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 #include <linux/swap.h>
  46 
  47 #include <asm/segment.h>
  48 #include <asm/pgtable.h>
  49 #include <asm/io.h>
  50 
  51 #define LOAD_INT(x) ((x) >> FSHIFT)
  52 #define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1-1)) * 100)
  53 
  54 #ifdef CONFIG_DEBUG_MALLOC
  55 int get_malloc(char * buffer);
  56 #endif
  57 
  58 
  59 static int read_core(struct inode * inode, struct file * file,char * buf, int count)
     /*  */
  60 {
  61         unsigned long p = file->f_pos, memsize;
  62         int read;
  63         int count1;
  64         char * pnt;
  65         struct user dump;
  66 #ifdef __i386__
  67 #       define FIRST_MAPPED     PAGE_SIZE       /* we don't have page 0 mapped on x86.. */
  68 #else
  69 #       define FIRST_MAPPED     0
  70 #endif
  71 
  72         memset(&dump, 0, sizeof(struct user));
  73         dump.magic = CMAGIC;
  74         dump.u_dsize = MAP_NR(high_memory);
  75 #ifdef __alpha__
  76         dump.start_data = PAGE_OFFSET;
  77 #endif
  78 
  79         if (count < 0)
  80                 return -EINVAL;
  81         memsize = MAP_NR(high_memory + PAGE_SIZE) << PAGE_SHIFT;
  82         if (p >= memsize)
  83                 return 0;
  84         if (count > memsize - p)
  85                 count = memsize - p;
  86         read = 0;
  87 
  88         if (p < sizeof(struct user) && count > 0) {
  89                 count1 = count;
  90                 if (p + count1 > sizeof(struct user))
  91                         count1 = sizeof(struct user)-p;
  92                 pnt = (char *) &dump + p;
  93                 memcpy_tofs(buf,(void *) pnt, count1);
  94                 buf += count1;
  95                 p += count1;
  96                 count -= count1;
  97                 read += count1;
  98         }
  99 
 100         while (count > 0 && p < PAGE_SIZE + FIRST_MAPPED) {
 101                 put_user(0,buf);
 102                 buf++;
 103                 p++;
 104                 count--;
 105                 read++;
 106         }
 107         memcpy_tofs(buf, (void *) (PAGE_OFFSET + p - PAGE_SIZE), count);
 108         read += count;
 109         file->f_pos += read;
 110         return read;
 111 }
 112 
 113 static struct file_operations proc_kcore_operations = {
 114         NULL,           /* lseek */
 115         read_core,
 116 };
 117 
 118 struct inode_operations proc_kcore_inode_operations = {
 119         &proc_kcore_operations, 
 120 };
 121 
 122 
 123 /*
 124  * This function accesses profiling information. The returned data is
 125  * binary: the sampling step and the actual contents of the profile
 126  * buffer. Use of the program readprofile is recommended in order to
 127  * get meaningful info out of these data.
 128  */
 129 static int read_profile(struct inode *inode, struct file *file, char *buf, int count)
     /*  */
 130 {
 131         unsigned long p = file->f_pos;
 132         int read;
 133         char * pnt;
 134         unsigned int sample_step = 1 << prof_shift;
 135 
 136         if (count < 0)
 137                 return -EINVAL;
 138         if (p >= (prof_len+1)*sizeof(unsigned int))
 139                 return 0;
 140         if (count > (prof_len+1)*sizeof(unsigned int) - p)
 141                 count = (prof_len+1)*sizeof(unsigned int) - p;
 142         read = 0;
 143 
 144         while (p < sizeof(unsigned int) && count > 0) {
 145                 put_user(*((char *)(&sample_step)+p),buf);
 146                 buf++; p++; count--; read++;
 147         }
 148         pnt = (char *)prof_buffer + p - sizeof(unsigned int);
 149         memcpy_tofs(buf,(void *)pnt,count);
 150         read += count;
 151         file->f_pos += read;
 152         return read;
 153 }
 154 
 155 /* Writing to /proc/profile resets the counters */
 156 static int write_profile(struct inode * inode, struct file * file, const char * buf, int count)
     /*  */
 157 {
 158     int i=prof_len;
 159 
 160     while (i--)
 161             prof_buffer[i]=0UL;
 162     return count;
 163 }
 164 
 165 static struct file_operations proc_profile_operations = {
 166         NULL,           /* lseek */
 167         read_profile,
 168         write_profile,
 169 };
 170 
 171 struct inode_operations proc_profile_inode_operations = {
 172         &proc_profile_operations, 
 173 };
 174 
 175 
 176 static int get_loadavg(char * buffer)
     /*  */
 177 {
 178         int a, b, c;
 179 
 180         a = avenrun[0] + (FIXED_1/200);
 181         b = avenrun[1] + (FIXED_1/200);
