1// SPDX-License-Identifier: GPL-2.0
2/*
3 * Core of Xen paravirt_ops implementation.
4 *
5 * This file contains the xen_paravirt_ops structure itself, and the
6 * implementations for:
7 * - privileged instructions
8 * - interrupt flags
9 * - segment operations
10 * - booting and setup
11 *
12 * Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
13 */
14
15#include <linux/cpu.h>
16#include <linux/kernel.h>
17#include <linux/init.h>
18#include <linux/smp.h>
19#include <linux/preempt.h>
20#include <linux/hardirq.h>
21#include <linux/percpu.h>
22#include <linux/delay.h>
23#include <linux/start_kernel.h>
24#include <linux/sched.h>
25#include <linux/kprobes.h>
26#include <linux/memblock.h>
27#include <linux/export.h>
28#include <linux/mm.h>
29#include <linux/page-flags.h>
30#include <linux/highmem.h>
31#include <linux/console.h>
32#include <linux/pci.h>
33#include <linux/gfp.h>
34#include <linux/edd.h>
35#include <linux/objtool.h>
36
37#include <xen/xen.h>
38#include <xen/events.h>
39#include <xen/interface/xen.h>
40#include <xen/interface/version.h>
41#include <xen/interface/physdev.h>
42#include <xen/interface/vcpu.h>
43#include <xen/interface/memory.h>
44#include <xen/interface/nmi.h>
45#include <xen/interface/xen-mca.h>
46#include <xen/features.h>
47#include <xen/page.h>
48#include <xen/hvc-console.h>
49#include <xen/acpi.h>
50
51#include <asm/paravirt.h>
52#include <asm/apic.h>
53#include <asm/page.h>
54#include <asm/xen/pci.h>
55#include <asm/xen/hypercall.h>
56#include <asm/xen/hypervisor.h>
57#include <asm/xen/cpuid.h>
58#include <asm/fixmap.h>
59#include <asm/processor.h>
60#include <asm/proto.h>
61#include <asm/msr-index.h>
62#include <asm/traps.h>
63#include <asm/setup.h>
64#include <asm/desc.h>
65#include <asm/pgalloc.h>
66#include <asm/tlbflush.h>
67#include <asm/reboot.h>
68#include <asm/stackprotector.h>
69#include <asm/hypervisor.h>
70#include <asm/mach_traps.h>
71#include <asm/mwait.h>
72#include <asm/pci_x86.h>
73#include <asm/cpu.h>
74#ifdef CONFIG_X86_IOPL_IOPERM
75#include <asm/io_bitmap.h>
76#endif
77
78#ifdef CONFIG_ACPI
79#include <linux/acpi.h>
80#include <asm/acpi.h>
81#include <acpi/pdc_intel.h>
82#include <acpi/processor.h>
83#include <xen/interface/platform.h>
84#endif
85
86#include "xen-ops.h"
87#include "mmu.h"
88#include "smp.h"
89#include "multicalls.h"
90#include "pmu.h"
91
92#include "../kernel/cpu/cpu.h" /* get_cpu_cap() */
93
94void *xen_initial_gdt;
95
96static int xen_cpu_up_prepare_pv(unsigned int cpu);
97static int xen_cpu_dead_pv(unsigned int cpu);
98
99struct tls_descs {
100	struct desc_struct desc[3];
101};
102
103/*
104 * Updating the 3 TLS descriptors in the GDT on every task switch is
105 * surprisingly expensive so we avoid updating them if they haven't
106 * changed.  Since Xen writes different descriptors than the one
107 * passed in the update_descriptor hypercall we keep shadow copies to
108 * compare against.
109 */
110static DEFINE_PER_CPU(struct tls_descs, shadow_tls_desc);
111
112static void __init xen_banner(void)
113{
114	unsigned version = HYPERVISOR_xen_version(XENVER_version, NULL);
115	struct xen_extraversion extra;
116	HYPERVISOR_xen_version(XENVER_extraversion, &extra);
117
118	pr_info("Booting paravirtualized kernel on %s\n", pv_info.name);
119	printk(KERN_INFO "Xen version: %d.%d%s%s\n",
120	       version >> 16, version & 0xffff, extra.extraversion,
121	       xen_feature(XENFEAT_mmu_pt_update_preserve_ad) ? " (preserve-AD)" : "");
122}
123
124static void __init xen_pv_init_platform(void)
125{
126	populate_extra_pte(fix_to_virt(FIX_PARAVIRT_BOOTMAP));
127
128	set_fixmap(FIX_PARAVIRT_BOOTMAP, xen_start_info->shared_info);
129	HYPERVISOR_shared_info = (void *)fix_to_virt(FIX_PARAVIRT_BOOTMAP);
130
131	/* xen clock uses per-cpu vcpu_info, need to init it for boot cpu */
132	xen_vcpu_info_reset(0);
133
134	/* pvclock is in shared info area */
135	xen_init_time_ops();
136}
137
138static void __init xen_pv_guest_late_init(void)
139{
140#ifndef CONFIG_SMP
141	/* Setup shared vcpu info for non-smp configurations */
142	xen_setup_vcpu_info_placement();
143#endif
144}
145
146/* Check if running on Xen version (major, minor) or later */
147bool
148xen_running_on_version_or_later(unsigned int major, unsigned int minor)
149{
150	unsigned int version;
151
152	if (!xen_domain())
153		return false;
154
155	version = HYPERVISOR_xen_version(XENVER_version, NULL);
156	if ((((version >> 16) == major) && ((version & 0xffff) >= minor)) ||
157		((version >> 16) > major))
158		return true;
159	return false;
160}
161
162static __read_mostly unsigned int cpuid_leaf5_ecx_val;
163static __read_mostly unsigned int cpuid_leaf5_edx_val;
164
165static void xen_cpuid(unsigned int *ax, unsigned int *bx,
166		      unsigned int *cx, unsigned int *dx)
167{
168	unsigned maskebx = ~0;
169
170	/*
171	 * Mask out inconvenient features, to try and disable as many
172	 * unsupported kernel subsystems as possible.
