xref: /openbmc/linux/drivers/hv/hv.c (revision aac5987a)
1 /*
2  * Copyright (c) 2009, Microsoft Corporation.
3  *
4  * This program is free software; you can redistribute it and/or modify it
5  * under the terms and conditions of the GNU General Public License,
6  * version 2, as published by the Free Software Foundation.
7  *
8  * This program is distributed in the hope it will be useful, but WITHOUT
9  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
10  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
11  * more details.
12  *
13  * You should have received a copy of the GNU General Public License along with
14  * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
15  * Place - Suite 330, Boston, MA 02111-1307 USA.
16  *
17  * Authors:
18  *   Haiyang Zhang <haiyangz@microsoft.com>
19  *   Hank Janssen  <hjanssen@microsoft.com>
20  *
21  */
22 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
23 
24 #include <linux/kernel.h>
25 #include <linux/mm.h>
26 #include <linux/slab.h>
27 #include <linux/vmalloc.h>
28 #include <linux/hyperv.h>
29 #include <linux/version.h>
30 #include <linux/interrupt.h>
31 #include <linux/clockchips.h>
32 #include <asm/hyperv.h>
33 #include <asm/mshyperv.h>
34 #include "hyperv_vmbus.h"
35 
36 /* The one and only */
37 struct hv_context hv_context = {
38 	.synic_initialized	= false,
39 };
40 
41 #define HV_TIMER_FREQUENCY (10 * 1000 * 1000) /* 100ns period */
42 #define HV_MAX_MAX_DELTA_TICKS 0xffffffff
43 #define HV_MIN_DELTA_TICKS 1
44 
45 /*
46  * hv_init - Main initialization routine.
47  *
48  * This routine must be called before any other routines in here are called
49  */
50 int hv_init(void)
51 {
52 	if (!hv_is_hypercall_page_setup())
53 		return -ENOTSUPP;
54 
55 	hv_context.cpu_context = alloc_percpu(struct hv_per_cpu_context);
56 	if (!hv_context.cpu_context)
57 		return -ENOMEM;
58 
59 	return 0;
60 }
61 
62 /*
63  * hv_post_message - Post a message using the hypervisor message IPC.
64  *
65  * This involves a hypercall.
66  */
67 int hv_post_message(union hv_connection_id connection_id,
68 		  enum hv_message_type message_type,
69 		  void *payload, size_t payload_size)
70 {
71 	struct hv_input_post_message *aligned_msg;
72 	struct hv_per_cpu_context *hv_cpu;
73 	u64 status;
74 
75 	if (payload_size > HV_MESSAGE_PAYLOAD_BYTE_COUNT)
76 		return -EMSGSIZE;
77 
78 	hv_cpu = get_cpu_ptr(hv_context.cpu_context);
79 	aligned_msg = hv_cpu->post_msg_page;
80 	aligned_msg->connectionid = connection_id;
81 	aligned_msg->reserved = 0;
82 	aligned_msg->message_type = message_type;
83 	aligned_msg->payload_size = payload_size;
84 	memcpy((void *)aligned_msg->payload, payload, payload_size);
85 	put_cpu_ptr(hv_cpu);
86 
87 	status = hv_do_hypercall(HVCALL_POST_MESSAGE, aligned_msg, NULL);
88 
89 	return status & 0xFFFF;
90 }
91 
92 static int hv_ce_set_next_event(unsigned long delta,
93 				struct clock_event_device *evt)
94 {
95 	u64 current_tick;
96 
97 	WARN_ON(!clockevent_state_oneshot(evt));
98 
99 	hv_get_current_tick(current_tick);
100 	current_tick += delta;
101 	hv_init_timer(HV_X64_MSR_STIMER0_COUNT, current_tick);
102 	return 0;
103 }
104 
105 static int hv_ce_shutdown(struct clock_event_device *evt)
106 {
107 	hv_init_timer(HV_X64_MSR_STIMER0_COUNT, 0);
108 	hv_init_timer_config(HV_X64_MSR_STIMER0_CONFIG, 0);
109 
110 	return 0;
111 }
112 
113 static int hv_ce_set_oneshot(struct clock_event_device *evt)
114 {
115 	union hv_timer_config timer_cfg;
116 
117 	timer_cfg.enable = 1;
118 	timer_cfg.auto_enable = 1;
119 	timer_cfg.sintx = VMBUS_MESSAGE_SINT;
120 	hv_init_timer_config(HV_X64_MSR_STIMER0_CONFIG, timer_cfg.as_uint64);
121 
122 	return 0;
123 }
124 
125 static void hv_init_clockevent_device(struct clock_event_device *dev, int cpu)
126 {
127 	dev->name = "Hyper-V clockevent";
128 	dev->features = CLOCK_EVT_FEAT_ONESHOT;
129 	dev->cpumask = cpumask_of(cpu);
130 	dev->rating = 1000;
131 	/*
132 	 * Avoid settint dev->owner = THIS_MODULE deliberately as doing so will
133 	 * result in clockevents_config_and_register() taking additional
134 	 * references to the hv_vmbus module making it impossible to unload.
