diff --git a/[refs] b/[refs]
index e07e65c6f680..18d053a2e864 100644
--- a/[refs]
+++ b/[refs]
@@ -1,2 +1,2 @@
 ---
-refs/heads/master: 07808b74e7dab1aa385e698795875337d72daf7d
+refs/heads/master: 47492d3667ec519172ab978bd8231b8c7152fa9d
diff --git a/trunk/arch/x86_64/kernel/pci-gart.c b/trunk/arch/x86_64/kernel/pci-gart.c
index 88be97c96987..2e28e855ec3c 100644
--- a/trunk/arch/x86_64/kernel/pci-gart.c
+++ b/trunk/arch/x86_64/kernel/pci-gart.c
@@ -220,6 +220,12 @@ dma_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_handle,
 	   uses the normal dma_mask for alloc_coherent. */
 	dma_mask &= *dev->dma_mask;
 
+	/* Why <=? Even when the mask is smaller than 4GB it is often larger 
+	   than 16MB and in this case we have a chance of finding fitting memory 
+	   in the next higher zone first. If not retry with true GFP_DMA. -AK */
+	if (dma_mask <= 0xffffffff)
+		gfp |= GFP_DMA32;
+
  again:
 	memory = dma_alloc_pages(dev, gfp, get_order(size));
 	if (memory == NULL)
@@ -245,7 +251,7 @@ dma_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_handle,
 				}
 
 				if (!(gfp & GFP_DMA)) { 
-					gfp |= GFP_DMA; 
+					gfp = (gfp & ~GFP_DMA32) | GFP_DMA;
 					goto again;
 				}
 				return NULL;