173	 */
174	switch (*ax) {
175	case CPUID_MWAIT_LEAF:
176		/* Synthesize the values.. */
177		*ax = 0;
178		*bx = 0;
179		*cx = cpuid_leaf5_ecx_val;
180		*dx = cpuid_leaf5_edx_val;
181		return;
182
183	case 0xb:
184		/* Suppress extended topology stuff */
185		maskebx = 0;
186		break;
187	}
188
189	asm(XEN_EMULATE_PREFIX "cpuid"
190		: "=a" (*ax),
191		  "=b" (*bx),
192		  "=c" (*cx),
193		  "=d" (*dx)
194		: "0" (*ax), "2" (*cx));
195
196	*bx &= maskebx;
197}
198STACK_FRAME_NON_STANDARD(xen_cpuid); /* XEN_EMULATE_PREFIX */
199
200static bool __init xen_check_mwait(void)
201{
202#ifdef CONFIG_ACPI
203	struct xen_platform_op op = {
204		.cmd			= XENPF_set_processor_pminfo,
205		.u.set_pminfo.id	= -1,
206		.u.set_pminfo.type	= XEN_PM_PDC,
207	};
208	uint32_t buf[3];
209	unsigned int ax, bx, cx, dx;
210	unsigned int mwait_mask;
211
212	/* We need to determine whether it is OK to expose the MWAIT
213	 * capability to the kernel to harvest deeper than C3 states from ACPI
214	 * _CST using the processor_harvest_xen.c module. For this to work, we
215	 * need to gather the MWAIT_LEAF values (which the cstate.c code
216	 * checks against). The hypervisor won't expose the MWAIT flag because
217	 * it would break backwards compatibility; so we will find out directly
218	 * from the hardware and hypercall.
219	 */
220	if (!xen_initial_domain())
221		return false;
222
223	/*
224	 * When running under platform earlier than Xen4.2, do not expose
225	 * mwait, to avoid the risk of loading native acpi pad driver
226	 */
227	if (!xen_running_on_version_or_later(4, 2))
228		return false;
229
230	ax = 1;
231	cx = 0;
232
233	native_cpuid(&ax, &bx, &cx, &dx);
234
235	mwait_mask = (1 << (X86_FEATURE_EST % 32)) |
236		     (1 << (X86_FEATURE_MWAIT % 32));
237
238	if ((cx & mwait_mask) != mwait_mask)
239		return false;
240
241	/* We need to emulate the MWAIT_LEAF and for that we need both
242	 * ecx and edx. The hypercall provides only partial information.
243	 */
244
245	ax = CPUID_MWAIT_LEAF;
246	bx = 0;
247	cx = 0;
248	dx = 0;
249
250	native_cpuid(&ax, &bx, &cx, &dx);
251
252	/* Ask the Hypervisor whether to clear ACPI_PDC_C_C2C3_FFH. If so,
253	 * don't expose MWAIT_LEAF and let ACPI pick the IOPORT version of C3.
254	 */
255	buf[0] = ACPI_PDC_REVISION_ID;
256	buf[1] = 1;
257	buf[2] = (ACPI_PDC_C_CAPABILITY_SMP | ACPI_PDC_EST_CAPABILITY_SWSMP);
258
259	set_xen_guest_handle(op.u.set_pminfo.pdc, buf);
260
261	if ((HYPERVISOR_platform_op(&op) == 0) &&
262	    (buf[2] & (ACPI_PDC_C_C1_FFH | ACPI_PDC_C_C2C3_FFH))) {
263		cpuid_leaf5_ecx_val = cx;
264		cpuid_leaf5_edx_val = dx;
265	}
266	return true;
267#else
268	return false;
269#endif
270}
271
272static bool __init xen_check_xsave(void)
273{
274	unsigned int cx, xsave_mask;
275
276	cx = cpuid_ecx(1);
277
278	xsave_mask = (1 << (X86_FEATURE_XSAVE % 32)) |
279		     (1 << (X86_FEATURE_OSXSAVE % 32));
280
281	/* Xen will set CR4.OSXSAVE if supported and not disabled by force */
282	return (cx & xsave_mask) == xsave_mask;
283}
284
285static void __init xen_init_capabilities(void)
286{
287	setup_force_cpu_cap(X86_FEATURE_XENPV);
288	setup_clear_cpu_cap(X86_FEATURE_DCA);
289	setup_clear_cpu_cap(X86_FEATURE_APERFMPERF);
290	setup_clear_cpu_cap(X86_FEATURE_MTRR);
291	setup_clear_cpu_cap(X86_FEATURE_ACC);
292	setup_clear_cpu_cap(X86_FEATURE_X2APIC);
293	setup_clear_cpu_cap(X86_FEATURE_SME);
294
295	/*
296	 * Xen PV would need some work to support PCID: CR3 handling as well
297	 * as xen_flush_tlb_others() would need updating.
298	 */
299	setup_clear_cpu_cap(X86_FEATURE_PCID);
300
301	if (!xen_initial_domain())
302		setup_clear_cpu_cap(X86_FEATURE_ACPI);
303
304	if (xen_check_mwait())
305		setup_force_cpu_cap(X86_FEATURE_MWAIT);
306	else
307		setup_clear_cpu_cap(X86_FEATURE_MWAIT);
308
309	if (!xen_check_xsave()) {
310		setup_clear_cpu_cap(X86_FEATURE_XSAVE);
311		setup_clear_cpu_cap(X86_FEATURE_OSXSAVE);
312	}
313}
314
315static void xen_set_debugreg(int reg, unsigned long val)
316{
317	HYPERVISOR_set_debugreg(reg, val);
318}
319
320static unsigned long xen_get_debugreg(int reg)
321{
322	return HYPERVISOR_get_debugreg(reg);
323}
324
325static void xen_end_context_switch(struct task_struct *next)
326{
327	xen_mc_flush();
328	paravirt_end_context_switch(next);
329}
330
331static unsigned long xen_store_tr(void)
332{
333	return 0;
334}
335
336/*
337 * Set the page permissions for a particular virtual address.  If the
338 * address is a vmalloc mapping (or other non-linear mapping), then
339 * find the linear mapping of the page and also set its protections to
340 * match.
341 */
342static void set_aliased_prot(void *v, pgprot_t prot)
343{
344	int level;
345	pte_t *ptep;
346	pte_t pte;
347	unsigned long pfn;
348	unsigned char dummy;
349	void *va;
350
351	ptep = lookup_address((unsigned long)v, &level);
352	BUG_ON(ptep == NULL);
353
354	pfn = pte_pfn(*ptep);
355	pte = pfn_pte(pfn, prot);
356
357	/*
358	 * Careful: update_va_mapping() will fail if the virtual address
359	 * we're poking isn't populated in the page tables.  We don't
360	 * need to worry about the direct map (that's always in the page
361	 * tables), but we need to be careful about vmap space.  In
362	 * particular, the top level page table can lazily propagate
363	 * entries between processes, so if we've switched mms since we
364	 * vmapped the target in the first place, we might not have the
365	 * top-level page table entry populated.
366	 *
367	 * We disable preemption because we want the same mm active when
368	 * we probe the target and when we issue the hypercall.  We'll
369	 * have the same nominal mm, but if we're a kernel thread, lazy
370	 * mm dropping could change our pgd.
