hv.c 10.7 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 * Copyright (c) 2009, Microsoft Corporation.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * This program is distributed in the hope it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
 * Place - Suite 330, Boston, MA 02111-1307 USA.
 *
 * Authors:
 *   Haiyang Zhang <haiyangz@microsoft.com>
 *   Hank Janssen  <hjanssen@microsoft.com>
 *
 */
22 23
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

24 25
#include <linux/kernel.h>
#include <linux/mm.h>
26
#include <linux/slab.h>
27
#include <linux/vmalloc.h>
28
#include <linux/hyperv.h>
29
#include <asm/hyperv.h>
30
#include "hyperv_vmbus.h"
31

32
/* The one and only */
33 34 35 36 37
struct hv_context hv_context = {
	.synic_initialized	= false,
	.hypercall_page		= NULL,
	.signal_event_param	= NULL,
	.signal_event_buffer	= NULL,
38 39
};

40
/*
41
 * query_hypervisor_presence
L
Lucas De Marchi 已提交
42
 * - Query the cpuid for presence of windows hypervisor
43
 */
44
static int query_hypervisor_presence(void)
45
{
46 47 48 49 50
	unsigned int eax;
	unsigned int ebx;
	unsigned int ecx;
	unsigned int edx;
	unsigned int op;
51

52 53 54 55
	eax = 0;
	ebx = 0;
	ecx = 0;
	edx = 0;
56
	op = HVCPUID_VERSION_FEATURES;
57
	cpuid(op, &eax, &ebx, &ecx, &edx);
58 59

	return ecx & HV_PRESENT_BIT;
60 61
}

62
/*
63
 * query_hypervisor_info - Get version info of the windows hypervisor
64
 */
65
static int query_hypervisor_info(void)
66 67 68 69 70
{
	unsigned int eax;
	unsigned int ebx;
	unsigned int ecx;
	unsigned int edx;
71
	unsigned int max_leaf;
72
	unsigned int op;
73

74 75 76 77 78 79 80 81
	/*
	* Its assumed that this is called after confirming that Viridian
	* is present. Query id and revision.
	*/
	eax = 0;
	ebx = 0;
	ecx = 0;
	edx = 0;
82
	op = HVCPUID_VENDOR_MAXFUNCTION;
83
	cpuid(op, &eax, &ebx, &ecx, &edx);
84

85
	max_leaf = eax;
86

87
	if (max_leaf >= HVCPUID_VERSION) {
88 89 90 91
		eax = 0;
		ebx = 0;
		ecx = 0;
		edx = 0;
92
		op = HVCPUID_VERSION;
93
		cpuid(op, &eax, &ebx, &ecx, &edx);
94
		pr_info("Hyper-V Host OS Build:%d-%d.%d-%d-%d.%d\n",
95 96 97 98 99 100 101
			    eax,
			    ebx >> 16,
			    ebx & 0xFFFF,
			    ecx,
			    edx >> 24,
			    edx & 0xFFFFFF);
	}
102
	return max_leaf;
103
}
104

105
/*
106
 * do_hypercall- Invoke the specified hypercall
107
 */
108
static u64 do_hypercall(u64 control, void *input, void *output)
109
{
110
#ifdef CONFIG_X86_64
111 112 113
	u64 hv_status = 0;
	u64 input_address = (input) ? virt_to_phys(input) : 0;
	u64 output_address = (output) ? virt_to_phys(output) : 0;
114
	void *hypercall_page = hv_context.hypercall_page;
115

116 117 118 119
	__asm__ __volatile__("mov %0, %%r8" : : "r" (output_address) : "r8");
	__asm__ __volatile__("call *%3" : "=a" (hv_status) :
			     "c" (control), "d" (input_address),
			     "m" (hypercall_page));
120

