enlighten.c 39.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
/*
 * Core of Xen paravirt_ops implementation.
 *
 * This file contains the xen_paravirt_ops structure itself, and the
 * implementations for:
 * - privileged instructions
 * - interrupt flags
 * - segment operations
 * - booting and setup
 *
 * Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
 */

#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/smp.h>
#include <linux/preempt.h>
18
#include <linux/hardirq.h>
19 20 21 22 23 24
#include <linux/percpu.h>
#include <linux/delay.h>
#include <linux/start_kernel.h>
#include <linux/sched.h>
#include <linux/bootmem.h>
#include <linux/module.h>
25 26 27
#include <linux/mm.h>
#include <linux/page-flags.h>
#include <linux/highmem.h>
28
#include <linux/console.h>
29 30 31 32 33 34

#include <xen/interface/xen.h>
#include <xen/interface/physdev.h>
#include <xen/interface/vcpu.h>
#include <xen/features.h>
#include <xen/page.h>
J
Jeremy Fitzhardinge 已提交
35
#include <xen/hvc-console.h>
36 37 38 39 40 41 42

#include <asm/paravirt.h>
#include <asm/page.h>
#include <asm/xen/hypercall.h>
#include <asm/xen/hypervisor.h>
#include <asm/fixmap.h>
#include <asm/processor.h>
43
#include <asm/msr-index.h>
44 45 46
#include <asm/setup.h>
#include <asm/desc.h>
#include <asm/pgtable.h>
J
Jeremy Fitzhardinge 已提交
47
#include <asm/tlbflush.h>
J
Jeremy Fitzhardinge 已提交
48
#include <asm/reboot.h>
49 50

#include "xen-ops.h"
J
Jeremy Fitzhardinge 已提交
51
#include "mmu.h"
52 53 54 55 56 57
#include "multicalls.h"

EXPORT_SYMBOL_GPL(hypercall_page);

DEFINE_PER_CPU(struct vcpu_info *, xen_vcpu);
DEFINE_PER_CPU(struct vcpu_info, xen_vcpu_info);
58

59 60 61
enum xen_domain_type xen_domain_type = XEN_NATIVE;
EXPORT_SYMBOL_GPL(xen_domain_type);

62 63 64 65 66 67 68 69 70 71 72 73
/*
 * Identity map, in addition to plain kernel map.  This needs to be
 * large enough to allocate page table pages to allocate the rest.
 * Each page can map 2MB.
 */
static pte_t level1_ident_pgt[PTRS_PER_PTE * 4] __page_aligned_bss;

#ifdef CONFIG_X86_64
/* l3 pud for userspace vsyscall mapping */
static pud_t level3_user_vsyscall[PTRS_PER_PUD] __page_aligned_bss;
#endif /* CONFIG_X86_64 */

74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
/*
 * Note about cr3 (pagetable base) values:
 *
 * xen_cr3 contains the current logical cr3 value; it contains the
 * last set cr3.  This may not be the current effective cr3, because
 * its update may be being lazily deferred.  However, a vcpu looking
 * at its own cr3 can use this value knowing that it everything will
 * be self-consistent.
 *
 * xen_current_cr3 contains the actual vcpu cr3; it is set once the
 * hypercall to set the vcpu cr3 is complete (so it may be a little
 * out of date, but it will never be set early).  If one vcpu is
 * looking at another vcpu's cr3 value, it should use this variable.
 */
DEFINE_PER_CPU(unsigned long, xen_cr3);	 /* cr3 stored as physaddr */
DEFINE_PER_CPU(unsigned long, xen_current_cr3);	 /* actual vcpu cr3 */
90 91 92 93

struct start_info *xen_start_info;
EXPORT_SYMBOL_GPL(xen_start_info);

94
struct shared_info xen_dummy_shared_info;
95 96 97 98 99

/*
 * Point at some empty memory to start with. We map the real shared_info
 * page as soon as fixmap is up and running.
 */
100
struct shared_info *HYPERVISOR_shared_info = (void *)&xen_dummy_shared_info;
101 102 103 104 105 106 107 108 109 110 111 112 113 114

/*
 * Flag to determine whether vcpu info placement is available on all
 * VCPUs.  We assume it is to start with, and then set it to zero on
 * the first failure.  This is because it can succeed on some VCPUs
 * and not others, since it can involve hypervisor memory allocation,
 * or because the guest failed to guarantee all the appropriate
 * constraints on all VCPUs (ie buffer can't cross a page boundary).
 *
 * Note that any particular CPU may be using a placed vcpu structure,
 * but we can only optimise if the all are.
 *
 * 0: not available, 1: available
 */
115 116 117 118 119 120 121 122
static int have_vcpu_info_placement =
#ifdef CONFIG_X86_32
	1
#else
	0
#endif
	;

123

124
static void xen_vcpu_setup(int cpu)
125
{
126 127 128 129
	struct vcpu_register_vcpu_info info;
	int err;
	struct vcpu_info *vcpup;

130
	BUG_ON(HYPERVISOR_shared_info == &xen_dummy_shared_info);
131
	per_cpu(xen_vcpu, cpu) = &HYPERVISOR_shared_info->vcpu_info[cpu];
132 133 134 135 136 137 138 139 140

	if (!have_vcpu_info_placement)
		return;		/* already tested, not available */

	vcpup = &per_cpu(xen_vcpu_info, cpu);

	info.mfn = virt_to_mfn(vcpup);
	info.offset = offset_in_page(vcpup);

141
	printk(KERN_DEBUG "trying to map vcpu_info %d at %p, mfn %llx, offset %d\n",
142 143 144 145 146 147 148 149 150 151 152 153 154 155
	       cpu, vcpup, info.mfn, info.offset);

	/* Check to see if the hypervisor will put the vcpu_info
	   structure where we want it, which allows direct access via
	   a percpu-variable. */
	err = HYPERVISOR_vcpu_op(VCPUOP_register_vcpu_info, cpu, &info);

	if (err) {
		printk(KERN_DEBUG "register_vcpu_info failed: err=%d\n", err);
		have_vcpu_info_placement = 0;
	} else {
		/* This cpu is using the registered vcpu info, even if
		   later ones fail to. */
		per_cpu(xen_vcpu, cpu) = vcpup;
156

157 158 159
		printk(KERN_DEBUG "cpu %d using vcpu_info at %p\n",
		       cpu, vcpup);
	}
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
/*
 * On restore, set the vcpu placement up again.
 * If it fails, then we're in a bad state, since
 * we can't back out from using it...
 */
void xen_vcpu_restore(void)
{
	if (have_vcpu_info_placement) {
		int cpu;

		for_each_online_cpu(cpu) {
			bool other_cpu = (cpu != smp_processor_id());

			if (other_cpu &&
			    HYPERVISOR_vcpu_op(VCPUOP_down, cpu, NULL))
				BUG();

			xen_vcpu_setup(cpu);

			if (other_cpu &&
			    HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL))
				BUG();
		}

		BUG_ON(!have_vcpu_info_placement);
	}
}

190 191
static void __init xen_banner(void)
{
192 193 194 195
	unsigned version = HYPERVISOR_xen_version(XENVER_version, NULL);
	struct xen_extraversion extra;
	HYPERVISOR_xen_version(XENVER_extraversion, &extra);

196
	printk(KERN_INFO "Booting paravirtualized kernel on %s\n",
197
	       pv_info.name);
198 199
	printk(KERN_INFO "Xen version: %d.%d%s%s\n",
	       version >> 16, version & 0xffff, extra.extraversion,
200
	       xen_feature(XENFEAT_mmu_pt_update_preserve_ad) ? " (preserve-AD)" : "");
201 202
}

203 204
static void xen_cpuid(unsigned int *ax, unsigned int *bx,
		      unsigned int *cx, unsigned int *dx)
205 206 207 208 209 210 211
{
	unsigned maskedx = ~0;

