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 <linux/version.h>
30
#include <asm/hyperv.h>
31
#include "hyperv_vmbus.h"
32

33
/* The one and only */
34 35 36 37 38
struct hv_context hv_context = {
	.synic_initialized	= false,
	.hypercall_page		= NULL,
	.signal_event_param	= NULL,
	.signal_event_buffer	= NULL,
39 40
};

41
/*
42
 * query_hypervisor_presence
L
Lucas De Marchi 已提交
43
 * - Query the cpuid for presence of windows hypervisor
44
 */
45
static int query_hypervisor_presence(void)
46
{
47 48 49 50 51
	unsigned int eax;
	unsigned int ebx;
	unsigned int ecx;
	unsigned int edx;
	unsigned int op;
52

53 54 55 56
	eax = 0;
	ebx = 0;
	ecx = 0;
	edx = 0;
57
	op = HVCPUID_VERSION_FEATURES;
58
	cpuid(op, &eax, &ebx, &ecx, &edx);
59 60

	return ecx & HV_PRESENT_BIT;
61 62
}

63
/*
64
 * query_hypervisor_info - Get version info of the windows hypervisor
65
 */
66
static int query_hypervisor_info(void)
67 68 69 70 71
{
	unsigned int eax;
	unsigned int ebx;
	unsigned int ecx;
	unsigned int edx;
72
	unsigned int max_leaf;
73
	unsigned int op;
74

75 76 77 78 79 80 81 82
	/*
	* Its assumed that this is called after confirming that Viridian
	* is present. Query id and revision.
	*/
	eax = 0;
	ebx = 0;
	ecx = 0;
	edx = 0;
83
	op = HVCPUID_VENDOR_MAXFUNCTION;
84
	cpuid(op, &eax, &ebx, &ecx, &edx);
85

86
	max_leaf = eax;
87

88
	if (max_leaf >= HVCPUID_VERSION) {
89 90 91 92
		eax = 0;
		ebx = 0;
		ecx = 0;
		edx = 0;
93
		op = HVCPUID_VERSION;
94
		cpuid(op, &eax, &ebx, &ecx, &edx);
95
		pr_info("Hyper-V Host OS Build:%d-%d.%d-%d-%d.%d\n",
96 97 98 99 100 101 102
			    eax,
			    ebx >> 16,
			    ebx & 0xFFFF,
			    ecx,
			    edx >> 24,
			    edx & 0xFFFFFF);
	}
103
	return max_leaf;
104
}
105

106
/*
107
 * do_hypercall- Invoke the specified hypercall
108
 */
109
static u64 do_hypercall(u64 control, void *input, void *output)
110
{
111
#ifdef CONFIG_X86_64
112 113 114
	u64 hv_status = 0;
	u64 input_address = (input) ? virt_to_phys(input) : 0;
	u64 output_address = (output) ? virt_to_phys(output) : 0;
115
	void *hypercall_page = hv_context.hypercall_page;
116

117 118 119 120
	__asm__ __volatile__("mov %0, %%r8" : : "r" (output_address) : "r8");
	__asm__ __volatile__("call *%3" : "=a" (hv_status) :
			     "c" (control), "d" (input_address),
			     "m" (hypercall_page));
121

122
	return hv_status;
123 124 125

#else

126 127 128 129 130 131 132 133 134 135
	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;
136
	void *hypercall_page = hv_context.hypercall_page;
137

138 139 140 141 142
	__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));
143

144
	return hv_status_lo | ((u64)hv_status_hi << 32);
145
#endif /* !x86_64 */
146 147
}

148
/*
149
 * hv_init - Main initialization routine.
150 151 152
 *
 * This routine must be called before any other routines in here are called
 */
153
int hv_init(void)
154
{
155 156 157
	int max_leaf;
	union hv_x64_msr_hypercall_contents hypercall_msr;
	void *virtaddr = NULL;
158

159
	memset(hv_context.synic_event_page, 0, sizeof(void *) * NR_CPUS);
160
	memset(hv_context.synic_message_page, 0,
161
	       sizeof(void *) * NR_CPUS);
162

163
	if (!query_hypervisor_presence())
164
		goto cleanup;
165

166
	max_leaf = query_hypervisor_info();
167

168 169 170 171 172
	/*
	 * Write our OS ID.
	 */
	hv_context.guestid = generate_guest_id(0, LINUX_VERSION_CODE, 0);
	wrmsrl(HV_X64_MSR_GUEST_OS_ID, hv_context.guestid);
173

174
	/* See if the hypercall page is already set */
175
	rdmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
176

