Hi Marek,
2015-08-20 6:15 GMT+09:00 Marek Vasut marex@denx.de:
On Wednesday, August 19, 2015 at 02:02:18 PM, Tom Rini wrote:
On Wed, Aug 19, 2015 at 06:39:29AM +0200, Marek Vasut wrote:
On Tuesday, August 18, 2015 at 07:53:23 PM, Tom Rini wrote:
On Wed, Aug 12, 2015 at 07:31:55AM +0900, Masahiro Yamada wrote:
We have flipped CONFIG_SPL_DISABLE_OF_CONTROL. We have cleansing devices, $(SPL_) and CONFIG_IS_ENABLED(), so we are ready to clear
away the ugly logic in include/fdtdec.h: #ifdef CONFIG_OF_CONTROL # if defined(CONFIG_SPL_BUILD) && !defined(SPL_OF_CONTROL) # define OF_CONTROL 0 # else # define OF_CONTROL 1 # endif #else # define OF_CONTROL 0 #endif
Now CONFIG_IS_ENABLED(OF_CONTROL) is the substitute. It refers to CONFIG_OF_CONTROL for U-boot proper and CONFIG_SPL_OF_CONTROL for SPL.
Also, we no longer have to cancel CONFIG_OF_CONTROL in include/config_uncmd_spl.h and scripts/Makefile.spl.
Signed-off-by: Masahiro Yamada yamada.masahiro@socionext.com Reviewed-by: Tom Rini trini@konsulko.com Reviewed-by: Simon Glass sjg@chromium.org Acked-by: Linus Walleij linus.walleij@linaro.org
(With a slight change due to samsun PR, but applied cleanly after rename) Applied to u-boot/master, thanks!
And this just broke just about every SoCFPGA platform (try ie. socfpga_cyclone5) ;-) I poked into this a bit and noticed that adding #include <linux/kconfig.h> into include/configs/socfpga_common.h fixes the issue, but I'd like to check with you if this is the correct way of fixing it.
Broke it how? $ ./tools/buildman/buildman socfpga_cyclone5 boards.cfg is up to date. Nothing to do. Building current source for 1 boards (1 thread, 6 jobs per thread) 1 0 0 /1 socfpga_cyclone5
Looks like I wasn't really accurate and mainline actually managed to avoid this breakage. I only triggered it with a custom patch, which enables env in SPI NOR (see the spit below).
I suspect the problem is caused by tools/envcrc.c, which does include the board config file (which uses this CONFIG_IS_ENABLED()), but does not include a header file which defines it.
The question therefore is, shall we not used CONFIG_IS_ENABLED() in board config files or shall we include linux/kconfig.h in tools/envcrc.c ?
Our final goal is: We should not use any CONFIG_* in host tools. Actually, some host tools in U-boot depend on the board configuration, which is weird. If we could rip off CONFIG_* from them, neither CONFIG_IS_ENABLED() nor linux/kconfig.h is needed for host tools. I believe this is the right thing to do. Any volunteer?
What I can suggest now is: Do not use CONFIG_IS_ENABLED() in board config headers.
include/config/microblaze-generic.h does this:
#if !defined(CONFIG_OF_CONTROL) || \ (defined(CONFIG_SPL_BUILD) && !defined(CONFIG_SPL_OF_CONTROL)) /* ddr sdram - main memory */ # define CONFIG_SYS_SDRAM_BASE XILINX_RAM_START # define CONFIG_SYS_SDRAM_SIZE XILINX_RAM_SIZE #endif
This is indeed ugly, but works...