_index(queue->pending_cons++);
 886                 pending_idx = queue->pending_ring[index];
 887                 xenvif_tx_create_map_op(queue, pending_idx, txp, gop);
 888                 frag_set_pending_idx(&frags[shinfo->nr_frags], pending_idx);
 889         }
 890
 891         if (frag_overflow) {
 892
 893                 shinfo = skb_shinfo(nskb);
 894                 frags = shinfo->frags;
 895
 896                 for (shinfo->nr_frags = 0; shinfo->nr_frags < frag_overflow;
 897                      shinfo->nr_frags++, txp++, gop++) {
 898                         index = pending_index(queue->pending_cons++);
 899                         pending_idx = queue->pending_ring[index];
 900                         xenvif_tx_create_map_op(queue, pending_idx, txp, gop);
 901                         frag_set_pending_idx(&frags[shinfo->nr_frags],
 902                                              pending_idx);
 903                 }
 904
 905                 skb_shinfo(skb)->frag_list = nskb;
 906         }
 907
 908         return gop;
 909 }
 910
 911 static inline void xenvif_grant_handle_set(struct xenvif_queue *queue,
 912                                            u16 pending_idx,
 913                                            grant_handle_t handle)
 914 {
 915         if (unlikely(queue->grant_tx_handle[pending_idx] !=
 916                      NETBACK_INVALID_HANDLE)) {
 917                 netdev_err(queue->vif->dev,
 918                            "Trying to overwrite active handle! pending_idx: 0x%x\n",
 919                            pending_idx);
 920                 BUG();
 921         }
 922         queue->grant_tx_handle[pending_idx] = handle;
 923 }
 924
 925 static inline void xenvif_grant_handle_reset(struct xenvif_queue *queue,
 926                                              u16 pending_idx)
 927 {
 928         if (unlikely(queue->grant_tx_handle[pending_idx] ==
 929                      NETBACK_INVALID_HANDLE)) {
 930                 netdev_err(queue->vif->dev,
 931                            "Trying to unmap invalid handle! pending_idx: 0x%x\n",
 932                            pending_idx);
 933                 BUG();
 934         }
 935         queue->grant_tx_handle[pending_idx] = NETBACK_INVALID_HANDLE;
 936 }
 937
 938 static int xenvif_tx_check_gop(struct xenvif_queue *queue,
 939                                struct sk_buff *skb,
 940                                struct gnttab_map_grant_ref **gopp_map,
 941                                struct gnttab_copy **gopp_copy)
 942 {
 943         struct gnttab_map_grant_ref *gop_map = *gopp_map;
 944         u16 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
 945         /* This always points to the shinfo of the skb being checked, which
 946          * could be either the first or the one on the frag_list
 947          */
 948         struct skb_shared_info *shinfo = skb_shinfo(skb);
 949         /* If this is non-NULL, we are currently checking the frag_list skb, and
 950          * this points to the shinfo of the first one
 951          */
 952         struct skb_shared_info *first_shinfo = NULL;
 953         int nr_frags = shinfo->nr_frags;
 954         const bool sharedslot = nr_frags &&
 955                                 frag_get_pending_idx(&shinfo->frags[0]) == pending_idx;
 956         int i, err;
 957
 958         /* Check status of header. */
 959         err = (*gopp_copy)->status;
 960         if (unlikely(err)) {
 961                 if (net_ratelimit())
 962                         netdev_dbg(queue->vif->dev,
 963                                    "Grant copy of header failed! status: %d pending_idx: %u ref: %u\n",
 964                                    (*gopp_copy)->status,
 965                                    pending_idx,
 966                                    (*gopp_copy)->source.u.ref);
 967                 /* The first frag might still have this slot mapped */
 968                 if (!sharedslot)
 969                         xenvif_idx_release(queue, pending_idx,
 970                                            XEN_NETIF_RSP_ERROR);
 971         }
 972         (*gopp_copy)++;
 973
 974 check_frags:
 975         for (i = 0; i < nr_frags; i++, gop_map++) {
 976                 int j, newerr;
 977
 978                 pending_idx = frag_get_pending_idx(&shinfo->frags[i]);
 979
 980                 /* Check error status: if okay then remember grant handle. */
 981                 newerr = gop_map->status;
 982
 983                 if (likely(!newerr)) {
 984                         xenvif_grant_handle_set(queue,
 985                                                 pending_idx,
 986                                                 gop_map->handle);
 987                         /* Had a previous error? Invalidate this fragment. */
 988                         if (unlikely(err)) {
 989                                 xenvif_idx_unmap(queue, pending_idx);
 990                                 /* If the mapping of the first frag was OK, but
 991                                  * the header's copy failed, and they are
 992                                  * sharing a slot, send an error
 993                                  */
 994                                 if (i == 0 && sharedslot)
 995                                         xenvif_idx_release(queue, pending_idx,
 996                                                            XEN_NETIF_RSP_ERROR);
 997                                 else
 998                                         xenvif_idx_release(queue, pending_idx,
 999                                                            XEN_NETIF_RSP_OKAY);
1000                         }
1001                         continue;
1002                 }
1003
1004                 /* Error on this fragment: respond to client with an error. */
1005                 if (net_ratelimit())
1006                         netdev_dbg(queue->vif->dev,
1007                                    "Grant map of %d. frag failed! status: %d pending_idx: %u ref: %u\n",
1008                                    i,
1009                                    gop_map->status,
1010                                    pending_idx,
1011                                    gop_map->ref);
1012
1013                 xenvif_idx_release(queue, pending_idx, XEN_NETIF_RSP_ERROR);
1014
1015                 /* Not the first error? Preceding frags already invalidated. */
1016                 if (err)
1017                         continue;
1018
1019                 /* First error: if the header haven't shared a slot with the
1020                  * first frag, release it as well.
