linux/arch/x86/kernel/paravirt.c
<<
>>
Prefs
   1/*  Paravirtualization interfaces
   2    Copyright (C) 2006 Rusty Russell IBM Corporation
   3
   4    This program is free software; you can redistribute it and/or modify
   5    it under the terms of the GNU General Public License as published by
   6    the Free Software Foundation; either version 2 of the License, or
   7    (at your option) any later version.
   8
   9    This program is distributed in the hope that it will be useful,
  10    but WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12    GNU General Public License for more details.
  13
  14    You should have received a copy of the GNU General Public License
  15    along with this program; if not, write to the Free Software
  16    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#include <linux/errno.h>
  22#include <linux/module.h>
  23#include <linux/efi.h>
  24#include <linux/bcd.h>
  25#include <linux/highmem.h>
  26#include <linux/kprobes.h>
  27
  28#include <asm/bug.h>
  29#include <asm/paravirt.h>
  30#include <asm/debugreg.h>
  31#include <asm/desc.h>
  32#include <asm/setup.h>
  33#include <asm/pgtable.h>
  34#include <asm/time.h>
  35#include <asm/pgalloc.h>
  36#include <asm/irq.h>
  37#include <asm/delay.h>
  38#include <asm/fixmap.h>
  39#include <asm/apic.h>
  40#include <asm/tlbflush.h>
  41#include <asm/timer.h>
  42#include <asm/special_insns.h>
  43
  44/* nop stub */
  45void _paravirt_nop(void)
  46{
  47}
  48
  49/* identity function, which can be inlined */
  50u32 _paravirt_ident_32(u32 x)
  51{
  52        return x;
  53}
  54
  55u64 _paravirt_ident_64(u64 x)
  56{
  57        return x;
  58}
  59
  60void __init default_banner(void)
  61{
  62        printk(KERN_INFO "Booting paravirtualized kernel on %s\n",
  63               pv_info.name);
  64}
  65
  66/* Undefined instruction for dealing with missing ops pointers. */
  67static const unsigned char ud2a[] = { 0x0f, 0x0b };
  68
  69unsigned paravirt_patch_nop(void)
  70{
  71        return 0;
  72}
  73
  74unsigned paravirt_patch_ignore(unsigned len)
  75{
  76        return len;
  77}
  78
  79struct branch {
  80        unsigned char opcode;
  81        u32 delta;
  82} __attribute__((packed));
  83
  84unsigned paravirt_patch_call(void *insnbuf,
  85                             const void *target, u16 tgt_clobbers,
  86                             unsigned long addr, u16 site_clobbers,
  87                             unsigned len)
  88{
  89        struct branch *b = insnbuf;
  90        unsigned long delta = (unsigned long)target - (addr+5);
  91
  92        if (tgt_clobbers & ~site_clobbers)
  93                return len;     /* target would clobber too much for this site */
  94        if (len < 5)
  95                return len;     /* call too long for patch site */
  96
  97        b->opcode = 0xe8; /* call */
  98        b->delta = delta;
  99        BUILD_BUG_ON(sizeof(*b) != 5);
 100
 101        return 5;
 102}
 103
 104unsigned paravirt_patch_jmp(void *insnbuf, const void *target,
 105                            unsigned long addr, unsigned len)
 106{
 107        struct branch *b = insnbuf;
 108        unsigned long delta = (unsigned long)target - (addr+5);
 109
 110        if (len < 5)
 111                return len;     /* call too long for patch site */
 112
 113        b->opcode = 0xe9;       /* jmp */
 114        b->delta = delta;
 115
 116        return 5;
 117}
 118
 119/* Neat trick to map patch type back to the call within the
 120 * corresponding structure. */
 121static void *get_call_destination(u8 type)
 122{
 123        struct paravirt_patch_template tmpl = {
 124                .pv_init_ops = pv_init_ops,
 125                .pv_time_ops = pv_time_ops,
