A new (and hopefully definitive) CPE is to be used for TF-A:
cpe:2.3:o:trustedfirmware:trusted_firmware-a:
as reported in TF-A directly[1]. NVD correctly reports[2] the new CPE
deprecates the "old" one Buildroot was using.
CVEs for projects under the TrustedFirmware umbrella are now seemingly
handled under the CPE vendor "trustedfirmware".
[1] efd7b189fa%5E%21/
[2] https://nvd.nist.gov/products/cpe/detail/2E1BD3E8-DF65-42E3-A0BA-747137D6DEF2
Signed-off-by: Quentin Schulz <quentin.schulz@cherry.de>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Commit 53bb7eafb2 ("package/arm-trusted-firmware: add
ARM_TRUSTED_FIRMWARE_CPE_ID_*") added logic to set the cpe-id version from
the package version by dropping a v or lts- prefix, but it used subst
instead of patsubst - So it would replace "v" and "lts-" anywhere in the
version and not only in the beginning of it.
Change to patsubst to fix that.
E.G. with the following defconfig:
BR2_aarch64=y
BR2_TARGET_ARM_TRUSTED_FIRMWARE=y
BR2_TARGET_ARM_TRUSTED_FIRMWARE_CUSTOM_VERSION=y
BR2_TARGET_ARM_TRUSTED_FIRMWARE_CUSTOM_VERSION_VALUE="version-with-v-or-lts-in-it"
before:
make printvars VARS=ARM_TRUSTED_FIRMWARE_CPE_ID_VERSION
ARM_TRUSTED_FIRMWARE_CPE_ID_VERSION=ersion-with--or-in-it
after:
make printvars VARS=ARM_TRUSTED_FIRMWARE_CPE_ID_VERSION
ARM_TRUSTED_FIRMWARE_CPE_ID_VERSION=ersion-with-v-or-lts-in-it
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
Arm Trusted Firmware (TF-A) can be used to load U-Boot or another
bootloader, which in turn loads the Linux kernel. However, TF-A is
capable of loading the kernel directly. To this end, we need to define
the BL33 and BL33_CFG compile options containing, respectively, the
zImage and the DTB.
This config introduces a new config option,
BR2_TARGET_ARM_TRUSTED_FIRMWARE_LINUX_AS_BL33, which sets the BL33 and
BL33_CFG parameters, and ensures that the kernel is built before the
TF-A by having linux as a _DEPENDENCY of the TF-A.
Signed-off-by: Jakob Kastelic <kastelic.jakob@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
The ATF package was not providing any license file when a custom
Linux version was selected.
Fix this by adding a Kconfig option to specify the license file,
with a default value set to the commonly used license paths.
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
[Arnout: use a single Kconfig option with conditional prompt]
Signed-off-by: Arnout Vandecappelle <arnout@rnout.be>
When building the arm-trusted-firmware, if the host environment has a value
configured in the BL31 variable such as the following:
export BL31=/tmp/bl31.elf
This will cause the build of the bl31.elf to be skipped leading to the
following build error:
make[1]: Nothing to be done for 'bl31'.
And then:
readelf: Error: './output/build/arm-trusted-firmware-custom/build/versal/release/bl31/bl31.elf': No such file
To fix this, clear the BL31 variable in the MAKE_OPTS, so that building the
arm-trusted-firmware will build regardless of the host environment.
Signed-off-by: Neal Frager <neal.frager@amd.com>
Tested-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Reviewed-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
When Building arm-trusted-firmware for the Macchiatobin platform
(a80x0_mcbin), which depends on the mv-ddr-marvell package, the build fails
complaining that this package's folder "does not contain valid
mv-ddr-marvell git repository".
This is expected under Buildroot, where we use intermediate archives.
