hv_init.c 11 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-only
2
3
4
5
6
7
8
9
/*
 * X86 specific Hyper-V initialization code.
 *
 * Copyright (C) 2016, Microsoft, Inc.
 *
 * Author : K. Y. Srinivasan <kys@microsoft.com>
 */

10
#include <linux/efi.h>
11
#include <linux/types.h>
12
13
#include <asm/apic.h>
#include <asm/desc.h>
14
#include <asm/hypervisor.h>
15
#include <asm/hyperv-tlfs.h>
16
17
18
19
#include <asm/mshyperv.h>
#include <linux/version.h>
#include <linux/vmalloc.h>
#include <linux/mm.h>
20
#include <linux/hyperv.h>
Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
21
22
#include <linux/slab.h>
#include <linux/cpuhotplug.h>
23
#include <clocksource/hyperv_timer.h>
24

25
26
void *hv_hypercall_pg;
EXPORT_SYMBOL_GPL(hv_hypercall_pg);
27

Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
28
29
30
u32 *hv_vp_index;
EXPORT_SYMBOL_GPL(hv_vp_index);

31
32
33
struct hv_vp_assist_page **hv_vp_assist_page;
EXPORT_SYMBOL_GPL(hv_vp_assist_page);

34
35
36
void  __percpu **hyperv_pcpu_input_arg;
EXPORT_SYMBOL_GPL(hyperv_pcpu_input_arg);

37
u32 hv_max_vp_index;
38
EXPORT_SYMBOL_GPL(hv_max_vp_index);
39

40
41
42
43
44
45
46
47
48
49
50
51
52
53
void *hv_alloc_hyperv_page(void)
{
	BUILD_BUG_ON(PAGE_SIZE != HV_HYP_PAGE_SIZE);

	return (void *)__get_free_page(GFP_KERNEL);
}
EXPORT_SYMBOL_GPL(hv_alloc_hyperv_page);

void hv_free_hyperv_page(unsigned long addr)
{
	free_page(addr);
}
EXPORT_SYMBOL_GPL(hv_free_hyperv_page);

Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
54
55
56
static int hv_cpu_init(unsigned int cpu)
{
	u64 msr_vp_index;
57
	struct hv_vp_assist_page **hvp = &hv_vp_assist_page[smp_processor_id()];
58
	void **input_arg;
59
	struct page *pg;
60
61

	input_arg = (void **)this_cpu_ptr(hyperv_pcpu_input_arg);
62
63
64
65
	pg = alloc_page(GFP_KERNEL);
	if (unlikely(!pg))
		return -ENOMEM;
	*input_arg = page_address(pg);
Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
66
67
68
69
70

	hv_get_vp_index(msr_vp_index);

	hv_vp_index[smp_processor_id()] = msr_vp_index;

71
72
73
	if (msr_vp_index > hv_max_vp_index)
		hv_max_vp_index = msr_vp_index;

74
75
76
	if (!hv_vp_assist_page)
		return 0;

77
78
79
80
81
82
83
84
85
86
87
	/*
	 * The VP ASSIST PAGE is an "overlay" page (see Hyper-V TLFS's Section
	 * 5.2.1 "GPA Overlay Pages"). Here it must be zeroed out to make sure
	 * we always write the EOI MSR in hv_apic_eoi_write() *after* the
	 * EOI optimization is disabled in hv_cpu_die(), otherwise a CPU may
	 * not be stopped in the case of CPU offlining and the VM will hang.
	 */
	if (!*hvp) {
		*hvp = __vmalloc(PAGE_SIZE, GFP_KERNEL | __GFP_ZERO,
				 PAGE_KERNEL);
	}
88
89
90
91
92
93
94
95
96
97
98

	if (*hvp) {
		u64 val;

		val = vmalloc_to_pfn(*hvp);
		val = (val << HV_X64_MSR_VP_ASSIST_PAGE_ADDRESS_SHIFT) |
			HV_X64_MSR_VP_ASSIST_PAGE_ENABLE;

		wrmsrl(HV_X64_MSR_VP_ASSIST_PAGE, val);
	}

Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
99
100
101
	return 0;
}

102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
static void (*hv_reenlightenment_cb)(void);

static void hv_reenlightenment_notify(struct work_struct *dummy)
{
	struct hv_tsc_emulation_status emu_status;

	rdmsrl(HV_X64_MSR_TSC_EMULATION_STATUS, *(u64 *)&emu_status);

