xref: /openbmc/linux/net/9p/trans_xen.c (revision 965f22bc)
1 /*
2  * linux/fs/9p/trans_xen
3  *
4  * Xen transport layer.
5  *
6  * Copyright (C) 2017 by Stefano Stabellini <stefano@aporeto.com>
7  *
8  * This program is free software; you can redistribute it and/or
9  * modify it under the terms of the GNU General Public License version 2
10  * as published by the Free Software Foundation; or, when distributed
11  * separately from the Linux kernel or incorporated into other
12  * software packages, subject to the following license:
13  *
14  * Permission is hereby granted, free of charge, to any person obtaining a copy
15  * of this source file (the "Software"), to deal in the Software without
16  * restriction, including without limitation the rights to use, copy, modify,
17  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
18  * and to permit persons to whom the Software is furnished to do so, subject to
19  * the following conditions:
20  *
21  * The above copyright notice and this permission notice shall be included in
22  * all copies or substantial portions of the Software.
23  *
24  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
25  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
26  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
27  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
28  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
29  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
30  * IN THE SOFTWARE.
31  */
32 
33 #include <xen/events.h>
34 #include <xen/grant_table.h>
35 #include <xen/xen.h>
36 #include <xen/xenbus.h>
37 #include <xen/interface/io/9pfs.h>
38 
39 #include <linux/module.h>
40 #include <linux/spinlock.h>
41 #include <net/9p/9p.h>
42 #include <net/9p/client.h>
43 #include <net/9p/transport.h>
44 
45 #define XEN_9PFS_NUM_RINGS 2
46 #define XEN_9PFS_RING_ORDER 6
47 #define XEN_9PFS_RING_SIZE  XEN_FLEX_RING_SIZE(XEN_9PFS_RING_ORDER)
48 
49 struct xen_9pfs_header {
50 	uint32_t size;
51 	uint8_t id;
52 	uint16_t tag;
53 
54 	/* uint8_t sdata[]; */
55 } __attribute__((packed));
56 
57 /* One per ring, more than one per 9pfs share */
58 struct xen_9pfs_dataring {
59 	struct xen_9pfs_front_priv *priv;
60 
61 	struct xen_9pfs_data_intf *intf;
62 	grant_ref_t ref;
63 	int evtchn;
64 	int irq;
65 	/* protect a ring from concurrent accesses */
66 	spinlock_t lock;
67 
68 	struct xen_9pfs_data data;
69 	wait_queue_head_t wq;
70 	struct work_struct work;
71 };
72 
73 /* One per 9pfs share */
74 struct xen_9pfs_front_priv {
75 	struct list_head list;
76 	struct xenbus_device *dev;
77 	char *tag;
78 	struct p9_client *client;
79 
80 	int num_rings;
81 	struct xen_9pfs_dataring *rings;
82 };
83 
84 static LIST_HEAD(xen_9pfs_devs);
85 static DEFINE_RWLOCK(xen_9pfs_lock);
86 
87 /* We don't currently allow canceling of requests */
88 static int p9_xen_cancel(struct p9_client *client, struct p9_req_t *req)
89 {
90 	return 1;
91 }
92 
93 static int p9_xen_create(struct p9_client *client, const char *addr, char *args)
94 {
95 	struct xen_9pfs_front_priv *priv;
96 
97 	if (addr == NULL)
98 		return -EINVAL;
99 
100 	read_lock(&xen_9pfs_lock);
101 	list_for_each_entry(priv, &xen_9pfs_devs, list) {
102 		if (!strcmp(priv->tag, addr)) {
103 			priv->client = client;
104 			read_unlock(&xen_9pfs_lock);
105 			return 0;
106 		}
107 	}
108 	read_unlock(&xen_9pfs_lock);
109 	return -EINVAL;
110 }
111 
112 static void p9_xen_close(struct p9_client *client)
113 {
114 	struct xen_9pfs_front_priv *priv;
115 
116 	read_lock(&xen_9pfs_lock);
117 	list_for_each_entry(priv, &xen_9pfs_devs, list) {
118 		if (priv->client == client) {
119 			priv->client = NULL;
