C++: Reduce re-evaluation

[MathiasVP]

This PR removes most of the re-evaluation of predicates from previous stages. In particular, the dataflow stage made use of various internal predicates from IR generation which caused some unfortunate re-evaluation.

There were also some other stage-related issues where various predicates in dataflow were cached, but not put inside a cached module, which produced unnecessary stages.

Unfortunately, this was easier said than done. It required a large refactoring of the dataflow files because I we have made some fairly arbitrary choices as to what belonged in DataFlowUtil.qll and DataFlowPrivate.qll which made it hard to unify the stages. I tried my best to make this refactoring commit-by-commit reviewable.

Most of this PR has been guided by the output of Schack's stageoverlap.py script. There should be no changes to any resulting tuples.

DCA looks great. We shave off about 4-5% of the total analysis time 🎉 Additionally, my experience is that this PR removes a lot of recomputation of DataFlowUtil.qll/DataFlowPrivate.qll predicates when I modify an isSource or isSink in a query.

[Copilot]

Pull request overview

This PR refactors the C++ dataflow implementation to reduce re-evaluation of predicates across compilation stages, achieving approximately 4-5% improvement in total analysis time. The changes reorganize how cached predicates are structured and introduce a new DataFlowNodes.qll file to centralize node definitions, thereby reducing stage overlap.

Changes:

• Introduces DataFlowNodes.qll to centralize dataflow node definitions, moving classes like Node, ParameterNode, VariableNode, etc. from DataFlowUtil.qll
• Wraps cached predicates in Cached modules with appropriate stage collapsing to prevent unnecessary recomputation
• Adds new cached predicates in IRConstruction.qll to expose translation information without causing re-evaluation
• Reorganizes TaintTrackingUtil.qll to wrap predicates in a Cached module for proper stage management
• Moves several cached predicates from DataFlowPrivate.qll into the Cached module for better stage control

Reviewed changes

Copilot reviewed 17 out of 17 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
FlowSummaryNode.ql	Adds import of DataFlowNodes for test access to node classes
CaptureModels.qll	Updates references from DataFlow::FieldAddress to DataFlowNodes::FieldAddress
IRConstruction.qll	Adds cached predicates to expose translation information (allocation extents, transparent conversions, etc.)
TaintTrackingUtil.qll	Wraps taint tracking predicates in a Cached module with stage collapsing
SsaImplCommon.qll	Moves utility predicates into Cached module for proper stage management
SsaImpl.qll	Delegates to IRConstruction for initialization target address
PrintIRUtilities.qll, PrintIRLocalFlow.qll, PrintIRFieldFlowSteps.qll	Adds DataFlowNodes imports for access to node classes
ModelUtil.qll	Adds DataFlowNodes import
ExprNodes.qll	Uses IRConstruction cached predicates instead of direct translation class access
DataFlowUtil.qll	Major refactoring: moves node definitions to DataFlowNodes.qll, imports DataFlowNodes::Public
DataFlowPrivate.qll	Consolidates cached predicates (toString, getLocation, newtype definitions) into Cached module
DataFlowNodes.qll	New file containing node class definitions, Content types, and related predicates
FlowSummaryImpl.qll	Updates toString call to use toStringImpl()
ExternalFlow.qll	Adds pragma[nomagic] for performance optimization

[jketema]

Doesn't look ideal that we now need to import something internal here. Does this query do something odd that requires this?

[MathiasVP]

Yes: 18dea55

I'm not sure why we did it this way, but it imports internal dataflow nodes to find definitions outside a loop. I think we could do this with the proper SSA API we have these days, but I didnt want to change that in this PR

[MathiasVP]

For standard queries you still only have to import semmle.code.cpp.dataflow.new.DataFlow

[jketema]

LGTM. As far as I can tell this is really just moving around stuff (and adding some missing comments). DCA numbers look ok.