Fast Model support for LAVA
This package contains low level code for managing fast models in LAVA.
Currently it is designed to manage NFS mounts, setting up the root filesystem
(exporting the provided tarball to the mount point and blessing it with
fast-model specific changes, if required). It also contains code for starting
the fast model runtime, monitoring execution and eventual shutdown (or forced
shutdown, if required).
Support for lava-serial plugin is being considered (to be able to get a
LAVA-Serial access to a running fast model emulation)
* Working installation of FastModelsPortfolio 7.0 from ARM
* Root access on local machine (to manipulate NFS shares, gained
automatically via sudo)
* Bleeding edge version of json-document and more mature versions
of lava-tool and their dependencies (in the upcoming release also
a bleeding edge version of lava-serial)
(I made this up as that part does not work yet).
To start a fast model on your machine run the following command:
$ lava fast-model start --kernel=uImage --initrd=initrd --rootfs=roottfs.tar.gz
Unpacking rootfs... done
Setting up NFS share... done
Starting fast model... done
You can connect to the serial console with
$ lava serial console --network localhost:5001
Pressing CTRL-C will terminate this simulation
Shutting down... done
Removing NFS share... done
Removing root filesystem... done
To use fast models in your code you can try the following example.
(I just made it up, it does not work yet)
from lava.fastmodels.simulators import LinuxSimulator
from lava.fastmodels.models import ModelLibrary
def on_line(self, line):
# Get a library, any path works (as long as you have models there)
library = ModelLibrary('/srv/lava/instances/dev/var/lib/lava/fastmodels/models')
# Get a model from the library
model = library['A15x1']
simulator = LinuxSimulator(model, rootfs="rootfs.tar.gz", uImage="uImage")
serial = simulator.serial_console
TODO: Brief introduction on what you do with files - including link to relevant help section.