elfload.c 91.4 KB
Newer Older
1
/* This is the Linux kernel elf-loading code, ported into user space */
2 3
#include <sys/time.h>
#include <sys/param.h>
4 5 6 7 8 9 10

#include <stdio.h>
#include <sys/types.h>
#include <fcntl.h>
#include <errno.h>
#include <unistd.h>
#include <sys/mman.h>
11
#include <sys/resource.h>
12 13
#include <stdlib.h>
#include <string.h>
14
#include <time.h>
15

B
bellard 已提交
16
#include "qemu.h"
17
#include "disas/disas.h"
18

M
malc 已提交
19
#ifdef _ARCH_PPC64
M
malc 已提交
20 21 22
#undef ARCH_DLINFO
#undef ELF_PLATFORM
#undef ELF_HWCAP
23
#undef ELF_HWCAP2
M
malc 已提交
24 25 26 27 28
#undef ELF_CLASS
#undef ELF_DATA
#undef ELF_ARCH
#endif

29 30
#define ELF_OSABI   ELFOSABI_SYSV

31 32 33 34 35 36 37 38
/* from personality.h */

/*
 * Flags for bug emulation.
 *
 * These occupy the top three bytes.
 */
enum {
39 40 41 42 43 44 45 46 47 48 49
    ADDR_NO_RANDOMIZE = 0x0040000,      /* disable randomization of VA space */
    FDPIC_FUNCPTRS =    0x0080000,      /* userspace function ptrs point to
                                           descriptors (signal handling) */
    MMAP_PAGE_ZERO =    0x0100000,
    ADDR_COMPAT_LAYOUT = 0x0200000,
    READ_IMPLIES_EXEC = 0x0400000,
    ADDR_LIMIT_32BIT =  0x0800000,
    SHORT_INODE =       0x1000000,
    WHOLE_SECONDS =     0x2000000,
    STICKY_TIMEOUTS =   0x4000000,
    ADDR_LIMIT_3GB =    0x8000000,
50 51 52 53 54 55 56 57 58
};

/*
 * Personality types.
 *
 * These go in the low byte.  Avoid using the top bit, it will
 * conflict with error returns.
 */
enum {
59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81
    PER_LINUX =         0x0000,
    PER_LINUX_32BIT =   0x0000 | ADDR_LIMIT_32BIT,
    PER_LINUX_FDPIC =   0x0000 | FDPIC_FUNCPTRS,
    PER_SVR4 =          0x0001 | STICKY_TIMEOUTS | MMAP_PAGE_ZERO,
    PER_SVR3 =          0x0002 | STICKY_TIMEOUTS | SHORT_INODE,
    PER_SCOSVR3 =       0x0003 | STICKY_TIMEOUTS | WHOLE_SECONDS | SHORT_INODE,
    PER_OSR5 =          0x0003 | STICKY_TIMEOUTS | WHOLE_SECONDS,
    PER_WYSEV386 =      0x0004 | STICKY_TIMEOUTS | SHORT_INODE,
    PER_ISCR4 =         0x0005 | STICKY_TIMEOUTS,
    PER_BSD =           0x0006,
    PER_SUNOS =         0x0006 | STICKY_TIMEOUTS,
    PER_XENIX =         0x0007 | STICKY_TIMEOUTS | SHORT_INODE,
    PER_LINUX32 =       0x0008,
    PER_LINUX32_3GB =   0x0008 | ADDR_LIMIT_3GB,
    PER_IRIX32 =        0x0009 | STICKY_TIMEOUTS,/* IRIX5 32-bit */
    PER_IRIXN32 =       0x000a | STICKY_TIMEOUTS,/* IRIX6 new 32-bit */
    PER_IRIX64 =        0x000b | STICKY_TIMEOUTS,/* IRIX6 64-bit */
    PER_RISCOS =        0x000c,
    PER_SOLARIS =       0x000d | STICKY_TIMEOUTS,
    PER_UW7 =           0x000e | STICKY_TIMEOUTS | MMAP_PAGE_ZERO,
    PER_OSF4 =          0x000f,                  /* OSF/1 v4 */
    PER_HPUX =          0x0010,
    PER_MASK =          0x00ff,
82 83 84 85 86
};

/*
 * Return the base personality without flags.
 */
87
#define personality(pers)       (pers & PER_MASK)
88

89 90 91 92 93 94 95 96 97 98
/* this flag is uneffective under linux too, should be deleted */
#ifndef MAP_DENYWRITE
#define MAP_DENYWRITE 0
#endif

/* should probably go in elf.h */
#ifndef ELIBBAD
#define ELIBBAD 80
#endif

99 100 101 102 103 104
#ifdef TARGET_WORDS_BIGENDIAN
#define ELF_DATA        ELFDATA2MSB
#else
#define ELF_DATA        ELFDATA2LSB
#endif

105
#ifdef TARGET_ABI_MIPSN32
106 107
typedef abi_ullong      target_elf_greg_t;
#define tswapreg(ptr)   tswap64(ptr)
108 109 110 111 112
#else
typedef abi_ulong       target_elf_greg_t;
#define tswapreg(ptr)   tswapal(ptr)
#endif

113
#ifdef USE_UID16
114 115
typedef abi_ushort      target_uid_t;
typedef abi_ushort      target_gid_t;
116
#else
117 118
typedef abi_uint        target_uid_t;
typedef abi_uint        target_gid_t;
119
#endif
120
typedef abi_int         target_pid_t;
121

122 123
#ifdef TARGET_I386

124 125 126 127 128
#define ELF_PLATFORM get_elf_platform()

static const char *get_elf_platform(void)
{
    static char elf_platform[] = "i386";
129
    int family = object_property_get_int(OBJECT(thread_cpu), "family", NULL);
130 131 132 133 134 135 136 137 138 139 140
    if (family > 6)
        family = 6;
    if (family >= 3)
        elf_platform[1] = '0' + family;
    return elf_platform;
}

#define ELF_HWCAP get_elf_hwcap()

static uint32_t get_elf_hwcap(void)
{
141 142 143
    X86CPU *cpu = X86_CPU(thread_cpu);

    return cpu->env.features[FEAT_1_EDX];
144 145
}

146 147 148 149 150 151 152 153 154 155 156 157 158 159
#ifdef TARGET_X86_64
#define ELF_START_MMAP 0x2aaaaab000ULL
#define elf_check_arch(x) ( ((x) == ELF_ARCH) )

#define ELF_CLASS      ELFCLASS64
#define ELF_ARCH       EM_X86_64

static inline void init_thread(struct target_pt_regs *regs, struct image_info *infop)
{
    regs->rax = 0;
    regs->rsp = infop->start_stack;
    regs->rip = infop->entry;
}

160
#define ELF_NREG    27
A
Anthony Liguori 已提交
161
typedef target_elf_greg_t  target_elf_gregset_t[ELF_NREG];
162 163 164 165 166 167 168 169

/*
 * Note that ELF_NREG should be 29 as there should be place for
 * TRAPNO and ERR "registers" as well but linux doesn't dump
 * those.
 *
 * See linux kernel: arch/x86/include/asm/elf.h
 */
170
static void elf_core_copy_regs(target_elf_gregset_t *regs, const CPUX86State *env)
171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
{
    (*regs)[0] = env->regs[15];
    (*regs)[1] = env->regs[14];
    (*regs)[2] = env->regs[13];
    (*regs)[3] = env->regs[12];
    (*regs)[4] = env->regs[R_EBP];
    (*regs)[5] = env->regs[R_EBX];
    (*regs)[6] = env->regs[11];
    (*regs)[7] = env->regs[10];
    (*regs)[8] = env->regs[9];
    (*regs)[9] = env->regs[8];
    (*regs)[10] = env->regs[R_EAX];
    (*regs)[11] = env->regs[R_ECX];
    (*regs)[12] = env->regs[R_EDX];
    (*regs)[13] = env->regs[R_ESI];
    (*regs)[14] = env->regs[R_EDI];
    (*regs)[15] = env->regs[R_EAX]; /* XXX */
    (*regs)[16] = env->eip;
    (*regs)[17] = env->segs[R_CS].selector & 0xffff;
    (*regs)[18] = env->eflags;
    (*regs)[19] = env->regs[R_ESP];
    (*regs)[20] = env->segs[R_SS].selector & 0xffff;
    (*regs)[21] = env->segs[R_FS].selector & 0xffff;
    (*regs)[22] = env->segs[R_GS].selector & 0xffff;
    (*regs)[23] = env->segs[R_DS].selector & 0xffff;
    (*regs)[24] = env->segs[R_ES].selector & 0xffff;
    (*regs)[25] = env->segs[R_FS].selector & 0xffff;
    (*regs)[26] = env->segs[R_GS].selector & 0xffff;
}

201 202
#else

203 204 205 206 207 208 209 210 211 212
#define ELF_START_MMAP 0x80000000

/*
 * This is used to ensure we don't load something for the wrong architecture.
 */
#define elf_check_arch(x) ( ((x) == EM_386) || ((x) == EM_486) )

/*
 * These are used to set parameters in the core dumps.
 */
213 214
#define ELF_CLASS       ELFCLASS32
#define ELF_ARCH        EM_386
215

216 217
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
B
bellard 已提交
218 219 220
{
    regs->esp = infop->start_stack;
    regs->eip = infop->entry;
P
pbrook 已提交
221 222 223 224 225 226 227 228 229

    /* SVR4/i386 ABI (pages 3-31, 3-32) says that when the program
       starts %edx contains a pointer to a function which might be
       registered using `atexit'.  This provides a mean for the
       dynamic linker to call DT_FINI functions for shared libraries
       that have been loaded before the code runs.

       A value of 0 tells we have no such handler.  */
    regs->edx = 0;
B
bellard 已提交
230
}
231 232

#define ELF_NREG    17
A
Anthony Liguori 已提交
233
typedef target_elf_greg_t  target_elf_gregset_t[ELF_NREG];
234 235 236 237 238 239 240 241

/*
 * Note that ELF_NREG should be 19 as there should be place for
 * TRAPNO and ERR "registers" as well but linux doesn't dump
 * those.
 *
 * See linux kernel: arch/x86/include/asm/elf.h
 */
242
static void elf_core_copy_regs(target_elf_gregset_t *regs, const CPUX86State *env)
243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
{
    (*regs)[0] = env->regs[R_EBX];
    (*regs)[1] = env->regs[R_ECX];
    (*regs)[2] = env->regs[R_EDX];
    (*regs)[3] = env->regs[R_ESI];
    (*regs)[4] = env->regs[R_EDI];
    (*regs)[5] = env->regs[R_EBP];
    (*regs)[6] = env->regs[R_EAX];
    (*regs)[7] = env->segs[R_DS].selector & 0xffff;
    (*regs)[8] = env->segs[R_ES].selector & 0xffff;
    (*regs)[9] = env->segs[R_FS].selector & 0xffff;
    (*regs)[10] = env->segs[R_GS].selector & 0xffff;
    (*regs)[11] = env->regs[R_EAX]; /* XXX */
    (*regs)[12] = env->eip;
    (*regs)[13] = env->segs[R_CS].selector & 0xffff;
    (*regs)[14] = env->eflags;
    (*regs)[15] = env->regs[R_ESP];
    (*regs)[16] = env->segs[R_SS].selector & 0xffff;
}
262
#endif
B
bellard 已提交
263

264
#define USE_ELF_CORE_DUMP
265
#define ELF_EXEC_PAGESIZE       4096
B
bellard 已提交
266 267 268 269 270

#endif

#ifdef TARGET_ARM

271 272 273
#ifndef TARGET_AARCH64
/* 32 bit ARM definitions */

B
bellard 已提交
274 275
#define ELF_START_MMAP 0x80000000

A
Alexander Graf 已提交
276
#define elf_check_arch(x) ((x) == ELF_MACHINE)
B
bellard 已提交
277

A
Alexander Graf 已提交
278
#define ELF_ARCH        ELF_MACHINE
279
#define ELF_CLASS       ELFCLASS32
B
bellard 已提交
280

281 282
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
B
bellard 已提交
283
{
284
    abi_long stack = infop->start_stack;
B
bellard 已提交
285
    memset(regs, 0, sizeof(*regs));
A
Alexander Graf 已提交
286

B
bellard 已提交
287
    regs->ARM_cpsr = 0x10;
288
    if (infop->entry & 1)
289
        regs->ARM_cpsr |= CPSR_T;
290
    regs->ARM_pc = infop->entry & 0xfffffffe;
B
bellard 已提交
291
    regs->ARM_sp = infop->start_stack;
292 293 294
    /* FIXME - what to for failure of get_user()? */
    get_user_ual(regs->ARM_r2, stack + 8); /* envp */
    get_user_ual(regs->ARM_r1, stack + 4); /* envp */
B
bellard 已提交
295
    /* XXX: it seems that r0 is zeroed after ! */
P
pbrook 已提交
296 297
    regs->ARM_r0 = 0;
    /* For uClinux PIC binaries.  */
298
    /* XXX: Linux does this only on ARM with no MMU (do we care ?) */
P
pbrook 已提交
299
    regs->ARM_r10 = infop->start_data;
B
bellard 已提交
300 301
}

302
#define ELF_NREG    18
A
Anthony Liguori 已提交
303
typedef target_elf_greg_t  target_elf_gregset_t[ELF_NREG];
304

305
static void elf_core_copy_regs(target_elf_gregset_t *regs, const CPUARMState *env)
306
{
307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325
    (*regs)[0] = tswapreg(env->regs[0]);
    (*regs)[1] = tswapreg(env->regs[1]);
    (*regs)[2] = tswapreg(env->regs[2]);
    (*regs)[3] = tswapreg(env->regs[3]);
    (*regs)[4] = tswapreg(env->regs[4]);
    (*regs)[5] = tswapreg(env->regs[5]);
    (*regs)[6] = tswapreg(env->regs[6]);
    (*regs)[7] = tswapreg(env->regs[7]);
    (*regs)[8] = tswapreg(env->regs[8]);
    (*regs)[9] = tswapreg(env->regs[9]);
    (*regs)[10] = tswapreg(env->regs[10]);
    (*regs)[11] = tswapreg(env->regs[11]);
    (*regs)[12] = tswapreg(env->regs[12]);
    (*regs)[13] = tswapreg(env->regs[13]);
    (*regs)[14] = tswapreg(env->regs[14]);
    (*regs)[15] = tswapreg(env->regs[15]);

    (*regs)[16] = tswapreg(cpsr_read((CPUARMState *)env));
    (*regs)[17] = tswapreg(env->regs[0]); /* XXX */
326 327
}

328
#define USE_ELF_CORE_DUMP
329
#define ELF_EXEC_PAGESIZE       4096
330

B
bellard 已提交
331 332
enum
{
333 334 335 336 337 338 339 340 341 342
    ARM_HWCAP_ARM_SWP       = 1 << 0,
    ARM_HWCAP_ARM_HALF      = 1 << 1,
    ARM_HWCAP_ARM_THUMB     = 1 << 2,
    ARM_HWCAP_ARM_26BIT     = 1 << 3,
    ARM_HWCAP_ARM_FAST_MULT = 1 << 4,
    ARM_HWCAP_ARM_FPA       = 1 << 5,
    ARM_HWCAP_ARM_VFP       = 1 << 6,
    ARM_HWCAP_ARM_EDSP      = 1 << 7,
    ARM_HWCAP_ARM_JAVA      = 1 << 8,
    ARM_HWCAP_ARM_IWMMXT    = 1 << 9,
343 344 345 346 347
    ARM_HWCAP_ARM_CRUNCH    = 1 << 10,
    ARM_HWCAP_ARM_THUMBEE   = 1 << 11,
    ARM_HWCAP_ARM_NEON      = 1 << 12,
    ARM_HWCAP_ARM_VFPv3     = 1 << 13,
    ARM_HWCAP_ARM_VFPv3D16  = 1 << 14,
348 349 350 351 352 353 354
    ARM_HWCAP_ARM_TLS       = 1 << 15,
    ARM_HWCAP_ARM_VFPv4     = 1 << 16,
    ARM_HWCAP_ARM_IDIVA     = 1 << 17,
    ARM_HWCAP_ARM_IDIVT     = 1 << 18,
    ARM_HWCAP_ARM_VFPD32    = 1 << 19,
    ARM_HWCAP_ARM_LPAE      = 1 << 20,
    ARM_HWCAP_ARM_EVTSTRM   = 1 << 21,
B
bellard 已提交
355 356
};

