[PATCH 0/4] Various build dependency fixes

[PATCH 0/4] Various build dependency fixes

[Jan Kiszka]

See patches for details.

Jan

CC: Andrew F. Davis <afd@ti.com>
CC: Heinrich Schuchardt <xypron.glpk@gmx.de>
CC: Jan Kiszka <jan.kiszka@siemens.com>
CC: Jean-Jacques Hiblot <jjhiblot@ti.com>

Jan Kiszka (4):
  kbuild: efi: Avoid rebuilding efi targets
  kbuild: spl: Fix parallel build
  kbuild: spl: Add shrunk arch-dtbs to targets list
  kbuild: arm: Fix duplicate builds of dtbs

 arch/arm/dts/Makefile             |  6 ++++--
 arch/arm/mach-k3/config_secure.mk | 19 ++++++++++++++-----
 lib/efi_loader/Makefile           |  1 +
 scripts/Makefile.lib              |  2 ++
 scripts/Makefile.spl              |  7 +++++--
 5 files changed, 26 insertions(+), 9 deletions(-)

--
2.26.1

[PATCH 1/4] kbuild: efi: Avoid rebuilding efi targets

[Jan Kiszka]

From: Jan Kiszka <jan.kiszka@siemens.com>

Add a couple of missing targets so that helloworld and other efi targets
are not needlessly rebuilt.

CC: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 lib/efi_loader/Makefile | 1 +
 scripts/Makefile.lib    | 2 ++
 2 files changed, 3 insertions(+)

diff --git a/lib/efi_loader/Makefile b/lib/efi_loader/Makefile
index eff3c25ec3..84d61df55b 100644
--- a/lib/efi_loader/Makefile
+++ b/lib/efi_loader/Makefile
@@ -17,6 +17,7 @@ CFLAGS_REMOVE_helloworld.o := $(CFLAGS_NON_EFI)

 ifneq ($(CONFIG_CMD_BOOTEFI_HELLO_COMPILE),)
 always += helloworld.efi
+targets += helloworld.o
 endif

 obj-$(CONFIG_CMD_BOOTEFI_HELLO) += helloworld_efi.o
diff --git a/scripts/Makefile.lib b/scripts/Makefile.lib
index 63fbadd757..734001c952 100644
--- a/scripts/Makefile.lib
+++ b/scripts/Makefile.lib
@@ -418,6 +418,8 @@ $(obj)/efi_reloc.o: $(srctree)/arch/$(ARCH)/lib/$(EFI_RELOC:.o=.c) $(recordmcoun
 $(obj)/%_efi.so: $(obj)/%.o $(obj)/efi_crt0.o $(obj)/efi_reloc.o $(obj)/efi_freestanding.o
 	$(call cmd,efi_ld)

+targets += $(obj)/efi_crt0.o $(obj)/efi_reloc.o $(obj)/efi_freestanding.o
+
 # ACPI
 # ---------------------------------------------------------------------------
 #
--
2.26.1

[PATCH 2/4] kbuild: spl: Fix parallel build

[Jan Kiszka]

From: Jan Kiszka <jan.kiszka@siemens.com>

The dts dir must exists when running this rule.

That missing dependency broke e.g. "make -j" for the am65x targets.

Fixes: 2f57c95100f2 ("spl: dm: Make it possible for the SPL to pick its own DTB from a FIT")
CC: Jean-Jacques Hiblot <jjhiblot@ti.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 scripts/Makefile.spl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/Makefile.spl b/scripts/Makefile.spl
index 6741ef911e..63ce5caf23 100644
--- a/scripts/Makefile.spl
+++ b/scripts/Makefile.spl
@@ -461,7 +461,7 @@ dtbs:

 SHRUNK_ARCH_DTB = $(patsubst %,$(obj)/dts/%.dtb,$(subst ",,$(CONFIG_SPL_OF_LIST)))
 .SECONDEXPANSION:
-$(SHRUNK_ARCH_DTB): $$(patsubst $(obj)/dts/%, arch/$(ARCH)/dts/%, $$@)
+$(SHRUNK_ARCH_DTB): $$(patsubst $(obj)/dts/%, arch/$(ARCH)/dts/%, $$@) dts_dir
 	$(call if_changed,fdtgrep)

 MKIMAGEFLAGS_$(SPL_BIN).multidtb.fit = -f auto -A $(ARCH) -T firmware -C none -O u-boot \
--
2.26.1

[PATCH 3/4] kbuild: spl: Add shrunk arch-dtbs to targets list

[Jan Kiszka]

From: Jan Kiszka <jan.kiszka@siemens.com>

This avoids needless rebuilding.

