Suspend2 isn't invasive.

[Nigel Cunningham]

On Sun, 2006-12-03 at 11:12 +0100, Tollef Fog Heen wrote:
> * Nigel Cunningham 
> 
> (Please respect the Mail-Followup-To; if I don't read a list I post
> to, I explicitly request Ccs.)

Sorry for the mess up. Hopefully this one will go right.

> | Ok. That's quite a lot of progress since I last checked. When you say
> | suspending to files, do you mean to swapfiles or to ordinary files? Does
> | it support multiple swap devices now?
> 
> Swapfiles.  It doesn't look like it supports multiple devices, but
> adding that does look like a SMOP.

Ok. So it's not everything Suspend2 supports then.

> | > There's nothing inherent in uswsusp's design which makes it have to
> | > try to free lots of memory.  At least, nothing I could immediately
> | > see?
> | 
> | The maximum image size of 1/2 of memory.
> 
> This seems to be less of a problem when you enable compression,
> doesn't it?  (I don't see where the limitation in the code is, but I
> haven't looked deeply at the suspend code; I've just added usplash
> support)

Compression is a completely separate issue. If you have, say, 1GB of
RAM, with [u]swsusp, you're only ever going to get a 512MB image at the
most. Compression might mean that only 256MB gets saved, but you're
still always going to be throwing away up to half the contents of RAM
before you even begin to write the image. Depending on what you're using
that 1/2GB for, it can make for a far less responsive system post
resume, and it can also greatly increase the chances of failing to
suspend (because you fail to free enough RAM).

Regards,

Nigel