	/*
	 * Mask out inconvenient features, to try and disable as many
	 * unsupported kernel subsystems as possible.
	 */
212
	if (*ax == 1)
213 214
		maskedx = ~((1 << X86_FEATURE_APIC) |  /* disable APIC */
			    (1 << X86_FEATURE_ACPI) |  /* disable ACPI */
215 216
			    (1 << X86_FEATURE_MCE)  |  /* disable MCE */
			    (1 << X86_FEATURE_MCA)  |  /* disable MCA */
217 218 219
			    (1 << X86_FEATURE_ACC));   /* thermal monitoring */

	asm(XEN_EMULATE_PREFIX "cpuid"
220 221 222 223 224 225
		: "=a" (*ax),
		  "=b" (*bx),
		  "=c" (*cx),
		  "=d" (*dx)
		: "0" (*ax), "2" (*cx));
	*dx &= maskedx;
226 227 228 229 230 231 232 233 234 235 236 237
}

static void xen_set_debugreg(int reg, unsigned long val)
{
	HYPERVISOR_set_debugreg(reg, val);
}

static unsigned long xen_get_debugreg(int reg)
{
	return HYPERVISOR_get_debugreg(reg);
}

238
static void xen_leave_lazy(void)
239
{
240
	paravirt_leave_lazy(paravirt_get_lazy_mode());
241 242 243 244 245 246 247 248
	xen_mc_flush();
}

static unsigned long xen_store_tr(void)
{
	return 0;
}

249
/*
250 251 252 253
 * Set the page permissions for a particular virtual address.  If the
 * address is a vmalloc mapping (or other non-linear mapping), then
 * find the linear mapping of the page and also set its protections to
 * match.
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283
 */
static void set_aliased_prot(void *v, pgprot_t prot)
{
	int level;
	pte_t *ptep;
	pte_t pte;
	unsigned long pfn;
	struct page *page;

	ptep = lookup_address((unsigned long)v, &level);
	BUG_ON(ptep == NULL);

	pfn = pte_pfn(*ptep);
	page = pfn_to_page(pfn);

	pte = pfn_pte(pfn, prot);

	if (HYPERVISOR_update_va_mapping((unsigned long)v, pte, 0))
		BUG();

	if (!PageHighMem(page)) {
		void *av = __va(PFN_PHYS(pfn));

		if (av != v)
			if (HYPERVISOR_update_va_mapping((unsigned long)av, pte, 0))
				BUG();
	} else
		kmap_flush_unused();
}

284 285
static void xen_alloc_ldt(struct desc_struct *ldt, unsigned entries)
{
286
	const unsigned entries_per_page = PAGE_SIZE / LDT_ENTRY_SIZE;
287 288
	int i;

289 290
	for(i = 0; i < entries; i += entries_per_page)
		set_aliased_prot(ldt + i, PAGE_KERNEL_RO);
291 292 293 294
}

static void xen_free_ldt(struct desc_struct *ldt, unsigned entries)
{
295
	const unsigned entries_per_page = PAGE_SIZE / LDT_ENTRY_SIZE;
296 297
	int i;

298 299
	for(i = 0; i < entries; i += entries_per_page)
		set_aliased_prot(ldt + i, PAGE_KERNEL);
300 301
}

302 303 304 305 306 307 308
static void xen_set_ldt(const void *addr, unsigned entries)
{
	struct mmuext_op *op;
	struct multicall_space mcs = xen_mc_entry(sizeof(*op));

	op = mcs.args;
	op->cmd = MMUEXT_SET_LDT;
309
	op->arg1.linear_addr = (unsigned long)addr;
310 311 312 313 314 315 316
	op->arg2.nr_ents = entries;

	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_CPU);
}

317
static void xen_load_gdt(const struct desc_ptr *dtr)
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356
{
	unsigned long *frames;
	unsigned long va = dtr->address;
	unsigned int size = dtr->size + 1;
	unsigned pages = (size + PAGE_SIZE - 1) / PAGE_SIZE;
	int f;
	struct multicall_space mcs;

	/* A GDT can be up to 64k in size, which corresponds to 8192
	   8-byte entries, or 16 4k pages.. */

	BUG_ON(size > 65536);
	BUG_ON(va & ~PAGE_MASK);

	mcs = xen_mc_entry(sizeof(*frames) * pages);
	frames = mcs.args;

	for (f = 0; va < dtr->address + size; va += PAGE_SIZE, f++) {
		frames[f] = virt_to_mfn(va);
		make_lowmem_page_readonly((void *)va);
	}

	MULTI_set_gdt(mcs.mc, frames, size / sizeof(struct desc_struct));

	xen_mc_issue(PARAVIRT_LAZY_CPU);
}

static void load_TLS_descriptor(struct thread_struct *t,
				unsigned int cpu, unsigned int i)
{
	struct desc_struct *gdt = get_cpu_gdt_table(cpu);
	xmaddr_t maddr = virt_to_machine(&gdt[GDT_ENTRY_TLS_MIN+i]);
	struct multicall_space mc = __xen_mc_entry(0);

	MULTI_update_descriptor(mc.mc, maddr.maddr, t->tls_array[i]);
}

static void xen_load_tls(struct thread_struct *t, unsigned int cpu)
{
357 358 359 360 361 362 363 364
	/*
	 * XXX sleazy hack: If we're being called in a lazy-cpu zone,
	 * it means we're in a context switch, and %gs has just been
	 * saved.  This means we can zero it out to prevent faults on
	 * exit from the hypervisor if the next process has no %gs.
	 * Either way, it has been saved, and the new value will get
	 * loaded properly.  This will go away as soon as Xen has been
	 * modified to not save/restore %gs for normal hypercalls.
365 366 367 368 369 370 371 372
	 *
	 * On x86_64, this hack is not used for %gs, because gs points
	 * to KERNEL_GS_BASE (and uses it for PDA references), so we
	 * must not zero %gs on x86_64
	 *
	 * For x86_64, we need to zero %fs, otherwise we may get an
	 * exception between the new %fs descriptor being loaded and
	 * %fs being effectively cleared at __switch_to().
373
	 */
374 375
	if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_CPU) {
#ifdef CONFIG_X86_32
376
		loadsegment(gs, 0);
377 378 379 380 381 382 383 384 385 386 387 388
#else
		loadsegment(fs, 0);
#endif
	}

	xen_mc_batch();

	load_TLS_descriptor(t, cpu, 0);
	load_TLS_descriptor(t, cpu, 1);
	load_TLS_descriptor(t, cpu, 2);

	xen_mc_issue(PARAVIRT_LAZY_CPU);
389 390
}

391 392 393 394 395
#ifdef CONFIG_X86_64
static void xen_load_gs_index(unsigned int idx)
{
	if (HYPERVISOR_set_segment_base(SEGBASE_GS_USER_SEL, idx))
		BUG();
396
}
397
#endif
398 399

static void xen_write_ldt_entry(struct desc_struct *dt, int entrynum,
400
				const void *ptr)
401
{
402
	xmaddr_t mach_lp = arbitrary_virt_to_machine(&dt[entrynum]);
403
	u64 entry = *(u64 *)ptr;
404

405 406
	preempt_disable();

407 408 409
	xen_mc_flush();
	if (HYPERVISOR_update_descriptor(mach_lp.maddr, entry))
		BUG();
410 411

	preempt_enable();
412 413
}

414
static int cvt_gate_to_trap(int vector, const gate_desc *val,
415 416
			    struct trap_info *info)
{
417
	if (val->type != 0xf && val->type != 0xe)
418 419 420
		return 0;

	info->vector = vector;
421 422 423
	info->address = gate_offset(*val);
	info->cs = gate_segment(*val);
	info->flags = val->dpl;
424
	/* interrupt gates clear IF */
425
	if (val->type == 0xe)
426 427 428 429 430 431
		info->flags |= 4;

	return 1;
}

/* Locations of each CPU's IDT */
432
static DEFINE_PER_CPU(struct desc_ptr, idt_desc);
433 434 435

