kqemu.c 20.4 KB
Newer Older
B
bellard 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
/*
 *  KQEMU support
 * 
 *  Copyright (c) 2005 Fabrice Bellard
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library 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
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */
#include "config.h"
#ifdef _WIN32
#include <windows.h>
23
#include <winioctl.h>
B
bellard 已提交
24 25 26
#else
#include <sys/types.h>
#include <sys/mman.h>
27
#include <sys/ioctl.h>
B
bellard 已提交
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
#endif
#include <stdlib.h>
#include <stdio.h>
#include <stdarg.h>
#include <string.h>
#include <errno.h>
#include <unistd.h>
#include <inttypes.h>

#include "cpu.h"
#include "exec-all.h"

#ifdef USE_KQEMU

#define DEBUG
B
bellard 已提交
43
//#define PROFILE
B
bellard 已提交
44 45 46 47 48

#include <unistd.h>
#include <fcntl.h>
#include "kqemu/kqemu.h"

B
bellard 已提交
49 50 51 52
/* compatibility stuff */
#ifndef KQEMU_RET_SYSCALL
#define KQEMU_RET_SYSCALL   0x0300 /* syscall insn */
#endif
B
bellard 已提交
53 54 55 56
#ifndef KQEMU_MAX_RAM_PAGES_TO_UPDATE
#define KQEMU_MAX_RAM_PAGES_TO_UPDATE 512
#define KQEMU_RAM_PAGES_UPDATE_ALL (KQEMU_MAX_RAM_PAGES_TO_UPDATE + 1)
#endif
B
bellard 已提交
57

58 59 60
#ifdef _WIN32
#define KQEMU_DEVICE "\\\\.\\kqemu"
#else
B
bellard 已提交
61
#define KQEMU_DEVICE "/dev/kqemu"
62 63 64 65 66 67 68 69 70 71 72
#endif

#ifdef _WIN32
#define KQEMU_INVALID_FD INVALID_HANDLE_VALUE
HANDLE kqemu_fd = KQEMU_INVALID_FD;
#define kqemu_closefd(x) CloseHandle(x)
#else
#define KQEMU_INVALID_FD -1
int kqemu_fd = KQEMU_INVALID_FD;
#define kqemu_closefd(x) close(x)
#endif
B
bellard 已提交
73 74 75 76

int kqemu_allowed = 1;
unsigned long *pages_to_flush;
unsigned int nb_pages_to_flush;
B
bellard 已提交
77 78
unsigned long *ram_pages_to_update;
unsigned int nb_ram_pages_to_update;
B
bellard 已提交
79 80 81 82 83 84 85
extern uint32_t **l1_phys_map;

#define cpuid(index, eax, ebx, ecx, edx) \
  asm volatile ("cpuid" \
                : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx) \
                : "0" (index))

B
bellard 已提交
86 87 88 89 90 91
#ifdef __x86_64__
static int is_cpuid_supported(void)
{
    return 1;
}
#else
B
bellard 已提交
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107
static int is_cpuid_supported(void)
{
    int v0, v1;
    asm volatile ("pushf\n"
                  "popl %0\n"
                  "movl %0, %1\n"
                  "xorl $0x00200000, %0\n"
                  "pushl %0\n"
                  "popf\n"
                  "pushf\n"
                  "popl %0\n"
                  : "=a" (v0), "=d" (v1)
                  :
                  : "cc");
    return (v0 != v1);
}
B
bellard 已提交
108
#endif
B
bellard 已提交
109 110 111 112 113 114 115 116 117 118 119 120 121

static void kqemu_update_cpuid(CPUState *env)
{
    int critical_features_mask, features;
    uint32_t eax, ebx, ecx, edx;

