fs/proc/array.c

/* */
This source file includes following definitions.
read_core
read_profile
write_profile
get_loadavg
get_kstat
get_uptime
get_meminfo
get_version
get_cmdline
get_task
get_phys_addr
get_array
get_env
get_arg
get_wchan
task_name
task_state
task_mem
task_sig
get_status
get_stat
statm_pte_range
statm_pmd_range
statm_pgd_range
get_statm
read_maps
get_root_array
get_process_array
fill_array
array_read
arraylong_read
   1 /*
   2  *  linux/fs/proc/array.c
   3  *
   4  *  Copyright (C) 1992  by Linus Torvalds
   5  *  based on ideas by Darren Senn
   6  *
   7  * Fixes:
   8  * Michael. K. Johnson: stat,statm extensions.
   9  *                      <johnsonm@stolaf.edu>
  10  *
  11  * Pauline Middelink :  Made cmdline,envline only break at '\0's, to
  12  *                      make sure SET_PROCTITLE works. Also removed
  13  *                      bad '!' which forced address recalculation for
  14  *                      EVERY character on the current page.
  15  *                      <middelin@polyware.iaf.nl>
  16  *
  17  * Danny ter Haar    :  added cpuinfo 
  18  *                      <dth@cistron.nl>
  19  *
  20  * Alessandro Rubini :  profile extension.
  21  *                      <rubini@ipvvis.unipv.it>
  22  *
  23  * Jeff Tranter      :  added BogoMips field to cpuinfo
  24  *                      <Jeff_Tranter@Mitel.COM>
  25  *
  26  * Bruno Haible      :  remove 4K limit for the maps file
  27  * <haible@ma2s2.mathematik.uni-karlsruhe.de>
  28  */
  29 
  30 #include <linux/types.h>
  31 #include <linux/errno.h>
  32 #include <linux/sched.h>
  33 #include <linux/kernel.h>
  34 #include <linux/kernel_stat.h>
  35 #include <linux/tty.h>
  36 #include <linux/user.h>
  37 #include <linux/a.out.h>
  38 #include <linux/string.h>
  39 #include <linux/mman.h>
  40 #include <linux/proc_fs.h>
  41 #include <linux/ioport.h>
  42 #include <linux/config.h>
  43 #include <linux/mm.h>
  44 #include <linux/pagemap.h>
  45 #ifdef CONFIG_APM
  46 #include <linux/apm_bios.h>
  47 #endif
  48 #include <linux/swap.h>
  49 
  50 #include <asm/segment.h>
  51 #include <asm/pgtable.h>
  52 #include <asm/io.h>
  53 
  54 #define LOAD_INT(x) ((x) >> FSHIFT)
  55 #define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1-1)) * 100)
  56 
  57 #ifdef CONFIG_DEBUG_MALLOC
  58 int get_malloc(char * buffer);
  59 #endif
  60 
  61 
  62 static int read_core(struct inode * inode, struct file * file,char * buf, int count)
     /*  */
  63 {
  64         unsigned long p = file->f_pos, memsize;
  65         int read;
  66         int count1;
  67         char * pnt;
  68         struct user dump;
  69 #ifdef __i386__
  70 #       define FIRST_MAPPED     PAGE_SIZE       /* we don't have page 0 mapped on x86.. */
  71 #else
  72 #       define FIRST_MAPPED     0
  73 #endif
  74 
  75         memset(&dump, 0, sizeof(struct user));
  76         dump.magic = CMAGIC;
  77         dump.u_dsize = MAP_NR(high_memory);
  78 #ifdef __alpha__
  79         dump.start_data = PAGE_OFFSET;
  80 #endif
  81 
  82         if (count < 0)
  83                 return -EINVAL;
  84         memsize = MAP_NR(high_memory + PAGE_SIZE) << PAGE_SHIFT;
  85         if (p >= memsize)
  86                 return 0;
  87         if (count > memsize - p)
  88                 count = memsize - p;
  89         read = 0;
  90 
  91         if (p < sizeof(struct user) && count > 0) {
  92                 count1 = count;
  93                 if (p + count1 > sizeof(struct user))
  94                         count1 = sizeof(struct user)-p;
  95                 pnt = (char *) &dump + p;
  96                 memcpy_tofs(buf,(void *) pnt, count1);
  97                 buf += count1;
  98                 p += count1;
  99                 count -= count1;
 100                 read += count1;
 101         }
 102 
 103         while (count > 0 && p < PAGE_SIZE + FIRST_MAPPED) {
 104                 put_user(0,buf);
 105                 buf++;
 106                 p++;
 107                 count--;
 108                 read++;
 109         }
 110         memcpy_tofs(buf, (void *) (PAGE_OFFSET + p - PAGE_SIZE), count);
 111         read += count;
 112         file->f_pos += read;
 113         return read;
 114 }
 115 
 116 static struct file_operations proc_kcore_operations = {
 117         NULL,           /* lseek */
 118         read_core,
 119 };
 120 
 121 struct inode_operations proc_kcore_inode_operations = {
 122         &proc_kcore_operations, 
 123 };
 124 
 125 
 126 /*
 127  * This function accesses profiling information. The returned data is
 128  * binary: the sampling step and the actual contents of the profile
 129  * buffer. Use of the program readprofile is recommended in order to
 130  * get meaningful info out of these data.