120 			read_unlock(&xen_9pfs_lock);
121 			return;
122 		}
123 	}
124 	read_unlock(&xen_9pfs_lock);
125 }
126 
127 static bool p9_xen_write_todo(struct xen_9pfs_dataring *ring, RING_IDX size)
128 {
129 	RING_IDX cons, prod;
130 
131 	cons = ring->intf->out_cons;
132 	prod = ring->intf->out_prod;
133 	virt_mb();
134 
135 	return XEN_9PFS_RING_SIZE -
136 		xen_9pfs_queued(prod, cons, XEN_9PFS_RING_SIZE) >= size;
137 }
138 
139 static int p9_xen_request(struct p9_client *client, struct p9_req_t *p9_req)
140 {
141 	struct xen_9pfs_front_priv *priv = NULL;
142 	RING_IDX cons, prod, masked_cons, masked_prod;
143 	unsigned long flags;
144 	u32 size = p9_req->tc->size;
145 	struct xen_9pfs_dataring *ring;
146 	int num;
147 
148 	read_lock(&xen_9pfs_lock);
149 	list_for_each_entry(priv, &xen_9pfs_devs, list) {
150 		if (priv->client == client)
151 			break;
152 	}
153 	read_unlock(&xen_9pfs_lock);
154 	if (!priv || priv->client != client)
155 		return -EINVAL;
156 
157 	num = p9_req->tc->tag % priv->num_rings;
158 	ring = &priv->rings[num];
159 
160 again:
161 	while (wait_event_killable(ring->wq,
162 				   p9_xen_write_todo(ring, size)) != 0)
163 		;
164 
165 	spin_lock_irqsave(&ring->lock, flags);
166 	cons = ring->intf->out_cons;
167 	prod = ring->intf->out_prod;
168 	virt_mb();
169 
170 	if (XEN_9PFS_RING_SIZE - xen_9pfs_queued(prod, cons,
171 						 XEN_9PFS_RING_SIZE) < size) {
172 		spin_unlock_irqrestore(&ring->lock, flags);
173 		goto again;
174 	}
175 
176 	masked_prod = xen_9pfs_mask(prod, XEN_9PFS_RING_SIZE);
177 	masked_cons = xen_9pfs_mask(cons, XEN_9PFS_RING_SIZE);
178 
179 	xen_9pfs_write_packet(ring->data.out, p9_req->tc->sdata, size,
180 			      &masked_prod, masked_cons, XEN_9PFS_RING_SIZE);
181 
182 	p9_req->status = REQ_STATUS_SENT;
183 	virt_wmb();			/* write ring before updating pointer */
184 	prod += size;
185 	ring->intf->out_prod = prod;
186 	spin_unlock_irqrestore(&ring->lock, flags);
187 	notify_remote_via_irq(ring->irq);
188 
189 	return 0;
190 }
191 
192 static void p9_xen_response(struct work_struct *work)
193 {
194 	struct xen_9pfs_front_priv *priv;
195 	struct xen_9pfs_dataring *ring;
196 	RING_IDX cons, prod, masked_cons, masked_prod;
197 	struct xen_9pfs_header h;
198 	struct p9_req_t *req;
199 	int status;
200 
201 	ring = container_of(work, struct xen_9pfs_dataring, work);
202 	priv = ring->priv;
203 
204 	while (1) {
205 		cons = ring->intf->in_cons;
206 		prod = ring->intf->in_prod;
207 		virt_rmb();
208 
209 		if (xen_9pfs_queued(prod, cons, XEN_9PFS_RING_SIZE) <
210 		    sizeof(h)) {
211 			notify_remote_via_irq(ring->irq);
212 			return;
213 		}
214 
215 		masked_prod = xen_9pfs_mask(prod, XEN_9PFS_RING_SIZE);
216 		masked_cons = xen_9pfs_mask(cons, XEN_9PFS_RING_SIZE);
217 
218 		/* First, read just the header */
219 		xen_9pfs_read_packet(&h, ring->data.in, sizeof(h),
220 				     masked_prod, &masked_cons,
221 				     XEN_9PFS_RING_SIZE);
222 
223 		req = p9_tag_lookup(priv->client, h.tag);
224 		if (!req || req->status != REQ_STATUS_SENT) {
225 			dev_warn(&priv->dev->dev, "Wrong req tag=%x\n", h.tag);
226 			cons += h.size;
227 			virt_mb();
228 			ring->intf->in_cons = cons;
229 			continue;
230 		}
231 
232 		memcpy(req->rc, &h, sizeof(h));
233 		req->rc->offset = 0;
234 
235 		masked_cons = xen_9pfs_mask(cons, XEN_9PFS_RING_SIZE);
236 		/* Then, read the whole packet (including the header) */
237 		xen_9pfs_read_packet(req->rc->sdata, ring->data.in, h.size,
238 				     masked_prod, &masked_cons,
239 				     XEN_9PFS_RING_SIZE);
240 
241 		virt_mb();
242 		cons += h.size;
243 		ring->intf->in_cons = cons;
244 
245 		status = (req->status != REQ_STATUS_ERROR) ?