 182         c = avenrun[2] + (FIXED_1/200);
 183         return sprintf(buffer,"%d.%02d %d.%02d %d.%02d %d/%d %d\n",
 184                 LOAD_INT(a), LOAD_FRAC(a),
 185                 LOAD_INT(b), LOAD_FRAC(b),
 186                 LOAD_INT(c), LOAD_FRAC(c),
 187                 nr_running, nr_tasks, last_pid);
 188 }
 189 
 190 static int get_kstat(char * buffer)
     /*  */
 191 {
 192         int i, len;
 193         unsigned sum = 0;
 194         extern unsigned long total_forks;
 195 
 196         for (i = 0 ; i < NR_IRQS ; i++)
 197                 sum += kstat.interrupts[i];
 198         len = sprintf(buffer,
 199                 "cpu  %u %u %u %lu\n"
 200                 "disk %u %u %u %u\n"
 201                 "disk_rio %u %u %u %u\n"
 202                 "disk_wio %u %u %u %u\n"
 203                 "disk_rblk %u %u %u %u\n"
 204                 "disk_wblk %u %u %u %u\n"
 205                 "page %u %u\n"
 206                 "swap %u %u\n"
 207                 "intr %u",
 208                 kstat.cpu_user,
 209                 kstat.cpu_nice,
 210                 kstat.cpu_system,
 211                 jiffies - (kstat.cpu_user + kstat.cpu_nice + kstat.cpu_system),
 212                 kstat.dk_drive[0], kstat.dk_drive[1],
 213                 kstat.dk_drive[2], kstat.dk_drive[3],
 214                 kstat.dk_drive_rio[0], kstat.dk_drive_rio[1],
 215                 kstat.dk_drive_rio[2], kstat.dk_drive_rio[3],
 216                 kstat.dk_drive_wio[0], kstat.dk_drive_wio[1],
 217                 kstat.dk_drive_wio[2], kstat.dk_drive_wio[3],
 218                 kstat.dk_drive_rblk[0], kstat.dk_drive_rblk[1],
 219                 kstat.dk_drive_rblk[2], kstat.dk_drive_rblk[3],
 220                 kstat.dk_drive_wblk[0], kstat.dk_drive_wblk[1],
 221                 kstat.dk_drive_wblk[2], kstat.dk_drive_wblk[3],
 222                 kstat.pgpgin,
 223                 kstat.pgpgout,
 224                 kstat.pswpin,
 225                 kstat.pswpout,
 226                 sum);
 227         for (i = 0 ; i < NR_IRQS ; i++)
 228                 len += sprintf(buffer + len, " %u", kstat.interrupts[i]);
 229         len += sprintf(buffer + len,
 230                 "\nctxt %u\n"
 231                 "btime %lu\n"
 232                 "processes %lu\n",
 233                 kstat.context_swtch,
 234                 xtime.tv_sec - jiffies / HZ,
 235                 total_forks);
 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 #elif defined(__sparc__)
 474 # define PT_REG(reg)            (PAGE_SIZE - sizeof(struct pt_regs)     \
 475                                  + (long)&((struct pt_regs *)0)->reg)
 476 # define KSTK_EIP(tsk)  (*(unsigned long *)(tsk->kernel_stack_page + PT_REG(pc)))
 477 # define KSTK_ESP(tsk)  (*(unsigned long *)(tsk->kernel_stack_page + PT_REG(u_regs[UREG_FP])))
 478 #endif
 479 
 480 /* Gcc optimizes away "strlen(x)" for constant x */
 481 #define ADDBUF(buffer, string) \
 482 do { memcpy(buffer, string, strlen(string)); \
 483      buffer += strlen(string); } while (0)
 484 
 485 static inline char * task_name(struct task_struct *p, char * buf)
     /*  */
 486 {
 487         int i;
 488         char * name;
 489 
 490         ADDBUF(buf, "Name:\t");
 491         name = p->comm;
 492         i = sizeof(p->comm);
 493         do {
 494                 unsigned char c = *name;
 495                 name++;
 496                 i--;
 497                 *buf = c;
 498                 if (!c)
 499                         break;
 500                 if (c == '\\') {
 501                         buf[1] = c;
 502                         buf += 2;
 503                         continue;
 504                 }
 505                 if (c == '\n') {
 506                         buf[0] = '\\';
 507                         buf[1] = 'n';
 508                         buf += 2;
 509                         continue;
 510                 }
 511                 buf++;
 512         } while (i);
 513         *buf = '\n';
 514         return buf+1;
 515 }
 516 
 517 static inline char * task_state(struct task_struct *p, char *buffer)
     /*  */
 518 {
 519 #define NR_STATES (sizeof(states)/sizeof(const char *))
 520         unsigned int n = p->state;
 521         static const char * states[] = {
 522                 "R (running)",
 523                 "S (sleeping)",
 524                 "D (disk sleep)",
 525                 "Z (zombie)",
 526                 "T (stopped)",
 527                 "W (paging)",
 528                 ". Huh?"
