diff options
-rw-r--r-- | Documentation/kernel-parameters.txt | 45 | ||||
-rw-r--r-- | arch/x86/kernel/acpi/sleep.c | 2 | ||||
-rw-r--r-- | arch/x86/kernel/e820.c | 21 | ||||
-rw-r--r-- | drivers/acpi/main.c | 67 | ||||
-rw-r--r-- | include/linux/acpi.h | 1 | ||||
-rw-r--r-- | include/linux/suspend.h | 13 | ||||
-rw-r--r-- | kernel/power/disk.c | 6 | ||||
-rw-r--r-- | kernel/power/snapshot.c | 370 | ||||
-rw-r--r-- | kernel/power/swsusp.c | 122 |
9 files changed, 446 insertions, 201 deletions
diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index 136f02842de9..f3a89fac2202 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt | |||
@@ -150,16 +150,20 @@ and is between 256 and 4096 characters. It is defined in the file | |||
150 | default: 0 | 150 | default: 0 |
151 | 151 | ||
152 | acpi_sleep= [HW,ACPI] Sleep options | 152 | acpi_sleep= [HW,ACPI] Sleep options |
153 | Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig, old_ordering } | 153 | Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig, |
154 | See Documentation/power/video.txt for s3_bios and s3_mode. | 154 | old_ordering, s4_nonvs } |
155 | See Documentation/power/video.txt for information on | ||
156 | s3_bios and s3_mode. | ||
155 | s3_beep is for debugging; it makes the PC's speaker beep | 157 | s3_beep is for debugging; it makes the PC's speaker beep |
156 | as soon as the kernel's real-mode entry point is called. | 158 | as soon as the kernel's real-mode entry point is called. |
157 | s4_nohwsig prevents ACPI hardware signature from being | 159 | s4_nohwsig prevents ACPI hardware signature from being |
158 | used during resume from hibernation. | 160 | used during resume from hibernation. |
159 | old_ordering causes the ACPI 1.0 ordering of the _PTS | 161 | old_ordering causes the ACPI 1.0 ordering of the _PTS |
160 | control method, wrt putting devices into low power | 162 | control method, with respect to putting devices into |
161 | states, to be enforced (the ACPI 2.0 ordering of _PTS is | 163 | low power states, to be enforced (the ACPI 2.0 ordering |
162 | used by default). | 164 | of _PTS is used by default). |
165 | s4_nonvs prevents the kernel from saving/restoring the | ||
166 | ACPI NVS memory during hibernation. | ||
163 | 167 | ||
164 | acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode | 168 | acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode |
165 | Format: { level | edge | high | low } | 169 | Format: { level | edge | high | low } |
@@ -194,7 +198,7 @@ and is between 256 and 4096 characters. It is defined in the file | |||
194 | acpi_skip_timer_override [HW,ACPI] | 198 | acpi_skip_timer_override [HW,ACPI] |
195 | Recognize and ignore IRQ0/pin2 Interrupt Override. | 199 | Recognize and ignore IRQ0/pin2 Interrupt Override. |
196 | For broken nForce2 BIOS resulting in XT-PIC timer. | 200 | For broken nForce2 BIOS resulting in XT-PIC timer. |
197 | acpi_use_timer_override [HW,ACPI} | 201 | acpi_use_timer_override [HW,ACPI] |
198 | Use timer override. For some broken Nvidia NF5 boards | 202 | Use timer override. For some broken Nvidia NF5 boards |
199 | that require a timer override, but don't have | 203 | that require a timer override, but don't have |
200 | HPET | 204 | HPET |
@@ -861,17 +865,19 @@ and is between 256 and 4096 characters. It is defined in the file | |||
861 | See Documentation/ide/ide.txt. | 865 | See Documentation/ide/ide.txt. |
862 | 866 | ||
863 | idle= [X86] | 867 | idle= [X86] |
864 | Format: idle=poll or idle=mwait, idle=halt, idle=nomwait | 868 | Format: idle=poll, idle=mwait, idle=halt, idle=nomwait |
865 | Poll forces a polling idle loop that can slightly improves the performance | 869 | Poll forces a polling idle loop that can slightly |
866 | of waking up a idle CPU, but will use a lot of power and make the system | 870 | improve the performance of waking up a idle CPU, but |
867 | run hot. Not recommended. | 871 | will use a lot of power and make the system run hot. |
868 | idle=mwait. On systems which support MONITOR/MWAIT but the kernel chose | 872 | Not recommended. |
869 | to not use it because it doesn't save as much power as a normal idle | 873 | idle=mwait: On systems which support MONITOR/MWAIT but |
870 | loop use the MONITOR/MWAIT idle loop anyways. Performance should be the same | 874 | the kernel chose to not use it because it doesn't save |
871 | as idle=poll. | 875 | as much power as a normal idle loop, use the |
872 | idle=halt. Halt is forced to be used for CPU idle. | 876 | MONITOR/MWAIT idle loop anyways. Performance should be |
877 | the same as idle=poll. | ||
878 | idle=halt: Halt is forced to be used for CPU idle. | ||
873 | In such case C2/C3 won't be used again. | 879 | In such case C2/C3 won't be used again. |
874 | idle=nomwait. Disable mwait for CPU C-states | 880 | idle=nomwait: Disable mwait for CPU C-states |
875 | 881 | ||
876 | ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem | 882 | ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem |
877 | Claim all unknown PCI IDE storage controllers. | 883 | Claim all unknown PCI IDE storage controllers. |
@@ -1053,8 +1059,8 @@ and is between 256 and 4096 characters. It is defined in the file | |||
1053 | lapic [X86-32,APIC] Enable the local APIC even if BIOS | 1059 | lapic [X86-32,APIC] Enable the local APIC even if BIOS |
1054 | disabled it. | 1060 | disabled it. |
1055 | 1061 | ||
1056 | lapic_timer_c2_ok [X86-32,x86-64,APIC] trust the local apic timer in | 1062 | lapic_timer_c2_ok [X86-32,x86-64,APIC] trust the local apic timer |
1057 | C2 power state. | 1063 | in C2 power state. |
1058 | 1064 | ||
1059 | libata.dma= [LIBATA] DMA control | 1065 | libata.dma= [LIBATA] DMA control |
1060 | libata.dma=0 Disable all PATA and SATA DMA | 1066 | libata.dma=0 Disable all PATA and SATA DMA |
@@ -2242,7 +2248,8 @@ and is between 256 and 4096 characters. It is defined in the file | |||
2242 | 2248 | ||
2243 | thermal.psv= [HW,ACPI] | 2249 | thermal.psv= [HW,ACPI] |
2244 | -1: disable all passive trip points | 2250 | -1: disable all passive trip points |
2245 | <degrees C>: override all passive trip points to this value | 2251 | <degrees C>: override all passive trip points to this |
2252 | value | ||
2246 | 2253 | ||
2247 | thermal.tzp= [HW,ACPI] | 2254 | thermal.tzp= [HW,ACPI] |
