Commit 4403b406d4369a275d483ece6ddee0088cc0d592
1 parent
e013e13bf6
Exists in
master
and in
7 other branches
Revert "Call init_workqueues before pre smp initcalls."
This reverts commit a802dd0eb5fc97a50cf1abb1f788a8f6cc5db635 by moving the call to init_workqueues() back where it belongs - after SMP has been initialized. It also moves stop_machine_init() - which needs workqueues - to a later phase using a core_initcall() instead of early_initcall(). That should satisfy all ordering requirements, and was apparently the reason why init_workqueues() was moved to be too early. Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Showing 2 changed files with 2 additions and 3 deletions Inline Diff
init/main.c
1 | /* | 1 | /* |
2 | * linux/init/main.c | 2 | * linux/init/main.c |
3 | * | 3 | * |
4 | * Copyright (C) 1991, 1992 Linus Torvalds | 4 | * Copyright (C) 1991, 1992 Linus Torvalds |
5 | * | 5 | * |
6 | * GK 2/5/95 - Changed to support mounting root fs via NFS | 6 | * GK 2/5/95 - Changed to support mounting root fs via NFS |
7 | * Added initrd & change_root: Werner Almesberger & Hans Lermen, Feb '96 | 7 | * Added initrd & change_root: Werner Almesberger & Hans Lermen, Feb '96 |
8 | * Moan early if gcc is old, avoiding bogus kernels - Paul Gortmaker, May '96 | 8 | * Moan early if gcc is old, avoiding bogus kernels - Paul Gortmaker, May '96 |
9 | * Simplified starting of init: Michael A. Griffith <grif@acm.org> | 9 | * Simplified starting of init: Michael A. Griffith <grif@acm.org> |
10 | */ | 10 | */ |
11 | 11 | ||
12 | #include <linux/types.h> | 12 | #include <linux/types.h> |
13 | #include <linux/module.h> | 13 | #include <linux/module.h> |
14 | #include <linux/proc_fs.h> | 14 | #include <linux/proc_fs.h> |
15 | #include <linux/kernel.h> | 15 | #include <linux/kernel.h> |
16 | #include <linux/syscalls.h> | 16 | #include <linux/syscalls.h> |
17 | #include <linux/string.h> | 17 | #include <linux/string.h> |
18 | #include <linux/ctype.h> | 18 | #include <linux/ctype.h> |
19 | #include <linux/delay.h> | 19 | #include <linux/delay.h> |
20 | #include <linux/utsname.h> | 20 | #include <linux/utsname.h> |
21 | #include <linux/ioport.h> | 21 | #include <linux/ioport.h> |
22 | #include <linux/init.h> | 22 | #include <linux/init.h> |
23 | #include <linux/smp_lock.h> | 23 | #include <linux/smp_lock.h> |
24 | #include <linux/initrd.h> | 24 | #include <linux/initrd.h> |
25 | #include <linux/bootmem.h> | 25 | #include <linux/bootmem.h> |
26 | #include <linux/tty.h> | 26 | #include <linux/tty.h> |
27 | #include <linux/gfp.h> | 27 | #include <linux/gfp.h> |
28 | #include <linux/percpu.h> | 28 | #include <linux/percpu.h> |
29 | #include <linux/kmod.h> | 29 | #include <linux/kmod.h> |
30 | #include <linux/vmalloc.h> | 30 | #include <linux/vmalloc.h> |
31 | #include <linux/kernel_stat.h> | 31 | #include <linux/kernel_stat.h> |
32 | #include <linux/start_kernel.h> | 32 | #include <linux/start_kernel.h> |
33 | #include <linux/security.h> | 33 | #include <linux/security.h> |
34 | #include <linux/smp.h> | 34 | #include <linux/smp.h> |
35 | #include <linux/workqueue.h> | 35 | #include <linux/workqueue.h> |
36 | #include <linux/profile.h> | 36 | #include <linux/profile.h> |
37 | #include <linux/rcupdate.h> | 37 | #include <linux/rcupdate.h> |
38 | #include <linux/moduleparam.h> | 38 | #include <linux/moduleparam.h> |
39 | #include <linux/kallsyms.h> | 39 | #include <linux/kallsyms.h> |
40 | #include <linux/writeback.h> | 40 | #include <linux/writeback.h> |
41 | #include <linux/cpu.h> | 41 | #include <linux/cpu.h> |
42 | #include <linux/cpuset.h> | 42 | #include <linux/cpuset.h> |
43 | #include <linux/cgroup.h> | 43 | #include <linux/cgroup.h> |
44 | #include <linux/efi.h> | 44 | #include <linux/efi.h> |
45 | #include <linux/tick.h> | 45 | #include <linux/tick.h> |
46 | #include <linux/interrupt.h> | 46 | #include <linux/interrupt.h> |
47 | #include <linux/taskstats_kern.h> | 47 | #include <linux/taskstats_kern.h> |
48 | #include <linux/delayacct.h> | 48 | #include <linux/delayacct.h> |
49 | #include <linux/unistd.h> | 49 | #include <linux/unistd.h> |
50 | #include <linux/rmap.h> | 50 | #include <linux/rmap.h> |
51 | #include <linux/mempolicy.h> | 51 | #include <linux/mempolicy.h> |
52 | #include <linux/key.h> | 52 | #include <linux/key.h> |
53 | #include <linux/unwind.h> | 53 | #include <linux/unwind.h> |
54 | #include <linux/buffer_head.h> | 54 | #include <linux/buffer_head.h> |
55 | #include <linux/page_cgroup.h> | 55 | #include <linux/page_cgroup.h> |
56 | #include <linux/debug_locks.h> | 56 | #include <linux/debug_locks.h> |
57 | #include <linux/debugobjects.h> | 57 | #include <linux/debugobjects.h> |
58 | #include <linux/lockdep.h> | 58 | #include <linux/lockdep.h> |
59 | #include <linux/pid_namespace.h> | 59 | #include <linux/pid_namespace.h> |
60 | #include <linux/device.h> | 60 | #include <linux/device.h> |
61 | #include <linux/kthread.h> | 61 | #include <linux/kthread.h> |
62 | #include <linux/sched.h> | 62 | #include <linux/sched.h> |
63 | #include <linux/signal.h> | 63 | #include <linux/signal.h> |
64 | #include <linux/idr.h> | 64 | #include <linux/idr.h> |
65 | #include <linux/ftrace.h> | 65 | #include <linux/ftrace.h> |
66 | 66 | ||
67 | #include <asm/io.h> | 67 | #include <asm/io.h> |
68 | #include <asm/bugs.h> | 68 | #include <asm/bugs.h> |
69 | #include <asm/setup.h> | 69 | #include <asm/setup.h> |
70 | #include <asm/sections.h> | 70 | #include <asm/sections.h> |
71 | #include <asm/cacheflush.h> | 71 | #include <asm/cacheflush.h> |
72 | 72 | ||
73 | #ifdef CONFIG_X86_LOCAL_APIC | 73 | #ifdef CONFIG_X86_LOCAL_APIC |
74 | #include <asm/smp.h> | 74 | #include <asm/smp.h> |
75 | #endif | 75 | #endif |
76 | 76 | ||
77 | /* | 77 | /* |
78 | * This is one of the first .c files built. Error out early if we have compiler | 78 | * This is one of the first .c files built. Error out early if we have compiler |
79 | * trouble. | 79 | * trouble. |
80 | */ | 80 | */ |
81 | 81 | ||
82 | #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0 | 82 | #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0 |
83 | #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended. | 83 | #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended. |
84 | #endif | 84 | #endif |
85 | 85 | ||
86 | static int kernel_init(void *); | 86 | static int kernel_init(void *); |
87 | 87 | ||
88 | extern void init_IRQ(void); | 88 | extern void init_IRQ(void); |
89 | extern void fork_init(unsigned long); | 89 | extern void fork_init(unsigned long); |
90 | extern void mca_init(void); | 90 | extern void mca_init(void); |
91 | extern void sbus_init(void); | 91 | extern void sbus_init(void); |
92 | extern void prio_tree_init(void); | 92 | extern void prio_tree_init(void); |
93 | extern void radix_tree_init(void); | 93 | extern void radix_tree_init(void); |
94 | extern void free_initmem(void); | 94 | extern void free_initmem(void); |
95 | #ifdef CONFIG_ACPI | 95 | #ifdef CONFIG_ACPI |
96 | extern void acpi_early_init(void); | 96 | extern void acpi_early_init(void); |
97 | #else | 97 | #else |
98 | static inline void acpi_early_init(void) { } | 98 | static inline void acpi_early_init(void) { } |
99 | #endif | 99 | #endif |
100 | #ifndef CONFIG_DEBUG_RODATA | 100 | #ifndef CONFIG_DEBUG_RODATA |
101 | static inline void mark_rodata_ro(void) { } | 101 | static inline void mark_rodata_ro(void) { } |
102 | #endif | 102 | #endif |
103 | 103 | ||
104 | #ifdef CONFIG_TC | 104 | #ifdef CONFIG_TC |
105 | extern void tc_init(void); | 105 | extern void tc_init(void); |
106 | #endif | 106 | #endif |
107 | 107 | ||
108 | enum system_states system_state; | 108 | enum system_states system_state; |
109 | EXPORT_SYMBOL(system_state); | 109 | EXPORT_SYMBOL(system_state); |
110 | 110 | ||
111 | /* | 111 | /* |
112 | * Boot command-line arguments | 112 | * Boot command-line arguments |
113 | */ | 113 | */ |
114 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | 114 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT |
115 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | 115 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT |
116 | 116 | ||
117 | extern void time_init(void); | 117 | extern void time_init(void); |
