PM: Fixes related to suspend and hibernation for 2.6.29
Subsystems (8)
cover PM: Fixes related to suspend and hibernation for 2.6.29
Hi Len,
The following are fixes related to suspend and hibernation I have collected over the last couple of weeks. Some of them are in -mm. I think they all are 2.6.29 material.
1/9 and 2/9 fix hibernation regression caused by the async boot changes, I hope Greg will not object if they go through your tree.
3/9 fixes a long-standing hibernation issue that cause the resume partition to become unuseable after two consecutive failing attempts to hibernate.
4/9 fixes build of kprobes.c with CONFIG_PM unset.
5/9 and 6/9 clean up memory freeing code used during hibernation.
7/9 fixes a bug related to the use of PM notifiers during hibernation with the help of the user land interface.
8/9 and 9/9 fix the handling of consoles during suspend/hibernation.
Please add these patches to the suspend branch and push to Linus for 2.6.29.
Thanks,
Rafael
Signature · 3 lines
Everyone knows that debugging is twice as hard as writing a program in the first place. So if you're as clever as you can be when you write it, how will you ever debug it? --- Brian Kernighan
PATCH 1/9 Consolidate driver_probe_done() loops into one place
From: Arjan van de Ven <[email protected]>
there's a few places that currently loop over driver_probe_done(), and I'm about to add another one. This patch abstracts it into a helper to reduce duplication.
--- drivers/base/dd.c | 17 +++++++++++++++++ include/linux/device.h | 2 ++ init/do_mounts.c | 13 +++++++++---- init/do_mounts_md.c | 5 +++-- 4 files changed, 31 insertions(+), 6 deletions(-)
drivers/base/dd.c +17 -0
@@ -18,9 +18,11 @@*/ #include <linux/device.h> +#include <linux/delay.h> #include <linux/module.h> #include <linux/kthread.h> #include <linux/wait.h> +#include <linux/async.h> #include "base.h" #include "power/power.h"
@@ -168,6 +170,21 @@ int driver_probe_done(void) }/** +* wait_for_device_probe +* Wait for device probing to be completed. +* +* Note: this function polls at 100 msec intervals. +*/ +int wait_for_device_probe (void ) +{ +/* wait for the known devices to complete their probing */ +while (driver_probe_done () !=0 ) +msleep (100 ); +async_synchronize_full (); +return 0 ; +} + +/** * driver_probe_device - attempt to bind device & driver together * @drv: driver to bind a device to * @dev: device to try to bind to the driver
include/linux/device.h +2 -0
@@ -147,6 +147,8 @@ extern void put_driver(struct device_driextern struct device_driver *driver_find (const char *name,struct bus_type *bus);extern int driver_probe_done (void ); +extern int wait_for_device_probe (void ); +/* sysfs interface for exporting driver attributes */
init/do_mounts.c +9 -4
@@ -370,10 +370,14 @@ void __init prepare_namespace(void)ssleep (root_delay); } -/* wait for the known devices to complete their probing */ -while (driver_probe_done () !=0 ) -msleep (100 ); -async_synchronize_full (); +/* +* wait for the known devices to complete their probing +* +* Note: this is a potential source of long boot delays. +* For example, it is not atypical to wait 5 seconds here +* for the touchpad of a laptop to initialize. +*/ +wait_for_device_probe ();md_run_setup ();
@@ -399,6 +403,7 @@ void __init prepare_namespace(void)while (driver_probe_done () !=0 || (ROOT_DEV =name_to_dev_t (saved_root_name)) ==0 )msleep (100 ); +async_synchronize_full (); } is_floppy =MAJOR (ROOT_DEV ) ==FLOPPY_MAJOR ;
init/do_mounts_md.c +3 -2
@@ -281,8 +281,9 @@ static void __init autodetect_raid(void)*/ printk (KERN_INFO "md: Waiting for all devices to be available before autodetect\n" );printk (KERN_INFO "md: If you don't use raid, use raid=noautodetect\n" ); -while (driver_probe_done () <0 ) -msleep (100 ); + +wait_for_device_probe (); + fd =sys_open ("/dev/md0" ,0 ,0 );if (fd >=0 ) {sys_ioctl (fd,RAID_AUTORUN , raid_autopart);
Re: [PATCH 1/9] Consolidate driver_probe_done() loops into one place
Quoted 8 lines · Rafael J. Wysocki
From: Arjan van de Ven <[email protected]>
there's a few places that currently loop over
driver_probe_done(), and I'm about to add another one. This patch abstracts it into a helper to reduce duplication.Signed-off-by Arjan van de Ven <[email protected]>Signed-off-by Rafael J. Wysocki <[email protected]>
Fine with me if this goes through Len's tree.
thanks,
greg k-h
Re: [PATCH 1/9] Consolidate driver_probe_done() loops into one place
In reply to Rafael J. WysockiHi,
Quoted 5 lines · Rafael J. Wysocki <rjw <at> sisk.pl> writes:
From: Arjan van de Ven <arjan <at> linux.intel.com>
there's a few places that currently loop over
driver_probe_done(), and I'm about to add another one. This patch abstracts it into a helper to reduce duplication.