Fixes: 2f57c95100f2 ("spl: dm: Make it possible for the SPL to pick its own DTB from a FIT")
CC: Jean-Jacques Hiblot <jjhiblot@ti.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 scripts/Makefile.spl | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/scripts/Makefile.spl b/scripts/Makefile.spl
index 63ce5caf23..e6d56a1286 100644
--- a/scripts/Makefile.spl
+++ b/scripts/Makefile.spl
@@ -459,11 +459,14 @@ dtbs:
 # information in a variable so we can use it in if_changed and friends.
 .PHONY: $(PHONY)

-SHRUNK_ARCH_DTB = $(patsubst %,$(obj)/dts/%.dtb,$(subst ",,$(CONFIG_SPL_OF_LIST)))
+SPL_OF_LIST_TARGETS = $(patsubst %,dts/%.dtb,$(subst ",,$(CONFIG_SPL_OF_LIST)))
+SHRUNK_ARCH_DTB = $(addprefix $(obj)/,$(SPL_OF_LIST_TARGETS))
 .SECONDEXPANSION:
 $(SHRUNK_ARCH_DTB): $$(patsubst $(obj)/dts/%, arch/$(ARCH)/dts/%, $$@) dts_dir
 	$(call if_changed,fdtgrep)

+targets += $(SPL_OF_LIST_TARGETS)
+
 MKIMAGEFLAGS_$(SPL_BIN).multidtb.fit = -f auto -A $(ARCH) -T firmware -C none -O u-boot \
 	-n "Multi DTB fit image for $(SPL_BIN)" -E \
 	$(patsubst %,-b %,$(SHRUNK_ARCH_DTB))
--
2.26.1

[PATCH 4/4] kbuild: arm: Fix duplicate builds of dtbs

[Jan Kiszka]

From: Jan Kiszka <jan.kiszka@siemens.com>

Build the secured board dtbs (.dtb_HS) as part of the regular dtb build
on CONFIG_TI_SECURE_DEVICE targets. This avoids rebuilding them,
possibly overwriting artifacts that are in use, as it is done so far.

In the same run, fix needless rebuilding of the secured spl dtb.

Fixes: 508369672ca3 ("arm: mach-k3: Add secure device build support")
CC: Andrew F. Davis <afd@ti.com>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 arch/arm/dts/Makefile             |  6 ++++--
 arch/arm/mach-k3/config_secure.mk | 19 ++++++++++++++-----
 2 files changed, 18 insertions(+), 7 deletions(-)

diff --git a/arch/arm/dts/Makefile b/arch/arm/dts/Makefile
index 2c123bd6da..b68e9c0726 100644
--- a/arch/arm/dts/Makefile
+++ b/arch/arm/dts/Makefile
@@ -1,5 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0+

+include $(srctree)/arch/arm/mach-k3/config_secure.mk
+
 dtb-$(CONFIG_TARGET_SMARTWEB) += at91sam9260-smartweb.dtb
 dtb-$(CONFIG_TARGET_TAURUS) += at91sam9g20-taurus.dtb
 dtb-$(CONFIG_TARGET_CORVUS) += at91sam9g45-corvus.dtb
@@ -927,13 +929,13 @@ dtb-$(CONFIG_TARGET_DURIAN) += phytium-durian.dtb

 dtb-$(CONFIG_TARGET_PRESIDIO_ASIC) += ca-presidio-engboard.dtb

-targets += $(dtb-y)
+targets += $(dtb-y) $(TI_SECURE_DTBS)

