1MPROTECT(2) Linux Programmer's Manual MPROTECT(2)
2
6 mprotect, pkey_mprotect - set protection on a region of memory
7
9 #include <sys/mman.h>
10 int mprotect(void *addr, size_t len, int prot);
12 #define _GNU_SOURCE /* See feature_test_macros(7) */
14 #include <sys/mman.h>
15 int pkey_mprotect(void *addr, size_t len, int prot, int pkey);
17
19 mprotect() changes the access protections for the calling process's
20 memory pages containing any part of the address range in the interval
21 [addr, addr+len-1]. addr must be aligned to a page boundary.
22 If the calling process tries to access memory in a manner that violates
24 the protections, then the kernel generates a SIGSEGV signal for the
25 process.
26 prot is a combination of the following access flags: PROT_NONE or a
28 bitwise-or of the other values in the following list:
29 PROT_NONE
31 The memory cannot be accessed at all.
32 PROT_READ
34 The memory can be read.
35 PROT_WRITE
37 The memory can be modified.
38 PROT_EXEC
40 The memory can be executed.
41 PROT_SEM (since Linux 2.5.7)
43 The memory can be used for atomic operations. This flag was in‐
44 troduced as part of the futex(2) implementation (in order to
45 guarantee the ability to perform atomic operations required by
46 commands such as FUTEX_WAIT), but is not currently used in on
47 any architecture.
48 PROT_SAO (since Linux 2.6.26)
50 The memory should have strong access ordering. This feature is
51 specific to the PowerPC architecture (version 2.06 of the archi‐
52 tecture specification adds the SAO CPU feature, and it is avail‐
53 able on POWER 7 or PowerPC A2, for example).
54 Additionally (since Linux 2.6.0), prot can have one of the following
56 flags set:
57 PROT_GROWSUP
59 Apply the protection mode up to the end of a mapping that grows
60 upwards. (Such mappings are created for the stack area on ar‐
61 chitectures—for example, HP-PARISC—that have an upwardly growing
62 stack.)
63 PROT_GROWSDOWN
65 Apply the protection mode down to the beginning of a mapping
66 that grows downward (which should be a stack segment or a seg‐
67 ment mapped with the MAP_GROWSDOWN flag set).
68 Like mprotect(), pkey_mprotect() changes the protection on the pages
70 specified by addr and len. The pkey argument specifies the protection
71 key (see pkeys(7)) to assign to the memory. The protection key must be
72 allocated with pkey_alloc(2) before it is passed to pkey_mprotect().
73 For an example of the use of this system call, see pkeys(7).
74
76 On success, mprotect() and pkey_mprotect() return zero. On error,
77 these system calls return -1, and errno is set appropriately.
78
80 EACCES The memory cannot be given the specified access. This can hap‐
81 pen, for example, if you mmap(2) a file to which you have read-
82 only access, then ask mprotect() to mark it PROT_WRITE.
83 EINVAL addr is not a valid pointer, or not a multiple of the system
85 page size.
86 EINVAL (pkey_mprotect()) pkey has not been allocated with pkey_alloc(2)
88 EINVAL Both PROT_GROWSUP and PROT_GROWSDOWN were specified in prot.
90 EINVAL Invalid flags specified in prot.
92 EINVAL (PowerPC architecture) PROT_SAO was specified in prot, but SAO
94 hardware feature is not available.
95 ENOMEM Internal kernel structures could not be allocated.
97 ENOMEM Addresses in the range [addr, addr+len-1] are invalid for the
99 address space of the process, or specify one or more pages that
100 are not mapped. (Before kernel 2.4.19, the error EFAULT was in‐
101 correctly produced for these cases.)
102 ENOMEM Changing the protection of a memory region would result in the
104 total number of mappings with distinct attributes (e.g., read
105 versus read/write protection) exceeding the allowed maximum.
106 (For example, making the protection of a range PROT_READ in the
107 middle of a region currently protected as PROT_READ|PROT_WRITE
108 would result in three mappings: two read/write mappings at each
109 end and a read-only mapping in the middle.)
110
112 pkey_mprotect() first appeared in Linux 4.9; library support was added
113 in glibc 2.27.
