Re: [PATCH] gurantee DMA area for alloc_bootmem_low() ver. 2.

[Andi Kleen <ak@suse.de>]

Andrew Morton <akpm@osdl.org> writes:
> > 
> >  To avoid this panic, following patch confirm allocated area, and retry
> >  if it is not in DMA.
> >  I tested this patch on my Tiger 4 and our new server.
> 
> It kills my x86_64 box:


Funny I ran into a similar problem recently. On a multi node x86-64
system when swiotlb is forced (normally those are AMD systems which
use the AMD hardware IOMMU) the bootmem_alloc in swiotlb.c would
allocate from the last node. Why? Because alloc_bootmem just
does for_each_pgdat() and tries each node and the pgdat list
starts with the highest node going down to the lowest.

I just changed the ordering of the pgdat list that made bootmem 
work again.

-Andi

Index: linux/mm/bootmem.c
=================================--- linux.orig/mm/bootmem.c
+++ linux/mm/bootmem.c
@@ -61,9 +61,17 @@ static unsigned long __init init_bootmem
 {
 	bootmem_data_t *bdata = pgdat->bdata;
 	unsigned long mapsize = ((end - start)+7)/8;
+	static struct pglist_data *pgdat_last;
 
-	pgdat->pgdat_next = pgdat_list;
-	pgdat_list = pgdat;
+	pgdat->pgdat_next = NULL;
+	/* Add new nodes last so that bootmem always starts 
+	   searching in the first nodes, not the last ones */
+	if (pgdat_last)
+		pgdat_last->pgdat_next = pgdat;
+	else {
+		pgdat_list = pgdat; 	
+		pgdat_last = pgdat;
+	}
 
 	mapsize = ALIGN(mapsize, sizeof(long));
 	bdata->node_bootmem_map = phys_to_virt(mapstart << PAGE_SHIFT);