 529         };
 530 
 531         if (n >= NR_STATES)
 532                 n = NR_STATES-1;
 533 
 534         buffer += sprintf(buffer,
 535                 "State:\t%s\n"
 536                 "Pid:\t%d\n"
 537                 "PPid:\t%d\n"
 538                 "Uid:\t%d\t%d\t%d\t%d\n"
 539                 "Gid:\t%d\t%d\t%d\t%d\n",
 540                 states[n],
 541                 p->pid, p->p_pptr->pid,
 542                 p->uid, p->euid, p->suid, p->fsuid,
 543                 p->gid, p->egid, p->sgid, p->fsgid);
 544         return buffer;
 545 }
 546 
 547 static inline char * task_mem(struct task_struct *p, char *buffer)
     /*  */
 548 {
 549         struct mm_struct * mm = p->mm;
 550 
 551         if (mm && mm != &init_mm) {
 552                 struct vm_area_struct * vma = mm->mmap;
 553                 unsigned long data = 0, stack = 0;
 554                 unsigned long exec = 0, lib = 0;
 555 
 556                 for (vma = mm->mmap; vma; vma = vma->vm_next) {
 557                         unsigned long len = (vma->vm_end - vma->vm_start) >> 10;
 558                         if (!vma->vm_inode) {
 559                                 data += len;
 560                                 if (vma->vm_flags & VM_GROWSDOWN)
 561                                         stack += len;
 562                                 continue;
 563                         }
 564                         if (vma->vm_flags & VM_WRITE)
 565                                 continue;
 566                         if (vma->vm_flags & VM_EXEC) {
 567                                 exec += len;
 568                                 if (vma->vm_flags & VM_EXECUTABLE)
 569                                         continue;
 570                                 lib += len;
 571                         }
 572                 }       
 573                 buffer += sprintf(buffer,
 574                         "VmSize:\t%8lu kB\n"
 575                         "VmLck:\t%8lu kB\n"
 576                         "VmRSS:\t%8lu kB\n"
 577                         "VmData:\t%8lu kB\n"
 578                         "VmStk:\t%8lu kB\n"
 579                         "VmExe:\t%8lu kB\n"
 580                         "VmLib:\t%8lu kB\n",
 581                         mm->total_vm << (PAGE_SHIFT-10),
 582                         mm->locked_vm << (PAGE_SHIFT-10),
 583                         mm->rss << (PAGE_SHIFT-10),
 584                         data - stack, stack,
 585                         exec - lib, lib);
 586         }
 587         return buffer;
 588 }
 589 
 590 static inline char * task_sig(struct task_struct *p, char *buffer)
     /*  */
 591 {
 592         buffer += sprintf(buffer,
 593                 "SigPnd:\t%08lx\n"
 594                 "SigBlk:\t%08lx\n",
 595                 p->signal, p->blocked);
 596         if (p->sig) {
 597                 struct sigaction * action = p->sig->action;
 598                 unsigned long sig_ign = 0, sig_caught = 0;
 599                 unsigned long bit = 1;
 600                 int i;
 601 
 602                 for (i = 0; i < 32; i++) {
 603                         switch((unsigned long) action->sa_handler) {
 604                                 case 0:
 605                                         break;
 606                                 case 1:
 607                                         sig_ign |= bit;
 608                                         break;
 609                                 default:
 610                                         sig_caught |= bit;
 611                         }
 612                         bit <<= 1;
 613                         action++;
 614                 }
 615 
 616                 buffer += sprintf(buffer,
 617                         "SigIgn:\t%08lx\n"
 618                         "SigCgt:\t%08lx\n",
 619                         sig_ign, sig_caught);
 620         }
 621         return buffer;
 622 }
 623 
 624 static int get_status(int pid, char * buffer)
     /*  */
 625 {
 626         char * orig = buffer;
 627         struct task_struct ** p = get_task(pid), *tsk;
 628 
 629         if (!p || (tsk = *p) == NULL)
 630                 return 0;
 631         buffer = task_name(tsk, buffer);
 632         buffer = task_state(tsk, buffer);
 633         buffer = task_mem(tsk, buffer);