 126                .pv_cpu_ops = pv_cpu_ops,
 127                .pv_irq_ops = pv_irq_ops,
 128                .pv_apic_ops = pv_apic_ops,
 129                .pv_mmu_ops = pv_mmu_ops,
 130#ifdef CONFIG_PARAVIRT_SPINLOCKS
 131                .pv_lock_ops = pv_lock_ops,
 132#endif
 133        };
 134        return *((void **)&tmpl + type);
 135}
 136
 137unsigned paravirt_patch_default(u8 type, u16 clobbers, void *insnbuf,
 138                                unsigned long addr, unsigned len)
 139{
 140        void *opfunc = get_call_destination(type);
 141        unsigned ret;
 142
 143        if (opfunc == NULL)
 144                /* If there's no function, patch it with a ud2a (BUG) */
 145                ret = paravirt_patch_insns(insnbuf, len, ud2a, ud2a+sizeof(ud2a));
 146        else if (opfunc == _paravirt_nop)
 147                /* If the operation is a nop, then nop the callsite */
 148                ret = paravirt_patch_nop();
 149
 150        /* identity functions just return their single argument */
 151        else if (opfunc == _paravirt_ident_32)
 152                ret = paravirt_patch_ident_32(insnbuf, len);
 153        else if (opfunc == _paravirt_ident_64)
 154                ret = paravirt_patch_ident_64(insnbuf, len);
 155
 156        else if (type == PARAVIRT_PATCH(pv_cpu_ops.iret) ||
 157                 type == PARAVIRT_PATCH(pv_cpu_ops.irq_enable_sysexit) ||
 158                 type == PARAVIRT_PATCH(pv_cpu_ops.usergs_sysret32) ||
 159                 type == PARAVIRT_PATCH(pv_cpu_ops.usergs_sysret64))
 160                /* If operation requires a jmp, then jmp */
 161                ret = paravirt_patch_jmp(insnbuf, opfunc, addr, len);
 162        else
 163                /* Otherwise call the function; assume target could
 164                   clobber any caller-save reg */
 165                ret = paravirt_patch_call(insnbuf, opfunc, CLBR_ANY,
 166                                          addr, clobbers, len);
 167
 168        return ret;
 169}
 170
 171unsigned paravirt_patch_insns(void *insnbuf, unsigned len,
 172                              const char *start, const char *end)
 173{
 174        unsigned insn_len = end - start;
 175
 176        if (insn_len > len || start == NULL)
 177                insn_len = len;
 178        else
 179                memcpy(insnbuf, start, insn_len);
 180
 181        return insn_len;
 182}
 183
 184static void native_flush_tlb(void)
 185{
 186        __native_flush_tlb();
 187}
 188
 189/*
 190 * Global pages have to be flushed a bit differently. Not a real
 191 * performance problem because this does not happen often.
 192 */
 193static void native_flush_tlb_global(void)
 194{
 195        __native_flush_tlb_global();
 196}
 197
 198static void native_flush_tlb_single(unsigned long addr)
 199{
 200        __native_flush_tlb_single(addr);
 201}
 202
 203struct static_key paravirt_steal_enabled;
 204struct static_key paravirt_steal_rq_enabled;
 205
 206static u64 native_steal_clock(int cpu)
 207{
 208        return 0;
 209}
 210
 211/* These are in entry.S */
 212extern void native_iret(void);
 213extern void native_irq_enable_sysexit(void);
 214extern void native_usergs_sysret32(void);
 215extern void native_usergs_sysret64(void);
 216
 217static struct resource reserve_ioports = {
 218        .start = 0,
 219        .end = IO_SPACE_LIMIT,
 220        .name = "paravirt-ioport",
 221        .flags = IORESOURCE_IO | IORESOURCE_BUSY,
 222};
 223
 224/*
 225 * Reserve the whole legacy IO space to prevent any legacy drivers
 226 * from wasting time probing for their hardware.  This is a fairly
 227 * brute-force approach to disabling all non-virtual drivers.
 228 *
 229 * Note that this must be called very early to have any effect.
