Resuming from a crashdump

[David Schultz]

On Tue, Jan 25, 2005, Bruce M Simpson wrote:
> If we could take a clean subsystem-by-subsystem approach to marshaling
> kernel state to disk, that would be good. What gives me particular pain
> here is dealing with things like the filesystem. How does one deal with
> open files, etc, with pending I/O?

You just write an image of physical memory to disk, and the data
structures behind the open files come back exactly as they were
before.  ;-)  The hard part, I think, is getting all the device
drivers to understand how to reinitialize the hardware in exactly
the same state it was in before the suspend.

BTW, contrary to what Matt said, I don't think network
connections are a big deal.  If you turn your computer on after a
long plane flight to Tahiti, and the kernel still thinks you have
active TCP connections, it will quickly discover that the
remote host doesn't think so and take appropriate action.
I'm pretty sure Windows doesn't do anything special with network
connections, because a friend of mine says he manages to turn off
his laptop, walk to the other side of a building and down a few
floors, turn the laptop back on, and still be logged in to a
remote server via SSH.