mm/memory_hotplug: Fix updating the node span

We recently started updating the node span based on the zone span to avoid touching uninitialized memmaps.

Currently, we will always detect the node span to start at 0, meaning a node can easily span too many pages. pgdat_is_empty() will still work correctly if all zones span no pages. We should skip over all zones without spanned pages and properly handle the first detected zone that spans pages.

Unfortunately, in contrast to the zone span (/proc/zoneinfo), the node span cannot easily be inspected and tested. The node span gives no real guarantees when an architecture supports memory hotplug, meaning it can easily contain holes or span pages of different nodes.

The node span is not really used after init on architectures that support memory hotplug. E.g., we use it in mm/memory_hotplug.c:try_offline_node() and in mm/kmemleak.c:kmemleak_scan(). These users seem to be fine.

---

Luckily, this patch made me realize that try_offline_node() is completely broken:
- We easily touch the (garbage) memmap of offline sections
- We will not properly handle the case where we have different NIDs within a single section
This needs proper fixing. We will have to look at the memory block nid of offline memory blocks and scan all pages (or rather pageblocks) of online memory blocks.

---
 mm/memory_hotplug.c | 8 ++++++++
 1 file changed, 8 insertions(+)

index 561371ead39a..0140c20837b6 100644
@@ -442,6 +442,14 @@ static void update_pgdat_span(struct pglist_data *pgdat)
 					     zone->spanned_pages;
 
 		/* No need to lock the zones, they can't change. */
+		if (!zone->spanned_pages)
+			continue;
+		if (!node_end_pfn) {
+			node_start_pfn = zone->zone_start_pfn;
+			node_end_pfn = zone_end_pfn;
+			continue;
+		}
+
 		if (zone_end_pfn > node_end_pfn)
 			node_end_pfn = zone_end_pfn;
 		if (zone->zone_start_pfn < node_start_pfn)

2.21.0

@Andrew, can we also give this a churn, we should try to get this into 5.4 due to

$ git tag --contains 00d6c019b5bc
[...]
v5.4-rc5

Thanks!

Thanks,

David / dhildenb