371	 *
372	 * Out of an abundance of caution, this uses __get_user() to fault
373	 * in the target address just in case there's some obscure case
374	 * in which the target address isn't readable.
375	 */
376
377	preempt_disable();
378
379	copy_from_kernel_nofault(&dummy, v, 1);
380
381	if (HYPERVISOR_update_va_mapping((unsigned long)v, pte, 0))
382		BUG();
383
384	va = __va(PFN_PHYS(pfn));
385
386	if (va != v && HYPERVISOR_update_va_mapping((unsigned long)va, pte, 0))
387		BUG();
388
389	preempt_enable();
390}
391
392static void xen_alloc_ldt(struct desc_struct *ldt, unsigned entries)
393{
394	const unsigned entries_per_page = PAGE_SIZE / LDT_ENTRY_SIZE;
395	int i;
396
397	/*
398	 * We need to mark the all aliases of the LDT pages RO.  We
399	 * don't need to call vm_flush_aliases(), though, since that's
400	 * only responsible for flushing aliases out the TLBs, not the
401	 * page tables, and Xen will flush the TLB for us if needed.
402	 *
403	 * To avoid confusing future readers: none of this is necessary
404	 * to load the LDT.  The hypervisor only checks this when the
405	 * LDT is faulted in due to subsequent descriptor access.
406	 */
407
408	for (i = 0; i < entries; i += entries_per_page)
409		set_aliased_prot(ldt + i, PAGE_KERNEL_RO);
410}
411
412static void xen_free_ldt(struct desc_struct *ldt, unsigned entries)
413{
414	const unsigned entries_per_page = PAGE_SIZE / LDT_ENTRY_SIZE;
415	int i;
416
417	for (i = 0; i < entries; i += entries_per_page)
418		set_aliased_prot(ldt + i, PAGE_KERNEL);
419}
420
421static void xen_set_ldt(const void *addr, unsigned entries)
422{
423	struct mmuext_op *op;
424	struct multicall_space mcs = xen_mc_entry(sizeof(*op));
425
426	trace_xen_cpu_set_ldt(addr, entries);
427
428	op = mcs.args;
429	op->cmd = MMUEXT_SET_LDT;
430	op->arg1.linear_addr = (unsigned long)addr;
431	op->arg2.nr_ents = entries;
432
433	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
434
435	xen_mc_issue(PARAVIRT_LAZY_CPU);
436}
437
438static void xen_load_gdt(const struct desc_ptr *dtr)
439{
440	unsigned long va = dtr->address;
441	unsigned int size = dtr->size + 1;
442	unsigned long pfn, mfn;
443	int level;
444	pte_t *ptep;
445	void *virt;
446
447	/* @size should be at most GDT_SIZE which is smaller than PAGE_SIZE. */
448	BUG_ON(size > PAGE_SIZE);
449	BUG_ON(va & ~PAGE_MASK);
450
451	/*
452	 * The GDT is per-cpu and is in the percpu data area.
453	 * That can be virtually mapped, so we need to do a
454	 * page-walk to get the underlying MFN for the
455	 * hypercall.  The page can also be in the kernel's
456	 * linear range, so we need to RO that mapping too.
457	 */
458	ptep = lookup_address(va, &level);
459	BUG_ON(ptep == NULL);
460
461	pfn = pte_pfn(*ptep);
462	mfn = pfn_to_mfn(pfn);
463	virt = __va(PFN_PHYS(pfn));
464
465	make_lowmem_page_readonly((void *)va);
466	make_lowmem_page_readonly(virt);
467
468	if (HYPERVISOR_set_gdt(&mfn, size / sizeof(struct desc_struct)))
469		BUG();
470}
471
472/*
473 * load_gdt for early boot, when the gdt is only mapped once
474 */
475static void __init xen_load_gdt_boot(const struct desc_ptr *dtr)
476{
477	unsigned long va = dtr->address;
478	unsigned int size = dtr->size + 1;
479	unsigned long pfn, mfn;
480	pte_t pte;
481
482	/* @size should be at most GDT_SIZE which is smaller than PAGE_SIZE. */
483	BUG_ON(size > PAGE_SIZE);
484	BUG_ON(va & ~PAGE_MASK);
485
486	pfn = virt_to_pfn(va);
487	mfn = pfn_to_mfn(pfn);
488
489	pte = pfn_pte(pfn, PAGE_KERNEL_RO);
490
491	if (HYPERVISOR_update_va_mapping((unsigned long)va, pte, 0))
492		BUG();
493
494	if (HYPERVISOR_set_gdt(&mfn, size / sizeof(struct desc_struct)))
495		BUG();
496}
497
498static inline bool desc_equal(const struct desc_struct *d1,
499			      const struct desc_struct *d2)
500{
501	return !memcmp(d1, d2, sizeof(*d1));
502}
503
504static void load_TLS_descriptor(struct thread_struct *t,
505				unsigned int cpu, unsigned int i)
506{
507	struct desc_struct *shadow = &per_cpu(shadow_tls_desc, cpu).desc[i];
508	struct desc_struct *gdt;
509	xmaddr_t maddr;
510	struct multicall_space mc;
511
512	if (desc_equal(shadow, &t->tls_array[i]))
513		return;
514
515	*shadow = t->tls_array[i];
516
517	gdt = get_cpu_gdt_rw(cpu);
518	maddr = arbitrary_virt_to_machine(&gdt[GDT_ENTRY_TLS_MIN+i]);
519	mc = __xen_mc_entry(0);
520
521	MULTI_update_descriptor(mc.mc, maddr.maddr, t->tls_array[i]);
522}
523
524static void xen_load_tls(struct thread_struct *t, unsigned int cpu)
525{
526	/*
527	 * In lazy mode we need to zero %fs, otherwise we may get an
528	 * exception between the new %fs descriptor being loaded and
529	 * %fs being effectively cleared at __switch_to().
530	 */
531	if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_CPU)
532		loadsegment(fs, 0);
533
534	xen_mc_batch();
535
536	load_TLS_descriptor(t, cpu, 0);
537	load_TLS_descriptor(t, cpu, 1);
538	load_TLS_descriptor(t, cpu, 2);
539
540	xen_mc_issue(PARAVIRT_LAZY_CPU);
541}
542
543static void xen_load_gs_index(unsigned int idx)
544{
545	if (HYPERVISOR_set_segment_base(SEGBASE_GS_USER_SEL, idx))
546		BUG();
547}
548
549static void xen_write_ldt_entry(struct desc_struct *dt, int entrynum,
550				const void *ptr)
551{
552	xmaddr_t mach_lp = arbitrary_virt_to_machine(&dt[entrynum]);
553	u64 entry = *(u64 *)ptr;
554
555	trace_xen_cpu_write_ldt_entry(dt, entrynum, entry);
556
557	preempt_disable();
558
559	xen_mc_flush();
560	if (HYPERVISOR_update_descriptor(mach_lp.maddr, entry))
561		BUG();
562
563	preempt_enable();
564}
565
566void noist_exc_debug(struct pt_regs *regs);
567
568DEFINE_IDTENTRY_RAW(xenpv_exc_nmi)
569{
570	/* On Xen PV, NMI doesn't use IST.  The C part is the sane as native. */
571	exc_nmi(regs);
572}
573
574DEFINE_IDTENTRY_RAW(xenpv_exc_debug)
575{
576	/*
577	 * There's no IST on Xen PV, but we still need to dispatch
578	 * to the correct handler.
