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         extern unsigned long total_forks;
 198 
 199         for (i = 0 ; i < NR_IRQS ; i++)
 200                 sum += kstat.interrupts[i];
 201         len = sprintf(buffer,
 202                 "cpu  %u %u %u %lu\n"
 203                 "disk %u %u %u %u\n"
 204                 "disk_rio %u %u %u %u\n"
 205                 "disk_wio %u %u %u %u\n"
 206                 "disk_rblk %u %u %u %u\n"
 207                 "disk_wblk %u %u %u %u\n"
 208                 "page %u %u\n"
 209                 "swap %u %u\n"
 210                 "intr %u",
 211                 kstat.cpu_user,
 212                 kstat.cpu_nice,
 213                 kstat.cpu_system,
 214                 jiffies - (kstat.cpu_user + kstat.cpu_nice + kstat.cpu_system),
 215                 kstat.dk_drive[0], kstat.dk_drive[1],
 216                 kstat.dk_drive[2], kstat.dk_drive[3],
 217                 kstat.dk_drive_rio[0], kstat.dk_drive_rio[1],
 218                 kstat.dk_drive_rio[2], kstat.dk_drive_rio[3],
 219                 kstat.dk_drive_wio[0], kstat.dk_drive_wio[1],
 220                 kstat.dk_drive_wio[2], kstat.dk_drive_wio[3],
 221                 kstat.dk_drive_rblk[0], kstat.dk_drive_rblk[1],
 222                 kstat.dk_drive_rblk[2], kstat.dk_drive_rblk[3],
 223                 kstat.dk_drive_wblk[0], kstat.dk_drive_wblk[1],
 224                 kstat.dk_drive_wblk[2], kstat.dk_drive_wblk[3],
 225                 kstat.pgpgin,
 226                 kstat.pgpgout,
 227                 kstat.pswpin,
 228                 kstat.pswpout,
 229                 sum);
 230         for (i = 0 ; i < NR_IRQS ; i++)
 231                 len += sprintf(buffer + len, " %u", kstat.interrupts[i]);
 232         len += sprintf(buffer + len,
 233                 "\nctxt %u\n"
 234                 "btime %lu\n"
 235                 "processes %lu\n",
 236                 kstat.context_swtch,
 237                 xtime.tv_sec - jiffies / HZ,
 238                 total_forks);
 239         return len;
 240 }
 241 
 242 
 243 static int get_uptime(char * buffer)
     /*  */
 244 {
 245         unsigned long uptime;
 246         unsigned long idle;
 247 
 248         uptime = jiffies;
 249         idle = task[0]->utime + task[0]->stime;
 250 
 251         /* The formula for the fraction parts really is ((t * 100) / HZ) % 100, but
 252            that would overflow about every five days at HZ == 100.
 253            Therefore the identity a = (a / b) * b + a % b is used so that it is
 254            calculated as (((t / HZ) * 100) + ((t % HZ) * 100) / HZ) % 100.
 255            The part in front of the '+' always evaluates as 0 (mod 100). All divisions
 256            in the above formulas are truncating. For HZ being a power of 10, the
 257            calculations simplify to the version in the #else part (if the printf
 258            format is adapted to the same number of digits as zeroes in HZ.
 259          */
 260 #if HZ!=100
 261         return sprintf(buffer,"%lu.%02lu %lu.%02lu\n",
 262                 uptime / HZ,
 263                 (((uptime % HZ) * 100) / HZ) % 100,
 264                 idle / HZ,
 265                 (((idle % HZ) * 100) / HZ) % 100);
 266 #else
 267         return sprintf(buffer,"%lu.%02lu %lu.%02lu\n",
 268                 uptime / HZ,
 269                 uptime % HZ,
 270                 idle / HZ,
 271                 idle % HZ);
 272 #endif
 273 }
 274 
 275 static int get_meminfo(char * buffer)
     /*  */
 276 {
 277         struct sysinfo i;
 278         int len;
 279 
 280         si_meminfo(&i);
 281         si_swapinfo(&i);
 282         len = sprintf(buffer, "        total:    used:    free:  shared: buffers:  cached:\n"
 283                 "Mem:  %8lu %8lu %8lu %8lu %8lu %8lu\n"
 284                 "Swap: %8lu %8lu %8lu\n",
 285                 i.totalram, i.totalram-i.freeram, i.freeram, i.sharedram, i.bufferram, page_cache_size*PAGE_SIZE,
 286                 i.totalswap, i.totalswap-i.freeswap, i.freeswap);
 287         /*
 288          * Tagged format, for easy grepping and expansion. The above will go away
 289          * eventually, once the tools have been updated.
