Merge branch 'upstream-net26' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
[linux-2.6] / fs / proc / proc_misc.c
1 /*
2  *  linux/fs/proc/proc_misc.c
3  *
4  *  linux/fs/proc/array.c
5  *  Copyright (C) 1992  by Linus Torvalds
6  *  based on ideas by Darren Senn
7  *
8  *  This used to be the part of array.c. See the rest of history and credits
9  *  there. I took this into a separate file and switched the thing to generic
10  *  proc_file_inode_operations, leaving in array.c only per-process stuff.
11  *  Inumbers allocation made dynamic (via create_proc_entry()).  AV, May 1999.
12  *
13  * Changes:
14  * Fulton Green      :  Encapsulated position metric calculations.
15  *                      <kernel@FultonGreen.com>
16  */
17
18 #include <linux/types.h>
19 #include <linux/errno.h>
20 #include <linux/time.h>
21 #include <linux/kernel.h>
22 #include <linux/kernel_stat.h>
23 #include <linux/fs.h>
24 #include <linux/tty.h>
25 #include <linux/string.h>
26 #include <linux/mman.h>
27 #include <linux/proc_fs.h>
28 #include <linux/ioport.h>
29 #include <linux/mm.h>
30 #include <linux/mmzone.h>
31 #include <linux/pagemap.h>
32 #include <linux/interrupt.h>
33 #include <linux/swap.h>
34 #include <linux/slab.h>
35 #include <linux/genhd.h>
36 #include <linux/smp.h>
37 #include <linux/signal.h>
38 #include <linux/module.h>
39 #include <linux/init.h>
40 #include <linux/seq_file.h>
41 #include <linux/times.h>
42 #include <linux/profile.h>
43 #include <linux/utsname.h>
44 #include <linux/blkdev.h>
45 #include <linux/hugetlb.h>
46 #include <linux/jiffies.h>
47 #include <linux/sysrq.h>
48 #include <linux/vmalloc.h>
49 #include <linux/crash_dump.h>
50 #include <linux/pid_namespace.h>
51 #include <linux/bootmem.h>
52 #include <asm/uaccess.h>
53 #include <asm/pgtable.h>
54 #include <asm/io.h>
55 #include <asm/tlb.h>
56 #include <asm/div64.h>
57 #include "internal.h"
58
59 #define LOAD_INT(x) ((x) >> FSHIFT)
60 #define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1-1)) * 100)
61 /*
62  * Warning: stuff below (imported functions) assumes that its output will fit
63  * into one page. For some of those functions it may be wrong. Moreover, we
64  * have a way to deal with that gracefully. Right now I used straightforward
65  * wrappers, but this needs further analysis wrt potential overflows.
66  */
67 extern int get_hardware_list(char *);
68 extern int get_stram_list(char *);
69 extern int get_exec_domain_list(char *);
70 extern int get_dma_list(char *);
71
72 static int proc_calc_metrics(char *page, char **start, off_t off,
73                                  int count, int *eof, int len)
74 {
75         if (len <= off+count) *eof = 1;
76         *start = page + off;
77         len -= off;
78         if (len>count) len = count;
79         if (len<0) len = 0;
80         return len;
81 }
82
83 static int loadavg_read_proc(char *page, char **start, off_t off,
84                                  int count, int *eof, void *data)
85 {
86         int a, b, c;
87         int len;
88         unsigned long seq;
89
90         do {
91                 seq = read_seqbegin(&xtime_lock);
92                 a = avenrun[0] + (FIXED_1/200);
93                 b = avenrun[1] + (FIXED_1/200);
94                 c = avenrun[2] + (FIXED_1/200);
95         } while (read_seqretry(&xtime_lock, seq));
96
97         len = sprintf(page,"%d.%02d %d.%02d %d.%02d %ld/%d %d\n",
98                 LOAD_INT(a), LOAD_FRAC(a),
99                 LOAD_INT(b), LOAD_FRAC(b),
100                 LOAD_INT(c), LOAD_FRAC(c),
101                 nr_running(), nr_threads,
102                 task_active_pid_ns(current)->last_pid);
103         return proc_calc_metrics(page, start, off, count, eof, len);
104 }
105
106 static int uptime_read_proc(char *page, char **start, off_t off,
107                                  int count, int *eof, void *data)
108 {
109         struct timespec uptime;
110         struct timespec idle;
111         int len;
112         cputime_t idletime = cputime_add(init_task.utime, init_task.stime);
113
114         do_posix_clock_monotonic_gettime(&uptime);
115         monotonic_to_bootbased(&uptime);
116         cputime_to_timespec(idletime, &idle);
117         len = sprintf(page,"%lu.%02lu %lu.%02lu\n",
118                         (unsigned long) uptime.tv_sec,
119                         (uptime.tv_nsec / (NSEC_PER_SEC / 100)),
120                         (unsigned long) idle.tv_sec,
121                         (idle.tv_nsec / (NSEC_PER_SEC / 100)));
122
123         return proc_calc_metrics(page, start, off, count, eof, len);
124 }
125
126 static int meminfo_read_proc(char *page, char **start, off_t off,
127                                  int count, int *eof, void *data)
128 {
129         struct sysinfo i;
130         int len;
131         unsigned long committed;
132         unsigned long allowed;
133         struct vmalloc_info vmi;
134         long cached;
135
136 /*
137  * display in kilobytes.
