diff options
Diffstat (limited to 'llvm_tools')
| -rw-r--r-- | llvm_tools/README.md | 44 | ||||
| -rwxr-xr-x | llvm_tools/bisect_clang_crashes.py | 139 | ||||
| -rwxr-xr-x | llvm_tools/bisect_clang_crashes_unittest.py | 92 | ||||
| -rwxr-xr-x | llvm_tools/cherrypick_cl.py | 34 | ||||
| -rwxr-xr-x | llvm_tools/fetch_cros_sdk_rolls.py | 112 | ||||
| -rwxr-xr-x | llvm_tools/get_llvm_hash.py | 46 | ||||
| -rwxr-xr-x | llvm_tools/get_llvm_hash_unittest.py | 5 | ||||
| -rwxr-xr-x | llvm_tools/git_llvm_rev.py | 14 | ||||
| -rwxr-xr-x | llvm_tools/git_llvm_rev_test.py | 30 | ||||
| -rwxr-xr-x | llvm_tools/llvm_bisection.py | 278 | ||||
| -rwxr-xr-x | llvm_tools/llvm_bisection_unittest.py | 624 | ||||
| -rw-r--r-- | llvm_tools/llvm_project.py | 11 | ||||
| -rwxr-xr-x | llvm_tools/nightly_revert_checker.py | 9 | ||||
| -rwxr-xr-x | llvm_tools/upload_lexan_crashes_to_forcey.py | 258 | ||||
| -rwxr-xr-x | llvm_tools/upload_lexan_crashes_to_forcey_test.py | 122 |
15 files changed, 1208 insertions, 610 deletions
diff --git a/llvm_tools/README.md b/llvm_tools/README.md index 1c1862d8..783ec22d 100644 --- a/llvm_tools/README.md +++ b/llvm_tools/README.md @@ -278,6 +278,7 @@ $ ./auto_llvm_bisection.py --start_rev 369410 --end_rev 369420 \ --last_tested /abs/path/to/last_tested_file.json \ --extra_change_lists 513590 1394249 \ --options latest-toolchain nochromesdk \ + --chroot_path /path/to/chromeos/chroot \ --builder eve-release-tryjob ``` @@ -488,7 +489,7 @@ these synthesized numbers and git SHAs. Usage should be straightforward: 6f635f90929da9545dd696071a829a1a42f84b30 ~> ./git_llvm_rev.py --llvm_dir llvm-project-copy/ --sha 6f635f90929da9545dd696071a829a1a42f84b30 r380000 -~> ./git_llvm_rev.py --llvm_dir llvm-project-copy/ --sha origin/master +~> ./git_llvm_rev.py --llvm_dir llvm-project-copy/ --sha origin/some-branch r387778 ``` @@ -550,3 +551,44 @@ PYTHONPATH=../ ./nightly_revert_checker.py \ --llvm_dir llvm-project-copy \ --chromeos_dir ../../../../ ``` + +### `bisect_clang_crashes.py` + +This script downloads clang crash diagnoses from +gs://chromeos-toolchain-artifacts/clang-crash-diagnoses and sends them to 4c for +bisection. + +Usage example: + +``` +$ ./bisect_clang_crashes.py --4c 4c-cli --state_file ./output/state.json +``` + +The above command downloads the artifacts of clang crash diagnoses and send them +to 4c server for bisection. The summary of submitted jobs will be saved in +output/state.json under the current path. The output directory will be created +automatically if it does not exist yet. To get more information of the submitted +jobs, please refer to go/4c-cli. + +### `upload_lexan_crashes_to_forcey.py` + +This script downloads clang crash diagnoses from Lexan's bucket and sends them +to 4c for bisection. + +Usage example: + +``` +$ ./upload_lexan_crashes_to_forcey.py --4c 4c-cli \ + --state_file ./output/state.json +``` + +The above command downloads the artifacts of clang crash diagnoses and send them +to 4c server for bisection. The summary of submitted jobs will be saved in +output/state.json under the current path. The output directory will be created +automatically if it does not exist yet. To get more information of the submitted +jobs, please refer to go/4c-cli. + +Note that it's recommended to 'seed' the state file with a most recent upload +date. This can be done by running this tool *once* with a `--last_date` flag. +This flag has the script override whatever's in the state file (if anything) and +start submitting all crashes uploaded starting at the given day. diff --git a/llvm_tools/bisect_clang_crashes.py b/llvm_tools/bisect_clang_crashes.py new file mode 100755 index 00000000..e8ee2ab6 --- /dev/null +++ b/llvm_tools/bisect_clang_crashes.py @@ -0,0 +1,139 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- +# Copyright 2020 The Chromium OS Authors. All rights reserved. +# Use of this source code is governed by a BSD-style license that can be +# found in the LICENSE file. + +"""Fetches and submits the artifacts from Chrome OS toolchain's crash bucket. +""" + +# pylint: disable=cros-logging-import + +import argparse +import glob +import json +import logging +import os +import os.path +import shutil +import subprocess +import sys + +import chroot + + +def get_artifacts(pattern): + results = subprocess.check_output(['gsutil.py', 'ls', pattern], + stderr=subprocess.STDOUT, + encoding='utf-8') + return sorted(l.strip() for l in results.splitlines()) + + +def get_crash_reproducers(working_dir): + results = [] + for src in [ + f for f in glob.glob('%s/*.c*' % working_dir) + if f.split('.')[-1] in ['c', 'cc', 'cpp'] + ]: + script = '.'.join(src.split('.')[:-1]) + '.sh' + if not os.path.exists(script): + logging.warning('could not find the matching script of %s', src) + else: + results.append((src, script)) + return results + + +def submit_crash_to_forcey(forcey: str, temporary_directory: str, + buildbucket_id: str, url: str) -> None: + dest_dir = os.path.join(temporary_directory, buildbucket_id) + dest_file = os.path.join(dest_dir, os.path.basename(url)) + logging.info('Downloading and submitting %r...', url) + subprocess.check_output(['gsutil.py', 'cp', url, dest_file], + stderr=subprocess.STDOUT) + subprocess.check_output(['tar', '-xJf', dest_file], cwd=dest_dir) + for src, script in get_crash_reproducers(dest_dir): + subprocess.check_output([ + forcey, 'reduce', '-wait=false', '-note', + '%s:%s' % (url, src), '-sh_file', script, '-src_file', src + ]) + + +def main(argv): + chroot.VerifyOutsideChroot() + logging.basicConfig( + format='%(asctime)s: %(levelname)s: %(filename)s:%(lineno)d: %(message)s', + level=logging.INFO, + ) + cur_dir = os.path.dirname(os.path.abspath(__file__)) + parser = argparse.ArgumentParser(description=__doc__) + parser.add_argument( + '--4c', dest='forcey', required=True, help='Path to a 4c client binary') + parser.add_argument( + '--state_file', + default=os.path.join(cur_dir, 'chromeos-state.json'), + help='The path to the state file.') + parser.add_argument( + '--nocleanup', + action='store_false', + dest='cleanup', + help='Keep temporary files created after the script finishes.') + opts = parser.parse_args(argv) + + state_file = os.path.abspath(opts.state_file) + os.makedirs(os.path.dirname(state_file), exist_ok=True) + temporary_directory = '/tmp/bisect_clang_crashes' + os.makedirs(temporary_directory, exist_ok=True) + urls = get_artifacts('gs://chromeos-toolchain-artifacts/clang-crash-diagnoses' + '/**/*clang_crash_diagnoses.tar.xz') + logging.info('%d crash URLs found', len(urls)) + + visited = {} + if os.path.exists(state_file): + buildbucket_ids = {url.split('/')[-2] for url in urls} + with open(state_file, encoding='utf-8') as f: + data = json.load(f) + visited = {k: v for k, v in data.items() if k in buildbucket_ids} + logging.info('Successfully loaded %d previously-submitted crashes', + len(visited)) + + try: + for url in urls: + splits = url.split('/') + buildbucket_id = splits[-2] + # Skip the builds that has been processed + if buildbucket_id in visited: + continue + submit_crash_to_forcey( + forcey=opts.forcey, + temporary_directory=temporary_directory, + buildbucket_id=buildbucket_id, + url=url, + ) + visited[buildbucket_id] = url + + exception_in_flight = False + except: + exception_in_flight = True + raise + finally: + if exception_in_flight: + # This is best-effort. If the machine powers off or similar, we'll just + # resubmit the same crashes, which is suboptimal, but otherwise + # acceptable. + logging.error('Something went wrong; attempting to save our work...') + else: + logging.info('Persisting state...') + + tmp_state_file = state_file + '.tmp' + with open(tmp_state_file, 'w', encoding='utf-8') as f: + json.dump(visited, f, indent=2) + os.rename(tmp_state_file, state_file) + + logging.info('State successfully persisted') + + if opts.cleanup: + shutil.rmtree(temporary_directory) + + +if __name__ == '__main__': + sys.exit(main(sys.argv[1:])) diff --git a/llvm_tools/bisect_clang_crashes_unittest.py b/llvm_tools/bisect_clang_crashes_unittest.py new file mode 100755 index 00000000..c9143450 --- /dev/null +++ b/llvm_tools/bisect_clang_crashes_unittest.py @@ -0,0 +1,92 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- +# Copyright 2020 The Chromium OS Authors. All rights reserved. +# Use of this source code is governed by a BSD-style license that can be +# found in the LICENSE file. + +"""Tests for bisect_clang_crashes.""" + +# pylint: disable=cros-logging-import +import glob +import logging +import os.path +import subprocess +import unittest +import unittest.mock as mock + +import bisect_clang_crashes + + +class Test(unittest.TestCase): + """Tests for bisect_clang_crashes.""" + + class _SilencingFilter(object): + """Silences all log messages. + + Also collects info about log messages that would've been emitted. + """ + + def __init__(self): + self.messages = [] + + def filter(self, record): + self.messages.append(record.getMessage()) + return 0 + + @mock.patch.object(subprocess, 'check_output') + def test_get_artifacts(self, mock_gsutil_ls): + pattern = 'gs://chromeos-toolchain-artifacts/clang-crash-diagnoses/' \ + '**/*clang_crash_diagnoses.tar.xz' + mock_gsutil_ls.return_value = 'artifact1\nartifact2\nartifact3' + results = bisect_clang_crashes.get_artifacts(pattern) + self.assertEqual(results, ['artifact1', 'artifact2', 'artifact3']) + mock_gsutil_ls.assert_called_once_with(['gsutil.py', 'ls', pattern], + stderr=subprocess.STDOUT, + encoding='utf-8') + + @mock.patch.object(os.path, 'exists') + @mock.patch.object(glob, 'glob') + def test_get_crash_reproducers_succeed(self, mock_file_search, + mock_file_check): + working_dir = 'SomeDirectory' + mock_file_search.return_value = ['a.c', 'b.cpp', 'c.cc'] + mock_file_check.side_effect = [True, True, True] + results = bisect_clang_crashes.get_crash_reproducers(working_dir) + mock_file_search.assert_called_once_with('%s/*.c*' % working_dir) + self.assertEqual(mock_file_check.call_count, 3) + self.assertEqual(mock_file_check.call_args_list[0], mock.call('a.sh')) + self.assertEqual(mock_file_check.call_args_list[1], mock.call('b.sh')) + self.assertEqual(mock_file_check.call_args_list[2], mock.call('c.sh')) + self.assertEqual(results, [('a.c', 'a.sh'), ('b.cpp', 'b.sh'), + ('c.cc', 