118 | /* Default late time init is NULL. archs can override this later. */ | 118 | /* Default late time init is NULL. archs can override this later. */ |
119 | void (*late_time_init)(void); | 119 | void (*late_time_init)(void); |
120 | extern void softirq_init(void); | 120 | extern void softirq_init(void); |
121 | 121 | ||
122 | /* Untouched command line saved by arch-specific code. */ | 122 | /* Untouched command line saved by arch-specific code. */ |
123 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | 123 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; |
124 | /* Untouched saved command line (eg. for /proc) */ | 124 | /* Untouched saved command line (eg. for /proc) */ |
125 | char *saved_command_line; | 125 | char *saved_command_line; |
126 | /* Command line for parameter parsing */ | 126 | /* Command line for parameter parsing */ |
127 | static char *static_command_line; | 127 | static char *static_command_line; |
128 | 128 | ||
129 | static char *execute_command; | 129 | static char *execute_command; |
130 | static char *ramdisk_execute_command; | 130 | static char *ramdisk_execute_command; |
131 | 131 | ||
132 | #ifdef CONFIG_SMP | 132 | #ifdef CONFIG_SMP |
133 | /* Setup configured maximum number of CPUs to activate */ | 133 | /* Setup configured maximum number of CPUs to activate */ |
134 | unsigned int __initdata setup_max_cpus = NR_CPUS; | 134 | unsigned int __initdata setup_max_cpus = NR_CPUS; |
135 | 135 | ||
136 | /* | 136 | /* |
137 | * Setup routine for controlling SMP activation | 137 | * Setup routine for controlling SMP activation |
138 | * | 138 | * |
139 | * Command-line option of "nosmp" or "maxcpus=0" will disable SMP | 139 | * Command-line option of "nosmp" or "maxcpus=0" will disable SMP |
140 | * activation entirely (the MPS table probe still happens, though). | 140 | * activation entirely (the MPS table probe still happens, though). |
141 | * | 141 | * |
142 | * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer | 142 | * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer |
143 | * greater than 0, limits the maximum number of CPUs activated in | 143 | * greater than 0, limits the maximum number of CPUs activated in |
144 | * SMP mode to <NUM>. | 144 | * SMP mode to <NUM>. |
145 | */ | 145 | */ |
146 | #ifndef CONFIG_X86_IO_APIC | 146 | #ifndef CONFIG_X86_IO_APIC |
147 | static inline void disable_ioapic_setup(void) {}; | 147 | static inline void disable_ioapic_setup(void) {}; |
148 | #endif | 148 | #endif |
149 | 149 | ||
150 | static int __init nosmp(char *str) | 150 | static int __init nosmp(char *str) |
151 | { | 151 | { |
152 | setup_max_cpus = 0; | 152 | setup_max_cpus = 0; |
153 | disable_ioapic_setup(); | 153 | disable_ioapic_setup(); |
154 | return 0; | 154 | return 0; |
155 | } | 155 | } |
156 | 156 | ||
157 | early_param("nosmp", nosmp); | 157 | early_param("nosmp", nosmp); |
158 | 158 | ||
159 | static int __init maxcpus(char *str) | 159 | static int __init maxcpus(char *str) |
160 | { | 160 | { |
161 | get_option(&str, &setup_max_cpus); | 161 | get_option(&str, &setup_max_cpus); |
162 | if (setup_max_cpus == 0) | 162 | if (setup_max_cpus == 0) |
163 | disable_ioapic_setup(); | 163 | disable_ioapic_setup(); |
164 | 164 | ||
165 | return 0; | 165 | return 0; |
166 | } | 166 | } |
167 | 167 | ||
168 | early_param("maxcpus", maxcpus); | 168 | early_param("maxcpus", maxcpus); |
169 | #else | 169 | #else |
170 | #define setup_max_cpus NR_CPUS | 170 | #define setup_max_cpus NR_CPUS |
171 | #endif | 171 | #endif |
172 | 172 | ||
173 | /* | 173 | /* |
174 | * If set, this is an indication to the drivers that reset the underlying | 174 | * If set, this is an indication to the drivers that reset the underlying |
175 | * device before going ahead with the initialization otherwise driver might | 175 | * device before going ahead with the initialization otherwise driver might |
176 | * rely on the BIOS and skip the reset operation. | 176 | * rely on the BIOS and skip the reset operation. |
177 | * | 177 | * |
178 | * This is useful if kernel is booting in an unreliable environment. | 178 | * This is useful if kernel is booting in an unreliable environment. |
179 | * For ex. kdump situaiton where previous kernel has crashed, BIOS has been | 179 | * For ex. kdump situaiton where previous kernel has crashed, BIOS has been |
180 | * skipped and devices will be in unknown state. | 180 | * skipped and devices will be in unknown state. |
181 | */ | 181 | */ |
182 | unsigned int reset_devices; | 182 | unsigned int reset_devices; |
183 | EXPORT_SYMBOL(reset_devices); | 183 | EXPORT_SYMBOL(reset_devices); |
184 | 184 | ||
185 | static int __init set_reset_devices(char *str) | 185 | static int __init set_reset_devices(char *str) |
186 | { | 186 | { |
187 | reset_devices = 1; | 187 | reset_devices = 1; |
188 | return 1; | 188 | return 1; |
189 | } | 189 | } |
190 | 190 | ||
191 | __setup("reset_devices", set_reset_devices); | 191 | __setup("reset_devices", set_reset_devices); |
192 | 192 | ||
193 | static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; | 193 | static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; |
194 | char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | 194 | char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; |
195 | static const char *panic_later, *panic_param; | 195 | static const char *panic_later, *panic_param; |
196 | 196 | ||
197 | extern struct obs_kernel_param __setup_start[], __setup_end[]; | 197 | extern struct obs_kernel_param __setup_start[], __setup_end[]; |
198 | 198 | ||
199 | static int __init obsolete_checksetup(char *line) | 199 | static int __init obsolete_checksetup(char *line) |
200 | { | 200 | { |
201 | struct obs_kernel_param *p; | 201 | struct obs_kernel_param *p; |
202 | int had_early_param = 0; | 202 | int had_early_param = 0; |
203 | 203 | ||
204 | p = __setup_start; | 204 | p = __setup_start; |
205 | do { | 205 | do { |
206 | int n = strlen(p->str); | 206 | int n = strlen(p->str); |
207 | if (!strncmp(line, p->str, n)) { | 207 | if (!strncmp(line, p->str, n)) { |
208 | if (p->early) { | 208 | if (p->early) { |
209 | /* Already done in parse_early_param? | 209 | /* Already done in parse_early_param? |
210 | * (Needs exact match on param part). | 210 | * (Needs exact match on param part). |
211 | * Keep iterating, as we can have early | 211 | * Keep iterating, as we can have early |
212 | * params and __setups of same names 8( */ | 212 | * params and __setups of same names 8( */ |
213 | if (line[n] == '\0' || line[n] == '=') | 213 | if (line[n] == '\0' || line[n] == '=') |
214 | had_early_param = 1; | 214 | had_early_param = 1; |
215 | } else if (!p->setup_func) { | 215 | } else if (!p->setup_func) { |
216 | printk(KERN_WARNING "Parameter %s is obsolete," | 216 | printk(KERN_WARNING "Parameter %s is obsolete," |
217 | " ignored\n", p->str); | 217 | " ignored\n", p->str); |
218 | return 1; | 218 | return 1; |
219 | } else if (p->setup_func(line + n)) | 219 | } else if (p->setup_func(line + n)) |
220 | return 1; | 220 | return 1; |
221 | } | 221 | } |
222 | p++; | 222 | p++; |
223 | } while (p < __setup_end); | 223 | } while (p < __setup_end); |
224 | 224 | ||
225 | return had_early_param; | 225 | return had_early_param; |
226 | } | 226 | } |
227 | 227 | ||
228 | /* | 228 | /* |
229 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | 229 | * This should be approx 2 Bo*oMips to start (note initial shift), and will |
230 | * still work even if initially too large, it will just take slightly longer | 230 | * still work even if initially too large, it will just take slightly longer |
231 | */ | 231 | */ |
232 | unsigned long loops_per_jiffy = (1<<12); | 232 | unsigned long loops_per_jiffy = (1<<12); |
233 | 233 | ||
234 | EXPORT_SYMBOL(loops_per_jiffy); | 234 | EXPORT_SYMBOL(loops_per_jiffy); |
235 | 235 | ||
236 | static int __init debug_kernel(char *str) | 236 | static int __init debug_kernel(char *str) |
237 | { | 237 | { |
238 | console_loglevel = 10; | 238 | console_loglevel = 10; |
239 | return 0; | 239 | return 0; |
240 | } | 240 | } |
241 | 241 | ||