 131  */
 132 static int read_profile(struct inode *inode, struct file *file, char *buf, int count)
     /*  */
 133 {
 134         unsigned long p = file->f_pos;
 135         int read;
 136         char * pnt;
 137         unsigned int sample_step = 1 << prof_shift;
 138 
 139         if (count < 0)
 140                 return -EINVAL;
 141         if (p >= (prof_len+1)*sizeof(unsigned int))
 142                 return 0;
 143         if (count > (prof_len+1)*sizeof(unsigned int) - p)
 144                 count = (prof_len+1)*sizeof(unsigned int) - p;
 145         read = 0;
 146 
 147         while (p < sizeof(unsigned int) && count > 0) {
 148                 put_user(*((char *)(&sample_step)+p),buf);
 149                 buf++; p++; count--; read++;
 150         }
 151         pnt = (char *)prof_buffer + p - sizeof(unsigned int);
 152         memcpy_tofs(buf,(void *)pnt,count);
 153         read += count;
 154         file->f_pos += read;
 155         return read;
 156 }
 157 
 158 /* Writing to /proc/profile resets the counters */
 159 static int write_profile(struct inode * inode, struct file * file, const char * buf, int count)
     /*  */
 160 {
 161     int i=prof_len;
 162 
 163     while (i--)
 164             prof_buffer[i]=0UL;
 165     return count;
 166 }
 167 
 168 static struct file_operations proc_profile_operations = {
 169         NULL,           /* lseek */
 170         read_profile,
 171         write_profile,
 172 };
 173 
 174 struct inode_operations proc_profile_inode_operations = {
 175         &proc_profile_operations, 
 176 };
 177 
 178 
 179 static int get_loadavg(char * buffer)
     /*  */
 180 {
 181         int a, b, c;
 182 
 183         a = avenrun[0] + (FIXED_1/200);
 184         b = avenrun[1] + (FIXED_1/200);
 185         c = avenrun[2] + (FIXED_1/200);
 186         return sprintf(buffer,"%d.%02d %d.%02d %d.%02d %d/%d\n",
 187                 LOAD_INT(a), LOAD_FRAC(a),
 188                 LOAD_INT(b), LOAD_FRAC(b),
 189                 LOAD_INT(c), LOAD_FRAC(c),
 190                 nr_running, nr_tasks);
 191 }
 192 
 193 static int get_kstat(char * buffer)
     /*  */
 194 {
 195         int i, len;
 196         unsigned sum = 0;
 197 
 198         for (i = 0 ; i < NR_IRQS ; i++)
 199                 sum += kstat.interrupts[i];
 200         len = sprintf(buffer,
 201                 "cpu  %u %u %u %lu\n"
 202                 "disk %u %u %u %u\n"
 203                 "disk_rio %u %u %u %u\n"
 204                 "disk_wio %u %u %u %u\n"
 205                 "disk_rblk %u %u %u %u\n"
 206                 "disk_wblk %u %u %u %u\n"
 207                 "page %u %u\n"
 208                 "swap %u %u\n"
 209                 "intr %u",
 210                 kstat.cpu_user,
 211                 kstat.cpu_nice,
 212                 kstat.cpu_system,
 213                 jiffies - (kstat.cpu_user + kstat.cpu_nice + kstat.cpu_system),
 214                 kstat.dk_drive[0], kstat.dk_drive[1],
 215                 kstat.dk_drive[2], kstat.dk_drive[3],
 216                 kstat.dk_drive_rio[0], kstat.dk_drive_rio[1],
 217                 kstat.dk_drive_rio[2], kstat.dk_drive_rio[3],
 218                 kstat.dk_drive_wio[0], kstat.dk_drive_wio[1],
 219                 kstat.dk_drive_wio[2], kstat.dk_drive_wio[3],
 220                 kstat.dk_drive_rblk[0], kstat.dk_drive_rblk[1],
 221                 kstat.dk_drive_rblk[2], kstat.dk_drive_rblk[3],
 222                 kstat.dk_drive_wblk[0], kstat.dk_drive_wblk[1],
 223                 kstat.dk_drive_wblk[2], kstat.dk_drive_wblk[3],
 224                 kstat.pgpgin,
 225                 kstat.pgpgout,
 226                 kstat.pswpin,
 227                 kstat.pswpout,
 228                 sum);
 229         for (i = 0 ; i < NR_IRQS ; i++)
 230                 len += sprintf(buffer + len, " %u", kstat.interrupts[i]);
 231         len += sprintf(buffer + len,
 232                 "\nctxt %u\n"
 233                 "btime %lu\n",
 234                 kstat.context_swtch,
 235                 xtime.tv_sec - jiffies / HZ);
 236         return len;
 237 }
 238 
 239 
 240 static int get_uptime(char * buffer)
     /*  */
 241 {
 242         unsigned long uptime;
 243         unsigned long idle;
 244 
 245         uptime = jiffies;
 246         idle = task[0]->utime + task[0]->stime;
 247 
 248         /* The formula for the fraction parts really is ((t * 100) / HZ) % 100, but
 249            that would overflow about every five days at HZ == 100.
 250            Therefore the identity a = (a / b) * b + a % b is used so that it is
 251            calculated as (((t / HZ) * 100) + ((t % HZ) * 100) / HZ) % 100.
 252            The part in front of the '+' always evaluates as 0 (mod 100). All divisions
 253            in the above formulas are truncating. For HZ being a power of 10, the
 254            calculations simplify to the version in the #else part (if the printf
 255            format is adapted to the same number of digits as zeroes in HZ.