138  */
139 #define K(x) ((x) << (PAGE_SHIFT - 10))
140         si_meminfo(&i);
141         si_swapinfo(&i);
142         committed = atomic_read(&vm_committed_space);
143         allowed = ((totalram_pages - hugetlb_total_pages())
144                 * sysctl_overcommit_ratio / 100) + total_swap_pages;
145
146         cached = global_page_state(NR_FILE_PAGES) -
147                         total_swapcache_pages - i.bufferram;
148         if (cached < 0)
149                 cached = 0;
150
151         get_vmalloc_info(&vmi);
152
153         /*
154          * Tagged format, for easy grepping and expansion.
155          */
156         len = sprintf(page,
157                 "MemTotal:     %8lu kB\n"
158                 "MemFree:      %8lu kB\n"
159                 "Buffers:      %8lu kB\n"
160                 "Cached:       %8lu kB\n"
161                 "SwapCached:   %8lu kB\n"
162                 "Active:       %8lu kB\n"
163                 "Inactive:     %8lu kB\n"
164 #ifdef CONFIG_HIGHMEM
165                 "HighTotal:    %8lu kB\n"
166                 "HighFree:     %8lu kB\n"
167                 "LowTotal:     %8lu kB\n"
168                 "LowFree:      %8lu kB\n"
169 #endif
170                 "SwapTotal:    %8lu kB\n"
171                 "SwapFree:     %8lu kB\n"
172                 "Dirty:        %8lu kB\n"
173                 "Writeback:    %8lu kB\n"
174                 "AnonPages:    %8lu kB\n"
175                 "Mapped:       %8lu kB\n"
176                 "Slab:         %8lu kB\n"
177                 "SReclaimable: %8lu kB\n"
178                 "SUnreclaim:   %8lu kB\n"
179                 "PageTables:   %8lu kB\n"
180                 "NFS_Unstable: %8lu kB\n"
181                 "Bounce:       %8lu kB\n"
182                 "CommitLimit:  %8lu kB\n"
183                 "Committed_AS: %8lu kB\n"
184                 "VmallocTotal: %8lu kB\n"
185                 "VmallocUsed:  %8lu kB\n"
186                 "VmallocChunk: %8lu kB\n",
187                 K(i.totalram),
188                 K(i.freeram),
189                 K(i.bufferram),
190                 K(cached),
191                 K(total_swapcache_pages),
192                 K(global_page_state(NR_ACTIVE)),
193                 K(global_page_state(NR_INACTIVE)),
194 #ifdef CONFIG_HIGHMEM
195                 K(i.totalhigh),
196                 K(i.freehigh),
197                 K(i.totalram-i.totalhigh),
198                 K(i.freeram-i.freehigh),
199 #endif
200                 K(i.totalswap),
201                 K(i.freeswap),
202                 K(global_page_state(NR_FILE_DIRTY)),
203                 K(global_page_state(NR_WRITEBACK)),
204                 K(global_page_state(NR_ANON_PAGES)),
205                 K(global_page_state(NR_FILE_MAPPED)),
206                 K(global_page_state(NR_SLAB_RECLAIMABLE) +
207                                 global_page_state(NR_SLAB_UNRECLAIMABLE)),
208                 K(global_page_state(NR_SLAB_RECLAIMABLE)),
209                 K(global_page_state(NR_SLAB_UNRECLAIMABLE)),
210                 K(global_page_state(NR_PAGETABLE)),
211                 K(global_page_state(NR_UNSTABLE_NFS)),
212                 K(global_page_state(NR_BOUNCE)),
213                 K(allowed),
214                 K(committed),
215                 (unsigned long)VMALLOC_TOTAL >> 10,
216                 vmi.used >> 10,
217                 vmi.largest_chunk >> 10
218                 );
219
220                 len += hugetlb_report_meminfo(page + len);
221
222         return proc_calc_metrics(page, start, off, count, eof, len);
223 #undef K
224 }
225
226 extern const struct seq_operations fragmentation_op;
227 static int fragmentation_open(struct inode *inode, struct file *file)
228 {
229         (void)inode;
230         return seq_open(file, &fragmentation_op);
231 }
232
233 static const struct file_operations fragmentation_file_operations = {
234         .open           = fragmentation_open,
235         .read           = seq_read,
236         .llseek         = seq_lseek,
