Merge pull request #14350 from asgerf/shared/deduplicate-path-graph

Shared: Add DataFlow::DeduplicatePathGraph
This commit is contained in:
Asger F
2024-12-18 14:04:29 +01:00
committed by GitHub
7 changed files with 469 additions and 80 deletions

View File

@@ -0,0 +1,34 @@
import java.util.function.Function;
class A {
String source() { return ""; }
void sink(String s) { }
String propagateState(String s, String state) {
return "";
}
void foo() {
Function<String, String> lambda = Math.random() > 0.5
? x -> propagateState(x, "A")
: x -> propagateState(x, "B");
String step0 = source();
String step1 = lambda.apply(step0);
String step2 = lambda.apply(step1);
sink(step2);
}
void bar() {
Function<String, String> lambda =
(x -> Math.random() > 0.5 ? propagateState(x, "A") : propagateState(x, "B"));
String step0 = source();
String step1 = lambda.apply(step0);
String step2 = lambda.apply(step1);
sink(step2);
}
}

View File

@@ -0,0 +1,79 @@
nodes
| A.java:14:9:14:9 | x : String | semmle.label | x : String |
| A.java:14:14:14:35 | propagateState(...) : String | semmle.label | propagateState(...) : String |
| A.java:14:29:14:29 | x : String | semmle.label | x : String |
| A.java:15:9:15:9 | x : String | semmle.label | x : String |
| A.java:15:14:15:35 | propagateState(...) : String | semmle.label | propagateState(...) : String |
| A.java:15:29:15:29 | x : String | semmle.label | x : String |
| A.java:17:20:17:27 | source(...) : String | semmle.label | source(...) : String |
| A.java:18:20:18:38 | apply(...) : String | semmle.label | apply(...) : String |
| A.java:18:20:18:38 | apply(...) : String | semmle.label | apply(...) : String |
| A.java:18:33:18:37 | step0 : String | semmle.label | step0 : String |
| A.java:19:20:19:38 | apply(...) : String | semmle.label | apply(...) : String |
| A.java:19:33:19:37 | step1 : String | semmle.label | step1 : String |
| A.java:19:33:19:37 | step1 : String | semmle.label | step1 : String |
| A.java:21:10:21:14 | step2 | semmle.label | step2 |
| A.java:26:8:26:8 | x : String | semmle.label | x : String |
| A.java:26:8:26:8 | x : String | semmle.label | x : String |
| A.java:26:13:26:81 | ...?...:... : String | semmle.label | ...?...:... : String |
| A.java:26:13:26:81 | ...?...:... : String | semmle.label | ...?...:... : String |
| A.java:26:35:26:56 | propagateState(...) : String | semmle.label | propagateState(...) : String |
| A.java:26:50:26:50 | x : String | semmle.label | x : String |
| A.java:26:60:26:81 | propagateState(...) : String | semmle.label | propagateState(...) : String |
| A.java:26:75:26:75 | x : String | semmle.label | x : String |
| A.java:28:20:28:27 | source(...) : String | semmle.label | source(...) : String |
| A.java:29:20:29:38 | apply(...) : String | semmle.label | apply(...) : String |
| A.java:29:20:29:38 | apply(...) : String | semmle.label | apply(...) : String |
| A.java:29:33:29:37 | step0 : String | semmle.label | step0 : String |
| A.java:30:20:30:38 | apply(...) : String | semmle.label | apply(...) : String |
| A.java:30:33:30:37 | step1 : String | semmle.label | step1 : String |
| A.java:30:33:30:37 | step1 : String | semmle.label | step1 : String |
| A.java:32:10:32:14 | step2 | semmle.label | step2 |
edges
| A.java:14:9:14:9 | x : String | A.java:14:29:14:29 | x : String | provenance | |
| A.java:14:29:14:29 | x : String | A.java:14:14:14:35 | propagateState(...) : String | provenance | Config |
| A.java:15:9:15:9 | x : String | A.java:15:29:15:29 | x : String | provenance | |
| A.java:15:29:15:29 | x : String | A.java:15:14:15:35 | propagateState(...) : String | provenance | Config |
| A.java:17:20:17:27 | source(...) : String | A.java:18:33:18:37 | step0 : String | provenance | |
| A.java:18:20:18:38 | apply(...) : String | A.java:19:33:19:37 | step1 : String | provenance | |
| A.java:18:20:18:38 | apply(...) : String | A.java:19:33:19:37 | step1 : String | provenance | |
| A.java:18:33:18:37 | step0 : String | A.java:14:9:14:9 | x : String | provenance | |
| A.java:18:33:18:37 | step0 : String | A.java:15:9:15:9 | x : String | provenance | |
