vserver 2.0 rc7
[linux-2.6.git] / 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/tty.h>
24 #include <linux/string.h>
25 #include <linux/mman.h>
26 #include <linux/proc_fs.h>
27 #include <linux/ioport.h>
28 #include <linux/config.h>
29 #include <linux/mm.h>
30 #include <linux/mmzone.h>
31 #include <linux/pagemap.h>
32 #include <linux/swap.h>
33 #include <linux/slab.h>
34 #include <linux/smp.h>
35 #include <linux/signal.h>
36 #include <linux/module.h>
37 #include <linux/init.h>
38 #include <linux/smp_lock.h>
39 #include <linux/seq_file.h>
40 #include <linux/times.h>
41 #include <linux/profile.h>
42 #include <linux/blkdev.h>
43 #include <linux/hugetlb.h>
44 #include <linux/jiffies.h>
45 #include <linux/sysrq.h>
46 #include <linux/vmalloc.h>
47 #include <asm/uaccess.h>
48 #include <asm/pgtable.h>
49 #include <asm/io.h>
50 #include <asm/tlb.h>
51 #include <asm/div64.h>
52 #include "internal.h"
53
54 #include <linux/vs_cvirt.h>
55
56 #define LOAD_INT(x) ((x) >> FSHIFT)
57 #define LOAD_FRAC(x) LOAD_INT(((x) & (FIXED_1-1)) * 100)
58 /*
59  * Warning: stuff below (imported functions) assumes that its output will fit
60  * into one page. For some of those functions it may be wrong. Moreover, we
61  * have a way to deal with that gracefully. Right now I used straightforward
62  * wrappers, but this needs further analysis wrt potential overflows.
63  */
64 extern int get_hardware_list(char *);
65 extern int get_stram_list(char *);
66 extern int get_chrdev_list(char *);
67 extern int get_filesystem_list(char *);
68 extern int get_exec_domain_list(char *);
69 extern int get_dma_list(char *);
70 extern int get_locks_status (char *, char **, off_t, int);
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         unsigned int running, threads;
87         int a, b, c;
88         int len;
89
90         if (vx_flags(VXF_VIRT_LOAD, 0)) {
91                 struct vx_info *vxi = current->vx_info;
92
93                 a = vxi->cvirt.load[0] + (FIXED_1/200);
94                 b = vxi->cvirt.load[1] + (FIXED_1/200);
95                 c = vxi->cvirt.load[2] + (FIXED_1/200);
96
97                 running = atomic_read(&vxi->cvirt.nr_running);
98                 threads = atomic_read(&vxi->cvirt.nr_threads);
99         } else {
100                 a = avenrun[0] + (FIXED_1/200);
101                 b = avenrun[1] + (FIXED_1/200);
102                 c = avenrun[2] + (FIXED_1/200);
103
104                 running = nr_running();
105                 threads = nr_threads;
106         }
107         len = sprintf(page,"%d.%02d %d.%02d %d.%02d %d/%d %d\n",
108                 LOAD_INT(a), LOAD_FRAC(a),
109                 LOAD_INT(b), LOAD_FRAC(b),
110                 LOAD_INT(c), LOAD_FRAC(c),
111                 running, threads, last_pid);
112         return proc_calc_metrics(page, start, off, count, eof, len);
113 }
114
115 static int uptime_read_proc(char *page, char **start, off_t off,
116                                  int count, int *eof, void *data)
117 {
118         struct timespec uptime;
119         struct timespec idle;
120         int len;
121         cputime_t idletime = cputime_add(init_task.utime, init_task.stime);
122
123         do_posix_clock_monotonic_gettime(&uptime);
124         cputime_to_timespec(idletime, &idle);
125         if (vx_flags(VXF_VIRT_UPTIME, 0))
126                 vx_vsi_uptime(&uptime, &idle);
127
128         len = sprintf(page,"%lu.%02lu %lu.%02lu\n",
129                         (unsigned long) uptime.tv_sec,
130                         (uptime.tv_nsec / (NSEC_PER_SEC / 100)),
131                         (unsigned long) idle.tv_sec,
132                         (idle.tv_nsec / (NSEC_PER_SEC / 100)));
133
134         return proc_calc_metrics(page, start, off, count, eof, len);
135 }
136
137 static int meminfo_read_proc(char *page, char **start, off_t off,
138                                  int count, int *eof, void *data)
139 {
140         struct sysinfo i;
141         int len;
142         struct page_state ps;
143         unsigned long inactive;
144         unsigned long active;
145         unsigned long free;
146         unsigned long committed;
147         unsigned long allowed;
148         struct vmalloc_info vmi;
149         long cached;
150
151         get_page_state(&ps);
152         get_zone_counts(&active, &inactive, &free);
153
154 /*
155  * display in kilobytes.