1021                  */
1022                 if (!sharedslot)
1023                         xenvif_idx_release(queue,
1024                                            XENVIF_TX_CB(skb)->pending_idx,
1025                                            XEN_NETIF_RSP_OKAY);
1026
1027                 /* Invalidate preceding fragments of this skb. */
1028                 for (j = 0; j < i; j++) {
1029                         pending_idx = frag_get_pending_idx(&shinfo->frags[j]);
1030                         xenvif_idx_unmap(queue, pending_idx);
1031                         xenvif_idx_release(queue, pending_idx,
1032                                            XEN_NETIF_RSP_OKAY);
1033                 }
1034
1035                 /* And if we found the error while checking the frag_list, unmap
1036                  * the first skb's frags
1037                  */
1038                 if (first_shinfo) {
1039                         for (j = 0; j < first_shinfo->nr_frags; j++) {
1040                                 pending_idx = frag_get_pending_idx(&first_shinfo->frags[j]);
1041                                 xenvif_idx_unmap(queue, pending_idx);
1042                                 xenvif_idx_release(queue, pending_idx,
1043                                                    XEN_NETIF_RSP_OKAY);
1044                         }
1045                 }
1046
1047                 /* Remember the error: invalidate all subsequent fragments. */
1048                 err = newerr;
1049         }
1050
1051         if (skb_has_frag_list(skb) && !first_shinfo) {
1052                 first_shinfo = skb_shinfo(skb);
1053                 shinfo = skb_shinfo(skb_shinfo(skb)->frag_list);
1054                 nr_frags = shinfo->nr_frags;
1055
1056                 goto check_frags;
1057         }
1058
1059         *gopp_map = gop_map;
1060         return err;
1061 }
1062
1063 static void xenvif_fill_frags(struct xenvif_queue *queue, struct sk_buff *skb)
1064 {
1065         struct skb_shared_info *shinfo = skb_shinfo(skb);
1066         int nr_frags = shinfo->nr_frags;
1067         int i;
1068         u16 prev_pending_idx = INVALID_PENDING_IDX;
1069
1070         for (i = 0; i < nr_frags; i++) {
1071                 skb_frag_t *frag = shinfo->frags + i;
1072                 struct xen_netif_tx_request *txp;
1073                 struct page *page;
1074                 u16 pending_idx;
1075
1076                 pending_idx = frag_get_pending_idx(frag);
1077
1078                 /* If this is not the first frag, chain it to the previous*/
1079                 if (prev_pending_idx == INVALID_PENDING_IDX)
1080                         skb_shinfo(skb)->destructor_arg =
1081                                 &callback_param(queue, pending_idx);
1082                 else
1083                         callback_param(queue, prev_pending_idx).ctx =
1084                                 &callback_param(queue, pending_idx);
1085
1086                 callback_param(queue, pending_idx).ctx = NULL;
1087                 prev_pending_idx = pending_idx;
1088
1089                 txp = &queue->pending_tx_info[pending_idx].req;
1090                 page = virt_to_page(idx_to_kaddr(queue, pending_idx));
1091                 __skb_fill_page_desc(skb, i, page, txp->offset, txp->size);
1092                 skb->len += txp->size;
1093                 skb->data_len += txp->size;
1094                 skb->truesize += txp->size;
1095
1096                 /* Take an extra reference to offset network stack's put_page */
1097                 get_page(queue->mmap_pages[pending_idx]);
1098         }
1099 }
1100
1101 static int xenvif_get_extras(struct xenvif_queue *queue,
1102                                 struct xen_netif_extra_info *extras,
1103                                 int work_to_do)
1104 {
1105         struct xen_netif_extra_info extra;
1106         RING_IDX cons = queue->tx.req_cons;
1107
1108         do {
1109                 if (unlikely(work_to_do-- <= 0)) {
1110                         netdev_err(queue->vif->dev, "Missing extra info\n");
1111                         xenvif_fatal_tx_err(queue->vif);
1112                         return -EBADR;
1113                 }
1114
1115                 memcpy(&extra, RING_GET_REQUEST(&queue->tx, cons),
1116                        sizeof(extra));
1117                 if (unlikely(!extra.type ||
1118                              extra.type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
1119                         queue->tx.req_cons = ++cons;
1120                         netdev_err(queue->vif->dev,
1121                                    "Invalid extra type: %d\n", extra.type);
1122                         xenvif_fatal_tx_err(queue->vif);
1123                         return -EINVAL;
1124                 }
1125
1126                 memcpy(&extras[extra.type - 1], &extra, sizeof(extra));
1127                 queue->tx.req_cons = ++cons;
1128         } while (extra.flags & XEN_NETIF_EXTRA_FLAG_MORE);
1129
1130         return work_to_do;
1131 }
1132
1133 static int xenvif_set_skb_gso(struct xenvif *vif,
1134                               struct sk_buff *skb,
1135                               struct xen_netif_extra_info *gso)
1136 {
1137         if (!gso->u.gso.size) {
1138                 netdev_err(vif->dev, "GSO size must not be zero.\n");
1139                 xenvif_fatal_tx_err(vif);
1140                 return -EINVAL;
1141         }
1142
1143         switch (gso->u.gso.type) {
1144         case XEN_NETIF_GSO_TYPE_TCPV4:
1145                 skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
1146                 break;
1147         case XEN_NETIF_GSO_TYPE_TCPV6:
1148                 skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
1149                 break;
1150         default:
1151                 netdev_err(vif->dev, "Bad GSO type %d.\n", gso->u.gso.type);
1152                 xenvif_fatal_tx_err(vif);
1153                 return -EINVAL;
1154         }
1155
1156         skb_shinfo(skb)->gso_size = gso->u.gso.size;
1157         /* gso_segs will be calculated later */
1158
1159         return 0;
1160 }
1161
1162 static int checksum_setup(struct xenvif_queue *queue, struct sk_buff *skb)
1163 {
1164         bool recalculate_partial_csum = false;
1165
1166         /* A GSO SKB must be CHECKSUM_PARTIAL. However some buggy
1167          * peers can fail to set NETRXF_csum_blank when sending a GSO
1168          * frame. In this case force the SKB to CHECKSUM_PARTIAL and
1169          * recalculate the partial checksum.
1170          */
1171         if (skb->ip_summed != CHECKSUM_PARTIAL && skb_is_gso(skb)) {
1172                 queue->stats.rx_gso_checksum_fixup++;
1173                 skb->ip_summed = CHECKSUM_PARTIAL;
1174                 recalculate_partial_csum = true;
1175         }
1176
1177         /* A non-CHECKSUM_PARTIAL SKB does not require setup. */
1178         if (skb->ip_summed != CHECKSUM_PARTIAL)
1179                 return 0;
1180
1181         return skb_checksum_setup(skb, recalculate_partial_csum);
1182 }
1183
1184 static bool tx_credit_exceeded(struct xenvif_queue *queue, unsigned size)
1185 {
1186         u64 now = get_jiffies_64();
1187         u64 next_credit = queue->credit_window_start +
1188                 msecs_to_jiffies(queue->credit_usec / 1000);
1189
1190         /* Timer could already be pending in rare cases. */
1191         if (timer_pending(&queue->credit_timeout))
1192                 return true;
1193
1194         /* Passed the point where we can replenish credit? */
1195         if (time_after_eq64(now, next_credit)) {
1196                 queue->credit_window_start = now;
1197                 tx_add_credit(queue);
1198         }
1199
1200         /* Still too big to send right now? Set a callback. */
1201         if (size > queue->remaining_credit) {
1202                 queue->credit_timeout.data     =
1203                         (unsigned long)queue;
1204                 mod_timer(&queue->credit_timeout,
1205                           next_credit);
1206                 queue->credit_window_start = next_credit;
1207
1208                 return true;
1209         }
1210
1211         return false;
1212 }
1213
1214 /* No locking is required in xenvif_mcast_add/del() as they are
1215  * only ever invoked from NAPI poll. An RCU list is used because
1216  * xenvif_mcast_match() is called asynchronously, during start_xmit.
