* Jerome Forissier  :  added per-CPU time information to /proc/stat
 *                      and /proc/<pid>/cpu extension
 *                      <forissier@isia.cma.fr>
 *   - Incorporation and non-SMP safe operation
 *   of forissier patch in 2.1.78 by
 *   Hans Marcus <crowbar@concepts.nl>
 *
 * aeb@cwi.nl        :  /proc/partitions
 */

#include <linux/types.h>
#include <linux/errno.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/kernel_stat.h>
#include <linux/tty.h>
#include <linux/user.h>
#include <linux/a.out.h>
#include <linux/string.h>
#include <linux/mman.h>
#include <linux/proc_fs.h>
#include <linux/ioport.h>
#include <linux/config.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/swap.h>
#include <linux/slab.h>
#include <linux/smp.h>
#include <linux/signal.h>

#include <asm/uaccess.h>
#include <asm/pgtable.h>
#include <asm/io.h>

#define LOAD_INT(x) ((x) >> FSHIFT)
#define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1-1)) * 100)

#ifdef CONFIG_DEBUG_MALLOC
int get_malloc(char * buffer);
#endif
 

static ssize_t read_core(struct file * file, char * buf,
    size_t count, loff_t *ppos)
{
 unsigned long p = *ppos, memsize;
 ssize_t read;
 ssize_t count1;
 char * pnt;
 struct user dump;
#if defined (__i386__) || defined (__mc68000__)
# define FIRST_MAPPED PAGE_SIZE /* we don't have page 0 mapped on x86.. */
#else
# define FIRST_MAPPED 0
#endif

 memset(&dump, 0, sizeof(struct user));
 dump.magic = CMAGIC;
 dump.u_dsize = max_mapnr;
#ifdef __alpha__
 dump.start_data = PAGE_OFFSET;
#endif

 memsize = (max_mapnr + 1) << PAGE_SHIFT;
 if (p >= memsize)
  return 0;
 if (count > memsize - p)
  count = memsize - p;
 read = 0;

 if (p < sizeof(struct user) && count > 0) {
  count1 = count;
  if (p + count1 > sizeof(struct user))
   count1 = sizeof(struct user)-p;
  pnt = (char *) &dump + p;
  copy_to_user(buf,(void *) pnt, count1);
  buf += count1;
  p += count1;
  count -= count1;
  read += count1;
 }

 if (count > 0 && p < PAGE_SIZE + FIRST_MAPPED) {
  count1 = PAGE_SIZE + FIRST_MAPPED - p;
  if (count1 > count)
   count1 = count;
  clear_user(buf, count1);
  buf += count1;
  p += count1;
  count -= count1;
  read += count1;
 }
 if (count > 0) {
  copy_to_user(buf, (void *) (PAGE_OFFSET+p-PAGE_SIZE), count);
  read += count;
 }
 *ppos += read;
 return read;
}

static struct file_operations proc_kcore_operations = {
 NULL,           /* lseek */
 read_core,
};

struct inode_operations proc_kcore_inode_operations = {
 &proc_kcore_operations,
};

/*
 * This function accesses profiling information. The returned data is
 * binary: the sampling step and the actual contents of the profile
 * buffer. Use of the program readprofile is recommended in order to
 * get meaningful info out of these data.
 */
static ssize_t read_profile(struct file *file, char *buf,
       size_t count, loff_t *ppos)
{
 unsigned long p = *ppos;
 ssize_t read;
 char * pnt;
 unsigned int sample_step = 1 << prof_shift;

 if (p >= (prof_len+1)*sizeof(unsigned int))
  return 0;
 if (count > (prof_len+1)*sizeof(unsigned int) - p)
  count = (prof_len+1)*sizeof(unsigned int) - p;
 read = 0;

 while (p < sizeof(unsigned int) && count > 0) {
  put_user(*((char *)(&sample_step)+p),buf);
  buf++; p++; count--; read++;
 }
 pnt = (char *)prof_buffer + p - sizeof(unsigned int);
 copy_to_user(buf,(void *)pnt,count);
 read += count;
 *ppos += read;
 return read;
}

/*
 * Writing to /proc/profile resets the counters
 *
 * Writing a 'profiling multiplier' value into it also re-sets the profiling
 * interrupt frequency, on architectures that support this.
 */
static ssize_t write_profile(struct file * file, const char * buf,
        size_t count, loff_t *ppos)
{
#ifdef __SMP__
 extern int setup_profiling_timer (unsigned int multiplier);

 if (count==sizeof(int)) {
  unsigned int multiplier;

  if (copy_from_user(&multiplier, buf, sizeof(int)))
   return -EFAULT;

  if (setup_profiling_timer(multiplier))
   return -EINVAL;
 }
#endif
 
 memset(prof_buffer, 0, prof_len * sizeof(*prof_buffer));
 return count;
}

static struct file_operations proc_profile_operations = {
 NULL,           /* lseek */
 read_profile,
 write_profile,
};

struct inode_operations proc_profile_inode_operations = {
 &proc_profile_operations,
};
 

static int get_loadavg(char * buffer)
{
 int a, b, c;

 a = avenrun[0] + (FIXED_1/200);
 b = avenrun[1] + (FIXED_1/200);
 c = avenrun[2] + (FIXED_1/200);
 return sprintf(buffer,"%d.%02d %d.%02d %d.%02d %d/%d %d\n",
  LOAD_INT(a), LOAD_FRAC(a),
  LOAD_INT(b), LOAD_FRAC(b),
  LOAD_INT(c), LOAD_FRAC(c),
  nr_running, nr_tasks, last_pid);
}

static int get_kstat(char * buffer)
{
 int i, len;
 unsigned sum = 0;
 extern unsigned long total_forks;
 unsigned long ticks;

 ticks = jiffies * smp_num_cpus;
 for (i = 0 ; i < NR_IRQS ; i++)
  sum += kstat_irqs(i);