579	 */
580	if (user_mode(regs))
581		noist_exc_debug(regs);
582	else
583		exc_debug(regs);
584}
585
586DEFINE_IDTENTRY_RAW(exc_xen_unknown_trap)
587{
588	/* This should never happen and there is no way to handle it. */
589	instrumentation_begin();
590	pr_err("Unknown trap in Xen PV mode.");
591	BUG();
592	instrumentation_end();
593}
594
595struct trap_array_entry {
596	void (*orig)(void);
597	void (*xen)(void);
598	bool ist_okay;
599};
600
601#define TRAP_ENTRY(func, ist_ok) {			\
602	.orig		= asm_##func,			\
603	.xen		= xen_asm_##func,		\
604	.ist_okay	= ist_ok }
605
606#define TRAP_ENTRY_REDIR(func, ist_ok) {		\
607	.orig		= asm_##func,			\
608	.xen		= xen_asm_xenpv_##func,		\
609	.ist_okay	= ist_ok }
610
611static struct trap_array_entry trap_array[] = {
612	TRAP_ENTRY_REDIR(exc_debug,			true  ),
613	TRAP_ENTRY(exc_double_fault,			true  ),
614#ifdef CONFIG_X86_MCE
615	TRAP_ENTRY(exc_machine_check,			true  ),
616#endif
617	TRAP_ENTRY_REDIR(exc_nmi,			true  ),
618	TRAP_ENTRY(exc_int3,				false ),
619	TRAP_ENTRY(exc_overflow,			false ),
620#ifdef CONFIG_IA32_EMULATION
621	{ entry_INT80_compat,          xen_entry_INT80_compat,          false },
622#endif
623	TRAP_ENTRY(exc_page_fault,			false ),
624	TRAP_ENTRY(exc_divide_error,			false ),
625	TRAP_ENTRY(exc_bounds,				false ),
626	TRAP_ENTRY(exc_invalid_op,			false ),
627	TRAP_ENTRY(exc_device_not_available,		false ),
628	TRAP_ENTRY(exc_coproc_segment_overrun,		false ),
629	TRAP_ENTRY(exc_invalid_tss,			false ),
630	TRAP_ENTRY(exc_segment_not_present,		false ),
631	TRAP_ENTRY(exc_stack_segment,			false ),
632	TRAP_ENTRY(exc_general_protection,		false ),
633	TRAP_ENTRY(exc_spurious_interrupt_bug,		false ),
634	TRAP_ENTRY(exc_coprocessor_error,		false ),
635	TRAP_ENTRY(exc_alignment_check,			false ),
636	TRAP_ENTRY(exc_simd_coprocessor_error,		false ),
637};
638
639static bool __ref get_trap_addr(void **addr, unsigned int ist)
640{
641	unsigned int nr;
642	bool ist_okay = false;
643	bool found = false;
644
645	/*
646	 * Replace trap handler addresses by Xen specific ones.
647	 * Check for known traps using IST and whitelist them.
648	 * The debugger ones are the only ones we care about.
649	 * Xen will handle faults like double_fault, so we should never see
650	 * them.  Warn if there's an unexpected IST-using fault handler.
651	 */
652	for (nr = 0; nr < ARRAY_SIZE(trap_array); nr++) {
653		struct trap_array_entry *entry = trap_array + nr;
654
655		if (*addr == entry->orig) {
656			*addr = entry->xen;
657			ist_okay = entry->ist_okay;
658			found = true;
659			break;
660		}
661	}
662
663	if (nr == ARRAY_SIZE(trap_array) &&
664	    *addr >= (void *)early_idt_handler_array[0] &&
665	    *addr < (void *)early_idt_handler_array[NUM_EXCEPTION_VECTORS]) {
666		nr = (*addr - (void *)early_idt_handler_array[0]) /
667		     EARLY_IDT_HANDLER_SIZE;
668		*addr = (void *)xen_early_idt_handler_array[nr];
669		found = true;
670	}
671
672	if (!found)
673		*addr = (void *)xen_asm_exc_xen_unknown_trap;
674
675	if (WARN_ON(found && ist != 0 && !ist_okay))
676		return false;
677
678	return true;
679}
680
681static int cvt_gate_to_trap(int vector, const gate_desc *val,
682			    struct trap_info *info)
683{
684	unsigned long addr;
685
686	if (val->bits.type != GATE_TRAP && val->bits.type != GATE_INTERRUPT)
687		return 0;
688
689	info->vector = vector;
690
691	addr = gate_offset(val);
692	if (!get_trap_addr((void **)&addr, val->bits.ist))
693		return 0;
694	info->address = addr;
695
696	info->cs = gate_segment(val);
697	info->flags = val->bits.dpl;
698	/* interrupt gates clear IF */
699	if (val->bits.type == GATE_INTERRUPT)
700		info->flags |= 1 << 2;
701
702	return 1;
703}
704
705/* Locations of each CPU's IDT */
706static DEFINE_PER_CPU(struct desc_ptr, idt_desc);
707
708/* Set an IDT entry.  If the entry is part of the current IDT, then
709   also update Xen. */
710static void xen_write_idt_entry(gate_desc *dt, int entrynum, const gate_desc *g)
711{
712	unsigned long p = (unsigned long)&dt[entrynum];
713	unsigned long start, end;
714
715	trace_xen_cpu_write_idt_entry(dt, entrynum, g);
716
717	preempt_disable();
718
719	start = __this_cpu_read(idt_desc.address);
720	end = start + __this_cpu_read(idt_desc.size) + 1;
721
722	xen_mc_flush();
723
724	native_write_idt_entry(dt, entrynum, g);
725
726	if (p >= start && (p + 8) <= end) {
727		struct trap_info info[2];
728
729		info[1].address = 0;
730
731		if (cvt_gate_to_trap(entrynum, g, &info[0]))
732			if (HYPERVISOR_set_trap_table(info))
733				BUG();
734	}
735
736	preempt_enable();
737}
738
739static unsigned xen_convert_trap_info(const struct desc_ptr *desc,
740				      struct trap_info *traps, bool full)
741{
742	unsigned in, out, count;
743
744	count = (desc->size+1) / sizeof(gate_desc);
745	BUG_ON(count > 256);
746
747	for (in = out = 0; in < count; in++) {
748		gate_desc *entry = (gate_desc *)(desc->address) + in;
749
750		if (cvt_gate_to_trap(in, entry, &traps[out]) || full)
751			out++;
752	}
753
754	return out;
755}
756
757void xen_copy_trap_info(struct trap_info *traps)
758{
759	const struct desc_ptr *desc = this_cpu_ptr(&idt_desc);
760
761	xen_convert_trap_info(desc, traps, true);
762}
763
764/* Load a new IDT into Xen.  In principle this can be per-CPU, so we