 290          */
 291         return len + sprintf(buffer+len,
 292                 "MemTotal:  %8lu kB\n"
 293                 "MemFree:   %8lu kB\n"
 294                 "MemShared: %8lu kB\n"
 295                 "Buffers:   %8lu kB\n"
 296                 "Cached:    %8lu kB\n"
 297                 "SwapTotal: %8lu kB\n"
 298                 "SwapFree:  %8lu kB\n",
 299                 i.totalram >> 10,
 300                 i.freeram >> 10,
 301                 i.sharedram >> 10,
 302                 i.bufferram >> 10,
 303                 page_cache_size << (PAGE_SHIFT - 10),
 304                 i.totalswap >> 10,
 305                 i.freeswap >> 10);
 306 }
 307 
 308 static int get_version(char * buffer)
     /*  */
 309 {
 310         extern char *linux_banner;
 311 
 312         strcpy(buffer, linux_banner);
 313         return strlen(buffer);
 314 }
 315 
 316 static int get_cmdline(char * buffer)
     /*  */
 317 {
 318         extern char saved_command_line[];
 319 
 320         return sprintf(buffer, "%s\n", saved_command_line);
 321 }
 322 
 323 static struct task_struct ** get_task(pid_t pid)
     /*  */
 324 {
 325         struct task_struct ** p;
 326 
 327         p = task;
 328         while (++p < task+NR_TASKS) {
 329                 if (*p && (*p)->pid == pid)
 330                         return p;
 331         }
 332         return NULL;
 333 }
 334 
 335 static unsigned long get_phys_addr(struct task_struct * p, unsigned long ptr)
     /*  */
 336 {
 337         pgd_t *page_dir;
 338         pmd_t *page_middle;
 339         pte_t pte;
 340 
 341         if (!p || !p->mm || ptr >= TASK_SIZE)
 342                 return 0;
 343         page_dir = pgd_offset(p->mm,ptr);
 344         if (pgd_none(*page_dir))
 345                 return 0;
 346         if (pgd_bad(*page_dir)) {
 347                 printk("bad page directory entry %08lx\n", pgd_val(*page_dir));
 348                 pgd_clear(page_dir);
 349                 return 0;
 350         }
 351         page_middle = pmd_offset(page_dir,ptr);
 352         if (pmd_none(*page_middle))
 353                 return 0;
 354         if (pmd_bad(*page_middle)) {
 355                 printk("bad page middle entry %08lx\n", pmd_val(*page_middle));
 356                 pmd_clear(page_middle);
 357                 return 0;
 358         }
 359         pte = *pte_offset(page_middle,ptr);
 360         if (!pte_present(pte))
 361                 return 0;
 362         return pte_page(pte) + (ptr & ~PAGE_MASK);
 363 }
 364 
 365 static int get_array(struct task_struct ** p, unsigned long start, unsigned long end, char * buffer)
     /*  */
 366 {
 367         unsigned long addr;
 368         int size = 0, result = 0;
 369         char c;
 370 
 371         if (start >= end)
 372                 return result;
 373         for (;;) {
 374                 addr = get_phys_addr(*p, start);
 375                 if (!addr)
 376                         goto ready;
 377                 do {
 378                         c = *(char *) addr;
 379                         if (!c)
 380                                 result = size;
 381                         if (size < PAGE_SIZE)
 382                                 buffer[size++] = c;
 383                         else
 384                                 goto ready;
 385                         addr++;
 386                         start++;
 387                         if (!c && start >= end)
 388                                 goto ready;
 389                 } while (addr & ~PAGE_MASK);
 390         }
 391 ready:
 392         /* remove the trailing blanks, used to fill out argv,envp space */
 393         while (result>0 && buffer[result-1]==' ')
 394                 result--;
 395         return result;
 396 }
 397 
 398 static int get_env(int pid, char * buffer)
     /*  */
 399 {
 400         struct task_struct ** p = get_task(pid);
 401 
 402         if (!p || !*p || !(*p)->mm)
 403                 return 0;
 404         return get_array(p, (*p)->mm->env_start, (*p)->mm->env_end, buffer);
 405 }
 406 
 407 static int get_arg(int pid, char * buffer)
     /*  */
 408 {
 409         struct task_struct ** p = get_task(pid);
 410 
 411         if (!p || !*p || !(*p)->mm)
 412                 return 0;
 413         return get_array(p, (*p)->mm->arg_start, (*p)->mm->arg_end, buffer);
 414 }
 415 
 416 static unsigned long get_wchan(struct task_struct *p)
     /*  */
 417 {
 418         if (!p || p == current || p->state == TASK_RUNNING)
 419                 return 0;
 420 #if defined(__i386__)
 421         {
 422                 unsigned long ebp, eip;
 423                 unsigned long stack_page;
 424                 int count = 0;
 425 
 426                 stack_page = p->kernel_stack_page;
 427                 if (!stack_page)
 428                         return 0;
 429                 ebp = p->tss.ebp;
 430                 do {
 431                         if (ebp < stack_page || ebp >= 4092+stack_page)
 432                                 return 0;
 433                         eip = *(unsigned long *) (ebp+4);
 434                         if ((void *)eip != sleep_on &&
 435                             (void *)eip != interruptible_sleep_on)
 436                                 return eip;
 437                         ebp = *(unsigned long *) ebp;
 438                 } while (count++ < 16);
 439         }
 440 #elif defined(__alpha__)
 441         /*
 442          * This one depends on the frame size of schedule().  Do a
 443          * "disass schedule" in gdb to find the frame size.  Also, the
 444          * code assumes that sleep_on() follows immediately after
 445          * interruptible_sleep_on() and that add_timer() follows
 446          * immediately after interruptible_sleep().  Ugly, isn't it?
