GitLab

this gets rid of the non-fatal build error 'No names found, cannot describe anything'

These all do the same thing except with different hosts or NICs. Combine them
in one script running all configurations.
- gem5_i40e_pair
- qemu_i40e_pair
- qemu_e1000_pair

Helps quickly determine available CPU features in host simulators.

Simply installs vanilla simics under sims/external/simics. Because of licensing
restrictions we cannot automatically download simics from the intel page.
Instead the makefile will prompt the user to download to files and add them to
the simics directory.

This commit does not include any simbricks integration at all yet.
Co-authored-by: Antoine Kaufmann <antoinek@mpi-sws.org>

This enables other environments (in other code bases) to use a different
definition or disable the checks. Came up in the context of Simics.
Co-authored-by: Antoine Kaufmann <antoinek@mpi-sws.org>

While we configure gem5 and qemu to directly load the kernel, simics will use a
proper EFI-based firmware for boot.

Currently we invoke m5 checksum for gem5 but nothing at all for other
simulators. This just adds an echo "ready for checkpoint". We use this in
simics to generate a checkpoint, but it is also sane default for simulators not
supporting checkpoints, as at least the logs will show what's going on.

Previously the default was m5 exit with a special case for QEMU. Here we flip
this in anticipation of Simics support.

Instead of first converting to strings. Should reduce memory consumption when
dealing with large logs.

yapf takes a long time to run when this directory contains large output files

drastically speeds up simulation performance in the presence of a lot of output

The newer linux drivers now use 16b instead of 32b descriptors. So far
this leads to an abort. This adds proper support for 16b descriptors.

Depending on the order some of the constructors get executed, instances of the
logger can end up with a null pointer for the runner. This led to a segfault
when calling TimePs() on that nullptr.

Here we fix this in two steps: first the logger takes a device pointer, which
in turn contains a pointer to the runner that will eventually be initialized.
Also fix up the logger to simply use 0 as a timestamp for log events before the
runner is there.

This adds the physical memory region to the kernel memory manager in a second
(emulated) numa node.

Co-authored-by: Hejing Li <hajeongee@gmail.com>