'c.sh')]) + + @mock.patch.object(os.path, 'exists') + @mock.patch.object(glob, 'glob') + def test_get_crash_reproducers_no_matching_script(self, mock_file_search, + mock_file_check): + + def silence_logging(): + root = logging.getLogger() + filt = self._SilencingFilter() + root.addFilter(filt) + self.addCleanup(root.removeFilter, filt) + return filt + + log_filter = silence_logging() + working_dir = 'SomeDirectory' + mock_file_search.return_value = ['a.c', 'b.cpp', 'c.cc'] + mock_file_check.side_effect = [True, False, True] + results = bisect_clang_crashes.get_crash_reproducers(working_dir) + mock_file_search.assert_called_once_with('%s/*.c*' % working_dir) + self.assertEqual(mock_file_check.call_count, 3) + self.assertEqual(mock_file_check.call_args_list[0], mock.call('a.sh')) + self.assertEqual(mock_file_check.call_args_list[1], mock.call('b.sh')) + self.assertEqual(mock_file_check.call_args_list[2], mock.call('c.sh')) + self.assertEqual(results, [('a.c', 'a.sh'), ('c.cc', 'c.sh')]) + self.assertTrue( + any('could not find the matching script of b.cpp' in x + for x in log_filter.messages), log_filter.messages) + + +if __name__ == '__main__': + unittest.main() diff --git a/llvm_tools/cherrypick_cl.py b/llvm_tools/cherrypick_cl.py index 78330d19..9e306725 100755 --- a/llvm_tools/cherrypick_cl.py +++ b/llvm_tools/cherrypick_cl.py @@ -170,32 +170,34 @@ def main(): help='Automatically create a CL if specified') args = parser.parse_args() - llvm_config = git_llvm_rev.LLVMConfig( - remote='origin', dir=get_llvm_hash.GetAndUpdateLLVMProjectInLLVMTools()) - llvm_symlink = chroot.ConvertChrootPathsToAbsolutePaths( args.chroot_path, chroot.GetChrootEbuildPaths(args.chroot_path, ['sys-devel/llvm']))[0] + llvm_symlink_dir = os.path.dirname(llvm_symlink) + + git_status = subprocess.check_output(['git', 'status', '-s'], + cwd=llvm_symlink_dir, + encoding='utf-8') + if git_status: + raise ValueError('Uncommited changes detected in %s' % + os.path.dirname(os.path.dirname(llvm_symlink_dir))) + start_sha = args.start_sha if start_sha == 'llvm': - start_sha = parse_ebuild_for_assignment( - os.path.dirname(llvm_symlink), 'LLVM_HASH') + start_sha = parse_ebuild_for_assignment(llvm_symlink_dir, 'LLVM_HASH') elif start_sha == 'llvm-next': - start_sha = parse_ebuild_for_assignment( - os.path.dirname(llvm_symlink), 'LLVM_NEXT_HASH') + start_sha = parse_ebuild_for_assignment(llvm_symlink_dir, 'LLVM_NEXT_HASH') logging.info('Base llvm hash == %s', start_sha) + llvm_config = git_llvm_rev.LLVMConfig( + remote='origin', dir=get_llvm_hash.GetAndUpdateLLVMProjectInLLVMTools()) + start_sha = resolve_llvm_ref(llvm_config.dir, start_sha) start_rev = git_llvm_rev.translate_sha_to_rev(llvm_config, start_sha) if args.create_cl: branch = 'cherry-pick' - symlink = os.path.dirname( - chroot.GetChrootEbuildPaths(args.chroot_path, ['sys-devel/llvm'])[0]) - symlink = chroot.ConvertChrootPathsToAbsolutePaths(args.chroot_path, - [symlink])[0] - symlink_dir = os.path.dirname(symlink) - git.CreateBranch(symlink_dir, branch) + git.CreateBranch(llvm_symlink_dir, branch) symlinks_to_uprev = [] commit_messages = [ 'llvm: cherry-pick CLs from upstream\n', @@ -221,9 +223,9 @@ def main(): relative_patches_dir = 'cherry' if package == 'llvm' else '' patches_dir = os.path.join(symlink_dir, 'files', relative_patches_dir) logging.info('Cherrypicking %s (%s) into %s', rev, sha, package) + add_cherrypick(patches_json_path, patches_dir, relative_patches_dir, start_rev, llvm_config.dir, rev, sha, package) - if args.create_cl: symlinks_to_uprev.extend(symlinks) commit_messages.extend([ @@ -240,8 +242,8 @@ def main(): for symlink in symlinks_to_uprev: update_chromeos_llvm_hash.UprevEbuildSymlink(symlink) subprocess.check_output(['git', 'add', '--all'], cwd=symlink_dir) - git.UploadChanges(symlink_dir, branch, commit_messages) - git.DeleteBranch(symlink_dir, branch) + git.UploadChanges(llvm_symlink_dir, branch, commit_messages) + git.DeleteBranch(llvm_symlink_dir, branch) if __name__ == '__main__': diff --git a/llvm_tools/fetch_cros_sdk_rolls.py b/llvm_tools/fetch_cros_sdk_rolls.py new file mode 100755 index 00000000..42af678a --- /dev/null +++ b/llvm_tools/fetch_cros_sdk_rolls.py @@ -0,0 +1,112 @@ +#!/usr/bin/env python3 +# Copyright 2020 The Chromium OS Authors. All rights reserved. +# Use of this source code is governed by a BSD-style license that can be +# found in the LICENSE file. + +"""Gets info about completed chromiumos-sdk runs. + +Moreover, this script exists to get versions of published sdk tarballs in +gs://chromiumos-sdk/. The hope is that it'll help answer the question "when did +the toolchain ebuild ${x} go live?" +""" + +# pylint: disable=cros-logging-import + +import argparse +import json +import logging +import os +import shutil +import subprocess +import sys +import tempfile +from typing import Dict, List +from pathlib import Path + + +def fetch_all_sdk_manifest_paths() -> List[str]: + """Fetches all paths of SDK manifests; newer = later in the return value.""" + results = subprocess.run( + ['gsutil', 'ls', 'gs://chromiumos-sdk/cros-sdk-20??.*.Manifest'], + check=True, + stdout=subprocess.PIPE, + encoding='utf-8', + ).stdout + # These are named so that sorted order == newest last. + return sorted(x.strip() for x in results.splitlines()) + + +def fetch_manifests_into(into_dir: Path, manifests: List[str]): + # Wrap this in a `try` block because gsutil likes to print to stdout *and* + # stderr even on success, so we silence them & only print on failure. + try: + subprocess.run( + [ + 'gsutil', + '-m', + 'cp', + '-I', + str(into_dir), + ], + check=True, + input='\n'.join(manifests), + stdout=subprocess.PIPE, + stderr=subprocess.STDOUT, + encoding='utf-8', + ) + except subprocess.CalledProcessError as e: + logging.exception('gsutil failed; output:\n%s', e.stdout) + + +def load_manifest_versions(manifest: Path) -> Dict[str, str]: + with manifest.open(encoding='utf-8') as f: + raw_versions = json.load(f) + + # We get a dict of list of lists of versions and some other metadata, e.g. + # {"foo/bar": [["1.2.3", {}]]} + # Trim out the metadata. + return {k: v[0][0] for k, v in raw_versions['packages'].items()} + + +def main(): + parser = argparse.ArgumentParser( + description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter) + parser.add_argument( + '-d', '--debug', action='store_true', help='Emit debugging output') + parser.add_argument( + '-n', + '--number', + type=int, + default=20, + help='Number of recent manifests to fetch info about. 0 means unlimited.') + args = parser.parse_args() + + is_debug = args.debug + logging.basicConfig(level=logging.DEBUG if is_debug else logging.INFO) + + logging.debug('Fetching SDK manifests') + manifest_paths = fetch_all_sdk_manifest_paths() + logging.debug('%d SDK manifests fetched', len(manifest_paths)) + + number = args.number + if number: + manifest_paths = manifest_paths[-number:] + + tempdir = Path(tempfile.mkdtemp(prefix='cros-sdk-rolls')) + try: + logging.debug('Working in tempdir %r', tempdir) + fetch_manifests_into(tempdir, manifest_paths) + + for path in manifest_paths: + basename = os.path.basename(path) + versions = load_manifest_versions(tempdir.joinpath(basename)) + print(f'{basename}: {versions["sys-devel/llvm"]}') + finally: + if is_debug: + logging.debug('Keeping around tempdir %r to aid debugging', tempdir) + else: + shutil.rmtree(tempdir) + + +if __name__ == '__main__': + sys.exit(main()) diff --git a/llvm_tools/get_llvm_hash.py b/llvm_tools/get_llvm_hash.py index a5b5429e..4c479962 100755 --- a/llvm_tools/get_llvm_hash.py +++ b/llvm_tools/get_llvm_hash.py @@ -17,7 +17,8 @@ import tempfile from contextlib import contextmanager import git_llvm_rev -from subprocess_helpers import CheckCommand, check_output +from subprocess_helpers import CheckCommand +from subprocess_helpers import check_output _LLVM_GIT_URL = ('https://chromium.googlesource.com/external/github.com/llvm' '/llvm-project') @@ -39,7 +40,7 @@ def GetVersionFrom(src_dir, git_hash): version = git_llvm_rev.translate_sha_to_rev( git_llvm_rev.LLVMConfig(remote='origin', dir=src_dir), git_hash) # Note: branches aren't supported - assert version.branch == 'master', version.branch + assert version.branch == git_llvm_rev.MAIN_BRANCH, version.branch return version.number @@ -59,7 +60,7 @@ def GetGitHashFrom(src_dir, version): return git_llvm_rev.translate_rev_to_sha( git_llvm_rev.LLVMConfig(remote='origin', dir=src_dir), - git_llvm_rev.Rev(branch='master', number=version)) + git_llvm_rev.Rev(branch=git_llvm_rev.MAIN_BRANCH, number=version)) @contextmanager @@ -85,13 +86,10 @@ def CreateTempLLVMRepo(temp_dir): """ abs_path_to_llvm_project_dir = GetAndUpdateLLVMProjectInLLVMTools() - - add_worktree_cmd = [ + CheckCommand([ 'git', '-C', abs_path_to_llvm_project_dir, 'worktree', 'add', '--detach', - temp_dir, 'master' - ] - - CheckCommand(add_worktree_cmd) + temp_dir, git_llvm_rev.MAIN_BRANCH + ]) try: yield temp_dir @@ -117,7 +115,7 @@ def GetAndUpdateLLVMProjectInLLVMTools(): Raises: ValueError: LLVM repo (in 'llvm-project-copy' dir.) has changes or failed to - checkout to master or failed to fetch from chromium mirror of LLVM. + checkout to main or failed to fetch from chromium mirror of LLVM. """ abs_path_to_llvm_tools_dir = os.path.dirname(os.path.abspath(__file__)) @@ -127,7 +125,8 @@ def GetAndUpdateLLVMProjectInLLVMTools(): if not os.path.isdir(abs_path_to_llvm_project_dir): print( - 'Checking out LLVM from scratch. This could take a while...', + 'Checking out LLVM from scratch. This could take a while...