177
	virtaddr = __vmalloc(PAGE_SIZE, GFP_KERNEL, PAGE_KERNEL_EXEC);
178

179
	if (!virtaddr)
180
		goto cleanup;
181

182
	hypercall_msr.enable = 1;
183

184 185
	hypercall_msr.guest_physical_address = vmalloc_to_pfn(virtaddr);
	wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
186 187

	/* Confirm that hypercall page did get setup. */
188 189
	hypercall_msr.as_uint64 = 0;
	rdmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
190

191
	if (!hypercall_msr.enable)
192
		goto cleanup;
193

194
	hv_context.hypercall_page = virtaddr;
195

196
	/* Setup the global signal event param for the signal event hypercall */
197
	hv_context.signal_event_buffer =
198 199
			kmalloc(sizeof(struct hv_input_signal_event_buffer),
				GFP_KERNEL);
200
	if (!hv_context.signal_event_buffer)
201
		goto cleanup;
202

203
	hv_context.signal_event_param =
204
		(struct hv_input_signal_event *)
205
			(ALIGN((unsigned long)
206
				  hv_context.signal_event_buffer,
207
				  HV_HYPERCALL_PARAM_ALIGN));
208 209
	hv_context.signal_event_param->connectionid.asu32 = 0;
	hv_context.signal_event_param->connectionid.u.id =
210
						VMBUS_EVENT_CONNECTION_ID;
211 212
	hv_context.signal_event_param->flag_number = 0;
	hv_context.signal_event_param->rsvdz = 0;
213

214
	return 0;
215

216
cleanup:
217 218 219 220
	if (virtaddr) {
		if (hypercall_msr.enable) {
			hypercall_msr.as_uint64 = 0;
			wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
221 222
		}

223
		vfree(virtaddr);
224
	}
225 226

	return -ENOTSUPP;
227 228
}

229
/*
230
 * hv_cleanup - Cleanup routine.
231 232 233
 *
 * This routine is called normally during driver unloading or exiting.
 */
234
void hv_cleanup(void)
235
{
236
	union hv_x64_msr_hypercall_contents hypercall_msr;
237

238 239 240
	/* Reset our OS id */
	wrmsrl(HV_X64_MSR_GUEST_OS_ID, 0);

241 242 243
	kfree(hv_context.signal_event_buffer);
	hv_context.signal_event_buffer = NULL;
	hv_context.signal_event_param = NULL;
244

245
	if (hv_context.hypercall_page) {
246 247
		hypercall_msr.as_uint64 = 0;
		wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
248 249
		vfree(hv_context.hypercall_page);
		hv_context.hypercall_page = NULL;
250 251 252
	}
}

253
/*
254
 * hv_post_message - Post a message using the hypervisor message IPC.
255 256 257
 *
 * This involves a hypercall.
 */
258
int hv_post_message(union hv_connection_id connection_id,
259 260
		  enum hv_message_type message_type,
		  void *payload, size_t payload_size)
261
{
262
	struct aligned_input {
263
		u64 alignment8;
264
		struct hv_input_post_message msg;
265 266
	};

267
	struct hv_input_post_message *aligned_msg;
268
	u16 status;
269
	unsigned long addr;
270

271
	if (payload_size > HV_MESSAGE_PAYLOAD_BYTE_COUNT)
272
		return -EMSGSIZE;
273

274
	addr = (unsigned long)kmalloc(sizeof(struct aligned_input), GFP_ATOMIC);
275
	if (!addr)
276
		return -ENOMEM;
277

278
	aligned_msg = (struct hv_input_post_message *)
279
			(ALIGN(addr, HV_HYPERCALL_PARAM_ALIGN));
280

281 282 283 284
	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);
285

286 287
	status = do_hypercall(HVCALL_POST_MESSAGE, aligned_msg, NULL)
		& 0xFFFF;
288

289
	kfree((void *)addr);
290 291 292 293 294

	return status;
}


295
/*
296 297
 * hv_signal_event -
 * Signal an event on the specified connection using the hypervisor event IPC.
298 299 300
 *
 * This involves a hypercall.
 */
301
u16 hv_signal_event(void)
302
{
303
	u16 status;
304

305
	status = do_hypercall(HVCALL_SIGNAL_EVENT,
306
			       hv_context.signal_event_param,
307
			       NULL) & 0xFFFF;
308 309 310
	return status;
}

311
/*
312
 * hv_synic_init - Initialize the Synthethic Interrupt Controller.
313 314 315 316 317
 *
 * 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.
 */
318
void hv_synic_init(void *irqarg)
319
{
320
	u64 version;
321 322
	union hv_synic_simp simp;
	union hv_synic_siefp siefp;
323
	union hv_synic_sint shared_sint;
324
	union hv_synic_scontrol sctrl;
325