2248 | Specify global default ACPI thermal zone polling rate | 2255 | Specify global default ACPI thermal zone polling rate |
diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index 806b4e9051b4..707c1f6f95fa 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c | |||
@@ -159,6 +159,8 @@ static int __init acpi_sleep_setup(char *str) | |||
159 | #endif | 159 | #endif |
160 | if (strncmp(str, "old_ordering", 12) == 0) | 160 | if (strncmp(str, "old_ordering", 12) == 0) |
161 | acpi_old_suspend_ordering(); | 161 | acpi_old_suspend_ordering(); |
162 | if (strncmp(str, "s4_nonvs", 8) == 0) | ||
163 | acpi_s4_no_nvs(); | ||
162 | str = strchr(str, ','); | 164 | str = strchr(str, ','); |
163 | if (str != NULL) | 165 | if (str != NULL) |
164 | str += strspn(str, ", \t"); | 166 | str += strspn(str, ", \t"); |
diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index 7aafeb5263ef..74c6a21fdc8c 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c | |||
@@ -665,6 +665,27 @@ void __init e820_mark_nosave_regions(unsigned long limit_pfn) | |||
665 | } | 665 | } |
666 | #endif | 666 | #endif |
667 | 667 | ||
668 | #ifdef CONFIG_HIBERNATION | ||
669 | /** | ||
670 | * Mark ACPI NVS memory region, so that we can save/restore it during | ||
671 | * hibernation and the subsequent resume. | ||
672 | */ | ||
673 | static int __init e820_mark_nvs_memory(void) | ||
674 | { | ||
675 | int i; | ||
676 | |||
677 | for (i = 0; i < e820.nr_map; i++) { | ||
678 | struct e820entry *ei = &e820.map[i]; | ||
679 | |||
680 | if (ei->type == E820_NVS) | ||
681 | hibernate_nvs_register(ei->addr, ei->size); | ||
682 | } | ||
683 | |||
684 | return 0; | ||
685 | } | ||
686 | core_initcall(e820_mark_nvs_memory); | ||
687 | #endif | ||
688 | |||
668 | /* | 689 | /* |
669 | * Early reserved memory areas. | 690 | * Early reserved memory areas. |
670 | */ | 691 | */ |
diff --git a/drivers/acpi/main.c b/drivers/acpi/main.c index d8242772de92..7e3c609cbef2 100644 --- a/drivers/acpi/main.c +++ b/drivers/acpi/main.c | |||
@@ -101,6 +101,19 @@ void __init acpi_old_suspend_ordering(void) | |||
101 | * cases. | 101 | * cases. |
102 | */ | 102 | */ |
103 | static bool set_sci_en_on_resume; | 103 | static bool set_sci_en_on_resume; |
104 | /* | ||
105 | * The ACPI specification wants us to save NVS memory regions during hibernation | ||
106 | * and to restore them during the subsequent resume. However, it is not certain | ||
107 | * if this mechanism is going to work on all machines, so we allow the user to | ||
108 | * disable this mechanism using the 'acpi_sleep=s4_nonvs' kernel command line | ||
109 | * option. | ||
110 | */ | ||
111 | static bool s4_no_nvs; | ||
112 | |||
113 | void __init acpi_s4_no_nvs(void) | ||
114 | { | ||
115 | s4_no_nvs = true; | ||
116 | } | ||
104 | 117 | ||
105 | /** | 118 | /** |
106 | * acpi_pm_disable_gpes - Disable the GPEs. | 119 | * acpi_pm_disable_gpes - Disable the GPEs. |
@@ -394,9 +407,25 @@ void __init acpi_no_s4_hw_signature(void) | |||
394 | 407 | ||
395 | static int acpi_hibernation_begin(void) | 408 | static int acpi_hibernation_begin(void) |
396 | { | 409 | { |
397 | acpi_target_sleep_state = ACPI_STATE_S4; | 410 | int error; |
398 | acpi_sleep_tts_switch(acpi_target_sleep_state); | 411 | |
399 | return 0; | 412 | error = s4_no_nvs ? 0 : hibernate_nvs_alloc(); |
413 | if (!error) { | ||
414 | acpi_target_sleep_state = ACPI_STATE_S4; | ||
415 | acpi_sleep_tts_switch(acpi_target_sleep_state); | ||
416 | } | ||
417 | |||
418 | return error; | ||
419 | } | ||
420 | |||
421 | static int acpi_hibernation_pre_snapshot(void) | ||
422 | { | ||
423 | int error = acpi_pm_prepare(); | ||
424 | |||
425 | if (!error) | ||
426 | hibernate_nvs_save(); | ||
427 | |||
428 | return error; | ||
400 | } | 429 | } |
401 | 430 | ||
402 | static int acpi_hibernation_enter(void) | 431 | static int acpi_hibernation_enter(void) |
@@ -417,6 +446,12 @@ static int acpi_hibernation_enter(void) | |||
417 | return ACPI_SUCCESS(status) ? 0 : -EFAULT; | 446 | return ACPI_SUCCESS(status) ? 0 : -EFAULT; |
418 | } | 447 | } |
419 | 448 | ||
449 | static void acpi_hibernation_finish(void) | ||
450 | { | ||
451 | hibernate_nvs_free(); | ||
452 | acpi_pm_finish(); | ||
453 | } | ||
454 | |||
420 | static void acpi_hibernation_leave(void) | 455 | static void acpi_hibernation_leave(void) |
421 | { | 456 | { |
422 | /* | 457 | /* |
@@ -432,6 +467,8 @@ static void acpi_hibernation_leave(void) | |||
432 | "cannot resume!\n"); | 467 | "cannot resume!\n"); |
433 | panic("ACPI S4 hardware signature mismatch"); | 468 | panic("ACPI S4 hardware signature mismatch"); |
434 | } | 469 | } |
470 | /* Restore the NVS memory area */ | ||
471 | hibernate_nvs_restore(); | ||
435 | } | 472 | } |
436 | 473 | ||
437 | static void acpi_pm_enable_gpes(void) | 474 | static void acpi_pm_enable_gpes(void) |
@@ -442,8 +479,8 @@ static void acpi_pm_enable_gpes(void) | |||
442 | static struct platform_hibernation_ops acpi_hibernation_ops = { | 479 | static struct platform_hibernation_ops acpi_hibernation_ops = { |
443 | .begin = acpi_hibernation_begin, | 480 | .begin = acpi_hibernation_begin, |
444 | .end = acpi_pm_end, | 481 | .end = acpi_pm_end, |
445 | .pre_snapshot = acpi_pm_prepare, | 482 | .pre_snapshot = acpi_hibernation_pre_snapshot, |
446 | .finish = acpi_pm_finish, | 483 | .finish = acpi_hibernation_finish, |
447 | .prepare = acpi_pm_prepare, | 484 | .prepare = acpi_pm_prepare, |
448 | .enter = acpi_hibernation_enter, | 485 | .enter = acpi_hibernation_enter, |
449 | .leave = acpi_hibernation_leave, | 486 | .leave = acpi_hibernation_leave, |
@@ -469,8 +506,22 @@ static int acpi_hibernation_begin_old(void) | |||
469 | 506 | ||
470 | error = acpi_sleep_prepare(ACPI_STATE_S4); | 507 | error = acpi_sleep_prepare(ACPI_STATE_S4); |
471 | 508 | ||
509 | if (!error) { | ||
510 | if (!s4_no_nvs) | ||
511 | error = hibernate_nvs_alloc(); | ||
512 | if (!error) | ||
513 | acpi_target_sleep_state = ACPI_STATE_S4; | ||
514 | } | ||
515 | return error; | ||
516 | } | ||
517 | |||
518 | static int acpi_hibernation_pre_snapshot_old(void) | ||
519 | { | ||
520 | int error = acpi_pm_disable_gpes(); | ||
521 | |||
472 | if (!error) | 522 | if (!error) |
473 | acpi_target_sleep_state = ACPI_STATE_S4; | 523 | hibernate_nvs_save(); |
524 | |||
