kconfig: turnaround into single .config
Subsystems (10)
v3 cover kconfig: turnaround into single .config
Masahiro Yamada (6):
ARM: UniPhier: set CONFIG_SYS_MALLOC_F to the global default value
malloc_f: fix broken .config caused by CONFIG_SYS_MALLOC_F
malloc_f: add ARCH_MALLOC_F_LEN to specify SoC-default malloc_f_len
kconfig: switch to single .config configuration
kconfig: remove unneeded dependency on !SPL_BUILD
malloc_f: enable SYS_MALLOC_F by default if DM is on
Simon Glass (1):
kconfig: Adjust ordering so that defaults work as expected
Kconfig | 32 +--- arch/arm/Kconfig | 7 +- arch/arm/cpu/arm1176/bcm2835/Kconfig | 6 +- arch/arm/cpu/armv7/exynos/Kconfig | 30 ++-- arch/arm/cpu/armv7/omap3/Kconfig | 12 +- arch/arm/cpu/armv7/s5pc1xx/Kconfig | 4 +- arch/arm/cpu/armv7/tegra-common/Kconfig | 20 ++- arch/arm/cpu/armv7/uniphier/Kconfig | 8 - arch/x86/Kconfig | 6 +- board/amcc/canyonlands/Kconfig | 8 - board/compulab/cm_t335/Kconfig | 6 +- board/gumstix/pepper/Kconfig | 6 +- board/isee/igep0033/Kconfig | 6 +- board/phytec/pcm051/Kconfig | 6 +- board/samsung/goni/Kconfig | 6 +- board/samsung/smdkc100/Kconfig | 6 +- board/silica/pengwyn/Kconfig | 6 +- board/ti/am335x/Kconfig | 12 +- common/Kconfig | 1 - config.mk | 5 + configs/ph1_ld4_defconfig | 2 +- configs/ph1_pro4_defconfig | 2 +- configs/ph1_sld8_defconfig | 2 +- doc/README.kconfig | 132 ++++------------ drivers/core/Kconfig | 16 +- drivers/mtd/nand/Kconfig | 6 +- dts/Kconfig | 1 - include/config_uncmd_spl.h | 13 ++ scripts/Makefile.autoconf | 36 +++-- scripts/Makefile.build | 3 +- scripts/Makefile.spl | 10 +- scripts/Makefile.uncmd_spl | 18 +++ scripts/multiconfig.sh | 261 +------------------------------- 33 files changed, 183 insertions(+), 512 deletions(-) create mode 100644 scripts/Makefile.uncmd_spl
Signature · 1 line
1.9.1
PATCH v3 1/7 ARM: UniPhier: set CONFIG_SYS_MALLOC_F to the global default value
It is true that malloc is necessary for Driver Model before relocation, but there is no good reason to reserve the malloc space more than enough. The default value 0x400 works well.
---
Changes in v3: None
Changes in v2: None
arch/arm/cpu/armv7/uniphier/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/cpu/armv7/uniphier/Kconfig +1 -1
@@ -52,7 +52,7 @@ config SYS_MALLOC_F default y config SYS_MALLOC_F_LEN - default 0x2000 + default 0x400 config CMD_PINMON bool "Enable boot mode pins monitor command"
Signature · 1 line
1.9.1
PATCH v3 2/7 malloc_f: fix broken .config caused by CONFIG_SYS_MALLOC_F
In reply to Masahiro YamadaSince commit b724bd7d6349 (dm: Kconfig: Move CONFIG_SYS_MALLOC_F_LEN to Kconfig), the ".config" created by the configuration has been wrong.
For example, the following is a snippet of the ".config" generated by "make beaver_defconfig":
--------------->8-----------------
CONFIG_CC_OPTIMIZE_FOR_SIZE=y
# CONFIG_SYS_MALLOC_F is not set
CONFIG_SYS_MALLOC_F_LEN=0x1800
# CONFIG_EXPERT is not set
---------------8<-----------------
CONFIG_SYS_MALLOC_F_LEN is supposed to depend on CONFIG_SYS_MALLOC_F (see the top level Kconfig), but the ".config" above is not actually following that dependency.
This is caused by two mistakes of commit b724bd7d6349.
[1] Wrong default value of CONFIG_SYS_MALLOC_F CONFIG_SYS_MALLOC_F is a boolean option, but its default value is set to 0x400.
[2] Missing "if SYS_MALLOC_F" in the default setting in each Kconfig For example, arch/arm/cpu/armv7/tegra-common/Kconfig has the line "default 0x1800" for SYS_MALLOC_F_LEN. It must be described as "default 0x1800 if SYS_MALLOC_F" to follow the dependency.
Those two bugs together create such a broken ".config".
Unfortunately, even if we correct both [1] and [2], the value of CONFIG_SYS_MALLOC_F_LEN is not set as we expect.
The "default 0x1800 if SYS_MALLOC_F" would be simply ignored because the "default 0x400" in the top level Kconfig is parsed first.
Notice that if multiple default lines appear for the same CONFIG, the first one takes precedence.
This commit corrects [1] and [2] so as not to create a broken .config file in any cases. Note that the default values in arch/arm/cpu/armv7/tegra-common/Kconfig and arch/x86/Kconfig are not working at this moment. This will be solved by the next commit.
The default value 0x400 is redundant for OMAP, Exynos, UniPhier, etc. They can be simply removed.
There are still redundant "CONFIG_SYS_MALLOC_F_LEN=0x400" in many defconfig files, but this commit is not touching them.
---
Changes in v3:
- Drop comments "This is meaningless ..."
- Update commit description
Changes in v2: None
Kconfig | 1 - arch/arm/cpu/armv7/exynos/Kconfig | 3 --- arch/arm/cpu/armv7/omap3/Kconfig | 3 --- arch/arm/cpu/armv7/tegra-common/Kconfig | 2 +- arch/arm/cpu/armv7/uniphier/Kconfig | 3 --- arch/x86/Kconfig | 2 +- board/amcc/canyonlands/Kconfig | 4 ---- board/ti/am335x/Kconfig | 3 --- 8 files changed, 2 insertions(+), 19 deletions(-)
Kconfig +0 -1
@@ -58,7 +58,6 @@ config CC_OPTIMIZE_FOR_SIZE config SYS_MALLOC_F bool "Enable malloc() pool before relocation" - default 0x400 help Before relocation memory is very limited on many platforms. Still, we can provide a small malloc() pool if needed. Driver model in
arch/arm/cpu/armv7/exynos/Kconfig +0 -3
@@ -83,9 +83,6 @@ config DM_GPIO config SYS_MALLOC_F default y if !SPL_BUILD -config SYS_MALLOC_F_LEN - default 0x400 if !SPL_BUILD - source "board/samsung/smdkv310/Kconfig" source "board/samsung/trats/Kconfig" source "board/samsung/universal_c210/Kconfig"
arch/arm/cpu/armv7/omap3/Kconfig +0 -3
@@ -105,9 +105,6 @@ config DM_SERIAL config SYS_MALLOC_F default y if DM && !SPL_BUILD -config SYS_MALLOC_F_LEN - default 0x400 if DM && !SPL_BUILD - config SYS_SOC default "omap3"
arch/arm/cpu/armv7/tegra-common/Kconfig +1 -1
@@ -21,7 +21,7 @@ config SYS_MALLOC_F default y config SYS_MALLOC_F_LEN - default 0x1800 + default 0x1800 if SYS_MALLOC_F config USE_PRIVATE_LIBGCC default y if SPL_BUILD
arch/arm/cpu/armv7/uniphier/Kconfig +0 -3
@@ -51,9 +51,6 @@ endchoice config SYS_MALLOC_F default y -config SYS_MALLOC_F_LEN - default 0x400 - config CMD_PINMON bool "Enable boot mode pins monitor command" default y
arch/x86/Kconfig +1 -1
@@ -80,7 +80,7 @@ config SYS_MALLOC_F default y config SYS_MALLOC_F_LEN - default 0x800 + default 0x800 if SYS_MALLOC_F config RAMBASE hex
board/amcc/canyonlands/Kconfig +0 -4
@@ -43,8 +43,4 @@ config SYS_MALLOC_F bool default y -config SYS_MALLOC_F_LEN - hex - default 0x400 - endif
board/ti/am335x/Kconfig +0 -3
@@ -50,7 +50,4 @@ config DM_SERIAL config SYS_MALLOC_F default y if DM && !SPL_BUILD -config SYS_MALLOC_F_LEN - default 0x400 if DM && !SPL_BUILD - endif
Signature · 1 line
1.9.1
PATCH v3 2/7 malloc_f: fix broken .config caused by CONFIG_SYS_MALLOC_F
Hi Masahiro,
Quoted 50 lines · Masahiro Yamada
Since commit
b724bd7d6349(dm:Kconfig: MoveCONFIG_SYS_MALLOC_F_LENtoKconfig), the ".config" created by the configuration has been wrong.For example, the following is a snippet of the ".config" generated by "make
beaver_defconfig":--------------->8-----------------
CONFIG_CC_OPTIMIZE_FOR_SIZE=y
#CONFIG_SYS_MALLOC_Fis not set
CONFIG_SYS_MALLOC_F_LEN=0x1800
#CONFIG_EXPERTis not set
---------------8<-----------------
CONFIG_SYS_MALLOC_F_LENis supposed to depend onCONFIG_SYS_MALLOC_F(see the top levelKconfig), but the ".config" above is not actually following that dependency.This is caused by two mistakes of commit
b724bd7d6349.[1] Wrong default value of
CONFIG_SYS_MALLOC_FCONFIG_SYS_MALLOC_Fis a boolean option, but its default value is set to0x400.[2] Missing "if
SYS_MALLOC_F" in the default setting in eachKconfigFor example,arch/arm/cpu/armv7/tegra-common/Kconfighas the line "default0x1800" forSYS_MALLOC_F_LEN. It must be described as "default0x1800ifSYS_MALLOC_F" to follow the dependency.Those two bugs together create such a broken ".config".
Unfortunately, even if we correct both [1] and [2], the value of
CONFIG_SYS_MALLOC_F_LENis not set as we expect.
The "default0x1800ifSYS_MALLOC_F" would be simply ignored because the "default0x400" in the top levelKconfigis parsed first.Notice that if multiple default lines appear for the same CONFIG, the first one takes precedence.
This commit corrects [1] and [2] so as not to create a broken .config file in any cases. Note that the default values in
arch/arm/cpu/armv7/tegra-common/Kconfigandarch/x86/Kconfigare not working at this moment. This will be solved by the next commit.The default value
0x400is redundant for OMAP, Exynos, UniPhier, etc. They can be simply removed.There are still redundant "
CONFIG_SYS_MALLOC_F_LEN=0x400" in many defconfig files, but this commit is not touching them.
Quoted 8 lines
Signed-off-by Masahiro Yamada <[email protected]>---
Changes in v3:
- Drop comments "This is meaningless ..."
- Update commit descriptionChanges in v2: None
I think it might have been better for this patch to merge with patch 4, since patch 3 fixes one of these problem. But it's not important.
PATCH v3 3/7 kconfig: Adjust ordering so that defaults work as expected
In reply to Masahiro YamadaFrom: Simon Glass <[email protected]>
At present defaults in arch-specific Kconfig files are ignored if the top-level item comes ahead of it in include order. This means that it is not possible to have a U-Boot default that architectures and boards can override. This does not seem very useful.
Move the include earlier to support this.
---
Changes in v3:
- The same as what Simon posted
http://patchwork.ozlabs.org/patch/441670/
Included in this series to clarify the order
Changes in v2: None
Kconfig | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
Kconfig +3 -2
@@ -12,6 +12,9 @@ config KCONFIG_OBJDIR string option env="KCONFIG_OBJDIR" +# Allow defaults in arch-specific code to override any given here +source "arch/Kconfig" + menu "General setup" config LOCALVERSION
@@ -171,8 +174,6 @@ config SYS_CLK_FREQ endmenu # Boot images -source "arch/Kconfig" - source "common/Kconfig" source "dts/Kconfig"
Signature · 1 line
1.9.1
PATCH v3 4/7 malloc_f: add ARCH_MALLOC_F_LEN to specify SoC-default malloc_f_len
In reply to Masahiro YamadaNow the default value of CONFIG_SYS_MALLOC_F_LEN can be overridden by SoC Kconfig file, but we still have to add the same conditional "if SYS_MALLOC_F" to every default. Otherwise, a broken .config file could be generated. It is too painful.
This commit intends to solve it by introducing SoC-default ARCH_MALLOC_F_LEN. This works as follows:
- If ARCH_MALLOC_F_LEN is defined, it overrides the default of SYS_MALLOC_F_LEN. If you have the SoC-optimized value, you might wish to do this.
- If ARCH_MALLOC_F_LEN is not defined, SYS_MALLOC_F_LEN is default to 0x400. This is the global default and it should work well enough in most cases.
- You can still change SYS_MALLOC_F_LEN per board, although you do not probably have to do so. Per-board default should be saved into each defconfig.
---
Changes in v3:
- Newly added
Changes in v2: None
Kconfig | 1 + arch/arm/cpu/armv7/tegra-common/Kconfig | 5 +++-- arch/x86/Kconfig | 5 +++-- 3 files changed, 7 insertions(+), 4 deletions(-)
Kconfig +1 -0
@@ -70,6 +70,7 @@ config SYS_MALLOC_F config SYS_MALLOC_F_LEN hex "Size of malloc() pool before relocation" depends on SYS_MALLOC_F + default ARCH_MALLOC_F_LEN if ARCH_MALLOC_F_LEN default 0x400 help Before relocation memory is very limited on many platforms. Still,
arch/arm/cpu/armv7/tegra-common/Kconfig +3 -2
@@ -20,8 +20,9 @@ endchoice config SYS_MALLOC_F default y -config SYS_MALLOC_F_LEN - default 0x1800 if SYS_MALLOC_F +config ARCH_MALLOC_F_LEN + hex + default 0x1800 config USE_PRIVATE_LIBGCC default y if SPL_BUILD
arch/x86/Kconfig +3 -2
@@ -79,8 +79,9 @@ config DM_SERIAL config SYS_MALLOC_F default y -config SYS_MALLOC_F_LEN - default 0x800 if SYS_MALLOC_F +config ARCH_MALLOC_F_LEN + hex + default 0x800 config RAMBASE hex
Signature · 1 line
1.9.1
PATCH v3 4/7 malloc_f: add ARCH_MALLOC_F_LEN to specify SoC-default malloc_f_len
Hi Masahiro,
Quoted 16 lines · Masahiro Yamada
Now the default value of
CONFIG_SYS_MALLOC_F_LENcan be overridden by SoCKconfigfile, but we still have to add the same conditional "ifSYS_MALLOC_F" to every default. Otherwise, a broken .config file could be generated. It is too painful.This commit intends to solve it by introducing SoC-default
ARCH_MALLOC_F_LEN. This works as follows:- If
ARCH_MALLOC_F_LENis defined, it overrides the default ofSYS_MALLOC_F_LEN. If you have the SoC-optimized value, you might wish to do this.- If
ARCH_MALLOC_F_LENis not defined,SYS_MALLOC_F_LENis default to0x400. This is the global default and it should work well enough in most cases.