765   hold a spinlock to protect the static traps[] array (static because
766   it avoids allocation, and saves stack space). */
767static void xen_load_idt(const struct desc_ptr *desc)
768{
769	static DEFINE_SPINLOCK(lock);
770	static struct trap_info traps[257];
771	static const struct trap_info zero = { };
772	unsigned out;
773
774	trace_xen_cpu_load_idt(desc);
775
776	spin_lock(&lock);
777
778	memcpy(this_cpu_ptr(&idt_desc), desc, sizeof(idt_desc));
779
780	out = xen_convert_trap_info(desc, traps, false);
781	traps[out] = zero;
782
783	xen_mc_flush();
784	if (HYPERVISOR_set_trap_table(traps))
785		BUG();
786
787	spin_unlock(&lock);
788}
789
790/* Write a GDT descriptor entry.  Ignore LDT descriptors, since
791   they're handled differently. */
792static void xen_write_gdt_entry(struct desc_struct *dt, int entry,
793				const void *desc, int type)
794{
795	trace_xen_cpu_write_gdt_entry(dt, entry, desc, type);
796
797	preempt_disable();
798
799	switch (type) {
800	case DESC_LDT:
801	case DESC_TSS:
802		/* ignore */
803		break;
804
805	default: {
806		xmaddr_t maddr = arbitrary_virt_to_machine(&dt[entry]);
807
808		xen_mc_flush();
809		if (HYPERVISOR_update_descriptor(maddr.maddr, *(u64 *)desc))
810			BUG();
811	}
812
813	}
814
815	preempt_enable();
816}
817
818/*
819 * Version of write_gdt_entry for use at early boot-time needed to
820 * update an entry as simply as possible.
821 */
822static void __init xen_write_gdt_entry_boot(struct desc_struct *dt, int entry,
823					    const void *desc, int type)
824{
825	trace_xen_cpu_write_gdt_entry(dt, entry, desc, type);
826
827	switch (type) {
828	case DESC_LDT:
829	case DESC_TSS:
830		/* ignore */
831		break;
832
833	default: {
834		xmaddr_t maddr = virt_to_machine(&dt[entry]);
835
836		if (HYPERVISOR_update_descriptor(maddr.maddr, *(u64 *)desc))
837			dt[entry] = *(struct desc_struct *)desc;
838	}
839
840	}
841}
842
843static void xen_load_sp0(unsigned long sp0)
844{
845	struct multicall_space mcs;
846
847	mcs = xen_mc_entry(0);
848	MULTI_stack_switch(mcs.mc, __KERNEL_DS, sp0);
849	xen_mc_issue(PARAVIRT_LAZY_CPU);
850	this_cpu_write(cpu_tss_rw.x86_tss.sp0, sp0);
851}
852
853#ifdef CONFIG_X86_IOPL_IOPERM
854static void xen_invalidate_io_bitmap(void)
855{
856	struct physdev_set_iobitmap iobitmap = {
857		.bitmap = NULL,
858		.nr_ports = 0,
859	};
860
861	native_tss_invalidate_io_bitmap();
862	HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap, &iobitmap);
863}
864
865static void xen_update_io_bitmap(void)
866{
867	struct physdev_set_iobitmap iobitmap;
868	struct tss_struct *tss = this_cpu_ptr(&cpu_tss_rw);
869
870	native_tss_update_io_bitmap();
871
872	iobitmap.bitmap = (uint8_t *)(&tss->x86_tss) +
873			  tss->x86_tss.io_bitmap_base;
874	if (tss->x86_tss.io_bitmap_base == IO_BITMAP_OFFSET_INVALID)
875		iobitmap.nr_ports = 0;
876	else
877		iobitmap.nr_ports = IO_BITMAP_BITS;
878
879	HYPERVISOR_physdev_op(PHYSDEVOP_set_iobitmap, &iobitmap);
880}
881#endif
882
883static void xen_io_delay(void)
884{
885}
886
887static DEFINE_PER_CPU(unsigned long, xen_cr0_value);
888
889static unsigned long xen_read_cr0(void)
890{
891	unsigned long cr0 = this_cpu_read(xen_cr0_value);
892
893	if (unlikely(cr0 == 0)) {
894		cr0 = native_read_cr0();
895		this_cpu_write(xen_cr0_value, cr0);
896	}
897
898	return cr0;
899}
900
901static void xen_write_cr0(unsigned long cr0)
902{
903	struct multicall_space mcs;
904
905	this_cpu_write(xen_cr0_value, cr0);
906
907	/* Only pay attention to cr0.TS; everything else is
908	   ignored. */
909	mcs = xen_mc_entry(0);
910
911	MULTI_fpu_taskswitch(mcs.mc, (cr0 & X86_CR0_TS) != 0);
912
913	xen_mc_issue(PARAVIRT_LAZY_CPU);
914}
915
916static void xen_write_cr4(unsigned long cr4)
917{
918	cr4 &= ~(X86_CR4_PGE | X86_CR4_PSE | X86_CR4_PCE);
919
920	native_write_cr4(cr4);
921}
922
923static u64 xen_read_msr_safe(unsigned int msr, int *err)
924{
925	u64 val;
926
927	if (pmu_msr_read(msr, &val, err))
928		return val;
929
930	val = native_read_msr_safe(msr, err);
931	switch (msr) {
932	case MSR_IA32_APICBASE:
933		val &= ~X2APIC_ENABLE;
934		break;
935	}
936	return val;
937}
938
939static int xen_write_msr_safe(unsigned int msr, unsigned low, unsigned high)
940{
941	int ret;
942	unsigned int which;
943	u64 base;
944
945	ret = 0;
946
947	switch (msr) {
948	case MSR_FS_BASE:		which = SEGBASE_FS; goto set;
949	case MSR_KERNEL_GS_BASE:	which = SEGBASE_GS_USER; goto set;
950	case MSR_GS_BASE:		which = SEGBASE_GS_KERNEL; goto set;
951
952	set:
953		base = ((u64)high << 32) | low;
954		if (HYPERVISOR_set_segment_base(which, base) != 0)
955			ret = -EIO;
956		break;
957
958	case MSR_STAR:
959	case MSR_CSTAR:
960	case MSR_LSTAR:
961	case MSR_SYSCALL_MASK:
962	case MSR_IA32_SYSENTER_CS:
963	case MSR_IA32_SYSENTER_ESP:
964	case MSR_IA32_SYSENTER_EIP:
965		/* Fast syscall setup is all done in hypercalls, so
966		   these are all ignored.  Stub them out here to stop
967		   Xen console noise. */
968		break;
969
970	default:
971		if (!pmu_msr_write(msr, low, high, &ret))
972			ret = native_write_msr_safe(msr, low, high);
973	}
974
975	return ret;
976}
977
978static u64 xen_read_msr(unsigned int msr)
979{
980	/*
981	 * This will silently swallow a #GP from RDMSR.  It may be worth
982	 * changing that.