\n' + '(This should only need to be done once, though.)', file=sys.stderr) os.mkdir(abs_path_to_llvm_project_dir) @@ -143,15 +142,11 @@ def GetAndUpdateLLVMProjectInLLVMTools(): raise ValueError('LLVM repo in %s has changes, please remove.' % abs_path_to_llvm_project_dir) - checkout_to_master_cmd = [ - 'git', '-C', abs_path_to_llvm_project_dir, 'checkout', 'master' - ] - - CheckCommand(checkout_to_master_cmd) - - update_master_cmd = ['git', '-C', abs_path_to_llvm_project_dir, 'pull'] - - CheckCommand(update_master_cmd) + CheckCommand([ + 'git', '-C', abs_path_to_llvm_project_dir, 'checkout', + git_llvm_rev.MAIN_BRANCH + ]) + CheckCommand(['git', '-C', abs_path_to_llvm_project_dir, 'pull']) return abs_path_to_llvm_project_dir @@ -298,14 +293,9 @@ class LLVMHash(object): def GetTopOfTrunkGitHash(self): """Gets the latest git hash from top of trunk of LLVM.""" - path_to_master_branch = 'refs/heads/master' - - llvm_tot_git_hash_cmd = [ - 'git', 'ls-remote', _LLVM_GIT_URL, path_to_master_branch - ] - - llvm_tot_git_hash = check_output(llvm_tot_git_hash_cmd) - + path_to_main_branch = 'refs/heads/master' + llvm_tot_git_hash = check_output( + ['git', 'ls-remote', _LLVM_GIT_URL, path_to_main_branch]) return llvm_tot_git_hash.rstrip().split()[0] diff --git a/llvm_tools/get_llvm_hash_unittest.py b/llvm_tools/get_llvm_hash_unittest.py index c828f433..2e56aed5 100755 --- a/llvm_tools/get_llvm_hash_unittest.py +++ b/llvm_tools/get_llvm_hash_unittest.py @@ -8,12 +8,11 @@ from __future__ import print_function -import get_llvm_hash import subprocess import unittest import unittest.mock as mock -import test_helpers +import get_llvm_hash from get_llvm_hash import LLVMHash # We grab protected stuff from get_llvm_hash. That's OK. @@ -87,7 +86,7 @@ class TestGetLLVMHash(unittest.TestCase): @mock.patch.object(subprocess, 'check_output') def testSuccessfullyGetGitHashFromToTOfLLVM(self, mock_check_output): - mock_check_output.return_value = 'a123testhash1 path/to/master\n' + mock_check_output.return_value = 'a123testhash1 path/to/main\n' self.assertEqual(LLVMHash().GetTopOfTrunkGitHash(), 'a123testhash1') mock_check_output.assert_called_once() diff --git a/llvm_tools/git_llvm_rev.py b/llvm_tools/git_llvm_rev.py index 8eefcdce..07209f1e 100755 --- a/llvm_tools/git_llvm_rev.py +++ b/llvm_tools/git_llvm_rev.py @@ -18,6 +18,8 @@ import subprocess import sys import typing as t +MAIN_BRANCH = 'master' + # Note that after base_llvm_sha, we reach The Wild West(TM) of commits. # So reasonable input that could break us includes: # @@ -52,9 +54,9 @@ class Rev(t.NamedTuple('Rev', (('branch', str), ('number', int)))): # pairs. # # We support r${commits_since_base_commit} as shorthand for - # (master, r${commits_since_base_commit}). + # (main, r${commits_since_base_commit}). if rev.startswith('r'): - branch_name = 'master' + branch_name = MAIN_BRANCH rev_string = rev[1:] else: match = re.match(r'\((.+), r(\d+)\)', rev) @@ -67,7 +69,7 @@ class Rev(t.NamedTuple('Rev', (('branch', str), ('number', int)))): def __str__(self) -> str: branch_name, number = self - if branch_name == 'master': + if branch_name == MAIN_BRANCH: return 'r%d' % number return '(%s, r%d)' % (branch_name, number) @@ -141,7 +143,7 @@ def translate_sha_to_rev(llvm_config: LLVMConfig, sha_or_ref: str) -> Rev: cwd=llvm_config.dir, ) count = int(result.strip()) - return Rev(branch='master', number=count + base_llvm_revision) + return Rev(branch=MAIN_BRANCH, number=count + base_llvm_revision) # Otherwise, either: # - |merge_base| is |sha| (we have a guaranteed llvm-svn number on |sha|) @@ -150,7 +152,7 @@ def translate_sha_to_rev(llvm_config: LLVMConfig, sha_or_ref: str) -> Rev: merge_base_number = translate_prebase_sha_to_rev_number( llvm_config, merge_base) if merge_base == sha: - return Rev(branch='master', number=merge_base_number) + return Rev(branch=MAIN_BRANCH, number=merge_base_number) distance_from_base = check_output( [ @@ -270,7 +272,7 @@ def translate_rev_to_sha(llvm_config: LLVMConfig, rev: Rev) -> str: """ branch, number = rev - if branch == 'master': + if branch == MAIN_BRANCH: if number < base_llvm_revision: return translate_prebase_rev_to_sha(llvm_config, rev) base_sha = base_llvm_sha diff --git a/llvm_tools/git_llvm_rev_test.py b/llvm_tools/git_llvm_rev_test.py index 1e38f589..74280c5d 100755 --- a/llvm_tools/git_llvm_rev_test.py +++ b/llvm_tools/git_llvm_rev_test.py @@ -6,12 +6,11 @@ """Tests for git_llvm_rev.""" -from __future__ import print_function - import unittest import git_llvm_rev import llvm_project +from git_llvm_rev import MAIN_BRANCH def get_llvm_config() -> git_llvm_rev.LLVMConfig: @@ -32,17 +31,17 @@ class Test(unittest.TestCase): def test_sha_to_rev_on_base_sha_works(self) -> None: sha = self.rev_to_sha_with_round_trip( git_llvm_rev.Rev( - branch='master', number=git_llvm_rev.base_llvm_revision)) + branch=MAIN_BRANCH, number=git_llvm_rev.base_llvm_revision)) self.assertEqual(sha, git_llvm_rev.base_llvm_sha) def test_sha_to_rev_prior_to_base_rev_works(self) -> None: sha = self.rev_to_sha_with_round_trip( - git_llvm_rev.Rev(branch='master', number=375000)) + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=375000)) self.assertEqual(sha, '2f6da767f13b8fd81f840c211d405fea32ac9db7') def test_sha_to_rev_after_base_rev_works(self) -> None: sha = self.rev_to_sha_with_round_trip( - git_llvm_rev.Rev(branch='master', number=375506)) + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=375506)) self.assertEqual(sha, '3bf7fddeb05655d9baed4cc69e13535c677ed1dd') def test_llvm_svn_parsing_runs_ignore_reverts(self) -> None: @@ -51,18 +50,19 @@ class Test(unittest.TestCase): # Commit which performed the revert sha = self.rev_to_sha_with_round_trip( - git_llvm_rev.Rev(branch='master', number=374895)) + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=374895)) self.assertEqual(sha, '1731fc88d1fa1fa55edd056db73a339b415dd5d6') # Commit that was reverted sha = self.rev_to_sha_with_round_trip( - git_llvm_rev.Rev(branch='master', number=374841)) + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=374841)) self.assertEqual(sha, '2a1386c81de504b5bda44fbecf3f7b4cdfd748fc') def test_imaginary_revs_raise(self) -> None: with self.assertRaises(ValueError) as r: git_llvm_rev.translate_rev_to_sha( - get_llvm_config(), git_llvm_rev.Rev(branch='master', number=9999999)) + get_llvm_config(), + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=9999999)) self.assertIn('Try updating your tree?', str(r.exception)) @@ -71,15 +71,15 @@ class Test(unittest.TestCase): # properties about it. merge_sha_rev_number = 4496 + git_llvm_rev.base_llvm_revision sha = self.rev_to_sha_with_round_trip( - git_llvm_rev.Rev(branch='master', number=merge_sha_rev_number)) + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=merge_sha_rev_number)) self.assertEqual(sha, '0f0d0ed1c78f1a80139a1f2133fad5284691a121') sha = self.rev_to_sha_with_round_trip( - git_llvm_rev.Rev(branch='master', number=merge_sha_rev_number - 1)) + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=merge_sha_rev_number - 1)) self.assertEqual(sha, '6f635f90929da9545dd696071a829a1a42f84b30') sha = self.rev_to_sha_with_round_trip( - git_llvm_rev.Rev(branch='master', number=merge_sha_rev_number + 1)) + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=merge_sha_rev_number + 1)) self.assertEqual(sha, '199700a5cfeedf227619f966aa3125cef18bc958') # NOTE: The below tests have _zz_ in their name as an optimization. Iterating @@ -101,11 +101,11 @@ class Test(unittest.TestCase): backing_sha = 'c89a3d78f43d81b9cff7b9248772ddf14d21b749' sha = self.rev_to_sha_with_round_trip( - git_llvm_rev.Rev(branch='master', number=rev_number)) + git_llvm_rev.Rev(branch=MAIN_BRANCH, number=rev_number)) self.assertEqual(sha, backing_sha) - # Note that this won't round-trip: since this commit is on the master - # branch, we'll pick master for this. That's fine + # Note that this won't round-trip: since this commit is on the main + # branch, we'll pick main for this. That's fine. sha = git_llvm_rev.translate_rev_to_sha( get_llvm_config(), git_llvm_rev.Rev(branch='release/9.x', number=rev_number)) @@ -113,7 +113,7 @@ class Test(unittest.TestCase): def test_zz_branch_revs_work_after_merge_points(self) -> None: # Picking the commit on the 9.x branch after the merge-base for that + - # master. Note that this is where llvm-svn numbers should diverge from + # main. Note that this is where llvm-svn numbers should diverge from # ours, and are therefore untrustworthy. The commit for this *does* have a # different `llvm-svn:` string than we should have. sha = self.rev_to_sha_with_round_trip( diff --git a/llvm_tools/llvm_bisection.py b/llvm_tools/llvm_bisection.py index 2772ca48..c8d694cd 100755 --- a/llvm_tools/llvm_bisection.py +++ b/llvm_tools/llvm_bisection.py @@ -13,10 +13,12 @@ import enum import errno import json import os +import subprocess import sys import chroot import get_llvm_hash +import git_llvm_rev import modify_a_tryjob import update_tryjob_status @@ -28,11 +30,6 @@ class BisectionExitStatus(enum.Enum): BISECTION_COMPLETE = 126 -def is_file_and_json(json_file): - """Validates that the file exists and is a JSON file.""" - return os.path.isfile(json_file) and json_file.endswith('.json') - - def GetCommandLineArgs(): """Parses the command line for the command line arguments.""" @@ -121,30 +118,27 @@ def GetCommandLineArgs(): help='display contents of a command to the terminal ' '(default: %(default)s)') + # Add argument for whether to display command contents to `stdout`. + parser.add_argument( + '--nocleanup', + action='store_false', + dest='cleanup', + help='Abandon CLs created for bisectoin') + args_output = parser.parse_args() assert args_output.start_rev < args_output.end_rev, ( - 'Start revision %d is >= end revision %d' % (args_output.start_rev, - args_output.end_rev)) + 'Start revision %d is >= end revision %d' % + (args_output.start_rev, args_output.end_rev)) if args_output.last_tested and not args_output.last_tested.endswith('.json'): - raise ValueError( - 'Filed provided %s does not end in ".json"' % args_output.last_tested) + raise ValueError('Filed provided %s does not end in ".json"' % + args_output.last_tested) return args_output -def _ValidateStartAndEndAgainstJSONStartAndEnd(start, end, json_start, - json_end): - """Valides that the command line arguments are the same as the JSON.""" - - if start != json_start or end != json_end: - raise ValueError('The start %d or the end %d version provided is ' - 'different than "start" %d or "end" %d in the .JSON ' - 'file' % (start, end, json_start, json_end)) - - -def GetStartAndEndRevision(start, end, tryjobs): +def GetRemainingRange(start, end, tryjobs): """Gets the start and end intervals in 'json_file'. Args: @@ -230,145 +224,56 @@ def GetStartAndEndRevision(start, end, tryjobs): return good_rev, bad_rev, pending_revisions, skip_revisions -def GetRevisionsBetweenBisection(start, end, parallel, src_path, - pending_revisions, skip_revisions): - """Gets the revisions between 'start' and 'end'. - - Sometimes, the LLVM source