1217  */
1218
1219 static int xenvif_mcast_add(struct xenvif *vif, const u8 *addr)
1220 {
1221         struct xenvif_mcast_addr *mcast;
1222
1223         if (vif->fe_mcast_count == XEN_NETBK_MCAST_MAX) {
1224                 if (net_ratelimit())
1225                         netdev_err(vif->dev,
1226                                    "Too many multicast addresses\n");
1227                 return -ENOSPC;
1228         }
1229
1230         mcast = kzalloc(sizeof(*mcast), GFP_ATOMIC);
1231         if (!mcast)
1232                 return -ENOMEM;
1233
1234         ether_addr_copy(mcast->addr, addr);
1235         list_add_tail_rcu(&mcast->entry, &vif->fe_mcast_addr);
1236         vif->fe_mcast_count++;
1237
1238         return 0;
1239 }
1240
1241 static void xenvif_mcast_del(struct xenvif *vif, const u8 *addr)
1242 {
1243         struct xenvif_mcast_addr *mcast;
1244
1245         list_for_each_entry_rcu(mcast, &vif->fe_mcast_addr, entry) {
1246                 if (ether_addr_equal(addr, mcast->addr)) {
1247                         --vif->fe_mcast_count;
1248                         list_del_rcu(&mcast->entry);
1249                         kfree_rcu(mcast, rcu);
1250                         break;
1251                 }
1252         }
1253 }
1254
1255 bool xenvif_mcast_match(struct xenvif *vif, const u8 *addr)
1256 {
1257         struct xenvif_mcast_addr *mcast;
1258
1259         rcu_read_lock();
1260         list_for_each_entry_rcu(mcast, &vif->fe_mcast_addr, entry) {
1261                 if (ether_addr_equal(addr, mcast->addr)) {
1262                         rcu_read_unlock();
1263                         return true;
1264                 }
1265         }
1266         rcu_read_unlock();
1267
1268         return false;
1269 }
1270
1271 void xenvif_mcast_addr_list_free(struct xenvif *vif)
1272 {
1273         /* No need for locking or RCU here. NAPI poll and TX queue
1274          * are stopped.
1275          */
1276         while (!list_empty(&vif->fe_mcast_addr)) {
1277                 struct xenvif_mcast_addr *mcast;
1278
1279                 mcast = list_first_entry(&vif->fe_mcast_addr,
1280                                          struct xenvif_mcast_addr,
1281                                          entry);
1282                 --vif->fe_mcast_count;
1283                 list_del(&mcast->entry);
1284                 kfree(mcast);
1285         }
1286 }
1287
1288 static void xenvif_tx_build_gops(struct xenvif_queue *queue,
1289                                      int budget,
1290                                      unsigned *copy_ops,
1291                                      unsigned *map_ops)
1292 {
1293         struct gnttab_map_grant_ref *gop = queue->tx_map_ops;
1294         struct sk_buff *skb, *nskb;
1295         int ret;
1296         unsigned int frag_overflow;
1297
1298         while (skb_queue_len(&queue->tx_queue) < budget) {
1299                 struct xen_netif_tx_request txreq;
1300                 struct xen_netif_tx_request txfrags[XEN_NETBK_LEGACY_SLOTS_MAX];
1301                 struct xen_netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX-1];
1302                 u16 pending_idx;
1303                 RING_IDX idx;
1304                 int work_to_do;
1305                 unsigned int data_len;
1306                 pending_ring_idx_t index;
1307
1308                 if (queue->tx.sring->req_prod - queue->tx.req_cons >
1309                     XEN_NETIF_TX_RING_SIZE) {
1310                         netdev_err(queue->vif->dev,
1311                                    "Impossible number of requests. "
1312                                    "req_prod %d, req_cons %d, size %ld\n",
1313                                    queue->tx.sring->req_prod, queue->tx.req_cons,
1314                                    XEN_NETIF_TX_RING_SIZE);
1315                         xenvif_fatal_tx_err(queue->vif);
1316                         break;
1317                 }
1318
1319                 work_to_do = RING_HAS_UNCONSUMED_REQUESTS(&queue->tx);
1320                 if (!work_to_do)
1321                         break;
1322
1323                 idx = queue->tx.req_cons;
1324                 rmb(); /* Ensure that we see the request before we copy it. */
1325                 memcpy(&txreq, RING_GET_REQUEST(&queue->tx, idx), sizeof(txreq));
1326
1327                 /* Credit-based scheduling. */
1328                 if (txreq.size > queue->remaining_credit &&
1329                     tx_credit_exceeded(queue, txreq.size))
1330                         break;
1331
1332                 queue->remaining_credit -= txreq.size;
1333
1334                 work_to_do--;
1335                 queue->tx.req_cons = ++idx;
1336
1337                 memset(extras, 0, sizeof(extras));
1338                 if (txreq.flags & XEN_NETTXF_extra_info) {
1339                         work_to_do = xenvif_get_extras(queue, extras,
1340                                                        work_to_do);
1341                         idx = queue->tx.req_cons;
1342                         if (unlikely(work_to_do < 0))
1343                                 break;
1344                 }
1345
1346                 if (extras[XEN_NETIF_EXTRA_TYPE_MCAST_ADD - 1].type) {
1347                         struct xen_netif_extra_info *extra;
1348
1349                         extra = &extras[XEN_NETIF_EXTRA_TYPE_MCAST_ADD - 1];
1350                         ret = xenvif_mcast_add(queue->vif, extra->u.mcast.addr);
1351
1352                         make_tx_response(queue, &txreq,
1353                                          (ret == 0) ?
