Re: [gentoo-portage-dev] Speeding up Tree Verification
Hi, On 29-06-2020 21:13:43 -0500, Sid Spry wrote: > Hello, > > I have some runnable pseudocode outlining a faster tree verification > algorithm. > Before I create patches I'd like to see if there is any guidance on making the > changes as unobtrusive as possible. If the radical change in algorithm is > acceptable I can work on adding the changes. > > Instead of composing any kind of structured data out of the portage tree my > algorithm just lists all files and then optionally batches them out to > threads. > There is a noticeable speedup by eliding the tree traversal operations which > can be seen when running the algorithm with a single thread and comparing it > to > the current algorithm in gemato (which should still be discussed here?). I remember something that gemato used to use multiple threads, but because it totally saturated disk-IO, it was brought back to a single thread. People were complaining about unusable systems. In any case, can you share your performance results? What speedup did you see, on warm and hot FS caches? Which type of disk do you use? You could compare against qmanifest, which uses OpenMP-based paralllelism while verifying the tree. On SSDs this does help. Thanks, Fabian > > Some simple tests like counting all objects traversed and verified returns the > same(ish). Once it is put into portage it could be tested in detail. > > There is also my partial attempt at removing the brittle interface to GnuPG > (it's not as if the current code is badly designed, just that parsing the > output of GnuPG directly is likely not the best idea). > > Needs gemato, dnspython, and requests. Slightly better than random code > because > I took inspiration from the existing gemato classes. > > ```python (veriftree.py) > #!/usr/bin/env python3 > import os, sys, zlib, hashlib, tempfile, shutil, timeit > import subprocess > from typing import List > from pprint import pprint > > from gemato.manifest import ( > ManifestFile, > ManifestFileEntry, > ) > from wkd import ( > check_domain_signature, > hash_localpart, > build_web_key_uri, > stream_to_file > ) > from fetchmedia import ( > OpenPGPEnvironment, > setup_verification_environment > ) > > # 0. Top level directory (repository) contains Manifest, a PGP signature of > #blake2b and sha512 hashes of Manifest.files.gz. > # 1. Manifest.files contains hashes of each category Manifest.gz. > # 2. The category Manifest contains hashes of each package Manifest. > # 3. The package Manifest contains hashes of each package file. > #Must be aware of PMS, e.g. aux tag specifies a file in files/. > > # 0. Check signature of repo Manifest. > # 1. Merge items in Manifest.files, each category Manifest, and each package > #Manifest into one big list. The path must be made absolute. > # 2. Distribute items to threads. > > # To check operation compare directory tree to files appearing in all > # ManifestRecords. > > class ManifestTree(object): > __slots__ = ['_directory', '_manifest_list', '_manifest_records', > '_manifest_results'] > > def __init__(self, directory: str): > self._directory = directory > # Tuples of (base_path, full_path). > self._manifest_list = [] > self._manifest_records = [] > self._manifest_results = [] > > def build_manifest_list(self): > for path, dirs, files in os.walk(self._directory): > #if 'glsa' in path or 'news' in path: > #if 'metadata' in path: > #continue # Skip the metadata directory for now. > # It contains a repository. Current algo barfs on Manifest > # containing only sig. > > if 'Manifest.files.gz' in files: > self._manifest_list += [(path, path + '/Manifest.files.gz')] > if 'Manifest.gz' in files: > self._manifest_list += [(path, path + '/Manifest.gz')] > > if path == self._directory: > continue # Skip the repo manifest. Order matters, fix > eventually. > if 'Manifest' in files: > self._manifest_list += [(path, path + '/Manifest')] > > def parse_manifests(self): > td = tempfile.TemporaryDirectory(dir='./') > for manifest in self._manifest_list: > def inner(): > if manifest[1].endswith('.gz'): > name = 'Manifest.files' # Need to also handle Manifest.gz. > path = '{0}/{1}'.format(td.name, name) > subprocess.run(['sh', '-c', 'gunzip -c {0} > {1}' > .format(manifest[1], path)]) > for line in open(path): > mr = ManifestRecord(line) > mr.make_absolute(manifest[0]) > self._manifest_records += [mr] > else: > for line in open(manifest[1]): >
Re: [gentoo-portage-dev] Re: Speeding up Tree Verification
On 6/29/20 7:15 PM, Sid Spry wrote: > On Mon, Jun 29, 2020, at 9:13 PM, Sid Spry wrote: >> Hello, >> >> I have some runnable pseudocode outlining a faster tree verification >> algorithm. > > Ah, right. It's worth noting that even faster than this algorithm is simply > verifying > a .tar.xz. Is that totally off the table? I realize it doesn't fit every > usecase, but it > seems to be faster in both sync and verification time. We've already got support for that with sync-type = webrsync. However, I imagine sync-type = git is even better. All of the types are covered here: https://wiki.gentoo.org/wiki/Portage_Security -- Thanks, Zac
[gentoo-portage-dev] Re: Speeding up Tree Verification
On Mon, Jun 29, 2020, at 9:13 PM, Sid Spry wrote: > Hello, > > I have some runnable pseudocode outlining a faster tree verification > algorithm. Ah, right. It's worth noting that even faster than this algorithm is simply verifying a .tar.xz. Is that totally off the table? I realize it doesn't fit every usecase, but it seems to be faster in both sync and verification time.