246 			REQ_STATUS_RCVD : REQ_STATUS_ERROR;
247 
248 		p9_client_cb(priv->client, req, status);
249 	}
250 }
251 
252 static irqreturn_t xen_9pfs_front_event_handler(int irq, void *r)
253 {
254 	struct xen_9pfs_dataring *ring = r;
255 
256 	if (!ring || !ring->priv->client) {
257 		/* ignore spurious interrupt */
258 		return IRQ_HANDLED;
259 	}
260 
261 	wake_up_interruptible(&ring->wq);
262 	schedule_work(&ring->work);
263 
264 	return IRQ_HANDLED;
265 }
266 
267 static struct p9_trans_module p9_xen_trans = {
268 	.name = "xen",
269 	.maxsize = 1 << (XEN_9PFS_RING_ORDER + XEN_PAGE_SHIFT),
270 	.def = 1,
271 	.create = p9_xen_create,
272 	.close = p9_xen_close,
273 	.request = p9_xen_request,
274 	.cancel = p9_xen_cancel,
275 	.owner = THIS_MODULE,
276 };
277 
278 static const struct xenbus_device_id xen_9pfs_front_ids[] = {
279 	{ "9pfs" },
280 	{ "" }
281 };
282 
283 static void xen_9pfs_front_free(struct xen_9pfs_front_priv *priv)
284 {
285 	int i, j;
286 
287 	write_lock(&xen_9pfs_lock);
288 	list_del(&priv->list);
289 	write_unlock(&xen_9pfs_lock);
290 
291 	for (i = 0; i < priv->num_rings; i++) {
292 		if (!priv->rings[i].intf)
293 			break;
294 		if (priv->rings[i].irq > 0)
295 			unbind_from_irqhandler(priv->rings[i].irq, priv->dev);
296 		if (priv->rings[i].data.in) {
297 			for (j = 0; j < (1 << XEN_9PFS_RING_ORDER); j++) {
298 				grant_ref_t ref;
299 
300 				ref = priv->rings[i].intf->ref[j];
301 				gnttab_end_foreign_access(ref, 0, 0);
302 			}
303 			free_pages((unsigned long)priv->rings[i].data.in,
304 				   XEN_9PFS_RING_ORDER -
305 				   (PAGE_SHIFT - XEN_PAGE_SHIFT));
306 		}
307 		gnttab_end_foreign_access(priv->rings[i].ref, 0, 0);
308 		free_page((unsigned long)priv->rings[i].intf);
309 	}
310 	kfree(priv->rings);
311 	kfree(priv->tag);
312 	kfree(priv);
313 }
314 
315 static int xen_9pfs_front_remove(struct xenbus_device *dev)
316 {
317 	struct xen_9pfs_front_priv *priv = dev_get_drvdata(&dev->dev);
318 
319 	dev_set_drvdata(&dev->dev, NULL);
320 	xen_9pfs_front_free(priv);
321 	return 0;
322 }
323 
324 static int xen_9pfs_front_alloc_dataring(struct xenbus_device *dev,
325 					 struct xen_9pfs_dataring *ring)
326 {
327 	int i = 0;
328 	int ret = -ENOMEM;
329 	void *bytes = NULL;
330 
331 	init_waitqueue_head(&ring->wq);
332 	spin_lock_init(&ring->lock);
333 	INIT_WORK(&ring->work, p9_xen_response);
334 
335 	ring->intf = (struct xen_9pfs_data_intf *)get_zeroed_page(GFP_KERNEL);
336 	if (!ring->intf)
337 		return ret;
338 	ret = gnttab_grant_foreign_access(dev->otherend_id,
339 					  virt_to_gfn(ring->intf), 0);
340 	if (ret < 0)
341 		goto out;
342 	ring->ref = ret;
343 	bytes = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
344 			XEN_9PFS_RING_ORDER - (PAGE_SHIFT - XEN_PAGE_SHIFT));
345 	if (!bytes) {
346 		ret = -ENOMEM;
347 		goto out;
348 	}
349 	for (; i < (1 << XEN_9PFS_RING_ORDER); i++) {