 # Add any required device tree compiler flags here
 DTC_FLAGS +=

 PHONY += dtbs
-dtbs: $(addprefix $(obj)/, $(dtb-y))
+dtbs: $(addprefix $(obj)/, $(dtb-y) $(TI_SECURE_DTBS))
 	@:

 clean-files := *.dtb *.dtbo *_HS
diff --git a/arch/arm/mach-k3/config_secure.mk b/arch/arm/mach-k3/config_secure.mk
index 6d63c57665..d9141e10a0 100644
--- a/arch/arm/mach-k3/config_secure.mk
+++ b/arch/arm/mach-k3/config_secure.mk
@@ -26,7 +26,12 @@ endif
 $(obj)/u-boot-spl-nodtb.bin_HS: $(obj)/u-boot-spl-nodtb.bin FORCE
 	$(call if_changed,k3secureimg)

-tispl.bin_HS: $(obj)/u-boot-spl-nodtb.bin_HS $(patsubst %,$(obj)/dts/%.dtb_HS,$(subst ",,$(CONFIG_SPL_OF_LIST))) $(SPL_ITS) FORCE
+SPL_OF_LIST_TARGETS = $(patsubst %,dts/%.dtb,$(subst ",,$(CONFIG_SPL_OF_LIST)))
+SPL_OF_LIST_TARGETS_HS = $(addsuffix _HS,$(SPL_OF_LIST_TARGETS))
+
+targets += $(SPL_OF_LIST_TARGETS) $(SPL_OF_LIST_TARGETS_HS)
+
+tispl.bin_HS: $(obj)/u-boot-spl-nodtb.bin_HS $(addprefix $(obj)/,$(SPL_OF_LIST_TARGETS_HS)) $(SPL_ITS) FORCE
 	$(call if_changed,mkfitimage)

 MKIMAGEFLAGS_u-boot.img_HS = -f auto -A $(ARCH) -T firmware -C none -O u-boot \
@@ -34,11 +39,15 @@ MKIMAGEFLAGS_u-boot.img_HS = -f auto -A $(ARCH) -T firmware -C none -O u-boot \
 	-n "U-Boot $(UBOOTRELEASE) for $(BOARD) board" -E \
 	$(patsubst %,-b arch/$(ARCH)/dts/%.dtb_HS,$(subst ",,$(CONFIG_OF_LIST)))

-OF_LIST_TARGETS = $(patsubst %,arch/$(ARCH)/dts/%.dtb,$(subst ",,$(CONFIG_OF_LIST)))
-$(OF_LIST_TARGETS): dtbs
-
 u-boot-nodtb.bin_HS: u-boot-nodtb.bin FORCE
 	$(call if_changed,k3secureimg)

-u-boot.img_HS: u-boot-nodtb.bin_HS u-boot.img $(patsubst %.dtb,%.dtb_HS,$(OF_LIST_TARGETS)) FORCE
+u-boot.img_HS: u-boot-nodtb.bin_HS u-boot.img dtbs FORCE
 	$(call if_changed,mkimage)
+
+# Used when included by arch-dts makefile
+-include include/config/auto.conf
+
+ifeq ($(CONFIG_TI_SECURE_DEVICE),y)
+TI_SECURE_DTBS = $(addsuffix _HS, $(dtb-y))
+endif
--
2.26.1

[PATCH 1/4] kbuild: efi: Avoid rebuilding efi targets

[Heinrich Schuchardt]

On 5/4/20 2:38 PM, Jan Kiszka wrote:
> From: Jan Kiszka <jan.kiszka@siemens.com>
>
> Add a couple of missing targets so that helloworld and other efi targets
> are not needlessly rebuilt.
>
> CC: Heinrich Schuchardt <xypron.glpk@gmx.de>
> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Tested-by: Heinrich Schuchardt <xypron.glpk@gmx.de>