242 | static int __init quiet_kernel(char *str) | 242 | static int __init quiet_kernel(char *str) |
243 | { | 243 | { |
244 | console_loglevel = 4; | 244 | console_loglevel = 4; |
245 | return 0; | 245 | return 0; |
246 | } | 246 | } |
247 | 247 | ||
248 | early_param("debug", debug_kernel); | 248 | early_param("debug", debug_kernel); |
249 | early_param("quiet", quiet_kernel); | 249 | early_param("quiet", quiet_kernel); |
250 | 250 | ||
251 | static int __init loglevel(char *str) | 251 | static int __init loglevel(char *str) |
252 | { | 252 | { |
253 | get_option(&str, &console_loglevel); | 253 | get_option(&str, &console_loglevel); |
254 | return 0; | 254 | return 0; |
255 | } | 255 | } |
256 | 256 | ||
257 | early_param("loglevel", loglevel); | 257 | early_param("loglevel", loglevel); |
258 | 258 | ||
259 | /* | 259 | /* |
260 | * Unknown boot options get handed to init, unless they look like | 260 | * Unknown boot options get handed to init, unless they look like |
261 | * failed parameters | 261 | * failed parameters |
262 | */ | 262 | */ |
263 | static int __init unknown_bootoption(char *param, char *val) | 263 | static int __init unknown_bootoption(char *param, char *val) |
264 | { | 264 | { |
265 | /* Change NUL term back to "=", to make "param" the whole string. */ | 265 | /* Change NUL term back to "=", to make "param" the whole string. */ |
266 | if (val) { | 266 | if (val) { |
267 | /* param=val or param="val"? */ | 267 | /* param=val or param="val"? */ |
268 | if (val == param+strlen(param)+1) | 268 | if (val == param+strlen(param)+1) |
269 | val[-1] = '='; | 269 | val[-1] = '='; |
270 | else if (val == param+strlen(param)+2) { | 270 | else if (val == param+strlen(param)+2) { |
271 | val[-2] = '='; | 271 | val[-2] = '='; |
272 | memmove(val-1, val, strlen(val)+1); | 272 | memmove(val-1, val, strlen(val)+1); |
273 | val--; | 273 | val--; |
274 | } else | 274 | } else |
275 | BUG(); | 275 | BUG(); |
276 | } | 276 | } |
277 | 277 | ||
278 | /* Handle obsolete-style parameters */ | 278 | /* Handle obsolete-style parameters */ |
279 | if (obsolete_checksetup(param)) | 279 | if (obsolete_checksetup(param)) |
280 | return 0; | 280 | return 0; |
281 | 281 | ||
282 | /* | 282 | /* |
283 | * Preemptive maintenance for "why didn't my misspelled command | 283 | * Preemptive maintenance for "why didn't my misspelled command |
284 | * line work?" | 284 | * line work?" |
285 | */ | 285 | */ |
286 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) { | 286 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) { |
287 | printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param); | 287 | printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param); |
288 | return 0; | 288 | return 0; |
289 | } | 289 | } |
290 | 290 | ||
291 | if (panic_later) | 291 | if (panic_later) |
292 | return 0; | 292 | return 0; |
293 | 293 | ||
294 | if (val) { | 294 | if (val) { |
295 | /* Environment option */ | 295 | /* Environment option */ |
296 | unsigned int i; | 296 | unsigned int i; |
297 | for (i = 0; envp_init[i]; i++) { | 297 | for (i = 0; envp_init[i]; i++) { |
298 | if (i == MAX_INIT_ENVS) { | 298 | if (i == MAX_INIT_ENVS) { |
299 | panic_later = "Too many boot env vars at `%s'"; | 299 | panic_later = "Too many boot env vars at `%s'"; |
300 | panic_param = param; | 300 | panic_param = param; |
301 | } | 301 | } |
302 | if (!strncmp(param, envp_init[i], val - param)) | 302 | if (!strncmp(param, envp_init[i], val - param)) |
303 | break; | 303 | break; |
304 | } | 304 | } |
305 | envp_init[i] = param; | 305 | envp_init[i] = param; |
306 | } else { | 306 | } else { |
307 | /* Command line option */ | 307 | /* Command line option */ |
308 | unsigned int i; | 308 | unsigned int i; |
309 | for (i = 0; argv_init[i]; i++) { | 309 | for (i = 0; argv_init[i]; i++) { |
310 | if (i == MAX_INIT_ARGS) { | 310 | if (i == MAX_INIT_ARGS) { |
311 | panic_later = "Too many boot init vars at `%s'"; | 311 | panic_later = "Too many boot init vars at `%s'"; |
312 | panic_param = param; | 312 | panic_param = param; |
313 | } | 313 | } |
314 | } | 314 | } |
315 | argv_init[i] = param; | 315 | argv_init[i] = param; |
316 | } | 316 | } |
317 | return 0; | 317 | return 0; |
318 | } | 318 | } |
319 | 319 | ||
320 | #ifdef CONFIG_DEBUG_PAGEALLOC | 320 | #ifdef CONFIG_DEBUG_PAGEALLOC |
321 | int __read_mostly debug_pagealloc_enabled = 0; | 321 | int __read_mostly debug_pagealloc_enabled = 0; |
322 | #endif | 322 | #endif |
323 | 323 | ||
324 | static int __init init_setup(char *str) | 324 | static int __init init_setup(char *str) |
325 | { | 325 | { |
326 | unsigned int i; | 326 | unsigned int i; |
327 | 327 | ||
328 | execute_command = str; | 328 | execute_command = str; |
329 | /* | 329 | /* |
330 | * In case LILO is going to boot us with default command line, | 330 | * In case LILO is going to boot us with default command line, |
331 | * it prepends "auto" before the whole cmdline which makes | 331 | * it prepends "auto" before the whole cmdline which makes |
332 | * the shell think it should execute a script with such name. | 332 | * the shell think it should execute a script with such name. |
333 | * So we ignore all arguments entered _before_ init=... [MJ] | 333 | * So we ignore all arguments entered _before_ init=... [MJ] |
334 | */ | 334 | */ |
335 | for (i = 1; i < MAX_INIT_ARGS; i++) | 335 | for (i = 1; i < MAX_INIT_ARGS; i++) |
336 | argv_init[i] = NULL; | 336 | argv_init[i] = NULL; |
337 | return 1; | 337 | return 1; |
338 | } | 338 | } |
339 | __setup("init=", init_setup); | 339 | __setup("init=", init_setup); |
340 | 340 | ||
341 | static int __init rdinit_setup(char *str) | 341 | static int __init rdinit_setup(char *str) |
342 | { | 342 | { |
343 | unsigned int i; | 343 | unsigned int i; |
344 | 344 | ||
345 | ramdisk_execute_command = str; | 345 | ramdisk_execute_command = str; |
346 | /* See "auto" comment in init_setup */ | 346 | /* See "auto" comment in init_setup */ |
347 | for (i = 1; i < MAX_INIT_ARGS; i++) | 347 | for (i = 1; i < MAX_INIT_ARGS; i++) |
348 | argv_init[i] = NULL; | 348 | argv_init[i] = NULL; |
349 | return 1; | 349 | return 1; |
350 | } | 350 | } |
351 | __setup("rdinit=", rdinit_setup); | 351 | __setup("rdinit=", rdinit_setup); |
352 | 352 | ||
353 | #ifndef CONFIG_SMP | 353 | #ifndef CONFIG_SMP |
354 | 354 | ||
355 | #ifdef CONFIG_X86_LOCAL_APIC | 355 | #ifdef CONFIG_X86_LOCAL_APIC |
356 | static void __init smp_init(void) | 356 | static void __init smp_init(void) |
357 | { | 357 | { |
358 | APIC_init_uniprocessor(); | 358 | APIC_init_uniprocessor(); |
359 | } | 359 | } |
360 | #else | 360 | #else |
361 | #define smp_init() do { } while (0) | 361 | #define smp_init() do { } while (0) |
362 | #endif | 362 | #endif |
363 | 363 | ||
364 | static inline void setup_per_cpu_areas(void) { } | 364 | static inline void setup_per_cpu_areas(void) { } |
365 | static inline void setup_nr_cpu_ids(void) { } | 365 | static inline void setup_nr_cpu_ids(void) { } |
366 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } | 366 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } |
367 | 367 | ||
368 | #else | 368 | #else |
369 | 369 | ||
370 | #if NR_CPUS > BITS_PER_LONG | 370 | #if NR_CPUS > BITS_PER_LONG |
371 | cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL; | 371 | cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL; |
372 | EXPORT_SYMBOL(cpu_mask_all); | 372 | EXPORT_SYMBOL(cpu_mask_all); |
373 | #endif | 373 | #endif |
374 | 374 | ||
375 | /* Setup number of possible processor ids */ | 375 | /* Setup number of possible processor ids */ |
376 | int nr_cpu_ids __read_mostly = NR_CPUS; | 376 | int nr_cpu_ids __read_mostly = NR_CPUS; |
377 | EXPORT_SYMBOL(nr_cpu_ids); | 377 | EXPORT_SYMBOL(nr_cpu_ids); |
378 | 378 | ||
379 | /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */ | 379 | /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */ |
380 | static void __init setup_nr_cpu_ids(void) | 380 | static void __init setup_nr_cpu_ids(void) |
381 | { | 381 | { |
382 | int cpu, highest_cpu = 0; | 382 | int cpu, highest_cpu = 0; |
383 | 383 | ||
384 | for_each_possible_cpu(cpu) | 384 | for_each_possible_cpu(cpu) |
385 | highest_cpu = cpu; | 385 | highest_cpu = cpu; |
