[PATCH 0/5] lib: reduce SPL size
Several configuration options inject code into SPL even if it is not needed.
Heinrich Schuchardt (5): lib: do not build OID registry in SPL lib: date functions in SPL common: image_sign_info helper functions in SPL lib: do not provide hexdump in SPL dlmalloc: remove unit test support in SPL
Kconfig | 2 +- common/Kconfig | 11 +++++++++++ common/Makefile | 2 +- common/dlmalloc.c | 2 +- lib/Makefile | 6 +++--- lib/hexdump.c | 2 +- 6 files changed, 18 insertions(+), 7 deletions(-)
-- 2.25.1
The OID registry is only used by crypto functions that are not built in SPL. So we should not build it in SPL.
Fixes: a9b45e6e8382 ("lib: add oid registry utility") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de --- lib/Makefile | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/lib/Makefile b/lib/Makefile index 32bf3f3693..5d4bb3232c 100644 --- a/lib/Makefile +++ b/lib/Makefile @@ -119,6 +119,7 @@ obj-$(CONFIG_$(SPL_TPL_)STRTO) += strto.o else # Main U-Boot always uses the full printf support obj-y += vsprintf.o strto.o +obj-$(CONFIG_OID_REGISTRY) += oid_registry.o endif
obj-y += date.o @@ -127,8 +128,6 @@ obj-$(CONFIG_LIB_ELF) += elf.o # # Build a fast OID lookup registry from include/linux/oid_registry.h # -obj-$(CONFIG_OID_REGISTRY) += oid_registry.o - $(obj)/oid_registry.o: $(obj)/oid_registry_data.c
$(obj)/oid_registry_data.c: $(srctree)/include/linux/oid_registry.h \ -- 2.25.1
On Wed, Apr 15, 2020 at 06:46:19PM +0200, Heinrich Schuchardt wrote:
The OID registry is only used by crypto functions that are not built in SPL. So we should not build it in SPL.
Fixes: a9b45e6e8382 ("lib: add oid registry utility") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
lib/Makefile | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
This problem shows up with the full secure boot patch series, yes? Thanks.
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:19PM +0200, Heinrich Schuchardt wrote:
The OID registry is only used by crypto functions that are not built in SPL. So we should not build it in SPL.
Fixes: a9b45e6e8382 ("lib: add oid registry utility") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
lib/Makefile | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
This problem shows up with the full secure boot patch series, yes? Thanks.
Take origin/master:
pine64-lts_defconfig + CONFIG_RSA=y + CONFIG_ASYMMETRIC_KEY_TYPE=y + CONFIG_ASYMMETRIC_PUBLIC_KEY_SUBTYPE=y + CONFIG_RSA_PUBLIC_KEY_PARSER=y
and you will see spl/lib/oid_registry.o being built.
These settings will in future be needed for UEFI secure boot.
Best regards
Heinrich
On Thu, Apr 16, 2020 at 05:49:13PM +0200, Heinrich Schuchardt wrote:
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:19PM +0200, Heinrich Schuchardt wrote:
The OID registry is only used by crypto functions that are not built in SPL. So we should not build it in SPL.
Fixes: a9b45e6e8382 ("lib: add oid registry utility") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
lib/Makefile | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
This problem shows up with the full secure boot patch series, yes? Thanks.
Take origin/master:
pine64-lts_defconfig
- CONFIG_RSA=y
- CONFIG_ASYMMETRIC_KEY_TYPE=y
- CONFIG_ASYMMETRIC_PUBLIC_KEY_SUBTYPE=y
- CONFIG_RSA_PUBLIC_KEY_PARSER=y
and you will see spl/lib/oid_registry.o being built.
These settings will in future be needed for UEFI secure boot.
Thanks.
Reviewed-by: Tom Rini trini@konsulko.com
On Wed, Apr 15, 2020 at 06:46:19PM +0200, Heinrich Schuchardt wrote:
The OID registry is only used by crypto functions that are not built in SPL. So we should not build it in SPL.
Fixes: a9b45e6e8382 ("lib: add oid registry utility") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de Reviewed-by: Tom Rini trini@konsulko.com
Applied to u-boot/master, thanks!
Date functions are only needed in SPL if SPL contains RTC drivers.
Fixes: 05429b6cf5b3 ("rtc: move date.c from drivers/rtc/ to lib/") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de --- lib/Makefile | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/lib/Makefile b/lib/Makefile index 5d4bb3232c..dbea69f52c 100644 --- a/lib/Makefile +++ b/lib/Makefile @@ -116,13 +116,14 @@ else obj-$(CONFIG_$(SPL_TPL_)SPRINTF) += vsprintf.o endif obj-$(CONFIG_$(SPL_TPL_)STRTO) += strto.o +obj-$(CONFIG_$(SPL_TPL_)RTC_SUPPORT) += date.o else # Main U-Boot always uses the full printf support obj-y += vsprintf.o strto.o obj-$(CONFIG_OID_REGISTRY) += oid_registry.o +obj-y += date.o endif
-obj-y += date.o obj-$(CONFIG_LIB_ELF) += elf.o
# -- 2.25.1
On Wed, Apr 15, 2020 at 06:46:20PM +0200, Heinrich Schuchardt wrote:
Date functions are only needed in SPL if SPL contains RTC drivers.