357 358 359 360 361 362 363 364
enum {
    ARM_HWCAP2_ARM_AES      = 1 << 0,
    ARM_HWCAP2_ARM_PMULL    = 1 << 1,
    ARM_HWCAP2_ARM_SHA1     = 1 << 2,
    ARM_HWCAP2_ARM_SHA2     = 1 << 3,
    ARM_HWCAP2_ARM_CRC32    = 1 << 4,
};

365 366
/* The commpage only exists for 32 bit kernels */

367 368 369 370 371 372 373 374 375 376 377
#define TARGET_HAS_VALIDATE_GUEST_SPACE
/* Return 1 if the proposed guest space is suitable for the guest.
 * Return 0 if the proposed guest space isn't suitable, but another
 * address space should be tried.
 * Return -1 if there is no way the proposed guest space can be
 * valid regardless of the base.
 * The guest code may leave a page mapped and populate it if the
 * address is suitable.
 */
static int validate_guest_space(unsigned long guest_base,
                                unsigned long guest_size)
378 379 380 381 382 383 384
{
    unsigned long real_start, test_page_addr;

    /* We need to check that we can force a fault on access to the
     * commpage at 0xffff0fxx
     */
    test_page_addr = guest_base + (0xffff0f00 & qemu_host_page_mask);
385 386 387 388 389 390 391 392 393

    /* If the commpage lies within the already allocated guest space,
     * then there is no way we can allocate it.
     */
    if (test_page_addr >= guest_base
        && test_page_addr <= (guest_base + guest_size)) {
        return -1;
    }

394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425
    /* Note it needs to be writeable to let us initialise it */
    real_start = (unsigned long)
                 mmap((void *)test_page_addr, qemu_host_page_size,
                     PROT_READ | PROT_WRITE,
                     MAP_ANONYMOUS | MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);

    /* If we can't map it then try another address */
    if (real_start == -1ul) {
        return 0;
    }

    if (real_start != test_page_addr) {
        /* OS didn't put the page where we asked - unmap and reject */
        munmap((void *)real_start, qemu_host_page_size);
        return 0;
    }

    /* Leave the page mapped
     * Populate it (mmap should have left it all 0'd)
     */

    /* Kernel helper versions */
    __put_user(5, (uint32_t *)g2h(0xffff0ffcul));

    /* Now it's populated make it RO */
    if (mprotect((void *)test_page_addr, qemu_host_page_size, PROT_READ)) {
        perror("Protecting guest commpage");
        exit(-1);
    }

    return 1; /* All good */
}
426 427

#define ELF_HWCAP get_elf_hwcap()
428
#define ELF_HWCAP2 get_elf_hwcap2()
429 430 431

static uint32_t get_elf_hwcap(void)
{
432
    ARMCPU *cpu = ARM_CPU(thread_cpu);
433 434 435 436 437 438 439 440 441
    uint32_t hwcaps = 0;

    hwcaps |= ARM_HWCAP_ARM_SWP;
    hwcaps |= ARM_HWCAP_ARM_HALF;
    hwcaps |= ARM_HWCAP_ARM_THUMB;
    hwcaps |= ARM_HWCAP_ARM_FAST_MULT;

    /* probe for the extra features */
#define GET_FEATURE(feat, hwcap) \
442
    do { if (arm_feature(&cpu->env, feat)) { hwcaps |= hwcap; } } while (0)
443 444
    /* EDSP is in v5TE and above, but all our v5 CPUs are v5TE */
    GET_FEATURE(ARM_FEATURE_V5, ARM_HWCAP_ARM_EDSP);
445 446 447 448 449
    GET_FEATURE(ARM_FEATURE_VFP, ARM_HWCAP_ARM_VFP);
    GET_FEATURE(ARM_FEATURE_IWMMXT, ARM_HWCAP_ARM_IWMMXT);
    GET_FEATURE(ARM_FEATURE_THUMB2EE, ARM_HWCAP_ARM_THUMBEE);
    GET_FEATURE(ARM_FEATURE_NEON, ARM_HWCAP_ARM_NEON);
    GET_FEATURE(ARM_FEATURE_VFP3, ARM_HWCAP_ARM_VFPv3);
450 451 452 453 454 455 456 457 458 459 460
    GET_FEATURE(ARM_FEATURE_V6K, ARM_HWCAP_ARM_TLS);
    GET_FEATURE(ARM_FEATURE_VFP4, ARM_HWCAP_ARM_VFPv4);
    GET_FEATURE(ARM_FEATURE_ARM_DIV, ARM_HWCAP_ARM_IDIVA);
    GET_FEATURE(ARM_FEATURE_THUMB_DIV, ARM_HWCAP_ARM_IDIVT);
    /* All QEMU's VFPv3 CPUs have 32 registers, see VFP_DREG in translate.c.
     * Note that the ARM_HWCAP_ARM_VFPv3D16 bit is always the inverse of
     * ARM_HWCAP_ARM_VFPD32 (and so always clear for QEMU); it is unrelated
     * to our VFP_FP16 feature bit.
     */
    GET_FEATURE(ARM_FEATURE_VFP3, ARM_HWCAP_ARM_VFPD32);
    GET_FEATURE(ARM_FEATURE_LPAE, ARM_HWCAP_ARM_LPAE);
461 462 463

    return hwcaps;
}
B
bellard 已提交
464

465 466 467 468 469 470
static uint32_t get_elf_hwcap2(void)
{
    ARMCPU *cpu = ARM_CPU(thread_cpu);
    uint32_t hwcaps = 0;

    GET_FEATURE(ARM_FEATURE_V8_AES, ARM_HWCAP2_ARM_AES);
471
    GET_FEATURE(ARM_FEATURE_V8_PMULL, ARM_HWCAP2_ARM_PMULL);
472 473
    GET_FEATURE(ARM_FEATURE_V8_SHA1, ARM_HWCAP2_ARM_SHA1);
    GET_FEATURE(ARM_FEATURE_V8_SHA256, ARM_HWCAP2_ARM_SHA2);
474 475 476 477 478 479
    GET_FEATURE(ARM_FEATURE_CRC, ARM_HWCAP2_ARM_CRC32);
    return hwcaps;
}

#undef GET_FEATURE

480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549
#else
/* 64 bit ARM definitions */
#define ELF_START_MMAP 0x80000000

#define elf_check_arch(x) ((x) == ELF_MACHINE)

#define ELF_ARCH        ELF_MACHINE
#define ELF_CLASS       ELFCLASS64
#define ELF_PLATFORM    "aarch64"

static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
{
    abi_long stack = infop->start_stack;
    memset(regs, 0, sizeof(*regs));

    regs->pc = infop->entry & ~0x3ULL;
    regs->sp = stack;
}

#define ELF_NREG    34
typedef target_elf_greg_t  target_elf_gregset_t[ELF_NREG];

static void elf_core_copy_regs(target_elf_gregset_t *regs,
                               const CPUARMState *env)
{
    int i;

    for (i = 0; i < 32; i++) {
        (*regs)[i] = tswapreg(env->xregs[i]);
    }
    (*regs)[32] = tswapreg(env->pc);
    (*regs)[33] = tswapreg(pstate_read((CPUARMState *)env));
}

#define USE_ELF_CORE_DUMP
#define ELF_EXEC_PAGESIZE       4096

enum {
    ARM_HWCAP_A64_FP            = 1 << 0,
    ARM_HWCAP_A64_ASIMD         = 1 << 1,
    ARM_HWCAP_A64_EVTSTRM       = 1 << 2,
    ARM_HWCAP_A64_AES           = 1 << 3,
    ARM_HWCAP_A64_PMULL         = 1 << 4,
    ARM_HWCAP_A64_SHA1          = 1 << 5,
    ARM_HWCAP_A64_SHA2          = 1 << 6,
    ARM_HWCAP_A64_CRC32         = 1 << 7,
};

#define ELF_HWCAP get_elf_hwcap()

static uint32_t get_elf_hwcap(void)
{
    ARMCPU *cpu = ARM_CPU(thread_cpu);
    uint32_t hwcaps = 0;

    hwcaps |= ARM_HWCAP_A64_FP;
    hwcaps |= ARM_HWCAP_A64_ASIMD;

    /* probe for the extra features */
#define GET_FEATURE(feat, hwcap) \
    do { if (arm_feature(&cpu->env, feat)) { hwcaps |= hwcap; } } while (0)
    GET_FEATURE(ARM_FEATURE_V8_AES, ARM_HWCAP_A64_PMULL);
#undef GET_FEATURE

    return hwcaps;
}

#endif /* not TARGET_AARCH64 */
#endif /* TARGET_ARM */
550

551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578
#ifdef TARGET_UNICORE32

#define ELF_START_MMAP          0x80000000

#define elf_check_arch(x)       ((x) == EM_UNICORE32)

#define ELF_CLASS               ELFCLASS32
#define ELF_DATA                ELFDATA2LSB
#define ELF_ARCH                EM_UNICORE32

static inline void init_thread(struct target_pt_regs *regs,
        struct image_info *infop)
{
    abi_long stack = infop->start_stack;
    memset(regs, 0, sizeof(*regs));
    regs->UC32_REG_asr = 0x10;
    regs->UC32_REG_pc = infop->entry & 0xfffffffe;
    regs->UC32_REG_sp = infop->start_stack;
    /* FIXME - what to for failure of get_user()? */
    get_user_ual(regs->UC32_REG_02, stack + 8); /* envp */
    get_user_ual(regs->UC32_REG_01, stack + 4); /* envp */
    /* XXX: it seems that r0 is zeroed after ! */
    regs->UC32_REG_00 = 0;
}

#define ELF_NREG    34
typedef target_elf_greg_t  target_elf_gregset_t[ELF_NREG];

579
static void elf_core_copy_regs(target_elf_gregset_t *regs, const CPUUniCore32State *env)
580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613
{
    (*regs)[0] = env->regs[0];
    (*regs)[1] = env->regs[1];
    (*regs)[2] = env->regs[2];
    (*regs)[3] = env->regs[3];
    (*regs)[4] = env->regs[4];
    (*regs)[5] = env->regs[5];
    (*regs)[6] = env->regs[6];
    (*regs)[7] = env->regs[7];
    (*regs)[8] = env->regs[8];
    (*regs)[9] = env->regs[9];
    (*regs)[10] = env->regs[10];
    (*regs)[11] = env->regs[11];
    (*regs)[12] = env->regs[12];
    (*regs)[13] = env->regs[13];
    (*regs)[14] = env->regs[14];
    (*regs)[15] = env->regs[15];
    (*regs)[16] = env->regs[16];
    (*regs)[17] = env->regs[17];
    (*regs)[18] = env->regs[18];
    (*regs)[19] = env->regs[19];
    (*regs)[20] = env->regs[20];
    (*regs)[21] = env->regs[21];
    (*regs)[22] = env->regs[22];
    (*regs)[23] = env->regs[23];
    (*regs)[24] = env->regs[24];
    (*regs)[25] = env->regs[25];
    (*regs)[26] = env->regs[26];
    (*regs)[27] = env->regs[27];
    (*regs)[28] = env->regs[28];
    (*regs)[29] = env->regs[29];
    (*regs)[30] = env->regs[30];
    (*regs)[31] = env->regs[31];

614
    (*regs)[32] = cpu_asr_read((CPUUniCore32State *)env);
615 616 617 618 619 620 621 622 623 624
    (*regs)[33] = env->regs[0]; /* XXX */
}

#define USE_ELF_CORE_DUMP
#define ELF_EXEC_PAGESIZE               4096

#define ELF_HWCAP                       (UC32_HWCAP_CMOV | UC32_HWCAP_UCF64)

#endif

625
#ifdef TARGET_SPARC
B
bellard 已提交
626
#ifdef TARGET_SPARC64
627 628

#define ELF_START_MMAP 0x80000000
629 630
#define ELF_HWCAP  (HWCAP_SPARC_FLUSH | HWCAP_SPARC_STBAR | HWCAP_SPARC_SWAP \
                    | HWCAP_SPARC_MULDIV | HWCAP_SPARC_V9)
631
#ifndef TARGET_ABI32
632
#define elf_check_arch(x) ( (x) == EM_SPARCV9 || (x) == EM_SPARC32PLUS )
633 634 635
#else
#define elf_check_arch(x) ( (x) == EM_SPARC32PLUS || (x) == EM_SPARC )
#endif
636

B
bellard 已提交
637
#define ELF_CLASS   ELFCLASS64
638 639
#define ELF_ARCH    EM_SPARCV9

640
#define STACK_BIAS              2047
B
bellard 已提交
641

642 643
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
B
bellard 已提交
644
{
645
#ifndef TARGET_ABI32
B
bellard 已提交
646
    regs->tstate = 0;
647
#endif
B
bellard 已提交
648 649 650
    regs->pc = infop->entry;
    regs->npc = regs->pc + 4;
    regs->y = 0;
651 652 653
#ifdef TARGET_ABI32
    regs->u_regs[14] = infop->start_stack - 16 * 4;
#else
654 655 656 657
    if (personality(infop->personality) == PER_LINUX32)
        regs->u_regs[14] = infop->start_stack - 16 * 4;
    else
        regs->u_regs[14] = infop->start_stack - 16 * 8 - STACK_BIAS;
658
#endif
B
bellard 已提交
659 660 661 662
}

#else
#define ELF_START_MMAP 0x80000000
663 664
#define ELF_HWCAP  (HWCAP_SPARC_FLUSH | HWCAP_SPARC_STBAR | HWCAP_SPARC_SWAP \
                    | HWCAP_SPARC_MULDIV)
B
bellard 已提交
665 666
#define elf_check_arch(x) ( (x) == EM_SPARC )

667 668 669
#define ELF_CLASS   ELFCLASS32
#define ELF_ARCH    EM_SPARC

670 671
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
672
{
673 674 675 676 677
    regs->psr = 0;
    regs->pc = infop->entry;
    regs->npc = regs->pc + 4;
    regs->y = 0;
    regs->u_regs[14] = infop->start_stack - 16 * 4;
678 679
}

B
bellard 已提交
680
#endif
681 682
#endif

683 684 685 686
#ifdef TARGET_PPC

#define ELF_START_MMAP 0x80000000

687
#if defined(TARGET_PPC64) && !defined(TARGET_ABI32)
688 689 690

#define elf_check_arch(x) ( (x) == EM_PPC64 )

691
#define ELF_CLASS       ELFCLASS64
692 693 694

#else

695 696
#define elf_check_arch(x) ( (x) == EM_PPC )