/* Set an IDT entry.  If the entry is part of the current IDT, then
   also update Xen. */
436
static void xen_write_idt_entry(gate_desc *dt, int entrynum, const gate_desc *g)
437 438
{
	unsigned long p = (unsigned long)&dt[entrynum];
439 440 441 442 443 444
	unsigned long start, end;

	preempt_disable();

	start = __get_cpu_var(idt_desc).address;
	end = start + __get_cpu_var(idt_desc).size + 1;
445 446 447

	xen_mc_flush();

448
	native_write_idt_entry(dt, entrynum, g);
449 450 451 452 453 454

	if (p >= start && (p + 8) <= end) {
		struct trap_info info[2];

		info[1].address = 0;

455
		if (cvt_gate_to_trap(entrynum, g, &info[0]))
456 457 458
			if (HYPERVISOR_set_trap_table(info))
				BUG();
	}
459 460

	preempt_enable();
461 462
}

463
static void xen_convert_trap_info(const struct desc_ptr *desc,
J
Jeremy Fitzhardinge 已提交
464
				  struct trap_info *traps)
465 466 467
{
	unsigned in, out, count;

468
	count = (desc->size+1) / sizeof(gate_desc);
469 470 471
	BUG_ON(count > 256);

	for (in = out = 0; in < count; in++) {
472
		gate_desc *entry = (gate_desc*)(desc->address) + in;
473

474
		if (cvt_gate_to_trap(in, entry, &traps[out]))
475 476 477
			out++;
	}
	traps[out].address = 0;
J
Jeremy Fitzhardinge 已提交
478 479 480 481
}

void xen_copy_trap_info(struct trap_info *traps)
{
482
	const struct desc_ptr *desc = &__get_cpu_var(idt_desc);
J
Jeremy Fitzhardinge 已提交
483 484 485 486 487 488 489

	xen_convert_trap_info(desc, traps);
}

/* Load a new IDT into Xen.  In principle this can be per-CPU, so we
   hold a spinlock to protect the static traps[] array (static because
   it avoids allocation, and saves stack space). */
490
static void xen_load_idt(const struct desc_ptr *desc)
J
Jeremy Fitzhardinge 已提交
491 492 493 494 495 496
{
	static DEFINE_SPINLOCK(lock);
	static struct trap_info traps[257];

	spin_lock(&lock);

497 498
	__get_cpu_var(idt_desc) = *desc;

J
Jeremy Fitzhardinge 已提交
499
	xen_convert_trap_info(desc, traps);
500 501 502 503 504 505 506 507 508 509 510

	xen_mc_flush();
	if (HYPERVISOR_set_trap_table(traps))
		BUG();

	spin_unlock(&lock);
}

/* Write a GDT descriptor entry.  Ignore LDT descriptors, since
   they're handled differently. */
static void xen_write_gdt_entry(struct desc_struct *dt, int entry,
511
				const void *desc, int type)
512
{
513 514
	preempt_disable();

515 516 517
	switch (type) {
	case DESC_LDT:
	case DESC_TSS:
518 519 520 521 522 523 524
		/* ignore */
		break;

	default: {
		xmaddr_t maddr = virt_to_machine(&dt[entry]);

		xen_mc_flush();
525
		if (HYPERVISOR_update_descriptor(maddr.maddr, *(u64 *)desc))
526 527 528 529
			BUG();
	}

	}
530 531

	preempt_enable();
532 533
}

534
static void xen_load_sp0(struct tss_struct *tss,
535
			 struct thread_struct *thread)
536 537
{
	struct multicall_space mcs = xen_mc_entry(0);
538
	MULTI_stack_switch(mcs.mc, __KERNEL_DS, thread->sp0);
539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555
	xen_mc_issue(PARAVIRT_LAZY_CPU);
}

static void xen_set_iopl_mask(unsigned mask)
{
	struct physdev_set_iopl set_iopl;

	/* Force the change at ring 0. */
	set_iopl.iopl = (mask == 0) ? 1 : (mask >> 12) & 3;
	HYPERVISOR_physdev_op(PHYSDEVOP_set_iopl, &set_iopl);
}

static void xen_io_delay(void)
{
}

#ifdef CONFIG_X86_LOCAL_APIC
556
static u32 xen_apic_read(unsigned long reg)
557 558 559
{
	return 0;
}
J
Jeremy Fitzhardinge 已提交
560

561
static void xen_apic_write(unsigned long reg, u32 val)
J
Jeremy Fitzhardinge 已提交
562 563 564 565
{
	/* Warn to see if there's any stray references */
	WARN_ON(1);
}
566 567 568 569
#endif

static void xen_flush_tlb(void)
{
J
Jeremy Fitzhardinge 已提交
570
	struct mmuext_op *op;
571 572 573 574 575
	struct multicall_space mcs;

	preempt_disable();

	mcs = xen_mc_entry(sizeof(*op));
576

J
Jeremy Fitzhardinge 已提交
577 578 579 580 581
	op = mcs.args;
	op->cmd = MMUEXT_TLB_FLUSH_LOCAL;
	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
582 583

	preempt_enable();
584 585 586 587
}

static void xen_flush_tlb_single(unsigned long addr)
{
J
Jeremy Fitzhardinge 已提交
588
	struct mmuext_op *op;
589 590 591
	struct multicall_space mcs;

	preempt_disable();
592

593
	mcs = xen_mc_entry(sizeof(*op));
J
Jeremy Fitzhardinge 已提交
594 595 596 597 598 599
	op = mcs.args;
	op->cmd = MMUEXT_INVLPG_LOCAL;
	op->arg1.linear_addr = addr & PAGE_MASK;
	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
600 601

	preempt_enable();
602 603
}

J
Jeremy Fitzhardinge 已提交
604 605 606
static void xen_flush_tlb_others(const cpumask_t *cpus, struct mm_struct *mm,
				 unsigned long va)
{
J
Jeremy Fitzhardinge 已提交
607 608 609 610
	struct {
		struct mmuext_op op;
		cpumask_t mask;
	} *args;
J
Jeremy Fitzhardinge 已提交
611
	cpumask_t cpumask = *cpus;
J
Jeremy Fitzhardinge 已提交
612
	struct multicall_space mcs;
J
Jeremy Fitzhardinge 已提交
613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628

	/*
	 * A couple of (to be removed) sanity checks:
	 *
	 * - current CPU must not be in mask
	 * - mask must exist :)
	 */
	BUG_ON(cpus_empty(cpumask));
	BUG_ON(cpu_isset(smp_processor_id(), cpumask));
	BUG_ON(!mm);

	/* If a CPU which we ran on has gone down, OK. */
	cpus_and(cpumask, cpumask, cpu_online_map);
	if (cpus_empty(cpumask))
		return;

J
Jeremy Fitzhardinge 已提交
629 630 631 632 633
	mcs = xen_mc_entry(sizeof(*args));
	args = mcs.args;
	args->mask = cpumask;
	args->op.arg2.vcpumask = &args->mask;

J
Jeremy Fitzhardinge 已提交
634
	if (va == TLB_FLUSH_ALL) {
J
Jeremy Fitzhardinge 已提交
635
		args->op.cmd = MMUEXT_TLB_FLUSH_MULTI;
J
Jeremy Fitzhardinge 已提交
636
	} else {
J
Jeremy Fitzhardinge 已提交
637 638
		args->op.cmd = MMUEXT_INVLPG_MULTI;
		args->op.arg1.linear_addr = va;
J
Jeremy Fitzhardinge 已提交
639 640
	}

J
Jeremy Fitzhardinge 已提交
641 642 643
	MULTI_mmuext_op(mcs.mc, &args->op, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
J
Jeremy Fitzhardinge 已提交
644 645
}

646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669
static void xen_clts(void)
{
	struct multicall_space mcs;

	mcs = xen_mc_entry(0);

	MULTI_fpu_taskswitch(mcs.mc, 0);

	xen_mc_issue(PARAVIRT_LAZY_CPU);
}

static void xen_write_cr0(unsigned long cr0)
{
	struct multicall_space mcs;

	/* Only pay attention to cr0.TS; everything else is
	   ignored. */
	mcs = xen_mc_entry(0);