1354                                          XEN_NETIF_RSP_OKAY :
1355                                          XEN_NETIF_RSP_ERROR);
1356                         push_tx_responses(queue);
1357                         continue;
1358                 }
1359
1360                 if (extras[XEN_NETIF_EXTRA_TYPE_MCAST_DEL - 1].type) {
1361                         struct xen_netif_extra_info *extra;
1362
1363                         extra = &extras[XEN_NETIF_EXTRA_TYPE_MCAST_DEL - 1];
1364                         xenvif_mcast_del(queue->vif, extra->u.mcast.addr);
1365
1366                         make_tx_response(queue, &txreq, XEN_NETIF_RSP_OKAY);
1367                         push_tx_responses(queue);
1368                         continue;
1369                 }
1370
1371                 ret = xenvif_count_requests(queue, &txreq, txfrags, work_to_do);
1372                 if (unlikely(ret < 0))
1373                         break;
1374
1375                 idx += ret;
1376
1377                 if (unlikely(txreq.size < ETH_HLEN)) {
1378                         netdev_dbg(queue->vif->dev,
1379                                    "Bad packet size: %d\n", txreq.size);
1380                         xenvif_tx_err(queue, &txreq, idx);
1381                         break;
1382                 }
1383
1384                 /* No crossing a page as the payload mustn't fragment. */
1385                 if (unlikely((txreq.offset + txreq.size) > XEN_PAGE_SIZE)) {
1386                         netdev_err(queue->vif->dev,
1387                                    "txreq.offset: %u, size: %u, end: %lu\n",
1388                                    txreq.offset, txreq.size,
1389                                    (unsigned long)(txreq.offset&~XEN_PAGE_MASK) + txreq.size);
1390                         xenvif_fatal_tx_err(queue->vif);
1391                         break;
1392                 }
1393
1394                 index = pending_index(queue->pending_cons);
1395                 pending_idx = queue->pending_ring[index];
1396
1397                 data_len = (txreq.size > XEN_NETBACK_TX_COPY_LEN &&
1398                             ret < XEN_NETBK_LEGACY_SLOTS_MAX) ?
1399                         XEN_NETBACK_TX_COPY_LEN : txreq.size;
1400
1401                 skb = xenvif_alloc_skb(data_len);
1402                 if (unlikely(skb == NULL)) {
1403                         netdev_dbg(queue->vif->dev,
1404                                    "Can't allocate a skb in start_xmit.\n");
1405                         xenvif_tx_err(queue, &txreq, idx);
1406                         break;
1407                 }
1408
1409                 skb_shinfo(skb)->nr_frags = ret;
1410                 if (data_len < txreq.size)
1411                         skb_shinfo(skb)->nr_frags++;
1412                 /* At this point shinfo->nr_frags is in fact the number of
1413                  * slots, which can be as large as XEN_NETBK_LEGACY_SLOTS_MAX.
1414                  */
1415                 frag_overflow = 0;
1416                 nskb = NULL;
1417                 if (skb_shinfo(skb)->nr_frags > MAX_SKB_FRAGS) {
1418                         frag_overflow = skb_shinfo(skb)->nr_frags - MAX_SKB_FRAGS;
1419                         BUG_ON(frag_overflow > MAX_SKB_FRAGS);
1420                         skb_shinfo(skb)->nr_frags = MAX_SKB_FRAGS;
1421                         nskb = xenvif_alloc_skb(0);
1422                         if (unlikely(nskb == NULL)) {
1423                                 kfree_skb(skb);
1424                                 xenvif_tx_err(queue, &txreq, idx);
1425                                 if (net_ratelimit())
1426                                         netdev_err(queue->vif->dev,
1427                                                    "Can't allocate the frag_list skb.\n");
1428                                 break;
1429                         }
1430                 }
1431
1432                 if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
1433                         struct xen_netif_extra_info *gso;
1434                         gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];
1435
1436                         if (xenvif_set_skb_gso(queue->vif, skb, gso)) {
1437                                 /* Failure in xenvif_set_skb_gso is fatal. */
1438                                 kfree_skb(skb);
1439                                 kfree_skb(nskb);
1440                                 break;
1441                         }
1442                 }
1443
1444                 XENVIF_TX_CB(skb)->pending_idx = pending_idx;
1445
1446                 __skb_put(skb, data_len);
1447                 queue->tx_copy_ops[*copy_ops].source.u.ref = txreq.gref;
1448                 queue->tx_copy_ops[*copy_ops].source.domid = queue->vif->domid;
1449                 queue->tx_copy_ops[*copy_ops].source.offset = txreq.offset;
1450
1451                 queue->tx_copy_ops[*copy_ops].dest.u.gmfn =
1452                         virt_to_gfn(skb->data);
1453                 queue->tx_copy_ops[*copy_ops].dest.domid = DOMID_SELF;
1454                 queue->tx_copy_ops[*copy_ops].dest.offset =
1455                         offset_in_page(skb->data) & ~XEN_PAGE_MASK;
1456
1457                 queue->tx_copy_ops[*copy_ops].len = data_len;
1458                 queue->tx_copy_ops[*copy_ops].flags = GNTCOPY_source_gref;
1459
1460                 (*copy_ops)++;
1461
1462                 if (data_len < txreq.size) {
1463                         frag_set_pending_idx(&skb_shinfo(skb)->frags[0],
1464                                              pending_idx);
1465                         xenvif_tx_create_map_op(queue, pending_idx, &txreq, gop);
1466                         gop++;
1467                 } else {
1468                         frag_set_pending_idx(&skb_shinfo(skb)->frags[0],
1469                                              INVALID_PENDING_IDX);
1470                         memcpy(&queue->pending_tx_info[pending_idx].req, &txreq,
1471                                sizeof(txreq));
1472                 }
1473
1474                 queue->pending_cons++;
1475
1476                 gop = xenvif_get_requests(queue, skb, txfrags, gop,
1477                                           frag_overflow, nskb);
1478
1479                 __skb_queue_tail(&queue->tx_queue, skb);
1480
1481                 queue->tx.req_cons = idx;
1482
1483                 if (((gop-queue->tx_map_ops) >= ARRAY_SIZE(queue->tx_map_ops)) ||
1484                     (*copy_ops >= ARRAY_SIZE(queue->tx_copy_ops)))
1485                         break;
1486         }
1487
1488         (*map_ops) = gop - queue->tx_map_ops;
1489         return;
1490 }
1491
1492 /* Consolidate skb with a frag_list into a brand new one with local pages on
1493  * frags. Returns 0 or -ENOMEM if can't allocate new pages.
