Re: PATCH: Address lots of pending pm_message_t changes

[Bernard Blackham <bernard@blackham.com.au>]

[Trimmed Cc]

On Tue, Feb 15, 2005 at 11:15:56AM +1100, Nigel Cunningham wrote:
> > Well, yes, if you switch pm_message_t into struct. But we are not yet
> > ready to do that... it is going to be typedefed to u32 for 2.6.11...
> 
> Ah. So I haven't realised that Bernard took your patches wholesale,
> which is why we're fixing the compile errors too :>
> 
> Okay then, I guess the whole thing isn't urgent then?

I was taking the whole shebang in order to differentiate between
PMSG_FREEZE and PMSG_SUSPEND - they're currently typedef'd to the
same thing (3), so drivers such as ide-disk can't decide whether or
not they need to spin down for the atomic copy or for powering off.
(Otherwise you'll find the HDD spinning down and up mid-suspend).

I believe vanilla swsusp passed "3" as the power state hence the HDD
spun down and up anyway, so there were no regressions there, just
bugs. Software Suspend 2 passed "4" as the power state which
ide-disk.c treated as "flush caches, but don't spin down". This is
what broke when the new typedefs went in, but would be fixed when
they're complete (whole shebang).

Bernard.

-- 
 Bernard Blackham <bernard at blackham dot com dot au>