fs/binfmt_elf.c

   1 /*
   2  * linux/fs/binfmt_elf.c
   3  *
   4  * These are the functions used to load ELF format executables as used
   5  * on SVr4 machines.  Information on the format may be found in the book
   6  * "UNIX SYSTEM V RELEASE 4 Programmers Guide: Ansi C and Programming Support
   7  * Tools".
   8  *
   9  * Copyright 1993, 1994: Eric Youngdale (ericy@cais.com).
  10  */
  11
  12 #include <linux/module.h>
  13 #include <linux/kernel.h>
  14 #include <linux/fs.h>
  15 #include <linux/mm.h>
  16 #include <linux/mman.h>
  17 #include <linux/errno.h>
  18 #include <linux/signal.h>
  19 #include <linux/binfmts.h>
  20 #include <linux/string.h>
  21 #include <linux/file.h>
  22 #include <linux/slab.h>
  23 #include <linux/personality.h>
  24 #include <linux/elfcore.h>
  25 #include <linux/init.h>
  26 #include <linux/highuid.h>
  27 #include <linux/compiler.h>
  28 #include <linux/highmem.h>
  29 #include <linux/pagemap.h>
  30 #include <linux/security.h>
  31 #include <linux/random.h>
  32 #include <linux/elf.h>
  33 #include <linux/utsname.h>
  34 #include <asm/uaccess.h>
  35 #include <asm/param.h>
  36 #include <asm/page.h>
  37
  38 static int load_elf_binary(struct linux_binprm *bprm, struct pt_regs *regs);
  39 static int load_elf_library(struct file *);
  40 static unsigned long elf_map(struct file *, unsigned long, struct elf_phdr *,
  41                                 int, int, unsigned long);
  42
  43 /*
  44  * If we don't support core dumping, then supply a NULL so we
  45  * don't even try.
  46  */
  47 #if defined(USE_ELF_CORE_DUMP) && defined(CONFIG_ELF_CORE)
  48 static int elf_core_dump(long signr, struct pt_regs *regs, struct file *file, unsigned long limit);
  49 #else
  50 #define elf_core_dump   NULL
  51 #endif
  52
  53 #if ELF_EXEC_PAGESIZE > PAGE_SIZE
  54 #define ELF_MIN_ALIGN   ELF_EXEC_PAGESIZE
  55 #else
  56 #define ELF_MIN_ALIGN   PAGE_SIZE
  57 #endif
  58
  59 #ifndef ELF_CORE_EFLAGS
  60 #define ELF_CORE_EFLAGS 0
  61 #endif
  62
  63 #define ELF_PAGESTART(_v) ((_v) & ~(unsigned long)(ELF_MIN_ALIGN-1))
  64 #define ELF_PAGEOFFSET(_v) ((_v) & (ELF_MIN_ALIGN-1))
  65 #define ELF_PAGEALIGN(_v) (((_v) + ELF_MIN_ALIGN - 1) & ~(ELF_MIN_ALIGN - 1))
  66
  67 static struct linux_binfmt elf_format = {
  68                 .module         = THIS_MODULE,
  69                 .load_binary    = load_elf_binary,
  70                 .load_shlib     = load_elf_library,
  71                 .core_dump      = elf_core_dump,
  72                 .min_coredump   = ELF_EXEC_PAGESIZE,
  73                 .hasvdso        = 1
  74 };
  75
  76 #define BAD_ADDR(x) ((unsigned long)(x) >= TASK_SIZE)
  77
  78 static int set_brk(unsigned long start, unsigned long end)
  79 {
  80         start = ELF_PAGEALIGN(start);
  81         end = ELF_PAGEALIGN(end);
  82         if (end > start) {
  83                 unsigned long addr;
  84                 down_write(&current->mm->mmap_sem);
  85                 addr = do_brk(start, end - start);
  86                 up_write(&current->mm->mmap_sem);
  87                 if (BAD_ADDR(addr))
  88                         return addr;
  89         }
  90         current->mm->start_brk = current->mm->brk = end;
  91         return 0;
  92 }
  93
  94 /* We need to explicitly zero any fractional pages
  95    after the data section (i.e. bss).  This would
  96    contain the junk from the file that should not
  97    be in memory
  98  */
  99 static int padzero(unsigned long elf_bss)
 100 {
 101         unsigned long nbyte;
 102
 103         nbyte = ELF_PAGEOFFSET(elf_bss);
 104         if (nbyte) {
 105                 nbyte = ELF_MIN_ALIGN - nbyte;
 106                 if (clear_user((void __user *) elf_bss, nbyte))
 107                         return -EFAULT;
 108         }
 109         return 0;
 110 }
 111
 112 /* Let's use some macros to make this stack manipulation a little clearer */
 113 #ifdef CONFIG_STACK_GROWSUP
 114 #define STACK_ADD(sp, items) ((elf_addr_t __user *)(sp) + (items))
 115 #define STACK_ROUND(sp, items) \
 116         ((15 + (unsigned long) ((sp) + (items))) &~ 15UL)
 117 #define STACK_ALLOC(sp, len) ({ \
 118         elf_addr_t __user *old_sp = (elf_addr_t __user *)sp; sp += len; \
 119         old_sp; })
 120 #else
 121 #define STACK_ADD(sp, items) ((elf_addr_t __user *)(sp) - (items))
 122 #define STACK_ROUND(sp, items) \
 123         (((unsigned long) (sp - items)) &~ 15UL)
 124 #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
 125 #endif
 126
 127 #ifndef ELF_BASE_PLATFORM
 128 /*
 129  * AT_BASE_PLATFORM indicates the "real" hardware/microarchitecture.
 130  * If the arch defines ELF_BASE_PLATFORM (in asm/elf.h), the value
 131  * will be copied to the user stack in the same manner as AT_PLATFORM.
 132  */
 133 #define ELF_BASE_PLATFORM NULL
 134 #endif
 135
 136 static int
 137 create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 138                 unsigned long load_addr, unsigned long interp_load_addr)
 139 {
 140         unsigned long p = bprm->p;
 141         int argc = bprm->argc;
 142         int envc = bprm->envc;
 143         elf_addr_t __user *argv;
 144         elf_addr_t __user *envp;
 145         elf_addr_t __user *sp;
 146         elf_addr_t __user *u_platform;
 147         elf_addr_t __user *u_base_platform;
 148         elf_addr_t __user *u_rand_bytes;
 149         const char *k_platform = ELF_PLATFORM;
 150         const char *k_base_platform = ELF_BASE_PLATFORM;
 151         unsigned char k_rand_bytes[16];
 152         int items;
 153         elf_addr_t *elf_info;
 154         int ei_index = 0;
 155         const struct cred *cred = current_cred();
 156         struct vm_area_struct *vma;
 157
 158         /*
 159          * In some cases (e.g. Hyper-Threading), we want to avoid L1
 160          * evictions by the processes running on the same package. One
 161          * thing we can do is to shuffle the initial stack for them.
 162          */
 163
 164         p = arch_align_stack(p);
 165
 166         /*
 167          * If this architecture has a platform capability string, copy it
 168          * to userspace.  In some cases (Sparc), this info is impossible
 169          * for userspace to get any other way, in others (i386) it is
 170          * merely difficult.
 171          */
 172         u_platform = NULL;
 173         if (k_platform) {
 174                 size_t len = strlen(k_platform) + 1;
 175
 176                 u_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
 177                 if (__copy_to_user(u_platform, k_platform, len))
 178                         return -EFAULT;
 179         }
 180
 181         /*
 182          * If this architecture has a "base" platform capability
 183          * string, copy it to userspace.
 184          */
 185         u_base_platform = NULL;
 186         if (k_base_platform) {
 187                 size_t len = strlen(k_base_platform) + 1;
 188
 189                 u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
 190                 if (__copy_to_user(u_base_platform, k_base_platform, len))
 191                         return -EFAULT;
 192         }
 193
 194         /*
 195          * Generate 16 random bytes for userspace PRNG seeding.
 196          */
 197         get_random_bytes(k_rand_bytes, sizeof(k_rand_bytes));
 198         u_rand_bytes = (elf_addr_t __user *)
 199                        STACK_ALLOC(p, sizeof(k_rand_bytes));
 200         if (__copy_to_user(u_rand_bytes, k_rand_bytes, sizeof(k_rand_bytes)))
 201                 return -EFAULT;
 202
 203         /* Create the ELF interpreter info */
 204         elf_info = (elf_addr_t *)current->mm->saved_auxv;
 205         /* update AT_VECTOR_SIZE_BASE if the number of NEW_AUX_ENT() changes */
 206 #define NEW_AUX_ENT(id, val) \
 207         do { \
 208                 elf_info[ei_index++] = id; \
 209                 elf_info[ei_index++] = val; \
 210         } while (0)
 211
 212 #ifdef ARCH_DLINFO
 213         /*
 214          * ARCH_DLINFO must come first so PPC can do its special alignment of
 215          * AUXV.