1494  */
1495 static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *skb)
1496 {
1497         unsigned int offset = skb_headlen(skb);
1498         skb_frag_t frags[MAX_SKB_FRAGS];
1499         int i, f;
1500         struct ubuf_info *uarg;
1501         struct sk_buff *nskb = skb_shinfo(skb)->frag_list;
1502
1503         queue->stats.tx_zerocopy_sent += 2;
1504         queue->stats.tx_frag_overflow++;
1505
1506         xenvif_fill_frags(queue, nskb);
1507         /* Subtract frags size, we will correct it later */
1508         skb->truesize -= skb->data_len;
1509         skb->len += nskb->len;
1510         skb->data_len += nskb->len;
1511
1512         /* create a brand new frags array and coalesce there */
1513         for (i = 0; offset < skb->len; i++) {
1514                 struct page *page;
1515                 unsigned int len;
1516
1517                 BUG_ON(i >= MAX_SKB_FRAGS);
1518                 page = alloc_page(GFP_ATOMIC);
1519                 if (!page) {
1520                         int j;
1521                         skb->truesize += skb->data_len;
1522                         for (j = 0; j < i; j++)
1523                                 put_page(frags[j].page.p);
1524                         return -ENOMEM;
1525                 }
1526
1527                 if (offset + PAGE_SIZE < skb->len)
1528                         len = PAGE_SIZE;
1529                 else
1530                         len = skb->len - offset;
1531                 if (skb_copy_bits(skb, offset, page_address(page), len))
1532                         BUG();
1533
1534                 offset += len;
1535                 frags[i].page.p = page;
1536                 frags[i].page_offset = 0;
1537                 skb_frag_size_set(&frags[i], len);
1538         }
1539
1540         /* Copied all the bits from the frag list -- free it. */
1541         skb_frag_list_init(skb);
1542         xenvif_skb_zerocopy_prepare(queue, nskb);
1543         kfree_skb(nskb);
1544
1545         /* Release all the original (foreign) frags. */
1546         for (f = 0; f < skb_shinfo(skb)->nr_frags; f++)
1547                 skb_frag_unref(skb, f);
1548         uarg = skb_shinfo(skb)->destructor_arg;
1549         /* increase inflight counter to offset decrement in callback */
1550         atomic_inc(&queue->inflight_packets);
1551         uarg->callback(uarg, true);
1552         skb_shinfo(skb)->destructor_arg = NULL;
1553
1554         /* Fill the skb with the new (local) frags. */
1555         memcpy(skb_shinfo(skb)->frags, frags, i * sizeof(skb_frag_t));
1556         skb_shinfo(skb)->nr_frags = i;
1557         skb->truesize += i * PAGE_SIZE;
1558
1559         return 0;
1560 }
1561
1562 static int xenvif_tx_submit(struct xenvif_queue *queue)
1563 {
1564         struct gnttab_map_grant_ref *gop_map = queue->tx_map_ops;
1565         struct gnttab_copy *gop_copy = queue->tx_copy_ops;
1566         struct sk_buff *skb;
1567         int work_done = 0;
1568
1569         while ((skb = __skb_dequeue(&queue->tx_queue)) != NULL) {
1570                 struct xen_netif_tx_request *txp;
1571                 u16 pending_idx;
1572                 unsigned data_len;
1573
1574                 pending_idx = XENVIF_TX_CB(skb)->pending_idx;
1575                 txp = &queue->pending_tx_info[pending_idx].req;
1576
1577                 /* Check the remap error code. */
1578                 if (unlikely(xenvif_tx_check_gop(queue, skb, &gop_map, &gop_copy))) {
1579                         /* If there was an error, xenvif_tx_check_gop is
1580                          * expected to release all the frags which were mapped,
1581                          * so kfree_skb shouldn't do it again
1582                          */
1583                         skb_shinfo(skb)->nr_frags = 0;
1584                         if (skb_has_frag_list(skb)) {
1585                                 struct sk_buff *nskb =
1586                                                 skb_shinfo(skb)->frag_list;
1587                                 skb_shinfo(nskb)->nr_frags = 0;
1588                         }
1589                         kfree_skb(skb);
1590                         continue;
1591                 }
1592
1593                 data_len = skb->len;
1594                 callback_param(queue, pending_idx).ctx = NULL;
1595                 if (data_len < txp->size) {
1596                         /* Append the packet payload as a fragment. */
1597                         txp->offset += data_len;
1598                         txp->size -= data_len;
1599                 } else {
1600                         /* Schedule a response immediately. */
1601                         xenvif_idx_release(queue, pending_idx,
1602                                            XEN_NETIF_RSP_OKAY);
1603                 }
1604
1605                 if (txp->flags & XEN_NETTXF_csum_blank)
1606                         skb->ip_summed = CHECKSUM_PARTIAL;
1607                 else if (txp->flags & XEN_NETTXF_data_validated)
1608                         skb->ip_summed = CHECKSUM_UNNECESSARY;
1609
1610                 xenvif_fill_frags(queue, skb);
1611
1612                 if (unlikely(skb_has_frag_list(skb))) {
1613                         if (xenvif_handle_frag_list(queue, skb)) {
1614                                 if (net_ratelimit())
1615                                         netdev_err(queue->vif->dev,
1616                                                    "Not enough memory to consolidate frag_list!\n");
1617                                 xenvif_skb_zerocopy_prepare(queue, skb);
1618                                 kfree_skb(skb);
1619                                 continue;
1620                         }
1621                 }
1622
1623                 skb->dev      = queue->vif->dev;
1624                 skb->protocol = eth_type_trans(skb, skb->dev);
1625                 skb_reset_network_header(skb);
1626
1627                 if (checksum_setup(queue, skb)) {
1628                         netdev_dbg(queue->vif->dev,
1629                                    "Can't setup checksum in net_tx_action\n");
1630                         /* We have to set this flag to trigger the callback */
1631                         if (skb_shinfo(skb)->destructor_arg)
1632                                 xenvif_skb_zerocopy_prepare(queue, skb);
1633                         kfree_skb(skb);
1634                         continue;
1635                 }
1636
1637                 skb_probe_transport_header(skb, 0);
1638
1639                 /* If the packet is GSO then we will have just set up the
1640                  * transport header offset in checksum_setup so it's now
1641                  * straightforward to calculate gso_segs.
1642                  */
1643                 if (skb_is_gso(skb)) {
1644                         int mss = skb_shinfo(skb)->gso_size;
1645                         int hdrlen = skb_transport_header(skb) -
1646                                 skb_mac_header(skb) +
1647                                 tcp_hdrlen(skb);
1648
1649                         skb_shinfo(skb)->gso_segs =
1650                                 DIV_ROUND_UP(skb->len - hdrlen, mss);
1651                 }
1652
1653                 queue->stats.rx_bytes += skb->len;
1654                 queue->stats.rx_packets++;
1655
1656                 work_done++;
1657
1658                 /* Set this flag right before netif_receive_skb, otherwise
1659                  * someone might think this packet already left netback, and
1660                  * do a skb_copy_ubufs while we are still in control of the
1661                  * skb. E.g. the __pskb_pull_tail earlier can do such thing.