[snip]
Quoted 8 lines
drivers/base/dd.c +8
-0+int wait_for_device_probe (void ) +{ +/* wait for the known devices to complete their probing */ +while (driver_probe_done () !=0 ) +msleep (100 ); +async_synchronize_full (); +return 0 ; +}
[snip]
Quoted 4 lines
Index:
linux-2.6/init/do_mounts.c
===================================================================init/do_mounts.c +0
-0
[snip]
init/do_mounts.c void __init prepare_namespace(void) · line 407
Quoted 6 lines
init/do_mounts.c +1
-0@@ -399,6 +403,7 @@ void __init prepare_namespace(void)while (driver_probe_done () !=0 || (ROOT_DEV =name_to_dev_t (saved_root_name)) ==0 )msleep (100 ); +async_synchronize_full (); }
Note that even this loop can be consolidated. The while loop above can be equivalently rewritten as
while ((ROOT_DEV = name_to_dev_t(saved_root_name)) == 0)
msleep(100);
while (driver_probe_done() != 0)
msleep(100);which, with your patch applied is equivalent to
while ((ROOT_DEV = name_to_dev_t(saved_root_name)) == 0)
msleep(100);
wait_for_device_probe();- Hari
PATCH 2/9 PM/resume: wait for device probing to finish
In reply to Rafael J. WysockiFrom: Arjan van de Ven <[email protected]>
the resume code does not currently wait for device probing to finish. Even without async function calls this is dicey and not correct, but with async function calls during the boot sequence this is going to get hit more...
This patch adds the synchronization using the newly introduced helper.
--- kernel/power/disk.c | 11 +++++++++++ 1 file changed, 11 insertions(+)
kernel/power/disk.c +11 -0
@@ -595,6 +595,12 @@ static int software_resume(void)unsigned int flags;/* +* If the user said "noresume".. bail out early. +*/ +if (noresume) +return 0 ; + +/* * name_to_dev_t() below takes a sysfs buffer mutex when sysfs * is configured into the kernel. Since the regular hibernate * trigger path is via sysfs which takes a buffer mutex before
@@ -610,6 +616,11 @@ static int software_resume(void)mutex_unlock (&pm_mutex);return -ENOENT ; } +/* +* Some device discovery might still be in progress; we need +* to wait for this to finish. +*/ +wait_for_device_probe (); swsusp_resume_device =name_to_dev_t (resume_file);pr_debug ("PM: Resume from partition %s\n" , resume_file); }else {
PATCH 2/9 PM/resume: wait for device probing to finish
In reply to Rafael J. WysockiFrom: Arjan van de Ven <[email protected]>
the resume code does not currently wait for device probing to finish. Even without async function calls this is dicey and not correct, but with async function calls during the boot sequence this is going to get hit more...
This patch adds the synchronization using the newly introduced helper.
--- kernel/power/disk.c | 11 +++++++++++ 1 file changed, 11 insertions(+)
kernel/power/disk.c +11 -0
@@ -595,6 +595,12 @@ static int software_resume(void)unsigned int flags;/* +* If the user said "noresume".. bail out early. +*/ +if (noresume) +return 0 ; + +/* * name_to_dev_t() below takes a sysfs buffer mutex when sysfs * is configured into the kernel. Since the regular hibernate * trigger path is via sysfs which takes a buffer mutex before
@@ -610,6 +616,11 @@ static int software_resume(void)mutex_unlock (&pm_mutex);return -ENOENT ; } +/* +* Some device discovery might still be in progress; we need +* to wait for this to finish. +*/ +wait_for_device_probe (); swsusp_resume_device =name_to_dev_t (resume_file);pr_debug ("PM: Resume from partition %s\n" , resume_file); }else {
Re: [PATCH 2/9] PM/resume: wait for device probing to finish
Quoted 11 lines · Rafael J. Wysocki
From: Arjan van de Ven <[email protected]>
the resume code does not currently wait for device probing to finish. Even without async function calls this is dicey and not correct, but with async function calls during the boot sequence this is going to get hit more...
This patch adds the synchronization using the newly introduced helper.
Signed-off-by Arjan van de Ven <[email protected]>Signed-off-by Rafael J. Wysocki <[email protected]>
Fine with me if this goes through Len's tree.
thanks,
greg k-h
Re: [PATCH 2/9] PM/resume: wait for device probing to finish
In reply to Rafael J. WysockiQuoted 11 lines · Rafael J. Wysocki
From: Arjan van de Ven <[email protected]>
the resume code does not currently wait for device probing to finish. Even without async function calls this is dicey and not correct, but with async function calls during the boot sequence this is going to get hit more...