 216          * update AT_VECTOR_SIZE_ARCH if the number of NEW_AUX_ENT() in
 217          * ARCH_DLINFO changes
 218          */
 219         ARCH_DLINFO;
 220 #endif
 221         NEW_AUX_ENT(AT_HWCAP, ELF_HWCAP);
 222         NEW_AUX_ENT(AT_PAGESZ, ELF_EXEC_PAGESIZE);
 223         NEW_AUX_ENT(AT_CLKTCK, CLOCKS_PER_SEC);
 224         NEW_AUX_ENT(AT_PHDR, load_addr + exec->e_phoff);
 225         NEW_AUX_ENT(AT_PHENT, sizeof(struct elf_phdr));
 226         NEW_AUX_ENT(AT_PHNUM, exec->e_phnum);
 227         NEW_AUX_ENT(AT_BASE, interp_load_addr);
 228         NEW_AUX_ENT(AT_FLAGS, 0);
 229         NEW_AUX_ENT(AT_ENTRY, exec->e_entry);
 230         NEW_AUX_ENT(AT_UID, cred->uid);
 231         NEW_AUX_ENT(AT_EUID, cred->euid);
 232         NEW_AUX_ENT(AT_GID, cred->gid);
 233         NEW_AUX_ENT(AT_EGID, cred->egid);
 234         NEW_AUX_ENT(AT_SECURE, security_bprm_secureexec(bprm));
 235         NEW_AUX_ENT(AT_RANDOM, (elf_addr_t)(unsigned long)u_rand_bytes);
 236         NEW_AUX_ENT(AT_EXECFN, bprm->exec);
 237         if (k_platform) {
 238                 NEW_AUX_ENT(AT_PLATFORM,
 239                             (elf_addr_t)(unsigned long)u_platform);
 240         }
 241         if (k_base_platform) {
 242                 NEW_AUX_ENT(AT_BASE_PLATFORM,
 243                             (elf_addr_t)(unsigned long)u_base_platform);
 244         }
 245         if (bprm->interp_flags & BINPRM_FLAGS_EXECFD) {
 246                 NEW_AUX_ENT(AT_EXECFD, bprm->interp_data);
 247         }
 248 #undef NEW_AUX_ENT
 249         /* AT_NULL is zero; clear the rest too */
 250         memset(&elf_info[ei_index], 0,
 251                sizeof current->mm->saved_auxv - ei_index * sizeof elf_info[0]);
 252
 253         /* And advance past the AT_NULL entry.  */
 254         ei_index += 2;
 255
 256         sp = STACK_ADD(p, ei_index);
 257
 258         items = (argc + 1) + (envc + 1) + 1;
 259         bprm->p = STACK_ROUND(sp, items);
 260
 261         /* Point sp at the lowest address on the stack */
 262 #ifdef CONFIG_STACK_GROWSUP
 263         sp = (elf_addr_t __user *)bprm->p - items - ei_index;
 264         bprm->exec = (unsigned long)sp; /* XXX: PARISC HACK */
 265 #else
 266         sp = (elf_addr_t __user *)bprm->p;
 267 #endif
 268
 269
 270         /*
 271          * Grow the stack manually; some architectures have a limit on how
 272          * far ahead a user-space access may be in order to grow the stack.
 273          */
 274         vma = find_extend_vma(current->mm, bprm->p);
 275         if (!vma)
 276                 return -EFAULT;
 277
 278         /* Now, let's put argc (and argv, envp if appropriate) on the stack */
 279         if (__put_user(argc, sp++))
 280                 return -EFAULT;
 281         argv = sp;
 282         envp = argv + argc + 1;
 283
 284         /* Populate argv and envp */
 285         p = current->mm->arg_end = current->mm->arg_start;
 286         while (argc-- > 0) {
 287                 size_t len;
 288                 if (__put_user((elf_addr_t)p, argv++))
 289                         return -EFAULT;
 290                 len = strnlen_user((void __user *)p, MAX_ARG_STRLEN);
 291                 if (!len || len > MAX_ARG_STRLEN)
 292                         return -EINVAL;
 293                 p += len;
 294         }
 295         if (__put_user(0, argv))
 296                 return -EFAULT;
 297         current->mm->arg_end = current->mm->env_start = p;
 298         while (envc-- > 0) {
 299                 size_t len;
 300                 if (__put_user((elf_addr_t)p, envp++))
 301                         return -EFAULT;
 302                 len = strnlen_user((void __user *)p, MAX_ARG_STRLEN);
 303                 if (!len || len > MAX_ARG_STRLEN)
 304                         return -EINVAL;
 305                 p += len;
 306         }
 307         if (__put_user(0, envp))
 308                 return -EFAULT;
 309         current->mm->env_end = p;
 310
 311         /* Put the elf_info on the stack in the right place.  */
 312         sp = (elf_addr_t __user *)envp + 1;
 313         if (copy_to_user(sp, elf_info, ei_index * sizeof(elf_addr_t)))
 314                 return -EFAULT;
 315         return 0;
 316 }
 317
 318 #ifndef elf_map
 319
 320 static unsigned long elf_map(struct file *filep, unsigned long addr,
 321                 struct elf_phdr *eppnt, int prot, int type,
 322                 unsigned long total_size)
 323 {
 324         unsigned long map_addr;
 325         unsigned long size = eppnt->p_filesz + ELF_PAGEOFFSET(eppnt->p_vaddr);
 326         unsigned long off = eppnt->p_offset - ELF_PAGEOFFSET(eppnt->p_vaddr);
 327         addr = ELF_PAGESTART(addr);
 328         size = ELF_PAGEALIGN(size);
 329
 330         /* mmap() will return -EINVAL if given a zero size, but a
 331          * segment with zero filesize is perfectly valid */
 332         if (!size)
 333                 return addr;
 334
 335         down_write(&current->mm->mmap_sem);
 336         /*
 337         * total_size is the size of the ELF (interpreter) image.
 338         * The _first_ mmap needs to know the full size, otherwise
 339         * randomization might put this image into an overlapping
 340         * position with the ELF binary image. (since size < total_size)
 341         * So we first map the 'big' image - and unmap the remainder at
 342         * the end. (which unmap is needed for ELF images with holes.)
 343         */
 344         if (total_size) {
 345                 total_size = ELF_PAGEALIGN(total_size);
 346                 map_addr = do_mmap(filep, addr, total_size, prot, type, off);
 347                 if (!BAD_ADDR(map_addr))
 348                         do_munmap(current->mm, map_addr+size, total_size-size);
 349         } else
 350                 map_addr = do_mmap(filep, addr, size, prot, type, off);
 351
 352         up_write(&current->mm->mmap_sem);
 353         return(map_addr);
 354 }
 355
 356 #endif /* !elf_map */
 357
 358 static unsigned long total_mapping_size(struct elf_phdr *cmds, int nr)
 359 {
 360         int i, first_idx = -1, last_idx = -1;
 361
 362         for (i = 0; i < nr; i++) {
 363                 if (cmds[i].p_type == PT_LOAD) {
 364                         last_idx = i;
 365                         if (first_idx == -1)
 366                                 first_idx = i;
 367                 }
 368         }
 369         if (first_idx == -1)
 370                 return 0;
 371
 372         return cmds[last_idx].p_vaddr + cmds[last_idx].p_memsz -
 373                                 ELF_PAGESTART(cmds[first_idx].p_vaddr);
 374 }
 375
 376
 377 /* This is much more generalized than the library routine read function,
 378    so we keep this separate.  Technically the library read function
 379    is only provided so that we can read a.out libraries that have
 380    an ELF header */
 381
 382 static unsigned long load_elf_interp(struct elfhdr *interp_elf_ex,
 383                 struct file *interpreter, unsigned long *interp_map_addr,
 384                 unsigned long no_base)
 385 {
 386         struct elf_phdr *elf_phdata;
 387         struct elf_phdr *eppnt;
 388         unsigned long load_addr = 0;
 389         int load_addr_set = 0;
 390         unsigned long last_bss = 0, elf_bss = 0;
 391         unsigned long error = ~0UL;
 392         unsigned long total_size;
 393         int retval, i, size;
 394
 395         /* First of all, some simple consistency checks */
 396         if (interp_elf_ex->e_type != ET_EXEC &&
 397             interp_elf_ex->e_type != ET_DYN)
 398                 goto out;
 399         if (!elf_check_arch(interp_elf_ex))
 400                 goto out;
 401         if (!interpreter->f_op || !interpreter->f_op->mmap)
 402                 goto out;
 403
 404         /*
 405          * If the size of this structure has changed, then punt, since
 406          * we will be doing the wrong thing.
 407          */
 408         if (interp_elf_ex->e_phentsize != sizeof(struct elf_phdr))
 409                 goto out;
 410         if (interp_elf_ex->e_phnum < 1 ||
 411                 interp_elf_ex->e_phnum > 65536U / sizeof(struct elf_phdr))
 412                 goto out;
 413
 414         /* Now read in all of the header information */
 415         size = sizeof(struct elf_phdr) * interp_elf_ex->e_phnum;
 416         if (size > ELF_MIN_ALIGN)
 417                 goto out;
 418         elf_phdata = kmalloc(size, GFP_KERNEL);
 419         if (!elf_phdata)
 420                 goto out;
 421
 422         retval = kernel_read(interpreter, interp_elf_ex->e_phoff,
 423                              (char *)elf_phdata,size);
 424         error = -EIO;
 425         if (retval != size) {
 426                 if (retval < 0)
 427                         error = retval;
 428                 goto out_close;
 429         }
 430
 431         total_size = total_mapping_size(elf_phdata, interp_elf_ex->e_phnum);
 432         if (!total_size) {
 433                 error = -EINVAL;
 434                 goto out_close;
 435         }
 436
 437         eppnt = elf_phdata;
 438         for (i = 0; i < interp_elf_ex->e_phnum; i++, eppnt++) {
 439                 if (eppnt->p_type == PT_LOAD) {
 440                         int elf_type = MAP_PRIVATE | MAP_DENYWRITE;
 441                         int elf_prot = 0;
 442                         unsigned long vaddr = 0;
 443                         unsigned long k, map_addr;
 444
 445                         if (eppnt->p_flags & PF_R)
 446                                 elf_prot = PROT_READ;
 447                         if (eppnt->p_flags & PF_W)
 448                                 elf_prot |= PROT_WRITE;
 449                         if (eppnt->p_flags & PF_X)
 450                                 elf_prot |= PROT_EXEC;
 451                         vaddr = eppnt->p_vaddr;
 452                         if (interp_elf_ex->e_type == ET_EXEC || load_addr_set)
 453                                 elf_type |= MAP_FIXED;
 454                         else if (no_base && interp_elf_ex->e_type == ET_DYN)
 455                                 load_addr = -vaddr;
 456
 457                         map_addr = elf_map(interpreter, load_addr + vaddr,
 458                                         eppnt, elf_prot, elf_type, total_size);
 459                         total_size = 0;
 460                         if (!*interp_map_addr)
 461                                 *interp_map_addr = map_addr;
 462                         error = map_addr;
 463                         if (BAD_ADDR(map_addr))
 464                                 goto out_close;
 465
 466                         if (!load_addr_set &&
 467                             interp_elf_ex->e_type == ET_DYN) {
 468                                 load_addr = map_addr - ELF_PAGESTART(vaddr);
 469                                 load_addr_set = 1;
 470                         }
 471
 472                         /*
 473                          * Check to see if the section's size will overflow the
 474                          * allowed task size. Note that p_filesz must always be
 475                          * <= p_memsize so it's only necessary to check p_memsz.