 447          * Maybe adding a wchan field to task_struct would be better,
 448          * after all...
 449          */
 450         {
 451             unsigned long schedule_frame;
 452             unsigned long pc;
 453 
 454             pc = thread_saved_pc(&p->tss);
 455             if (pc >= (unsigned long) interruptible_sleep_on && pc < (unsigned long) add_timer) {
 456                 schedule_frame = ((unsigned long *)p->tss.ksp)[6];
 457                 return ((unsigned long *)schedule_frame)[12];
 458             }
 459             return pc;
 460         }
 461 #endif
 462         return 0;
 463 }
 464 
 465 #if defined(__i386__)
 466 # define KSTK_EIP(tsk)  (((unsigned long *)tsk->kernel_stack_page)[1019])
 467 # define KSTK_ESP(tsk)  (((unsigned long *)tsk->kernel_stack_page)[1022])
 468 #elif defined(__alpha__)
 469   /*
 470    * See arch/alpha/kernel/ptrace.c for details.
 471    */
 472 # define PT_REG(reg)            (PAGE_SIZE - sizeof(struct pt_regs)     \
 473                                  + (long)&((struct pt_regs *)0)->reg)
 474 # define KSTK_EIP(tsk)  (*(unsigned long *)(tsk->kernel_stack_page + PT_REG(pc)))
 475 # define KSTK_ESP(tsk)  ((tsk) == current ? rdusp() : (tsk)->tss.usp)
 476 #endif
 477 
 478 /* Gcc optimizes away "strlen(x)" for constant x */
 479 #define ADDBUF(buffer, string) \
 480 do { memcpy(buffer, string, strlen(string)); \
 481      buffer += strlen(string); } while (0)
 482 
 483 static inline char * task_name(struct task_struct *p, char * buf)
     /*  */
 484 {
 485         int i;
 486         char * name;
 487 
 488         ADDBUF(buf, "Name:\t");
 489         name = p->comm;
 490         i = sizeof(p->comm);
 491         do {
 492                 unsigned char c = *name;
 493                 name++;
 494                 i--;
 495                 *buf = c;
 496                 if (!c)
 497                         break;
 498                 if (c == '\\') {
 499                         buf[1] = c;
 500                         buf += 2;
 501                         continue;
 502                 }
 503                 if (c == '\n') {
 504                         buf[0] = '\\';
 505                         buf[1] = 'n';
 506                         buf += 2;
 507                         continue;
 508                 }
 509                 buf++;
 510         } while (i);
 511         *buf = '\n';
 512         return buf+1;
 513 }
 514 
 515 static inline char * task_state(struct task_struct *p, char *buffer)
     /*  */
 516 {
 517 #define NR_STATES (sizeof(states)/sizeof(const char *))
 518         unsigned int n = p->state;
 519         static const char * states[] = {
 520                 "R (running)",
 521                 "S (sleeping)",
 522                 "D (disk sleep)",
 523                 "Z (zombie)",
 524                 "T (stopped)",
 525                 "W (paging)",
 526                 ". Huh?"