 230 */
 231int paravirt_disable_iospace(void)
 232{
 233        return request_resource(&ioport_resource, &reserve_ioports);
 234}
 235
 236static DEFINE_PER_CPU(enum paravirt_lazy_mode, paravirt_lazy_mode) = PARAVIRT_LAZY_NONE;
 237
 238static inline void enter_lazy(enum paravirt_lazy_mode mode)
 239{
 240        BUG_ON(this_cpu_read(paravirt_lazy_mode) != PARAVIRT_LAZY_NONE);
 241
 242        this_cpu_write(paravirt_lazy_mode, mode);
 243}
 244
 245static void leave_lazy(enum paravirt_lazy_mode mode)
 246{
 247        BUG_ON(this_cpu_read(paravirt_lazy_mode) != mode);
 248
 249        this_cpu_write(paravirt_lazy_mode, PARAVIRT_LAZY_NONE);
 250}
 251
 252void paravirt_enter_lazy_mmu(void)
 253{
 254        enter_lazy(PARAVIRT_LAZY_MMU);
 255}
 256
 257void paravirt_leave_lazy_mmu(void)
 258{
 259        leave_lazy(PARAVIRT_LAZY_MMU);
 260}
 261
 262void paravirt_flush_lazy_mmu(void)
 263{
 264        preempt_disable();
 265
 266        if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
 267                arch_leave_lazy_mmu_mode();
 268                arch_enter_lazy_mmu_mode();
 269        }
 270
 271        preempt_enable();
 272}
 273
 274void paravirt_start_context_switch(struct task_struct *prev)
 275{
 276        BUG_ON(preemptible());
 277
 278        if (this_cpu_read(paravirt_lazy_mode) == PARAVIRT_LAZY_MMU) {
 279                arch_leave_lazy_mmu_mode();
 280                set_ti_thread_flag(task_thread_info(prev), TIF_LAZY_MMU_UPDATES);
 281        }
 282        enter_lazy(PARAVIRT_LAZY_CPU);
 283}
 284
 285void paravirt_end_context_switch(struct task_struct *next)
 286{
 287        BUG_ON(preemptible());
 288
 289        leave_lazy(PARAVIRT_LAZY_CPU);
 290
 291        if (test_and_clear_ti_thread_flag(task_thread_info(next), TIF_LAZY_MMU_UPDATES))
 292                arch_enter_lazy_mmu_mode();
 293}
 294
 295enum paravirt_lazy_mode paravirt_get_lazy_mode(void)
 296{
 297        if (in_interrupt())
 298                return PARAVIRT_LAZY_NONE;
 299
 300        return this_cpu_read(paravirt_lazy_mode);
 301}
 302
 303struct pv_info pv_info = {
 304        .name = "bare hardware",
 305        .paravirt_enabled = 0,
 306        .kernel_rpl = 0,
 307        .shared_kernel_pmd = 1, /* Only used when CONFIG_X86_PAE is set */
 308
 309#ifdef CONFIG_X86_64
 310        .extra_user_64bit_cs = __USER_CS,
 311#endif
 312};
 313
 314struct pv_init_ops pv_init_ops = {
 315        .patch = native_patch,
 316};
 317
 318struct pv_time_ops pv_time_ops = {
 319        .sched_clock = native_sched_clock,
 320        .steal_clock = native_steal_clock,
 321};
 322
 323__visible struct pv_irq_ops pv_irq_ops = {
 324        .save_fl = __PV_IS_CALLEE_SAVE(native_save_fl),
 325        .restore_fl = __PV_IS_CALLEE_SAVE(native_restore_fl),
 326        .irq_disable = __PV_IS_CALLEE_SAVE(native_irq_disable),
 327        .irq_enable = __PV_IS_CALLEE_SAVE(native_irq_enable),
 328        .safe_halt = native_safe_halt,
 329        .halt = native_halt,
 330#ifdef CONFIG_X86_64
 331        .adjust_exception_frame = paravirt_nop,
 332#endif
 333};
 334
 335__visible struct pv_cpu_ops pv_cpu_ops = {
 336        .cpuid = native_cpuid,
 337        .get_debugreg = native_get_debugreg,
 338        .set_debugreg = native_set_debugreg,
 339        .clts = native_clts,
 340        .read_cr0 = native_read_cr0,
 341        .write_cr0 = native_write_cr0,
 342        .read_cr4 = native_read_cr4,
 343        .read_cr4_safe = native_read_cr4_safe,
 344        .write_cr4 = native_write_cr4,
 345#ifdef CONFIG_X86_64
 346        .read_cr8 = native_read_cr8,
 347        .write_cr8 = native_write_cr8,
 348#endif
 349        .wbinvd = native_wbinvd,
 350        .read_msr = native_read_msr_safe,
 351        .write_msr = native_write_msr_safe,
 352        .read_tsc = native_read_tsc,
 353        .read_pmc = native_read_pmc,
 354        .read_tscp = native_read_tscp,
 355        .load_tr_desc = native_load_tr_desc,
 356        .set_ldt = native_set_ldt,
 357        .load_gdt = native_load_gdt,
 358        .load_idt = native_load_idt,
 359        .store_idt = native_store_idt,
 360        .store_tr = native_store_tr,
