1 /* SPDX-License-Identifier: LGPL-2.1+ */
10 #ifndef FUSE_USE_VERSION
11 #define FUSE_USE_VERSION 30
14 #ifndef FUSE_USE_VERSION
15 #define FUSE_USE_VERSION 26
19 #define _FILE_OFFSET_BITS 64
21 #define __STDC_FORMAT_MACROS
39 #include <linux/magic.h>
40 #include <linux/sched.h>
41 #include <sys/epoll.h>
43 #include <sys/mount.h>
44 #include <sys/param.h>
45 #include <sys/socket.h>
46 #include <sys/syscall.h>
47 #include <sys/sysinfo.h>
51 #include "cgroup_fuse.h"
52 #include "cgroups/cgroup.h"
53 #include "cgroups/cgroup_utils.h"
54 #include "cpuset_parse.h"
55 #include "lxcfs_fuse_compat.h"
56 #include "memory_utils.h"
57 #include "proc_loadavg.h"
58 #include "proc_cpuview.h"
62 uint64_t hierarchical_memory_limit
;
63 uint64_t hierarchical_memsw_limit
;
66 uint64_t total_rss_huge
;
68 uint64_t total_mapped_file
;
70 uint64_t total_writeback
;
72 uint64_t total_pgpgin
;
73 uint64_t total_pgpgout
;
74 uint64_t total_pgfault
;
75 uint64_t total_pgmajfault
;
76 uint64_t total_inactive_anon
;
77 uint64_t total_active_anon
;
78 uint64_t total_inactive_file
;
79 uint64_t total_active_file
;
80 uint64_t total_unevictable
;
83 __lxcfs_fuse_ops
int proc_getattr(const char *path
, struct stat
*sb
)
87 memset(sb
, 0, sizeof(struct stat
));
88 if (clock_gettime(CLOCK_REALTIME
, &now
) < 0)
91 sb
->st_uid
= sb
->st_gid
= 0;
92 sb
->st_atim
= sb
->st_mtim
= sb
->st_ctim
= now
;
93 if (strcmp(path
, "/proc") == 0) {
94 sb
->st_mode
= S_IFDIR
| 00555;
99 if (strcmp(path
, "/proc/meminfo") == 0 ||
100 strcmp(path
, "/proc/cpuinfo") == 0 ||
101 strcmp(path
, "/proc/uptime") == 0 ||
102 strcmp(path
, "/proc/stat") == 0 ||
103 strcmp(path
, "/proc/diskstats") == 0 ||
104 strcmp(path
, "/proc/swaps") == 0 ||
105 strcmp(path
, "/proc/loadavg") == 0 ||
106 strcmp(path
, "/proc/slabinfo") == 0) {
108 sb
->st_mode
= S_IFREG
| 00444;
116 __lxcfs_fuse_ops
int proc_readdir(const char *path
, void *buf
,
117 fuse_fill_dir_t filler
, off_t offset
,
118 struct fuse_file_info
*fi
)
120 if (DIR_FILLER(filler
, buf
, ".", NULL
, 0) != 0 ||
121 DIR_FILLER(filler
, buf
, "..", NULL
, 0) != 0 ||
122 DIR_FILLER(filler
, buf
, "cpuinfo", NULL
, 0) != 0 ||
123 DIR_FILLER(filler
, buf
, "meminfo", NULL
, 0) != 0 ||
124 DIR_FILLER(filler
, buf
, "stat", NULL
, 0) != 0 ||
125 DIR_FILLER(filler
, buf
, "uptime", NULL
, 0) != 0 ||
126 DIR_FILLER(filler
, buf
, "diskstats", NULL
, 0) != 0 ||
127 DIR_FILLER(filler
, buf
, "swaps", NULL
, 0) != 0 ||
128 DIR_FILLER(filler
, buf
, "loadavg", NULL
, 0) != 0 ||
129 DIR_FILLER(filler
, buf
, "slabinfo", NULL
, 0) != 0)
135 static off_t
get_procfile_size(const char *path
)
137 __do_fclose
FILE *f
= NULL
;
138 __do_free
char *line
= NULL
;
140 ssize_t sz
, answer
= 0;
142 f
= fopen(path
, "re");
146 while ((sz
= getline(&line
, &len
, f
)) != -1)
152 __lxcfs_fuse_ops
int proc_open(const char *path
, struct fuse_file_info
*fi
)
154 __do_free
struct file_info
*info
= NULL
;
157 if (strcmp(path
, "/proc/meminfo") == 0)
158 type
= LXC_TYPE_PROC_MEMINFO
;
159 else if (strcmp(path
, "/proc/cpuinfo") == 0)
160 type
= LXC_TYPE_PROC_CPUINFO
;
161 else if (strcmp(path
, "/proc/uptime") == 0)
162 type
= LXC_TYPE_PROC_UPTIME
;
163 else if (strcmp(path
, "/proc/stat") == 0)
164 type
= LXC_TYPE_PROC_STAT
;
165 else if (strcmp(path
, "/proc/diskstats") == 0)
166 type
= LXC_TYPE_PROC_DISKSTATS
;
167 else if (strcmp(path
, "/proc/swaps") == 0)
168 type
= LXC_TYPE_PROC_SWAPS
;
169 else if (strcmp(path
, "/proc/loadavg") == 0)
170 type
= LXC_TYPE_PROC_LOADAVG
;
171 else if (strcmp(path
, "/proc/slabinfo") == 0)
172 type
= LXC_TYPE_PROC_SLABINFO
;
176 info
= zalloc(sizeof(*info
));
182 info
->buflen
= get_procfile_size(path
) + BUF_RESERVE_SIZE
;
184 info
->buf
= zalloc(info
->buflen
);
187 /* set actual size to buffer size */
188 info
->size
= info
->buflen
;
190 fi
->fh
= PTR_TO_UINT64(move_ptr(info
));
194 __lxcfs_fuse_ops
int proc_access(const char *path
, int mask
)
196 if (strcmp(path
, "/proc") == 0 && access(path
, R_OK
) == 0)
199 /* these are all read-only */
200 if ((mask
& ~R_OK
) != 0)
206 __lxcfs_fuse_ops
int proc_release(const char *path
, struct fuse_file_info
*fi
)
208 do_release_file_info(fi
);
212 static uint64_t get_memlimit(const char *cgroup
, bool swap
)
214 __do_free
char *memlimit_str
= NULL
;
215 uint64_t memlimit
= 0;
219 ret
= cgroup_ops
->get_memory_swap_max(cgroup_ops
, cgroup
, &memlimit_str
);
221 ret
= cgroup_ops
->get_memory_max(cgroup_ops
, cgroup
, &memlimit_str
);
222 if (ret
> 0 && memlimit_str
[0] && safe_uint64(memlimit_str
, &memlimit
, 10) < 0)
223 lxcfs_error("Failed to convert memlimit %s", memlimit_str
);
229 * This function taken from glibc-2.32, as POSIX dirname("/some-dir") will
230 * return "/some-dir" as opposed to "/", which breaks `get_min_memlimit()`
232 static char *gnu_dirname(char *path
)
234 static const char dot
[] = ".";
238 last_slash
= path
!= NULL
? strrchr(path
, '/') : NULL
;
240 if (last_slash
!= NULL
&& last_slash
!= path
&& last_slash
[1] == '\0') {
241 /* Determine whether all remaining characters are slashes. */
244 for (runp
= last_slash
; runp
!= path
; --runp
)
248 /* The '/' is the last character, we have to look further. */
250 last_slash
= memrchr(path
, '/', runp
- path
);
253 if (last_slash
!= NULL
) {
254 /* Determine whether all remaining characters are slashes. */
257 for (runp
= last_slash
; runp
!= path
; --runp
)
261 /* Terminate the path. */
264 * The last slash is the first character in the string.
