Added

* The CPU Template can be set with an API call on PUT /machine-config. The only
  available template is T2.
* Hyperthreading can be enabled/disabled with an API call on PUT
  /machine-config. By default, hyperthreading is disabled.
* Added boot time performance test (tests/performance/test_boottime.py).
* Added Rate Limiter for Virtio block and net devices. The Rate Limiter uses
  two token buckets to limit rate on bytes/s and ops/s. The rate limiter can be
  (optionally) configured per drive with a PUT on /drives/{drive_id} and per
  network interface with a PUT on /network-interface/{iface_id}.
* Implemented pre-boot PUT updates for /boot-source, /drives, /network-interfaces
  and /vsock.
* Added integration tests for PUT updates.

Changed

* Moved the API definition (swagger/firecracker-beta.yaml) to the api_server i
  crate.
* Removed "console=ttyS0" and added "8250.nr_uarts=0" to the default kernel
  command line to decrease the boot time.
* Changed the CPU topology to have all logical CPUs on a single socket.
* Removed the upper bound on CPU count as with musl there is no good way to get
  the total number of logical processors on a host.
* Build time tests now print the full output of commands.
* Disabled the Performance Monitor Unit and the Turbo Boost.
* Check the expected KVM capabilities before starting the VM.
* Logs now have timestamps.

Fixed

* testrun.sh can run on platforms with more than one package manager by setting
  the package manager via a command line parameter (-p).
* Allow correct set up of multiple network-interfaces with autogenerated MAC.
* Fixed sporadic bug in virtio which was causing lost packages.
* Don't allow PUT requests with empty body on /machine-config.
* Deny PUT operations after the microvm boots (exception: the temporarily fix
  for live resize of block devices).

Removed

* Removed examples crate. This used to have a Python example of starting
  Firecracker. This is replaced by test_api.py integration tests.
* Removed helper scripts for getting coverage and coding style errors. These
  were replaced by test_coverage.py and test_style.py test integration tests.
* Removed --vmm-no-api command line option. Firecracker can only be started via
  the API.

Added

* Users can interrogate the Machine Configuration (i.e. vcpu count and memory size) using a GET request on /machine-config.
* The logging system can be configured through the API using a PUT on /logger.
* Block devices support live resize by calling PUT with the same parameters as when the block was created.
* Release builds have Link Time Optimization (LTO) enabled.
* Firecracker is built with musl, resulting in a statically linked binary.
* More in-tree integration tests were added as part of the continuous integration system.

Changed

* The vcpu count is enforced to 1 or an even number.
* The Swagger definition of rate limiters was updated.
* Syslog-enabled logs were replaced with a hostfile backed mechanism.

Fixed

* The host topology of the CPU and the caches is not leaked into the microvm anymore.
* Boot time was improved by advertising the availability of the TSC deadline timer.
* Fixed an issue which prevented Firecracker from working on 4.14 (or newer) host kernels.
* Specifying the MAC address for an interface through the API is optional.

Removed

* Removed support for attaching vsock devices.
* Removed support for building Firecracker with glibc.

[0.1.1]

Changed:

* Users can now specify the MAC address of a guest network interface via the
PUT network interface API request. Previously, the guest MAC address
parameter was ignored.

Fixed:

* Fixed a guest memory allocation issue, which previously led to a
potentially significant memory chunk being wasted.
* Fixed an issue which caused compilation problems, due to a compatibility
breaking transitive dependency in the tokio suite of crates.