474 | return error; | 525 | return error; |
475 | } | 526 | } |
476 | 527 | ||
@@ -481,8 +532,8 @@ static int acpi_hibernation_begin_old(void) | |||
481 | static struct platform_hibernation_ops acpi_hibernation_ops_old = { | 532 | static struct platform_hibernation_ops acpi_hibernation_ops_old = { |
482 | .begin = acpi_hibernation_begin_old, | 533 | .begin = acpi_hibernation_begin_old, |
483 | .end = acpi_pm_end, | 534 | .end = acpi_pm_end, |
484 | .pre_snapshot = acpi_pm_disable_gpes, | 535 | .pre_snapshot = acpi_hibernation_pre_snapshot_old, |
485 | .finish = acpi_pm_finish, | 536 | .finish = acpi_hibernation_finish, |
486 | .prepare = acpi_pm_disable_gpes, | 537 | .prepare = acpi_pm_disable_gpes, |
487 | .enter = acpi_hibernation_enter, | 538 | .enter = acpi_hibernation_enter, |
488 | .leave = acpi_hibernation_leave, | 539 | .leave = acpi_hibernation_leave, |
diff --git a/include/linux/acpi.h b/include/linux/acpi.h index fba8051fb297..dfa0a5356c53 100644 --- a/include/linux/acpi.h +++ b/include/linux/acpi.h | |||
@@ -270,6 +270,7 @@ int acpi_check_mem_region(resource_size_t start, resource_size_t n, | |||
270 | #ifdef CONFIG_PM_SLEEP | 270 | #ifdef CONFIG_PM_SLEEP |
271 | void __init acpi_no_s4_hw_signature(void); | 271 | void __init acpi_no_s4_hw_signature(void); |
272 | void __init acpi_old_suspend_ordering(void); | 272 | void __init acpi_old_suspend_ordering(void); |
273 | void __init acpi_s4_no_nvs(void); | ||
273 | #endif /* CONFIG_PM_SLEEP */ | 274 | #endif /* CONFIG_PM_SLEEP */ |
274 | #else /* CONFIG_ACPI */ | 275 | #else /* CONFIG_ACPI */ |
275 | 276 | ||
diff --git a/include/linux/suspend.h b/include/linux/suspend.h index 2ce8207686e2..2b409c44db83 100644 --- a/include/linux/suspend.h +++ b/include/linux/suspend.h | |||
@@ -232,6 +232,11 @@ extern unsigned long get_safe_page(gfp_t gfp_mask); | |||
232 | 232 | ||
233 | extern void hibernation_set_ops(struct platform_hibernation_ops *ops); | 233 | extern void hibernation_set_ops(struct platform_hibernation_ops *ops); |
234 | extern int hibernate(void); | 234 | extern int hibernate(void); |
235 | extern int hibernate_nvs_register(unsigned long start, unsigned long size); | ||
236 | extern int hibernate_nvs_alloc(void); | ||
237 | extern void hibernate_nvs_free(void); | ||
238 | extern void hibernate_nvs_save(void); | ||
239 | extern void hibernate_nvs_restore(void); | ||
235 | #else /* CONFIG_HIBERNATION */ | 240 | #else /* CONFIG_HIBERNATION */ |
236 | static inline int swsusp_page_is_forbidden(struct page *p) { return 0; } | 241 | static inline int swsusp_page_is_forbidden(struct page *p) { return 0; } |
237 | static inline void swsusp_set_page_free(struct page *p) {} | 242 | static inline void swsusp_set_page_free(struct page *p) {} |
@@ -239,6 +244,14 @@ static inline void swsusp_unset_page_free(struct page *p) {} | |||
239 | 244 | ||
240 | static inline void hibernation_set_ops(struct platform_hibernation_ops *ops) {} | 245 | static inline void hibernation_set_ops(struct platform_hibernation_ops *ops) {} |
241 | static inline int hibernate(void) { return -ENOSYS; } | 246 | static inline int hibernate(void) { return -ENOSYS; } |
247 | static inline int hibernate_nvs_register(unsigned long a, unsigned long b) | ||
248 | { | ||
249 | return 0; | ||
250 | } | ||
251 | static inline int hibernate_nvs_alloc(void) { return 0; } | ||
252 | static inline void hibernate_nvs_free(void) {} | ||
253 | static inline void hibernate_nvs_save(void) {} | ||
254 | static inline void hibernate_nvs_restore(void) {} | ||
242 | #endif /* CONFIG_HIBERNATION */ | 255 | #endif /* CONFIG_HIBERNATION */ |
243 | 256 | ||
244 | #ifdef CONFIG_PM_SLEEP | 257 | #ifdef CONFIG_PM_SLEEP |
diff --git a/kernel/power/disk.c b/kernel/power/disk.c index c9d74083746f..096fe4899ea4 100644 --- a/kernel/power/disk.c +++ b/kernel/power/disk.c | |||
@@ -259,12 +259,12 @@ int hibernation_snapshot(int platform_mode) | |||
259 | { | 259 | { |
260 | int error, ftrace_save; | 260 | int error, ftrace_save; |
261 | 261 | ||
262 | /* Free memory before shutting down devices. */ | 262 | error = platform_begin(platform_mode); |
263 | error = swsusp_shrink_memory(); | ||
264 | if (error) | 263 | if (error) |
265 | return error; | 264 | return error; |
266 | 265 | ||
267 | error = platform_begin(platform_mode); | 266 | /* Free memory before shutting down devices. */ |
267 | error = swsusp_shrink_memory(); | ||
268 | if (error) | 268 | if (error) |
269 | goto Close; | 269 | goto Close; |
270 | 270 | ||
diff --git a/kernel/power/snapshot.c b/kernel/power/snapshot.c index 5d2ab836e998..f5fc2d7680f2 100644 --- a/kernel/power/snapshot.c +++ b/kernel/power/snapshot.c | |||
@@ -25,6 +25,7 @@ | |||
25 | #include <linux/syscalls.h> | 25 | #include <linux/syscalls.h> |
26 | #include <linux/console.h> | 26 | #include <linux/console.h> |
27 | #include <linux/highmem.h> | 27 | #include <linux/highmem.h> |
28 | #include <linux/list.h> | ||
28 | 29 | ||
29 | #include <asm/uaccess.h> | 30 | #include <asm/uaccess.h> |
30 | #include <asm/mmu_context.h> | 31 | #include <asm/mmu_context.h> |
@@ -192,12 +193,6 @@ static void *chain_alloc(struct chain_allocator *ca, unsigned int size) | |||
192 | return ret; | 193 | return ret; |
193 | } | 194 | } |
194 | 195 | ||
195 | static void chain_free(struct chain_allocator *ca, int clear_page_nosave) | ||
196 | { | ||
197 | free_list_of_pages(ca->chain, clear_page_nosave); | ||
198 | memset(ca, 0, sizeof(struct chain_allocator)); | ||
199 | } | ||
200 | |||
201 | /** | 196 | /** |
202 | * Data types related to memory bitmaps. | 197 | * Data types related to memory bitmaps. |
203 | * | 198 | * |
@@ -233,7 +228,7 @@ static void chain_free(struct chain_allocator *ca, int clear_page_nosave) | |||
233 | #define BM_BITS_PER_BLOCK (PAGE_SIZE << 3) | 228 | #define BM_BITS_PER_BLOCK (PAGE_SIZE << 3) |
234 | 229 | ||
235 | struct bm_block { | 230 | struct bm_block { |
236 | struct bm_block *next; /* next element of the list */ | 231 | struct list_head hook; /* hook into a list of bitmap blocks */ |
237 | unsigned long start_pfn; /* pfn represented by the first bit */ | 232 | unsigned long start_pfn; /* pfn represented by the first bit */ |
238 | unsigned long end_pfn; /* pfn represented by the last bit plus 1 */ | 233 | unsigned long end_pfn; /* pfn represented by the last bit plus 1 */ |