697
#define ELF_CLASS       ELFCLASS32
698 699 700

#endif

701
#define ELF_ARCH        EM_PPC
702

N
Nathan Froyd 已提交
703 704 705
/* Feature masks for the Aux Vector Hardware Capabilities (AT_HWCAP).
   See arch/powerpc/include/asm/cputable.h.  */
enum {
706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734
    QEMU_PPC_FEATURE_32 = 0x80000000,
    QEMU_PPC_FEATURE_64 = 0x40000000,
    QEMU_PPC_FEATURE_601_INSTR = 0x20000000,
    QEMU_PPC_FEATURE_HAS_ALTIVEC = 0x10000000,
    QEMU_PPC_FEATURE_HAS_FPU = 0x08000000,
    QEMU_PPC_FEATURE_HAS_MMU = 0x04000000,
    QEMU_PPC_FEATURE_HAS_4xxMAC = 0x02000000,
    QEMU_PPC_FEATURE_UNIFIED_CACHE = 0x01000000,
    QEMU_PPC_FEATURE_HAS_SPE = 0x00800000,
    QEMU_PPC_FEATURE_HAS_EFP_SINGLE = 0x00400000,
    QEMU_PPC_FEATURE_HAS_EFP_DOUBLE = 0x00200000,
    QEMU_PPC_FEATURE_NO_TB = 0x00100000,
    QEMU_PPC_FEATURE_POWER4 = 0x00080000,
    QEMU_PPC_FEATURE_POWER5 = 0x00040000,
    QEMU_PPC_FEATURE_POWER5_PLUS = 0x00020000,
    QEMU_PPC_FEATURE_CELL = 0x00010000,
    QEMU_PPC_FEATURE_BOOKE = 0x00008000,
    QEMU_PPC_FEATURE_SMT = 0x00004000,
    QEMU_PPC_FEATURE_ICACHE_SNOOP = 0x00002000,
    QEMU_PPC_FEATURE_ARCH_2_05 = 0x00001000,
    QEMU_PPC_FEATURE_PA6T = 0x00000800,
    QEMU_PPC_FEATURE_HAS_DFP = 0x00000400,
    QEMU_PPC_FEATURE_POWER6_EXT = 0x00000200,
    QEMU_PPC_FEATURE_ARCH_2_06 = 0x00000100,
    QEMU_PPC_FEATURE_HAS_VSX = 0x00000080,
    QEMU_PPC_FEATURE_PSERIES_PERFMON_COMPAT = 0x00000040,

    QEMU_PPC_FEATURE_TRUE_LE = 0x00000002,
    QEMU_PPC_FEATURE_PPC_LE = 0x00000001,
N
Nathan Froyd 已提交
735 736 737 738 739 740
};

#define ELF_HWCAP get_elf_hwcap()

static uint32_t get_elf_hwcap(void)
{
741
    PowerPCCPU *cpu = POWERPC_CPU(thread_cpu);
N
Nathan Froyd 已提交
742 743 744 745
    uint32_t features = 0;

    /* We don't have to be terribly complete here; the high points are
       Altivec/FP/SPE support.  Anything else is just a bonus.  */
746
#define GET_FEATURE(flag, feature)                                      \
747
    do { if (cpu->env.insns_flags & flag) { features |= feature; } } while (0)
748 749 750 751 752 753 754 755
    GET_FEATURE(PPC_64B, QEMU_PPC_FEATURE_64);
    GET_FEATURE(PPC_FLOAT, QEMU_PPC_FEATURE_HAS_FPU);
    GET_FEATURE(PPC_ALTIVEC, QEMU_PPC_FEATURE_HAS_ALTIVEC);
    GET_FEATURE(PPC_SPE, QEMU_PPC_FEATURE_HAS_SPE);
    GET_FEATURE(PPC_SPE_SINGLE, QEMU_PPC_FEATURE_HAS_EFP_SINGLE);
    GET_FEATURE(PPC_SPE_DOUBLE, QEMU_PPC_FEATURE_HAS_EFP_DOUBLE);
    GET_FEATURE(PPC_BOOKE, QEMU_PPC_FEATURE_BOOKE);
    GET_FEATURE(PPC_405_MAC, QEMU_PPC_FEATURE_HAS_4xxMAC);
N
Nathan Froyd 已提交
756 757 758 759 760
#undef GET_FEATURE

    return features;
}

761 762 763 764 765 766 767 768 769
/*
 * The requirements here are:
 * - keep the final alignment of sp (sp & 0xf)
 * - make sure the 32-bit value at the first 16 byte aligned position of
 *   AUXV is greater than 16 for glibc compatibility.
 *   AT_IGNOREPPC is used for that.
 * - for compatibility with glibc ARCH_DLINFO must always be defined on PPC,
 *   even if DLINFO_ARCH_ITEMS goes to zero or is undefined.
 */
B
bellard 已提交
770
#define DLINFO_ARCH_ITEMS       5
771 772 773 774 775 776 777 778 779 780 781
#define ARCH_DLINFO                                     \
    do {                                                \
        NEW_AUX_ENT(AT_DCACHEBSIZE, 0x20);              \
        NEW_AUX_ENT(AT_ICACHEBSIZE, 0x20);              \
        NEW_AUX_ENT(AT_UCACHEBSIZE, 0);                 \
        /*                                              \
         * Now handle glibc compatibility.              \
         */                                             \
        NEW_AUX_ENT(AT_IGNOREPPC, AT_IGNOREPPC);        \
        NEW_AUX_ENT(AT_IGNOREPPC, AT_IGNOREPPC);        \
    } while (0)
782

783 784 785
static inline void init_thread(struct target_pt_regs *_regs, struct image_info *infop)
{
    _regs->gpr[1] = infop->start_stack;
786
#if defined(TARGET_PPC64) && !defined(TARGET_ABI32)
787 788
    _regs->gpr[2] = ldq_raw(infop->entry + 8) + infop->load_bias;
    infop->entry = ldq_raw(infop->entry) + infop->load_bias;
789
#endif
790 791 792
    _regs->nip = infop->entry;
}

793 794 795 796
/* See linux kernel: arch/powerpc/include/asm/elf.h.  */
#define ELF_NREG 48
typedef target_elf_greg_t target_elf_gregset_t[ELF_NREG];

797
static void elf_core_copy_regs(target_elf_gregset_t *regs, const CPUPPCState *env)
798 799 800 801 802
{
    int i;
    target_ulong ccr = 0;

    for (i = 0; i < ARRAY_SIZE(env->gpr); i++) {
803
        (*regs)[i] = tswapreg(env->gpr[i]);
804 805
    }

806 807 808 809 810
    (*regs)[32] = tswapreg(env->nip);
    (*regs)[33] = tswapreg(env->msr);
    (*regs)[35] = tswapreg(env->ctr);
    (*regs)[36] = tswapreg(env->lr);
    (*regs)[37] = tswapreg(env->xer);
811 812 813 814

    for (i = 0; i < ARRAY_SIZE(env->crf); i++) {
        ccr |= env->crf[i] << (32 - ((i + 1) * 4));
    }
815
    (*regs)[38] = tswapreg(ccr);
816 817 818
}

#define USE_ELF_CORE_DUMP
819
#define ELF_EXEC_PAGESIZE       4096
820 821 822

#endif

B
bellard 已提交
823 824 825 826 827 828
#ifdef TARGET_MIPS

#define ELF_START_MMAP 0x80000000

#define elf_check_arch(x) ( (x) == EM_MIPS )

T
ths 已提交
829 830 831
#ifdef TARGET_MIPS64
#define ELF_CLASS   ELFCLASS64
#else
B
bellard 已提交
832
#define ELF_CLASS   ELFCLASS32
T
ths 已提交
833
#endif
B
bellard 已提交
834 835
#define ELF_ARCH    EM_MIPS

836 837
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
B
bellard 已提交
838
{
T
ths 已提交
839
    regs->cp0_status = 2 << CP0St_KSU;
B
bellard 已提交
840 841 842 843
    regs->cp0_epc = infop->entry;
    regs->regs[29] = infop->start_stack;
}

844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
/* See linux kernel: arch/mips/include/asm/elf.h.  */
#define ELF_NREG 45
typedef target_elf_greg_t target_elf_gregset_t[ELF_NREG];

/* See linux kernel: arch/mips/include/asm/reg.h.  */
enum {
#ifdef TARGET_MIPS64
    TARGET_EF_R0 = 0,
#else
    TARGET_EF_R0 = 6,
#endif
    TARGET_EF_R26 = TARGET_EF_R0 + 26,
    TARGET_EF_R27 = TARGET_EF_R0 + 27,
    TARGET_EF_LO = TARGET_EF_R0 + 32,
    TARGET_EF_HI = TARGET_EF_R0 + 33,
    TARGET_EF_CP0_EPC = TARGET_EF_R0 + 34,
    TARGET_EF_CP0_BADVADDR = TARGET_EF_R0 + 35,
    TARGET_EF_CP0_STATUS = TARGET_EF_R0 + 36,
    TARGET_EF_CP0_CAUSE = TARGET_EF_R0 + 37
};

/* See linux kernel: arch/mips/kernel/process.c:elf_dump_regs.  */
866
static void elf_core_copy_regs(target_elf_gregset_t *regs, const CPUMIPSState *env)
867 868 869 870 871 872 873 874 875
{
    int i;

    for (i = 0; i < TARGET_EF_R0; i++) {
        (*regs)[i] = 0;
    }
    (*regs)[TARGET_EF_R0] = 0;

    for (i = 1; i < ARRAY_SIZE(env->active_tc.gpr); i++) {
876
        (*regs)[TARGET_EF_R0 + i] = tswapreg(env->active_tc.gpr[i]);
877 878 879 880
    }

    (*regs)[TARGET_EF_R26] = 0;
    (*regs)[TARGET_EF_R27] = 0;
881 882 883 884 885 886
    (*regs)[TARGET_EF_LO] = tswapreg(env->active_tc.LO[0]);
    (*regs)[TARGET_EF_HI] = tswapreg(env->active_tc.HI[0]);
    (*regs)[TARGET_EF_CP0_EPC] = tswapreg(env->active_tc.PC);
    (*regs)[TARGET_EF_CP0_BADVADDR] = tswapreg(env->CP0_BadVAddr);
    (*regs)[TARGET_EF_CP0_STATUS] = tswapreg(env->CP0_Status);
    (*regs)[TARGET_EF_CP0_CAUSE] = tswapreg(env->CP0_Cause);
887 888 889
}

#define USE_ELF_CORE_DUMP
T
ths 已提交
890 891
#define ELF_EXEC_PAGESIZE        4096

B
bellard 已提交
892 893
#endif /* TARGET_MIPS */

894 895 896 897
#ifdef TARGET_MICROBLAZE

#define ELF_START_MMAP 0x80000000

898
#define elf_check_arch(x) ( (x) == EM_MICROBLAZE || (x) == EM_MICROBLAZE_OLD)
899 900

#define ELF_CLASS   ELFCLASS32
901
#define ELF_ARCH    EM_MICROBLAZE
902

903 904
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
905 906 907 908 909 910 911 912
{
    regs->pc = infop->entry;
    regs->r1 = infop->start_stack;

}

#define ELF_EXEC_PAGESIZE        4096

913 914 915 916 917
#define USE_ELF_CORE_DUMP
#define ELF_NREG 38
typedef target_elf_greg_t target_elf_gregset_t[ELF_NREG];

/* See linux kernel: arch/mips/kernel/process.c:elf_dump_regs.  */
918
static void elf_core_copy_regs(target_elf_gregset_t *regs, const CPUMBState *env)
919 920 921 922
{
    int i, pos = 0;

    for (i = 0; i < 32; i++) {
923
        (*regs)[pos++] = tswapreg(env->regs[i]);
924 925 926
    }

    for (i = 0; i < 6; i++) {
927
        (*regs)[pos++] = tswapreg(env->sregs[i]);
928 929 930
    }
}

931 932
#endif /* TARGET_MICROBLAZE */

J
Jia Liu 已提交
933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962
#ifdef TARGET_OPENRISC

#define ELF_START_MMAP 0x08000000

#define elf_check_arch(x) ((x) == EM_OPENRISC)

#define ELF_ARCH EM_OPENRISC
#define ELF_CLASS ELFCLASS32
#define ELF_DATA  ELFDATA2MSB

static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
{
    regs->pc = infop->entry;
    regs->gpr[1] = infop->start_stack;
}

#define USE_ELF_CORE_DUMP
#define ELF_EXEC_PAGESIZE 8192

/* See linux kernel arch/openrisc/include/asm/elf.h.  */
#define ELF_NREG 34 /* gprs and pc, sr */
typedef target_elf_greg_t target_elf_gregset_t[ELF_NREG];

static void elf_core_copy_regs(target_elf_gregset_t *regs,
                               const CPUOpenRISCState *env)
{
    int i;

    for (i = 0; i < 32; i++) {
963
        (*regs)[i] = tswapreg(env->gpr[i]);
J
Jia Liu 已提交
964 965
    }

966 967
    (*regs)[32] = tswapreg(env->pc);
    (*regs)[33] = tswapreg(env->sr);
J
Jia Liu 已提交
968 969 970 971 972 973
}
#define ELF_HWCAP 0
#define ELF_PLATFORM NULL

#endif /* TARGET_OPENRISC */

B
bellard 已提交
974 975 976 977 978 979 980 981 982
#ifdef TARGET_SH4

#define ELF_START_MMAP 0x80000000

#define elf_check_arch(x) ( (x) == EM_SH )

#define ELF_CLASS ELFCLASS32
#define ELF_ARCH  EM_SH

983 984
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
B
bellard 已提交
985
{
986 987 988
    /* Check other registers XXXXX */
    regs->pc = infop->entry;
    regs->regs[15] = infop->start_stack;
B
bellard 已提交
989 990
}

991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
/* See linux kernel: arch/sh/include/asm/elf.h.  */
#define ELF_NREG 23
typedef target_elf_greg_t target_elf_gregset_t[ELF_NREG];

/* See linux kernel: arch/sh/include/asm/ptrace.h.  */
enum {
    TARGET_REG_PC = 16,
    TARGET_REG_PR = 17,
    TARGET_REG_SR = 18,
    TARGET_REG_GBR = 19,
    TARGET_REG_MACH = 20,
    TARGET_REG_MACL = 21,
    TARGET_REG_SYSCALL = 22
};

1006
static inline void elf_core_copy_regs(target_elf_gregset_t *regs,
1007
                                      const CPUSH4State *env)
1008 1009 1010 1011
{
    int i;

    for (i = 0; i < 16; i++) {
1012
        (*regs[i]) = tswapreg(env->gregs[i]);
1013 1014
    }

1015 1016 1017 1018 1019 1020
    (*regs)[TARGET_REG_PC] = tswapreg(env->pc);
    (*regs)[TARGET_REG_PR] = tswapreg(env->pr);
    (*regs)[TARGET_REG_SR] = tswapreg(env->sr);
    (*regs)[TARGET_REG_GBR] = tswapreg(env->gbr);
    (*regs)[TARGET_REG_MACH] = tswapreg(env->mach);
    (*regs)[TARGET_REG_MACL] = tswapreg(env->macl);
1021 1022 1023 1024
    (*regs)[TARGET_REG_SYSCALL] = 0; /* FIXME */
}

#define USE_ELF_CORE_DUMP
B
bellard 已提交
1025 1026 1027 1028
#define ELF_EXEC_PAGESIZE        4096

#endif

1029 1030 1031 1032 1033 1034 1035 1036 1037
#ifdef TARGET_CRIS

#define ELF_START_MMAP 0x80000000

#define elf_check_arch(x) ( (x) == EM_CRIS )

#define ELF_CLASS ELFCLASS32
#define ELF_ARCH  EM_CRIS

1038 1039
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
1040
{
1041
    regs->erp = infop->entry;
1042 1043 1044 1045 1046 1047
}

#define ELF_EXEC_PAGESIZE        8192

#endif

P
pbrook 已提交
1048 1049 1050 1051 1052 1053
#ifdef TARGET_M68K

#define ELF_START_MMAP 0x80000000

#define elf_check_arch(x) ( (x) == EM_68K )

1054 1055
#define ELF_CLASS       ELFCLASS32
#define ELF_ARCH        EM_68K
P
pbrook 已提交
1056 1057

/* ??? Does this need to do anything?
1058
   #define ELF_PLAT_INIT(_r) */
P
pbrook 已提交
1059

1060 1061
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
P
pbrook 已提交
1062 1063 1064 1065 1066 1067
{
    regs->usp = infop->start_stack;
    regs->sr = 0;
    regs->pc = infop->entry;
}

