ia64: fix build failure caused by memory model changes
authorMike Rapoport <rppt@linux.ibm.com>
Fri, 18 Dec 2020 16:35:50 +0000 (18:35 +0200)
committerMike Rapoport <rppt@linux.ibm.com>
Sun, 17 Jan 2021 11:31:09 +0000 (13:31 +0200)
The change of ia64's default memory model to SPARSEMEM causes defconfig
build to fail:

  CC      kernel/async.o
In file included from include/linux/numa.h:25,
                 from include/linux/async.h:13,
                 from kernel/async.c:47:
arch/ia64/include/asm/sparsemem.h:14:40: warning: "PAGE_SHIFT" is not defined, evaluates to 0 [-Wundef]
   14 | #if ((CONFIG_FORCE_MAX_ZONEORDER - 1 + PAGE_SHIFT) > SECTION_SIZE_BITS)
      |                                        ^~~~~~~~~~
In file included from include/linux/gfp.h:6,
                 from include/linux/xarray.h:14,
                 from include/linux/radix-tree.h:19,
                 from include/linux/idr.h:15,
                 from include/linux/kernfs.h:13,
                 from include/linux/sysfs.h:16,
                 from include/linux/kobject.h:20,
                 from include/linux/energy_model.h:7,
                 from include/linux/device.h:16,
                 from include/linux/async.h:14,
                 from kernel/async.c:47:
include/linux/mmzone.h:1156:2: error: #error Allocator MAX_ORDER exceeds SECTION_SIZE
 1156 | #error Allocator MAX_ORDER exceeds SECTION_SIZE
      |  ^~~~~

The error cause is the missing definition of PAGE_SHIFT in the calculation
of SECTION_SIZE_BITS.

Add include of <asm/page.h> to arch/ia64/include/asm/sparsemem.h to solve
the problem.

Fixes: 214496cb1870 ("ia64: make SPARSEMEM default and disable DISCONTIGMEM")
Reported-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
arch/ia64/include/asm/sparsemem.h

index dd8c166..42ed524 100644 (file)
@@ -3,6 +3,7 @@
 #define _ASM_IA64_SPARSEMEM_H
 
 #ifdef CONFIG_SPARSEMEM
+#include <asm/page.h>
 /*
  * SECTION_SIZE_BITS            2^N: how big each section will be
  * MAX_PHYSMEM_BITS             2^N: how much memory we can have in that space