Validate builder provenance and guard detached package jobs

This commit is contained in:
Codex
2026-10-13 04:22:23 +03:00
parent 440fe42464
commit 7a23508e0c
2 changed files with 374 additions and 11 deletions
+206
View File
@@ -1,6 +1,9 @@
import hashlib
import contextlib
import io
import importlib.util
import json
import lzma
import os
from pathlib import Path
import tempfile
@@ -109,6 +112,209 @@ class BuilderTests(unittest.TestCase):
other = self.module.Builder(self.project, self.project / 'other-build', self.project / 'cache')
self.assertNotEqual(self.builder().runroot, other.runroot)
def existing_builder(self, source=None, cache=None, layer=None, snapshot='old-snapshot'):
"""A real executable models only the external Podman boundary."""
source = source or self.project
cache = cache or self.project / 'cache'
image = self.project / 'image'
image.mkdir(exist_ok=True)
inputs = {'stage3': {'url': 'https://example.test/stage.tar.xz',
'sha256': hashlib.sha256(lzma.compress(b'original layer')).hexdigest()},
'portage': {'filename': 'repo.tar.gz', 'commit': 'old-snapshot'}}
(image / 'sources.json').write_text(json.dumps(inputs))
builder = self.builder()
(builder.work / 'podman').mkdir(parents=True)
fixture = self.project / 'podman-fixture.json'
fixture.write_text(json.dumps({'Image': 'original-image-id',
'Mounts': [{'Source': str(source), 'Destination': '/hydra-source', 'RW': False},
{'Source': str(cache), 'Destination': '/hydra-cache', 'RW': False}],
'RootFS': {'Layers': [layer or 'sha256:' + hashlib.sha256(b'original layer').hexdigest()]},
'snapshot': snapshot}))
executable = self.project / 'fixture-podman'
executable.write_text('''#!/usr/bin/env python3
import json, pathlib, sys
root = pathlib.Path(__file__).parent
data = json.loads((root / 'podman-fixture.json').read_text())
args = sys.argv[7:]
if 'exists' in args:
sys.exit(0)
if 'inspect' in args:
print(json.dumps([data]))
elif args[0] == 'top':
print('HPID ARGS')
if data.get('active_job'):
print('123 /usr/bin/python3 /usr/lib/python-exec/python3/emerge --update package')
elif args[0] == 'cp':
snapshot = data.get('snapshot_pending') if args[1].endswith('.hydra-snapshot-input') else data['snapshot']
if snapshot is None:
sys.exit(1)
pathlib.Path(args[-1]).write_text(snapshot + '\\n')
elif args[0] == 'exec' and data.get('stopped'):
sys.exit(125)
elif args[0] == 'exec' and 'packages.lock' in args[-1]:
sys.exit(75 if data.get('job_lock') else 0)
elif args[0] == 'exec' and args[-1].startswith('cat '):
snapshot = data['snapshot'] or (data.get('snapshot_pending') if '.hydra-snapshot-input' in args[-1] else None)
if snapshot is None:
sys.exit(1)
print(snapshot)
else:
if args[0] == 'start':
data['stopped'] = False
(root / 'podman-fixture.json').write_text(json.dumps(data))
with (root / 'mutations.jsonl').open('a') as stream:
stream.write(json.dumps(args) + '\\n')
''')
executable.chmod(0o755)
builder.podman = str(executable)
with builder.locked():
(builder.cache / 'stage.tar.xz').write_bytes(lzma.compress(b'original layer'))
return builder, inputs
def assert_prepare_refused(self, builder, inputs, reason):
with patch.object(builder, 'fetch', return_value=inputs):
with self.assertRaisesRegex(self.module.BuildError, reason):
builder.prepare()
self.assertFalse((self.project / 'mutations.jsonl').exists(), 'refusal must precede container/config mutations')
self.assertNotIn('prepare', builder.manifest(), 'refusal must not record false provenance')
def test_other_worktree_mount_is_refused_before_preparation(self):
builder, inputs = self.existing_builder(source=self.project / 'other-worktree')
self.assert_prepare_refused(builder, inputs, 'source')
def test_other_cache_mount_is_refused_before_preparation(self):
builder, inputs = self.existing_builder(cache=self.project / 'other-cache')
self.assert_prepare_refused(builder, inputs, 'cache')
def test_changed_stage3_pin_refuses_old_container(self):
builder, inputs = self.existing_builder()
archive = lzma.compress(b'new stage layer')
(builder.cache / 'stage.tar.xz').write_bytes(archive)
inputs['stage3']['sha256'] = hashlib.sha256(archive).hexdigest()
self.assert_prepare_refused(builder, inputs, 'stage3')
def test_changed_snapshot_pin_refuses_existing_tree(self):
builder, inputs = self.existing_builder()
inputs['portage']['commit'] = 'new-snapshot'
self.assert_prepare_refused(builder, inputs, 'snapshot')
def test_matching_existing_builder_retains_compatible_provenance(self):
builder, inputs = self.existing_builder()
with patch.object(builder, 'fetch', return_value=inputs):
builder.prepare()
self.assertEqual(builder.manifest()['prepare'].get('stage3_sha256'), inputs['stage3']['sha256'])
self.assertEqual(builder.manifest()['prepare']['snapshot'], 'old-snapshot')
def test_status_invalidates_missing_package_inventory(self):
builder = self.builder()
with builder.locked():