 476                          */
 477                         k = load_addr + eppnt->p_vaddr;
 478                         if (BAD_ADDR(k) ||
 479                             eppnt->p_filesz > eppnt->p_memsz ||
 480                             eppnt->p_memsz > TASK_SIZE ||
 481                             TASK_SIZE - eppnt->p_memsz < k) {
 482                                 error = -ENOMEM;
 483                                 goto out_close;
 484                         }
 485
 486                         /*
 487                          * Find the end of the file mapping for this phdr, and
 488                          * keep track of the largest address we see for this.
 489                          */
 490                         k = load_addr + eppnt->p_vaddr + eppnt->p_filesz;
 491                         if (k > elf_bss)
 492                                 elf_bss = k;
 493
 494                         /*
 495                          * Do the same thing for the memory mapping - between
 496                          * elf_bss and last_bss is the bss section.
 497                          */
 498                         k = load_addr + eppnt->p_memsz + eppnt->p_vaddr;
 499                         if (k > last_bss)
 500                                 last_bss = k;
 501                 }
 502         }
 503
 504         if (last_bss > elf_bss) {
 505                 /*
 506                  * Now fill out the bss section.  First pad the last page up
 507                  * to the page boundary, and then perform a mmap to make sure
 508                  * that there are zero-mapped pages up to and including the
 509                  * last bss page.
 510                  */
 511                 if (padzero(elf_bss)) {
 512                         error = -EFAULT;
 513                         goto out_close;
 514                 }
 515
 516                 /* What we have mapped so far */
 517                 elf_bss = ELF_PAGESTART(elf_bss + ELF_MIN_ALIGN - 1);
 518
 519                 /* Map the last of the bss segment */
 520                 down_write(&current->mm->mmap_sem);
 521                 error = do_brk(elf_bss, last_bss - elf_bss);
 522                 up_write(&current->mm->mmap_sem);
 523                 if (BAD_ADDR(error))
 524                         goto out_close;
 525         }
 526
 527         error = load_addr;
 528
 529 out_close:
 530         kfree(elf_phdata);
 531 out:
 532         return error;
 533 }
 534
 535 /*
 536  * These are the functions used to load ELF style executables and shared
 537  * libraries.  There is no binary dependent code anywhere else.
 538  */
 539
 540 #define INTERPRETER_NONE 0
 541 #define INTERPRETER_ELF 2
 542
 543 #ifndef STACK_RND_MASK
 544 #define STACK_RND_MASK (0x7ff >> (PAGE_SHIFT - 12))     /* 8MB of VA */
 545 #endif
 546
 547 static unsigned long randomize_stack_top(unsigned long stack_top)
 548 {
 549         unsigned int random_variable = 0;
 550
 551         if ((current->flags & PF_RANDOMIZE) &&
 552                 !(current->personality & ADDR_NO_RANDOMIZE)) {
 553                 random_variable = get_random_int() & STACK_RND_MASK;
 554                 random_variable <<= PAGE_SHIFT;
 555         }
 556 #ifdef CONFIG_STACK_GROWSUP
 557         return PAGE_ALIGN(stack_top) + random_variable;
 558 #else
 559         return PAGE_ALIGN(stack_top) - random_variable;
 560 #endif
 561 }
 562
 563 static int load_elf_binary(struct linux_binprm *bprm, struct pt_regs *regs)
 564 {
 565         struct file *interpreter = NULL; /* to shut gcc up */
 566         unsigned long load_addr = 0, load_bias = 0;
 567         int load_addr_set = 0;
 568         char * elf_interpreter = NULL;
 569         unsigned long error;
 570         struct elf_phdr *elf_ppnt, *elf_phdata;
 571         unsigned long elf_bss, elf_brk;
 572         int retval, i;
 573         unsigned int size;
 574         unsigned long elf_entry;
 575         unsigned long interp_load_addr = 0;
 576         unsigned long start_code, end_code, start_data, end_data;
 577         unsigned long reloc_func_desc = 0;
 578         int executable_stack = EXSTACK_DEFAULT;
 579         unsigned long def_flags = 0;
 580         struct {
 581                 struct elfhdr elf_ex;
 582                 struct elfhdr interp_elf_ex;
 583         } *loc;
 584
 585         loc = kmalloc(sizeof(*loc), GFP_KERNEL);
 586         if (!loc) {
 587                 retval = -ENOMEM;
 588                 goto out_ret;
 589         }
 590
 591         /* Get the exec-header */
 592         loc->elf_ex = *((struct elfhdr *)bprm->buf);
 593
 594         retval = -ENOEXEC;
 595         /* First of all, some simple consistency checks */
 596         if (memcmp(loc->elf_ex.e_ident, ELFMAG, SELFMAG) != 0)
 597                 goto out;
 598
 599         if (loc->elf_ex.e_type != ET_EXEC && loc->elf_ex.e_type != ET_DYN)
 600                 goto out;
 601         if (!elf_check_arch(&loc->elf_ex))
 602                 goto out;
 603         if (!bprm->file->f_op||!bprm->file->f_op->mmap)
 604                 goto out;
 605
 606         /* Now read in all of the header information */
 607         if (loc->elf_ex.e_phentsize != sizeof(struct elf_phdr))
 608                 goto out;
 609         if (loc->elf_ex.e_phnum < 1 ||
 610                 loc->elf_ex.e_phnum > 65536U / sizeof(struct elf_phdr))
 611                 goto out;
 612         size = loc->elf_ex.e_phnum * sizeof(struct elf_phdr);
 613         retval = -ENOMEM;
 614         elf_phdata = kmalloc(size, GFP_KERNEL);
 615         if (!elf_phdata)
 616                 goto out;
 617
 618         retval = kernel_read(bprm->file, loc->elf_ex.e_phoff,
 619                              (char *)elf_phdata, size);
 620         if (retval != size) {
 621                 if (retval >= 0)
 622                         retval = -EIO;
 623                 goto out_free_ph;
 624         }
 625
 626         elf_ppnt = elf_phdata;
 627         elf_bss = 0;
 628         elf_brk = 0;
 629
 630         start_code = ~0UL;
 631         end_code = 0;
 632         start_data = 0;
 633         end_data = 0;
 634
 635         for (i = 0; i < loc->elf_ex.e_phnum; i++) {
 636                 if (elf_ppnt->p_type == PT_INTERP) {
 637                         /* This is the program interpreter used for
 638                          * shared libraries - for now assume that this
 639                          * is an a.out format binary
 640                          */
 641                         retval = -ENOEXEC;
 642                         if (elf_ppnt->p_filesz > PATH_MAX ||
 643                             elf_ppnt->p_filesz < 2)
 644                                 goto out_free_ph;
 645
 646                         retval = -ENOMEM;
 647                         elf_interpreter = kmalloc(elf_ppnt->p_filesz,
 648                                                   GFP_KERNEL);
 649                         if (!elf_interpreter)
 650                                 goto out_free_ph;
 651
 652                         retval = kernel_read(bprm->file, elf_ppnt->p_offset,
 653                                              elf_interpreter,
 654                                              elf_ppnt->p_filesz);
 655                         if (retval != elf_ppnt->p_filesz) {
 656                                 if (retval >= 0)
 657                                         retval = -EIO;
 658                                 goto out_free_interp;
 659                         }
 660                         /* make sure path is NULL terminated */
 661                         retval = -ENOEXEC;
 662                         if (elf_interpreter[elf_ppnt->p_filesz - 1] != '\0')
 663                                 goto out_free_interp;
 664
 665                         interpreter = open_exec(elf_interpreter);
 666                         retval = PTR_ERR(interpreter);
 667                         if (IS_ERR(interpreter))
 668                                 goto out_free_interp;
 669
 670                         /*
 671                          * If the binary is not readable then enforce
 672                          * mm->dumpable = 0 regardless of the interpreter's
 673                          * permissions.
