paravirt.c 12.0 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
/*  Paravirtualization interfaces
    Copyright (C) 2006 Rusty Russell IBM Corporation

    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation; either version 2 of the License, or
    (at your option) any later version.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with this program; if not, write to the Free Software
    Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
17 18

    2007 - x86_64 support added by Glauber de Oliveira Costa, Red Hat Inc
19
*/
20

21 22 23 24
#include <linux/errno.h>
#include <linux/module.h>
#include <linux/efi.h>
#include <linux/bcd.h>
25
#include <linux/highmem.h>
26
#include <linux/kprobes.h>
27 28 29

#include <asm/bug.h>
#include <asm/paravirt.h>
30
#include <asm/debugreg.h>
31 32
#include <asm/desc.h>
#include <asm/setup.h>
33
#include <asm/pgtable.h>
34
#include <asm/time.h>
35
#include <asm/pgalloc.h>
36 37
#include <asm/irq.h>
#include <asm/delay.h>
38 39
#include <asm/fixmap.h>
#include <asm/apic.h>
40
#include <asm/tlbflush.h>
41
#include <asm/timer.h>
42
#include <asm/special_insns.h>
43

44 45 46 47 48 49 50 51 52 53 54 55
/*
 * nop stub, which must not clobber anything *including the stack* to
 * avoid confusing the entry prologues.
 */
extern void _paravirt_nop(void);
asm (".pushsection .entry.text, \"ax\"\n"
     ".global _paravirt_nop\n"
     "_paravirt_nop:\n\t"
     "ret\n\t"
     ".size _paravirt_nop, . - _paravirt_nop\n\t"
     ".type _paravirt_nop, @function\n\t"
     ".popsection");
56

57 58 59 60 61 62 63 64 65 66 67
/* identity function, which can be inlined */
u32 _paravirt_ident_32(u32 x)
{
	return x;
}

u64 _paravirt_ident_64(u64 x)
{
	return x;
}

68
void __init default_banner(void)
69 70
{
	printk(KERN_INFO "Booting paravirtualized kernel on %s\n",
71
	       pv_info.name);
72 73
}

74 75
/* Undefined instruction for dealing with missing ops pointers. */
static const unsigned char ud2a[] = { 0x0f, 0x0b };
76

77 78 79 80 81 82 83 84 85 86
unsigned paravirt_patch_nop(void)
{
	return 0;
}

unsigned paravirt_patch_ignore(unsigned len)
{
	return len;
}

87 88 89 90 91
struct branch {
	unsigned char opcode;
	u32 delta;
} __attribute__((packed));

92 93 94
unsigned paravirt_patch_call(void *insnbuf,
			     const void *target, u16 tgt_clobbers,
			     unsigned long addr, u16 site_clobbers,
95 96
			     unsigned len)
{
97 98
	struct branch *b = insnbuf;
	unsigned long delta = (unsigned long)target - (addr+5);
99 100 101 102 103

	if (tgt_clobbers & ~site_clobbers)
		return len;	/* target would clobber too much for this site */
	if (len < 5)
		return len;	/* call too long for patch site */
104

105 106 107
	b->opcode = 0xe8; /* call */
	b->delta = delta;
	BUILD_BUG_ON(sizeof(*b) != 5);
108

109 110 111
	return 5;
}

112
unsigned paravirt_patch_jmp(void *insnbuf, const void *target,
113
			    unsigned long addr, unsigned len)
114
{
115 116
	struct branch *b = insnbuf;
	unsigned long delta = (unsigned long)target - (addr+5);
117 118 119 120

	if (len < 5)
		return len;	/* call too long for patch site */

121 122
	b->opcode = 0xe9;	/* jmp */
	b->delta = delta;
123 124 125 126

	return 5;
}

127 128 129 130 131 132 133 134 135 136 137
/* Neat trick to map patch type back to the call within the
 * corresponding structure. */
static void *get_call_destination(u8 type)
{
	struct paravirt_patch_template tmpl = {
		.pv_init_ops = pv_init_ops,
		.pv_time_ops = pv_time_ops,
		.pv_cpu_ops = pv_cpu_ops,
		.pv_irq_ops = pv_irq_ops,
		.pv_apic_ops = pv_apic_ops,
		.pv_mmu_ops = pv_mmu_ops,
138
#ifdef CONFIG_PARAVIRT_SPINLOCKS
139
		.pv_lock_ops = pv_lock_ops,
140
#endif
141 142 143 144
	};
	return *((void **)&tmpl + type);
}