The issue can be reproduced with the commands:
cat >.config <<EOF
BR2_aarch64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TARGET_ARM_TRUSTED_FIRMWARE=y
BR2_TARGET_ARM_TRUSTED_FIRMWARE_PLATFORM="a80x0_mcbin"
BR2_TARGET_ARM_TRUSTED_FIRMWARE_EDK2_AS_BL33=y
BR2_TARGET_BINARIES_MARVELL=y
BR2_TARGET_EDK2=y
BR2_TARGET_EDK2_PLATFORM_SOLIDRUN_ARMADA80X0MCBIN=y
BR2_TARGET_MV_DDR_MARVELL=y
EOF
make olddefconfig
make
The build is failing with the error message:
plat/marvell/armada/a8k/common/ble/ble.mk:34: *** "'MV_DDR_PATH=/buildroot/output/build/mv-ddr-marvell-d5acc10c287e40cc2feeb28710b92e45c93c702c' was specified, but '/buildroot/output/build/mv-ddr-marvell-d5acc10c287e40cc2feeb28710b92e45c93c702c' does not contain valid mv-ddr-marvell git repository". Stop.
Add patches to fix the build for this platform, for a few versions of TF-A
(v2.6, v2.7, v2.8, lts-v2.8.20, v2.9, v2.10, lts-v2.10.5, v2.11, v2.12 and
lts-v2.12.1).
Signed-off-by: Vincent Stehlé <vincent.stehle@arm.com>
Cc: Dick Olsson <hi@senzilla.io>
Cc: Sergey Matyukevich <geomatsi@gmail.com>
[Julien: add commands to reproduce the issue]
Signed-off-by: Julien Olivain <ju.o@free.fr>
Change all Kconfig entries for packages in the "Bootloaders" menu to
submenus.
Currently many bootloaders and firmwares in this menu use a 'config'
keyword and thus menuconfig presents their sub-options as a flat list in
the same menu. This is not very convenient as some have a lot of
sub-options, and in (more and more common) configurations with multiple
such packages enabled the menuconfig screen becomes very crowded. Moreover
options of 'string' type don't have a visible indentation when their value
exceeds a few characters, thus making it visually hard to grasp which
package the various options belong to.
Non-bootloader packages having suboptions usually use a 'menuconfig'
keywork to have sub-options in a submenu. Do the same for packages in the
"Bootloaders" menu which have more than one sub-option.
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Reviewed-by: Neal Frager <neal.frager@amd.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Allow to enable BR2_TARGET_ARM_TRUSTED_FIRMWARE_NEEDS_ARM32_TOOLCHAIN
option on aarch64 hosts using HOST_ARM_GNU_TOOLCHAIN_SUPPORTS, in order
to build an ATF firmwares that require a pre-built bare metal toolchain
(rockpro64_defconfig).
[Romain: improve commit log]
Signed-off-by: Bryce Johnson <bryce@redpinelabs.com>
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Arm Trusted Firmware (TF-A) v2.11 fails to build in some situations
with the error message:
ld: build/qemu/release/bl1/bl1.elf: error: PHDR segment not covered by LOAD segment
This error can be reproduced with the commands:
make qemu_aarch64_ebbr_defconfig
utils/config --set-str BR2_TARGET_ARM_TRUSTED_FIRMWARE_CUSTOM_VERSION_VALUE v2.11
make olddefconfig
make
This issue was reported in [1].
This error is coming from a check that was made more strict since
binutils ld 2.34. This error message is normally related to dynamic
linker, so it should normally not apply to a package like TF-A.
When BR2_SHARED_LIBS=y (shared libraries only) and BR2_PIC_PIE=y
(Build code with PIC/PIE), the Buildroot toolchain-wrapper will try
to enable position-independent code/executables. See [2]. This
configuration is a common default.
Arm Trusted Firmware (TF-A) build system, on its side, tries to detect
if the toolchain enables PIE automatically. It does so by checking if
--enable-default-pie is in the output of "$(CC) -v". If found, the TF-A
build system tries to disable PIE globally (it can be explicitly
enabled again with the ENABLE_PIE=1 build variable, in some specific
configurations). This detection mechanism is not working with the
Buildroot toolchain wrapper which is enabling PIE silently. See [3].
