Merge branch 'main' into error-refactor

tests/README.md

@@ -393,3 +393,31 @@ uvm.ssh.run("ls")
 snap = uvm.snapshot_full()
 uvm.help.tmux_ssh()
 ```
+
+It supports a number of options, you can check with `devtool sandbox --
+--help`.
+
+### Running outside of Docker
+
+Running without Docker
+
+```
+source /etc/os-release
+case $ID-$VERSION_ID in
+amzn-2)
+    sudo yum remove -y python3
+    sudo amazon-linux-extras install -y python3.8
+    sudo ln -sv /usr/bin/python3.8 /usr/bin/python3
+    sudo ln -sv /usr/bin/pip3.8 /usr/bin/pip3
+esac
+
+sudo pip3 install pytest ipython requests psutil tenacity filelock "urllib3<2.0" requests_unixsocket
+
+sudo env PYTHONPATH=tests HOME=$HOME ~/.local/bin/ipython3 -i ./tools/sandbox.py -- --binary-dir ../repro/v1.4.1
+```
+
+> :warning: **Notice this runs as root!**
+
+```python
+!id
+```

tests/conftest.py

@@ -176,18 +176,6 @@ def test_fc_session_root_path():
     yield fc_session_root_path
 
 
-@pytest.fixture(scope="session")
-def bin_cloner_path(test_fc_session_root_path):
-    """Build a binary that `clone`s into the jailer.
-
-    It's necessary because Python doesn't interface well with the `clone()`
-    syscall directly.
-    """
-    cloner_bin_path = os.path.join(test_fc_session_root_path, "newpid_cloner")
-    build_tools.gcc_compile("host_tools/newpid_cloner.c", cloner_bin_path)
-    yield cloner_bin_path
-
-
 @pytest.fixture(scope="session")
 def bin_vsock_path(test_fc_session_root_path):
     """Build a simple vsock client/server application."""
@@ -276,7 +264,7 @@ def fc_tmp_path(test_fc_session_root_path):
 
 
 @pytest.fixture()
-def microvm_factory(fc_tmp_path, bin_cloner_path, request):
+def microvm_factory(fc_tmp_path, request, record_property):
     """Fixture to create microvms simply.
 
     In order to avoid running out of space when instantiating many microvms,
@@ -290,10 +278,9 @@ def microvm_factory(fc_tmp_path, bin_cloner_path, request):
         jailer_binary_path = Path(binary_dir) / "jailer"
     else:
         fc_binary_path, jailer_binary_path = build_tools.get_firecracker_binaries()
+    record_property("firecracker_bin", str(fc_binary_path))
 
-    uvm_factory = MicroVMFactory(
-        fc_tmp_path, bin_cloner_path, fc_binary_path, jailer_binary_path
-    )
+    uvm_factory = MicroVMFactory(fc_tmp_path, fc_binary_path, jailer_binary_path)
     yield uvm_factory
     uvm_factory.kill()
     shutil.rmtree(fc_tmp_path)

tests/framework/defs.py

@@ -8,9 +8,6 @@
 # Firecracker's binary name
 FC_BINARY_NAME = "firecracker"
 
-# Jailer's binary name
-JAILER_BINARY_NAME = "jailer"
-
 # The Firecracker sources workspace dir
 FC_WORKSPACE_DIR = Path(__file__).parent.parent.parent.resolve()
 
@@ -32,10 +29,6 @@
 # Absolute path to the test results folder
 TEST_RESULTS_DIR = FC_WORKSPACE_DIR / "test_results"
 
-# Name of the file that stores firecracker's PID when launched by jailer with
-#  `--new-pid-ns`.
-FC_PID_FILE_NAME = "firecracker.pid"
-
 # The minimum required host kernel version for which io_uring is supported in
 # Firecracker.
 MIN_KERNEL_VERSION_FOR_IO_URING = "5.10.51"

tests/framework/jailer.py

@@ -7,7 +7,7 @@
 import stat
 from pathlib import Path
 
-from retry.api import retry_call
+from tenacity import Retrying, retry_if_exception_type, stop_after_delay
 
 from framework import defs, utils
 from framework.defs import FC_BINARY_NAME
@@ -226,13 +226,13 @@ def cleanup(self):
                 # Obtain the tasks from each cgroup and wait on them before
                 # removing the microvm's associated cgroup folder.
                 try:
-                    retry_call(
-                        f=self._kill_cgroup_tasks,
-                        fargs=[controller],
-                        exceptions=TimeoutError,
-                        max_delay=5,
-                        logger=None,
-                    )
+                    for attempt in Retrying(
+                        retry=retry_if_exception_type(TimeoutError),
+                        stop=stop_after_delay(5),
+                        reraise=True,
+                    ):
+                        with attempt:
+                            self._kill_cgroup_tasks(controller)
                 except TimeoutError:
                     pass
 
@@ -271,3 +271,8 @@ def _kill_cgroup_tasks(self, controller):
             if os.path.exists("/proc/{}".format(task)):
                 raise TimeoutError
         return True
+
+    @property
+    def pid_file(self):
+        """Return the PID file of the jailed process"""
+        return Path(self.chroot_path()) / (self.exec_file.name + ".pid")