#ifdef __SMP__
 len = sprintf(buffer,
  "cpu  %u %u %u %lu\n",
  kstat.cpu_user,
  kstat.cpu_nice,
  kstat.cpu_system,
  jiffies*smp_num_cpus - (kstat.cpu_user + kstat.cpu_nice + kstat.cpu_system));
 for (i = 0 ; i < smp_num_cpus; i++)
  len += sprintf(buffer + len, "cpu%d %u %u %u %lu\n",
   i,
   kstat.per_cpu_user[cpu_logical_map(i)],
   kstat.per_cpu_nice[cpu_logical_map(i)],
   kstat.per_cpu_system[cpu_logical_map(i)],
   jiffies - (  kstat.per_cpu_user[cpu_logical_map(i)] \
              + kstat.per_cpu_nice[cpu_logical_map(i)] \
              + kstat.per_cpu_system[cpu_logical_map(i)]));
 len += sprintf(buffer + len,
  "disk %u %u %u %u\n"
  "disk_rio %u %u %u %u\n"
  "disk_wio %u %u %u %u\n"
  "disk_rblk %u %u %u %u\n"
  "disk_wblk %u %u %u %u\n"
  "page %u %u\n"
  "swap %u %u\n"
  "intr %u",
#else
 len = sprintf(buffer,
  "cpu  %u %u %u %lu\n"
  "disk %u %u %u %u\n"
  "disk_rio %u %u %u %u\n"
  "disk_wio %u %u %u %u\n"
  "disk_rblk %u %u %u %u\n"
  "disk_wblk %u %u %u %u\n"
  "page %u %u\n"
  "swap %u %u\n"
  "intr %u",
  kstat.cpu_user,
  kstat.cpu_nice,
  kstat.cpu_system,
  ticks - (kstat.cpu_user + kstat.cpu_nice + kstat.cpu_system),
#endif
  kstat.dk_drive[0], kstat.dk_drive[1],
  kstat.dk_drive[2], kstat.dk_drive[3],
  kstat.dk_drive_rio[0], kstat.dk_drive_rio[1],
  kstat.dk_drive_rio[2], kstat.dk_drive_rio[3],
  kstat.dk_drive_wio[0], kstat.dk_drive_wio[1],
  kstat.dk_drive_wio[2], kstat.dk_drive_wio[3],
  kstat.dk_drive_rblk[0], kstat.dk_drive_rblk[1],
  kstat.dk_drive_rblk[2], kstat.dk_drive_rblk[3],
  kstat.dk_drive_wblk[0], kstat.dk_drive_wblk[1],
  kstat.dk_drive_wblk[2], kstat.dk_drive_wblk[3],
  kstat.pgpgin,
  kstat.pgpgout,
  kstat.pswpin,
  kstat.pswpout,
  sum);
 for (i = 0 ; i < NR_IRQS ; i++)
  len += sprintf(buffer + len, " %u", kstat_irqs(i));
 len += sprintf(buffer + len,
  "\nctxt %u\n"
  "btime %lu\n"
  "processes %lu\n",
  kstat.context_swtch,
  xtime.tv_sec - jiffies / HZ,
  total_forks);
 return len;
}
 

static int get_uptime(char * buffer)
{
 unsigned long uptime;
 unsigned long idle;

 uptime = jiffies;
 idle = task[0]->times.tms_utime + task[0]->times.tms_stime;

 /* The formula for the fraction parts really is ((t * 100) / HZ) % 100, but
    that would overflow about every five days at HZ == 100.
    Therefore the identity a = (a / b) * b + a % b is used so that it is
    calculated as (((t / HZ) * 100) + ((t % HZ) * 100) / HZ) % 100.
    The part in front of the '+' always evaluates as 0 (mod 100). All divisions
    in the above formulas are truncating. For HZ being a power of 10, the
    calculations simplify to the version in the #else part (if the printf
    format is adapted to the same number of digits as zeroes in HZ.
  */
#if HZ!=100
 return sprintf(buffer,"%lu.%02lu %lu.%02lu\n",
  uptime / HZ,
  (((uptime % HZ) * 100) / HZ) % 100,
  idle / HZ,
  (((idle % HZ) * 100) / HZ) % 100);
#else
 return sprintf(buffer,"%lu.%02lu %lu.%02lu\n",
  uptime / HZ,
  uptime % HZ,
  idle / HZ,
  idle % HZ);
#endif
}

static int get_meminfo(char * buffer)
{
 struct sysinfo i;
 int len;

 si_meminfo(&i);
 si_swapinfo(&i);
 len = sprintf(buffer, "        total:    used:    free:  shared: buffers:  cached:\n"
  "Mem:  %8lu %8lu %8lu %8lu %8lu %8lu\n"
  "Swap: %8lu %8lu %8lu\n",
  i.totalram, i.totalram-i.freeram, i.freeram, i.sharedram, i.bufferram, page_cache_size*PAGE_SIZE,
  i.totalswap, i.totalswap-i.freeswap, i.freeswap);
 /*
  * Tagged format, for easy grepping and expansion. The above will go away
  * eventually, once the tools have been updated.
  */
 return len + sprintf(buffer+len,
  "MemTotal:  %8lu kB\n"
  "MemFree:   %8lu kB\n"
  "MemShared: %8lu kB\n"
  "Buffers:   %8lu kB\n"
  "Cached:    %8lu kB\n"
  "SwapTotal: %8lu kB\n"
  "SwapFree:  %8lu kB\n",
  i.totalram >> 10,
  i.freeram >> 10,
  i.sharedram >> 10,
  i.bufferram >> 10,
  page_cache_size << (PAGE_SHIFT - 10),
  i.totalswap >> 10,
  i.freeswap >> 10);
}

static int get_version(char * buffer)
{
 extern char *linux_banner;

 strcpy(buffer, linux_banner);
 return strlen(buffer);
}

static int get_cmdline(char * buffer)
{
 extern char saved_command_line[];