239 | unsigned long *data; /* bitmap representing pages */ | 234 | unsigned long *data; /* bitmap representing pages */ |
@@ -244,24 +239,15 @@ static inline unsigned long bm_block_bits(struct bm_block *bb) | |||
244 | return bb->end_pfn - bb->start_pfn; | 239 | return bb->end_pfn - bb->start_pfn; |
245 | } | 240 | } |
246 | 241 | ||
247 | struct zone_bitmap { | ||
248 | struct zone_bitmap *next; /* next element of the list */ | ||
249 | unsigned long start_pfn; /* minimal pfn in this zone */ | ||
250 | unsigned long end_pfn; /* maximal pfn in this zone plus 1 */ | ||
251 | struct bm_block *bm_blocks; /* list of bitmap blocks */ | ||
252 | struct bm_block *cur_block; /* recently used bitmap block */ | ||
253 | }; | ||
254 | |||
255 | /* strcut bm_position is used for browsing memory bitmaps */ | 242 | /* strcut bm_position is used for browsing memory bitmaps */ |
256 | 243 | ||
257 | struct bm_position { | 244 | struct bm_position { |
258 | struct zone_bitmap *zone_bm; | ||
259 | struct bm_block *block; | 245 | struct bm_block *block; |
260 | int bit; | 246 | int bit; |
261 | }; | 247 | }; |
262 | 248 | ||
263 | struct memory_bitmap { | 249 | struct memory_bitmap { |
264 | struct zone_bitmap *zone_bm_list; /* list of zone bitmaps */ | 250 | struct list_head blocks; /* list of bitmap blocks */ |
265 | struct linked_page *p_list; /* list of pages used to store zone | 251 | struct linked_page *p_list; /* list of pages used to store zone |
266 | * bitmap objects and bitmap block | 252 | * bitmap objects and bitmap block |
267 | * objects | 253 | * objects |
@@ -273,11 +259,7 @@ struct memory_bitmap { | |||
273 | 259 | ||
274 | static void memory_bm_position_reset(struct memory_bitmap *bm) | 260 | static void memory_bm_position_reset(struct memory_bitmap *bm) |
275 | { | 261 | { |
276 | struct zone_bitmap *zone_bm; | 262 | bm->cur.block = list_entry(bm->blocks.next, struct bm_block, hook); |
277 | |||
278 | zone_bm = bm->zone_bm_list; | ||
279 | bm->cur.zone_bm = zone_bm; | ||
280 | bm->cur.block = zone_bm->bm_blocks; | ||
281 | bm->cur.bit = 0; | 263 | bm->cur.bit = 0; |
282 | } | 264 | } |
283 | 265 | ||
@@ -285,151 +267,184 @@ static void memory_bm_free(struct memory_bitmap *bm, int clear_nosave_free); | |||
285 | 267 | ||
286 | /** | 268 | /** |
287 | * create_bm_block_list - create a list of block bitmap objects | 269 | * create_bm_block_list - create a list of block bitmap objects |
270 | * @nr_blocks - number of blocks to allocate | ||
271 | * @list - list to put the allocated blocks into | ||
272 | * @ca - chain allocator to be used for allocating memory | ||
288 | */ | 273 | */ |
289 | 274 | static int create_bm_block_list(unsigned long pages, | |
290 | static inline struct bm_block * | 275 | struct list_head *list, |
291 | create_bm_block_list(unsigned int nr_blocks, struct chain_allocator *ca) | 276 | struct chain_allocator *ca) |
292 | { | 277 | { |
293 | struct bm_block *bblist = NULL; | 278 | unsigned int nr_blocks = DIV_ROUND_UP(pages, BM_BITS_PER_BLOCK); |
294 | 279 | ||
295 | while (nr_blocks-- > 0) { | 280 | while (nr_blocks-- > 0) { |
296 | struct bm_block *bb; | 281 | struct bm_block *bb; |
297 | 282 | ||
298 | bb = chain_alloc(ca, sizeof(struct bm_block)); | 283 | bb = chain_alloc(ca, sizeof(struct bm_block)); |
299 | if (!bb) | 284 | if (!bb) |
300 | return NULL; | 285 | return -ENOMEM; |
301 | 286 | list_add(&bb->hook, list); | |
302 | bb->next = bblist; | ||
303 | bblist = bb; | ||
304 | } | 287 | } |
305 | return bblist; | 288 | |
289 | return 0; | ||
306 | } | 290 | } |
307 | 291 | ||
292 | struct mem_extent { | ||
293 | struct list_head hook; | ||
294 | unsigned long start; | ||
295 | unsigned long end; | ||
296 | }; | ||
297 | |||
308 | /** | 298 | /** |
309 | * create_zone_bm_list - create a list of zone bitmap objects | 299 | * free_mem_extents - free a list of memory extents |
300 | * @list - list of extents to empty | ||
310 | */ | 301 | */ |
302 | static void free_mem_extents(struct list_head *list) | ||
303 | { | ||
304 | struct mem_extent *ext, *aux; | ||
311 | 305 | ||
312 | static inline struct zone_bitmap * | 306 | list_for_each_entry_safe(ext, aux, list, hook) { |
313 | create_zone_bm_list(unsigned int nr_zones, struct chain_allocator *ca) | 307 | list_del(&ext->hook); |
308 | kfree(ext); | ||
309 | } | ||
310 | } | ||
311 | |||
312 | /** | ||
313 | * create_mem_extents - create a list of memory extents representing | ||
314 | * contiguous ranges of PFNs | ||
315 | * @list - list to put the extents into | ||
316 | * @gfp_mask - mask to use for memory allocations | ||
317 | */ | ||
318 | static int create_mem_extents(struct list_head *list, gfp_t gfp_mask) | ||
314 | { | 319 | { |
315 | struct zone_bitmap *zbmlist = NULL; | 320 | struct zone *zone; |
316 | 321 | ||
317 | while (nr_zones-- > 0) { | 322 | INIT_LIST_HEAD(list); |
318 | struct zone_bitmap *zbm; | ||
319 | 323 | ||
320 | zbm = chain_alloc(ca, sizeof(struct zone_bitmap)); | 324 | for_each_zone(zone) { |
321 | if (!zbm) | 325 | unsigned long zone_start, zone_end; |
322 | return NULL; | 326 | struct mem_extent *ext, *cur, *aux; |
327 | |||
328 | if (!populated_zone(zone)) | ||
329 | continue; | ||
323 | 330 | ||
324 | zbm->next = zbmlist; | 331 | zone_start = zone->zone_start_pfn; |
325 | zbmlist = zbm; | 332 | zone_end = zone->zone_start_pfn + zone->spanned_pages; |
333 | |||
334 | list_for_each_entry(ext, list, hook) | ||
335 | if (zone_start <= ext->end) | ||
336 | break; | ||
337 | |||
338 | if (&ext->hook == list || zone_end < ext->start) { | ||
339 | /* New extent is necessary */ | ||
340 | struct mem_extent *new_ext; | ||
341 | |||
342 | new_ext = kzalloc(sizeof(struct mem_extent), gfp_mask); | ||
343 | if (!new_ext) { | ||
344 | free_mem_extents(list); | ||
345 | return -ENOMEM; | ||
346 | } | ||
347 | new_ext->start = zone_start; | ||
348 | new_ext->end = zone_end; | ||
349 | list_add_tail(&new_ext->hook, &ext->hook); | ||
350 | continue; | ||
351 | } | ||
352 | |||
353 | /* Merge this zone's range of PFNs with the existing one */ | ||
354 | if (zone_start < ext->start) | ||
355 | ext->start = zone_start; | ||
356 | if (zone_end > ext->end) | ||
357 | ext->end = zone_end; | ||
358 | |||
359 | /* More merging may be possible */ | ||
360 | cur = ext; | ||