265 * We have to return "/". As a special case we have to
266 * return "//" if there are exactly two slashes at the
267 * beginning of the string. See XBD 4.10 Path Name
268 * Resolution for more information
270 if (last_slash
== path
+ 1)
273 last_slash
= path
+ 1;
277 last_slash
[0] = '\0';
280 * This assignment is ill-designed but the XPG specs require to
281 * return a string containing "." in any case no directory part
282 * is found and so a static and constant string is required.
290 static uint64_t get_min_memlimit(const char *cgroup
, bool swap
)
292 __do_free
char *copy
= NULL
;
293 uint64_t memlimit
= 0, retlimit
= 0;
295 copy
= strdup(cgroup
);
297 return log_error_errno(0, ENOMEM
, "Failed to allocate memory");
299 retlimit
= get_memlimit(copy
, swap
);
302 * If the cgroup doesn't start with / (probably won't happen), dirname()
303 * will terminate with "" instead of "/"
305 while (*copy
&& strcmp(copy
, "/") != 0) {
308 it
= gnu_dirname(it
);
309 memlimit
= get_memlimit(it
, swap
);
310 if (memlimit
> 0 && memlimit
< retlimit
)
317 static inline bool startswith(const char *line
, const char *pref
)
319 return strncmp(line
, pref
, strlen(pref
)) == 0;
322 static int proc_swaps_read(char *buf
, size_t size
, off_t offset
,
323 struct fuse_file_info
*fi
)
325 __do_free
char *cgroup
= NULL
, *memusage_str
= NULL
,
326 *memswusage_str
= NULL
, *memswpriority_str
= NULL
;
327 struct fuse_context
*fc
= fuse_get_context();
328 struct lxcfs_opts
*opts
= (struct lxcfs_opts
*)fuse_get_context()->private_data
;
329 bool wants_swap
= opts
&& !opts
->swap_off
&& liblxcfs_can_use_swap();
330 struct file_info
*d
= INTTYPE_TO_PTR(fi
->fh
);
331 uint64_t memswlimit
= 0, memlimit
= 0, memusage
= 0, memswusage
= 0,
332 swtotal
= 0, swusage
= 0, memswpriority
= 1,
333 hostswtotal
= 0, hostswfree
= 0;
334 ssize_t total_len
= 0;
336 char *cache
= d
->buf
;
338 __do_free
char *line
= NULL
;
339 __do_free
void *fopen_cache
= NULL
;
340 __do_fclose
FILE *f
= NULL
;
346 if (offset
> d
->size
)
352 left
= d
->size
- offset
;
353 total_len
= left
> size
? size
: left
;
354 memcpy(buf
, cache
+ offset
, total_len
);
359 pid_t initpid
= lookup_initpid_in_store(fc
->pid
);
360 if (initpid
<= 1 || is_shared_pidns(initpid
))
363 cgroup
= get_pid_cgroup(initpid
, "memory");
365 return read_file_fuse("/proc/swaps", buf
, size
, d
);
366 prune_init_slice(cgroup
);
368 memlimit
= get_min_memlimit(cgroup
, false);
370 ret
= cgroup_ops
->get_memory_current(cgroup_ops
, cgroup
, &memusage_str
);
374 if (safe_uint64(memusage_str
, &memusage
, 10) < 0)
375 lxcfs_error("Failed to convert memusage %s", memusage_str
);
378 memswlimit
= get_min_memlimit(cgroup
, true);
379 if (memswlimit
> 0) {
380 ret
= cgroup_ops
->get_memory_swap_current(cgroup_ops
, cgroup
, &memswusage_str
);
381 if (ret
>= 0 && safe_uint64(memswusage_str
, &memswusage
, 10) == 0) {
382 if (memlimit
> memswlimit
)
385 swtotal
= (memswlimit
- memlimit
) / 1024;
386 if (memusage
> memswusage
|| swtotal
== 0)
389 swusage
= (memswusage
- memusage
) / 1024;
392 ret
= cgroup_ops
->get_memory_swappiness(cgroup_ops
, cgroup
, &memswpriority_str
);
394 safe_uint64(memswpriority_str
, &memswpriority
, 10);
398 total_len
= snprintf(d
->buf
, d
->size
, "Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
400 /* Read host total and free values */
401 f
= fopen_cached("/proc/meminfo", "re", &fopen_cache
);
405 while (getline(&line
, &linelen
, f
) != -1) {
406 if (startswith(line
, "SwapTotal:"))
407 sscanf(line
, "SwapTotal: %8" PRIu64
" kB", &hostswtotal
);
408 else if (startswith(line
, "SwapFree:"))
409 sscanf(line
, "SwapFree: %8" PRIu64
" kB", &hostswfree
);
413 /* The total amount of swap is always reported to be the
414 lesser of the RAM+SWAP limit or the SWAP device size.
