On Tuesday, 2 March 2021 11:41:52 PM AEDT Jason Gunthorpe wrote:

quoted

However try_to_protect() scans the PTEs again under the PTL so checking 

the 

quoted

mapping of interest actually gets replaced during the rmap walk seems like 

a 

quoted

reasonable solution. Thanks for the comments.

It does seem cleaner if you can manage it, the notifier will still be
needd to program the HW though

Checking during the rmap walk wasn't hard but ultimately pointless. As you say 
a range notifier and lock is required to program the hardware, which requires 
checking the mappings with a mmu notifier sequence anyway.

 - Alistair