Data flow: Prune parameter-self flow in stage 1 - #18333
Conversation
1837862 to
5cef673
Compare
5cef673 to
06ba814
Compare
There was a problem hiding this comment.
Copilot wasn't able to review any files in this pull request.
Files not reviewed (2)
- shared/dataflow/codeql/dataflow/internal/DataFlowImpl.qll: Language not supported
- shared/dataflow/codeql/dataflow/internal/DataFlowImplCommon.qll: Language not supported
Tip: Copilot code review supports C#, Go, Java, JavaScript, Markdown, Python, Ruby and TypeScript, with more languages coming soon. Learn more
| pragma[nomagic] | ||
| private predicate fwdFlowInParam(DataFlowCall call, ParamNodeEx p, Cc cc) { | ||
| fwdFlowIn(call, _, cc, p) | ||
| } |
There was a problem hiding this comment.
The columns call and p ought to functionally determine arg on almost all cases, so I think it's better to inline this projection rather than materialising additional tuples.
aschackmull
left a comment
There was a problem hiding this comment.
IIRC you were experimenting with elimination of the allowParameterReturnInSelf concept? Of course, if we're not doing that then this change LGTM (modulo one minor comment).
Yes, I was, but performance turned out to be terrible :-( |
Example tuple counts (~10 % reduction in stage 1):
Before
After