Vijayenthiran Subramaniam authored Oct 05, 2018 and Arvind Chauhan committed Oct 12, 2018

Print uart log file name before launching the model. This is useful when
there are multiple log files and want to display the latest one.

Change-Id: I11441185cc813671a14c8091531a4ae28201f745
Signed-off-by: Vijayenthiran Subramaniam <vijayenthiran.subramaniam@arm.com>

Thomas Abraham authored Oct 05, 2018 and Arvind Chauhan committed Oct 12, 2018

Improve the user help message of the busybox_boot command for the SGI platforms.
The help message now includes examples of the command usage.

Change-Id: Idc41c31eb8b2437e851baef5e234da7785e631f3
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

Thomas Abraham authored Sep 07, 2018 and Arvind Chauhan committed Oct 12, 2018

Add support for automatted termination of the test case. The script
waits for the busybox boot to complete and then kills the model.

Change-Id: Iaeea280f162092cc4031c4a442bd0121efbde25a
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

Thomas Abraham authored Sep 07, 2018 and Arvind Chauhan committed Oct 12, 2018

Allow for automation using the -j option. With -j=1 as the parameter,
the model is launched as a background process, thus allowing the
calling test case scripts to continue execution and check for
completion of the test case. With -j=0, the model is launched as a
foreground process.

Change-Id: Icfc4df685cf23ed59532054e2d6cf162b4e2938c
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

Thomas Abraham authored Sep 07, 2018 and Arvind Chauhan committed Oct 12, 2018

Add common utility functions to support checking of boot completion
and kiiling of the model.

Change-Id: Ia18df8334ada713ace65a77b1a180703f21af540
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

vishwanathahg authored Sep 04, 2018 and Arvind Chauhan committed Oct 12, 2018

Create or use an existing 64KB NOR flag image and supply it as
the back store for the NOR1 interface.

Change-Id: I67ac67eeaa296b03e6ec2e465ddd35bf19b90d59
Signed-off-by: Vishwanatha HG <vishwanatha.hg@arm.com>
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

vishwanathahg authored Sep 04, 2018 and Arvind Chauhan committed Oct 12, 2018

SGI platforms can be supplied with NOR flash images of 64KB
in size. These NOR flash images can be loaded to NOR1 or
the NOR2 flash interface. Add a new helper function to
create a 64KB nor flash image.

Change-Id: I5522642df7ebac1677b315a6033ff52ebf2f87fc
Signed-off-by: Vishwanatha HG <vishwanatha.hg@arm.com>
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

Thomas Abraham authored Sep 04, 2018 and Arvind Chauhan committed Oct 12, 2018

Use the function 'find_tap_interface' to determine a available tap
network interface.

Change-Id: Id5587122fb614ea05d61b2e8aff512c08716cefe
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

Thomas Abraham authored Sep 03, 2018 and Arvind Chauhan committed Oct 12, 2018

To allow for auto detection of the tap interface, add a new function
that parses the available network interfaces and finds a tap interface
from it.

Change-Id: Id07b39fc983b2e42d4b1d26ad9e61c9b018c46e3
Signed-off-by: Samarth Parikh <samarth.parikh@arm.com>
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

Thomas Abraham authored Aug 17, 2018 and Arvind Chauhan committed Oct 12, 2018

The existing structure of the sgi575 platform model-scripts is not
extensible when adding support for additional sgi platforms and
test cases. So the sgi platform model-scripts have been reorganized
as below

    model-scripts
         |
          ---sgi
              |
              |--- distro-boot.sh
              |--- busybox-boot.sh
               --- platforms
                       |
                        --- sgi575
                              |
                               --- run_model.sh

The new 'sgi' directory will hold all the test scripts. The new
'platforms' directory will hold the platform specific run_model.sh
script and there is a directory for each supported platform.

The distro-boot.sh is now moved to 'sgi' directory and the changes
required to align with this new structure has been included in this
script. There is also a new 'busybox-boot.sh' script as well to be
used for busybox boot tests.

Change-Id: I78445bf59f0690900f60a819500ab455e1c14179
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

Thomas Abraham authored Aug 16, 2018 and Arvind Chauhan committed Oct 12, 2018

The SGI-575 platform's run_model.sh script has lot of unused code
and needs to be refreshed a bit to adapt the changes in the command
line options. So remove the unused portions and update the script.

Change-Id: Ib034afc83212d618f6c61eb55ad6f12c4ac4b0ef
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

vishwanathahg authored Aug 03, 2018 and Arvind Chauhan committed Oct 12, 2018

The '-a' option allow to pass addtional model parameters to sgi575 run_model.sh file.

Change-Id: I76c5518836def7add46d0b26be43971a041bc804
Signed-off-by: Vishwanatha HG <vishwanatha.hg@arm.com>

Thomas Abraham authored Aug 06, 2018 and Arvind Chauhan committed Oct 12, 2018

Update the run-model script to align with the changes in
output paths listed in build script config files.

Change-Id: I83836bd92eea90a9c287925c78e369ba1a50da2b
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

Thomas Abraham authored Aug 05, 2018 and Arvind Chauhan committed Oct 12, 2018

To avoid passing a MODEL parameter for every run, add a default
model path to run_model.sh.

