drivers/net/xen-netback/common.h

   1 /*
   2  * This program is free software; you can redistribute it and/or
   3  * modify it under the terms of the GNU General Public License version 2
   4  * as published by the Free Software Foundation; or, when distributed
   5  * separately from the Linux kernel or incorporated into other
   6  * software packages, subject to the following license:
   7  *
   8  * Permission is hereby granted, free of charge, to any person obtaining a copy
   9  * of this source file (the "Software"), to deal in the Software without
  10  * restriction, including without limitation the rights to use, copy, modify,
  11  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
  12  * and to permit persons to whom the Software is furnished to do so, subject to
  13  * the following conditions:
  14  *
  15  * The above copyright notice and this permission notice shall be included in
  16  * all copies or substantial portions of the Software.
  17  *
  18  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  19  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  20  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  21  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  22  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  23  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  24  * IN THE SOFTWARE.
  25  */
  26
  27 #ifndef __XEN_NETBACK__COMMON_H__
  28 #define __XEN_NETBACK__COMMON_H__
  29
  30 #define pr_fmt(fmt) KBUILD_MODNAME ":%s: " fmt, __func__
  31
  32 #include <linux/module.h>
  33 #include <linux/interrupt.h>
  34 #include <linux/slab.h>
  35 #include <linux/ip.h>
  36 #include <linux/in.h>
  37 #include <linux/io.h>
  38 #include <linux/netdevice.h>
  39 #include <linux/etherdevice.h>
  40 #include <linux/wait.h>
  41 #include <linux/sched.h>
  42
  43 #include <xen/interface/io/netif.h>
  44 #include <xen/interface/grant_table.h>
  45 #include <xen/grant_table.h>
  46 #include <xen/xenbus.h>
  47 #include <linux/debugfs.h>
  48
  49 typedef unsigned int pending_ring_idx_t;
  50 #define INVALID_PENDING_RING_IDX (~0U)
  51
  52 struct pending_tx_info {
  53         struct xen_netif_tx_request req; /* tx request */
  54         /* Callback data for released SKBs. The callback is always
  55          * xenvif_zerocopy_callback, desc contains the pending_idx, which is
  56          * also an index in pending_tx_info array. It is initialized in
  57          * xenvif_alloc and it never changes.
  58          * skb_shinfo(skb)->destructor_arg points to the first mapped slot's
  59          * callback_struct in this array of struct pending_tx_info's, then ctx
  60          * to the next, or NULL if there is no more slot for this skb.
  61          * ubuf_to_vif is a helper which finds the struct xenvif from a pointer
  62          * to this field.
  63          */
  64         struct ubuf_info callback_struct;
  65 };
  66
  67 #define XEN_NETIF_TX_RING_SIZE __CONST_RING_SIZE(xen_netif_tx, PAGE_SIZE)
  68 #define XEN_NETIF_RX_RING_SIZE __CONST_RING_SIZE(xen_netif_rx, PAGE_SIZE)
  69
  70 struct xenvif_rx_meta {
  71         int id;
  72         int size;
  73         int gso_type;
  74         int gso_size;
  75 };
  76
  77 #define GSO_BIT(type) \
  78         (1 << XEN_NETIF_GSO_TYPE_ ## type)
  79
  80 /* Discriminate from any valid pending_idx value. */
  81 #define INVALID_PENDING_IDX 0xFFFF
  82
  83 #define MAX_BUFFER_OFFSET PAGE_SIZE
  84
  85 #define MAX_PENDING_REQS XEN_NETIF_TX_RING_SIZE
  86
  87 /* It's possible for an skb to have a maximal number of frags
  88  * but still be less than MAX_BUFFER_OFFSET in size. Thus the
  89  * worst-case number of copy operations is MAX_SKB_FRAGS per
  90  * ring slot.
  91  */
  92 #define MAX_GRANT_COPY_OPS (MAX_SKB_FRAGS * XEN_NETIF_RX_RING_SIZE)
  93
  94 #define NETBACK_INVALID_HANDLE -1
  95
  96 /* To avoid confusion, we define XEN_NETBK_LEGACY_SLOTS_MAX indicating
  97  * the maximum slots a valid packet can use. Now this value is defined
  98  * to be XEN_NETIF_NR_SLOTS_MIN, which is supposed to be supported by
  99  * all backend.
 100  */
 101 #define XEN_NETBK_LEGACY_SLOTS_MAX XEN_NETIF_NR_SLOTS_MIN
 102
 103 /* Queue name is interface name with "-qNNN" appended */
 104 #define QUEUE_NAME_SIZE (IFNAMSIZ + 5)
 105
 106 /* IRQ name is queue name with "-tx" or "-rx" appended */
 107 #define IRQ_NAME_SIZE (QUEUE_NAME_SIZE + 3)
 108
 109 struct xenvif;
 110
 111 struct xenvif_stats {
 112         /* Stats fields to be updated per-queue.
 113          * A subset of struct net_device_stats that contains only the
 114          * fields that are updated in netback.c for each queue.
 115          */
 116         unsigned int rx_bytes;
 117         unsigned int rx_packets;
 118         unsigned int tx_bytes;
 119         unsigned int tx_packets;
 120
 121         /* Additional stats used by xenvif */
 122         unsigned long rx_gso_checksum_fixup;
 123         unsigned long tx_zerocopy_sent;
 124         unsigned long tx_zerocopy_success;
 125         unsigned long tx_zerocopy_fail;
 126         unsigned long tx_frag_overflow;
 127 };
 128
 129 struct xenvif_queue { /* Per-queue data for xenvif */
 130         unsigned int id; /* Queue ID, 0-based */
 131         char name[QUEUE_NAME_SIZE]; /* DEVNAME-qN */
 132         struct xenvif *vif; /* Parent VIF */
 133
 134         /* Use NAPI for guest TX */
 135         struct napi_struct napi;
 136         /* When feature-split-event-channels = 0, tx_irq = rx_irq. */
 137         unsigned int tx_irq;
 138         /* Only used when feature-split-event-channels = 1 */
 139         char tx_irq_name[IRQ_NAME_SIZE]; /* DEVNAME-qN-tx */
 140         struct xen_netif_tx_back_ring tx;
 141         struct sk_buff_head tx_queue;
 142         struct page *mmap_pages[MAX_PENDING_REQS];
 143         pending_ring_idx_t pending_prod;
 144         pending_ring_idx_t pending_cons;
 145         u16 pending_ring[MAX_PENDING_REQS];
 146         struct pending_tx_info pending_tx_info[MAX_PENDING_REQS];
 147         grant_handle_t grant_tx_handle[MAX_PENDING_REQS];
 148
 149         struct gnttab_copy tx_copy_ops[MAX_PENDING_REQS];
 150         struct gnttab_map_grant_ref tx_map_ops[MAX_PENDING_REQS];
 151         struct gnttab_unmap_grant_ref tx_unmap_ops[MAX_PENDING_REQS];
 152         /* passed to gnttab_[un]map_refs with pages under (un)mapping */
 153         struct page *pages_to_map[MAX_PENDING_REQS];
 154         struct page *pages_to_unmap[MAX_PENDING_REQS];
 155
 156         /* This prevents zerocopy callbacks  to race over dealloc_ring */
 157         spinlock_t callback_lock;
 158         /* This prevents dealloc thread and NAPI instance to race over response
 159          * creation and pending_ring in xenvif_idx_release. In xenvif_tx_err
 160          * it only protect response creation
 161          */
 162         spinlock_t response_lock;
 163         pending_ring_idx_t dealloc_prod;
 164         pending_ring_idx_t dealloc_cons;
 165         u16 dealloc_ring[MAX_PENDING_REQS];
 166         struct task_struct *dealloc_task;
 167         wait_queue_head_t dealloc_wq;
 168         atomic_t inflight_packets;
 169
 170         /* Use kthread for guest RX */
 171         struct task_struct *task;
 172         wait_queue_head_t wq;
 173         /* When feature-split-event-channels = 0, tx_irq = rx_irq. */
 174         unsigned int rx_irq;
 175         /* Only used when feature-split-event-channels = 1 */
 176         char rx_irq_name[IRQ_NAME_SIZE]; /* DEVNAME-qN-rx */
 177         struct xen_netif_rx_back_ring rx;
 178         struct sk_buff_head rx_queue;
 179
 180         unsigned int rx_queue_max;
 181         unsigned int rx_queue_len;
 182         unsigned long last_rx_time;
 183         bool stalled;
 184
 185         struct gnttab_copy grant_copy_op[MAX_GRANT_COPY_OPS];
 186
 187         /* We create one meta structure per ring request we consume, so
 188          * the maximum number is the same as the ring size.
 189          */
 190         struct xenvif_rx_meta meta[XEN_NETIF_RX_RING_SIZE];
 191
 192         /* Transmit shaping: allow 'credit_bytes' every 'credit_usec'. */
 193         unsigned long   credit_bytes;
 194         unsigned long   credit_usec;
 195         unsigned long   remaining_credit;
 196         struct timer_list credit_timeout;
 197         u64 credit_window_start;
 198
 199         /* Statistics */
 200         struct xenvif_stats stats;
 201 };
 202
 203 /* Maximum number of Rx slots a to-guest packet may use, including the
 204  * slot needed for GSO meta-data.
 205  */
 206 #define XEN_NETBK_RX_SLOTS_MAX (MAX_SKB_FRAGS + 1)
 207
 208 enum state_bit_shift {
 209         /* This bit marks that the vif is connected */
 210         VIF_STATUS_CONNECTED,
 211 };
 212
 213 struct xenvif {
 214         /* Unique identifier for this interface. */
 215         domid_t          domid;
 216         unsigned int     handle;
 217
 218         u8               fe_dev_addr[6];
 219
 220         /* Frontend feature information. */
 221         int gso_mask;
 222         int gso_prefix_mask;
 223
 224         u8 can_sg:1;
 225         u8 ip_csum:1;
 226         u8 ipv6_csum:1;
 227
 228         /* Is this interface disabled? True when backend discovers
 229          * frontend is rogue.
