Adds external packages repository for buildroot containing:
* Packing for optee_client, optee_examples, optee_test and optee_benchmark
* Configs and a script to assembly defconfigs needed for buildroot

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

Switches to xtest-clean-common optee-examples-clean-common instead of
xtest-clean optee-examples-clean in the optee-os-clean-common. This is
needed in case the platform make file doesn't provide these targets.

Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

The 'strace-clean' target runs the 'clean' target in the strace project
directly, and afterwards removes the Makefile and configure scripts as well.
This is the same strategy used by other platforms using strace (hikey.mk and
hikey960.mk), however other platforms check for the existence of the Makefile
before executing $(MAKE), allowing a build to be cleaned multiple times.
Without this check, subsequent invocations of 'strace-clean' (or simply
'clean') fail as there is no Makefile to use.

Fixes:

$ make strace-clean
[...]
$ make strace-clean
[...]
make[1]: Entering directory '/home/goatshriek/optee-armv8/strace'
make[1]: *** No rule to make target 'clean'.  Stop.

Signed-off-by: Joel Anderson <jeander@vt.edu>
Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>

Like other platforms (fvp.mk, hikey.mk, hikey960.mk, hikey_debian.mk)
define the edk2-call macro to invoke the EDK2 build script rather than
$(MAKE) directly. Indeed, the edk2-clean-common target in common.mk
expects $(edk2-call) to take the 'cleanall' argument, which the EDK2
build script does accept, but the EDK2 makefile doesn't (there is no
cleanall rule in EDK2).

Fixes:

 $ make clean
 [...]
 make[1]: Entering directory '/home/jerome/optee_repo_qemu_v8/edk2'
 make[1]: *** No rule to make target 'cleanall'.  Stop.

Another option to fix the above error would be to change the edk2-clean
target to no use edk2-clean-common and call "$(MAKE) clean" directly.
But then, it would be inconsistent with other platforms and the EDK2
build script may be a better interface than raw make (more stable).

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reported-by: Joel Anderson <jeander@vt.edu>
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

With the recent QEMU v7 changes in manifest.git:
  918f04199c ("qemu: general cleanup and remote changes")
where the remote "github" was introduced, we introduced a regression for
the QEMU v7 Travis builds. The remote name "github" has been used in
Travis when setting up a remote for the build.git itself. By renaming
the remote we do not get any naming collisions and the Travis build
script will be prepared for similar changes in manifest.git for the
other platforms.

Signed-off-by: Joakim Bech <joakim.bech@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Fix benchmark build regression, which was introduced by output path changes in
optee_benchmark last patches.

Reviewed-by: Joakim Bech <joakim.bech@linaro.org>
Signed-off-by: Igor Opaniuk <igor.opaniuk@linaro.org>

Multiple personality support in strace is not something that we make use
of in OP-TEE for the moment, so instead of putting more prerequisites on
our builds let us just disable this feature in strace.

Signed-off-by: Joakim Bech <joakim.bech@linaro.org>
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Joakim Bech <joakim.bech@linaro.org>

Signed-off-by: Joakim Bech <joakim.bech@linaro.org>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>

QEMU v8 and FVP are both now working with upstream edk2 and
edk2-platforms. QEMU v8's dsc file is located in the edk2 folder whilst
the dsc for FVP is in the edk2-platforms folder. Due to this we had to
export the WORKSPACE variable differently otherwise we would get either
compiler errors or we would get the "Build" folder created on the top
level (something that we do not want).

Signed-off-by: Joakim Bech <joakim.bech@linaro.org>
Tested-by: Joakim Bech <joakim.bech@linaro.org> (QEMU v8, FVP)
Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>

We are switching from using the old semihosting based way of loading and
running OP-TEE, kernel, root fs etc on FVP to instead use a proper boot
image similar to how it is working on real devices. A difference here
compared to the past is that we are using Grub here to load the kernel.