 634         buffer = task_sig(tsk, buffer);
 635         return buffer - orig;
 636 }
 637 
 638 static int get_stat(int pid, char * buffer)
     /*  */
 639 {
 640         struct task_struct ** p = get_task(pid), *tsk;
 641         unsigned long sigignore=0, sigcatch=0, wchan;
 642         unsigned long vsize, eip, esp;
 643         long priority, nice;
 644         int i,tty_pgrp;
 645         char state;
 646 
 647         if (!p || (tsk = *p) == NULL)
 648                 return 0;
 649         if (tsk->state < 0 || tsk->state > 5)
 650                 state = '.';
 651         else
 652                 state = "RSDZTW"[tsk->state];
 653         vsize = eip = esp = 0;
 654         if (tsk->mm && tsk->mm != &init_mm) {
 655                 struct vm_area_struct *vma = tsk->mm->mmap;
 656                 while (vma) {
 657                         vsize += vma->vm_end - vma->vm_start;
 658                         vma = vma->vm_next;
 659                 }
 660                 if (tsk->kernel_stack_page) {
 661                         eip = KSTK_EIP(tsk);
 662                         esp = KSTK_ESP(tsk);
 663                 }
 664         }
 665         wchan = get_wchan(tsk);
 666         if (tsk->sig) {
 667                 unsigned long bit = 1;
 668                 for(i=0; i<32; ++i) {
 669                         switch((unsigned long) tsk->sig->action[i].sa_handler) {
 670                                 case 0:
 671                                         break;
 672                                 case 1:
 673                                         sigignore |= bit;
 674                                         break;
 675                                 default:
 676                                         sigcatch |= bit;
 677                         }
 678                         bit <<= 1;
 679                 }
 680         }
 681         if (tsk->tty)
 682                 tty_pgrp = tsk->tty->pgrp;
 683         else
 684                 tty_pgrp = -1;
 685 
 686         /* scale priority and nice values from timeslices to -20..20 */
 687         /* to make it look like a "normal" unix priority/nice value  */
 688         priority = tsk->counter;
 689         priority = 20 - (priority * 10 + DEF_PRIORITY / 2) / DEF_PRIORITY;
 690         nice = tsk->priority;
 691         nice = 20 - (nice * 20 + DEF_PRIORITY / 2) / DEF_PRIORITY;
 692 
 693         return sprintf(buffer,"%d (%s) %c %d %d %d %d %d %lu %lu \
 694 %lu %lu %lu %lu %lu %ld %ld %ld %ld %ld %ld %lu %lu %ld %lu %lu %lu %lu %lu \
 695 %lu %lu %lu %lu %lu %lu %lu %lu\n",
 696                 pid,
 697                 tsk->comm,
 698                 state,
 699                 tsk->p_pptr->pid,
 700                 tsk->pgrp,
 701                 tsk->session,
 702                 tsk->tty ? kdev_t_to_nr(tsk->tty->device) : 0,
 703                 tty_pgrp,
 704                 tsk->flags,
 705                 tsk->min_flt,
 706                 tsk->cmin_flt,
 707                 tsk->maj_flt,
 708                 tsk->cmaj_flt,
 709                 tsk->utime,
 710                 tsk->stime,
 711                 tsk->cutime,
 712                 tsk->cstime,
 713                 priority,
 714                 nice,
 715                 tsk->timeout,
 716                 tsk->it_real_value,
 717                 tsk->start_time,
 718                 vsize,
 719                 tsk->mm ? tsk->mm->rss : 0, /* you might want to shift this left 3 */
 720                 tsk->rlim ? tsk->rlim[RLIMIT_RSS].rlim_cur : 0,
 721                 tsk->mm ? tsk->mm->start_code : 0,
 722                 tsk->mm ? tsk->mm->end_code : 0,
 723                 tsk->mm ? tsk->mm->start_stack : 0,
 724                 esp,
 725                 eip,
 726                 tsk->signal,
 727                 tsk->blocked,
 728                 sigignore,
 729                 sigcatch,
 730                 wchan,
 731                 tsk->nswap,
 732                 tsk->cnswap);
 733 }
 734                 
 735 static inline void statm_pte_range(pmd_t * pmd, unsigned long address, unsigned long size,
     /*  */
 736         int * pages, int * shared, int * dirty, int * total)
 737 {
 738         pte_t * pte;
 739         unsigned long end;
 740 
 741         if (pmd_none(*pmd))
 742                 return;
 743         if (pmd_bad(*pmd)) {