 256          */
 257 #if HZ!=100
 258         return sprintf(buffer,"%lu.%02lu %lu.%02lu\n",
 259                 uptime / HZ,
 260                 (((uptime % HZ) * 100) / HZ) % 100,
 261                 idle / HZ,
 262                 (((idle % HZ) * 100) / HZ) % 100);
 263 #else
 264         return sprintf(buffer,"%lu.%02lu %lu.%02lu\n",
 265                 uptime / HZ,
 266                 uptime % HZ,
 267                 idle / HZ,
 268                 idle % HZ);
 269 #endif
 270 }
 271 
 272 static int get_meminfo(char * buffer)
     /*  */
 273 {
 274         struct sysinfo i;
 275         int len;
 276 
 277         si_meminfo(&i);
 278         si_swapinfo(&i);
 279         len = sprintf(buffer, "        total:    used:    free:  shared: buffers:  cached:\n"
 280                 "Mem:  %8lu %8lu %8lu %8lu %8lu %8lu\n"
 281                 "Swap: %8lu %8lu %8lu\n",
 282                 i.totalram, i.totalram-i.freeram, i.freeram, i.sharedram, i.bufferram, page_cache_size*PAGE_SIZE,
 283                 i.totalswap, i.totalswap-i.freeswap, i.freeswap);
 284         /*
 285          * Tagged format, for easy grepping and expansion. The above will go away
 286          * eventually, once the tools have been updated.
 287          */
 288         return len + sprintf(buffer+len,
 289                 "MemTotal:  %8lu kB\n"
 290                 "MemFree:   %8lu kB\n"
 291                 "MemShared: %8lu kB\n"
 292                 "Buffers:   %8lu kB\n"
 293                 "Cached:    %8lu kB\n"
 294                 "SwapTotal: %8lu kB\n"
 295                 "SwapFree:  %8lu kB\n",
 296                 i.totalram >> 10,
 297                 i.freeram >> 10,
 298                 i.sharedram >> 10,
 299                 i.bufferram >> 10,
 300                 page_cache_size << (PAGE_SHIFT - 10),
 301                 i.totalswap >> 10,
 302                 i.freeswap >> 10);
 303 }
 304 
 305 static int get_version(char * buffer)
     /*  */
 306 {
 307         extern char *linux_banner;
 308 
 309         strcpy(buffer, linux_banner);
 310         return strlen(buffer);
 311 }
 312 
 313 static int get_cmdline(char * buffer)
     /*  */
 314 {
 315         extern char saved_command_line[];
 316 
 317         return sprintf(buffer, "%s\n", saved_command_line);
 318 }
 319 
 320 static struct task_struct ** get_task(pid_t pid)
     /*  */
 321 {
 322         struct task_struct ** p;
 323 
 324         p = task;
 325         while (++p < task+NR_TASKS) {
 326                 if (*p && (*p)->pid == pid)
 327                         return p;
 328         }
 329         return NULL;
 330 }
 331 
 332 static unsigned long get_phys_addr(struct task_struct * p, unsigned long ptr)
     /*  */
 333 {
 334         pgd_t *page_dir;
 335         pmd_t *page_middle;
 336         pte_t pte;
 337 
 338         if (!p || !p->mm || ptr >= TASK_SIZE)
 339                 return 0;
 340         page_dir = pgd_offset(p->mm,ptr);
 341         if (pgd_none(*page_dir))
 342                 return 0;
 343         if (pgd_bad(*page_dir)) {
 344                 printk("bad page directory entry %08lx\n", pgd_val(*page_dir));
 345                 pgd_clear(page_dir);
 346                 return 0;
 347         }
 348         page_middle = pmd_offset(page_dir,ptr);
 349         if (pmd_none(*page_middle))
 350                 return 0;
 351         if (pmd_bad(*page_middle)) {
 352                 printk("bad page middle entry %08lx\n", pmd_val(*page_middle));
 353                 pmd_clear(page_middle);
 354                 return 0;
 355         }
 356         pte = *pte_offset(page_middle,ptr);
 357         if (!pte_present(pte))
 358                 return 0;
 359         return pte_page(pte) + (ptr & ~PAGE_MASK);
 360 }
 361 
 362 static int get_array(struct task_struct ** p, unsigned long start, unsigned long end, char * buffer)
     /*  */
 363 {
 364         unsigned long addr;
 365         int size = 0, result = 0;
 366         char c;
 367 
 368         if (start >= end)
 369                 return result;
 370         for (;;) {
 371                 addr = get_phys_addr(*p, start);
 372                 if (!addr)
 373                         goto ready;
 374                 do {
 375                         c = *(char *) addr;
 376                         if (!c)
 377                                 result = size;
 378                         if (size < PAGE_SIZE)
 379                                 buffer[size++] = c;
 380                         else
 381                                 goto ready;
 382                         addr++;
 383                         start++;
 384                         if (!c && start >= end)
 385                                 goto ready;
 386                 } while (addr & ~PAGE_MASK);
 387         }
 388 ready:
 389         /* remove the trailing blanks, used to fill out argv,envp space */
 390         while (result>0 && buffer[result-1]==' ')
 391                 result--;
 392         return result;
 393 }
 394 
 395 static int get_env(int pid, char * buffer)
     /*  */
 396 {
 397         struct task_struct ** p = get_task(pid);
 398 
 399         if (!p || !*p || !(*p)->mm)
 400                 return 0;
 401         return get_array(p, (*p)->mm->env_start, (*p)->mm->env_end, buffer);
 402 }
 403 
 404 static int get_arg(int pid, char * buffer)
     /*  */
 405 {
 406         struct task_struct ** p = get_task(pid);
 407 
 408         if (!p || !*p || !(*p)->mm)
 409                 return 0;
 410         return get_array(p, (*p)->mm->arg_start, (*p)->mm->arg_end, buffer);
 411 }
 412 
 413 static unsigned long get_wchan(struct task_struct *p)
     /*  */
 414 {
 415         if (!p || p == current || p->state == TASK_RUNNING)
 416                 return 0;
 417 #if defined(__i386__)
 418         {
 419                 unsigned long ebp, eip;
 420                 unsigned long stack_page;
 421                 int count = 0;
 422 
 423                 stack_page = p->kernel_stack_page;
 424                 if (!stack_page)
 425                         return 0;
 426                 ebp = p->tss.ebp;
 427                 do {
 428                         if (ebp < stack_page || ebp >= 4092+stack_page)
 429                                 return 0;
 430                         eip = *(unsigned long *) (ebp+4);
 431                         if ((void *)eip != sleep_on &&
 432                             (void *)eip != interruptible_sleep_on)
 433                                 return eip;
 434                         ebp = *(unsigned long *) ebp;
 435                 } while (count++ < 16);
 436         }
 437 #elif defined(__alpha__)
 438         /*
 439          * This one depends on the frame size of schedule().  Do a
 440          * "disass schedule" in gdb to find the frame size.  Also, the
 441          * code assumes that sleep_on() follows immediately after
 442          * interruptible_sleep_on() and that add_timer() follows
 443          * immediately after interruptible_sleep().  Ugly, isn't it?