1068 1069 1070 1071
/* See linux kernel: arch/m68k/include/asm/elf.h.  */
#define ELF_NREG 20
typedef target_elf_greg_t target_elf_gregset_t[ELF_NREG];

1072
static void elf_core_copy_regs(target_elf_gregset_t *regs, const CPUM68KState *env)
1073
{
1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
    (*regs)[0] = tswapreg(env->dregs[1]);
    (*regs)[1] = tswapreg(env->dregs[2]);
    (*regs)[2] = tswapreg(env->dregs[3]);
    (*regs)[3] = tswapreg(env->dregs[4]);
    (*regs)[4] = tswapreg(env->dregs[5]);
    (*regs)[5] = tswapreg(env->dregs[6]);
    (*regs)[6] = tswapreg(env->dregs[7]);
    (*regs)[7] = tswapreg(env->aregs[0]);
    (*regs)[8] = tswapreg(env->aregs[1]);
    (*regs)[9] = tswapreg(env->aregs[2]);
    (*regs)[10] = tswapreg(env->aregs[3]);
    (*regs)[11] = tswapreg(env->aregs[4]);
    (*regs)[12] = tswapreg(env->aregs[5]);
    (*regs)[13] = tswapreg(env->aregs[6]);
    (*regs)[14] = tswapreg(env->dregs[0]);
    (*regs)[15] = tswapreg(env->aregs[7]);
    (*regs)[16] = tswapreg(env->dregs[0]); /* FIXME: orig_d0 */
    (*regs)[17] = tswapreg(env->sr);
    (*regs)[18] = tswapreg(env->pc);
1093 1094 1095 1096
    (*regs)[19] = 0;  /* FIXME: regs->format | regs->vector */
}

#define USE_ELF_CORE_DUMP
1097
#define ELF_EXEC_PAGESIZE       8192
P
pbrook 已提交
1098 1099 1100

#endif

1101 1102 1103 1104 1105 1106 1107 1108 1109
#ifdef TARGET_ALPHA

#define ELF_START_MMAP (0x30000000000ULL)

#define elf_check_arch(x) ( (x) == ELF_ARCH )

#define ELF_CLASS      ELFCLASS64
#define ELF_ARCH       EM_ALPHA

1110 1111
static inline void init_thread(struct target_pt_regs *regs,
                               struct image_info *infop)
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121
{
    regs->pc = infop->entry;
    regs->ps = 8;
    regs->usp = infop->start_stack;
}

#define ELF_EXEC_PAGESIZE        8192

#endif /* TARGET_ALPHA */

U
Ulrich Hecht 已提交
1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140
#ifdef TARGET_S390X

#define ELF_START_MMAP (0x20000000000ULL)

#define elf_check_arch(x) ( (x) == ELF_ARCH )

#define ELF_CLASS	ELFCLASS64
#define ELF_DATA	ELFDATA2MSB
#define ELF_ARCH	EM_S390

static inline void init_thread(struct target_pt_regs *regs, struct image_info *infop)
{
    regs->psw.addr = infop->entry;
    regs->psw.mask = PSW_MASK_64 | PSW_MASK_32;
    regs->gprs[15] = infop->start_stack;
}

#endif /* TARGET_S390X */

1141 1142 1143 1144 1145 1146 1147 1148
#ifndef ELF_PLATFORM
#define ELF_PLATFORM (NULL)
#endif

#ifndef ELF_HWCAP
#define ELF_HWCAP 0
#endif

1149
#ifdef TARGET_ABI32
1150
#undef ELF_CLASS
1151
#define ELF_CLASS ELFCLASS32
1152 1153 1154 1155
#undef bswaptls
#define bswaptls(ptr) bswap32s(ptr)
#endif

1156
#include "elf.h"
B
bellard 已提交
1157 1158 1159

struct exec
{
1160 1161 1162 1163 1164 1165 1166 1167
    unsigned int a_info;   /* Use macros N_MAGIC, etc for access */
    unsigned int a_text;   /* length of text, in bytes */
    unsigned int a_data;   /* length of data, in bytes */
    unsigned int a_bss;    /* length of uninitialized data area, in bytes */
    unsigned int a_syms;   /* length of symbol table data in file, in bytes */
    unsigned int a_entry;  /* start address */
    unsigned int a_trsize; /* length of relocation info for text, in bytes */
    unsigned int a_drsize; /* length of relocation info for data, in bytes */
B
bellard 已提交
1168 1169 1170 1171 1172 1173 1174 1175 1176
};


#define N_MAGIC(exec) ((exec).a_info & 0xffff)
#define OMAGIC 0407
#define NMAGIC 0410
#define ZMAGIC 0413
#define QMAGIC 0314

1177
/* Necessary parameters */
B
bellard 已提交
1178 1179 1180
#define TARGET_ELF_EXEC_PAGESIZE TARGET_PAGE_SIZE
#define TARGET_ELF_PAGESTART(_v) ((_v) & ~(unsigned long)(TARGET_ELF_EXEC_PAGESIZE-1))
#define TARGET_ELF_PAGEOFFSET(_v) ((_v) & (TARGET_ELF_EXEC_PAGESIZE-1))
1181

J
James Hogan 已提交
1182
#define DLINFO_ITEMS 14
1183

B
bellard 已提交
1184 1185
static inline void memcpy_fromfs(void * to, const void * from, unsigned long n)
{
1186
    memcpy(to, from, n);
B
bellard 已提交
1187
}
1188

1189
#ifdef BSWAP_NEEDED
B
bellard 已提交
1190
static void bswap_ehdr(struct elfhdr *ehdr)
1191
{
1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204
    bswap16s(&ehdr->e_type);            /* Object file type */
    bswap16s(&ehdr->e_machine);         /* Architecture */
    bswap32s(&ehdr->e_version);         /* Object file version */
    bswaptls(&ehdr->e_entry);           /* Entry point virtual address */
    bswaptls(&ehdr->e_phoff);           /* Program header table file offset */
    bswaptls(&ehdr->e_shoff);           /* Section header table file offset */
    bswap32s(&ehdr->e_flags);           /* Processor-specific flags */
    bswap16s(&ehdr->e_ehsize);          /* ELF header size in bytes */
    bswap16s(&ehdr->e_phentsize);       /* Program header table entry size */
    bswap16s(&ehdr->e_phnum);           /* Program header table entry count */
    bswap16s(&ehdr->e_shentsize);       /* Section header table entry size */
    bswap16s(&ehdr->e_shnum);           /* Section header table entry count */
    bswap16s(&ehdr->e_shstrndx);        /* Section header string table index */
1205 1206
}

1207
static void bswap_phdr(struct elf_phdr *phdr, int phnum)
1208
{
1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
    int i;
    for (i = 0; i < phnum; ++i, ++phdr) {
        bswap32s(&phdr->p_type);        /* Segment type */
        bswap32s(&phdr->p_flags);       /* Segment flags */
        bswaptls(&phdr->p_offset);      /* Segment file offset */
        bswaptls(&phdr->p_vaddr);       /* Segment virtual address */
        bswaptls(&phdr->p_paddr);       /* Segment physical address */
        bswaptls(&phdr->p_filesz);      /* Segment size in file */
        bswaptls(&phdr->p_memsz);       /* Segment size in memory */
        bswaptls(&phdr->p_align);       /* Segment alignment */
    }
1220
}
B
bellard 已提交
1221

1222
static void bswap_shdr(struct elf_shdr *shdr, int shnum)
B
bellard 已提交
1223
{
1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
    int i;
    for (i = 0; i < shnum; ++i, ++shdr) {
        bswap32s(&shdr->sh_name);
        bswap32s(&shdr->sh_type);
        bswaptls(&shdr->sh_flags);
        bswaptls(&shdr->sh_addr);
        bswaptls(&shdr->sh_offset);
        bswaptls(&shdr->sh_size);
        bswap32s(&shdr->sh_link);
        bswap32s(&shdr->sh_info);
        bswaptls(&shdr->sh_addralign);
        bswaptls(&shdr->sh_entsize);
    }
B
bellard 已提交
1237 1238
}

1239
static void bswap_sym(struct elf_sym *sym)
B
bellard 已提交
1240 1241
{
    bswap32s(&sym->st_name);
1242 1243
    bswaptls(&sym->st_value);
    bswaptls(&sym->st_size);
B
bellard 已提交
1244 1245
    bswap16s(&sym->st_shndx);
}
1246 1247 1248 1249 1250
#else
static inline void bswap_ehdr(struct elfhdr *ehdr) { }
static inline void bswap_phdr(struct elf_phdr *phdr, int phnum) { }
static inline void bswap_shdr(struct elf_shdr *shdr, int shnum) { }
static inline void bswap_sym(struct elf_sym *sym) { }
1251 1252
#endif

1253
#ifdef USE_ELF_CORE_DUMP
1254
static int elf_core_dump(int, const CPUArchState *);
1255
#endif /* USE_ELF_CORE_DUMP */
1256
static void load_symbols(struct elfhdr *hdr, int fd, abi_ulong load_bias);
1257

1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281
/* Verify the portions of EHDR within E_IDENT for the target.
   This can be performed before bswapping the entire header.  */
static bool elf_check_ident(struct elfhdr *ehdr)
{
    return (ehdr->e_ident[EI_MAG0] == ELFMAG0
            && ehdr->e_ident[EI_MAG1] == ELFMAG1
            && ehdr->e_ident[EI_MAG2] == ELFMAG2
            && ehdr->e_ident[EI_MAG3] == ELFMAG3
            && ehdr->e_ident[EI_CLASS] == ELF_CLASS
            && ehdr->e_ident[EI_DATA] == ELF_DATA
            && ehdr->e_ident[EI_VERSION] == EV_CURRENT);
}

/* Verify the portions of EHDR outside of E_IDENT for the target.
   This has to wait until after bswapping the header.  */
static bool elf_check_ehdr(struct elfhdr *ehdr)
{
    return (elf_check_arch(ehdr->e_machine)
            && ehdr->e_ehsize == sizeof(struct elfhdr)
            && ehdr->e_phentsize == sizeof(struct elf_phdr)
            && ehdr->e_shentsize == sizeof(struct elf_shdr)
            && (ehdr->e_type == ET_EXEC || ehdr->e_type == ET_DYN));
}

1282
/*
P
pbrook 已提交
1283
 * 'copy_elf_strings()' copies argument/envelope strings from user
1284 1285 1286 1287
 * memory to free pages in kernel mem. These are in a format ready
 * to be put directly into the top of new user memory.
 *
 */
1288 1289
static abi_ulong copy_elf_strings(int argc,char ** argv, void **page,
                                  abi_ulong p)
1290 1291 1292 1293 1294
{
    char *tmp, *tmp1, *pag = NULL;
    int len, offset = 0;

    if (!p) {
1295
        return 0;       /* bullet-proofing */
1296 1297
    }
    while (argc-- > 0) {
1298 1299
        tmp = argv[argc];
        if (!tmp) {
1300 1301 1302
            fprintf(stderr, "VFS: argc is wrong");
            exit(-1);
        }
1303
        tmp1 = tmp;
1304 1305 1306 1307 1308 1309 1310 1311 1312
        while (*tmp++);
        len = tmp - tmp1;
        if (p < len) {  /* this shouldn't happen - 128kB */
            return 0;
        }
        while (len) {
            --p; --tmp; --len;
            if (--offset < 0) {
                offset = p % TARGET_PAGE_SIZE;
1313
                pag = (char *)page[p/TARGET_PAGE_SIZE];
B
bellard 已提交
1314
                if (!pag) {
1315
                    pag = g_try_malloc0(TARGET_PAGE_SIZE);
1316
                    page[p/TARGET_PAGE_SIZE] = pag;
B
bellard 已提交
1317 1318
                    if (!pag)
                        return 0;
1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332
                }
            }
            if (len == 0 || offset == 0) {
                *(pag + offset) = *tmp;
            }
            else {
                int bytes_to_copy = (len > offset) ? offset : len;
                tmp -= bytes_to_copy;
                p -= bytes_to_copy;
                offset -= bytes_to_copy;
                len -= bytes_to_copy;
                memcpy_fromfs(pag + offset, tmp, bytes_to_copy + 1);
            }
        }
1333 1334 1335 1336
    }
    return p;
}

1337 1338
static abi_ulong setup_arg_pages(abi_ulong p, struct linux_binprm *bprm,
                                 struct image_info *info)
1339
{
1340
    abi_ulong stack_base, size, error, guard;
1341 1342
    int i;

B
bellard 已提交
1343
    /* Create enough stack to hold everything.  If we don't use
1344
       it for args, we'll use it for something else.  */
1345
    size = guest_stack_size;
1346
    if (size < MAX_ARG_PAGES*TARGET_PAGE_SIZE) {
B
bellard 已提交
1347
        size = MAX_ARG_PAGES*TARGET_PAGE_SIZE;
1348 1349 1350 1351 1352 1353 1354 1355
    }
    guard = TARGET_PAGE_SIZE;
    if (guard < qemu_real_host_page_size) {
        guard = qemu_real_host_page_size;
    }

    error = target_mmap(0, size + guard, PROT_READ | PROT_WRITE,
                        MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
B
bellard 已提交
1356
    if (error == -1) {
1357
        perror("mmap stack");
B
bellard 已提交
1358 1359
        exit(-1);
    }
1360

1361 1362 1363 1364 1365
    /* We reserve one extra page at the top of the stack as guard.  */
    target_mprotect(error, guard, PROT_NONE);

    info->stack_limit = error + guard;
    stack_base = info->stack_limit + size - MAX_ARG_PAGES*TARGET_PAGE_SIZE;
1366
    p += stack_base;
B
bellard 已提交
1367

1368
    for (i = 0 ; i < MAX_ARG_PAGES ; i++) {
1369 1370
        if (bprm->page[i]) {
            info->rss++;
1371
            /* FIXME - check return value of memcpy_to_target() for failure */
1372
            memcpy_to_target(stack_base, bprm->page[i], TARGET_PAGE_SIZE);
1373
            g_free(bprm->page[i]);
1374
        }
1375
        stack_base += TARGET_PAGE_SIZE;
1376 1377 1378 1379
    }
    return p;
}

1380 1381 1382
/* Map and zero the bss.  We need to explicitly zero any fractional pages
   after the data section (i.e. bss).  */
static void zero_bss(abi_ulong elf_bss, abi_ulong last_bss, int prot)
1383
{
1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409
    uintptr_t host_start, host_map_start, host_end;

    last_bss = TARGET_PAGE_ALIGN(last_bss);

    /* ??? There is confusion between qemu_real_host_page_size and
       qemu_host_page_size here and elsewhere in target_mmap, which
       may lead to the end of the data section mapping from the file
       not being mapped.  At least there was an explicit test and
       comment for that here, suggesting that "the file size must
       be known".  The comment probably pre-dates the introduction
       of the fstat system call in target_mmap which does in fact
       find out the size.  What isn't clear is if the workaround
       here is still actually needed.  For now, continue with it,
       but merge it with the "normal" mmap that would allocate the bss.  */

    host_start = (uintptr_t) g2h(elf_bss);
    host_end = (uintptr_t) g2h(last_bss);
    host_map_start = (host_start + qemu_real_host_page_size - 1);
    host_map_start &= -qemu_real_host_page_size;

    if (host_map_start < host_end) {
        void *p = mmap((void *)host_map_start, host_end - host_map_start,
                       prot, MAP_FIXED | MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
        if (p == MAP_FAILED) {
            perror("cannot mmap brk");
            exit(-1);
1410 1411
        }

1412 1413 1414 1415
        /* Since we didn't use target_mmap, make sure to record
           the validity of the pages with qemu.  */
        page_set_flags(elf_bss & TARGET_PAGE_MASK, last_bss, prot|PAGE_VALID);
    }
1416

1417 1418 1419 1420
    if (host_start < host_map_start) {
        memset((void *)host_start, 0, host_map_start - host_start);
    }
}
1421