Fixes: 05429b6cf5b3 ("rtc: move date.c from drivers/rtc/ to lib/") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:20PM +0200, Heinrich Schuchardt wrote:
Date functions are only needed in SPL if SPL contains RTC drivers.
Fixes: 05429b6cf5b3 ("rtc: move date.c from drivers/rtc/ to lib/") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
We are compiling ./spl/lib/date.o for no good reason since said patch. It may be that the when creating u-boot-spl.bin from u-boot-spl this gets deleted again due to linker optimization.
Best regards
Heinrich
On Thu, Apr 16, 2020 at 06:30:48PM +0200, Heinrich Schuchardt wrote:
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:20PM +0200, Heinrich Schuchardt wrote:
Date functions are only needed in SPL if SPL contains RTC drivers.
Fixes: 05429b6cf5b3 ("rtc: move date.c from drivers/rtc/ to lib/") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
We are compiling ./spl/lib/date.o for no good reason since said patch. It may be that the when creating u-boot-spl.bin from u-boot-spl this gets deleted again due to linker optimization.
It is discarded at link time, yes. I'm generally dis-inclined to grab patches like this when it adds more to the Makefile without a size savings.
Do not build image_sign_info helper functions in SPL if not needed.
Fixes: b983cc2da0ba ("lib: rsa: decouple rsa from FIT image verification") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de --- Kconfig | 2 +- common/Kconfig | 11 +++++++++++ common/Makefile | 2 +- 3 files changed, 13 insertions(+), 2 deletions(-)
diff --git a/Kconfig b/Kconfig index 1b0b6999d8..6038e93afb 100644 --- a/Kconfig +++ b/Kconfig @@ -447,7 +447,7 @@ config SPL_FIT_SIGNATURE select SPL_FIT select SPL_RSA select SPL_RSA_VERIFY - select IMAGE_SIGN_INFO + select SPL_IMAGE_SIGN_INFO
config SPL_LOAD_FIT bool "Enable SPL loading U-Boot as a FIT (basic fitImage features)" diff --git a/common/Kconfig b/common/Kconfig index 3072651082..75d9dfdc33 100644 --- a/common/Kconfig +++ b/common/Kconfig @@ -1053,3 +1053,14 @@ config IMAGE_SIGN_INFO select SHA256 help Enable image_sign_info helper functions. + +if IMAGE_SIGN_INFO + +config SPL_IMAGE_SIGN_INFO + bool + select SHA1 + select SHA256 + help + Enable image_sign_info helper functions in SPL. + +endif diff --git a/common/Makefile b/common/Makefile index 702f2396cf..ad030e05eb 100644 --- a/common/Makefile +++ b/common/Makefile @@ -112,7 +112,7 @@ obj-$(CONFIG_ANDROID_BOOT_IMAGE) += image-android.o image-android-dt.o obj-$(CONFIG_$(SPL_TPL_)OF_LIBFDT) += image-fdt.o obj-$(CONFIG_$(SPL_TPL_)FIT) += image-fit.o obj-$(CONFIG_$(SPL_)MULTI_DTB_FIT) += boot_fit.o common_fit.o -obj-$(CONFIG_IMAGE_SIGN_INFO) += image-sig.o +obj-$(CONFIG_$(SPL_TPL_)IMAGE_SIGN_INFO) += image-sig.o obj-$(CONFIG_$(SPL_TPL_)FIT_SIGNATURE) += image-fit-sig.o obj-$(CONFIG_$(SPL_TPL_)FIT_CIPHER) += image-cipher.o obj-$(CONFIG_IO_TRACE) += iotrace.o -- 2.25.1
On Wed, Apr 15, 2020 at 06:46:21PM +0200, Heinrich Schuchardt wrote:
Do not build image_sign_info helper functions in SPL if not needed.
Fixes: b983cc2da0ba ("lib: rsa: decouple rsa from FIT image verification") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
Reviewed-by: Tom Rini trini@konsulko.com
On Wed, Apr 15, 2020 at 06:46:21PM +0200, Heinrich Schuchardt wrote:
Do not build image_sign_info helper functions in SPL if not needed.
Fixes: b983cc2da0ba ("lib: rsa: decouple rsa from FIT image verification") Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de Reviewed-by: Tom Rini trini@konsulko.com
Applied to u-boot/master, thanks!
