On Sun, May 06, 2018 at 05:37:25PM +1000, Nicholas Piggin wrote: > The radix fault handler will unmap a page table if it wants to install > a huge page there instead. It will just clear that page table entry, > free the page, and flush the pwc. This is a pretty inadequate commit description. You are describing what the code does before your patch, but you don't say why that is bad or what your patch does and why that is better. In particular, I don't see how we could ever get the situation where we are replacing a page table page with a huge page and there are any valid leaf PTEs in (or below) that page table page. I accept that I am leaking the page table pages but I don't see why tlbies would be necessary. What scenario did you have in mind where we would need to do the full unmap, that is, where would would find any valid leaf PTEs in a page table page that we are replacing? Paul. -- To unsubscribe from this list: send the line "unsubscribe kvm-ppc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
