Bug#365368: acknowledged by developer (Re: Bug#365368: powersaved mistaken about resume= parameter)
jacob wrote: On Thu, May 04, 2006 at 01:05:37AM +0200, Michael Biebl wrote: jacob wrote: On Wed, May 03, 2006 at 01:03:25PM -0700, Debian Bug Tracking System wrote: From: Michael Biebl [EMAIL PROTECTED] jacob wrote: On Tue, May 02, 2006 at 03:37:05PM +0200, Michael Biebl wrote: jacob wrote: Package: powersaved Version: 0.12.11-1 Severity: normal powersaved complains that there is no resume= boot option, when I try to suspend to disk. However: $ cat /sys/power/resume 0:0 Ok, here is the source of the problem. Your resume partition is not correctly set (0:0 basically means, no resume partition). So powersaved is actually working correctly (that's why I'm closing this bug). The question now is, why the resume partition is not correctly set. I'm not a initrd expert, but a quick google search revealed [1]. Maybe your problem is similar. Could be that one of the initrd scripts does not set /sys/power/resume correctly. You could try to comment out SUSPEND in /etc/mkinitramfs/conf.d/resume and update the initrd (update-initramfs), maybe then /sys/power/resume is not modified by the scripts in the initrd. Light finally dawns. It was something even more insidious. I was suspending before I got resume working, with a ramdisk created by yaird. Yaird doesn't properly support resume yet. When I discovered this, I installed initramfs-tools, updated my ramdisk, and suspended. It resumed properly on reboot. I hadn't rebooted since. So it never set /sys/power/resume. After a reboot, /sys/power/resume says 3:2, as it should. Thanks for your help, Michael. Hopefully, no one else follows this odd sequence, but if they do, then this bug report should help them. Great to hear that it works now. If you want you could file a wishlist bug against yaird to support setting the resume partition as does initramfs-tools. Cheers, Michael signature.asc Description: OpenPGP digital signature
Bug#365368: acknowledged by developer (Re: Bug#365368: powersaved mistaken about resume= parameter)
On Wed, May 03, 2006 at 01:03:25PM -0700, Debian Bug Tracking System wrote: From: Michael Biebl [EMAIL PROTECTED] jacob wrote: On Tue, May 02, 2006 at 03:37:05PM +0200, Michael Biebl wrote: jacob wrote: Package: powersaved Version: 0.12.11-1 Severity: normal powersaved complains that there is no resume= boot option, when I try to suspend to disk. However: (root) /boot$ grep resume= /boot/grub/menu.lst ## e.g. defoptions=vga=791 resume=/dev/hda5 # defoptions=resume=/dev/hda2 kernel /boot/vmlinuz-2.6.16-dsdt-64m root=/dev/hda1 ro resume=/dev/hda2 kernel /boot/vmlinuz-2.6.16-1-k7 root=/dev/hda1 ro resume=/dev/hda2 (root) /boot$ cat /proc/cmdline root=/dev/hda1 ro resume=/dev/hda2 Setting SUSPEND2DISK_SKIP_RESUME_CHECK to yes, as mentioned in the error message (although not found in /etc/powersave/sleep), corrects the problem. Hi Jacob, could you please send me the output from cat /sys/power/resume? powersaved uses this method for determining the swap partition as there are several ways to specify the default resume partition (kernel config, boot pararameter, initrd parameter). Please also send me the log file /var/log/suspend2disk.log. $ cat /sys/power/resume 0:0 Ok, here is the source of the problem. Your resume partition is not correctly set (0:0 basically means, no resume partition). So powersaved is actually working correctly (that's why I'm closing this bug). The question now is, why the resume partition is not correctly set. Memory info: total used free sharedbuffers cached Mem:710696 302760 407936 0 0 228288 -/+ buffers/cache: 74472 636224 Swap:0 0 0 Interesting. Seems as if the swap partition is not activated. Do you have a /dev/hda2 noneswapsw 0 0 line in your /etc/fstab? Hmm... that's odd. $ free total used free sharedbuffers cached Mem: 710696 330852 379844 0 0 263396 -/+ buffers/cache:67456 643240 Swap:0 0 0 $ sudo /sbin/swapon -a $ free total used free sharedbuffers cached Mem: 710696 330940 379756 0 4 263512 -/+ buffers/cache:67424 643272 Swap: 506036 0 506036 $ cat /sys/power/resume 0:0 $ Another possible reason could be, that the resume partition is not correctly set in the initrd (You are using a Debian kernel which uses an initial ramdisk). Could you please check /etc/mkinitramfs/conf.d/resume and also /etc/mkinitramfs/initramfs.conf if they have a bogus RESUME=... $ grep RESUME /etc/mkinitramfs/conf.d/resume RESUME=/dev/hda2 $ grep RESUME /etc/mkinitramfs/initramfs.conf # RESUME: [ /dev/hda2 | /dev/sdb2 ] #RESUME= $ sudo powersave -U $ gives the same The resume partition is not set up... message. $ cat /sys/power/resume 0:0 $ So powersave *is* doing the right thing, based on the sysfs entries. So, maybe this is a kernel bug? Jacob -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#365368: acknowledged by developer (Re: Bug#365368: powersaved mistaken about resume= parameter)