    /* the following features are kept identical on the host and
       target cpus because they are important for user code. Strictly
       speaking, only SSE really matters because the OS must support
       it if the user code uses it. */
    critical_features_mask = 
        CPUID_CMOV | CPUID_CX8 | 
        CPUID_FXSR | CPUID_MMX | CPUID_SSE | 
B
bellard 已提交
122
        CPUID_SSE2 | CPUID_SEP;
B
bellard 已提交
123 124 125 126 127 128
    if (!is_cpuid_supported()) {
        features = 0;
    } else {
        cpuid(1, eax, ebx, ecx, edx);
        features = edx;
    }
B
bellard 已提交
129 130 131 132 133 134
#ifdef __x86_64__
    /* NOTE: on x86_64 CPUs, SYSENTER is not supported in
       compatibility mode, so in order to have the best performances
       it is better not to use it */
    features &= ~CPUID_SEP;
#endif
B
bellard 已提交
135 136 137 138 139 140 141 142 143 144 145
    env->cpuid_features = (env->cpuid_features & ~critical_features_mask) |
        (features & critical_features_mask);
    /* XXX: we could update more of the target CPUID state so that the
       non accelerated code sees exactly the same CPU features as the
       accelerated code */
}

int kqemu_init(CPUState *env)
{
    struct kqemu_init init;
    int ret, version;
146 147 148
#ifdef _WIN32
    DWORD temp;
#endif
B
bellard 已提交
149 150 151 152

    if (!kqemu_allowed)
        return -1;

153 154 155 156 157 158
#ifdef _WIN32
    kqemu_fd = CreateFile(KQEMU_DEVICE, GENERIC_WRITE | GENERIC_READ,
                          FILE_SHARE_READ | FILE_SHARE_WRITE,
                          NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL,
                          NULL);
#else
B
bellard 已提交
159
    kqemu_fd = open(KQEMU_DEVICE, O_RDWR);
160 161
#endif
    if (kqemu_fd == KQEMU_INVALID_FD) {
B
bellard 已提交
162 163 164 165
        fprintf(stderr, "Could not open '%s' - QEMU acceleration layer not activated\n", KQEMU_DEVICE);
        return -1;
    }
    version = 0;
166 167 168 169
#ifdef _WIN32
    DeviceIoControl(kqemu_fd, KQEMU_GET_VERSION, NULL, 0,
                    &version, sizeof(version), &temp, NULL);
#else
B
bellard 已提交
170
    ioctl(kqemu_fd, KQEMU_GET_VERSION, &version);
171
#endif
B
bellard 已提交
172 173 174 175 176 177 178 179 180 181 182
    if (version != KQEMU_VERSION) {
        fprintf(stderr, "Version mismatch between kqemu module and qemu (%08x %08x) - disabling kqemu use\n",
                version, KQEMU_VERSION);
        goto fail;
    }

    pages_to_flush = qemu_vmalloc(KQEMU_MAX_PAGES_TO_FLUSH * 
                                  sizeof(unsigned long));
    if (!pages_to_flush)
        goto fail;

B
bellard 已提交
183 184 185 186 187
    ram_pages_to_update = qemu_vmalloc(KQEMU_MAX_RAM_PAGES_TO_UPDATE * 
                                       sizeof(unsigned long));
    if (!ram_pages_to_update)
        goto fail;

B
bellard 已提交
188 189 190 191 192
    init.ram_base = phys_ram_base;
    init.ram_size = phys_ram_size;
    init.ram_dirty = phys_ram_dirty;
    init.phys_to_ram_map = l1_phys_map;
    init.pages_to_flush = pages_to_flush;
B
bellard 已提交
193 194 195
#if KQEMU_VERSION >= 0x010200
    init.ram_pages_to_update = ram_pages_to_update;
#endif
196 197 198 199
#ifdef _WIN32
    ret = DeviceIoControl(kqemu_fd, KQEMU_INIT, &init, sizeof(init),
                          NULL, 0, &temp, NULL) == TRUE ? 0 : -1;
#else
B
bellard 已提交
200
    ret = ioctl(kqemu_fd, KQEMU_INIT, &init);
201
#endif
B
bellard 已提交
202 203 204
    if (ret < 0) {
        fprintf(stderr, "Error %d while initializing QEMU acceleration layer - disabling it for now\n", ret);
    fail:
205 206
        kqemu_closefd(kqemu_fd);
        kqemu_fd = KQEMU_INVALID_FD;
B
bellard 已提交
207 208 209 210 211
        return -1;
    }
    kqemu_update_cpuid(env);
    env->kqemu_enabled = 1;
    nb_pages_to_flush = 0;
B
bellard 已提交
212
    nb_ram_pages_to_update = 0;
B
bellard 已提交
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238
    return 0;
}

