xref: /openbmc/linux/net/9p/trans_xen.c (revision 78beef62)
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 	p9_req_put(p9_req);
189 
190 	return 0;
191 }
192 
193 static void p9_xen_response(struct work_struct *work)
194 {
195 	struct xen_9pfs_front_priv *priv;
196 	struct xen_9pfs_dataring *ring;
197 	RING_IDX cons, prod, masked_cons, masked_prod;
198 	struct xen_9pfs_header h;
199 	struct p9_req_t *req;
200 	int status;
201 
202 	ring = container_of(work, struct xen_9pfs_dataring, work);
203 	priv = ring->priv;
204 
205 	while (1) {
206 		cons = ring->intf->in_cons;
207 		prod = ring->intf->in_prod;
208 		virt_rmb();
209 
210 		if (xen_9pfs_queued(prod, cons, XEN_9PFS_RING_SIZE) <
211 		    sizeof(h)) {
212 			notify_remote_via_irq(ring->irq);
213 			return;
214 		}
215 
216 		masked_prod = xen_9pfs_mask(prod, XEN_9PFS_RING_SIZE);
217 		masked_cons = xen_9pfs_mask(cons, XEN_9PFS_RING_SIZE);
218 
219 		/* First, read just the header */
220 		xen_9pfs_read_packet(&h, ring->data.in, sizeof(h),
221 				     masked_prod, &masked_cons,
222 				     XEN_9PFS_RING_SIZE);
223 
224 		req = p9_tag_lookup(priv->client, h.tag);
225 		if (!req || req->status != REQ_STATUS_SENT) {
226 			dev_warn(&priv->dev->dev, "Wrong req tag=%x\n", h.tag);
227 			cons += h.size;
228 			virt_mb();
229 			ring->intf->in_cons = cons;
230 			continue;
231 		}
232 
233 		memcpy(&req->rc, &h, sizeof(h));
234 		req->rc.offset = 0;
235 
236 		masked_cons = xen_9pfs_mask(cons, XEN_9PFS_RING_SIZE);
237 		/* Then, read the whole packet (including the header) */
238 		xen_9pfs_read_packet(req->rc.sdata, ring->data.in, h.size,
239 				     masked_prod, &masked_cons,
240 				     XEN_9PFS_RING_SIZE);
241 
242 		virt_mb();
243 		cons += h.size;
244 		ring->intf->in_cons = cons;
245 
246 		status = (req->status != REQ_STATUS_ERROR) ?
247 			REQ_STATUS_RCVD : REQ_STATUS_ERROR;
248 
249 		p9_client_cb(priv->client, req, status);
250 	}
251 }
252 
253 static irqreturn_t xen_9pfs_front_event_handler(int irq, void *r)
254 {
255 	struct xen_9pfs_dataring *ring = r;
256 
257 	if (!ring || !ring->priv->client) {
258 		/* ignore spurious interrupt */
259 		return IRQ_HANDLED;
260 	}
261 
262 	wake_up_interruptible(&ring->wq);
263 	schedule_work(&ring->work);
264 
265 	return IRQ_HANDLED;
266 }
267 
268 static struct p9_trans_module p9_xen_trans = {
269 	.name = "xen",
270 	.maxsize = 1 << (XEN_9PFS_RING_ORDER + XEN_PAGE_SHIFT),
271 	.def = 1,
272 	.create = p9_xen_create,
273 	.close = p9_xen_close,
274 	.request = p9_xen_request,
275 	.cancel = p9_xen_cancel,
276 	.owner = THIS_MODULE,
277 };
278 
279 static const struct xenbus_device_id xen_9pfs_front_ids[] = {
280 	{ "9pfs" },
281 	{ "" }
282 };
283 
284 static void xen_9pfs_front_free(struct xen_9pfs_front_priv *priv)
285 {
286 	int i, j;
287 
288 	write_lock(&xen_9pfs_lock);
289 	list_del(&priv->list);
290 	write_unlock(&xen_9pfs_lock);
291 
292 	for (i = 0; i < priv->num_rings; i++) {
293 		if (!priv->rings[i].intf)
294 			break;
295 		if (priv->rings[i].irq > 0)
296 			unbind_from_irqhandler(priv->rings[i].irq, priv->dev);
297 		if (priv->rings[i].data.in) {
298 			for (j = 0; j < (1 << XEN_9PFS_RING_ORDER); j++) {
299 				grant_ref_t ref;
300 
301 				ref = priv->rings[i].intf->ref[j];
302 				gnttab_end_foreign_access(ref, 0, 0);
303 			}
304 			free_pages((unsigned long)priv->rings[i].data.in,
305 				   XEN_9PFS_RING_ORDER -
306 				   (PAGE_SHIFT - XEN_PAGE_SHIFT));
307 		}
308 		gnttab_end_foreign_access(priv->rings[i].ref, 0, 0);
309 		free_page((unsigned long)priv->rings[i].intf);