> ---
>  lib/efi_loader/Makefile | 1 +
>  scripts/Makefile.lib    | 2 ++
>  2 files changed, 3 insertions(+)
>
> diff --git a/lib/efi_loader/Makefile b/lib/efi_loader/Makefile
> index eff3c25ec3..84d61df55b 100644
> --- a/lib/efi_loader/Makefile
> +++ b/lib/efi_loader/Makefile
> @@ -17,6 +17,7 @@ CFLAGS_REMOVE_helloworld.o := $(CFLAGS_NON_EFI)
>
>  ifneq ($(CONFIG_CMD_BOOTEFI_HELLO_COMPILE),)
>  always += helloworld.efi
> +targets += helloworld.o
>  endif
>
>  obj-$(CONFIG_CMD_BOOTEFI_HELLO) += helloworld_efi.o
> diff --git a/scripts/Makefile.lib b/scripts/Makefile.lib
> index 63fbadd757..734001c952 100644
> --- a/scripts/Makefile.lib
> +++ b/scripts/Makefile.lib
> @@ -418,6 +418,8 @@ $(obj)/efi_reloc.o: $(srctree)/arch/$(ARCH)/lib/$(EFI_RELOC:.o=.c) $(recordmcoun
>  $(obj)/%_efi.so: $(obj)/%.o $(obj)/efi_crt0.o $(obj)/efi_reloc.o $(obj)/efi_freestanding.o
>  	$(call cmd,efi_ld)
>
> +targets += $(obj)/efi_crt0.o $(obj)/efi_reloc.o $(obj)/efi_freestanding.o
> +
>  # ACPI
>  # ---------------------------------------------------------------------------
>  #
> --
> 2.26.1
>

[PATCH 4/4] kbuild: arm: Fix duplicate builds of dtbs

[Andrew F. Davis]

On 5/4/20 8:38 AM, Jan Kiszka wrote:
> From: Jan Kiszka <jan.kiszka@siemens.com>
> 
> Build the secured board dtbs (.dtb_HS) as part of the regular dtb build
> on CONFIG_TI_SECURE_DEVICE targets. This avoids rebuilding them,
> possibly overwriting artifacts that are in use, as it is done so far.
> 
> In the same run, fix needless rebuilding of the secured spl dtb.
> 
> Fixes: 508369672ca3 ("arm: mach-k3: Add secure device build support")
> CC: Andrew F. Davis <afd@ti.com>
> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
> ---
>  arch/arm/dts/Makefile             |  6 ++++--
>  arch/arm/mach-k3/config_secure.mk | 19 ++++++++++++++-----
>  2 files changed, 18 insertions(+), 7 deletions(-)
> 
> diff --git a/arch/arm/dts/Makefile b/arch/arm/dts/Makefile
> index 2c123bd6da..b68e9c0726 100644
> --- a/arch/arm/dts/Makefile
> +++ b/arch/arm/dts/Makefile
> @@ -1,5 +1,7 @@
>  # SPDX-License-Identifier: GPL-2.0+
> 
> +include $(srctree)/arch/arm/mach-k3/config_secure.mk
> +


This is hacky, we should not in the top level dts makefile have a
dependency on a specific platform configuration file.

What about mach-omap/config_secure.mk, etc..

The negative of building dtbs twice for HS is very minor compared to the
extra complexity this patch adds. It will be much more difficult to
clean this up later.