void kqemu_flush_page(CPUState *env, target_ulong addr)
{
#ifdef DEBUG
    if (loglevel & CPU_LOG_INT) {
        fprintf(logfile, "kqemu_flush_page: addr=" TARGET_FMT_lx "\n", addr);
    }
#endif
    if (nb_pages_to_flush >= KQEMU_MAX_PAGES_TO_FLUSH)
        nb_pages_to_flush = KQEMU_FLUSH_ALL;
    else
        pages_to_flush[nb_pages_to_flush++] = addr;
}

void kqemu_flush(CPUState *env, int global)
{
#ifdef DEBUG
    if (loglevel & CPU_LOG_INT) {
        fprintf(logfile, "kqemu_flush:\n");
    }
#endif
    nb_pages_to_flush = KQEMU_FLUSH_ALL;
}

B
bellard 已提交
239 240 241 242 243 244 245
void kqemu_set_notdirty(CPUState *env, ram_addr_t ram_addr)
{
#ifdef DEBUG
    if (loglevel & CPU_LOG_INT) {
        fprintf(logfile, "kqemu_set_notdirty: addr=%08lx\n", ram_addr);
    }
#endif
B
bellard 已提交
246 247 248
    /* we only track transitions to dirty state */
    if (phys_ram_dirty[ram_addr >> TARGET_PAGE_BITS] != 0xff)
        return;
B
bellard 已提交
249 250 251 252 253 254
    if (nb_ram_pages_to_update >= KQEMU_MAX_RAM_PAGES_TO_UPDATE)
        nb_ram_pages_to_update = KQEMU_RAM_PAGES_UPDATE_ALL;
    else
        ram_pages_to_update[nb_ram_pages_to_update++] = ram_addr;
}

B
bellard 已提交
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
struct fpstate {
    uint16_t fpuc;
    uint16_t dummy1;
    uint16_t fpus;
    uint16_t dummy2;
    uint16_t fptag;
    uint16_t dummy3;

    uint32_t fpip;
    uint32_t fpcs;
    uint32_t fpoo;
    uint32_t fpos;
    uint8_t fpregs1[8 * 10];
};

struct fpxstate {
    uint16_t fpuc;
    uint16_t fpus;
    uint16_t fptag;
    uint16_t fop;
    uint32_t fpuip;
    uint16_t cs_sel;
    uint16_t dummy0;
    uint32_t fpudp;
    uint16_t ds_sel;
    uint16_t dummy1;
    uint32_t mxcsr;
    uint32_t mxcsr_mask;
    uint8_t fpregs1[8 * 16];
B
bellard 已提交
284 285
    uint8_t xmm_regs[16 * 16];
    uint8_t dummy2[96];
B
bellard 已提交
286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360
};

static struct fpxstate fpx1 __attribute__((aligned(16)));

static void restore_native_fp_frstor(CPUState *env)
{
    int fptag, i, j;
    struct fpstate fp1, *fp = &fp1;
    
    fp->fpuc = env->fpuc;
    fp->fpus = (env->fpus & ~0x3800) | (env->fpstt & 0x7) << 11;
    fptag = 0;
    for (i=7; i>=0; i--) {
	fptag <<= 2;
	if (env->fptags[i]) {
            fptag |= 3;
        } else {
            /* the FPU automatically computes it */
        }
    }
    fp->fptag = fptag;
    j = env->fpstt;
    for(i = 0;i < 8; i++) {
        memcpy(&fp->fpregs1[i * 10], &env->fpregs[j].d, 10);
        j = (j + 1) & 7;
    }
    asm volatile ("frstor %0" : "=m" (*fp));
}
 
static void save_native_fp_fsave(CPUState *env)
{
    int fptag, i, j;
    uint16_t fpuc;
    struct fpstate fp1, *fp = &fp1;