 444          * Maybe adding a wchan field to task_struct would be better,
 445          * after all...
 446          */
 447         {
 448             unsigned long schedule_frame;
 449             unsigned long pc;
 450 
 451             pc = thread_saved_pc(&p->tss);
 452             if (pc >= (unsigned long) interruptible_sleep_on && pc < (unsigned long) add_timer) {
 453                 schedule_frame = ((unsigned long *)p->tss.ksp)[6];
 454                 return ((unsigned long *)schedule_frame)[12];
 455             }
 456             return pc;
 457         }
 458 #endif
 459         return 0;
 460 }
 461 
 462 #if defined(__i386__)
 463 # define KSTK_EIP(tsk)  (((unsigned long *)tsk->kernel_stack_page)[1019])
 464 # define KSTK_ESP(tsk)  (((unsigned long *)tsk->kernel_stack_page)[1022])
 465 #elif defined(__alpha__)
 466   /*
 467    * See arch/alpha/kernel/ptrace.c for details.
 468    */
 469 # define PT_REG(reg)            (PAGE_SIZE - sizeof(struct pt_regs)     \
 470                                  + (long)&((struct pt_regs *)0)->reg)
 471 # define KSTK_EIP(tsk)  (*(unsigned long *)(tsk->kernel_stack_page + PT_REG(pc)))
 472 # define KSTK_ESP(tsk)  ((tsk) == current ? rdusp() : (tsk)->tss.usp)
 473 #endif
 474 
 475 /* Gcc optimizes away "strlen(x)" for constant x */
 476 #define ADDBUF(buffer, string) \
 477 do { memcpy(buffer, string, strlen(string)); \
 478      buffer += strlen(string); } while (0)
 479 
 480 static inline char * task_name(struct task_struct *p, char * buf)
     /*  */
 481 {
 482         int i;
 483         char * name;
 484 
 485         ADDBUF(buf, "Name:\t");
 486         name = p->comm;
 487         i = sizeof(p->comm);
 488         do {
 489                 unsigned char c = *name;
 490                 name++;
 491                 i--;
 492                 *buf = c;
 493                 if (!c)
 494                         break;
 495                 if (c == '\\') {
 496                         buf[1] = c;
 497                         buf += 2;
 498                         continue;
 499                 }
 500                 if (c == '\n') {
 501                         buf[0] = '\\';
 502                         buf[1] = 'n';
 503                         buf += 2;
 504                         continue;
 505                 }
 506                 buf++;
 507         } while (i);
 508         *buf = '\n';
 509         return buf+1;
 510 }
 511 
 512 static inline char * task_state(struct task_struct *p, char *buffer)
     /*  */
 513 {
 514 #define NR_STATES (sizeof(states)/sizeof(const char *))
 515         unsigned int n = p->state;
 516         static const char * states[] = {
 517                 "R (running)",
 518                 "S (sleeping)",
 519                 "D (disk sleep)",
 520                 "Z (zombie)",
 521                 "T (stopped)",
 522                 "W (paging)",
 523                 ". Huh?"
 524         };
 525 
 526         if (n >= NR_STATES)
 527                 n = NR_STATES-1;
 528 
 529         buffer += sprintf(buffer,
 530                 "State:\t%s\n"
 531                 "Pid:\t%d\n"
 532                 "PPid:\t%d\n"
 533                 "Uid:\t%d\t%d\t%d\t%d\n"
 534                 "Gid:\t%d\t%d\t%d\t%d\n",
 535                 states[n],
 536                 p->pid, p->p_pptr->pid,
 537                 p->uid, p->euid, p->suid, p->fsuid,
 538                 p->gid, p->egid, p->sgid, p->fsgid);
 539         return buffer;
 540 }
 541 
 542 static inline char * task_mem(struct task_struct *p, char *buffer)
     /*  */
 543 {
 544         struct mm_struct * mm = p->mm;
 545 
 546         if (mm && mm != &init_mm) {
 547                 struct vm_area_struct * vma = mm->mmap;
 548                 unsigned long data = 0, stack = 0;
 549                 unsigned long exec = 0, lib = 0;
 550 
 551                 for (vma = mm->mmap; vma; vma = vma->vm_next) {
 552                         unsigned long len = (vma->vm_end - vma->vm_start) >> 10;
 553                         if (!vma->vm_inode) {
 554                                 data += len;
 555                                 if (vma->vm_flags & VM_GROWSDOWN)
 556                                         stack += len;
 557                                 continue;
 558                         }
 559                         if (vma->vm_flags & VM_WRITE)
 560                                 continue;
 561                         if (vma->vm_flags & VM_EXEC) {
 562                                 exec += len;
 563                                 if (vma->vm_flags & VM_EXECUTABLE)
 564                                         continue;
 565                                 lib += len;
 566                         }
 567                 }       
 568                 buffer += sprintf(buffer,
 569                         "VmSize:\t%8lu kB\n"
 570                         "VmLck:\t%8lu kB\n"
 571                         "VmRSS:\t%8lu kB\n"
 572                         "VmData:\t%8lu kB\n"
 573                         "VmStk:\t%8lu kB\n"
 574                         "VmExe:\t%8lu kB\n"
 575                         "VmLib:\t%8lu kB\n",
 576                         mm->total_vm << (PAGE_SHIFT-10),
 577                         mm->locked_vm << (PAGE_SHIFT-10),
 578                         mm->rss << (PAGE_SHIFT-10),
 579                         data - stack, stack,
