On Sun, Nov 14, 2004 at 07:44:38PM +0100, Michael Karbach wrote:
Hi,
I try (as usual) to suspend-to-disk with ACPI and powersave on SuSE-9.2. The messages from the system is appended below. Up to now this kind of suspend has worked pretty well, and I am really did not know what has happend. May be this is related to the last you-update I have done a few days ago?
probably not.
/var/log/message:
Nov 14 19:29:02 wptnb8 kernel: Freeing memory... done (17088 pages freed) Nov 14 19:29:02 wptnb8 kernel: PM: Attempting to suspend to disk. Nov 14 19:29:02 wptnb8 kernel: PM: snapshotting memory. Nov 14 19:29:02 wptnb8 kernel: swsusp: critical section: Nov 14 19:29:02 wptnb8 kernel: swsusp: Saving Highmem Nov 14 19:29:02 wptnb8 kernel: [nosave pfn 0x3be]<7>[nosave pfn 0x3bf]swsusp: Need to copy 131684 pages Nov 14 19:29:02 wptnb8 kernel: suspend: (pages needed: 131684 + 512 free: 64892) Nov 14 19:29:02 wptnb8 kernel: swsusp: Not enough free pages: Have 64892
There is not enough memory free to do the atomic copy. The machine needs 131684 pages but it has only 64892...
Nov 14 19:29:02 wptnb8 kernel: swsusp: Restoring Highmem [...] Nov 14 19:29:03 wptnb8 kernel: Restarting tasks... done
...so it resumes immediately. What is strange is the demand for so much memory for suspend. Do you have some very memory-intensive applications running? Note that there is a second error that can happen at suspend: there is enough free memory, but the routine needs an high order allocation (consecutive memory pages) and sometimes memory is fragmented and this allocation (and with it the whole suspend) fails. In this case you get a "page allocation failure" followed by a call backtrace in the logs. Right now there is not much you can do about this but there is work going on to avoid this allocation failures or at least work aroung them. -- Stefan Seyfried