145 146
unsigned paravirt_patch_default(u8 type, u16 clobbers, void *insnbuf,
				unsigned long addr, unsigned len)
147
{
148
	void *opfunc = get_call_destination(type);
149 150 151 152
	unsigned ret;

	if (opfunc == NULL)
		/* If there's no function, patch it with a ud2a (BUG) */
153
		ret = paravirt_patch_insns(insnbuf, len, ud2a, ud2a+sizeof(ud2a));
154
	else if (opfunc == _paravirt_nop)
155 156
		/* If the operation is a nop, then nop the callsite */
		ret = paravirt_patch_nop();
157 158 159 160 161 162 163

	/* identity functions just return their single argument */
	else if (opfunc == _paravirt_ident_32)
		ret = paravirt_patch_ident_32(insnbuf, len);
	else if (opfunc == _paravirt_ident_64)
		ret = paravirt_patch_ident_64(insnbuf, len);

164
	else if (type == PARAVIRT_PATCH(pv_cpu_ops.iret) ||
165 166
		 type == PARAVIRT_PATCH(pv_cpu_ops.usergs_sysret32) ||
		 type == PARAVIRT_PATCH(pv_cpu_ops.usergs_sysret64))
167
		/* If operation requires a jmp, then jmp */
168
		ret = paravirt_patch_jmp(insnbuf, opfunc, addr, len);
169 170 171
	else
		/* Otherwise call the function; assume target could
		   clobber any caller-save reg */
172 173
		ret = paravirt_patch_call(insnbuf, opfunc, CLBR_ANY,
					  addr, clobbers, len);
174 175 176 177

	return ret;
}

178
unsigned paravirt_patch_insns(void *insnbuf, unsigned len,
179 180 181
			      const char *start, const char *end)
{
	unsigned insn_len = end - start;
182

183 184 185
	if (insn_len > len || start == NULL)
		insn_len = len;
	else
186
		memcpy(insnbuf, start, insn_len);
187 188 189 190

	return insn_len;
}

191
static void native_flush_tlb(void)
192 193 194 195 196 197 198 199
{
	__native_flush_tlb();
}

/*
 * Global pages have to be flushed a bit differently. Not a real
 * performance problem because this does not happen often.
 */
200
static void native_flush_tlb_global(void)
201 202 203 204
{
	__native_flush_tlb_global();
}

205
static void native_flush_tlb_single(unsigned long addr)
206 207 208 209
{
	__native_flush_tlb_single(addr);
}

210 211
struct static_key paravirt_steal_enabled;
struct static_key paravirt_steal_rq_enabled;
212 213 214 215 216 217

static u64 native_steal_clock(int cpu)
{
	return 0;
}

218
/* These are in entry.S */
219
extern void native_iret(void);
220 221
extern void native_usergs_sysret32(void);
extern void native_usergs_sysret64(void);
222

223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238
static struct resource reserve_ioports = {
	.start = 0,
	.end = IO_SPACE_LIMIT,
	.name = "paravirt-ioport",
	.flags = IORESOURCE_IO | IORESOURCE_BUSY,
};

/*
 * Reserve the whole legacy IO space to prevent any legacy drivers
 * from wasting time probing for their hardware.  This is a fairly
 * brute-force approach to disabling all non-virtual drivers.
 *
 * Note that this must be called very early to have any effect.
 */
int paravirt_disable_iospace(void)
{
239
	return request_resource(&ioport_resource, &reserve_ioports);
240 241
}

242 243 244 245
static DEFINE_PER_CPU(enum paravirt_lazy_mode, paravirt_lazy_mode) = PARAVIRT_LAZY_NONE;

static inline void enter_lazy(enum paravirt_lazy_mode mode)
{
246
	BUG_ON(this_cpu_read(paravirt_lazy_mode) != PARAVIRT_LAZY_NONE);
247

248
	this_cpu_write(paravirt_lazy_mode, mode);
249 250
}

251
static void leave_lazy(enum paravirt_lazy_mode mode)
252
{
253
	BUG_ON(this_cpu_read(paravirt_lazy_mode) != mode);
254

255
	this_cpu_write(paravirt_lazy_mode, PARAVIRT_LAZY_NONE);
256 257 258 259 260 261 262 263 264
}

void paravirt_enter_lazy_mmu(void)
{
	enter_lazy(PARAVIRT_LAZY_MMU);
}

void paravirt_leave_lazy_mmu(void)
{
265
	leave_lazy(PARAVIRT_LAZY_MMU);
266 267
}

