Re: [RFC v2 4/4] mm: fallback for offline nodes in alloc_pages_node

[Vlastimil Babka]

On 07/25/2015 01:06 AM, David Rientjes wrote:
> On Fri, 24 Jul 2015, Vlastimil Babka wrote:
>
> > Because I didn't think you would suggest the "nid = numa_mem_id()" for
> > !node_online(nid) fixup would happen only for CONFIG_DEBUG_VM kernels. But it
> > seems that you do suggest that? I would understand if the fixup (correcting an
> > offline node to some that's online) was done regardless of DEBUG_VM, and
> > DEBUG_VM just switched between silent and noisy fixup. But having a debug option
> > alter the outcome seems wrong?
>
> Hmm, not sure why this is surprising, I don't expect people to deploy
> production kernels with CONFIG_DEBUG_VM enabled, it's far too expensive.
> I was expecting they would enable it for, well... debug :)

But is there any other place that does such thing for debug builds?

> In that case, if nid is a valid node but offline, then the nid =
> numa_mem_id() fixup seems fine to allow the kernel to continue debugging.
>
> When a node is offlined as a result of memory hotplug, the pgdat doesn't
> get freed so it can be onlined later.  Thus, alloc_pages_node() with an
> offline node and !CONFIG_DEBUG_VM may not panic.  If it does, this can
> probably be removed because we're covered.

I've checked, but can't say I understand the hotplug code completely... 
but it seems there are two cases
- the node was never online, and the nid passed to alloc_pages_node() is 
clearly bogus. Then there's no pgdat and it should crash on NULL pointer 
dereference. VM_WARN_ON() in __alloc_pages_node() will already catch 
this and provide more details as to what caused the crash. Fixup would 
allow "continue debugging", but it seems that having configured e.g. a 
crashdump to inspect is a better way to debug than letting the kernel 
continue?
- the node has been online in the past, so the nid pointing to an 
offline node might be due to a race with offlining. It shouldn't crash, 
and most likely the zonelist that is left untouched by the offlining 
(AFAICS) will allow fallback to other nodes. Unless there is a nodemask 
of __GFP_THIS_NODE, in which case allocation fails. Again, VM_WARN_ON() 
in __alloc_pages_node() will warn us already. I doubt the fixup is 
needed here?

So I would just drop this patch. We already have the debug warning in 
__alloc_pages_node(), and a fixup is imho just confusing.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/