156  */
157 #define K(x) ((x) << (PAGE_SHIFT - 10))
158         si_meminfo(&i);
159         si_swapinfo(&i);
160         committed = atomic_read(&vm_committed_space);
161         allowed = ((totalram_pages - hugetlb_total_pages())
162                 * sysctl_overcommit_ratio / 100) + total_swap_pages;
163
164         cached = get_page_cache_size() - total_swapcache_pages - i.bufferram;
165         if (cached < 0)
166                 cached = 0;
167
168         get_vmalloc_info(&vmi);
169
170         /*
171          * Tagged format, for easy grepping and expansion.
172          */
173         len = sprintf(page,
174                 "MemTotal:     %8lu kB\n"
175                 "MemFree:      %8lu kB\n"
176                 "Buffers:      %8lu kB\n"
177                 "Cached:       %8lu kB\n"
178                 "SwapCached:   %8lu kB\n"
179                 "Active:       %8lu kB\n"
180                 "Inactive:     %8lu kB\n"
181                 "HighTotal:    %8lu kB\n"
182                 "HighFree:     %8lu kB\n"
183                 "LowTotal:     %8lu kB\n"
184                 "LowFree:      %8lu kB\n"
185                 "SwapTotal:    %8lu kB\n"
186                 "SwapFree:     %8lu kB\n"
187                 "Dirty:        %8lu kB\n"
188                 "Writeback:    %8lu kB\n"
189                 "Mapped:       %8lu kB\n"
190                 "Slab:         %8lu kB\n"
191                 "CommitLimit:  %8lu kB\n"
192                 "Committed_AS: %8lu kB\n"
193                 "PageTables:   %8lu kB\n"
194                 "VmallocTotal: %8lu kB\n"
195                 "VmallocUsed:  %8lu kB\n"
196                 "VmallocChunk: %8lu kB\n",
197                 K(i.totalram),
198                 K(i.freeram),
199                 K(i.bufferram),
200                 K(cached),
201                 K(total_swapcache_pages),
202                 K(active),
203                 K(inactive),
204                 K(i.totalhigh),
205                 K(i.freehigh),
206                 K(i.totalram-i.totalhigh),
207                 K(i.freeram-i.freehigh),
208                 K(i.totalswap),
209                 K(i.freeswap),
210                 K(ps.nr_dirty),
211                 K(ps.nr_writeback),
212                 K(ps.nr_mapped),
213                 K(ps.nr_slab),
214                 K(allowed),
215                 K(committed),
216                 K(ps.nr_page_table_pages),
217                 (unsigned long)VMALLOC_TOTAL >> 10,
218                 vmi.used >> 10,
219                 vmi.largest_chunk >> 10
220                 );
221
222                 len += hugetlb_report_meminfo(page + len);
223
224         return proc_calc_metrics(page, start, off, count, eof, len);
225 #undef K
226 }
227
228 extern struct seq_operations fragmentation_op;
229 static int fragmentation_open(struct inode *inode, struct file *file)
230 {
231         (void)inode;
232         return seq_open(file, &fragmentation_op);
233 }
234
235 static struct file_operations fragmentation_file_operations = {
236         .open           = fragmentation_open,
237         .read           = seq_read,
238         .llseek         = seq_lseek,
239         .release        = seq_release,
240 };
241
242 static int version_read_proc(char *page, char **start, off_t off,
243                                  int count, int *eof, void *data)
244 {
245         int len;
246
247         len = sprintf(page, vx_linux_banner,
248                 vx_new_uts(release),
249                 vx_new_uts(version));
250         return proc_calc_metrics(page, start, off, count, eof, len);
251 }
252
253 extern struct seq_operations cpuinfo_op;
254 static int cpuinfo_open(struct inode *inode, struct file *file)