 361        .load_tls = native_load_tls,
 362#ifdef CONFIG_X86_64
 363        .load_gs_index = native_load_gs_index,
 364#endif
 365        .write_ldt_entry = native_write_ldt_entry,
 366        .write_gdt_entry = native_write_gdt_entry,
 367        .write_idt_entry = native_write_idt_entry,
 368
 369        .alloc_ldt = paravirt_nop,
 370        .free_ldt = paravirt_nop,
 371
 372        .load_sp0 = native_load_sp0,
 373
 374#if defined(CONFIG_X86_32) || defined(CONFIG_IA32_EMULATION)
 375        .irq_enable_sysexit = native_irq_enable_sysexit,
 376#endif
 377#ifdef CONFIG_X86_64
 378#ifdef CONFIG_IA32_EMULATION
 379        .usergs_sysret32 = native_usergs_sysret32,
 380#endif
 381        .usergs_sysret64 = native_usergs_sysret64,
 382#endif
 383        .iret = native_iret,
 384        .swapgs = native_swapgs,
 385
 386        .set_iopl_mask = native_set_iopl_mask,
 387        .io_delay = native_io_delay,
 388
 389        .start_context_switch = paravirt_nop,
 390        .end_context_switch = paravirt_nop,
 391};
 392
 393/* At this point, native_get/set_debugreg has real function entries */
 394NOKPROBE_SYMBOL(native_get_debugreg);
 395NOKPROBE_SYMBOL(native_set_debugreg);
 396NOKPROBE_SYMBOL(native_load_idt);
 397
 398struct pv_apic_ops pv_apic_ops = {
 399#ifdef CONFIG_X86_LOCAL_APIC
 400        .startup_ipi_hook = paravirt_nop,
 401#endif
 402};
 403
 404#if defined(CONFIG_X86_32) && !defined(CONFIG_X86_PAE)
 405/* 32-bit pagetable entries */
 406#define PTE_IDENT       __PV_IS_CALLEE_SAVE(_paravirt_ident_32)
 407#else
 408/* 64-bit pagetable entries */
 409#define PTE_IDENT       __PV_IS_CALLEE_SAVE(_paravirt_ident_64)
 410#endif
 411
 412struct pv_mmu_ops pv_mmu_ops = {
 413
 414        .read_cr2 = native_read_cr2,
 415        .write_cr2 = native_write_cr2,
 416        .read_cr3 = native_read_cr3,
 417        .write_cr3 = native_write_cr3,
 418
 419        .flush_tlb_user = native_flush_tlb,
 420        .flush_tlb_kernel = native_flush_tlb_global,
 421        .flush_tlb_single = native_flush_tlb_single,
 422        .flush_tlb_others = native_flush_tlb_others,
 423
 424        .pgd_alloc = __paravirt_pgd_alloc,
 425        .pgd_free = paravirt_nop,
 426
 427        .alloc_pte = paravirt_nop,
 428        .alloc_pmd = paravirt_nop,
 429        .alloc_pud = paravirt_nop,
 430        .release_pte = paravirt_nop,
 431        .release_pmd = paravirt_nop,
 432        .release_pud = paravirt_nop,
 433
 434        .set_pte = native_set_pte,
 435        .set_pte_at = native_set_pte_at,
 436        .set_pmd = native_set_pmd,
 437        .set_pmd_at = native_set_pmd_at,
 438        .pte_update = paravirt_nop,
 439        .pte_update_defer = paravirt_nop,
 440        .pmd_update = paravirt_nop,
 441        .pmd_update_defer = paravirt_nop,
 442
 443        .ptep_modify_prot_start = __ptep_modify_prot_start,
 444        .ptep_modify_prot_commit = __ptep_modify_prot_commit,
 445
 446#if PAGETABLE_LEVELS >= 3
 447#ifdef CONFIG_X86_PAE
 448        .set_pte_atomic = native_set_pte_atomic,
 449        .pte_clear = native_pte_clear,
 450        .pmd_clear = native_pmd_clear,
 451#endif
 452        .set_pud = native_set_pud,
 453
 454        .pmd_val = PTE_IDENT,
 455        .make_pmd = PTE_IDENT,
 456
 457#if PAGETABLE_LEVELS == 4
 458        .pud_val = PTE_IDENT,
 459        .make_pud = PTE_IDENT,
 460
 461        .set_pgd = native_set_pgd,
 462#endif
 463#endif /* PAGETABLE_LEVELS >= 3 */
 464
 465        .pte_val = PTE_IDENT,
 466        .pgd_val = PTE_IDENT,
 467
 468        .make_pte = PTE_IDENT,
 469        .make_pgd = PTE_IDENT,
 470
 471        .dup_mmap = paravirt_nop,
 472        .exit_mmap = paravirt_nop,
 473        .activate_mm = paravirt_nop,
 474
 475        .lazy_mode = {
 476                .enter = paravirt_nop,
 477                .leave = paravirt_nop,
 478                .flush = paravirt_nop,
 479        },
 480
 481        .set_fixmap = native_set_fixmap,
 482};
 483
 484EXPORT_SYMBOL_GPL(pv_time_ops);
 485EXPORT_SYMBOL    (pv_cpu_ops);
 486EXPORT_SYMBOL    (pv_mmu_ops);
 487EXPORT_SYMBOL_GPL(pv_apic_ops);
 488EXPORT_SYMBOL_GPL(pv_info);
 489EXPORT_SYMBOL    (pv_irq_ops);
 490