Quoted 6 lines
- You can still change
SYS_MALLOC_F_LENper board, although you do not probably have to do so. Per-board default should be saved into each defconfig.Signed-off-by Masahiro Yamada <[email protected]>---
Seems like a reasonable solution to me. I have to admit I prefer to adding another option, but if that is the only way to avoid the broken Kconfig (which I think works fine but could be quite confusing for people) then so be it.
PATCH v3 4/7 malloc_f: add ARCH_MALLOC_F_LEN to specify SoC-default malloc_f_len
In reply to Masahiro Yamada2015-02-20 14:24 GMT+09:00 Masahiro Yamada <[email protected]>:
Quoted 16 lines
Now the default value of
CONFIG_SYS_MALLOC_F_LENcan be overridden by SoCKconfigfile, but we still have to add the same conditional "ifSYS_MALLOC_F" to every default. Otherwise, a broken .config file could be generated. It is too painful.This commit intends to solve it by introducing SoC-default
ARCH_MALLOC_F_LEN. This works as follows:- If
ARCH_MALLOC_F_LENis defined, it overrides the default ofSYS_MALLOC_F_LEN. If you have the SoC-optimized value, you might wish to do this.- If
ARCH_MALLOC_F_LENis not defined,SYS_MALLOC_F_LENis default to0x400. This is the global default and it should work well enough in most cases.
Quoted 6 lines
- You can still change
SYS_MALLOC_F_LENper board, although you do not probably have to do so. Per-board default should be saved into each defconfig.Signed-off-by Masahiro Yamada <[email protected]>---
Sorry, this patch does not work at all.
I was completely misunderstanding.
I take it back.
Signature · 2 lines
Best Regards Masahiro Yamada
PATCH v3 5/7 kconfig: switch to single .config configuration
In reply to Masahiro YamadaWhen Kconfig for U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.
It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of Kconfig is lost.
[2] It is too painful to change both ".config" and "spl/.config". Sunxi guys started to work around this problem by creating a new configuration target. Commit cbdd9a9737cc (sunxi: kconfig: Add %_felconfig rule to enable FEL build of sunxi platforms.) added "make *_felconfig" to enable CONFIG_SPL_FEL on both images. Changing the configuration of multiple images in one command is a generic demand. The current implementation cannot propose any good solution about this.
[3] Kconfig files are getting ugly and difficult to understand. Commit b724bd7d6349 (dm: Kconfig: Move CONFIG_SYS_MALLOC_F_LEN to Kconfig) has sprinkled "if !SPL_BUILD" over the Kconfig files.
[4] The build system got more complicated than it should be. To adjust Linux-originated Kconfig to U-Boot, the helper script "scripts/multiconfig.sh" was introduced. Writing a complicated text processor is a shell script sometimes caused problems.
Now I believe the "single .config" will serve us better. With it, all the problems above would go away. Instead, we will have to add some CONFIG_SPL_* (and CONFIG_TPL_*) options such as CONFIG_SPL_DM, but we will not have much. Anyway, this is what we do now in scripts/Makefile.spl.
I admit my mistake with my apology and this commit switches to the single .config configuration.
It is not so difficult to do that:
- Remove unnecessary processings from scripts/multiconfig.sh This file will remain for a while to support the current defconfig format. It will be removed after more cleanups are done.
- Adjust some makefiles and Kconfigs
- Add some entries to include/config_uncmd_spl.h and the new file scripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.
- update doc/README.kconfig
More cleaning up patches will follow this.
---
Changes in v3: None
Changes in v2:
- Fix typos in commit description
Kconfig | 14 -- arch/arm/cpu/armv7/tegra-common/Kconfig | 2 +- arch/arm/cpu/armv7/uniphier/Kconfig | 1 - config.mk | 5 + configs/ph1_ld4_defconfig | 2 +- configs/ph1_pro4_defconfig | 2 +- configs/ph1_sld8_defconfig | 2 +- doc/README.kconfig | 132 ++++------------ drivers/mtd/nand/Kconfig | 2 +- include/config_uncmd_spl.h | 9 ++ scripts/Makefile.autoconf | 36 +++-- scripts/Makefile.build | 3 +- scripts/Makefile.spl | 10 +- scripts/Makefile.uncmd_spl | 16 ++ scripts/multiconfig.sh | 261 +------------------------------- 15 files changed, 108 insertions(+), 389 deletions(-) create mode 100644 scripts/Makefile.uncmd_spl
Kconfig +0 -14
@@ -8,10 +8,6 @@ config UBOOTVERSION string option env="UBOOTVERSION" -config KCONFIG_OBJDIR - string - option env="KCONFIG_OBJDIR" - # Allow defaults in arch-specific code to override any given here source "arch/Kconfig"
@@ -90,16 +86,6 @@ endmenu # General setup menu "Boot images" -config SPL_BUILD - bool - depends on $KCONFIG_OBJDIR="spl" || $KCONFIG_OBJDIR="tpl" - default y - -config TPL_BUILD - bool - depends on $KCONFIG_OBJDIR="tpl" - default y - config SUPPORT_SPL bool
arch/arm/cpu/armv7/tegra-common/Kconfig +1 -1
@@ -25,7 +25,7 @@ config ARCH_MALLOC_F_LEN default 0x1800 config USE_PRIVATE_LIBGCC - default y if SPL_BUILD + default y config DM default y if !SPL_BUILD
arch/arm/cpu/armv7/uniphier/Kconfig +0 -1
@@ -68,7 +68,6 @@ config CMD_DDRPHY_DUMP choice prompt "DDR3 Frequency select" - depends on SPL_BUILD config DDR_FREQ_1600 bool "DDR3 1600"
config.mk +5 -0
@@ -24,6 +24,11 @@ VENDOR := ARCH := $(CONFIG_SYS_ARCH:"%"=%) CPU := $(CONFIG_SYS_CPU:"%"=%) +ifdef CONFIG_SPL_BUILD +ifdef CONFIG_TEGRA +CPU := arm720t +endif +endif BOARD := $(CONFIG_SYS_BOARD:"%"=%) ifneq ($(CONFIG_SYS_VENDOR),) VENDOR := $(CONFIG_SYS_VENDOR:"%"=%)
configs/ph1_ld4_defconfig +1 -1
@@ -39,4 +39,4 @@ CONFIG_DM_I2C=y CONFIG_USB=y CONFIG_USB_EHCI_HCD=y CONFIG_USB_STORAGE=y -S:CONFIG_SPL_NAND_DENALI=y +CONFIG_SPL_NAND_DENALI=y
configs/ph1_pro4_defconfig +1 -1
@@ -39,4 +39,4 @@ CONFIG_DM_I2C=y CONFIG_USB=y CONFIG_USB_EHCI_HCD=y CONFIG_USB_STORAGE=y -S:CONFIG_SPL_NAND_DENALI=y +CONFIG_SPL_NAND_DENALI=y
configs/ph1_sld8_defconfig +1 -1
@@ -39,4 +39,4 @@ CONFIG_DM_I2C=y CONFIG_USB=y CONFIG_USB_EHCI_HCD=y CONFIG_USB_STORAGE=y -S:CONFIG_SPL_NAND_DENALI=y +CONFIG_SPL_NAND_DENALI=y
doc/README.kconfig +32 -100
163 lines not on this page · raw message
drivers/mtd/nand/Kconfig +1 -1
@@ -36,7 +36,7 @@ config NAND_DENALI_SPARE_AREA_SKIP_BYTES endif -if SPL_BUILD +if SPL config SPL_NAND_DENALI bool "Support Denali NAND controller for SPL"
include/config_uncmd_spl.h +9 -0
@@ -20,5 +20,14 @@#undef CONFIG_CMD_SNTP#undef CONFIG_CMD_TFTPPUT#undef CONFIG_CMD_TFTPSRV +#undef CONFIG_OF_CONTROL + +#ifndef CONFIG_SPL_DM +#undef CONFIG_DM_SERIAL +#undef CONFIG_DM_GPIO +#undef CONFIG_DM_I2C +#undef CONFIG_DM_SPI +#endif +#endif /* CONFIG_SPL_BUILD */ #endif /* __CONFIG_UNCMD_SPL_H__ */
scripts/Makefile.autoconf +26 -10
@@ -7,9 +7,17 @@ # (= When we move all CONFIGs from header files to Kconfig) # this makefile can be deleted. -# obj is "include" or "spl/include" or "tpl/include" -# for non-SPL, SPL, TPL, respectively -include $(obj)/config/auto.conf +__all: include/autoconf.mk include/autoconf.mk.dep + +ifeq ($(shell grep -q '^CONFIG_SPL=y' include/config/auto.conf 2>/dev/null && echo y),y) +__all: spl/include/autoconf.mk +endif + +ifeq ($(shell grep -q '^CONFIG_TPL=y' include/config/auto.conf 2>/dev/null && echo y),y) +__all: tpl/include/autoconf.mk +endif + +include include/config/auto.conf include scripts/Kbuild.include
@@ -22,7 +30,6 @@ CPP = $(CC) -E include config.mk UBOOTINCLUDE := \ - -I$(obj) \ -Iinclude \ $(if $(KBUILD_SRC), -I$(srctree)/include) \ -I$(srctree)/arch/$(ARCH)/include \
@@ -48,10 +55,10 @@ include/autoconf.mk.dep: FORCE # same CONFIG macros quiet_cmd_autoconf = GEN $@ cmd_autoconf = \ - $(CPP) $(c_flags) -DDO_DEPS_ONLY -dM $(srctree)/include/common.h > [email protected] && { \ + $(CPP) $(c_flags) $2 -DDO_DEPS_ONLY -dM $(srctree)/include/common.h > [email protected] && { \ sed -n -f $(srctree)/tools/scripts/define2mk.sed [email protected] | \ while read line; do \ - if ! grep -q "$${line%=*}=" $(obj)/config/auto.conf; then \ + if ! grep -q "$${line%=*}=" include/config/auto.conf; then \ echo "$$line"; \ fi \ done > $@; \
@@ -60,10 +67,19 @@ quiet_cmd_autoconf = GEN $@ rm [email protected]; false; \ } -$(obj)/autoconf.mk: FORCE +include/autoconf.mk: FORCE $(call cmd,autoconf) -include/autoconf.mk include/autoconf.mk.dep: include/config.h +spl/include/autoconf.mk: FORCE + $(Q)mkdir -p $(dir $@) + $(call cmd,autoconf,-DCONFIG_SPL_BUILD) + +tpl/include/autoconf.mk: FORCE + $(Q)mkdir -p $(dir $@) + $(call cmd,autoconf,-DCONFIG_SPL_BUILD -DCONFIG_TPL_BUILD) + +include/autoconf.mk include/autoconf.mk.dep \ + spl/include/autoconf.mk tpl/include/autoconf.mk: include/config.h # include/config.h # Prior to Kconfig, it was generated by mkconfig. Now it is created here.
@@ -75,10 +91,10 @@ define filechk_config_h done; \ echo \#define CONFIG_BOARDDIR board/$(if $(VENDOR),$(VENDOR)/)$(BOARD);\ echo \#include \<config_defaults.h\>; \ + echo \#include \<config_uncmd_spl.h\>; \ echo \#include \<configs/$(CONFIG_SYS_CONFIG_NAME).h\>; \ echo \#include \<asm/config.h\>; \ - echo \#include \<config_fallbacks.h\>; \ - echo \#include \<config_uncmd_spl.h\>; ) + echo \#include \<config_fallbacks.h\>;) endef include/config.h: scripts/Makefile.autoconf create_symlink FORCE
scripts/Makefile.build +2 -1
@@ -41,8 +41,9 @@ subdir-ccflags-y := # Read auto.conf if it exists, otherwise ignore # Modified for U-Boot --include $(prefix)/include/config/auto.conf +-include include/config/auto.conf -include $(prefix)/include/autoconf.mk +include scripts/Makefile.uncmd_spl include scripts/Kbuild.include
scripts/Makefile.spl +6 -4
@@ -21,13 +21,15 @@ _dummy := $(shell [ -d $(obj) ] || mkdir -p $(obj)) include $(srctree)/scripts/Kbuild.include -UBOOTINCLUDE := -I$(obj)/include $(UBOOTINCLUDE) - --include $(obj)/include/config/auto.conf +-include include/config/auto.conf -include $(obj)/include/autoconf.mk +KBUILD_CPPFLAGS += -DCONFIG_SPL_BUILD +ifeq ($(CONFIG_TPL_BUILD),y) +KBUILD_CPPFLAGS += -DCONFIG_TPL_BUILD +endif + ifeq ($(CONFIG_TPL_BUILD),y) -export CONFIG_TPL_BUILD SPL_BIN := u-boot-tpl else SPL_BIN := u-boot-spl
scripts/Makefile.uncmd_spl new file +16 -0
@@ -0,0 +1,16 @@ +# Makefile version of include/config_uncmd_spl.h +# +# Invent a better way + +ifdef CONFIG_SPL_BUILD +CONFIG_OF_CONTROL= + +ifndef CONFIG_SPL_DM +CONFIG_DM_SERIAL= +CONFIG_DM_GPIO= +CONIFG_DM_I2C= +CONFIG_DM_SPI= +CONFIG_DM_SPI_FLASH= +endif + +endif
scripts/multiconfig.sh +7 -254
330 lines not on this page · raw message
Signature · 1 line
1.9.1
PATCH v3 5/7 kconfig: switch to single .config configuration
Quoted 59 lines · Masahiro Yamada
When
Kconfigfor U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of
Kconfigis lost.[2] It is too painful to change both ".config" and "spl/.config". Sunxi guys started to work around this problem by creating a new configuration target. Commit
cbdd9a9737cc(sunxi: kconfig: Add %_felconfig rule to enable FEL build of sunxi platforms.) added "make*_felconfig" to enableCONFIG_SPL_FELon both images. Changing the configuration of multiple images in one command is a generic demand. The current implementation cannot propose any good solution about this.[3]
Kconfigfiles are getting ugly and difficult to understand. Commitb724bd7d6349(dm:Kconfig: MoveCONFIG_SYS_MALLOC_F_LENtoKconfig) has sprinkled "if !SPL_BUILD" over theKconfigfiles.[4] The build system got more complicated than it should be. To adjust Linux-originated
Kconfigto U-Boot, the helper script "scripts/multiconfig.sh" was introduced. Writing a complicated text processor is a shell script sometimes caused problems.Now I believe the "single .config" will serve us better. With it, all the problems above would go away. Instead, we will have to add some CONFIG_SPL_* (and CONFIG_TPL_*) options such as
CONFIG_SPL_DM, but we will not have much. Anyway, this is what we do now inscripts/Makefile.spl.I admit my mistake with my apology and this commit switches to the single .config configuration.