326
	u32 irq_vector = *((u32 *)(irqarg));
327
	int cpu = smp_processor_id();
328

329
	if (!hv_context.hypercall_page)
330
		return;
331

332
	/* Check the version */
333
	rdmsrl(HV_X64_MSR_SVERSION, version);
334

335 336
	hv_context.synic_message_page[cpu] =
		(void *)get_zeroed_page(GFP_ATOMIC);
337

338
	if (hv_context.synic_message_page[cpu] == NULL) {
339
		pr_err("Unable to allocate SYNIC message page\n");
340
		goto cleanup;
341
	}
342

343 344
	hv_context.synic_event_page[cpu] =
		(void *)get_zeroed_page(GFP_ATOMIC);
345

346
	if (hv_context.synic_event_page[cpu] == NULL) {
347
		pr_err("Unable to allocate SYNIC event page\n");
348
		goto cleanup;
349
	}
350

351
	/* Setup the Synic's message page */
352 353
	rdmsrl(HV_X64_MSR_SIMP, simp.as_uint64);
	simp.simp_enabled = 1;
354
	simp.base_simp_gpa = virt_to_phys(hv_context.synic_message_page[cpu])
355
		>> PAGE_SHIFT;
356

357
	wrmsrl(HV_X64_MSR_SIMP, simp.as_uint64);
358

359
	/* Setup the Synic's event page */
360 361
	rdmsrl(HV_X64_MSR_SIEFP, siefp.as_uint64);
	siefp.siefp_enabled = 1;
362
	siefp.base_siefp_gpa = virt_to_phys(hv_context.synic_event_page[cpu])
363 364
		>> PAGE_SHIFT;

365
	wrmsrl(HV_X64_MSR_SIEFP, siefp.as_uint64);
366 367

	/* Setup the shared SINT. */
368
	rdmsrl(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT, shared_sint.as_uint64);
369

370 371 372
	shared_sint.as_uint64 = 0;
	shared_sint.vector = irq_vector; /* HV_SHARED_SINT_IDT_VECTOR + 0x20; */
	shared_sint.masked = false;
373
	shared_sint.auto_eoi = false;
374

375
	wrmsrl(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT, shared_sint.as_uint64);
376

377
	/* Enable the global synic bit */
378 379
	rdmsrl(HV_X64_MSR_SCONTROL, sctrl.as_uint64);
	sctrl.enable = 1;
380

381
	wrmsrl(HV_X64_MSR_SCONTROL, sctrl.as_uint64);
382

383
	hv_context.synic_initialized = true;
384
	return;
385

386
cleanup:
387
	if (hv_context.synic_event_page[cpu])
388
		free_page((unsigned long)hv_context.synic_event_page[cpu]);
389

390
	if (hv_context.synic_message_page[cpu])
391
		free_page((unsigned long)hv_context.synic_message_page[cpu]);
392
	return;
393 394
}

395
/*
396
 * hv_synic_cleanup - Cleanup routine for hv_synic_init().
397
 */
398
void hv_synic_cleanup(void *arg)
399
{
400
	union hv_synic_sint shared_sint;
401 402
	union hv_synic_simp simp;
	union hv_synic_siefp siefp;
403
	int cpu = smp_processor_id();
404

405
	if (!hv_context.synic_initialized)
406 407
		return;

408
	rdmsrl(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT, shared_sint.as_uint64);
409

410
	shared_sint.masked = 1;
411

412
	/* Need to correctly cleanup in the case of SMP!!! */
413
	/* Disable the interrupt */
414
	wrmsrl(HV_X64_MSR_SINT0 + VMBUS_MESSAGE_SINT, shared_sint.as_uint64);
415

416 417 418
	rdmsrl(HV_X64_MSR_SIMP, simp.as_uint64);
	simp.simp_enabled = 0;
	simp.base_simp_gpa = 0;
419

420
	wrmsrl(HV_X64_MSR_SIMP, simp.as_uint64);
421

422 423 424
	rdmsrl(HV_X64_MSR_SIEFP, siefp.as_uint64);
	siefp.siefp_enabled = 0;
	siefp.base_siefp_gpa = 0;
425

426
	wrmsrl(HV_X64_MSR_SIEFP, siefp.as_uint64);
427

428 429
	free_page((unsigned long)hv_context.synic_message_page[cpu]);
	free_page((unsigned long)hv_context.synic_event_page[cpu]);
430
}