 527         };
 528 
 529         if (n >= NR_STATES)
 530                 n = NR_STATES-1;
 531 
 532         buffer += sprintf(buffer,
 533                 "State:\t%s\n"
 534                 "Pid:\t%d\n"
 535                 "PPid:\t%d\n"
 536                 "Uid:\t%d\t%d\t%d\t%d\n"
 537                 "Gid:\t%d\t%d\t%d\t%d\n",
 538                 states[n],
 539                 p->pid, p->p_pptr->pid,
 540                 p->uid, p->euid, p->suid, p->fsuid,
 541                 p->gid, p->egid, p->sgid, p->fsgid);
 542         return buffer;
 543 }
 544 
 545 static inline char * task_mem(struct task_struct *p, char *buffer)
     /*  */
 546 {
 547         struct mm_struct * mm = p->mm;
 548 
 549         if (mm && mm != &init_mm) {
 550                 struct vm_area_struct * vma = mm->mmap;
 551                 unsigned long data = 0, stack = 0;
 552                 unsigned long exec = 0, lib = 0;
 553 
 554                 for (vma = mm->mmap; vma; vma = vma->vm_next) {
 555                         unsigned long len = (vma->vm_end - vma->vm_start) >> 10;
 556                         if (!vma->vm_inode) {
 557                                 data += len;
 558                                 if (vma->vm_flags & VM_GROWSDOWN)
 559                                         stack += len;
 560                                 continue;
 561                         }
 562                         if (vma->vm_flags & VM_WRITE)
 563                                 continue;
 564                         if (vma->vm_flags & VM_EXEC) {
 565                                 exec += len;
 566                                 if (vma->vm_flags & VM_EXECUTABLE)
 567                                         continue;
 568                                 lib += len;
 569                         }
 570                 }       
 571                 buffer += sprintf(buffer,
 572                         "VmSize:\t%8lu kB\n"
 573                         "VmLck:\t%8lu kB\n"
 574                         "VmRSS:\t%8lu kB\n"
 575                         "VmData:\t%8lu kB\n"
 576                         "VmStk:\t%8lu kB\n"
 577                         "VmExe:\t%8lu kB\n"
 578                         "VmLib:\t%8lu kB\n",
 579                         mm->total_vm << (PAGE_SHIFT-10),
 580                         mm->locked_vm << (PAGE_SHIFT-10),
 581                         mm->rss << (PAGE_SHIFT-10),
 582                         data - stack, stack,
 583                         exec - lib, lib);
 584         }
 585         return buffer;
 586 }
 587 
 588 static inline char * task_sig(struct task_struct *p, char *buffer)
     /*  */
 589 {
 590         buffer += sprintf(buffer,
 591                 "SigPnd:\t%08lx\n"
 592                 "SigBlk:\t%08lx\n",
 593                 p->signal, p->blocked);
 594         if (p->sig) {
 595                 struct sigaction * action = p->sig->action;
 596                 unsigned long sig_ign = 0, sig_caught = 0;
 597                 unsigned long bit = 1;
 598                 int i;
 599 
 600                 for (i = 0; i < 32; i++) {
 601                         switch((unsigned long) action->sa_handler) {
 602                                 case 0:
 603                                         break;
 604                                 case 1:
 605                                         sig_ign |= bit;
 606                                         break;
 607                                 default:
 608                                         sig_caught |= bit;
 609                         }
 610                         bit <<= 1;
 611                         action++;
 612                 }
 613                 
 614                 buffer += sprintf(buffer,
 615                         "SigIgn:\t%08lx\n"
 616                         "SigCgt:\t%08lx\n",
 617                         sig_ign, sig_caught);
 618         }
 619         return buffer;
 620 }
 621 
 622 static int get_status(int pid, char * buffer)
     /*  */
 623 {
 624         char * orig = buffer;
 625         struct task_struct ** p = get_task(pid), *tsk;
 626 
 627         if (!p || (tsk = *p) == NULL)
 628                 return 0;
 629         buffer = task_name(tsk, buffer);
 630         buffer = task_state(tsk, buffer);
 631         buffer = task_mem(tsk, buffer);
 632         buffer = task_sig(tsk, buffer);
 633         return buffer - orig;
 634 }
 635 
 636 static int get_stat(int pid, char * buffer)
     /*  */
 637 {
 638         struct task_struct ** p = get_task(pid), *tsk;
 639         unsigned long sigignore=0, sigcatch=0, wchan;
 640         unsigned long vsize, eip, esp;
 641         long priority, nice;
 642         int i,tty_pgrp;
 643         char state;
 644 
 645         if (!p || (tsk = *p) == NULL)
 646                 return 0;
 647         if (tsk->state < 0 || tsk->state > 5)
 648                 state = '.';
 649         else
 650                 state = "RSDZTW"[tsk->state];
 651         vsize = eip = esp = 0;
 652         if (tsk->mm && tsk->mm != &init_mm) {
 653                 struct vm_area_struct *vma = tsk->mm->mmap;
 654                 while (vma) {
 655                         vsize += vma->vm_end - vma->vm_start;
 656                         vma = vma->vm_next;
 657                 }
 658                 if (tsk->kernel_stack_page) {
 659                         eip = KSTK_EIP(tsk);
 660                         esp = KSTK_ESP(tsk);
 661                 }
 662         }
 663         wchan = get_wchan(tsk);