 674                          */
 675                         if (file_permission(interpreter, MAY_READ) < 0)
 676                                 bprm->interp_flags |= BINPRM_FLAGS_ENFORCE_NONDUMP;
 677
 678                         retval = kernel_read(interpreter, 0, bprm->buf,
 679                                              BINPRM_BUF_SIZE);
 680                         if (retval != BINPRM_BUF_SIZE) {
 681                                 if (retval >= 0)
 682                                         retval = -EIO;
 683                                 goto out_free_dentry;
 684                         }
 685
 686                         /* Get the exec headers */
 687                         loc->interp_elf_ex = *((struct elfhdr *)bprm->buf);
 688                         break;
 689                 }
 690                 elf_ppnt++;
 691         }
 692
 693         elf_ppnt = elf_phdata;
 694         for (i = 0; i < loc->elf_ex.e_phnum; i++, elf_ppnt++)
 695                 if (elf_ppnt->p_type == PT_GNU_STACK) {
 696                         if (elf_ppnt->p_flags & PF_X)
 697                                 executable_stack = EXSTACK_ENABLE_X;
 698                         else
 699                                 executable_stack = EXSTACK_DISABLE_X;
 700                         break;
 701                 }
 702
 703         /* Some simple consistency checks for the interpreter */
 704         if (elf_interpreter) {
 705                 retval = -ELIBBAD;
 706                 /* Not an ELF interpreter */
 707                 if (memcmp(loc->interp_elf_ex.e_ident, ELFMAG, SELFMAG) != 0)
 708                         goto out_free_dentry;
 709                 /* Verify the interpreter has a valid arch */
 710                 if (!elf_check_arch(&loc->interp_elf_ex))
 711                         goto out_free_dentry;
 712         }
 713
 714         /* Flush all traces of the currently running executable */
 715         retval = flush_old_exec(bprm);
 716         if (retval)
 717                 goto out_free_dentry;
 718
 719         /* OK, This is the point of no return */
 720         current->flags &= ~PF_FORKNOEXEC;
 721         current->mm->def_flags = def_flags;
 722
 723         /* Do this immediately, since STACK_TOP as used in setup_arg_pages
 724            may depend on the personality.  */
 725         SET_PERSONALITY(loc->elf_ex);
 726         if (elf_read_implies_exec(loc->elf_ex, executable_stack))
 727                 current->personality |= READ_IMPLIES_EXEC;
 728
 729         if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space)
 730                 current->flags |= PF_RANDOMIZE;
 731
 732         setup_new_exec(bprm);
 733
 734         /* Do this so that we can load the interpreter, if need be.  We will
 735            change some of these later */
 736         current->mm->free_area_cache = current->mm->mmap_base;
 737         current->mm->cached_hole_size = 0;
 738         retval = setup_arg_pages(bprm, randomize_stack_top(STACK_TOP),
 739                                  executable_stack);
 740         if (retval < 0) {
 741                 send_sig(SIGKILL, current, 0);
 742                 goto out_free_dentry;
 743         }
 744
 745         current->mm->start_stack = bprm->p;
 746
 747         /* Now we do a little grungy work by mmaping the ELF image into
 748            the correct location in memory. */
 749         for(i = 0, elf_ppnt = elf_phdata;
 750             i < loc->elf_ex.e_phnum; i++, elf_ppnt++) {
 751                 int elf_prot = 0, elf_flags;
 752                 unsigned long k, vaddr;
 753
 754                 if (elf_ppnt->p_type != PT_LOAD)
 755                         continue;
 756
 757                 if (unlikely (elf_brk > elf_bss)) {
 758                         unsigned long nbyte;
 759
 760                         /* There was a PT_LOAD segment with p_memsz > p_filesz
 761                            before this one. Map anonymous pages, if needed,
 762                            and clear the area.  */
 763                         retval = set_brk (elf_bss + load_bias,
 764                                           elf_brk + load_bias);
 765                         if (retval) {
 766                                 send_sig(SIGKILL, current, 0);
 767                                 goto out_free_dentry;
 768                         }
 769                         nbyte = ELF_PAGEOFFSET(elf_bss);
 770                         if (nbyte) {
 771                                 nbyte = ELF_MIN_ALIGN - nbyte;
 772                                 if (nbyte > elf_brk - elf_bss)
 773                                         nbyte = elf_brk - elf_bss;
 774                                 if (clear_user((void __user *)elf_bss +
 775                                                         load_bias, nbyte)) {
 776                                         /*
 777                                          * This bss-zeroing can fail if the ELF
 778                                          * file specifies odd protections. So
 779                                          * we don't check the return value
 780                                          */
 781                                 }
 782                         }
 783                 }
 784
 785                 if (elf_ppnt->p_flags & PF_R)
 786                         elf_prot |= PROT_READ;
 787                 if (elf_ppnt->p_flags & PF_W)
 788                         elf_prot |= PROT_WRITE;
 789                 if (elf_ppnt->p_flags & PF_X)
 790                         elf_prot |= PROT_EXEC;
 791
 792                 elf_flags = MAP_PRIVATE | MAP_DENYWRITE | MAP_EXECUTABLE;
 793
 794                 vaddr = elf_ppnt->p_vaddr;
 795                 if (loc->elf_ex.e_type == ET_EXEC || load_addr_set) {
 796                         elf_flags |= MAP_FIXED;
 797                 } else if (loc->elf_ex.e_type == ET_DYN) {
 798                         /* Try and get dynamic programs out of the way of the
 799                          * default mmap base, as well as whatever program they
 800                          * might try to exec.  This is because the brk will
 801                          * follow the loader, and is not movable.  */
 802 #ifdef CONFIG_X86
 803                         load_bias = 0;
 804 #else
 805                         load_bias = ELF_PAGESTART(ELF_ET_DYN_BASE - vaddr);
 806 #endif
 807                 }
 808
 809                 error = elf_map(bprm->file, load_bias + vaddr, elf_ppnt,
 810                                 elf_prot, elf_flags, 0);
 811                 if (BAD_ADDR(error)) {
 812                         send_sig(SIGKILL, current, 0);
 813                         retval = IS_ERR((void *)error) ?
 814                                 PTR_ERR((void*)error) : -EINVAL;
 815                         goto out_free_dentry;
 816                 }
 817
 818                 if (!load_addr_set) {
 819                         load_addr_set = 1;
 820                         load_addr = (elf_ppnt->p_vaddr - elf_ppnt->p_offset);
 821                         if (loc->elf_ex.e_type == ET_DYN) {
 822                                 load_bias += error -
 823                                              ELF_PAGESTART(load_bias + vaddr);
 824                                 load_addr += load_bias;
 825                                 reloc_func_desc = load_bias;
 826                         }
 827                 }
 828                 k = elf_ppnt->p_vaddr;
 829                 if (k < start_code)
 830                         start_code = k;
 831                 if (start_data < k)
 832                         start_data = k;
 833
 834                 /*
 835                  * Check to see if the section's size will overflow the
 836                  * allowed task size. Note that p_filesz must always be
 837                  * <= p_memsz so it is only necessary to check p_memsz.
 838                  */
 839                 if (BAD_ADDR(k) || elf_ppnt->p_filesz > elf_ppnt->p_memsz ||
 840                     elf_ppnt->p_memsz > TASK_SIZE ||
 841                     TASK_SIZE - elf_ppnt->p_memsz < k) {
 842                         /* set_brk can never work. Avoid overflows. */
 843                         send_sig(SIGKILL, current, 0);
 844                         retval = -EINVAL;
 845                         goto out_free_dentry;
 846                 }
 847
 848                 k = elf_ppnt->p_vaddr + elf_ppnt->p_filesz;
 849
 850                 if (k > elf_bss)
 851                         elf_bss = k;
 852                 if ((elf_ppnt->p_flags & PF_X) && end_code < k)
 853                         end_code = k;
 854                 if (end_data < k)
 855                         end_data = k;
 856                 k = elf_ppnt->p_vaddr + elf_ppnt->p_memsz;
 857                 if (k > elf_brk)
 858                         elf_brk = k;
 859         }
 860
 861         loc->elf_ex.e_entry += load_bias;
 862         elf_bss += load_bias;
 863         elf_brk += load_bias;
 864         start_code += load_bias;
 865         end_code += load_bias;
 866         start_data += load_bias;
 867         end_data += load_bias;
 868
 869         /* Calling set_brk effectively mmaps the pages that we need
 870          * for the bss and break sections.  We must do this before
 871          * mapping in the interpreter, to make sure it doesn't wind
 872          * up getting placed where the bss needs to go.
 873          */
 874         retval = set_brk(elf_bss, elf_brk);
 875         if (retval) {
 876                 send_sig(SIGKILL, current, 0);
 877                 goto out_free_dentry;
 878         }
 879         if (likely(elf_bss != elf_brk) && unlikely(padzero(elf_bss))) {
 880                 send_sig(SIGSEGV, current, 0);
 881                 retval = -EFAULT; /* Nobody gets to see this, but.. */
 882                 goto out_free_dentry;
 883         }
 884
 885         if (elf_interpreter) {
 886                 unsigned long uninitialized_var(interp_map_addr);
 887
 888                 elf_entry = load_elf_interp(&loc->interp_elf_ex,
 889                                             interpreter,
 890                                             &interp_map_addr,
 891                                             load_bias);
 892                 if (!IS_ERR((void *)elf_entry)) {
 893                         /*
 894                          * load_elf_interp() returns relocation
 895                          * adjustment
 896                          */
 897                         interp_load_addr = elf_entry;
 898                         elf_entry += loc->interp_elf_ex.e_entry;
 899                 }
 900                 if (BAD_ADDR(elf_entry)) {
 901                         force_sig(SIGSEGV, current);
 902                         retval = IS_ERR((void *)elf_entry) ?
 903                                         (int)elf_entry : -EINVAL;
 904                         goto out_free_dentry;
 905                 }
 906                 reloc_func_desc = interp_load_addr;
 907
 908                 allow_write_access(interpreter);
 909                 fput(interpreter);
 910                 kfree(elf_interpreter);
 911         } else {
 912                 elf_entry = loc->elf_ex.e_entry;
 913                 if (BAD_ADDR(elf_entry)) {
 914                         force_sig(SIGSEGV, current);
 915                         retval = -EINVAL;
 916                         goto out_free_dentry;
 917                 }
 918         }
 919
 920         kfree(elf_phdata);
 921
 922         set_binfmt(&elf_format);
 923
 924 #ifdef ARCH_HAS_SETUP_ADDITIONAL_PAGES
 925         retval = arch_setup_additional_pages(bprm, !!elf_interpreter);
 926         if (retval < 0) {
 927                 send_sig(SIGKILL, current, 0);
 928                 goto out;
 929         }
 930 #endif /* ARCH_HAS_SETUP_ADDITIONAL_PAGES */
 931
 932         install_exec_creds(bprm);
 933         current->flags &= ~PF_FORKNOEXEC;
 934         retval = create_elf_tables(bprm, &loc->elf_ex,
 935                           load_addr, interp_load_addr);
 936         if (retval < 0) {
 937                 send_sig(SIGKILL, current, 0);
 938                 goto out;
 939         }
 940         /* N.B. passed_fileno might not be initialized? */
 941         current->mm->end_code = end_code;
 942         current->mm->start_code = start_code;
 943         current->mm->start_data = start_data;
 944         current->mm->end_data = end_data;
 945         current->mm->start_stack = bprm->p;
 946
 947 #ifdef arch_randomize_brk
 948         if ((current->flags & PF_RANDOMIZE) && (randomize_va_space > 1))
 949                 current->mm->brk = current->mm->start_brk =
 950                         arch_randomize_brk(current->mm);
 951 #endif
 952
 953         if (current->personality & MMAP_PAGE_ZERO) {
 954                 /* Why this, you ask???  Well SVr4 maps page 0 as read-only,
 955                    and some applications "depend" upon this behavior.
 956                    Since we do not have the power to recompile these, we
 957                    emulate the SVr4 behavior. Sigh. */
 958                 down_write(&current->mm->mmap_sem);
 959                 error = do_mmap(NULL, 0, PAGE_SIZE, PROT_READ | PROT_EXEC,
 960                                 MAP_FIXED | MAP_PRIVATE, 0);
 961                 up_write(&current->mm->mmap_sem);
 962         }
 963
 964 #ifdef ELF_PLAT_INIT
 965         /*
 966          * The ABI may specify that certain registers be set up in special
 967          * ways (on i386 %edx is the address of a DT_FINI function, for
 968          * example.  In addition, it may also specify (eg, PowerPC64 ELF)
 969          * that the e_entry field is the address of the function descriptor
 970          * for the startup routine, rather than the address of the startup
 971          * routine itself.  This macro performs whatever initialization to
 972          * the regs structure is required as well as any relocations to the
 973          * function descriptor entries when executing dynamically links apps.
