[U-Boot] kbuild: wrong version_autogenerated.h

[Andreas Bießmann]

Dear Masahiro Yamada,

I noticed a wrongly generated version_autogenerated.h with my default 
workflow:

---8<---
andreas at andreas-pc % PATH=$AVR32_PATH:$PATH BUILD_NCPUS=2 
BUILD_NBUILDS=6 CROSS_COMPILE=avr32-linux- 
MAKEALL_LOGDIR=/tmp/makeall_logs BUILD_DIR=/tmp/makeall_buildavr32 
./MAKEALL grasshopper
Configuring for grasshopper board...
    text	   data	    bss	    dec	    hex	filename
  105427	   9518	  21388	 136333	  2148d 
/tmp/makeall_buildavr32/grasshopper/u-boot

--------------------- SUMMARY ----------------------------
Boards compiled: 0
----------------------------------------------------------
PATH=$AVR32_PATH:$PATH BUILD_NCPUS=2 BUILD_NBUILDS=6 CROSS_COMPILE= = = 
    10,69s user 2,68s system 198% cpu 6,740 total
andreas at andreas-pc % cat 
/tmp/makeall_buildavr32/grasshopper/include/generated/version_autogenerated.h
#define PLAIN_VERSION ""
#define U_BOOT_VERSION "U-Boot " PLAIN_VERSION
#define CC_VERSION_STRING "avr32-linux-gcc (OSELAS.Toolchain-2011.11.0) 
4.4.7"
#define LD_VERSION_STRING "GNU ld (GNU Binutils) 2.20.1.20100303"
andreas at andreas-pc %
--->8---

However using make directly is working correctly:

---8<---
andreas at andreas-pc % PATH=$AVR32_PATH:$PATH  CROSS_COMPILE=avr32-linux- 
make -s O=/tmp/build_grasshopper grasshopper_config all
Configuring for grasshopper board...
PATH=$AVR32_PATH:$PATH CROSS_COMPILE=avr32-linux- make -s 
grasshopper_config  9,84s user 2,66s system 81% cpu 15,406 total
andreas at andreas-pc % cat 
/tmp/build_grasshopper/include/generated/version_autogenerated.h
#define PLAIN_VERSION "2014.04-rc1-00277-g180e065"
#define U_BOOT_VERSION "U-Boot " PLAIN_VERSION
#define CC_VERSION_STRING "avr32-linux-gcc (OSELAS.Toolchain-2011.11.0) 
4.4.7"
#define LD_VERSION_STRING "GNU ld (GNU Binutils) 2.20.1.20100303"
andreas at andreas-pc %
--->8---

I haven't seen anything about that on the list, is it a known bug? Is 
MAKEALL still a tool to use or is it outdated?
At the moment I have not the time to dive into that problem, maybe I 
find some minutes next week.

Best regards

Andreas Bießmann