983	 */
984	int err;
985
986	return xen_read_msr_safe(msr, &err);
987}
988
989static void xen_write_msr(unsigned int msr, unsigned low, unsigned high)
990{
991	/*
992	 * This will silently swallow a #GP from WRMSR.  It may be worth
993	 * changing that.
994	 */
995	xen_write_msr_safe(msr, low, high);
996}
997
998/* This is called once we have the cpu_possible_mask */
999void __init xen_setup_vcpu_info_placement(void)
1000{
1001	int cpu;
1002
1003	for_each_possible_cpu(cpu) {
1004		/* Set up direct vCPU id mapping for PV guests. */
1005		per_cpu(xen_vcpu_id, cpu) = cpu;
1006
1007		/*
1008		 * xen_vcpu_setup(cpu) can fail  -- in which case it
1009		 * falls back to the shared_info version for cpus
1010		 * where xen_vcpu_nr(cpu) < MAX_VIRT_CPUS.
1011		 *
1012		 * xen_cpu_up_prepare_pv() handles the rest by failing
1013		 * them in hotplug.
1014		 */
1015		(void) xen_vcpu_setup(cpu);
1016	}
1017
1018	/*
1019	 * xen_vcpu_setup managed to place the vcpu_info within the
1020	 * percpu area for all cpus, so make use of it.
1021	 */
1022	if (xen_have_vcpu_info_placement) {
1023		pv_ops.irq.save_fl = __PV_IS_CALLEE_SAVE(xen_save_fl_direct);
1024		pv_ops.irq.restore_fl =
1025			__PV_IS_CALLEE_SAVE(xen_restore_fl_direct);
1026		pv_ops.irq.irq_disable =
1027			__PV_IS_CALLEE_SAVE(xen_irq_disable_direct);
1028		pv_ops.irq.irq_enable =
1029			__PV_IS_CALLEE_SAVE(xen_irq_enable_direct);
1030		pv_ops.mmu.read_cr2 =
1031			__PV_IS_CALLEE_SAVE(xen_read_cr2_direct);
1032	}
1033}
1034
1035static const struct pv_info xen_info __initconst = {
1036	.extra_user_64bit_cs = FLAT_USER_CS64,
1037	.name = "Xen",
1038};
1039
1040static const struct pv_cpu_ops xen_cpu_ops __initconst = {
1041	.cpuid = xen_cpuid,
1042
1043	.set_debugreg = xen_set_debugreg,
1044	.get_debugreg = xen_get_debugreg,
1045
1046	.read_cr0 = xen_read_cr0,
1047	.write_cr0 = xen_write_cr0,
1048
1049	.write_cr4 = xen_write_cr4,
1050
1051	.wbinvd = native_wbinvd,
1052
1053	.read_msr = xen_read_msr,
1054	.write_msr = xen_write_msr,
1055
1056	.read_msr_safe = xen_read_msr_safe,
1057	.write_msr_safe = xen_write_msr_safe,
1058
1059	.read_pmc = xen_read_pmc,
1060
1061	.iret = xen_iret,
1062	.usergs_sysret64 = xen_sysret64,
1063
1064	.load_tr_desc = paravirt_nop,
1065	.set_ldt = xen_set_ldt,
1066	.load_gdt = xen_load_gdt,
1067	.load_idt = xen_load_idt,
1068	.load_tls = xen_load_tls,
1069	.load_gs_index = xen_load_gs_index,
1070
1071	.alloc_ldt = xen_alloc_ldt,
1072	.free_ldt = xen_free_ldt,
1073
1074	.store_tr = xen_store_tr,
1075
1076	.write_ldt_entry = xen_write_ldt_entry,
1077	.write_gdt_entry = xen_write_gdt_entry,
1078	.write_idt_entry = xen_write_idt_entry,
1079	.load_sp0 = xen_load_sp0,
1080
1081#ifdef CONFIG_X86_IOPL_IOPERM
1082	.invalidate_io_bitmap = xen_invalidate_io_bitmap,
1083	.update_io_bitmap = xen_update_io_bitmap,
1084#endif
1085	.io_delay = xen_io_delay,
1086
1087	.start_context_switch = paravirt_start_context_switch,
1088	.end_context_switch = xen_end_context_switch,
1089};
1090
1091static void xen_restart(char *msg)
1092{
1093	xen_reboot(SHUTDOWN_reboot);
1094}
1095
1096static void xen_machine_halt(void)
1097{
1098	xen_reboot(SHUTDOWN_poweroff);
1099}
1100
1101static void xen_machine_power_off(void)
1102{
1103	if (pm_power_off)
1104		pm_power_off();
1105	xen_reboot(SHUTDOWN_poweroff);
1106}
1107
1108static void xen_crash_shutdown(struct pt_regs *regs)
1109{
1110	xen_reboot(SHUTDOWN_crash);
1111}
1112
1113static const struct machine_ops xen_machine_ops __initconst = {
1114	.restart = xen_restart,
1115	.halt = xen_machine_halt,
1116	.power_off = xen_machine_power_off,
1117	.shutdown = xen_machine_halt,
1118	.crash_shutdown = xen_crash_shutdown,
1119	.emergency_restart = xen_emergency_restart,
1120};
1121
1122static unsigned char xen_get_nmi_reason(void)
1123{
1124	unsigned char reason = 0;
1125
1126	/* Construct a value which looks like it came from port 0x61. */
1127	if (test_bit(_XEN_NMIREASON_io_error,
1128		     &HYPERVISOR_shared_info->arch.nmi_reason))
1129		reason |= NMI_REASON_IOCHK;
1130	if (test_bit(_XEN_NMIREASON_pci_serr,
1131		     &HYPERVISOR_shared_info->arch.nmi_reason))
1132		reason |= NMI_REASON_SERR;
1133
1134	return reason;
1135}
1136
1137static void __init xen_boot_params_init_edd(void)
1138{
1139#if IS_ENABLED(CONFIG_EDD)
1140	struct xen_platform_op op;
1141	struct edd_info *edd_info;
1142	u32 *mbr_signature;
1143	unsigned nr;
1144	int ret;
1145