SPL should not be enlarged by building with CONFIG_HEXDUMP=y.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de --- lib/hexdump.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/hexdump.c b/lib/hexdump.c index bf14b5bdbd..a3f219a874 100644 --- a/lib/hexdump.c +++ b/lib/hexdump.c @@ -18,7 +18,7 @@ const char hex_asc[] = "0123456789abcdef"; const char hex_asc_upper[] = "0123456789ABCDEF";
-#ifdef CONFIG_HEXDUMP +#if CONFIG_IS_ENABLED(HEXDUMP) /** * hex_dump_to_buffer - convert a blob of data to "hex ASCII" in memory * @buf: data blob to dump -- 2.25.1
On Wed, Apr 15, 2020 at 06:46:22PM +0200, Heinrich Schuchardt wrote:
SPL should not be enlarged by building with CONFIG_HEXDUMP=y.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
lib/hexdump.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:22PM +0200, Heinrich Schuchardt wrote:
SPL should not be enlarged by building with CONFIG_HEXDUMP=y.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
lib/hexdump.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
I checked for origin/master pine64_defconfig
with HEXDUMP 1493728 spl/u-boot-spl
without HEXDUMP 1485712 spl/u-boot-spl
Currently no driver is actually calling the hexdump function is SPL. So the code seems to be removed in the final binary.
Best regards
Heinrich
On Thu, Apr 16, 2020 at 06:27:00PM +0200, Heinrich Schuchardt wrote:
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:22PM +0200, Heinrich Schuchardt wrote:
SPL should not be enlarged by building with CONFIG_HEXDUMP=y.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
lib/hexdump.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
I checked for origin/master pine64_defconfig
with HEXDUMP 1493728 spl/u-boot-spl
without HEXDUMP 1485712 spl/u-boot-spl
Currently no driver is actually calling the hexdump function is SPL. So the code seems to be removed in the final binary.
OK, thanks.
Reviewed-by: Tom Rini trini@konsulko.com
On Wed, Apr 15, 2020 at 06:46:22PM +0200, Heinrich Schuchardt wrote:
SPL should not be enlarged by building with CONFIG_HEXDUMP=y.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de Reviewed-by: Tom Rini trini@konsulko.com
Applied to u-boot/master, thanks!
We cannot run unit tests in SPL. So remove the unit test support.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de --- common/dlmalloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/common/dlmalloc.c b/common/dlmalloc.c index dade68faf7..db5ab55ed3 100644 --- a/common/dlmalloc.c +++ b/common/dlmalloc.c @@ -1,6 +1,6 @@ #include <common.h>
-#if defined(CONFIG_UNIT_TEST) +#if CONFIG_IS_ENABLED(UNIT_TEST) #define DEBUG #endif
-- 2.25.1
On Wed, Apr 15, 2020 at 06:46:23PM +0200, Heinrich Schuchardt wrote:
We cannot run unit tests in SPL. So remove the unit test support.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
common/dlmalloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:23PM +0200, Heinrich Schuchardt wrote:
We cannot run unit tests in SPL. So remove the unit test support.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
common/dlmalloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
pine64-lts_defconfig + CONFIG_UNIT_TEST=y:
without patch: 1487904 spl/u-boot-spl
with patch: 1485712 spl/u-boot-spl
The difference seems to be in the size of debug sections.
Best regards
Heinrich
On Thu, Apr 16, 2020 at 05:37:56PM +0200, Heinrich Schuchardt wrote:
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:23PM +0200, Heinrich Schuchardt wrote:
We cannot run unit tests in SPL. So remove the unit test support.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
common/dlmalloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
pine64-lts_defconfig + CONFIG_UNIT_TEST=y:
without patch: 1487904 spl/u-boot-spl
with patch: 1485712 spl/u-boot-spl
The difference seems to be in the size of debug sections.
Interesting, what toolchain are you using? That doesn't show up here. Thanks!
On 16.04.20 19:00, Tom Rini wrote:
On Thu, Apr 16, 2020 at 05:37:56PM +0200, Heinrich Schuchardt wrote:
On 4/16/20 4:32 AM, Tom Rini wrote:
On Wed, Apr 15, 2020 at 06:46:23PM +0200, Heinrich Schuchardt wrote:
We cannot run unit tests in SPL. So remove the unit test support.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de --- common/dlmalloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Does this show up as a problem somewhere? I don't see this resulting in size savings anywhere, thanks.
pine64-lts_defconfig + CONFIG_UNIT_TEST=y:
without patch: 1487904 spl/u-boot-spl
with patch: 1485712 spl/u-boot-spl
The difference seems to be in the size of debug sections.
Interesting, what toolchain are you using? That doesn't show up here. Thanks!
I am on Debian Bullseye gcc-9.2.
As you already remarked replying to another patch the linker is stripping the differences. My idea is that we should only compile the necessary.
Best regards
Heinrich
On Wed, Apr 15, 2020 at 06:46:23PM +0200, Heinrich Schuchardt wrote:
We cannot run unit tests in SPL. So remove the unit test support.
Signed-off-by: Heinrich Schuchardt xypron.glpk@gmx.de
Applied to u-boot/master, thanks!
participants (2)
-
Heinrich Schuchardt -
Tom Rini