350 		ret = gnttab_grant_foreign_access(
351 				dev->otherend_id, virt_to_gfn(bytes) + i, 0);
352 		if (ret < 0)
353 			goto out;
354 		ring->intf->ref[i] = ret;
355 	}
356 	ring->intf->ring_order = XEN_9PFS_RING_ORDER;
357 	ring->data.in = bytes;
358 	ring->data.out = bytes + XEN_9PFS_RING_SIZE;
359 
360 	ret = xenbus_alloc_evtchn(dev, &ring->evtchn);
361 	if (ret)
362 		goto out;
363 	ring->irq = bind_evtchn_to_irqhandler(ring->evtchn,
364 					      xen_9pfs_front_event_handler,
365 					      0, "xen_9pfs-frontend", ring);
366 	if (ring->irq >= 0)
367 		return 0;
368 
369 	xenbus_free_evtchn(dev, ring->evtchn);
370 	ret = ring->irq;
371 out:
372 	if (bytes) {
373 		for (i--; i >= 0; i--)
374 			gnttab_end_foreign_access(ring->intf->ref[i], 0, 0);
375 		free_pages((unsigned long)bytes,
376 			   XEN_9PFS_RING_ORDER -
377 			   (PAGE_SHIFT - XEN_PAGE_SHIFT));
378 	}
379 	gnttab_end_foreign_access(ring->ref, 0, 0);
380 	free_page((unsigned long)ring->intf);
381 	return ret;
382 }
383 
384 static int xen_9pfs_front_probe(struct xenbus_device *dev,
385 				const struct xenbus_device_id *id)
386 {
387 	int ret, i;
388 	struct xenbus_transaction xbt;
389 	struct xen_9pfs_front_priv *priv = NULL;
390 	char *versions;
391 	unsigned int max_rings, max_ring_order, len = 0;
392 
393 	versions = xenbus_read(XBT_NIL, dev->otherend, "versions", &len);
394 	if (!len)
395 		return -EINVAL;
396 	if (strcmp(versions, "1")) {
397 		kfree(versions);
398 		return -EINVAL;
399 	}
400 	kfree(versions);
401 	max_rings = xenbus_read_unsigned(dev->otherend, "max-rings", 0);
402 	if (max_rings < XEN_9PFS_NUM_RINGS)
403 		return -EINVAL;
404 	max_ring_order = xenbus_read_unsigned(dev->otherend,
405 					      "max-ring-page-order", 0);
406 	if (max_ring_order < XEN_9PFS_RING_ORDER)
407 		return -EINVAL;
408 
409 	priv = kzalloc(sizeof(*priv), GFP_KERNEL);
410 	if (!priv)
411 		return -ENOMEM;
412 
413 	priv->dev = dev;
414 	priv->num_rings = XEN_9PFS_NUM_RINGS;
415 	priv->rings = kcalloc(priv->num_rings, sizeof(*priv->rings),
416 			      GFP_KERNEL);
417 	if (!priv->rings) {
418 		kfree(priv);
419 		return -ENOMEM;
420 	}
421 
422 	for (i = 0; i < priv->num_rings; i++) {
423 		priv->rings[i].priv = priv;
424 		ret = xen_9pfs_front_alloc_dataring(dev, &priv->rings[i]);
425 		if (ret < 0)
426 			goto error;
427 	}
428 
429  again:
430 	ret = xenbus_transaction_start(&xbt);
431 	if (ret) {
432 		xenbus_dev_fatal(dev, ret, "starting transaction");
433 		goto error;
434 	}
435 	ret = xenbus_printf(xbt, dev->nodename, "version", "%u", 1);
436 	if (ret)
437 		goto error_xenbus;
438 	ret = xenbus_printf(xbt, dev->nodename, "num-rings", "%u",
439 			    priv->num_rings);
440 	if (ret)
441 		goto error_xenbus;
442 	for (i = 0; i < priv->num_rings; i++) {
443 		char str[16];
444 
445 		BUILD_BUG_ON(XEN_9PFS_NUM_RINGS > 9);
446 		sprintf(str, "ring-ref%u", i);