	MULTI_fpu_taskswitch(mcs.mc, (cr0 & X86_CR0_TS) != 0);

	xen_mc_issue(PARAVIRT_LAZY_CPU);
}

670 671 672 673 674
static void xen_write_cr2(unsigned long cr2)
{
	x86_read_percpu(xen_vcpu)->arch.cr2 = cr2;
}

675 676 677 678 679
static unsigned long xen_read_cr2(void)
{
	return x86_read_percpu(xen_vcpu)->arch.cr2;
}

680 681 682 683 684
static unsigned long xen_read_cr2_direct(void)
{
	return x86_read_percpu(xen_vcpu_info.arch.cr2);
}

685 686
static void xen_write_cr4(unsigned long cr4)
{
687 688 689 690
	cr4 &= ~X86_CR4_PGE;
	cr4 &= ~X86_CR4_PSE;

	native_write_cr4(cr4);
691 692 693 694 695 696 697
}

static unsigned long xen_read_cr3(void)
{
	return x86_read_percpu(xen_cr3);
}

698 699 700 701 702
static void set_current_cr3(void *v)
{
	x86_write_percpu(xen_current_cr3, (unsigned long)v);
}

703
static void __xen_write_cr3(bool kernel, unsigned long cr3)
704
{
705 706
	struct mmuext_op *op;
	struct multicall_space mcs;
707
	unsigned long mfn;
708

709 710 711 712
	if (cr3)
		mfn = pfn_to_mfn(PFN_DOWN(cr3));
	else
		mfn = 0;
713

714
	WARN_ON(mfn == 0 && kernel);
715

716
	mcs = __xen_mc_entry(sizeof(*op));
717

718
	op = mcs.args;
719
	op->cmd = kernel ? MMUEXT_NEW_BASEPTR : MMUEXT_NEW_USER_BASEPTR;
720
	op->arg1.mfn = mfn;
721

722
	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
723

724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753
	if (kernel) {
		x86_write_percpu(xen_cr3, cr3);

		/* Update xen_current_cr3 once the batch has actually
		   been submitted. */
		xen_mc_callback(set_current_cr3, (void *)cr3);
	}
}

static void xen_write_cr3(unsigned long cr3)
{
	BUG_ON(preemptible());

	xen_mc_batch();  /* disables interrupts */

	/* Update while interrupts are disabled, so its atomic with
	   respect to ipis */
	x86_write_percpu(xen_cr3, cr3);

	__xen_write_cr3(true, cr3);

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(__va(cr3));
		if (user_pgd)
			__xen_write_cr3(false, __pa(user_pgd));
		else
			__xen_write_cr3(false, 0);
	}
#endif
754

755
	xen_mc_issue(PARAVIRT_LAZY_CPU);  /* interrupts restored */
756 757
}

758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778
static int xen_write_msr_safe(unsigned int msr, unsigned low, unsigned high)
{
	int ret;

	ret = 0;

	switch(msr) {
#ifdef CONFIG_X86_64
		unsigned which;
		u64 base;

	case MSR_FS_BASE:		which = SEGBASE_FS; goto set;
	case MSR_KERNEL_GS_BASE:	which = SEGBASE_GS_USER; goto set;
	case MSR_GS_BASE:		which = SEGBASE_GS_KERNEL; goto set;

	set:
		base = ((u64)high << 32) | low;
		if (HYPERVISOR_set_segment_base(which, base) != 0)
			ret = -EFAULT;
		break;
#endif
J
Jeremy Fitzhardinge 已提交
779 780 781 782 783 784 785 786 787 788 789 790 791

	case MSR_STAR:
	case MSR_CSTAR:
	case MSR_LSTAR:
	case MSR_SYSCALL_MASK:
	case MSR_IA32_SYSENTER_CS:
	case MSR_IA32_SYSENTER_ESP:
	case MSR_IA32_SYSENTER_EIP:
		/* Fast syscall setup is all done in hypercalls, so
		   these are all ignored.  Stub them out here to stop
		   Xen console noise. */
		break;

792 793 794 795 796 797 798
	default:
		ret = native_write_msr_safe(msr, low, high);
	}

	return ret;
}

799 800
/* Early in boot, while setting up the initial pagetable, assume
   everything is pinned. */
801
static __init void xen_alloc_pte_init(struct mm_struct *mm, u32 pfn)
802
{
803
#ifdef CONFIG_FLATMEM
804
	BUG_ON(mem_map);	/* should only be used early */
805
#endif
806 807 808
	make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
}

809
/* Early release_pte assumes that all pts are pinned, since there's
810
   only init_mm and anything attached to that is pinned. */
811
static void xen_release_pte_init(u32 pfn)
812 813 814 815
{
	make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
}

816
static void pin_pagetable_pfn(unsigned cmd, unsigned long pfn)
817 818
{
	struct mmuext_op op;
819
	op.cmd = cmd;
820 821 822 823 824
	op.arg1.mfn = pfn_to_mfn(pfn);
	if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
		BUG();
}

825 826
/* This needs to make sure the new pte page is pinned iff its being
   attached to a pinned pagetable. */
827
static void xen_alloc_ptpage(struct mm_struct *mm, u32 pfn, unsigned level)
828
{
829
	struct page *page = pfn_to_page(pfn);
830

831 832 833
	if (PagePinned(virt_to_page(mm->pgd))) {
		SetPagePinned(page);

834
		if (!PageHighMem(page)) {
835
			make_lowmem_page_readonly(__va(PFN_PHYS((unsigned long)pfn)));
836
			if (level == PT_PTE && USE_SPLIT_PTLOCKS)
837
				pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);
838
		} else
839 840 841 842
			/* make sure there are no stray mappings of
			   this page */
			kmap_flush_unused();
	}
843 844
}

845
static void xen_alloc_pte(struct mm_struct *mm, u32 pfn)
846
{
847
	xen_alloc_ptpage(mm, pfn, PT_PTE);
848 849
}

850
static void xen_alloc_pmd(struct mm_struct *mm, u32 pfn)
851
{
852
	xen_alloc_ptpage(mm, pfn, PT_PMD);
853 854
}

855 856 857 858 859 860 861 862 863 864
static int xen_pgd_alloc(struct mm_struct *mm)
{
	pgd_t *pgd = mm->pgd;
	int ret = 0;

	BUG_ON(PagePinned(virt_to_page(pgd)));

#ifdef CONFIG_X86_64
	{
		struct page *page = virt_to_page(pgd);
865
		pgd_t *user_pgd;
866 867 868

		BUG_ON(page->private != 0);

869 870 871 872 873 874 875 876 877 878
		ret = -ENOMEM;

		user_pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
		page->private = (unsigned long)user_pgd;

		if (user_pgd != NULL) {
			user_pgd[pgd_index(VSYSCALL_START)] =
				__pgd(__pa(level3_user_vsyscall) | _PAGE_TABLE);
			ret = 0;
		}
879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896

		BUG_ON(PagePinned(virt_to_page(xen_get_user_pgd(pgd))));
	}
#endif

	return ret;
}

static void xen_pgd_free(struct mm_struct *mm, pgd_t *pgd)
{
#ifdef CONFIG_X86_64
	pgd_t *user_pgd = xen_get_user_pgd(pgd);

	if (user_pgd)
		free_page((unsigned long)user_pgd);
#endif
}

897
/* This should never happen until we're OK to use struct page */
898
static void xen_release_ptpage(u32 pfn, unsigned level)
899
{
900 901 902
	struct page *page = pfn_to_page(pfn);

	if (PagePinned(page)) {
903
		if (!PageHighMem(page)) {
904
			if (level == PT_PTE && USE_SPLIT_PTLOCKS)
905
				pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);
906
			make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
907
		}
908
		ClearPagePinned(page);
909
	}
910 911
}

912
static void xen_release_pte(u32 pfn)
913 914 915 916
{
	xen_release_ptpage(pfn, PT_PTE);
}

917
static void xen_release_pmd(u32 pfn)
918 919 920 921
{
	xen_release_ptpage(pfn, PT_PMD);
}