121
	return hv_status;
122 123 124

#else

125 126 127 128 129 130 131 132 133 134
	u32 control_hi = control >> 32;
	u32 control_lo = control & 0xFFFFFFFF;
	u32 hv_status_hi = 1;
	u32 hv_status_lo = 1;
	u64 input_address = (input) ? virt_to_phys(input) : 0;
	u32 input_address_hi = input_address >> 32;
	u32 input_address_lo = input_address & 0xFFFFFFFF;
	u64 output_address = (output) ? virt_to_phys(output) : 0;
	u32 output_address_hi = output_address >> 32;
	u32 output_address_lo = output_address & 0xFFFFFFFF;
135
	void *hypercall_page = hv_context.hypercall_page;
136

137 138 139 140 141
	__asm__ __volatile__ ("call *%8" : "=d"(hv_status_hi),
			      "=a"(hv_status_lo) : "d" (control_hi),
			      "a" (control_lo), "b" (input_address_hi),
			      "c" (input_address_lo), "D"(output_address_hi),
			      "S"(output_address_lo), "m" (hypercall_page));
142

143
	return hv_status_lo | ((u64)hv_status_hi << 32);
144
#endif /* !x86_64 */
145 146
}

147
/*
148
 * hv_init - Main initialization routine.
149 150 151
 *
 * This routine must be called before any other routines in here are called
 */
152
int hv_init(void)
153
{
154 155 156
	int max_leaf;
	union hv_x64_msr_hypercall_contents hypercall_msr;
	void *virtaddr = NULL;
157

158 159 160
	memset(hv_context.synic_event_page, 0, sizeof(void *) * MAX_NUM_CPUS);
	memset(hv_context.synic_message_page, 0,
	       sizeof(void *) * MAX_NUM_CPUS);
161

162
	if (!query_hypervisor_presence())
163
		goto cleanup;
164

165
	max_leaf = query_hypervisor_info();
166

167
	rdmsrl(HV_X64_MSR_GUEST_OS_ID, hv_context.guestid);
168

169
	if (hv_context.guestid != 0)
170
		goto cleanup;
171

172 173
	/* Write our OS info */
	wrmsrl(HV_X64_MSR_GUEST_OS_ID, HV_LINUX_GUEST_ID);
174
	hv_context.guestid = HV_LINUX_GUEST_ID;
175

176
	/* See if the hypercall page is already set */
177
	rdmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
178

179
	virtaddr = __vmalloc(PAGE_SIZE, GFP_KERNEL, PAGE_KERNEL_EXEC);
180

181
	if (!virtaddr)
182
		goto cleanup;
183

184
	hypercall_msr.enable = 1;
185

186 187
	hypercall_msr.guest_physical_address = vmalloc_to_pfn(virtaddr);
	wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
188 189

	/* Confirm that hypercall page did get setup. */
190 191
	hypercall_msr.as_uint64 = 0;
	rdmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
192

193
	if (!hypercall_msr.enable)
194
		goto cleanup;
195

196
	hv_context.hypercall_page = virtaddr;
197

198
	/* Setup the global signal event param for the signal event hypercall */
199
	hv_context.signal_event_buffer =
200 201
			kmalloc(sizeof(struct hv_input_signal_event_buffer),
				GFP_KERNEL);
202
	if (!hv_context.signal_event_buffer)
203
		goto cleanup;
204

205
	hv_context.signal_event_param =
206
		(struct hv_input_signal_event *)
207
			(ALIGN((unsigned long)
208
				  hv_context.signal_event_buffer,
209
				  HV_HYPERCALL_PARAM_ALIGN));
210 211
	hv_context.signal_event_param->connectionid.asu32 = 0;
	hv_context.signal_event_param->connectionid.u.id =
212
						VMBUS_EVENT_CONNECTION_ID;
213 214
	hv_context.signal_event_param->flag_number = 0;
	hv_context.signal_event_param->rsvdz = 0;
215

216
	return 0;
217

218
cleanup:
219 220 221 222
	if (virtaddr) {
		if (hypercall_msr.enable) {
			hypercall_msr.as_uint64 = 0;
			wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
223 224
		}