310 	}
311 	kfree(priv->rings);
312 	kfree(priv->tag);
313 	kfree(priv);
314 }
315 
316 static int xen_9pfs_front_remove(struct xenbus_device *dev)
317 {
318 	struct xen_9pfs_front_priv *priv = dev_get_drvdata(&dev->dev);
319 
320 	dev_set_drvdata(&dev->dev, NULL);
321 	xen_9pfs_front_free(priv);
322 	return 0;
323 }
324 
325 static int xen_9pfs_front_alloc_dataring(struct xenbus_device *dev,
326 					 struct xen_9pfs_dataring *ring)
327 {
328 	int i = 0;
329 	int ret = -ENOMEM;
330 	void *bytes = NULL;
331 
332 	init_waitqueue_head(&ring->wq);
333 	spin_lock_init(&ring->lock);
334 	INIT_WORK(&ring->work, p9_xen_response);
335 
336 	ring->intf = (struct xen_9pfs_data_intf *)get_zeroed_page(GFP_KERNEL);
337 	if (!ring->intf)
338 		return ret;
339 	ret = gnttab_grant_foreign_access(dev->otherend_id,
340 					  virt_to_gfn(ring->intf), 0);
341 	if (ret < 0)
342 		goto out;
343 	ring->ref = ret;
344 	bytes = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
345 			XEN_9PFS_RING_ORDER - (PAGE_SHIFT - XEN_PAGE_SHIFT));
346 	if (!bytes) {
347 		ret = -ENOMEM;
348 		goto out;
349 	}
350 	for (; i < (1 << XEN_9PFS_RING_ORDER); i++) {
351 		ret = gnttab_grant_foreign_access(
352 				dev->otherend_id, virt_to_gfn(bytes) + i, 0);
353 		if (ret < 0)
354 			goto out;
355 		ring->intf->ref[i] = ret;
356 	}
357 	ring->intf->ring_order = XEN_9PFS_RING_ORDER;
358 	ring->data.in = bytes;
359 	ring->data.out = bytes + XEN_9PFS_RING_SIZE;
360 
361 	ret = xenbus_alloc_evtchn(dev, &ring->evtchn);
362 	if (ret)
363 		goto out;
364 	ring->irq = bind_evtchn_to_irqhandler(ring->evtchn,
365 					      xen_9pfs_front_event_handler,
366 					      0, "xen_9pfs-frontend", ring);
367 	if (ring->irq >= 0)
368 		return 0;
369 
370 	xenbus_free_evtchn(dev, ring->evtchn);
371 	ret = ring->irq;
372 out:
373 	if (bytes) {
374 		for (i--; i >= 0; i--)
375 			gnttab_end_foreign_access(ring->intf->ref[i], 0, 0);
376 		free_pages((unsigned long)bytes,
377 			   XEN_9PFS_RING_ORDER -
378 			   (PAGE_SHIFT - XEN_PAGE_SHIFT));
379 	}
380 	gnttab_end_foreign_access(ring->ref, 0, 0);
381 	free_page((unsigned long)ring->intf);
382 	return ret;
383 }
384 
385 static int xen_9pfs_front_probe(struct xenbus_device *dev,
386 				const struct xenbus_device_id *id)
387 {
388 	int ret, i;
389 	struct xenbus_transaction xbt;
390 	struct xen_9pfs_front_priv *priv = NULL;
391 	char *versions;
392 	unsigned int max_rings, max_ring_order, len = 0;
393 
394 	versions = xenbus_read(XBT_NIL, dev->otherend, "versions", &len);
395 	if (IS_ERR(versions))
396 		return PTR_ERR(versions);
397 	if (strcmp(versions, "1")) {
398 		kfree(versions);
399 		return -EINVAL;
400 	}
401 	kfree(versions);
402 	max_rings = xenbus_read_unsigned(dev->otherend, "max-rings", 0);
403 	if (max_rings < XEN_9PFS_NUM_RINGS)
404 		return -EINVAL;
405 	max_ring_order = xenbus_read_unsigned(dev->otherend,
406 					      "max-ring-page-order", 0);
407 	if (max_ring_order < XEN_9PFS_RING_ORDER)
408 		return -EINVAL;
409 
410 	priv = kzalloc(sizeof(*priv), GFP_KERNEL);
411 	if (!priv)
412 		return -ENOMEM;
413 
414 	priv->dev = dev;
415 	priv->num_rings = XEN_9PFS_NUM_RINGS;
416 	priv->rings = kcalloc(priv->num_rings, sizeof(*priv->rings),
417 			      GFP_KERNEL);
418 	if (!priv->rings) {
419 		kfree(priv);
420 		return -ENOMEM;
421 	}
422 
423 	for (i = 0; i < priv->num_rings; i++) {
424 		priv->rings[i].priv = priv;
425 		ret = xen_9pfs_front_alloc_dataring(dev, &priv->rings[i]);
426 		if (ret < 0)
427 			goto error;
428 	}
429 
430  again:
431 	ret = xenbus_transaction_start(&xbt);
432 	if (ret) {
433 		xenbus_dev_fatal(dev, ret, "starting transaction");
434 		goto error;
435 	}
436 	ret = xenbus_printf(xbt, dev->nodename, "version", "%u", 1);
437 	if (ret)
438 		goto error_xenbus;
439 	ret = xenbus_printf(xbt, dev->nodename, "num-rings", "%u",
440 			    priv->num_rings);
441 	if (ret)
442 		goto error_xenbus;
443 	for (i = 0; i < priv->num_rings; i++) {
444 		char str[16];
445 
446 		BUILD_BUG_ON(XEN_9PFS_NUM_RINGS > 9);
447 		sprintf(str, "ring-ref%u", i);
448 		ret = xenbus_printf(xbt, dev->nodename, str, "%d",
449 				    priv->rings[i].ref);
450 		if (ret)
451 			goto error_xenbus;
452 
453 		sprintf(str, "event-channel-%u", i);
454 		ret = xenbus_printf(xbt, dev->nodename, str, "%u",
455 				    priv->rings[i].evtchn);
456 		if (ret)
457 			goto error_xenbus;
458 	}
459 	priv->tag = xenbus_read(xbt, dev->nodename, "tag", NULL);
460 	if (IS_ERR(priv->tag)) {
461 		ret = PTR_ERR(priv->tag);
462 		goto error_xenbus;
463 	}
464 	ret = xenbus_transaction_end(xbt, 0);
465 	if (ret) {
466 		if (ret == -EAGAIN)
467 			goto again;
468 		xenbus_dev_fatal(dev, ret, "completing transaction");
469 		goto error;
470 	}
471 
472 	write_lock(&xen_9pfs_lock);
473 	list_add_tail(&priv->list, &xen_9pfs_devs);
474 	write_unlock(&xen_9pfs_lock);
475 	dev_set_drvdata(&dev->dev, priv);
476 	xenbus_switch_state(dev, XenbusStateInitialised);
477 
478 	return 0;
479 
480  error_xenbus:
481 	xenbus_transaction_end(xbt, 1);
482 	xenbus_dev_fatal(dev, ret, "writing xenstore");
483  error:
484 	dev_set_drvdata(&dev->dev, NULL);
485 	xen_9pfs_front_free(priv);
486 	return ret;
487 }
488 
489 static int xen_9pfs_front_resume(struct xenbus_device *dev)
490 {
491 	dev_warn(&dev->dev, "suspend/resume unsupported\n");
492 	return 0;
493 }
494 
495 static void xen_9pfs_front_changed(struct xenbus_device *dev,
496 				   enum xenbus_state backend_state)
497 {
498 	switch (backend_state) {
499 	case XenbusStateReconfiguring:
500 	case XenbusStateReconfigured:
501 	case XenbusStateInitialising:
502 	case XenbusStateInitialised:
503 	case XenbusStateUnknown:
504 		break;
505 
506 	case XenbusStateInitWait:
507 		break;
508 
509 	case XenbusStateConnected:
510 		xenbus_switch_state(dev, XenbusStateConnected);
511 		break;
512 
513 	case XenbusStateClosed:
514 		if (dev->state == XenbusStateClosed)
515 			break;
516 		/* fall through - Missed the backend's CLOSING state */
517 	case XenbusStateClosing:
518 		xenbus_frontend_closed(dev);
519 		break;
520 	}
521 }
522 
523 static struct xenbus_driver xen_9pfs_front_driver = {
524 	.ids = xen_9pfs_front_ids,
525 	.probe = xen_9pfs_front_probe,
526 	.remove = xen_9pfs_front_remove,
527 	.resume = xen_9pfs_front_resume,
528 	.otherend_changed = xen_9pfs_front_changed,
529 };
530 
531 static int p9_trans_xen_init(void)
532 {
533 	int rc;
534 
535 	if (!xen_domain())
536 		return -ENODEV;
537 
538 	pr_info("Initialising Xen transport for 9pfs\n");
539 
540 	v9fs_register_trans(&p9_xen_trans);
541 	rc = xenbus_register_frontend(&xen_9pfs_front_driver);
542 	if (rc)
543 		v9fs_unregister_trans(&p9_xen_trans);
544 
545 	return rc;
546 }
547 module_init(p9_trans_xen_init);
548 
549 static void p9_trans_xen_exit(void)
550 {
551 	v9fs_unregister_trans(&p9_xen_trans);
552 	return xenbus_unregister_driver(&xen_9pfs_front_driver);
553 }
554 module_exit(p9_trans_xen_exit);
555 
556 MODULE_AUTHOR("Stefano Stabellini <stefano@aporeto.com>");
557 MODULE_DESCRIPTION("Xen Transport for 9P");
558 MODULE_LICENSE("GPL");
559