kcmp(2) - c7

1KCMP(2)                    Linux Programmer's Manual                   KCMP(2)
2
3
4

NAME

6       kcmp  -  compare  two  processes  to  determine  if they share a kernel
7       resource
8

SYNOPSIS

10       #include <linux/kcmp.h>
11
12       int kcmp(pid_t pid1, pid_t pid2, int type,
13                unsigned long idx1, unsigned long idx2);
14
15       Note: There is no glibc wrapper for this system call; see NOTES.
16

DESCRIPTION

18       The kcmp() system call can be used to check whether the  two  processes
19       identified  by  pid1  and  pid2 share a kernel resource such as virtual
20       memory, file descriptors, and so on.
21
22       The type argument specifies which resource is to be compared in the two
23       processes.  It has one of the following values:
24
25       KCMP_FILE
26              Check  whether a file descriptor idx1 in the process pid1 refers
27              to the same open file description (see open(2)) as file descrip‐
28              tor idx2 in the process pid2.
29
30       KCMP_FILES
31              Check  whether  the  process  share  the  same  set of open file
32              descriptors.  The arguments idx1 and idx2 are ignored.
33
34       KCMP_FS
35              Check whether the processes share the same file system  informa‐
36              tion (i.e., file mode creation mask, working directory, and file
37              system root).  The arguments idx1 and idx2 are ignored.
38
39       KCMP_IO
40              Check whether the processes share I/O  context.   The  arguments
41              idx1 and idx2 are ignored.
42
43       KCMP_SIGHAND
44              Check  whether the processes share the same table of signal dis‐
45              positions.  The arguments idx1 and idx2 are ignored.
46
47       KCMP_SYSVSEM
48              Check whether the processes share the same list of System V sem‐
49              aphore  undo  operations.   The  arguments  idx1  and  idx2  are
50              ignored.
51
52       KCMP_VM
53              Check whether the processes share the same address  space.   The
54              arguments idx1 and idx2 are ignored.
55
56       Note the kcmp() is not protected against false positives which may have
57       place if tasks are running.  Which means one should  stop  tasks  being
58       inspected with this syscall to obtain meaningful results.
59

RETURN VALUE

61       The return value of a successful call to kcmp() is simply the result of
62       arithmetic comparison of kernel  pointers  (when  the  kernel  compares
63       resources, it uses their memory addresses).
64
65       The  easiest way to explain is to consider an example.  Suppose that v1
66       and v2 are the addresses of  appropriate  resources,  then  the  return
67       value is one of the following:
68
69           0   v1  is equal to v2; in other words, the two processes share the
70               resource.
71
72           1   v1 is less than v2.
73
74           2   v1 is greater than v2.
75
76           3   v1 is not equal to v2, but ordering information is unavailable.
77
78       On error, -1 is returned, and errno is set appropriately.
79
80       kcmp () was designed to return values suitable for  sorting.   This  is
81       particularly  handy  if  one  needs  to  compare a large number of file
82       descriptors.
83

ERRORS

85       EBADF  type is KCMP_FILE and fd1 or fd2 is not an open file descriptor.
86
87       EINVAL type is invalid.
88
89       EPERM  Insufficient  permission  to  inspect  process  resources.   The
90              CAP_SYS_PTRACE  capability is required to inspect processes that
91              you do not own.
92
93       ESRCH  Process pid1 or pid2 does not exist.
94

VERSIONS

96       The kcmp() system call first appeared in Linux 3.5.
97

CONFORMING TO

99       kcmp() is Linux specific and should not be used in programs intended to
100       be portable.
101

NOTES

103       Glibc  does  not  provide a wrapper for this system call; call it using
104       syscall(2).
105
106       This system call is available only if the kernel  was  configured  with
107       CONFIG_CHECKPOINT_RESTORE.   The main use of the system call is for the
108       checkpoint/restore in user space (CRIU) feature.   The  alternative  to
109       this system call would have been to expose suitable process information
110       via the proc(5) file system; this was deemed to be unsuitable for secu‐
111       rity reasons.
112
113       See  clone(2)  for  some background information on the shared resources
114       referred to on this page.
115

COLOPHON

120       This page is part of release 3.53 of the Linux  man-pages  project.   A
121       description  of  the project, and information about reporting bugs, can
122       be found at http://www.kernel.org/doc/man-pages/.
123
124
125
126Linux                             2013-01-27                           KCMP(2)