Re: [gentoo-user] Weird hibernate problem
On Tue, Sep 11, 2012 at 11:52 AM, Timur Aydin wrote: > On 09/11/12 19:08, Canek Peláez Valdés wrote: >> Mmmh. You didn't set CONFIG_PM_STD_PARTITION: >> > > I specified the partition on the kernel command line: > > ta@bonsai ~/uclinux_2011R1/db1/uclinux-dist $ cat /boot/grub/grub.conf > timeout 30 > default 0 > splashimage=(hd0,0)/boot/grub/splash.xpm.gz > > title Gentoo > root (hd0,0) > kernel /vmlinuz root=/dev/sda3 resume=/dev/sda6 > > title Gentoo.old > root (hd0,0) > kernel /vmlinuz.old root=/dev/sda3 resume=/dev/sda6 That is for resuming; not for actually hibernating. >> CONFIG_PM_STD_PARTITION="" >> >> Set it to /dev/sda6, and see if it helps. Also, can we have a look at >> yout /etc/fstab file? > > Here is /etc/fstab: > > # > > > # NOTE: If your BOOT partition is ReiserFS, add the notail option to opts. > /dev/sda1 /boot ext2noatime 1 2 > /dev/sda2 /home ext3noatime 0 2 > /dev/sda3 / ext3noatime 0 1 > /dev/sda5 noneswapsw 0 0 > /dev/sda7 /backup ext3noatime 0 2 > /dev/cdrom /mnt/cdrom autonoauto,ro 0 0 > > I want to attach the ps ax output before and after hibernate, but I am > concerned about filesystem corruption with all those kworker threads > lurking around... I hibernate my desktop all the time. I only have a swap partition, and it's both set in CONFIG_PM_STD_PARTITION and in my /etc/fstab: CONFIG_PM_STD_PARTITION="/dev/sdc1" # LABEL=Gentoo/ ext4noatime 0 1 LABEL=Home /home ext4noatime,auto 0 2 LABEL=Swap noneswapsw 0 0 shm /dev/shmtmpfs nodev,nosuid,noexec 0 0 tmpfs /tmptmpfs defaults,nosuid,size=100% 0 0 # swaplabel /dev/sdc1 LABEL: Swap UUID: 28a9478e-8b53-4591-8fec-64512eeadf6c Regards. -- Canek Peláez Valdés Posgrado en Ciencia e Ingeniería de la Computación Universidad Nacional Autónoma de México
Re: [gentoo-user] Weird hibernate problem
On 09/11/12 19:08, Canek Peláez Valdés wrote: > Mmmh. You didn't set CONFIG_PM_STD_PARTITION: > I specified the partition on the kernel command line: ta@bonsai ~/uclinux_2011R1/db1/uclinux-dist $ cat /boot/grub/grub.conf timeout 30 default 0 splashimage=(hd0,0)/boot/grub/splash.xpm.gz title Gentoo root (hd0,0) kernel /vmlinuz root=/dev/sda3 resume=/dev/sda6 title Gentoo.old root (hd0,0) kernel /vmlinuz.old root=/dev/sda3 resume=/dev/sda6 > CONFIG_PM_STD_PARTITION="" > > Set it to /dev/sda6, and see if it helps. Also, can we have a look at > yout /etc/fstab file? Here is /etc/fstab: # # NOTE: If your BOOT partition is ReiserFS, add the notail option to opts. /dev/sda1 /boot ext2noatime 1 2 /dev/sda2 /home ext3noatime 0 2 /dev/sda3 / ext3noatime 0 1 /dev/sda5 noneswapsw 0 0 /dev/sda7 /backup ext3noatime 0 2 /dev/cdrom /mnt/cdrom autonoauto,ro 0 0 I want to attach the ps ax output before and after hibernate, but I am concerned about filesystem corruption with all those kworker threads lurking around... Timur
Re: [gentoo-user] Weird hibernate problem
On Tue, Sep 11, 2012 at 2:04 AM, Timur Aydin wrote: > On 9/11/2012 6:31 AM, W.Kenworthy wrote: >> Hi Timur, we need a lot more information: >> >> what kernel version >> in kernel or ToI hibernation >> are you using genkernel >> separate /usr >> lvm >> >> and anything else applicable. >> >> Hibernation can be a pig to get going. >> > > Hi, here is the additional information: > > System is ~x86 with 3.5.3 kernel that I compiled myself. Config is > attached. Not sure whether I am using in kernel or ToI hibernation. But > to hibernate, I have initially tried using the command line tools of > pm-utils (pm-hibernate). In the end, I wanted to configure KDE so that I > can quickly hibernate my development system to conserve electricity. > /usr is not separate, but /home is on a separate partition. Here are the > partitions: > > /dev/sda1 (/boot) > /dev/sda2 (/home) > /dev/sda3 (/) > /dev/sda5 (swap) > /dev/sda6 (swap, used for hibernation) > /dev/sda7 (/backup) > > /dev/sdb1 (1. raid disk) > /dev/sdc1 (2. raid disk) > /dev/sdd1 (3. raid disk) > /dev/sde1 (4. raid disk) > /dev/sdf1 (5. raid disk) > > I did a few more tests. If I keep suspending the system and waking it > up, the number of kworker, migrate and ksoftirq threads is increasing. Mmmh. You didn't set CONFIG_PM_STD_PARTITION: CONFIG_PM_STD_PARTITION="" Set it to /dev/sda6, and see if it helps. Also, can we have a look at yout /etc/fstab file? Regards. -- Canek Peláez Valdés Posgrado en Ciencia e Ingeniería de la Computación Universidad Nacional Autónoma de México
Re: [gentoo-user] Weird hibernate problem
