From 8da4861b3860c505e39341b4135c21f67569e4d8 Mon Sep 17 00:00:00 2001 From: Kuang-che Wu Date: Tue, 22 Oct 2024 21:04:41 +0800 Subject: subcmds: reduce multiprocessing serialization overhead Follow the same approach as 39ffd9977e to reduce serialization overhead. Below benchmarks are tested with 2.7k projects on my workstation (warm cache). git tracing is disabled for benchmark. (seconds) | v2.48 | v2.48 | this CL | this CL | | -j32 | | -j32 ----------------------------------------------------------- with clean tree state: branches (none) | 5.6 | 5.9 | 1.0 | 0.9 status (clean) | 21.3 | 9.4 | 19.4 | 4.7 diff (none) | 7.6 | 7.2 | 5.7 | 2.2 prune (none) | 5.7 | 6.1 | 1.3 | 1.2 abandon (none) | 19.4 | 18.6 | 0.9 | 0.8 upload (none) | 19.7 | 18.7 | 0.9 | 0.8 forall -c true | 7.5 | 7.6 | 0.6 | 0.6 forall -c "git log -1" | 11.3 | 11.1 | 0.6 | 0.6 with branches: start BRANCH --all | 21.9 | 20.3 | 13.6 | 2.6 checkout BRANCH | 29.1 | 27.8 | 1.1 | 1.0 branches (2) | 28.0 | 28.6 | 1.5 | 1.3 abandon BRANCH | 29.2 | 27.5 | 9.7 | 2.2 Bug: b/371638995 Change-Id: I53989a3d1e43063587b3f52f852b1c2c56b49412 Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/440221 Reviewed-by: Josip Sokcevic Tested-by: Kuang-che Wu Commit-Queue: Kuang-che Wu --- subcmds/start.py | 40 +++++++++++++++++++++++++--------------- 1 file changed, 25 insertions(+), 15 deletions(-) (limited to 'subcmds/start.py') diff --git a/subcmds/start.py b/subcmds/start.py index 56008f42..6dca7e4e 100644 --- a/subcmds/start.py +++ b/subcmds/start.py @@ -21,7 +21,6 @@ from error import RepoExitError from git_command import git from git_config import IsImmutable from progress import Progress -from project import Project from repo_logging import RepoLogger @@ -29,7 +28,7 @@ logger = RepoLogger(__file__) class ExecuteOneResult(NamedTuple): - project: Project + project_idx: int error: Exception @@ -80,18 +79,20 @@ revision specified in the manifest. if not git.check_ref_format("heads/%s" % nb): self.OptionParser.error("'%s' is not a valid name" % nb) - def _ExecuteOne(self, revision, nb, project): + @classmethod + def _ExecuteOne(cls, revision, nb, default_revisionExpr, project_idx): """Start one project.""" # If the current revision is immutable, such as a SHA1, a tag or # a change, then we can't push back to it. Substitute with # dest_branch, if defined; or with manifest default revision instead. branch_merge = "" error = None + project = cls.get_parallel_context()["projects"][project_idx] if IsImmutable(project.revisionExpr): if project.dest_branch: branch_merge = project.dest_branch else: - branch_merge = self.manifest.default.revisionExpr + branch_merge = default_revisionExpr try: project.StartBranch( @@ -100,7 +101,7 @@ revision specified in the manifest. except Exception as e: logger.error("error: unable to checkout %s: %s", project.name, e) error = e - return ExecuteOneResult(project, error) + return ExecuteOneResult(project_idx, error) def Execute(self, opt, args): nb = args[0] @@ -120,19 +121,28 @@ revision specified in the manifest. def _ProcessResults(_pool, pm, results): for result in results: if result.error: - err_projects.append(result.project) + project = all_projects[result.project_idx] + err_projects.append(project) err.append(result.error) pm.update(msg="") - self.ExecuteInParallel( - opt.jobs, - functools.partial(self._ExecuteOne, opt.revision, nb), - all_projects, - callback=_ProcessResults, - output=Progress( - f"Starting {nb}", len(all_projects), quiet=opt.quiet - ), - ) + with self.ParallelContext(): + self.get_parallel_context()["projects"] = all_projects + self.ExecuteInParallel( + opt.jobs, + functools.partial( + self._ExecuteOne, + opt.revision, + nb, + self.manifest.default.revisionExpr, + ), + range(len(all_projects)), + callback=_ProcessResults, + output=Progress( + f"Starting {nb}", len(all_projects), quiet=opt.quiet + ), + chunksize=1, + ) if err_projects: for p in err_projects: -- cgit v1.2.3-54-g00ecf