1 /* Rewritten by Rusty Russell, on the backs of many others...
2 Copyright (C) 2001 Rusty Russell, 2002 Rusty Russell IBM.
3
4 This program is free software; you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation; either version 2 of the License, or
7 (at your option) any later version.
8
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
13
14 You should have received a copy of the GNU General Public License
15 along with this program; if not, write to the Free Software
16 Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17 */
18 #include <linux/ftrace.h>
19 #include <linux/memory.h>
20 #include <linux/extable.h>
21 #include <linux/module.h>
22 #include <linux/mutex.h>
23 #include <linux/init.h>
24 #include <linux/kprobes.h>
25 #include <linux/filter.h>
26
27 #include <asm/sections.h>
28 #include <linux/uaccess.h>
29
30 /*
31 * mutex protecting text section modification (dynamic code patching).
32 * some users need to sleep (allocating memory...) while they hold this lock.
33 *
34 * Note: Also protects SMP-alternatives modification on x86.
35 *
36 * NOT exported to modules - patching kernel text is a really delicate matter.
37 */
38 DEFINE_MUTEX(text_mutex);
39
40 extern struct exception_table_entry __start___ex_table[];
41 extern struct exception_table_entry __stop___ex_table[];
42
43 /* Cleared by build time tools if the table is already sorted. */
44 u32 __initdata __visible main_extable_sort_needed = 1;
45
46 /* Sort the kernel's built-in exception table */
sort_main_extable(void)47 void __init sort_main_extable(void)
48 {
49 if (main_extable_sort_needed && __stop___ex_table > __start___ex_table) {
50 pr_notice("Sorting __ex_table...\n");
51 sort_extable(__start___ex_table, __stop___ex_table);
52 }
53 }
54
55 /* Given an address, look for it in the exception tables. */
search_exception_tables(unsigned long addr)56 const struct exception_table_entry *search_exception_tables(unsigned long addr)
57 {
58 const struct exception_table_entry *e;
59
60 e = search_extable(__start___ex_table,
61 __stop___ex_table - __start___ex_table, addr);
62 if (!e)
63 e = search_module_extables(addr);
64 return e;
65 }
66
init_kernel_text(unsigned long addr)67 int init_kernel_text(unsigned long addr)
68 {
69 if (addr >= (unsigned long)_sinittext &&
70 addr < (unsigned long)_einittext)
71 return 1;
72 return 0;
73 }
74
core_kernel_text(unsigned long addr)75 int notrace core_kernel_text(unsigned long addr)
76 {
77 if (addr >= (unsigned long)_stext &&
78 addr < (unsigned long)_etext)
79 return 1;
80
81 if (system_state < SYSTEM_RUNNING &&
82 init_kernel_text(addr))
83 return 1;
84 return 0;
85 }
86
87 /**
88 * core_kernel_data - tell if addr points to kernel data
89 * @addr: address to test
90 *
91 * Returns true if @addr passed in is from the core kernel data
92 * section.
93 *
94 * Note: On some archs it may return true for core RODATA, and false
95 * for others. But will always be true for core RW data.
96 */
core_kernel_data(unsigned long addr)97 int core_kernel_data(unsigned long addr)
98 {
99 if (addr >= (unsigned long)_sdata &&
100 addr < (unsigned long)_edata)
101 return 1;
102 return 0;
103 }
104
__kernel_text_address(unsigned long addr)105 int __kernel_text_address(unsigned long addr)
106 {
107 if (kernel_text_address(addr))
108 return 1;
109 /*
110 * There might be init symbols in saved stacktraces.
111 * Give those symbols a chance to be printed in
112 * backtraces (such as lockdep traces).
113 *
114 * Since we are after the module-symbols check, there's
115 * no danger of address overlap:
116 */
117 if (init_kernel_text(addr))
118 return 1;
119 return 0;
120 }
121
kernel_text_address(unsigned long addr)122 int kernel_text_address(unsigned long addr)
123 {
124 bool no_rcu;
125 int ret = 1;
126
127 if (core_kernel_text(addr))
128 return 1;
129
130 /*
131 * If a stack dump happens while RCU is not watching, then
132 * RCU needs to be notified that it requires to start
133 * watching again. This can happen either by tracing that
134 * triggers a stack trace, or a WARN() that happens during
135 * coming back from idle, or cpu on or offlining.
136 *
137 * is_module_text_address() as well as the kprobe slots
138 * and is_bpf_text_address() require RCU to be watching.
139 */
140 no_rcu = !rcu_is_watching();
141
142 /* Treat this like an NMI as it can happen anywhere */
143 if (no_rcu)
144 rcu_nmi_enter();
145
146 if (is_module_text_address(addr))
147 goto out;
148 if (is_ftrace_trampoline(addr))
149 goto out;
150 if (is_kprobe_optinsn_slot(addr) || is_kprobe_insn_slot(addr))
151 goto out;
152 if (is_bpf_text_address(addr))
153 goto out;
154 ret = 0;
155 out:
156 if (no_rcu)
157 rcu_nmi_exit();
158
159 return ret;
160 }
161
162 /*
163 * On some architectures (PPC64, IA64) function pointers
164 * are actually only tokens to some data that then holds the
165 * real function address. As a result, to find if a function
166 * pointer is part of the kernel text, we need to do some
167 * special dereferencing first.
168 */
func_ptr_is_kernel_text(void * ptr)169 int func_ptr_is_kernel_text(void *ptr)
170 {
171 unsigned long addr;
172 addr = (unsigned long) dereference_function_descriptor(ptr);
173 if (core_kernel_text(addr))
174 return 1;
175 return is_module_text_address(addr);
176 }
177
