I am using linux mint debian edition (LMDE). If I hibernate the system shortly after logging in, it might succeed. After some operations, it takes two times of attempts of s2disk to hibernate successfully. $ dmesg [142528.245031] PM: Creating hibernation image: [142528.248008] PM: Need to copy 530249 pages [142528.248008] PM: Normal pages needed: 530249 + 1024, available pages: 518040 [142528.248008] PM: Not enough free memory [142528.248008] PM: Error -12 creating hibernation image [142528.248008] CPU0: Thermal monitoring handled by SMI [142528.248008] Enabling non-boot CPUs ... [142528.248008] smpboot: Booting Node 0 Processor 1 APIC 0x1 [142528.144009] CPU1: Thermal monitoring handled by SMI [142528.256966] CPU1 is up [142528.261997] ACPI: Waking up from system sleep state S4 Supposedly, the free space in the swap may not be adequate. I then swapped off and then on to clean up the swap space to make all swap space available, which is confirmed by "free -h" command. Then, s2disk still needs 2 attempts to hibernate successfully.
Could you try the latest upstream kernel v3.13-rc1?
Is there easier way to help you debug this ? The linux environment is a production system for me and I don't want to risk damaging it by upgrading to a release-candidate kernel.
You can't upgrade kernel? So how can you test a fix patch? So far, I know a fix about hibernation swap allocation fix in the v3.13. This may fix the issue. http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=fd432b9f8c7c88428a4635b9f5a9c6e174df6e36
Because I had only 4G RAM and I used 64-bit LMDE, that might not fix it. I am not familiar with the procedure. I found the most forefront Debian sid used the kernel 3.11-2, still lesser than v3.13. So it would involve re-compile and re-configure the kernel by myself right ?
(In reply to Eric Lan from comment #4) > Because I had only 4G RAM and I used 64-bit LMDE, that might not fix it. > > I am not familiar with the procedure. I found the most forefront Debian sid > used the kernel 3.11-2, still lesser than v3.13. So it would involve > re-compile and re-configure the kernel by myself right ? Yes, you can follow this link to rebuild kernel. http://kernelnewbies.org/KernelBuild
It doesn't have anything to do with swap, it may be that we didn't preallocate/free enough memory during the start of hibernation. Please attach full dmesg after a failed hibernation.
Created attachment 118501 [details] dmesg after failed "s2disk" dmesg after failed "s2disk" (the first time)
*** This bug has been marked as a duplicate of bug 47931 ***