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

DEFINITIONS