Andrew


>  dtb-$(CONFIG_TARGET_SMARTWEB) += at91sam9260-smartweb.dtb
>  dtb-$(CONFIG_TARGET_TAURUS) += at91sam9g20-taurus.dtb
>  dtb-$(CONFIG_TARGET_CORVUS) += at91sam9g45-corvus.dtb
> @@ -927,13 +929,13 @@ dtb-$(CONFIG_TARGET_DURIAN) += phytium-durian.dtb
> 
>  dtb-$(CONFIG_TARGET_PRESIDIO_ASIC) += ca-presidio-engboard.dtb
> 
> -targets += $(dtb-y)
> +targets += $(dtb-y) $(TI_SECURE_DTBS)
> 
>  # Add any required device tree compiler flags here
>  DTC_FLAGS +=
> 
>  PHONY += dtbs
> -dtbs: $(addprefix $(obj)/, $(dtb-y))
> +dtbs: $(addprefix $(obj)/, $(dtb-y) $(TI_SECURE_DTBS))
>  	@:
> 
>  clean-files := *.dtb *.dtbo *_HS
> diff --git a/arch/arm/mach-k3/config_secure.mk b/arch/arm/mach-k3/config_secure.mk
> index 6d63c57665..d9141e10a0 100644
> --- a/arch/arm/mach-k3/config_secure.mk
> +++ b/arch/arm/mach-k3/config_secure.mk
> @@ -26,7 +26,12 @@ endif
>  $(obj)/u-boot-spl-nodtb.bin_HS: $(obj)/u-boot-spl-nodtb.bin FORCE
>  	$(call if_changed,k3secureimg)
> 
> -tispl.bin_HS: $(obj)/u-boot-spl-nodtb.bin_HS $(patsubst %,$(obj)/dts/%.dtb_HS,$(subst ",,$(CONFIG_SPL_OF_LIST))) $(SPL_ITS) FORCE
> +SPL_OF_LIST_TARGETS = $(patsubst %,dts/%.dtb,$(subst ",,$(CONFIG_SPL_OF_LIST)))
> +SPL_OF_LIST_TARGETS_HS = $(addsuffix _HS,$(SPL_OF_LIST_TARGETS))
> +
> +targets += $(SPL_OF_LIST_TARGETS) $(SPL_OF_LIST_TARGETS_HS)
> +
> +tispl.bin_HS: $(obj)/u-boot-spl-nodtb.bin_HS $(addprefix $(obj)/,$(SPL_OF_LIST_TARGETS_HS)) $(SPL_ITS) FORCE
>  	$(call if_changed,mkfitimage)
> 
>  MKIMAGEFLAGS_u-boot.img_HS = -f auto -A $(ARCH) -T firmware -C none -O u-boot \
> @@ -34,11 +39,15 @@ MKIMAGEFLAGS_u-boot.img_HS = -f auto -A $(ARCH) -T firmware -C none -O u-boot \
>  	-n "U-Boot $(UBOOTRELEASE) for $(BOARD) board" -E \
>  	$(patsubst %,-b arch/$(ARCH)/dts/%.dtb_HS,$(subst ",,$(CONFIG_OF_LIST)))
> 
> -OF_LIST_TARGETS = $(patsubst %,arch/$(ARCH)/dts/%.dtb,$(subst ",,$(CONFIG_OF_LIST)))
> -$(OF_LIST_TARGETS): dtbs
> -
>  u-boot-nodtb.bin_HS: u-boot-nodtb.bin FORCE
>  	$(call if_changed,k3secureimg)
> 
> -u-boot.img_HS: u-boot-nodtb.bin_HS u-boot.img $(patsubst %.dtb,%.dtb_HS,$(OF_LIST_TARGETS)) FORCE
> +u-boot.img_HS: u-boot-nodtb.bin_HS u-boot.img dtbs FORCE
>  	$(call if_changed,mkimage)
> +
> +# Used when included by arch-dts makefile
> +-include include/config/auto.conf
> +
> +ifeq ($(CONFIG_TI_SECURE_DEVICE),y)
> +TI_SECURE_DTBS = $(addsuffix _HS, $(dtb-y))
> +endif
> --
> 2.26.1
> 

[PATCH 4/4] kbuild: arm: Fix duplicate builds of dtbs

[Jan Kiszka]

On 08.05.20 17:40, Andrew F. Davis wrote:
> On 5/4/20 8:38 AM, Jan Kiszka wrote:
>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>
>> Build the secured board dtbs (.dtb_HS) as part of the regular dtb build
>> on CONFIG_TI_SECURE_DEVICE targets. This avoids rebuilding them,
>> possibly overwriting artifacts that are in use, as it is done so far.
>>
>> In the same run, fix needless rebuilding of the secured spl dtb.
>>
>> Fixes: 508369672ca3 ("arm: mach-k3: Add secure device build support")
>> CC: Andrew F. Davis <afd@ti.com>
>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>> ---
>>   arch/arm/dts/Makefile             |  6 ++++--
>>   arch/arm/mach-k3/config_secure.mk | 19 ++++++++++++++-----
>>   2 files changed, 18 insertions(+), 7 deletions(-)
>>
>> diff --git a/arch/arm/dts/Makefile b/arch/arm/dts/Makefile
>> index 2c123bd6da..b68e9c0726 100644
>> --- a/arch/arm/dts/Makefile
>> +++ b/arch/arm/dts/Makefile
>> @@ -1,5 +1,7 @@
>>   # SPDX-License-Identifier: GPL-2.0+
>>
>> +include $(srctree)/arch/arm/mach-k3/config_secure.mk
>> +
>
>
> This is hacky, we should not in the top level dts makefile have a
> dependency on a specific platform configuration file.
>
> What about mach-omap/config_secure.mk, etc..