This patch adds the synchronization using the newly introduced helper.
Signed-off-by Arjan van de Ven <[email protected]>Signed-off-by Rafael J. Wysocki <[email protected]>
Fine with me if this goes through Len's tree.
thanks,
greg k-h
PATCH 3/9 PM/hibernate: fix "swap breaks after hibernation failures"
In reply to Rafael J. WysockiFrom: Alan Jenkins <[email protected]>
http://bugzilla.kernel.org/show_bug.cgi?id=12239
The image writing code dropped a reference to the current swap device. This doesn't show up if the hibernation succeeds - because it doesn't affect the image which gets resumed. But it means multiple _failed_ hibernations end up freeing the swap device while it is still use!
swsusp_write() finds the block device for the swap file using swap_type_of(). It then uses blkdev_get() / blkdev_put() to open and close the block device.
Unfortunately, blkdev_get() assumes ownership of the inode of the block_device passed to it. So blkdev_put() calls iput() on the inode. This is by design and other callers expect this behaviour. The fix is for swap_type_of() to take a reference on the inode using bdget().
--- mm/swapfile.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
mm/swapfile.c +2 -2
@@ -635,7 +635,7 @@ int swap_type_of(dev_t device, sector_tif (!bdev) {if (bdev_p) - *bdev_p = sis->bdev; + *bdev_p =bdget (sis->bdev->bd_dev);spin_unlock (&swap_lock);return i;
@@ -647,7 +647,7 @@ int swap_type_of(dev_t device, sector_t struct swap_extent, list);if (se->start_block == offset) {if (bdev_p) - *bdev_p = sis->bdev; + *bdev_p =bdget (sis->bdev->bd_dev);spin_unlock (&swap_lock);bdput (bdev);
PATCH 4/9 PM: fix build for CONFIG_PM unset
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Compilation of kprobes.c with CONFIG_PM unset is broken due to some broken config dependncies. Fix that.
---
kernel/Makefile | 1 + kernel/power/Makefile | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-)
kernel/Makefile +1 -0
@@ -51,6 +51,7 @@ obj-$(CONFIG_UID16) += uid16.o obj-$(CONFIG_MODULES) += module.o obj-$(CONFIG_KALLSYMS) += kallsyms.o obj-$(CONFIG_PM) += power/ +obj-$(CONFIG_FREEZER) += power/ obj-$(CONFIG_BSD_PROCESS_ACCT) += acct.o obj-$(CONFIG_KEXEC) += kexec.o obj-$(CONFIG_BACKTRACE_SELF_TEST) += backtracetest.o
kernel/power/Makefile +1 -1
@@ -3,7 +3,7 @@ ifeq ($(CONFIG_PM_DEBUG),y) EXTRA_CFLAGS += -DDEBUG endif -obj-y := main.o +obj-$(CONFIG_PM) += main.o obj-$(CONFIG_PM_SLEEP) += console.o obj-$(CONFIG_FREEZER) += process.o obj-$(CONFIG_HIBERNATION) += swsusp.o disk.o snapshot.o swap.o user.o
PATCH 5/9 swsusp: dont fiddle with swappiness
In reply to Rafael J. WysockiFrom: Johannes Weiner <[email protected]>
sc.swappiness is not used in the swsusp memory shrinking path, do not set it.
---
mm/vmscan.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-)
mm/vmscan.c +1 -4
@@ -2112,7 +2112,6 @@ unsigned long shrink_all_memory(unsigned .may_swap = 0, .swap_cluster_max = nr_pages, .may_writepage = 1, - .swappiness = vm_swappiness, .isolate_pages = isolate_pages_global, };
@@ -2146,10 +2145,8 @@ unsigned long shrink_all_memory(unsignedint prio;/* Force reclaiming mapped pages in the passes #3 and #4 */ -if (pass >2 ) { +if (pass >2 ) sc.may_swap =1 ; - sc.swappiness =100 ; - }for (prio =DEF_PRIORITY ; prio >=0 ; prio--) {unsigned long nr_to_scan = nr_pages - ret;
PATCH 5/9 swsusp: dont fiddle with swappiness
In reply to Rafael J. WysockiFrom: Johannes Weiner <[email protected]>
sc.swappiness is not used in the swsusp memory shrinking path, do not set it.
