Re: [ptxdist] [PATCH v3] tf-a: new package for ARM trusted firmware A

[Philipp Zabel <p.zabel@pengutronix.de>]

Hi,

On Wed, 2020-02-19 at 11:42 +0100, Ahmad Fatoum wrote:
> Trusted Firmware-A (TF-A) is a reference implementation of secure world
> software for Arm A-Profile architectures (Armv8-A and Armv7-A).
> 
> Cc: Alejandro Vazquez <avazquez.dev@gmail.com>
> Signed-off-by: Rouven Czerwinski <rouven@czerwinskis.de>
> Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
> ---
> v2 -> v3:
>   - sorted version and md5 Kconfig options to top (mol)
>   - Added trailing spaces to align arrow after main prompt (mol)
>   - Changed TF_A_VERSION help text to indicate that the version can be any
>     git commitsh
>   - Removed default value for TF_A_ARTIFACTS (mol)
>   - integrated TF_A_VERSION into BUILD_STRING of resulting artifacts
>     (Guillermo, mtr)
>   - unconditional error if TF_A_ARTIFACTS is empty instead of warning later on
>     (mol)
>   - removed diplicate Prepare header (mol)
>   - resorted make variable definitions
>   - avoid unportable install --target-directory (Guillermo)
>   - documented support for globbing in TF_A_ARTIFACTS
>   - don't glob delete in IMAGEDIR, instead expand the glob in the build
>     directory first
> v1 -> v2:
>   - Made TF_A_ARCH_MAJOR configurable to support 32 bit ARMv8 (Guillermo)
>   - Replaces stm32mp-specific TF_A_DTB with TF_A_EXTRA_ARGS to contain
>     all board/vendor specific options
>   - removed reference to no longer existing CREDITS file
>   - removed TF_A_MAKE_OPT contents that are set elsewhere
>   - reduced uses of += in favor of directly appending to the string
>   - delete old build directory in prepare instead of compile
>   - use default compile stage (Guillermo)
>   - install artifacts to sysroot /usr/lib/firmware in install stage
>   - install artifacts to IMAGEDIR in targetinstall
>   - fix clean stage to delete proper artifacts
> ---
>  platforms/tf-a.in | 140 ++++++++++++++++++++++++++++++++++++++++++++++
>  rules/tf-a.make   | 118 ++++++++++++++++++++++++++++++++++++++
>  2 files changed, 258 insertions(+)
>  create mode 100644 platforms/tf-a.in
>  create mode 100644 rules/tf-a.make
> 
> diff --git a/platforms/tf-a.in b/platforms/tf-a.in
> new file mode 100644
> index 000000000000..75dad9d159f4
> --- /dev/null
> +++ b/platforms/tf-a.in
> @@ -0,0 +1,140 @@
[...]
> +config TF_A_VERSION
> +	string
> +	default "v2.2"
> +	prompt "TF-A version"
> +	help
> +	  Enter the TF-A git commitsh you want to build. Usally a tagged release
                                   ^                       ^
Some missing vowels here. Also, the git documentation uses spells it
"commit-ish" [1], but in most places it just says "commit".

[1] https://git-scm.com/docs/gitglossary

[...]
> diff --git a/rules/tf-a.make b/rules/tf-a.make
> new file mode 100644
> index 000000000000..bb7ba9e96822
> --- /dev/null
> +++ b/rules/tf-a.make
> @@ -0,0 +1,118 @@
[...]
> +#
> +# Paths and names
> +#
> +TF_A_VERSION	:= $(call remove_quotes,$(PTXCONF_TF_A_VERSION))
> +TF_A_MD5	:= $(call remove_quotes,$(PTXCONF_TF_A_MD5))
> +TF_A		:= tf-a-$(TF_A_VERSION)
> +TF_A_SUFFIX	:= tar.gz
> +TF_A_URL	:= https://git.trustedfirmware.org/TF-A/trusted-firmware-a.git/snapshot/$(TF_A_VERSION).$(TF_A_SUFFIX)
> +TF_A_SOURCE	:= $(SRCDIR)/$(TF_A).$(TF_A_SUFFIX)
> +TF_A_DIR	:= $(BUILDDIR)/$(TF_A)
> +TF_A_LICENSE    := BSD-3-Clause

What about the other licenses listed in docs/licenses.rst, such as NCSA,
and Zlib? Does any of this code make it into the TF-A binary?

> +
> +# ----------------------------------------------------------------------------
> +# Prepare
> +# ----------------------------------------------------------------------------
> +
> +TF_A_WRAPPER_BLACKLIST := \
> +	TARGET_HARDEN_RELRO \
> +	TARGET_HARDEN_BINDNOW \
> +	TARGET_HARDEN_PIE \
> +	TARGET_DEBUG \
> +	TARGET_BUILD_ID
> +
> +# TF_A_DEBUG=1

Is this a leftover? I see no other packages with a commented out
FOO_DEBUG=1 line.

regards
Philipp

_______________________________________________
ptxdist mailing list
ptxdist@pengutronix.de