It is not so difficult to do that:
- Remove unnecessary processings from
scripts/multiconfig.shThis file will remain for a while to support the current defconfig format. It will be removed after more cleanups are done.- Adjust some makefiles and Kconfigs
- Add some entries to
include/config_uncmd_spl.hand the new filescripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.
Quoted 5 lines
- update doc/README.kconfig
More cleaning up patches will follow this.
Signed-off-by Masahiro Yamada <[email protected]>
I notice with seaboard there is a bit of a code size different. But I don't think this matters.
06: kconfig: switch to single .config configuration
arm: (for 1/1 boards) all -839.0 bss +4.0 rodata -27.0
spl/u-boot-spl:all +16.0 spl/u-boot-spl:data +16.0 text -816.0
seaboard : all -839 bss +4 rodata -27
spl/u-boot-spl:all +16 spl/u-boot-spl:data +16 text -816
u-boot: add: 1/-11, grow: 0/-1 bytes: 4/-1352 (-1348)
function old new delta
__div0 - 4 +4
__aeabi_idiv0 12 - -12
__lshrdi3 24 - -24
__ashldi3 24 - -24
__aeabi_llsr 24 - -24
__aeabi_llsl 24 - -24
__ashrdi3 26 - -26
__aeabi_uidivmod 26 - -26
__aeabi_lasr 26 - -26
__aeabi_idivmod 26 - -26
raise 28 - -28
__udivsi3 604 152 -452
__divsi3 660 - -660
spl-u-boot-spl: add: 0/0, grow: 1/0 bytes: 16/0 (16)
function old new delta
gdata 184 200 +16
Regards,
Simon
PATCH v3 5/7 kconfig: switch to single .config configuration
In reply to Masahiro YamadaQuoted 11 lines · Masahiro Yamada
When
Kconfigfor U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
Quoted 5 lines
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of
Kconfigis lost.
Sharing can happen in the defconfig with "+S:"...
What sort of dependencies are people wanting? Would it be possible to modify kconfig to import SPL .config into the main config (or vice versa?) with a name prefix so that dependencies could happen, without sacrificing the ability to set symbols independently?
Or as Ian suggested, have only the main config be user-editable, but still let select/depends turn certain things on/off for the auto-generated SPL config.
Quoted 8 lines
[2] It is too painful to change both ".config" and "spl/.config". Sunxi guys started to work around this problem by creating a new configuration target. Commit
cbdd9a9737cc(sunxi: kconfig: Add %_felconfig rule to enable FEL build of sunxi platforms.) added "make*_felconfig" to enableCONFIG_SPL_FELon both images. Changing the configuration of multiple images in one command is a generic demand. The current implementation cannot propose any good solution about this.
How about defconfig fragments? Instead of having script infrastructure specifically for CONFIG_SPL_FEL, merge a fragment containing "+S:CONFIG_SPL_FEL".
Quoted 3 lines
[3]
Kconfigfiles are getting ugly and difficult to understand. Commitb724bd7d6349(dm:Kconfig: MoveCONFIG_SYS_MALLOC_F_LENtoKconfig) has sprinkled "if !SPL_BUILD" over theKconfigfiles.
It seems like the root cause of this sprinkling is wanting to use default y to avoid touching a bunch of defconfig files, but not wanting to do the default y at the toplevel Kconfig. Maybe better tooling for bulk defconfig updates would help. In any case, couldn't you do CONFIG_SPL_DM currently, by making DM depend on "!SPL_BUILD || SPL_DM", without fundamentally changing the SPL kconfig infrastructure?
Why do symbols like LOCALVERSION and CC_OPTIMIZE_FOR_SIZE depend on ! SPL_BUILD?
Quoted 10 lines
[4] The build system got more complicated than it should be. To adjust Linux-originated
Kconfigto U-Boot, the helper script "scripts/multiconfig.sh" was introduced. Writing a complicated text processor is a shell script sometimes caused problems.Now I believe the "single .config" will serve us better. With it, all the problems above would go away. Instead, we will have to add some CONFIG_SPL_* (and CONFIG_TPL_*) options such as
CONFIG_SPL_DM, but we will not have much. Anyway, this is what we do now inscripts/Makefile.spl.
I had been hoping that the split configs would let us get rid of many of the CONFIG_SPL_* options that we already have.
How will TPL be handled? Are you going to duplicate all the SPL symbols? Or just avoid ever kconfigizing them?
Quoted 6 lines
- Add some entries to
include/config_uncmd_spl.hand the new filescripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.
How is uncmd_spl better than "!SPL_BUILD"?
-Scott
PATCH v3 5/7 kconfig: switch to single .config configuration
Hi Scott,
Quoted 12 lines · On Mon, 23 Feb 2015 19:22:51 -0600 Scott Wood <[email protected]> wrote:
Quoted 11 lines · Masahiro Yamada
When
Kconfigfor U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
Quoted 7 lines
Quoted 5 lines
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of
Kconfigis lost.Sharing can happen in the defconfig with "+S:"...
Yes, it can as for "make *_defconfig".
If we modify some options in .config for example by "make menuconfig", we also modify some in spl/.config correspondingly.
Users are responsible for configure .config and spl/.config in sync in the sane combination.
Quoted 4 lines
What sort of dependencies are people wanting? Would it be possible to modify kconfig to import SPL .config into the main config (or vice versa?) with a name prefix so that dependencies could happen, without sacrificing the ability to set symbols independently?
To have independent symboles coexist in a single .config, I can only suggest to duplicate options like CONFIG_FOO=0x100CONFIG_SPL_FOO=0x200CONFIG_TPL_FOO=0x300
Quoted 3 lines
Or as Ian suggested, have only the main config be user-editable, but still let select/depends turn certain things on/off for the auto-generated SPL config.
I guess it is possible for boolean options, but impossible to set hex/int options independently.
BTW, Ian's idea had been already achieved by include/config_uncmd_spl.h
Quoted 9 lines
Quoted 8 lines
[2] It is too painful to change both ".config" and "spl/.config". Sunxi guys started to work around this problem by creating a new configuration target. Commit
cbdd9a9737cc(sunxi: kconfig: Add %_felconfig rule to enable FEL build of sunxi platforms.) added "make*_felconfig" to enableCONFIG_SPL_FELon both images. Changing the configuration of multiple images in one command is a generic demand. The current implementation cannot propose any good solution about this.
Quoted 3 lines
How about defconfig fragments? Instead of having script infrastructure specifically for
CONFIG_SPL_FEL, merge a fragment containing "+S:CONFIG_SPL_FEL".
Do you mean something like this?
U-boot proper : common/.config + .config SPL : common/.config + spl/.config TPL : common/.config + tpl/.config
Quoted 8 lines
Quoted 3 lines
[3]
Kconfigfiles are getting ugly and difficult to understand. Commitb724bd7d6349(dm:Kconfig: MoveCONFIG_SYS_MALLOC_F_LENtoKconfig) has sprinkled "if !SPL_BUILD" over theKconfigfiles.It seems like the root cause of this sprinkling is wanting to use default y to avoid touching a bunch of defconfig files, but not wanting to do the default y at the toplevel
Kconfig. Maybe better tooling for bulk defconfig updates would help.
Yes. If we could move the default settings into defconfig files (and defconfig is just for that purpose), this problem would go away. But, in the duscussion with Simon and Alexey, we understood maintaining many defconfigs in sync is a pain.
Quoted 3 lines
In any case, couldn't you do
CONFIG_SPL_DMcurrently, by making DM depend on "!SPL_BUILD ||SPL_DM", without fundamentally changing the SPL kconfig infrastructure?
As for the Driver Model options, the dependency descriptions will get ugly, but we won't carry them so long.
In a long run, all the boards will be converted and eventually CONFIG_DM will bocome the default.
Quoted 2 lines
Why do symbols like LOCALVERSION and
CC_OPTIMIZE_FOR_SIZEdepend on !SPL_BUILD?
These two options are used by the top-level Makefile and it is automatically propagated to spl/*.
It is harmless to define them again in spl/.config, but meaningless.
Quoted 11 lines
Quoted 10 lines
[4] The build system got more complicated than it should be. To adjust Linux-originated
Kconfigto U-Boot, the helper script "scripts/multiconfig.sh" was introduced. Writing a complicated text processor is a shell script sometimes caused problems.Now I believe the "single .config" will serve us better. With it, all the problems above would go away. Instead, we will have to add some CONFIG_SPL_* (and CONFIG_TPL_*) options such as
CONFIG_SPL_DM, but we will not have much. Anyway, this is what we do now inscripts/Makefile.spl.
Quoted 5 lines
I had been hoping that the split configs would let us get rid of many of the CONFIG_SPL_* options that we already have.
How will TPL be handled? Are you going to duplicate all the SPL symbols? Or just avoid ever kconfigizing them?
Not all, but I expect some duplicated CONFIG_TPL_* such as CONFIG_TPL_TEXT_BASE.
Currently, U-Boot runs SPL, TPL, and U-Boot proper in this order, but in hindsight, it might have been better to run
TPL, SPL, and U-Boot proper, in this order.
In 4KB memory footprint, it is impossible to include Driver Model. It would be a really ad-hoc implementation.
In the former order, we need CONFIG_TPL_DM,
but in the latter, we can save it.
I know TPL means "Third Program Loader", but
can we perhaps swap the order
if we assume TPL is the abbreviation of "Tiny Program Loader" ?
Quoted 8 lines
Quoted 6 lines
- Add some entries to
include/config_uncmd_spl.hand the new filescripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.How is
uncmd_splbetter than "!SPL_BUILD"?
We can use Kconfig as it is in Linux.
Best Regards
Masahiro Yamada
PATCH v3 5/7 kconfig: switch to single .config configuration
Quoted 41 lines · Masahiro Yamada
Hi Scott,
Quoted 12 lines · On Mon, 23 Feb 2015 19:22:51 -0600 Scott Wood <[email protected]> wrote:
Quoted 11 lines · Masahiro Yamada
When
Kconfigfor U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
Quoted 7 lines
Quoted 5 lines
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of
Kconfigis lost.Sharing can happen in the defconfig with "+S:"...
Yes, it can as for "make
*_defconfig".If we modify some options in .config for example by "make menuconfig", we also modify some in spl/.config correspondingly.
Users are responsible for configure .config and spl/.config in sync in the sane combination.
Quoted 4 lines
What sort of dependencies are people wanting? Would it be possible to modify kconfig to import SPL .config into the main config (or vice versa?) with a name prefix so that dependencies could happen, without sacrificing the ability to set symbols independently?
Quoted 5 lines
To have independent symboles coexist in a single .config, I can only suggest to duplicate options like
CONFIG_FOO=0x100CONFIG_SPL_FOO=0x200CONFIG_TPL_FOO=0x300
What I meant was a way to keep the configs separate, but automatically import the CONFIG_FOO from the SPL .config as CONFIG_SPL_FOO (or some other prefix that doesn't conflict with SPL-specific options).
Quoted 6 lines
Quoted 3 lines
Or as Ian suggested, have only the main config be user-editable, but still let select/depends turn certain things on/off for the auto-generated SPL config.
I guess it is possible for boolean options, but impossible to set hex/int options independently.
How many hex/int options are there, that need to be different in SPL versus the main U-Boot? Having a few CONFIG_SPL_xxx for those is better than having a bunch.
Quoted 1 line
BTW, Ian's idea had been already achieved by
include/config_uncmd_spl.h
So, the answer is to avoid kconfig and go back to using the preprocessor for configuration? :-(
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines
[2] It is too painful to change both ".config" and "spl/.config". Sunxi guys started to work around this problem by creating a new configuration target. Commit
cbdd9a9737cc(sunxi: kconfig: Add %_felconfig rule to enable FEL build of sunxi platforms.) added "make*_felconfig" to enableCONFIG_SPL_FELon both images. Changing the configuration of multiple images in one command is a generic demand. The current implementation cannot propose any good solution about this.
Quoted 8 lines
Quoted 3 lines
How about defconfig fragments? Instead of having script infrastructure specifically for
CONFIG_SPL_FEL, merge a fragment containing "+S:CONFIG_SPL_FEL".Do you mean something like this?
U-boot proper : common/.config + .config SPL : common/.config + spl/.config TPL : common/.config + tpl/.config
No, I meant having a fragment containing only "+S:CONFIG_SPL_FEL" that could be merged into any other config.