922 923 924 925 926 927 928 929 930 931 932 933
#if PAGETABLE_LEVELS == 4
static void xen_alloc_pud(struct mm_struct *mm, u32 pfn)
{
	xen_alloc_ptpage(mm, pfn, PT_PUD);
}

static void xen_release_pud(u32 pfn)
{
	xen_release_ptpage(pfn, PT_PUD);
}
#endif

934 935
#ifdef CONFIG_HIGHPTE
static void *xen_kmap_atomic_pte(struct page *page, enum km_type type)
936
{
937 938 939 940 941 942 943 944 945 946 947
	pgprot_t prot = PAGE_KERNEL;

	if (PagePinned(page))
		prot = PAGE_KERNEL_RO;

	if (0 && PageHighMem(page))
		printk("mapping highpte %lx type %d prot %s\n",
		       page_to_pfn(page), type,
		       (unsigned long)pgprot_val(prot) & _PAGE_RW ? "WRITE" : "READ");

	return kmap_atomic_prot(page, type, prot);
948
}
949
#endif
950

951
#ifdef CONFIG_X86_32
952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969
static __init pte_t mask_rw_pte(pte_t *ptep, pte_t pte)
{
	/* If there's an existing pte, then don't allow _PAGE_RW to be set */
	if (pte_val_ma(*ptep) & _PAGE_PRESENT)
		pte = __pte_ma(((pte_val_ma(*ptep) & _PAGE_RW) | ~_PAGE_RW) &
			       pte_val_ma(pte));

	return pte;
}

/* Init-time set_pte while constructing initial pagetables, which
   doesn't allow RO pagetable pages to be remapped RW */
static __init void xen_set_pte_init(pte_t *ptep, pte_t pte)
{
	pte = mask_rw_pte(ptep, pte);

	xen_set_pte(ptep, pte);
}
970
#endif
971

972 973 974 975
static __init void xen_pagetable_setup_start(pgd_t *base)
{
}

976
void xen_setup_shared_info(void)
977 978
{
	if (!xen_feature(XENFEAT_auto_translated_physmap)) {
979 980 981 982 983
		set_fixmap(FIX_PARAVIRT_BOOTMAP,
			   xen_start_info->shared_info);

		HYPERVISOR_shared_info =
			(struct shared_info *)fix_to_virt(FIX_PARAVIRT_BOOTMAP);
984 985 986 987
	} else
		HYPERVISOR_shared_info =
			(struct shared_info *)__va(xen_start_info->shared_info);

988 989 990 991
#ifndef CONFIG_SMP
	/* In UP this is as good a place as any to set up shared info */
	xen_setup_vcpu_info_placement();
#endif
J
Jeremy Fitzhardinge 已提交
992 993

	xen_setup_mfn_list_list();
994 995 996 997
}

static __init void xen_pagetable_setup_done(pgd_t *base)
{
998
	xen_setup_shared_info();
999
}
1000

1001 1002
static __init void xen_post_allocator_init(void)
{
1003
	pv_mmu_ops.set_pte = xen_set_pte;
1004 1005
	pv_mmu_ops.set_pmd = xen_set_pmd;
	pv_mmu_ops.set_pud = xen_set_pud;
1006 1007 1008
#if PAGETABLE_LEVELS == 4
	pv_mmu_ops.set_pgd = xen_set_pgd;
#endif
1009

1010 1011
	/* This will work as long as patching hasn't happened yet
	   (which it hasn't) */
1012 1013 1014 1015
	pv_mmu_ops.alloc_pte = xen_alloc_pte;
	pv_mmu_ops.alloc_pmd = xen_alloc_pmd;
	pv_mmu_ops.release_pte = xen_release_pte;
	pv_mmu_ops.release_pmd = xen_release_pmd;
1016 1017 1018 1019
#if PAGETABLE_LEVELS == 4
	pv_mmu_ops.alloc_pud = xen_alloc_pud;
	pv_mmu_ops.release_pud = xen_release_pud;
#endif
1020

1021 1022 1023
#ifdef CONFIG_X86_64
	SetPagePinned(virt_to_page(level3_user_vsyscall));
#endif
1024 1025 1026
	xen_mark_init_mm_pinned();
}

1027
/* This is called once we have the cpu_possible_map */
1028
void xen_setup_vcpu_info_placement(void)
1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
{
	int cpu;

	for_each_possible_cpu(cpu)
		xen_vcpu_setup(cpu);

	/* xen_vcpu_setup managed to place the vcpu_info within the
	   percpu area for all cpus, so make use of it */
	if (have_vcpu_info_placement) {
		printk(KERN_INFO "Xen: using vcpu_info placement\n");

1040 1041 1042 1043 1044
		pv_irq_ops.save_fl = xen_save_fl_direct;
		pv_irq_ops.restore_fl = xen_restore_fl_direct;
		pv_irq_ops.irq_disable = xen_irq_disable_direct;
		pv_irq_ops.irq_enable = xen_irq_enable_direct;
		pv_mmu_ops.read_cr2 = xen_read_cr2_direct;
1045
	}
1046 1047
}

1048 1049
static unsigned xen_patch(u8 type, u16 clobbers, void *insnbuf,
			  unsigned long addr, unsigned len)
1050 1051 1052 1053 1054 1055
{
	char *start, *end, *reloc;
	unsigned ret;

	start = end = reloc = NULL;

1056 1057
#define SITE(op, x)							\
	case PARAVIRT_PATCH(op.x):					\
1058 1059 1060 1061 1062 1063 1064 1065
	if (have_vcpu_info_placement) {					\
		start = (char *)xen_##x##_direct;			\
		end = xen_##x##_direct_end;				\
		reloc = xen_##x##_direct_reloc;				\
	}								\
	goto patch_site

	switch (type) {
1066 1067 1068 1069
		SITE(pv_irq_ops, irq_enable);
		SITE(pv_irq_ops, irq_disable);
		SITE(pv_irq_ops, save_fl);
		SITE(pv_irq_ops, restore_fl);
1070 1071 1072 1073 1074 1075
#undef SITE

	patch_site:
		if (start == NULL || (end-start) > len)
			goto default_patch;

1076
		ret = paravirt_patch_insns(insnbuf, len, start, end);
1077 1078 1079 1080 1081 1082 1083

		/* Note: because reloc is assigned from something that
		   appears to be an array, gcc assumes it's non-null,
		   but doesn't know its relationship with start and
		   end. */
		if (reloc > start && reloc < end) {
			int reloc_off = reloc - start;
1084 1085
			long *relocp = (long *)(insnbuf + reloc_off);
			long delta = start - (char *)addr;
1086 1087 1088 1089 1090 1091 1092

			*relocp += delta;
		}
		break;

	default_patch:
	default:
1093 1094
		ret = paravirt_patch_default(type, clobbers, insnbuf,
					     addr, len);
1095 1096 1097 1098 1099 1100
		break;
	}

	return ret;
}

1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
static void xen_set_fixmap(unsigned idx, unsigned long phys, pgprot_t prot)
{
	pte_t pte;

	phys >>= PAGE_SHIFT;

	switch (idx) {
	case FIX_BTMAP_END ... FIX_BTMAP_BEGIN:
#ifdef CONFIG_X86_F00F_BUG
	case FIX_F00F_IDT:
#endif
1112
#ifdef CONFIG_X86_32
1113 1114
	case FIX_WP_TEST:
	case FIX_VDSO:
1115
# ifdef CONFIG_HIGHMEM
1116
	case FIX_KMAP_BEGIN ... FIX_KMAP_END:
1117
# endif
1118 1119 1120
#else
	case VSYSCALL_LAST_PAGE ... VSYSCALL_FIRST_PAGE:
#endif
1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132
#ifdef CONFIG_X86_LOCAL_APIC
	case FIX_APIC_BASE:	/* maps dummy local APIC */
#endif
		pte = pfn_pte(phys, prot);
		break;

	default:
		pte = mfn_pte(phys, prot);
		break;
	}

	__native_set_fixmap(idx, pte);
1133 1134 1135 1136 1137 1138 1139 1140 1141

#ifdef CONFIG_X86_64
	/* Replicate changes to map the vsyscall page into the user
	   pagetable vsyscall mapping. */
	if (idx >= VSYSCALL_LAST_PAGE && idx <= VSYSCALL_FIRST_PAGE) {
		unsigned long vaddr = __fix_to_virt(idx);
		set_pte_vaddr_pud(level3_user_vsyscall, vaddr, pte);
	}
#endif
1142 1143
}