 return sprintf(buffer, "%s\n", saved_command_line);
}

static unsigned long get_phys_addr(struct task_struct * p, unsigned long ptr)
{
 pgd_t *page_dir;
 pmd_t *page_middle;
 pte_t pte;

 if (!p || !p->mm || ptr >= TASK_SIZE)
  return 0;
 /* Check for NULL pgd .. shouldn't happen! */
 if (!p->mm->pgd) {
  printk("get_phys_addr: pid %d has NULL pgd!\n", p->pid);
  return 0;
 }

 page_dir = pgd_offset(p->mm,ptr);
 if (pgd_none(*page_dir))
  return 0;
 if (pgd_bad(*page_dir)) {
  printk("bad page directory entry %08lx\n", pgd_val(*page_dir));
  pgd_clear(page_dir);
  return 0;
 }
 page_middle = pmd_offset(page_dir,ptr);
 if (pmd_none(*page_middle))
  return 0;
 if (pmd_bad(*page_middle)) {
  printk("bad page middle entry %08lx\n", pmd_val(*page_middle));
  pmd_clear(page_middle);
  return 0;
 }
 pte = *pte_offset(page_middle,ptr);
 if (!pte_present(pte))
  return 0;
 return pte_page(pte) + (ptr & ~PAGE_MASK);
}

static int get_array(struct task_struct *p, unsigned long start, unsigned long end, char * buffer)
{
 unsigned long addr;
 int size = 0, result = 0;
 char c;

 if (start >= end)
  return result;
 for (;;) {
  addr = get_phys_addr(p, start);
  if (!addr)
   return result;
  do {
   c = *(char *) addr;
   if (!c)
    result = size;
   if (size < PAGE_SIZE)
    buffer[size++] = c;
   else
    return result;
   addr++;
   start++;
   if (!c && start >= end)
    return result;
  } while (addr & ~PAGE_MASK);
 }
 return result;
}

static int get_env(int pid, char * buffer)
{
 struct task_struct *p;
 
 read_lock(&tasklist_lock);
 p = find_task_by_pid(pid);
 read_unlock(&tasklist_lock); /* FIXME!! This should be done after the last use */

 if (!p || !p->mm)
  return 0;
 return get_array(p, p->mm->env_start, p->mm->env_end, buffer);
}

static int get_arg(int pid, char * buffer)
{
 struct task_struct *p;

 read_lock(&tasklist_lock);
 p = find_task_by_pid(pid);
 read_unlock(&tasklist_lock); /* FIXME!! This should be done after the last use */
 if (!p || !p->mm)
  return 0;
 return get_array(p, p->mm->arg_start, p->mm->arg_end, buffer);
}

/*
 * These bracket the sleeping functions..
 */
extern void scheduling_functions_start_here(void);
extern void scheduling_functions_end_here(void);
#define first_sched ((unsigned long) scheduling_functions_start_here)
#define last_sched ((unsigned long) scheduling_functions_end_here)

static unsigned long get_wchan(struct task_struct *p)
{
 if (!p || p == current || p->state == TASK_RUNNING)
  return 0;
#if defined(__i386__)
 {
  unsigned long ebp, esp, eip;
  unsigned long stack_page;
  int count = 0;

  stack_page = (unsigned long)p;
  esp = p->tss.esp;
  if (!stack_page || esp < stack_page || esp >= 8188+stack_page)
   return 0;
  /* include/asm-i386/system.h:switch_to() pushes ebp last. */
  ebp = *(unsigned long *) esp;
  do {
   if (ebp < stack_page || ebp >= 8188+stack_page)
    return 0;
   eip = *(unsigned long *) (ebp+4);
   if (eip < first_sched || eip >= last_sched)
    return eip;
   ebp = *(unsigned long *) ebp;
  } while (count++ < 16);
 }
#elif defined(__alpha__)
 /*
  * This one depends on the frame size of schedule().  Do a
  * "disass schedule" in gdb to find the frame size.  Also, the
  * code assumes that sleep_on() follows immediately after
  * interruptible_sleep_on() and that add_timer() follows
  * immediately after interruptible_sleep().  Ugly, isn't it?
  * Maybe adding a wchan field to task_struct would be better,
  * after all...
  */
 {
     unsigned long schedule_frame;
     unsigned long pc;

     pc = thread_saved_pc(&p->tss);
     if (pc >= first_sched && pc < last_sched) {
  schedule_frame = ((unsigned long *)p->tss.ksp)[6];
  return ((unsigned long *)schedule_frame)[12];
     }
     return pc;
 }
#elif defined(__mc68000__)
 {
     unsigned long fp, pc;
     unsigned long stack_page;
     int count = 0;

     stack_page = (unsigned long)p;
     fp = ((struct switch_stack *)p->tss.ksp)->a6;
     do {
      if (fp < stack_page+sizeof(struct task_struct) ||
   fp >= 8184+stack_page)
       return 0;
      pc = ((unsigned long *)fp)[1];
  /* FIXME: This depends on the order of these functions. */
      if (pc < first_sched || pc >= last_sched)
        return pc;
      fp = *(unsigned long *) fp;
     } while (count++ < 16);
 }
#elif defined(__powerpc__)
 return (p->tss.wchan);
#elif defined (CONFIG_ARM)
 {
  unsigned long fp, lr;
  unsigned long stack_page;
  int count = 0;

  stack_page = 4096 + (unsigned long)p;
  fp = get_css_fp (&p->tss);
  do {
   if (fp < stack_page || fp > 4092+stack_page)
    return 0;
   lr = pc_pointer (((unsigned long *)fp)[-1]);
   if (lr < first_sched || lr > last_sched)
    return lr;
   fp = *(unsigned long *) (fp - 12);
  } while (count ++ < 16);
 }
#elif defined (__sparc__)
 {
  unsigned long pc, fp, bias = 0;
  unsigned long task_base = (unsigned long) p;
  struct reg_window *rw;
  int count = 0;