 744                 printk("statm_pte_range: bad pmd (%08lx)\n", pmd_val(*pmd));
 745                 pmd_clear(pmd);
 746                 return;
 747         }
 748         pte = pte_offset(pmd, address);
 749         address &= ~PMD_MASK;
 750         end = address + size;
 751         if (end > PMD_SIZE)
 752                 end = PMD_SIZE;
 753         do {
 754                 pte_t page = *pte;
 755 
 756                 address += PAGE_SIZE;
 757                 pte++;
 758                 if (pte_none(page))
 759                         continue;
 760                 ++*total;
 761                 if (!pte_present(page))
 762                         continue;
 763                 ++*pages;
 764                 if (pte_dirty(page))
 765                         ++*dirty;
 766                 if (pte_page(page) >= high_memory)
 767                         continue;
 768                 if (mem_map[MAP_NR(pte_page(page))].count > 1)
 769                         ++*shared;
 770         } while (address < end);
 771 }
 772 
 773 static inline void statm_pmd_range(pgd_t * pgd, unsigned long address, unsigned long size,
     /*  */
 774         int * pages, int * shared, int * dirty, int * total)
 775 {
 776         pmd_t * pmd;
 777         unsigned long end;
 778 
 779         if (pgd_none(*pgd))
 780                 return;
 781         if (pgd_bad(*pgd)) {
 782                 printk("statm_pmd_range: bad pgd (%08lx)\n", pgd_val(*pgd));
 783                 pgd_clear(pgd);
 784                 return;
 785         }
 786         pmd = pmd_offset(pgd, address);
 787         address &= ~PGDIR_MASK;
 788         end = address + size;
 789         if (end > PGDIR_SIZE)
 790                 end = PGDIR_SIZE;
 791         do {
 792                 statm_pte_range(pmd, address, end - address, pages, shared, dirty, total);
 793                 address = (address + PMD_SIZE) & PMD_MASK;
 794                 pmd++;
 795         } while (address < end);
 796 }
 797 
 798 static void statm_pgd_range(pgd_t * pgd, unsigned long address, unsigned long end,
     /*  */
 799         int * pages, int * shared, int * dirty, int * total)
 800 {
 801         while (address < end) {
 802                 statm_pmd_range(pgd, address, end - address, pages, shared, dirty, total);
 803                 address = (address + PGDIR_SIZE) & PGDIR_MASK;
 804                 pgd++;
 805         }
 806 }
 807 
 808 static int get_statm(int pid, char * buffer)
     /*  */
 809 {
 810         struct task_struct ** p = get_task(pid), *tsk;
 811         int size=0, resident=0, share=0, trs=0, lrs=0, drs=0, dt=0;
 812 
 813         if (!p || (tsk = *p) == NULL)
 814                 return 0;
 815         if (tsk->mm && tsk->mm != &init_mm) {
 816                 struct vm_area_struct * vma = tsk->mm->mmap;
 817 
 818                 while (vma) {
 819                         pgd_t *pgd = pgd_offset(tsk->mm, vma->vm_start);
 820                         int pages = 0, shared = 0, dirty = 0, total = 0;
 821 
 822                         statm_pgd_range(pgd, vma->vm_start, vma->vm_end, &pages, &shared, &dirty, &total);
 823                         resident += pages;
 824                         share += shared;
 825                         dt += dirty;
 826                         size += total;
 827                         if (vma->vm_flags & VM_EXECUTABLE)
 828                                 trs += pages;   /* text */
 829                         else if (vma->vm_flags & VM_GROWSDOWN)
 830                                 drs += pages;   /* stack */
 831                         else if (vma->vm_end > 0x60000000)
 832                                 lrs += pages;   /* library */
 833                         else
 834                                 drs += pages;
 835                         vma = vma->vm_next;
 836                 }
 837         }
 838         return sprintf(buffer,"%d %d %d %d %d %d %d\n",
 839                        size, resident, share, trs, lrs, drs, dt);
 840 }
 841 
 842 /*
 843  * The way we support synthetic files > 4K
 844  * - without storing their contents in some buffer and
 845  * - without walking through the entire synthetic file until we reach the
 846  *   position of the requested data
 847  * is to cleverly encode the current position in the file's f_pos field.
