There’s many ways to point out the issues with the patch without being a jerk.

Yes, if you don't mind pointing out again those exact same issues again, because the same person (or potentially someone else) did the same mistake again, as they failed to understand the gravity of the issue again. And again, again, again.

...or alternatively you give the person a good smacking. That's what Torvalds did, while pointing out those issues again. Carrot and stick

maybe there could’ve been more useful conversations about the concerns (re: tar) that were brought up in the previous message.

Likely not - that tar example was brought to highlight that Torvalds' suggestion would cause a regression; that's it. The discussion itself reached a dead end, the solution wouldn't be to keep the conversation about that, but someone submitting a patch that would neither cause said regression nor misuse the VSF functions.