1144
static const struct pv_info xen_info __initdata = {
1145 1146 1147 1148
	.paravirt_enabled = 1,
	.shared_kernel_pmd = 0,

	.name = "Xen",
1149
};
1150

1151
static const struct pv_init_ops xen_init_ops __initdata = {
1152
	.patch = xen_patch,
1153

1154
	.banner = xen_banner,
1155 1156
	.memory_setup = xen_memory_setup,
	.arch_setup = xen_arch_setup,
1157
	.post_allocator_init = xen_post_allocator_init,
1158
};
1159

1160
static const struct pv_time_ops xen_time_ops __initdata = {
J
Jeremy Fitzhardinge 已提交
1161
	.time_init = xen_time_init,
1162

J
Jeremy Fitzhardinge 已提交
1163 1164
	.set_wallclock = xen_set_wallclock,
	.get_wallclock = xen_get_wallclock,
1165
	.get_tsc_khz = xen_tsc_khz,
J
Jeremy Fitzhardinge 已提交
1166
	.sched_clock = xen_sched_clock,
1167
};
J
Jeremy Fitzhardinge 已提交
1168

1169
static const struct pv_cpu_ops xen_cpu_ops __initdata = {
1170 1171 1172 1173 1174
	.cpuid = xen_cpuid,

	.set_debugreg = xen_set_debugreg,
	.get_debugreg = xen_get_debugreg,

1175
	.clts = xen_clts,
1176 1177

	.read_cr0 = native_read_cr0,
1178
	.write_cr0 = xen_write_cr0,
1179 1180 1181 1182 1183 1184 1185 1186

	.read_cr4 = native_read_cr4,
	.read_cr4_safe = native_read_cr4_safe,
	.write_cr4 = xen_write_cr4,

	.wbinvd = native_wbinvd,

	.read_msr = native_read_msr_safe,
1187
	.write_msr = xen_write_msr_safe,
1188 1189 1190
	.read_tsc = native_read_tsc,
	.read_pmc = native_read_pmc,

1191
	.iret = xen_iret,
1192
	.irq_enable_sysexit = xen_sysexit,
1193 1194 1195 1196
#ifdef CONFIG_X86_64
	.usergs_sysret32 = xen_sysret32,
	.usergs_sysret64 = xen_sysret64,
#endif
1197 1198 1199 1200 1201 1202

	.load_tr_desc = paravirt_nop,
	.set_ldt = xen_set_ldt,
	.load_gdt = xen_load_gdt,
	.load_idt = xen_load_idt,
	.load_tls = xen_load_tls,
1203 1204 1205
#ifdef CONFIG_X86_64
	.load_gs_index = xen_load_gs_index,
#endif
1206

1207 1208 1209
	.alloc_ldt = xen_alloc_ldt,
	.free_ldt = xen_free_ldt,

1210 1211 1212 1213 1214 1215 1216
	.store_gdt = native_store_gdt,
	.store_idt = native_store_idt,
	.store_tr = xen_store_tr,

	.write_ldt_entry = xen_write_ldt_entry,
	.write_gdt_entry = xen_write_gdt_entry,
	.write_idt_entry = xen_write_idt_entry,
1217
	.load_sp0 = xen_load_sp0,
1218 1219 1220 1221

	.set_iopl_mask = xen_set_iopl_mask,
	.io_delay = xen_io_delay,

J
Jeremy Fitzhardinge 已提交
1222 1223 1224
	/* Xen takes care of %gs when switching to usermode for us */
	.swapgs = paravirt_nop,

1225 1226 1227 1228
	.lazy_mode = {
		.enter = paravirt_enter_lazy_cpu,
		.leave = xen_leave_lazy,
	},
1229 1230 1231
};

static const struct pv_apic_ops xen_apic_ops __initdata = {
1232
#ifdef CONFIG_X86_LOCAL_APIC
J
Jeremy Fitzhardinge 已提交
1233
	.apic_write = xen_apic_write,
1234 1235 1236 1237 1238
	.apic_read = xen_apic_read,
	.setup_boot_clock = paravirt_nop,
	.setup_secondary_clock = paravirt_nop,
	.startup_ipi_hook = paravirt_nop,
#endif
1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
};

static const struct pv_mmu_ops xen_mmu_ops __initdata = {
	.pagetable_setup_start = xen_pagetable_setup_start,
	.pagetable_setup_done = xen_pagetable_setup_done,

	.read_cr2 = xen_read_cr2,
	.write_cr2 = xen_write_cr2,

	.read_cr3 = xen_read_cr3,
	.write_cr3 = xen_write_cr3,
1250 1251 1252 1253

	.flush_tlb_user = xen_flush_tlb,
	.flush_tlb_kernel = xen_flush_tlb,
	.flush_tlb_single = xen_flush_tlb_single,
J
Jeremy Fitzhardinge 已提交
1254
	.flush_tlb_others = xen_flush_tlb_others,
1255 1256 1257 1258

	.pte_update = paravirt_nop,
	.pte_update_defer = paravirt_nop,

1259 1260
	.pgd_alloc = xen_pgd_alloc,
	.pgd_free = xen_pgd_free,
1261

1262 1263 1264 1265 1266
	.alloc_pte = xen_alloc_pte_init,
	.release_pte = xen_release_pte_init,
	.alloc_pmd = xen_alloc_pte_init,
	.alloc_pmd_clone = paravirt_nop,
	.release_pmd = xen_release_pte_init,
1267 1268 1269 1270

#ifdef CONFIG_HIGHPTE
	.kmap_atomic_pte = xen_kmap_atomic_pte,
#endif
1271

1272 1273 1274
#ifdef CONFIG_X86_64
	.set_pte = xen_set_pte,
#else
1275
	.set_pte = xen_set_pte_init,
1276
#endif
J
Jeremy Fitzhardinge 已提交
1277
	.set_pte_at = xen_set_pte_at,
1278
	.set_pmd = xen_set_pmd_hyper,
J
Jeremy Fitzhardinge 已提交
1279

1280 1281 1282
	.ptep_modify_prot_start = __ptep_modify_prot_start,
	.ptep_modify_prot_commit = __ptep_modify_prot_commit,

J
Jeremy Fitzhardinge 已提交
1283
	.pte_val = xen_pte_val,
1284
	.pte_flags = native_pte_flags,
J
Jeremy Fitzhardinge 已提交
1285 1286 1287 1288 1289
	.pgd_val = xen_pgd_val,

	.make_pte = xen_make_pte,
	.make_pgd = xen_make_pgd,

1290
#ifdef CONFIG_X86_PAE
J
Jeremy Fitzhardinge 已提交
1291 1292 1293 1294
	.set_pte_atomic = xen_set_pte_atomic,
	.set_pte_present = xen_set_pte_at,
	.pte_clear = xen_pte_clear,
	.pmd_clear = xen_pmd_clear,
1295 1296
#endif	/* CONFIG_X86_PAE */
	.set_pud = xen_set_pud_hyper,
J
Jeremy Fitzhardinge 已提交
1297 1298 1299 1300

	.make_pmd = xen_make_pmd,
	.pmd_val = xen_pmd_val,

1301 1302 1303 1304 1305 1306 1307 1308 1309
#if PAGETABLE_LEVELS == 4
	.pud_val = xen_pud_val,
	.make_pud = xen_make_pud,
	.set_pgd = xen_set_pgd_hyper,

	.alloc_pud = xen_alloc_pte_init,
	.release_pud = xen_release_pte_init,
#endif	/* PAGETABLE_LEVELS == 4 */