 974          */
 975         ELF_PLAT_INIT(regs, reloc_func_desc);
 976 #endif
 977
 978         start_thread(regs, elf_entry, bprm->p);
 979         retval = 0;
 980 out:
 981         kfree(loc);
 982 out_ret:
 983         return retval;
 984
 985         /* error cleanup */
 986 out_free_dentry:
 987         allow_write_access(interpreter);
 988         if (interpreter)
 989                 fput(interpreter);
 990 out_free_interp:
 991         kfree(elf_interpreter);
 992 out_free_ph:
 993         kfree(elf_phdata);
 994         goto out;
 995 }
 996
 997 /* This is really simpleminded and specialized - we are loading an
 998    a.out library that is given an ELF header. */
 999 static int load_elf_library(struct file *file)
1000 {
1001         struct elf_phdr *elf_phdata;
1002         struct elf_phdr *eppnt;
1003         unsigned long elf_bss, bss, len;
1004         int retval, error, i, j;
1005         struct elfhdr elf_ex;
1006
1007         error = -ENOEXEC;
1008         retval = kernel_read(file, 0, (char *)&elf_ex, sizeof(elf_ex));
1009         if (retval != sizeof(elf_ex))
1010                 goto out;
1011
1012         if (memcmp(elf_ex.e_ident, ELFMAG, SELFMAG) != 0)
1013                 goto out;
1014
1015         /* First of all, some simple consistency checks */
1016         if (elf_ex.e_type != ET_EXEC || elf_ex.e_phnum > 2 ||
1017             !elf_check_arch(&elf_ex) || !file->f_op || !file->f_op->mmap)
1018                 goto out;
1019
1020         /* Now read in all of the header information */
1021
1022         j = sizeof(struct elf_phdr) * elf_ex.e_phnum;
1023         /* j < ELF_MIN_ALIGN because elf_ex.e_phnum <= 2 */
1024
1025         error = -ENOMEM;
1026         elf_phdata = kmalloc(j, GFP_KERNEL);
1027         if (!elf_phdata)
1028                 goto out;
1029
1030         eppnt = elf_phdata;
1031         error = -ENOEXEC;
1032         retval = kernel_read(file, elf_ex.e_phoff, (char *)eppnt, j);
1033         if (retval != j)
1034                 goto out_free_ph;
1035
1036         for (j = 0, i = 0; i<elf_ex.e_phnum; i++)
1037                 if ((eppnt + i)->p_type == PT_LOAD)
1038                         j++;
1039         if (j != 1)
1040                 goto out_free_ph;
1041
1042         while (eppnt->p_type != PT_LOAD)
1043                 eppnt++;
1044
1045         /* Now use mmap to map the library into memory. */
1046         down_write(&current->mm->mmap_sem);
1047         error = do_mmap(file,
1048                         ELF_PAGESTART(eppnt->p_vaddr),
1049                         (eppnt->p_filesz +
1050                          ELF_PAGEOFFSET(eppnt->p_vaddr)),
1051                         PROT_READ | PROT_WRITE | PROT_EXEC,
1052                         MAP_FIXED | MAP_PRIVATE | MAP_DENYWRITE,
1053                         (eppnt->p_offset -
1054                          ELF_PAGEOFFSET(eppnt->p_vaddr)));
1055         up_write(&current->mm->mmap_sem);
1056         if (error != ELF_PAGESTART(eppnt->p_vaddr))
1057                 goto out_free_ph;
1058
1059         elf_bss = eppnt->p_vaddr + eppnt->p_filesz;
1060         if (padzero(elf_bss)) {
1061                 error = -EFAULT;
1062                 goto out_free_ph;
1063         }
1064
1065         len = ELF_PAGESTART(eppnt->p_filesz + eppnt->p_vaddr +
1066                             ELF_MIN_ALIGN - 1);
1067         bss = eppnt->p_memsz + eppnt->p_vaddr;
1068         if (bss > len) {
1069                 down_write(&current->mm->mmap_sem);
1070                 do_brk(len, bss - len);
1071                 up_write(&current->mm->mmap_sem);
1072         }
1073         error = 0;
1074
1075 out_free_ph:
1076         kfree(elf_phdata);
1077 out:
1078         return error;
1079 }
1080
1081 /*
1082  * Note that some platforms still use traditional core dumps and not
1083  * the ELF core dump.  Each platform can select it as appropriate.
1084  */
1085 #if defined(USE_ELF_CORE_DUMP) && defined(CONFIG_ELF_CORE)
1086
1087 /*
1088  * ELF core dumper
1089  *
1090  * Modelled on fs/exec.c:aout_core_dump()
1091  * Jeremy Fitzhardinge <jeremy@sw.oz.au>
1092  */
1093 /*
1094  * These are the only things you should do on a core-file: use only these
1095  * functions to write out all the necessary info.
1096  */
1097 static int dump_write(struct file *file, const void *addr, int nr)
1098 {
1099         return file->f_op->write(file, addr, nr, &file->f_pos) == nr;
1100 }
1101
1102 static int dump_seek(struct file *file, loff_t off)
1103 {
1104         if (file->f_op->llseek && file->f_op->llseek != no_llseek) {
1105                 if (file->f_op->llseek(file, off, SEEK_CUR) < 0)
1106                         return 0;
1107         } else {
1108                 char *buf = (char *)get_zeroed_page(GFP_KERNEL);
1109                 if (!buf)
1110                         return 0;
1111                 while (off > 0) {
1112                         unsigned long n = off;
1113                         if (n > PAGE_SIZE)
1114                                 n = PAGE_SIZE;
1115                         if (!dump_write(file, buf, n))
1116                                 return 0;
1117                         off -= n;
1118                 }
1119                 free_page((unsigned long)buf);
1120         }
1121         return 1;
1122 }
1123
1124 /*
1125  * Decide what to dump of a segment, part, all or none.
1126  */
1127 static unsigned long vma_dump_size(struct vm_area_struct *vma,
1128                                    unsigned long mm_flags)
1129 {
1130 #define FILTER(type)    (mm_flags & (1UL << MMF_DUMP_##type))
1131
1132         /* The vma can be set up to tell us the answer directly.  */
1133         if (vma->vm_flags & VM_ALWAYSDUMP)
1134                 goto whole;
1135
1136         /* Hugetlb memory check */
1137         if (vma->vm_flags & VM_HUGETLB) {
1138                 if ((vma->vm_flags & VM_SHARED) && FILTER(HUGETLB_SHARED))
1139                         goto whole;
1140                 if (!(vma->vm_flags & VM_SHARED) && FILTER(HUGETLB_PRIVATE))
1141                         goto whole;
1142         }
1143
1144         /* Do not dump I/O mapped devices or special mappings */
1145         if (vma->vm_flags & (VM_IO | VM_RESERVED))
1146                 return 0;
1147
1148         /* By default, dump shared memory if mapped from an anonymous file. */
1149         if (vma->vm_flags & VM_SHARED) {
1150                 if (vma->vm_file->f_path.dentry->d_inode->i_nlink == 0 ?
1151                     FILTER(ANON_SHARED) : FILTER(MAPPED_SHARED))
1152                         goto whole;
1153                 return 0;
1154         }
1155
1156         /* Dump segments that have been written to.  */
1157         if (vma->anon_vma && FILTER(ANON_PRIVATE))
1158                 goto whole;
1159         if (vma->vm_file == NULL)
1160                 return 0;
1161
1162         if (FILTER(MAPPED_PRIVATE))
1163                 goto whole;
1164
1165         /*
1166          * If this looks like the beginning of a DSO or executable mapping,
1167          * check for an ELF header.  If we find one, dump the first page to
1168          * aid in determining what was mapped here.
1169          */
1170         if (FILTER(ELF_HEADERS) &&
1171             vma->vm_pgoff == 0 && (vma->vm_flags & VM_READ)) {
1172                 u32 __user *header = (u32 __user *) vma->vm_start;
1173                 u32 word;
1174                 mm_segment_t fs = get_fs();
1175                 /*
1176                  * Doing it this way gets the constant folded by GCC.
1177                  */
1178                 union {
1179                         u32 cmp;
1180                         char elfmag[SELFMAG];
1181                 } magic;
1182                 BUILD_BUG_ON(SELFMAG != sizeof word);
1183                 magic.elfmag[EI_MAG0] = ELFMAG0;
1184                 magic.elfmag[EI_MAG1] = ELFMAG1;
1185                 magic.elfmag[EI_MAG2] = ELFMAG2;
1186                 magic.elfmag[EI_MAG3] = ELFMAG3;
1187                 /*
1188                  * Switch to the user "segment" for get_user(),
1189                  * then put back what elf_core_dump() had in place.