#ifdef __sparc_v9__
  bias = STACK_BIAS;
#endif
      fp = p->tss.ksp + bias;
  do {
   /* Bogus frame pointer? */
   if (fp < (task_base + sizeof(struct task_struct)) ||
       fp >= (task_base + (2 * PAGE_SIZE)))
    break;
   rw = (struct reg_window *) fp;
   pc = rw->ins[7];
   if (pc < first_sched || pc >= last_sched)
    return pc;
   fp = rw->ins[6] + bias;
  } while (++count < 16);
 }
#endif
 return 0;
}

#if defined(__i386__)
# define KSTK_EIP(tsk) (((unsigned long *)(4096+(unsigned long)(tsk)))[1019])
# define KSTK_ESP(tsk) (((unsigned long *)(4096+(unsigned long)(tsk)))[1022])
#elif defined(__alpha__)
  /*
   * See arch/alpha/kernel/ptrace.c for details.
   */
# define PT_REG(reg)  (PAGE_SIZE - sizeof(struct pt_regs) \
     + (long)&((struct pt_regs *)0)->reg)
# define KSTK_EIP(tsk) \
    (*(unsigned long *)(PT_REG(pc) + PAGE_SIZE + (unsigned long)(tsk)))
# define KSTK_ESP(tsk) ((tsk) == current ? rdusp() : (tsk)->tss.usp)
#elif defined(CONFIG_ARM)
# define KSTK_EIP(tsk) (((unsigned long *)(4096+(unsigned long)(tsk)))[1022])
# define KSTK_ESP(tsk) (((unsigned long *)(4096+(unsigned long)(tsk)))[1020])
#elif defined(__mc68000__)
#define KSTK_EIP(tsk) \
    ({   \
 unsigned long eip = 0;  \
  if ((tsk)->tss.esp0 > PAGE_SIZE && \
     MAP_NR((tsk)->tss.esp0) < max_mapnr) \
       eip = ((struct pt_regs *) (tsk)->tss.esp0)->pc;  \
        eip; })
#define KSTK_ESP(tsk) ((tsk) == current ? rdusp() : (tsk)->tss.usp)
#elif defined(__powerpc__)
#define KSTK_EIP(tsk) ((tsk)->tss.regs->nip)
#define KSTK_ESP(tsk) ((tsk)->tss.regs->gpr[1])
#elif defined (__sparc_v9__)
# define KSTK_EIP(tsk)  ((tsk)->tss.kregs->tpc)
# define KSTK_ESP(tsk)  ((tsk)->tss.kregs->u_regs[UREG_FP])
#elif defined(__sparc__)
# define KSTK_EIP(tsk)  ((tsk)->tss.kregs->pc)
# define KSTK_ESP(tsk)  ((tsk)->tss.kregs->u_regs[UREG_FP])
#endif

/* Gcc optimizes away "strlen(x)" for constant x */
#define ADDBUF(buffer, string) \
do { memcpy(buffer, string, strlen(string)); \
     buffer += strlen(string); } while (0)

static inline char * task_name(struct task_struct *p, char * buf)
{
 int i;
 char * name;

 ADDBUF(buf, "Name:\t");
 name = p->comm;
 i = sizeof(p->comm);
 do {
  unsigned char c = *name;
  name++;
  i--;
  *buf = c;
  if (!c)
   break;
  if (c == '\\') {
   buf[1] = c;
   buf += 2;
   continue;
  }
  if (c == '\n') {
   buf[0] = '\\';
   buf[1] = 'n';
   buf += 2;
   continue;
  }
  buf++;
 } while (i);
 *buf = '\n';
 return buf+1;
}

/*
 * The task state array is a strange "bitmap" of
 * reasons to sleep. Thus "running" is zero, and
 * you can test for combinations of others with
 * simple bit tests.
 */
static const char *task_state_array[] = {
 "R (running)",  /*  0 */
 "S (sleeping)",  /*  1 */
 "D (disk sleep)", /*  2 */
 "Z (zombie)",  /*  4 */
 "T (stopped)",  /*  8 */
 "W (paging)"  /* 16 */
};

static inline const char * get_task_state(struct task_struct *tsk)
{
 unsigned int state = tsk->state & (TASK_RUNNING |
        TASK_INTERRUPTIBLE |
        TASK_UNINTERRUPTIBLE |
        TASK_ZOMBIE |
        TASK_STOPPED |
        TASK_SWAPPING);
 const char **p = &task_state_array[0];

 while (state) {
  p++;
  state >>= 1;
 }
 return *p;
}

static inline char * task_state(struct task_struct *p, char *buffer)
{
 buffer += sprintf(buffer,
  "State:\t%s\n"
  "Pid:\t%d\n"
  "PPid:\t%d\n"
  "Uid:\t%d\t%d\t%d\t%d\n"
  "Gid:\t%d\t%d\t%d\t%d\n",
  get_task_state(p),
  p->pid, p->p_pptr->pid,
  p->uid, p->euid, p->suid, p->fsuid,
  p->gid, p->egid, p->sgid, p->fsgid);
 return buffer;
}

static inline char * task_mem(struct task_struct *p, char *buffer)
{
 struct mm_struct * mm = p->mm;

 if (mm && mm != &init_mm) {
  struct vm_area_struct * vma = mm->mmap;
  unsigned long data = 0, stack = 0;
  unsigned long exec = 0, lib = 0;

  for (vma = mm->mmap; vma; vma = vma->vm_next) {
   unsigned long len = (vma->vm_end - vma->vm_start) >> 10;
   if (!vma->vm_file) {
    data += len;
    if (vma->vm_flags & VM_GROWSDOWN)
     stack += len;
    continue;
   }
   if (vma->vm_flags & VM_WRITE)
    continue;
   if (vma->vm_flags & VM_EXEC) {
    exec += len;
    if (vma->vm_flags & VM_EXECUTABLE)
     continue;
    lib += len;
   }
  }
  buffer += sprintf(buffer,
   "VmSize:\t%8lu kB\n"
   "VmLck:\t%8lu kB\n"
   "VmRSS:\t%8lu kB\n"
   "VmData:\t%8lu kB\n"
   "VmStk:\t%8lu kB\n"
   "VmExe:\t%8lu kB\n"
   "VmLib:\t%8lu kB\n",
   mm->total_vm << (PAGE_SHIFT-10),
   mm->locked_vm << (PAGE_SHIFT-10),
   mm->rss << (PAGE_SHIFT-10),
   data - stack, stack,
   exec - lib, lib);
 }
 return buffer;
}