| A.java:18:33:18:37 | step0 : String | A.java:18:20:18:38 | apply(...) : String | provenance | Config |
| A.java:18:33:18:37 | step0 : String | A.java:18:20:18:38 | apply(...) : String | provenance | Config |
| A.java:19:20:19:38 | apply(...) : String | A.java:21:10:21:14 | step2 | provenance | |
| A.java:19:33:19:37 | step1 : String | A.java:14:9:14:9 | x : String | provenance | |
| A.java:19:33:19:37 | step1 : String | A.java:15:9:15:9 | x : String | provenance | |
| A.java:19:33:19:37 | step1 : String | A.java:19:20:19:38 | apply(...) : String | provenance | Config |
| A.java:19:33:19:37 | step1 : String | A.java:19:20:19:38 | apply(...) : String | provenance | Config |
| A.java:26:8:26:8 | x : String | A.java:26:50:26:50 | x : String | provenance | |
| A.java:26:8:26:8 | x : String | A.java:26:75:26:75 | x : String | provenance | |
| A.java:26:35:26:56 | propagateState(...) : String | A.java:26:13:26:81 | ...?...:... : String | provenance | |
| A.java:26:50:26:50 | x : String | A.java:26:35:26:56 | propagateState(...) : String | provenance | Config |
| A.java:26:60:26:81 | propagateState(...) : String | A.java:26:13:26:81 | ...?...:... : String | provenance | |
| A.java:26:75:26:75 | x : String | A.java:26:60:26:81 | propagateState(...) : String | provenance | Config |
| A.java:28:20:28:27 | source(...) : String | A.java:29:33:29:37 | step0 : String | provenance | |
| A.java:29:20:29:38 | apply(...) : String | A.java:30:33:30:37 | step1 : String | provenance | |
| A.java:29:20:29:38 | apply(...) : String | A.java:30:33:30:37 | step1 : String | provenance | |
| A.java:29:33:29:37 | step0 : String | A.java:26:8:26:8 | x : String | provenance | |
| A.java:29:33:29:37 | step0 : String | A.java:26:8:26:8 | x : String | provenance | |
| A.java:29:33:29:37 | step0 : String | A.java:29:20:29:38 | apply(...) : String | provenance | Config |
| A.java:29:33:29:37 | step0 : String | A.java:29:20:29:38 | apply(...) : String | provenance | Config |
| A.java:30:20:30:38 | apply(...) : String | A.java:32:10:32:14 | step2 | provenance | |
| A.java:30:33:30:37 | step1 : String | A.java:26:8:26:8 | x : String | provenance | |
| A.java:30:33:30:37 | step1 : String | A.java:26:8:26:8 | x : String | provenance | |
| A.java:30:33:30:37 | step1 : String | A.java:30:20:30:38 | apply(...) : String | provenance | Config |
| A.java:30:33:30:37 | step1 : String | A.java:30:20:30:38 | apply(...) : String | provenance | Config |
subpaths
| A.java:18:33:18:37 | step0 : String | A.java:14:9:14:9 | x : String | A.java:14:14:14:35 | propagateState(...) : String | A.java:18:20:18:38 | apply(...) : String |
| A.java:18:33:18:37 | step0 : String | A.java:15:9:15:9 | x : String | A.java:15:14:15:35 | propagateState(...) : String | A.java:18:20:18:38 | apply(...) : String |
| A.java:19:33:19:37 | step1 : String | A.java:14:9:14:9 | x : String | A.java:14:14:14:35 | propagateState(...) : String | A.java:19:20:19:38 | apply(...) : String |
| A.java:19:33:19:37 | step1 : String | A.java:15:9:15:9 | x : String | A.java:15:14:15:35 | propagateState(...) : String | A.java:19:20:19:38 | apply(...) : String |
| A.java:29:33:29:37 | step0 : String | A.java:26:8:26:8 | x : String | A.java:26:13:26:81 | ...?...:... : String | A.java:29:20:29:38 | apply(...) : String |
| A.java:29:33:29:37 | step0 : String | A.java:26:8:26:8 | x : String | A.java:26:13:26:81 | ...?...:... : String | A.java:29:20:29:38 | apply(...) : String |
| A.java:30:33:30:37 | step1 : String | A.java:26:8:26:8 | x : String | A.java:26:13:26:81 | ...?...:... : String | A.java:30:20:30:38 | apply(...) : String |
| A.java:30:33:30:37 | step1 : String | A.java:26:8:26:8 | x : String | A.java:26:13:26:81 | ...?...:... : String | A.java:30:20:30:38 | apply(...) : String |
spuriousFlow
#select
| A.java:17:20:17:27 | source(...) : String | A.java:17:20:17:27 | source(...) : String | A.java:21:10:21:14 | step2 | Flow |
| A.java:28:20:28:27 | source(...) : String | A.java:28:20:28:27 | source(...) : String | A.java:32:10:32:14 | step2 | Flow |