1146	edd_info = boot_params.eddbuf;
1147	mbr_signature = boot_params.edd_mbr_sig_buffer;
1148
1149	op.cmd = XENPF_firmware_info;
1150
1151	op.u.firmware_info.type = XEN_FW_DISK_INFO;
1152	for (nr = 0; nr < EDDMAXNR; nr++) {
1153		struct edd_info *info = edd_info + nr;
1154
1155		op.u.firmware_info.index = nr;
1156		info->params.length = sizeof(info->params);
1157		set_xen_guest_handle(op.u.firmware_info.u.disk_info.edd_params,
1158				     &info->params);
1159		ret = HYPERVISOR_platform_op(&op);
1160		if (ret)
1161			break;
1162
1163#define C(x) info->x = op.u.firmware_info.u.disk_info.x
1164		C(device);
1165		C(version);
1166		C(interface_support);
1167		C(legacy_max_cylinder);
1168		C(legacy_max_head);
1169		C(legacy_sectors_per_track);
1170#undef C
1171	}
1172	boot_params.eddbuf_entries = nr;
1173
1174	op.u.firmware_info.type = XEN_FW_DISK_MBR_SIGNATURE;
1175	for (nr = 0; nr < EDD_MBR_SIG_MAX; nr++) {
1176		op.u.firmware_info.index = nr;
1177		ret = HYPERVISOR_platform_op(&op);
1178		if (ret)
1179			break;
1180		mbr_signature[nr] = op.u.firmware_info.u.disk_mbr_signature.mbr_signature;
1181	}
1182	boot_params.edd_mbr_sig_buf_entries = nr;
1183#endif
1184}
1185
1186/*
1187 * Set up the GDT and segment registers for -fstack-protector.  Until
1188 * we do this, we have to be careful not to call any stack-protected
1189 * function, which is most of the kernel.
1190 */
1191static void __init xen_setup_gdt(int cpu)
1192{
1193	pv_ops.cpu.write_gdt_entry = xen_write_gdt_entry_boot;
1194	pv_ops.cpu.load_gdt = xen_load_gdt_boot;
1195
1196	setup_stack_canary_segment(cpu);
1197	switch_to_new_gdt(cpu);
1198
1199	pv_ops.cpu.write_gdt_entry = xen_write_gdt_entry;
1200	pv_ops.cpu.load_gdt = xen_load_gdt;
1201}
1202
1203static void __init xen_dom0_set_legacy_features(void)
1204{
1205	x86_platform.legacy.rtc = 1;
1206}
1207
1208static void __init xen_domu_set_legacy_features(void)
1209{
1210	x86_platform.legacy.rtc = 0;
1211}
1212
1213/* First C function to be called on Xen boot */
1214asmlinkage __visible void __init xen_start_kernel(void)
1215{
1216	struct physdev_set_iopl set_iopl;
1217	unsigned long initrd_start = 0;
1218	int rc;
1219
1220	if (!xen_start_info)
1221		return;
1222
1223	xen_domain_type = XEN_PV_DOMAIN;
1224	xen_start_flags = xen_start_info->flags;
1225
1226	xen_setup_features();
1227
1228	/* Install Xen paravirt ops */
1229	pv_info = xen_info;
1230	pv_ops.init.patch = paravirt_patch_default;
1231	pv_ops.cpu = xen_cpu_ops;
1232	xen_init_irq_ops();
1233
1234	/*
1235	 * Setup xen_vcpu early because it is needed for
1236	 * local_irq_disable(), irqs_disabled(), e.g. in printk().
1237	 *
1238	 * Don't do the full vcpu_info placement stuff until we have
1239	 * the cpu_possible_mask and a non-dummy shared_info.
1240	 */
1241	xen_vcpu_info_reset(0);
1242
1243	x86_platform.get_nmi_reason = xen_get_nmi_reason;
1244
1245	x86_init.resources.memory_setup = xen_memory_setup;
1246	x86_init.irqs.intr_mode_select	= x86_init_noop;
1247	x86_init.irqs.intr_mode_init	= x86_init_noop;
1248	x86_init.oem.arch_setup = xen_arch_setup;
1249	x86_init.oem.banner = xen_banner;
1250	x86_init.hyper.init_platform = xen_pv_init_platform;
1251	x86_init.hyper.guest_late_init = xen_pv_guest_late_init;
1252
1253	/*
1254	 * Set up some pagetable state before starting to set any ptes.
1255	 */
1256
1257	xen_setup_machphys_mapping();
1258	xen_init_mmu_ops();
1259
1260	/* Prevent unwanted bits from being set in PTEs. */
1261	__supported_pte_mask &= ~_PAGE_GLOBAL;
1262	__default_kernel_pte_mask &= ~_PAGE_GLOBAL;
1263
1264	/*
1265	 * Prevent page tables from being allocated in highmem, even
1266	 * if CONFIG_HIGHPTE is enabled.
1267	 */
1268	__userpte_alloc_gfp &= ~__GFP_HIGHMEM;
1269
1270	/* Get mfn list */
1271	xen_build_dynamic_phys_to_machine();
1272
1273	/* Work out if we support NX */
1274	get_cpu_cap(&boot_cpu_data);
1275	x86_configure_nx();
1276
1277	/*
1278	 * Set up kernel GDT and segment registers, mainly so that
1279	 * -fstack-protector code can be executed.
1280	 */
1281	xen_setup_gdt(0);
1282
1283	/* Determine virtual and physical address sizes */
1284	get_cpu_address_sizes(&boot_cpu_data);
1285
1286	/* Let's presume PV guests always boot on vCPU with id 0. */
1287	per_cpu(xen_vcpu_id, 0) = 0;
1288
1289	idt_setup_early_handler();
1290
1291	xen_init_capabilities();
1292
1293#ifdef CONFIG_X86_LOCAL_APIC
1294	/*
1295	 * set up the basic apic ops.