 580                         exec - lib, lib);
 581         }
 582         return buffer;
 583 }
 584 
 585 static inline char * task_sig(struct task_struct *p, char *buffer)
     /*  */
 586 {
 587         buffer += sprintf(buffer,
 588                 "SigPnd:\t%08lx\n"
 589                 "SigBlk:\t%08lx\n",
 590                 p->signal, p->blocked);
 591         if (p->sig) {
 592                 struct sigaction * action = p->sig->action;
 593                 unsigned long sig_ign = 0, sig_caught = 0;
 594                 unsigned long bit = 1;
 595                 int i;
 596 
 597                 for (i = 0; i < 32; i++) {
 598                         switch((unsigned long) action->sa_handler) {
 599                                 case 0:
 600                                         break;
 601                                 case 1:
 602                                         sig_ign |= bit;
 603                                         break;
 604                                 default:
 605                                         sig_caught |= bit;
 606                         }
 607                         bit <<= 1;
 608                         action++;
 609                 }
 610                 
 611                 buffer += sprintf(buffer,
 612                         "SigIgn:\t%08lx\n"
 613                         "SigCgt:\t%08lx\n",
 614                         sig_ign, sig_caught);
 615         }
 616         return buffer;
 617 }
 618 
 619 static int get_status(int pid, char * buffer)
     /*  */
 620 {
 621         char * orig = buffer;
 622         struct task_struct ** p = get_task(pid), *tsk;
 623 
 624         if (!p || (tsk = *p) == NULL)
 625                 return 0;
 626         buffer = task_name(tsk, buffer);
 627         buffer = task_state(tsk, buffer);
 628         buffer = task_mem(tsk, buffer);
 629         buffer = task_sig(tsk, buffer);
 630         return buffer - orig;
 631 }
 632 
 633 static int get_stat(int pid, char * buffer)
     /*  */
 634 {
 635         struct task_struct ** p = get_task(pid), *tsk;
 636         unsigned long sigignore=0, sigcatch=0, wchan;
 637         unsigned long vsize, eip, esp;
 638         long priority, nice;
 639         int i,tty_pgrp;
 640         char state;
 641 
 642         if (!p || (tsk = *p) == NULL)
 643                 return 0;
 644         if (tsk->state < 0 || tsk->state > 5)
 645                 state = '.';
 646         else
 647                 state = "RSDZTW"[tsk->state];
 648         vsize = eip = esp = 0;
 649         if (tsk->mm && tsk->mm != &init_mm) {
 650                 struct vm_area_struct *vma = tsk->mm->mmap;
 651                 while (vma) {
 652                         vsize += vma->vm_end - vma->vm_start;
 653                         vma = vma->vm_next;
 654                 }
 655                 if (tsk->kernel_stack_page) {
 656                         eip = KSTK_EIP(tsk);
 657                         esp = KSTK_ESP(tsk);
 658                 }
 659         }
 660         wchan = get_wchan(tsk);
 661         if (tsk->sig) {
 662                 unsigned long bit = 1;
 663                 for(i=0; i<32; ++i) {
 664                         switch((unsigned long) tsk->sig->action[i].sa_handler) {
 665                                 case 0:
 666                                         break;
 667                                 case 1:
 668                                         sigignore |= bit;
 669                                         break;
 670                                 default:
 671                                         sigcatch |= bit;
 672                         }
 673                         bit <<= 1;
 674                 }
 675         }
 676         if (tsk->tty)
 677                 tty_pgrp = tsk->tty->pgrp;
 678         else
 679                 tty_pgrp = -1;
 680 
 681         /* scale priority and nice values from timeslices to -20..20 */
 682         /* to make it look like a "normal" unix priority/nice value  */
 683         priority = tsk->counter;
 684         priority = 20 - (priority * 10 + DEF_PRIORITY / 2) / DEF_PRIORITY;
 685         nice = tsk->priority;
 686         nice = 20 - (nice * 20 + DEF_PRIORITY / 2) / DEF_PRIORITY;
 687 
 688         return sprintf(buffer,"%d (%s) %c %d %d %d %d %d %lu %lu \
 689 %lu %lu %lu %ld %ld %ld %ld %ld %ld %lu %lu %ld %lu %lu %lu %lu %lu %lu %lu %lu %lu \
 690 %lu %lu %lu %lu\n",
 691                 pid,
 692                 tsk->comm,
 693                 state,
 694                 tsk->p_pptr->pid,
 695                 tsk->pgrp,
 696                 tsk->session,
 697                 tsk->tty ? kdev_t_to_nr(tsk->tty->device) : 0,
 698                 tty_pgrp,
 699                 tsk->flags,
 700                 tsk->min_flt,
 701                 tsk->cmin_flt,
 702                 tsk->maj_flt,
 703                 tsk->cmaj_flt,
 704                 tsk->utime,
 705                 tsk->stime,
 706                 tsk->cutime,
 707                 tsk->cstime,
 708                 priority,
 709                 nice,
 710                 tsk->timeout,
 711                 tsk->it_real_value,
 712                 tsk->start_time,
 713                 vsize,
 714                 tsk->mm ? tsk->mm->rss : 0, /* you might want to shift this left 3 */
 715                 tsk->rlim ? tsk->rlim[RLIMIT_RSS].rlim_cur : 0,