 664         if (tsk->sig) {
 665                 unsigned long bit = 1;
 666                 for(i=0; i<32; ++i) {
 667                         switch((unsigned long) tsk->sig->action[i].sa_handler) {
 668                                 case 0:
 669                                         break;
 670                                 case 1:
 671                                         sigignore |= bit;
 672                                         break;
 673                                 default:
 674                                         sigcatch |= bit;
 675                         }
 676                         bit <<= 1;
 677                 }
 678         }
 679         if (tsk->tty)
 680                 tty_pgrp = tsk->tty->pgrp;
 681         else
 682                 tty_pgrp = -1;
 683 
 684         /* scale priority and nice values from timeslices to -20..20 */
 685         /* to make it look like a "normal" unix priority/nice value  */
 686         priority = tsk->counter;
 687         priority = 20 - (priority * 10 + DEF_PRIORITY / 2) / DEF_PRIORITY;
 688         nice = tsk->priority;
 689         nice = 20 - (nice * 20 + DEF_PRIORITY / 2) / DEF_PRIORITY;
 690 
 691         return sprintf(buffer,"%d (%s) %c %d %d %d %d %d %lu %lu \
 692 %lu %lu %lu %ld %ld %ld %ld %ld %ld %lu %lu %ld %lu %lu %lu %lu %lu %lu %lu %lu %lu \
 693 %lu %lu %lu %lu\n",
 694                 pid,
 695                 tsk->comm,
 696                 state,
 697                 tsk->p_pptr->pid,
 698                 tsk->pgrp,
 699                 tsk->session,
 700                 tsk->tty ? kdev_t_to_nr(tsk->tty->device) : 0,
 701                 tty_pgrp,
 702                 tsk->flags,
 703                 tsk->min_flt,
 704                 tsk->cmin_flt,
 705                 tsk->maj_flt,
 706                 tsk->cmaj_flt,
 707                 tsk->utime,
 708                 tsk->stime,
 709                 tsk->cutime,
 710                 tsk->cstime,
 711                 priority,
 712                 nice,
 713                 tsk->timeout,
 714                 tsk->it_real_value,
 715                 tsk->start_time,
 716                 vsize,
 717                 tsk->mm ? tsk->mm->rss : 0, /* you might want to shift this left 3 */
 718                 tsk->rlim ? tsk->rlim[RLIMIT_RSS].rlim_cur : 0,
 719                 tsk->mm ? tsk->mm->start_code : 0,
 720                 tsk->mm ? tsk->mm->end_code : 0,
 721                 tsk->mm ? tsk->mm->start_stack : 0,
 722                 esp,
 723                 eip,
 724                 tsk->signal,
 725                 tsk->blocked,
 726                 sigignore,
 727                 sigcatch,
 728                 wchan);
 729 }
 730                 
 731 static inline void statm_pte_range(pmd_t * pmd, unsigned long address, unsigned long size,
     /*  */
 732         int * pages, int * shared, int * dirty, int * total)
 733 {
 734         pte_t * pte;
 735         unsigned long end;
 736 
 737         if (pmd_none(*pmd))
 738                 return;
 739         if (pmd_bad(*pmd)) {
 740                 printk("statm_pte_range: bad pmd (%08lx)\n", pmd_val(*pmd));
 741                 pmd_clear(pmd);
 742                 return;
 743         }
 744         pte = pte_offset(pmd, address);
 745         address &= ~PMD_MASK;
 746         end = address + size;
 747         if (end > PMD_SIZE)
 748                 end = PMD_SIZE;
 749         do {
 750                 pte_t page = *pte;
 751 
 752                 address += PAGE_SIZE;
 753                 pte++;
 754                 if (pte_none(page))
 755                         continue;
 756                 ++*total;
 757                 if (!pte_present(page))
 758                         continue;
 759                 ++*pages;
 760                 if (pte_dirty(page))
 761                         ++*dirty;
 762                 if (pte_page(page) >= high_memory)
 763                         continue;
 764                 if (mem_map[MAP_NR(pte_page(page))].count > 1)
 765                         ++*shared;
 766         } while (address < end);
 767 }
 768 
 769 static inline void statm_pmd_range(pgd_t * pgd, unsigned long address, unsigned long size,
     /*  */
 770         int * pages, int * shared, int * dirty, int * total)
 771 {
 772         pmd_t * pmd;
 773         unsigned long end;
 774 
 775         if (pgd_none(*pgd))
 776                 return;
 777         if (pgd_bad(*pgd)) {
 778                 printk("statm_pmd_range: bad pgd (%08lx)\n", pgd_val(*pgd));
 779                 pgd_clear(pgd);
 780                 return;
 781         }
 782         pmd = pmd_offset(pgd, address);
 783         address &= ~PGDIR_MASK;
 784         end = address + size;
 785         if (end > PGDIR_SIZE)
 786                 end = PGDIR_SIZE;
 787         do {
 788                 statm_pte_range(pmd, address, end - address, pages, shared, dirty, total);
 789                 address = (address + PMD_SIZE) & PMD_MASK;
 790                 pmd++;
 791         } while (address < end);
 792 }
 793 
 794 static void statm_pgd_range(pgd_t * pgd, unsigned long address, unsigned long end,
     /*  */
 795         int * pages, int * shared, int * dirty, int * total)
 796 {
 797         while (address < end) {
 798                 statm_pmd_range(pgd, address, end - address, pages, shared, dirty, total);
 799                 address = (address + PGDIR_SIZE) & PGDIR_MASK;
 800                 pgd++;
 801         }
 802 }
 803 