415 This is because the kernel can swap as much as it
416 wants and not only up to swtotal. */
417 swtotal
= memlimit
/ 1024 + swtotal
;
418 if (hostswtotal
< swtotal
) {
419 swtotal
= hostswtotal
;
422 /* When swappiness is 0, pretend we can't swap. */
423 if (memswpriority
== 0) {
429 l
= snprintf(d
->buf
+ total_len
, d
->size
- total_len
,
430 "none%*svirtual\t\t%" PRIu64
"\t%" PRIu64
"\t0\n",
431 36, " ", swtotal
, swusage
);
435 if (total_len
< 0 || l
< 0)
436 return log_error(0, "Failed writing to cache");
439 d
->size
= (int)total_len
;
441 if (total_len
> size
)
443 memcpy(buf
, d
->buf
, total_len
);
448 static void get_blkio_io_value(char *str
, unsigned major
, unsigned minor
,
449 char *iotype
, uint64_t *v
)
456 snprintf(key
, 32, "%u:%u %s", major
, minor
, iotype
);
461 if (startswith(str
, key
)) {
462 sscanf(str
+ len
, "%lu", v
);
465 eol
= strchr(str
, '\n');
472 struct lxcfs_diskstats
{
473 unsigned int major
; /* 1 - major number */
474 unsigned int minor
; /* 2 - minor mumber */
475 char dev_name
[72]; /* 3 - device name */
476 uint64_t read
; /* 4 - reads completed successfully */
477 uint64_t read_merged
; /* 5 - reads merged */
478 uint64_t read_sectors
; /* 6 - sectors read */
479 uint64_t read_ticks
; /* 7 - time spent reading (ms) */
480 uint64_t write
; /* 8 - writes completed */
481 uint64_t write_merged
; /* 9 - writes merged */
482 uint64_t write_sectors
; /* 10 - sectors written */
483 uint64_t write_ticks
; /* 11 - time spent writing (ms) */
484 uint64_t ios_pgr
; /* 12 - I/Os currently in progress */
485 uint64_t total_ticks
; /* 13 - time spent doing I/Os (ms) */
486 uint64_t rq_ticks
; /* 14 - weighted time spent doing I/Os (ms) */
487 uint64_t discard
; /* 15 - discards completed successfully (4.18+) */
488 uint64_t discard_merged
; /* 16 - discards merged (4.18+) */
489 uint64_t discard_sectors
; /* 17 - sectors discarded (4.18+) */
490 uint64_t discard_ticks
; /* 18 - time spent discarding (4.18+) */
493 static int proc_diskstats_read(char *buf
, size_t size
, off_t offset
,
494 struct fuse_file_info
*fi
)
496 __do_free
char *cg
= NULL
, *io_serviced_str
= NULL
,
497 *io_merged_str
= NULL
, *io_service_bytes_str
= NULL
,
498 *io_wait_time_str
= NULL
, *io_service_time_str
= NULL
,
500 __do_free
void *fopen_cache
= NULL
;
501 __do_fclose
FILE *f
= NULL
;
502 struct fuse_context
*fc
= fuse_get_context();
503 struct file_info
*d
= INTTYPE_TO_PTR(fi
->fh
);
504 struct lxcfs_diskstats stats
= {};
506 uint64_t read_service_time
, write_service_time
, discard_service_time
, read_wait_time
,
507 write_wait_time
, discard_wait_time
;
508 char *cache
= d
->buf
;
509 size_t cache_size
= d
->buflen
;
510 size_t linelen
= 0, total_len
= 0;
517 if (offset
> d
->size
)
523 left
= d
->size
- offset
;
524 total_len
= left
> size
? size
: left
;
525 memcpy(buf
, cache
+ offset
, total_len
);
530 pid_t initpid
= lookup_initpid_in_store(fc
->pid
);
531 if (initpid
<= 1 || is_shared_pidns(initpid
))
534 cg
= get_pid_cgroup(initpid
, "blkio");
536 return read_file_fuse("/proc/diskstats", buf
, size
, d
);
537 prune_init_slice(cg
);
539 ret
= cgroup_ops
->get_io_serviced(cgroup_ops
, cg
, &io_serviced_str
);
541 if (ret
== -EOPNOTSUPP
)
542 return read_file_fuse("/proc/diskstats", buf
, size
, d
);
545 ret
= cgroup_ops
->get_io_merged(cgroup_ops
, cg
, &io_merged_str
);
547 if (ret
== -EOPNOTSUPP
)
548 return read_file_fuse("/proc/diskstats", buf
, size
, d
);
551 ret
= cgroup_ops
->get_io_service_bytes(cgroup_ops
, cg
, &io_service_bytes_str
);
553 if (ret
== -EOPNOTSUPP
)
554 return read_file_fuse("/proc/diskstats", buf
, size
, d
);
557 ret
= cgroup_ops
->get_io_wait_time(cgroup_ops
, cg
, &io_wait_time_str
);
559 if (ret
== -EOPNOTSUPP
)
560 return read_file_fuse("/proc/diskstats", buf
, size
, d
);
563 ret
= cgroup_ops
->get_io_service_time(cgroup_ops
, cg
, &io_service_time_str
);
565 if (ret
== -EOPNOTSUPP
)
566 return read_file_fuse("/proc/diskstats", buf
, size
, d
);
569 f
= fopen_cached("/proc/diskstats", "re", &fopen_cache
);
573 while (getline(&line
, &linelen
, f
) != -1) {
577 i
= sscanf(line
, "%u %u %71s", &stats
.major
, &stats
.minor
, stats
.dev_name
);
581 get_blkio_io_value(io_serviced_str
, stats
.major
, stats
.minor
, "Read", &stats
.read
);
582 get_blkio_io_value(io_serviced_str
, stats
.major
, stats
.minor
, "Write", &stats
.write
);
583 get_blkio_io_value(io_serviced_str
, stats
.major
, stats
.minor
, "Discard", &stats
.discard
);
585 get_blkio_io_value(io_merged_str
, stats
.major
, stats
.minor
, "Read", &stats
.read_merged
);
586 get_blkio_io_value(io_merged_str
, stats
.major
, stats
.minor
, "Write", &stats
.write_merged
);
587 get_blkio_io_value(io_merged_str
, stats
.major
, stats
.minor
, "Discard", &stats
.discard_merged
);
589 get_blkio_io_value(io_service_bytes_str
, stats
.major
, stats
.minor
, "Read", &stats
.read_sectors
);
590 stats
.read_sectors
= stats
.read_sectors
/ 512;
591 get_blkio_io_value(io_service_bytes_str
, stats
.major
, stats
.minor
, "Write", &stats
.write_sectors
);
592 stats
.write_sectors
= stats
.write_sectors
/ 512;
593 get_blkio_io_value(io_service_bytes_str
, stats
.major
, stats
.minor
, "Discard", &stats
.discard_sectors
);
594 stats
.discard_sectors
= stats
.discard_sectors
/ 512;
596 get_blkio_io_value(io_service_time_str
, stats
.major
, stats
.minor
, "Read", &read_service_time
);
597 read_service_time
= read_service_time
/ 1000000;
598 get_blkio_io_value(io_wait_time_str
, stats
.major
, stats
.minor
, "Read", &read_wait_time
);
599 read_wait_time
= read_wait_time
/ 1000000;
600 stats
.read_ticks
= read_service_time
+ read_wait_time
;
602 get_blkio_io_value(io_service_time_str
, stats
.major
, stats
.minor
, "Write", &write_service_time
);
603 write_service_time