builder.record('packages')
output = io.StringIO()
with contextlib.redirect_stdout(output):
builder.status()
self.assertFalse(json.loads(output.getvalue())['packages']['complete'])
def test_status_cli_runs_during_active_build_without_mutations(self):
builder = self.builder()
self.module.subprocess.run(['git', 'init', '--quiet', str(self.project)], check=True)
output = io.StringIO()
with builder.locked():
builder.record('packages')
manifest_before = (builder.work / 'manifest.json').read_bytes()
with patch.object(self.module, '__file__', str(self.project / 'hydra/build.py')):
with contextlib.redirect_stdout(output):
result = self.module.main(['status'])
self.assertEqual(result, 0, 'status must not acquire the exclusive build lock')
data = json.loads(output.getvalue())
self.assertTrue(data['running'])
self.assertFalse(data['packages']['complete'])
self.assertEqual((builder.work / 'manifest.json').read_bytes(), manifest_before)
def test_detached_emerge_refuses_preparation_before_fetch_or_writes(self):
builder, inputs = self.existing_builder()
fixture = self.project / 'podman-fixture.json'
data = json.loads(fixture.read_text())
data['active_job'] = True
fixture.write_text(json.dumps(data))
with patch.object(builder, 'fetch', return_value=inputs) as fetch:
with self.assertRaisesRegex(self.module.BuildError, 'package job'):
builder.prepare()
fetch.assert_not_called()
self.assertFalse((self.project / 'mutations.jsonl').exists())
def test_contained_job_lock_refuses_preparation_before_fetch(self):
builder, inputs = self.existing_builder()
fixture = self.project / 'podman-fixture.json'
data = json.loads(fixture.read_text())
data['job_lock'] = True
fixture.write_text(json.dumps(data))
with patch.object(builder, 'fetch', return_value=inputs) as fetch:
with self.assertRaisesRegex(self.module.BuildError, 'package job'):
builder.prepare()
fetch.assert_not_called()
def test_status_detects_detached_emerge_without_outer_lock(self):
builder, inputs = self.existing_builder()
fixture = self.project / 'podman-fixture.json'
data = json.loads(fixture.read_text())
data['active_job'] = True
fixture.write_text(json.dumps(data))
output = io.StringIO()
with contextlib.redirect_stdout(output):
builder.status()
status = json.loads(output.getvalue())
self.assertTrue(status['running'])
self.assertTrue(status['package_job_running'])
self.assertFalse(status['mutation_lock_held'])
def test_package_job_records_failure_and_log_without_host_collection(self):
self.assertTrue(hasattr(self.module, 'package_job_script'), 'durable job runner is required')
script = self.module.package_job_script(['/bin/sh', '-c', 'echo diagnostic; exit 7'],
self.project / 'job', self.project / 'pkgdb', 'inputs-one')
result = self.module.subprocess.run(['python3', '-c', script])
self.assertEqual(result.returncode, 7)
status = json.loads((self.project / 'job/packages.status.json').read_text())
self.assertFalse(status['running'])
self.assertEqual(status['exit_code'], 7)
self.assertIn('diagnostic', (self.project / 'job/packages.log').read_text())
self.assertFalse((self.project / 'job/packages.json').exists())
def test_package_job_persists_inventory_after_success_without_host_collection(self):
self.assertTrue(hasattr(self.module, 'package_job_script'), 'durable job runner is required')
package = self.project / 'pkgdb/category/example-1.0'
package.mkdir(parents=True)
(package / 'USE').write_text('wayland opengl')
(package / 'BUILD_TIME').write_text('123')
script = self.module.package_job_script(['/bin/sh', '-c', 'exit 0'], self.project / 'job',
self.project / 'pkgdb', 'inputs-one')
result = self.module.subprocess.run(['python3', '-c', script])
self.assertEqual(result.returncode, 0)
self.assertEqual(json.loads((self.project / 'job/packages.json').read_text()),
[{'package': 'category/example-1.0', 'use': ['wayland', 'opengl'], 'build_time': '123'}])
self.assertEqual(json.loads((self.project / 'job/packages.status.json').read_text())['exit_code'], 0)
def test_matching_stopped_builder_checks_provenance_before_start(self):
builder, inputs = self.existing_builder()
fixture = self.project / 'podman-fixture.json'
data = json.loads(fixture.read_text())
data['stopped'] = True
fixture.write_text(json.dumps(data))
with patch.object(builder, 'fetch', return_value=inputs):
try:
builder.prepare()
except self.module.BuildError as error:
self.fail(f'Compatible stopped builder must resume: {error}')
self.assertTrue(builder.manifest()['prepare']['complete'])
def test_unfinished_snapshot_is_not_reported_as_prepared(self):
builder, inputs = self.existing_builder(snapshot=None)
fixture = self.project / 'podman-fixture.json'
data = json.loads(fixture.read_text())
data['snapshot_pending'] = 'old-snapshot'
fixture.write_text(json.dumps(data))
builder.record('prepare')
self.assertFalse(builder.complete('prepare'))
if __name__ == '__main__':
unittest.main()