255 {
256         return seq_open(file, &cpuinfo_op);
257 }
258 static struct file_operations proc_cpuinfo_operations = {
259         .open           = cpuinfo_open,
260         .read           = seq_read,
261         .llseek         = seq_lseek,
262         .release        = seq_release,
263 };
264
265 extern struct seq_operations vmstat_op;
266 static int vmstat_open(struct inode *inode, struct file *file)
267 {
268         return seq_open(file, &vmstat_op);
269 }
270 static struct file_operations proc_vmstat_file_operations = {
271         .open           = vmstat_open,
272         .read           = seq_read,
273         .llseek         = seq_lseek,
274         .release        = seq_release,
275 };
276
277 #ifdef CONFIG_PROC_HARDWARE
278 static int hardware_read_proc(char *page, char **start, off_t off,
279                                  int count, int *eof, void *data)
280 {
281         int len = get_hardware_list(page);
282         return proc_calc_metrics(page, start, off, count, eof, len);
283 }
284 #endif
285
286 #ifdef CONFIG_STRAM_PROC
287 static int stram_read_proc(char *page, char **start, off_t off,
288                                  int count, int *eof, void *data)
289 {
290         int len = get_stram_list(page);
291         return proc_calc_metrics(page, start, off, count, eof, len);
292 }
293 #endif
294
295 extern struct seq_operations partitions_op;
296 static int partitions_open(struct inode *inode, struct file *file)
297 {
298         return seq_open(file, &partitions_op);
299 }
300 static struct file_operations proc_partitions_operations = {
301         .open           = partitions_open,
302         .read           = seq_read,
303         .llseek         = seq_lseek,
304         .release        = seq_release,
305 };
306
307 extern struct seq_operations diskstats_op;
308 static int diskstats_open(struct inode *inode, struct file *file)
309 {
310         return seq_open(file, &diskstats_op);
311 }
312 static struct file_operations proc_diskstats_operations = {
313         .open           = diskstats_open,
314         .read           = seq_read,
315         .llseek         = seq_lseek,
316         .release        = seq_release,
317 };
318
319 #ifdef CONFIG_MODULES
320 extern struct seq_operations modules_op;
321 static int modules_open(struct inode *inode, struct file *file)
322 {
323         return seq_open(file, &modules_op);
324 }
325 static struct file_operations proc_modules_operations = {
326         .open           = modules_open,
327         .read           = seq_read,
328         .llseek         = seq_lseek,
329         .release        = seq_release,
330 };
331 #endif
332
333 extern struct seq_operations slabinfo_op;
334 extern ssize_t slabinfo_write(struct file *, const char __user *, size_t, loff_t *);
335 static int slabinfo_open(struct inode *inode, struct file *file)
336 {
337         return seq_open(file, &slabinfo_op);
338 }
339 static struct file_operations proc_slabinfo_operations = {
340         .open           = slabinfo_open,
341         .read           = seq_read,
342         .write          = slabinfo_write,
343         .llseek         = seq_lseek,
344         .release        = seq_release,
345 };
346
347 static int show_stat(struct seq_file *p, void *v)
348 {
349         int i;
350         unsigned long jif;
351         cputime64_t user, nice, system, idle, iowait, irq, softirq, steal;
352         u64 sum = 0;
353
354         user = nice = system = idle = iowait =
355                 irq = softirq = steal = cputime64_zero;
356         jif = - wall_to_monotonic.tv_sec;
357         if (wall_to_monotonic.tv_nsec)
358                 --jif;
359
360         for_each_cpu(i) {