tree's revisions do not increment by 1 (there is - a jump), so need to construct a list of all revisions that are NOT missing - between 'start' and 'end'. Then, the step amount (i.e. length of the list - divided by ('parallel' + 1)) will be used for indexing into the list. - - Args: - start: The start revision. - end: The end revision. - parallel: The number of tryjobs to create between 'start' and 'end'. - src_path: The absolute path to the LLVM source tree to use. - pending_revisions: A set containing 'pending' revisions that are between - 'start' and 'end'. - skip_revisions: A set containing revisions between 'start' and 'end' that - are to be skipped. - - Returns: - A list of revisions between 'start' and 'end'. - """ - - valid_revisions = [] - - # Start at ('start' + 1) because 'start' is the good revision. - # - # FIXME: Searching for each revision from ('start' + 1) up to 'end' in the - # LLVM source tree is a quadratic algorithm. It's a good idea to optimize - # this. - for cur_revision in range(start + 1, end): - try: - if cur_revision not in pending_revisions and \ - cur_revision not in skip_revisions: - # Verify that the current revision exists by finding its corresponding - # git hash in the LLVM source tree. - get_llvm_hash.GetGitHashFrom(src_path, cur_revision) - valid_revisions.append(cur_revision) - except ValueError: - # Could not find the git hash for the current revision. - continue - - # ('parallel' + 1) so that the last revision in the list is not close to - # 'end' (have a bit more coverage). - index_step = len(valid_revisions) // (parallel + 1) - - if not index_step: - index_step = 1 - - result = [valid_revisions[index] \ - for index in range(0, len(valid_revisions), index_step)] - - return result - - -def GetRevisionsListAndHashList(start, end, parallel, src_path, - pending_revisions, skip_revisions): +def GetCommitsBetween(start, end, parallel, src_path, pending_revisions, + skip_revisions): """Determines the revisions between start and end.""" - new_llvm = get_llvm_hash.LLVMHash() + with get_llvm_hash.LLVMHash().CreateTempDirectory() as temp_dir: + # We have guaranteed contiguous revision numbers after this, + # and that guarnatee simplifies things considerably, so we don't + # support anything before it. + assert start >= git_llvm_rev.base_llvm_revision, f'{start} was too long ago' - with new_llvm.CreateTempDirectory() as temp_dir: with get_llvm_hash.CreateTempLLVMRepo(temp_dir) as new_repo: if not src_path: src_path = new_repo - - # Get a list of revisions between start and end. - revisions = GetRevisionsBetweenBisection( - start, end, parallel, src_path, pending_revisions, skip_revisions) - + index_step = (end - (start + 1)) // (parallel + 1) + if not index_step: + index_step = 1 + revisions = [ + rev for rev in range(start + 1, end, index_step) + if rev not in pending_revisions and rev not in skip_revisions + ] git_hashes = [ get_llvm_hash.GetGitHashFrom(src_path, rev) for rev in revisions ] + return revisions, git_hashes - return revisions, git_hashes - - -def DieWithNoRevisionsError(start, end, skip_revisions, pending_revisions): - """Raises a ValueError exception with useful information.""" - - no_revisions_message = ('No revisions between start %d and end ' - '%d to create tryjobs' % (start, end)) - if pending_revisions: - no_revisions_message += '\nThe following tryjobs are pending:\n' \ - + '\n'.join(str(rev) for rev in pending_revisions) - - if skip_revisions: - no_revisions_message += '\nThe following tryjobs were skipped:\n' \ - + '\n'.join(str(rev) for rev in skip_revisions) - - raise ValueError(no_revisions_message) - - -def CheckForExistingTryjobsInRevisionsToLaunch(revisions, jobs): - """Checks if a revision in 'revisions' exists in 'jobs' list.""" - - for rev in revisions: - if update_tryjob_status.FindTryjobIndex(rev, jobs) is not None: - raise ValueError('Revision %d exists already in "jobs"' % rev) - - -def UpdateBisection(revisions, git_hashes, bisect_contents, last_tested, - update_packages, chroot_path, patch_metadata_file, - extra_change_lists, options, builder, verbose): +def Bisect(revisions, git_hashes, bisect_state, last_tested, update_packages, + chroot_path, patch_metadata_file, extra_change_lists, options, + builder, verbose): """Adds tryjobs and updates the status file with the new tryjobs.""" try: for svn_revision, git_hash in zip(revisions, git_hashes): - tryjob_dict = modify_a_tryjob.AddTryjob( - update_packages, git_hash, svn_revision, chroot_path, - patch_metadata_file, extra_change_lists, options, builder, verbose, - svn_revision) + tryjob_dict = modify_a_tryjob.AddTryjob(update_packages, git_hash, + svn_revision, chroot_path, + patch_metadata_file, + extra_change_lists, options, + builder, verbose, svn_revision) - bisect_contents['jobs'].append(tryjob_dict) + bisect_state['jobs'].append(tryjob_dict) finally: # Do not want to lose progress if there is an exception. if last_tested: new_file = '%s.new' % last_tested with open(new_file, 'w') as json_file: - json.dump(bisect_contents, json_file, indent=4, separators=(',', ': ')) + json.dump(bisect_state, json_file, indent=4, separators=(',', ': ')) os.rename(new_file, last_tested) -def _NoteCompletedBisection(last_tested, src_path, end): - """Prints that bisection is complete.""" - - print('Finished bisecting for %s' % last_tested) - - if src_path: - bad_llvm_hash = get_llvm_hash.GetGitHashFrom(src_path, end) - else: - bad_llvm_hash = get_llvm_hash.LLVMHash().GetLLVMHash(end) - - print( - 'The bad revision is %d and its commit hash is %s' % (end, bad_llvm_hash)) - - def LoadStatusFile(last_tested, start, end): """Loads the status file for bisection.""" @@ -383,37 +288,36 @@ def LoadStatusFile(last_tested, start, end): def main(args_output): - """Bisects LLVM based off of a .JSON file. + """Bisects LLVM commits. Raises: AssertionError: The script was run inside the chroot. """ chroot.VerifyOutsideChroot() - update_packages = [ 'sys-devel/llvm', 'sys-libs/compiler-rt', 'sys-libs/libcxx', 'sys-libs/libcxxabi', 'sys-libs/llvm-libunwind' ] - patch_metadata_file = 'PATCHES.json' - start = args_output.start_rev end = args_output.end_rev - bisect_contents = LoadStatusFile(args_output.last_tested, start, end) - - _ValidateStartAndEndAgainstJSONStartAndEnd( - start, end, bisect_contents['start'], bisect_contents['end']) + bisect_state = LoadStatusFile(args_output.last_tested, start, end) + if start != bisect_state['start'] or end != bisect_state['end']: + raise ValueError(f'The start {start} or the end {end} version provided is ' + f'different than "start" {bisect_state["start"]} or "end" ' + f'{bisect_state["end"]} in the .JSON file') # Pending and skipped revisions are between 'start_revision' and # 'end_revision'. start_revision, end_revision, pending_revisions, skip_revisions = \ - GetStartAndEndRevision(start, end, bisect_contents['jobs']) + GetRemainingRange(start, end, bisect_state['jobs']) - revisions, git_hashes = GetRevisionsListAndHashList( - start_revision, end_revision, args_output.parallel, args_output.src_path, - pending_revisions, skip_revisions) + revisions, git_hashes = GetCommitsBetween(start_revision, end_revision, + args_output.parallel, + args_output.src_path, + pending_revisions, skip_revisions) # No more revisions between 'start_revision' and 'end_revision', so # bisection is complete. @@ -421,39 +325,61 @@ def main(args_output): # This is determined by finding all valid revisions between 'start_revision' # and 'end_revision' and that are NOT in the 'pending' and 'skipped' set. if not revisions: - # Successfully completed bisection where there are 2 cases: - # 1) 'start_revision' and 'end_revision' are back-to-back (example: - # 'start_revision' is 369410 and 'end_revision' is 369411). - # - # 2) 'start_revision' and 'end_revision' are NOT back-to-back, so there must - # be tryjobs in between which are labeled as 'skip' for their 'status' - # value. - # - # In either case, there are no 'pending' jobs. - if not pending_revisions: - _NoteCompletedBisection(args_output.last_tested, args_output.src_path, - end_revision) + if pending_revisions: + # Some tryjobs are not finished which may change the actual bad + # commit/revision when those tryjobs are finished. + no_revisions_message = (f'No revisions between start {start_revision} ' + f'and end {end_revision} to create tryjobs\n') - if skip_revisions: - skip_revisions_message = ('\nThe following revisions were skipped:\n' + - '\n'.join(str(rev) for rev in skip_revisions)) + if pending_revisions: + no_revisions_message += ( + 'The following tryjobs are pending:\n' + + '\n'.join(str(rev) for rev in pending_revisions) + '\n') - print(skip_revisions_message) - - return BisectionExitStatus.BISECTION_COMPLETE.value - - # Some tryjobs are not finished which may change the actual bad - # commit/revision when those tryjobs are finished. - DieWithNoRevisionsError(start_revision, end_revision, skip_revisions, - pending_revisions) - - CheckForExistingTryjobsInRevisionsToLaunch(revisions, bisect_contents['jobs']) + if skip_revisions: + no_revisions_message += ('The following tryjobs were skipped:\n' + + '\n'.join(str(rev) for rev in skip_revisions) + + '\n') + + raise ValueError(no_revisions_message) + + print(f'Finished bisecting for {args_output.last_tested}') + if args_output.src_path: + bad_llvm_hash = get_llvm_hash.GetGitHashFrom(args_output.src_path, + end_revision) + else: + bad_llvm_hash = get_llvm_hash.LLVMHash().GetLLVMHash(end_revision) + print(f'The bad revision is {end_revision} and its commit hash is ' + f'{bad_llvm_hash}') + if skip_revisions: + skip_revisions_message = ('\nThe following revisions were skipped:\n' + + '\n'.join(str(rev) for rev in skip_revisions)) + print(skip_revisions_message) + + if args_output.cleanup: + # Abondon all the CLs created for bisection + gerrit = os.path.join(args_output.chroot_path, 'chromite/bin/gerrit') + for build in bisect_state['jobs']: + try: + subprocess.check_output([gerrit, 'abandon', build['cl']], + stderr=subprocess.STDOUT, + encoding='utf-8') + except subprocess.CalledProcessError as err: + # the CL may have been abandoned + if 'chromite.lib.gob_util.GOBError' not in err.output: + raise + + return BisectionExitStatus.BISECTION_COMPLETE.value - UpdateBisection(revisions, git_hashes, bisect_contents, - args_output.last_tested, update_packages, - args_output.chroot_path, patch_metadata_file, - args_output.extra_change_lists, args_output.options, - args_output.builder, args_output.verbose) + for rev in revisions: + if update_tryjob_status.FindTryjobIndex(rev, + bisect_state['jobs']) is not None: + raise ValueError(f'Revision {rev} exists already in "jobs"') + + Bisect(revisions, git_hashes, bisect_state, args_output.last_tested, + update_packages, args_output.chroot_path, patch_metadata_file, + args_output.extra_change_lists, args_output.options, + args_output.builder, args_output.verbose) if __name__ == '__main__': diff --git a/llvm_tools/llvm_bisection_unittest.py b/llvm_tools/llvm_bisection_unittest.py index e730293b..a40770a5 100755 --- a/llvm_tools/llvm_bisection_unittest.py +++ b/llvm_tools/llvm_bisection_unittest.py @@ -11,50 +11,57 @@ from __future__ import print_function import json +import os +import subprocess import unittest import unittest.mock as mock import chroot import get_llvm_hash +import git_llvm_rev import llvm_bisection import modify_a_tryjob import test_helpers -import update_tryjob_status class LLVMBisectionTest(unittest.TestCase): """Unittests for LLVM bisection.""" - def testStartAndEndDoNotMatchJsonStartAndEnd(self): + def testGetRemainingRangePassed(self): start = 100 end = 150 - json_start = 110 - json_end = 150 - - # Verify the exception is raised when the start and end revision for LLVM - # bisection do not match the .JSON's 'start' and 'end' values. - with self.assertRaises(ValueError) as err: - llvm_bisection._ValidateStartAndEndAgainstJSONStartAndEnd( - start, end, json_start, json_end) - - expected_error_message = ('The start %d or the end %d version provided is ' - 'different than "start" %d or "end" %d in the ' - '.JSON file' % (start, end, json_start, json_end)) - - self.assertEqual(str(err.exception), expected_error_message) - - def testStartAndEndMatchJsonStartAndEnd(self): - start = 100 - end = 150 + test_tryjobs = [{ + 'rev': 110, + 'status': 'good', + 'link': 'https://some_tryjob_1_url.com' + }, { + 'rev': 120, + 'status': 'good', + 'link': 'https://some_tryjob_2_url.com' + }, { + 'rev': 130, + 'status': 'pending', + 'link': 'https://some_tryjob_3_url.com' + }, { + 'rev': 135, + 'status': 'skip', + 'link': 'https://some_tryjob_4_url.com' + }, { + 'rev': 140, + 'status': 'bad', + 'link': 'https://some_tryjob_5_url.com' + }] - json_start = 100 - json_end = 150 + # Tuple consists of the new good revision, the new bad revision, a set of + # 'pending' revisions, and a set of 'skip' revisions. + expected_revisions_tuple = 120, 140, {130}, {135} - llvm_bisection._ValidateStartAndEndAgainstJSONStartAndEnd( - start, end, json_start, json_end) + self.assertEqual( + llvm_bisection.GetRemainingRange(start, end, test_tryjobs), + expected_revisions_tuple) - def testTryjobStatusIsMissing(self): + def testGetRemainingRangeFailedWithMissingStatus(self): start = 100 end = 150 @@ -72,18 +79,14 @@ class LLVMBisectionTest(unittest.TestCase): 'link': 'https://some_tryjob_3_url.com' }] - # Verify the exception is raised when a tryjob does not have a value for - # the 'status' key or the 'status' key is missing. with self.assertRaises(ValueError) as err: - llvm_bisection.GetStartAndEndRevision(start, end, test_tryjobs) + llvm_bisection.GetRemainingRange(start, end, test_tryjobs) - expected_error_message = ( - '"status" is missing or has no value, please ' - 'go to %s and update it' % test_tryjobs[1]['link']) + error_message = ('"status" is missing or has no value, please ' + 'go to %s and update it' % test_tryjobs[1]['link']) + self.assertEqual(str(err.exception), error_message) - self.assertEqual(str(err.exception), expected_error_message) - - def testGoodRevisionGreaterThanBadRevision(self): + def testGetRemainingRangeFailedWithInvalidRange(self): start = 100 end = 150 @@ -101,206 +104,68 @@ class LLVMBisectionTest(unittest.TestCase): 'link': 'https://some_tryjob_3_url.com' }] - # Verify the exception is raised when the new 'start' revision is greater - # than the new 'bad' revision for bisection (i.e. bisection is broken). with self.assertRaises(AssertionError) as err: - llvm_bisection.GetStartAndEndRevision(start, end, test_tryjobs) + llvm_bisection.GetRemainingRange(start, end, test_tryjobs) - expected_error_message = ( - 'Bisection is broken because %d (good) is >= ' - '%d (bad)' % (test_tryjobs[2]['rev'], test_tryjobs[0]['rev'])) + expected_error_message = ('Bisection is broken because %d (good) is >= ' + '%d (bad)' % + (test_tryjobs[2]['rev'], test_tryjobs[0]['rev'])) self.assertEqual(str(err.exception), expected_error_message) - def testSuccessfullyGetNewStartAndNewEndRevision(self): - start = 100 - end = 150 - - test_tryjobs = [{ - 'rev': 110, - 'status': 'good', - 'link': 'https://some_tryjob_1_url.com' - }, { - 'rev': 120, - 'status': 'good', - 'link': 'https://some_tryjob_2_url.com' - }, { - 'rev': 130, - 'status': 'pending', - 'link': 'https://some_tryjob_3_url.com' - }, { - 'rev': 135, - 'status': 'skip', - 'link': 'https://some_tryjob_4_url.com' - }, { - 'rev': 140, - 'status': 'bad', - 'link': 'https://some_tryjob_5_url.com' - }] - - # Tuple consists of the new good revision, the new bad revision, a set of - # 'pending' revisions, and a set of 'skip' revisions. - expected_revisions_tuple = 120, 140, {130}, {135} - - self.assertTupleEqual( - llvm_bisection.GetStartAndEndRevision(start, end, test_tryjobs), - expected_revisions_tuple) - @mock.patch.object(get_llvm_hash, 'GetGitHashFrom') - def testNoRevisionsBetweenStartAndEnd(self, mock_get_git_hash): - start = 100 - end = 110 - - test_pending_revisions = {107} - test_skip_revisions = {101, 102, 103, 104, 108, 109} - - # Simulate behavior of `GetGitHashFrom()` when the revision does not - # exist in the LLVM source tree. - def MockGetGitHashForRevisionRaiseException(_src_path, _revision): - raise ValueError('Revision does not exist') - - mock_get_git_hash.side_effect = MockGetGitHashForRevisionRaiseException - - parallel = 3 - - abs_path_to_src = '/abs/path/to/src' - - self.assertListEqual( - llvm_bisection.GetRevisionsBetweenBisection( - start, end, parallel, abs_path_to_src, test_pending_revisions, - test_skip_revisions), []) - - # Assume llvm_bisection module has imported GetGitHashFrom - @mock.patch.object(get_llvm_hash, 'GetGitHashFrom') - def testSuccessfullyRetrievedRevisionsBetweenStartAndEnd( - self, mock_get_git_hash): - - start = 100 - end = 110 - - test_pending_revisions = set() - test_skip_revisions = {101, 102, 103, 104, 106, 108, 109} - + def testGetCommitsBetweenPassed(self, mock_get_git_hash): + start = git_llvm_rev.base_llvm_revision + end = start + 10 + test_pending_revisions = {start + 7} + test_skip_revisions = { + start + 1, start + 2, start + 4, start + 8, start + 9 + } parallel = 3 - abs_path_to_src = '/abs/path/to/src' - # Valid revision that exist in the LLVM source tree between 'start' and - # 'end' and were not in the 'pending' set or 'skip' set. - expected_revisions_between_start_and_end = [105, 107] - - self.assertListEqual( - llvm_bisection.GetRevisionsBetweenBisection( - start, end, parallel, abs_path_to_src, test_pending_revisions, - test_skip_revisions), expected_revisions_between_start_and_end) - - self.assertEqual(mock_get_git_hash.call_count, 2) + revs = ['a123testhash3', 'a123testhash5'] + mock_get_git_hash.side_effect = revs - # Simulate behavior of `GetGitHashFrom()` when successfully retrieved - # a list git hashes for each revision in the revisions list. - # Assume llvm_bisection module has imported GetGitHashFrom - @mock.patch.object(get_llvm_hash, 'GetGitHashFrom') - # Simulate behavior of `GetRevisionsBetweenBisection()` when successfully - # retrieved a list of valid revisions between 'start' and 'end'. - @mock.patch.object(llvm_bisection, 'GetRevisionsBetweenBisection') - # Simulate behavior of `CreatTempLLVMRepo()` when successfully created a - # worktree when a source path was not provided. - @mock.patch.object(get_llvm_hash, 'CreateTempLLVMRepo') - def testSuccessfullyGetRevisionsListAndHashList( - self, mock_create_temp_llvm_repo, mock_get_revisions_between_bisection, - mock_get_git_hash): - - expected_revisions_and_hash_tuple = ([102, 105, 108], [ - 'a123testhash1', 'a123testhash2', 'a123testhash3' - ]) - - @test_helpers.CallCountsToMockFunctions - def MockGetGitHashForRevision(call_count, _src_path, _rev): - # Simulate retrieving the git hash for the revision. - if call_count < 3: - return expected_revisions_and_hash_tuple[1][call_count] - - assert False, 'Called `GetGitHashFrom()` more than expected.' - - temp_worktree = '/abs/path/to/tmpDir' - - mock_create_temp_llvm_repo.return_value.__enter__.return_value.name = \ - temp_worktree - - # Simulate the valid revisions list. - mock_get_revisions_between_bisection.return_value = \ - expected_revisions_and_hash_tuple[0] - - # Simulate behavior of `GetGitHashFrom()` by using the testing - # function. - mock_get_git_hash.side_effect = MockGetGitHashForRevision + git_hashes = [ + git_llvm_rev.base_llvm_revision + 3, git_llvm_rev.base_llvm_revision + 5 + ] - start = 100 - end = 110 - parallel = 3 - src_path = None - pending_revisions = {103, 104} - skip_revisions = {101, 106, 107, 109} - - self.assertTupleEqual( - llvm_bisection.GetRevisionsListAndHashList( - start, end, parallel, src_path, pending_revisions, skip_revisions), - expected_revisions_and_hash_tuple) - - mock_get_revisions_between_bisection.assert_called_once() - - self.assertEqual(mock_get_git_hash.call_count, 3) + self.assertEqual( + llvm_bisection.GetCommitsBetween(start, end, parallel, abs_path_to_src, + test_pending_revisions, + test_skip_revisions), + (git_hashes, revs)) - def testSuccessfullyDieWithNoRevisionsError(self): + def testLoadStatusFilePassedWithExistingFile(self): start = 100 - end = 110 - - pending_revisions = {105, 108} - skip_revisions = {101, 102, 103, 104, 106, 107, 109} - - expected_no_revisions_message = ('No revisions between start %d and end ' - '%d to create tryjobs' % (start, end)) - - expected_no_revisions_message += '\nThe following tryjobs are pending:\n' \ - + '\n'.join(str(rev) for rev in pending_revisions) - - expected_no_revisions_message += '\nThe following tryjobs were skipped:\n' \ - + '\n'.join(str(rev) for