 716                 tsk->mm ? tsk->mm->start_code : 0,
 717                 tsk->mm ? tsk->mm->end_code : 0,
 718                 tsk->mm ? tsk->mm->start_stack : 0,
 719                 esp,
 720                 eip,
 721                 tsk->signal,
 722                 tsk->blocked,
 723                 sigignore,
 724                 sigcatch,
 725                 wchan);
 726 }
 727                 
 728 static inline void statm_pte_range(pmd_t * pmd, unsigned long address, unsigned long size,
     /*  */
 729         int * pages, int * shared, int * dirty, int * total)
 730 {
 731         pte_t * pte;
 732         unsigned long end;
 733 
 734         if (pmd_none(*pmd))
 735                 return;
 736         if (pmd_bad(*pmd)) {
 737                 printk("statm_pte_range: bad pmd (%08lx)\n", pmd_val(*pmd));
 738                 pmd_clear(pmd);
 739                 return;
 740         }
 741         pte = pte_offset(pmd, address);
 742         address &= ~PMD_MASK;
 743         end = address + size;
 744         if (end > PMD_SIZE)
 745                 end = PMD_SIZE;
 746         do {
 747                 pte_t page = *pte;
 748 
 749                 address += PAGE_SIZE;
 750                 pte++;
 751                 if (pte_none(page))
 752                         continue;
 753                 ++*total;
 754                 if (!pte_present(page))
 755                         continue;
 756                 ++*pages;
 757                 if (pte_dirty(page))
 758                         ++*dirty;
 759                 if (pte_page(page) >= high_memory)
 760                         continue;
 761                 if (mem_map[MAP_NR(pte_page(page))].count > 1)
 762                         ++*shared;
 763         } while (address < end);
 764 }
 765 
 766 static inline void statm_pmd_range(pgd_t * pgd, unsigned long address, unsigned long size,
     /*  */
 767         int * pages, int * shared, int * dirty, int * total)
 768 {
 769         pmd_t * pmd;
 770         unsigned long end;
 771 
 772         if (pgd_none(*pgd))
 773                 return;
 774         if (pgd_bad(*pgd)) {
 775                 printk("statm_pmd_range: bad pgd (%08lx)\n", pgd_val(*pgd));
 776                 pgd_clear(pgd);
 777                 return;
 778         }
 779         pmd = pmd_offset(pgd, address);
 780         address &= ~PGDIR_MASK;
 781         end = address + size;
 782         if (end > PGDIR_SIZE)
 783                 end = PGDIR_SIZE;
 784         do {
 785                 statm_pte_range(pmd, address, end - address, pages, shared, dirty, total);
 786                 address = (address + PMD_SIZE) & PMD_MASK;
 787                 pmd++;
 788         } while (address < end);
 789 }
 790 
 791 static void statm_pgd_range(pgd_t * pgd, unsigned long address, unsigned long end,
     /*  */
 792         int * pages, int * shared, int * dirty, int * total)
 793 {
 794         while (address < end) {
 795                 statm_pmd_range(pgd, address, end - address, pages, shared, dirty, total);
 796                 address = (address + PGDIR_SIZE) & PGDIR_MASK;
 797                 pgd++;
 798         }
 799 }
 800 
 801 static int get_statm(int pid, char * buffer)
     /*  */
 802 {
 803         struct task_struct ** p = get_task(pid), *tsk;
 804         int size=0, resident=0, share=0, trs=0, lrs=0, drs=0, dt=0;
 805 
 806         if (!p || (tsk = *p) == NULL)
 807                 return 0;
 808         if (tsk->mm && tsk->mm != &init_mm) {
 809                 struct vm_area_struct * vma = tsk->mm->mmap;
 810 
 811                 while (vma) {
 812                         pgd_t *pgd = pgd_offset(tsk->mm, vma->vm_start);
 813                         int pages = 0, shared = 0, dirty = 0, total = 0;
 814 
 815                         statm_pgd_range(pgd, vma->vm_start, vma->vm_end, &pages, &shared, &dirty, &total);
 816                         resident += pages;
 817                         share += shared;
 818                         dt += dirty;
 819                         size += total;
 820                         if (vma->vm_flags & VM_EXECUTABLE)
 821                                 trs += pages;   /* text */
 822                         else if (vma->vm_flags & VM_GROWSDOWN)
 823                                 drs += pages;   /* stack */
 824                         else if (vma->vm_end > 0x60000000)
 825                                 lrs += pages;   /* library */
 826                         else
 827                                 drs += pages;
 828                         vma = vma->vm_next;
 829                 }
 830         }
 831         return sprintf(buffer,"%d %d %d %d %d %d %d\n",
 832                        size, resident, share, trs, lrs, drs, dt);
 833 }
 834 
 835 /*
 836  * The way we support synthetic files > 4K
 837  * - without storing their contents in some buffer and
 838  * - without walking through the entire synthetic file until we reach the
 839  *   position of the requested data
 840  * is to cleverly encode the current position in the file's f_pos field.
 841  * There is no requirement that a read() call which returns `count' bytes
 842  * of data increases f_pos by exactly `count'.
 843  *
 844  * This idea is Linus' one. Bruno implemented it.
 845  */
 846 
 847 /*
 848  * For the /proc/<pid>/maps file, we use fixed length records, each containing
 849  * a single line.
 850  */
 851 #define MAPS_LINE_LENGTH        1024
 852 #define MAPS_LINE_SHIFT         10
 853 /*
 854  * f_pos = (number of the vma in the task->mm->mmap list) * MAPS_LINE_LENGTH
 855  *         + (index into the line)
 856  */
 857 #define MAPS_LINE_FORMAT          "%08lx-%08lx %s %08lx %s %lu\n"
 858 #define MAPS_LINE_MAX   49 /* sum of 8  1  8  1 4 1 8 1 5 1 10 1 */
 859 
 860 static int read_maps (int pid, struct file * file, char * buf, int count)
     /*  */
 861 {
 862         struct task_struct ** p = get_task(pid);
 863         char * destptr;
 864         loff_t lineno;
 865         int column;
 866         struct vm_area_struct * map;
 867         int i;
 868 
 869         if (!p || !*p)
 870                 return -EINVAL;
 871 
 872         if (!(*p)->mm || (*p)->mm == &init_mm || count == 0)
 873                 return 0;
 874 
 875         /* decode f_pos */
 876         lineno = file->f_pos >> MAPS_LINE_SHIFT;
 877         column = file->f_pos & (MAPS_LINE_LENGTH-1);
 878 
 879         /* quickly go to line lineno */
 880         for (map = (*p)->mm->mmap, i = 0; map && (i < lineno); map = map->vm_next, i++)
 881                 continue;
 882 
 883         destptr = buf;
 884 
 885         for ( ; map ; ) {
 886                 /* produce the next line */
 887                 char line[MAPS_LINE_MAX+1];
 888                 char str[5], *cp = str;
 889                 int flags;
 890                 kdev_t dev;
 891                 unsigned long ino;
 892                 int len;
 893 
 894                 flags = map->vm_flags;
 895 
 896                 *cp++ = flags & VM_READ ? 'r' : '-';
 897                 *cp++ = flags & VM_WRITE ? 'w' : '-';
 898                 *cp++ = flags & VM_EXEC ? 'x' : '-';
 899                 *cp++ = flags & VM_MAYSHARE ? 's' : 'p';
 900                 *cp++ = 0;
 901 
 902                 if (map->vm_inode != NULL) {
 903                         dev = map->vm_inode->i_dev;
 904                         ino = map->vm_inode->i_ino;
 905                 } else {
 906                         dev = 0;
 907                         ino = 0;
 908                 }
 909 
 910                 len = sprintf(line, MAPS_LINE_FORMAT,
 911                               map->vm_start, map->vm_end, str, map->vm_offset,
 912                               kdevname(dev), ino);
 913 
 914                 if (column >= len) {
 915                         column = 0; /* continue with next line at column 0 */
 916                         lineno++;
 917                         map = map->vm_next;
 918                         continue;
 919                 }
 920 
 921                 i = len-column;
 922                 if (i > count)
 923                         i = count;
 924                 memcpy_tofs(destptr, line+column, i);
 925                 destptr += i; count -= i;
 926                 column += i;
 927                 if (column >= len) {
 928                         column = 0; /* next time: next line at column 0 */
 929                         lineno++;
 930                         map = map->vm_next;
 931                 }
 932 
 933                 /* done? */
 934                 if (count == 0)
 935                         break;
 936 
 937                 /* By writing to user space, we might have slept.
 938                  * Stop the loop, to avoid a race condition.
 939                  */
 940                 if (*p != current)
 941                         break;
 942         }
 943 
 944         /* encode f_pos */
 945         file->f_pos = (lineno << MAPS_LINE_SHIFT) + column;
 946 
 947         return destptr-buf;
 948 }
 949 
 950 #ifdef CONFIG_MODULES
 951 extern int get_module_list(char *);
 952 extern int get_ksyms_list(char *, char **, off_t, int);
 953 #endif
 954 extern int get_device_list(char *);
 955 extern int get_filesystem_list(char *);
 956 extern int get_filesystem_info( char * );
 957 extern int get_irq_list(char *);
 958 extern int get_dma_list(char *);
 959 extern int get_cpuinfo(char *);
 960 extern int get_pci_list(char*);
 961 extern int get_md_status (char *);
 962 #ifdef __SMP_PROF__
 963 extern int get_smp_prof_list(char *);
 964 #endif
 965 
 966 static int get_root_array(char * page, int type, char **start, off_t offset, int length)
     /*  */
 967 {
 968         switch (type) {
 969                 case PROC_LOADAVG:
 970                         return get_loadavg(page);
 971 
 972                 case PROC_UPTIME:
 973                         return get_uptime(page);
 974 
 975                 case PROC_MEMINFO:
 976                         return get_meminfo(page);
 977 
 978 #ifdef CONFIG_PCI
 979                 case PROC_PCI:
 980                         return get_pci_list(page);
 981 #endif
 982                         
 983                 case PROC_CPUINFO:
 984                         return get_cpuinfo(page);
 985 
 986                 case PROC_VERSION:
 987                         return get_version(page);
 988 
 989 #ifdef CONFIG_DEBUG_MALLOC
 990                 case PROC_MALLOC:
 991                         return get_malloc(page);
 992 #endif
 993 
 994 #ifdef CONFIG_MODULES
 995                 case PROC_MODULES:
 996                         return get_module_list(page);
 997 
 998                 case PROC_KSYMS:
 999                         return get_ksyms_list(page, start, offset, length);
1000 #endif
1001 
1002                 case PROC_STAT:
1003                         return get_kstat(page);
1004 
1005                 case PROC_DEVICES:
1006                         return get_device_list(page);
1007 
1008                 case PROC_INTERRUPTS:
1009                         return get_irq_list(page);
1010 
1011                 case PROC_FILESYSTEMS:
1012                         return get_filesystem_list(page);
1013 
1014                 case PROC_DMA:
1015                         return get_dma_list(page);
1016 
1017                 case PROC_IOPORTS:
1018                         return get_ioport_list(page);
1019 #ifdef CONFIG_BLK_DEV_MD
1020                 case PROC_MD:
1021                         return get_md_status(page);
1022 #endif
1023 #ifdef CONFIG_APM
1024                 case PROC_APM:
1025                         return apm_proc(page);
1026 #endif
1027 #ifdef __SMP_PROF__
1028                 case PROC_SMP_PROF:
1029                         return get_smp_prof_list(page);
1030 #endif
1031                 case PROC_CMDLINE:
1032                         return get_cmdline(page);
1033 