View File

@@ -0,0 +1,84 @@
/**
* @kind path-problem
*/
import java
import semmle.code.java.dataflow.DataFlow
import DataFlow
MethodCall propagateCall(string state) {
result.getMethod().getName() = "propagateState" and
state = result.getArgument(1).(StringLiteral).getValue()
}
module TestConfig implements StateConfigSig {
class FlowState = string;
predicate isSource(Node n, FlowState state) {
n.asExpr().(MethodCall).getMethod().getName() = "source" and state = ["A", "B"]
}
predicate isSink(Node n, FlowState state) {
n.asExpr() = any(MethodCall acc | acc.getMethod().getName() = "sink").getAnArgument() and
state = ["A", "B"]
}
predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {
exists(MethodCall call |
call = propagateCall(state1) and
state2 = state1 and
node1.asExpr() = call.getArgument(0) and
node2.asExpr() = call
)
}
}
module TestFlow = DataFlow::GlobalWithState<TestConfig>;
module Graph = DataFlow::DeduplicatePathGraph<TestFlow::PathNode, TestFlow::PathGraph>;
/**
* Holds if `node` is reachable from a call to `propagateState` with the given `state` argument.
* `call` indicates if a call step was taken (i.e. into a subpath).
*
* We use this to check if one `propagateState` can flow out of another, which is not allowed.
*/
predicate reachableFromPropagate(Graph::PathNode node, string state, boolean call) {
node.getNode().asExpr() = propagateCall(state) and call = false
or
exists(Graph::PathNode prev | reachableFromPropagate(prev, state, call) |
Graph::edges(prev, node, _, _) and
not Graph::subpaths(prev, node, _, _) // argument-passing edges are handled separately
or
Graph::subpaths(prev, _, _, node) // arg -> out (should be included in 'edges' but handle the case here for clarity)
)
or
exists(Graph::PathNode prev |
reachableFromPropagate(prev, state, _) and
Graph::subpaths(prev, node, _, _) and // arg -> parameter
call = true
)
or
exists(Graph::PathNode prev |
reachableFromPropagate(prev, state, call) and
Graph::subpaths(_, _, prev, node) and // return -> out
call = false
)
}
/**
* Holds if `node` is the return value of a `propagateState` call that appears to be reachable
* with a different state than the one propagated by the call, indicating spurious flow resulting from
* merging path nodes.
*/
query predicate spuriousFlow(Graph::PathNode node, string state1, string state2) {
reachableFromPropagate(node, state1, _) and
node.getNode().asExpr() = propagateCall(state2) and
state1 != state2
}
import Graph::PathGraph
from Graph::PathNode source, Graph::PathNode sink
where TestFlow::flowPath(source.getAnOriginalPathNode(), sink.getAnOriginalPathNode())
select source, source, sink, "Flow"

View File

@@ -16,20 +16,9 @@
private import codeql.ruby.AST
private import codeql.ruby.security.CodeInjectionQuery
import CodeInjectionFlow::PathGraph
import DataFlow::DeduplicatePathGraph<CodeInjectionFlow::PathNode, CodeInjectionFlow::PathGraph>
from CodeInjectionFlow::PathNode source, CodeInjectionFlow::PathNode sink, Source sourceNode
where
CodeInjectionFlow::flowPath(source, sink) and
sourceNode = source.getNode() and
// removing duplications of the same path, but different flow-labels.
sink =
min(CodeInjectionFlow::PathNode otherSink |
CodeInjectionFlow::flowPath(any(CodeInjectionFlow::PathNode s | s.getNode() = sourceNode),
otherSink) and
otherSink.getNode() = sink.getNode()
|
otherSink order by otherSink.getState().getStringRepresentation()
)
select sink.getNode(), source, sink, "This code execution depends on a $@.", sourceNode,
from PathNode source, PathNode sink
where CodeInjectionFlow::flowPath(source.getAnOriginalPathNode(), sink.getAnOriginalPathNode())
select sink.getNode(), source, sink, "This code execution depends on a $@.", source.getNode(),
"user-provided value"