	/* Don't issue the callback if TSC accesses are not emulated */
	if (hv_reenlightenment_cb && emu_status.inprogress)
		hv_reenlightenment_cb();
}
static DECLARE_DELAYED_WORK(hv_reenlightenment_work, hv_reenlightenment_notify);

void hyperv_stop_tsc_emulation(void)
{
	u64 freq;
	struct hv_tsc_emulation_status emu_status;

	rdmsrl(HV_X64_MSR_TSC_EMULATION_STATUS, *(u64 *)&emu_status);
	emu_status.inprogress = 0;
	wrmsrl(HV_X64_MSR_TSC_EMULATION_STATUS, *(u64 *)&emu_status);

	rdmsrl(HV_X64_MSR_TSC_FREQUENCY, freq);
	tsc_khz = div64_u64(freq, 1000);
}
EXPORT_SYMBOL_GPL(hyperv_stop_tsc_emulation);

static inline bool hv_reenlightenment_available(void)
{
	/*
	 * Check for required features and priviliges to make TSC frequency
	 * change notifications work.
	 */
	return ms_hyperv.features & HV_X64_ACCESS_FREQUENCY_MSRS &&
		ms_hyperv.misc_features & HV_FEATURE_FREQUENCY_MSRS_AVAILABLE &&
		ms_hyperv.features & HV_X64_ACCESS_REENLIGHTENMENT;
}

__visible void __irq_entry hyperv_reenlightenment_intr(struct pt_regs *regs)
{
	entering_ack_irq();

145
146
	inc_irq_stat(irq_hv_reenlightenment_count);

147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
	schedule_delayed_work(&hv_reenlightenment_work, HZ/10);

	exiting_irq();
}

void set_hv_tscchange_cb(void (*cb)(void))
{
	struct hv_reenlightenment_control re_ctrl = {
		.vector = HYPERV_REENLIGHTENMENT_VECTOR,
		.enabled = 1,
		.target_vp = hv_vp_index[smp_processor_id()]
	};
	struct hv_tsc_emulation_control emu_ctrl = {.enabled = 1};

	if (!hv_reenlightenment_available()) {
		pr_warn("Hyper-V: reenlightenment support is unavailable\n");
		return;
	}

	hv_reenlightenment_cb = cb;

	/* Make sure callback is registered before we write to MSRs */
	wmb();

	wrmsrl(HV_X64_MSR_REENLIGHTENMENT_CONTROL, *((u64 *)&re_ctrl));
	wrmsrl(HV_X64_MSR_TSC_EMULATION_CONTROL, *((u64 *)&emu_ctrl));
}
EXPORT_SYMBOL_GPL(set_hv_tscchange_cb);

void clear_hv_tscchange_cb(void)
{
	struct hv_reenlightenment_control re_ctrl;

	if (!hv_reenlightenment_available())
		return;

	rdmsrl(HV_X64_MSR_REENLIGHTENMENT_CONTROL, *(u64 *)&re_ctrl);
	re_ctrl.enabled = 0;
	wrmsrl(HV_X64_MSR_REENLIGHTENMENT_CONTROL, *(u64 *)&re_ctrl);

	hv_reenlightenment_cb = NULL;
}
EXPORT_SYMBOL_GPL(clear_hv_tscchange_cb);

191
192
193
194
static int hv_cpu_die(unsigned int cpu)
{
	struct hv_reenlightenment_control re_ctrl;
	unsigned int new_cpu;
195
196
197
198
199
200
201
202
203
204
	unsigned long flags;
	void **input_arg;
	void *input_pg = NULL;

	local_irq_save(flags);
	input_arg = (void **)this_cpu_ptr(hyperv_pcpu_input_arg);
	input_pg = *input_arg;
	*input_arg = NULL;
	local_irq_restore(flags);
	free_page((unsigned long)input_pg);
205

206
207
208
	if (hv_vp_assist_page && hv_vp_assist_page[cpu])
		wrmsrl(HV_X64_MSR_VP_ASSIST_PAGE, 0);

