layers: do not try to unmount if not mounted at all on delete

[runcom]

This check has been wrongly removed with #198
The check must stay as it's now part of the Stop/Delete API so reintroduce it back
This fixes #233 and the associated CRI-O issues
This PR + cri-o/cri-o#1910 fully fix the issue

@rhatdan @nalind @sboeuf @chavafg PTAL

I'm going to revendor c/storage in CRI-O to full fix crio after this is merged

Signed-off-by: Antonio Murdaca [email protected]

[rhatdan]

I was attempting to look at Unmount to see if we should fail safely in there, but I think it gets a little complicated. So this patch LGTM
@vrothberg @nalind @mtrmac PTAL

[TomSweeneyRedHat]

LGTM, definitely want other head nods on this.

I was worried that we could get stuck in the loop here, but the Unmount function looks like it will eventually error out or get the count below zero.

[vrothberg]

Code LGTM but I am picky with commit messages.

@runcom, could you add some description of why the change is necessary and add also link the other issues/PRs?

[runcom]

repushed with something in commit message

@runcom, could you add some description of why the change is necessary and add also link the other issues/PRs?

that's the commit title "layers: do not try to unmount if not mounted at all on delete", that's why it's necessary

[vrothberg]

LGTM, thanks :) (I can't merge here)

that's the commit title "layers: do not try to unmount if not mounted at all on delete", that's why it's necessary

I like the title as it really explains the change which is great for changelogs but I am currently on a crusade against commits without a message (see https://github.com/containers/common-files/blob/master/CONTRIBUTING.md#describe-your-changes-in-commit-messages). Github has turned most git histories into a very bad shape as most information is spread over PRs and issues.

[runcom]

I like the title as it really explains the change which is great for changelogs but I am currently on a crusade against commits without a message (see https://github.com/containers/common-files/blob/master/CONTRIBUTING.md#describe-your-changes-in-commit-messages). Github has turned most git histories into a very bad shape as most information is spread over PRs and issues.

I concur with that yeah, kata-containers have a commit verifier which also makes sure that the body of a commit is not empty and there's always an associated issue that a given commit is fixing so track is kept. Maybe one day we'll have something like that as well :)

[vrothberg]

@TomSweeneyRedHat and I were philosophizing about Github PR templates for all github.com/containers projects to link to https://github.com/containers/common-files/blob/master/CONTRIBUTING.md. But a bot sounds really neat.

[rhatdan]

LGTM