Hi Akashi,

On 09/07/18 06:49, AKASHI Takahiro wrote:

On Tue, Jul 03, 2018 at 05:36:24PM +0100, James Morse wrote:

quoted

On 23/06/18 03:20, AKASHI Takahiro wrote:

quoted

Memblock list is another source for usable system memory layout.
A merged new arch_kexec_walk_mem() will walk through either io resource
list or memblock list depending on CONFIG_ARCH_DISCARD_MEMBLOCK so that
arm64, in addition to powerpc, will be able to utilize this generic
function for kexec_file.

quoted

diff --git a/arch/powerpc/kernel/machine_kexec_file_64.c b/arch/powerpc/kernel/machine_kexec_file_64.c
index 0bd23dc789a4..3d4be91786ce 100644
--- a/arch/powerpc/kernel/machine_kexec_file_64.c
+++ b/arch/powerpc/kernel/machine_kexec_file_64.c

quoted

quoted

diff --git a/kernel/kexec_file.c b/kernel/kexec_file.c
index 63c7ce1c0c3e..563acd1c9a61 100644
--- a/kernel/kexec_file.c
+++ b/kernel/kexec_file.c

@@ -16,6 +16,7 @@
 #include <linux/file.h>
 #include <linux/slab.h>
 #include <linux/kexec.h>
+#include <linux/memblock.h>
 #include <linux/mutex.h>
 #include <linux/list.h>
 #include <linux/fs.h>

@@ -501,6 +502,53 @@ static int locate_mem_hole_callback(struct resource *res, void *arg)
 	return locate_mem_hole_bottom_up(start, end, kbuf);
 }
 
+#if defined(CONFIG_HAVE_MEMBLOCK) && !defined(CONFIG_ARCH_DISCARD_MEMBLOCK)

The only caller is also guarded by these same ifdefs. Can't we remove this and
rely on the compilers dead-code elimination to remove this function when its not
needed?

I don't think we can remove this #ifdef.
"for_each_free_mem_range[_reverse]()" is defined under CONFIG_HAVE_MEMBLOCK
in memblock.h. If some architecture wants to support KEXEC_FILE but
doesn't have HAVE_MEMBLOCK, compiling kexec_file.c will fail.

Ah, I'd missed this, turns out memblock isn't ubiquitous!


Thanks,

James