Tom Lane wrote: > > "Recently dead" means "still live to somebody", so those tids better not > change either. But I don't think that's what he meant. I'm more > worried about the deadlock possibilities inherent in trying to upgrade a > buffer lock. We do not have deadlock detection for LWLocks.
I am guessing he is going to have to release the lock, then ask for an exclusive one.
Yes, thats what is done. Since we try to prune the HOT-update chain even in the SELECT path, we upgrade the lock only if we are sure that there is atleast one tuple that can be removed from the chain or the root needs to be fixed (broken ctid chain for some reason).