 804 static int get_statm(int pid, char * buffer)
     /*  */
 805 {
 806         struct task_struct ** p = get_task(pid), *tsk;
 807         int size=0, resident=0, share=0, trs=0, lrs=0, drs=0, dt=0;
 808 
 809         if (!p || (tsk = *p) == NULL)
 810                 return 0;
 811         if (tsk->mm && tsk->mm != &init_mm) {
 812                 struct vm_area_struct * vma = tsk->mm->mmap;
 813 
 814                 while (vma) {
 815                         pgd_t *pgd = pgd_offset(tsk->mm, vma->vm_start);
 816                         int pages = 0, shared = 0, dirty = 0, total = 0;
 817 
 818                         statm_pgd_range(pgd, vma->vm_start, vma->vm_end, &pages, &shared, &dirty, &total);
 819                         resident += pages;
 820                         share += shared;
 821                         dt += dirty;
 822                         size += total;
 823                         if (vma->vm_flags & VM_EXECUTABLE)
 824                                 trs += pages;   /* text */
 825                         else if (vma->vm_flags & VM_GROWSDOWN)
 826                                 drs += pages;   /* stack */
 827                         else if (vma->vm_end > 0x60000000)
 828                                 lrs += pages;   /* library */
 829                         else
 830                                 drs += pages;
 831                         vma = vma->vm_next;
 832                 }
 833         }
 834         return sprintf(buffer,"%d %d %d %d %d %d %d\n",
 835                        size, resident, share, trs, lrs, drs, dt);
 836 }
 837 
 838 /*
 839  * The way we support synthetic files > 4K
 840  * - without storing their contents in some buffer and
 841  * - without walking through the entire synthetic file until we reach the
 842  *   position of the requested data
 843  * is to cleverly encode the current position in the file's f_pos field.
 844  * There is no requirement that a read() call which returns `count' bytes
 845  * of data increases f_pos by exactly `count'.
 846  *
 847  * This idea is Linus' one. Bruno implemented it.
 848  */
 849 
 850 /*
 851  * For the /proc/<pid>/maps file, we use fixed length records, each containing
 852  * a single line.
 853  */
 854 #define MAPS_LINE_LENGTH        1024
 855 #define MAPS_LINE_SHIFT         10
 856 /*
 857  * f_pos = (number of the vma in the task->mm->mmap list) * MAPS_LINE_LENGTH
 858  *         + (index into the line)
 859  */
 860 #define MAPS_LINE_FORMAT          "%08lx-%08lx %s %08lx %s %lu\n"
 861 #define MAPS_LINE_MAX   49 /* sum of 8  1  8  1 4 1 8 1 5 1 10 1 */
 862 
 863 static int read_maps (int pid, struct file * file, char * buf, int count)
     /*  */
 864 {
 865         struct task_struct ** p = get_task(pid);
 866         char * destptr;
 867         loff_t lineno;
 868         int column;
 869         struct vm_area_struct * map;
 870         int i;
 871 
 872         if (!p || !*p)
 873                 return -EINVAL;
 874 
 875         if (!(*p)->mm || (*p)->mm == &init_mm || count == 0)
 876                 return 0;
 877 
 878         /* decode f_pos */
 879         lineno = file->f_pos >> MAPS_LINE_SHIFT;
 880         column = file->f_pos & (MAPS_LINE_LENGTH-1);
 881 
 882         /* quickly go to line lineno */
 883         for (map = (*p)->mm->mmap, i = 0; map && (i < lineno); map = map->vm_next, i++)
 884                 continue;
 885 
 886         destptr = buf;
 887 
 888         for ( ; map ; ) {
 889                 /* produce the next line */
 890                 char line[MAPS_LINE_MAX+1];
 891                 char str[5], *cp = str;
 892                 int flags;
 893                 kdev_t dev;
 894                 unsigned long ino;
 895                 int len;
 896 
 897                 flags = map->vm_flags;
 898 
 899                 *cp++ = flags & VM_READ ? 'r' : '-';
 900                 *cp++ = flags & VM_WRITE ? 'w' : '-';
 901                 *cp++ = flags & VM_EXEC ? 'x' : '-';
 902                 *cp++ = flags & VM_MAYSHARE ? 's' : 'p';
 903                 *cp++ = 0;
 904 
 905                 if (map->vm_inode != NULL) {
 906                         dev = map->vm_inode->i_dev;
 907                         ino = map->vm_inode->i_ino;
 908                 } else {
 909                         dev = 0;
 910                         ino = 0;
 911                 }
 912 
 913                 len = sprintf(line, MAPS_LINE_FORMAT,
 914                               map->vm_start, map->vm_end, str, map->vm_offset,
 915                               kdevname(dev), ino);
 916 
 917                 if (column >= len) {
 918                         column = 0; /* continue with next line at column 0 */
 919                         lineno++;
 920                         map = map->vm_next;
 921                         continue;
 922                 }
 923 
 924                 i = len-column;
 925                 if (i > count)
 926                         i = count;
 927                 memcpy_tofs(destptr, line+column, i);
 928                 destptr += i; count -= i;
 929                 column += i;
 930                 if (column >= len) {
 931                         column = 0; /* next time: next line at column 0 */
 932                         lineno++;
 933                         map = map->vm_next;
 934                 }
 935 
 936                 /* done? */
 937                 if (count == 0)
 938                         break;
 939 
 940                 /* By writing to user space, we might have slept.