---
mm/vmscan.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-)
mm/vmscan.c +1 -4
@@ -2112,7 +2112,6 @@ unsigned long shrink_all_memory(unsigned .may_swap = 0, .swap_cluster_max = nr_pages, .may_writepage = 1, - .swappiness = vm_swappiness, .isolate_pages = isolate_pages_global, };
@@ -2146,10 +2145,8 @@ unsigned long shrink_all_memory(unsignedint prio;/* Force reclaiming mapped pages in the passes #3 and #4 */ -if (pass >2 ) { +if (pass >2 ) sc.may_swap =1 ; - sc.swappiness =100 ; - }for (prio =DEF_PRIORITY ; prio >=0 ; prio--) {unsigned long nr_to_scan = nr_pages - ret;
PATCH 6/9 swsusp: clean up shrink_all_zones()
In reply to Rafael J. WysockiFrom: Johannes Weiner <[email protected]>
Move local variables to innermost possible scopes and use local variables to cache calculations/reads done more than once.
No change in functionality (intended).
---
mm/vmscan.c | 23 +++++++++++------------ 1 file changed, 11 insertions(+), 12 deletions(-)
mm/vmscan.c +11 -12
@@ -2057,31 +2057,31 @@ static unsigned long shrink_all_zones(un int pass, struct scan_control *sc) { struct zone *zone; -unsigned long nr_to_scan, ret =0 ; -enum lru_list l; +unsigned long ret =0 ;for_each_zone (zone) { +enum lru_list l;if (!populated_zone (zone))continue ; -if (zone_is_all_unreclaimable (zone) && prio !=DEF_PRIORITY )continue ;for_each_evictable_lru (l) { +enum zone_stat_item ls =NR_LRU_BASE + l; +unsigned long lru_pages =zone_page_state (zone, ls); +/* For pass = 0, we don't shrink the active list */ -if (pass ==0 && - (l ==LRU_ACTIVE || l ==LRU_ACTIVE_FILE )) +if (pass ==0 && (l ==LRU_ACTIVE_ANON || + l ==LRU_ACTIVE_FILE ))continue ; - zone->lru[l].nr_scan += - (zone_page_state (zone,NR_LRU_BASE + l) - >> prio) +1 ; + zone->lru[l].nr_scan += (lru_pages >> prio) +1 ;if (zone->lru[l].nr_scan >= nr_pages || pass >3 ) { +unsigned long nr_to_scan; + zone->lru[l].nr_scan =0 ; - nr_to_scan =min (nr_pages, -zone_page_state (zone, -NR_LRU_BASE + l)); + nr_to_scan =min (nr_pages, lru_pages); ret +=shrink_list (l, nr_to_scan, zone, sc, prio);if (ret >= nr_pages)
@@ -2089,7 +2089,6 @@ static unsigned long shrink_all_zones(un } } } -return ret; }
PATCH 6/9 swsusp: clean up shrink_all_zones()
In reply to Rafael J. WysockiFrom: Johannes Weiner <[email protected]>
Move local variables to innermost possible scopes and use local variables to cache calculations/reads done more than once.
No change in functionality (intended).
---
mm/vmscan.c | 23 +++++++++++------------ 1 file changed, 11 insertions(+), 12 deletions(-)
mm/vmscan.c +11 -12
@@ -2057,31 +2057,31 @@ static unsigned long shrink_all_zones(un int pass, struct scan_control *sc) { struct zone *zone; -unsigned long nr_to_scan, ret =0 ; -enum lru_list l; +unsigned long ret =0 ;for_each_zone (zone) { +enum lru_list l;if (!populated_zone (zone))continue ; -if (zone_is_all_unreclaimable (zone) && prio !=DEF_PRIORITY )continue ;for_each_evictable_lru (l) { +enum zone_stat_item ls =NR_LRU_BASE + l; +unsigned long lru_pages =zone_page_state (zone, ls); +/* For pass = 0, we don't shrink the active list */ -if (pass ==0 && - (l ==LRU_ACTIVE || l ==LRU_ACTIVE_FILE )) +if (pass ==0 && (l ==LRU_ACTIVE_ANON || + l ==LRU_ACTIVE_FILE ))continue ; - zone->lru[l].nr_scan += - (zone_page_state (zone,NR_LRU_BASE + l) - >> prio) +1 ; + zone->lru[l].nr_scan += (lru_pages >> prio) +1 ;if (zone->lru[l].nr_scan >= nr_pages || pass >3 ) { +unsigned long nr_to_scan; + zone->lru[l].nr_scan =0 ; - nr_to_scan =min (nr_pages, -zone_page_state (zone, -NR_LRU_BASE + l)); + nr_to_scan =min (nr_pages, lru_pages); ret +=shrink_list (l, nr_to_scan, zone, sc, prio);if (ret >= nr_pages)
@@ -2089,7 +2089,6 @@ static unsigned long shrink_all_zones(un } } } -return ret; }
PATCH 7/9 PM: Fix pm_notifiers during user mode hibernation
In reply to Rafael J. WysockiFrom: Andrey Borzenkov <[email protected]>
Snapshot device is opened with O_RDONLY during suspend and O_WRONLY durig resume. Make sure we also call notifiers with correct parameter telling them what we are really doing.