Quoted 9 lines
Quoted 8 lines
Quoted 3 lines
[3]
Kconfigfiles are getting ugly and difficult to understand. Commitb724bd7d6349(dm:Kconfig: MoveCONFIG_SYS_MALLOC_F_LENtoKconfig) has sprinkled "if !SPL_BUILD" over theKconfigfiles.It seems like the root cause of this sprinkling is wanting to use default y to avoid touching a bunch of defconfig files, but not wanting to do the default y at the toplevel
Kconfig. Maybe better tooling for bulk defconfig updates would help.
Quoted 4 lines
Yes. If we could move the default settings into defconfig files (and defconfig is just for that purpose), this problem would go away. But, in the duscussion with Simon and Alexey, we understood maintaining many defconfigs in sync is a pain.
I think that's a problem that needs to be solved regardless of SPL.
Quoted 8 lines
Quoted 3 lines
In any case, couldn't you do
CONFIG_SPL_DMcurrently, by making DM depend on "!SPL_BUILD ||SPL_DM", without fundamentally changing the SPL kconfig infrastructure?As for the Driver Model options, the dependency descriptions will get ugly, but we won't carry them so long.
In a long run, all the boards will be converted and eventuallyCONFIG_DMwill bocome the default.
...so it's not a very good example of why the current situation must change.
Quoted 7 lines
Quoted 2 lines
Why do symbols like LOCALVERSION and
CC_OPTIMIZE_FOR_SIZEdepend on !SPL_BUILD?These two options are used by the top-level
Makefileand it is automatically propagated to spl/*.It is harmless to define them again in spl/.config, but meaningless.
...so not all of the existing !SPL_BUILD instances in Kconfig need to be there.
Quoted 11 lines
Quoted 11 lines
Quoted 10 lines
[4] The build system got more complicated than it should be. To adjust Linux-originated
Kconfigto U-Boot, the helper script "scripts/multiconfig.sh" was introduced. Writing a complicated text processor is a shell script sometimes caused problems.Now I believe the "single .config" will serve us better. With it, all the problems above would go away. Instead, we will have to add some CONFIG_SPL_* (and CONFIG_TPL_*) options such as
CONFIG_SPL_DM, but we will not have much. Anyway, this is what we do now inscripts/Makefile.spl.
Quoted 7 lines
Quoted 5 lines
I had been hoping that the split configs would let us get rid of many of the CONFIG_SPL_* options that we already have.
How will TPL be handled? Are you going to duplicate all the SPL symbols? Or just avoid ever kconfigizing them?
Not all, but I expect some duplicated CONFIG_TPL_* such as
CONFIG_TPL_TEXT_BASE.
I'm not talking about TEXT_BASE. I'm talking about stuff like this:
#ifdef CONFIG_TPL_BUILD
#define CONFIG_SPL_NAND_BOOT
#define CONFIG_SPL_FLUSH_IMAGE
#define CONFIG_SPL_ENV_SUPPORT
#define CONFIG_SPL_NAND_INIT
#define CONFIG_SPL_SERIAL_SUPPORT
#define CONFIG_SPL_LIBGENERIC_SUPPORT
#define CONFIG_SPL_LIBCOMMON_SUPPORT
#define CONFIG_SPL_I2C_SUPPORT
#define CONFIG_SPL_NAND_SUPPORT
#define CONFIG_SPL_MPC8XXX_INIT_DDR_SUPPORT
#define CONFIG_SPL_COMMON_INIT_DDR
#define CONFIG_SPL_MAX_SIZE (128 << 10)
#define CONFIG_SPL_TEXT_BASE 0xf8f81000
#define CONFIG_SYS_MPC85XX_NO_RESETVEC
#define CONFIG_SYS_NAND_U_BOOT_SIZE (832 << 10) #define CONFIG_SYS_NAND_U_BOOT_DST (0x11000000) #define CONFIG_SYS_NAND_U_BOOT_START (0x11000000) #define CONFIG_SYS_NAND_U_BOOT_OFFS ((128 + 128) << 10)
#elif defined(CONFIG_SPL_BUILD)
#define CONFIG_SPL_INIT_MINIMAL
#define CONFIG_SPL_SERIAL_SUPPORT
#define CONFIG_SPL_NAND_SUPPORT
#define CONFIG_SPL_FLUSH_IMAGE
#define CONFIG_SPL_TEXT_BASE 0xff800000 #define CONFIG_SPL_MAX_SIZE 4096 #define CONFIG_SYS_NAND_U_BOOT_SIZE (128 << 10) #define CONFIG_SYS_NAND_U_BOOT_DST 0xf8f80000 #define CONFIG_SYS_NAND_U_BOOT_START 0xf8f80000 #define CONFIG_SYS_NAND_U_BOOT_OFFS (128 << 10)
#endif
If symbols like CONFIG_SPL_I2C_SUPPORT or CONFIG_SPL_COMMON_INIT_DDR get kconfigized, how would you handle them being in TPL but not SPL?
Quoted 3 lines
Currently, U-Boot runs SPL, TPL, and U-Boot proper in this order, but in hindsight, it might have been better to run
TPL, SPL, and U-Boot proper, in this order.
TPL is just makefile infrastructure for inserting an extra stage. It doesn't refer to the contents.
Quoted 2 lines
In 4KB memory footprint, it is impossible to include Driver Model. It would be a really ad-hoc implementation.
"Is", not "would be". And this applies to some SPL targets without TPL as well.
Quoted 6 lines
In the former order, we need
CONFIG_TPL_DM,
but in the latter, we can save it.I know TPL means "Third Program Loader", but
can we perhaps swap the order
if we assume TPL is the abbreviation of "Tiny Program Loader" ?
If you redefine TPL to mean SPL that doesn't use certain code, you'll end up with targets that have TPL but no SPL. Are you sure this is simplifying anything?
Quoted 10 lines
Quoted 8 lines
Quoted 6 lines
- Add some entries to
include/config_uncmd_spl.hand the new filescripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.How is
uncmd_splbetter than "!SPL_BUILD"?We can use
Kconfigas it is in Linux.
Not after this patch.
-Scott
PATCH v3 5/7 kconfig: switch to single .config configuration
Hi Scott,
Quoted 48 lines · On Tue, 24 Feb 2015 18:17:59 -0600 Scott Wood <[email protected]> wrote:
Quoted 41 lines · Masahiro Yamada
Hi Scott,
Quoted 12 lines · On Mon, 23 Feb 2015 19:22:51 -0600 Scott Wood <[email protected]> wrote:
Quoted 11 lines · Masahiro Yamada
When
Kconfigfor U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
Quoted 7 lines
Quoted 5 lines
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of
Kconfigis lost.Sharing can happen in the defconfig with "+S:"...
Yes, it can as for "make
*_defconfig".If we modify some options in .config for example by "make menuconfig", we also modify some in spl/.config correspondingly.
Users are responsible for configure .config and spl/.config in sync in the sane combination.
Quoted 4 lines
What sort of dependencies are people wanting? Would it be possible to modify kconfig to import SPL .config into the main config (or vice versa?) with a name prefix so that dependencies could happen, without sacrificing the ability to set symbols independently?
Quoted 5 lines
To have independent symboles coexist in a single .config, I can only suggest to duplicate options like
CONFIG_FOO=0x100CONFIG_SPL_FOO=0x200CONFIG_TPL_FOO=0x300
Quoted 3 lines
What I meant was a way to keep the configs separate, but automatically import the
CONFIG_FOOfrom the SPL .config asCONFIG_SPL_FOO(or some other prefix that doesn't conflict with SPL-specific options).
What is the benefit of doing this?
Quoted 10 lines
Quoted 6 lines
Quoted 3 lines
Or as Ian suggested, have only the main config be user-editable, but still let select/depends turn certain things on/off for the auto-generated SPL config.
I guess it is possible for boolean options, but impossible to set hex/int options independently.
How many hex/int options are there, that need to be different in SPL versus the main U-Boot? Having a few
CONFIG_SPL_xxxfor those is better than having a bunch.
OK.
But, I do not think we need to tweak the Kconfig just for saving boolean options.
Quoted 4 lines
Quoted 1 line
BTW, Ian's idea had been already achieved by
include/config_uncmd_spl.hSo, the answer is to avoid kconfig and go back to using the preprocessor for configuration? :-(
I am not saying I prefer the preprocessor.
Indeed, include/config_uncmd_spl.h is ugly,
so I'd like to propose a better solution.
If we introduce CONFIG_SPL_DM, for example, the ifdef conditional in source files will be like this:
#if (!defined(CONFIG_SPL_BUILD) && defined(CONFIG_DM)) || \
(defined(CONFIG_SPL_BUILD) && defined(CONFIG_SPL_DM))[Driver Model Code]
#else
[Non Driver Model Code]
#endifThis is too ugly to be written in each conditional.
So, I want to describe like this:
#if IS_ENABLED_CONFIG(DM)
[Driver Model Code]
#else
[Non Driver Model Code]
#endifI will post some patches later on.
Quoted 13 lines
Quoted 13 lines
Quoted 12 lines
Quoted 8 lines
[2] It is too painful to change both ".config" and "spl/.config". Sunxi guys started to work around this problem by creating a new configuration target. Commit
cbdd9a9737cc(sunxi: kconfig: Add %_felconfig rule to enable FEL build of sunxi platforms.) added "make*_felconfig" to enableCONFIG_SPL_FELon both images. Changing the configuration of multiple images in one command is a generic demand. The current implementation cannot propose any good solution about this.How about defconfig fragments? Instead of having script infrastructure specifically for
CONFIG_SPL_FEL, merge a fragment containing "+S:CONFIG_SPL_FEL".
Quoted 7 lines
Quoted 4 lines
Do you mean something like this?
U-boot proper : common/.config + .config SPL : common/.config + spl/.config TPL : common/.config + tpl/.configNo, I meant having a fragment containing only "+S:CONFIG_SPL_FEL" that could be merged into any other config.
So, the fragment is something like the _common_ .config, right?
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines
Quoted 3 lines
[3]
Kconfigfiles are getting ugly and difficult to understand. Commitb724bd7d6349(dm:Kconfig: MoveCONFIG_SYS_MALLOC_F_LENtoKconfig) has sprinkled "if !SPL_BUILD" over theKconfigfiles.It seems like the root cause of this sprinkling is wanting to use default y to avoid touching a bunch of defconfig files, but not wanting to do the default y at the toplevel
Kconfig. Maybe better tooling for bulk defconfig updates would help.
Quoted 6 lines
Quoted 4 lines
Yes. If we could move the default settings into defconfig files (and defconfig is just for that purpose), this problem would go away. But, in the duscussion with Simon and Alexey, we understood maintaining many defconfigs in sync is a pain.
I think that's a problem that needs to be solved regardless of SPL.
Agree.
I think we can live the defaults in Kconfig, but I am still searching for a different solution.
Quoted 11 lines
Quoted 8 lines
Quoted 3 lines
In any case, couldn't you do
CONFIG_SPL_DMcurrently, by making DM depend on "!SPL_BUILD ||SPL_DM", without fundamentally changing the SPL kconfig infrastructure?As for the Driver Model options, the dependency descriptions will get ugly, but we won't carry them so long.
In a long run, all the boards will be converted and eventuallyCONFIG_DMwill bocome the default....so it's not a very good example of why the current situation must change.
Right, but we still have many other options that can be enabled on SPL.
Quoted 25 lines
Quoted 7 lines
Quoted 2 lines
Why do symbols like LOCALVERSION and
CC_OPTIMIZE_FOR_SIZEdepend on !SPL_BUILD?These two options are used by the top-level
Makefileand it is automatically propagated to spl/*.It is harmless to define them again in spl/.config, but meaningless.
...so not all of the existing !SPL_BUILD instances in
Kconfigneed to be there.Quoted 14 lines
Quoted 14 lines
Quoted 10 lines
[4] The build system got more complicated than it should be. To adjust Linux-originated
Kconfigto U-Boot, the helper script "scripts/multiconfig.sh" was introduced. Writing a complicated text processor is a shell script sometimes caused problems.Now I believe the "single .config" will serve us better. With it, all the problems above would go away. Instead, we will have to add some CONFIG_SPL_* (and CONFIG_TPL_*) options such as
CONFIG_SPL_DM, but we will not have much. Anyway, this is what we do now inscripts/Makefile.spl.I had been hoping that the split configs would let us get rid of many of the CONFIG_SPL_* options that we already have.
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
How will TPL be handled? Are you going to duplicate all the SPL symbols? Or just avoid ever kconfigizing them?
Not all, but I expect some duplicated CONFIG_TPL_* such as
CONFIG_TPL_TEXT_BASE.I'm not talking about
TEXT_BASE. I'm talking about stuff like this:
We have to add some CONFIG_TPL_*, but we will just have 20.
Quoted 30 lines
#ifdef
CONFIG_TPL_BUILD
#defineCONFIG_SPL_NAND_BOOT
#defineCONFIG_SPL_FLUSH_IMAGE
#defineCONFIG_SPL_ENV_SUPPORT
#defineCONFIG_SPL_NAND_INIT
#defineCONFIG_SPL_SERIAL_SUPPORT
#defineCONFIG_SPL_LIBGENERIC_SUPPORT
#defineCONFIG_SPL_LIBCOMMON_SUPPORT
#defineCONFIG_SPL_I2C_SUPPORT
#defineCONFIG_SPL_NAND_SUPPORT
#defineCONFIG_SPL_MPC8XXX_INIT_DDR_SUPPORT
#defineCONFIG_SPL_COMMON_INIT_DDR
#defineCONFIG_SPL_MAX_SIZE(128 << 10)
#defineCONFIG_SPL_TEXT_BASE0xf8f81000
#defineCONFIG_SYS_MPC85XX_NO_RESETVEC#define CONFIG_SYS_NAND_U_BOOT_SIZE (832 << 10) #define CONFIG_SYS_NAND_U_BOOT_DST (0x11000000) #define CONFIG_SYS_NAND_U_BOOT_START (0x11000000) #define CONFIG_SYS_NAND_U_BOOT_OFFS ((128 + 128) << 10)#elif
defined(CONFIG_SPL_BUILD)
#defineCONFIG_SPL_INIT_MINIMAL
#defineCONFIG_SPL_SERIAL_SUPPORT
#defineCONFIG_SPL_NAND_SUPPORT
#defineCONFIG_SPL_FLUSH_IMAGE#define CONFIG_SPL_TEXT_BASE 0xff800000 #define CONFIG_SPL_MAX_SIZE 4096 #define CONFIG_SYS_NAND_U_BOOT_SIZE (128 << 10) #define CONFIG_SYS_NAND_U_BOOT_DST 0xf8f80000 #define CONFIG_SYS_NAND_U_BOOT_START 0xf8f80000 #define CONFIG_SYS_NAND_U_BOOT_OFFS (128 << 10)
Quoted 4 lines
#endif
If symbols like
CONFIG_SPL_I2C_SUPPORTorCONFIG_SPL_COMMON_INIT_DDRget kconfigized, how would you handle them being in TPL but not SPL?
We can add CONFIG_TPL_* if necessary.
As I said, if we swap the order of SPL and TPL, we will be able to save CONFIG_TPL_* defines.
Quoted 16 lines
Quoted 3 lines
Currently, U-Boot runs SPL, TPL, and U-Boot proper in this order, but in hindsight, it might have been better to run
TPL, SPL, and U-Boot proper, in this order.TPL is just makefile infrastructure for inserting an extra stage. It doesn't refer to the contents.
Quoted 2 lines
In 4KB memory footprint, it is impossible to include Driver Model. It would be a really ad-hoc implementation.
"Is", not "would be". And this applies to some SPL targets without TPL as well.
Quoted 3 lines
In the former order, we need
CONFIG_TPL_DM,
but in the latter, we can save it.
Quoted 7 lines
Quoted 3 lines
I know TPL means "Third Program Loader", but
can we perhaps swap the order
if we assume TPL is the abbreviation of "Tiny Program Loader" ?If you redefine TPL to mean SPL that doesn't use certain code, you'll end up with targets that have TPL but no SPL. Are you sure this is simplifying anything?
Sorry, I can't get it.
What I expect is like follows:
CONFIG_TPL still depends on CONFIG_SPL.
We have three options for the boot procedure:
[1] U-Boot-proper (CONFIG_SPL is not defined)
[2] SPL + U-Boot-proper (CONFIG_SPL is defined)
[3] TPL + SPL + U-Boot-proper (CONFIG_SPL and CONFIG_TPL are defined)
The image size: TPL < SPL < U-Boot-proper
Driver Model and some other features are available on SPL if CONFIG_SPL_* is defined.
Almost no systematic infrastructure is available on TPL, so we will have
very small number of CONFIG_TPL_*.
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
- Add some entries to
include/config_uncmd_spl.hand the new filescripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.
Quoted 6 lines
Quoted 3 lines
Quoted 1 line
How is
uncmd_splbetter than "!SPL_BUILD"?We can use
Kconfigas it is in Linux.Not after this patch.
Right, I need to do more cleanups for that.
Best Regards
Masahiro Yamada
PATCH v3 5/7 kconfig: switch to single .config configuration
Hi,
Quoted 105 lines · Masahiro Yamada
Hi Scott,
Quoted 48 lines · On Tue, 24 Feb 2015 18:17:59 -0600 Scott Wood <[email protected]> wrote:
Quoted 41 lines · Masahiro Yamada
Hi Scott,
Quoted 12 lines · On Mon, 23 Feb 2015 19:22:51 -0600 Scott Wood <[email protected]> wrote:
Quoted 11 lines · Masahiro Yamada
When
Kconfigfor U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
Quoted 7 lines
Quoted 5 lines
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of
Kconfigis lost.Sharing can happen in the defconfig with "+S:"...
Yes, it can as for "make
*_defconfig".If we modify some options in .config for example by "make menuconfig", we also modify some in spl/.config correspondingly.
Users are responsible for configure .config and spl/.config in sync in the sane combination.
Quoted 4 lines
What sort of dependencies are people wanting? Would it be possible to modify kconfig to import SPL .config into the main config (or vice versa?) with a name prefix so that dependencies could happen, without sacrificing the ability to set symbols independently?
Quoted 5 lines
To have independent symboles coexist in a single .config, I can only suggest to duplicate options like
CONFIG_FOO=0x100CONFIG_SPL_FOO=0x200CONFIG_TPL_FOO=0x300Quoted 3 lines
What I meant was a way to keep the configs separate, but automatically import the
CONFIG_FOOfrom the SPL .config asCONFIG_SPL_FOO(or some other prefix that doesn't conflict with SPL-specific options).What is the benefit of doing this?
Quoted 10 lines
Quoted 6 lines
Quoted 3 lines
Or as Ian suggested, have only the main config be user-editable, but still let select/depends turn certain things on/off for the auto-generated SPL config.
I guess it is possible for boolean options, but impossible to set hex/int options independently.
How many hex/int options are there, that need to be different in SPL versus the main U-Boot? Having a few
CONFIG_SPL_xxxfor those is better than having a bunch.OK.
But, I do not think we need to tweak theKconfigjust for saving boolean options.Quoted 4 lines
Quoted 1 line
BTW, Ian's idea had been already achieved by
include/config_uncmd_spl.hSo, the answer is to avoid kconfig and go back to using the preprocessor for configuration? :-(
I am not saying I prefer the preprocessor.
Indeed,
include/config_uncmd_spl.his ugly,
so I'd like to propose a better solution.If we introduce
CONFIG_SPL_DM, for example, the ifdef conditional in source files will be like this:#if (!defined(CONFIG_SPL_BUILD) && defined(CONFIG_DM)) || \ (defined(CONFIG_SPL_BUILD) && defined(CONFIG_SPL_DM))[Driver Model Code]
#else [Non Driver Model Code] #endifThis is too ugly to be written in each conditional.
So, I want to describe like this:
Quoted 7 lines
#if IS_ENABLED_CONFIG(DM) [Driver Model Code] #else [Non Driver Model Code] #endifI will post some patches later on.
BTW it would be nice if we could compile the code even if it is not included (e.g with the compile-time IS_ENABLED(CONFIG_...)). Also for driver model I would like it to be supported the same in both SPL and U-Boot proper, to avoid this sort of thing. We are very close to merging the driver model SPL support.
[snip]
Regards,
Simon
PATCH v3 5/7 kconfig: switch to single .config configuration
In reply to Masahiro YamadaQuoted 54 lines · Masahiro Yamada
Hi Scott,
Quoted 48 lines · On Tue, 24 Feb 2015 18:17:59 -0600 Scott Wood <[email protected]> wrote:
Quoted 46 lines · Masahiro Yamada
Hi Scott,
Quoted 12 lines · On Mon, 23 Feb 2015 19:22:51 -0600 Scott Wood <[email protected]> wrote:
Quoted 11 lines · Masahiro Yamada
When
Kconfigfor U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
Quoted 7 lines
Quoted 5 lines
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of
Kconfigis lost.Sharing can happen in the defconfig with "+S:"...
Yes, it can as for "make
*_defconfig".If we modify some options in .config for example by "make menuconfig", we also modify some in spl/.config correspondingly.
Users are responsible for configure .config and spl/.config in sync in the sane combination.
Quoted 4 lines
What sort of dependencies are people wanting? Would it be possible to modify kconfig to import SPL .config into the main config (or vice versa?) with a name prefix so that dependencies could happen, without sacrificing the ability to set symbols independently?
To have independent symboles coexist in a single .config, I can only suggest to duplicate options like
CONFIG_FOO=0x100CONFIG_SPL_FOO=0x200CONFIG_TPL_FOO=0x300
Quoted 5 lines
Quoted 3 lines
What I meant was a way to keep the configs separate, but automatically import the
CONFIG_FOOfrom the SPL .config asCONFIG_SPL_FOO(or some other prefix that doesn't conflict with SPL-specific options).What is the benefit of doing this?
So that you can express dependencies in the main U-Boot on SPL symbols, which you said was one of the problems that motivated this change.
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
Or as Ian suggested, have only the main config be user-editable, but still let select/depends turn certain things on/off for the auto-generated SPL config.
I guess it is possible for boolean options, but impossible to set hex/int options independently.
Quoted 6 lines
Quoted 3 lines
How many hex/int options are there, that need to be different in SPL versus the main U-Boot? Having a few
CONFIG_SPL_xxxfor those is better than having a bunch.OK.
But, I do not think we need to tweak theKconfigjust for saving boolean options.
Most options are boolean (especially if you ignore hardware description that isn't going to differ between SPL and non-SPL). I think it makes a lot of sense to not want to duplicate them, and especially to not complicate each place that ifdefs on the symbol by having to check for SPL.
Quoted 16 lines
Quoted 4 lines
Quoted 1 line
BTW, Ian's idea had been already achieved by
include/config_uncmd_spl.hSo, the answer is to avoid kconfig and go back to using the preprocessor for configuration? :-(
I am not saying I prefer the preprocessor.
Indeed,
include/config_uncmd_spl.his ugly,
so I'd like to propose a better solution.If we introduce
CONFIG_SPL_DM, for example, the ifdef conditional in source files will be like this:#if (!defined(CONFIG_SPL_BUILD) && defined(CONFIG_DM)) || \ (defined(CONFIG_SPL_BUILD) && defined(CONFIG_SPL_DM))
Quoted 7 lines
[Driver Model Code]
#else [Non Driver Model Code] #endifThis is too ugly to be written in each conditional.
That's not what I was suggesting.
I was suggesting something like:
config DM
bool "Enable Driver Model" depends on !SPL_BUILD || SPL_DM ...
Then the ifdef would just be
#ifdef CONFIG_DM
[Driver Model Code]
#else
[Non Driver Model Code]
#endif
Quoted 18 lines
Quoted 18 lines
Quoted 17 lines
Quoted 9 lines
Quoted 8 lines
[2] It is too painful to change both ".config" and "spl/.config". Sunxi guys started to work around this problem by creating a new configuration target. Commit
cbdd9a9737cc(sunxi: kconfig: Add %_felconfig rule to enable FEL build of sunxi platforms.) added "make*_felconfig" to enableCONFIG_SPL_FELon both images. Changing the configuration of multiple images in one command is a generic demand. The current implementation cannot propose any good solution about this.Quoted 3 lines
How about defconfig fragments? Instead of having script infrastructure specifically for
CONFIG_SPL_FEL, merge a fragment containing "+S:CONFIG_SPL_FEL".Do you mean something like this?
U-boot proper : common/.config + .config SPL : common/.config + spl/.config TPL : common/.config + tpl/.config
Quoted 4 lines
Quoted 2 lines
No, I meant having a fragment containing only "+S:CONFIG_SPL_FEL" that could be merged into any other config.
So, the fragment is something like the _common_ .config, right?
The fragment would be part of a library of common config tweaks that users can apply -- providing similar convenience as _felconfig but with generic infrastructure.
Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 10 lines
[4] The build system got more complicated than it should be. To adjust Linux-originated
Kconfigto U-Boot, the helper script "scripts/multiconfig.sh" was introduced. Writing a complicated text processor is a shell script sometimes caused problems.Now I believe the "single .config" will serve us better. With it, all the problems above would go away. Instead, we will have to add some CONFIG_SPL_* (and CONFIG_TPL_*) options such as
CONFIG_SPL_DM, but we will not have much. Anyway, this is what we do now inscripts/Makefile.spl.I had been hoping that the split configs would let us get rid of many of the CONFIG_SPL_* options that we already have.
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
How will TPL be handled? Are you going to duplicate all the SPL symbols? Or just avoid ever kconfigizing them?
Not all, but I expect some duplicated CONFIG_TPL_* such as
CONFIG_TPL_TEXT_BASE.I'm not talking about
TEXT_BASE. I'm talking about stuff like this:We have to add some CONFIG_TPL_*, but we will just have 20.
Just? And how much makefile crud to check for those options when building? To achieve what?
Quoted 18 lines
Quoted 3 lines
If you redefine TPL to mean SPL that doesn't use certain code, you'll end up with targets that have TPL but no SPL. Are you sure this is simplifying anything?
Sorry, I can't get it.
What I expect is like follows:
CONFIG_TPLstill depends onCONFIG_SPL.We have three options for the boot procedure:
[1] U-Boot-proper (
CONFIG_SPLis not defined)[2] SPL + U-Boot-proper (
CONFIG_SPLis defined)[3] TPL + SPL + U-Boot-proper (
CONFIG_SPLandCONFIG_TPLare defined)
Quoted 6 lines
The image size: TPL < SPL < U-Boot-proper
Driver Model and some other features are available on SPL if CONFIG_SPL_* is defined.
Almost no systematic infrastructure is available on TPL, so we will have
very small number of CONFIG_TPL_*.
So instead of CONFIG_TPL_* we'd have more hardcoded makefile hackery to disable things in TPL -- and we'd have to duplicate that hackery for SPLs that are just as small as what you're now calling TPL, if you're not going to allow TPL without SPL.
-Scott
PATCH v3 5/7 kconfig: switch to single .config configuration
In reply to Scott WoodHi Scott,
Quoted 62 lines · Scott Wood
Quoted 11 lines · Masahiro Yamada
When
Kconfigfor U-boot was examined, one of the biggest issues was how to support multiple images (Normal, SPL, TPL). There were actually two options, "single .config" and "multiple .config". After some discussions and thought experiments, I chose the latter, i.e. to create ".config", "spl/.config", "tpl/.config" for Normal, SPL, TPL, respectively.It is true that the "multiple .config" strategy provided us the maximum flexibility and helped to avoid duplicating CONFIGs among Normal, SPL, TPL, but I have noticed some fatal problems:
Quoted 5 lines
[1] It is impossible to share CONFIG options across the images. If you change the configuration of Main image, you often have to adjust some SPL configurations correspondingly. Currently, we cannot handle the dependencies between them. It means one of the biggest advantages of
Kconfigis lost.Sharing can happen in the defconfig with "+S:"...
What sort of dependencies are people wanting? Would it be possible to modify kconfig to import SPL .config into the main config (or vice versa?) with a name prefix so that dependencies could happen, without sacrificing the ability to set symbols independently?
Or as Ian suggested, have only the main config be user-editable, but still let select/depends turn certain things on/off for the auto-generated SPL config.
Quoted 8 lines
[2] It is too painful to change both ".config" and "spl/.config". Sunxi guys started to work around this problem by creating a new configuration target. Commit
cbdd9a9737cc(sunxi: kconfig: Add %_felconfig rule to enable FEL build of sunxi platforms.) added "make*_felconfig" to enableCONFIG_SPL_FELon both images. Changing the configuration of multiple images in one command is a generic demand. The current implementation cannot propose any good solution about this.How about defconfig fragments? Instead of having script infrastructure specifically for
CONFIG_SPL_FEL, merge a fragment containing "+S:CONFIG_SPL_FEL".Quoted 3 lines
[3]
Kconfigfiles are getting ugly and difficult to understand. Commitb724bd7d6349(dm:Kconfig: MoveCONFIG_SYS_MALLOC_F_LENtoKconfig) has sprinkled "if !SPL_BUILD" over theKconfigfiles.It seems like the root cause of this sprinkling is wanting to use default y to avoid touching a bunch of defconfig files, but not wanting to do the default y at the toplevel
Kconfig. Maybe better tooling for bulk defconfig updates would help. In any case, couldn't you doCONFIG_SPL_DMcurrently, by making DM depend on "!SPL_BUILD ||SPL_DM", without fundamentally changing the SPL kconfig infrastructure?Why do symbols like LOCALVERSION and
CC_OPTIMIZE_FOR_SIZEdepend on !SPL_BUILD?Quoted 5 lines
[4] The build system got more complicated than it should be. To adjust Linux-originated
Kconfigto U-Boot, the helper script "scripts/multiconfig.sh" was introduced. Writing a complicated text processor is a shell script sometimes caused problems.
Quoted 8 lines
Quoted 5 lines
Now I believe the "single .config" will serve us better. With it, all the problems above would go away. Instead, we will have to add some CONFIG_SPL_* (and CONFIG_TPL_*) options such as
CONFIG_SPL_DM, but we will not have much. Anyway, this is what we do now inscripts/Makefile.spl.I had been hoping that the split configs would let us get rid of many of the CONFIG_SPL_* options that we already have.
As Masahiro says this will happen as we remove one by one the various SPL special cases. For example, when driver model supports SPL and U-Boot then we won't need that option - you will either use driver model (or not) for both. I suspect other options will go the same way.
But if things are split into multiple configs it is really hard to compare them and keep them in sync.
Once the dust settles I will be sending a series to remove the special cases around CONFIG_SPL_DM.
Quoted 4 lines
How will TPL be handled? Are you going to duplicate all the SPL symbols? Or just avoid ever kconfigizing them?
Quoted 8 lines
Quoted 6 lines
- Add some entries to
include/config_uncmd_spl.hand the new filescripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.How is
uncmd_splbetter than "!SPL_BUILD"?
Regards,
Simon
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
In reply to Masahiro YamadaNow CONFIG_SPL_BUILD is not defined in Kconfig, so "!depends on SPL_BUILD" and "if !SPL_BUILD" are redundant.
---
Changes in v3: None
Changes in v2:
- Fix a typo in commit description
Kconfig | 12 ++---------- arch/arm/Kconfig | 7 +++---- arch/arm/cpu/arm1176/bcm2835/Kconfig | 6 +++--- arch/arm/cpu/armv7/exynos/Kconfig | 26 +++++++++++++------------- arch/arm/cpu/armv7/omap3/Kconfig | 8 ++++---- arch/arm/cpu/armv7/s5pc1xx/Kconfig | 4 ++-- arch/arm/cpu/armv7/tegra-common/Kconfig | 12 ++++++------ arch/arm/cpu/armv7/uniphier/Kconfig | 1 - board/compulab/cm_t335/Kconfig | 6 +++--- board/gumstix/pepper/Kconfig | 6 +++--- board/isee/igep0033/Kconfig | 6 +++--- board/phytec/pcm051/Kconfig | 6 +++--- board/samsung/goni/Kconfig | 6 +++--- board/samsung/smdkc100/Kconfig | 6 +++--- board/silica/pengwyn/Kconfig | 6 +++--- board/ti/am335x/Kconfig | 8 ++++---- common/Kconfig | 1 - drivers/core/Kconfig | 16 ++++++---------- drivers/mtd/nand/Kconfig | 4 ---- dts/Kconfig | 1 - include/config_uncmd_spl.h | 4 ++++ scripts/Makefile.uncmd_spl | 4 +++- 22 files changed, 71 insertions(+), 85 deletions(-)
Kconfig +2 -10
@@ -15,7 +15,6 @@ menu "General setup" config LOCALVERSION string "Local version - append to U-Boot release" - depends on !SPL_BUILD help Append an extra string to the end of your U-Boot version. This will show up on your boot log, for example.
@@ -26,7 +25,6 @@ config LOCALVERSION config LOCALVERSION_AUTO bool "Automatically append version information to the version string" - depends on !SPL_BUILD default y help This will try to automatically determine if the current tree is a
@@ -47,7 +45,6 @@ config LOCALVERSION_AUTO config CC_OPTIMIZE_FOR_SIZE bool "Optimize for size" - depends on !SPL_BUILD default y help Enabling this option will pass "-Os" instead of "-O2" to gcc
@@ -95,23 +92,19 @@ config SUPPORT_TPL config SPL bool depends on SUPPORT_SPL - prompt "Enable SPL" if !SPL_BUILD - default y if SPL_BUILD + prompt "Enable SPL" help If you want to build SPL as well as the normal image, say Y. config TPL bool depends on SPL && SUPPORT_TPL - prompt "Enable TPL" if !SPL_BUILD - default y if TPL_BUILD - default n + prompt "Enable TPL" help If you want to build TPL as well as the normal image and SPL, say Y. config FIT bool "Support Flattened Image Tree" - depends on !SPL_BUILD help This option allows to boot the new uImage structrure, Flattened Image Tree. FIT is formally a FDT, which can include
@@ -135,7 +128,6 @@ config FIT_SIGNATURE config SYS_EXTRA_OPTIONS string "Extra Options (DEPRECATED)" - depends on !SPL_BUILD help The old configuration infrastructure (= mkconfig + boards.cfg) provided the extra options field. If you have something like
arch/arm/Kconfig +3 -4
@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7 config TARGET_VEXPRESS64_AEMV8A bool "Support vexpress_aemv8a"
@@ -833,7 +832,7 @@ config ARCH_UNIPHIER select CPU_V7 select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL endchoice
arch/arm/cpu/arm1176/bcm2835/Kconfig +3 -3
@@ -1,12 +1,12 @@ if TARGET_RPI config DM - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y endif
arch/arm/cpu/armv7/exynos/Kconfig +13 -13
@@ -6,7 +6,7 @@ choice config TARGET_SMDKV310 select SUPPORT_SPL bool "Exynos4210 SMDKV310 board" - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL config TARGET_TRATS bool "Exynos4210 Trats board"
@@ -33,32 +33,32 @@ config TARGET_ARNDALE select CPU_V7_HAS_NONSEC select CPU_V7_HAS_VIRT select SUPPORT_SPL - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL config TARGET_SMDK5250 bool "SMDK5250 board" select SUPPORT_SPL - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL config TARGET_SNOW bool "Snow board" select SUPPORT_SPL - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL config TARGET_SMDK5420 bool "SMDK5420 board" select SUPPORT_SPL - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL config TARGET_PEACH_PI bool "Peach Pi board" select SUPPORT_SPL - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL config TARGET_PEACH_PIT bool "Peach Pit board" select SUPPORT_SPL - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL endchoice
@@ -66,22 +66,22 @@ config SYS_SOC default "exynos" config DM - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y config DM_SPI - default y if !SPL_BUILD + default y config DM_SPI_FLASH - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y config SYS_MALLOC_F - default y if !SPL_BUILD + default y source "board/samsung/smdkv310/Kconfig" source "board/samsung/trats/Kconfig"
arch/arm/cpu/armv7/omap3/Kconfig +4 -4
@@ -94,16 +94,16 @@ config TARGET_TWISTER endchoice config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if DM && !SPL_BUILD + default y if DM config DM_SERIAL - default y if DM && !SPL_BUILD + default y if DM config SYS_MALLOC_F - default y if DM && !SPL_BUILD + default y if DM config SYS_SOC default "omap3"
arch/arm/cpu/armv7/s5pc1xx/Kconfig +2 -2
@@ -5,11 +5,11 @@ choice config TARGET_S5P_GONI bool "S5P Goni board" - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL config TARGET_SMDKC100 bool "Support smdkc100 board" - select OF_CONTROL if !SPL_BUILD + select OF_CONTROL endchoice
arch/arm/cpu/armv7/tegra-common/Kconfig +6 -6
@@ -28,22 +28,22 @@ config USE_PRIVATE_LIBGCC default y config DM - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y config DM_SPI - default y if !SPL_BUILD + default y config DM_SPI_FLASH - default y if !SPL_BUILD + default y config DM_I2C - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y source "arch/arm/cpu/armv7/tegra20/Kconfig" source "arch/arm/cpu/armv7/tegra30/Kconfig"
arch/arm/cpu/armv7/uniphier/Kconfig +0 -1
@@ -61,7 +61,6 @@ config CMD_PINMON config CMD_DDRPHY_DUMP bool "Enable dump command of DDR PHY parameters" - depends on !SPL_BUILD help The command "ddrphy" shows the resulting parameters of DDR PHY training; it is useful for the evaluation of DDR PHY training.
board/compulab/cm_t335/Kconfig +3 -3
@@ -13,12 +13,12 @@ config SYS_CONFIG_NAME default "cm_t335" config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y endif
board/gumstix/pepper/Kconfig +3 -3
@@ -13,12 +13,12 @@ config SYS_CONFIG_NAME default "pepper" config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y endif
board/isee/igep0033/Kconfig +3 -3
@@ -13,12 +13,12 @@ config SYS_CONFIG_NAME default "am335x_igep0033" config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y endif
board/phytec/pcm051/Kconfig +3 -3
@@ -13,12 +13,12 @@ config SYS_CONFIG_NAME default "pcm051" config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y endif
board/samsung/goni/Kconfig +3 -3
@@ -13,12 +13,12 @@ config SYS_CONFIG_NAME default "s5p_goni" config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y endif
board/samsung/smdkc100/Kconfig +3 -3
@@ -13,12 +13,12 @@ config SYS_CONFIG_NAME default "smdkc100" config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y endif
board/silica/pengwyn/Kconfig +3 -3
@@ -13,12 +13,12 @@ config SYS_CONFIG_NAME default "pengwyn" config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if !SPL_BUILD + default y config DM_SERIAL - default y if !SPL_BUILD + default y endif
board/ti/am335x/Kconfig +4 -4
@@ -39,15 +39,15 @@ config NOR_BOOT NOR for environment. config DM - default y if !SPL_BUILD + default y config DM_GPIO - default y if DM && !SPL_BUILD + default y if DM config DM_SERIAL - default y if DM && !SPL_BUILD + default y if DM config SYS_MALLOC_F - default y if DM && !SPL_BUILD + default y if DM endif
common/Kconfig +0 -1
@@ -1,5 +1,4 @@ menu "Command line interface" - depends on !SPL_BUILD config HUSH_PARSER bool "Use hush shell"
drivers/core/Kconfig +6 -10
@@ -1,6 +1,5 @@ config DM bool "Enable Driver Model" - depends on !SPL_BUILD help This config option enables Driver Model. This brings in the core support, including scanning of platform data on start-up. If
@@ -22,31 +21,28 @@ config SPL_DM config DM_WARN bool "Enable warnings in driver model" + depends on DM + default y help The dm_warn() function can use up quite a bit of space for its strings. By default this is disabled for SPL builds to save space. This will cause dm_warn() to be compiled out - it will do nothing when called. - depends on DM - default y if !SPL_BUILD - default n if SPL_BUILD config DM_DEVICE_REMOVE bool "Support device removal" + depends on DM + default y help We can save some code space by dropping support for removing a device. This is not normally required in SPL, so by default this option is disabled for SPL. - depends on DM - default y if !SPL_BUILD - default n if SPL_BUILD config DM_STDIO bool "Support stdio registration" + depends on DM + default y help Normally serial drivers register with stdio so that they can be used as normal output devices. In SPL we don't normally use stdio, so we can omit this feature. - depends on DM - default y if !SPL_BUILD - default n if SPL_BUILD
drivers/mtd/nand/Kconfig +0 -4
@@ -6,8 +6,6 @@ config SYS_NAND_SELF_INIT This option, if enabled, provides more flexible and linux-like NAND initialization process. -if !SPL_BUILD - config NAND_DENALI bool "Support Denali NAND controller" select SYS_NAND_SELF_INIT
@@ -34,8 +32,6 @@ config NAND_DENALI_SPARE_AREA_SKIP_BYTES of OOB area before last ECC sector data starts. This is potentially used to preserve the bad block marker in the OOB area. -endif - if SPL config SPL_NAND_DENALI
dts/Kconfig +0 -1
@@ -9,7 +9,6 @@ config SUPPORT_OF_CONTROL bool menu "Device Tree Control" - depends on !SPL_BUILD depends on SUPPORT_OF_CONTROL config OF_CONTROL
include/config_uncmd_spl.h +4 -0
@@ -29,5 +29,9 @@#undef CONFIG_DM_SPI#endif +#undef CONFIG_DM_WARN +#undef CONFIG_DM_DEVICE_REMOVE +#undef CONFIG_DM_STDIO +#endif /* CONFIG_SPL_BUILD */ #endif /* __CONFIG_UNCMD_SPL_H__ */
scripts/Makefile.uncmd_spl +3 -1
@@ -1,6 +1,6 @@ # Makefile version of include/config_uncmd_spl.h # -# Invent a better way +# TODO: Invent a better way ifdef CONFIG_SPL_BUILD CONFIG_OF_CONTROL=
@@ -13,4 +13,6 @@ CONFIG_DM_SPI= CONFIG_DM_SPI_FLASH= endif +CONFIG_DM_DEVICE_REMOVE= + endif
Signature · 1 line
1.9.1
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
+Stephen
Hi Masahiro,
Quoted 34 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.Signed-off-by Masahiro Yamada <[email protected]>---
Changes in v3: None
Changes in v2:
- Fix a typo in commit descriptionKconfig | 12 ++---------- arch/arm/Kconfig | 7 +++---- arch/arm/cpu/arm1176/bcm2835/Kconfig | 6 +++--- arch/arm/cpu/armv7/exynos/Kconfig | 26 +++++++++++++------------- arch/arm/cpu/armv7/omap3/Kconfig | 8 ++++---- arch/arm/cpu/armv7/s5pc1xx/Kconfig | 4 ++-- arch/arm/cpu/armv7/tegra-common/Kconfig | 12 ++++++------ arch/arm/cpu/armv7/uniphier/Kconfig | 1 - board/compulab/cm_t335/Kconfig | 6 +++--- board/gumstix/pepper/Kconfig | 6 +++--- board/isee/igep0033/Kconfig | 6 +++--- board/phytec/pcm051/Kconfig | 6 +++--- board/samsung/goni/Kconfig | 6 +++--- board/samsung/smdkc100/Kconfig | 6 +++--- board/silica/pengwyn/Kconfig | 6 +++--- board/ti/am335x/Kconfig | 8 ++++---- common/Kconfig | 1 - drivers/core/Kconfig | 16 ++++++---------- drivers/mtd/nand/Kconfig | 4 ---- dts/Kconfig | 1 - include/config_uncmd_spl.h | 4 ++++ scripts/Makefile.uncmd_spl | 4 +++- 22 files changed, 71 insertions(+), 85 deletions(-)
[snip]
arch/arm/Kconfig config TEGRA · line 743
Quoted 13 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7
Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
Regards,
Simon
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Quoted 17 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 12 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-1@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD
Quoted 8 lines
arch/arm/Kconfig config TEGRA · line 743
Quoted 4 lines
arch/arm/Kconfig +2
-2- select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
Yes, on Tegra SPL runs on an ARMv4t and main U-Boot on some ARMv7 CPU. The change above is actively incorrect.
The OF_CONTROL change is probably incorrect too; we certainly don't intend to use OF_CONTROL in the SPL (there's really nothing to control in the SPL) - I just don't know if enabling that feature will cause any issue. Things to look out for would be bloat of the SPL binary so that it didn't fit into the space before the main binary's TEXT_BASE, since the two get concatenated together into a single binary that's loaded into RAM, and XIP'd.
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Hi,
Quoted 36 lines · Stephen Warren
Quoted 22 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 15 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-2@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILDQuoted 8 lines
arch/arm/Kconfig config TEGRA · line 743
Quoted 3 lines
arch/arm/Kconfig +2
-1- select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
Yes, on Tegra SPL runs on an ARMv4t and main U-Boot on some ARMv7 CPU. The change above is actively incorrect.
Quoted 6 lines
The
OF_CONTROLchange is probably incorrect too; we certainly don't intend to useOF_CONTROLin the SPL (there's really nothing to control in the SPL)
- I just don't know if enabling that feature will cause any issue. Things to look out for would be bloat of the SPL binary so that it didn't fit into the space before the main binary'sTEXT_BASE, since the two get concatenated together into a single binary that's loaded into RAM, and XIP'd.
This at least is not a problem with this patch.
I'll make some time to take a look at this in the next few days.
Regards,
Simon
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 34 lines
Hi,
Quoted 31 lines · Stephen Warren
Quoted 30 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 10 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in Kconfig.
So, we only define CPU_V7, for the main processors.
For SPL, we override the "CPU" in config.mk
ifdef CONFIG_SPL_BUILD
ifdef CONFIG_TEGRA
CPU := arm720t
endif
endif
I know what you might be saying is, this is too ugly. Yes.
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Quoted 10 lines
Quoted 9 lines
Yes, on Tegra SPL runs on an ARMv4t and main U-Boot on some ARMv7 CPU. The change above is actively incorrect.
The
OF_CONTROLchange is probably incorrect too; we certainly don't intend to useOF_CONTROLin the SPL (there's really nothing to control in the SPL)
- I just don't know if enabling that feature will cause any issue. Things to look out for would be bloat of the SPL binary so that it didn't fit into the space before the main binary'sTEXT_BASE, since the two get concatenated together into a single binary that's loaded into RAM, and XIP'd.
Quoted 4 lines
This at least is not a problem with this patch.
I'll make some time to take a look at this in the next few days.
As for CONFIG_OF_CONTROL, I think my answer is already in the git-description. I wrote as follows:
- Add some entries to include/config_uncmd_spl.h and the new file scripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.
Signature · 2 lines
Best Regards Masahiro Yamada
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Hi Masahiro,
Quoted 54 lines · Masahiro YAMADA
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 34 lines
Hi,
Quoted 31 lines · Stephen Warren
Quoted 30 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 10 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in
Kconfig.So, we only define
CPU_V7, for the main processors.For SPL, we override the "CPU" in config.mk
ifdef
CONFIG_SPL_BUILD
ifdefCONFIG_TEGRA
CPU := arm720t
endif
endif
Quoted 5 lines
I know what you might be saying is, this is too ugly. Yes.
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Yes I saw that, I understand now. So SPL_BUILD is no longer available in Kconfig, but is still available in Makefiles, right?
Quoted 21 lines
Quoted 13 lines
Quoted 6 lines
Yes, on Tegra SPL runs on an ARMv4t and main U-Boot on some ARMv7 CPU. The change above is actively incorrect.
Quoted 6 lines
The
OF_CONTROLchange is probably incorrect too; we certainly don't intend to useOF_CONTROLin the SPL (there's really nothing to control in the SPL)
- I just don't know if enabling that feature will cause any issue. Things to look out for would be bloat of the SPL binary so that it didn't fit into the space before the main binary'sTEXT_BASE, since the two get concatenated together into a single binary that's loaded into RAM, and XIP'd.Quoted 4 lines
This at least is not a problem with this patch.
I'll make some time to take a look at this in the next few days.
As for
CONFIG_OF_CONTROL, I think my answer is already in the git-description. I wrote as follows:
Quoted 6 lines
- Add some entries to
include/config_uncmd_spl.hand the new filescripts/Makefile.uncmd_spl. Some CONFIG options that are not supported on SPL must be disabled because one .config is shared between SPL and U-Boot proper going forward. I know this is not a beautiful solution and I think we can do better, but let's see how much we will have to describe them.
So I thnk we are OK. I will give it all a test on Tegra.
Regards,
Simon
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Hi Simon,
2015-02-21 11:28 GMT+09:00 Simon Glass <[email protected]>:
Quoted 57 lines
Hi Masahiro,
Quoted 54 lines · Masahiro YAMADA
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 34 lines
Hi,
Quoted 31 lines · Stephen Warren
Quoted 30 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 10 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in
Kconfig.So, we only define
CPU_V7, for the main processors.For SPL, we override the "CPU" in config.mk
ifdef
CONFIG_SPL_BUILD
ifdefCONFIG_TEGRA
CPU := arm720t
endif
endif
Quoted 8 lines
Quoted 5 lines
I know what you might be saying is, this is too ugly. Yes.
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Yes I saw that, I understand now. So
SPL_BUILDis no longer available inKconfig, but is still available in Makefiles, right?
Yes, exactly!
Signature · 2 lines
Best Regards Masahiro Yamada
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Hi Masahiro,
Quoted 63 lines · Masahiro YAMADA
Hi Simon,
2015-02-21 11:28 GMT+09:00 Simon Glass <[email protected]>:
Quoted 59 lines
Hi Masahiro,
Quoted 56 lines · Masahiro YAMADA
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 34 lines
Hi,
Quoted 31 lines · Stephen Warren
Quoted 30 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 10 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in
Kconfig.So, we only define
CPU_V7, for the main processors.For SPL, we override the "CPU" in config.mk
ifdef
CONFIG_SPL_BUILD
ifdefCONFIG_TEGRA
CPU := arm720t
endif
endifI know what you might be saying is, this is too ugly. Yes.
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Yes I saw that, I understand now. So
SPL_BUILDis no longer available inKconfig, but is still available in Makefiles, right?Yes, exactly!
This all works fine on Tegra for me. However I like to suggest dropping a few patches in this series.
I don't think it is worth using ARCH_MALLOC_F_LEN. In fact for me the Tegra defconfig looks OK and SPL is built correctly.
My remaining question is about that Tegra seems to want USE_PRIVATE_LIBGCC for SPL but not for U-Boot. I'm not sure why, nor whether it matters. It seems to work find using it for both.
I have pushed my tested tree to u-boot-x86 in branch single-kconfig-for-masahiro.
It contains only these patches:
ARM: UniPhier: set CONFIG_SYS_MALLOC_F to the global default value
kconfig: Adjust ordering so that defaults work as expected
kconfig: switch to single .config configuration
kconfig: remove unneeded dependency on !SPL_BUILD
Perhaps we can look at the others later once we get things moved over? I think the SYS_MALLOC_F thing is not that big a deal and not urgent to resolve if we still have a problem.
Regards,
Simon
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Quoted 78 lines · Simon Glass
Hi Masahiro,
Quoted 63 lines · Masahiro YAMADA
Hi Simon,
2015-02-21 11:28 GMT+09:00 Simon Glass <[email protected]>:
Quoted 59 lines
Hi Masahiro,
Quoted 56 lines · Masahiro YAMADA
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 34 lines
Hi,
Quoted 31 lines · Stephen Warren
Quoted 30 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 10 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in
Kconfig.So, we only define
CPU_V7, for the main processors.For SPL, we override the "CPU" in config.mk
ifdef
CONFIG_SPL_BUILD
ifdefCONFIG_TEGRA
CPU := arm720t
endif
endifI know what you might be saying is, this is too ugly. Yes.
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Yes I saw that, I understand now. So
SPL_BUILDis no longer available inKconfig, but is still available in Makefiles, right?Yes, exactly!
This all works fine on Tegra for me. However I like to suggest dropping a few patches in this series.
Quoted 6 lines
I don't think it is worth using
ARCH_MALLOC_F_LEN. In fact for me the Tegra defconfig looks OK and SPL is built correctly.My remaining question is about that Tegra seems to want
USE_PRIVATE_LIBGCCfor SPL but not for U-Boot. I'm not sure why, nor whether it matters. It seems to work find using it for both.
Depending on the toolchain, we actively need USE_PRIVATE_LIBGCC for SPL, and don't /need/ it for non-SPL. However, enabling USE_PRIVATE_LIBGCC for non-SPL likely won't hurt.
The issue is that the libgcc bundled with most compilers is for ARMv7 (since we tend to use ARMv7 compilers, since the main U-boot is built for ARMv7). That bundled libgcc won't work on the ARMv4 that runs the SPL, so we need USE_PRIVATE_LIBGCC there. The private libgcc bundled with U-Boot should work fine when built for either CPU, so it is OK to always use it, rather than only use it when strictly needed.
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Hi Stephen,
Quoted 110 lines · Stephen Warren
Quoted 97 lines · Simon Glass
Hi Masahiro,
Quoted 82 lines · Masahiro YAMADA
Hi Simon,
2015-02-21 11:28 GMT+09:00 Simon Glass <[email protected]>:
Quoted 77 lines
Hi Masahiro,
Quoted 71 lines · Masahiro YAMADA
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 44 lines
Hi,
Quoted 39 lines · Stephen Warren
Quoted 37 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 12 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the
SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only
reason that Tegra has SPL. Doesn't this change with this commit?No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in
Kconfig.So, we only define
CPU_V7, for the main processors.For SPL, we override the "CPU" in config.mk
ifdef
CONFIG_SPL_BUILD
ifdefCONFIG_TEGRA
CPU := arm720t
endif
endifI know what you might be saying is, this is too ugly. Yes.
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Quoted 5 lines
Quoted 2 lines
Yes I saw that, I understand now. So
SPL_BUILDis no longer available inKconfig, but is still available in Makefiles, right?Yes, exactly!
This all works fine on Tegra for me. However I like to suggest dropping a few patches in this series.
Quoted 6 lines
I don't think it is worth using
ARCH_MALLOC_F_LEN. In fact for me the Tegra defconfig looks OK and SPL is built correctly.My remaining question is about that Tegra seems to want
USE_PRIVATE_LIBGCCfor SPL but not for U-Boot. I'm not sure why, nor whether it matters. It seems to work find using it for both.Depending on the toolchain, we actively need
USE_PRIVATE_LIBGCCfor SPL, and don't /need/ it for non-SPL. However, enablingUSE_PRIVATE_LIBGCCfor non-SPL likely won't hurt.
Quoted 6 lines
The issue is that the libgcc bundled with most compilers is for ARMv7 (since we tend to use ARMv7 compilers, since the main U-boot is built for ARMv7). That bundled libgcc won't work on the ARMv4 that runs the SPL, so we need
USE_PRIVATE_LIBGCCthere. The private libgcc bundled with U-Boot should work fine when built for either CPU, so it is OK to always use it, rather than only use it when strictly needed.
Ah yes, I think I knew that once. So in short Masahiro's patch here should be fine.
Regards,
Simon
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Hi Simon, Stephen,
Quoted 120 lines · On Mon, 23 Feb 2015 10:44:54 -0700 Simon Glass <[email protected]> wrote:
Hi Stephen,
Quoted 110 lines · Stephen Warren
Quoted 97 lines · Simon Glass
Hi Masahiro,
Quoted 82 lines · Masahiro YAMADA
Hi Simon,
2015-02-21 11:28 GMT+09:00 Simon Glass <[email protected]>:
Quoted 77 lines
Hi Masahiro,
Quoted 71 lines · Masahiro YAMADA
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 44 lines
Hi,
Quoted 39 lines · Stephen Warren
Quoted 37 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 12 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the
SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only
reason that Tegra has SPL. Doesn't this change with this commit?No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in
Kconfig.So, we only define
CPU_V7, for the main processors.For SPL, we override the "CPU" in config.mk
ifdef
CONFIG_SPL_BUILD
ifdefCONFIG_TEGRA
CPU := arm720t
endif
endifI know what you might be saying is, this is too ugly. Yes.
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Quoted 5 lines
Quoted 2 lines
Yes I saw that, I understand now. So
SPL_BUILDis no longer available inKconfig, but is still available in Makefiles, right?Yes, exactly!
This all works fine on Tegra for me. However I like to suggest dropping a few patches in this series.
Quoted 6 lines
I don't think it is worth using
ARCH_MALLOC_F_LEN. In fact for me the Tegra defconfig looks OK and SPL is built correctly.My remaining question is about that Tegra seems to want
USE_PRIVATE_LIBGCCfor SPL but not for U-Boot. I'm not sure why, nor whether it matters. It seems to work find using it for both.Depending on the toolchain, we actively need
USE_PRIVATE_LIBGCCfor SPL, and don't /need/ it for non-SPL. However, enablingUSE_PRIVATE_LIBGCCfor non-SPL likely won't hurt.Quoted 6 lines
The issue is that the libgcc bundled with most compilers is for ARMv7 (since we tend to use ARMv7 compilers, since the main U-boot is built for ARMv7). That bundled libgcc won't work on the ARMv4 that runs the SPL, so we need
USE_PRIVATE_LIBGCCthere. The private libgcc bundled with U-Boot should work fine when built for either CPU, so it is OK to always use it, rather than only use it when strictly needed.
Quoted 3 lines
Ah yes, I think I knew that once. So in short Masahiro's patch here should be fine.
Yes, Stephen explained all about my intention.
I think CONFIG_USE_PRIVATE_LIBGCC is also necessary for Raspberry Pi 1 for example.
Moreover, I had already posted this patch:
http://patchwork.ozlabs.org/patch/438360/
I'd like to expand the private library to all the ARM boards.
Linux Kernel includes the library in its source tree.
I think it is generally a good idea to reduce the depencendy on particular toolchains. Agree?
Best Regards
Masahiro Yamada
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
Quoted 127 lines · Masahiro Yamada
Hi Simon, Stephen,
Quoted 120 lines · On Mon, 23 Feb 2015 10:44:54 -0700 Simon Glass <[email protected]> wrote:
Hi Stephen,
Quoted 116 lines · Stephen Warren
Quoted 97 lines · Simon Glass
Hi Masahiro,
Quoted 82 lines · Masahiro YAMADA
Hi Simon,
2015-02-21 11:28 GMT+09:00 Simon Glass <[email protected]>:
Quoted 77 lines
Hi Masahiro,
Quoted 71 lines · Masahiro YAMADA
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 44 lines
Hi,
Quoted 39 lines · Stephen Warren
Quoted 37 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 12 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the
SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only
reason that Tegra has SPL. Doesn't this change with this commit?No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in
Kconfig.So, we only define
CPU_V7, for the main processors.For SPL, we override the "CPU" in config.mk
ifdef
CONFIG_SPL_BUILD
ifdefCONFIG_TEGRA
CPU := arm720t
endif
endifI know what you might be saying is, this is too ugly. Yes.
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Quoted 5 lines
Quoted 2 lines
Yes I saw that, I understand now. So
SPL_BUILDis no longer available inKconfig, but is still available in Makefiles, right?Yes, exactly!
This all works fine on Tegra for me. However I like to suggest dropping a few patches in this series.
Quoted 6 lines
I don't think it is worth using
ARCH_MALLOC_F_LEN. In fact for me the Tegra defconfig looks OK and SPL is built correctly.My remaining question is about that Tegra seems to want
USE_PRIVATE_LIBGCCfor SPL but not for U-Boot. I'm not sure why, nor whether it matters. It seems to work find using it for both.Depending on the toolchain, we actively need
USE_PRIVATE_LIBGCCfor SPL, and don't /need/ it for non-SPL. However, enablingUSE_PRIVATE_LIBGCCfor non-SPL likely won't hurt.The issue is that the libgcc bundled with most compilers is for ARMv7 (since we tend to use ARMv7 compilers, since the main U-boot is built for ARMv7). That bundled libgcc won't work on the ARMv4 that runs the SPL, so we need
USE_PRIVATE_LIBGCCthere. The private libgcc bundled with U-Boot should work fine when built for either CPU, so it is OK to always use it, rather than only use it when strictly needed.
Quoted 8 lines
Quoted 3 lines
Ah yes, I think I knew that once. So in short Masahiro's patch here should be fine.
Yes, Stephen explained all about my intention.
I think
CONFIG_USE_PRIVATE_LIBGCCis also necessary for Raspberry Pi 1 for example.
I can't remember if it's already set for the Pi. If not, the toolchains I use happen not to need it:-)
Quoted 10 lines
Moreover, I had already posted this patch:
http://patchwork.ozlabs.org/patch/438360/I'd like to expand the private library to all the ARM boards.
Linux Kernel includes the library in its source tree.
I think it is generally a good idea to reduce the depencendy on particular toolchains. Agree?
I tend to agree. However, in the past, Wolfgang Denk has argued against (ever?) using that option, claiming people should just use the correct toolchain.
PATCH v3 6/7 kconfig: remove unneeded dependency on !SPL_BUILD
In reply to Simon GlassHi Simon,
Quoted 90 lines · On Mon, 23 Feb 2015 07:02:25 -0700 Simon Glass <[email protected]> wrote:
Hi Masahiro,
Quoted 63 lines · Masahiro YAMADA
Hi Simon,
2015-02-21 11:28 GMT+09:00 Simon Glass <[email protected]>:
Quoted 59 lines
Hi Masahiro,
Quoted 56 lines · Masahiro YAMADA
Hi Simon, Stephen,
2015-02-21 3:39 GMT+09:00 Simon Glass <[email protected]>:
Quoted 34 lines
Hi,
Quoted 31 lines · Stephen Warren
Quoted 30 lines · Simon Glass
+Stephen
Hi Masahiro,
Quoted 10 lines · Masahiro Yamada
Now
CONFIG_SPL_BUILDis not defined inKconfig, so "!depends onSPL_BUILD" and "if !SPL_BUILD" are redundant.arch/arm/Kconfig +0
-0arch/arm/Kconfig config TEGRA · line 743
Quoted 8 lines
arch/arm/Kconfig +2
-3@@ -739,9 +739,8 @@ config TEGRA bool "NVIDIA Tegra" select SUPPORT_SPL select SPL - select OF_CONTROL if !SPL_BUILD - select CPU_ARM720T if SPL_BUILD - select CPU_V7 if !SPL_BUILD + select OF_CONTROL + select CPU_V7Sorry if I have missed something here. On Tegra most unfortunately the SPL uses ARMv4t and U-Boot proper uses ARMv7. In fact that is the only reason that Tegra has SPL. Doesn't this change with this commit?
No. I think behavior is still the same as before.
In a single .config, we cannot define two CPUs in
Kconfig.So, we only define
CPU_V7, for the main processors.For SPL, we override the "CPU" in config.mk
ifdef
CONFIG_SPL_BUILD
ifdefCONFIG_TEGRA
CPU := arm720t
endif
endifI know what you might be saying is, this is too ugly. Yes.
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
I think we can do a little better with further rafactoring, but the basic idea is, SPL of Tegra is a special case.
Yes I saw that, I understand now. So
SPL_BUILDis no longer available inKconfig, but is still available in Makefiles, right?Yes, exactly!
This all works fine on Tegra for me. However I like to suggest dropping a few patches in this series.
I don't think it is worth using
ARCH_MALLOC_F_LEN. In fact for me the Tegra defconfig looks OK and SPL is built correctly.My remaining question is about that Tegra seems to want
USE_PRIVATE_LIBGCCfor SPL but not for U-Boot. I'm not sure why, nor whether it matters. It seems to work find using it for both.I have pushed my tested tree to u-boot-x86 in branch single-kconfig-for-masahiro.
It contains only these patches:
Quoted 8 lines
ARM: UniPhier: set
CONFIG_SYS_MALLOC_Fto the global default value
kconfig: Adjust ordering so that defaults work as expected
kconfig: switch to single .config configuration
kconfig: remove unneeded dependency on !SPL_BUILDPerhaps we can look at the others later once we get things moved over? I think the
SYS_MALLOC_Fthing is not that big a deal and not urgent to resolve if we still have a problem.
OK. Let's go with these four.
Including SYS_MALLOC_F things makes the point of this series blurred.
I have just posted v5.
Mashiro Yamada
PATCH v3 7/7 malloc_f: enable SYS_MALLOC_F by default if DM is on
In reply to Masahiro YamadaTo use Derive Model before relocation, CONFIG_SYS_MALLOC_F must be enabled. This should probably be a common requirement for all the boards with Driver Model implementation. Let's handle it globally rather than per-SoC or per-board.
---
Changes in v3:
- Newly added
Changes in v2: None
Kconfig | 1 + arch/arm/cpu/armv7/exynos/Kconfig | 3 --- arch/arm/cpu/armv7/omap3/Kconfig | 3 --- arch/arm/cpu/armv7/tegra-common/Kconfig | 3 --- arch/arm/cpu/armv7/uniphier/Kconfig | 3 --- arch/x86/Kconfig | 3 --- board/amcc/canyonlands/Kconfig | 4 ---- board/ti/am335x/Kconfig | 3 --- 8 files changed, 1 insertion(+), 22 deletions(-)
Kconfig +1 -0
@@ -54,6 +54,7 @@ config CC_OPTIMIZE_FOR_SIZE config SYS_MALLOC_F bool "Enable malloc() pool before relocation" + default y if DM help Before relocation memory is very limited on many platforms. Still, we can provide a small malloc() pool if needed. Driver model in
arch/arm/cpu/armv7/exynos/Kconfig +0 -3
@@ -80,9 +80,6 @@ config DM_SPI_FLASH config DM_GPIO default y -config SYS_MALLOC_F - default y - source "board/samsung/smdkv310/Kconfig" source "board/samsung/trats/Kconfig" source "board/samsung/universal_c210/Kconfig"
arch/arm/cpu/armv7/omap3/Kconfig +0 -3
@@ -102,9 +102,6 @@ config DM_GPIO config DM_SERIAL default y if DM -config SYS_MALLOC_F - default y if DM - config SYS_SOC default "omap3"
arch/arm/cpu/armv7/tegra-common/Kconfig +0 -3
@@ -17,9 +17,6 @@ config TEGRA124 endchoice -config SYS_MALLOC_F - default y - config ARCH_MALLOC_F_LEN hex default 0x1800
arch/arm/cpu/armv7/uniphier/Kconfig +0 -3
@@ -48,9 +48,6 @@ config DCC_MICRO_SUPPORT_CARD endchoice -config SYS_MALLOC_F - default y - config CMD_PINMON bool "Enable boot mode pins monitor command" default y
arch/x86/Kconfig +0 -3
@@ -76,9 +76,6 @@ config DM_GPIO config DM_SERIAL default y -config SYS_MALLOC_F - default y - config ARCH_MALLOC_F_LEN hex default 0x800
board/amcc/canyonlands/Kconfig +0 -4
@@ -39,8 +39,4 @@ config DM config DM_SERIAL default y -config SYS_MALLOC_F - bool - default y - endif
board/ti/am335x/Kconfig +0 -3
@@ -47,7 +47,4 @@ config DM_GPIO config DM_SERIAL default y if DM -config SYS_MALLOC_F - default y if DM - endif
Signature · 1 line
1.9.1
PATCH v3 7/7 malloc_f: enable SYS_MALLOC_F by default if DM is on
Hi Masahiro,
Quoted 1 line · Masahiro Yamada
To use Derive Model before relocation,
CONFIG_SYS_MALLOC_F
Driver
Quoted 3 lines
must be enabled. This should probably be a common requirement for all the boards with Driver Model implementation. Let's handle it globally rather than per-SoC or per-board.
This seems reasonable to me.
Quoted 127 lines
Signed-off-by Masahiro Yamada <[email protected]>---
Changes in v3:
- Newly addedChanges in v2: None
Kconfig | 1 + arch/arm/cpu/armv7/exynos/Kconfig | 3 --- arch/arm/cpu/armv7/omap3/Kconfig | 3 --- arch/arm/cpu/armv7/tegra-common/Kconfig | 3 --- arch/arm/cpu/armv7/uniphier/Kconfig | 3 --- arch/x86/Kconfig | 3 --- board/amcc/canyonlands/Kconfig | 4 ---- board/ti/am335x/Kconfig | 3 --- 8 files changed, 1 insertion(+), 22 deletions(-)Kconfig +1
-0@@ -54,6 +54,7 @@ config CC_OPTIMIZE_FOR_SIZE config SYS_MALLOC_F bool "Enable malloc() pool before relocation" + default y if DM help Before relocation memory is very limited on many platforms. Still, we can provide a small malloc() pool if needed. Driver model inarch/arm/cpu/armv7/exynos/Kconfig +0
-3@@ -80,9 +80,6 @@ config DM_SPI_FLASH config DM_GPIO default y -config SYS_MALLOC_F - default y - source "board/samsung/smdkv310/Kconfig" source "board/samsung/trats/Kconfig" source "board/samsung/universal_c210/Kconfig"arch/arm/cpu/armv7/omap3/Kconfig +0
-3@@ -102,9 +102,6 @@ config DM_GPIO config DM_SERIAL default y if DM -config SYS_MALLOC_F - default y if DM - config SYS_SOC default "omap3"arch/arm/cpu/armv7/tegra-common/Kconfig +0
-3@@ -17,9 +17,6 @@ config TEGRA124 endchoice -config SYS_MALLOC_F - default y - config ARCH_MALLOC_F_LEN hex default 0x1800arch/arm/cpu/armv7/uniphier/Kconfig +0
-3@@ -48,9 +48,6 @@ config DCC_MICRO_SUPPORT_CARD endchoice -config SYS_MALLOC_F - default y - config CMD_PINMON bool "Enable boot mode pins monitor command" default yarch/x86/Kconfig +0
-3@@ -76,9 +76,6 @@ config DM_GPIO config DM_SERIAL default y -config SYS_MALLOC_F - default y - config ARCH_MALLOC_F_LEN hex default 0x800board/amcc/canyonlands/Kconfig +0
-4@@ -39,8 +39,4 @@ config DM config DM_SERIAL default y -config SYS_MALLOC_F - bool - default y - endifboard/ti/am335x/Kconfig +0
-4@@ -47,7 +47,4 @@ config DM_GPIO config DM_SERIAL default y if DM -config SYS_MALLOC_F - default y if DM - endif --
Quoted 2 lines
1.9.1
Regards,
Simon