Commit 1061ed6c "boot/arm-trusted-firmware: add -fno-PIE to CFLAGS"
added the option -fno-PIE in CFLAGS for that reason. See [4].
TF-A >= v2.11 now needs the same treatment for LDFLAGS. This is
because TF-A switched the default linker from "ld" to "gcc", in
upstream commit [5]. This change makes the Buildroot toolchain wrapper
to enable PIE at link, without passing the "--no-dynamic-linker" option
that would normally avoids this ld error.
Also, even if there is no defconfigs using the
BR2_TARGET_ARM_TRUSTED_FIRMWARE_LATEST_VERSION config directive, it is
worth mentioning that the Buildroot TF-A "latest version" was updated
to v2.11 in commit 9c50759cd "boot/arm-trusted-firmware: bump to
v2.11". See [6]. This latest version is the default choice. So
Buildroot is subject to generate this build failure. This can be
reproduced with the commands:
cat <<EOF >.config
BR2_aarch64=y
BR2_TARGET_ARM_TRUSTED_FIRMWARE=y
BR2_TARGET_ARM_TRUSTED_FIRMWARE_PLATFORM="qemu"
BR2_TOOLCHAIN_EXTERNAL=y
EOF
make olddefconfig
make arm-trusted-firmware
This commit fixes the issue by adding the option "-no-pie" in LDFLAGS.
This will prevent the Buildroot toolchain wrapper to enable PIE, for
versions using "gcc" as a linker. This change should also remain
compatible with older TF-A < 2.11 using "ld" as a linker, since
"-no-pie" is also a valid ld option.
Fixes:
ld: build/qemu/release/bl1/bl1.elf: error: PHDR segment not covered by LOAD segment
[1] https://github.com/TrustedFirmware-A/trusted-firmware-a/issues/26
[2] https://gitlab.com/buildroot.org/buildroot/-/blob/2024.08-rc3/toolchain/toolchain-wrapper.c?ref_type=tags#L403
[3] https://git.trustedfirmware.org/plugins/gitiles/TF-A/trusted-firmware-a.git/+/refs/tags/v2.11.0/Makefile#711
[4] 1061ed6c62
[5] 2f1c5e7eb1%5E%21/
[6] 9c50759cd1
Reported-by: Vincent Stehlé <vincent.stehle@arm.com>
Tested-by: Vincent Stehlé <vincent.stehle@arm.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
This commit adds license information, and license file hashes for LTS
versions, by using per-version directories.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
There is a new lts-v2.10.x branch. Add it to the selection
Signed-off-by: Bryan Brattlof <bb@ti.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
bump the lts-v2.8.x branch to v2.8.20. Our patch still applies.
Signed-off-by: Bryan Brattlof <bb@ti.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
ARM_TRUSTED_FIRMWARE_SITE is currently pointing to the deprecated
ARM-software github read-only mirror[0] which is no longer publishing
new tags. They do have a newer github mirror under the TrustedFirmware-A
organization[1] which continues to receive tag updates we could use
however because of the way github generates tarballs changing the SITE
to point to TrustedFormware-A changes the pre-calculated hash values
for every version for everyone.
Without much way around changing all hash values if we want these latest
tags, lets drop the github mirrors and move to the official git
source[2] so we can download real tarballs which shouldn't change even
if the ARM_TRUSTED_FIRMWARE_SITE moves again and hopefully preventing
this from ever happening again.
[2] https://git.trustedfirmware.org/TF-A/trusted-firmware-a.git/
[1] https://github.com/TrustedFirmware-A/trusted-firmware-a
[0] https://github.com/ARM-software/arm-trusted-firmware
Signed-off-by: Bryan Brattlof <bb@ti.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Commit
c6ab9328e1 ("boot/arm-trusted-firmware:
change BR_NO_CHECK_HASH_FOR logic") was written too hastily, and is
actually broken as it does the reverse of what it should do. It really
should disable the check when we're using a custom version/git/tarball
of TF-A, not the opposite.