135 	 */
136 
137 	dev->set_state_shutdown = hv_ce_shutdown;
138 	dev->set_state_oneshot = hv_ce_set_oneshot;
139 	dev->set_next_event = hv_ce_set_next_event;
140 }
141 
142 
143 int hv_synic_alloc(void)
144 {
145 	int cpu;
146 
147 	hv_context.hv_numa_map = kzalloc(sizeof(struct cpumask) * nr_node_ids,
148 					 GFP_ATOMIC);
149 	if (hv_context.hv_numa_map == NULL) {
150 		pr_err("Unable to allocate NUMA map\n");
151 		goto err;
152 	}
153 
154 	for_each_present_cpu(cpu) {
155 		struct hv_per_cpu_context *hv_cpu
156 			= per_cpu_ptr(hv_context.cpu_context, cpu);
157 
158 		memset(hv_cpu, 0, sizeof(*hv_cpu));
159 		tasklet_init(&hv_cpu->msg_dpc,
160 			     vmbus_on_msg_dpc, (unsigned long) hv_cpu);
161 
162 		hv_cpu->clk_evt = kzalloc(sizeof(struct clock_event_device),
163 					  GFP_KERNEL);
164 		if (hv_cpu->clk_evt == NULL) {
165 			pr_err("Unable to allocate clock event device\n");
166 			goto err;
167 		}
168 		hv_init_clockevent_device(hv_cpu->clk_evt, cpu);
169 
170 		hv_cpu->synic_message_page =
171 			(void *)get_zeroed_page(GFP_ATOMIC);
172 		if (hv_cpu->synic_message_page == NULL) {
173 			pr_err("Unable to allocate SYNIC message page\n");
174 			goto err;
175 		}
176 
177 		hv_cpu->synic_event_page = (void *)get_zeroed_page(GFP_ATOMIC);
178 		if (hv_cpu->synic_event_page == NULL) {
179 			pr_err("Unable to allocate SYNIC event page\n");
180 			goto err;
181 		}
182 
183 		hv_cpu->post_msg_page = (void *)get_zeroed_page(GFP_ATOMIC);
184 		if (hv_cpu->post_msg_page == NULL) {
185 			pr_err("Unable to allocate post msg page\n");
186 			goto err;
187 		}
188 
189 		INIT_LIST_HEAD(&hv_cpu->chan_list);
190 	}
191 
192 	return 0;
193 err:
194 	return -ENOMEM;
195 }
196 
197 
198 void hv_synic_free(void)
199 {
200 	int cpu;
201 
202 	for_each_present_cpu(cpu) {
203 		struct hv_per_cpu_context *hv_cpu
204 			= per_cpu_ptr(hv_context.cpu_context, cpu);
205 
206 		if (hv_cpu->synic_event_page)
207 			free_page((unsigned long)hv_cpu->synic_event_page);
208 		if (hv_cpu->synic_message_page)
209 			free_page((unsigned long)hv_cpu->synic_message_page);
210 		if (hv_cpu->post_msg_page)
211 			free_page((unsigned long)hv_cpu->post_msg_page);
212 	}
213 
214 	kfree(hv_context.hv_numa_map);
215 }
216 
217 /*
218  * hv_synic_init - Initialize the Synthethic Interrupt Controller.
219  *
220  * If it is already initialized by another entity (ie x2v shim), we need to
221  * retrieve the initialized message and event pages.  Otherwise, we create and
222  * initialize the message and event pages.
223  */
224 int hv_synic_init(unsigned int cpu)
225 {
226 	struct hv_per_cpu_context *hv_cpu
227 		= per_cpu_ptr(hv_context.cpu_context, cpu);
228 	union hv_synic_simp simp;
229 	union hv_synic_siefp siefp;
230 	union hv_synic_sint shared_sint;
231 	union hv_synic_scontrol sctrl;
232 	u64 vp_index;
233 
234 	/* Setup the Synic's message page */
235 	hv_get_simp(simp.as_uint64);
236 	simp.simp_enabled = 1;
237 	simp.base_simp_gpa = virt_to_phys(hv_cpu->synic_message_page)
238 		>> PAGE_SHIFT;
239 
240 	hv_set_simp(simp.as_uint64);
241 
242 	/* Setup the Synic's event page */
243 	hv_get_siefp(siefp.as_uint64);
244 	siefp.siefp_enabled = 1;
245 	siefp.base_siefp_gpa = virt_to_phys(hv_cpu->synic_event_page)
246 		>> PAGE_SHIFT;
247 
248 	hv_set_siefp(siefp.as_uint64);
249 
250 	/* Setup the shared SINT. */
251 	hv_get_synint_state(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT,
252 			    shared_sint.as_uint64);
253 
254 	shared_sint.as_uint64 = 0;
255 	shared_sint.vector = HYPERVISOR_CALLBACK_VECTOR;
256 	shared_sint.masked = false;
257 	shared_sint.auto_eoi = true;
258 
259 	hv_set_synint_state(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT,
260 			    shared_sint.as_uint64);
261 
262 	/* Enable the global synic bit */
263 	hv_get_synic_state(sctrl.as_uint64);
264 	sctrl.enable = 1;
265 
266 	hv_set_synic_state(sctrl.as_uint64);
267 
268 	hv_context.synic_initialized = true;
269 
270 	/*
271 	 * Setup the mapping between Hyper-V's notion
272 	 * of cpuid and Linux' notion of cpuid.
