2005-12-13 13:22:17

by Yasunori Goto

Subject: [Patch] Fix calculation of grow_pgdat_span() in mm/memory_hotplug.c

Dave-san.
CC: Andrew-san.

I realized 2.6.15-rc5 still has a bug for memory hotplug.
The calculation for node_spanned_pages at grow_pgdat_span() is
clearly wrong. This is patch for it.

(Please see grow_zone_span() to compare. It is correct.)

Thanks.

Signed-off-by: Yasunori Goto <[email protected]>

Index: zone_reclaim/mm/memory_hotplug.c
===================================================================
--- zone_reclaim.orig/mm/memory_hotplug.c 2005-12-13 21:38:16.000000000 +0900
+++ zone_reclaim/mm/memory_hotplug.c 2005-12-13 21:39:14.000000000 +0900
@@ -104,7 +104,7 @@ static void grow_pgdat_span(struct pglis
pgdat->node_start_pfn = start_pfn;

if (end_pfn > old_pgdat_end_pfn)
- pgdat->node_spanned_pages = end_pfn - pgdat->node_spanned_pages;
+ pgdat->node_spanned_pages = end_pfn - pgdat->node_start_pfn;
}

int online_pages(unsigned long pfn, unsigned long nr_pages)

--
Yasunori Goto

2005-12-13 16:49:39

by Dave Hansen

Subject: Re: [Patch] Fix calculation of grow_pgdat_span() in mm/memory_hotplug.c

On Tue, 2005-12-13 at 22:20 +0900, Yasunori Goto wrote:
> Dave-san.
> CC: Andrew-san.
>
> I realized 2.6.15-rc5 still has a bug for memory hotplug.
> The calculation for node_spanned_pages at grow_pgdat_span() is
> clearly wrong. This is patch for it.

Clearly wrong is the term for it. Thanks for the fix.

-- Dave