static void collect_sigign_sigcatch(struct task_struct *p, sigset_t *ign,
        sigset_t *catch)
{
 struct k_sigaction *k;
 int i;

 sigemptyset(ign);
 sigemptyset(catch);

 if (p->sig) {
  k = p->sig->action;
  for (i = 1; i <= _NSIG; ++i, ++k) {
   if (k->sa.sa_handler == SIG_IGN)
    sigaddset(ign, i);
   else if (k->sa.sa_handler != SIG_DFL)
    sigaddset(catch, i);
  }
 }
}

static inline char * task_sig(struct task_struct *p, char *buffer)
{
 sigset_t ign, catch;

 buffer += sprintf(buffer, "SigPnd:\t");
 buffer = render_sigset_t(&p->signal, buffer);
 *buffer++ = '\n';
 buffer += sprintf(buffer, "SigBlk:\t");
 buffer = render_sigset_t(&p->blocked, buffer);
 *buffer++ = '\n';

 collect_sigign_sigcatch(p, &ign, &catch);
 buffer += sprintf(buffer, "SigIgn:\t");
 buffer = render_sigset_t(&ign, buffer);
 *buffer++ = '\n';
 buffer += sprintf(buffer, "SigCgt:\t"); /* Linux 2.0 uses "SigCgt" */
 buffer = render_sigset_t(&catch, buffer);
 *buffer++ = '\n';

 return buffer;
}

extern inline char *task_cap(struct task_struct *p, char *buffer)
{
    return buffer + sprintf(buffer, "CapInh:\t%016x\n"
       "CapPrm:\t%016x\n"
       "CapEff:\t%016x\n",
       p->cap_inheritable.cap,
       p->cap_permitted.cap,
       p->cap_effective.cap);
}
 

static int get_status(int pid, char * buffer)
{
 char * orig = buffer;
 struct task_struct *tsk;

 read_lock(&tasklist_lock);
 tsk = find_task_by_pid(pid);
 read_unlock(&tasklist_lock); /* FIXME!! This should be done after the last use */
 if (!tsk)
  return 0;
 buffer = task_name(tsk, buffer);
 buffer = task_state(tsk, buffer);
 buffer = task_mem(tsk, buffer);
 buffer = task_sig(tsk, buffer);
 buffer = task_cap(tsk, buffer);
 return buffer - orig;
}

static int get_stat(int pid, char * buffer)
{
 struct task_struct *tsk;
 unsigned long vsize, eip, esp, wchan;
 long priority, nice;
 int tty_pgrp;
 sigset_t sigign, sigcatch;
 char state;

 read_lock(&tasklist_lock);
 tsk = find_task_by_pid(pid);
 read_unlock(&tasklist_lock); /* FIXME!! This should be done after the last use */
 if (!tsk)
  return 0;
 state = *get_task_state(tsk);
 vsize = eip = esp = 0;
 if (tsk->mm && tsk->mm != &init_mm) {
  struct vm_area_struct *vma = tsk->mm->mmap;
  while (vma) {
   vsize += vma->vm_end - vma->vm_start;
   vma = vma->vm_next;
  }
  eip = KSTK_EIP(tsk);
  esp = KSTK_ESP(tsk);
 }

 wchan = get_wchan(tsk);

 collect_sigign_sigcatch(tsk, &sigign, &sigcatch);

 if (tsk->tty)
  tty_pgrp = tsk->tty->pgrp;
 else
  tty_pgrp = -1;

 /* scale priority and nice values from timeslices to -20..20 */
 /* to make it look like a "normal" Unix priority/nice value  */
 priority = tsk->counter;
 priority = 20 - (priority * 10 + DEF_PRIORITY / 2) / DEF_PRIORITY;
 nice = tsk->priority;
 nice = 20 - (nice * 20 + DEF_PRIORITY / 2) / DEF_PRIORITY;

 return sprintf(buffer,"%d (%s) %c %d %d %d %d %d %lu %lu \
%lu %lu %lu %lu %lu %ld %ld %ld %ld %ld %ld %lu %lu %ld %lu %lu %lu %lu %lu \
%lu %lu %lu %lu %lu %lu %lu %lu %d\n",
  pid,
  tsk->comm,
  state,
  tsk->p_pptr->pid,
  tsk->pgrp,
  tsk->session,
         tsk->tty ? kdev_t_to_nr(tsk->tty->device) : 0,
  tty_pgrp,
  tsk->flags,
  tsk->min_flt,
  tsk->cmin_flt,
  tsk->maj_flt,
  tsk->cmaj_flt,
  tsk->times.tms_utime,
  tsk->times.tms_stime,
  tsk->times.tms_cutime,
  tsk->times.tms_cstime,
  priority,
  nice,
  tsk->timeout,
  tsk->it_real_value,
  tsk->start_time,
  vsize,
  tsk->mm ? tsk->mm->rss : 0, /* you might want to shift this left 3 */
  tsk->rlim ? tsk->rlim[RLIMIT_RSS].rlim_cur : 0,
  tsk->mm ? tsk->mm->start_code : 0,
  tsk->mm ? tsk->mm->end_code : 0,
  tsk->mm ? tsk->mm->start_stack : 0,
  esp,
  eip,
  /* The signal information here is obsolete.
   * It must be decimal for Linux 2.0 compatibility.
   * Use /proc/#/status for real-time signals.
   */
  tsk->signal .sig[0] & 0x7fffffffUL,
  tsk->blocked.sig[0] & 0x7fffffffUL,
  sigign      .sig[0] & 0x7fffffffUL,
  sigcatch    .sig[0] & 0x7fffffffUL,
  wchan,
  tsk->nswap,
  tsk->cnswap,
  tsk->exit_signal);
}
 
