Re: [GIT PULL] MM updates for 6.19-rc1

[Linus Torvalds]

On Fri, 5 Dec 2025 at 18:54, Jason Gunthorpe <jgg@xxxxxxxxxx> wrote:
>
> Reverting it now works well enough too. Let's see if Ankit can prepare
> a new revision and perhaps Alex can forward it next week or very early
> rc1 after LPC.

So adding that memory poisoning logic at mmap time looks crazy to me.
The mmap has *nothing* to do with memory poisoning. The memory is
there whether it is mapped or not, and admittedly it's harder to
_access_ when not mapped, but it still has nothing to do with mmap.

As far as I can tell, that memory poisoning logic should be
added/removed when the device is registered or unregistered, and there
is nothing about mmap that should have any effect what-so-ever.

                     Linus