1662                  */
1663                 if (skb_shinfo(skb)->destructor_arg) {
1664                         xenvif_skb_zerocopy_prepare(queue, skb);
1665                         queue->stats.tx_zerocopy_sent++;
1666                 }
1667
1668                 netif_receive_skb(skb);
1669         }
1670
1671         return work_done;
1672 }
1673
1674 void xenvif_zerocopy_callback(struct ubuf_info *ubuf, bool zerocopy_success)
1675 {
1676         unsigned long flags;
1677         pending_ring_idx_t index;
1678         struct xenvif_queue *queue = ubuf_to_queue(ubuf);
1679
1680         /* This is the only place where we grab this lock, to protect callbacks
1681          * from each other.
1682          */
1683         spin_lock_irqsave(&queue->callback_lock, flags);
1684         do {
1685                 u16 pending_idx = ubuf->desc;
1686                 ubuf = (struct ubuf_info *) ubuf->ctx;
1687                 BUG_ON(queue->dealloc_prod - queue->dealloc_cons >=
1688                         MAX_PENDING_REQS);
1689                 index = pending_index(queue->dealloc_prod);
1690                 queue->dealloc_ring[index] = pending_idx;
1691                 /* Sync with xenvif_tx_dealloc_action:
1692                  * insert idx then incr producer.
1693                  */
1694                 smp_wmb();
1695                 queue->dealloc_prod++;
1696         } while (ubuf);
1697         spin_unlock_irqrestore(&queue->callback_lock, flags);
1698
1699         if (likely(zerocopy_success))
1700                 queue->stats.tx_zerocopy_success++;
1701         else
1702                 queue->stats.tx_zerocopy_fail++;
1703         xenvif_skb_zerocopy_complete(queue);
1704 }
1705
1706 static inline void xenvif_tx_dealloc_action(struct xenvif_queue *queue)
1707 {
1708         struct gnttab_unmap_grant_ref *gop;
1709         pending_ring_idx_t dc, dp;
1710         u16 pending_idx, pending_idx_release[MAX_PENDING_REQS];
1711         unsigned int i = 0;
1712
1713         dc = queue->dealloc_cons;
1714         gop = queue->tx_unmap_ops;
1715
1716         /* Free up any grants we have finished using */
1717         do {
1718                 dp = queue->dealloc_prod;
1719
1720                 /* Ensure we see all indices enqueued by all
1721                  * xenvif_zerocopy_callback().
1722                  */
1723                 smp_rmb();
1724
1725                 while (dc != dp) {
1726                         BUG_ON(gop - queue->tx_unmap_ops >= MAX_PENDING_REQS);
1727                         pending_idx =
1728                                 queue->dealloc_ring[pending_index(dc++)];
1729
1730                         pending_idx_release[gop - queue->tx_unmap_ops] =
1731                                 pending_idx;
1732                         queue->pages_to_unmap[gop - queue->tx_unmap_ops] =
1733                                 queue->mmap_pages[pending_idx];
1734                         gnttab_set_unmap_op(gop,
1735                                             idx_to_kaddr(queue, pending_idx),
1736                                             GNTMAP_host_map,
1737                                             queue->grant_tx_handle[pending_idx]);
1738                         xenvif_grant_handle_reset(queue, pending_idx);
1739                         ++gop;
1740                 }
1741
1742         } while (dp != queue->dealloc_prod);
1743
1744         queue->dealloc_cons = dc;
1745
1746         if (gop - queue->tx_unmap_ops > 0) {
1747                 int ret;
1748                 ret = gnttab_unmap_refs(queue->tx_unmap_ops,
1749                                         NULL,
1750                                         queue->pages_to_unmap,
1751                                         gop - queue->tx_unmap_ops);
1752                 if (ret) {
1753                         netdev_err(queue->vif->dev, "Unmap fail: nr_ops %tu ret %d\n",
1754                                    gop - queue->tx_unmap_ops, ret);
1755                         for (i = 0; i < gop - queue->tx_unmap_ops; ++i) {
1756                                 if (gop[i].status != GNTST_okay)
1757                                         netdev_err(queue->vif->dev,
1758                                                    " host_addr: 0x%llx handle: 0x%x status: %d\n",
1759                                                    gop[i].host_addr,
1760                                                    gop[i].handle,
1761                                                    gop[i].status);
1762                         }
1763                         BUG();
1764                 }
1765         }
1766
1767         for (i = 0; i < gop - queue->tx_unmap_ops; ++i)
1768                 xenvif_idx_release(queue, pending_idx_release[i],
1769                                    XEN_NETIF_RSP_OKAY);
1770 }
1771
1772
1773 /* Called after netfront has transmitted */
1774 int xenvif_tx_action(struct xenvif_queue *queue, int budget)
1775 {
1776         unsigned nr_mops, nr_cops = 0;
1777         int work_done, ret;
1778
1779         if (unlikely(!tx_work_todo(queue)))
1780                 return 0;
1781
1782         xenvif_tx_build_gops(queue, budget, &nr_cops, &nr_mops);
1783
1784         if (nr_cops == 0)
1785                 return 0;
1786
1787         gnttab_batch_copy(queue->tx_copy_ops, nr_cops);
1788         if (nr_mops != 0) {
1789                 ret = gnttab_map_refs(queue->tx_map_ops,
1790                                       NULL,
1791                                       queue->pages_to_map,
1792                                       nr_mops);
1793                 BUG_ON(ret);
1794         }
1795
1796         work_done = xenvif_tx_submit(queue);
1797
1798         return work_done;
1799 }
1800
1801 static void xenvif_idx_release(struct xenvif_queue *queue, u16 pending_idx,
1802                                u8 status)
1803 {
1804         struct pending_tx_info *pending_tx_info;
1805         pending_ring_idx_t index;
1806         unsigned long flags;
1807
1808         pending_tx_info = &queue->pending_tx_info[pending_idx];
1809
1810         spin_lock_irqsave(&queue->response_lock, flags);
1811
1812         make_tx_response(queue, &pending_tx_info->req, status);
1813
1814         /* Release the pending index before pusing the Tx response so
1815          * its available before a new Tx request is pushed by the
1816          * frontend.
