On Mon 11-06-18 11:23:18, Xie XiuQi wrote:
> Hi Michal,
>
> On 2018/6/7 20:21, Michal Hocko wrote:
> > On Thu 07-06-18 19:55:53, Hanjun Guo wrote:
> >> On 2018/6/7 18:55, Michal Hocko wrote:
> > [...]
> >>> I am not sure I have the full context but pci_acpi_scan_root calls
> >>> kzalloc_node(sizeof(*info), GFP_KERNEL, node)
> >>> and that should fall back to whatever node that is online. Offline node
> >>> shouldn't keep any pages behind. So there must be something else going
> >>> on here and the patch is not the right way to handle it. What does
> >>> faddr2line __alloc_pages_nodemask+0xf0 tells on this kernel?
> >>
> >> The whole context is:
> >>
> >> The system is booted with a NUMA node has no memory attaching to it
> >> (memory-less NUMA node), also with NR_CPUS less than CPUs presented
> >> in MADT, so CPUs on this memory-less node are not brought up, and
> >> this NUMA node will not be online (but SRAT presents this NUMA node);
> >>
> >> Devices attaching to this NUMA node such as PCI host bridge still
> >> return the valid NUMA node via _PXM, but actually that valid NUMA node
> >> is not online which lead to this issue.
> >
> > But we should have other numa nodes on the zonelists so the allocator
> > should fall back to other node. If the zonelist is not intiailized
> > properly, though, then this can indeed show up as a problem. Knowing
> > which exact place has blown up would help get a better picture...
> >
>
> I specific a non-exist node to allocate memory using kzalloc_node,
> and got this following error message.
>
> And I found out there is just a VM_WARN, but it does not prevent the memory
> allocation continue.
>
> This nid would be use to access NODE_DADA(nid), so if nid is invalid,
> it would cause oops here.
>
> 459 /*
> 460 * Allocate pages, preferring the node given as nid. The node must be valid and
> 461 * online. For more general interface, see alloc_pages_node().
> 462 */
> 463 static inline struct page *
> 464 __alloc_pages_node(int nid, gfp_t gfp_mask, unsigned int order)
> 465 {
> 466 VM_BUG_ON(nid < 0 || nid >= MAX_NUMNODES);
> 467 VM_WARN_ON(!node_online(nid));
> 468
> 469 return __alloc_pages(gfp_mask, order, nid);
> 470 }
> 471
>
> (I wrote a ko, to allocate memory on a non-exist node using kzalloc_node().)
OK, so this is an artificialy broken code, right. You shouldn't get a
non-existent node via standard APIs AFAICS. The original report was
about an existing node which is offline AFAIU. That would be a different
case. If I am missing something and there are legitimate users that try
to allocate from non-existing nodes then we should handle that in
node_zonelist.
[...]
--
Michal Hocko
SUSE Labs