[gentoo-portage-dev] Speeding up Tree Verification
Hello, I have some runnable pseudocode outlining a faster tree verification algorithm. Before I create patches I'd like to see if there is any guidance on making the changes as unobtrusive as possible. If the radical change in algorithm is acceptable I can work on adding the changes. Instead of composing any kind of structured data out of the portage tree my algorithm just lists all files and then optionally batches them out to threads. There is a noticeable speedup by eliding the tree traversal operations which can be seen when running the algorithm with a single thread and comparing it to the current algorithm in gemato (which should still be discussed here?). Some simple tests like counting all objects traversed and verified returns the same(ish). Once it is put into portage it could be tested in detail. There is also my partial attempt at removing the brittle interface to GnuPG (it's not as if the current code is badly designed, just that parsing the output of GnuPG directly is likely not the best idea). Needs gemato, dnspython, and requests. Slightly better than random code because I took inspiration from the existing gemato classes. ```python (veriftree.py) #!/usr/bin/env python3 import os, sys, zlib, hashlib, tempfile, shutil, timeit import subprocess from typing import List from pprint import pprint from gemato.manifest import ( ManifestFile, ManifestFileEntry, ) from wkd import ( check_domain_signature, hash_localpart, build_web_key_uri, stream_to_file ) from fetchmedia import ( OpenPGPEnvironment, setup_verification_environment ) # 0. Top level directory (repository) contains Manifest, a PGP signature of #blake2b and sha512 hashes of Manifest.files.gz. # 1. Manifest.files contains hashes of each category Manifest.gz. # 2. The category Manifest contains hashes of each package Manifest. # 3. The package Manifest contains hashes of each package file. #Must be aware of PMS, e.g. aux tag specifies a file in files/. # 0. Check signature of repo Manifest. # 1. Merge items in Manifest.files, each category Manifest, and each package #Manifest into one big list. The path must be made absolute. # 2. Distribute items to threads. # To check operation compare directory tree to files appearing in all # ManifestRecords. class ManifestTree(object): __slots__ = ['_directory', '_manifest_list', '_manifest_records', '_manifest_results'] def __init__(self, directory: str): self._directory = directory # Tuples of (base_path, full_path). self._manifest_list = [] self._manifest_records = [] self._manifest_results = [] def build_manifest_list(self): for path, dirs, files in os.walk(self._directory): #if 'glsa' in path or 'news' in path: #if 'metadata' in path: #continue # Skip the metadata directory for now. # It contains a repository. Current algo barfs on Manifest # containing only sig. if 'Manifest.files.gz' in files: self._manifest_list += [(path, path + '/Manifest.files.gz')] if 'Manifest.gz' in files: self._manifest_list += [(path, path + '/Manifest.gz')] if path == self._directory: continue # Skip the repo manifest. Order matters, fix eventually. if 'Manifest' in files: self._manifest_list += [(path, path + '/Manifest')] def parse_manifests(self): td = tempfile.TemporaryDirectory(dir='./') for manifest in self._manifest_list: def inner(): if manifest[1].endswith('.gz'): name = 'Manifest.files' # Need to also handle Manifest.gz. path = '{0}/{1}'.format(td.name, name) subprocess.run(['sh', '-c', 'gunzip -c {0} > {1}' .format(manifest[1], path)]) for line in open(path): mr = ManifestRecord(line) mr.make_absolute(manifest[0]) self._manifest_records += [mr] else: for line in open(manifest[1]): if line.startswith('-'): return # Skip the signed manifest. mr = ManifestRecord(line) mr.make_absolute(manifest[0]) self._manifest_records += [mr] inner() def verify_manifests(self): for record in self._manifest_records: self._manifest_results += [record.verify()] class ManifestRecord(object): __slots__ = ['_tag', '_abs_path', '_path', '_size', '_hashes'] def __init__(self, line: str=None): self._tag = None self._abs_path = None self._path = None self._size = None self._hashes = [] if line: self.from_string(line) def from_s