Thanks to Romain for pointing out the problem, and providing an IRL
patch to solve the issue.
Reported-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Patch 0001-feat-build-add-support-for-new-binutils-versions.patch as
found in v2.8 patch dir has already been applied in lts-v2.8.13
Patch 0002-build-tools-avoid-unnecessary-link.patch from v2.8 has
been included as 0001-build-tools-avoid-unnecessary-link.patch for
lts-v2.8.13
Signed-off-by: Marleen Vos <marleen.vos@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Currently, arm-trusted-firmware checks the hash only if
BR2_TARGET_ARM_TRUSTED_FIRMWARE_LATEST_VERSION=y. As we're about to
introduce other version options for which the hash checking is needed,
let's reverse the logic, and instead ignore hash checking if we're
using a custom version or custom git or custom tarball.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Patch 0001-build-tools-avoid-unnecessary-link.patch that was applied to
v2.9 is already in v2.10, so no new patch dir for v2.10 is added
Signed-off-by: Marleen Vos <marleen.vos@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/5834758777
Commit e7d16c35a (boot/arm-trusted-firmware: fix the RPATH of fiptool) tried
to fix the build of host-fiptool, but forgot to pass HOST_CFLAGS.
On hosts without (compatible) openssl development headers, this breaks
the build when it cannot find the openssl headers:
fiptool_platform.h:19:11: fatal error: openssl/sha.h: No such file or directory
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
The arm-trusted-firmware package builds a host tool called "fiptool",
which is used during the build process of arm-trusted-firmware
itself. This tool links against the OpenSSL host library, and
therefore needs to be built with the correct RPATH pointing to
$HOST_DIR/lib.
This is why commit a957d9a90a
("boot/arm-trusted-firmware: build fiptool separately with dependency
o n host-openssl") added the ARM_TRUSTED_FIRMWARE_BUILD_FIPTOOL
variable, which builds the fiptool tool first, with the right
variables set, before invoking the full build of TF-A. This ensured
that fiptool was built with the correct RPATH.
However, more recent versions of TF-A have modified their Makefile
machinery, and fiptool is being rebuilt even if it was built
before. Unfortunately, this rebuild is no longer done with the right
flags, so we end up with a fiptool binary that no longer has the right
RPATH, and fiptool fails to find the OpenSSL libraries from
$HOST_DIR/lib.
In order to fix this, we take a different approach: we do not build
fiptool separately first, but we inject the necessary flags through
the HOSTCC variable. Indeed, there's no HOST_LDFLAGS or HOST_LDLIBS
variable or similar that would allow us to pass the -Wl,-rpath flag
that is needed. Shoe-horning this flag into HOSTCC gets the job done,
and actually simplifies our arm-trusted-firmware.mk.
This patch break the compatibility with version prior to 1.4 (upstream
commit 72610c4102990 ("build: Introduce HOSTCC flag")). v1.4 is very old
(July 2017), not used anymore in-tree and probably not used anymore
outside the tree.
Signed-off-by: Louis Chauvet <louis.chauvet@bootlin.com>
Co-authored-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
TF-A prints a version string at boot which includes the version number as
specified in the Makefile and additional "build" information, specified by
the BUILD_STRING parameter:
https://trustedfirmware-a.readthedocs.io/en/v2.9/getting_started/build-options.html
BUILD_STRING: Input string for VERSION_STRING, which allows the TF-A build
to be uniquely identified. Defaults to the current git commit id.
This is implemented as:
# Default build string (git branch and commit)
ifeq (${BUILD_STRING},)
BUILD_STRING := $(shell git describe --always --dirty --tags 2> /dev/null)
endif
https://github.com/ARM-software/arm-trusted-firmware/blob/master/Makefile#L225-L228
Which is never correct in the context of Buildroot, as the TF-A build
directory is never a git repo, and git will move up the directory tree and
possibly and (possibly) pick up the git version of Buildroot instead.
To fix that, explicitly set BUILD_STRING to the version of the package.
