On 08/14/2013 11:36 PM, Greg Kroah-Hartman wrote:

On Wed, Aug 14, 2013 at 03:14:11AM -0700, Guenter Roeck wrote:

quoted

On 08/14/2013 01:26 AM, Geert Uytterhoeven wrote:

quoted

On Wed, Aug 14, 2013 at 12:36 AM, Guenter Roeck [off-list ref] wrote:

quoted

quoted

quoted

cris/...:
In file included from include/linux/page-flags.h:8:0,
                   from kernel/bounds.c:9:
include/linux/types.h:25:1: error: unknown type name '__kernel_ino_t'
include/linux/types.h:29:1: error: unknown type name '__kernel_off_t'
include/linux/types.h:30:1: error: unknown type name '__kernel_pid_t'
include/linux/types.h:31:1: error: unknown type name '__kernel_daddr_t'
include/linux/types.h:33:1: error: unknown type name '__kernel_suseconds_t'

That one might be tricky (or simple if someone knows what is wrong).

Fixed in v3.5-rc1:

commit 74f077d2a7651409c44bb323471f219a4b0d2aab
Author: Jiri Slaby [off-list ref]
Date:   Mon Apr 2 13:40:17 2012 +0200

      cris: posix_types.h, include asm-generic/posix_types.h

It does fix above error, but then there is another error:

    AS      arch/cris/arch-v10/lib/checksum.o
In file included from <command-line>:4:0:
/home/groeck/src/linux-stable/include/linux/kconfig.h:23:0: error: syntax error in macro parameter list
make[1]: *** [arch/cris/arch-v10/lib/checksum.o] Error 1
make[1]: *** Waiting for unfinished jobs....
    AS      arch/cris/arch-v10/lib/checksumcopy.o
In file included from <command-line>:4:0:
/home/groeck/src/linux-stable/include/linux/kconfig.h:23:0: error: syntax error in macro parameter list
make[1]: *** [arch/cris/arch-v10/lib/checksumcopy.o] Error 1

commit 7b91747d42a1012e3781dd09fa638d113809e3fd
Author: Paul Gortmaker [off-list ref]
Date:   Wed Apr 18 21:58:43 2012 +0200

      cris: Remove old legacy "-traditional" flag from arch-v10/lib/Makefile

Great, thanks for tracking this down.

With both patches applied, we are almost there. Next error:

kernel/built-in.o: In function `core_kernel_data':
(.text+0x12e38): undefined reference to `_sdata'
make: *** [vmlinux] Error 1

This has been fixed with commit 473e162e (CRIS: Add _sdata to vmlinux.lds.S).

So, in summary, we have

473e162e CRIS: Add _sdata to vmlinux.lds.S
69349c2d cris: Remove old legacy "-traditional" flag from arch-v10/lib/Makefile
7b91747d cris: posix_types.h, include asm-generic/posix_types.h

With those three patches applied, cris targets build in 3.4.y.

Wonderful, I've now applied all of these.

Anything else I'm missing for 3.4?

Besides the now failing mips buids;

Building arm:allmodconfig ... failed
--------------
Error log:
.config:25:warning: symbol value '' invalid for PHYS_OFFSET
make[2]: *** [silentoldconfig] Error 1
make[1]: *** [silentoldconfig] Error 2
make: *** No rule to make target `include/config/auto.conf', needed by `include/config/kernel.release'.  Stop.

That was "fixed" by a complete arm Kconfig overhaul. I don't think you want
that in 3.4.
--------------

Building frv:defconfig ... failed
--------------
Error log:
[ ... ]
kernel/fork.c: In function 'free_task':
kernel/fork.c:174:2: error: implicit declaration of function 'free_task_struct' [-Werror=implicit-function-declaration]
kernel/fork.c: In function 'dup_task_struct':
kernel/fork.c:266:2: error: implicit declaration of function 'alloc_task_struct_node' [-Werror=implicit-function-declaration]
kernel/fork.c:266:6: warning: assignment makes pointer from integer without a cast [enabled by default]
cc1: some warnings being treated as errors
make[1]: *** [kernel/fork.o] Error 1

I already sent you a sha with a fix for that.

---------------
Building powerpc:chroma_defconfig ... failed
--------------
Error log:
arch/powerpc/mm/numa.c: In function 'do_init_bootmem':
arch/powerpc/mm/numa.c:529:19: error: 'aa.arrays' may be used uninitialized in this function [-Werror=maybe-uninitialized]
arch/powerpc/mm/numa.c:642:22: note: 'aa.arrays' was declared here
cc1: all warnings being treated as errors
make[1]: *** [arch/powerpc/mm/numa.o] Error 1
make: *** [arch/powerpc/mm] Error 2
make: *** Waiting for unfinished jobs....
--------------

That's all. Assuming you revert the mips patch (MIPS: Rewrite pfn_valid to work in modules, too)
and apply the frv patch, we are down to three build failures out of 64 builds. Not too bad.

I now see an occasional qemu runtime crash on the x86 target on 3.4. It doesn't happen
all the time, so it is hard to determine if it is a new problem or an old one.
I'll try to find out. The crash is seen during reboot in native_nmi_stop_other_cpus(),
if that rings a bell with anyone.

Guenter