386 | 386 | ||
387 | nr_cpu_ids = highest_cpu + 1; | 387 | nr_cpu_ids = highest_cpu + 1; |
388 | } | 388 | } |
389 | 389 | ||
390 | #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA | 390 | #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA |
391 | unsigned long __per_cpu_offset[NR_CPUS] __read_mostly; | 391 | unsigned long __per_cpu_offset[NR_CPUS] __read_mostly; |
392 | 392 | ||
393 | EXPORT_SYMBOL(__per_cpu_offset); | 393 | EXPORT_SYMBOL(__per_cpu_offset); |
394 | 394 | ||
395 | static void __init setup_per_cpu_areas(void) | 395 | static void __init setup_per_cpu_areas(void) |
396 | { | 396 | { |
397 | unsigned long size, i; | 397 | unsigned long size, i; |
398 | char *ptr; | 398 | char *ptr; |
399 | unsigned long nr_possible_cpus = num_possible_cpus(); | 399 | unsigned long nr_possible_cpus = num_possible_cpus(); |
400 | 400 | ||
401 | /* Copy section for each CPU (we discard the original) */ | 401 | /* Copy section for each CPU (we discard the original) */ |
402 | size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE); | 402 | size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE); |
403 | ptr = alloc_bootmem_pages(size * nr_possible_cpus); | 403 | ptr = alloc_bootmem_pages(size * nr_possible_cpus); |
404 | 404 | ||
405 | for_each_possible_cpu(i) { | 405 | for_each_possible_cpu(i) { |
406 | __per_cpu_offset[i] = ptr - __per_cpu_start; | 406 | __per_cpu_offset[i] = ptr - __per_cpu_start; |
407 | memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); | 407 | memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); |
408 | ptr += size; | 408 | ptr += size; |
409 | } | 409 | } |
410 | } | 410 | } |
411 | #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */ | 411 | #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */ |
412 | 412 | ||
413 | /* Called by boot processor to activate the rest. */ | 413 | /* Called by boot processor to activate the rest. */ |
414 | static void __init smp_init(void) | 414 | static void __init smp_init(void) |
415 | { | 415 | { |
416 | unsigned int cpu; | 416 | unsigned int cpu; |
417 | 417 | ||
418 | /* | 418 | /* |
419 | * Set up the current CPU as possible to migrate to. | 419 | * Set up the current CPU as possible to migrate to. |
420 | * The other ones will be done by cpu_up/cpu_down() | 420 | * The other ones will be done by cpu_up/cpu_down() |
421 | */ | 421 | */ |
422 | cpu = smp_processor_id(); | 422 | cpu = smp_processor_id(); |
423 | cpu_set(cpu, cpu_active_map); | 423 | cpu_set(cpu, cpu_active_map); |
424 | 424 | ||
425 | /* FIXME: This should be done in userspace --RR */ | 425 | /* FIXME: This should be done in userspace --RR */ |
426 | for_each_present_cpu(cpu) { | 426 | for_each_present_cpu(cpu) { |
427 | if (num_online_cpus() >= setup_max_cpus) | 427 | if (num_online_cpus() >= setup_max_cpus) |
428 | break; | 428 | break; |
429 | if (!cpu_online(cpu)) | 429 | if (!cpu_online(cpu)) |
430 | cpu_up(cpu); | 430 | cpu_up(cpu); |
431 | } | 431 | } |
432 | 432 | ||
433 | /* Any cleanup work */ | 433 | /* Any cleanup work */ |
434 | printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus()); | 434 | printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus()); |
435 | smp_cpus_done(setup_max_cpus); | 435 | smp_cpus_done(setup_max_cpus); |
436 | } | 436 | } |
437 | 437 | ||
438 | #endif | 438 | #endif |
439 | 439 | ||
440 | /* | 440 | /* |
441 | * We need to store the untouched command line for future reference. | 441 | * We need to store the untouched command line for future reference. |
442 | * We also need to store the touched command line since the parameter | 442 | * We also need to store the touched command line since the parameter |
443 | * parsing is performed in place, and we should allow a component to | 443 | * parsing is performed in place, and we should allow a component to |
444 | * store reference of name/value for future reference. | 444 | * store reference of name/value for future reference. |
445 | */ | 445 | */ |
446 | static void __init setup_command_line(char *command_line) | 446 | static void __init setup_command_line(char *command_line) |
447 | { | 447 | { |
448 | saved_command_line = alloc_bootmem(strlen (boot_command_line)+1); | 448 | saved_command_line = alloc_bootmem(strlen (boot_command_line)+1); |
449 | static_command_line = alloc_bootmem(strlen (command_line)+1); | 449 | static_command_line = alloc_bootmem(strlen (command_line)+1); |
450 | strcpy (saved_command_line, boot_command_line); | 450 | strcpy (saved_command_line, boot_command_line); |
451 | strcpy (static_command_line, command_line); | 451 | strcpy (static_command_line, command_line); |
452 | } | 452 | } |
453 | 453 | ||
454 | /* | 454 | /* |
455 | * We need to finalize in a non-__init function or else race conditions | 455 | * We need to finalize in a non-__init function or else race conditions |
456 | * between the root thread and the init thread may cause start_kernel to | 456 | * between the root thread and the init thread may cause start_kernel to |
457 | * be reaped by free_initmem before the root thread has proceeded to | 457 | * be reaped by free_initmem before the root thread has proceeded to |
458 | * cpu_idle. | 458 | * cpu_idle. |
459 | * | 459 | * |
460 | * gcc-3.4 accidentally inlines this function, so use noinline. | 460 | * gcc-3.4 accidentally inlines this function, so use noinline. |
461 | */ | 461 | */ |
462 | 462 | ||
463 | static void noinline __init_refok rest_init(void) | 463 | static void noinline __init_refok rest_init(void) |
464 | __releases(kernel_lock) | 464 | __releases(kernel_lock) |
465 | { | 465 | { |
466 | int pid; | 466 | int pid; |
467 | 467 | ||
468 | kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND); | 468 | kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND); |
469 | numa_default_policy(); | 469 | numa_default_policy(); |
470 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); | 470 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); |
471 | kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); | 471 | kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); |
472 | unlock_kernel(); | 472 | unlock_kernel(); |
473 | 473 | ||
474 | /* | 474 | /* |
475 | * The boot idle thread must execute schedule() | 475 | * The boot idle thread must execute schedule() |
476 | * at least once to get things moving: | 476 | * at least once to get things moving: |
477 | */ | 477 | */ |
478 | init_idle_bootup_task(current); | 478 | init_idle_bootup_task(current); |
479 | preempt_enable_no_resched(); | 479 | preempt_enable_no_resched(); |
480 | schedule(); | 480 | schedule(); |
481 | preempt_disable(); | 481 | preempt_disable(); |
482 | 482 | ||
483 | /* Call into cpu_idle with preempt disabled */ | 483 | /* Call into cpu_idle with preempt disabled */ |
484 | cpu_idle(); | 484 | cpu_idle(); |
485 | } | 485 | } |
486 | 486 | ||
487 | /* Check for early params. */ | 487 | /* Check for early params. */ |
488 | static int __init do_early_param(char *param, char *val) | 488 | static int __init do_early_param(char *param, char *val) |
489 | { | 489 | { |
490 | struct obs_kernel_param *p; | 490 | struct obs_kernel_param *p; |
491 | 491 | ||
492 | for (p = __setup_start; p < __setup_end; p++) { | 492 | for (p = __setup_start; p < __setup_end; p++) { |
493 | if ((p->early && strcmp(param, p->str) == 0) || | 493 | if ((p->early && strcmp(param, p->str) == 0) || |
494 | (strcmp(param, "console") == 0 && | 494 | (strcmp(param, "console") == 0 && |
495 | strcmp(p->str, "earlycon") == 0) | 495 | strcmp(p->str, "earlycon") == 0) |
496 | ) { | 496 | ) { |
497 | if (p->setup_func(val) != 0) | 497 | if (p->setup_func(val) != 0) |
498 | printk(KERN_WARNING | 498 | printk(KERN_WARNING |
499 | "Malformed early option '%s'\n", param); | 499 | "Malformed early option '%s'\n", param); |
500 | } | 500 | } |
501 | } | 501 | } |
502 | /* We accept everything at this stage. */ | 502 | /* We accept everything at this stage. */ |
503 | return 0; | 503 | return 0; |
504 | } | 504 | } |
505 | 505 | ||
506 | /* Arch code calls this early on, or if not, just before other parsing. */ | 506 | /* Arch code calls this early on, or if not, just before other parsing. */ |