1190                  */
1191                 set_fs(USER_DS);
1192                 if (unlikely(get_user(word, header)))
1193                         word = 0;
1194                 set_fs(fs);
1195                 if (word == magic.cmp)
1196                         return PAGE_SIZE;
1197         }
1198
1199 #undef  FILTER
1200
1201         return 0;
1202
1203 whole:
1204         return vma->vm_end - vma->vm_start;
1205 }
1206
1207 /* An ELF note in memory */
1208 struct memelfnote
1209 {
1210         const char *name;
1211         int type;
1212         unsigned int datasz;
1213         void *data;
1214 };
1215
1216 static int notesize(struct memelfnote *en)
1217 {
1218         int sz;
1219
1220         sz = sizeof(struct elf_note);
1221         sz += roundup(strlen(en->name) + 1, 4);
1222         sz += roundup(en->datasz, 4);
1223
1224         return sz;
1225 }
1226
1227 #define DUMP_WRITE(addr, nr, foffset)   \
1228         do { if (!dump_write(file, (addr), (nr))) return 0; *foffset += (nr); } while(0)
1229
1230 static int alignfile(struct file *file, loff_t *foffset)
1231 {
1232         static const char buf[4] = { 0, };
1233         DUMP_WRITE(buf, roundup(*foffset, 4) - *foffset, foffset);
1234         return 1;
1235 }
1236
1237 static int writenote(struct memelfnote *men, struct file *file,
1238                         loff_t *foffset)
1239 {
1240         struct elf_note en;
1241         en.n_namesz = strlen(men->name) + 1;
1242         en.n_descsz = men->datasz;
1243         en.n_type = men->type;
1244
1245         DUMP_WRITE(&en, sizeof(en), foffset);
1246         DUMP_WRITE(men->name, en.n_namesz, foffset);
1247         if (!alignfile(file, foffset))
1248                 return 0;
1249         DUMP_WRITE(men->data, men->datasz, foffset);
1250         if (!alignfile(file, foffset))
1251                 return 0;
1252
1253         return 1;
1254 }
1255 #undef DUMP_WRITE
1256
1257 #define DUMP_WRITE(addr, nr)    \
1258         if ((size += (nr)) > limit || !dump_write(file, (addr), (nr))) \
1259                 goto end_coredump;
1260
1261 static void fill_elf_header(struct elfhdr *elf, int segs,
1262                             u16 machine, u32 flags, u8 osabi)
1263 {
1264         memset(elf, 0, sizeof(*elf));
1265
1266         memcpy(elf->e_ident, ELFMAG, SELFMAG);
1267         elf->e_ident[EI_CLASS] = ELF_CLASS;
1268         elf->e_ident[EI_DATA] = ELF_DATA;
1269         elf->e_ident[EI_VERSION] = EV_CURRENT;
1270         elf->e_ident[EI_OSABI] = ELF_OSABI;
1271
1272         elf->e_type = ET_CORE;
1273         elf->e_machine = machine;
1274         elf->e_version = EV_CURRENT;
1275         elf->e_phoff = sizeof(struct elfhdr);
1276         elf->e_flags = flags;
1277         elf->e_ehsize = sizeof(struct elfhdr);
1278         elf->e_phentsize = sizeof(struct elf_phdr);
1279         elf->e_phnum = segs;
1280
1281         return;
1282 }
1283
1284 static void fill_elf_note_phdr(struct elf_phdr *phdr, int sz, loff_t offset)
1285 {
1286         phdr->p_type = PT_NOTE;
1287         phdr->p_offset = offset;
1288         phdr->p_vaddr = 0;
1289         phdr->p_paddr = 0;
1290         phdr->p_filesz = sz;
1291         phdr->p_memsz = 0;
1292         phdr->p_flags = 0;
1293         phdr->p_align = 0;
1294         return;
1295 }
1296
1297 static void fill_note(struct memelfnote *note, const char *name, int type,
1298                 unsigned int sz, void *data)
1299 {
1300         note->name = name;
1301         note->type = type;
1302         note->datasz = sz;
1303         note->data = data;
1304         return;
1305 }
1306
1307 /*
1308  * fill up all the fields in prstatus from the given task struct, except
1309  * registers which need to be filled up separately.
1310  */
1311 static void fill_prstatus(struct elf_prstatus *prstatus,
1312                 struct task_struct *p, long signr)
1313 {
1314         prstatus->pr_info.si_signo = prstatus->pr_cursig = signr;
1315         prstatus->pr_sigpend = p->pending.signal.sig[0];
1316         prstatus->pr_sighold = p->blocked.sig[0];
1317         rcu_read_lock();
1318         prstatus->pr_ppid = task_pid_vnr(rcu_dereference(p->real_parent));
1319         rcu_read_unlock();
1320         prstatus->pr_pid = task_pid_vnr(p);
1321         prstatus->pr_pgrp = task_pgrp_vnr(p);
1322         prstatus->pr_sid = task_session_vnr(p);
1323         if (thread_group_leader(p)) {
1324                 struct task_cputime cputime;
1325
1326                 /*
1327                  * This is the record for the group leader.  It shows the
1328                  * group-wide total, not its individual thread total.
1329                  */
1330                 thread_group_cputime(p, &cputime);
1331                 cputime_to_timeval(cputime.utime, &prstatus->pr_utime);
1332                 cputime_to_timeval(cputime.stime, &prstatus->pr_stime);
1333         } else {
1334                 cputime_to_timeval(p->utime, &prstatus->pr_utime);
1335                 cputime_to_timeval(p->stime, &prstatus->pr_stime);
1336         }
1337         cputime_to_timeval(p->signal->cutime, &prstatus->pr_cutime);
1338         cputime_to_timeval(p->signal->cstime, &prstatus->pr_cstime);
1339 }
1340
1341 static int fill_psinfo(struct elf_prpsinfo *psinfo, struct task_struct *p,
1342                        struct mm_struct *mm)
1343 {
1344         const struct cred *cred;
1345         unsigned int i, len;
1346
1347         /* first copy the parameters from user space */
1348         memset(psinfo, 0, sizeof(struct elf_prpsinfo));
1349
1350         len = mm->arg_end - mm->arg_start;
1351         if (len >= ELF_PRARGSZ)
1352                 len = ELF_PRARGSZ-1;
1353         if (copy_from_user(&psinfo->pr_psargs,
1354                            (const char __user *)mm->arg_start, len))
1355                 return -EFAULT;
1356         for(i = 0; i < len; i++)
1357                 if (psinfo->pr_psargs[i] == 0)
1358                         psinfo->pr_psargs[i] = ' ';
1359         psinfo->pr_psargs[len] = 0;
1360
1361         rcu_read_lock();
1362         psinfo->pr_ppid = task_pid_vnr(rcu_dereference(p->real_parent));
1363         rcu_read_unlock();
1364         psinfo->pr_pid = task_pid_vnr(p);
1365         psinfo->pr_pgrp = task_pgrp_vnr(p);
1366         psinfo->pr_sid = task_session_vnr(p);
1367
1368         i = p->state ? ffz(~p->state) + 1 : 0;
1369         psinfo->pr_state = i;
1370         psinfo->pr_sname = (i > 5) ? '.' : "RSDTZW"[i];
1371         psinfo->pr_zomb = psinfo->pr_sname == 'Z';
1372         psinfo->pr_nice = task_nice(p);
1373         psinfo->pr_flag = p->flags;
1374         rcu_read_lock();
1375         cred = __task_cred(p);
1376         SET_UID(psinfo->pr_uid, cred->uid);
1377         SET_GID(psinfo->pr_gid, cred->gid);
1378         rcu_read_unlock();
1379         strncpy(psinfo->pr_fname, p->comm, sizeof(psinfo->pr_fname));
1380
1381         return 0;
1382 }
1383
1384 static void fill_auxv_note(struct memelfnote *note, struct mm_struct *mm)
1385 {
1386         elf_addr_t *auxv = (elf_addr_t *) mm->saved_auxv;
1387         int i = 0;
1388         do
1389                 i += 2;
1390         while (auxv[i - 2] != AT_NULL);
1391         fill_note(note, "CORE", NT_AUXV, i * sizeof(elf_addr_t), auxv);
1392 }
1393
1394 #ifdef CORE_DUMP_USE_REGSET
1395 #include <linux/regset.h>
1396
1397 struct elf_thread_core_info {
1398         struct elf_thread_core_info *next;
1399         struct task_struct *task;
1400         struct elf_prstatus prstatus;
1401         struct memelfnote notes[0];
1402 };
1403
1404 struct elf_note_info {
1405         struct elf_thread_core_info *thread;
1406         struct memelfnote psinfo;
1407         struct memelfnote auxv;
1408         size_t size;
1409         int thread_notes;
1410 };
1411
1412 /*
1413  * When a regset has a writeback hook, we call it on each thread before
1414  * dumping user memory.  On register window machines, this makes sure the
1415  * user memory backing the register data is up to date before we read it.
1416  */
1417 static void do_thread_regset_writeback(struct task_struct *task,
1418                                        const struct user_regset *regset)
1419 {
1420         if (regset->writeback)
1421                 regset->writeback(task, regset, 1);
1422 }
1423
1424 static int fill_thread_core_info(struct elf_thread_core_info *t,
1425                                  const struct user_regset_view *view,
1426                                  long signr, size_t *total)
1427 {
1428         unsigned int i;
1429
1430         /*
1431          * NT_PRSTATUS is the one special case, because the regset data
1432          * goes into the pr_reg field inside the note contents, rather
1433          * than being the whole note contents.  We fill the reset in here.
1434          * We assume that regset 0 is NT_PRSTATUS.
1435          */
1436         fill_prstatus(&t->prstatus, t->task, signr);
1437         (void) view->regsets[0].get(t->task, &view->regsets[0],
1438                                     0, sizeof(t->prstatus.pr_reg),
1439                                     &t->prstatus.pr_reg, NULL);
1440
1441         fill_note(&t->notes[0], "CORE", NT_PRSTATUS,
1442                   sizeof(t->prstatus), &t->prstatus);
1443         *total += notesize(&t->notes[0]);
1444
1445         do_thread_regset_writeback(t->task, &view->regsets[0]);
1446
1447         /*
1448          * Each other regset might generate a note too.  For each regset
1449          * that has no core_note_type or is inactive, we leave t->notes[i]
1450          * all zero and we'll know to skip writing it later.
1451          */
1452         for (i = 1; i < view->n; ++i) {
1453                 const struct user_regset *regset = &view->regsets[i];
1454                 do_thread_regset_writeback(t->task, regset);
1455                 if (regset->core_note_type &&
1456                     (!regset->active || regset->active(t->task, regset))) {
1457                         int ret;
1458                         size_t size = regset->n * regset->size;
1459                         void *data = kmalloc(size, GFP_KERNEL);
1460                         if (unlikely(!data))
1461                                 return 0;
1462                         ret = regset->get(t->task, regset,
1463                                           0, size, data, NULL);
1464                         if (unlikely(ret))
1465                                 kfree(data);
1466                         else {
1467                                 if (regset->core_note_type != NT_PRFPREG)
1468                                         fill_note(&t->notes[i], "LINUX",
1469                                                   regset->core_note_type,
1470                                                   size, data);
1471                                 else {
1472                                         t->prstatus.pr_fpvalid = 1;
1473                                         fill_note(&t->notes[i], "CORE",
1474                                                   NT_PRFPREG, size, data);
1475                                 }
1476                                 *total += notesize(&t->notes[i]);
1477                         }
1478                 }
1479         }
1480
1481         return 1;
1482 }
1483
1484 static int fill_note_info(struct elfhdr *elf, int phdrs,
1485                           struct elf_note_info *info,
1486                           long signr, struct pt_regs *regs)
1487 {
1488         struct task_struct *dump_task = current;
1489         const struct user_regset_view *view = task_user_regset_view(dump_task);
1490         struct elf_thread_core_info *t;
1491         struct elf_prpsinfo *psinfo;
1492         struct core_thread *ct;
1493         unsigned int i;
1494
1495         info->size = 0;
1496         info->thread = NULL;
1497
1498         psinfo = kmalloc(sizeof(*psinfo), GFP_KERNEL);
1499         if (psinfo == NULL)
1500                 return 0;
1501
1502         fill_note(&info->psinfo, "CORE", NT_PRPSINFO, sizeof(*psinfo), psinfo);
1503
1504         /*
1505          * Figure out how many notes we're going to need for each thread.