237         .release        = seq_release,
238 };
239
240 extern const struct seq_operations pagetypeinfo_op;
241 static int pagetypeinfo_open(struct inode *inode, struct file *file)
242 {
243         return seq_open(file, &pagetypeinfo_op);
244 }
245
246 static const struct file_operations pagetypeinfo_file_ops = {
247         .open           = pagetypeinfo_open,
248         .read           = seq_read,
249         .llseek         = seq_lseek,
250         .release        = seq_release,
251 };
252
253 extern const struct seq_operations zoneinfo_op;
254 static int zoneinfo_open(struct inode *inode, struct file *file)
255 {
256         return seq_open(file, &zoneinfo_op);
257 }
258
259 static const struct file_operations proc_zoneinfo_file_operations = {
260         .open           = zoneinfo_open,
261         .read           = seq_read,
262         .llseek         = seq_lseek,
263         .release        = seq_release,
264 };
265
266 static int version_read_proc(char *page, char **start, off_t off,
267                                  int count, int *eof, void *data)
268 {
269         int len;
270
271         len = snprintf(page, PAGE_SIZE, linux_proc_banner,
272                 utsname()->sysname,
273                 utsname()->release,
274                 utsname()->version);
275         return proc_calc_metrics(page, start, off, count, eof, len);
276 }
277
278 extern const struct seq_operations cpuinfo_op;
279 static int cpuinfo_open(struct inode *inode, struct file *file)
280 {
281         return seq_open(file, &cpuinfo_op);
282 }
283
284 static const struct file_operations proc_cpuinfo_operations = {
285         .open           = cpuinfo_open,
286         .read           = seq_read,
287         .llseek         = seq_lseek,
288         .release        = seq_release,
289 };
290
291 static int devinfo_show(struct seq_file *f, void *v)
292 {
293         int i = *(loff_t *) v;
294
295         if (i < CHRDEV_MAJOR_HASH_SIZE) {
296                 if (i == 0)
297                         seq_printf(f, "Character devices:\n");
298                 chrdev_show(f, i);
299         }
300 #ifdef CONFIG_BLOCK
301         else {
302                 i -= CHRDEV_MAJOR_HASH_SIZE;
303                 if (i == 0)
304                         seq_printf(f, "\nBlock devices:\n");
305                 blkdev_show(f, i);
306         }
307 #endif
308         return 0;
309 }
310
311 static void *devinfo_start(struct seq_file *f, loff_t *pos)
312 {
313         if (*pos < (BLKDEV_MAJOR_HASH_SIZE + CHRDEV_MAJOR_HASH_SIZE))
314                 return pos;
315         return NULL;
316 }
317
318 static void *devinfo_next(struct seq_file *f, void *v, loff_t *pos)
319 {
320         (*pos)++;
321         if (*pos >= (BLKDEV_MAJOR_HASH_SIZE + CHRDEV_MAJOR_HASH_SIZE))
322                 return NULL;
323         return pos;
324 }
325
326 static void devinfo_stop(struct seq_file *f, void *v)
327 {
328         /* Nothing to do */
329 }
330
331 static const struct seq_operations devinfo_ops = {
332         .start = devinfo_start,
333         .next  = devinfo_next,
334         .stop  = devinfo_stop,
335         .show  = devinfo_show
336 };
337
338 static int devinfo_open(struct inode *inode, struct file *filp)
339 {
340         return seq_open(filp, &devinfo_ops);
341 }
342
343 static const struct file_operations proc_devinfo_operations = {
344         .open           = devinfo_open,
345         .read           = seq_read,
346         .llseek         = seq_lseek,
347         .release        = seq_release,
348 };
349
350 extern const struct seq_operations vmstat_op;
351 static int vmstat_open(struct inode *inode, struct file *file)
352 {
353         return seq_open(file, &vmstat_op);
354 }
355 static const struct file_operations proc_vmstat_file_operations = {
356         .open           = vmstat_open,
357         .read           = seq_read,
358         .llseek         = seq_lseek,
359         .release        = seq_release,
360 };
361
362 #ifdef CONFIG_PROC_HARDWARE
