Suspend to disk 3428

No. Why will the swap parbreastion contents be trashed? If it doesn't get trashed during a normal VM pageout, why would it be impossible for swsup to do the same?

What's the problem?

Easy. Imagine it this way: swsup simply suspends all processes (like "kill -STOP") and then causes all the memory pages of all processes to be swapped out. Then, what remains? The kernel code and kernel data structures. swsusp can then allocate *free* space from the swap device, and save the kernel code&data to the free space. (Then, a few "pointers" need to be stored in a certain location -- in the swap signature -- to tell swsup where to load back those kernel pages upon resume.)

Of course, this is a very oversimplified picture. Just to let you understand how it is possible for swsusp to store things into swap without trashing the active swap contents.

No. You can use your swap parbreastion as usual. And if you have 2 swap parbreastions, you can have 2 different suspend-resume contexts, meaning that you can select freely which one to resume from. (But there could be problems with the consistency of data in other file parbreastions, if these parbreastions are shared among the different suspend-resume contexts.)

The VM pager won't trash the used blocks on the swap parbreastions. I can't see any reason that swsusp would.

--