 230          */
 231         bool disabled;
 232         unsigned long status;
 233         unsigned long drain_timeout;
 234         unsigned long stall_timeout;
 235
 236         /* Queues */
 237         struct xenvif_queue *queues;
 238         unsigned int num_queues; /* active queues, resource allocated */
 239         unsigned int stalled_queues;
 240
 241         spinlock_t lock;
 242
 243 #ifdef CONFIG_DEBUG_FS
 244         struct dentry *xenvif_dbg_root;
 245 #endif
 246
 247         /* Miscellaneous private stuff. */
 248         struct net_device *dev;
 249 };
 250
 251 struct xenvif_rx_cb {
 252         unsigned long expires;
 253         int meta_slots_used;
 254 };
 255
 256 #define XENVIF_RX_CB(skb) ((struct xenvif_rx_cb *)(skb)->cb)
 257
 258 static inline struct xenbus_device *xenvif_to_xenbus_device(struct xenvif *vif)
 259 {
 260         return to_xenbus_device(vif->dev->dev.parent);
 261 }
 262
 263 struct xenvif *xenvif_alloc(struct device *parent,
 264                             domid_t domid,
 265                             unsigned int handle);
 266
 267 int xenvif_init_queue(struct xenvif_queue *queue);
 268 void xenvif_deinit_queue(struct xenvif_queue *queue);
 269
 270 int xenvif_connect(struct xenvif_queue *queue, unsigned long tx_ring_ref,
 271                    unsigned long rx_ring_ref, unsigned int tx_evtchn,
 272                    unsigned int rx_evtchn);
 273 void xenvif_disconnect(struct xenvif *vif);
 274 void xenvif_free(struct xenvif *vif);
 275
 276 int xenvif_xenbus_init(void);
 277 void xenvif_xenbus_fini(void);
 278
 279 int xenvif_schedulable(struct xenvif *vif);
 280
 281 int xenvif_queue_stopped(struct xenvif_queue *queue);
 282 void xenvif_wake_queue(struct xenvif_queue *queue);
 283
 284 /* (Un)Map communication rings. */
 285 void xenvif_unmap_frontend_rings(struct xenvif_queue *queue);
 286 int xenvif_map_frontend_rings(struct xenvif_queue *queue,
 287                               grant_ref_t tx_ring_ref,
 288                               grant_ref_t rx_ring_ref);
 289
 290 /* Check for SKBs from frontend and schedule backend processing */
 291 void xenvif_napi_schedule_or_enable_events(struct xenvif_queue *queue);
 292
 293 /* Prevent the device from generating any further traffic. */
 294 void xenvif_carrier_off(struct xenvif *vif);
 295
 296 int xenvif_tx_action(struct xenvif_queue *queue, int budget);
 297
 298 int xenvif_kthread_guest_rx(void *data);
 299 void xenvif_kick_thread(struct xenvif_queue *queue);
 300
 301 int xenvif_dealloc_kthread(void *data);
 302
 303 void xenvif_rx_queue_tail(struct xenvif_queue *queue, struct sk_buff *skb);
 304
 305 /* Determine whether the needed number of slots (req) are available,
 306  * and set req_event if not.
 307  */
 308 bool xenvif_rx_ring_slots_available(struct xenvif_queue *queue, int needed);
 309
 310 void xenvif_carrier_on(struct xenvif *vif);
 311
 312 /* Callback from stack when TX packet can be released */
 313 void xenvif_zerocopy_callback(struct ubuf_info *ubuf, bool zerocopy_success);
 314
 315 /* Unmap a pending page and release it back to the guest */
 316 void xenvif_idx_unmap(struct xenvif_queue *queue, u16 pending_idx);
 317
 318 static inline pending_ring_idx_t nr_pending_reqs(struct xenvif_queue *queue)
 319 {
 320         return MAX_PENDING_REQS -
 321                 queue->pending_prod + queue->pending_cons;
 322 }
 323
 324 /* Callback from stack when TX packet can be released */
 325 void xenvif_zerocopy_callback(struct ubuf_info *ubuf, bool zerocopy_success);
 326
 327 irqreturn_t xenvif_interrupt(int irq, void *dev_id);
 328
 329 extern bool separate_tx_rx_irq;
 330
 331 extern unsigned int rx_drain_timeout_msecs;
 332 extern unsigned int rx_stall_timeout_msecs;
 333 extern unsigned int xenvif_max_queues;
 334
 335 #ifdef CONFIG_DEBUG_FS
 336 extern struct dentry *xen_netback_dbg_root;
 337 #endif
 338
 339 void xenvif_skb_zerocopy_prepare(struct xenvif_queue *queue,
 340                                  struct sk_buff *skb);
 341 void xenvif_skb_zerocopy_complete(struct xenvif_queue *queue);
 342
 343 #endif /* __XEN_NETBACK__COMMON_H__ */