114
116 mprotect(): POSIX.1-2001, POSIX.1-2008, SVr4. POSIX says that the be‐
117 havior of mprotect() is unspecified if it is applied to a region of
118 memory that was not obtained via mmap(2).
119 pkey_mprotect() is a nonportable Linux extension.
121
123 On Linux, it is always permissible to call mprotect() on any address in
124 a process's address space (except for the kernel vsyscall area). In
125 particular, it can be used to change existing code mappings to be
126 writable.
127 Whether PROT_EXEC has any effect different from PROT_READ depends on
129 processor architecture, kernel version, and process state. If READ_IM‐
130 PLIES_EXEC is set in the process's personality flags (see personal‐
131 ity(2)), specifying PROT_READ will implicitly add PROT_EXEC.
132 On some hardware architectures (e.g., i386), PROT_WRITE implies
134 PROT_READ.
135 POSIX.1 says that an implementation may permit access other than that
137 specified in prot, but at a minimum can allow write access only if
138 PROT_WRITE has been set, and must not allow any access if PROT_NONE has
139 been set.
140 Applications should be careful when mixing use of mprotect() and
142 pkey_mprotect(). On x86, when mprotect() is used with prot set to
143 PROT_EXEC a pkey may be allocated and set on the memory implicitly by
144 the kernel, but only when the pkey was 0 previously.
145 On systems that do not support protection keys in hardware, pkey_mpro‐
147 tect() may still be used, but pkey must be set to -1. When called this
148 way, the operation of pkey_mprotect() is equivalent to mprotect().
149
151 The program below demonstrates the use of mprotect(). The program al‐
152 locates four pages of memory, makes the third of these pages read-only,
153 and then executes a loop that walks upward through the allocated region
154 modifying bytes.
155 An example of what we might see when running the program is the follow‐
157 ing:
158 $ ./a.out
160 Start of region: 0x804c000
161 Got SIGSEGV at address: 0x804e000
162 Program source
164 #include <unistd.h>
166 #include <signal.h>
167 #include <stdio.h>
168 #include <malloc.h>
169 #include <stdlib.h>
170 #include <errno.h>
171 #include <sys/mman.h>
172 #define handle_error(msg) \
174 do { perror(msg); exit(EXIT_FAILURE); } while (0)
175 static char *buffer;
177 static void
179 handler(int sig, siginfo_t *si, void *unused)
180 {
181 /* Note: calling printf() from a signal handler is not safe
182 (and should not be done in production programs), since
183 printf() is not async-signal-safe; see signal-safety(7).
184 Nevertheless, we use printf() here as a simple way of
185 showing that the handler was called. */
186 printf("Got SIGSEGV at address: %p\n", si->si_addr);
188 exit(EXIT_FAILURE);
189 }
190 int
192 main(int argc, char *argv[])
193 {
194 int pagesize;
195 struct sigaction sa;
196 sa.sa_flags = SA_SIGINFO;
198 sigemptyset(&sa.sa_mask);
199 sa.sa_sigaction = handler;
200 if (sigaction(SIGSEGV, &sa, NULL) == -1)
201 handle_error("sigaction");
202 pagesize = sysconf(_SC_PAGE_SIZE);
204 if (pagesize == -1)
205 handle_error("sysconf");
206 /* Allocate a buffer aligned on a page boundary;
208 initial protection is PROT_READ | PROT_WRITE */
209 buffer = memalign(pagesize, 4 * pagesize);
211 if (buffer == NULL)
212 handle_error("memalign");
213 printf("Start of region: %p\n", buffer);
215 if (mprotect(buffer + pagesize * 2, pagesize,
217 PROT_READ) == -1)
218 handle_error("mprotect");
219 for (char *p = buffer ; ; )
221 *(p++) = 'a';
222 printf("Loop completed\n"); /* Should never happen */
224 exit(EXIT_SUCCESS);
225 }
226
228 mmap(2), sysconf(3), pkeys(7)
229
231 This page is part of release 5.10 of the Linux man-pages project. A
232 description of the project, information about reporting bugs, and the
233 latest version of this page, can be found at
234 https://www.kernel.org/doc/man-pages/.
235Linux 2020-11-01 MPROTECT(2)