static inline void statm_pte_range(pmd_t * pmd, unsigned long address, unsigned long size,
 int * pages, int * shared, int * dirty, int * total)
{
 pte_t * pte;
 unsigned long end;

 if (pmd_none(*pmd))
  return;
 if (pmd_bad(*pmd)) {
  printk("statm_pte_range: bad pmd (%08lx)\n", pmd_val(*pmd));
  pmd_clear(pmd);
  return;
 }
 pte = pte_offset(pmd, address);
 address &= ~PMD_MASK;
 end = address + size;
 if (end > PMD_SIZE)
  end = PMD_SIZE;
 do {
  pte_t page = *pte;

  address += PAGE_SIZE;
  pte++;
  if (pte_none(page))
   continue;
  ++*total;
  if (!pte_present(page))
   continue;
  ++*pages;
  if (pte_dirty(page))
   ++*dirty;
  if (MAP_NR(pte_page(page)) >= max_mapnr)
   continue;
  if (atomic_read(&mem_map[MAP_NR(pte_page(page))].count) > 1)
   ++*shared;
 } while (address < end);
}

static inline void statm_pmd_range(pgd_t * pgd, unsigned long address, unsigned long size,
 int * pages, int * shared, int * dirty, int * total)
{
 pmd_t * pmd;
 unsigned long end;

 if (pgd_none(*pgd))
  return;
 if (pgd_bad(*pgd)) {
  printk("statm_pmd_range: bad pgd (%08lx)\n", pgd_val(*pgd));
  pgd_clear(pgd);
  return;
 }
 pmd = pmd_offset(pgd, address);
 address &= ~PGDIR_MASK;
 end = address + size;
 if (end > PGDIR_SIZE)
  end = PGDIR_SIZE;
 do {
  statm_pte_range(pmd, address, end - address, pages, shared, dirty, total);
  address = (address + PMD_SIZE) & PMD_MASK;
  pmd++;
 } while (address < end);
}

static void statm_pgd_range(pgd_t * pgd, unsigned long address, unsigned long end,
 int * pages, int * shared, int * dirty, int * total)
{
 while (address < end) {
  statm_pmd_range(pgd, address, end - address, pages, shared, dirty, total);
  address = (address + PGDIR_SIZE) & PGDIR_MASK;
  pgd++;
 }
}

static int get_statm(int pid, char * buffer)
{
 struct task_struct *tsk;
 int size=0, resident=0, share=0, trs=0, lrs=0, drs=0, dt=0;

 read_lock(&tasklist_lock);
 tsk = find_task_by_pid(pid);
 read_unlock(&tasklist_lock); /* FIXME!! This should be done after the last use */
 if (!tsk)
  return 0;
 if (tsk->mm && tsk->mm != &init_mm) {
  struct vm_area_struct * vma = tsk->mm->mmap;

  while (vma) {
   pgd_t *pgd = pgd_offset(tsk->mm, vma->vm_start);
   int pages = 0, shared = 0, dirty = 0, total = 0;

   statm_pgd_range(pgd, vma->vm_start, vma->vm_end, &pages, &shared, &dirty, &total);
   resident += pages;
   share += shared;
   dt += dirty;
   size += total;
   if (vma->vm_flags & VM_EXECUTABLE)
    trs += pages; /* text */
   else if (vma->vm_flags & VM_GROWSDOWN)
    drs += pages; /* stack */
   else if (vma->vm_end > 0x60000000)
    lrs += pages; /* library */
   else
    drs += pages;
   vma = vma->vm_next;
  }
 }
 return sprintf(buffer,"%d %d %d %d %d %d %d\n",
         size, resident, share, trs, lrs, drs, dt);
}

/*
 * The way we support synthetic files > 4K
 * - without storing their contents in some buffer and
 * - without walking through the entire synthetic file until we reach the
 *   position of the requested data
 * is to cleverly encode the current position in the file's f_pos field.
 * There is no requirement that a read() call which returns `count' bytes
 * of data increases f_pos by exactly `count'.
 *
 * This idea is Linus' one. Bruno implemented it.
 */

/*
 * For the /proc/<pid>/maps file, we use fixed length records, each containing
 * a single line.
 */
#define MAPS_LINE_LENGTH 4096
#define MAPS_LINE_SHIFT  12
/*
 * f_pos = (number of the vma in the task->mm->mmap list) * MAPS_LINE_LENGTH
 *         + (index into the line)
 */
/* for systems with sizeof(void*) == 4: */
#define MAPS_LINE_FORMAT4   "%08lx-%08lx %s %08lx %s %lu"
#define MAPS_LINE_MAX4 49 /* sum of 8  1  8  1 4 1 8 1 5 1 10 1 */

/* for systems with sizeof(void*) == 8: */
#define MAPS_LINE_FORMAT8   "%016lx-%016lx %s %016lx %s %lu"
#define MAPS_LINE_MAX8 73 /* sum of 16  1  16  1 4 1 16 1 5 1 10 1 */

#define MAPS_LINE_MAX MAPS_LINE_MAX8
 

static ssize_t read_maps (int pid, struct file * file, char * buf,
     size_t count, loff_t *ppos)
{
 struct task_struct *p;
 struct vm_area_struct * map, * next;
 char * destptr = buf, * buffer;
 loff_t lineno;
 ssize_t column, i;
 int volatile_task;
 long retval;

 /*
  * We might sleep getting the page, so get it first.
  */
 retval = -ENOMEM;
 buffer = (char*)__get_free_page(GFP_KERNEL);
 if (!buffer)
  goto out;

 retval = -EINVAL;
 read_lock(&tasklist_lock);
 p = find_task_by_pid(pid);
 read_unlock(&tasklist_lock); /* FIXME!! This should be done after the last use */
 if (!p)
  goto freepage_out;

 if (!p->mm || p->mm == &init_mm || count == 0)
  goto getlen_out;

