On 04/04/2017 06:50 AM, Vijay Kilari wrote:
Hi Julien,
Hello Vijay,
On Mon, Apr 3, 2017 at 3:31 PM, Julien Grall <julien.gr...@arm.com> wrote:
On 28/03/17 13:35, vijay.kil...@gmail.com wrote:
diff --git a/xen/include/asm-arm/arm32/mm.h
b/xen/include/asm-arm/arm32/mm.h
new file mode 100644
index 0000000..e93d9df
--- /dev/null
+++ b/xen/include/asm-arm/arm32/mm.h
@@ -0,0 +1,20 @@
+#ifndef __ARM_ARM32_MM_H__
+#define __ARM_ARM32_MM_H__
+
+/* On ARM only xenheap memory is directly mapped. Hence return false. */
By reading this comment some people will wonder why you don't check whether
the mfn is in xenheap then. As mentioned above, I am ok if you always return
false here. But you need to explain why.
Is this ok?
"On ARM32, all the RAM is not mapped by Xen, instead it is mapped by xenheap.
So DIRECTMAP_VIRT region is not mapped.
Hence we return always false when mfn is checked on DIRECTMAP_VIRT region."
This does not make any sense. You don't map the RAM using xenheap,
xenheap is the only portion of the RAM always mapped. Furthermore, as I
said there are no "DIRECTMAP_VIRT" region on ARM, have a grep at this
keyword for ARM32...
How about:
"Only a limited amount of RAM, called xenheap, is always mapped on
ARM32. For convenience always return false."
Cheers,
--
Julien Grall
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