Change-Id: I67d2787fb0818c9f4549c0c93f630fe3cbf82f70
Signed-off-by: Thomas Abraham <thomas.abraham@arm.com>

SCPv2 build for SGI575 doesn't implement MCP RAM

Change-Id: I3d972a2e168a4714d759674e277d6881fd6d829b
Signed-off-by: Arvind Chauhan <arvind.chauhan@arm.com>

Change-Id: Ia8ea050117aeb1d003e492cc994171af1807bb0e
Signed-off-by: Nariman Poushin <nariman.poushin@linaro.org>

Change-Id: I183b85244bd2ebac6772a94579ded7d4b6b89ab7
Signed-off-by: Nariman Poushin <nariman.poushin@linaro.org>

Signed-off-by: Manoj Kumar <manoj.kumar3@arm.com>

So far for FVP, parameter MODEL is exported with model name but
sgi575 expects the path name instead of model. This patch makes the
run_model.sh of sgi575 to be inline with FVP run_model.sh

Change-Id: I015eb05f713b1f2abfb942382827db3ee4465bcb
Signed-off-by: Manoj Kumar <manoj.kumar3@arm.com>

This is to account for the issue of the model defaulting to the highest
architecture spec available, which causes an assert in arm-tf.

As per:
https://github.com/ARM-software/tf-issues/issues/552



Signed-off-by: Nariman Poushin <nariman.poushin@linaro.org>

Signed-off-by: Nariman Poushin <nariman.poushin@linaro.org>

This is to prepare for the addition of model scripts for other platforms

Signed-off-by: Nariman Poushin <nariman.poushin@linaro.org>

The AEMv8 FVP model can be launched in Aarch32 mode by setting
"CONFIG64=0" for each core.

Use the "--aarch32" parameter to determine that we want to launch the
model in Aarch32 mode.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

Add the ability to specifiy commandline parameters to the script.

Other than "help", there are no parameters added in this patch.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

The Cortex-A32 model is very similar to the AEMv8 model and uses a
subset of commandline parameters to launch it.

The main difference is that the Cortex-A32 model only has 1 cluster and
therefore doesn't uspport the "-C cluster" options.

The model is Aarch32 mode only, but it uses the same DTB as the 64-bit
AEMv8 model.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

The FDT was located at 0x83000000 but this address is reserved for
OPTEE.

To prevent a clash, move the FDT to 0x82000000, which is also suitable
for 32-bit kernels.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

The Foundation Model RTC runs too fast, so use the --use-real-time
parameter to enforce realistic timing.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

Save the uart logs for the AEMv8 models.  The Foundation models don't
have a option for storing the uart logs.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

The ARMLT kernel tree for 16.03 use DTS files with GICv3 support enabled
by default.  The files have also been renamed.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

Pass the model parameter so that shutdown really does shut down the
model, rather than reboot it.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

If the users specifies NET=1 when launching the model, the networking
params will be appended to the model launch commandline.

When using the AEMv8 models, if the MAC address is not supplied, a MAC
address will be generated.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

If you are trying to boot the fvp-uefi variant, you don't need a
kernel/dtb/initrd, so make these files optional.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

Cache state modelling has proven to be frustrating for users who think
their system is broken when it takes such a long time to boot, eg, to
the Android GUI.

Disabling cache state modelling provides a massive increase in speed.

Users can set the environment variable to 1 if they wish to enable cache
state modelling, eg:

    export CACHE_STATE_MODELLED=1

Or set it when they run the model, eg:

    CACHE_STATE_MODELLED=1 ./model-scripts/run_model.sh <output dir>

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

Sandrine Bailleux authored Jul 24, 2015 and Ryan Harkin committed Jul 24, 2015

Cache state modelling is slower, so some users may wish to disable it,
but we enable by default to provide better test coverage.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

Sandrine Bailleux authored Jul 23, 2015 and Ryan Harkin committed Jul 24, 2015

Make some of the model parameters configurable through environment
variables and assign default values to them. The default values
have been chosen according to the recommended configuration in the
ARM Trusted Firmware User Guide.

The following FVP parameters are now configurable:

 - secure memory
   default: disabled
   Was unconditionally disabled previously.

 - cache state modelled [Base AEM FVP only]
   default: disabled
   Was unconditionally disabled previously.

 - GICv3 mode [Foundation FVP only]
   default: enabled

Sandrine Bailleux authored Jul 23, 2015 and Ryan Harkin committed Jul 24, 2015

run_model.sh differentiates between the Base AEM and Foundation FVPs
by looking at the model version string. It uses the variable MODEL
to get the path to the FVP binary. However, when using the Foundation
FVP, it then uses the variable FOUNDATION_MODEL, forcing the user
to define both MODEL and FOUNDATION_MODEL variables.

This patch fixes the issue by using the MODEL variable all the time.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>

Added the initial version of the run_model.sh script.

This model allows the user to specify variables for most parameters used
when running the models, or to choose sensible defaults where possible.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>