hohn/codeql
1
0
Fork 0
mirror of https://github.com/github/codeql.git synced 2026-04-25 08:45:14 +02:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #15260 from hvitved/dataflow/may-benefit-from-cctx-simplify

Browse Source 
Data flow: Remove column from `mayBenefitFromCallContext`
This commit is contained in:
Tom Hvitved
2024-01-10 11:43:15 +01:00
committed by GitHub
parent ac8e92eec5 f90201eb56
commit c9cf2a899c
17 changed files with 85 additions and 126 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
docs/ql-libraries/dataflow/dataflow.md
View File

@@ -588,18 +588,17 @@ However, joining the virtual dispatch relation with itself in this way is
usually way too big to be feasible. Instead, the relation above should only be
defined for those values of `call` for which the set of resulting dispatch
targets might be reduced. To do this, define the set of `call`s that might for
some reason benefit from a call context as the following predicate (the `c`
column should be `call.getEnclosingCallable()`):
some reason benefit from a call context as the following predicate:
```ql
predicate mayBenefitFromCallContext(DataFlowCall call, DataFlowCallable c)
predicate mayBenefitFromCallContext(DataFlowCall call)
```
And then define `DataFlowCallable viableImplInCallContext(DataFlowCall call,
DataFlowCall ctx)` as sketched above, but restricted to
`mayBenefitFromCallContext(call, _)`.
`mayBenefitFromCallContext(call)`.
The shared implementation will then compare counts of virtual dispatch targets
using `viableCallable` and `viableImplInCallContext` for each `call` in
`mayBenefitFromCallContext(call, _)` and track call contexts during flow
`mayBenefitFromCallContext(call)` and track call contexts during flow
calculation when differences in these counts show an improved precision in
further calls.