Pass it before BR2_TARGET_ARM_TRUSTED_FIRMWARE_ADDITIONAL_VARIABLES so it
can still be overridden by the user if needed.
strings images/tf-a-* | egrep '^v2\.'
v2.9(debug):2023.08-859-g99d78b52a0
v2.9(debug):v2.9
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
We now have 3 different options to select an image as the BL33 stage
of TF-A: Barebox, U-Boot or EDK2. Technically speaking, they are
mutually exclusive: they all specify a BL33= variable to the TF-A
build, and TF-A can only support a single BL33 stage.
However, as pointed out by Vincent Fazio in [0] there is nothing that
prevents selecting Barebox, U-Boot and EDK2 together, even though it
doesn't make sense.
To address this, this commit introduces a choice...endchoice block,
into which the Barebox, U-Boot and EDK2 options are moved. An
additional "none" option is added, which is the default, and
corresponds to not having any BL33 image.
Since we keep the same name for the options, no legacy handling is
necessary.
[0] https://lore.kernel.org/buildroot/PH1P110MB1603A4AA1638838DA56BAA069FDA9@PH1P110MB1603.NAMP110.PROD.OUTLOOK.COM/
Reported-by: Vincent Fazio <vfazio@xes-inc.com>
Cc: Vincent Fazio <vfazio@xes-inc.com>
Cc: Casey Reeves <casey@xogium.me>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Tested-by: Vincent Fazio <vfazio@xes-inc.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
When building a fip firmware (BR2_TARGET_ARM_TRUSTED_FIRMWARE_FIP=y), the
TF-A build recipe starts by building the host program fiptool with the
proper build environment variables. Then the main TF-A target firmware
build step takes place, with the expectation that the fiptool program will
be used under the hood if necessary.
In TF-A, the build recipe for the host program fiptool has subtly changed
after v2.7, in commit cf2dd17ddda2 ("refactor(security): add OpenSSL 1.x
compatibility"). This change has the effect to force re-linking fiptool
each time.
If we try to build with Buildroot a fip firmware with a TF-A version after
v2.7 comprising the aforementioned change, the fiptool program is forcibly
re-linked during the main firmware build step. This happens without the
proper build environment variables and consequently, if openssl is not
installed on the host, the libcrypto shared library will not be found by
the linker and the link will fail with the following error:
/usr/bin/ld: cannot find -lcrypto: No such file or directory
A patch has been integrated into TF-A to avoid re-linking fiptool when not
necessary, which should solve the problem starting with version v2.10. Add
that patch in Buildroot for versions v2.8 and v2.9, to repair the build in
the cases described above.
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/4664845767
Signed-off-by: Vincent Stehlé <vincent.stehle@arm.com>
Cc: Dick Olsson <hi@senzilla.io>
Cc: Sergey Matyukevich <geomatsi@gmail.com>
[yann.morin.1998@free.fr:
- don't use symlink in v2.9, just copy patch
- fix numbering in v2.8 which now has two patches
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Since version 2022.04.0, Barebox now supports a generic image built
for platforms, to be used as BL33 to generate FIP images. This patch
makes it available as a choice of BL33 to be used with ATF and is
based on the similar work put into U-Boot as BL33.
Signed-off-by: Casey Reeves <casey@xogium.me>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
ATF version 2.2 and older does not disable SSP when
ENABLE_STACK_PROTECTOR is not set. This is because the compiler enables
SSP by default, and ATF does not pass -fno-stack-protector to the
compiler. Upstream commit 7af195e29a42 ("Disable stack protection
explicitly") fixed the issue for v2.3 and newer.
Add -fno-stack-protector in CFLAGS when
BR2_TARGET_ARM_TRUSTED_FIRMWARE_SSP is not set to fix older ATF
versions.