 848  * There is no requirement that a read() call which returns `count' bytes
 849  * of data increases f_pos by exactly `count'.
 850  *
 851  * This idea is Linus' one. Bruno implemented it.
 852  */
 853 
 854 /*
 855  * For the /proc/<pid>/maps file, we use fixed length records, each containing
 856  * a single line.
 857  */
 858 #define MAPS_LINE_LENGTH        1024
 859 #define MAPS_LINE_SHIFT         10
 860 /*
 861  * f_pos = (number of the vma in the task->mm->mmap list) * MAPS_LINE_LENGTH
 862  *         + (index into the line)
 863  */
 864 #define MAPS_LINE_FORMAT          "%08lx-%08lx %s %08lx %s %lu\n"
 865 #define MAPS_LINE_MAX   49 /* sum of 8  1  8  1 4 1 8 1 5 1 10 1 */
 866 
 867 static int read_maps (int pid, struct file * file, char * buf, int count)
     /*  */
 868 {
 869         struct task_struct ** p = get_task(pid);
 870         char * destptr;
 871         loff_t lineno;
 872         int column;
 873         struct vm_area_struct * map;
 874         int i;
 875 
 876         if (!p || !*p)
 877                 return -EINVAL;
 878 
 879         if (!(*p)->mm || (*p)->mm == &init_mm || count == 0)
 880                 return 0;
 881 
 882         /* decode f_pos */
 883         lineno = file->f_pos >> MAPS_LINE_SHIFT;
 884         column = file->f_pos & (MAPS_LINE_LENGTH-1);
 885 
 886         /* quickly go to line lineno */
 887         for (map = (*p)->mm->mmap, i = 0; map && (i < lineno); map = map->vm_next, i++)
 888                 continue;
 889 
 890         destptr = buf;
 891 
 892         for ( ; map ; ) {
 893                 /* produce the next line */
 894                 char line[MAPS_LINE_MAX+1];
 895                 char str[5], *cp = str;
 896                 int flags;
 897                 kdev_t dev;
 898                 unsigned long ino;
 899                 int len;
 900 
 901                 flags = map->vm_flags;
 902 
 903                 *cp++ = flags & VM_READ ? 'r' : '-';
 904                 *cp++ = flags & VM_WRITE ? 'w' : '-';
 905                 *cp++ = flags & VM_EXEC ? 'x' : '-';
 906                 *cp++ = flags & VM_MAYSHARE ? 's' : 'p';
 907                 *cp++ = 0;
 908 
 909                 if (map->vm_inode != NULL) {
 910                         dev = map->vm_inode->i_dev;
 911                         ino = map->vm_inode->i_ino;
 912                 } else {
 913                         dev = 0;
 914                         ino = 0;
 915                 }
 916 
 917                 len = sprintf(line, MAPS_LINE_FORMAT,
 918                               map->vm_start, map->vm_end, str, map->vm_offset,
 919                               kdevname(dev), ino);
 920 
 921                 if (column >= len) {
 922                         column = 0; /* continue with next line at column 0 */
 923                         lineno++;
 924                         map = map->vm_next;
 925                         continue;
 926                 }
 927 
 928                 i = len-column;
 929                 if (i > count)
 930                         i = count;
 931                 memcpy_tofs(destptr, line+column, i);
 932                 destptr += i; count -= i;
 933                 column += i;
 934                 if (column >= len) {
 935                         column = 0; /* next time: next line at column 0 */
 936                         lineno++;
 937                         map = map->vm_next;
 938                 }
 939 
 940                 /* done? */
 941                 if (count == 0)
 942                         break;
 943 
 944                 /* By writing to user space, we might have slept.
 945                  * Stop the loop, to avoid a race condition.