361                 int j;
362
363                 user = cputime64_add(user, kstat_cpu(i).cpustat.user);
364                 nice = cputime64_add(nice, kstat_cpu(i).cpustat.nice);
365                 system = cputime64_add(system, kstat_cpu(i).cpustat.system);
366                 idle = cputime64_add(idle, kstat_cpu(i).cpustat.idle);
367                 iowait = cputime64_add(iowait, kstat_cpu(i).cpustat.iowait);
368                 irq = cputime64_add(irq, kstat_cpu(i).cpustat.irq);
369                 softirq = cputime64_add(softirq, kstat_cpu(i).cpustat.softirq);
370                 steal = cputime64_add(steal, kstat_cpu(i).cpustat.steal);
371                 for (j = 0 ; j < NR_IRQS ; j++)
372                         sum += kstat_cpu(i).irqs[j];
373         }
374
375         seq_printf(p, "cpu  %llu %llu %llu %llu %llu %llu %llu %llu\n",
376                 (unsigned long long)cputime64_to_clock_t(user),
377                 (unsigned long long)cputime64_to_clock_t(nice),
378                 (unsigned long long)cputime64_to_clock_t(system),
379                 (unsigned long long)cputime64_to_clock_t(idle),
380                 (unsigned long long)cputime64_to_clock_t(iowait),
381                 (unsigned long long)cputime64_to_clock_t(irq),
382                 (unsigned long long)cputime64_to_clock_t(softirq),
383                 (unsigned long long)cputime64_to_clock_t(steal));
384         for_each_online_cpu(i) {
385
386                 /* Copy values here to work around gcc-2.95.3, gcc-2.96 */
387                 user = kstat_cpu(i).cpustat.user;
388                 nice = kstat_cpu(i).cpustat.nice;
389                 system = kstat_cpu(i).cpustat.system;
390                 idle = kstat_cpu(i).cpustat.idle;
391                 iowait = kstat_cpu(i).cpustat.iowait;
392                 irq = kstat_cpu(i).cpustat.irq;
393                 softirq = kstat_cpu(i).cpustat.softirq;
394                 steal = kstat_cpu(i).cpustat.steal;
395                 seq_printf(p, "cpu%d %llu %llu %llu %llu %llu %llu %llu %llu\n",
396                         i,
397                         (unsigned long long)cputime64_to_clock_t(user),
398                         (unsigned long long)cputime64_to_clock_t(nice),
399                         (unsigned long long)cputime64_to_clock_t(system),
400                         (unsigned long long)cputime64_to_clock_t(idle),
401                         (unsigned long long)cputime64_to_clock_t(iowait),
402                         (unsigned long long)cputime64_to_clock_t(irq),
403                         (unsigned long long)cputime64_to_clock_t(softirq),
404                         (unsigned long long)cputime64_to_clock_t(steal));
405         }
406         seq_printf(p, "intr %llu", (unsigned long long)sum);
407
408 #if !defined(CONFIG_PPC64) && !defined(CONFIG_ALPHA)
409         for (i = 0; i < NR_IRQS; i++)
410                 seq_printf(p, " %u", kstat_irqs(i));
411 #endif
412
413         seq_printf(p,
414                 "\nctxt %llu\n"
415                 "btime %lu\n"
416                 "processes %lu\n"
417                 "procs_running %lu\n"
418                 "procs_blocked %lu\n",
419                 nr_context_switches(),
420                 (unsigned long)jif,
421                 total_forks,
422                 nr_running(),
423                 nr_iowait());
424
425         return 0;
426 }
427
428 static int stat_open(struct inode *inode, struct file *file)
429 {
430         unsigned size = 4096 * (1 + num_possible_cpus() / 32);
431         char *buf;
432         struct seq_file *m;
433         int res;
434
435         /* don't ask for more than the kmalloc() max size, currently 128 KB */