363 static int hardware_read_proc(char *page, char **start, off_t off,
364                                  int count, int *eof, void *data)
365 {
366         int len = get_hardware_list(page);
367         return proc_calc_metrics(page, start, off, count, eof, len);
368 }
369 #endif
370
371 #ifdef CONFIG_STRAM_PROC
372 static int stram_read_proc(char *page, char **start, off_t off,
373                                  int count, int *eof, void *data)
374 {
375         int len = get_stram_list(page);
376         return proc_calc_metrics(page, start, off, count, eof, len);
377 }
378 #endif
379
380 #ifdef CONFIG_BLOCK
381 static int partitions_open(struct inode *inode, struct file *file)
382 {
383         return seq_open(file, &partitions_op);
384 }
385 static const struct file_operations proc_partitions_operations = {
386         .open           = partitions_open,
387         .read           = seq_read,
388         .llseek         = seq_lseek,
389         .release        = seq_release,
390 };
391
392 static int diskstats_open(struct inode *inode, struct file *file)
393 {
394         return seq_open(file, &diskstats_op);
395 }
396 static const struct file_operations proc_diskstats_operations = {
397         .open           = diskstats_open,
398         .read           = seq_read,
399         .llseek         = seq_lseek,
400         .release        = seq_release,
401 };
402 #endif
403
404 #ifdef CONFIG_MODULES
405 extern const struct seq_operations modules_op;
406 static int modules_open(struct inode *inode, struct file *file)
407 {
408         return seq_open(file, &modules_op);
409 }
410 static const struct file_operations proc_modules_operations = {
411         .open           = modules_open,
412         .read           = seq_read,
413         .llseek         = seq_lseek,
414         .release        = seq_release,
415 };
416 #endif
417
418 #ifdef CONFIG_SLABINFO
419 static int slabinfo_open(struct inode *inode, struct file *file)
420 {
421         return seq_open(file, &slabinfo_op);
422 }
423 static const struct file_operations proc_slabinfo_operations = {
424         .open           = slabinfo_open,
425         .read           = seq_read,
426         .write          = slabinfo_write,
427         .llseek         = seq_lseek,
428         .release        = seq_release,
429 };
430
431 #ifdef CONFIG_DEBUG_SLAB_LEAK
432 extern const struct seq_operations slabstats_op;
433 static int slabstats_open(struct inode *inode, struct file *file)
434 {
435         unsigned long *n = kzalloc(PAGE_SIZE, GFP_KERNEL);
436         int ret = -ENOMEM;
437         if (n) {
438                 ret = seq_open(file, &slabstats_op);
439                 if (!ret) {
440                         struct seq_file *m = file->private_data;
441                         *n = PAGE_SIZE / (2 * sizeof(unsigned long));
442                         m->private = n;
443                         n = NULL;
444                 }
445                 kfree(n);
446         }
447         return ret;
448 }
449
450 static const struct file_operations proc_slabstats_operations = {
451         .open           = slabstats_open,
452         .read           = seq_read,
453         .llseek         = seq_lseek,
454         .release        = seq_release_private,
455 };
456 #endif
457 #endif
458
459 static int show_stat(struct seq_file *p, void *v)
460 {
461         int i;
462         unsigned long jif;
463         cputime64_t user, nice, system, idle, iowait, irq, softirq, steal;
464         cputime64_t guest;
465         u64 sum = 0;
466         struct timespec boottime;
467         unsigned int *per_irq_sum;
468
469         per_irq_sum = kzalloc(sizeof(unsigned int)*NR_IRQS, GFP_KERNEL);
470         if (!per_irq_sum)
471                 return -ENOMEM;
472
473         user = nice = system = idle = iowait =
474                 irq = softirq = steal = cputime64_zero;
475         guest = cputime64_zero;
476         getboottime(&boottime);
477         jif = boottime.tv_sec;
478
479         for_each_possible_cpu(i) {