225
		vfree(virtaddr);
226
	}
227 228

	return -ENOTSUPP;
229 230
}

231
/*
232
 * hv_cleanup - Cleanup routine.
233 234 235
 *
 * This routine is called normally during driver unloading or exiting.
 */
236
void hv_cleanup(void)
237
{
238
	union hv_x64_msr_hypercall_contents hypercall_msr;
239

240 241 242
	/* Reset our OS id */
	wrmsrl(HV_X64_MSR_GUEST_OS_ID, 0);

243 244 245
	kfree(hv_context.signal_event_buffer);
	hv_context.signal_event_buffer = NULL;
	hv_context.signal_event_param = NULL;
246

247
	if (hv_context.hypercall_page) {
248 249
		hypercall_msr.as_uint64 = 0;
		wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
250 251
		vfree(hv_context.hypercall_page);
		hv_context.hypercall_page = NULL;
252 253 254
	}
}

255
/*
256
 * hv_post_message - Post a message using the hypervisor message IPC.
257 258 259
 *
 * This involves a hypercall.
 */
260
u16 hv_post_message(union hv_connection_id connection_id,
261 262
		  enum hv_message_type message_type,
		  void *payload, size_t payload_size)
263
{
264
	struct aligned_input {
265
		u64 alignment8;
266
		struct hv_input_post_message msg;
267 268
	};

269
	struct hv_input_post_message *aligned_msg;
270
	u16 status;
271
	unsigned long addr;
272

273
	if (payload_size > HV_MESSAGE_PAYLOAD_BYTE_COUNT)
274
		return -EMSGSIZE;
275

276
	addr = (unsigned long)kmalloc(sizeof(struct aligned_input), GFP_ATOMIC);
277
	if (!addr)
278
		return -ENOMEM;
279

280
	aligned_msg = (struct hv_input_post_message *)
281
			(ALIGN(addr, HV_HYPERCALL_PARAM_ALIGN));
282

283 284 285 286
	aligned_msg->connectionid = connection_id;
	aligned_msg->message_type = message_type;
	aligned_msg->payload_size = payload_size;
	memcpy((void *)aligned_msg->payload, payload, payload_size);
287

288 289
	status = do_hypercall(HVCALL_POST_MESSAGE, aligned_msg, NULL)
		& 0xFFFF;
290

291
	kfree((void *)addr);
292 293 294 295 296

	return status;
}


297
/*
298 299
 * hv_signal_event -
 * Signal an event on the specified connection using the hypervisor event IPC.
300 301 302
 *
 * This involves a hypercall.
 */
303
u16 hv_signal_event(void)
304
{
305
	u16 status;
306

307
	status = do_hypercall(HVCALL_SIGNAL_EVENT,
308
			       hv_context.signal_event_param,
309
			       NULL) & 0xFFFF;
310 311 312
	return status;
}

313
/*
314
 * hv_synic_init - Initialize the Synthethic Interrupt Controller.
315 316 317 318 319
 *
 * If it is already initialized by another entity (ie x2v shim), we need to
 * retrieve the initialized message and event pages.  Otherwise, we create and
 * initialize the message and event pages.
 */
320
void hv_synic_init(void *irqarg)
321
{
322
	u64 version;
323 324
	union hv_synic_simp simp;
	union hv_synic_siefp siefp;
325
	union hv_synic_sint shared_sint;
326
	union hv_synic_scontrol sctrl;
327

328
	u32 irq_vector = *((u32 *)(irqarg));
329
	int cpu = smp_processor_id();
330

331
	if (!hv_context.hypercall_page)
332
		return;
333

334
	/* Check the version */
335
	rdmsrl(HV_X64_MSR_SVERSION, version);
336

337 338
	hv_context.synic_message_page[cpu] =
		(void *)get_zeroed_page(GFP_ATOMIC);
339