361 | list_for_each_entry_safe_continue(cur, aux, list, hook) { | ||
362 | if (zone_end < cur->start) | ||
363 | break; | ||
364 | if (zone_end < cur->end) | ||
365 | ext->end = cur->end; | ||
366 | list_del(&cur->hook); | ||
367 | kfree(cur); | ||
368 | } | ||
326 | } | 369 | } |
327 | return zbmlist; | 370 | |
371 | return 0; | ||
328 | } | 372 | } |
329 | 373 | ||
330 | /** | 374 | /** |
331 | * memory_bm_create - allocate memory for a memory bitmap | 375 | * memory_bm_create - allocate memory for a memory bitmap |
332 | */ | 376 | */ |
333 | |||
334 | static int | 377 | static int |
335 | memory_bm_create(struct memory_bitmap *bm, gfp_t gfp_mask, int safe_needed) | 378 | memory_bm_create(struct memory_bitmap *bm, gfp_t gfp_mask, int safe_needed) |
336 | { | 379 | { |
337 | struct chain_allocator ca; | 380 | struct chain_allocator ca; |
338 | struct zone *zone; | 381 | struct list_head mem_extents; |
339 | struct zone_bitmap *zone_bm; | 382 | struct mem_extent *ext; |
340 | struct bm_block *bb; | 383 | int error; |
341 | unsigned int nr; | ||
342 | 384 | ||
343 | chain_init(&ca, gfp_mask, safe_needed); | 385 | chain_init(&ca, gfp_mask, safe_needed); |
386 | INIT_LIST_HEAD(&bm->blocks); | ||
344 | 387 | ||
345 | /* Compute the number of zones */ | 388 | error = create_mem_extents(&mem_extents, gfp_mask); |
346 | nr = 0; | 389 | if (error) |
347 | for_each_zone(zone) | 390 | return error; |
348 | if (populated_zone(zone)) | ||
349 | nr++; | ||
350 | |||
351 | /* Allocate the list of zones bitmap objects */ | ||
352 | zone_bm = create_zone_bm_list(nr, &ca); | ||
353 | bm->zone_bm_list = zone_bm; | ||
354 | if (!zone_bm) { | ||
355 | chain_free(&ca, PG_UNSAFE_CLEAR); | ||
356 | return -ENOMEM; | ||
357 | } | ||
358 | |||
359 | /* Initialize the zone bitmap objects */ | ||
360 | for_each_zone(zone) { | ||
361 | unsigned long pfn; | ||
362 | 391 | ||
363 | if (!populated_zone(zone)) | 392 | list_for_each_entry(ext, &mem_extents, hook) { |
364 | continue; | 393 | struct bm_block *bb; |
394 | unsigned long pfn = ext->start; | ||
395 | unsigned long pages = ext->end - ext->start; | ||
365 | 396 | ||
366 | zone_bm->start_pfn = zone->zone_start_pfn; | 397 | bb = list_entry(bm->blocks.prev, struct bm_block, hook); |
367 | zone_bm->end_pfn = zone->zone_start_pfn + zone->spanned_pages; | ||
368 | /* Allocate the list of bitmap block objects */ | ||
369 | nr = DIV_ROUND_UP(zone->spanned_pages, BM_BITS_PER_BLOCK); | ||
370 | bb = create_bm_block_list(nr, &ca); | ||
371 | zone_bm->bm_blocks = bb; | ||
372 | zone_bm->cur_block = bb; | ||
373 | if (!bb) | ||
374 | goto Free; | ||
375 | 398 | ||
376 | nr = zone->spanned_pages; | 399 | error = create_bm_block_list(pages, bm->blocks.prev, &ca); |
377 | pfn = zone->zone_start_pfn; | 400 | if (error) |
378 | /* Initialize the bitmap block objects */ | 401 | goto Error; |
379 | while (bb) { | ||
380 | unsigned long *ptr; | ||
381 | 402 | ||
382 | ptr = get_image_page(gfp_mask, safe_needed); | 403 | list_for_each_entry_continue(bb, &bm->blocks, hook) { |
383 | bb->data = ptr; | 404 | bb->data = get_image_page(gfp_mask, safe_needed); |
384 | if (!ptr) | 405 | if (!bb->data) { |
385 | goto Free; | 406 | error = -ENOMEM; |
407 | goto Error; | ||
408 | } | ||
386 | 409 | ||
387 | bb->start_pfn = pfn; | 410 | bb->start_pfn = pfn; |
388 | if (nr >= BM_BITS_PER_BLOCK) { | 411 | if (pages >= BM_BITS_PER_BLOCK) { |
389 | pfn += BM_BITS_PER_BLOCK; | 412 | pfn += BM_BITS_PER_BLOCK; |
390 | nr -= BM_BITS_PER_BLOCK; | 413 | pages -= BM_BITS_PER_BLOCK; |
391 | } else { | 414 | } else { |
392 | /* This is executed only once in the loop */ | 415 | /* This is executed only once in the loop */ |
393 | pfn += nr; | 416 | pfn += pages; |
394 | } | 417 | } |
395 | bb->end_pfn = pfn; | 418 | bb->end_pfn = pfn; |
396 | bb = bb->next; | ||
397 | } | 419 | } |
398 | zone_bm = zone_bm->next; | ||
399 | } | 420 | } |
421 | |||
400 | bm->p_list = ca.chain; | 422 | bm->p_list = ca.chain; |
401 | memory_bm_position_reset(bm); | 423 | memory_bm_position_reset(bm); |
402 | return 0; | 424 | Exit: |
425 | free_mem_extents(&mem_extents); | ||
426 | return error; | ||
403 | 427 | ||
404 | Free: | 428 | Error: |
405 | bm->p_list = ca.chain; | 429 | bm->p_list = ca.chain; |
406 | memory_bm_free(bm, PG_UNSAFE_CLEAR); | 430 | memory_bm_free(bm, PG_UNSAFE_CLEAR); |
407 | return -ENOMEM; | 431 | goto Exit; |
408 | } | 432 | } |
409 | 433 | ||
410 | /** | 434 | /** |
411 | * memory_bm_free - free memory occupied by the memory bitmap @bm | 435 | * memory_bm_free - free memory occupied by the memory bitmap @bm |
412 | */ | 436 | */ |
413 | |||
414 | static void memory_bm_free(struct memory_bitmap *bm, int clear_nosave_free) | 437 | static void memory_bm_free(struct memory_bitmap *bm, int clear_nosave_free) |
415 | { | 438 | { |
416 | struct zone_bitmap *zone_bm; | 439 | struct bm_block *bb; |
417 | 440 | ||
418 | /* Free the list of bit blocks for each zone_bitmap object */ | 441 | list_for_each_entry(bb, &bm->blocks, hook) |
419 | zone_bm = bm->zone_bm_list; | 442 | if (bb->data) |
420 | while (zone_bm) { | 443 | free_image_page(bb->data, clear_nosave_free); |
421 | struct bm_block *bb; | ||
422 | 444 | ||
423 | bb = zone_bm->bm_blocks; | ||
424 | while (bb) { | ||
425 | if (bb->data) | ||
426 | free_image_page(bb->data, clear_nosave_free); | ||
427 | bb = bb->next; | ||
428 | } | ||
429 | zone_bm = zone_bm->next; | ||
430 | } | ||
431 | free_list_of_pages(bm->p_list, clear_nosave_free); | 445 | free_list_of_pages(bm->p_list, clear_nosave_free); |
432 | bm->zone_bm_list = NULL; | 446 | |
447 | INIT_LIST_HEAD(&bm->blocks); | ||
433 | } | 448 | } |
434 | 449 | ||
435 | /** | 450 | /** |
@@ -437,38 +452,33 @@ static void memory_bm_free(struct memory_bitmap *bm, int clear_nosave_free) | |||
437 | * to given pfn. The cur_zone_bm member of @bm and the cur_block member | 452 | * to given pfn. The cur_zone_bm member of @bm and the cur_block member |
438 | * of @bm->cur_zone_bm are updated. | 453 | * of @bm->cur_zone_bm are updated. |
439 | */ | 454 | */ |
440 | |||
441 | static int memory_bm_find_bit(struct memory_bitmap *bm, unsigned long pfn, | 455 | static int memory_bm_find_bit(struct memory_bitmap *bm, unsigned long pfn, |
442 | void **addr, unsigned int *bit_nr) | 456 | void **addr, unsigned int *bit_nr) |