480                 int j;
481
482                 user = cputime64_add(user, kstat_cpu(i).cpustat.user);
483                 nice = cputime64_add(nice, kstat_cpu(i).cpustat.nice);
484                 system = cputime64_add(system, kstat_cpu(i).cpustat.system);
485                 idle = cputime64_add(idle, kstat_cpu(i).cpustat.idle);
486                 iowait = cputime64_add(iowait, kstat_cpu(i).cpustat.iowait);
487                 irq = cputime64_add(irq, kstat_cpu(i).cpustat.irq);
488                 softirq = cputime64_add(softirq, kstat_cpu(i).cpustat.softirq);
489                 steal = cputime64_add(steal, kstat_cpu(i).cpustat.steal);
490                 guest = cputime64_add(guest, kstat_cpu(i).cpustat.guest);
491                 for (j = 0; j < NR_IRQS; j++) {
492                         unsigned int temp = kstat_cpu(i).irqs[j];
493                         sum += temp;
494                         per_irq_sum[j] += temp;
495                 }
496         }
497
498         seq_printf(p, "cpu  %llu %llu %llu %llu %llu %llu %llu %llu %llu\n",
499                 (unsigned long long)cputime64_to_clock_t(user),
500                 (unsigned long long)cputime64_to_clock_t(nice),
501                 (unsigned long long)cputime64_to_clock_t(system),
502                 (unsigned long long)cputime64_to_clock_t(idle),
503                 (unsigned long long)cputime64_to_clock_t(iowait),
504                 (unsigned long long)cputime64_to_clock_t(irq),
505                 (unsigned long long)cputime64_to_clock_t(softirq),
506                 (unsigned long long)cputime64_to_clock_t(steal),
507                 (unsigned long long)cputime64_to_clock_t(guest));
508         for_each_online_cpu(i) {
509
510                 /* Copy values here to work around gcc-2.95.3, gcc-2.96 */
511                 user = kstat_cpu(i).cpustat.user;
512                 nice = kstat_cpu(i).cpustat.nice;
513                 system = kstat_cpu(i).cpustat.system;
514                 idle = kstat_cpu(i).cpustat.idle;
515                 iowait = kstat_cpu(i).cpustat.iowait;
516                 irq = kstat_cpu(i).cpustat.irq;
517                 softirq = kstat_cpu(i).cpustat.softirq;
518                 steal = kstat_cpu(i).cpustat.steal;
519                 guest = kstat_cpu(i).cpustat.guest;
520                 seq_printf(p,
521                         "cpu%d %llu %llu %llu %llu %llu %llu %llu %llu %llu\n",
522                         i,
523                         (unsigned long long)cputime64_to_clock_t(user),
524                         (unsigned long long)cputime64_to_clock_t(nice),
525                         (unsigned long long)cputime64_to_clock_t(system),
526                         (unsigned long long)cputime64_to_clock_t(idle),
527                         (unsigned long long)cputime64_to_clock_t(iowait),
528                         (unsigned long long)cputime64_to_clock_t(irq),
529                         (unsigned long long)cputime64_to_clock_t(softirq),
530                         (unsigned long long)cputime64_to_clock_t(steal),
531                         (unsigned long long)cputime64_to_clock_t(guest));
532         }
533         seq_printf(p, "intr %llu", (unsigned long long)sum);
534
535         for (i = 0; i < NR_IRQS; i++)
536                 seq_printf(p, " %u", per_irq_sum[i]);
537
538         seq_printf(p,
539                 "\nctxt %llu\n"
540                 "btime %lu\n"
541                 "processes %lu\n"
542                 "procs_running %lu\n"
543                 "procs_blocked %lu\n",
544                 nr_context_switches(),
545                 (unsigned long)jif,
546                 total_forks,
547                 nr_running(),
548                 nr_iowait());
549
550         kfree(per_irq_sum);
551         return 0;
552 }
553
554 static int stat_open(struct inode *inode, struct file *file)
555 {
556         unsigned size = 4096 * (1 + num_possible_cpus() / 32);
557         char *buf;
558         struct seq_file *m;
559         int res;