447 		ret = xenbus_printf(xbt, dev->nodename, str, "%d",
448 				    priv->rings[i].ref);
449 		if (ret)
450 			goto error_xenbus;
451 
452 		sprintf(str, "event-channel-%u", i);
453 		ret = xenbus_printf(xbt, dev->nodename, str, "%u",
454 				    priv->rings[i].evtchn);
455 		if (ret)
456 			goto error_xenbus;
457 	}
458 	priv->tag = xenbus_read(xbt, dev->nodename, "tag", NULL);
459 	if (IS_ERR(priv->tag)) {
460 		ret = PTR_ERR(priv->tag);
461 		goto error_xenbus;
462 	}
463 	ret = xenbus_transaction_end(xbt, 0);
464 	if (ret) {
465 		if (ret == -EAGAIN)
466 			goto again;
467 		xenbus_dev_fatal(dev, ret, "completing transaction");
468 		goto error;
469 	}
470 
471 	write_lock(&xen_9pfs_lock);
472 	list_add_tail(&priv->list, &xen_9pfs_devs);
473 	write_unlock(&xen_9pfs_lock);
474 	dev_set_drvdata(&dev->dev, priv);
475 	xenbus_switch_state(dev, XenbusStateInitialised);
476 
477 	return 0;
478 
479  error_xenbus:
480 	xenbus_transaction_end(xbt, 1);
481 	xenbus_dev_fatal(dev, ret, "writing xenstore");
482  error:
483 	dev_set_drvdata(&dev->dev, NULL);
484 	xen_9pfs_front_free(priv);
485 	return ret;
486 }
487 
488 static int xen_9pfs_front_resume(struct xenbus_device *dev)
489 {
490 	dev_warn(&dev->dev, "suspend/resume unsupported\n");
491 	return 0;
492 }
493 
494 static void xen_9pfs_front_changed(struct xenbus_device *dev,
495 				   enum xenbus_state backend_state)
496 {
497 	switch (backend_state) {
498 	case XenbusStateReconfiguring:
499 	case XenbusStateReconfigured:
500 	case XenbusStateInitialising:
501 	case XenbusStateInitialised:
502 	case XenbusStateUnknown:
503 		break;
504 
505 	case XenbusStateInitWait:
506 		break;
507 
508 	case XenbusStateConnected:
509 		xenbus_switch_state(dev, XenbusStateConnected);
510 		break;
511 
512 	case XenbusStateClosed:
513 		if (dev->state == XenbusStateClosed)
514 			break;
515 		/* Missed the backend's CLOSING state -- fallthrough */
516 	case XenbusStateClosing:
517 		xenbus_frontend_closed(dev);
518 		break;
519 	}
520 }
521 
522 static struct xenbus_driver xen_9pfs_front_driver = {
523 	.ids = xen_9pfs_front_ids,
524 	.probe = xen_9pfs_front_probe,
525 	.remove = xen_9pfs_front_remove,
526 	.resume = xen_9pfs_front_resume,
527 	.otherend_changed = xen_9pfs_front_changed,
528 };
529 
530 static int p9_trans_xen_init(void)
531 {
532 	if (!xen_domain())
533 		return -ENODEV;
534 
535 	pr_info("Initialising Xen transport for 9pfs\n");
536 
537 	v9fs_register_trans(&p9_xen_trans);
538 	return xenbus_register_frontend(&xen_9pfs_front_driver);
539 }
540 module_init(p9_trans_xen_init);
541 
542 static void p9_trans_xen_exit(void)
543 {
544 	v9fs_unregister_trans(&p9_xen_trans);
545 	return xenbus_unregister_driver(&xen_9pfs_front_driver);
546 }
547 module_exit(p9_trans_xen_exit);
548 
549 MODULE_AUTHOR("Stefano Stabellini <stefano@aporeto.com>");
550 MODULE_DESCRIPTION("Xen Transport for 9P");
551 MODULE_LICENSE("GPL");
552