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/3301821171
Cc: Dick Olsson <hi@senzilla.io>
Cc: Sergey Matyukevich <geomatsi@gmail.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
SSP support requires support in ATF platform code. Not all platforms
implement plat_get_stack_protector_canary() hook. The result is build
failure:
(.text.asm.update_stack_protector_canary+0x4): undefined reference to `plat_get_stack_protector_canary'
Commit cf176128ec ("boot/arm-trusted-firmware: add SSP option")
originally introduces this issue. But then commit ccac9a5bbb
("boot/arm-trusted-firmware: don't force ENABLE_STACK_PROTECTOR") hid
the problem by effectively disabling SSP for all platforms. So only
after commit 09acc7cbc9 ("boot/arm-trusted-firmware: fix SSP
support") the issue showed up.
Make SSP an opt-in for platform that actually provide the
plat_get_stack_protector_canary() hook.
Cc: Sergey Matyukevich <geomatsi@gmail.com>
Cc: Dick Olsson <hi@senzilla.io>
Tested-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Commit ccac9a5bbb ("boot/arm-trusted-firmware: don't force
ENABLE_STACK_PROTECTOR") fixed a build failure but also effectively
disabled SSP entirely for ATF. This is because ENABLE_STACK_PROTECTOR is
set to 0 unconditionally in make_helpers/defaults.mk, overwriting any
environment set value. So we must pass ENABLE_STACK_PROTECTOR in
MAKE_OPTS for it to be effective. But to avoid said build failure we
can't pass ENABLE_STACK_PROTECTOR=0.
Only pass ENABLE_STACK_PROTECTOR when
BR2_TARGET_ARM_TRUSTED_FIRMWARE_SSP is enabled. Drop SSP_LEVEL value for
the !BR2_TARGET_ARM_TRUSTED_FIRMWARE_SSP case which is now unused.
Cc: Dick Olsson <hi@senzilla.io>
Cc: Sergey Matyukevich <geomatsi@gmail.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This patch adds the option BR2_TARGET_ARM_TRUSTED_FIRMWARE_RCW that
allows TF-A to encapsulate a pre-loaded RCW (Reset Configuration Word)
file into BL2.
Upcoming NXP QorIQ family board needs this option.
Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Reviewed-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Older versions of TF-A don't support setting CFLAGS on the make command
line. They use 'CFLAGS += ...' in the Makefile, which is still
overridden by the CFLAGS on the command line.
Fix this by moving the CFLAGS setting to the environment. Both older and
newer versions of TF-A handle this correctly.
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/2781800954
and a large number of other defconfigs
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Kory Maincent <kory.maincent@bootlin.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
It should be possible to compile ARM_TRUSTED_FIRMWARE without u-boot or EDK2.
For example, one might want to produce "bl31.bin" for use as an init stub for
the Raspberry Pi 4.
Signed-off-by: Christian Stewart <christian@paral.in>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
This patch updates the condition to handle custom tarballs as specified by
the configuration. This change is made to have cleaner condition and for
consistency.
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
There are two extra ending parentheses to be removed.
They are present from the beggining of TF-A package. I suppose extra
ending parenthese does not bring any issue, but it seems relevant to
remove them.
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Bumped the default version of the "arm-trusted-firmware" (a.k.a. TF-A)
from 2.5 to 2.7: updated the config and the tarball checksum.
Work-around CVE-2022-23960.
Updated the "qemu_aarch64_sbsa_defconfig" accordingly: it was using an
"arm-trusted-firmware" v2.4, it nows selects version 2.7.
Updated the license checksum because the license file changed
slightly: it mentions an additional file ("irq.h") released under a
dual GPL or MIT license.
Tested with the "qemu_aarch64_sbsa_defconfig" and QEMU: it boots
without new warning or error message.
Release notes:
<https://trustedfirmware-a.readthedocs.io/en/latest/change-log.html>
Signed-off-by: Olivier L'Heureux <olivier.lheureux@mind.be>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Some platforms such as TI's k3 outputs a TARGET_BOARD directory
in their ATF build output paths after PLATFORM.
$(@D)/build/<PLATFORM>/ does not contain the debug or release
directories for these platforms.