They don't need those special signing the k3 need. If they did, they
could be included here as well.

>
> The negative of building dtbs twice for HS is very minor compared to the
> extra complexity this patch adds. It will be much more difficult to
> clean this up later.

Overwriting a build artifacts that were already completed and may just
be in use while doing so is broken. This must be fixed.

As I wrote, I will add support for injecting public keys into the dtbs,
and that also revealed this issue.

Better suggestions welcome.

Jan

[PATCH 2/4] kbuild: spl: Fix parallel build

[Tom Rini]

On Mon, May 04, 2020 at 02:38:30PM +0200, Jan Kiszka wrote:

> From: Jan Kiszka <jan.kiszka@siemens.com>
> 
> The dts dir must exists when running this rule.
> 
> That missing dependency broke e.g. "make -j" for the am65x targets.
> 
> Fixes: 2f57c95100f2 ("spl: dm: Make it possible for the SPL to pick its own DTB from a FIT")
> CC: Jean-Jacques Hiblot <jjhiblot@ti.com>
> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Applied to u-boot/master, thanks!

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200514/00504ac6/attachment.sig>

[PATCH 3/4] kbuild: spl: Add shrunk arch-dtbs to targets list

[Tom Rini]

On Mon, May 04, 2020 at 02:38:31PM +0200, Jan Kiszka wrote:

> From: Jan Kiszka <jan.kiszka@siemens.com>
> 
> This avoids needless rebuilding.
> 
> Fixes: 2f57c95100f2 ("spl: dm: Make it possible for the SPL to pick its own DTB from a FIT")
> CC: Jean-Jacques Hiblot <jjhiblot@ti.com>
> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

Applied to u-boot/master, thanks!

[Aside, I'm assuming Andrew will reply back still on 4/4, but 2/4 and
3/4 are needed to fix building / testing right now]

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200514/8c3cfa71/attachment.sig>

[PATCH 4/4] kbuild: arm: Fix duplicate builds of dtbs

[Andrew F. Davis]

On 5/8/20 11:54 AM, Jan Kiszka wrote:
> On 08.05.20 17:40, Andrew F. Davis wrote:
>> On 5/4/20 8:38 AM, Jan Kiszka wrote:
>>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>>
>>> Build the secured board dtbs (.dtb_HS) as part of the regular dtb build
>>> on CONFIG_TI_SECURE_DEVICE targets. This avoids rebuilding them,
>>> possibly overwriting artifacts that are in use, as it is done so far.
>>>
>>> In the same run, fix needless rebuilding of the secured spl dtb.
>>>
>>> Fixes: 508369672ca3 ("arm: mach-k3: Add secure device build support")
>>> CC: Andrew F. Davis <afd@ti.com>
>>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>>> ---
>>> ? arch/arm/dts/Makefile???????????? |? 6 ++++--
>>> ? arch/arm/mach-k3/config_secure.mk | 19 ++++++++++++++-----
>>> ? 2 files changed, 18 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/arch/arm/dts/Makefile b/arch/arm/dts/Makefile
>>> index 2c123bd6da..b68e9c0726 100644
>>> --- a/arch/arm/dts/Makefile
>>> +++ b/arch/arm/dts/Makefile
>>> @@ -1,5 +1,7 @@
>>> ? # SPDX-License-Identifier: GPL-2.0+
>>>
>>> +include $(srctree)/arch/arm/mach-k3/config_secure.mk
>>> +
>>
>>
>> This is hacky, we should not in the top level dts makefile have a
>> dependency on a specific platform configuration file.
>>
>> What about mach-omap/config_secure.mk, etc..
> 
> They don't need those special signing the k3 need. If they did, they
> could be included here as well.
> 


I did everything the same in that file, so I don't see why it wouldn't
need the same treatment. But my point is that this will keep growing,
the fixes should be local to the mk files or everyone who does something
similar will need to add here.


>>
>> The negative of building dtbs twice for HS is very minor compared to the
>> extra complexity this patch adds. It will be much more difficult to
>> clean this up later.
> 
> Overwriting a build artifacts that were already completed and may just
> be in use while doing so is broken. This must be fixed.
> 


Absolutely agree it needs fixed, just not sure this just doesn't add to
the makefile complexity in the wrong way.


