MEN-3974

[oleorhagen]

No description provided.

[lluiscampos]

My concern is that this segfault can happen with production code, not only with this test. The test is anyway skipped in the Debian build with a custom patch, so this PR is yet another workaround then.

Have we figure out something in the underlying code? When I debugged it it was very strange: conn was actually nil, and that should not be possible!

[oleorhagen]

My concern is that this segfault can happen with production code, not only with this test. The test is anyway skipped in the Debian build with a custom patch, so this PR is yet another workaround then.

I see. The second commit can probably be skipped then.

Have we figure out something in the underlying code? When I debugged it it was very strange: conn was actually nil, and that should not be possible!

I was worried at first too, but:

It was because the error was not checked. If err != nil => conn = nil. So as long as our code does it's error checking properly, this is not an issue.

I added this too the tests also, as an extra: b0c8095

[oleorhagen]

Update was simply gofmt.

[lluiscampos]

I was worried at first too, but:

It was because the error was not checked. If err != nil => conn = nil. So as long as our code does it's error checking properly, this is not an issue.

That makes sense, thanks.

[oleorhagen]

@lluiscampos regarding the second commit. Leave it, or remove?

[lluiscampos]

@oleorhagen Leave it, it is harmless. I'll remove the other patch from Debian in the next update.