---
kernel/power/user.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
kernel/power/user.c +4 -4
@@ -95,15 +95,15 @@ static int snapshot_open(struct inode *i data->swap = swsusp_resume_device ?swap_type_of (swsusp_resume_device,0 ,NULL ) :-1 ; data->mode =O_RDONLY ; - error =pm_notifier_call_chain (PM_RESTORE_PREPARE ); + error =pm_notifier_call_chain (PM_HIBERNATION_PREPARE );if (error) -pm_notifier_call_chain (PM_POST_RESTORE ); +pm_notifier_call_chain (PM_POST_HIBERNATION ); }else { data->swap =-1 ; data->mode =O_WRONLY ; - error =pm_notifier_call_chain (PM_HIBERNATION_PREPARE ); + error =pm_notifier_call_chain (PM_RESTORE_PREPARE );if (error) -pm_notifier_call_chain (PM_POST_HIBERNATION ); +pm_notifier_call_chain (PM_POST_RESTORE ); }if (error)atomic_inc (&snapshot_device_available);
PATCH 7/9 PM: Fix pm_notifiers during user mode hibernation
In reply to Rafael J. WysockiFrom: Andrey Borzenkov <[email protected]>
Snapshot device is opened with O_RDONLY during suspend and O_WRONLY durig resume. Make sure we also call notifiers with correct parameter telling them what we are really doing.
---
kernel/power/user.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
kernel/power/user.c +4 -4
@@ -95,15 +95,15 @@ static int snapshot_open(struct inode *i data->swap = swsusp_resume_device ?swap_type_of (swsusp_resume_device,0 ,NULL ) :-1 ; data->mode =O_RDONLY ; - error =pm_notifier_call_chain (PM_RESTORE_PREPARE ); + error =pm_notifier_call_chain (PM_HIBERNATION_PREPARE );if (error) -pm_notifier_call_chain (PM_POST_RESTORE ); +pm_notifier_call_chain (PM_POST_HIBERNATION ); }else { data->swap =-1 ; data->mode =O_WRONLY ; - error =pm_notifier_call_chain (PM_HIBERNATION_PREPARE ); + error =pm_notifier_call_chain (PM_RESTORE_PREPARE );if (error) -pm_notifier_call_chain (PM_POST_HIBERNATION ); +pm_notifier_call_chain (PM_POST_RESTORE ); }if (error)atomic_inc (&snapshot_device_available);
PATCH 8/9 PM: Wait for console in resume
In reply to Rafael J. WysockiFrom: Arve Hjønnevåg <[email protected]>
Avoids later waking up to a blinking cursor if the device woke up and returned to sleep before the console switch happened.
---
kernel/power/console.c | 6 ++++++ 1 file changed, 6 insertions(+)
kernel/power/console.c +6 -0
@@ -78,6 +78,12 @@ void pm_restore_console(void) }set_console (orig_fgconsole);release_console_sem (); + +if (vt_waitactive (orig_fgconsole)) { +pr_debug ("Resume: Can't switch VCs." ); +return ; + } + kmsg_redirect = orig_kmsg; }#endif
PATCH 8/9 PM: Wait for console in resume
In reply to Rafael J. WysockiFrom: Arve Hjønnevåg <[email protected]>
Avoids later waking up to a blinking cursor if the device woke up and returned to sleep before the console switch happened.
---
kernel/power/console.c | 6 ++++++ 1 file changed, 6 insertions(+)
kernel/power/console.c +6 -0
@@ -78,6 +78,12 @@ void pm_restore_console(void) }set_console (orig_fgconsole);release_console_sem (); + +if (vt_waitactive (orig_fgconsole)) { +pr_debug ("Resume: Can't switch VCs." ); +return ; + } + kmsg_redirect = orig_kmsg; }#endif
PATCH 9/9 PM: Fix suspend_console and resume_console to use only one semaphore
In reply to Rafael J. WysockiFrom: Arve Hjønnevåg <[email protected]>
This fixes a race where a thread acquires the console while the console is suspended, and the console is resumed before this thread releases it. In this case, the secondary console semaphore would be left locked, and the primary semaphore would be released twice. This in turn would cause the console switch on suspend or resume to hang forever.
Note that suspend_console does not actually lock the console for clients that use acquire_console_sem, it only locks it for clients that use try_acquire_console_sem. If we change suspend_console to fully lock the console, then the kernel may deadlock on suspend. One client of try_acquire_console_sem is acquire_console_semaphore_for_printk, which uses it to prevent printk from using the console while it is suspended.