> As I wrote, I will add support for injecting public keys into the dtbs,
> and that also revealed this issue.
> 


If I understand right, you are adding the keys into the dtb then signing
them? Why not add them to the dts first, then nothing would need to
change with multiple builds of the dtbs.

Andrew


> Better suggestions welcome.
> 
> Jan

[PATCH 4/4] kbuild: arm: Fix duplicate builds of dtbs

[Jan Kiszka]

On 18.05.20 18:26, Andrew F. Davis wrote:
> On 5/8/20 11:54 AM, Jan Kiszka wrote:
>> On 08.05.20 17:40, Andrew F. Davis wrote:
>>> On 5/4/20 8:38 AM, Jan Kiszka wrote:
>>>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>>>
>>>> Build the secured board dtbs (.dtb_HS) as part of the regular dtb build
>>>> on CONFIG_TI_SECURE_DEVICE targets. This avoids rebuilding them,
>>>> possibly overwriting artifacts that are in use, as it is done so far.
>>>>
>>>> In the same run, fix needless rebuilding of the secured spl dtb.
>>>>
>>>> Fixes: 508369672ca3 ("arm: mach-k3: Add secure device build support")
>>>> CC: Andrew F. Davis <afd@ti.com>
>>>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>>>> ---
>>>>  ? arch/arm/dts/Makefile???????????? |? 6 ++++--
>>>>  ? arch/arm/mach-k3/config_secure.mk | 19 ++++++++++++++-----
>>>>  ? 2 files changed, 18 insertions(+), 7 deletions(-)
>>>>
>>>> diff --git a/arch/arm/dts/Makefile b/arch/arm/dts/Makefile
>>>> index 2c123bd6da..b68e9c0726 100644
>>>> --- a/arch/arm/dts/Makefile
>>>> +++ b/arch/arm/dts/Makefile
>>>> @@ -1,5 +1,7 @@
>>>>  ? # SPDX-License-Identifier: GPL-2.0+
>>>>
>>>> +include $(srctree)/arch/arm/mach-k3/config_secure.mk
>>>> +
>>>
>>>
>>> This is hacky, we should not in the top level dts makefile have a
>>> dependency on a specific platform configuration file.
>>>
>>> What about mach-omap/config_secure.mk, etc..
>>
>> They don't need those special signing the k3 need. If they did, they
>> could be included here as well.
>>
>
>
> I did everything the same in that file, so I don't see why it wouldn't
> need the same treatment. But my point is that this will keep growing,
> the fixes should be local to the mk files or everyone who does something
> similar will need to add here.
>

The problem is that dtbs are built in a separate make session. If you
want to address specific dtbs in a rule, you must do that in that
session, not outside like you did. That applies to signing for TI K3,
just like it would for bringing in public keys as discussed below.

I think you mentioned offlist that eliminating this nesting would help,
but I do not yet see if that can be easily done.

>
>>>
>>> The negative of building dtbs twice for HS is very minor compared to the
>>> extra complexity this patch adds. It will be much more difficult to
>>> clean this up later.
>>
>> Overwriting a build artifacts that were already completed and may just
>> be in use while doing so is broken. This must be fixed.
>>
>
>
> Absolutely agree it needs fixed, just not sure this just doesn't add to
> the makefile complexity in the wrong way.
>

As I said, I'm open for other approaches that resolve this.

>
>> As I wrote, I will add support for injecting public keys into the dtbs,
>> and that also revealed this issue.
>>
>
>
> If I understand right, you are adding the keys into the dtb then signing
> them? Why not add them to the dts first, then nothing would need to
> change with multiple builds of the dtbs.

You sign the chain from right to left, while loading is left to right:
the right-side image is signed while injecting the necessary keys/certs
into the left-side image that loads it. And then that is signed as well,
in the end often with a hardware-specific mechanism (unfortunately).

But dtbs are hardware-specific, keys are deployment-specific. That's why
they are injected into existing dtbs, rather than asking the user to
patch them into dts files in the source tree. Just today I ran into
another piece of code that models this signing and key injection outside
of U-Boot, with a lot of knowledge about the particular target. It would
be way nicer to solve that generically.

Jan