1817          */
1818         index = pending_index(queue->pending_prod++);
1819         queue->pending_ring[index] = pending_idx;
1820
1821         push_tx_responses(queue);
1822
1823         spin_unlock_irqrestore(&queue->response_lock, flags);
1824 }
1825
1826
1827 static void make_tx_response(struct xenvif_queue *queue,
1828                              struct xen_netif_tx_request *txp,
1829                              s8       st)
1830 {
1831         RING_IDX i = queue->tx.rsp_prod_pvt;
1832         struct xen_netif_tx_response *resp;
1833
1834         resp = RING_GET_RESPONSE(&queue->tx, i);
1835         resp->id     = txp->id;
1836         resp->status = st;
1837
1838         if (txp->flags & XEN_NETTXF_extra_info)
1839                 RING_GET_RESPONSE(&queue->tx, ++i)->status = XEN_NETIF_RSP_NULL;
1840
1841         queue->tx.rsp_prod_pvt = ++i;
1842 }
1843
1844 static void push_tx_responses(struct xenvif_queue *queue)
1845 {
1846         int notify;
1847
1848         RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&queue->tx, notify);
1849         if (notify)
1850                 notify_remote_via_irq(queue->tx_irq);
1851 }
1852
1853 static struct xen_netif_rx_response *make_rx_response(struct xenvif_queue *queue,
1854                                              u16      id,
1855                                              s8       st,
1856                                              u16      offset,
1857                                              u16      size,
1858                                              u16      flags)
1859 {
1860         RING_IDX i = queue->rx.rsp_prod_pvt;
1861         struct xen_netif_rx_response *resp;
1862
1863         resp = RING_GET_RESPONSE(&queue->rx, i);
1864         resp->offset     = offset;
1865         resp->flags      = flags;
1866         resp->id         = id;
1867         resp->status     = (s16)size;
1868         if (st < 0)
1869                 resp->status = (s16)st;
1870
1871         queue->rx.rsp_prod_pvt = ++i;
1872
1873         return resp;
1874 }
1875
1876 void xenvif_idx_unmap(struct xenvif_queue *queue, u16 pending_idx)
1877 {
1878         int ret;
1879         struct gnttab_unmap_grant_ref tx_unmap_op;
1880
1881         gnttab_set_unmap_op(&tx_unmap_op,
1882                             idx_to_kaddr(queue, pending_idx),
1883                             GNTMAP_host_map,
1884                             queue->grant_tx_handle[pending_idx]);
1885         xenvif_grant_handle_reset(queue, pending_idx);
1886
1887         ret = gnttab_unmap_refs(&tx_unmap_op, NULL,
1888                                 &queue->mmap_pages[pending_idx], 1);
1889         if (ret) {
1890                 netdev_err(queue->vif->dev,
1891                            "Unmap fail: ret: %d pending_idx: %d host_addr: %llx handle: 0x%x status: %d\n",
1892                            ret,
1893                            pending_idx,
1894                            tx_unmap_op.host_addr,
1895                            tx_unmap_op.handle,
1896                            tx_unmap_op.status);
1897                 BUG();
1898         }
1899 }
1900
1901 static inline int tx_work_todo(struct xenvif_queue *queue)
1902 {
1903         if (likely(RING_HAS_UNCONSUMED_REQUESTS(&queue->tx)))
1904                 return 1;
1905
1906         return 0;
1907 }
1908
1909 static inline bool tx_dealloc_work_todo(struct xenvif_queue *queue)
1910 {
1911         return queue->dealloc_cons != queue->dealloc_prod;
1912 }
1913
1914 void xenvif_unmap_frontend_rings(struct xenvif_queue *queue)
1915 {
1916         if (queue->tx.sring)
1917                 xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(queue->vif),
1918                                         queue->tx.sring);
1919         if (queue->rx.sring)
1920                 xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(queue->vif),
1921                                         queue->rx.sring);
1922 }
1923
1924 int xenvif_map_frontend_rings(struct xenvif_queue *queue,
1925                               grant_ref_t tx_ring_ref,
1926                               grant_ref_t rx_ring_ref)
1927 {
1928         void *addr;
1929         struct xen_netif_tx_sring *txs;
1930         struct xen_netif_rx_sring *rxs;
1931
1932         int err = -ENOMEM;
1933
1934         err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(queue->vif),
1935                                      &tx_ring_ref, 1, &addr);
1936         if (err)
1937                 goto err;
1938
1939         txs = (struct xen_netif_tx_sring *)addr;
1940         BACK_RING_INIT(&queue->tx, txs, XEN_PAGE_SIZE);
1941
1942         err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(queue->vif),
1943                                      &rx_ring_ref, 1, &addr);
1944         if (err)
1945                 goto err;
1946
1947         rxs = (struct xen_netif_rx_sring *)addr;
1948         BACK_RING_INIT(&queue->rx, rxs, XEN_PAGE_SIZE);
1949
1950         return 0;
1951
1952 err:
1953         xenvif_unmap_frontend_rings(queue);
1954         return err;
1955 }
1956
1957 static void xenvif_queue_carrier_off(struct xenvif_queue *queue)
1958 {
1959         struct xenvif *vif = queue->vif;
1960
1961         queue->stalled = true;
1962
1963         /* At least one queue has stalled? Disable the carrier. */
1964         spin_lock(&vif->lock);
1965         if (vif->stalled_queues++ == 0) {
1966                 netdev_info(vif->dev, "Guest Rx stalled");
1967                 netif_carrier_off(vif->dev);
1968         }
1969         spin_unlock(&vif->lock);
1970 }
1971
1972 static void xenvif_queue_carrier_on(struct xenvif_queue *queue)
1973 {
1974         struct xenvif *vif = queue->vif;
1975
1976         queue->last_rx_time = jiffies; /* Reset Rx stall detection. */
1977         queue->stalled = false;
1978
1979         /* All queues are ready? Enable the carrier. */
1980         spin_lock(&vif->lock);
1981         if (--vif->stalled_queues == 0) {
1982                 netdev_info(vif->dev, "Guest Rx ready");
1983                 netif_carrier_on(vif->dev);
1984         }
1985         spin_unlock(&vif->lock);
1986 }
1987
1988 static bool xenvif_rx_queue_stalled(struct xenvif_queue *queue)
1989 {
1990         RING_IDX prod, cons;
1991
1992         prod = queue->rx.sring->req_prod;
1993         cons = queue->rx.req_cons;
1994
1995         return !queue->stalled && prod - cons < 1
1996                 && time_after(jiffies,
1997                               queue->last_rx_time + queue->vif->stall_timeout);
1998 }
1999
2000 static bool xenvif_rx_queue_ready(struct xenvif_queue *queue)
2001 {
2002         RING_IDX prod, cons;
2003
2004         prod = queue->rx.sring->req_prod;
2005         cons = queue->rx.req_cons;
2006
2007         return queue->stalled && prod - cons >= 1;
2008 }
2009
2010 static bool xenvif_have_rx_work(struct xenvif_queue *queue)
2011 {
2012         return (!skb_queue_empty(&queue->rx_queue)
2013                 && xenvif_rx_ring_slots_available(queue))
2014                 || (queue->vif->stall_timeout &&
2015                     (xenvif_rx_queue_stalled(queue)
2016                      || xenvif_rx_queue_ready(queue)))
2017                 || kthread_should_stop()
2018                 || queue->vif->disabled;
2019 }
2020
2021 static long xenvif_rx_queue_timeout(struct xenvif_queue *queue)
2022 {
2023         struct sk_buff *skb;
2024         long timeout;
2025
2026         skb = skb_peek(&queue->rx_queue);
2027         if (!skb)
2028                 return MAX_SCHEDULE_TIMEOUT;
2029
2030         timeout = XENVIF_RX_CB(skb)->expires - jiffies;
2031         return timeout < 0 ? 0 : timeout;
2032 }
2033
2034 /* Wait until the guest Rx thread has work.