507 | void __init parse_early_param(void) | 507 | void __init parse_early_param(void) |
508 | { | 508 | { |
509 | static __initdata int done = 0; | 509 | static __initdata int done = 0; |
510 | static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; | 510 | static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; |
511 | 511 | ||
512 | if (done) | 512 | if (done) |
513 | return; | 513 | return; |
514 | 514 | ||
515 | /* All fall through to do_early_param. */ | 515 | /* All fall through to do_early_param. */ |
516 | strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); | 516 | strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); |
517 | parse_args("early options", tmp_cmdline, NULL, 0, do_early_param); | 517 | parse_args("early options", tmp_cmdline, NULL, 0, do_early_param); |
518 | done = 1; | 518 | done = 1; |
519 | } | 519 | } |
520 | 520 | ||
521 | /* | 521 | /* |
522 | * Activate the first processor. | 522 | * Activate the first processor. |
523 | */ | 523 | */ |
524 | 524 | ||
525 | static void __init boot_cpu_init(void) | 525 | static void __init boot_cpu_init(void) |
526 | { | 526 | { |
527 | int cpu = smp_processor_id(); | 527 | int cpu = smp_processor_id(); |
528 | /* Mark the boot cpu "present", "online" etc for SMP and UP case */ | 528 | /* Mark the boot cpu "present", "online" etc for SMP and UP case */ |
529 | cpu_set(cpu, cpu_online_map); | 529 | cpu_set(cpu, cpu_online_map); |
530 | cpu_set(cpu, cpu_present_map); | 530 | cpu_set(cpu, cpu_present_map); |
531 | cpu_set(cpu, cpu_possible_map); | 531 | cpu_set(cpu, cpu_possible_map); |
532 | } | 532 | } |
533 | 533 | ||
534 | void __init __weak smp_setup_processor_id(void) | 534 | void __init __weak smp_setup_processor_id(void) |
535 | { | 535 | { |
536 | } | 536 | } |
537 | 537 | ||
538 | void __init __weak thread_info_cache_init(void) | 538 | void __init __weak thread_info_cache_init(void) |
539 | { | 539 | { |
540 | } | 540 | } |
541 | 541 | ||
542 | asmlinkage void __init start_kernel(void) | 542 | asmlinkage void __init start_kernel(void) |
543 | { | 543 | { |
544 | char * command_line; | 544 | char * command_line; |
545 | extern struct kernel_param __start___param[], __stop___param[]; | 545 | extern struct kernel_param __start___param[], __stop___param[]; |
546 | 546 | ||
547 | smp_setup_processor_id(); | 547 | smp_setup_processor_id(); |
548 | 548 | ||
549 | /* | 549 | /* |
550 | * Need to run as early as possible, to initialize the | 550 | * Need to run as early as possible, to initialize the |
551 | * lockdep hash: | 551 | * lockdep hash: |
552 | */ | 552 | */ |
553 | unwind_init(); | 553 | unwind_init(); |
554 | lockdep_init(); | 554 | lockdep_init(); |
555 | debug_objects_early_init(); | 555 | debug_objects_early_init(); |
556 | cgroup_init_early(); | 556 | cgroup_init_early(); |
557 | 557 | ||
558 | local_irq_disable(); | 558 | local_irq_disable(); |
559 | early_boot_irqs_off(); | 559 | early_boot_irqs_off(); |
560 | early_init_irq_lock_class(); | 560 | early_init_irq_lock_class(); |
561 | 561 | ||
562 | /* | 562 | /* |
563 | * Interrupts are still disabled. Do necessary setups, then | 563 | * Interrupts are still disabled. Do necessary setups, then |
564 | * enable them | 564 | * enable them |
565 | */ | 565 | */ |
566 | lock_kernel(); | 566 | lock_kernel(); |
567 | tick_init(); | 567 | tick_init(); |
568 | boot_cpu_init(); | 568 | boot_cpu_init(); |
569 | page_address_init(); | 569 | page_address_init(); |
570 | printk(KERN_NOTICE); | 570 | printk(KERN_NOTICE); |
571 | printk(linux_banner); | 571 | printk(linux_banner); |
572 | setup_arch(&command_line); | 572 | setup_arch(&command_line); |
573 | mm_init_owner(&init_mm, &init_task); | 573 | mm_init_owner(&init_mm, &init_task); |
574 | setup_command_line(command_line); | 574 | setup_command_line(command_line); |
575 | unwind_setup(); | 575 | unwind_setup(); |
576 | setup_per_cpu_areas(); | 576 | setup_per_cpu_areas(); |
577 | setup_nr_cpu_ids(); | 577 | setup_nr_cpu_ids(); |
578 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ | 578 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ |
579 | 579 | ||
580 | /* | 580 | /* |
581 | * Set up the scheduler prior starting any interrupts (such as the | 581 | * Set up the scheduler prior starting any interrupts (such as the |
582 | * timer interrupt). Full topology setup happens at smp_init() | 582 | * timer interrupt). Full topology setup happens at smp_init() |
583 | * time - but meanwhile we still have a functioning scheduler. | 583 | * time - but meanwhile we still have a functioning scheduler. |
584 | */ | 584 | */ |
585 | sched_init(); | 585 | sched_init(); |
586 | /* | 586 | /* |
587 | * Disable preemption - early bootup scheduling is extremely | 587 | * Disable preemption - early bootup scheduling is extremely |
588 | * fragile until we cpu_idle() for the first time. | 588 | * fragile until we cpu_idle() for the first time. |
589 | */ | 589 | */ |
590 | preempt_disable(); | 590 | preempt_disable(); |
591 | build_all_zonelists(); | 591 | build_all_zonelists(); |
592 | page_alloc_init(); | 592 | page_alloc_init(); |
593 | printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line); | 593 | printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line); |
594 | parse_early_param(); | 594 | parse_early_param(); |
595 | parse_args("Booting kernel", static_command_line, __start___param, | 595 | parse_args("Booting kernel", static_command_line, __start___param, |
596 | __stop___param - __start___param, | 596 | __stop___param - __start___param, |
597 | &unknown_bootoption); | 597 | &unknown_bootoption); |
598 | if (!irqs_disabled()) { | 598 | if (!irqs_disabled()) { |
599 | printk(KERN_WARNING "start_kernel(): bug: interrupts were " | 599 | printk(KERN_WARNING "start_kernel(): bug: interrupts were " |
600 | "enabled *very* early, fixing it\n"); | 600 | "enabled *very* early, fixing it\n"); |
601 | local_irq_disable(); | 601 | local_irq_disable(); |
602 | } | 602 | } |
603 | sort_main_extable(); | 603 | sort_main_extable(); |
604 | trap_init(); | 604 | trap_init(); |
605 | rcu_init(); | 605 | rcu_init(); |
606 | init_IRQ(); | 606 | init_IRQ(); |
607 | pidhash_init(); | 607 | pidhash_init(); |
608 | init_timers(); | 608 | init_timers(); |
609 | hrtimers_init(); | 609 | hrtimers_init(); |
610 | softirq_init(); | 610 | softirq_init(); |
611 | timekeeping_init(); | 611 | timekeeping_init(); |
612 | time_init(); | 612 | time_init(); |
613 | sched_clock_init(); | 613 | sched_clock_init(); |
614 | profile_init(); | 614 | profile_init(); |
615 | if (!irqs_disabled()) | 615 | if (!irqs_disabled()) |
616 | printk("start_kernel(): bug: interrupts were enabled early\n"); | 616 | printk("start_kernel(): bug: interrupts were enabled early\n"); |
617 | early_boot_irqs_on(); | 617 | early_boot_irqs_on(); |
618 | local_irq_enable(); | 618 | local_irq_enable(); |
619 | 619 | ||
620 | /* | 620 | /* |
621 | * HACK ALERT! This is early. We're enabling the console before | 621 | * HACK ALERT! This is early. We're enabling the console before |
622 | * we've done PCI setups etc, and console_init() must be aware of | 622 | * we've done PCI setups etc, and console_init() must be aware of |
623 | * this. But we do want output early, in case something goes wrong. | 623 | * this. But we do want output early, in case something goes wrong. |
624 | */ | 624 | */ |
625 | console_init(); | 625 | console_init(); |
626 | if (panic_later) | 626 | if (panic_later) |
627 | panic(panic_later, panic_param); | 627 | panic(panic_later, panic_param); |
628 | 628 | ||
629 | lockdep_info(); | 629 | lockdep_info(); |
630 | 630 | ||
631 | /* | 631 | /* |
632 | * Need to run this when irqs are enabled, because it wants | 632 | * Need to run this when irqs are enabled, because it wants |
633 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | 633 | * to self-test [hard/soft]-irqs on/off lock inversion bugs |
634 | * too: | 634 | * too: |
635 | */ | 635 | */ |
636 | locking_selftest(); | 636 | locking_selftest(); |
637 | 637 | ||
638 | #ifdef CONFIG_BLK_DEV_INITRD | 638 | #ifdef CONFIG_BLK_DEV_INITRD |
639 | if (initrd_start && !initrd_below_start_ok && | 639 | if (initrd_start && !initrd_below_start_ok && |
640 | page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) { | 640 | page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) { |
641 | printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " | 641 | printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " |
642 | "disabling it.\n", | 642 | "disabling it.\n", |
643 | page_to_pfn(virt_to_page((void *)initrd_start)), | 643 | page_to_pfn(virt_to_page((void *)initrd_start)), |
644 | min_low_pfn); | 644 | min_low_pfn); |
645 | initrd_start = 0; | 645 | initrd_start = 0; |
646 | } | 646 | } |
647 | #endif | 647 | #endif |
648 | vmalloc_init(); | 648 | vmalloc_init(); |
649 | vfs_caches_init_early(); | 649 | vfs_caches_init_early(); |
650 | cpuset_init_early(); | 650 | cpuset_init_early(); |
651 | page_cgroup_init(); | 651 | page_cgroup_init(); |
652 | mem_init(); | 652 | mem_init(); |
653 | enable_debug_pagealloc(); | 653 | enable_debug_pagealloc(); |
654 | cpu_hotplug_init(); | 654 | cpu_hotplug_init(); |
655 | kmem_cache_init(); | 655 | kmem_cache_init(); |
656 | debug_objects_mem_init(); | 656 | debug_objects_mem_init(); |
657 | idr_init_cache(); | 657 | idr_init_cache(); |
658 | setup_per_cpu_pageset(); | 658 | setup_per_cpu_pageset(); |
659 | numa_policy_init(); | 659 | numa_policy_init(); |
660 | if (late_time_init) | 660 | if (late_time_init) |
661 | late_time_init(); | 661 | late_time_init(); |
662 | calibrate_delay(); | 662 | calibrate_delay(); |
663 | pidmap_init(); | 663 | pidmap_init(); |
664 | pgtable_cache_init(); | 664 | pgtable_cache_init(); |
665 | prio_tree_init(); | 665 | prio_tree_init(); |
666 | anon_vma_init(); | 666 | anon_vma_init(); |
667 | #ifdef CONFIG_X86 | 667 | #ifdef CONFIG_X86 |
668 | if (efi_enabled) | 668 | if (efi_enabled) |
669 | efi_enter_virtual_mode(); | 669 | efi_enter_virtual_mode(); |
670 | #endif | 670 | #endif |
671 | thread_info_cache_init(); | 671 | thread_info_cache_init(); |
672 | fork_init(num_physpages); | 672 | fork_init(num_physpages); |
673 | proc_caches_init(); | 673 | proc_caches_init(); |
674 | buffer_init(); | 674 | buffer_init(); |
675 | key_init(); | 675 | key_init(); |
676 | security_init(); | 676 | security_init(); |
677 | vfs_caches_init(num_physpages); | 677 | vfs_caches_init(num_physpages); |
678 | radix_tree_init(); | 678 | radix_tree_init(); |
679 | signals_init(); | 679 | signals_init(); |
680 | /* rootfs populating might need page-writeback */ | 680 | /* rootfs populating might need page-writeback */ |
681 | page_writeback_init(); | 681 | page_writeback_init(); |
682 | #ifdef CONFIG_PROC_FS | 682 | #ifdef CONFIG_PROC_FS |
683 | proc_root_init(); | 683 | proc_root_init(); |
684 | #endif | 684 | #endif |
685 | cgroup_init(); | 685 | cgroup_init(); |
686 | cpuset_init(); | 686 | cpuset_init(); |
687 | taskstats_init_early(); | 687 | taskstats_init_early(); |
688 | delayacct_init(); | 688 | delayacct_init(); |
689 | 689 | ||
690 | check_bugs(); | 690 | check_bugs(); |
691 | 691 | ||
692 | acpi_early_init(); /* before LAPIC and SMP init */ | 692 | acpi_early_init(); /* before LAPIC and SMP init */ |
693 | 693 | ||
694 | ftrace_init(); | 694 | ftrace_init(); |
695 | 695 | ||
696 | /* Do the rest non-__init'ed, we're now alive */ | 696 | /* Do the rest non-__init'ed, we're now alive */ |
697 | rest_init(); | 697 | rest_init(); |
698 | } | 698 | } |
699 | 699 | ||
700 | static int initcall_debug; | 700 | static int initcall_debug; |
701 | core_param(initcall_debug, initcall_debug, bool, 0644); | 701 | core_param(initcall_debug, initcall_debug, bool, 0644); |
702 | 702 | ||
703 | int do_one_initcall(initcall_t fn) | 703 | int do_one_initcall(initcall_t fn) |
704 | { | 704 | { |
705 | int count = preempt_count(); | 705 | int count = preempt_count(); |
706 | ktime_t delta; | 706 | ktime_t delta; |
707 | char msgbuf[64]; | 707 | char msgbuf[64]; |
708 | struct boot_trace it; | 708 | struct boot_trace it; |
709 | 709 | ||
710 | if (initcall_debug) { | 710 | if (initcall_debug) { |
711 | it.caller = task_pid_nr(current); | 711 | it.caller = task_pid_nr(current); |
712 | printk("calling %pF @ %i\n", fn, it.caller); | 712 | printk("calling %pF @ %i\n", fn, it.caller); |
713 | it.calltime = ktime_get(); | 713 | it.calltime = ktime_get(); |
714 | } | 714 | } |
715 | 715 | ||
716 | it.result = fn(); | 716 | it.result = fn(); |
717 | 717 | ||
718 | if (initcall_debug) { | 718 | if (initcall_debug) { |
719 | it.rettime = ktime_get(); | 719 | it.rettime = ktime_get(); |
720 | delta = ktime_sub(it.rettime, it.calltime); | 720 | delta = ktime_sub(it.rettime, it.calltime); |
721 | it.duration = (unsigned long long) delta.tv64 >> 10; | 721 | it.duration = (unsigned long long) delta.tv64 >> 10; |
722 | printk("initcall %pF returned %d after %Ld usecs\n", fn, | 722 | printk("initcall %pF returned %d after %Ld usecs\n", fn, |
723 | it.result, it.duration); | 723 | it.result, it.duration); |
724 | trace_boot(&it, fn); | 724 | trace_boot(&it, fn); |
725 | } | 725 | } |
726 | 726 | ||
727 | msgbuf[0] = 0; | 727 | msgbuf[0] = 0; |
728 | 728 | ||
729 | if (it.result && it.result != -ENODEV && initcall_debug) | 729 | if (it.result && it.result != -ENODEV && initcall_debug) |
730 | sprintf(msgbuf, "error code %d ", it.result); | 730 | sprintf(msgbuf, "error code %d ", it.result); |
731 | 731 | ||
732 | if (preempt_count() != count) { | 732 | if (preempt_count() != count) { |
733 | strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf)); | 733 | strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf)); |
734 | preempt_count() = count; | 734 | preempt_count() = count; |
735 | } | 735 | } |
736 | if (irqs_disabled()) { | 736 | if (irqs_disabled()) { |
737 | strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); | 737 | strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); |
738 | local_irq_enable(); | 738 | local_irq_enable(); |
739 | } | 739 | } |
740 | if (msgbuf[0]) { | 740 | if (msgbuf[0]) { |
741 | printk("initcall %pF returned with %s\n", fn, msgbuf); | 741 | printk("initcall %pF returned with %s\n", fn, msgbuf); |
742 | } | 742 | } |
743 | 743 | ||
744 | return it.result; | 744 | return it.result; |
745 | } | 745 | } |
746 | 746 | ||
747 | 747 | ||
748 | extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[]; | 748 | extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[]; |
749 | 749 | ||
750 | static void __init do_initcalls(void) | 750 | static void __init do_initcalls(void) |
751 | { | 751 | { |
752 | initcall_t *call; | 752 | initcall_t *call; |
753 | 753 | ||
754 | for (call = __early_initcall_end; call < __initcall_end; call++) | 754 | for (call = __early_initcall_end; call < __initcall_end; call++) |
755 | do_one_initcall(*call); | 755 | do_one_initcall(*call); |
756 | 756 | ||
757 | /* Make sure there is no pending stuff from the initcall sequence */ | 757 | /* Make sure there is no pending stuff from the initcall sequence */ |
758 | flush_scheduled_work(); | 758 | flush_scheduled_work(); |
759 | } | 759 | } |
760 | 760 | ||
761 | /* | 761 | /* |
762 | * Ok, the machine is now initialized. None of the devices | 762 | * Ok, the machine is now initialized. None of the devices |
763 | * have been touched yet, but the CPU subsystem is up and | 763 | * have been touched yet, but the CPU subsystem is up and |
764 | * running, and memory and process management works. | 764 | * running, and memory and process management works. |
765 | * | 765 | * |
766 | * Now we can finally start doing some real work.. | 766 | * Now we can finally start doing some real work.. |
767 | */ | 767 | */ |
768 | static void __init do_basic_setup(void) | 768 | static void __init do_basic_setup(void) |
769 | { | 769 | { |
770 | rcu_init_sched(); /* needed by module_init stage. */ | 770 | rcu_init_sched(); /* needed by module_init stage. */ |
771 | init_workqueues(); | ||
771 | usermodehelper_init(); | 772 | usermodehelper_init(); |
772 | driver_init(); | 773 | driver_init(); |
773 | init_irq_proc(); | 774 | init_irq_proc(); |
774 | do_initcalls(); | 775 | do_initcalls(); |
775 | } | 776 | } |
776 | 777 | ||
777 | static void __init do_pre_smp_initcalls(void) | 778 | static void __init do_pre_smp_initcalls(void) |
