[dm-devel] staged dm_internal_{suspend, resume} related changes for wider review

[Mike Snitzer]

On Wed, Nov 05 2014 at  9:37am -0500,
Mikulas Patocka <mpatocka at redhat.com> wrote:

> The patch series introduces two suspend mechanisms and it is unclear how 
> should they interact with each other.

And this point is not correct.  As you know dm_internal_suspend and
dm_internal_resume interface predates any of my changes.

That existing interface was extended them to be (mostly) fully formed
equivalents of dm_suspend() and dm_resume().

I say "mostly" because dm_internal_resume() doesn't call into the targets'
resume hooks because no existing callers (dm-stats or dm-thinp) need
to.  But obviously dm_resume() does need to so it passes @resume_targets
as true to __dm_resume().

I'm not trying to suggest there is a bug or bugs in this new code (you
already pointed out the locking issue across ioctls that I fixed).

But a bug doesn't implicitly mean this is an imperfect way forward --
if/when a bug is found we'll deal with it.. so feel free to pour over
this code to see if there is a bug or bugs.  I really do welcome your
review -- I would just like technical issues to be the focus of any
technical review.