Re: SOFTWARE_SUSPEND as a module

["Nigel Cunningham" <ncunningham@linuxmail.com>]

Hi.

On Fri, 23 Apr 2004 19:38:36 +1000, Herbert Xu  
<herbert@gondor.apana.org.au> wrote:
> What I've done is:
>
> * Split swsusp.c into swsusp-core.c and the rest based on whether it
> is called by do_magic() or not.
> * Added EXPORTs for the symbols needed by what's left of swsusp.
> * Added module parameter parsing.

Do you really need to make suspend a module to achieve this? I'll freely  
admit it might be nice to save a little bit of memory by modularising it,  
and it might save Pavel and I a reboot or two while developing, but for  
what you're after, couldn't you just move the call that starts resuming?

> As a side-effect it also allows you to resume from devices that couldn't
> be done before due to the need for user-space setup.  Examples are LVM
> and NBD.

LVM can be compiled in, can't it? Does it need to do some setup from an  
initrd?

>> prepared for it in userland? In such case you need to add
>> freeze_processes() to resume path.
>
> Is that really necessary if the user-space caller ensures that all
> disk accesses are shut down? After all the loading for resume will
> occur in the initrd before any disk activity has occured.

So long as the initrd doesn't mount any filesystems that were suspended,  
you should be okay, IIRC.

Nigel
-- 
Nigel Cunningham
C/- Westminster Presbyterian Church Belconnen
61 Templeton Street, Cook, ACT 2614, Australia.
+61 (2) 6251 7727 (wk)