778 | { | 779 | { |
779 | initcall_t *call; | 780 | initcall_t *call; |
780 | 781 | ||
781 | for (call = __initcall_start; call < __early_initcall_end; call++) | 782 | for (call = __initcall_start; call < __early_initcall_end; call++) |
782 | do_one_initcall(*call); | 783 | do_one_initcall(*call); |
783 | } | 784 | } |
784 | 785 | ||
785 | static void run_init_process(char *init_filename) | 786 | static void run_init_process(char *init_filename) |
786 | { | 787 | { |
787 | argv_init[0] = init_filename; | 788 | argv_init[0] = init_filename; |
788 | kernel_execve(init_filename, argv_init, envp_init); | 789 | kernel_execve(init_filename, argv_init, envp_init); |
789 | } | 790 | } |
790 | 791 | ||
791 | /* This is a non __init function. Force it to be noinline otherwise gcc | 792 | /* This is a non __init function. Force it to be noinline otherwise gcc |
792 | * makes it inline to init() and it becomes part of init.text section | 793 | * makes it inline to init() and it becomes part of init.text section |
793 | */ | 794 | */ |
794 | static int noinline init_post(void) | 795 | static int noinline init_post(void) |
795 | { | 796 | { |
796 | free_initmem(); | 797 | free_initmem(); |
797 | unlock_kernel(); | 798 | unlock_kernel(); |
798 | mark_rodata_ro(); | 799 | mark_rodata_ro(); |
799 | system_state = SYSTEM_RUNNING; | 800 | system_state = SYSTEM_RUNNING; |
800 | numa_default_policy(); | 801 | numa_default_policy(); |
801 | 802 | ||
802 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | 803 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) |
803 | printk(KERN_WARNING "Warning: unable to open an initial console.\n"); | 804 | printk(KERN_WARNING "Warning: unable to open an initial console.\n"); |
804 | 805 | ||
805 | (void) sys_dup(0); | 806 | (void) sys_dup(0); |
806 | (void) sys_dup(0); | 807 | (void) sys_dup(0); |
807 | 808 | ||
808 | current->signal->flags |= SIGNAL_UNKILLABLE; | 809 | current->signal->flags |= SIGNAL_UNKILLABLE; |
809 | 810 | ||
810 | if (ramdisk_execute_command) { | 811 | if (ramdisk_execute_command) { |
811 | run_init_process(ramdisk_execute_command); | 812 | run_init_process(ramdisk_execute_command); |
812 | printk(KERN_WARNING "Failed to execute %s\n", | 813 | printk(KERN_WARNING "Failed to execute %s\n", |
813 | ramdisk_execute_command); | 814 | ramdisk_execute_command); |
814 | } | 815 | } |
815 | 816 | ||
816 | /* | 817 | /* |
817 | * We try each of these until one succeeds. | 818 | * We try each of these until one succeeds. |
818 | * | 819 | * |
819 | * The Bourne shell can be used instead of init if we are | 820 | * The Bourne shell can be used instead of init if we are |
820 | * trying to recover a really broken machine. | 821 | * trying to recover a really broken machine. |
821 | */ | 822 | */ |
822 | if (execute_command) { | 823 | if (execute_command) { |
823 | run_init_process(execute_command); | 824 | run_init_process(execute_command); |
824 | printk(KERN_WARNING "Failed to execute %s. Attempting " | 825 | printk(KERN_WARNING "Failed to execute %s. Attempting " |
825 | "defaults...\n", execute_command); | 826 | "defaults...\n", execute_command); |
826 | } | 827 | } |
827 | run_init_process("/sbin/init"); | 828 | run_init_process("/sbin/init"); |
828 | run_init_process("/etc/init"); | 829 | run_init_process("/etc/init"); |
829 | run_init_process("/bin/init"); | 830 | run_init_process("/bin/init"); |
830 | run_init_process("/bin/sh"); | 831 | run_init_process("/bin/sh"); |
831 | 832 | ||
832 | panic("No init found. Try passing init= option to kernel."); | 833 | panic("No init found. Try passing init= option to kernel."); |
833 | } | 834 | } |
834 | 835 | ||
835 | static int __init kernel_init(void * unused) | 836 | static int __init kernel_init(void * unused) |
836 | { | 837 | { |
837 | lock_kernel(); | 838 | lock_kernel(); |
838 | /* | 839 | /* |
839 | * init can run on any cpu. | 840 | * init can run on any cpu. |
840 | */ | 841 | */ |
841 | set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR); | 842 | set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR); |
842 | /* | 843 | /* |
843 | * Tell the world that we're going to be the grim | 844 | * Tell the world that we're going to be the grim |
844 | * reaper of innocent orphaned children. | 845 | * reaper of innocent orphaned children. |
845 | * | 846 | * |
846 | * We don't want people to have to make incorrect | 847 | * We don't want people to have to make incorrect |
847 | * assumptions about where in the task array this | 848 | * assumptions about where in the task array this |
848 | * can be found. | 849 | * can be found. |
849 | */ | 850 | */ |
850 | init_pid_ns.child_reaper = current; | 851 | init_pid_ns.child_reaper = current; |
851 | 852 | ||
852 | cad_pid = task_pid(current); | 853 | cad_pid = task_pid(current); |
853 | |||
854 | init_workqueues(); | ||
855 | 854 | ||
856 | smp_prepare_cpus(setup_max_cpus); | 855 | smp_prepare_cpus(setup_max_cpus); |
857 | 856 | ||
858 | do_pre_smp_initcalls(); | 857 | do_pre_smp_initcalls(); |
859 | start_boot_trace(); | 858 | start_boot_trace(); |
860 | 859 | ||
861 | smp_init(); | 860 | smp_init(); |
862 | sched_init_smp(); | 861 | sched_init_smp(); |
863 | 862 | ||
864 | cpuset_init_smp(); | 863 | cpuset_init_smp(); |
865 | 864 | ||
866 | do_basic_setup(); | 865 | do_basic_setup(); |
867 | 866 | ||
868 | /* | 867 | /* |
869 | * check if there is an early userspace init. If yes, let it do all | 868 | * check if there is an early userspace init. If yes, let it do all |
870 | * the work | 869 | * the work |
871 | */ | 870 | */ |
872 | 871 | ||
873 | if (!ramdisk_execute_command) | 872 | if (!ramdisk_execute_command) |
874 | ramdisk_execute_command = "/init"; | 873 | ramdisk_execute_command = "/init"; |
875 | 874 | ||
876 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | 875 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { |
877 | ramdisk_execute_command = NULL; | 876 | ramdisk_execute_command = NULL; |
878 | prepare_namespace(); | 877 | prepare_namespace(); |
879 | } | 878 | } |
880 | 879 | ||
881 | /* | 880 | /* |
882 | * Ok, we have completed the initial bootup, and | 881 | * Ok, we have completed the initial bootup, and |
883 | * we're essentially up and running. Get rid of the | 882 | * we're essentially up and running. Get rid of the |
884 | * initmem segments and start the user-mode stuff.. | 883 | * initmem segments and start the user-mode stuff.. |
885 | */ | 884 | */ |
886 | stop_boot_trace(); | 885 | stop_boot_trace(); |
887 | init_post(); | 886 | init_post(); |
888 | return 0; | 887 | return 0; |
889 | } | 888 | } |
kernel/stop_machine.c
1 | /* Copyright 2008, 2005 Rusty Russell rusty@rustcorp.com.au IBM Corporation. | 1 | /* Copyright 2008, 2005 Rusty Russell rusty@rustcorp.com.au IBM Corporation. |
2 | * GPL v2 and any later version. | 2 | * GPL v2 and any later version. |
3 | */ | 3 | */ |
4 | #include <linux/cpu.h> | 4 | #include <linux/cpu.h> |
5 | #include <linux/err.h> | 5 | #include <linux/err.h> |
6 | #include <linux/kthread.h> | 6 | #include <linux/kthread.h> |
7 | #include <linux/module.h> | 7 | #include <linux/module.h> |
8 | #include <linux/sched.h> | 8 | #include <linux/sched.h> |
9 | #include <linux/stop_machine.h> | 9 | #include <linux/stop_machine.h> |
10 | #include <linux/syscalls.h> | 10 | #include <linux/syscalls.h> |
11 | #include <linux/interrupt.h> | 11 | #include <linux/interrupt.h> |
12 | 12 | ||
13 | #include <asm/atomic.h> | 13 | #include <asm/atomic.h> |
14 | #include <asm/uaccess.h> | 14 | #include <asm/uaccess.h> |
15 | 15 | ||
16 | /* This controls the threads on each CPU. */ | 16 | /* This controls the threads on each CPU. */ |
17 | enum stopmachine_state { | 17 | enum stopmachine_state { |
18 | /* Dummy starting state for thread. */ | 18 | /* Dummy starting state for thread. */ |
19 | STOPMACHINE_NONE, | 19 | STOPMACHINE_NONE, |
20 | /* Awaiting everyone to be scheduled. */ | 20 | /* Awaiting everyone to be scheduled. */ |
21 | STOPMACHINE_PREPARE, | 21 | STOPMACHINE_PREPARE, |
22 | /* Disable interrupts. */ | 22 | /* Disable interrupts. */ |
23 | STOPMACHINE_DISABLE_IRQ, | 23 | STOPMACHINE_DISABLE_IRQ, |