 946                  */
 947                 if (*p != current)
 948                         break;
 949         }
 950 
 951         /* encode f_pos */
 952         file->f_pos = (lineno << MAPS_LINE_SHIFT) + column;
 953 
 954         return destptr-buf;
 955 }
 956 
 957 #ifdef CONFIG_MODULES
 958 extern int get_module_list(char *);
 959 extern int get_ksyms_list(char *, char **, off_t, int);
 960 #endif
 961 extern int get_device_list(char *);
 962 extern int get_filesystem_list(char *);
 963 extern int get_filesystem_info( char * );
 964 extern int get_irq_list(char *);
 965 extern int get_dma_list(char *);
 966 extern int get_cpuinfo(char *);
 967 extern int get_pci_list(char*);
 968 extern int get_md_status (char *);
 969 extern int get_rtc_status (char *);
 970 extern int get_locks_status (char *);
 971 #ifdef __SMP_PROF__
 972 extern int get_smp_prof_list(char *);
 973 #endif
 974 
 975 static int get_root_array(char * page, int type, char **start, off_t offset, int length)
     /*  */
 976 {
 977         switch (type) {
 978                 case PROC_LOADAVG:
 979                         return get_loadavg(page);
 980 
 981                 case PROC_UPTIME:
 982                         return get_uptime(page);
 983 
 984                 case PROC_MEMINFO:
 985                         return get_meminfo(page);
 986 
 987 #ifdef CONFIG_PCI
 988                 case PROC_PCI:
 989                         return get_pci_list(page);
 990 #endif
 991                         
 992                 case PROC_CPUINFO:
 993                         return get_cpuinfo(page);
 994 
 995                 case PROC_VERSION:
 996                         return get_version(page);
 997 
 998 #ifdef CONFIG_DEBUG_MALLOC
 999                 case PROC_MALLOC:
1000                         return get_malloc(page);
1001 #endif
1002 
1003 #ifdef CONFIG_MODULES
1004                 case PROC_MODULES:
1005                         return get_module_list(page);
1006 
1007                 case PROC_KSYMS:
1008                         return get_ksyms_list(page, start, offset, length);
1009 #endif
1010 
1011                 case PROC_STAT:
1012                         return get_kstat(page);
1013 
1014                 case PROC_DEVICES:
1015                         return get_device_list(page);
1016 
1017                 case PROC_INTERRUPTS:
1018                         return get_irq_list(page);
1019 
1020                 case PROC_FILESYSTEMS:
1021                         return get_filesystem_list(page);
1022 
1023                 case PROC_DMA:
1024                         return get_dma_list(page);
1025 
1026                 case PROC_IOPORTS:
1027                         return get_ioport_list(page);
1028 #ifdef CONFIG_BLK_DEV_MD
1029                 case PROC_MD:
1030                         return get_md_status(page);
1031 #endif
1032 #ifdef __SMP_PROF__
1033                 case PROC_SMP_PROF:
1034                         return get_smp_prof_list(page);
1035 #endif
1036                 case PROC_CMDLINE:
1037                         return get_cmdline(page);
1038 
1039                 case PROC_MTAB:
1040                        return get_filesystem_info( page );
1041 #ifdef CONFIG_RTC
1042                 case PROC_RTC:
1043                         return get_rtc_status(page);
1044 #endif
1045                 case PROC_LOCKS:
1046                         return get_locks_status(page);
1047         }
1048         return -EBADF;
1049 }
1050 
1051 static int get_process_array(char * page, int pid, int type)
     /*  */
1052 {
1053         switch (type) {
1054                 case PROC_PID_STATUS:
1055                         return get_status(pid, page);
1056                 case PROC_PID_ENVIRON:
1057                         return get_env(pid, page);
1058                 case PROC_PID_CMDLINE:
1059                         return get_arg(pid, page);
1060                 case PROC_PID_STAT:
1061                         return get_stat(pid, page);
1062                 case PROC_PID_STATM:
1063                         return get_statm(pid, page);
1064         }
1065         return -EBADF;
1066 }
1067 
1068 
1069 static inline int fill_array(char * page, int pid, int type, char **start, off_t offset, int length)
     /*  */
1070 {
1071         if (pid)
1072                 return get_process_array(page, pid, type);
1073         return get_root_array(page, type, start, offset, length);
1074 }
1075 
1076 #define PROC_BLOCK_SIZE (3*1024)                /* 4K page size but our output routines use some slack for overruns */
1077 
1078 static int array_read(struct inode * inode, struct file * file,char * buf, int count)
     /*  */
1079 {
1080         unsigned long page;
1081         char *start;
1082         int length;
1083         int end;
1084         unsigned int type, pid;
1085         struct proc_dir_entry *dp;
1086 
1087         if (count < 0)
1088                 return -EINVAL;
1089         if (count > PROC_BLOCK_SIZE)
1090                 count = PROC_BLOCK_SIZE;
1091         if (!(page = __get_free_page(GFP_KERNEL)))