= write_service_time
/ 1000000;
604 get_blkio_io_value(io_wait_time_str
, stats
.major
, stats
.minor
, "Write", &write_wait_time
);
605 write_wait_time
= write_wait_time
/ 1000000;
606 stats
.write_ticks
= write_service_time
+ write_wait_time
;
608 get_blkio_io_value(io_service_time_str
, stats
.major
, stats
.minor
, "Discard", &discard_service_time
);
609 discard_service_time
= discard_service_time
/ 1000000;
610 get_blkio_io_value(io_wait_time_str
, stats
.major
, stats
.minor
, "Discard", &discard_wait_time
);
611 discard_wait_time
= discard_wait_time
/ 1000000;
612 stats
.discard_ticks
= discard_service_time
+ discard_wait_time
;
614 get_blkio_io_value(io_service_time_str
, stats
.major
, stats
.minor
, "Total", &stats
.total_ticks
);
615 stats
.total_ticks
= stats
.total_ticks
/ 1000000;
617 memset(lbuf
, 0, 256);
618 if (stats
.read
|| stats
.write
|| stats
.read_merged
|| stats
.write_merged
||
619 stats
.read_sectors
|| stats
.write_sectors
|| stats
.read_ticks
||
620 stats
.write_ticks
|| stats
.ios_pgr
|| stats
.total_ticks
|| stats
.rq_ticks
||
621 stats
.discard_merged
|| stats
.discard_sectors
|| stats
.discard_ticks
)
622 snprintf(lbuf
, 256, "%u %u %s %lu %lu %lu %lu %lu %lu %lu %lu %lu %lu %lu %lu %lu %lu\n",
637 stats
.discard_merged
,
638 stats
.discard_sectors
,
639 stats
.discard_ticks
);
643 l
= snprintf(cache
, cache_size
, "%s", lbuf
);
645 return log_error(0, "Failed to write cache");
647 return log_error(0, "Write to cache was truncated");
656 if (total_len
> size
)
658 memcpy(buf
, d
->buf
, total_len
);
664 static inline void iwashere(void)
666 mknod("/tmp/lxcfs-iwashere", S_IFREG
, 0644);
671 * This function retrieves the busy time of a group of tasks by looking at
672 * cpuacct.usage. Unfortunately, this only makes sense when the container has
673 * been given it's own cpuacct cgroup. If not, this function will take the busy
674 * time of all other taks that do not actually belong to the container into
675 * account as well. If someone has a clever solution for this please send a
678 static double get_reaper_busy(pid_t task
)
680 __do_free
char *cgroup
= NULL
, *usage_str
= NULL
;
684 initpid
= lookup_initpid_in_store(task
);
688 cgroup
= get_pid_cgroup(initpid
, "cpuacct");
691 prune_init_slice(cgroup
);
693 if (!cgroup_ops
->get(cgroup_ops
, "cpuacct", cgroup
, "cpuacct.usage", &usage_str
))
696 if (safe_uint64(usage_str
, &usage
, 10) < 0)
697 lxcfs_error("Failed to convert usage %s", usage_str
);
699 return ((double)usage
/ 1000000000);
702 static uint64_t get_reaper_start_time(pid_t pid
)
704 __do_free
void *fopen_cache
= NULL
;
705 __do_fclose
FILE *f
= NULL
;
708 char path
[STRLITERALLEN("/proc/") + LXCFS_NUMSTRLEN64
+
709 STRLITERALLEN("/stat") + 1];
712 qpid
= lookup_initpid_in_store(pid
);
714 return ret_errno(EINVAL
);
716 ret
= snprintf(path
, sizeof(path
), "/proc/%d/stat", qpid
);
717 if (ret
< 0 || (size_t)ret
>= sizeof(path
))
718 return ret_errno(EINVAL
);
720 f
= fopen_cached(path
, "re", &fopen_cache
);
722 return ret_errno(EINVAL
);
724 /* Note that the *scanf() argument supression requires that length
725 * modifiers such as "l" are omitted. Otherwise some compilers will yell
726 * at us. It's like telling someone you're not married and then asking
727 * if you can bring your wife to the party.
729 ret
= fscanf(f
, "%*d " /* (1) pid %d */
730 "%*s " /* (2) comm %s */
731 "%*c " /* (3) state %c */
732 "%*d " /* (4) ppid %d */
733 "%*d " /* (5) pgrp %d */
734 "%*d " /* (6) session %d */
735 "%*d " /* (7) tty_nr %d */
736 "%*d " /* (8) tpgid %d */
737 "%*u " /* (9) flags %u */
738 "%*u " /* (10) minflt %lu */
739 "%*u " /* (11) cminflt %lu */
740 "%*u " /* (12) majflt %lu */
741 "%*u " /* (13) cmajflt %lu */
742 "%*u " /* (14) utime %lu */
743 "%*u " /* (15) stime %lu */
744 "%*d " /* (16) cutime %ld */
745 "%*d " /* (17) cstime %ld */
746 "%*d " /* (18) priority %ld */
747 "%*d " /* (19) nice %ld */
748 "%*d " /* (20) num_threads %ld */
749 "%*d " /* (21) itrealvalue %ld */
750 "%" PRIu64
, /* (22) starttime %llu */
753 return ret_errno(EINVAL
);
755 return ret_set_errno(starttime
, 0);
758 static double get_reaper_start_time_in_sec(pid_t pid
)
760 uint64_t clockticks
, ticks_per_sec
;
764 clockticks
= get_reaper_start_time(pid
);
766 return log_debug(0, "Failed to retrieve start time of pid %d", pid
);
768 ret
= sysconf(_SC_CLK_TCK
);
770 return log_debug(0, "Failed to determine number of clock ticks in a second");
772 ticks_per_sec
= (uint64_t)ret
;
773 res
= (double)clockticks
/ ticks_per_sec
;
777 static double get_reaper_age(pid_t pid
)
780 double procstart
, procage
;
783 * We need to substract the time the process has started since system
784 * boot minus the time when the system has started to get the actual
787 procstart
= get_reaper_start_time_in_sec(pid
);
791 struct timespec spec
;
793 ret
= clock_gettime(CLOCK_BOOTTIME
, &spec
);
797 uptime_ms
= (spec
.tv_sec
* 1000) + (spec
.tv_nsec
* 1e-6);
798 procage
= (uptime_ms
- (procstart
* 1000)) / 1000;
805 * We read /proc/uptime and reuse its second field.
806 * For the first field, we use the mtime for the reaper for
807 * the calling pid as returned by getreaperage
809 static int proc_uptime_read(char *buf
, size_t size
, off_t offset
,
810 struct fuse_file_info
*fi
)
812 struct fuse_context
*fc
= fuse_get_context();
813 struct file_info
*d
= INTTYPE_TO_PTR(fi
->fh
);
814 char *cache
= d
->buf
;
815 ssize_t total_len
= 0, ret
= 0;
816 double busytime
, idletime
, reaperage
;
825 if (offset
> d
->size
)
831 left
= d
->size
- offset
;
832 total_len
= left
> size
? size
: left
;
833 memcpy(buf
, cache
+ offset
, total_len
);
838 reaperage
= get_reaper_age(fc
->pid
);
840 * To understand why this is done, please read the comment to the
841 * get_reaper_busy() function.