1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448
#ifdef CONFIG_USE_FDPIC
static abi_ulong loader_build_fdpic_loadmap(struct image_info *info, abi_ulong sp)
{
    uint16_t n;
    struct elf32_fdpic_loadseg *loadsegs = info->loadsegs;

    /* elf32_fdpic_loadseg */
    n = info->nsegs;
    while (n--) {
        sp -= 12;
        put_user_u32(loadsegs[n].addr, sp+0);
        put_user_u32(loadsegs[n].p_vaddr, sp+4);
        put_user_u32(loadsegs[n].p_memsz, sp+8);
    }

    /* elf32_fdpic_loadmap */
    sp -= 4;
    put_user_u16(0, sp+0); /* version */
    put_user_u16(info->nsegs, sp+2); /* nsegs */

    info->personality = PER_LINUX_FDPIC;
    info->loadmap_addr = sp;

    return sp;
}
#endif

1449
static abi_ulong create_elf_tables(abi_ulong p, int argc, int envc,
1450 1451 1452
                                   struct elfhdr *exec,
                                   struct image_info *info,
                                   struct image_info *interp_info)
1453
{
1454
    abi_ulong sp;
A
Alexander Graf 已提交
1455
    abi_ulong sp_auxv;
1456
    int size;
1457 1458 1459
    int i;
    abi_ulong u_rand_bytes;
    uint8_t k_rand_bytes[16];
1460 1461 1462 1463 1464
    abi_ulong u_platform;
    const char *k_platform;
    const int n = sizeof(elf_addr_t);

    sp = p;
1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479

#ifdef CONFIG_USE_FDPIC
    /* Needs to be before we load the env/argc/... */
    if (elf_is_fdpic(exec)) {
        /* Need 4 byte alignment for these structs */
        sp &= ~3;
        sp = loader_build_fdpic_loadmap(info, sp);
        info->other_info = interp_info;
        if (interp_info) {
            interp_info->other_info = info;
            sp = loader_build_fdpic_loadmap(interp_info, sp);
        }
    }
#endif

1480 1481 1482 1483 1484 1485 1486 1487 1488
    u_platform = 0;
    k_platform = ELF_PLATFORM;
    if (k_platform) {
        size_t len = strlen(k_platform) + 1;
        sp -= (len + n - 1) & ~(n - 1);
        u_platform = sp;
        /* FIXME - check return value of memcpy_to_target() for failure */
        memcpy_to_target(sp, k_platform, len);
    }
1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502

    /*
     * Generate 16 random bytes for userspace PRNG seeding (not
     * cryptically secure but it's not the aim of QEMU).
     */
    srand((unsigned int) time(NULL));
    for (i = 0; i < 16; i++) {
        k_rand_bytes[i] = rand();
    }
    sp -= 16;
    u_rand_bytes = sp;
    /* FIXME - check return value of memcpy_to_target() for failure */
    memcpy_to_target(sp, k_rand_bytes, 16);

1503 1504 1505 1506 1507 1508 1509
    /*
     * Force 16 byte _final_ alignment here for generality.
     */
    sp = sp &~ (abi_ulong)15;
    size = (DLINFO_ITEMS + 1) * 2;
    if (k_platform)
        size += 2;
1510
#ifdef DLINFO_ARCH_ITEMS
1511
    size += DLINFO_ARCH_ITEMS * 2;
1512 1513 1514
#endif
#ifdef ELF_HWCAP2
    size += 2;
1515
#endif
1516
    size += envc + argc + 2;
1517
    size += 1;  /* argc itself */
1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529
    size *= n;
    if (size & 15)
        sp -= 16 - (size & 15);

    /* This is correct because Linux defines
     * elf_addr_t as Elf32_Off / Elf64_Off
     */
#define NEW_AUX_ENT(id, val) do {               \
        sp -= n; put_user_ual(val, sp);         \
        sp -= n; put_user_ual(id, sp);          \
    } while(0)

A
Alexander Graf 已提交
1530
    sp_auxv = sp;
1531 1532 1533
    NEW_AUX_ENT (AT_NULL, 0);

    /* There must be exactly DLINFO_ITEMS entries here.  */
1534
    NEW_AUX_ENT(AT_PHDR, (abi_ulong)(info->load_addr + exec->e_phoff));
1535 1536 1537
    NEW_AUX_ENT(AT_PHENT, (abi_ulong)(sizeof (struct elf_phdr)));
    NEW_AUX_ENT(AT_PHNUM, (abi_ulong)(exec->e_phnum));
    NEW_AUX_ENT(AT_PAGESZ, (abi_ulong)(TARGET_PAGE_SIZE));
1538
    NEW_AUX_ENT(AT_BASE, (abi_ulong)(interp_info ? interp_info->load_addr : 0));
1539
    NEW_AUX_ENT(AT_FLAGS, (abi_ulong)0);
1540
    NEW_AUX_ENT(AT_ENTRY, info->entry);
1541 1542 1543 1544 1545 1546
    NEW_AUX_ENT(AT_UID, (abi_ulong) getuid());
    NEW_AUX_ENT(AT_EUID, (abi_ulong) geteuid());
    NEW_AUX_ENT(AT_GID, (abi_ulong) getgid());
    NEW_AUX_ENT(AT_EGID, (abi_ulong) getegid());
    NEW_AUX_ENT(AT_HWCAP, (abi_ulong) ELF_HWCAP);
    NEW_AUX_ENT(AT_CLKTCK, (abi_ulong) sysconf(_SC_CLK_TCK));
1547 1548
    NEW_AUX_ENT(AT_RANDOM, (abi_ulong) u_rand_bytes);

1549 1550 1551 1552
#ifdef ELF_HWCAP2
    NEW_AUX_ENT(AT_HWCAP2, (abi_ulong) ELF_HWCAP2);
#endif

1553 1554
    if (k_platform)
        NEW_AUX_ENT(AT_PLATFORM, u_platform);
1555
#ifdef ARCH_DLINFO
1556 1557 1558 1559 1560
    /*
     * ARCH_DLINFO must come last so platform specific code can enforce
     * special alignment requirements on the AUXV if necessary (eg. PPC).
     */
    ARCH_DLINFO;
1561 1562 1563
#endif
#undef NEW_AUX_ENT

1564
    info->saved_auxv = sp;
A
Alexander Graf 已提交
1565
    info->auxv_len = sp_auxv - sp;
1566

1567
    sp = loader_build_argptr(envc, argc, sp, p, 0);
1568 1569
    /* Check the right amount of stack was allocated for auxvec, envp & argv. */
    assert(sp_auxv - sp == size);
1570
    return sp;
1571 1572
}

1573
#ifndef TARGET_HAS_VALIDATE_GUEST_SPACE
1574
/* If the guest doesn't have a validation function just agree */
1575 1576
static int validate_guest_space(unsigned long guest_base,
                                unsigned long guest_size)
1577 1578 1579 1580 1581
{
    return 1;
}
#endif

1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594
unsigned long init_guest_space(unsigned long host_start,
                               unsigned long host_size,
                               unsigned long guest_start,
                               bool fixed)
{
    unsigned long current_start, real_start;
    int flags;

    assert(host_start || host_size);

    /* If just a starting address is given, then just verify that
     * address.  */
    if (host_start && !host_size) {
1595
        if (validate_guest_space(host_start, host_size) == 1) {
1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
            return host_start;
        } else {
            return (unsigned long)-1;
        }
    }

    /* Setup the initial flags and start address.  */
    current_start = host_start & qemu_host_page_mask;
    flags = MAP_ANONYMOUS | MAP_PRIVATE | MAP_NORESERVE;
    if (fixed) {
        flags |= MAP_FIXED;
    }

    /* Otherwise, a non-zero size region of memory needs to be mapped
     * and validated.  */
    while (1) {
1612 1613
        unsigned long real_size = host_size;

1614 1615 1616 1617 1618 1619 1620 1621 1622 1623
        /* Do not use mmap_find_vma here because that is limited to the
         * guest address space.  We are going to make the
         * guest address space fit whatever we're given.
         */
        real_start = (unsigned long)
            mmap((void *)current_start, host_size, PROT_NONE, flags, -1, 0);
        if (real_start == (unsigned long)-1) {
            return (unsigned long)-1;
        }

1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645
        /* Ensure the address is properly aligned.  */
        if (real_start & ~qemu_host_page_mask) {
            munmap((void *)real_start, host_size);
            real_size = host_size + qemu_host_page_size;
            real_start = (unsigned long)
                mmap((void *)real_start, real_size, PROT_NONE, flags, -1, 0);
            if (real_start == (unsigned long)-1) {
                return (unsigned long)-1;
            }
            real_start = HOST_PAGE_ALIGN(real_start);
        }

        /* Check to see if the address is valid.  */
        if (!host_start || real_start == current_start) {
            int valid = validate_guest_space(real_start - guest_start,
                                             real_size);
            if (valid == 1) {
                break;
            } else if (valid == -1) {
                return (unsigned long)-1;
            }
            /* valid == 0, so try again. */
1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666
        }

        /* That address didn't work.  Unmap and try a different one.
         * The address the host picked because is typically right at
         * the top of the host address space and leaves the guest with
         * no usable address space.  Resort to a linear search.  We
         * already compensated for mmap_min_addr, so this should not
         * happen often.  Probably means we got unlucky and host
         * address space randomization put a shared library somewhere
         * inconvenient.
         */
        munmap((void *)real_start, host_size);
        current_start += qemu_host_page_size;
        if (host_start == current_start) {
            /* Theoretically possible if host doesn't have any suitably
             * aligned areas.  Normally the first mmap will fail.
             */
            return (unsigned long)-1;
        }
    }

1667 1668
    qemu_log("Reserved 0x%lx bytes of guest address space\n", host_size);

1669 1670 1671
    return real_start;
}

1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
static void probe_guest_base(const char *image_name,
                             abi_ulong loaddr, abi_ulong hiaddr)
{
    /* Probe for a suitable guest base address, if the user has not set
     * it explicitly, and set guest_base appropriately.
     * In case of error we will print a suitable message and exit.
     */
#if defined(CONFIG_USE_GUEST_BASE)
    const char *errmsg;
    if (!have_guest_base && !reserved_va) {
        unsigned long host_start, real_start, host_size;

        /* Round addresses to page boundaries.  */
        loaddr &= qemu_host_page_mask;
        hiaddr = HOST_PAGE_ALIGN(hiaddr);

        if (loaddr < mmap_min_addr) {
            host_start = HOST_PAGE_ALIGN(mmap_min_addr);
        } else {
            host_start = loaddr;
            if (host_start != loaddr) {
                errmsg = "Address overflow loading ELF binary";
                goto exit_errmsg;
            }
        }
        host_size = hiaddr - loaddr;
1698 1699 1700 1701 1702 1703 1704 1705

        /* Setup the initial guest memory space with ranges gleaned from
         * the ELF image that is being loaded.
         */
        real_start = init_guest_space(host_start, host_size, loaddr, false);
        if (real_start == (unsigned long)-1) {
            errmsg = "Unable to find space for application";
            goto exit_errmsg;
1706
        }
1707 1708
        guest_base = real_start - loaddr;

1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721
        qemu_log("Relocating guest address space from 0x"
                 TARGET_ABI_FMT_lx " to 0x%lx\n",
                 loaddr, real_start);
    }
    return;

exit_errmsg:
    fprintf(stderr, "%s: %s\n", image_name, errmsg);
    exit(-1);
#endif
}


1722
/* Load an ELF image into the address space.
1723

1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734
   IMAGE_NAME is the filename of the image, to use in error messages.
   IMAGE_FD is the open file descriptor for the image.

   BPRM_BUF is a copy of the beginning of the file; this of course
   contains the elf file header at offset 0.  It is assumed that this
   buffer is sufficiently aligned to present no problems to the host
   in accessing data at aligned offsets within the buffer.

   On return: INFO values will be filled in, as necessary or available.  */

static void load_elf_image(const char *image_name, int image_fd,
1735
                           struct image_info *info, char **pinterp_name,
1736
                           char bprm_buf[BPRM_BUF_SIZE])
1737
{
1738 1739 1740 1741 1742
    struct elfhdr *ehdr = (struct elfhdr *)bprm_buf;
    struct elf_phdr *phdr;
    abi_ulong load_addr, load_bias, loaddr, hiaddr, error;
    int i, retval;
    const char *errmsg;
1743

1744 1745 1746 1747 1748 1749 1750 1751
    /* First of all, some simple consistency checks */
    errmsg = "Invalid ELF image for this architecture";
    if (!elf_check_ident(ehdr)) {
        goto exit_errmsg;
    }
    bswap_ehdr(ehdr);
    if (!elf_check_ehdr(ehdr)) {
        goto exit_errmsg;
1752
    }
1753

1754 1755 1756
    i = ehdr->e_phnum * sizeof(struct elf_phdr);
    if (ehdr->e_phoff + i <= BPRM_BUF_SIZE) {
        phdr = (struct elf_phdr *)(bprm_buf + ehdr->e_phoff);
1757
    } else {
1758 1759
        phdr = (struct elf_phdr *) alloca(i);
        retval = pread(image_fd, phdr, i, ehdr->e_phoff);
1760
        if (retval != i) {
1761
            goto exit_read;
1762
        }
1763
    }
1764
    bswap_phdr(phdr, ehdr->e_phnum);
B
bellard 已提交
1765

1766 1767 1768 1769 1770
#ifdef CONFIG_USE_FDPIC
    info->nsegs = 0;
    info->pt_dynamic_addr = 0;
#endif

1771 1772 1773
    /* Find the maximum size of the image and allocate an appropriate
       amount of memory to handle that.  */
    loaddr = -1, hiaddr = 0;
1774 1775 1776
    for (i = 0; i < ehdr->e_phnum; ++i) {
        if (phdr[i].p_type == PT_LOAD) {
            abi_ulong a = phdr[i].p_vaddr;
1777 1778 1779
            if (a < loaddr) {
                loaddr = a;
            }
1780
            a += phdr[i].p_memsz;
1781 1782 1783
            if (a > hiaddr) {
                hiaddr = a;
            }
1784 1785 1786
#ifdef CONFIG_USE_FDPIC
            ++info->nsegs;
#endif
1787 1788 1789 1790
        }
    }

    load_addr = loaddr;
1791
    if (ehdr->e_type == ET_DYN) {
1792 1793 1794 1795 1796 1797 1798 1799 1800
        /* The image indicates that it can be loaded anywhere.  Find a
           location that can hold the memory space required.  If the
           image is pre-linked, LOADDR will be non-zero.  Since we do
           not supply MAP_FIXED here we'll use that address if and
           only if it remains available.  */
        load_addr = target_mmap(loaddr, hiaddr - loaddr, PROT_NONE,
                                MAP_PRIVATE | MAP_ANON | MAP_NORESERVE,
                                -1, 0);
        if (load_addr == -1) {
1801
            goto exit_perror;
1802
        }
1803 1804 1805 1806
    } else if (pinterp_name != NULL) {
        /* This is the main executable.  Make sure that the low
           address does not conflict with MMAP_MIN_ADDR or the
           QEMU application itself.  */
1807
        probe_guest_base(image_name, loaddr, hiaddr);
1808
    }
1809
    load_bias = load_addr - loaddr;
1810

1811 1812 1813
#ifdef CONFIG_USE_FDPIC
    {
        struct elf32_fdpic_loadseg *loadsegs = info->loadsegs =
1814
            g_malloc(sizeof(*loadsegs) * info->nsegs);
1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831

        for (i = 0; i < ehdr->e_phnum; ++i) {
            switch (phdr[i].p_type) {
            case PT_DYNAMIC:
                info->pt_dynamic_addr = phdr[i].p_vaddr + load_bias;
                break;
            case PT_LOAD:
                loadsegs->addr = phdr[i].p_vaddr + load_bias;
                loadsegs->p_vaddr = phdr[i].p_vaddr;
                loadsegs->p_memsz = phdr[i].p_memsz;
                ++loadsegs;
                break;
            }
        }
    }
#endif