209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
	if (hv_reenlightenment_cb == NULL)
		return 0;

	rdmsrl(HV_X64_MSR_REENLIGHTENMENT_CONTROL, *((u64 *)&re_ctrl));
	if (re_ctrl.target_vp == hv_vp_index[cpu]) {
		/* Reassign to some other online CPU */
		new_cpu = cpumask_any_but(cpu_online_mask, cpu);

		re_ctrl.target_vp = hv_vp_index[new_cpu];
		wrmsrl(HV_X64_MSR_REENLIGHTENMENT_CONTROL, *((u64 *)&re_ctrl));
	}

	return 0;
}

224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
static int __init hv_pci_init(void)
{
	int gen2vm = efi_enabled(EFI_BOOT);

	/*
	 * For Generation-2 VM, we exit from pci_arch_init() by returning 0.
	 * The purpose is to suppress the harmless warning:
	 * "PCI: Fatal: No config space access function found"
	 */
	if (gen2vm)
		return 0;

	/* For Generation-1 VM, we'll proceed in pci_arch_init().  */
	return 1;
}

240
241
242
243
244
/*
 * This function is to be invoked early in the boot sequence after the
 * hypervisor has been detected.
 *
 * 1. Setup the hypercall page.
245
 * 2. Register Hyper-V specific clocksource.
246
 * 3. Setup Hyper-V specific APIC entry points.
247
 */
248
void __init hyperv_init(void)
249
{
250
	u64 guest_id, required_msrs;
251
	union hv_x64_msr_hypercall_contents hypercall_msr;
252
	int cpuhp, i;
253

254
	if (x86_hyper_type != X86_HYPER_MS_HYPERV)
255
256
		return;

257
258
259
260
261
262
263
	/* Absolutely required MSRs */
	required_msrs = HV_X64_MSR_HYPERCALL_AVAILABLE |
		HV_X64_MSR_VP_INDEX_AVAILABLE;

	if ((ms_hyperv.features & required_msrs) != required_msrs)
		return;

264
265
266
267
268
269
270
271
272
273
	/*
	 * Allocate the per-CPU state for the hypercall input arg.
	 * If this allocation fails, we will not be able to setup
	 * (per-CPU) hypercall input page and thus this failure is
	 * fatal on Hyper-V.
	 */
	hyperv_pcpu_input_arg = alloc_percpu(void  *);

	BUG_ON(hyperv_pcpu_input_arg == NULL);

Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
274
275
276
277
278
279
	/* Allocate percpu VP index */
	hv_vp_index = kmalloc_array(num_possible_cpus(), sizeof(*hv_vp_index),
				    GFP_KERNEL);
	if (!hv_vp_index)
		return;

280
281
282
	for (i = 0; i < num_possible_cpus(); i++)
		hv_vp_index[i] = VP_INVAL;

283
284
285
286
	hv_vp_assist_page = kcalloc(num_possible_cpus(),
				    sizeof(*hv_vp_assist_page), GFP_KERNEL);
	if (!hv_vp_assist_page) {
		ms_hyperv.hints &= ~HV_X64_ENLIGHTENED_VMCS_RECOMMENDED;
Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
287
		goto free_vp_index;
288
289
290
291
292
293
	}

	cpuhp = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "x86/hyperv_init:online",
				  hv_cpu_init, hv_cpu_die);
	if (cpuhp < 0)
		goto free_vp_assist_page;
Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
294

295
296
297
298
299
300
301
302
	/*
	 * Setup the hypercall page and enable hypercalls.
	 * 1. Register the guest ID
	 * 2. Enable the hypercall and register the hypercall page
	 */
	guest_id = generate_guest_id(0, LINUX_VERSION_CODE, 0);
	wrmsrl(HV_X64_MSR_GUEST_OS_ID, guest_id);

303
304
	hv_hypercall_pg  = __vmalloc(PAGE_SIZE, GFP_KERNEL, PAGE_KERNEL_RX);
	if (hv_hypercall_pg == NULL) {
305
		wrmsrl(HV_X64_MSR_GUEST_OS_ID, 0);
306
		goto remove_cpuhp_state;
307
308
309
310
	}

	rdmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
	hypercall_msr.enable = 1;
311
	hypercall_msr.guest_physical_address = vmalloc_to_pfn(hv_hypercall_pg);
312
	wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
313

314
315
	hv_apic_init();

316
317
	x86_init.pci.arch_init = hv_pci_init;

318
319
	/* Register Hyper-V specific clocksource */
	hv_init_clocksource();
Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
320
321
	return;

