delete_module(2)

1delete_module(2)              System Calls Manual             delete_module(2)
2
3
4

NAME

6       delete_module - unload a kernel module
7

LIBRARY

9       Standard C library (libc, -lc)
10

SYNOPSIS

12       #include <fcntl.h>            /* Definition of O_* constants */
13       #include <sys/syscall.h>      /* Definition of SYS_* constants */
14       #include <unistd.h>
15
16       int syscall(SYS_delete_module, const char *name, unsigned int flags);
17
18       Note:  glibc provides no wrapper for delete_module(), necessitating the
19       use of syscall(2).
20

DESCRIPTION

22       The delete_module() system call attempts to remove the unused  loadable
23       module  entry  identified by name.  If the module has an exit function,
24       then that function is executed before unloading the module.  The  flags
25       argument  is  used  to  modify  the behavior of the system call, as de‐
26       scribed below.  This system call requires privilege.
27
28       Module removal is attempted according to the following rules:
29
30       (1)  If there are other loaded modules that depend on (i.e.,  refer  to
31            symbols defined in) this module, then the call fails.
32
33       (2)  Otherwise, if the reference count for the module (i.e., the number
34            of processes currently using the module) is zero, then the  module
35            is immediately unloaded.
36
37       (3)  If  a  module has a nonzero reference count, then the behavior de‐
38            pends on the bits set in flags.  In normal usage (see NOTES),  the
39            O_NONBLOCK  flag is always specified, and the O_TRUNC flag may ad‐
40            ditionally be specified.
41
42            The various combinations for flags have the following effect:
43
44            flags == O_NONBLOCK
45                   The call returns immediately, with an error.
46
47            flags == (O_NONBLOCK | O_TRUNC)
48                   The module is unloaded immediately, regardless  of  whether
49                   it has a nonzero reference count.
50
51            (flags & O_NONBLOCK) == 0
52                   If  flags  does not specify O_NONBLOCK, the following steps
53                   occur:
54
55                   •  The module is marked so that no new references are  per‐
56                      mitted.
57
58                   •  If  the  module's reference count is nonzero, the caller
59                      is placed in an uninterruptible sleep state  (TASK_UNIN‐
60                      TERRUPTIBLE) until the reference count is zero, at which
61                      point the call unblocks.
62
63                   •  The module is unloaded in the usual way.
64
65       The O_TRUNC flag has one further effect on the rules  described  above.
66       By default, if a module has an init function but no exit function, then
67       an attempt to remove the module fails.  However, if O_TRUNC was  speci‐
68       fied, this requirement is bypassed.
69
70       Using  the O_TRUNC flag is dangerous!  If the kernel was not built with
71       CONFIG_MODULE_FORCE_UNLOAD, this flag is silently ignored.   (Normally,
72       CONFIG_MODULE_FORCE_UNLOAD  is  enabled.)   Using  this flag taints the
73       kernel (TAINT_FORCED_RMMOD).
74

RETURN VALUE

76       On success, zero is returned.  On error, -1 is returned  and  errno  is
77       set to indicate the error.
78

ERRORS

80       EBUSY  The module is not "live" (i.e., it is still being initialized or
81              is already marked for removal); or, the module has an init func‐
82              tion  but has no exit function, and O_TRUNC was not specified in
83              flags.
84
85       EFAULT name refers to a location outside the process's  accessible  ad‐
86              dress space.
87
88       ENOENT No module by that name exists.
89
90       EPERM  The  caller  was not privileged (did not have the CAP_SYS_MODULE
91              capability), or module unloading is disabled (see /proc/sys/ker‐
92              nel/modules_disabled in proc(5)).
93
94       EWOULDBLOCK
95              Other  modules  depend on this module; or, O_NONBLOCK was speci‐
96              fied in flags, but the reference count of this module is nonzero
97              and O_TRUNC was not specified in flags.
98

STANDARDS

100       Linux.
101

HISTORY

103       The delete_module() system call is not supported by glibc.  No declara‐
104       tion is provided in glibc headers, but, through  a  quirk  of  history,
105       glibc  versions  before  glibc  2.23  did export an ABI for this system
106       call.  Therefore, in order to employ this system call,  it  is  (before
107       glibc  2.23) sufficient to manually declare the interface in your code;
108       alternatively, you can invoke the system call using syscall(2).
109
110   Linux 2.4 and earlier
111       In Linux 2.4 and earlier, the system call took only one argument:
112
113          int delete_module(const char *name);
114
115       If name is NULL, all unused modules marked auto-clean are removed.
116
117       Some further details of differences in the behavior of  delete_module()
118       in  Linux  2.4  and  earlier are not currently explained in this manual
119       page.
120

NOTES

122       The uninterruptible sleep that may occur if O_NONBLOCK is omitted  from
123       flags  is  considered undesirable, because the sleeping process is left
124       in an unkillable state.  As at Linux 3.7, specifying O_NONBLOCK is  op‐
125       tional, but in future kernels it is likely to become mandatory.
126