2035  *
2036  * The timeout needs to be adjusted based on the current head of the
2037  * queue (and not just the head at the beginning).  In particular, if
2038  * the queue is initially empty an infinite timeout is used and this
2039  * needs to be reduced when a skb is queued.
2040  *
2041  * This cannot be done with wait_event_timeout() because it only
2042  * calculates the timeout once.
2043  */
2044 static void xenvif_wait_for_rx_work(struct xenvif_queue *queue)
2045 {
2046         DEFINE_WAIT(wait);
2047
2048         if (xenvif_have_rx_work(queue))
2049                 return;
2050
2051         for (;;) {
2052                 long ret;
2053
2054                 prepare_to_wait(&queue->wq, &wait, TASK_INTERRUPTIBLE);
2055                 if (xenvif_have_rx_work(queue))
2056                         break;
2057                 ret = schedule_timeout(xenvif_rx_queue_timeout(queue));
2058                 if (!ret)
2059                         break;
2060         }
2061         finish_wait(&queue->wq, &wait);
2062 }
2063
2064 int xenvif_kthread_guest_rx(void *data)
2065 {
2066         struct xenvif_queue *queue = data;
2067         struct xenvif *vif = queue->vif;
2068
2069         if (!vif->stall_timeout)
2070                 xenvif_queue_carrier_on(queue);
2071
2072         for (;;) {
2073                 xenvif_wait_for_rx_work(queue);
2074
2075                 if (kthread_should_stop())
2076                         break;
2077
2078                 /* This frontend is found to be rogue, disable it in
2079                  * kthread context. Currently this is only set when
2080                  * netback finds out frontend sends malformed packet,
2081                  * but we cannot disable the interface in softirq
2082                  * context so we defer it here, if this thread is
2083                  * associated with queue 0.
2084                  */
2085                 if (unlikely(vif->disabled && queue->id == 0)) {
2086                         xenvif_carrier_off(vif);
2087                         break;
2088                 }
2089
2090                 if (!skb_queue_empty(&queue->rx_queue))
2091                         xenvif_rx_action(queue);
2092
2093                 /* If the guest hasn't provided any Rx slots for a
2094                  * while it's probably not responsive, drop the
2095                  * carrier so packets are dropped earlier.
2096                  */
2097                 if (vif->stall_timeout) {
2098                         if (xenvif_rx_queue_stalled(queue))
2099                                 xenvif_queue_carrier_off(queue);
2100                         else if (xenvif_rx_queue_ready(queue))
2101                                 xenvif_queue_carrier_on(queue);
2102                 }
2103
2104                 /* Queued packets may have foreign pages from other
2105                  * domains.  These cannot be queued indefinitely as
2106                  * this would starve guests of grant refs and transmit
2107                  * slots.
2108                  */
2109                 xenvif_rx_queue_drop_expired(queue);
2110
2111                 xenvif_rx_queue_maybe_wake(queue);
2112
2113                 cond_resched();
2114         }
2115
2116         /* Bin any remaining skbs */
2117         xenvif_rx_queue_purge(queue);
2118
2119         return 0;
2120 }
2121
2122 static bool xenvif_dealloc_kthread_should_stop(struct xenvif_queue *queue)
2123 {
2124         /* Dealloc thread must remain running until all inflight
2125          * packets complete.
2126          */
2127         return kthread_should_stop() &&
2128                 !atomic_read(&queue->inflight_packets);
2129 }
2130
2131 int xenvif_dealloc_kthread(void *data)
2132 {
2133         struct xenvif_queue *queue = data;
2134
2135         for (;;) {
2136                 wait_event_interruptible(queue->dealloc_wq,
2137                                          tx_dealloc_work_todo(queue) ||
2138                                          xenvif_dealloc_kthread_should_stop(queue));
2139                 if (xenvif_dealloc_kthread_should_stop(queue))
2140                         break;
2141
2142                 xenvif_tx_dealloc_action(queue);
2143                 cond_resched();
2144         }
2145
2146         /* Unmap anything remaining*/
2147         if (tx_dealloc_work_todo(queue))
2148                 xenvif_tx_dealloc_action(queue);
2149
2150         return 0;
2151 }
2152
2153 static int __init netback_init(void)
2154 {
2155         int rc = 0;
2156
2157         if (!xen_domain())
2158                 return -ENODEV;
2159
2160         /* Allow as many queues as there are CPUs if user has not
2161          * specified a value.
2162          */
2163         if (xenvif_max_queues == 0)
2164                 xenvif_max_queues = num_online_cpus();
2165
2166         if (fatal_skb_slots < XEN_NETBK_LEGACY_SLOTS_MAX) {
2167                 pr_info("fatal_skb_slots too small (%d), bump it to XEN_NETBK_LEGACY_SLOTS_MAX (%d)\n",
2168                         fatal_skb_slots, XEN_NETBK_LEGACY_SLOTS_MAX);
2169                 fatal_skb_slots = XEN_NETBK_LEGACY_SLOTS_MAX;
2170         }
2171
2172         rc = xenvif_xenbus_init();
2173         if (rc)
2174                 goto failed_init;
2175
2176 #ifdef CONFIG_DEBUG_FS
2177         xen_netback_dbg_root = debugfs_create_dir("xen-netback", NULL);
2178         if (IS_ERR_OR_NULL(xen_netback_dbg_root))
2179                 pr_warn("Init of debugfs returned %ld!\n",
2180                         PTR_ERR(xen_netback_dbg_root));
2181 #endif /* CONFIG_DEBUG_FS */
2182
2183         return 0;
2184
2185 failed_init:
2186         return rc;
2187 }
2188
2189 module_init(netback_init);
2190
2191 static void __exit netback_fini(void)
2192 {
2193 #ifdef CONFIG_DEBUG_FS
2194         if (!IS_ERR_OR_NULL(xen_netback_dbg_root))
2195                 debugfs_remove_recursive(xen_netback_dbg_root);
2196 #endif /* CONFIG_DEBUG_FS */
2197         xenvif_xenbus_fini();
2198 }
2199 module_exit(netback_fini);
2200
2201 MODULE_LICENSE("Dual BSD/GPL");
2202 MODULE_ALIAS("xen-backend:vif");