Skip to content

Latest commit

 

History

History
34 lines (21 loc) · 1.11 KB

scaling-arcus.md

File metadata and controls

34 lines (21 loc) · 1.11 KB
Raw

Analyzing ARCUS At Scale

Batching over Remote Hosts

The scripts in tools/angr/scripts/gnu-parallel-analysis let you run ARCUS's analysis.py over one or more remote hosts using GNU Parallel.

Guide:

  1. Make sure the "master" host you're running these scripts on can access all the remote hosts via SSH.

  2. Run setup-host.sh for each host. Ex: ./setup-host.sh server.example.org [email protected]:user/my-repo.git

  3. Write a hosts file at tools/angr/scripts/gnu-parallel-analysis/hosts. See GNU Parallel's documentation for more details.

  4. Place all the traces you want to analyze in one directory and then navigate to where you want the results to be stored (ideally, another empty directory). Note: you may want to use tools/pt/cmppath to filter out similar traces.

  5. Run run.sh.

Run Script Usage

run.sh <traces_dir> <git_branch> [extra_analysis_args ...]

EXAMPLES:

# basic run
./run.sh /path/to/traces-dir master

# pass additional arguments to analysis.py
./run.sh /path/to/traces-dir master --explore --logging=40