Re: [PATCH 1/2] sunrpc: make rpc_wait_bit_killable handle freeze events

[Tejun Heo <tj@kernel.org>]

Hello, Jeff.

On Mon, Nov 21, 2011 at 03:36:15PM -0500, Jeff Layton wrote:
> I suppose you're suggesting something like this?
> 
> 	freezer_do_not_count();
> 	if (!try_to_freeze())
> 		schedule();
> 	freezer_count();

No,

	freezer_do_not_count();
	schedule();
	freezer_count();
	try_to_freeze();

freezer_count() may make %current eligible for freezing again, so it
has to check whether freezing condition is pending afterwards.

> The freezer is not really my forte', but I have to say that the whole
> do_not_count scheme seems "sketchy". Would we not be better off with
> something closer to the original method that I proposed, along with a
> new TASK_WAKEFREEZE state bit? Processes that are sleeping in
> uninterruptible sleep that are able to deal with freezer events could
> set that bit and fake_signal_wake_up could be changed to also wake
> processes with that bit set.

AFAICS, the above should be race-free w/ the pending freezer update.

  http://git.kernel.org/?p=linux/kernel/git/tj/misc.git;a=shortlog;h=refs/heads/pm-freezer

Thank you.

-- 
tejun