1034                 case PROC_MTAB:
1035                        return get_filesystem_info( page );
1036         }
1037         return -EBADF;
1038 }
1039 
1040 static int get_process_array(char * page, int pid, int type)
     /*  */
1041 {
1042         switch (type) {
1043                 case PROC_PID_STATUS:
1044                         return get_status(pid, page);
1045                 case PROC_PID_ENVIRON:
1046                         return get_env(pid, page);
1047                 case PROC_PID_CMDLINE:
1048                         return get_arg(pid, page);
1049                 case PROC_PID_STAT:
1050                         return get_stat(pid, page);
1051                 case PROC_PID_STATM:
1052                         return get_statm(pid, page);
1053         }
1054         return -EBADF;
1055 }
1056 
1057 
1058 static inline int fill_array(char * page, int pid, int type, char **start, off_t offset, int length)
     /*  */
1059 {
1060         if (pid)
1061                 return get_process_array(page, pid, type);
1062         return get_root_array(page, type, start, offset, length);
1063 }
1064 
1065 #define PROC_BLOCK_SIZE (3*1024)                /* 4K page size but our output routines use some slack for overruns */
1066 
1067 static int array_read(struct inode * inode, struct file * file,char * buf, int count)
     /*  */
1068 {
1069         unsigned long page;
1070         char *start;
1071         int length;
1072         int end;
1073         unsigned int type, pid;
1074 
1075         if (count < 0)
1076                 return -EINVAL;
1077         if (count > PROC_BLOCK_SIZE)
1078                 count = PROC_BLOCK_SIZE;
1079         if (!(page = __get_free_page(GFP_KERNEL)))
1080                 return -ENOMEM;
1081         type = inode->i_ino;
1082         pid = type >> 16;
1083         type &= 0x0000ffff;
1084         start = NULL;
1085         length = fill_array((char *) page, pid, type,
1086                             &start, file->f_pos, count);
1087         if (length < 0) {
1088                 free_page(page);
1089                 return length;
1090         }
1091         if (start != NULL) {
1092                 /* We have had block-adjusting processing! */
1093                 memcpy_tofs(buf, start, length);
1094                 file->f_pos += length;
1095                 count = length;
1096         } else {
1097                 /* Static 4kB (or whatever) block capacity */
1098                 if (file->f_pos >= length) {
1099                         free_page(page);
1100                         return 0;
1101                 }
1102                 if (count + file->f_pos > length)
1103                         count = length - file->f_pos;
1104                 end = count + file->f_pos;
1105                 memcpy_tofs(buf, (char *) page + file->f_pos, count);
1106                 file->f_pos = end;
1107         }
1108         free_page(page);
1109         return count;
1110 }
1111 
1112 static struct file_operations proc_array_operations = {
1113         NULL,           /* array_lseek */
1114         array_read,
1115         NULL,           /* array_write */
1116         NULL,           /* array_readdir */
1117         NULL,           /* array_select */
1118         NULL,           /* array_ioctl */
1119         NULL,           /* mmap */
1120         NULL,           /* no special open code */
1121         NULL,           /* no special release code */
1122         NULL            /* can't fsync */
1123 };
1124 
1125 struct inode_operations proc_array_inode_operations = {
1126         &proc_array_operations, /* default base directory file-ops */
1127         NULL,                   /* create */
1128         NULL,                   /* lookup */
1129         NULL,                   /* link */
1130         NULL,                   /* unlink */
1131         NULL,                   /* symlink */
1132         NULL,                   /* mkdir */
1133         NULL,                   /* rmdir */
1134         NULL,                   /* mknod */
1135         NULL,                   /* rename */
1136         NULL,                   /* readlink */
1137         NULL,                   /* follow_link */
1138         NULL,                   /* readpage */
1139         NULL,                   /* writepage */
1140         NULL,                   /* bmap */
1141         NULL,                   /* truncate */
1142         NULL                    /* permission */
1143 };
1144 
1145 static int arraylong_read (struct inode * inode, struct file * file, char * buf, int count)
     /*  */
1146 {
1147         unsigned int pid = inode->i_ino >> 16;
1148         unsigned int type = inode->i_ino & 0x0000ffff;
1149 
1150         if (count < 0)
1151                 return -EINVAL;
1152 
1153         switch (type) {
1154                 case PROC_PID_MAPS:
1155                         return read_maps(pid, file, buf, count);
1156         }
1157         return -EINVAL;
1158 }
1159 
1160 static struct file_operations proc_arraylong_operations = {
1161         NULL,           /* array_lseek */
1162         arraylong_read,
1163         NULL,           /* array_write */
1164         NULL,           /* array_readdir */
1165         NULL,           /* array_select */
1166         NULL,           /* array_ioctl */
1167         NULL,           /* mmap */
1168         NULL,           /* no special open code */
1169         NULL,           /* no special release code */
1170         NULL            /* can't fsync */
1171 };
1172 
1173 struct inode_operations proc_arraylong_inode_operations = {
1174         &proc_arraylong_operations,     /* default base directory file-ops */
1175         NULL,                   /* create */
1176         NULL,                   /* lookup */
1177         NULL,                   /* link */
1178         NULL,                   /* unlink */
1179         NULL,                   /* symlink */
1180         NULL,                   /* mkdir */
1181         NULL,                   /* rmdir */
1182         NULL,                   /* mknod */
1183         NULL,                   /* rename */
1184         NULL,                   /* readlink */
1185         NULL,                   /* follow_link */
1186         NULL,                   /* readpage */
1187         NULL,                   /* writepage */
1188         NULL,                   /* bmap */
1189         NULL,                   /* truncate */
1190         NULL                    /* permission */
1191 };
/* */
root/fs/proc/array.c

DEFINITIONS
