Skip to main content

A Semgrep scan is having a problem - what next?

If a Semgrep scan is failing or running slowly, try the following steps to investigate:

  1. Update Semgrep to the latest version, if you are not currently running the latest version. Some errors result from an older version of Semgrep being used with newer rules.
  2. Re-run the scan with either the -v/--verbose or --debug (extremely verbose) flags. These options provide more information about what is failing.
  • When using verbose logs, also set --max-log-list-entries to 0 or any negative value to see the full output; otherwise lists of rules and files are suppressed if there are more than 100 entries in the list.
  1. If you are running cross-file (interfile) analysis in the scan, remove any options starting with --pro, or run Semgrep with --oss-only. This allows isolation of any issues related to cross-file analysis, and often speeds up a scan or reduces memory usage.
info

Semgrep verbose or debug logs can be quite lengthy. To prevent flooding your terminal and preserve the logs for analysis, you can redirect all output to a file with semgrep [OPTIONS] [TARGETS]... &> semgrep.log. See also How to collect logs when running Semgrep in CLI.

Memory usage issues (OOM errors)​

Memory usage is a common issue with scans, especially in memory-constrained environments such as continuous integration (CI) providers. Semgrep may exit with code -11 (or -9), which are the POSIX signals raised to cause the crash.

  • Try increasing the memory available if you are working in a container or managed instance where you can manage the amount of memory.
  • Use the --max-memory LIMIT option for your Semgrep run. This option stops a rule/file scan if it reaches the set limit, and moves to the next rule / file.
    • If you are running an interfile scan, this option also falls back to the OSS engine if the interfile pre-processing stage requires more than this amount of memory.
  • Run Semgrep in single-threaded mode with --jobs 1. This reduces the amount of memory used compared to running multiple jobs.
  • Try increasing your stack limit, if a limit is set for the context where you invoke Semgrep (ulimit -s [limit]).

Slow scans​

The first step to improving Semgrep's speed is limiting its run to only the files you care about. Most commonly, it's limited using a .semgrepignore file. See Ignoring files, folders, or parts of code.

After addressing files to ignore:

  • If you suspect the presence of a large file slowing Semgrep's analysis, decrease the maximum size of files scanned with --max-target-bytes BYTES.
  • Run Semgrep with the --time flag. This outputs a list of the rules and files that took the longest.
    • Identify the slowest files from the list. You may find that you can add some of those files to your ignore list as well.
    • Identify the slowest rules from the list. You may find that some of them don't apply to your codebase and can be skipped.

Adjusting timeouts​

Semgrep has several timeout settings that affect scan duration and can be adjusted to optimize scan behavior:

  • --timeout: Similar to --max-memory, --timeout affects the behavior of the scan when running a single rule on a single file. It defaults to 5 seconds. Typical values range from 3 seconds (favors faster scans, but more timeouts) to 30 seconds (slower scans, fewer timeouts).
  • --timeout-threshold: The number of attempts made to run a single rule on a single file, if it times out due to the --timeout limit. It defaults to 3. Decreasing the value may speed up scans but cause more timeouts.
  • --interfile-timeout: If you are running an interfile scan, this is the maximum amount of time in seconds to spend on interfile analysis before falling back to the OSS Engine. Defaults to 3 hours (10800 seconds) for scans using semgrep ci. Otherwise, the default is no maximum time (continue with cross-file analysis until the scan completes).

401 error when scanning with Semgrep Registry rules​

If you receive a 401 when scanning using registry rules (for example, with --config auto), try the following:

  1. Run semgrep logout.
  2. Attempt the scan again.

If this is successful, your stored local token was invalid. Use semgrep login to log in again and receive a fresh token.

Scan failures with analysis errors​

Analysis or parsing errors usually only affect a particular rule, file, or language. If your scan encounters an analysis issue, using verbose logging can provide you with helpful error details, such as:

metavariable-pattern failed because we lack range info for $X, please file a bug report

If the error you receive is not that specific, try one of these options:

  1. Use --exclude-rule to exclude a rule from the scan. This allows isolating the problem to the particular rule.
  • If you are running Semgrep in CI with Semgrep AppSec Platform, and don't need to run the rule, you can also disable the rule.
  1. Use --exclude to exclude a file or files from the scan. You can use wildcards in file exclusions to exclude files matching particular patterns.
  2. Use --include with a pattern specifying a path or an extension for a particular language, to limit the scan to that path, or to files in that language.

Reporting crashes or analysis errors​

Once you have isolated the issue:

  1. Identify the rule, file, and lines (if available) where Semgrep encountered the error.
  2. Determine whether you can share a minimal example of the code or rule that is causing the issue.
  • If the issue occurs with cross-file analysis, or the code is internal or sensitive and cannot be sufficiently redacted, reach out for help, and include what you've determined so far.
  • Otherwise, share the issue details and related code with Semgrep via https://github.com/semgrep/semgrep/issues.

If you are encountering memory usage issues, please include in your report:

  • The total size of the files
  • The number of files being scanned
  • The maximum memory used by Semgrep (an estimate from top is fine)
  • The system specifications

Not finding what you need in this doc? Ask questions in our Community Slack group, or see Support for other ways to get help.