1832 1833 1834 1835 1836 1837 1838 1839
    info->load_bias = load_bias;
    info->load_addr = load_addr;
    info->entry = ehdr->e_entry + load_bias;
    info->start_code = -1;
    info->end_code = 0;
    info->start_data = -1;
    info->end_data = 0;
    info->brk = 0;
P
Paul Brook 已提交
1840
    info->elf_flags = ehdr->e_flags;
1841 1842 1843

    for (i = 0; i < ehdr->e_phnum; i++) {
        struct elf_phdr *eppnt = phdr + i;
1844
        if (eppnt->p_type == PT_LOAD) {
1845
            abi_ulong vaddr, vaddr_po, vaddr_ps, vaddr_ef, vaddr_em;
1846 1847 1848 1849 1850 1851
            int elf_prot = 0;

            if (eppnt->p_flags & PF_R) elf_prot =  PROT_READ;
            if (eppnt->p_flags & PF_W) elf_prot |= PROT_WRITE;
            if (eppnt->p_flags & PF_X) elf_prot |= PROT_EXEC;

1852 1853 1854 1855 1856 1857
            vaddr = load_bias + eppnt->p_vaddr;
            vaddr_po = TARGET_ELF_PAGEOFFSET(vaddr);
            vaddr_ps = TARGET_ELF_PAGESTART(vaddr);

            error = target_mmap(vaddr_ps, eppnt->p_filesz + vaddr_po,
                                elf_prot, MAP_PRIVATE | MAP_FIXED,
1858
                                image_fd, eppnt->p_offset - vaddr_po);
B
bellard 已提交
1859
            if (error == -1) {
1860
                goto exit_perror;
B
bellard 已提交
1861 1862
            }

1863 1864
            vaddr_ef = vaddr + eppnt->p_filesz;
            vaddr_em = vaddr + eppnt->p_memsz;
1865

1866
            /* If the load segment requests extra zeros (e.g. bss), map it.  */
1867 1868
            if (vaddr_ef < vaddr_em) {
                zero_bss(vaddr_ef, vaddr_em, elf_prot);
1869
            }
1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890

            /* Find the full program boundaries.  */
            if (elf_prot & PROT_EXEC) {
                if (vaddr < info->start_code) {
                    info->start_code = vaddr;
                }
                if (vaddr_ef > info->end_code) {
                    info->end_code = vaddr_ef;
                }
            }
            if (elf_prot & PROT_WRITE) {
                if (vaddr < info->start_data) {
                    info->start_data = vaddr;
                }
                if (vaddr_ef > info->end_data) {
                    info->end_data = vaddr_ef;
                }
                if (vaddr_em > info->brk) {
                    info->brk = vaddr_em;
                }
            }
1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917
        } else if (eppnt->p_type == PT_INTERP && pinterp_name) {
            char *interp_name;

            if (*pinterp_name) {
                errmsg = "Multiple PT_INTERP entries";
                goto exit_errmsg;
            }
            interp_name = malloc(eppnt->p_filesz);
            if (!interp_name) {
                goto exit_perror;
            }

            if (eppnt->p_offset + eppnt->p_filesz <= BPRM_BUF_SIZE) {
                memcpy(interp_name, bprm_buf + eppnt->p_offset,
                       eppnt->p_filesz);
            } else {
                retval = pread(image_fd, interp_name, eppnt->p_filesz,
                               eppnt->p_offset);
                if (retval != eppnt->p_filesz) {
                    goto exit_perror;
                }
            }
            if (interp_name[eppnt->p_filesz - 1] != 0) {
                errmsg = "Invalid PT_INTERP entry";
                goto exit_errmsg;
            }
            *pinterp_name = interp_name;
1918
        }
1919
    }
1920

1921 1922 1923 1924 1925 1926
    if (info->end_data == 0) {
        info->start_data = info->end_code;
        info->end_data = info->end_code;
        info->brk = info->end_code;
    }

1927
    if (qemu_log_enabled()) {
1928
        load_symbols(ehdr, image_fd, load_bias);
1929
    }
1930

1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954
    close(image_fd);
    return;

 exit_read:
    if (retval >= 0) {
        errmsg = "Incomplete read of file header";
        goto exit_errmsg;
    }
 exit_perror:
    errmsg = strerror(errno);
 exit_errmsg:
    fprintf(stderr, "%s: %s\n", image_name, errmsg);
    exit(-1);
}

static void load_elf_interp(const char *filename, struct image_info *info,
                            char bprm_buf[BPRM_BUF_SIZE])
{
    int fd, retval;

    fd = open(path(filename), O_RDONLY);
    if (fd < 0) {
        goto exit_perror;
    }
1955

1956 1957 1958 1959 1960 1961 1962 1963
    retval = read(fd, bprm_buf, BPRM_BUF_SIZE);
    if (retval < 0) {
        goto exit_perror;
    }
    if (retval < BPRM_BUF_SIZE) {
        memset(bprm_buf + retval, 0, BPRM_BUF_SIZE - retval);
    }

1964
    load_elf_image(filename, fd, info, NULL, bprm_buf);
1965 1966 1967 1968 1969
    return;

 exit_perror:
    fprintf(stderr, "%s: %s\n", filename, strerror(errno));
    exit(-1);
1970 1971
}

1972 1973
static int symfind(const void *s0, const void *s1)
{
1974
    target_ulong addr = *(target_ulong *)s0;
1975 1976
    struct elf_sym *sym = (struct elf_sym *)s1;
    int result = 0;
1977
    if (addr < sym->st_value) {
1978
        result = -1;
1979
    } else if (addr >= sym->st_value + sym->st_size) {
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
        result = 1;
    }
    return result;
}

static const char *lookup_symbolxx(struct syminfo *s, target_ulong orig_addr)
{
#if ELF_CLASS == ELFCLASS32
    struct elf_sym *syms = s->disas_symtab.elf32;
#else
    struct elf_sym *syms = s->disas_symtab.elf64;
#endif

    // binary search
    struct elf_sym *sym;

1996
    sym = bsearch(&orig_addr, syms, s->disas_num_syms, sizeof(*syms), symfind);
B
Blue Swirl 已提交
1997
    if (sym != NULL) {
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
        return s->disas_strtab + sym->st_name;
    }

    return "";
}

/* FIXME: This should use elf_ops.h  */
static int symcmp(const void *s0, const void *s1)
{
    struct elf_sym *sym0 = (struct elf_sym *)s0;
    struct elf_sym *sym1 = (struct elf_sym *)s1;
    return (sym0->st_value < sym1->st_value)
        ? -1
        : ((sym0->st_value > sym1->st_value) ? 1 : 0);
}

B
bellard 已提交
2014
/* Best attempt to load symbols from this ELF object. */
2015
static void load_symbols(struct elfhdr *hdr, int fd, abi_ulong load_bias)
B
bellard 已提交
2016
{
2017 2018
    int i, shnum, nsyms, sym_idx = 0, str_idx = 0;
    struct elf_shdr *shdr;
2019 2020 2021
    char *strings = NULL;
    struct syminfo *s = NULL;
    struct elf_sym *new_syms, *syms = NULL;
B
bellard 已提交
2022

2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034
    shnum = hdr->e_shnum;
    i = shnum * sizeof(struct elf_shdr);
    shdr = (struct elf_shdr *)alloca(i);
    if (pread(fd, shdr, i, hdr->e_shoff) != i) {
        return;
    }

    bswap_shdr(shdr, shnum);
    for (i = 0; i < shnum; ++i) {
        if (shdr[i].sh_type == SHT_SYMTAB) {
            sym_idx = i;
            str_idx = shdr[i].sh_link;
2035 2036
            goto found;
        }
B
bellard 已提交
2037
    }
2038 2039 2040

    /* There will be no symbol table if the file was stripped.  */
    return;
B
bellard 已提交
2041 2042

 found:
2043
    /* Now know where the strtab and symtab are.  Snarf them.  */
B
bellard 已提交
2044
    s = malloc(sizeof(*s));
2045
    if (!s) {
2046
        goto give_up;
2047
    }
2048

2049 2050 2051
    i = shdr[str_idx].sh_size;
    s->disas_strtab = strings = malloc(i);
    if (!strings || pread(fd, strings, i, shdr[str_idx].sh_offset) != i) {
2052
        goto give_up;
2053
    }
2054

2055 2056 2057
    i = shdr[sym_idx].sh_size;
    syms = malloc(i);
    if (!syms || pread(fd, syms, i, shdr[sym_idx].sh_offset) != i) {
2058
        goto give_up;
2059
    }
2060

2061 2062
    nsyms = i / sizeof(struct elf_sym);
    for (i = 0; i < nsyms; ) {
2063
        bswap_sym(syms + i);
2064 2065 2066 2067 2068
        /* Throw away entries which we do not need.  */
        if (syms[i].st_shndx == SHN_UNDEF
            || syms[i].st_shndx >= SHN_LORESERVE
            || ELF_ST_TYPE(syms[i].st_info) != STT_FUNC) {
            if (i < --nsyms) {
2069 2070
                syms[i] = syms[nsyms];
            }
2071
        } else {
2072
#if defined(TARGET_ARM) || defined (TARGET_MIPS)
2073 2074
            /* The bottom address bit marks a Thumb or MIPS16 symbol.  */
            syms[i].st_value &= ~(target_ulong)1;
B
blueswir1 已提交
2075
#endif
2076 2077 2078
            syms[i].st_value += load_bias;
            i++;
        }
B
blueswir1 已提交
2079
    }
2080

2081 2082 2083 2084 2085
    /* No "useful" symbol.  */
    if (nsyms == 0) {
        goto give_up;
    }

2086 2087 2088 2089
    /* Attempt to free the storage associated with the local symbols
       that we threw away.  Whether or not this has any effect on the
       memory allocation depends on the malloc implementation and how
       many symbols we managed to discard.  */
2090 2091
    new_syms = realloc(syms, nsyms * sizeof(*syms));
    if (new_syms == NULL) {
2092
        goto give_up;
2093
    }
2094
    syms = new_syms;
2095

2096
    qsort(syms, nsyms, sizeof(*syms), symcmp);
B
bellard 已提交
2097

2098 2099 2100 2101 2102 2103
    s->disas_num_syms = nsyms;
#if ELF_CLASS == ELFCLASS32
    s->disas_symtab.elf32 = syms;
#else
    s->disas_symtab.elf64 = syms;
#endif
2104
    s->lookup_symbol = lookup_symbolxx;
B
bellard 已提交
2105 2106
    s->next = syminfos;
    syminfos = s;
2107 2108 2109 2110 2111 2112 2113

    return;

give_up:
    free(s);
    free(strings);
    free(syms);
B
bellard 已提交
2114
}
2115

2116
int load_elf_binary(struct linux_binprm *bprm, struct image_info *info)
2117
{
2118
    struct image_info interp_info;
2119
    struct elfhdr elf_ex;
2120
    char *elf_interpreter = NULL;
2121

2122 2123 2124 2125 2126 2127
    info->start_mmap = (abi_ulong)ELF_START_MMAP;
    info->mmap = 0;
    info->rss = 0;

    load_elf_image(bprm->filename, bprm->fd, info,
                   &elf_interpreter, bprm->buf);
2128

2129 2130 2131 2132
    /* ??? We need a copy of the elf header for passing to create_elf_tables.
       If we do nothing, we'll have overwritten this when we re-use bprm->buf
       when we load the interpreter.  */
    elf_ex = *(struct elfhdr *)bprm->buf;
2133

P
pbrook 已提交
2134 2135 2136 2137
    bprm->p = copy_elf_strings(1, &bprm->filename, bprm->page, bprm->p);
    bprm->p = copy_elf_strings(bprm->envc,bprm->envp,bprm->page,bprm->p);
    bprm->p = copy_elf_strings(bprm->argc,bprm->argv,bprm->page,bprm->p);
    if (!bprm->p) {
2138 2139
        fprintf(stderr, "%s: %s\n", bprm->filename, strerror(E2BIG));
        exit(-1);
P
Paul Brook 已提交
2140 2141
    }

2142 2143 2144 2145
    /* Do this so that we can load the interpreter, if need be.  We will
       change some of these later */
    bprm->p = setup_arg_pages(bprm->p, bprm, info);

2146 2147
    if (elf_interpreter) {
        load_elf_interp(elf_interpreter, &interp_info, bprm->buf);
2148

2149 2150 2151 2152 2153 2154
        /* If the program interpreter is one of these two, then assume
           an iBCS2 image.  Otherwise assume a native linux image.  */

        if (strcmp(elf_interpreter, "/usr/lib/libc.so.1") == 0
            || strcmp(elf_interpreter, "/usr/lib/ld.so.1") == 0) {
            info->personality = PER_SVR4;
2155

2156 2157 2158 2159 2160 2161 2162
            /* Why this, you ask???  Well SVr4 maps page 0 as read-only,
               and some applications "depend" upon this behavior.  Since
               we do not have the power to recompile these, we emulate
               the SVr4 behavior.  Sigh.  */
            target_mmap(0, qemu_host_page_size, PROT_READ | PROT_EXEC,
                        MAP_FIXED | MAP_PRIVATE, -1, 0);
        }
2163 2164
    }

2165 2166 2167 2168 2169
    bprm->p = create_elf_tables(bprm->p, bprm->argc, bprm->envc, &elf_ex,
                                info, (elf_interpreter ? &interp_info : NULL));
    info->start_stack = bprm->p;

    /* If we have an interpreter, set that as the program's entry point.
2170
       Copy the load_bias as well, to help PPC64 interpret the entry
2171 2172 2173
       point as a function descriptor.  Do this after creating elf tables
       so that we copy the original program entry point into the AUXV.  */
    if (elf_interpreter) {
2174
        info->load_bias = interp_info.load_bias;
2175
        info->entry = interp_info.entry;
2176
        free(elf_interpreter);
2177
    }
2178

2179 2180 2181 2182
#ifdef USE_ELF_CORE_DUMP
    bprm->core_dump = &elf_core_dump;
#endif

2183 2184 2185
    return 0;
}

2186 2187 2188
#ifdef USE_ELF_CORE_DUMP
/*
 * Definitions to generate Intel SVR4-like core files.
L
Laurent Desnogues 已提交
2189
 * These mostly have the same names as the SVR4 types with "target_elf_"
2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200
 * tacked on the front to prevent clashes with linux definitions,
 * and the typedef forms have been avoided.  This is mostly like
 * the SVR4 structure, but more Linuxy, with things that Linux does
 * not support and which gdb doesn't really use excluded.
 *
 * Fields we don't dump (their contents is zero) in linux-user qemu
 * are marked with XXX.
 *
 * Core dump code is copied from linux kernel (fs/binfmt_elf.c).
 *
 * Porting ELF coredump for target is (quite) simple process.  First you
2201
 * define USE_ELF_CORE_DUMP in target ELF code (where init_thread() for
2202 2203 2204 2205 2206 2207 2208
 * the target resides):
 *
 * #define USE_ELF_CORE_DUMP
 *
 * Next you define type of register set used for dumping.  ELF specification
 * says that it needs to be array of elf_greg_t that has size of ELF_NREG.
 *
A
Anthony Liguori 已提交
2209
 * typedef <target_regtype> target_elf_greg_t;
2210
 * #define ELF_NREG <number of registers>
A
Anthony Liguori 已提交
2211
 * typedef taret_elf_greg_t target_elf_gregset_t[ELF_NREG];
2212 2213 2214 2215
 *
 * Last step is to implement target specific function that copies registers
 * from given cpu into just specified register set.  Prototype is:
 *
A
Anthony Liguori 已提交
2216
 * static void elf_core_copy_regs(taret_elf_gregset_t *regs,
2217
 *                                const CPUArchState *env);
2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232
 *
 * Parameters:
 *     regs - copy register values into here (allocated and zeroed by caller)
 *     env - copy registers from here
 *
 * Example for ARM target is provided in this file.
 */

