--=_courier-8706-1047952301-0001-2
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
GFP_ZONEMASK is set up as 0xf, meaning the low four bits specify the
zone type. As there are only 3 zone types (DMA, NORMAL, & HIGHMEM), and
only 2 of them (DMA & HIGHMEM) have flags (NORMAL is the default), this
is wrong. This simple patch changes one comment, and changes the value
of GFP_ZONEMASK from 0xf to 0x3.
I'm not sure if this was specified this way to allow for future
expansion, or what. If so, please ignore this. If anyone can tell me
why this is set to four bits, I'm all ears.
Cheers!
-Matt
--=_courier-8706-1047952301-0001-2
Content-Type: text/plain; name="GFP_ZONEMASK_fix-2.5.65.patch"; charset=iso-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
filename="GFP_ZONEMASK_fix-2.5.65.patch"
diff -Nur --exclude-from=/usr/src/.dontdiff linux-2.5.64-vanilla/include/linux/gfp.h linux-2.5.64-gfp_zonemask_fix/include/linux/gfp.h
--- linux-2.5.64-vanilla/include/linux/gfp.h Tue Mar 4 19:29:03 2003
+++ linux-2.5.64-gfp_zonemask_fix/include/linux/gfp.h Mon Mar 17 14:16:28 2003
@@ -7,7 +7,7 @@
/*
* GFP bitmasks..
*/
-/* Zone modifiers in GFP_ZONEMASK (see linux/mmzone.h - low four bits) */
+/* Zone modifiers in GFP_ZONEMASK (see linux/mmzone.h - low two bits) */
#define __GFP_DMA 0x01
#define __GFP_HIGHMEM 0x02
diff -Nur --exclude-from=/usr/src/.dontdiff linux-2.5.64-vanilla/include/linux/mmzone.h linux-2.5.64-gfp_zonemask_fix/include/linux/mmzone.h
--- linux-2.5.64-vanilla/include/linux/mmzone.h Tue Mar 4 19:29:22 2003
+++ linux-2.5.64-gfp_zonemask_fix/include/linux/mmzone.h Mon Mar 17 14:16:28 2003
@@ -162,7 +162,7 @@
struct zone *zones[MAX_NUMNODES * MAX_NR_ZONES + 1]; // NULL delimited
};
-#define GFP_ZONEMASK 0x0f
+#define GFP_ZONEMASK 0x03
/*
* The pg_data_t structure is used in machines with CONFIG_DISCONTIGMEM
--=_courier-8706-1047952301-0001-2--