843 idletime
= reaperage
;
844 busytime
= get_reaper_busy(fc
->pid
);
845 if (reaperage
>= busytime
)
846 idletime
= reaperage
- busytime
;
848 ret
= snprintf(d
->buf
, d
->buflen
, "%.2lf %.2lf\n", reaperage
, idletime
);
849 if (ret
< 0 || ret
>= d
->buflen
)
850 return read_file_fuse("/proc/uptime", buf
, size
, d
);
855 if (total_len
> size
)
857 memcpy(buf
, d
->buf
, total_len
);
862 #define CPUALL_MAX_SIZE (BUF_RESERVE_SIZE / 2)
863 static int proc_stat_read(char *buf
, size_t size
, off_t offset
,
864 struct fuse_file_info
*fi
)
866 __do_free
char *cg
= NULL
, *cpuset
= NULL
, *line
= NULL
;
867 __do_free
void *fopen_cache
= NULL
;
868 __do_free
struct cpuacct_usage
*cg_cpu_usage
= NULL
;
869 __do_fclose
FILE *f
= NULL
;
870 struct fuse_context
*fc
= fuse_get_context();
871 struct lxcfs_opts
*opts
= (struct lxcfs_opts
*)fc
->private_data
;
872 struct file_info
*d
= INTTYPE_TO_PTR(fi
->fh
);
873 size_t linelen
= 0, total_len
= 0;
874 int curcpu
= -1; /* cpu numbering starts at 0 */
876 uint64_t user
= 0, nice
= 0, system
= 0, idle
= 0, iowait
= 0, irq
= 0,
877 softirq
= 0, steal
= 0, guest
= 0, guest_nice
= 0;
878 uint64_t user_sum
= 0, nice_sum
= 0, system_sum
= 0, idle_sum
= 0,
879 iowait_sum
= 0, irq_sum
= 0, softirq_sum
= 0, steal_sum
= 0,
880 guest_sum
= 0, guest_nice_sum
= 0;
881 char cpuall
[CPUALL_MAX_SIZE
];
882 /* reserve for cpu all */
883 char *cache
= d
->buf
+ CPUALL_MAX_SIZE
;
884 size_t cache_size
= d
->buflen
- CPUALL_MAX_SIZE
;
885 int cg_cpu_usage_size
= 0;
890 if (offset
> d
->size
)
896 left
= d
->size
- offset
;
897 total_len
= left
> size
? size
: left
;
898 memcpy(buf
, d
->buf
+ offset
, total_len
);
903 pid_t initpid
= lookup_initpid_in_store(fc
->pid
);
904 if (initpid
<= 1 || is_shared_pidns(initpid
))
908 * when container run with host pid namespace initpid == 1, cgroup will "/"
909 * we should return host os's /proc contents.
910 * in some case cpuacct_usage.all in "/" will larger then /proc/stat
913 return read_file_fuse("/proc/stat", buf
, size
, d
);
915 cg
= get_pid_cgroup(initpid
, "cpuset");
917 return read_file_fuse("/proc/stat", buf
, size
, d
);
918 prune_init_slice(cg
);
920 cpuset
= get_cpuset(cg
);
924 f
= fopen_cached("/proc/stat", "re", &fopen_cache
);
928 /* Skip first system cpu line. */
929 if (getline(&line
, &linelen
, f
) < 0)
930 return log_error(0, "proc_stat_read read first line failed");
933 * Read cpuacct.usage_all for all CPUs.
934 * If the cpuacct cgroup is present, it is used to calculate the container's
935 * CPU usage. If not, values from the host's /proc/stat are used.
937 if (read_cpuacct_usage_all(cg
, cpuset
, &cg_cpu_usage
, &cg_cpu_usage_size
) == 0) {
938 if (cgroup_ops
->can_use_cpuview(cgroup_ops
) && opts
&& opts
->use_cfs
) {
939 total_len
= cpuview_proc_stat(cg
, cpuset
, cg_cpu_usage
,
940 cg_cpu_usage_size
, f
,
945 lxcfs_v("proc_stat_read failed to read from cpuacct, falling back to the host's /proc/stat");
948 while (getline(&line
, &linelen
, f
) != -1) {
950 char cpu_char
[10]; /* That's a lot of cores */
952 uint64_t all_used
, cg_used
, new_idle
;
955 if (strlen(line
) == 0)
957 if (sscanf(line
, "cpu%9[^ ]", cpu_char
) != 1) {
958 /* not a ^cpuN line containing a number N, just print it */
959 l
= snprintf(cache
, cache_size
, "%s", line
);
961 return log_error(0, "Failed to write cache");
963 return log_error(0, "Write to cache was truncated");
972 if (sscanf(cpu_char
, "%d", &physcpu
) != 1)
975 if (!cpu_in_cpuset(physcpu
, cpuset
))
980 ret
= sscanf(line
, "%*s %lu %lu %lu %lu %lu %lu %lu %lu %lu %lu",
991 if (ret
!= 10 || !cg_cpu_usage
) {
992 c
= strchr(line
, ' ');
996 l
= snprintf(cache
, cache_size
, "cpu%d%s", curcpu
, c
);
998 return log_error(0, "Failed to write cache");
1000 return log_error(0, "Write to cache was truncated");
1011 if (physcpu
>= cg_cpu_usage_size
)
1014 all_used
= user
+ nice
+ system
+ iowait
+ irq
+ softirq
+ steal
+ guest
+ guest_nice
;
1015 cg_used
= cg_cpu_usage
[physcpu
].user
+ cg_cpu_usage
[physcpu
].system
;
1017 if (all_used
>= cg_used
) {
1018 new_idle
= idle
+ (all_used
- cg_used
);
1021 lxcfs_error("cpu%d from %s has unexpected cpu time: %" PRIu64
" in /proc/stat, %" PRIu64
" in cpuacct.usage_all; unable to determine idle time",
1022 curcpu
, cg
, all_used
, cg_used
);
1026 l
= snprintf(cache
, cache_size
,
1027 "cpu%d %" PRIu64
" 0 %" PRIu64
" %" PRIu64
" 0 0 0 0 0 0\n",
1028 curcpu
, cg_cpu_usage
[physcpu
].user
,
1029 cg_cpu_usage
[physcpu
].system
, new_idle
);
1031 return log_error(0, "Failed to write cache");
1032 if (l
>= cache_size
)
1033 return log_error(0, "Write to cache was truncated");
1039 user_sum
+= cg_cpu_usage
[physcpu
].user
;
1040 system_sum
+= cg_cpu_usage
[physcpu
].system
;
1041 idle_sum
+= new_idle
;
1045 system_sum
+= system
;
1047 iowait_sum
+= iowait
;
1049 softirq_sum
+= softirq
;
1052 guest_nice_sum
+= guest_nice
;
1058 int cpuall_len
= snprintf(cpuall
, CPUALL_MAX_SIZE
, "cpu %lu %lu %lu %lu %lu %lu %lu %lu %lu %lu\n",
1069 if (cpuall_len
> 0 && cpuall_len
< CPUALL_MAX_SIZE
) {
1070 memcpy(cache
, cpuall
, cpuall_len
);
1071 cache
+= cpuall_len
;
1073 /* shouldn't happen */
1074 lxcfs_error("proc_stat_read copy cpuall failed, cpuall_len=%d", cpuall_len
);
1078 memmove(cache
, d
->buf
+ CPUALL_MAX_SIZE
, total_len
);
1079 total_len
+= cpuall_len
;
1083 d
->size
= total_len
;
1084 if (total_len
> size
)
1087 memcpy(buf
, d
->buf
, total_len
);
1091 /* Note that "memory.stat" in cgroup2 is hierarchical by default. */