340
	if (hv_context.synic_message_page[cpu] == NULL) {
341
		pr_err("Unable to allocate SYNIC message page\n");
342
		goto cleanup;
343
	}
344

345 346
	hv_context.synic_event_page[cpu] =
		(void *)get_zeroed_page(GFP_ATOMIC);
347

348
	if (hv_context.synic_event_page[cpu] == NULL) {
349
		pr_err("Unable to allocate SYNIC event page\n");
350
		goto cleanup;
351
	}
352

353
	/* Setup the Synic's message page */
354 355
	rdmsrl(HV_X64_MSR_SIMP, simp.as_uint64);
	simp.simp_enabled = 1;
356
	simp.base_simp_gpa = virt_to_phys(hv_context.synic_message_page[cpu])
357
		>> PAGE_SHIFT;
358

359
	wrmsrl(HV_X64_MSR_SIMP, simp.as_uint64);
360

361
	/* Setup the Synic's event page */
362 363
	rdmsrl(HV_X64_MSR_SIEFP, siefp.as_uint64);
	siefp.siefp_enabled = 1;
364
	siefp.base_siefp_gpa = virt_to_phys(hv_context.synic_event_page[cpu])
365 366
		>> PAGE_SHIFT;

367
	wrmsrl(HV_X64_MSR_SIEFP, siefp.as_uint64);
368 369

	/* Setup the shared SINT. */
370
	rdmsrl(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT, shared_sint.as_uint64);
371

372 373 374
	shared_sint.as_uint64 = 0;
	shared_sint.vector = irq_vector; /* HV_SHARED_SINT_IDT_VECTOR + 0x20; */
	shared_sint.masked = false;
375
	shared_sint.auto_eoi = false;
376

377
	wrmsrl(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT, shared_sint.as_uint64);
378

379
	/* Enable the global synic bit */
380 381
	rdmsrl(HV_X64_MSR_SCONTROL, sctrl.as_uint64);
	sctrl.enable = 1;
382

383
	wrmsrl(HV_X64_MSR_SCONTROL, sctrl.as_uint64);
384

385
	hv_context.synic_initialized = true;
386
	return;
387

388
cleanup:
389
	if (hv_context.synic_event_page[cpu])
390
		free_page((unsigned long)hv_context.synic_event_page[cpu]);
391

392
	if (hv_context.synic_message_page[cpu])
393
		free_page((unsigned long)hv_context.synic_message_page[cpu]);
394
	return;
395 396
}

397
/*
398
 * hv_synic_cleanup - Cleanup routine for hv_synic_init().
399
 */
400
void hv_synic_cleanup(void *arg)
401
{
402
	union hv_synic_sint shared_sint;
403 404
	union hv_synic_simp simp;
	union hv_synic_siefp siefp;
405
	int cpu = smp_processor_id();
406

407
	if (!hv_context.synic_initialized)
408 409
		return;

410
	rdmsrl(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT, shared_sint.as_uint64);
411

412
	shared_sint.masked = 1;
413

414
	/* Need to correctly cleanup in the case of SMP!!! */
415
	/* Disable the interrupt */
416
	wrmsrl(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT, shared_sint.as_uint64);
417

418 419 420
	rdmsrl(HV_X64_MSR_SIMP, simp.as_uint64);
	simp.simp_enabled = 0;
	simp.base_simp_gpa = 0;
421

422
	wrmsrl(HV_X64_MSR_SIMP, simp.as_uint64);
423

424 425 426
	rdmsrl(HV_X64_MSR_SIEFP, siefp.as_uint64);
	siefp.siefp_enabled = 0;
	siefp.base_siefp_gpa = 0;
427

428
	wrmsrl(HV_X64_MSR_SIEFP, siefp.as_uint64);
429

430 431
	free_page((unsigned long)hv_context.synic_message_page[cpu]);
	free_page((unsigned long)hv_context.synic_event_page[cpu]);
432
}