/* An ELF note in memory */
struct memelfnote {
    const char *name;
    size_t     namesz;
    size_t     namesz_rounded;
    int        type;
    size_t     datasz;
2233
    size_t     datasz_rounded;
2234 2235 2236 2237
    void       *data;
    size_t     notesz;
};

L
Laurent Desnogues 已提交
2238
struct target_elf_siginfo {
2239 2240 2241
    abi_int    si_signo; /* signal number */
    abi_int    si_code;  /* extra code */
    abi_int    si_errno; /* errno */
2242 2243
};

L
Laurent Desnogues 已提交
2244 2245
struct target_elf_prstatus {
    struct target_elf_siginfo pr_info;      /* Info associated with signal */
2246
    abi_short          pr_cursig;    /* Current signal */
2247 2248
    abi_ulong          pr_sigpend;   /* XXX */
    abi_ulong          pr_sighold;   /* XXX */
A
Anthony Liguori 已提交
2249 2250 2251 2252
    target_pid_t       pr_pid;
    target_pid_t       pr_ppid;
    target_pid_t       pr_pgrp;
    target_pid_t       pr_sid;
2253 2254 2255 2256
    struct target_timeval pr_utime;  /* XXX User time */
    struct target_timeval pr_stime;  /* XXX System time */
    struct target_timeval pr_cutime; /* XXX Cumulative user time */
    struct target_timeval pr_cstime; /* XXX Cumulative system time */
A
Anthony Liguori 已提交
2257
    target_elf_gregset_t      pr_reg;       /* GP registers */
2258
    abi_int            pr_fpvalid;   /* XXX */
2259 2260 2261 2262
};

#define ELF_PRARGSZ     (80) /* Number of chars for args */

L
Laurent Desnogues 已提交
2263
struct target_elf_prpsinfo {
2264 2265 2266 2267
    char         pr_state;       /* numeric process state */
    char         pr_sname;       /* char for pr_state */
    char         pr_zomb;        /* zombie */
    char         pr_nice;        /* nice val */
2268
    abi_ulong    pr_flag;        /* flags */
A
Anthony Liguori 已提交
2269 2270 2271
    target_uid_t pr_uid;
    target_gid_t pr_gid;
    target_pid_t pr_pid, pr_ppid, pr_pgrp, pr_sid;
2272 2273 2274 2275 2276 2277 2278
    /* Lots missing */
    char    pr_fname[16];           /* filename of executable */
    char    pr_psargs[ELF_PRARGSZ]; /* initial part of arg list */
};

/* Here is the structure in which status of each thread is captured. */
struct elf_thread_status {
B
Blue Swirl 已提交
2279
    QTAILQ_ENTRY(elf_thread_status)  ets_link;
L
Laurent Desnogues 已提交
2280
    struct target_elf_prstatus prstatus;   /* NT_PRSTATUS */
2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291
#if 0
    elf_fpregset_t fpu;             /* NT_PRFPREG */
    struct task_struct *thread;
    elf_fpxregset_t xfpu;           /* ELF_CORE_XFPREG_TYPE */
#endif
    struct memelfnote notes[1];
    int num_notes;
};

struct elf_note_info {
    struct memelfnote   *notes;
L
Laurent Desnogues 已提交
2292 2293
    struct target_elf_prstatus *prstatus;  /* NT_PRSTATUS */
    struct target_elf_prpsinfo *psinfo;    /* NT_PRPSINFO */
2294

B
Blue Swirl 已提交
2295
    QTAILQ_HEAD(thread_list_head, elf_thread_status) thread_list;
2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312
#if 0
    /*
     * Current version of ELF coredump doesn't support
     * dumping fp regs etc.
     */
    elf_fpregset_t *fpu;
    elf_fpxregset_t *xfpu;
    int thread_status_size;
#endif
    int notes_size;
    int numnote;
};

struct vm_area_struct {
    abi_ulong   vma_start;  /* start vaddr of memory region */
    abi_ulong   vma_end;    /* end vaddr of memory region */
    abi_ulong   vma_flags;  /* protection etc. flags for the region */
B
Blue Swirl 已提交
2313
    QTAILQ_ENTRY(vm_area_struct) vma_link;
2314 2315 2316
};

struct mm_struct {
B
Blue Swirl 已提交
2317
    QTAILQ_HEAD(, vm_area_struct) mm_mmap;
2318 2319 2320 2321 2322 2323
    int mm_count;           /* number of mappings */
};

static struct mm_struct *vma_init(void);
static void vma_delete(struct mm_struct *);
static int vma_add_mapping(struct mm_struct *, abi_ulong,
2324
                           abi_ulong, abi_ulong);
2325 2326 2327 2328
static int vma_get_mapping_count(const struct mm_struct *);
static struct vm_area_struct *vma_first(const struct mm_struct *);
static struct vm_area_struct *vma_next(struct vm_area_struct *);
static abi_ulong vma_dump_size(const struct vm_area_struct *);
P
Paul Brook 已提交
2329
static int vma_walker(void *priv, abi_ulong start, abi_ulong end,
2330
                      unsigned long flags);
2331 2332 2333

static void fill_elf_header(struct elfhdr *, int, uint16_t, uint32_t);
static void fill_note(struct memelfnote *, const char *, int,
2334
                      unsigned int, void *);
L
Laurent Desnogues 已提交
2335 2336
static void fill_prstatus(struct target_elf_prstatus *, const TaskState *, int);
static int fill_psinfo(struct target_elf_prpsinfo *, const TaskState *);
2337 2338 2339 2340
static void fill_auxv_note(struct memelfnote *, const TaskState *);
static void fill_elf_note_phdr(struct elf_phdr *, int, off_t);
static size_t note_size(const struct memelfnote *);
static void free_note_info(struct elf_note_info *);
2341 2342
static int fill_note_info(struct elf_note_info *, long, const CPUArchState *);
static void fill_thread_info(struct elf_note_info *, const CPUArchState *);
2343 2344 2345 2346 2347 2348 2349
static int core_dump_filename(const TaskState *, char *, size_t);

static int dump_write(int, const void *, size_t);
static int write_note(struct memelfnote *, int);
static int write_note_info(struct elf_note_info *, int);

#ifdef BSWAP_NEEDED
L
Laurent Desnogues 已提交
2350
static void bswap_prstatus(struct target_elf_prstatus *prstatus)
2351
{
2352 2353 2354
    prstatus->pr_info.si_signo = tswap32(prstatus->pr_info.si_signo);
    prstatus->pr_info.si_code = tswap32(prstatus->pr_info.si_code);
    prstatus->pr_info.si_errno = tswap32(prstatus->pr_info.si_errno);
2355
    prstatus->pr_cursig = tswap16(prstatus->pr_cursig);
2356 2357
    prstatus->pr_sigpend = tswapal(prstatus->pr_sigpend);
    prstatus->pr_sighold = tswapal(prstatus->pr_sighold);
2358 2359 2360 2361 2362 2363 2364 2365 2366
    prstatus->pr_pid = tswap32(prstatus->pr_pid);
    prstatus->pr_ppid = tswap32(prstatus->pr_ppid);
    prstatus->pr_pgrp = tswap32(prstatus->pr_pgrp);
    prstatus->pr_sid = tswap32(prstatus->pr_sid);
    /* cpu times are not filled, so we skip them */
    /* regs should be in correct format already */
    prstatus->pr_fpvalid = tswap32(prstatus->pr_fpvalid);
}

L
Laurent Desnogues 已提交
2367
static void bswap_psinfo(struct target_elf_prpsinfo *psinfo)
2368
{
2369
    psinfo->pr_flag = tswapal(psinfo->pr_flag);
2370 2371 2372 2373 2374 2375 2376
    psinfo->pr_uid = tswap16(psinfo->pr_uid);
    psinfo->pr_gid = tswap16(psinfo->pr_gid);
    psinfo->pr_pid = tswap32(psinfo->pr_pid);
    psinfo->pr_ppid = tswap32(psinfo->pr_ppid);
    psinfo->pr_pgrp = tswap32(psinfo->pr_pgrp);
    psinfo->pr_sid = tswap32(psinfo->pr_sid);
}
2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387

static void bswap_note(struct elf_note *en)
{
    bswap32s(&en->n_namesz);
    bswap32s(&en->n_descsz);
    bswap32s(&en->n_type);
}
#else
static inline void bswap_prstatus(struct target_elf_prstatus *p) { }
static inline void bswap_psinfo(struct target_elf_prpsinfo *p) {}
static inline void bswap_note(struct elf_note *en) { }
2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
#endif /* BSWAP_NEEDED */

/*
 * Minimal support for linux memory regions.  These are needed
 * when we are finding out what memory exactly belongs to
 * emulated process.  No locks needed here, as long as
 * thread that received the signal is stopped.
 */

static struct mm_struct *vma_init(void)
{
    struct mm_struct *mm;

2401
    if ((mm = g_malloc(sizeof (*mm))) == NULL)
2402 2403 2404
        return (NULL);

    mm->mm_count = 0;
B
Blue Swirl 已提交
2405
    QTAILQ_INIT(&mm->mm_mmap);
2406 2407 2408 2409 2410 2411 2412 2413 2414

    return (mm);
}

static void vma_delete(struct mm_struct *mm)
{
    struct vm_area_struct *vma;

    while ((vma = vma_first(mm)) != NULL) {
B
Blue Swirl 已提交
2415
        QTAILQ_REMOVE(&mm->mm_mmap, vma, vma_link);
2416
        g_free(vma);
2417
    }
2418
    g_free(mm);
2419 2420 2421
}

static int vma_add_mapping(struct mm_struct *mm, abi_ulong start,
2422
                           abi_ulong end, abi_ulong flags)
2423 2424 2425
{
    struct vm_area_struct *vma;

2426
    if ((vma = g_malloc0(sizeof (*vma))) == NULL)
2427 2428 2429 2430 2431 2432
        return (-1);

    vma->vma_start = start;
    vma->vma_end = end;
    vma->vma_flags = flags;

B
Blue Swirl 已提交
2433
    QTAILQ_INSERT_TAIL(&mm->mm_mmap, vma, vma_link);
2434 2435 2436 2437 2438 2439 2440
    mm->mm_count++;

    return (0);
}

static struct vm_area_struct *vma_first(const struct mm_struct *mm)
{
B
Blue Swirl 已提交
2441
    return (QTAILQ_FIRST(&mm->mm_mmap));
2442 2443 2444 2445
}

static struct vm_area_struct *vma_next(struct vm_area_struct *vma)
{
B
Blue Swirl 已提交
2446
    return (QTAILQ_NEXT(vma, vma_link));
2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489
}

static int vma_get_mapping_count(const struct mm_struct *mm)
{
    return (mm->mm_count);
}

/*
 * Calculate file (dump) size of given memory region.
 */
static abi_ulong vma_dump_size(const struct vm_area_struct *vma)
{
    /* if we cannot even read the first page, skip it */
    if (!access_ok(VERIFY_READ, vma->vma_start, TARGET_PAGE_SIZE))
        return (0);

    /*
     * Usually we don't dump executable pages as they contain
     * non-writable code that debugger can read directly from
     * target library etc.  However, thread stacks are marked
     * also executable so we read in first page of given region
     * and check whether it contains elf header.  If there is
     * no elf header, we dump it.
     */
    if (vma->vma_flags & PROT_EXEC) {
        char page[TARGET_PAGE_SIZE];

        copy_from_user(page, vma->vma_start, sizeof (page));
        if ((page[EI_MAG0] == ELFMAG0) &&
            (page[EI_MAG1] == ELFMAG1) &&
            (page[EI_MAG2] == ELFMAG2) &&
            (page[EI_MAG3] == ELFMAG3)) {
            /*
             * Mappings are possibly from ELF binary.  Don't dump
             * them.
             */
            return (0);
        }
    }

    return (vma->vma_end - vma->vma_start);
}

P
Paul Brook 已提交
2490
static int vma_walker(void *priv, abi_ulong start, abi_ulong end,
2491
                      unsigned long flags)
2492 2493 2494 2495 2496 2497 2498 2499
{
    struct mm_struct *mm = (struct mm_struct *)priv;

    vma_add_mapping(mm, start, end, flags);
    return (0);
}

static void fill_note(struct memelfnote *note, const char *name, int type,
2500
                      unsigned int sz, void *data)
2501 2502 2503 2504 2505 2506 2507 2508
{
    unsigned int namesz;

    namesz = strlen(name) + 1;
    note->name = name;
    note->namesz = namesz;
    note->namesz_rounded = roundup(namesz, sizeof (int32_t));
    note->type = type;
2509 2510 2511
    note->datasz = sz;
    note->datasz_rounded = roundup(sz, sizeof (int32_t));

2512 2513 2514 2515 2516 2517 2518
    note->data = data;

    /*
     * We calculate rounded up note size here as specified by
     * ELF document.
     */
    note->notesz = sizeof (struct elf_note) +
2519
        note->namesz_rounded + note->datasz_rounded;
2520 2521 2522
}

static void fill_elf_header(struct elfhdr *elf, int segs, uint16_t machine,
2523
                            uint32_t flags)
2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555
{
    (void) memset(elf, 0, sizeof(*elf));

    (void) memcpy(elf->e_ident, ELFMAG, SELFMAG);
    elf->e_ident[EI_CLASS] = ELF_CLASS;
    elf->e_ident[EI_DATA] = ELF_DATA;
    elf->e_ident[EI_VERSION] = EV_CURRENT;
    elf->e_ident[EI_OSABI] = ELF_OSABI;

    elf->e_type = ET_CORE;
    elf->e_machine = machine;
    elf->e_version = EV_CURRENT;
    elf->e_phoff = sizeof(struct elfhdr);
    elf->e_flags = flags;
    elf->e_ehsize = sizeof(struct elfhdr);
    elf->e_phentsize = sizeof(struct elf_phdr);
    elf->e_phnum = segs;

    bswap_ehdr(elf);
}

static void fill_elf_note_phdr(struct elf_phdr *phdr, int sz, off_t offset)
{
    phdr->p_type = PT_NOTE;
    phdr->p_offset = offset;
    phdr->p_vaddr = 0;
    phdr->p_paddr = 0;
    phdr->p_filesz = sz;
    phdr->p_memsz = 0;
    phdr->p_flags = 0;
    phdr->p_align = 0;

2556
    bswap_phdr(phdr, 1);
2557 2558 2559 2560 2561 2562 2563
}

static size_t note_size(const struct memelfnote *note)
{
    return (note->notesz);
}

L
Laurent Desnogues 已提交
2564
static void fill_prstatus(struct target_elf_prstatus *prstatus,
2565
                          const TaskState *ts, int signr)
2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576
{
    (void) memset(prstatus, 0, sizeof (*prstatus));
    prstatus->pr_info.si_signo = prstatus->pr_cursig = signr;
    prstatus->pr_pid = ts->ts_tid;
    prstatus->pr_ppid = getppid();
    prstatus->pr_pgrp = getpgrp();
    prstatus->pr_sid = getsid(0);

    bswap_prstatus(prstatus);
}

L
Laurent Desnogues 已提交
2577
static int fill_psinfo(struct target_elf_prpsinfo *psinfo, const TaskState *ts)
2578
{
2579
    char *base_filename;
2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600
    unsigned int i, len;

    (void) memset(psinfo, 0, sizeof (*psinfo));

    len = ts->info->arg_end - ts->info->arg_start;
    if (len >= ELF_PRARGSZ)
        len = ELF_PRARGSZ - 1;
    if (copy_from_user(&psinfo->pr_psargs, ts->info->arg_start, len))
        return -EFAULT;
    for (i = 0; i < len; i++)
        if (psinfo->pr_psargs[i] == 0)
            psinfo->pr_psargs[i] = ' ';
    psinfo->pr_psargs[len] = 0;

    psinfo->pr_pid = getpid();
    psinfo->pr_ppid = getppid();
    psinfo->pr_pgrp = getpgrp();
    psinfo->pr_sid = getsid(0);
    psinfo->pr_uid = getuid();
    psinfo->pr_gid = getgid();