---
kernel/printk.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-)
kernel/printk.c +9 -6
@@ -73,7 +73,6 @@ EXPORT_SYMBOL(oops_in_progress);* driver system. */ static DECLARE_MUTEX (console_sem); -static DECLARE_MUTEX (secondary_console_sem);struct console *console_drivers;EXPORT_SYMBOL_GPL (console_drivers);
@@ -891,12 +890,14 @@ void suspend_console(void)printk ("Suspending console(s) (use no_console_suspend to debug)\n" );acquire_console_sem (); console_suspended =1 ; +up (&console_sem); }void resume_console (void ) {if (!console_suspend_enabled)return ; +down (&console_sem); console_suspended =0 ;release_console_sem (); }
@@ -912,11 +913,9 @@ void resume_console(void)void acquire_console_sem (void ) {BUG_ON (in_interrupt ()); -if (console_suspended) { -down (&secondary_console_sem); -return ; - }down (&console_sem); +if (console_suspended) +return ; console_locked =1 ; console_may_schedule =1 ; }
@@ -926,6 +925,10 @@ int try_acquire_console_sem(void) {if (down_trylock (&console_sem))return -1 ; +if (console_suspended) { +up (&console_sem); +return -1 ; + } console_locked =1 ; console_may_schedule =0 ;return 0 ;
@@ -979,7 +982,7 @@ void release_console_sem(void)unsigned wake_klogd =0 ;if (console_suspended) { -up (&secondary_console_sem); +up (&console_sem);return ; }
PATCH 9/9 PM: Fix suspend_console and resume_console to use only one semaphore
In reply to Rafael J. WysockiFrom: Arve Hjønnevåg <[email protected]>
This fixes a race where a thread acquires the console while the console is suspended, and the console is resumed before this thread releases it. In this case, the secondary console semaphore would be left locked, and the primary semaphore would be released twice. This in turn would cause the console switch on suspend or resume to hang forever.
Note that suspend_console does not actually lock the console for clients that use acquire_console_sem, it only locks it for clients that use try_acquire_console_sem. If we change suspend_console to fully lock the console, then the kernel may deadlock on suspend. One client of try_acquire_console_sem is acquire_console_semaphore_for_printk, which uses it to prevent printk from using the console while it is suspended.
---
kernel/printk.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-)
kernel/printk.c +9 -6
@@ -73,7 +73,6 @@ EXPORT_SYMBOL(oops_in_progress);* driver system. */ static DECLARE_MUTEX (console_sem); -static DECLARE_MUTEX (secondary_console_sem);struct console *console_drivers;EXPORT_SYMBOL_GPL (console_drivers);
@@ -891,12 +890,14 @@ void suspend_console(void)printk ("Suspending console(s) (use no_console_suspend to debug)\n" );acquire_console_sem (); console_suspended =1 ; +up (&console_sem); }void resume_console (void ) {if (!console_suspend_enabled)return ; +down (&console_sem); console_suspended =0 ;release_console_sem (); }
@@ -912,11 +913,9 @@ void resume_console(void)void acquire_console_sem (void ) {BUG_ON (in_interrupt ()); -if (console_suspended) { -down (&secondary_console_sem); -return ; - }down (&console_sem); +if (console_suspended) +return ; console_locked =1 ; console_may_schedule =1 ; }
@@ -926,6 +925,10 @@ int try_acquire_console_sem(void) {if (down_trylock (&console_sem))return -1 ; +if (console_suspended) { +up (&console_sem); +return -1 ; + } console_locked =1 ; console_may_schedule =0 ;return 0 ;
@@ -979,7 +982,7 @@ void release_console_sem(void)unsigned wake_klogd =0 ;if (console_suspended) { -up (&secondary_console_sem); +up (&console_sem);return ; }
Len Brown
reviewerQuoted 6 lines · Rafael J. Wysocki
Hi Len,
The following are fixes related to suspend and hibernation I have collected over the last couple of weeks. Some of them are in -mm. I think they all are 2.6.29 material.
Quoted 7 lines
1/9 and 2/9 fix hibernation regression caused by the async boot changes, I hope Greg will not object if they go through your tree.
3/9 fixes a long-standing hibernation issue that cause the resume partition to become unuseable after two consecutive failing attempts to hibernate.
4/9 fixes build of
kprobes.cwithCONFIG_PMunset.
looks like this one added power/ twice --
and looks like akpm sent a fixed version of this to linus already?
Quoted 8 lines
5/9 and 6/9 clean up memory freeing code used during hibernation.
7/9 fixes a bug related to the use of PM notifiers during hibernation with the help of the user land interface.
8/9 and 9/9 fix the handling of consoles during suspend/hibernation.
Please add these patches to the suspend branch and push to Linus for 2.6.29.
All except #4 are now on the suspend branch.
thanks,
--
Len Brown, Intel Open Source Technology Center
Len Brown
reviewerQuoted 6 lines · Rafael J. Wysocki
Hi Len,
The following are fixes related to suspend and hibernation I have collected over the last couple of weeks. Some of them are in -mm. I think they all are 2.6.29 material.
