xref: /openbmc/linux/drivers/net/xen-netback/common.h (revision d5e7cafd)
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__ */
344