560
561         /* don't ask for more than the kmalloc() max size, currently 128 KB */
562         if (size > 128 * 1024)
563                 size = 128 * 1024;
564         buf = kmalloc(size, GFP_KERNEL);
565         if (!buf)
566                 return -ENOMEM;
567
568         res = single_open(file, show_stat, NULL);
569         if (!res) {
570                 m = file->private_data;
571                 m->buf = buf;
572                 m->size = size;
573         } else
574                 kfree(buf);
575         return res;
576 }
577 static const struct file_operations proc_stat_operations = {
578         .open           = stat_open,
579         .read           = seq_read,
580         .llseek         = seq_lseek,
581         .release        = single_release,
582 };
583
584 /*
585  * /proc/interrupts
586  */
587 static void *int_seq_start(struct seq_file *f, loff_t *pos)
588 {
589         return (*pos <= NR_IRQS) ? pos : NULL;
590 }
591
592 static void *int_seq_next(struct seq_file *f, void *v, loff_t *pos)
593 {
594         (*pos)++;
595         if (*pos > NR_IRQS)
596                 return NULL;
597         return pos;
598 }
599
600 static void int_seq_stop(struct seq_file *f, void *v)
601 {
602         /* Nothing to do */
603 }
604
605
606 static const struct seq_operations int_seq_ops = {
607         .start = int_seq_start,
608         .next  = int_seq_next,
609         .stop  = int_seq_stop,
610         .show  = show_interrupts
611 };
612
613 static int interrupts_open(struct inode *inode, struct file *filp)
614 {
615         return seq_open(filp, &int_seq_ops);
616 }
617
618 static const struct file_operations proc_interrupts_operations = {
619         .open           = interrupts_open,
620         .read           = seq_read,
621         .llseek         = seq_lseek,
622         .release        = seq_release,
623 };
624
625 static int filesystems_read_proc(char *page, char **start, off_t off,
626                                  int count, int *eof, void *data)
627 {
628         int len = get_filesystem_list(page);
629         return proc_calc_metrics(page, start, off, count, eof, len);
630 }
631
632 static int cmdline_read_proc(char *page, char **start, off_t off,
633                                  int count, int *eof, void *data)
634 {
635         int len;
636
637         len = sprintf(page, "%s\n", saved_command_line);
638         return proc_calc_metrics(page, start, off, count, eof, len);
639 }
640
641 static int locks_open(struct inode *inode, struct file *filp)
642 {
643         return seq_open(filp, &locks_seq_operations);
644 }
645
646 static const struct file_operations proc_locks_operations = {
647         .open           = locks_open,
648         .read           = seq_read,
649         .llseek         = seq_lseek,
650         .release        = seq_release,
651 };
652
653 static int execdomains_read_proc(char *page, char **start, off_t off,
654                                  int count, int *eof, void *data)
655 {
656         int len = get_exec_domain_list(page);
657         return proc_calc_metrics(page, start, off, count, eof, len);
658 }
659
660 #ifdef CONFIG_MAGIC_SYSRQ
661 /*
662  * writing 'C' to /proc/sysrq-trigger is like sysrq-C
663  */
664 static ssize_t write_sysrq_trigger(struct file *file, const char __user *buf,
665                                    size_t count, loff_t *ppos)
666 {
667         if (count) {
668                 char c;
669
670                 if (get_user(c, buf))
671                         return -EFAULT;
672                 __handle_sysrq(c, NULL, 0);
673         }
674         return count;
675 }
676
677 static const struct file_operations proc_sysrq_trigger_operations = {
678         .write          = write_sysrq_trigger,
679 };
680 #endif
681
682 #ifdef CONFIG_PROC_PAGE_MONITOR
683 #define KPMSIZE sizeof(u64)
684 #define KPMMASK (KPMSIZE - 1)
685 /* /proc/kpagecount - an array exposing page counts
686  *
687  * Each entry is a u64 representing the corresponding
688  * physical page count.