2601 2602 2603 2604 2605
    base_filename = g_path_get_basename(ts->bprm->filename);
    /*
     * Using strncpy here is fine: at max-length,
     * this field is not NUL-terminated.
     */
2606
    (void) strncpy(psinfo->pr_fname, base_filename,
2607
                   sizeof(psinfo->pr_fname));
2608

2609
    g_free(base_filename);
2610 2611 2612 2613 2614 2615 2616 2617 2618
    bswap_psinfo(psinfo);
    return (0);
}

static void fill_auxv_note(struct memelfnote *note, const TaskState *ts)
{
    elf_addr_t auxv = (elf_addr_t)ts->info->saved_auxv;
    elf_addr_t orig_auxv = auxv;
    void *ptr;
A
Alexander Graf 已提交
2619
    int len = ts->info->auxv_len;
2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642

    /*
     * Auxiliary vector is stored in target process stack.  It contains
     * {type, value} pairs that we need to dump into note.  This is not
     * strictly necessary but we do it here for sake of completeness.
     */

    /* read in whole auxv vector and copy it to memelfnote */
    ptr = lock_user(VERIFY_READ, orig_auxv, len, 0);
    if (ptr != NULL) {
        fill_note(note, "CORE", NT_AUXV, len, ptr);
        unlock_user(ptr, auxv, len);
    }
}

/*
 * Constructs name of coredump file.  We have following convention
 * for the name:
 *     qemu_<basename-of-target-binary>_<date>-<time>_<pid>.core
 *
 * Returns 0 in case of success, -1 otherwise (errno is set).
 */
static int core_dump_filename(const TaskState *ts, char *buf,
2643
                              size_t bufsize)
2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654
{
    char timestamp[64];
    char *filename = NULL;
    char *base_filename = NULL;
    struct timeval tv;
    struct tm tm;

    assert(bufsize >= PATH_MAX);

    if (gettimeofday(&tv, NULL) < 0) {
        (void) fprintf(stderr, "unable to get current timestamp: %s",
2655
                       strerror(errno));
2656 2657 2658 2659 2660 2661
        return (-1);
    }

    filename = strdup(ts->bprm->filename);
    base_filename = strdup(basename(filename));
    (void) strftime(timestamp, sizeof (timestamp), "%Y%m%d-%H%M%S",
2662
                    localtime_r(&tv.tv_sec, &tm));
2663
    (void) snprintf(buf, bufsize, "qemu_%s_%s_%d.core",
2664
                    base_filename, timestamp, (int)getpid());
2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730
    free(base_filename);
    free(filename);

    return (0);
}

static int dump_write(int fd, const void *ptr, size_t size)
{
    const char *bufp = (const char *)ptr;
    ssize_t bytes_written, bytes_left;
    struct rlimit dumpsize;
    off_t pos;

    bytes_written = 0;
    getrlimit(RLIMIT_CORE, &dumpsize);
    if ((pos = lseek(fd, 0, SEEK_CUR))==-1) {
        if (errno == ESPIPE) { /* not a seekable stream */
            bytes_left = size;
        } else {
            return pos;
        }
    } else {
        if (dumpsize.rlim_cur <= pos) {
            return -1;
        } else if (dumpsize.rlim_cur == RLIM_INFINITY) {
            bytes_left = size;
        } else {
            size_t limit_left=dumpsize.rlim_cur - pos;
            bytes_left = limit_left >= size ? size : limit_left ;
        }
    }

    /*
     * In normal conditions, single write(2) should do but
     * in case of socket etc. this mechanism is more portable.
     */
    do {
        bytes_written = write(fd, bufp, bytes_left);
        if (bytes_written < 0) {
            if (errno == EINTR)
                continue;
            return (-1);
        } else if (bytes_written == 0) { /* eof */
            return (-1);
        }
        bufp += bytes_written;
        bytes_left -= bytes_written;
    } while (bytes_left > 0);

    return (0);
}

static int write_note(struct memelfnote *men, int fd)
{
    struct elf_note en;

    en.n_namesz = men->namesz;
    en.n_type = men->type;
    en.n_descsz = men->datasz;

    bswap_note(&en);

    if (dump_write(fd, &en, sizeof(en)) != 0)
        return (-1);
    if (dump_write(fd, men->name, men->namesz_rounded) != 0)
        return (-1);
2731
    if (dump_write(fd, men->data, men->datasz_rounded) != 0)
2732 2733 2734 2735 2736
        return (-1);

    return (0);
}

2737
static void fill_thread_info(struct elf_note_info *info, const CPUArchState *env)
2738
{
2739 2740
    CPUState *cpu = ENV_GET_CPU((CPUArchState *)env);
    TaskState *ts = (TaskState *)cpu->opaque;
2741 2742
    struct elf_thread_status *ets;

2743
    ets = g_malloc0(sizeof (*ets));
2744 2745 2746 2747
    ets->num_notes = 1; /* only prstatus is dumped */
    fill_prstatus(&ets->prstatus, ts, 0);
    elf_core_copy_regs(&ets->prstatus.pr_reg, env);
    fill_note(&ets->notes[0], "CORE", NT_PRSTATUS, sizeof (ets->prstatus),
2748
              &ets->prstatus);
2749

B
Blue Swirl 已提交
2750
    QTAILQ_INSERT_TAIL(&info->thread_list, ets, ets_link);
2751 2752 2753 2754

    info->notes_size += note_size(&ets->notes[0]);
}

2755 2756 2757 2758 2759 2760 2761 2762 2763 2764
static void init_note_info(struct elf_note_info *info)
{
    /* Initialize the elf_note_info structure so that it is at
     * least safe to call free_note_info() on it. Must be
     * called before calling fill_note_info().
     */
    memset(info, 0, sizeof (*info));
    QTAILQ_INIT(&info->thread_list);
}

2765
static int fill_note_info(struct elf_note_info *info,
2766
                          long signr, const CPUArchState *env)
2767 2768
{
#define NUMNOTES 3
2769 2770
    CPUState *cpu = ENV_GET_CPU((CPUArchState *)env);
    TaskState *ts = (TaskState *)cpu->opaque;
2771 2772
    int i;

2773
    info->notes = g_malloc0(NUMNOTES * sizeof (struct memelfnote));
2774 2775
    if (info->notes == NULL)
        return (-ENOMEM);
2776
    info->prstatus = g_malloc0(sizeof (*info->prstatus));
2777 2778
    if (info->prstatus == NULL)
        return (-ENOMEM);
2779
    info->psinfo = g_malloc0(sizeof (*info->psinfo));
2780 2781 2782 2783 2784 2785 2786 2787 2788 2789
    if (info->prstatus == NULL)
        return (-ENOMEM);

    /*
     * First fill in status (and registers) of current thread
     * including process info & aux vector.
     */
    fill_prstatus(info->prstatus, ts, signr);
    elf_core_copy_regs(&info->prstatus->pr_reg, env);
    fill_note(&info->notes[0], "CORE", NT_PRSTATUS,
2790
              sizeof (*info->prstatus), info->prstatus);
2791 2792
    fill_psinfo(info->psinfo, ts);
    fill_note(&info->notes[1], "CORE", NT_PRPSINFO,
2793
              sizeof (*info->psinfo), info->psinfo);
2794 2795 2796 2797 2798 2799 2800 2801 2802
    fill_auxv_note(&info->notes[2], ts);
    info->numnote = 3;

    info->notes_size = 0;
    for (i = 0; i < info->numnote; i++)
        info->notes_size += note_size(&info->notes[i]);

    /* read and fill status of all threads */
    cpu_list_lock();
A
Andreas Färber 已提交
2803
    CPU_FOREACH(cpu) {
2804
        if (cpu == thread_cpu) {
2805
            continue;
2806 2807
        }
        fill_thread_info(info, (CPUArchState *)cpu->env_ptr);
2808 2809 2810 2811 2812 2813 2814 2815 2816 2817
    }
    cpu_list_unlock();

    return (0);
}

static void free_note_info(struct elf_note_info *info)
{
    struct elf_thread_status *ets;

B
Blue Swirl 已提交
2818 2819 2820
    while (!QTAILQ_EMPTY(&info->thread_list)) {
        ets = QTAILQ_FIRST(&info->thread_list);
        QTAILQ_REMOVE(&info->thread_list, ets, ets_link);
2821
        g_free(ets);
2822 2823
    }

2824 2825 2826
    g_free(info->prstatus);
    g_free(info->psinfo);
    g_free(info->notes);
2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840
}

static int write_note_info(struct elf_note_info *info, int fd)
{
    struct elf_thread_status *ets;
    int i, error = 0;

    /* write prstatus, psinfo and auxv for current thread */
    for (i = 0; i < info->numnote; i++)
        if ((error = write_note(&info->notes[i], fd)) != 0)
            return (error);

    /* write prstatus for each thread */
    for (ets = info->thread_list.tqh_first; ets != NULL;
2841
         ets = ets->ets_link.tqe_next) {
2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891
        if ((error = write_note(&ets->notes[0], fd)) != 0)
            return (error);
    }

    return (0);
}

/*
 * Write out ELF coredump.
 *
 * See documentation of ELF object file format in:
 * http://www.caldera.com/developers/devspecs/gabi41.pdf
 *
 * Coredump format in linux is following:
 *
 * 0   +----------------------+         \
 *     | ELF header           | ET_CORE  |
 *     +----------------------+          |
 *     | ELF program headers  |          |--- headers
 *     | - NOTE section       |          |
 *     | - PT_LOAD sections   |          |
 *     +----------------------+         /
 *     | NOTEs:               |
 *     | - NT_PRSTATUS        |
 *     | - NT_PRSINFO         |
 *     | - NT_AUXV            |
 *     +----------------------+ <-- aligned to target page
 *     | Process memory dump  |
 *     :                      :
 *     .                      .
 *     :                      :
 *     |                      |
 *     +----------------------+
 *
 * NT_PRSTATUS -> struct elf_prstatus (per thread)
 * NT_PRSINFO  -> struct elf_prpsinfo
 * NT_AUXV is array of { type, value } pairs (see fill_auxv_note()).
 *
 * Format follows System V format as close as possible.  Current
 * version limitations are as follows:
 *     - no floating point registers are dumped
 *
 * Function returns 0 in case of success, negative errno otherwise.
 *
 * TODO: make this work also during runtime: it should be
 * possible to force coredump from running process and then
 * continue processing.  For example qemu could set up SIGUSR2
 * handler (provided that target process haven't registered
 * handler for that) that does the dump when signal is received.
 */
2892
static int elf_core_dump(int signr, const CPUArchState *env)
2893
{
2894 2895
    const CPUState *cpu = ENV_GET_CPU((CPUArchState *)env);
    const TaskState *ts = (const TaskState *)cpu->opaque;
2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906
    struct vm_area_struct *vma = NULL;
    char corefile[PATH_MAX];
    struct elf_note_info info;
    struct elfhdr elf;
    struct elf_phdr phdr;
    struct rlimit dumpsize;
    struct mm_struct *mm = NULL;
    off_t offset = 0, data_offset = 0;
    int segs = 0;
    int fd = -1;

2907 2908
    init_note_info(&info);

2909 2910 2911
    errno = 0;
    getrlimit(RLIMIT_CORE, &dumpsize);
    if (dumpsize.rlim_cur == 0)
2912
        return 0;
2913 2914 2915 2916 2917

    if (core_dump_filename(ts, corefile, sizeof (corefile)) < 0)
        return (-errno);

    if ((fd = open(corefile, O_WRONLY | O_CREAT,
2918
                   S_IRUSR|S_IWUSR|S_IRGRP|S_IROTH)) < 0)
2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957
        return (-errno);

    /*
     * Walk through target process memory mappings and
     * set up structure containing this information.  After
     * this point vma_xxx functions can be used.
     */
    if ((mm = vma_init()) == NULL)
        goto out;

    walk_memory_regions(mm, vma_walker);
    segs = vma_get_mapping_count(mm);

    /*
     * Construct valid coredump ELF header.  We also
     * add one more segment for notes.
     */
    fill_elf_header(&elf, segs + 1, ELF_MACHINE, 0);
    if (dump_write(fd, &elf, sizeof (elf)) != 0)
        goto out;

    /* fill in in-memory version of notes */
    if (fill_note_info(&info, signr, env) < 0)
        goto out;

    offset += sizeof (elf);                             /* elf header */
    offset += (segs + 1) * sizeof (struct elf_phdr);    /* program headers */

    /* write out notes program header */
    fill_elf_note_phdr(&phdr, info.notes_size, offset);

    offset += info.notes_size;
    if (dump_write(fd, &phdr, sizeof (phdr)) != 0)
        goto out;

    /*
     * ELF specification wants data to start at page boundary so
     * we align it here.
     */
2958
    data_offset = offset = roundup(offset, ELF_EXEC_PAGESIZE);
2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980

    /*
     * Write program headers for memory regions mapped in
     * the target process.
     */
    for (vma = vma_first(mm); vma != NULL; vma = vma_next(vma)) {
        (void) memset(&phdr, 0, sizeof (phdr));

        phdr.p_type = PT_LOAD;
        phdr.p_offset = offset;
        phdr.p_vaddr = vma->vma_start;
        phdr.p_paddr = 0;
        phdr.p_filesz = vma_dump_size(vma);
        offset += phdr.p_filesz;
        phdr.p_memsz = vma->vma_end - vma->vma_start;
        phdr.p_flags = vma->vma_flags & PROT_READ ? PF_R : 0;
        if (vma->vma_flags & PROT_WRITE)
            phdr.p_flags |= PF_W;
        if (vma->vma_flags & PROT_EXEC)
            phdr.p_flags |= PF_X;
        phdr.p_align = ELF_EXEC_PAGESIZE;

2981
        bswap_phdr(&phdr, 1);
2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005
        dump_write(fd, &phdr, sizeof (phdr));
    }

    /*
     * Next we write notes just after program headers.  No
     * alignment needed here.
     */
    if (write_note_info(&info, fd) < 0)
        goto out;

    /* align data to page boundary */
    if (lseek(fd, data_offset, SEEK_SET) != data_offset)
        goto out;

    /*
     * Finally we can dump process memory into corefile as well.
     */
    for (vma = vma_first(mm); vma != NULL; vma = vma_next(vma)) {
        abi_ulong addr;
        abi_ulong end;

        end = vma->vma_start + vma_dump_size(vma);

        for (addr = vma->vma_start; addr < end;
3006
             addr += TARGET_PAGE_SIZE) {
3007 3008 3009 3010 3011 3012 3013 3014 3015
            char page[TARGET_PAGE_SIZE];
            int error;

            /*
             *  Read in page from target process memory and
             *  write it to coredump file.
             */
            error = copy_from_user(page, addr, sizeof (page));
            if (error != 0) {
3016
                (void) fprintf(stderr, "unable to dump " TARGET_ABI_FMT_lx "\n",
3017
                               addr);
3018 3019 3020 3021 3022 3023 3024 3025
                errno = -error;
                goto out;
            }
            if (dump_write(fd, page, TARGET_PAGE_SIZE) < 0)
                goto out;
        }
    }

3026
 out:
3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037
    free_note_info(&info);
    if (mm != NULL)
        vma_delete(mm);
    (void) close(fd);

    if (errno != 0)
        return (-errno);
    return (0);
}
#endif /* USE_ELF_CORE_DUMP */

P
pbrook 已提交
3038 3039 3040 3041
void do_init_thread(struct target_pt_regs *regs, struct image_info *infop)
{
    init_thread(regs, infop);
}