J
Jeremy Fitzhardinge 已提交
1310 1311 1312 1313
	.activate_mm = xen_activate_mm,
	.dup_mmap = xen_dup_mmap,
	.exit_mmap = xen_exit_mmap,

1314 1315 1316 1317
	.lazy_mode = {
		.enter = paravirt_enter_lazy_mmu,
		.leave = xen_leave_lazy,
	},
1318 1319

	.set_fixmap = xen_set_fixmap,
1320 1321
};

J
Jeremy Fitzhardinge 已提交
1322 1323
static void xen_reboot(int reason)
{
J
Jeremy Fitzhardinge 已提交
1324 1325
	struct sched_shutdown r = { .reason = reason };

J
Jeremy Fitzhardinge 已提交
1326 1327 1328 1329
#ifdef CONFIG_SMP
	smp_send_stop();
#endif

J
Jeremy Fitzhardinge 已提交
1330
	if (HYPERVISOR_sched_op(SCHEDOP_shutdown, &r))
J
Jeremy Fitzhardinge 已提交
1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362
		BUG();
}

static void xen_restart(char *msg)
{
	xen_reboot(SHUTDOWN_reboot);
}

static void xen_emergency_restart(void)
{
	xen_reboot(SHUTDOWN_reboot);
}

static void xen_machine_halt(void)
{
	xen_reboot(SHUTDOWN_poweroff);
}

static void xen_crash_shutdown(struct pt_regs *regs)
{
	xen_reboot(SHUTDOWN_crash);
}

static const struct machine_ops __initdata xen_machine_ops = {
	.restart = xen_restart,
	.halt = xen_machine_halt,
	.power_off = xen_machine_halt,
	.shutdown = xen_machine_halt,
	.crash_shutdown = xen_crash_shutdown,
	.emergency_restart = xen_emergency_restart,
};

1363

1364 1365
static void __init xen_reserve_top(void)
{
1366
#ifdef CONFIG_X86_32
1367 1368 1369 1370 1371 1372 1373
	unsigned long top = HYPERVISOR_VIRT_START;
	struct xen_platform_parameters pp;

	if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0)
		top = pp.virt_start;

	reserve_top_address(-top + 2 * PAGE_SIZE);
1374
#endif	/* CONFIG_X86_32 */
1375 1376
}

J
Jeremy Fitzhardinge 已提交
1377 1378 1379 1380 1381 1382
/*
 * Like __va(), but returns address in the kernel mapping (which is
 * all we have until the physical memory mapping has been set up.
 */
static void *__ka(phys_addr_t paddr)
{
1383
#ifdef CONFIG_X86_64
J
Jeremy Fitzhardinge 已提交
1384
	return (void *)(paddr + __START_KERNEL_map);
1385 1386 1387
#else
	return __va(paddr);
#endif
1388 1389
}

J
Jeremy Fitzhardinge 已提交
1390 1391 1392 1393 1394
/* Convert a machine address to physical address */
static unsigned long m2p(phys_addr_t maddr)
{
	phys_addr_t paddr;

1395
	maddr &= PTE_PFN_MASK;
J
Jeremy Fitzhardinge 已提交
1396 1397 1398
	paddr = mfn_to_pfn(maddr >> PAGE_SHIFT) << PAGE_SHIFT;

	return paddr;
1399 1400
}

J
Jeremy Fitzhardinge 已提交
1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415
/* Convert a machine address to kernel virtual */
static void *m2v(phys_addr_t maddr)
{
	return __ka(m2p(maddr));
}

static void set_page_prot(void *addr, pgprot_t prot)
{
	unsigned long pfn = __pa(addr) >> PAGE_SHIFT;
	pte_t pte = pfn_pte(pfn, prot);

	if (HYPERVISOR_update_va_mapping((unsigned long)addr, pte, 0))
		BUG();
}

1416
static __init void xen_map_identity_early(pmd_t *pmd, unsigned long max_pfn)
1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427
{
	unsigned pmdidx, pteidx;
	unsigned ident_pte;
	unsigned long pfn;

	ident_pte = 0;
	pfn = 0;
	for(pmdidx = 0; pmdidx < PTRS_PER_PMD && pfn < max_pfn; pmdidx++) {
		pte_t *pte_page;

		/* Reuse or allocate a page of ptes */
1428 1429
		if (pmd_present(pmd[pmdidx]))
			pte_page = m2v(pmd[pmdidx].pmd);
1430 1431 1432 1433 1434 1435 1436 1437
		else {
			/* Check for free pte pages */
			if (ident_pte == ARRAY_SIZE(level1_ident_pgt))
				break;

			pte_page = &level1_ident_pgt[ident_pte];
			ident_pte += PTRS_PER_PTE;

1438
			pmd[pmdidx] = __pmd(__pa(pte_page) | _PAGE_TABLE);
1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457
		}

		/* Install mappings */
		for(pteidx = 0; pteidx < PTRS_PER_PTE; pteidx++, pfn++) {
			pte_t pte;

			if (pfn > max_pfn_mapped)
				max_pfn_mapped = pfn;

			if (!pte_none(pte_page[pteidx]))
				continue;

			pte = pfn_pte(pfn, PAGE_KERNEL_EXEC);
			pte_page[pteidx] = pte;
		}
	}

	for(pteidx = 0; pteidx < ident_pte; pteidx += PTRS_PER_PTE)
		set_page_prot(&level1_ident_pgt[pteidx], PAGE_KERNEL_RO);
1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471

	set_page_prot(pmd, PAGE_KERNEL_RO);
}

#ifdef CONFIG_X86_64
static void convert_pfn_mfn(void *v)
{
	pte_t *pte = v;
	int i;

	/* All levels are converted the same way, so just treat them
	   as ptes. */
	for(i = 0; i < PTRS_PER_PTE; i++)
		pte[i] = xen_make_pte(pte[i].pte);
1472 1473
}

J
Jeremy Fitzhardinge 已提交
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484
/*
 * Set up the inital kernel pagetable.
 *
 * We can construct this by grafting the Xen provided pagetable into
 * head_64.S's preconstructed pagetables.  We copy the Xen L2's into
 * level2_ident_pgt, level2_kernel_pgt and level2_fixmap_pgt.  This
 * means that only the kernel has a physical mapping to start with -
 * but that's enough to get __va working.  We need to fill in the rest
 * of the physical mapping once some sort of allocator has been set
 * up.
 */
1485
static __init pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn)
J
Jeremy Fitzhardinge 已提交
1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
{
	pud_t *l3;
	pmd_t *l2;

	/* Zap identity mapping */
	init_level4_pgt[0] = __pgd(0);

	/* Pre-constructed entries are in pfn, so convert to mfn */
	convert_pfn_mfn(init_level4_pgt);
	convert_pfn_mfn(level3_ident_pgt);
	convert_pfn_mfn(level3_kernel_pgt);

	l3 = m2v(pgd[pgd_index(__START_KERNEL_map)].pgd);
	l2 = m2v(l3[pud_index(__START_KERNEL_map)].pud);

	memcpy(level2_ident_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
	memcpy(level2_kernel_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);

	l3 = m2v(pgd[pgd_index(__START_KERNEL_map + PMD_SIZE)].pgd);
	l2 = m2v(l3[pud_index(__START_KERNEL_map + PMD_SIZE)].pud);
	memcpy(level2_fixmap_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);

1508
	/* Set up identity map */
1509
	xen_map_identity_early(level2_ident_pgt, max_pfn);
1510

J
Jeremy Fitzhardinge 已提交
1511 1512 1513 1514
	/* Make pagetable pieces RO */
	set_page_prot(init_level4_pgt, PAGE_KERNEL_RO);
	set_page_prot(level3_ident_pgt, PAGE_KERNEL_RO);
	set_page_prot(level3_kernel_pgt, PAGE_KERNEL_RO);
1515
	set_page_prot(level3_user_vsyscall, PAGE_KERNEL_RO);
J
Jeremy Fitzhardinge 已提交
1516 1517 1518 1519
	set_page_prot(level2_kernel_pgt, PAGE_KERNEL_RO);
	set_page_prot(level2_fixmap_pgt, PAGE_KERNEL_RO);