322
323
324
325
326
remove_cpuhp_state:
	cpuhp_remove_state(cpuhp);
free_vp_assist_page:
	kfree(hv_vp_assist_page);
	hv_vp_assist_page = NULL;
Vitaly Kuznetsov's avatar
Vitaly Kuznetsov committed
327
328
329
free_vp_index:
	kfree(hv_vp_index);
	hv_vp_index = NULL;
330
}
331

332
333
334
335
336
337
338
339
340
341
/*
 * This routine is called before kexec/kdump, it does the required cleanup.
 */
void hyperv_cleanup(void)
{
	union hv_x64_msr_hypercall_contents hypercall_msr;

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

342
343
344
345
346
347
348
	/*
	 * Reset hypercall page reference before reset the page,
	 * let hypercall operations fail safely rather than
	 * panic the kernel for using invalid hypercall page
	 */
	hv_hypercall_pg = NULL;

349
350
351
	/* Reset the hypercall page */
	hypercall_msr.as_uint64 = 0;
	wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
352
353
354
355

	/* Reset the TSC page */
	hypercall_msr.as_uint64 = 0;
	wrmsrl(HV_X64_MSR_REFERENCE_TSC, hypercall_msr.as_uint64);
356
357
358
}
EXPORT_SYMBOL_GPL(hyperv_cleanup);

359
void hyperv_report_panic(struct pt_regs *regs, long err)
360
361
{
	static bool panic_reported;
362
	u64 guest_id;
363
364
365
366
367
368
369
370
371
372

	/*
	 * We prefer to report panic on 'die' chain as we have proper
	 * registers to report, but if we miss it (e.g. on BUG()) we need
	 * to report it on 'panic'.
	 */
	if (panic_reported)
		return;
	panic_reported = true;

373
374
375
376
377
378
379
	rdmsrl(HV_X64_MSR_GUEST_OS_ID, guest_id);

	wrmsrl(HV_X64_MSR_CRASH_P0, err);
	wrmsrl(HV_X64_MSR_CRASH_P1, guest_id);
	wrmsrl(HV_X64_MSR_CRASH_P2, regs->ip);
	wrmsrl(HV_X64_MSR_CRASH_P3, regs->ax);
	wrmsrl(HV_X64_MSR_CRASH_P4, regs->sp);
380
381
382
383
384
385
386

	/*
	 * Let Hyper-V know there is crash data available
	 */
	wrmsrl(HV_X64_MSR_CRASH_CTL, HV_CRASH_CTL_CRASH_NOTIFY);
}
EXPORT_SYMBOL_GPL(hyperv_report_panic);
387

388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
/**
 * hyperv_report_panic_msg - report panic message to Hyper-V
 * @pa: physical address of the panic page containing the message
 * @size: size of the message in the page
 */
void hyperv_report_panic_msg(phys_addr_t pa, size_t size)
{
	/*
	 * P3 to contain the physical address of the panic page & P4 to
	 * contain the size of the panic data in that page. Rest of the
	 * registers are no-op when the NOTIFY_MSG flag is set.
	 */
	wrmsrl(HV_X64_MSR_CRASH_P0, 0);
	wrmsrl(HV_X64_MSR_CRASH_P1, 0);
	wrmsrl(HV_X64_MSR_CRASH_P2, 0);
	wrmsrl(HV_X64_MSR_CRASH_P3, pa);
	wrmsrl(HV_X64_MSR_CRASH_P4, size);

	/*
	 * Let Hyper-V know there is crash data available along with
	 * the panic message.
	 */
	wrmsrl(HV_X64_MSR_CRASH_CTL,
	       (HV_CRASH_CTL_CRASH_NOTIFY | HV_CRASH_CTL_CRASH_NOTIFY_MSG));
}
EXPORT_SYMBOL_GPL(hyperv_report_panic_msg);

415
bool hv_is_hyperv_initialized(void)
416
417
418
{
	union hv_x64_msr_hypercall_contents hypercall_msr;

419
420
421
422
423
424
425
426
427
428
429
	/*
	 * Ensure that we're really on Hyper-V, and not a KVM or Xen
	 * emulation of Hyper-V
	 */
	if (x86_hyper_type != X86_HYPER_MS_HYPERV)
		return false;

	/*
	 * Verify that earlier initialization succeeded by checking
	 * that the hypercall page is setup
	 */
430
431
432
	hypercall_msr.as_uint64 = 0;
	rdmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);

433
	return hypercall_msr.enable;
434
}
435
EXPORT_SYMBOL_GPL(hv_is_hyperv_initialized);