24 | /* Run the function */ | 24 | /* Run the function */ |
25 | STOPMACHINE_RUN, | 25 | STOPMACHINE_RUN, |
26 | /* Exit */ | 26 | /* Exit */ |
27 | STOPMACHINE_EXIT, | 27 | STOPMACHINE_EXIT, |
28 | }; | 28 | }; |
29 | static enum stopmachine_state state; | 29 | static enum stopmachine_state state; |
30 | 30 | ||
31 | struct stop_machine_data { | 31 | struct stop_machine_data { |
32 | int (*fn)(void *); | 32 | int (*fn)(void *); |
33 | void *data; | 33 | void *data; |
34 | int fnret; | 34 | int fnret; |
35 | }; | 35 | }; |
36 | 36 | ||
37 | /* Like num_online_cpus(), but hotplug cpu uses us, so we need this. */ | 37 | /* Like num_online_cpus(), but hotplug cpu uses us, so we need this. */ |
38 | static unsigned int num_threads; | 38 | static unsigned int num_threads; |
39 | static atomic_t thread_ack; | 39 | static atomic_t thread_ack; |
40 | static DEFINE_MUTEX(lock); | 40 | static DEFINE_MUTEX(lock); |
41 | 41 | ||
42 | static struct workqueue_struct *stop_machine_wq; | 42 | static struct workqueue_struct *stop_machine_wq; |
43 | static struct stop_machine_data active, idle; | 43 | static struct stop_machine_data active, idle; |
44 | static const cpumask_t *active_cpus; | 44 | static const cpumask_t *active_cpus; |
45 | static void *stop_machine_work; | 45 | static void *stop_machine_work; |
46 | 46 | ||
47 | static void set_state(enum stopmachine_state newstate) | 47 | static void set_state(enum stopmachine_state newstate) |
48 | { | 48 | { |
49 | /* Reset ack counter. */ | 49 | /* Reset ack counter. */ |
50 | atomic_set(&thread_ack, num_threads); | 50 | atomic_set(&thread_ack, num_threads); |
51 | smp_wmb(); | 51 | smp_wmb(); |
52 | state = newstate; | 52 | state = newstate; |
53 | } | 53 | } |
54 | 54 | ||
55 | /* Last one to ack a state moves to the next state. */ | 55 | /* Last one to ack a state moves to the next state. */ |
56 | static void ack_state(void) | 56 | static void ack_state(void) |
57 | { | 57 | { |
58 | if (atomic_dec_and_test(&thread_ack)) | 58 | if (atomic_dec_and_test(&thread_ack)) |
59 | set_state(state + 1); | 59 | set_state(state + 1); |
60 | } | 60 | } |
61 | 61 | ||
62 | /* This is the actual function which stops the CPU. It runs | 62 | /* This is the actual function which stops the CPU. It runs |
63 | * in the context of a dedicated stopmachine workqueue. */ | 63 | * in the context of a dedicated stopmachine workqueue. */ |
64 | static void stop_cpu(struct work_struct *unused) | 64 | static void stop_cpu(struct work_struct *unused) |
65 | { | 65 | { |
66 | enum stopmachine_state curstate = STOPMACHINE_NONE; | 66 | enum stopmachine_state curstate = STOPMACHINE_NONE; |
67 | struct stop_machine_data *smdata = &idle; | 67 | struct stop_machine_data *smdata = &idle; |
68 | int cpu = smp_processor_id(); | 68 | int cpu = smp_processor_id(); |
69 | int err; | 69 | int err; |
70 | 70 | ||
71 | if (!active_cpus) { | 71 | if (!active_cpus) { |
72 | if (cpu == first_cpu(cpu_online_map)) | 72 | if (cpu == first_cpu(cpu_online_map)) |
73 | smdata = &active; | 73 | smdata = &active; |
74 | } else { | 74 | } else { |
75 | if (cpu_isset(cpu, *active_cpus)) | 75 | if (cpu_isset(cpu, *active_cpus)) |
76 | smdata = &active; | 76 | smdata = &active; |
77 | } | 77 | } |
78 | /* Simple state machine */ | 78 | /* Simple state machine */ |
79 | do { | 79 | do { |
80 | /* Chill out and ensure we re-read stopmachine_state. */ | 80 | /* Chill out and ensure we re-read stopmachine_state. */ |
81 | cpu_relax(); | 81 | cpu_relax(); |
82 | if (state != curstate) { | 82 | if (state != curstate) { |
83 | curstate = state; | 83 | curstate = state; |
84 | switch (curstate) { | 84 | switch (curstate) { |
85 | case STOPMACHINE_DISABLE_IRQ: | 85 | case STOPMACHINE_DISABLE_IRQ: |
86 | local_irq_disable(); | 86 | local_irq_disable(); |
87 | hard_irq_disable(); | 87 | hard_irq_disable(); |
88 | break; | 88 | break; |
89 | case STOPMACHINE_RUN: | 89 | case STOPMACHINE_RUN: |
90 | /* On multiple CPUs only a single error code | 90 | /* On multiple CPUs only a single error code |
91 | * is needed to tell that something failed. */ | 91 | * is needed to tell that something failed. */ |
92 | err = smdata->fn(smdata->data); | 92 | err = smdata->fn(smdata->data); |
93 | if (err) | 93 | if (err) |
94 | smdata->fnret = err; | 94 | smdata->fnret = err; |
95 | break; | 95 | break; |
96 | default: | 96 | default: |
97 | break; | 97 | break; |
98 | } | 98 | } |
99 | ack_state(); | 99 | ack_state(); |
100 | } | 100 | } |
101 | } while (curstate != STOPMACHINE_EXIT); | 101 | } while (curstate != STOPMACHINE_EXIT); |
102 | 102 | ||
103 | local_irq_enable(); | 103 | local_irq_enable(); |
104 | } | 104 | } |
105 | 105 | ||
106 | /* Callback for CPUs which aren't supposed to do anything. */ | 106 | /* Callback for CPUs which aren't supposed to do anything. */ |
107 | static int chill(void *unused) | 107 | static int chill(void *unused) |
108 | { | 108 | { |
109 | return 0; | 109 | return 0; |
110 | } | 110 | } |
111 | 111 | ||
112 | int __stop_machine(int (*fn)(void *), void *data, const cpumask_t *cpus) | 112 | int __stop_machine(int (*fn)(void *), void *data, const cpumask_t *cpus) |
113 | { | 113 | { |
114 | struct work_struct *sm_work; | 114 | struct work_struct *sm_work; |
115 | int i; | 115 | int i; |
116 | 116 | ||
117 | /* Set up initial state. */ | 117 | /* Set up initial state. */ |
118 | mutex_lock(&lock); | 118 | mutex_lock(&lock); |
119 | num_threads = num_online_cpus(); | 119 | num_threads = num_online_cpus(); |
120 | active_cpus = cpus; | 120 | active_cpus = cpus; |
121 | active.fn = fn; | 121 | active.fn = fn; |
122 | active.data = data; | 122 | active.data = data; |
123 | active.fnret = 0; | 123 | active.fnret = 0; |
124 | idle.fn = chill; | 124 | idle.fn = chill; |
125 | idle.data = NULL; | 125 | idle.data = NULL; |
126 | 126 | ||
127 | set_state(STOPMACHINE_PREPARE); | 127 | set_state(STOPMACHINE_PREPARE); |
128 | 128 | ||
129 | /* Schedule the stop_cpu work on all cpus: hold this CPU so one | 129 | /* Schedule the stop_cpu work on all cpus: hold this CPU so one |
130 | * doesn't hit this CPU until we're ready. */ | 130 | * doesn't hit this CPU until we're ready. */ |
131 | get_cpu(); | 131 | get_cpu(); |
132 | for_each_online_cpu(i) { | 132 | for_each_online_cpu(i) { |
133 | sm_work = percpu_ptr(stop_machine_work, i); | 133 | sm_work = percpu_ptr(stop_machine_work, i); |
134 | INIT_WORK(sm_work, stop_cpu); | 134 | INIT_WORK(sm_work, stop_cpu); |
135 | queue_work_on(i, stop_machine_wq, sm_work); | 135 | queue_work_on(i, stop_machine_wq, sm_work); |
136 | } | 136 | } |
137 | /* This will release the thread on our CPU. */ | 137 | /* This will release the thread on our CPU. */ |
138 | put_cpu(); | 138 | put_cpu(); |
139 | flush_workqueue(stop_machine_wq); | 139 | flush_workqueue(stop_machine_wq); |
140 | mutex_unlock(&lock); | 140 | mutex_unlock(&lock); |
141 | return active.fnret; | 141 | return active.fnret; |
142 | } | 142 | } |
143 | 143 | ||
144 | int stop_machine(int (*fn)(void *), void *data, const cpumask_t *cpus) | 144 | int stop_machine(int (*fn)(void *), void *data, const cpumask_t *cpus) |
145 | { | 145 | { |
146 | int ret; | 146 | int ret; |
147 | 147 | ||
148 | /* No CPUs can come up or down during this. */ | 148 | /* No CPUs can come up or down during this. */ |
149 | get_online_cpus(); | 149 | get_online_cpus(); |
150 | ret = __stop_machine(fn, data, cpus); | 150 | ret = __stop_machine(fn, data, cpus); |
151 | put_online_cpus(); | 151 | put_online_cpus(); |
152 | 152 | ||
153 | return ret; | 153 | return ret; |
154 | } | 154 | } |
155 | EXPORT_SYMBOL_GPL(stop_machine); | 155 | EXPORT_SYMBOL_GPL(stop_machine); |
156 | 156 | ||
157 | static int __init stop_machine_init(void) | 157 | static int __init stop_machine_init(void) |
158 | { | 158 | { |
159 | stop_machine_wq = create_rt_workqueue("kstop"); | 159 | stop_machine_wq = create_rt_workqueue("kstop"); |
160 | stop_machine_work = alloc_percpu(struct work_struct); | 160 | stop_machine_work = alloc_percpu(struct work_struct); |
161 | return 0; | 161 | return 0; |
162 | } | 162 | } |
163 | early_initcall(stop_machine_init); | 163 | core_initcall(stop_machine_init); |
164 | 164 |