689  */
690 static ssize_t kpagecount_read(struct file *file, char __user *buf,
691                              size_t count, loff_t *ppos)
692 {
693         u64 __user *out = (u64 __user *)buf;
694         struct page *ppage;
695         unsigned long src = *ppos;
696         unsigned long pfn;
697         ssize_t ret = 0;
698         u64 pcount;
699
700         pfn = src / KPMSIZE;
701         count = min_t(size_t, count, (max_pfn * KPMSIZE) - src);
702         if (src & KPMMASK || count & KPMMASK)
703                 return -EIO;
704
705         while (count > 0) {
706                 ppage = NULL;
707                 if (pfn_valid(pfn))
708                         ppage = pfn_to_page(pfn);
709                 pfn++;
710                 if (!ppage)
711                         pcount = 0;
712                 else
713                         pcount = atomic_read(&ppage->_count);
714
715                 if (put_user(pcount, out++)) {
716                         ret = -EFAULT;
717                         break;
718                 }
719
720                 count -= KPMSIZE;
721         }
722
723         *ppos += (char __user *)out - buf;
724         if (!ret)
725                 ret = (char __user *)out - buf;
726         return ret;
727 }
728
729 static struct file_operations proc_kpagecount_operations = {
730         .llseek = mem_lseek,
731         .read = kpagecount_read,
732 };
733
734 /* /proc/kpageflags - an array exposing page flags
735  *
736  * Each entry is a u64 representing the corresponding
737  * physical page flags.
738  */
739
740 /* These macros are used to decouple internal flags from exported ones */
741
742 #define KPF_LOCKED     0
743 #define KPF_ERROR      1
744 #define KPF_REFERENCED 2
745 #define KPF_UPTODATE   3
746 #define KPF_DIRTY      4
747 #define KPF_LRU        5
748 #define KPF_ACTIVE     6
749 #define KPF_SLAB       7
750 #define KPF_WRITEBACK  8
751 #define KPF_RECLAIM    9
752 #define KPF_BUDDY     10
753
754 #define kpf_copy_bit(flags, srcpos, dstpos) (((flags >> srcpos) & 1) << dstpos)
755
756 static ssize_t kpageflags_read(struct file *file, char __user *buf,
757                              size_t count, loff_t *ppos)
758 {
759         u64 __user *out = (u64 __user *)buf;
760         struct page *ppage;
761         unsigned long src = *ppos;
762         unsigned long pfn;
763         ssize_t ret = 0;
764         u64 kflags, uflags;
765
766         pfn = src / KPMSIZE;
767         count = min_t(unsigned long, count, (max_pfn * KPMSIZE) - src);
768         if (src & KPMMASK || count & KPMMASK)
769                 return -EIO;
770
771         while (count > 0) {
772                 ppage = NULL;
773                 if (pfn_valid(pfn))
774                         ppage = pfn_to_page(pfn);
775                 pfn++;
776                 if (!ppage)
777                         kflags = 0;
778                 else
779                         kflags = ppage->flags;
780
781                 uflags = kpf_copy_bit(KPF_LOCKED, PG_locked, kflags) |
782                         kpf_copy_bit(kflags, KPF_ERROR, PG_error) |
783                         kpf_copy_bit(kflags, KPF_REFERENCED, PG_referenced) |
784                         kpf_copy_bit(kflags, KPF_UPTODATE, PG_uptodate) |
785                         kpf_copy_bit(kflags, KPF_DIRTY, PG_dirty) |
786                         kpf_copy_bit(kflags, KPF_LRU, PG_lru) |
787                         kpf_copy_bit(kflags, KPF_ACTIVE, PG_active) |
788                         kpf_copy_bit(kflags, KPF_SLAB, PG_slab) |
789                         kpf_copy_bit(kflags, KPF_WRITEBACK, PG_writeback) |
790                         kpf_copy_bit(kflags, KPF_RECLAIM, PG_reclaim) |
791                         kpf_copy_bit(kflags, KPF_BUDDY, PG_buddy);
792
793                 if (put_user(uflags, out++)) {
794                         ret = -EFAULT;
795                         break;
796                 }
797
798                 count -= KPMSIZE;
799         }
800
801         *ppos += (char __user *)out - buf;
802         if (!ret)
803                 ret = (char __user *)out - buf;
804         return ret;
805 }
806
807 static struct file_operations proc_kpageflags_operations = {
808         .llseek = mem_lseek,