436         if (size > 128 * 1024)
437                 size = 128 * 1024;
438         buf = kmalloc(size, GFP_KERNEL);
439         if (!buf)
440                 return -ENOMEM;
441
442         res = single_open(file, show_stat, NULL);
443         if (!res) {
444                 m = file->private_data;
445                 m->buf = buf;
446                 m->size = size;
447         } else
448                 kfree(buf);
449         return res;
450 }
451 static struct file_operations proc_stat_operations = {
452         .open           = stat_open,
453         .read           = seq_read,
454         .llseek         = seq_lseek,
455         .release        = single_release,
456 };
457
458 static int devices_read_proc(char *page, char **start, off_t off,
459                                  int count, int *eof, void *data)
460 {
461         int len = get_chrdev_list(page);
462         len += get_blkdev_list(page+len);
463         return proc_calc_metrics(page, start, off, count, eof, len);
464 }
465
466 /*
467  * /proc/interrupts
468  */
469 static void *int_seq_start(struct seq_file *f, loff_t *pos)
470 {
471         return (*pos <= NR_IRQS) ? pos : NULL;
472 }
473
474 static void *int_seq_next(struct seq_file *f, void *v, loff_t *pos)
475 {
476         (*pos)++;
477         if (*pos > NR_IRQS)
478                 return NULL;
479         return pos;
480 }
481
482 static void int_seq_stop(struct seq_file *f, void *v)
483 {
484         /* Nothing to do */
485 }
486
487
488 extern int show_interrupts(struct seq_file *f, void *v); /* In arch code */
489 static struct seq_operations int_seq_ops = {
490         .start = int_seq_start,
491         .next  = int_seq_next,
492         .stop  = int_seq_stop,
493         .show  = show_interrupts
494 };
495
496 static int interrupts_open(struct inode *inode, struct file *filp)
497 {
498         return seq_open(filp, &int_seq_ops);
499 }
500
501 static struct file_operations proc_interrupts_operations = {
502         .open           = interrupts_open,
503         .read           = seq_read,
504         .llseek         = seq_lseek,
505         .release        = seq_release,
506 };
507
508 static int filesystems_read_proc(char *page, char **start, off_t off,
509                                  int count, int *eof, void *data)
510 {
511         int len = get_filesystem_list(page);
512         return proc_calc_metrics(page, start, off, count, eof, len);
513 }
514
515 static int cmdline_read_proc(char *page, char **start, off_t off,
516                                  int count, int *eof, void *data)
517 {
518         int len;
519
520         len = sprintf(page, "%s\n", saved_command_line);
521         return proc_calc_metrics(page, start, off, count, eof, len);
522 }
523
524 static int locks_read_proc(char *page, char **start, off_t off,
525                                  int count, int *eof, void *data)
526 {
527         int len = get_locks_status(page, start, off, count);
528
529         if (len < count)
530                 *eof = 1;
531         return len;
532 }
533
534 static int execdomains_read_proc(char *page, char **start, off_t off,
535                                  int count, int *eof, void *data)
536 {
537         int len = get_exec_domain_list(page);
538         return proc_calc_metrics(page, start, off, count, eof, len);
539 }
540
541 #ifdef CONFIG_MAGIC_SYSRQ
542 /*
543  * writing 'C' to /proc/sysrq-trigger is like sysrq-C
544  */
545 static ssize_t write_sysrq_trigger(struct file *file, const char __user *buf,
546                                    size_t count, loff_t *ppos)
547 {
548         if (count) {
549                 char c;
550
551                 if (get_user(c, buf))