443 | { | 457 | { |
444 | struct zone_bitmap *zone_bm; | ||
445 | struct bm_block *bb; | 458 | struct bm_block *bb; |
446 | 459 | ||
447 | /* Check if the pfn is from the current zone */ | 460 | /* |
448 | zone_bm = bm->cur.zone_bm; | 461 | * Check if the pfn corresponds to the current bitmap block and find |
449 | if (pfn < zone_bm->start_pfn || pfn >= zone_bm->end_pfn) { | 462 | * the block where it fits if this is not the case. |
450 | zone_bm = bm->zone_bm_list; | 463 | */ |
451 | /* We don't assume that the zones are sorted by pfns */ | 464 | bb = bm->cur.block; |
452 | while (pfn < zone_bm->start_pfn || pfn >= zone_bm->end_pfn) { | ||
453 | zone_bm = zone_bm->next; | ||
454 | |||
455 | if (!zone_bm) | ||
456 | return -EFAULT; | ||
457 | } | ||
458 | bm->cur.zone_bm = zone_bm; | ||
459 | } | ||
460 | /* Check if the pfn corresponds to the current bitmap block */ | ||
461 | bb = zone_bm->cur_block; | ||
462 | if (pfn < bb->start_pfn) | 465 | if (pfn < bb->start_pfn) |
463 | bb = zone_bm->bm_blocks; | 466 | list_for_each_entry_continue_reverse(bb, &bm->blocks, hook) |
467 | if (pfn >= bb->start_pfn) | ||
468 | break; | ||
464 | 469 | ||
465 | while (pfn >= bb->end_pfn) { | 470 | if (pfn >= bb->end_pfn) |
466 | bb = bb->next; | 471 | list_for_each_entry_continue(bb, &bm->blocks, hook) |
472 | if (pfn >= bb->start_pfn && pfn < bb->end_pfn) | ||
473 | break; | ||
467 | 474 | ||
468 | BUG_ON(!bb); | 475 | if (&bb->hook == &bm->blocks) |
469 | } | 476 | return -EFAULT; |
470 | zone_bm->cur_block = bb; | 477 | |
478 | /* The block has been found */ | ||
479 | bm->cur.block = bb; | ||
471 | pfn -= bb->start_pfn; | 480 | pfn -= bb->start_pfn; |
481 | bm->cur.bit = pfn + 1; | ||
472 | *bit_nr = pfn; | 482 | *bit_nr = pfn; |
473 | *addr = bb->data; | 483 | *addr = bb->data; |
474 | return 0; | 484 | return 0; |
@@ -519,6 +529,14 @@ static int memory_bm_test_bit(struct memory_bitmap *bm, unsigned long pfn) | |||
519 | return test_bit(bit, addr); | 529 | return test_bit(bit, addr); |
520 | } | 530 | } |
521 | 531 | ||
532 | static bool memory_bm_pfn_present(struct memory_bitmap *bm, unsigned long pfn) | ||
533 | { | ||
534 | void *addr; | ||
535 | unsigned int bit; | ||
536 | |||
537 | return !memory_bm_find_bit(bm, pfn, &addr, &bit); | ||
538 | } | ||
539 | |||
522 | /** | 540 | /** |
523 | * memory_bm_next_pfn - find the pfn that corresponds to the next set bit | 541 | * memory_bm_next_pfn - find the pfn that corresponds to the next set bit |
524 | * in the bitmap @bm. If the pfn cannot be found, BM_END_OF_MAP is | 542 | * in the bitmap @bm. If the pfn cannot be found, BM_END_OF_MAP is |
@@ -530,29 +548,21 @@ static int memory_bm_test_bit(struct memory_bitmap *bm, unsigned long pfn) | |||
530 | 548 | ||
531 | static unsigned long memory_bm_next_pfn(struct memory_bitmap *bm) | 549 | static unsigned long memory_bm_next_pfn(struct memory_bitmap *bm) |
532 | { | 550 | { |
533 | struct zone_bitmap *zone_bm; | ||
534 | struct bm_block *bb; | 551 | struct bm_block *bb; |
535 | int bit; | 552 | int bit; |
536 | 553 | ||
554 | bb = bm->cur.block; | ||
537 | do { | 555 | do { |
538 | bb = bm->cur.block; | 556 | bit = bm->cur.bit; |
539 | do { | 557 | bit = find_next_bit(bb->data, bm_block_bits(bb), bit); |
540 | bit = bm->cur.bit; | 558 | if (bit < bm_block_bits(bb)) |
541 | bit = find_next_bit(bb->data, bm_block_bits(bb), bit); | 559 | goto Return_pfn; |
542 | if (bit < bm_block_bits(bb)) | 560 | |
543 | goto Return_pfn; | 561 | bb = list_entry(bb->hook.next, struct bm_block, hook); |
544 | 562 | bm->cur.block = bb; | |
545 | bb = bb->next; | 563 | bm->cur.bit = 0; |
546 | bm->cur.block = bb; | 564 | } while (&bb->hook != &bm->blocks); |
547 | bm->cur.bit = 0; | 565 | |
548 | } while (bb); | ||
549 | zone_bm = bm->cur.zone_bm->next; | ||
550 | if (zone_bm) { | ||
551 | bm->cur.zone_bm = zone_bm; | ||
552 | bm->cur.block = zone_bm->bm_blocks; | ||
553 | bm->cur.bit = 0; | ||
554 | } | ||
555 | } while (zone_bm); | ||
556 | memory_bm_position_reset(bm); | 566 | memory_bm_position_reset(bm); |
557 | return BM_END_OF_MAP; | 567 | return BM_END_OF_MAP; |
558 | 568 | ||
@@ -808,8 +818,7 @@ static unsigned int count_free_highmem_pages(void) | |||
808 | * We should save the page if it isn't Nosave or NosaveFree, or Reserved, | 818 | * We should save the page if it isn't Nosave or NosaveFree, or Reserved, |
809 | * and it isn't a part of a free chunk of pages. | 819 | * and it isn't a part of a free chunk of pages. |
810 | */ | 820 | */ |
811 | 821 | static struct page *saveable_highmem_page(struct zone *zone, unsigned long pfn) | |
812 | static struct page *saveable_highmem_page(unsigned long pfn) | ||
813 | { | 822 | { |
814 | struct page *page; | 823 | struct page *page; |
815 | 824 | ||
@@ -817,6 +826,8 @@ static struct page *saveable_highmem_page(unsigned long pfn) | |||
817 | return NULL; | 826 | return NULL; |
818 | 827 | ||
819 | page = pfn_to_page(pfn); | 828 | page = pfn_to_page(pfn); |
829 | if (page_zone(page) != zone) | ||
830 | return NULL; | ||
820 | 831 | ||
821 | BUG_ON(!PageHighMem(page)); | 832 | BUG_ON(!PageHighMem(page)); |
822 | 833 | ||
@@ -846,13 +857,16 @@ unsigned int count_highmem_pages(void) | |||
846 | mark_free_pages(zone); | 857 | mark_free_pages(zone); |
847 | max_zone_pfn = zone->zone_start_pfn + zone->spanned_pages; | 858 | max_zone_pfn = zone->zone_start_pfn + zone->spanned_pages; |
848 | for (pfn = zone->zone_start_pfn; pfn < max_zone_pfn; pfn++) | 859 | for (pfn = zone->zone_start_pfn; pfn < max_zone_pfn; pfn++) |
849 | if (saveable_highmem_page(pfn)) | 860 | if (saveable_highmem_page(zone, pfn)) |
850 | n++; | 861 | n++; |
851 | } | 862 | } |
852 | return n; | 863 | return n; |
853 | } | 864 | } |
854 | #else | 865 | #else |
855 | static inline void *saveable_highmem_page(unsigned long pfn) { return NULL; } | 866 | static inline void *saveable_highmem_page(struct zone *z, unsigned long p) |
867 | { | ||
868 | return NULL; | ||
869 | } | ||
856 | #endif /* CONFIG_HIGHMEM */ | 870 | #endif /* CONFIG_HIGHMEM */ |
857 | 871 | ||
858 | /** | 872 | /** |
@@ -863,8 +877,7 @@ static inline void *saveable_highmem_page(unsigned long pfn) { return NULL; } | |||
863 | * of pages statically defined as 'unsaveable', and it isn't a part of | 877 | * of pages statically defined as 'unsaveable', and it isn't a part of |