Quoted 7 lines
1/9 and 2/9 fix hibernation regression caused by the async boot changes, I hope Greg will not object if they go through your tree.
3/9 fixes a long-standing hibernation issue that cause the resume partition to become unuseable after two consecutive failing attempts to hibernate.
4/9 fixes build of
kprobes.cwithCONFIG_PMunset.
looks like this one added power/ twice --
and looks like akpm sent a fixed version of this to linus already?
Quoted 8 lines
5/9 and 6/9 clean up memory freeing code used during hibernation.
7/9 fixes a bug related to the use of PM notifiers during hibernation with the help of the user land interface.
8/9 and 9/9 fix the handling of consoles during suspend/hibernation.
Please add these patches to the suspend branch and push to Linus for 2.6.29.
All except #4 are now on the suspend branch.
thanks,
--
Len Brown, Intel Open Source Technology Center
Quoted 12 lines · Len Brown
Quoted 9 lines · Rafael J. Wysocki
Hi Len,
The following are fixes related to suspend and hibernation I have collected over the last couple of weeks. Some of them are in -mm. I think they all are 2.6.29 material.
1/9 and 2/9 fix hibernation regression caused by the async boot changes, I hope Greg will not object if they go through your tree.
Quoted 6 lines
Quoted 4 lines
3/9 fixes a long-standing hibernation issue that cause the resume partition to become unuseable after two consecutive failing attempts to hibernate.
4/9 fixes build of
kprobes.cwithCONFIG_PMunset.looks like this one added power/ twice --
That was on purpose.
Quoted 1 line
and looks like akpm sent a fixed version of this to linus already?
Yes, he did.
Quoted 10 lines
Quoted 8 lines
5/9 and 6/9 clean up memory freeing code used during hibernation.
7/9 fixes a bug related to the use of PM notifiers during hibernation with the help of the user land interface.
8/9 and 9/9 fix the handling of consoles during suspend/hibernation.
Please add these patches to the suspend branch and push to Linus for 2.6.29.
All except #4 are now on the suspend branch.
1/9 and 2/9 are also in the Greg's tree, AFAICS.
Greg, do you intend to push them to Linus? In that case I think it's better if Len drops them.
Rafael
Quoted 12 lines · Len Brown
Quoted 9 lines · Rafael J. Wysocki
Hi Len,
The following are fixes related to suspend and hibernation I have collected over the last couple of weeks. Some of them are in -mm. I think they all are 2.6.29 material.
1/9 and 2/9 fix hibernation regression caused by the async boot changes, I hope Greg will not object if they go through your tree.
Quoted 6 lines
Quoted 4 lines
3/9 fixes a long-standing hibernation issue that cause the resume partition to become unuseable after two consecutive failing attempts to hibernate.
4/9 fixes build of
kprobes.cwithCONFIG_PMunset.looks like this one added power/ twice --
That was on purpose.
Quoted 1 line
and looks like akpm sent a fixed version of this to linus already?
Yes, he did.
Quoted 10 lines
Quoted 8 lines
5/9 and 6/9 clean up memory freeing code used during hibernation.
7/9 fixes a bug related to the use of PM notifiers during hibernation with the help of the user land interface.
8/9 and 9/9 fix the handling of consoles during suspend/hibernation.
Please add these patches to the suspend branch and push to Linus for 2.6.29.
All except #4 are now on the suspend branch.
1/9 and 2/9 are also in the Greg's tree, AFAICS.
Greg, do you intend to push them to Linus? In that case I think it's better if Len drops them.
Rafael
Quoted 33 lines · Rafael J. Wysocki
Quoted 12 lines · Len Brown
Quoted 9 lines · Rafael J. Wysocki
Hi Len,
The following are fixes related to suspend and hibernation I have collected over the last couple of weeks. Some of them are in -mm. I think they all are 2.6.29 material.
1/9 and 2/9 fix hibernation regression caused by the async boot changes, I hope Greg will not object if they go through your tree.
Quoted 6 lines
Quoted 4 lines
3/9 fixes a long-standing hibernation issue that cause the resume partition to become unuseable after two consecutive failing attempts to hibernate.
4/9 fixes build of
kprobes.cwithCONFIG_PMunset.looks like this one added power/ twice --
That was on purpose.
Quoted 1 line
and looks like akpm sent a fixed version of this to linus already?
Yes, he did.
Quoted 7 lines
Quoted 7 lines
5/9 and 6/9 clean up memory freeing code used during hibernation.
7/9 fixes a bug related to the use of PM notifiers during hibernation with the help of the user land interface.
8/9 and 9/9 fix the handling of consoles during suspend/hibernation.
Quoted 8 lines
Quoted 3 lines
Quoted 1 line
Please add these patches to the suspend branch and push to Linus for 2.6.29.
All except #4 are now on the suspend branch.