	/* Pin down new L4 */
1520 1521
	pin_pagetable_pfn(MMUEXT_PIN_L4_TABLE,
			  PFN_DOWN(__pa_symbol(init_level4_pgt)));
J
Jeremy Fitzhardinge 已提交
1522 1523 1524 1525 1526 1527

	/* Unpin Xen-provided one */
	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

	/* Switch over */
	pgd = init_level4_pgt;
1528 1529 1530 1531 1532 1533 1534 1535 1536

	/*
	 * At this stage there can be no user pgd, and no page
	 * structure to attach it to, so make sure we just set kernel
	 * pgd.
	 */
	xen_mc_batch();
	__xen_write_cr3(true, __pa(pgd));
	xen_mc_issue(PARAVIRT_LAZY_CPU);
J
Jeremy Fitzhardinge 已提交
1537

1538 1539 1540 1541
	reserve_early(__pa(xen_start_info->pt_base),
		      __pa(xen_start_info->pt_base +
			   xen_start_info->nr_pt_frames * PAGE_SIZE),
		      "XEN PAGETABLES");
J
Jeremy Fitzhardinge 已提交
1542 1543 1544

	return pgd;
}
1545 1546 1547
#else	/* !CONFIG_X86_64 */
static pmd_t level2_kernel_pgt[PTRS_PER_PMD] __page_aligned_bss;

1548
static __init pgd_t *xen_setup_kernel_pagetable(pgd_t *pgd, unsigned long max_pfn)
J
Jeremy Fitzhardinge 已提交
1549
{
1550 1551
	pmd_t *kernel_pmd;

J
Jeremy Fitzhardinge 已提交
1552 1553 1554 1555
	init_pg_tables_start = __pa(pgd);
	init_pg_tables_end = __pa(pgd) + xen_start_info->nr_pt_frames*PAGE_SIZE;
	max_pfn_mapped = PFN_DOWN(init_pg_tables_end + 512*1024);

1556 1557
	kernel_pmd = m2v(pgd[KERNEL_PGD_BOUNDARY].pgd);
	memcpy(level2_kernel_pgt, kernel_pmd, sizeof(pmd_t) * PTRS_PER_PMD);
1558

1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575
	xen_map_identity_early(level2_kernel_pgt, max_pfn);

	memcpy(swapper_pg_dir, pgd, sizeof(pgd_t) * PTRS_PER_PGD);
	set_pgd(&swapper_pg_dir[KERNEL_PGD_BOUNDARY],
			__pgd(__pa(level2_kernel_pgt) | _PAGE_PRESENT));

	set_page_prot(level2_kernel_pgt, PAGE_KERNEL_RO);
	set_page_prot(swapper_pg_dir, PAGE_KERNEL_RO);
	set_page_prot(empty_zero_page, PAGE_KERNEL_RO);

	pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

	xen_write_cr3(__pa(swapper_pg_dir));

	pin_pagetable_pfn(MMUEXT_PIN_L3_TABLE, PFN_DOWN(__pa(swapper_pg_dir)));

	return swapper_pg_dir;
1576
}
J
Jeremy Fitzhardinge 已提交
1577
#endif	/* CONFIG_X86_64 */
1578

1579 1580 1581 1582 1583 1584 1585 1586
/* First C function to be called on Xen boot */
asmlinkage void __init xen_start_kernel(void)
{
	pgd_t *pgd;

	if (!xen_start_info)
		return;

1587 1588
	xen_domain_type = XEN_PV_DOMAIN;

J
Jeremy Fitzhardinge 已提交
1589
	BUG_ON(memcmp(xen_start_info->magic, "xen-3", 5) != 0);
1590

1591 1592
	xen_setup_features();

1593
	/* Install Xen paravirt ops */
1594 1595 1596 1597 1598 1599 1600
	pv_info = xen_info;
	pv_init_ops = xen_init_ops;
	pv_time_ops = xen_time_ops;
	pv_cpu_ops = xen_cpu_ops;
	pv_apic_ops = xen_apic_ops;
	pv_mmu_ops = xen_mmu_ops;

1601 1602
	xen_init_irq_ops();

1603 1604 1605 1606 1607
	if (xen_feature(XENFEAT_mmu_pt_update_preserve_ad)) {
		pv_mmu_ops.ptep_modify_prot_start = xen_ptep_modify_prot_start;
		pv_mmu_ops.ptep_modify_prot_commit = xen_ptep_modify_prot_commit;
	}

J
Jeremy Fitzhardinge 已提交
1608 1609
	machine_ops = xen_machine_ops;

1610 1611 1612
#ifdef CONFIG_X86_64
	/* Disable until direct per-cpu data access. */
	have_vcpu_info_placement = 0;
1613
	x86_64_init_pda();
J
Jeremy Fitzhardinge 已提交
1614
#endif
1615

1616
	xen_smp_init();
1617 1618 1619

	/* Get mfn list */
	if (!xen_feature(XENFEAT_auto_translated_physmap))
1620
		xen_build_dynamic_phys_to_machine();
1621 1622 1623

	pgd = (pgd_t *)xen_start_info->pt_base;

J
Jeremy Fitzhardinge 已提交
1624 1625
	/* Prevent unwanted bits from being set in PTEs. */
	__supported_pte_mask &= ~_PAGE_GLOBAL;
1626
	if (!xen_initial_domain())
J
Jeremy Fitzhardinge 已提交
1627
		__supported_pte_mask &= ~(_PAGE_PWT | _PAGE_PCD);
1628 1629

	/* Don't do the full vcpu_info placement stuff until we have a
1630
	   possible map and a non-dummy shared_info. */
1631
	per_cpu(xen_vcpu, 0) = &HYPERVISOR_shared_info->vcpu_info[0];
1632

J
Jeremy Fitzhardinge 已提交
1633
	xen_raw_console_write("mapping kernel into physical memory\n");
1634
	pgd = xen_setup_kernel_pagetable(pgd, xen_start_info->nr_pages);
1635

J
Jeremy Fitzhardinge 已提交
1636
	init_mm.pgd = pgd;
1637 1638 1639

	/* keep using Xen gdt for now; no urgent need to change it */

1640
	pv_info.kernel_rpl = 1;
1641
	if (xen_feature(XENFEAT_supervisor_mode_kernel))
1642
		pv_info.kernel_rpl = 0;
1643 1644

	/* set the limit of our address space */
1645
	xen_reserve_top();
1646

1647
#ifdef CONFIG_X86_32
1648 1649 1650 1651
	/* set up basic CPUID stuff */
	cpu_detect(&new_cpu_data);
	new_cpu_data.hard_math = 1;
	new_cpu_data.x86_capability[0] = cpuid_edx(1);
1652
#endif
1653 1654

	/* Poke various useful things into boot_params */
1655 1656 1657 1658
	boot_params.hdr.type_of_loader = (9 << 4) | 0;
	boot_params.hdr.ramdisk_image = xen_start_info->mod_start
		? __pa(xen_start_info->mod_start) : 0;
	boot_params.hdr.ramdisk_size = xen_start_info->mod_len;
1659
	boot_params.hdr.cmd_line_ptr = __pa(xen_start_info->cmd_line);
1660

1661
	if (!xen_initial_domain()) {
1662
		add_preferred_console("xenboot", 0, NULL);
1663
		add_preferred_console("tty", 0, NULL);
1664
		add_preferred_console("hvc", 0, NULL);
1665
	}
1666

J
Jeremy Fitzhardinge 已提交
1667 1668
	xen_raw_console_write("about to get started...\n");

1669
	/* Start the world */
1670
#ifdef CONFIG_X86_32
1671
	i386_start_kernel();
1672
#else
J
Jeremy Fitzhardinge 已提交
1673
	x86_64_start_reservations((char *)__pa_symbol(&boot_params));
1674
#endif
1675
}