 941                  * Stop the loop, to avoid a race condition.
 942                  */
 943                 if (*p != current)
 944                         break;
 945         }
 946 
 947         /* encode f_pos */
 948         file->f_pos = (lineno << MAPS_LINE_SHIFT) + column;
 949 
 950         return destptr-buf;
 951 }
 952 
 953 #ifdef CONFIG_MODULES
 954 extern int get_module_list(char *);
 955 extern int get_ksyms_list(char *, char **, off_t, int);
 956 #endif
 957 extern int get_device_list(char *);
 958 extern int get_filesystem_list(char *);
 959 extern int get_filesystem_info( char * );
 960 extern int get_irq_list(char *);
 961 extern int get_dma_list(char *);
 962 extern int get_cpuinfo(char *);
 963 extern int get_pci_list(char*);
 964 extern int get_md_status (char *);
 965 #ifdef __SMP_PROF__
 966 extern int get_smp_prof_list(char *);
 967 #endif
 968 
 969 static int get_root_array(char * page, int type, char **start, off_t offset, int length)
     /*  */
 970 {
 971         switch (type) {
 972                 case PROC_LOADAVG:
 973                         return get_loadavg(page);
 974 
 975                 case PROC_UPTIME:
 976                         return get_uptime(page);
 977 
 978                 case PROC_MEMINFO:
 979                         return get_meminfo(page);
 980 
 981 #ifdef CONFIG_PCI
 982                 case PROC_PCI:
 983                         return get_pci_list(page);
 984 #endif
 985                         
 986                 case PROC_CPUINFO:
 987                         return get_cpuinfo(page);
 988 
 989                 case PROC_VERSION:
 990                         return get_version(page);
 991 
 992 #ifdef CONFIG_DEBUG_MALLOC
 993                 case PROC_MALLOC:
 994                         return get_malloc(page);
 995 #endif
 996 
 997 #ifdef CONFIG_MODULES
 998                 case PROC_MODULES:
 999                         return get_module_list(page);
1000 
1001                 case PROC_KSYMS:
1002                         return get_ksyms_list(page, start, offset, length);
1003 #endif
1004 
1005                 case PROC_STAT:
1006                         return get_kstat(page);
1007 
1008                 case PROC_DEVICES:
1009                         return get_device_list(page);
1010 
1011                 case PROC_INTERRUPTS:
1012                         return get_irq_list(page);
1013 
1014                 case PROC_FILESYSTEMS:
1015                         return get_filesystem_list(page);
1016 
1017                 case PROC_DMA:
1018                         return get_dma_list(page);
1019 
1020                 case PROC_IOPORTS:
1021                         return get_ioport_list(page);
1022 #ifdef CONFIG_BLK_DEV_MD
1023                 case PROC_MD:
1024                         return get_md_status(page);
1025 #endif
1026 #ifdef CONFIG_APM
1027                 case PROC_APM:
1028                         return apm_proc(page);
1029 #endif
1030 #ifdef __SMP_PROF__
1031                 case PROC_SMP_PROF:
1032                         return get_smp_prof_list(page);
1033 #endif
1034                 case PROC_CMDLINE:
1035                         return get_cmdline(page);
1036 
1037                 case PROC_MTAB:
1038                        return get_filesystem_info( page );
1039         }
1040         return -EBADF;
1041 }
1042 
1043 static int get_process_array(char * page, int pid, int type)
     /*  */
1044 {
1045         switch (type) {
1046                 case PROC_PID_STATUS:
1047                         return get_status(pid, page);
1048                 case PROC_PID_ENVIRON:
1049                         return get_env(pid, page);
1050                 case PROC_PID_CMDLINE:
1051                         return get_arg(pid, page);
1052                 case PROC_PID_STAT:
1053                         return get_stat(pid, page);
1054                 case PROC_PID_STATM:
1055                         return get_statm(pid, page);
1056         }
1057         return -EBADF;
1058 }
1059 
1060 
1061 static inline int fill_array(char * page, int pid, int type, char **start, off_t offset, int length)
     /*  */
1062 {
1063         if (pid)
1064                 return get_process_array(page, pid, type);
1065         return get_root_array(page, type, start, offset, length);
1066 }
1067 
1068 #define PROC_BLOCK_SIZE (3*1024)                /* 4K page size but our output routines use some slack for overruns */
1069 
1070 static int array_read(struct inode * inode, struct file * file,char * buf, int count)
     /*  */
1071 {
1072         unsigned long page;
1073         char *start;
1074         int length;
1075         int end;
1076         unsigned int type, pid;
1077 
1078         if (count < 0)
1079                 return -EINVAL;
1080         if (count > PROC_BLOCK_SIZE)
1081                 count = PROC_BLOCK_SIZE;
