C#: mass enable diff-informed data flow #19661

d10c · 2025-06-03T18:25:55Z

An auto-generated patch that enables diff-informed data flow in the obvious cases.

Builds on #18344 and https://github.com/github/codeql-patch/pull/88

Copilot

Pull Request Overview

This PR auto-generates patches to enable diff-informed data flow by adding a default observeDiffInformedIncrementalMode predicate in numerous data-flow configuration modules.

Added predicate observeDiffInformedIncrementalMode() { any() } to all relevant DataFlow::ConfigSig modules.
Covers security, cryptography, and likely-bug query modules for incremental diff analysis.

Reviewed Changes

Copilot reviewed 26 out of 26 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
csharp/ql/src/Security Features/CWE-114/AssemblyPathInjection.ql	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/src/Security Features/CWE-091/XMLInjection.ql	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/src/Likely Bugs/LeapYear/UnsafeYearConstruction.ql	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/ZipSlipQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/XPathInjectionQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/UrlRedirectQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/TaintedPathQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/SqlInjectionQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/ResourceInjectionQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/RegexInjectionQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/ReDoSQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/MissingXMLValidationQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/LogForgingQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/LDAPInjectionQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/ExposureOfPrivateInformationQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/CommandInjectionQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/CodeInjectionQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/dataflow/CleartextStorageQuery.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/cryptography/HardcodedSymmetricEncryptionKey.qll	Added `observeDiffInformedIncrementalMode` predicate
csharp/ql/lib/semmle/code/csharp/security/cryptography/EncryptionKeyDataFlowQuery.qll	Added `observeDiffInformedIncrementalMode` predicate

Comments suppressed due to low confidence (2)

csharp/ql/src/Security Features/CWE-114/AssemblyPathInjection.ql:45

[nitpick] Add a brief comment above this predicate to explain its role in diff-informed incremental analysis, improving clarity for future maintainers.

predicate observeDiffInformedIncrementalMode() { any() }

csharp/ql/src/Security Features/CWE-114/AssemblyPathInjection.ql:45

There are no existing tests exercising the incremental diff mode; consider adding test cases to validate behavior when this predicate is active.

predicate observeDiffInformedIncrementalMode() { any() }

csharp/ql/src/Security Features/CWE-114/AssemblyPathInjection.ql

d10c · 2025-06-05T15:59:30Z

It turns out that some of the generated changes in the PRs were not correct, e.g. because they should have also generated a getASelected{Source,Sink}Location() override but didn't (see Chuan-kai's comment here). So for now I'm putting them back in Draft until I make sure (via the patch script) that we are correctly handling all 3 documented query patterns, starting with the simplest one (both source and sink are used as location sources). If you have already started reviewing the PRs, thank you (also for your patience) and stay tuned for an update as to what has changed in the meantime!

michaelnebel · 2025-06-06T11:22:32Z

@d10c : Great!
Thank you for doing this. A couple of questions

Do you expect there will be any configurations that will not be diff-informed?
How does it impact tests (I remember hearing something along the lines of that tests already worked with "diff-informed", if the tests are .qlref files)?

d10c · 2025-06-10T15:06:38Z

Update: no changes since last time I opened the PR. It turns out that it's sound (but not optimally performant) to leave getASelected{Source,Sink}Location() un-overridden, specifically in case of a select clause containing only one of source or sink but not both. The patch script currently does not differentiate between that case and the one in which both source and sink are present in the select clause. So I will re-open these PRs as they are, and generate an appropriate getASelected{Source,Sink}Location() override in a follow-up round of PRs.

d10c · 2025-06-10T15:11:48Z

Do you expect there will be any configurations that will not be diff-informed?

In this initial PR, we are skipping over the cases where a node other than source or sink is used as a location source in the select clause. To enable these cases, a non-empty override of getASelected{Source,Sink}Location() will be needed. Also, queries with more than one dataflow config are probably another case to be considered separately.

How does it impact tests (I remember hearing something along the lines of that tests already worked with "diff-informed", if the tests are .qlref files)?

It's true that in order to enable diff-informed testing of queries, the tests have to be .qlref files. I'm planning on doing that kind of rewrite (similar to this PR) in a later phase.

michaelnebel

Should we trigger a DCA run before merging? Are there any other special testing that is needed before merge?

michaelnebel · 2025-06-11T07:50:58Z

csharp/ql/lib/change-notes/2025-06-04-diff-informed.md

+---
+category: minorAnalysis
+---
+* A number of built-in C# queries can now run in diff-informed mode.


Has the documentation on, how to run queries in diff-informed mode been released? (basically my question is whether we should have a release note)

No, it hasn't been made public. So it probably makes sense to not have a release note on any of these PRs.

michaelnebel · 2025-06-11T08:01:39Z

csharp/ql/lib/semmle/code/csharp/security/dataflow/SqlInjectionQuery.qll

@@ -43,6 +43,8 @@ module SqlInjectionConfig implements DataFlow::ConfigSig {
   * `node` from the data flow graph.
   */
  predicate isBarrier(DataFlow::Node node) { node instanceof Sanitizer }
+
+  predicate observeDiffInformedIncrementalMode() { any() }


The related location for this query uses a PathNode instead of Node. Will that cause an issue?