864 | * a free chunk of pages. | 878 | * a free chunk of pages. |
865 | */ | 879 | */ |
866 | 880 | static struct page *saveable_page(struct zone *zone, unsigned long pfn) | |
867 | static struct page *saveable_page(unsigned long pfn) | ||
868 | { | 881 | { |
869 | struct page *page; | 882 | struct page *page; |
870 | 883 | ||
@@ -872,6 +885,8 @@ static struct page *saveable_page(unsigned long pfn) | |||
872 | return NULL; | 885 | return NULL; |
873 | 886 | ||
874 | page = pfn_to_page(pfn); | 887 | page = pfn_to_page(pfn); |
888 | if (page_zone(page) != zone) | ||
889 | return NULL; | ||
875 | 890 | ||
876 | BUG_ON(PageHighMem(page)); | 891 | BUG_ON(PageHighMem(page)); |
877 | 892 | ||
@@ -903,7 +918,7 @@ unsigned int count_data_pages(void) | |||
903 | mark_free_pages(zone); | 918 | mark_free_pages(zone); |
904 | max_zone_pfn = zone->zone_start_pfn + zone->spanned_pages; | 919 | max_zone_pfn = zone->zone_start_pfn + zone->spanned_pages; |
905 | for (pfn = zone->zone_start_pfn; pfn < max_zone_pfn; pfn++) | 920 | for (pfn = zone->zone_start_pfn; pfn < max_zone_pfn; pfn++) |
906 | if(saveable_page(pfn)) | 921 | if (saveable_page(zone, pfn)) |
907 | n++; | 922 | n++; |
908 | } | 923 | } |
909 | return n; | 924 | return n; |
@@ -944,7 +959,7 @@ static inline struct page * | |||
944 | page_is_saveable(struct zone *zone, unsigned long pfn) | 959 | page_is_saveable(struct zone *zone, unsigned long pfn) |
945 | { | 960 | { |
946 | return is_highmem(zone) ? | 961 | return is_highmem(zone) ? |
947 | saveable_highmem_page(pfn) : saveable_page(pfn); | 962 | saveable_highmem_page(zone, pfn) : saveable_page(zone, pfn); |
948 | } | 963 | } |
949 | 964 | ||
950 | static void copy_data_page(unsigned long dst_pfn, unsigned long src_pfn) | 965 | static void copy_data_page(unsigned long dst_pfn, unsigned long src_pfn) |
@@ -966,7 +981,7 @@ static void copy_data_page(unsigned long dst_pfn, unsigned long src_pfn) | |||
966 | * data modified by kmap_atomic() | 981 | * data modified by kmap_atomic() |
967 | */ | 982 | */ |
968 | safe_copy_page(buffer, s_page); | 983 | safe_copy_page(buffer, s_page); |
969 | dst = kmap_atomic(pfn_to_page(dst_pfn), KM_USER0); | 984 | dst = kmap_atomic(d_page, KM_USER0); |
970 | memcpy(dst, buffer, PAGE_SIZE); | 985 | memcpy(dst, buffer, PAGE_SIZE); |
971 | kunmap_atomic(dst, KM_USER0); | 986 | kunmap_atomic(dst, KM_USER0); |
972 | } else { | 987 | } else { |
@@ -975,7 +990,7 @@ static void copy_data_page(unsigned long dst_pfn, unsigned long src_pfn) | |||
975 | } | 990 | } |
976 | } | 991 | } |
977 | #else | 992 | #else |
978 | #define page_is_saveable(zone, pfn) saveable_page(pfn) | 993 | #define page_is_saveable(zone, pfn) saveable_page(zone, pfn) |
979 | 994 | ||
980 | static inline void copy_data_page(unsigned long dst_pfn, unsigned long src_pfn) | 995 | static inline void copy_data_page(unsigned long dst_pfn, unsigned long src_pfn) |
981 | { | 996 | { |
@@ -1459,9 +1474,7 @@ load_header(struct swsusp_info *info) | |||
1459 | * unpack_orig_pfns - for each element of @buf[] (1 page at a time) set | 1474 | * unpack_orig_pfns - for each element of @buf[] (1 page at a time) set |
1460 | * the corresponding bit in the memory bitmap @bm | 1475 | * the corresponding bit in the memory bitmap @bm |
1461 | */ | 1476 | */ |
1462 | 1477 | static int unpack_orig_pfns(unsigned long *buf, struct memory_bitmap *bm) | |
1463 | static inline void | ||
1464 | unpack_orig_pfns(unsigned long *buf, struct memory_bitmap *bm) | ||
1465 | { | 1478 | { |
1466 | int j; | 1479 | int j; |
1467 | 1480 | ||
@@ -1469,8 +1482,13 @@ unpack_orig_pfns(unsigned long *buf, struct memory_bitmap *bm) | |||
1469 | if (unlikely(buf[j] == BM_END_OF_MAP)) | 1482 | if (unlikely(buf[j] == BM_END_OF_MAP)) |
1470 | break; | 1483 | break; |
1471 | 1484 | ||
1472 | memory_bm_set_bit(bm, buf[j]); | 1485 | if (memory_bm_pfn_present(bm, buf[j])) |
1486 | memory_bm_set_bit(bm, buf[j]); | ||
1487 | else | ||
1488 | return -EFAULT; | ||
1473 | } | 1489 | } |
1490 | |||
1491 | return 0; | ||
1474 | } | 1492 | } |
1475 | 1493 | ||
1476 | /* List of "safe" pages that may be used to store data loaded from the suspend | 1494 | /* List of "safe" pages that may be used to store data loaded from the suspend |
@@ -1608,7 +1626,7 @@ get_highmem_page_buffer(struct page *page, struct chain_allocator *ca) | |||
1608 | pbe = chain_alloc(ca, sizeof(struct highmem_pbe)); | 1626 | pbe = chain_alloc(ca, sizeof(struct highmem_pbe)); |
1609 | if (!pbe) { | 1627 | if (!pbe) { |
1610 | swsusp_free(); | 1628 | swsusp_free(); |
1611 | return NULL; | 1629 | return ERR_PTR(-ENOMEM); |
1612 | } | 1630 | } |
1613 | pbe->orig_page = page; | 1631 | pbe->orig_page = page; |
1614 | if (safe_highmem_pages > 0) { | 1632 | if (safe_highmem_pages > 0) { |
@@ -1677,7 +1695,7 @@ prepare_highmem_image(struct memory_bitmap *bm, unsigned int *nr_highmem_p) | |||
1677 | static inline void * | 1695 | static inline void * |
1678 | get_highmem_page_buffer(struct page *page, struct chain_allocator *ca) | 1696 | get_highmem_page_buffer(struct page *page, struct chain_allocator *ca) |
1679 | { | 1697 | { |
1680 | return NULL; | 1698 | return ERR_PTR(-EINVAL); |
1681 | } | 1699 | } |
1682 | 1700 | ||
1683 | static inline void copy_last_highmem_page(void) {} | 1701 | static inline void copy_last_highmem_page(void) {} |
@@ -1788,8 +1806,13 @@ prepare_image(struct memory_bitmap *new_bm, struct memory_bitmap *bm) | |||
1788 | static void *get_buffer(struct memory_bitmap *bm, struct chain_allocator *ca) | 1806 | static void *get_buffer(struct memory_bitmap *bm, struct chain_allocator *ca) |
1789 | { | 1807 | { |
1790 | struct pbe *pbe; | 1808 | struct pbe *pbe; |
1791 | struct page *page = pfn_to_page(memory_bm_next_pfn(bm)); | 1809 | struct page *page; |
1810 | unsigned long pfn = memory_bm_next_pfn(bm); | ||
1792 | 1811 | ||
1812 | if (pfn == BM_END_OF_MAP) | ||
1813 | return ERR_PTR(-EFAULT); | ||
1814 | |||
1815 | page = pfn_to_page(pfn); | ||
1793 | if (PageHighMem(page)) | 1816 | if (PageHighMem(page)) |
1794 | return get_highmem_page_buffer(page, ca); | 1817 | return get_highmem_page_buffer(page, ca); |
1795 | 1818 | ||
@@ -1805,7 +1828,7 @@ static void *get_buffer(struct memory_bitmap *bm, struct chain_allocator *ca) | |||