1506          */
1507         info->thread_notes = 0;
1508         for (i = 0; i < view->n; ++i)
1509                 if (view->regsets[i].core_note_type != 0)
1510                         ++info->thread_notes;
1511
1512         /*
1513          * Sanity check.  We rely on regset 0 being in NT_PRSTATUS,
1514          * since it is our one special case.
1515          */
1516         if (unlikely(info->thread_notes == 0) ||
1517             unlikely(view->regsets[0].core_note_type != NT_PRSTATUS)) {
1518                 WARN_ON(1);
1519                 return 0;
1520         }
1521
1522         /*
1523          * Initialize the ELF file header.
1524          */
1525         fill_elf_header(elf, phdrs,
1526                         view->e_machine, view->e_flags, view->ei_osabi);
1527
1528         /*
1529          * Allocate a structure for each thread.
1530          */
1531         for (ct = &dump_task->mm->core_state->dumper; ct; ct = ct->next) {
1532                 t = kzalloc(offsetof(struct elf_thread_core_info,
1533                                      notes[info->thread_notes]),
1534                             GFP_KERNEL);
1535                 if (unlikely(!t))
1536                         return 0;
1537
1538                 t->task = ct->task;
1539                 if (ct->task == dump_task || !info->thread) {
1540                         t->next = info->thread;
1541                         info->thread = t;
1542                 } else {
1543                         /*
1544                          * Make sure to keep the original task at
1545                          * the head of the list.
1546                          */
1547                         t->next = info->thread->next;
1548                         info->thread->next = t;
1549                 }
1550         }
1551
1552         /*
1553          * Now fill in each thread's information.
1554          */
1555         for (t = info->thread; t != NULL; t = t->next)
1556                 if (!fill_thread_core_info(t, view, signr, &info->size))
1557                         return 0;
1558
1559         /*
1560          * Fill in the two process-wide notes.
1561          */
1562         fill_psinfo(psinfo, dump_task->group_leader, dump_task->mm);
1563         info->size += notesize(&info->psinfo);
1564
1565         fill_auxv_note(&info->auxv, current->mm);
1566         info->size += notesize(&info->auxv);
1567
1568         return 1;
1569 }
1570
1571 static size_t get_note_info_size(struct elf_note_info *info)
1572 {
1573         return info->size;
1574 }
1575
1576 /*
1577  * Write all the notes for each thread.  When writing the first thread, the
1578  * process-wide notes are interleaved after the first thread-specific note.
1579  */
1580 static int write_note_info(struct elf_note_info *info,
1581                            struct file *file, loff_t *foffset)
1582 {
1583         bool first = 1;
1584         struct elf_thread_core_info *t = info->thread;
1585
1586         do {
1587                 int i;
1588
1589                 if (!writenote(&t->notes[0], file, foffset))
1590                         return 0;
1591
1592                 if (first && !writenote(&info->psinfo, file, foffset))
1593                         return 0;
1594                 if (first && !writenote(&info->auxv, file, foffset))
1595                         return 0;
1596
1597                 for (i = 1; i < info->thread_notes; ++i)
1598                         if (t->notes[i].data &&
1599                             !writenote(&t->notes[i], file, foffset))
1600                                 return 0;
1601
1602                 first = 0;
1603                 t = t->next;
1604         } while (t);
1605
1606         return 1;
1607 }
1608
1609 static void free_note_info(struct elf_note_info *info)
1610 {
1611         struct elf_thread_core_info *threads = info->thread;
1612         while (threads) {
1613                 unsigned int i;
1614                 struct elf_thread_core_info *t = threads;
1615                 threads = t->next;
1616                 WARN_ON(t->notes[0].data && t->notes[0].data != &t->prstatus);
1617                 for (i = 1; i < info->thread_notes; ++i)
1618                         kfree(t->notes[i].data);
1619                 kfree(t);
1620         }
1621         kfree(info->psinfo.data);
1622 }
1623
1624 #else
1625
1626 /* Here is the structure in which status of each thread is captured. */
1627 struct elf_thread_status
1628 {
1629         struct list_head list;
1630         struct elf_prstatus prstatus;   /* NT_PRSTATUS */
1631         elf_fpregset_t fpu;             /* NT_PRFPREG */
1632         struct task_struct *thread;
1633 #ifdef ELF_CORE_COPY_XFPREGS
1634         elf_fpxregset_t xfpu;           /* ELF_CORE_XFPREG_TYPE */
1635 #endif
1636         struct memelfnote notes[3];
1637         int num_notes;
1638 };
1639
1640 /*
1641  * In order to add the specific thread information for the elf file format,
1642  * we need to keep a linked list of every threads pr_status and then create
1643  * a single section for them in the final core file.
1644  */
1645 static int elf_dump_thread_status(long signr, struct elf_thread_status *t)
1646 {
1647         int sz = 0;
1648         struct task_struct *p = t->thread;
1649         t->num_notes = 0;
1650
1651         fill_prstatus(&t->prstatus, p, signr);
1652         elf_core_copy_task_regs(p, &t->prstatus.pr_reg);
1653
1654         fill_note(&t->notes[0], "CORE", NT_PRSTATUS, sizeof(t->prstatus),
1655                   &(t->prstatus));
1656         t->num_notes++;
1657         sz += notesize(&t->notes[0]);
1658
1659         if ((t->prstatus.pr_fpvalid = elf_core_copy_task_fpregs(p, NULL,
1660                                                                 &t->fpu))) {
1661                 fill_note(&t->notes[1], "CORE", NT_PRFPREG, sizeof(t->fpu),
1662                           &(t->fpu));
1663                 t->num_notes++;
1664                 sz += notesize(&t->notes[1]);
1665         }
1666
1667 #ifdef ELF_CORE_COPY_XFPREGS
1668         if (elf_core_copy_task_xfpregs(p, &t->xfpu)) {
1669                 fill_note(&t->notes[2], "LINUX", ELF_CORE_XFPREG_TYPE,
1670                           sizeof(t->xfpu), &t->xfpu);
1671                 t->num_notes++;
1672                 sz += notesize(&t->notes[2]);
1673         }
1674 #endif
1675         return sz;
1676 }
1677
1678 struct elf_note_info {
1679         struct memelfnote *notes;
1680         struct elf_prstatus *prstatus;  /* NT_PRSTATUS */
1681         struct elf_prpsinfo *psinfo;    /* NT_PRPSINFO */
1682         struct list_head thread_list;
1683         elf_fpregset_t *fpu;
1684 #ifdef ELF_CORE_COPY_XFPREGS
1685         elf_fpxregset_t *xfpu;
1686 #endif
1687         int thread_status_size;
1688         int numnote;
1689 };
1690
1691 static int elf_note_info_init(struct elf_note_info *info)
1692 {
1693         memset(info, 0, sizeof(*info));
1694         INIT_LIST_HEAD(&info->thread_list);
1695
1696         /* Allocate space for six ELF notes */
1697         info->notes = kmalloc(6 * sizeof(struct memelfnote), GFP_KERNEL);
1698         if (!info->notes)
1699                 return 0;
1700         info->psinfo = kmalloc(sizeof(*info->psinfo), GFP_KERNEL);
1701         if (!info->psinfo)
1702                 goto notes_free;
1703         info->prstatus = kmalloc(sizeof(*info->prstatus), GFP_KERNEL);
1704         if (!info->prstatus)
1705                 goto psinfo_free;
1706         info->fpu = kmalloc(sizeof(*info->fpu), GFP_KERNEL);
1707         if (!info->fpu)
1708                 goto prstatus_free;
1709 #ifdef ELF_CORE_COPY_XFPREGS
1710         info->xfpu = kmalloc(sizeof(*info->xfpu), GFP_KERNEL);
1711         if (!info->xfpu)
1712                 goto fpu_free;
1713 #endif
1714         return 1;
1715 #ifdef ELF_CORE_COPY_XFPREGS
1716  fpu_free:
1717         kfree(info->fpu);
1718 #endif
1719  prstatus_free:
1720         kfree(info->prstatus);
1721  psinfo_free:
1722         kfree(info->psinfo);
1723  notes_free:
1724         kfree(info->notes);
1725         return 0;
1726 }
1727
1728 static int fill_note_info(struct elfhdr *elf, int phdrs,
1729                           struct elf_note_info *info,
1730                           long signr, struct pt_regs *regs)
1731 {
1732         struct list_head *t;
1733
1734         if (!elf_note_info_init(info))
1735                 return 0;
1736
1737         if (signr) {
1738                 struct core_thread *ct;
1739                 struct elf_thread_status *ets;
1740
1741                 for (ct = current->mm->core_state->dumper.next;
1742                                                 ct; ct = ct->next) {
1743                         ets = kzalloc(sizeof(*ets), GFP_KERNEL);
1744                         if (!ets)
1745                                 return 0;
1746
1747                         ets->thread = ct->task;
1748                         list_add(&ets->list, &info->thread_list);
1749                 }
1750
1751                 list_for_each(t, &info->thread_list) {
1752                         int sz;
1753
1754                         ets = list_entry(t, struct elf_thread_status, list);
1755                         sz = elf_dump_thread_status(signr, ets);
1756                         info->thread_status_size += sz;
1757                 }
1758         }
1759         /* now collect the dump for the current */
1760         memset(info->prstatus, 0, sizeof(*info->prstatus));
1761         fill_prstatus(info->prstatus, current, signr);
1762         elf_core_copy_regs(&info->prstatus->pr_reg, regs);
1763
1764         /* Set up header */
1765         fill_elf_header(elf, phdrs, ELF_ARCH, ELF_CORE_EFLAGS, ELF_OSABI);
1766
1767         /*
1768          * Set up the notes in similar form to SVR4 core dumps made
1769          * with info from their /proc.