jacob wrote: On Wed, May 03, 2006 at 01:03:25PM -0700, Debian Bug Tracking System wrote: From: Michael Biebl [EMAIL PROTECTED] jacob wrote: On Tue, May 02, 2006 at 03:37:05PM +0200, Michael Biebl wrote: jacob wrote: Package: powersaved Version: 0.12.11-1 Severity: normal powersaved complains that there is no resume= boot option, when I try to suspend to disk. However: (root) /boot$ grep resume= /boot/grub/menu.lst ## e.g. defoptions=vga=791 resume=/dev/hda5 # defoptions=resume=/dev/hda2 kernel /boot/vmlinuz-2.6.16-dsdt-64m root=/dev/hda1 ro resume=/dev/hda2 kernel /boot/vmlinuz-2.6.16-1-k7 root=/dev/hda1 ro resume=/dev/hda2 (root) /boot$ cat /proc/cmdline root=/dev/hda1 ro resume=/dev/hda2 Setting SUSPEND2DISK_SKIP_RESUME_CHECK to yes, as mentioned in the error message (although not found in /etc/powersave/sleep), corrects the problem. Hi Jacob, could you please send me the output from cat /sys/power/resume? powersaved uses this method for determining the swap partition as there are several ways to specify the default resume partition (kernel config, boot pararameter, initrd parameter). Please also send me the log file /var/log/suspend2disk.log. $ cat /sys/power/resume 0:0 Ok, here is the source of the problem. Your resume partition is not correctly set (0:0 basically means, no resume partition). So powersaved is actually working correctly (that's why I'm closing this bug). The question now is, why the resume partition is not correctly set. Memory info: total used free sharedbuffers cached Mem:710696 302760 407936 0 0 228288 -/+ buffers/cache: 74472 636224 Swap:0 0 0 Interesting. Seems as if the swap partition is not activated. Do you have a /dev/hda2 noneswapsw 0 0 line in your /etc/fstab? Hmm... that's odd. $ free total used free sharedbuffers cached Mem: 710696 330852 379844 0 0 263396 -/+ buffers/cache:67456 643240 Swap:0 0 0 $ sudo /sbin/swapon -a $ free total used free sharedbuffers cached Mem: 710696 330940 379756 0 4 263512 -/+ buffers/cache:67424 643272 Swap: 506036 0 506036 $ cat /sys/power/resume 0:0 $ Another possible reason could be, that the resume partition is not correctly set in the initrd (You are using a Debian kernel which uses an initial ramdisk). Could you please check /etc/mkinitramfs/conf.d/resume and also /etc/mkinitramfs/initramfs.conf if they have a bogus RESUME=... $ grep RESUME /etc/mkinitramfs/conf.d/resume RESUME=/dev/hda2 $ grep RESUME /etc/mkinitramfs/initramfs.conf # RESUME: [ /dev/hda2 | /dev/sdb2 ] #RESUME= $ sudo powersave -U $ gives the same The resume partition is not set up... message. $ cat /sys/power/resume 0:0 $ So powersave *is* doing the right thing, based on the sysfs entries. So, maybe this is a kernel bug? I'm not a initrd expert, but a quick google search revealed [1]. Maybe your problem is similar. Could be that one of the initrd scripts does not set /sys/power/resume correctly. You could try to comment out SUSPEND in /etc/mkinitramfs/conf.d/resume and update the initrd (update-initramfs), maybe then /sys/power/resume is not modified by the scripts in the initrd. Cheers, Michael [1] http://lists.debian.org/debian-kernel/2005/10/msg00924.html signature.asc Description: OpenPGP digital signature
Bug#365368: acknowledged by developer (Re: Bug#365368: powersaved mistaken about resume= parameter)
On Thu, May 04, 2006 at 01:05:37AM +0200, Michael Biebl wrote: jacob wrote: On Wed, May 03, 2006 at 01:03:25PM -0700, Debian Bug Tracking System wrote: From: Michael Biebl [EMAIL PROTECTED] jacob wrote: On Tue, May 02, 2006 at 03:37:05PM +0200, Michael Biebl wrote: jacob wrote: Package: powersaved Version: 0.12.11-1 Severity: normal powersaved complains that there is no resume= boot option, when I try to suspend to disk. However: $ cat /sys/power/resume 0:0 Ok, here is the source of the problem. Your resume partition is not correctly set (0:0 basically means, no resume partition). So powersaved is actually working correctly (that's why I'm closing this bug). The question now is, why the resume partition is not correctly set. I'm not a initrd expert, but a quick google search revealed [1]. Maybe your problem is similar. Could be that one of the initrd scripts does not set /sys/power/resume correctly. You could try to comment out SUSPEND in /etc/mkinitramfs/conf.d/resume and update the initrd (update-initramfs), maybe then /sys/power/resume is not modified by the scripts in the initrd. Light finally dawns. It was something even more insidious. I was suspending before I got resume working, with a ramdisk created by yaird. Yaird doesn't properly support resume yet. When I discovered this, I installed initramfs-tools, updated my ramdisk, and suspended. It resumed properly on reboot. I hadn't rebooted since. So it never set /sys/power/resume. After a reboot, /sys/power/resume says 3:2, as it should. Thanks for your help, Michael. Hopefully, no one else follows this odd sequence, but if they do, then this bug report should help them. Jacob -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]