552                         return -EFAULT;
553                 __handle_sysrq(c, NULL, NULL, 0);
554         }
555         return count;
556 }
557
558 static struct file_operations proc_sysrq_trigger_operations = {
559         .write          = write_sysrq_trigger,
560 };
561 #endif
562
563 struct proc_dir_entry *proc_root_kcore;
564
565 void create_seq_entry(char *name, mode_t mode, struct file_operations *f)
566 {
567         struct proc_dir_entry *entry;
568         entry = create_proc_entry(name, mode, NULL);
569         if (entry)
570                 entry->proc_fops = f;
571 }
572
573 void __init proc_misc_init(void)
574 {
575         struct proc_dir_entry *entry;
576         static struct {
577                 char *name;
578                 int (*read_proc)(char*,char**,off_t,int,int*,void*);
579         } *p, simple_ones[] = {
580                 {"loadavg",     loadavg_read_proc},
581                 {"uptime",      uptime_read_proc},
582                 {"meminfo",     meminfo_read_proc},
583                 {"version",     version_read_proc},
584 #ifdef CONFIG_PROC_HARDWARE
585                 {"hardware",    hardware_read_proc},
586 #endif
587 #ifdef CONFIG_STRAM_PROC
588                 {"stram",       stram_read_proc},
589 #endif
590                 {"devices",     devices_read_proc},
591                 {"filesystems", filesystems_read_proc},
592                 {"cmdline",     cmdline_read_proc},
593                 {"locks",       locks_read_proc},
594                 {"execdomains", execdomains_read_proc},
595                 {NULL,}
596         };
597         for (p = simple_ones; p->name; p++)
598                 create_proc_read_entry(p->name, 0, NULL, p->read_proc, NULL);
599
600         proc_symlink("mounts", NULL, "self/mounts");
601
602         /* And now for trickier ones */
603         entry = create_proc_entry("kmsg", S_IRUSR, &proc_root);
604         if (entry)
605                 entry->proc_fops = &proc_kmsg_operations;
606         create_seq_entry("cpuinfo", 0, &proc_cpuinfo_operations);
607         create_seq_entry("partitions", 0, &proc_partitions_operations);
608         create_seq_entry("stat", 0, &proc_stat_operations);
609         create_seq_entry("interrupts", 0, &proc_interrupts_operations);
610         create_seq_entry("slabinfo",S_IWUSR|S_IRUGO,&proc_slabinfo_operations);
611         create_seq_entry("buddyinfo",S_IRUGO, &fragmentation_file_operations);
612         create_seq_entry("vmstat",S_IRUGO, &proc_vmstat_file_operations);
613         create_seq_entry("diskstats", 0, &proc_diskstats_operations);
614 #ifdef CONFIG_MODULES
615         create_seq_entry("modules", 0, &proc_modules_operations);
616 #endif
617 #ifdef CONFIG_SCHEDSTATS
618         create_seq_entry("schedstat", 0, &proc_schedstat_operations);
619 #endif
620 #ifdef CONFIG_PROC_KCORE
621         proc_root_kcore = create_proc_entry("kcore", S_IRUSR, NULL);
622         if (proc_root_kcore) {
623                 proc_root_kcore->proc_fops = &proc_kcore_operations;
624                 proc_root_kcore->size =
625                                 (size_t)high_memory - PAGE_OFFSET + PAGE_SIZE;
626         }
627 #endif
628 #ifdef CONFIG_MAGIC_SYSRQ
629         entry = create_proc_entry("sysrq-trigger", S_IWUSR, NULL);
630         if (entry)
631                 entry->proc_fops = &proc_sysrq_trigger_operations;
632 #endif
633 #ifdef CONFIG_PPC32
634         {
635                 extern struct file_operations ppc_htab_operations;
636                 entry = create_proc_entry("ppc_htab", S_IRUGO|S_IWUSR, NULL);
637                 if (entry)
638                         entry->proc_fops = &ppc_htab_operations;
639         }
640 #endif
641 }