rev in skip_revisions) - - # Verify that an exception is raised when there are no revisions to launch - # tryjobs for between 'start' and 'end' and some tryjobs are 'pending'. - with self.assertRaises(ValueError) as err: - llvm_bisection.DieWithNoRevisionsError(start, end, skip_revisions, - pending_revisions) + end = 150 - self.assertEqual(str(err.exception), expected_no_revisions_message) + test_bisect_state = {'start': start, 'end': end, 'jobs': []} - # Simulate behavior of `FindTryjobIndex()` when the index of the tryjob was - # found. - @mock.patch.object(update_tryjob_status, 'FindTryjobIndex', return_value=0) - def testTryjobExistsInRevisionsToLaunch(self, mock_find_tryjob_index): - test_existing_jobs = [{'rev': 102, 'status': 'good'}] + # Simulate that the status file exists. + with test_helpers.CreateTemporaryJsonFile() as temp_json_file: + with open(temp_json_file, 'w') as f: + test_helpers.WritePrettyJsonFile(test_bisect_state, f) - revision_to_launch = [102] + self.assertEqual( + llvm_bisection.LoadStatusFile(temp_json_file, start, end), + test_bisect_state) - expected_revision_that_exists = 102 + def testLoadStatusFilePassedWithoutExistingFile(self): + start = 200 + end = 250 - with self.assertRaises(ValueError) as err: - llvm_bisection.CheckForExistingTryjobsInRevisionsToLaunch( - revision_to_launch, test_existing_jobs) + expected_bisect_state = {'start': start, 'end': end, 'jobs': []} - expected_found_tryjob_index_error_message = ( - 'Revision %d exists already ' - 'in "jobs"' % expected_revision_that_exists) + last_tested = '/abs/path/to/file_that_does_not_exist.json' self.assertEqual( - str(err.exception), expected_found_tryjob_index_error_message) - - mock_find_tryjob_index.assert_called_once() + llvm_bisection.LoadStatusFile(last_tested, start, end), + expected_bisect_state) @mock.patch.object(modify_a_tryjob, 'AddTryjob') - def testSuccessfullyUpdatedStatusFileWhenExceptionIsRaised( - self, mock_add_tryjob): + def testBisectPassed(self, mock_add_tryjob): git_hash_list = ['a123testhash1', 'a123testhash2', 'a123testhash3'] revisions_list = [102, 104, 106] @@ -343,11 +208,11 @@ class LLVMBisectionTest(unittest.TestCase): # Verify that the status file is updated when an exception happened when # attempting to launch a revision (i.e. progress is not lost). with self.assertRaises(ValueError) as err: - llvm_bisection.UpdateBisection( - revisions_list, git_hash_list, bisection_contents, temp_json_file, - packages, args_output.chroot_path, patch_file, - args_output.extra_change_lists, args_output.options, - args_output.builders, args_output.verbose) + llvm_bisection.Bisect(revisions_list, git_hash_list, bisection_contents, + temp_json_file, packages, args_output.chroot_path, + patch_file, args_output.extra_change_lists, + args_output.options, args_output.builders, + args_output.verbose) expected_bisection_contents = { 'start': @@ -368,121 +233,143 @@ class LLVMBisectionTest(unittest.TestCase): with open(temp_json_file) as f: json_contents = json.load(f) - self.assertDictEqual(json_contents, expected_bisection_contents) + self.assertEqual(json_contents, expected_bisection_contents) self.assertEqual(str(err.exception), 'Unable to launch tryjob') self.assertEqual(mock_add_tryjob.call_count, 3) - # Simulate behavior of `GetGitHashFrom()` when successfully retrieved - # the git hash of the bad revision. Assume llvm_bisection has imported - # GetGitHashFrom + @mock.patch.object(subprocess, 'check_output', return_value=None) @mock.patch.object( - get_llvm_hash, 'GetGitHashFrom', return_value='a123testhash4') - def testCompletedBisectionWhenProvidedSrcPath(self, mock_get_git_hash): - last_tested = '/some/last/tested_file.json' + get_llvm_hash.LLVMHash, 'GetLLVMHash', return_value='a123testhash4') + @mock.patch.object(llvm_bisection, 'GetCommitsBetween') + @mock.patch.object(llvm_bisection, 'GetRemainingRange') + @mock.patch.object(llvm_bisection, 'LoadStatusFile') + @mock.patch.object(chroot, 'VerifyOutsideChroot', return_value=True) + def testMainPassed(self, mock_outside_chroot, mock_load_status_file, + mock_get_range, mock_get_revision_and_hash_list, + _mock_get_bad_llvm_hash, mock_abandon_cl): - src_path = '/abs/path/to/src/path' + start = 500 + end = 502 + cl = 1 - # The bad revision. - end = 150 + bisect_state = { + 'start': start, + 'end': end, + 'jobs': [{ + 'rev': 501, + 'status': 'bad', + 'cl': cl + }] + } - llvm_bisection._NoteCompletedBisection(last_tested, src_path, end) + skip_revisions = {501} + pending_revisions = {} - mock_get_git_hash.assert_called_once() + mock_load_status_file.return_value = bisect_state - # Simulate behavior of `GetLLVMHash()` when successfully retrieved - # the git hash of the bad revision. - @mock.patch.object( - get_llvm_hash.LLVMHash, 'GetLLVMHash', return_value='a123testhash5') - def testCompletedBisectionWhenNotProvidedSrcPath(self, mock_get_git_hash): - last_tested = '/some/last/tested_file.json' + mock_get_range.return_value = (start, end, pending_revisions, + skip_revisions) - src_path = None + mock_get_revision_and_hash_list.return_value = [], [] - # The bad revision. - end = 200 + args_output = test_helpers.ArgsOutputTest() + args_output.start_rev = start + args_output.end_rev = end + args_output.parallel = 3 + args_output.src_path = None + args_output.chroot_path = 'somepath' + args_output.cleanup = True - llvm_bisection._NoteCompletedBisection(last_tested, src_path, end) + self.assertEqual( + llvm_bisection.main(args_output), + llvm_bisection.BisectionExitStatus.BISECTION_COMPLETE.value) - mock_get_git_hash.assert_called_once() + mock_outside_chroot.assert_called_once() - def testSuccessfullyLoadedStatusFile(self): - start = 100 - end = 150 + mock_load_status_file.assert_called_once() - test_bisect_contents = {'start': start, 'end': end, 'jobs': []} + mock_get_range.assert_called_once() - # Simulate that the status file exists. - with test_helpers.CreateTemporaryJsonFile() as temp_json_file: - with open(temp_json_file, 'w') as f: - test_helpers.WritePrettyJsonFile(test_bisect_contents, f) + mock_get_revision_and_hash_list.assert_called_once() - self.assertDictEqual( - llvm_bisection.LoadStatusFile(temp_json_file, start, end), - test_bisect_contents) + mock_abandon_cl.assert_called_once() + self.assertEqual( + mock_abandon_cl.call_args, + mock.call( + [ + os.path.join(args_output.chroot_path, 'chromite/bin/gerrit'), + 'abandon', + cl, + ], + stderr=subprocess.STDOUT, + encoding='utf-8', + )) - def testLoadedStatusFileThatDoesNotExist(self): - start = 200 - end = 250 + @mock.patch.object(llvm_bisection, 'LoadStatusFile') + @mock.patch.object(chroot, 'VerifyOutsideChroot', return_value=True) + def testMainFailedWithInvalidRange(self, mock_outside_chroot, + mock_load_status_file): - expected_bisect_contents = {'start': start, 'end': end, 'jobs': []} + start = 500 + end = 502 - last_tested = '/abs/path/to/file_that_does_not_exist.json' + bisect_state = { + 'start': start - 1, + 'end': end, + } - self.assertDictEqual( - llvm_bisection.LoadStatusFile(last_tested, start, end), - expected_bisect_contents) - - # Simulate behavior of `_NoteCompletedBisection()` when there are no more - # tryjobs to launch between start and end, so bisection is complete. - @mock.patch.object(llvm_bisection, '_NoteCompletedBisection') - @mock.patch.object(llvm_bisection, 'GetRevisionsListAndHashList') - @mock.patch.object(llvm_bisection, 'GetStartAndEndRevision') - # Simulate behavior of `_ValidateStartAndEndAgainstJSONStartAndEnd()` when - # both start and end revisions match. - @mock.patch.object(llvm_bisection, - '_ValidateStartAndEndAgainstJSONStartAndEnd') + mock_load_status_file.return_value = bisect_state + + args_output = test_helpers.ArgsOutputTest() + args_output.start_rev = start + args_output.end_rev = end + args_output.parallel = 3 + args_output.src_path = None + + with self.assertRaises(ValueError) as err: + llvm_bisection.main(args_output) + + error_message = (f'The start {start} or the end {end} version provided is ' + f'different than "start" {bisect_state["start"]} or "end" ' + f'{bisect_state["end"]} in the .JSON file') + + self.assertEqual(str(err.exception), error_message) + + mock_outside_chroot.assert_called_once() + + mock_load_status_file.assert_called_once() + + @mock.patch.object(llvm_bisection, 'GetCommitsBetween') + @mock.patch.object(llvm_bisection, 'GetRemainingRange') @mock.patch.object(llvm_bisection, 'LoadStatusFile') - # Simulate behavior of `VerifyOutsideChroot()` when successfully invoked the - # script outside of the chroot. @mock.patch.object(chroot, 'VerifyOutsideChroot', return_value=True) - def testSuccessfullyBisectedLLVM( - self, mock_outside_chroot, mock_load_status_file, - mock_validate_start_and_end, mock_get_start_and_end_revision, - mock_get_revision_and_hash_list, mock_note_completed_bisection): + def testMainFailedWithPendingBuilds(self, mock_outside_chroot, + mock_load_status_file, mock_get_range, + mock_get_revision_and_hash_list): start = 500 end = 502 + rev = 501 - bisect_contents = { + bisect_state = { 'start': start, 'end': end, 'jobs': [{ - 'rev': 501, - 'status': 'skip' + 'rev': rev, + 'status': 'pending' }] } - skip_revisions = {501} - pending_revisions = {} + skip_revisions = {} + pending_revisions = {rev} + + mock_load_status_file.return_value = bisect_state + + mock_get_range.return_value = (start, end, pending_revisions, + skip_revisions) - # Simulate behavior of `LoadStatusFile()` when successfully loaded the - # status file. - mock_load_status_file.return_value = bisect_contents - - # Simulate behavior of `GetStartAndEndRevision()` when successfully found - # the new start and end revision of the bisection. - # - # Returns new start revision, new end revision, a set of pending revisions, - # and a set of skip revisions. - mock_get_start_and_end_revision.return_value = (start, end, - pending_revisions, - skip_revisions) - - # Simulate behavior of `GetRevisionsListAndHashList()` when successfully - # retrieved valid revisions (along with their git hashes) between start and - # end (in this case, none). mock_get_revision_and_hash_list.return_value = [], [] args_output = test_helpers.ArgsOutputTest() @@ -491,110 +378,131 @@ class LLVMBisectionTest(unittest.TestCase): args_output.parallel = 3 args_output.src_path = None - self.assertEqual( - llvm_bisection.main(args_output), - llvm_bisection.BisectionExitStatus.BISECTION_COMPLETE.value) + with self.assertRaises(ValueError) as err: + llvm_bisection.main(args_output) + + error_message = (f'No revisions between start {start} and end {end} to ' + 'create tryjobs\nThe following tryjobs are pending:\n' + f'{rev}\n') + + self.assertEqual(str(err.exception), error_message) mock_outside_chroot.assert_called_once() mock_load_status_file.assert_called_once() - mock_validate_start_and_end.assert_called_once() - - mock_get_start_and_end_revision.assert_called_once() + mock_get_range.assert_called_once() mock_get_revision_and_hash_list.assert_called_once() - mock_note_completed_bisection.assert_called_once() - - @mock.patch.object(llvm_bisection, 'DieWithNoRevisionsError') - # Simulate behavior of `_NoteCompletedBisection()` when there are no more - # tryjobs to launch between start and end, so bisection is complete. - @mock.patch.object(llvm_bisection, 'GetRevisionsListAndHashList') - @mock.patch.object(llvm_bisection, 'GetStartAndEndRevision') - # Simulate behavior of `_ValidateStartAndEndAgainstJSONStartAndEnd()` when - # both start and end revisions match. - @mock.patch.object(llvm_bisection, - '_ValidateStartAndEndAgainstJSONStartAndEnd') + @mock.patch.object(llvm_bisection, 'GetCommitsBetween') + @mock.patch.object(llvm_bisection, 'GetRemainingRange') @mock.patch.object(llvm_bisection, 'LoadStatusFile') - # Simulate behavior of `VerifyOutsideChroot()` when successfully invoked the - # script outside of the chroot. @mock.patch.object(chroot, 'VerifyOutsideChroot', return_value=True) - def testNoMoreTryjobsToLaunch( - self, mock_outside_chroot, mock_load_status_file, - mock_validate_start_and_end, mock_get_start_and_end_revision, - mock_get_revision_and_hash_list, mock_die_with_no_revisions_error): + def testMainFailedWithDuplicateBuilds(self, mock_outside_chroot, + mock_load_status_file, mock_get_range, + mock_get_revision_and_hash_list): start = 500 end = 502 + rev = 501 + git_hash = 'a123testhash1' - bisect_contents = { + bisect_state = { 'start': start, 'end': end, 'jobs': [{ - 'rev': 501, + 'rev': rev, 'status': 'pending' }] } skip_revisions = {} - pending_revisions = {501} - - no_revisions_error_message = ('No more tryjobs to launch between %d and ' - '%d' % (start, end)) - - def MockNoRevisionsErrorException(_start, _end, _skip, _pending): - raise ValueError(no_revisions_error_message) - - # Simulate behavior of `LoadStatusFile()` when successfully loaded the - # status file. - mock_load_status_file.return_value = bisect_contents - - # Simulate behavior of `GetStartAndEndRevision()` when successfully found - # the new start and end revision of the bisection. - # - # Returns new start revision, new end revision, a set of pending revisions, - # and a set of skip revisions. - mock_get_start_and_end_revision.return_value = (start, end, - pending_revisions, - skip_revisions) - - # Simulate behavior of `GetRevisionsListAndHashList()` when successfully - # retrieved valid revisions (along with their git hashes) between start and - # end (in this case, none). - mock_get_revision_and_hash_list.return_value = [], [] + pending_revisions = {rev} + + mock_load_status_file.return_value = bisect_state - # Use the test function to simulate `DieWithNoRevisionsWithError()` - # behavior. - mock_die_with_no_revisions_error.side_effect = MockNoRevisionsErrorException + mock_get_range.return_value = (start, end, pending_revisions, + skip_revisions) + + mock_get_revision_and_hash_list.return_value = [rev], [git_hash] - # Simulate behavior of arguments passed into the command line. args_output = test_helpers.ArgsOutputTest() args_output.start_rev = start args_output.end_rev = end args_output.parallel = 3 args_output.src_path = None - # Verify the exception is raised when there are no more tryjobs to launch - # between start and end when there are tryjobs that are 'pending', so - # the actual bad revision can change when those tryjobs's 'status' are - # updated. with self.assertRaises(ValueError) as err: llvm_bisection.main(args_output) - self.assertEqual(str(err.exception), no_revisions_error_message) + error_message = ('Revision %d exists already in "jobs"' % rev) + self.assertEqual(str(err.exception), error_message) mock_outside_chroot.assert_called_once() mock_load_status_file.assert_called_once() - mock_validate_start_and_end.assert_called_once() - - mock_get_start_and_end_revision.assert_called_once() + mock_get_range.assert_called_once() mock_get_revision_and_hash_list.assert_called_once() - mock_die_with_no_revisions_error.assert_called_once() + @mock.patch.object(subprocess, 'check_output', return_value=None) + @mock.patch.object( + get_llvm_hash.LLVMHash, 'GetLLVMHash', return_value='a123testhash4') + @mock.patch.object(llvm_bisection, 'GetCommitsBetween') + @mock.patch.object(llvm_bisection, 'GetRemainingRange') + @mock.patch.object(llvm_bisection, 'LoadStatusFile') + @mock.patch.object(chroot, 'VerifyOutsideChroot', return_value=True) + def testMainFailedToAbandonCL(self, mock_outside_chroot, + mock_load_status_file, mock_get_range, + mock_get_revision_and_hash_list, + _mock_get_bad_llvm_hash, mock_abandon_cl): + + start = 500 + end = 502 + + bisect_state = { + 'start': start, + 'end': end, + 'jobs': [{ + 'rev': 501, + 'status': 'bad', + 'cl': 0 + }] + } + + skip_revisions = {501} + pending_revisions = {} + + mock_load_status_file.return_value = bisect_state + + mock_get_range.return_value = (start, end, pending_revisions, + skip_revisions) + + mock_get_revision_and_hash_list.return_value = ([], []) + + error_message = 'Error message.' + mock_abandon_cl.side_effect = subprocess.CalledProcessError( + returncode=1, cmd=[], output=error_message) + + args_output = test_helpers.ArgsOutputTest() + args_output.start_rev = start + args_output.end_rev = end + args_output.parallel = 3 + args_output.src_path = None + args_output.cleanup = True + + with self.assertRaises(subprocess.CalledProcessError) as err: + llvm_bisection.main(args_output) + + self.assertEqual(err.exception.output, error_message) + + mock_outside_chroot.assert_called_once() + + mock_load_status_file.assert_called_once() + + mock_get_range.assert_called_once() if __name__ == '__main__': diff --git a/llvm_tools/llvm_project.py b/llvm_tools/llvm_project.py index c171370c..7937729f 100644 --- a/llvm_tools/llvm_project.py +++ b/llvm_tools/llvm_project.py @@ -13,6 +13,7 @@ import subprocess import sys import get_llvm_hash +import git_llvm_rev def get_location() -> str: @@ -33,9 +34,13 @@ def ensure_up_to_date(): assert checkout == actual_checkout, '%s != %s' % (actual_checkout, checkout) commit_timestamp = subprocess.check_output( - ['git', 'log', '-n1', '--format=%ct', 'origin/master'], + [ + 'git', 'log', '-n1', '--format=%ct', + 'origin/' + git_llvm_rev.MAIN_BRANCH + ], cwd=checkout, - encoding='utf-8') + encoding='utf-8', + ) commit_time = datetime.datetime.fromtimestamp(int(commit_timestamp.strip())) now = datetime.datetime.now() @@ -52,7 +57,7 @@ def ensure_up_to_date(): (time_since_last_commit.days, checkout), file=sys.stderr) - result = subprocess.run(['git', 'fetch', 'origin'], cwd=checkout) + result = subprocess.run(['git', 'fetch', 'origin'], check=False, cwd=checkout) if result.returncode: print( 'Sync failed somehow; hoping that things are fresh enough, then...', diff --git a/llvm_tools/nightly_revert_checker.py b/llvm_tools/nightly_revert_checker.py index 71b5574b..3a23890a 100755 --- a/llvm_tools/nightly_revert_checker.py +++ b/llvm_tools/nightly_revert_checker.py @@ -27,6 +27,7 @@ import typing as t import cros_utils.email_sender as email_sender import cros_utils.tiny_render as tiny_render import get_llvm_hash +import git_llvm_rev import revert_checker State = t.Any @@ -44,12 +45,12 @@ def _find_interesting_android_shas( encoding='utf-8', ).strip() - master_legacy = get_llvm_merge_base('aosp/master-legacy') + main_legacy = get_llvm_merge_base('aosp/master-legacy') testing_upstream = get_llvm_merge_base('aosp/testing-upstream') - result = [('master-legacy', master_legacy)] + result = [('main-legacy', main_legacy)] # If these are the same SHA, there's no point in tracking both. - if master_legacy != testing_upstream: + if main_legacy != testing_upstream: result.append(('testing-upstream', testing_upstream)) return result @@ -270,7 +271,7 @@ def main(argv: t.List[str]) -> None: for friendly_name, sha in interesting_shas: logging.info('Finding reverts across %s (%s)', friendly_name, sha) all_reverts = revert_checker.find_reverts( - llvm_dir, sha, root='origin/master') + llvm_dir, sha, root='origin/' + git_llvm_rev.MAIN_BRANCH) logging.info('Detected the following revert(s) across %s:\n%s', friendly_name, pprint.pformat(all_reverts)) diff --git a/llvm_tools/upload_lexan_crashes_to_forcey.py b/llvm_tools/upload_lexan_crashes_to_forcey.py new file mode 100755 index 00000000..b93f51a7 --- /dev/null +++ b/llvm_tools/upload_lexan_crashes_to_forcey.py @@ -0,0 +1,258 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- +# Copyright 2020 The Chromium OS Authors. All rights reserved. +# Use of this source code is governed by a BSD-style license that can be +# found in the LICENSE file. + +"""Fetches and submits the latest test-cases from Lexan's crash bucket.""" + +# pylint: disable=cros-logging-import + +import argparse +import contextlib +import datetime +import json +import logging +import os +import shutil +import subprocess +import sys +import tempfile +from typing import Generator, List, Iterable + +gsurl_base = 'gs://chrome-clang-crash-reports/v1' + + +def gsutil_ls(loc: str) -> List[str]: + results = subprocess.run(['gsutil.py', 'ls', loc], + stdout=subprocess.PIPE, + check=True, + encoding='utf-8') + return [l.strip() for l in results.stdout.splitlines()] + + +def gsurl_ls_last_numbers(url: str) -> List[int]: + return sorted(int(x.rstrip('/').split('/')[-1]) for x in gsutil_ls(url)) + + +def get_available_year_numbers() -> List[int]: + return gsurl_ls_last_numbers(gsurl_base) + + +def get_available_month_numbers(year: int) -> List[int]: + return gsurl_ls_last_numbers(f'{gsurl_base}/{year}') + + +def get_available_day_numbers(year: int, month: int) -> List[int]: + return gsurl_ls_last_numbers(f'{gsurl_base}/{year}/{month:02d}') + + +def