    asm volatile ("fsave %0" : : "m" (*fp));
    env->fpuc = fp->fpuc;
    env->fpstt = (fp->fpus >> 11) & 7;
    env->fpus = fp->fpus & ~0x3800;
    fptag = fp->fptag;
    for(i = 0;i < 8; i++) {
        env->fptags[i] = ((fptag & 3) == 3);
        fptag >>= 2;
    }
    j = env->fpstt;
    for(i = 0;i < 8; i++) {
        memcpy(&env->fpregs[j].d, &fp->fpregs1[i * 10], 10);
        j = (j + 1) & 7;
    }
    /* we must restore the default rounding state */
    fpuc = 0x037f | (env->fpuc & (3 << 10));
    asm volatile("fldcw %0" : : "m" (fpuc));
}

static void restore_native_fp_fxrstor(CPUState *env)
{
    struct fpxstate *fp = &fpx1;
    int i, j, fptag;

    fp->fpuc = env->fpuc;
    fp->fpus = (env->fpus & ~0x3800) | (env->fpstt & 0x7) << 11;
    fptag = 0;
    for(i = 0; i < 8; i++)
        fptag |= (env->fptags[i] << i);
    fp->fptag = fptag ^ 0xff;

    j = env->fpstt;
    for(i = 0;i < 8; i++) {
        memcpy(&fp->fpregs1[i * 16], &env->fpregs[j].d, 10);
        j = (j + 1) & 7;
    }
    if (env->cpuid_features & CPUID_SSE) {
        fp->mxcsr = env->mxcsr;
        /* XXX: check if DAZ is not available */
        fp->mxcsr_mask = 0xffff;
B
bellard 已提交
361
        memcpy(fp->xmm_regs, env->xmm_regs, CPU_NB_REGS * 16);
B
bellard 已提交
362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386
    }
    asm volatile ("fxrstor %0" : "=m" (*fp));
}

static void save_native_fp_fxsave(CPUState *env)
{
    struct fpxstate *fp = &fpx1;
    int fptag, i, j;
    uint16_t fpuc;

    asm volatile ("fxsave %0" : : "m" (*fp));
    env->fpuc = fp->fpuc;
    env->fpstt = (fp->fpus >> 11) & 7;
    env->fpus = fp->fpus & ~0x3800;
    fptag = fp->fptag ^ 0xff;
    for(i = 0;i < 8; i++) {
        env->fptags[i] = (fptag >> i) & 1;
    }
    j = env->fpstt;
    for(i = 0;i < 8; i++) {
        memcpy(&env->fpregs[j].d, &fp->fpregs1[i * 16], 10);
        j = (j + 1) & 7;
    }
    if (env->cpuid_features & CPUID_SSE) {
        env->mxcsr = fp->mxcsr;
B
bellard 已提交
387
        memcpy(env->xmm_regs, fp->xmm_regs, CPU_NB_REGS * 16);
B
bellard 已提交
388 389 390 391 392 393 394 395
    }

    /* we must restore the default rounding state */
    asm volatile ("fninit");
    fpuc = 0x037f | (env->fpuc & (3 << 10));
    asm volatile("fldcw %0" : : "m" (fpuc));
}

B
bellard 已提交
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
static int do_syscall(CPUState *env,
                      struct kqemu_cpu_state *kenv)
{
    int selector;
    
    selector = (env->star >> 32) & 0xffff;
#ifdef __x86_64__
    if (env->hflags & HF_LMA_MASK) {
        env->regs[R_ECX] = kenv->next_eip;
        env->regs[11] = env->eflags;

        cpu_x86_set_cpl(env, 0);
        cpu_x86_load_seg_cache(env, R_CS, selector & 0xfffc, 
                               0, 0xffffffff, 
                               DESC_G_MASK | DESC_B_MASK | DESC_P_MASK |
                               DESC_S_MASK |
                               DESC_CS_MASK | DESC_R_MASK | DESC_A_MASK | DESC_L_MASK);
        cpu_x86_load_seg_cache(env, R_SS, (selector + 8) & 0xfffc, 
                               0, 0xffffffff,
                               DESC_G_MASK | DESC_B_MASK | DESC_P_MASK |
                               DESC_S_MASK |
                               DESC_W_MASK | DESC_A_MASK);
        env->eflags &= ~env->fmask;
        if (env->hflags & HF_CS64_MASK)
            env->eip = env->lstar;
        else
            env->eip = env->cstar;
    } else 
#endif
    {
        env->regs[R_ECX] = (uint32_t)kenv->next_eip;
        