They are under $(@D)/build/<PLATFORM>/<TARGET_BOARD>/
Signed-off-by: Xuanhao Shi <x-shi@ti.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
By default the toolchain-wrapper enable -fPIE to the build of all packages.
TF-A support Position Independent Executable(PIE) only in few build cases,
therefore it should be disable by default.
If you still want to enable PIE, TF-A provide a "ENABLE_PIE" build options
that will override the cflags for the supported cases.
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
[Peter: Only do so for BR2_PIC_PIE]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since v2.2 release (commits 8cc36aec912 "doc: De-duplicate readme and
license files" and 9f1622b018ab "doc: Move content out of readme and
create new index page "), the license.rst file at the root of the git
repo is only telling to look at docs/license.rst file.
Let's point the ARM_TRUSTED_FIRMWARE_LICENSE_FILES to the correct file
and modify the .hash file accordingly.
The comment has also been wrong since we bumped from version 1.4 to 2.2
in commit a757d173f1 (boot/arm-trusted-firmware: bump to version
2.2). Drop referencing an explicit version, so that is is never wrong
again.
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
[yann.morin.1998@free.fr: also fix the comment.]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Similarly to Uboot, this patch adds the ability to copy in and build
out-of-source device tree sources during an ATF build.
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Commit 5f432df7e2 ("boot/arm-trusted-firmware: change
ENABLE_STACK_PROTECTOR value when disabled") set
ENABLE_STACK_PROTECTOR=0 when disabled. But since we pass this value as
MAKE_OPT, the internal ATF logic that sets ENABLE_STACK_PROTECTOR again
based on its initial value breaks. This leads to build failure:
make[1]: *** [/builds/buildroot.org/buildroot/output/build/arm-trusted-firmware-v2.4/build/a80x0_mcbin/release/libc/assert.o] Error 1
aarch64-buildroot-linux-uclibc-gcc.br_real: error: unrecognized command-line option ‘-fstack-protector-0’; did you mean ‘-fstack-protector’?
Move ENABLE_STACK_PROTECTOR to make environment instead to allow make to
change its value.
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/1497663294
Cc: Dick Olsson <hi@senzilla.io>
Cc: Sergey Matyukevich <geomatsi@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since commit
cf176128ec ("boot/arm-trusted-firmware:
add SSP option"), we are passing ENABLE_STACK_PROTECTOR=none when we
want to disable SSP usage in TF-A. While this works fine in recent
versions of TF-A, older versions such as TF-A will end up passing
-fstack-protector-none in this situation, which fails as this is not a
valid gcc option (the valid gcc option is -fno-stack-protector).
To solve this, we pass ENABLE_STACK_PROTECTOR=0 which was in older
TF-A versions used to say "don't do anything with SSP", and is also
still supported in newer versions of TF-A.
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/1478738580
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Buildroot sets appropriate ENABLE_STACK_PROTECTOR build flag value based
on the toolchain global BR2_SSP_* options, and all packages are built
with that setting.
However it might not be always convenient to automatically infer TF-A
stack protection from the toolchain features. For instance, secure
memory constraints may become an issue and all the extra TF-A features
need to be tuned or disabled in order to shrink TF-A firmware image.
Besides, for any value other than "none", TF-A platform specific hook
'plat_get_stack_protector_canary' must be implemented. However this hook
is not implemented by all the platforms supported by TF-A. For instance,
Allwinner currently does not provide such a hook.
Add an new option that a user can toggle to enable or disable SSP in
their ATF build. If enabled, the SSP level is automatically inherited
from the global setting.
Signed-off-by: Sergey Matyukevich <geomatsi@gmail.com>
[yann.morin.1998@free.fr: simplify logic with a single boolean]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Release notes:
https://trustedfirmware-a.readthedocs.io/en/latest/change-log.html#version-2-5
The existing qemu_aarch64_sbsa_defconfig was using an unpinned version
of ATF, so to avoid any regression, it is pinned to the previous
version, 2.4.
Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
[yann.morin.1998@free.fr: pin ATF version in qemu_aarch64_sbsa_defconfig]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