268 269 270 271 272 273 274 275 276 277 278 279
void paravirt_flush_lazy_mmu(void)
{
	preempt_disable();

	if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
		arch_leave_lazy_mmu_mode();
		arch_enter_lazy_mmu_mode();
	}

	preempt_enable();
}

280
void paravirt_start_context_switch(struct task_struct *prev)
281
{
282 283
	BUG_ON(preemptible());

284
	if (this_cpu_read(paravirt_lazy_mode) == PARAVIRT_LAZY_MMU) {
285
		arch_leave_lazy_mmu_mode();
286
		set_ti_thread_flag(task_thread_info(prev), TIF_LAZY_MMU_UPDATES);
287
	}
288 289 290
	enter_lazy(PARAVIRT_LAZY_CPU);
}

291
void paravirt_end_context_switch(struct task_struct *next)
292
{
293 294
	BUG_ON(preemptible());

295 296
	leave_lazy(PARAVIRT_LAZY_CPU);

297
	if (test_and_clear_ti_thread_flag(task_thread_info(next), TIF_LAZY_MMU_UPDATES))
298
		arch_enter_lazy_mmu_mode();
299 300 301 302
}

enum paravirt_lazy_mode paravirt_get_lazy_mode(void)
{
303 304 305
	if (in_interrupt())
		return PARAVIRT_LAZY_NONE;

306
	return this_cpu_read(paravirt_lazy_mode);
307 308
}

309
struct pv_info pv_info = {
310 311 312
	.name = "bare hardware",
	.paravirt_enabled = 0,
	.kernel_rpl = 0,
313
	.shared_kernel_pmd = 1,	/* Only used when CONFIG_X86_PAE is set */
314 315 316 317

#ifdef CONFIG_X86_64
	.extra_user_64bit_cs = __USER_CS,
#endif
318
};
319

320 321 322 323 324 325
struct pv_init_ops pv_init_ops = {
	.patch = native_patch,
};

struct pv_time_ops pv_time_ops = {
	.sched_clock = native_sched_clock,
326
	.steal_clock = native_steal_clock,
327 328
};

329
__visible struct pv_irq_ops pv_irq_ops = {
330 331 332 333
	.save_fl = __PV_IS_CALLEE_SAVE(native_save_fl),
	.restore_fl = __PV_IS_CALLEE_SAVE(native_restore_fl),
	.irq_disable = __PV_IS_CALLEE_SAVE(native_irq_disable),
	.irq_enable = __PV_IS_CALLEE_SAVE(native_irq_enable),
334 335
	.safe_halt = native_safe_halt,
	.halt = native_halt,
336 337 338
#ifdef CONFIG_X86_64
	.adjust_exception_frame = paravirt_nop,
#endif
339
};
340

341
__visible struct pv_cpu_ops pv_cpu_ops = {
342 343 344 345 346 347 348 349 350
	.cpuid = native_cpuid,
	.get_debugreg = native_get_debugreg,
	.set_debugreg = native_set_debugreg,
	.clts = native_clts,
	.read_cr0 = native_read_cr0,
	.write_cr0 = native_write_cr0,
	.read_cr4 = native_read_cr4,
	.read_cr4_safe = native_read_cr4_safe,
	.write_cr4 = native_write_cr4,
351 352 353 354
#ifdef CONFIG_X86_64
	.read_cr8 = native_read_cr8,
	.write_cr8 = native_write_cr8,
#endif
355
	.wbinvd = native_wbinvd,
356 357
	.read_msr = native_read_msr_safe,
	.write_msr = native_write_msr_safe,
358 359 360 361 362 363 364 365
	.read_pmc = native_read_pmc,
	.load_tr_desc = native_load_tr_desc,
	.set_ldt = native_set_ldt,
	.load_gdt = native_load_gdt,
	.load_idt = native_load_idt,
	.store_idt = native_store_idt,
	.store_tr = native_store_tr,
	.load_tls = native_load_tls,
366 367 368
#ifdef CONFIG_X86_64
	.load_gs_index = native_load_gs_index,
#endif
369
	.write_ldt_entry = native_write_ldt_entry,
370
	.write_gdt_entry = native_write_gdt_entry,
371
	.write_idt_entry = native_write_idt_entry,
372 373 374 375

	.alloc_ldt = paravirt_nop,
	.free_ldt = paravirt_nop,

376
	.load_sp0 = native_load_sp0,
377

378
#ifdef CONFIG_X86_64
379
#ifdef CONFIG_IA32_EMULATION
380
	.usergs_sysret32 = native_usergs_sysret32,
381
#endif
382
	.usergs_sysret64 = native_usergs_sysret64,
383
#endif
384
	.iret = native_iret,
385
	.swapgs = native_swapgs,
386