1092 static bool cgroup_parse_memory_stat(const char *cgroup
, struct memory_stat
*mstat
)
1094 __do_close
int fd
= -EBADF
;
1095 __do_fclose
FILE *f
= NULL
;
1096 __do_free
char *line
= NULL
;
1097 __do_free
void *fdopen_cache
= NULL
;
1102 fd
= cgroup_ops
->get_memory_stats_fd(cgroup_ops
, cgroup
);
1106 f
= fdopen_cached(fd
, "re", &fdopen_cache
);
1110 unified
= pure_unified_layout(cgroup_ops
);
1111 while ((linelen
= getline(&line
, &len
, f
)) != -1) {
1112 if (!unified
&& startswith(line
, "hierarchical_memory_limit")) {
1113 sscanf(line
, "hierarchical_memory_limit %" PRIu64
, &(mstat
->hierarchical_memory_limit
));
1114 } else if (!unified
&& startswith(line
, "hierarchical_memsw_limit")) {
1115 sscanf(line
, "hierarchical_memsw_limit %" PRIu64
, &(mstat
->hierarchical_memsw_limit
));
1116 } else if (startswith(line
, unified
? "file" :"total_cache")) {
1117 sscanf(line
, unified
? "file %" PRIu64
: "total_cache %" PRIu64
, &(mstat
->total_cache
));
1118 } else if (!unified
&& startswith(line
, "total_rss")) {
1119 sscanf(line
, "total_rss %" PRIu64
, &(mstat
->total_rss
));
1120 } else if (!unified
&& startswith(line
, "total_rss_huge")) {
1121 sscanf(line
, "total_rss_huge %" PRIu64
, &(mstat
->total_rss_huge
));
1122 } else if (startswith(line
, unified
? "shmem" : "total_shmem")) {
1123 sscanf(line
, unified
? "shmem %" PRIu64
: "total_shmem %" PRIu64
, &(mstat
->total_shmem
));
1124 } else if (startswith(line
, unified
? "file_mapped" : "total_mapped_file")) {
1125 sscanf(line
, unified
? "file_mapped %" PRIu64
: "total_mapped_file %" PRIu64
, &(mstat
->total_mapped_file
));
1126 } else if (!unified
&& startswith(line
, "total_dirty")) {
1127 sscanf(line
, "total_dirty %" PRIu64
, &(mstat
->total_dirty
));
1128 } else if (!unified
&& startswith(line
, "total_writeback")) {
1129 sscanf(line
, "total_writeback %" PRIu64
, &(mstat
->total_writeback
));
1130 } else if (!unified
&& startswith(line
, "total_swap")) {
1131 sscanf(line
, "total_swap %" PRIu64
, &(mstat
->total_swap
));
1132 } else if (!unified
&& startswith(line
, "total_pgpgin")) {
1133 sscanf(line
, "total_pgpgin %" PRIu64
, &(mstat
->total_pgpgin
));
1134 } else if (!unified
&& startswith(line
, "total_pgpgout")) {
1135 sscanf(line
, "total_pgpgout %" PRIu64
, &(mstat
->total_pgpgout
));
1136 } else if (startswith(line
, unified
? "pgfault" : "total_pgfault")) {
1137 sscanf(line
, unified
? "pgfault %" PRIu64
: "total_pgfault %" PRIu64
, &(mstat
->total_pgfault
));
1138 } else if (startswith(line
, unified
? "pgmajfault" : "total_pgmajfault")) {
1139 sscanf(line
, unified
? "pgmajfault %" PRIu64
: "total_pgmajfault %" PRIu64
, &(mstat
->total_pgmajfault
));
1140 } else if (startswith(line
, unified
? "inactive_anon" : "total_inactive_anon")) {
1141 sscanf(line
, unified
? "inactive_anon %" PRIu64
: "total_inactive_anon %" PRIu64
, &(mstat
->total_inactive_anon
));
1142 } else if (startswith(line
, unified
? "active_anon" : "total_active_anon")) {
1143 sscanf(line
, unified
? "active_anon %" PRIu64
: "total_active_anon %" PRIu64
, &(mstat
->total_active_anon
));
1144 } else if (startswith(line
, unified
? "inactive_file" : "total_inactive_file")) {
1145 sscanf(line
, unified
? "inactive_file %" PRIu64
: "total_inactive_file %" PRIu64
, &(mstat
->total_inactive_file
));
1146 } else if (startswith(line
, unified
? "active_file" : "total_active_file")) {
1147 sscanf(line
, unified
? "active_file %" PRIu64
: "total_active_file %" PRIu64
, &(mstat
->total_active_file
));
1148 } else if (startswith(line
, unified
? "unevictable" : "total_unevictable")) {
1149 sscanf(line
, unified
? "unevictable %" PRIu64
: "total_unevictable %" PRIu64
, &(mstat
->total_unevictable
));
1156 static int proc_meminfo_read(char *buf
, size_t size
, off_t offset
,
1157 struct fuse_file_info
*fi
)
1159 __do_free
char *cgroup
= NULL
, *line
= NULL
, *memusage_str
= NULL
,
1160 *memswusage_str
= NULL
, *memswpriority_str
= NULL
;
1161 __do_free
void *fopen_cache
= NULL
;
1162 __do_fclose
FILE *f
= NULL
;
1163 struct fuse_context
*fc
= fuse_get_context();
1164 struct lxcfs_opts
*opts
= (struct lxcfs_opts
*)fuse_get_context()->private_data
;
1165 bool wants_swap
= opts
&& !opts
->swap_off
&& liblxcfs_can_use_swap();
1166 struct file_info
*d
= INTTYPE_TO_PTR(fi
->fh
);
1167 uint64_t memlimit
= 0, memusage
= 0, memswlimit
= 0, memswusage
= 0,
1168 hosttotal
= 0, swfree
= 0, swusage
= 0, swtotal
= 0,
1170 struct memory_stat mstat
= {};
1171 size_t linelen
= 0, total_len
= 0;
1172 char *cache
= d
->buf
;
1173 size_t cache_size
= d
->buflen
;
1179 if (offset
> d
->size
)
1185 left
= d
->size
- offset
;
1186 total_len
= left
> size
? size
: left
;
1187 memcpy(buf
, cache
+ offset
, total_len
);
1192 pid_t initpid
= lookup_initpid_in_store(fc
->pid
);
1193 if (initpid
<= 1 || is_shared_pidns(initpid
))
1196 cgroup
= get_pid_cgroup(initpid
, "memory");
1198 return read_file_fuse("/proc/meminfo", buf
, size
, d
);
1200 prune_init_slice(cgroup
);
1203 ret
= cgroup_ops
->get_memory_current(cgroup_ops
, cgroup
, &memusage_str
);
1205 return read_file_fuse("/proc/meminfo", buf
, size
, d
);
1207 if (safe_uint64(memusage_str
, &memusage
, 10) < 0)
1208 lxcfs_error("Failed to convert memusage %s", memusage_str
);
1210 if (!cgroup_parse_memory_stat(cgroup
, &mstat
))
1211 return read_file_fuse("/proc/meminfo", buf
, size
, d
);
1213 memlimit
= get_min_memlimit(cgroup
, false);
1216 * Following values are allowed to fail, because swapaccount might be
1217 * turned off for current kernel.
