On Thu, 2020-10-29 at 10:12 +0200, Mike Rapoport wrote: > This series goal was primarily to separate dependincies and make it > clearer what DEBUG_PAGEALLOC and what SET_DIRECT_MAP are. As it > turned > out, there is also some lack of consistency between architectures > that > implement either of this so I tried to improve this as well. > > Honestly, I don't know if a thread can be paused at the time > __vunmap() > left invalid pages, but it could, there is an issue on arm64 with > DEBUG_PAGEALLOC=n and this set fixes it. Ah, ok. So from this and the other thread, this is about the logic in arm's cpa for when it will try the un/map operations. I think the logic actually works currently. And this series introduces a problem on ARM similar to the one you are saying preexists. I put the details in the other thread.