387 388
	.set_iopl_mask = native_set_iopl_mask,
	.io_delay = native_io_delay,
389

390 391
	.start_context_switch = paravirt_nop,
	.end_context_switch = paravirt_nop,
392
};
393

394
/* At this point, native_get/set_debugreg has real function entries */
395
NOKPROBE_SYMBOL(native_get_debugreg);
396 397
NOKPROBE_SYMBOL(native_set_debugreg);
NOKPROBE_SYMBOL(native_load_idt);
398

399
struct pv_apic_ops pv_apic_ops = {
400
#ifdef CONFIG_X86_LOCAL_APIC
401
	.startup_ipi_hook = paravirt_nop,
402
#endif
403 404
};

405 406
#if defined(CONFIG_X86_32) && !defined(CONFIG_X86_PAE)
/* 32-bit pagetable entries */
407
#define PTE_IDENT	__PV_IS_CALLEE_SAVE(_paravirt_ident_32)
408 409
#else
/* 64-bit pagetable entries */
410
#define PTE_IDENT	__PV_IS_CALLEE_SAVE(_paravirt_ident_64)
411 412
#endif

413
struct pv_mmu_ops pv_mmu_ops = {
414

415 416 417 418 419
	.read_cr2 = native_read_cr2,
	.write_cr2 = native_write_cr2,
	.read_cr3 = native_read_cr3,
	.write_cr3 = native_write_cr3,

420 421 422
	.flush_tlb_user = native_flush_tlb,
	.flush_tlb_kernel = native_flush_tlb_global,
	.flush_tlb_single = native_flush_tlb_single,
423
	.flush_tlb_others = native_flush_tlb_others,
424

425 426 427
	.pgd_alloc = __paravirt_pgd_alloc,
	.pgd_free = paravirt_nop,

428 429
	.alloc_pte = paravirt_nop,
	.alloc_pmd = paravirt_nop,
430
	.alloc_pud = paravirt_nop,
431 432
	.release_pte = paravirt_nop,
	.release_pmd = paravirt_nop,
433
	.release_pud = paravirt_nop,
434

435 436 437
	.set_pte = native_set_pte,
	.set_pte_at = native_set_pte_at,
	.set_pmd = native_set_pmd,
A
Andrea Arcangeli 已提交
438
	.set_pmd_at = native_set_pmd_at,
439 440
	.pte_update = paravirt_nop,
	.pte_update_defer = paravirt_nop,
A
Andrea Arcangeli 已提交
441 442
	.pmd_update = paravirt_nop,
	.pmd_update_defer = paravirt_nop,
443

444 445 446
	.ptep_modify_prot_start = __ptep_modify_prot_start,
	.ptep_modify_prot_commit = __ptep_modify_prot_commit,

447
#if CONFIG_PGTABLE_LEVELS >= 3
448 449 450 451
#ifdef CONFIG_X86_PAE
	.set_pte_atomic = native_set_pte_atomic,
	.pte_clear = native_pte_clear,
	.pmd_clear = native_pmd_clear,
452 453
#endif
	.set_pud = native_set_pud,
454 455 456

	.pmd_val = PTE_IDENT,
	.make_pmd = PTE_IDENT,
457

458
#if CONFIG_PGTABLE_LEVELS == 4
459 460 461
	.pud_val = PTE_IDENT,
	.make_pud = PTE_IDENT,

462
	.set_pgd = native_set_pgd,
463
#endif
464
#endif /* CONFIG_PGTABLE_LEVELS >= 3 */
465

466 467
	.pte_val = PTE_IDENT,
	.pgd_val = PTE_IDENT,
468

469 470
	.make_pte = PTE_IDENT,
	.make_pgd = PTE_IDENT,
471

472 473 474
	.dup_mmap = paravirt_nop,
	.exit_mmap = paravirt_nop,
	.activate_mm = paravirt_nop,
475 476 477 478

	.lazy_mode = {
		.enter = paravirt_nop,
		.leave = paravirt_nop,
479
		.flush = paravirt_nop,
480
	},
481 482

	.set_fixmap = native_set_fixmap,
483
};
484

485
EXPORT_SYMBOL_GPL(pv_time_ops);
486 487
EXPORT_SYMBOL    (pv_cpu_ops);
EXPORT_SYMBOL    (pv_mmu_ops);
488 489 490
EXPORT_SYMBOL_GPL(pv_apic_ops);
EXPORT_SYMBOL_GPL(pv_info);
EXPORT_SYMBOL    (pv_irq_ops);