1770          */
1771
1772         fill_note(info->notes + 0, "CORE", NT_PRSTATUS,
1773                   sizeof(*info->prstatus), info->prstatus);
1774         fill_psinfo(info->psinfo, current->group_leader, current->mm);
1775         fill_note(info->notes + 1, "CORE", NT_PRPSINFO,
1776                   sizeof(*info->psinfo), info->psinfo);
1777
1778         info->numnote = 2;
1779
1780         fill_auxv_note(&info->notes[info->numnote++], current->mm);
1781
1782         /* Try to dump the FPU. */
1783         info->prstatus->pr_fpvalid = elf_core_copy_task_fpregs(current, regs,
1784                                                                info->fpu);
1785         if (info->prstatus->pr_fpvalid)
1786                 fill_note(info->notes + info->numnote++,
1787                           "CORE", NT_PRFPREG, sizeof(*info->fpu), info->fpu);
1788 #ifdef ELF_CORE_COPY_XFPREGS
1789         if (elf_core_copy_task_xfpregs(current, info->xfpu))
1790                 fill_note(info->notes + info->numnote++,
1791                           "LINUX", ELF_CORE_XFPREG_TYPE,
1792                           sizeof(*info->xfpu), info->xfpu);
1793 #endif
1794
1795         return 1;
1796 }
1797
1798 static size_t get_note_info_size(struct elf_note_info *info)
1799 {
1800         int sz = 0;
1801         int i;
1802
1803         for (i = 0; i < info->numnote; i++)
1804                 sz += notesize(info->notes + i);
1805
1806         sz += info->thread_status_size;
1807
1808         return sz;
1809 }
1810
1811 static int write_note_info(struct elf_note_info *info,
1812                            struct file *file, loff_t *foffset)
1813 {
1814         int i;
1815         struct list_head *t;
1816
1817         for (i = 0; i < info->numnote; i++)
1818                 if (!writenote(info->notes + i, file, foffset))
1819                         return 0;
1820
1821         /* write out the thread status notes section */
1822         list_for_each(t, &info->thread_list) {
1823                 struct elf_thread_status *tmp =
1824                                 list_entry(t, struct elf_thread_status, list);
1825
1826                 for (i = 0; i < tmp->num_notes; i++)
1827                         if (!writenote(&tmp->notes[i], file, foffset))
1828                                 return 0;
1829         }
1830
1831         return 1;
1832 }
1833
1834 static void free_note_info(struct elf_note_info *info)
1835 {
1836         while (!list_empty(&info->thread_list)) {
1837                 struct list_head *tmp = info->thread_list.next;
1838                 list_del(tmp);
1839                 kfree(list_entry(tmp, struct elf_thread_status, list));
1840         }
1841
1842         kfree(info->prstatus);
1843         kfree(info->psinfo);
1844         kfree(info->notes);
1845         kfree(info->fpu);
1846 #ifdef ELF_CORE_COPY_XFPREGS
1847         kfree(info->xfpu);
1848 #endif
1849 }
1850
1851 #endif
1852
1853 static struct vm_area_struct *first_vma(struct task_struct *tsk,
1854                                         struct vm_area_struct *gate_vma)
1855 {
1856         struct vm_area_struct *ret = tsk->mm->mmap;
1857
1858         if (ret)
1859                 return ret;
1860         return gate_vma;
1861 }
1862 /*
1863  * Helper function for iterating across a vma list.  It ensures that the caller
1864  * will visit `gate_vma' prior to terminating the search.
1865  */
1866 static struct vm_area_struct *next_vma(struct vm_area_struct *this_vma,
1867                                         struct vm_area_struct *gate_vma)
1868 {
1869         struct vm_area_struct *ret;
1870
1871         ret = this_vma->vm_next;
1872         if (ret)
1873                 return ret;
1874         if (this_vma == gate_vma)
1875                 return NULL;
1876         return gate_vma;
1877 }
1878
1879 /*
1880  * Actual dumper
1881  *
1882  * This is a two-pass process; first we find the offsets of the bits,
1883  * and then they are actually written out.  If we run out of core limit
1884  * we just truncate.
1885  */
1886 static int elf_core_dump(long signr, struct pt_regs *regs, struct file *file, unsigned long limit)
1887 {
1888         int has_dumped = 0;
1889         mm_segment_t fs;
1890         int segs;
1891         size_t size = 0;
1892         struct vm_area_struct *vma, *gate_vma;
1893         struct elfhdr *elf = NULL;
1894         loff_t offset = 0, dataoff, foffset;
1895         unsigned long mm_flags;
1896         struct elf_note_info info;
1897
1898         /*
1899          * We no longer stop all VM operations.
1900          *
1901          * This is because those proceses that could possibly change map_count
1902          * or the mmap / vma pages are now blocked in do_exit on current
1903          * finishing this core dump.
1904          *
1905          * Only ptrace can touch these memory addresses, but it doesn't change
1906          * the map_count or the pages allocated. So no possibility of crashing
1907          * exists while dumping the mm->vm_next areas to the core file.
1908          */
1909
1910         /* alloc memory for large data structures: too large to be on stack */
1911         elf = kmalloc(sizeof(*elf), GFP_KERNEL);
1912         if (!elf)
1913                 goto out;
1914         /*
1915          * The number of segs are recored into ELF header as 16bit value.
1916          * Please check DEFAULT_MAX_MAP_COUNT definition when you modify here.
1917          */
1918         segs = current->mm->map_count;
1919 #ifdef ELF_CORE_EXTRA_PHDRS
1920         segs += ELF_CORE_EXTRA_PHDRS;
1921 #endif
1922
1923         gate_vma = get_gate_vma(current);
1924         if (gate_vma != NULL)
1925                 segs++;
1926
1927         /*
1928          * Collect all the non-memory information about the process for the
1929          * notes.  This also sets up the file header.
1930          */
1931         if (!fill_note_info(elf, segs + 1, /* including notes section */
1932                             &info, signr, regs))
1933                 goto cleanup;
1934
1935         has_dumped = 1;
1936         current->flags |= PF_DUMPCORE;
1937
1938         fs = get_fs();
1939         set_fs(KERNEL_DS);
1940
1941         DUMP_WRITE(elf, sizeof(*elf));
1942         offset += sizeof(*elf);                         /* Elf header */
1943         offset += (segs + 1) * sizeof(struct elf_phdr); /* Program headers */
1944         foffset = offset;
1945
1946         /* Write notes phdr entry */
1947         {
1948                 struct elf_phdr phdr;
1949                 size_t sz = get_note_info_size(&info);
1950
1951                 sz += elf_coredump_extra_notes_size();
1952
1953                 fill_elf_note_phdr(&phdr, sz, offset);
1954                 offset += sz;
1955                 DUMP_WRITE(&phdr, sizeof(phdr));
1956         }
1957
1958         dataoff = offset = roundup(offset, ELF_EXEC_PAGESIZE);
1959
1960         /*
1961          * We must use the same mm->flags while dumping core to avoid
1962          * inconsistency between the program headers and bodies, otherwise an
1963          * unusable core file can be generated.
1964          */
1965         mm_flags = current->mm->flags;
1966
1967         /* Write program headers for segments dump */
1968         for (vma = first_vma(current, gate_vma); vma != NULL;
1969                         vma = next_vma(vma, gate_vma)) {
1970                 struct elf_phdr phdr;
1971
1972                 phdr.p_type = PT_LOAD;
1973                 phdr.p_offset = offset;
1974                 phdr.p_vaddr = vma->vm_start;
1975                 phdr.p_paddr = 0;
1976                 phdr.p_filesz = vma_dump_size(vma, mm_flags);
1977                 phdr.p_memsz = vma->vm_end - vma->vm_start;
1978                 offset += phdr.p_filesz;
1979                 phdr.p_flags = vma->vm_flags & VM_READ ? PF_R : 0;
1980                 if (vma->vm_flags & VM_WRITE)
1981                         phdr.p_flags |= PF_W;
1982                 if (vma->vm_flags & VM_EXEC)
1983                         phdr.p_flags |= PF_X;
1984                 phdr.p_align = ELF_EXEC_PAGESIZE;
1985
1986                 DUMP_WRITE(&phdr, sizeof(phdr));
1987         }
1988
1989 #ifdef ELF_CORE_WRITE_EXTRA_PHDRS
1990         ELF_CORE_WRITE_EXTRA_PHDRS;
1991 #endif
1992
1993         /* write out the notes section */
1994         if (!write_note_info(&info, file, &foffset))
1995                 goto end_coredump;
1996
1997         if (elf_coredump_extra_notes_write(file, &foffset))
1998                 goto end_coredump;
1999
2000         /* Align to page */
2001         if (!dump_seek(file, dataoff - foffset))
2002                 goto end_coredump;
2003
2004         for (vma = first_vma(current, gate_vma); vma != NULL;
2005                         vma = next_vma(vma, gate_vma)) {
2006                 unsigned long addr;
2007                 unsigned long end;
2008
2009                 end = vma->vm_start + vma_dump_size(vma, mm_flags);
2010
2011                 for (addr = vma->vm_start; addr < end; addr += PAGE_SIZE) {
2012                         struct page *page;
2013                         int stop;
2014
2015                         page = get_dump_page(addr);
2016                         if (page) {
2017                                 void *kaddr = kmap(page);
2018                                 stop = ((size += PAGE_SIZE) > limit) ||
2019                                         !dump_write(file, kaddr, PAGE_SIZE);
2020                                 kunmap(page);
2021                                 page_cache_release(page);
2022                         } else
2023                                 stop = !dump_seek(file, PAGE_SIZE);
2024                         if (stop)
2025                                 goto end_coredump;
2026                 }
2027         }
2028
2029 #ifdef ELF_CORE_WRITE_EXTRA_DATA
2030         ELF_CORE_WRITE_EXTRA_DATA;
2031 #endif
2032
2033 end_coredump:
2034         set_fs(fs);
2035
2036 cleanup:
2037         free_note_info(&info);
2038         kfree(elf);
2039 out:
2040         return has_dumped;
2041 }
2042
2043 #endif          /* USE_ELF_CORE_DUMP */
2044
2045 static int __init init_elf_binfmt(void)
2046 {
2047         return register_binfmt(&elf_format);
2048 }
2049
2050 static void __exit exit_elf_binfmt(void)
2051 {
2052         /* Remove the COFF and ELF loaders. */
2053         unregister_binfmt(&elf_format);
2054 }
2055
2056 core_initcall(init_elf_binfmt);
2057 module_exit(exit_elf_binfmt);
2058 MODULE_LICENSE("GPL");