 /* Check whether the mmaps could change if we sleep */
 volatile_task = (p != current || atomic_read(&p->mm->count) > 1);

 /* decode f_pos */
 lineno = *ppos >> MAPS_LINE_SHIFT;
 column = *ppos & (MAPS_LINE_LENGTH-1);

 /* quickly go to line lineno */
 for (map = p->mm->mmap, i = 0; map && (i < lineno); map = map->vm_next, i++)
  continue;

 for ( ; map ; map = next ) {
  /* produce the next line */
  char *line;
  char str[5], *cp = str;
  int flags;
  kdev_t dev;
  unsigned long ino;
  int maxlen = (sizeof(void*) == 4) ?
   MAPS_LINE_MAX4 :  MAPS_LINE_MAX8;
  int len;

  /*
   * Get the next vma now (but it won't be used if we sleep).
   */
  next = map->vm_next;
  flags = map->vm_flags;

  *cp++ = flags & VM_READ ? 'r' : '-';
  *cp++ = flags & VM_WRITE ? 'w' : '-';
  *cp++ = flags & VM_EXEC ? 'x' : '-';
  *cp++ = flags & VM_MAYSHARE ? 's' : 'p';
  *cp++ = 0;

  dev = 0;
  ino = 0;
  if (map->vm_file != NULL) {
   dev = map->vm_file->f_dentry->d_inode->i_dev;
   ino = map->vm_file->f_dentry->d_inode->i_ino;
   line = d_path(map->vm_file->f_dentry, buffer, PAGE_SIZE);
   buffer[PAGE_SIZE-1] = '\n';
   line -= maxlen;
   if(line < buffer)
    line = buffer;
  } else
   line = buffer;

  len = sprintf(line,
         sizeof(void*) == 4 ? MAPS_LINE_FORMAT4 : MAPS_LINE_FORMAT8,
         map->vm_start, map->vm_end, str, map->vm_offset,
         kdevname(dev), ino);

  if(map->vm_file) {
   for(i = len; i < maxlen; i++)
    line[i] = ' ';
   len = buffer + PAGE_SIZE - line;
  } else
   line[len++] = '\n';
  if (column >= len) {
   column = 0; /* continue with next line at column 0 */
   lineno++;
   continue; /* we haven't slept */
  }

  i = len-column;
  if (i > count)
   i = count;
  copy_to_user(destptr, line+column, i); /* may have slept */
  destptr += i;
  count   -= i;
  column  += i;
  if (column >= len) {
   column = 0; /* next time: next line at column 0 */
   lineno++;
  }

  /* done? */
  if (count == 0)
   break;

  /* By writing to user space, we might have slept.
   * Stop the loop, to avoid a race condition.
   */
  if (volatile_task)
   break;
 }

 /* encode f_pos */
 *ppos = (lineno << MAPS_LINE_SHIFT) + column;

getlen_out:
 retval = destptr - buf;

freepage_out:
 free_page((unsigned long)buffer);
out:
 return retval;
}

#ifdef __SMP__
static int get_pidcpu(int pid, char * buffer)
{
 struct task_struct * tsk = current ;
 int i, len;

 read_lock(&tasklist_lock);
 if (pid != tsk->pid)
  tsk = find_task_by_pid(pid);
 read_unlock(&tasklist_lock); /* FIXME!! This should be done after the last use */

 if (tsk == NULL)
  return 0;

 len = sprintf(buffer,
  "cpu  %lu %lu\n",
  tsk->times.tms_utime,
  tsk->times.tms_stime);
 
 for (i = 0 ; i < smp_num_cpus; i++)
  len += sprintf(buffer + len, "cpu%d %lu %lu\n",
   i,
   tsk->per_cpu_utime[cpu_logical_map(i)],
   tsk->per_cpu_stime[cpu_logical_map(i)]);

 return len;
}
#endif

#ifdef CONFIG_MODULES
extern int get_module_list(char *);
extern int get_ksyms_list(char *, char **, off_t, int);
#endif
extern int get_device_list(char *);
extern int get_partition_list(char *);
extern int get_filesystem_list(char *);
extern int get_filesystem_info( char * );
extern int get_irq_list(char *);
extern int get_dma_list(char *);
extern int get_cpuinfo(char *);
extern int get_pci_list(char *);
extern int get_md_status (char *);
extern int get_rtc_status (char *);
extern int get_locks_status (char *, char **, off_t, int);
extern int get_swaparea_info (char *);
extern int get_hardware_list(char *);
extern int get_stram_list(char *);

static long get_root_array(char * page, int type, char **start,
 off_t offset, unsigned long length)
{
 switch (type) {
  case PROC_LOADAVG:
   return get_loadavg(page);

  case PROC_UPTIME:
   return get_uptime(page);

  case PROC_MEMINFO:
   return get_meminfo(page);

#ifdef CONFIG_PCI_OLD_PROC
           case PROC_PCI:
   return get_pci_list(page);
#endif

#ifdef CONFIG_NUBUS
  case PROC_NUBUS:
   return get_nubus_list(page);
#endif
 
  case PROC_CPUINFO:
   return get_cpuinfo(page);

  case PROC_VERSION:
   return get_version(page);

#ifdef CONFIG_DEBUG_MALLOC
  case PROC_MALLOC:
   return get_malloc(page);
#endif