        cpu_x86_set_cpl(env, 0);
        cpu_x86_load_seg_cache(env, R_CS, selector & 0xfffc, 
                           0, 0xffffffff, 
                               DESC_G_MASK | DESC_B_MASK | DESC_P_MASK |
                               DESC_S_MASK |
                               DESC_CS_MASK | DESC_R_MASK | DESC_A_MASK);
        cpu_x86_load_seg_cache(env, R_SS, (selector + 8) & 0xfffc, 
                               0, 0xffffffff,
                               DESC_G_MASK | DESC_B_MASK | DESC_P_MASK |
                               DESC_S_MASK |
                               DESC_W_MASK | DESC_A_MASK);
        env->eflags &= ~(IF_MASK | RF_MASK | VM_MASK);
        env->eip = (uint32_t)env->star;
    }
    return 2;
}

B
bellard 已提交
445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541
#ifdef PROFILE

#define PC_REC_SIZE 1
#define PC_REC_HASH_BITS 16
#define PC_REC_HASH_SIZE (1 << PC_REC_HASH_BITS)

typedef struct PCRecord {
    unsigned long pc;
    int64_t count;
    struct PCRecord *next;
} PCRecord;

PCRecord *pc_rec_hash[PC_REC_HASH_SIZE];
int nb_pc_records;

void kqemu_record_pc(unsigned long pc)
{
    unsigned long h;
    PCRecord **pr, *r;

    h = pc / PC_REC_SIZE;
    h = h ^ (h >> PC_REC_HASH_BITS);
    h &= (PC_REC_HASH_SIZE - 1);
    pr = &pc_rec_hash[h];
    for(;;) {
        r = *pr;
        if (r == NULL)
            break;
        if (r->pc == pc) {
            r->count++;
            return;
        }
        pr = &r->next;
    }
    r = malloc(sizeof(PCRecord));
    r->count = 1;
    r->pc = pc;
    r->next = NULL;
    *pr = r;
    nb_pc_records++;
}

int pc_rec_cmp(const void *p1, const void *p2)
{
    PCRecord *r1 = *(PCRecord **)p1;
    PCRecord *r2 = *(PCRecord **)p2;
    if (r1->count < r2->count)
        return 1;
    else if (r1->count == r2->count)
        return 0;
    else
        return -1;
}

void kqemu_record_dump(void)
{
    PCRecord **pr, *r;
    int i, h;
    FILE *f;
    int64_t total, sum;

    pr = malloc(sizeof(PCRecord *) * nb_pc_records);
    i = 0;
    total = 0;
    for(h = 0; h < PC_REC_HASH_SIZE; h++) {
        for(r = pc_rec_hash[h]; r != NULL; r = r->next) {
            pr[i++] = r;
            total += r->count;
        }
    }
    qsort(pr, nb_pc_records, sizeof(PCRecord *), pc_rec_cmp);
    
    f = fopen("/tmp/kqemu.stats", "w");
    if (!f) {
        perror("/tmp/kqemu.stats");
        exit(1);
    }
    fprintf(f, "total: %lld\n", total);
    sum = 0;
    for(i = 0; i < nb_pc_records; i++) {
        r = pr[i];
        sum += r->count;
        fprintf(f, "%08lx: %lld %0.2f%% %0.2f%%\n", 
                r->pc, 
                r->count, 
                (double)r->count / (double)total * 100.0,
                (double)sum / (double)total * 100.0);
    }
    fclose(f);
    free(pr);
}
#else
void kqemu_record_dump(void)
{
}
#endif

B
bellard 已提交
542 543 544 545
int kqemu_cpu_exec(CPUState *env)
{
    struct kqemu_cpu_state kcpu_state, *kenv = &kcpu_state;
    int ret;
546 547 548
#ifdef _WIN32
    DWORD temp;
#endif
B
bellard 已提交
549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568

