> You could even let the fs code do the compression for
> you, when you have your swap space on a e2compr (see
> http://www.netspace.net.au/~reiter/e2compr/ ).
As I say, it would only work if the swapping code used the mmap
interface of readpage and write, as e2compr doesn't provide a block
interface.
If people want to try compressed swapping, have a look at DouBle, by
Jean-Marc Verbavatz. It provides a compressed block device. You tell
it the name of a block device or (fixed-size) regular file, some
compression parameters (compression algorithm, cluster size, storage
granularity), and a (conservative) expected compression ratio. It
then provides a block device (say /dev/dbl0) that has more blocks than
the file or device where the data is actually being stored.
URL for DouBle is <ftp://achaz.saclay.cea.fr/pub/double/>.
There's also a (old, last time I looked) version on sunsite.
I imagine that the same optimisations apply to that as apply to any
compressed swapping system, so why don't people have a look at that
and make any necessary optimisations to that.
I don't think there's much more point in continuing this thread.
pjm.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu