On Sun, Jul 12, 2020 at 7:53 PM Joel Fernandes joel@joelfernandes.org wrote:
But I do feel like you figured out why the bug happened, now we're just discussing whether the patch is the right thing to do.
Yes.
Maybe saying "doing the pmd copies for the initial stack isn't important, so let's just note this as a special case and get rid of the WARN_ON()" might be an alternative solution.
Personally, I feel it is better to keep the warning just so in the future we can detect any bugs.
I don't disagree, the warning didn't happen to find a bug now, but it did fine a case we might be able to do better.
So now that I feel we understand the issue, and it's not a horrible problem, just a (very hard to trigger) warning, I don't think there's any huge hurry.
I think think I will - for now - change the WARN_ON() to WARN_ON_ONCE() (so that it doesn't floow the logs if somebody triggers this odd special case this malisiously), and add a note about how this happens to the code for posterito.
And if/when you figure out a better way to fix it, we can update the note.
Ok?
Linus