As far as I can tell, I don't think it matters - as this predicate only relates to filtering of source and sink nodes based on an extensible predicate that is pre-populated with a diff range (in which a source and sink should be located).

Since (AFAIK) PathNode and Node have the same location, it shouldn't make a difference.

michaelnebel · 2025-06-11T08:04:19Z

csharp/ql/src/Likely Bugs/LeapYear/UnsafeYearConstruction.ql

@@ -26,6 +26,8 @@ module UnsafeYearCreationFromArithmeticConfig implements DataFlow::ConfigSig {
      oc.getObjectType().getABaseType*().hasFullyQualifiedName("System", "DateTime")
    )
  }
+
+  predicate observeDiffInformedIncrementalMode() { any() }


Both the primary and related locations are of type PathNode. Is that a problem?

michaelnebel · 2025-06-11T08:05:18Z

csharp/ql/src/Security Features/CWE-114/AssemblyPathInjection.ql

@@ -41,6 +41,8 @@ module AssemblyPathInjectionConfig implements DataFlow::ConfigSig {
      name = "UnsafeLoadFrom" and arg = 0
    )
  }
+
+  predicate observeDiffInformedIncrementalMode() { any() }


Related location is PathNode.

michaelnebel · 2025-06-11T08:26:52Z

It's true that in order to enable diff-informed testing of queries, the tests have to be .qlref files. I'm planning on doing that kind of rewrite (similar to this PR) in a later phase.

Fantastic! We are also rewriting the test cases whenever we touch the queries (to use inline expectations) - so that would be great! 😄

d10c · 2025-06-11T16:31:16Z

As for DCA, it looks like we don't have sources set up with diffs, so we can't use DCA to test performance on diffs yet. A normal DCA run would then test before-and-after-this-PR without diff-informed mode, but those two runs should have the same behaviour. So I say we leave DCA testing out for now until I get the diff-informed sources in place.

An auto-generated patch that enables diff-informed data flow in the obvious cases. Builds on github#18344 and github/codeql-patch#88

michaelnebel · 2025-06-12T06:26:30Z

As for DCA, it looks like we don't have sources set up with diffs, so we can't use DCA to test performance on diffs yet. A normal DCA run would then test before-and-after-this-PR without diff-informed mode, but those two runs should have the same behaviour. So I say we leave DCA testing out for now until I get the diff-informed sources in place.

Yes, I just wanted to make sure that the introduction of the override didn't somehow affect performance negatively. I don't know, if introducing this override might for some obscure reason e.g. change some join-ordering that could affect performance.

michaelnebel

LGTM!

github-actions bot added the C# label Jun 3, 2025

d10c marked this pull request as ready for review June 4, 2025 11:32

Copilot AI review requested due to automatic review settings June 4, 2025 11:32

d10c requested a review from a team as a code owner June 4, 2025 11:32

Copilot AI reviewed Jun 4, 2025

View reviewed changes

csharp/ql/src/Security Features/CWE-114/AssemblyPathInjection.ql Show resolved Hide resolved

github-actions bot added the documentation label Jun 4, 2025

d10c marked this pull request as draft June 5, 2025 15:59

d10c marked this pull request as ready for review June 10, 2025 15:11

michaelnebel reviewed Jun 11, 2025

View reviewed changes

d10c added the no-change-note-required This PR does not need a change note label Jun 11, 2025

C#: mass enable diff-informed data flow

f2085c2

An auto-generated patch that enables diff-informed data flow in the obvious cases. Builds on github#18344 and github/codeql-patch#88

d10c force-pushed the d10c/csharp/diff-informed branch from 9c48c73 to f2085c2 Compare June 11, 2025 16:58

github-actions bot removed the documentation label Jun 11, 2025

michaelnebel approved these changes Jun 12, 2025

View reviewed changes

C#: mass enable diff-informed data flow #19661

Are you sure you want to change the base?

C#: mass enable diff-informed data flow #19661

Uh oh!

Conversation

d10c commented Jun 3, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Uh oh!

d10c commented Jun 5, 2025

Uh oh!

michaelnebel commented Jun 6, 2025

Uh oh!

d10c commented Jun 10, 2025

Uh oh!

d10c commented Jun 10, 2025

Uh oh!

michaelnebel left a comment

Choose a reason for hiding this comment

Uh oh!

michaelnebel Jun 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

d10c Jun 11, 2025

Choose a reason for hiding this comment

Uh oh!

michaelnebel Jun 11, 2025

Choose a reason for hiding this comment

Uh oh!

michaelnebel Jun 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

d10c Jun 11, 2025

Choose a reason for hiding this comment

Uh oh!

michaelnebel Jun 11, 2025

Choose a reason for hiding this comment

Uh oh!

michaelnebel Jun 11, 2025

Choose a reason for hiding this comment

Uh oh!

michaelnebel commented Jun 11, 2025

Uh oh!

d10c commented Jun 11, 2025

Uh oh!

michaelnebel commented Jun 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

michaelnebel left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

michaelnebel Jun 11, 2025 •

edited

Loading

michaelnebel Jun 11, 2025 •

edited

Loading

michaelnebel commented Jun 12, 2025 •

edited

Loading