1220 memswlimit
= get_min_memlimit(cgroup
, true);
1221 if (memswlimit
> 0) {
1222 ret
= cgroup_ops
->get_memory_swap_current(cgroup_ops
, cgroup
, &memswusage_str
);
1223 if (ret
>= 0 && safe_uint64(memswusage_str
, &memswusage
, 10) == 0) {
1224 if (memlimit
> memswlimit
)
1227 swtotal
= (memswlimit
- memlimit
) / 1024;
1228 if (memusage
> memswusage
|| swtotal
== 0)
1231 swusage
= (memswusage
- memusage
) / 1024;
1235 ret
= cgroup_ops
->get_memory_swappiness(cgroup_ops
, cgroup
, &memswpriority_str
);
1237 safe_uint64(memswpriority_str
, &memswpriority
, 10);
1240 f
= fopen_cached("/proc/meminfo", "re", &fopen_cache
);
1242 return read_file_fuse("/proc/meminfo", buf
, size
, d
);
1246 while (getline(&line
, &linelen
, f
) != -1) {
1248 char *printme
, lbuf
[100];
1250 memset(lbuf
, 0, 100);
1251 if (startswith(line
, "MemTotal:")) {
1252 sscanf(line
+sizeof("MemTotal:")-1, "%" PRIu64
, &hosttotal
);
1254 memlimit
= hosttotal
;
1256 if (hosttotal
< memlimit
)
1257 memlimit
= hosttotal
;
1258 snprintf(lbuf
, 100, "MemTotal: %8" PRIu64
" kB\n", memlimit
);
1260 } else if (startswith(line
, "MemFree:")) {
1261 snprintf(lbuf
, 100, "MemFree: %8" PRIu64
" kB\n", memlimit
- memusage
);
1263 } else if (startswith(line
, "MemAvailable:")) {
1264 snprintf(lbuf
, 100, "MemAvailable: %8" PRIu64
" kB\n", memlimit
- memusage
+ mstat
.total_cache
/ 1024);
1266 } else if (startswith(line
, "SwapTotal:")) {
1268 uint64_t hostswtotal
= 0;
1270 sscanf(line
+ STRLITERALLEN("SwapTotal:"), "%" PRIu64
, &hostswtotal
);
1272 /* The total amount of swap is always reported to be the
1273 lesser of the RAM+SWAP limit or the SWAP device size.
1274 This is because the kernel can swap as much as it
1275 wants and not only up to swtotal. */
1277 swtotal
= memlimit
+ swtotal
;
1278 if (hostswtotal
< swtotal
) {
1279 swtotal
= hostswtotal
;
1282 /* When swappiness is 0, pretend we can't swap. */
1283 if (memswpriority
== 0) {
1288 snprintf(lbuf
, 100, "SwapTotal: %8" PRIu64
" kB\n", swtotal
);
1290 } else if (startswith(line
, "SwapFree:")) {
1292 swfree
= swtotal
- swusage
;
1295 snprintf(lbuf
, 100, "SwapFree: %8" PRIu64
" kB\n", swfree
);
1297 } else if (startswith(line
, "Slab:")) {
1298 snprintf(lbuf
, 100, "Slab: %8" PRIu64
" kB\n", (uint64_t)0);
1300 } else if (startswith(line
, "Buffers:")) {
1301 snprintf(lbuf
, 100, "Buffers: %8" PRIu64
" kB\n", (uint64_t)0);
1303 } else if (startswith(line
, "Cached:")) {
1304 snprintf(lbuf
, 100, "Cached: %8" PRIu64
" kB\n",
1305 mstat
.total_cache
/ 1024);
1307 } else if (startswith(line
, "SwapCached:")) {
1308 snprintf(lbuf
, 100, "SwapCached: %8" PRIu64
" kB\n", (uint64_t)0);
1310 } else if (startswith(line
, "Active:")) {
1311 snprintf(lbuf
, 100, "Active: %8" PRIu64
" kB\n",
1312 (mstat
.total_active_anon
+
1313 mstat
.total_active_file
) /
1316 } else if (startswith(line
, "Inactive:")) {
1317 snprintf(lbuf
, 100, "Inactive: %8" PRIu64
" kB\n",
1318 (mstat
.total_inactive_anon
+
1319 mstat
.total_inactive_file
) /
1322 } else if (startswith(line
, "Active(anon):")) {
1323 snprintf(lbuf
, 100, "Active(anon): %8" PRIu64
" kB\n",
1324 mstat
.total_active_anon
/ 1024);
1326 } else if (startswith(line
, "Inactive(anon):")) {
1327 snprintf(lbuf
, 100, "Inactive(anon): %8" PRIu64
" kB\n",
1328 mstat
.total_inactive_anon
/ 1024);
1330 } else if (startswith(line
, "Active(file):")) {
1331 snprintf(lbuf
, 100, "Active(file): %8" PRIu64
" kB\n",
1332 mstat
.total_active_file
/ 1024);
1334 } else if (startswith(line
, "Inactive(file):")) {
1335 snprintf(lbuf
, 100, "Inactive(file): %8" PRIu64
" kB\n",
1336 mstat
.total_inactive_file
/ 1024);
1338 } else if (startswith(line
, "Unevictable:")) {
1339 snprintf(lbuf
, 100, "Unevictable: %8" PRIu64
" kB\n",
1340 mstat
.total_unevictable
/ 1024);
1342 } else if (startswith(line
, "Dirty:")) {
1343 snprintf(lbuf
, 100, "Dirty: %8" PRIu64
" kB\n",
1344 mstat
.total_dirty
/ 1024);
1346 } else if (startswith(line
, "Writeback:")) {
1347 snprintf(lbuf
, 100, "Writeback: %8" PRIu64
" kB\n",
1348 mstat
.total_writeback
/ 1024);
1350 } else if (startswith(line
, "AnonPages:")) {
1351 snprintf(lbuf
, 100, "AnonPages: %8" PRIu64
" kB\n",
1352 (mstat
.total_active_anon
+
1353 mstat
.total_inactive_anon
- mstat
.total_shmem
) /
1356 } else if (startswith(line
, "Mapped:")) {
1357 snprintf(lbuf
, 100, "Mapped: %8" PRIu64
" kB\n",
1358 mstat
.total_mapped_file
/ 1024);
1360 } else if (startswith(line
, "SReclaimable:")) {
