RHOAIENG-16517: chore(tests): add sandboxing so that Dockerfile builds can only access files we know they access (opendatahub-io#803)

Author: jiridanek

.github/workflows/build-notebooks-TEMPLATE.yaml

@@ -39,6 +39,13 @@ jobs:
 
       - uses: actions/checkout@v4
 
+      - run: mkdir -p $TMPDIR
+
+      # for bin/buildinputs in scripts/sandbox.py
+      - uses: actions/setup-go@v5
+        with:
+          cache-dependency-path: "**/*.sum"
+
       - name: Login to GitHub Container Registry
         uses: docker/login-action@v3
         with:

Makefile

@@ -69,7 +69,8 @@ define build_image
 		$(eval BUILD_ARGS := --build-arg BASE_IMAGE=$(BASE_IMAGE_NAME)),
 		$(eval BUILD_ARGS :=)
 	)
-	$(CONTAINER_ENGINE) build $(CONTAINER_BUILD_CACHE_ARGS) --tag $(IMAGE_NAME) --file $(2)/Dockerfile $(BUILD_ARGS) $(ROOT_DIR)/
+	$(ROOT_DIR)/scripts/sandbox.py --dockerfile '$(2)/Dockerfile' -- \
+		$(CONTAINER_ENGINE) build $(CONTAINER_BUILD_CACHE_ARGS) --tag $(IMAGE_NAME) --file '$(2)/Dockerfile' $(BUILD_ARGS) {}\;
 endef
 
 # Push function for the notebok image:

ci/cached-builds/gha_lvm_overlay.sh

@@ -7,7 +7,8 @@ set -Eeuo pipefail
 # This script creates file-backed volumes on /dev/root and /dev/sda1, then creates ext4 over both, and mounts it for our use
 # https://github.com/easimon/maximize-build-space/blob/master/action.yml
 
-root_reserve_mb=2048
+# root_reserve_mb=2048 was running out of disk space building cuda images
+root_reserve_mb=4096
 temp_reserve_mb=100
 swap_size_mb=4096
 

poetry.lock

@@ -13,6 +13,7 @@ python = "~3.12"
 [tool.poetry.group.dev.dependencies]
 pytest = "^8.2.2"
 pytest-subtests = "^0.12.1"
+pyfakefs = "^5.7.2"
 
 [build-system]
 requires = ["poetry-core"]

pyproject.toml

@@ -0,0 +1,77 @@
+#! /usr/bin/env python3
+
+import argparse
+import logging
+import pathlib
+import shutil
+import subprocess
+import sys
+import tempfile
+import json
+from typing import cast
+
+ROOT_DIR = pathlib.Path(__file__).parent.parent
+MAKE = shutil.which("gmake") or shutil.which("make")
+
+logging.basicConfig()
+logging.root.name = pathlib.Path(__file__).name
+
+
+class Args(argparse.Namespace):
+    dockerfile: pathlib.Path
+    remaining: list[str]
+
+
+def main() -> int:
+    p = argparse.ArgumentParser(allow_abbrev=False)
+    p.add_argument("--dockerfile", type=pathlib.Path, required=True)
+    p.add_argument('remaining', nargs=argparse.REMAINDER)
+
+    args = cast(Args, p.parse_args())
+
+    print(f"{__file__=} started with {args=}")
+
+    if not args.remaining or args.remaining[0] != "--":
+        print("must specify command to execute after double dashes at the end, such as `-- command --args ...`")
+        return 1
+    if not "{};" in args.remaining:
+        print("must give a `{};` parameter that will be replaced with new build context")
+        return 1
+
+    if not (ROOT_DIR / "bin/buildinputs").exists():
+        subprocess.check_call([MAKE, "bin/buildinputs"], cwd=ROOT_DIR)
+    stdout = subprocess.check_output([ROOT_DIR / "bin/buildinputs", str(args.dockerfile)],
+                                     text=True, cwd=ROOT_DIR)
+    prereqs = [pathlib.Path(file) for file in json.loads(stdout)] if stdout != "\n" else []
+    print(f"{prereqs=}")
+
+    with tempfile.TemporaryDirectory(delete=True) as tmpdir:
+        setup_sandbox(prereqs, pathlib.Path(tmpdir))
+        command = [arg if arg != "{};" else tmpdir for arg in args.remaining[1:]]
+        print(f"running {command=}")
+        try:
+            subprocess.check_call(command)
+        except subprocess.CalledProcessError as err:
+            logging.error("Failed to execute process, see errors logged above ^^^")
+            return err.returncode
+    return 0
+
+
+def setup_sandbox(prereqs: list[pathlib.Path], tmpdir: pathlib.Path):
+    # always adding .gitignore
+    gitignore = ROOT_DIR / ".gitignore"
+    if gitignore.exists():
+        shutil.copy(gitignore, tmpdir)
+
+    for dep in prereqs:
+        if dep.is_absolute():
+            dep = dep.relative_to(ROOT_DIR)
+        if dep.is_dir():
+            shutil.copytree(dep, tmpdir / dep, symlinks=False, dirs_exist_ok=True)
+        else:
+            (tmpdir / dep.parent).mkdir(parents=True, exist_ok=True)
+            shutil.copy(dep, tmpdir / dep.parent)
+
+
+if __name__ == '__main__':
+    sys.exit(main())