1805 | pbe = chain_alloc(ca, sizeof(struct pbe)); | 1828 | pbe = chain_alloc(ca, sizeof(struct pbe)); |
1806 | if (!pbe) { | 1829 | if (!pbe) { |
1807 | swsusp_free(); | 1830 | swsusp_free(); |
1808 | return NULL; | 1831 | return ERR_PTR(-ENOMEM); |
1809 | } | 1832 | } |
1810 | pbe->orig_address = page_address(page); | 1833 | pbe->orig_address = page_address(page); |
1811 | pbe->address = safe_pages_list; | 1834 | pbe->address = safe_pages_list; |
@@ -1868,7 +1891,10 @@ int snapshot_write_next(struct snapshot_handle *handle, size_t count) | |||
1868 | return error; | 1891 | return error; |
1869 | 1892 | ||
1870 | } else if (handle->prev <= nr_meta_pages) { | 1893 | } else if (handle->prev <= nr_meta_pages) { |
1871 | unpack_orig_pfns(buffer, ©_bm); | 1894 | error = unpack_orig_pfns(buffer, ©_bm); |
1895 | if (error) | ||
1896 | return error; | ||
1897 | |||
1872 | if (handle->prev == nr_meta_pages) { | 1898 | if (handle->prev == nr_meta_pages) { |
1873 | error = prepare_image(&orig_bm, ©_bm); | 1899 | error = prepare_image(&orig_bm, ©_bm); |
1874 | if (error) | 1900 | if (error) |
@@ -1879,12 +1905,14 @@ int snapshot_write_next(struct snapshot_handle *handle, size_t count) | |||
1879 | restore_pblist = NULL; | 1905 | restore_pblist = NULL; |
1880 | handle->buffer = get_buffer(&orig_bm, &ca); | 1906 | handle->buffer = get_buffer(&orig_bm, &ca); |
1881 | handle->sync_read = 0; | 1907 | handle->sync_read = 0; |
1882 | if (!handle->buffer) | 1908 | if (IS_ERR(handle->buffer)) |
1883 | return -ENOMEM; | 1909 | return PTR_ERR(handle->buffer); |
1884 | } | 1910 | } |
1885 | } else { | 1911 | } else { |
1886 | copy_last_highmem_page(); | 1912 | copy_last_highmem_page(); |
1887 | handle->buffer = get_buffer(&orig_bm, &ca); | 1913 | handle->buffer = get_buffer(&orig_bm, &ca); |
1914 | if (IS_ERR(handle->buffer)) | ||
1915 | return PTR_ERR(handle->buffer); | ||
1888 | if (handle->buffer != buffer) | 1916 | if (handle->buffer != buffer) |
1889 | handle->sync_read = 0; | 1917 | handle->sync_read = 0; |
1890 | } | 1918 | } |
diff --git a/kernel/power/swsusp.c b/kernel/power/swsusp.c index 023ff2a31d89..a92c91451559 100644 --- a/kernel/power/swsusp.c +++ b/kernel/power/swsusp.c | |||
@@ -262,3 +262,125 @@ int swsusp_shrink_memory(void) | |||
262 | 262 | ||
263 | return 0; | 263 | return 0; |
264 | } | 264 | } |
265 | |||
266 | /* | ||
267 | * Platforms, like ACPI, may want us to save some memory used by them during | ||
268 | * hibernation and to restore the contents of this memory during the subsequent | ||
269 | * resume. The code below implements a mechanism allowing us to do that. | ||
270 | */ | ||
271 | |||
272 | struct nvs_page { | ||
273 | unsigned long phys_start; | ||
274 | unsigned int size; | ||
275 | void *kaddr; | ||
276 | void *data; | ||
277 | struct list_head node; | ||
278 | }; | ||
279 | |||
280 | static LIST_HEAD(nvs_list); | ||
281 | |||
282 | /** | ||
283 | * hibernate_nvs_register - register platform NVS memory region to save | ||
284 | * @start - physical address of the region | ||
285 | * @size - size of the region | ||
286 | * | ||
287 | * The NVS region need not be page-aligned (both ends) and we arrange | ||
288 | * things so that the data from page-aligned addresses in this region will | ||
289 | * be copied into separate RAM pages. | ||
290 | */ | ||
291 | int hibernate_nvs_register(unsigned long start, unsigned long size) | ||
292 | { | ||
293 | struct nvs_page *entry, *next; | ||
294 | |||
295 | while (size > 0) { | ||
296 | unsigned int nr_bytes; | ||
297 | |||
298 | entry = kzalloc(sizeof(struct nvs_page), GFP_KERNEL); | ||
299 | if (!entry) | ||
300 | goto Error; | ||
301 | |||
302 | list_add_tail(&entry->node, &nvs_list); | ||
303 | entry->phys_start = start; | ||
304 | nr_bytes = PAGE_SIZE - (start & ~PAGE_MASK); | ||
305 | entry->size = (size < nr_bytes) ? size : nr_bytes; | ||
306 | |||
307 | start += entry->size; | ||
308 | size -= entry->size; | ||
309 | } | ||
310 | return 0; | ||
311 | |||
312 | Error: | ||
313 | list_for_each_entry_safe(entry, next, &nvs_list, node) { | ||
314 | list_del(&entry->node); | ||
315 | kfree(entry); | ||
316 | } | ||
317 | return -ENOMEM; | ||
318 | } | ||
319 | |||
320 | /** | ||
321 | * hibernate_nvs_free - free data pages allocated for saving NVS regions | ||
322 | */ | ||
323 | void hibernate_nvs_free(void) | ||
324 | { | ||
325 | struct nvs_page *entry; | ||
326 | |||
327 | list_for_each_entry(entry, &nvs_list, node) | ||
328 | if (entry->data) { | ||
329 | free_page((unsigned long)entry->data); | ||
330 | entry->data = NULL; | ||
331 | if (entry->kaddr) { | ||
332 | iounmap(entry->kaddr); | ||
333 | entry->kaddr = NULL; | ||
334 | } | ||
335 | } | ||
336 | } | ||
337 | |||
338 | /** | ||
339 | * hibernate_nvs_alloc - allocate memory necessary for saving NVS regions | ||
340 | */ | ||
341 | int hibernate_nvs_alloc(void) | ||
342 | { | ||
343 | struct nvs_page *entry; | ||
344 | |||
345 | list_for_each_entry(entry, &nvs_list, node) { | ||
346 | entry->data = (void *)__get_free_page(GFP_KERNEL); | ||
347 | if (!entry->data) { | ||
348 | hibernate_nvs_free(); | ||
349 | return -ENOMEM; | ||
350 | } | ||
351 | } | ||
352 | return 0; | ||
353 | } | ||
354 | |||
355 | /** | ||
356 | * hibernate_nvs_save - save NVS memory regions | ||
357 | */ | ||
358 | void hibernate_nvs_save(void) | ||
359 | { | ||
360 | struct nvs_page *entry; | ||
361 | |||
362 | printk(KERN_INFO "PM: Saving platform NVS memory\n"); | ||
363 | |||
364 | list_for_each_entry(entry, &nvs_list, node) | ||
365 | if (entry->data) { | ||
366 | entry->kaddr = ioremap(entry->phys_start, entry->size); | ||
367 | memcpy(entry->data, entry->kaddr, entry->size); | ||
368 | } | ||
369 | } | ||
370 | |||
371 | /** | ||
372 | * hibernate_nvs_restore - restore NVS memory regions | ||
373 | * | ||
374 | * This function is going to be called with interrupts disabled, so it | ||
375 | * cannot iounmap the virtual addresses used to access the NVS region. | ||
376 | */ | ||
377 | void hibernate_nvs_restore(void) | ||
378 | { | ||
379 | struct nvs_page *entry; | ||
380 | |||
381 | printk(KERN_INFO "PM: Restoring platform NVS memory\n"); | ||
382 | |||
383 | list_for_each_entry(entry, &nvs_list, node) | ||
384 | if (entry->data) | ||
385 | memcpy(entry->kaddr, entry->data, entry->size); | ||
386 | } | ||