1092                 return -ENOMEM;
1093         type = inode->i_ino;
1094         pid = type >> 16;
1095         type &= 0x0000ffff;
1096         start = NULL;
1097         dp = (struct proc_dir_entry *) inode->u.generic_ip;
1098         if (dp->get_info)
1099                 length = dp->get_info((char *)page, &start, file->f_pos,
1100                                       count, 0);
1101         else
1102                 length = fill_array((char *) page, pid, type,
1103                                     &start, file->f_pos, count);
1104         if (length < 0) {
1105                 free_page(page);
1106                 return length;
1107         }
1108         if (start != NULL) {
1109                 /* We have had block-adjusting processing! */
1110                 memcpy_tofs(buf, start, length);
1111                 file->f_pos += length;
1112                 count = length;
1113         } else {
1114                 /* Static 4kB (or whatever) block capacity */
1115                 if (file->f_pos >= length) {
1116                         free_page(page);
1117                         return 0;
1118                 }
1119                 if (count + file->f_pos > length)
1120                         count = length - file->f_pos;
1121                 end = count + file->f_pos;
1122                 memcpy_tofs(buf, (char *) page + file->f_pos, count);
1123                 file->f_pos = end;
1124         }
1125         free_page(page);
1126         return count;
1127 }
1128 
1129 static struct file_operations proc_array_operations = {
1130         NULL,           /* array_lseek */
1131         array_read,
1132         NULL,           /* array_write */
1133         NULL,           /* array_readdir */
1134         NULL,           /* array_select */
1135         NULL,           /* array_ioctl */
1136         NULL,           /* mmap */
1137         NULL,           /* no special open code */
1138         NULL,           /* no special release code */
1139         NULL            /* can't fsync */
1140 };
1141 
1142 struct inode_operations proc_array_inode_operations = {
1143         &proc_array_operations, /* default base directory file-ops */
1144         NULL,                   /* create */
1145         NULL,                   /* lookup */
1146         NULL,                   /* link */
1147         NULL,                   /* unlink */
1148         NULL,                   /* symlink */
1149         NULL,                   /* mkdir */
1150         NULL,                   /* rmdir */
1151         NULL,                   /* mknod */
1152         NULL,                   /* rename */
1153         NULL,                   /* readlink */
1154         NULL,                   /* follow_link */
1155         NULL,                   /* readpage */
1156         NULL,                   /* writepage */
1157         NULL,                   /* bmap */
1158         NULL,                   /* truncate */
1159         NULL                    /* permission */
1160 };
1161 
1162 static int arraylong_read (struct inode * inode, struct file * file, char * buf, int count)
     /*  */
1163 {
1164         unsigned int pid = inode->i_ino >> 16;
1165         unsigned int type = inode->i_ino & 0x0000ffff;
1166 
1167         if (count < 0)
1168                 return -EINVAL;
1169 
1170         switch (type) {
1171                 case PROC_PID_MAPS:
1172                         return read_maps(pid, file, buf, count);
1173         }
1174         return -EINVAL;
1175 }
1176 
1177 static struct file_operations proc_arraylong_operations = {
1178         NULL,           /* array_lseek */
1179         arraylong_read,
1180         NULL,           /* array_write */
1181         NULL,           /* array_readdir */
1182         NULL,           /* array_select */
1183         NULL,           /* array_ioctl */
1184         NULL,           /* mmap */
1185         NULL,           /* no special open code */
1186         NULL,           /* no special release code */
1187         NULL            /* can't fsync */
1188 };
1189 
1190 struct inode_operations proc_arraylong_inode_operations = {
1191         &proc_arraylong_operations,     /* default base directory file-ops */
1192         NULL,                   /* create */
1193         NULL,                   /* lookup */
1194         NULL,                   /* link */
1195         NULL,                   /* unlink */
1196         NULL,                   /* symlink */
1197         NULL,                   /* mkdir */
1198         NULL,                   /* rmdir */
1199         NULL,                   /* mknod */
1200         NULL,                   /* rename */
1201         NULL,                   /* readlink */
1202         NULL,                   /* follow_link */
1203         NULL,                   /* readpage */
1204         NULL,                   /* writepage */
1205         NULL,                   /* bmap */
1206         NULL,                   /* truncate */
1207         NULL                    /* permission */
1208 };
/* */
root/fs/proc/array.c

DEFINITIONS