scripts/sandbox.py

@@ -0,0 +1,55 @@
+#! /usr/bin/env python3
+
+import glob
+import logging
+import pathlib
+import sys
+import tempfile
+
+import pyfakefs.fake_filesystem
+
+from scripts.sandbox import setup_sandbox
+
+ROOT_DIR = pathlib.Path(__file__).parent.parent
+
+logging.basicConfig()
+logging.root.name = pathlib.Path(__file__).name
+
+class TestSandbox:
+    def test_filesystem_file(self, fs: pyfakefs.fake_filesystem.FakeFilesystem):
+        pathlib.Path("a").write_text("a")
+
+        with tempfile.TemporaryDirectory(delete=True) as tmpdir:
+            setup_sandbox([pathlib.Path("a")], pathlib.Path(tmpdir))
+            assert (pathlib.Path(tmpdir) / "a").is_file()
+
+    def test_filesystem_dir_with_file(self, fs: pyfakefs.fake_filesystem.FakeFilesystem):
+        pathlib.Path("a/").mkdir()
+        pathlib.Path("a/b").write_text("b")
+
+        with tempfile.TemporaryDirectory(delete=True) as tmpdir:
+            setup_sandbox([pathlib.Path("a")], pathlib.Path(tmpdir))
+            assert (pathlib.Path(tmpdir) / "a").is_dir()
+            assert (pathlib.Path(tmpdir) / "a" / "b").is_file()
+
+    def test_filesystem_dir_with_dir_with_file(self, fs: pyfakefs.fake_filesystem.FakeFilesystem):
+        pathlib.Path("a/b").mkdir(parents=True)
+        pathlib.Path("a/b/c").write_text("c")
+
+        with tempfile.TemporaryDirectory(delete=True) as tmpdir:
+            setup_sandbox([pathlib.Path("a")], pathlib.Path(tmpdir))
+            assert (pathlib.Path(tmpdir) / "a").is_dir()
+            assert (pathlib.Path(tmpdir) / "a" / "b").is_dir()
+            assert (pathlib.Path(tmpdir) / "a" / "b" / "c").is_file()
+
+    def test_filesystem_file_in_dir_in_dir(self, fs: pyfakefs.fake_filesystem.FakeFilesystem):
+        pathlib.Path("a/b").mkdir(parents=True)
+        pathlib.Path("a/b/c").write_text("c")
+
+        with tempfile.TemporaryDirectory(delete=True) as tmpdir:
+            setup_sandbox([pathlib.Path("a/b/c")], pathlib.Path(tmpdir))
+            for g in glob.glob("**/*", recursive=True):
+                logging.debug("%s", g)
+            assert (pathlib.Path(tmpdir) / "a").is_dir()
+            assert (pathlib.Path(tmpdir) / "a" / "b").is_dir()
+            assert (pathlib.Path(tmpdir) / "a" / "b" / "c").is_file()