1082         if (!(page = __get_free_page(GFP_KERNEL)))
1083                 return -ENOMEM;
1084         type = inode->i_ino;
1085         pid = type >> 16;
1086         type &= 0x0000ffff;
1087         start = NULL;
1088         length = fill_array((char *) page, pid, type,
1089                             &start, file->f_pos, count);
1090         if (length < 0) {
1091                 free_page(page);
1092                 return length;
1093         }
1094         if (start != NULL) {
1095                 /* We have had block-adjusting processing! */
1096                 memcpy_tofs(buf, start, length);
1097                 file->f_pos += length;
1098                 count = length;
1099         } else {
1100                 /* Static 4kB (or whatever) block capacity */
1101                 if (file->f_pos >= length) {
1102                         free_page(page);
1103                         return 0;
1104                 }
1105                 if (count + file->f_pos > length)
1106                         count = length - file->f_pos;
1107                 end = count + file->f_pos;
1108                 memcpy_tofs(buf, (char *) page + file->f_pos, count);
1109                 file->f_pos = end;
1110         }
1111         free_page(page);
1112         return count;
1113 }
1114 
1115 static struct file_operations proc_array_operations = {
1116         NULL,           /* array_lseek */
1117         array_read,
1118         NULL,           /* array_write */
1119         NULL,           /* array_readdir */
1120         NULL,           /* array_select */
1121         NULL,           /* array_ioctl */
1122         NULL,           /* mmap */
1123         NULL,           /* no special open code */
1124         NULL,           /* no special release code */
1125         NULL            /* can't fsync */
1126 };
1127 
1128 struct inode_operations proc_array_inode_operations = {
1129         &proc_array_operations, /* default base directory file-ops */
1130         NULL,                   /* create */
1131         NULL,                   /* lookup */
1132         NULL,                   /* link */
1133         NULL,                   /* unlink */
1134         NULL,                   /* symlink */
1135         NULL,                   /* mkdir */
1136         NULL,                   /* rmdir */
1137         NULL,                   /* mknod */
1138         NULL,                   /* rename */
1139         NULL,                   /* readlink */
1140         NULL,                   /* follow_link */
1141         NULL,                   /* readpage */
1142         NULL,                   /* writepage */
1143         NULL,                   /* bmap */
1144         NULL,                   /* truncate */
1145         NULL                    /* permission */
1146 };
1147 
1148 static int arraylong_read (struct inode * inode, struct file * file, char * buf, int count)
     /*  */
1149 {
1150         unsigned int pid = inode->i_ino >> 16;
1151         unsigned int type = inode->i_ino & 0x0000ffff;
1152 
1153         if (count < 0)
1154                 return -EINVAL;
1155 
1156         switch (type) {
1157                 case PROC_PID_MAPS:
1158                         return read_maps(pid, file, buf, count);
1159         }
1160         return -EINVAL;
1161 }
1162 
1163 static struct file_operations proc_arraylong_operations = {
1164         NULL,           /* array_lseek */
1165         arraylong_read,
1166         NULL,           /* array_write */
1167         NULL,           /* array_readdir */
1168         NULL,           /* array_select */
1169         NULL,           /* array_ioctl */
1170         NULL,           /* mmap */
1171         NULL,           /* no special open code */
1172         NULL,           /* no special release code */
1173         NULL            /* can't fsync */
1174 };
1175 
1176 struct inode_operations proc_arraylong_inode_operations = {
1177         &proc_arraylong_operations,     /* default base directory file-ops */
1178         NULL,                   /* create */
1179         NULL,                   /* lookup */
1180         NULL,                   /* link */
1181         NULL,                   /* unlink */
1182         NULL,                   /* symlink */
1183         NULL,                   /* mkdir */
1184         NULL,                   /* rmdir */
1185         NULL,                   /* mknod */
1186         NULL,                   /* rename */
1187         NULL,                   /* readlink */
1188         NULL,                   /* follow_link */
1189         NULL,                   /* readpage */
1190         NULL,                   /* writepage */
1191         NULL,                   /* bmap */
1192         NULL,                   /* truncate */
1193         NULL                    /* permission */
1194 };
/* */
root/fs/proc/array.c

DEFINITIONS