get_available_test_case_urls(year: int, month: int, day: int) -> List[str]: + return gsutil_ls(f'{gsurl_base}/{year}/{month:02d}/{day:02d}') + + +def test_cases_on_or_after(date: datetime.datetime + ) -> Generator[str, None, None]: + """Yields all test-cases submitted on or after the given date.""" + for year in get_available_year_numbers(): + if year < date.year: + continue + + for month in get_available_month_numbers(year): + if year == date.year and month < date.month: + continue + + for day in get_available_day_numbers(year, month): + when = datetime.date(year, month, day) + if when < date: + continue + + yield when, get_available_test_case_urls(year, month, day) + + +def to_ymd(date: datetime.date) -> str: + return date.strftime('%Y-%m-%d') + + +def from_ymd(date_str: str) -> datetime.date: + return datetime.datetime.strptime(date_str, '%Y-%m-%d').date() + + +def persist_state(seen_urls: Iterable[str], state_file: str, + current_date: datetime.date): + tmp_state_file = state_file + '.tmp' + with open(tmp_state_file, 'w', encoding='utf-8') as f: + json.dump( + { + 'already_seen': sorted(seen_urls), + 'most_recent_date': to_ymd(current_date), + }, + f, + ) + os.rename(tmp_state_file, state_file) + + +@contextlib.contextmanager +def temp_dir() -> Generator[str, None, None]: + loc = tempfile.mkdtemp('lexan-autosubmit') + try: + yield loc + finally: + shutil.rmtree(loc) + + +def submit_test_case(gs_url: str, cr_tool: str) -> None: + logging.info('Submitting %s', gs_url) + suffix = os.path.splitext(gs_url)[1] + with temp_dir() as tempdir: + target_name = 'test_case' + suffix + target = os.path.join(tempdir, target_name) + subprocess.run(['gsutil.py', 'cp', gs_url, target], check=True) + subprocess.run(['tar', 'xaf', target_name], check=True, cwd=tempdir) + os.unlink(target) + + # Sometimes (e.g., in + # gs://chrome-clang-crash-reports/v1/2020/03/27/ + # chromium.clang-ToTiOS-12754-GTXToolKit-2bfcde.tgz) + # we'll get `.crash` files. Unclear why, but let's filter them out anyway. + repro_files = [x for x in os.listdir(tempdir) if not x.endswith('.crash')] + assert len(repro_files) == 2, repro_files + if repro_files[0].endswith('.sh'): + sh_file, src_file = repro_files + assert not src_file.endswith('.sh'), repro_files + else: + src_file, sh_file = repro_files + assert sh_file.endswith('.sh'), repro_files + + subprocess.run( + [ + cr_tool, + 'reduce', + '-stream=false', + '-wait=false', + '-note', + gs_url, + '-sh_file', + os.path.join(tempdir, sh_file), + '-src_file', + os.path.join(tempdir, src_file), + ], + check=True, + ) + + +def submit_new_test_cases( + last_seen_test_cases: Iterable[str], + earliest_date_to_check: datetime.date, + forcey: str, + state_file_path: str, +) -> None: + """Submits new test-cases to forcey. + + This will persist state after each test-case is submitted. + + Args: + last_seen_test_cases: test-cases which have been submitted already, and + should be skipped if seen again. + earliest_date_to_check: the earliest date we should consider test-cases + from. + forcey: path to the forcey binary. + state_file_path: path to our state file. + """ + # `all_test_cases_seen` is the union of all test-cases seen on this and prior + # invocations. It guarantees, in all cases we care about, that we won't + # submit the same test-case twice. `test_cases_seen_this_invocation` is + # persisted as "all of the test-cases we've seen on this and prior + # invocations" if we successfully submit _all_ test-cases. + # + # Since you can visualize the test-cases this script considers as a sliding + # window that only moves forward, if we saw a test-case on a prior iteration + # but no longer see it, we'll never see it again (since it fell out of our + # sliding window by being too old). Hence, keeping it around is + # pointless. + # + # We only persist this minimized set of test-cases if _everything_ succeeds, + # since if something fails below, there's a chance that we haven't revisited + # test-cases that we've already seen. + all_test_cases_seen = set(last_seen_test_cases) + test_cases_seen_this_invocation = [] + most_recent_date = earliest_date_to_check + for date, candidates in test_cases_on_or_after(earliest_date_to_check): + most_recent_date = max(most_recent_date, date) + + for url in candidates: + test_cases_seen_this_invocation.append(url) + if url in all_test_cases_seen: + continue + + all_test_cases_seen.add(url) + submit_test_case(url, forcey) + + # Persisting on each iteration of this loop isn't free, but it's the + # easiest way to not resubmit test-cases, and it's good to keep in mind + # that: + # - the state file will be small (<12KB, since it only keeps a few days + # worth of test-cases after the first run) + # - in addition to this, we're downloading+unzipping+reuploading multiple + # MB of test-case bytes. + # + # So comparatively, the overhead here probably isn't an issue. + persist_state(all_test_cases_seen, state_file_path, most_recent_date) + + persist_state(test_cases_seen_this_invocation, state_file_path, + most_recent_date) + + +def main(argv: List[str]): + logging.basicConfig( + format='>> %(asctime)s: %(levelname)s: %(filename)s:%(lineno)d: ' + '%(message)s', + level=logging.INFO, + ) + + my_dir = os.path.dirname(os.path.abspath(__file__)) + + parser = argparse.ArgumentParser(description=__doc__) + parser.add_argument( + '--state_file', default=os.path.join(my_dir, 'lexan-state.json')) + parser.add_argument( + '--last_date', + help='The earliest date that we care about. All test cases from here ' + 'on will be picked up. Format is YYYY-MM-DD.') + parser.add_argument( + '--4c', dest='forcey', required=True, help='Path to a 4c client binary') + opts = parser.parse_args(argv) + + forcey = opts.forcey + state_file = opts.state_file + last_date_str = opts.last_date + + os.makedirs(os.path.dirname(state_file), 0o755, exist_ok=True) + + if last_date_str is None: + with open(state_file, encoding='utf-8') as f: + data = json.load(f) + most_recent_date = from_ymd(data['most_recent_date']) + submit_new_test_cases( + last_seen_test_cases=data['already_seen'], + # Note that we always subtract one day from this to avoid a race: + # uploads may appear slightly out-of-order (or builders may lag, or + # ...), so the last test-case uploaded for 2020/01/01 might appear + # _after_ the first test-case for 2020/01/02. Assuming that builders + # won't lag behind for over a day, the easiest way to handle this is to + # always check the previous and current days. + earliest_date_to_check=most_recent_date - datetime.timedelta(days=1), + forcey=forcey, + state_file_path=state_file, + ) + else: + submit_new_test_cases( + last_seen_test_cases=(), + earliest_date_to_check=from_ymd(last_date_str), + forcey=forcey, + state_file_path=state_file, + ) + + +if __name__ == '__main__': + sys.exit(main(sys.argv[1:])) diff --git a/llvm_tools/upload_lexan_crashes_to_forcey_test.py b/llvm_tools/upload_lexan_crashes_to_forcey_test.py new file mode 100755 index 00000000..3c9c0d4b --- /dev/null +++ b/llvm_tools/upload_lexan_crashes_to_forcey_test.py @@ -0,0 +1,122 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- +# Copyright 2020 The Chromium OS Authors. All rights reserved. +# Use of this source code is governed by a BSD-style license that can be +# found in the LICENSE file. + +"""Tests for upload_lexan_crashes_to_forcey.""" + +import datetime +import unittest +import unittest.mock + +import upload_lexan_crashes_to_forcey + + +class Test(unittest.TestCase): + """Tests for upload_lexan_crashes_to_forcey.""" + + def test_date_parsing_functions(self): + self.assertEqual( + datetime.date(2020, 2, 1), + upload_lexan_crashes_to_forcey.from_ymd('2020-02-01')) + + @unittest.mock.patch( + 'upload_lexan_crashes_to_forcey.test_cases_on_or_after', + return_value=( + ( + datetime.date(2020, 1, 1), + ('gs://test-case-1', 'gs://test-case-1.1'), + ), + (datetime.date(2020, 1, 2), ('gs://test-case-2',)), + (datetime.date(2020, 1, 1), ('gs://test-case-3',)), + (datetime.date(2020, 1, 4), ('gs://test-case-4',)), + )) + @unittest.mock.patch('upload_lexan_crashes_to_forcey.submit_test_case') + @unittest.mock.patch('upload_lexan_crashes_to_forcey.persist_state') + def test_new_test_case_submission_functions(self, persist_state_mock, + submit_test_case_mock, + test_cases_on_or_after_mock): + forcey_path = '/path/to/4c' + real_state_file_path = '/path/to/state/file' + earliest_date = datetime.date(2020, 1, 1) + + persist_state_calls = [] + + # Since the set this gets is mutated, we need to copy it somehow. + def persist_state_side_effect(test_cases_to_persist, state_file_path, + most_recent_date): + self.assertEqual(state_file_path, real_state_file_path) + persist_state_calls.append( + (sorted(test_cases_to_persist), most_recent_date)) + + persist_state_mock.side_effect = persist_state_side_effect + + upload_lexan_crashes_to_forcey.submit_new_test_cases( + last_seen_test_cases=( + 'gs://test-case-0', + 'gs://test-case-1', + ), + earliest_date_to_check=earliest_date, + forcey=forcey_path, + state_file_path=real_state_file_path, + ) + + test_cases_on_or_after_mock.assert_called_once_with(earliest_date) + self.assertEqual(submit_test_case_mock.call_args_list, [ + unittest.mock.call('gs://test-case-1.1', forcey_path), + unittest.mock.call('gs://test-case-2', forcey_path), + unittest.mock.call('gs://test-case-3', forcey_path), + unittest.mock.call('gs://test-case-4', forcey_path), + ]) + + self.assertEqual(persist_state_calls, [ + ( + ['gs://test-case-0', 'gs://test-case-1', 'gs://test-case-1.1'], + datetime.date(2020, 1, 1), + ), + ( + [ + 'gs://test-case-0', + 'gs://test-case-1', + 'gs://test-case-1.1', + 'gs://test-case-2', + ], + datetime.date(2020, 1, 2), + ), + ( + [ + 'gs://test-case-0', + 'gs://test-case-1', + 'gs://test-case-1.1', + 'gs://test-case-2', + 'gs://test-case-3', + ], + datetime.date(2020, 1, 2), + ), + ( + [ + 'gs://test-case-0', + 'gs://test-case-1', + 'gs://test-case-1.1', + 'gs://test-case-2', + 'gs://test-case-3', + 'gs://test-case-4', + ], + datetime.date(2020, 1, 4), + ), + ( + [ + 'gs://test-case-1', + 'gs://test-case-1.1', + 'gs://test-case-2', + 'gs://test-case-3', + 'gs://test-case-4', + ], + datetime.date(2020, 1, 4), + ), + ]) + + +if __name__ == '__main__': + unittest.main() |