On Fri 12-02-21 11:42:15, David Hildenbrand wrote: > On 12.02.21 11:33, Michal Hocko wrote: [...] > > I have to digest this but my first impression is that this is more heavy > > weight than it needs to. Pfn walkers should normally obey node range at > > least. The first pfn is usually excluded but I haven't seen real > > We've seen examples where this is not sufficient. Simple example: > > Have your physical memory end within a memory section. Easy via QEMU, just > do a "-m 4000M". The remaining part of the last section has fake/wrong > node/zone info. Does this really matter though. If those pages are reserved then nobody will touch them regardless of their node/zone ids. > Hotplug memory. The node/zone gets resized such that PFN walkers might > stumble over it. > > The basic idea is to make sure that any initialized/"online" pfn belongs to > exactly one node/zone and that the node/zone spans that PFN. Yeah, this sounds like a good idea but what is the poper node for hole between two ranges associated with a different nodes/zones? This will always be a random number. We should have a clear way to tell "do not touch those pages" and PageReserved sounds like a good way to tell that. > > problems with that. The VM_BUG_ON blowing up is really bad but as said > > above we can simply make it less offensive in presence of reserved pages > > as those shouldn't reach that path AFAICS normally. > > Andrea tried tried working around if via PG_reserved pages and it resulted > in quite some ugly code. Andrea also noted that we cannot rely on any random > page walker to do the right think when it comes to messed up node/zone info. I am sorry, I haven't followed previous discussions. Has the removal of the VM_BUG_ON been considered as an immediate workaround? -- Michal Hocko SUSE Labs