#ifdef CONFIG_MODULES
  case PROC_MODULES:
   return get_module_list(page);

  case PROC_KSYMS:
   return get_ksyms_list(page, start, offset, length);
#endif

  case PROC_STAT:
   return get_kstat(page);

  case PROC_SLABINFO:
   return get_slabinfo(page);

  case PROC_DEVICES:
   return get_device_list(page);

  case PROC_PARTITIONS:
   return get_partition_list(page);

  case PROC_INTERRUPTS:
   return get_irq_list(page);

  case PROC_FILESYSTEMS:
   return get_filesystem_list(page);

  case PROC_DMA:
   return get_dma_list(page);

  case PROC_IOPORTS:
   return get_ioport_list(page);
#ifdef CONFIG_BLK_DEV_MD
         case PROC_MD:
   return get_md_status(page);
#endif
  case PROC_CMDLINE:
   return get_cmdline(page);

                case PROC_MTAB:
                       return get_filesystem_info( page );

                case PROC_SWAP:
                       return get_swaparea_info(page);
#ifdef CONFIG_RTC
  case PROC_RTC:
   return get_rtc_status(page);
#endif
  case PROC_LOCKS:
   return get_locks_status(page, start, offset, length);
#ifdef CONFIG_PROC_HARDWARE
  case PROC_HARDWARE:
   return get_hardware_list(page);
#endif
#ifdef CONFIG_STRAM_PROC
  case PROC_STRAM:
   return get_stram_list(page);
#endif
 }
 return -EBADF;
}

static int get_process_array(char * page, int pid, int type)
{
 switch (type) {
  case PROC_PID_STATUS:
   return get_status(pid, page);
  case PROC_PID_ENVIRON:
   return get_env(pid, page);
  case PROC_PID_CMDLINE:
   return get_arg(pid, page);
  case PROC_PID_STAT:
   return get_stat(pid, page);
  case PROC_PID_STATM:
   return get_statm(pid, page);
#ifdef __SMP__
  case PROC_PID_CPU:
   return get_pidcpu(pid, page);
#endif
 }
 return -EBADF;
}
 

static inline int fill_array(char * page, int pid, int type, char **start, off_t offset, int length)
{
 if (pid)
  return get_process_array(page, pid, type);
 return get_root_array(page, type, start, offset, length);
}

#define PROC_BLOCK_SIZE (3*1024)  /* 4K page size but our output routines use some slack for overruns */

static ssize_t array_read(struct file * file, char * buf,
     size_t count, loff_t *ppos)
{
 struct inode * inode = file->f_dentry->d_inode;
 unsigned long page;
 char *start;
 ssize_t length;
 ssize_t end;
 unsigned int type, pid;
 struct proc_dir_entry *dp;

 if (count > PROC_BLOCK_SIZE)
  count = PROC_BLOCK_SIZE;
 if (!(page = __get_free_page(GFP_KERNEL)))
  return -ENOMEM;
 type = inode->i_ino;
 pid = type >> 16;
 type &= 0x0000ffff;
 start = NULL;
 dp = (struct proc_dir_entry *) inode->u.generic_ip;
 if (dp->get_info)
  length = dp->get_info((char *)page, &start, *ppos,
          count, 0);
 else
  length = fill_array((char *) page, pid, type,
        &start, *ppos, count);
 if (length < 0) {
  free_page(page);
  return length;
 }
 if (start != NULL) {
  /* We have had block-adjusting processing! */
  copy_to_user(buf, start, length);
  *ppos += length;
  count = length;
 } else {
  /* Static 4kB (or whatever) block capacity */
  if (*ppos >= length) {
   free_page(page);
   return 0;
  }
  if (count + *ppos > length)
   count = length - *ppos;
  end = count + *ppos;
  copy_to_user(buf, (char *) page + *ppos, count);
  *ppos = end;
 }
 free_page(page);
 return count;
}

static struct file_operations proc_array_operations = {
 NULL,  /* array_lseek */
 array_read,
 NULL,  /* array_write */
 NULL,  /* array_readdir */
 NULL,  /* array_poll */
 NULL,  /* array_ioctl */
 NULL,  /* mmap */
 NULL,  /* no special open code */
 NULL,  /* flush */
 NULL,  /* no special release code */
 NULL  /* can't fsync */
};

struct inode_operations proc_array_inode_operations = {
 &proc_array_operations, /* default base directory file-ops */
 NULL,   /* create */
 NULL,   /* lookup */
 NULL,   /* link */
 NULL,   /* unlink */
 NULL,   /* symlink */
 NULL,   /* mkdir */
 NULL,   /* rmdir */
 NULL,   /* mknod */
 NULL,   /* rename */
 NULL,   /* readlink */
 NULL,   /* follow_link */
 NULL,   /* readpage */
 NULL,   /* writepage */
 NULL,   /* bmap */
 NULL,   /* truncate */
 NULL   /* permission */
};

static ssize_t arraylong_read(struct file * file, char * buf,
         size_t count, loff_t *ppos)
{
 struct inode * inode = file->f_dentry->d_inode;
 unsigned int pid = inode->i_ino >> 16;
 unsigned int type = inode->i_ino & 0x0000ffff;

 switch (type) {
  case PROC_PID_MAPS:
   return read_maps(pid, file, buf, count, ppos);
 }
 return -EINVAL;
}

static struct file_operations proc_arraylong_operations = {
 NULL,  /* array_lseek */
 arraylong_read,
 NULL,  /* array_write */
 NULL,  /* array_readdir */
 NULL,  /* array_poll */
 NULL,  /* array_ioctl */
 NULL,  /* mmap */
 NULL,  /* no special open code */
 NULL,  /* flush */
 NULL,  /* no special release code */
 NULL  /* can't fsync */
};

struct inode_operations proc_arraylong_inode_operations = {
 &proc_arraylong_operations, /* default base directory file-ops */
 NULL,   /* create */
 NULL,   /* lookup */
 NULL,   /* link */
 NULL,   /* unlink */
 NULL,   /* symlink */
 NULL,   /* mkdir */
 NULL,   /* rmdir */
 NULL,   /* mknod */
 NULL,   /* rename */
 NULL,   /* readlink */
 NULL,   /* follow_link */
 NULL,   /* readpage */
 NULL,   /* writepage */
 NULL,   /* bmap */
 NULL,   /* truncate */
 NULL   /* permission */
};