#ifdef DEBUG
    if (loglevel & CPU_LOG_INT) {
        fprintf(logfile, "kqemu: cpu_exec: enter\n");
        cpu_dump_state(env, logfile, fprintf, 0);
    }
#endif
    memcpy(kenv->regs, env->regs, sizeof(kenv->regs));
    kenv->eip = env->eip;
    kenv->eflags = env->eflags;
    memcpy(&kenv->segs, &env->segs, sizeof(env->segs));
    memcpy(&kenv->ldt, &env->ldt, sizeof(env->ldt));
    memcpy(&kenv->tr, &env->tr, sizeof(env->tr));
    memcpy(&kenv->gdt, &env->gdt, sizeof(env->gdt));
    memcpy(&kenv->idt, &env->idt, sizeof(env->idt));
    kenv->cr0 = env->cr[0];
    kenv->cr2 = env->cr[2];
    kenv->cr3 = env->cr[3];
    kenv->cr4 = env->cr[4];
    kenv->a20_mask = env->a20_mask;
B
bellard 已提交
569
#if KQEMU_VERSION >= 0x010100
B
bellard 已提交
570 571
    kenv->efer = env->efer;
#endif
B
bellard 已提交
572 573 574 575 576 577 578 579 580 581 582 583 584
    if (env->dr[7] & 0xff) {
        kenv->dr7 = env->dr[7];
        kenv->dr0 = env->dr[0];
        kenv->dr1 = env->dr[1];
        kenv->dr2 = env->dr[2];
        kenv->dr3 = env->dr[3];
    } else {
        kenv->dr7 = 0;
    }
    kenv->dr6 = env->dr[6];
    kenv->cpl = 3;
    kenv->nb_pages_to_flush = nb_pages_to_flush;
    nb_pages_to_flush = 0;
B
bellard 已提交
585 586 587 588 589
#if KQEMU_VERSION >= 0x010200
    kenv->user_only = 1;
    kenv->nb_ram_pages_to_update = nb_ram_pages_to_update;
#endif
    nb_ram_pages_to_update = 0;
B
bellard 已提交
590 591 592 593 594 595 596 597
    
    if (!(kenv->cr0 & CR0_TS_MASK)) {
        if (env->cpuid_features & CPUID_FXSR)
            restore_native_fp_fxrstor(env);
        else
            restore_native_fp_frstor(env);
    }

598 599 600 601 602 603 604 605 606 607 608
#ifdef _WIN32
    DeviceIoControl(kqemu_fd, KQEMU_EXEC,
		    kenv, sizeof(struct kqemu_cpu_state),
		    kenv, sizeof(struct kqemu_cpu_state),
		    &temp, NULL);
    ret = kenv->retval;
#else
#if KQEMU_VERSION >= 0x010100
    ioctl(kqemu_fd, KQEMU_EXEC, kenv);
    ret = kenv->retval;
#else
B
bellard 已提交
609
    ret = ioctl(kqemu_fd, KQEMU_EXEC, kenv);
610 611
#endif
#endif
B
bellard 已提交
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636
    if (!(kenv->cr0 & CR0_TS_MASK)) {
        if (env->cpuid_features & CPUID_FXSR)
            save_native_fp_fxsave(env);
        else
            save_native_fp_fsave(env);
    }

    memcpy(env->regs, kenv->regs, sizeof(env->regs));
    env->eip = kenv->eip;
    env->eflags = kenv->eflags;
    memcpy(env->segs, kenv->segs, sizeof(env->segs));
#if 0
    /* no need to restore that */
    memcpy(env->ldt, kenv->ldt, sizeof(env->ldt));
    memcpy(env->tr, kenv->tr, sizeof(env->tr));
    memcpy(env->gdt, kenv->gdt, sizeof(env->gdt));
    memcpy(env->idt, kenv->idt, sizeof(env->idt));
    env->cr[0] = kenv->cr0;
    env->cr[3] = kenv->cr3;
    env->cr[4] = kenv->cr4;
    env->a20_mask = kenv->a20_mask;
#endif
    env->cr[2] = kenv->cr2;
    env->dr[6] = kenv->dr6;

B
bellard 已提交
637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679
#if KQEMU_VERSION >= 0x010200
    if (kenv->nb_ram_pages_to_update > 0) {
        cpu_tlb_update_dirty(env);
    }
#endif