1296	 */
1297	xen_init_apic();
1298#endif
1299
1300	if (xen_feature(XENFEAT_mmu_pt_update_preserve_ad)) {
1301		pv_ops.mmu.ptep_modify_prot_start =
1302			xen_ptep_modify_prot_start;
1303		pv_ops.mmu.ptep_modify_prot_commit =
1304			xen_ptep_modify_prot_commit;
1305	}
1306
1307	machine_ops = xen_machine_ops;
1308
1309	/*
1310	 * The only reliable way to retain the initial address of the
1311	 * percpu gdt_page is to remember it here, so we can go and
1312	 * mark it RW later, when the initial percpu area is freed.
1313	 */
1314	xen_initial_gdt = &per_cpu(gdt_page, 0);
1315
1316	xen_smp_init();
1317
1318#ifdef CONFIG_ACPI_NUMA
1319	/*
1320	 * The pages we from Xen are not related to machine pages, so
1321	 * any NUMA information the kernel tries to get from ACPI will
1322	 * be meaningless.  Prevent it from trying.
1323	 */
1324	disable_srat();
1325#endif
1326	WARN_ON(xen_cpuhp_setup(xen_cpu_up_prepare_pv, xen_cpu_dead_pv));
1327
1328	local_irq_disable();
1329	early_boot_irqs_disabled = true;
1330
1331	xen_raw_console_write("mapping kernel into physical memory\n");
1332	xen_setup_kernel_pagetable((pgd_t *)xen_start_info->pt_base,
1333				   xen_start_info->nr_pages);
1334	xen_reserve_special_pages();
1335
1336	/*
1337	 * We used to do this in xen_arch_setup, but that is too late
1338	 * on AMD were early_cpu_init (run before ->arch_setup()) calls
1339	 * early_amd_init which pokes 0xcf8 port.
1340	 */
1341	set_iopl.iopl = 1;
1342	rc = HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
1343	if (rc != 0)
1344		xen_raw_printk("physdev_op failed %d\n", rc);
1345
1346
1347	if (xen_start_info->mod_start) {
1348	    if (xen_start_info->flags & SIF_MOD_START_PFN)
1349		initrd_start = PFN_PHYS(xen_start_info->mod_start);
1350	    else
1351		initrd_start = __pa(xen_start_info->mod_start);
1352	}
1353
1354	/* Poke various useful things into boot_params */
1355	boot_params.hdr.type_of_loader = (9 << 4) | 0;
1356	boot_params.hdr.ramdisk_image = initrd_start;
1357	boot_params.hdr.ramdisk_size = xen_start_info->mod_len;
1358	boot_params.hdr.cmd_line_ptr = __pa(xen_start_info->cmd_line);
1359	boot_params.hdr.hardware_subarch = X86_SUBARCH_XEN;
1360
1361	if (!xen_initial_domain()) {
1362		add_preferred_console("xenboot", 0, NULL);
1363		if (pci_xen)
1364			x86_init.pci.arch_init = pci_xen_init;
1365		x86_platform.set_legacy_features =
1366				xen_domu_set_legacy_features;
1367	} else {
1368		const struct dom0_vga_console_info *info =
1369			(void *)((char *)xen_start_info +
1370				 xen_start_info->console.dom0.info_off);
1371		struct xen_platform_op op = {
1372			.cmd = XENPF_firmware_info,
1373			.interface_version = XENPF_INTERFACE_VERSION,
1374			.u.firmware_info.type = XEN_FW_KBD_SHIFT_FLAGS,
1375		};
1376
1377		x86_platform.set_legacy_features =
1378				xen_dom0_set_legacy_features;
1379		xen_init_vga(info, xen_start_info->console.dom0.info_size);
1380		xen_start_info->console.domU.mfn = 0;
1381		xen_start_info->console.domU.evtchn = 0;
1382
1383		if (HYPERVISOR_platform_op(&op) == 0)
1384			boot_params.kbd_status = op.u.firmware_info.u.kbd_shift_flags;
1385
1386		/* Make sure ACS will be enabled */
1387		pci_request_acs();
1388
1389		xen_acpi_sleep_register();
1390
1391		xen_boot_params_init_edd();
1392
1393#ifdef CONFIG_ACPI
1394		/*
1395		 * Disable selecting "Firmware First mode" for correctable
1396		 * memory errors, as this is the duty of the hypervisor to
1397		 * decide.
1398		 */
1399		acpi_disable_cmcff = 1;
1400#endif
1401	}
1402
1403	if (!boot_params.screen_info.orig_video_isVGA)
1404		add_preferred_console("tty", 0, NULL);
1405	add_preferred_console("hvc", 0, NULL);
1406	if (boot_params.screen_info.orig_video_isVGA)
1407		add_preferred_console("tty", 0, NULL);
1408
1409#ifdef CONFIG_PCI
1410	/* PCI BIOS service won't work from a PV guest. */
1411	pci_probe &= ~PCI_PROBE_BIOS;
1412#endif
1413	xen_raw_console_write("about to get started...\n");
1414
1415	/* We need this for printk timestamps */
1416	xen_setup_runstate_info(0);
1417
1418	xen_efi_init(&boot_params);
1419
1420	/* Start the world */
1421	cr4_init_shadow(); /* 32b kernel does this in i386_start_kernel() */
1422	x86_64_start_reservations((char *)__pa_symbol(&boot_params));
1423}
1424
1425static int xen_cpu_up_prepare_pv(unsigned int cpu)
1426{
1427	int rc;
1428
1429	if (per_cpu(xen_vcpu, cpu) == NULL)
1430		return -ENODEV;
1431
1432	xen_setup_timer(cpu);
1433
1434	rc = xen_smp_intr_init(cpu);
1435	if (rc) {
1436		WARN(1, "xen_smp_intr_init() for CPU %d failed: %d\n",
1437		     cpu, rc);
1438		return rc;
1439	}
1440
1441	rc = xen_smp_intr_init_pv(cpu);
1442	if (rc) {
1443		WARN(1, "xen_smp_intr_init_pv() for CPU %d failed: %d\n",
1444		     cpu, rc);
1445		return rc;
1446	}
1447
1448	return 0;
1449}
1450
1451static int xen_cpu_dead_pv(unsigned int cpu)
1452{
1453	xen_smp_intr_free(cpu);
1454	xen_smp_intr_free_pv(cpu);
1455
1456	xen_teardown_timer(cpu);
1457
1458	return 0;
1459}
1460
1461static uint32_t __init xen_platform_pv(void)
1462{
1463	if (xen_pv_domain())
1464		return xen_cpuid_base();
1465
1466	return 0;
1467}
1468
1469const __initconst struct hypervisor_x86 x86_hyper_xen_pv = {
1470	.name                   = "Xen PV",
1471	.detect                 = xen_platform_pv,
1472	.type			= X86_HYPER_XEN_PV,
1473	.runtime.pin_vcpu       = xen_pin_vcpu,
1474	.ignore_nopv		= true,
1475};
1476