Hi Timur, we need a lot more information: what kernel version in kernel or ToI hibernation are you using genkernel separate /usr lvm and anything else applicable. Hibernation can be a pig to get going. BillK -Original Message- From: Timur Aydin Reply-to: gentoo-user@lists.gentoo.org To: gentoo-user@lists.gentoo.org Subject: [gentoo-user] Weird hibernate problem Date: Mon, 10 Sep 2012 23:06:17 +0300 After corrupting my gentoo root filesystem system during hibernate experiments, I have finally finished the reinstallation. But hibernate still doesn't seem to work correctly. The symptoms are the same as during the experiments leading to the root fs corruption, but this time the root seems to remain intact. Here is what happened during the first experiment. I had an 8G swap partition as /dev/sda5 and a data partition as /dev/sda6. Thinking that pm-hibernate requires a dedicated, separate partition, I backed up /dev/sda6, turned off swap at /dev/sda5, deleted /dev/sda[56] and then created /dev/sda5 (8G), /dev/sda6 (8G) and /dev/sda7 (remaining size) Then I specified /dev/sda6 as the resume partition on the kernel command line. But when I did the pm-hibernate, the system powered off, and after reboot, the system seemed to have restored itself to the state it was at when I ran pm-hibernate. So it "seemed" to have worked, but the system was strangely unstable. There were many filesystem errors in the root partition and when I did a ps ax, I saw hundreds of kworker kernel threads lingering around. It was as if the hibernate image was slightly corrupted, but not enough to cause a complete lockup, but enough to cause there strange symptoms. I first thought this was related to using the swap partition as the resume destination. But after reinstalling gentoo, I again used a separate partition for hibernate, but I am still seeing the same symtoms. Many kworker kernel threads are sleeping. But this time, the root filesystem didn't have any error. Concerned that a filesystem corruption is imminent, I immediately turned off power. So, what could be causing these strange problems? Based on what I have read so far, the resume partition needs to be an active swap partition. This seems rather strange, because linux is using the swap partition for memory management as well. So shouldn't these be well separated to prevent corrupting each other? Hope someone can help me make sense of all of this...
Re: [gentoo-user] Weird hibernate problem
One unusual property of this system is that it has 5 additional SATA disks to be used for RAID experiments, in addition to the disk holding gentoo. So there are a total of 6 disks. But during these experiments, these 5 additional disks are not mounted. The motherboard is an Asus P8V68Z-VPRO/GEN3 with 8G RAM and a i7-2600K CPU, running 32bit gentoo. There are no peripherals attached to any of the motherboard expansion slots. -- Timur
Re: [gentoo-user] Weird hibernate problem
I have just tried again using pm-suspend and the same thing happens. Everything looks like it has worked, but when I do a ps ax, there are many (currently around 50) sleeping kernel threads. There are also a few extraneous "migration, ksoftirq" threads intermixed. -- Timur
[gentoo-user] Weird hibernate problem
After corrupting my gentoo root filesystem system during hibernate experiments, I have finally finished the reinstallation. But hibernate still doesn't seem to work correctly. The symptoms are the same as during the experiments leading to the root fs corruption, but this time the root seems to remain intact. Here is what happened during the first experiment. I had an 8G swap partition as /dev/sda5 and a data partition as /dev/sda6. Thinking that pm-hibernate requires a dedicated, separate partition, I backed up /dev/sda6, turned off swap at /dev/sda5, deleted /dev/sda[56] and then created /dev/sda5 (8G), /dev/sda6 (8G) and /dev/sda7 (remaining size) Then I specified /dev/sda6 as the resume partition on the kernel command line. But when I did the pm-hibernate, the system powered off, and after reboot, the system seemed to have restored itself to the state it was at when I ran pm-hibernate. So it "seemed" to have worked, but the system was strangely unstable. There were many filesystem errors in the root partition and when I did a ps ax, I saw hundreds of kworker kernel threads lingering around. It was as if the hibernate image was slightly corrupted, but not enough to cause a complete lockup, but enough to cause there strange symptoms. I first thought this was related to using the swap partition as the resume destination. But after reinstalling gentoo, I again used a separate partition for hibernate, but I am still seeing the same symtoms. Many kworker kernel threads are sleeping. But this time, the root filesystem didn't have any error. Concerned that a filesystem corruption is imminent, I immediately turned off power. So, what could be causing these strange problems? Based on what I have read so far, the resume partition needs to be an active swap partition. This seems rather strange, because linux is using the swap partition for memory management as well. So shouldn't these be well separated to prevent corrupting each other? Hope someone can help me make sense of all of this... -- Timur