273 	 * This array will be indexed using Linux cpuid.
274 	 */
275 	hv_get_vp_index(vp_index);
276 	hv_context.vp_index[cpu] = (u32)vp_index;
277 
278 	/*
279 	 * Register the per-cpu clockevent source.
280 	 */
281 	if (ms_hyperv.features & HV_X64_MSR_SYNTIMER_AVAILABLE)
282 		clockevents_config_and_register(hv_cpu->clk_evt,
283 						HV_TIMER_FREQUENCY,
284 						HV_MIN_DELTA_TICKS,
285 						HV_MAX_MAX_DELTA_TICKS);
286 	return 0;
287 }
288 
289 /*
290  * hv_synic_clockevents_cleanup - Cleanup clockevent devices
291  */
292 void hv_synic_clockevents_cleanup(void)
293 {
294 	int cpu;
295 
296 	if (!(ms_hyperv.features & HV_X64_MSR_SYNTIMER_AVAILABLE))
297 		return;
298 
299 	for_each_present_cpu(cpu) {
300 		struct hv_per_cpu_context *hv_cpu
301 			= per_cpu_ptr(hv_context.cpu_context, cpu);
302 
303 		clockevents_unbind_device(hv_cpu->clk_evt, cpu);
304 	}
305 }
306 
307 /*
308  * hv_synic_cleanup - Cleanup routine for hv_synic_init().
309  */
310 int hv_synic_cleanup(unsigned int cpu)
311 {
312 	union hv_synic_sint shared_sint;
313 	union hv_synic_simp simp;
314 	union hv_synic_siefp siefp;
315 	union hv_synic_scontrol sctrl;
316 	struct vmbus_channel *channel, *sc;
317 	bool channel_found = false;
318 	unsigned long flags;
319 
320 	if (!hv_context.synic_initialized)
321 		return -EFAULT;
322 
323 	/*
324 	 * Search for channels which are bound to the CPU we're about to
325 	 * cleanup. In case we find one and vmbus is still connected we need to
326 	 * fail, this will effectively prevent CPU offlining. There is no way
327 	 * we can re-bind channels to different CPUs for now.
328 	 */
329 	mutex_lock(&vmbus_connection.channel_mutex);
330 	list_for_each_entry(channel, &vmbus_connection.chn_list, listentry) {
331 		if (channel->target_cpu == cpu) {
332 			channel_found = true;
333 			break;
334 		}
335 		spin_lock_irqsave(&channel->lock, flags);
336 		list_for_each_entry(sc, &channel->sc_list, sc_list) {
337 			if (sc->target_cpu == cpu) {
338 				channel_found = true;
339 				break;
340 			}
341 		}
342 		spin_unlock_irqrestore(&channel->lock, flags);
343 		if (channel_found)
344 			break;
345 	}
346 	mutex_unlock(&vmbus_connection.channel_mutex);
347 
348 	if (channel_found && vmbus_connection.conn_state == CONNECTED)
349 		return -EBUSY;
350 
351 	/* Turn off clockevent device */
352 	if (ms_hyperv.features & HV_X64_MSR_SYNTIMER_AVAILABLE) {
353 		struct hv_per_cpu_context *hv_cpu
354 			= this_cpu_ptr(hv_context.cpu_context);
355 
356 		clockevents_unbind_device(hv_cpu->clk_evt, cpu);
357 		hv_ce_shutdown(hv_cpu->clk_evt);
358 		put_cpu_ptr(hv_cpu);
359 	}
360 
361 	hv_get_synint_state(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT,
362 			    shared_sint.as_uint64);
363 
364 	shared_sint.masked = 1;
365 
366 	/* Need to correctly cleanup in the case of SMP!!! */
367 	/* Disable the interrupt */
368 	hv_set_synint_state(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT,
369 			    shared_sint.as_uint64);
370 
371 	hv_get_simp(simp.as_uint64);
372 	simp.simp_enabled = 0;
373 	simp.base_simp_gpa = 0;
374 
375 	hv_set_simp(simp.as_uint64);
376 
377 	hv_get_siefp(siefp.as_uint64);
378 	siefp.siefp_enabled = 0;
379 	siefp.base_siefp_gpa = 0;
380 
381 	hv_set_siefp(siefp.as_uint64);
382 
383 	/* Disable the global synic bit */
384 	hv_get_synic_state(sctrl.as_uint64);
385 	sctrl.enable = 0;
386 	hv_set_synic_state(sctrl.as_uint64);
387 
388 	return 0;
389 }
390