Signed-off-by: Joakim Bech <joakim.bech@linaro.org>
Tested-by: Joakim Bech <joakim.bech@linaro.org> (FVP)
Reviewed-by: Victor Chong <victor.chong@linaro.org>

As a first step to update EDK2 to running latest  we are introducing
grub which will be a needed component when booting UEFI without
semihosting.

This particular patch will just bulid grub, it will still not be used
when running FVP (that is enabled in later patches in this series).

Signed-off-by: Joakim Bech <joakim.bech@linaro.org>
Reviewed-by: Victor Chong <victor.chong@linaro.org>

Signed-off-by: Alex Demyankov <alex.demyankov@gmail.com>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>

It's assumed we have a Juno R0 board. It isn't always the case. If someone
has another revision, he'll prefer to use the right device tree for his
board.

Signed-off-by: Fathi Boudra <fathi.boudra@linaro.org>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

ERROR_DEPRECATED=1 may cause build errors when ARM deprecates APIs
upstream. Since we're tracking upstream master, it can happen at any
time, and indeed we have just met this build issue:

 plat/qemu/qemu_bl31_setup.c:129:2: error: unknown field 'g0_interrupt_num' specified in initializer
   .g0_interrupt_num = ARRAY_SIZE(irq_sec_array),
   ^

Although we usually want to use the latest APIs, having a build that
can break randomly is not nice to our users. Therefore, remove the
ERROR_DEPRECATED flag from our Makefile to fix the compile error and
give us some time to adapt to upstream changes.

Of course, the weakness in this reasoning is that by removing
ERROR_DEPRECATED, we are likely to not even notice when an API has
changed. Anyway, this seems to be the most reasonable option given that
there is no WARN_DEPRECATED flag.

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>

The path to the executable in FVP Foundation model has changed, update
the path accordingly.

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

On the mtk8173 device there is partition which is ~25MB in size and that
area contains both Linux kernel and the root fs. When we stepped up from
Linux kernel 4.9 to 4.12 we got too big to fit in that area and
therefore we have tweaked the config and removed a couple of things not
being used in our setup.

Signed-off-by: Joakim Bech <joakim.bech@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

It seems that Shippable CI does not like running QEMU with 4 virtual
cores (or maybe we have a bug), because "make check" hangs when QEMU
is started with -smp 4. Revert to 1 core by default.

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>

The QEMU_SMP environment variable or makefile macro may be set to
control the number of cores used during "make check".

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>

Now that OP-TEE supports several cores in QEMU, use 4 of them when
running "make check". This is hopefully more likely to catch any
concurrency bug.

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>

Reviewed-by: Joakim Bech <joakim.bech@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

Adds missing atf-fb dependency to lloader.

Reviewed-by: Joakim Bech <joakim.bech@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

Drops the -O0 flag passed in CFLAGS when compiling ARM-TF. This is
needed to ensure that the binaries doesn't overflow their assigned
memory.

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

Drops the -O0 flag passed in CFLAGS when compiling ARM-TF. This is
needed to ensure that the binaries doesn't overflow their assigned
memory.

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

Drops the -O0 flag passed in CFLAGS when compiling ARM-TF. This is
needed to ensure that the binaries doesn't overflow their assigned
memory.

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

Drops the -O0 flag passed in CFLAGS when compiling ARM-TF. This is
needed to ensure that the binaries doesn't overflow their assigned
memory.

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

Drops the -O0 flag passed in CFLAGS when compiling ARM-TF. This is
needed to ensure that the binaries doesn't overflow their assigned
memory.

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>

The 'Known issues' section relates to QEMU, so make it a subsection
under 'QEMU'.

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>

l-loader manual build instructions have been integrated into makefiles

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Otherwise they accumulate in $(ROOT)

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Sync with other hikey*.mk files

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Add target to delete OP-TEE deb packages

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Use new l-loader repo in
https://github.com/96boards-hikey/l-loader


branch: testing/hikey960_v1.2

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

Sync with the newer instructions in hikey_debian

Signed-off-by: Victor Chong <victor.chong@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>