809         .read = kpageflags_read,
810 };
811 #endif /* CONFIG_PROC_PAGE_MONITOR */
812
813 struct proc_dir_entry *proc_root_kcore;
814
815 void create_seq_entry(char *name, mode_t mode, const struct file_operations *f)
816 {
817         struct proc_dir_entry *entry;
818         entry = create_proc_entry(name, mode, NULL);
819         if (entry)
820                 entry->proc_fops = f;
821 }
822
823 void __init proc_misc_init(void)
824 {
825         static struct {
826                 char *name;
827                 int (*read_proc)(char*,char**,off_t,int,int*,void*);
828         } *p, simple_ones[] = {
829                 {"loadavg",     loadavg_read_proc},
830                 {"uptime",      uptime_read_proc},
831                 {"meminfo",     meminfo_read_proc},
832                 {"version",     version_read_proc},
833 #ifdef CONFIG_PROC_HARDWARE
834                 {"hardware",    hardware_read_proc},
835 #endif
836 #ifdef CONFIG_STRAM_PROC
837                 {"stram",       stram_read_proc},
838 #endif
839                 {"filesystems", filesystems_read_proc},
840                 {"cmdline",     cmdline_read_proc},
841                 {"execdomains", execdomains_read_proc},
842                 {NULL,}
843         };
844         for (p = simple_ones; p->name; p++)
845                 create_proc_read_entry(p->name, 0, NULL, p->read_proc, NULL);
846
847         proc_symlink("mounts", NULL, "self/mounts");
848
849         /* And now for trickier ones */
850 #ifdef CONFIG_PRINTK
851         {
852                 struct proc_dir_entry *entry;
853                 entry = create_proc_entry("kmsg", S_IRUSR, &proc_root);
854                 if (entry)
855                         entry->proc_fops = &proc_kmsg_operations;
856         }
857 #endif
858         create_seq_entry("locks", 0, &proc_locks_operations);
859         create_seq_entry("devices", 0, &proc_devinfo_operations);
860         create_seq_entry("cpuinfo", 0, &proc_cpuinfo_operations);
861 #ifdef CONFIG_BLOCK
862         create_seq_entry("partitions", 0, &proc_partitions_operations);
863 #endif
864         create_seq_entry("stat", 0, &proc_stat_operations);
865         create_seq_entry("interrupts", 0, &proc_interrupts_operations);
866 #ifdef CONFIG_SLABINFO
867         create_seq_entry("slabinfo",S_IWUSR|S_IRUGO,&proc_slabinfo_operations);
868 #ifdef CONFIG_DEBUG_SLAB_LEAK
869         create_seq_entry("slab_allocators", 0 ,&proc_slabstats_operations);
870 #endif
871 #endif
872         create_seq_entry("buddyinfo",S_IRUGO, &fragmentation_file_operations);
873         create_seq_entry("pagetypeinfo", S_IRUGO, &pagetypeinfo_file_ops);
874         create_seq_entry("vmstat",S_IRUGO, &proc_vmstat_file_operations);
875         create_seq_entry("zoneinfo",S_IRUGO, &proc_zoneinfo_file_operations);
876 #ifdef CONFIG_BLOCK
877         create_seq_entry("diskstats", 0, &proc_diskstats_operations);
878 #endif
879 #ifdef CONFIG_MODULES
880         create_seq_entry("modules", 0, &proc_modules_operations);
881 #endif
882 #ifdef CONFIG_SCHEDSTATS
883         create_seq_entry("schedstat", 0, &proc_schedstat_operations);
884 #endif
885 #ifdef CONFIG_PROC_KCORE
886         proc_root_kcore = create_proc_entry("kcore", S_IRUSR, NULL);
887         if (proc_root_kcore) {
888                 proc_root_kcore->proc_fops = &proc_kcore_operations;
889                 proc_root_kcore->size =
890                                 (size_t)high_memory - PAGE_OFFSET + PAGE_SIZE;
891         }
892 #endif
893 #ifdef CONFIG_PROC_PAGE_MONITOR
894         create_seq_entry("kpagecount", S_IRUSR, &proc_kpagecount_operations);
895         create_seq_entry("kpageflags", S_IRUSR, &proc_kpageflags_operations);
896 #endif
897 #ifdef CONFIG_PROC_VMCORE
898         proc_vmcore = create_proc_entry("vmcore", S_IRUSR, NULL);
899         if (proc_vmcore)
900                 proc_vmcore->proc_fops = &proc_vmcore_operations;
901 #endif
902 #ifdef CONFIG_MAGIC_SYSRQ
903         {
904                 struct proc_dir_entry *entry;
905                 entry = create_proc_entry("sysrq-trigger", S_IWUSR, NULL);
906                 if (entry)
907                         entry->proc_fops = &proc_sysrq_trigger_operations;
908         }
909 #endif
910 }