[PATCH 0/7] [RFC] Sizing zones and holes in an architecture independent manner V2

Mel Gorman mel at csn.ul.ie
Thu Apr 13 19:52:08 EST 2006


On (13/04/06 00:20), Mel Gorman didst pronounce:
> This is V2 of the patchset. They have been boot tested on x86, ppc64
> and x86_64 but I still need to do a double check that zones are the
> same size before and after the patch on all arches. IA64 passed a
> basic compile-test. a driver program that fed in the values generated
> by IA64 to add_active_range(), zone_present_pages_in_node() and
> zone_absent_pages_in_node() seemed to generate expected values.

I didn't look at the test program output carefully enough! There was a
double counting of some holes because of a missing "if" - obvious in the
morning. Fix is this (applies on top of the debugging patch)

diff -rup -X /usr/src/patchset-0.5/bin//dontdiff linux-2.6.17-rc1-zonesizing-v6/mm/mem_init.c linux-2.6.17-rc1-zonesizing-v7/mm/mem_init.c
--- linux-2.6.17-rc1-zonesizing-v6/mm/mem_init.c	2006-04-13 10:30:50.000000000 +0100
+++ linux-2.6.17-rc1-zonesizing-v7/mm/mem_init.c	2006-04-13 10:37:11.000000000 +0100
@@ -761,9 +761,11 @@ unsigned long __init zone_absent_pages_i
 		}
 
 		/* Update the hole size cound and move on */
-		hole_pages += start_pfn - prev_end_pfn;
-		printk("Hole found index %d: %lu -> %lu\n",
-				i, prev_end_pfn, start_pfn);
+		if (start_pfn > arch_zone_lowest_possible_pfn[zone_type]) {
+			hole_pages += start_pfn - prev_end_pfn;
+			printk("Hole found index %d: %lu -> %lu\n",
+					i, prev_end_pfn, start_pfn);
+		}
 		prev_end_pfn = early_node_map[i].end_pfn;
 	}
 



More information about the Linuxppc-dev mailing list