    /* restore the hidden flags */
    {
        unsigned int new_hflags;
#ifdef TARGET_X86_64
        if ((env->hflags & HF_LMA_MASK) && 
            (env->segs[R_CS].flags & DESC_L_MASK)) {
            /* long mode */
            new_hflags = HF_CS32_MASK | HF_SS32_MASK | HF_CS64_MASK;
        } else
#endif
        {
            /* legacy / compatibility case */
            new_hflags = (env->segs[R_CS].flags & DESC_B_MASK)
                >> (DESC_B_SHIFT - HF_CS32_SHIFT);
            new_hflags |= (env->segs[R_SS].flags & DESC_B_MASK)
                >> (DESC_B_SHIFT - HF_SS32_SHIFT);
            if (!(env->cr[0] & CR0_PE_MASK) || 
                   (env->eflags & VM_MASK) ||
                   !(env->hflags & HF_CS32_MASK)) {
                /* XXX: try to avoid this test. The problem comes from the
                   fact that is real mode or vm86 mode we only modify the
                   'base' and 'selector' fields of the segment cache to go
                   faster. A solution may be to force addseg to one in
                   translate-i386.c. */
                new_hflags |= HF_ADDSEG_MASK;
            } else {
                new_hflags |= ((env->segs[R_DS].base | 
                                env->segs[R_ES].base |
                                env->segs[R_SS].base) != 0) << 
                    HF_ADDSEG_SHIFT;
            }
        }
        env->hflags = (env->hflags & 
           ~(HF_CS32_MASK | HF_SS32_MASK | HF_CS64_MASK | HF_ADDSEG_MASK)) |
            new_hflags;
    }

B
bellard 已提交
680 681 682 683 684
#ifdef DEBUG
    if (loglevel & CPU_LOG_INT) {
        fprintf(logfile, "kqemu: kqemu_cpu_exec: ret=0x%x\n", ret);
    }
#endif
B
bellard 已提交
685 686 687 688
    if (ret == KQEMU_RET_SYSCALL) {
        /* syscall instruction */
        return do_syscall(env, kenv);
    } else 
B
bellard 已提交
689 690 691 692 693 694
    if ((ret & 0xff00) == KQEMU_RET_INT) {
        env->exception_index = ret & 0xff;
        env->error_code = 0;
        env->exception_is_int = 1;
        env->exception_next_eip = kenv->next_eip;
#ifdef DEBUG
B
bellard 已提交
695 696 697 698 699
        if (loglevel & CPU_LOG_INT) {
            fprintf(logfile, "kqemu: interrupt v=%02x:\n", 
                    env->exception_index);
            cpu_dump_state(env, logfile, fprintf, 0);
        }
B
bellard 已提交
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715
#endif
        return 1;
    } else if ((ret & 0xff00) == KQEMU_RET_EXCEPTION) {
        env->exception_index = ret & 0xff;
        env->error_code = kenv->error_code;
        env->exception_is_int = 0;
        env->exception_next_eip = 0;
#ifdef DEBUG
        if (loglevel & CPU_LOG_INT) {
            fprintf(logfile, "kqemu: exception v=%02x e=%04x:\n",
                    env->exception_index, env->error_code);
            cpu_dump_state(env, logfile, fprintf, 0);
        }
#endif
        return 1;
    } else if (ret == KQEMU_RET_INTR) {
B
bellard 已提交
716 717 718 719 720
#ifdef DEBUG
        if (loglevel & CPU_LOG_INT) {
            cpu_dump_state(env, logfile, fprintf, 0);
        }
#endif
B
bellard 已提交
721 722
        return 0;
    } else if (ret == KQEMU_RET_SOFTMMU) { 
B
bellard 已提交
723 724 725 726 727 728 729 730
#ifdef PROFILE
        kqemu_record_pc(env->eip + env->segs[R_CS].base);
#endif
#ifdef DEBUG
        if (loglevel & CPU_LOG_INT) {
            cpu_dump_state(env, logfile, fprintf, 0);
        }
#endif
B
bellard 已提交
731 732 733 734 735 736 737 738 739 740
        return 2;
    } else {
        cpu_dump_state(env, stderr, fprintf, 0);
        fprintf(stderr, "Unsupported return value: 0x%x\n", ret);
        exit(1);
    }
    return 0;
}

#endif