On 20/05/11 19:34, Jon Dowland wrote:
On Fri, May 20, 2011 at 02:44:42PM +0700, Arief M Utama wrote:But, even so, I have done some test with "echo 'mem'> /sys/power/state" and "echo 'disk'> /sys/power/state" few times before, and it shows same behaviour with desktop initiated suspend and hibernate, "mem" suspend failed at resume, "disk" hibernate-resume working ok.Fundamentally, that implies a bug in the kernel for your particular hardware. Packages like uswsusp sometimes have more success because they apply kludges/quirks to work around the bug. These kludges/quirks are better handled in the kernel itself (sometimes hardware has quirks; other times it is a software error).
Update, and this might need to go also into Debian Suspend/Resume wiki,For suspend-resume, I've just discovered there's this bug notes in kernel's bugzilla:
https://bugzilla.kernel.org/show_bug.cgi?id=16396A changes regarding skip saving of nvs state on suspend apparently affected some systems. Affected system should be blacklisted in the table by looking at dmidecode output.
My system, Sony Vaio VGN-SR26GN, is affected, the solution for me currently is to add acpi_sleep=nonvs as boot parameter.
I've also attached my laptop's dmidecode output in there hopefully it will be included in the blacklist table in the next round.
Hope it helps anyone else out there :) Thanks all for the help and pointers. All the best. -arief