1/9 and 2/9 are also in the Greg's tree, AFAICS.
Greg, do you intend to push them to Linus? In that case I think it's better if Len drops them.
I was not planning to send them for .29, sorry I got confused. I'll drop them from my tree and Len can send them all to Linus now.
Is that ok?
thanks,
greg k-h
Quoted 33 lines · Rafael J. Wysocki
Quoted 12 lines · Len Brown
Quoted 9 lines · Rafael J. Wysocki
Hi Len,
The following are fixes related to suspend and hibernation I have collected over the last couple of weeks. Some of them are in -mm. I think they all are 2.6.29 material.
1/9 and 2/9 fix hibernation regression caused by the async boot changes, I hope Greg will not object if they go through your tree.
Quoted 6 lines
Quoted 4 lines
3/9 fixes a long-standing hibernation issue that cause the resume partition to become unuseable after two consecutive failing attempts to hibernate.
4/9 fixes build of
kprobes.cwithCONFIG_PMunset.looks like this one added power/ twice --
That was on purpose.
Quoted 1 line
and looks like akpm sent a fixed version of this to linus already?
Yes, he did.
Quoted 7 lines
Quoted 7 lines
5/9 and 6/9 clean up memory freeing code used during hibernation.
7/9 fixes a bug related to the use of PM notifiers during hibernation with the help of the user land interface.
8/9 and 9/9 fix the handling of consoles during suspend/hibernation.
Quoted 8 lines
Quoted 3 lines
Quoted 1 line
Please add these patches to the suspend branch and push to Linus for 2.6.29.
All except #4 are now on the suspend branch.
1/9 and 2/9 are also in the Greg's tree, AFAICS.
Greg, do you intend to push them to Linus? In that case I think it's better if Len drops them.
I was not planning to send them for .29, sorry I got confused. I'll drop them from my tree and Len can send them all to Linus now.
Is that ok?
thanks,
greg k-h
Quoted 3 lines · Greg KH
I was not planning to send them for .29, sorry I got confused. I'll drop them from my tree and Len can send them all to Linus now.
they really really need to go into .29, they are a fix for a regression.
This patch exists for 6 weeks now as a fix for the regression.. why is it so hard to get it sent to Linus???
Quoted 3 lines · Greg KH
I was not planning to send them for .29, sorry I got confused. I'll drop them from my tree and Len can send them all to Linus now.
they really really need to go into .29, they are a fix for a regression.
This patch exists for 6 weeks now as a fix for the regression.. why is it so hard to get it sent to Linus???
Quoted 8 lines · Arjan van de Ven
Quoted 2 lines · Greg KH
I was not planning to send them for .29, sorry I got confused. I'll drop them from my tree and Len can send them all to Linus now.
they really really need to go into .29, they are a fix for a regression.
This patch exists for 6 weeks now as a fix for the regression.. why is it so hard to get it sent to Linus???
I think it's because both of us (Len and I) think the other is doing it :(
Ok, so, Len, are you willing to send these to Linus this week?
thanks,
greg k-h
Quoted 8 lines · Arjan van de Ven
Quoted 2 lines · Greg KH
I was not planning to send them for .29, sorry I got confused. I'll drop them from my tree and Len can send them all to Linus now.
they really really need to go into .29, they are a fix for a regression.
This patch exists for 6 weeks now as a fix for the regression.. why is it so hard to get it sent to Linus???
I think it's because both of us (Len and I) think the other is doing it :(
Ok, so, Len, are you willing to send these to Linus this week?
thanks,
greg k-h
Quoted 2 lines · Greg KH
Ok, so, Len, are you willing to send these to Linus this week?
I took them all, and added both of you to the cc, so if/when things break and get bisected to this series, both of you should be cc'd on the discussion.
Linus
Len Brown
reviewerQuoted 4 lines
I took them all, and added both of you to the cc, so if/when things break and get bisected to this series, both of you should be cc'd on the discussion.
Okay, I'll delete them from my suspend branch.
thanks,
-Len
Len Brown
reviewerQuoted 4 lines
I took them all, and added both of you to the cc, so if/when things break and get bisected to this series, both of you should be cc'd on the discussion.
Okay, I'll delete them from my suspend branch.
thanks,
-Len
Quoted 8 lines · Linus Torvalds
Quoted 2 lines · Greg KH
Ok, so, Len, are you willing to send these to Linus this week?
I took them all, and added both of you to the cc, so if/when things break and get bisected to this series, both of you should be cc'd on the discussion.
Thanks Linus! :-)
Rafael
Quoted 8 lines · Linus Torvalds
Quoted 2 lines · Greg KH
Ok, so, Len, are you willing to send these to Linus this week?
I took them all, and added both of you to the cc, so if/when things break and get bisected to this series, both of you should be cc'd on the discussion.
Thanks Linus! :-)
Rafael