1361 snprintf(lbuf
, 100, "SReclaimable: %8" PRIu64
" kB\n", (uint64_t)0);
1363 } else if (startswith(line
, "SUnreclaim:")) {
1364 snprintf(lbuf
, 100, "SUnreclaim: %8" PRIu64
" kB\n", (uint64_t)0);
1366 } else if (startswith(line
, "Shmem:")) {
1367 snprintf(lbuf
, 100, "Shmem: %8" PRIu64
" kB\n",
1368 mstat
.total_shmem
/ 1024);
1370 } else if (startswith(line
, "ShmemHugePages:")) {
1371 snprintf(lbuf
, 100, "ShmemHugePages: %8" PRIu64
" kB\n", (uint64_t)0);
1373 } else if (startswith(line
, "ShmemPmdMapped:")) {
1374 snprintf(lbuf
, 100, "ShmemPmdMapped: %8" PRIu64
" kB\n", (uint64_t)0);
1376 } else if (startswith(line
, "AnonHugePages:")) {
1377 snprintf(lbuf
, 100, "AnonHugePages: %8" PRIu64
" kB\n",
1378 mstat
.total_rss_huge
/ 1024);
1384 l
= snprintf(cache
, cache_size
, "%s", printme
);
1386 return log_error(0, "Failed to write cache");
1387 if (l
>= cache_size
)
1388 return log_error(0, "Write to cache was truncated");
1396 d
->size
= total_len
;
1397 if (total_len
> size
)
1399 memcpy(buf
, d
->buf
, total_len
);
1404 static int proc_slabinfo_read(char *buf
, size_t size
, off_t offset
,
1405 struct fuse_file_info
*fi
)
1407 __do_free
char *cgroup
= NULL
, *line
= NULL
;
1408 __do_free
void *fopen_cache
= NULL
;
1409 __do_fclose
FILE *f
= NULL
;
1410 __do_close
int fd
= -EBADF
;
1411 struct fuse_context
*fc
= fuse_get_context();
1412 struct file_info
*d
= INTTYPE_TO_PTR(fi
->fh
);
1413 size_t linelen
= 0, total_len
= 0;
1414 char *cache
= d
->buf
;
1415 size_t cache_size
= d
->buflen
;
1421 if (offset
> d
->size
)
1427 left
= d
->size
- offset
;
1428 total_len
= left
> size
? size
: left
;
1429 memcpy(buf
, cache
+ offset
, total_len
);
1434 initpid
= lookup_initpid_in_store(fc
->pid
);
1435 if (initpid
<= 1 || is_shared_pidns(initpid
))
1438 cgroup
= get_pid_cgroup(initpid
, "memory");
1440 return read_file_fuse("/proc/slabinfo", buf
, size
, d
);
1442 prune_init_slice(cgroup
);
1444 fd
= cgroup_ops
->get_memory_slabinfo_fd(cgroup_ops
, cgroup
);
1446 return read_file_fuse("/proc/slabinfo", buf
, size
, d
);
1448 f
= fdopen_cached(fd
, "re", &fopen_cache
);
1450 return read_file_fuse("/proc/slabinfo", buf
, size
, d
);
1452 while (getline(&line
, &linelen
, f
) != -1) {
1453 ssize_t l
= snprintf(cache
, cache_size
, "%s", line
);
1455 return log_error(0, "Failed to write cache");
1456 if (l
>= cache_size
)
1457 return log_error(0, "Write to cache was truncated");
1465 d
->size
= total_len
;
1466 if (total_len
> size
)
1468 memcpy(buf
, d
->buf
, total_len
);
1473 __lxcfs_fuse_ops
int proc_read(const char *path
, char *buf
, size_t size
,
1474 off_t offset
, struct fuse_file_info
*fi
)
1476 struct file_info
*f
= INTTYPE_TO_PTR(fi
->fh
);
1479 case LXC_TYPE_PROC_MEMINFO
:
1480 if (liblxcfs_functional())
1481 return proc_meminfo_read(buf
, size
, offset
, fi
);
1483 return read_file_fuse_with_offset(LXC_TYPE_PROC_MEMINFO_PATH
,
1484 buf
, size
, offset
, f
);
1485 case LXC_TYPE_PROC_CPUINFO
:
1486 if (liblxcfs_functional())
1487 return proc_cpuinfo_read(buf
, size
, offset
, fi
);
1489 return read_file_fuse_with_offset(LXC_TYPE_PROC_CPUINFO_PATH
,
1490 buf
, size
, offset
, f
);
1491 case LXC_TYPE_PROC_UPTIME
:
1492 if (liblxcfs_functional())
1493 return proc_uptime_read(buf
, size
, offset
, fi
);
1495 return read_file_fuse_with_offset(LXC_TYPE_PROC_UPTIME_PATH
,
1496 buf
, size
, offset
, f
);
1497 case LXC_TYPE_PROC_STAT
:
1498 if (liblxcfs_functional())
1499 return proc_stat_read(buf
, size
, offset
, fi
);
1501 return read_file_fuse_with_offset(LXC_TYPE_PROC_STAT_PATH
, buf
,
1503 case LXC_TYPE_PROC_DISKSTATS
:
1504 if (liblxcfs_functional())
1505 return proc_diskstats_read(buf
, size
, offset
, fi
);
1507 return read_file_fuse_with_offset(LXC_TYPE_PROC_DISKSTATS_PATH
,
1508 buf
, size
, offset
, f
);
1509 case LXC_TYPE_PROC_SWAPS
:
1510 if (liblxcfs_functional())
1511 return proc_swaps_read(buf
, size
, offset
, fi
);
1513 return read_file_fuse_with_offset(LXC_TYPE_PROC_SWAPS_PATH
, buf
,
1515 case LXC_TYPE_PROC_LOADAVG
:
1516 if (liblxcfs_functional())
1517 return proc_loadavg_read(buf
, size
, offset
, fi
);
1519 return read_file_fuse_with_offset(LXC_TYPE_PROC_LOADAVG_PATH
,
1520 buf
, size
, offset
, f
);
1521 case LXC_TYPE_PROC_SLABINFO
:
1522 if (liblxcfs_functional())
1523 return proc_slabinfo_read(buf
, size
, offset
, fi
);
1525 return read_file_fuse_with_offset(LXC_TYPE_PROC_SLABINFO_PATH
,
1526 buf
, size
, offset
, f
);