View File

@@ -1,74 +1,18 @@
edges
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:8:10:8:13 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:8:10:8:13 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:20:20:20:23 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:20:20:20:23 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:23:21:23:24 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:23:21:23:24 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:29:15:29:18 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:32:19:32:22 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:38:24:38:27 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:38:24:38:27 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:41:40:41:43 | code | provenance | |
| CodeInjection.rb:5:12:5:17 | call to params | CodeInjection.rb:5:12:5:24 | ...[...] | provenance | |
| CodeInjection.rb:5:12:5:17 | call to params | CodeInjection.rb:5:12:5:24 | ...[...] | provenance | |
| CodeInjection.rb:5:12:5:24 | ...[...] | CodeInjection.rb:5:5:5:8 | code | provenance | |
| CodeInjection.rb:5:12:5:24 | ...[...] | CodeInjection.rb:5:5:5:8 | code | provenance | |
| CodeInjection.rb:38:24:38:27 | code | CodeInjection.rb:38:10:38:28 | call to escape | provenance | MaD:21 |
| CodeInjection.rb:38:24:38:27 | code | CodeInjection.rb:38:10:38:28 | call to escape | provenance | MaD:21 |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:80:16:80:19 | code | provenance | |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:86:10:86:37 | ... + ... | provenance | |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:86:22:86:25 | code | provenance | |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:88:10:88:32 | "prefix_#{...}_suffix" | provenance | AdditionalTaintStep |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:90:10:90:13 | code | provenance | |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:90:10:90:13 | code | provenance | |
| CodeInjection.rb:78:12:78:17 | call to params | CodeInjection.rb:78:12:78:24 | ...[...] | provenance | |
| CodeInjection.rb:78:12:78:17 | call to params | CodeInjection.rb:78:12:78:24 | ...[...] | provenance | |
| CodeInjection.rb:78:12:78:24 | ...[...] | CodeInjection.rb:78:5:78:8 | code | provenance | |
| CodeInjection.rb:78:12:78:24 | ...[...] | CodeInjection.rb:78:5:78:8 | code | provenance | |
| CodeInjection.rb:86:10:86:25 | ... + ... [element] | CodeInjection.rb:86:10:86:37 | ... + ... | provenance | |
| CodeInjection.rb:86:22:86:25 | code | CodeInjection.rb:86:10:86:25 | ... + ... [element] | provenance | |
| CodeInjection.rb:101:3:102:5 | self in index [@foo] | CodeInjection.rb:111:3:113:5 | self in baz [@foo] | provenance | |
| CodeInjection.rb:101:3:102:5 | self in index [@foo] | CodeInjection.rb:111:3:113:5 | self in baz [@foo] | provenance | |
| CodeInjection.rb:105:5:105:8 | [post] self [@foo] | CodeInjection.rb:108:3:109:5 | self in bar [@foo] | provenance | |
| CodeInjection.rb:105:5:105:8 | [post] self [@foo] | CodeInjection.rb:108:3:109:5 | self in bar [@foo] | provenance | |
| CodeInjection.rb:105:12:105:17 | call to params | CodeInjection.rb:105:12:105:23 | ...[...] | provenance | |
| CodeInjection.rb:105:12:105:17 | call to params | CodeInjection.rb:105:12:105:23 | ...[...] | provenance | |
| CodeInjection.rb:105:12:105:23 | ...[...] | CodeInjection.rb:105:5:105:8 | [post] self [@foo] | provenance | |
| CodeInjection.rb:105:12:105:23 | ...[...] | CodeInjection.rb:105:5:105:8 | [post] self [@foo] | provenance | |
| CodeInjection.rb:108:3:109:5 | self in bar [@foo] | CodeInjection.rb:101:3:102:5 | self in index [@foo] | provenance | |
| CodeInjection.rb:108:3:109:5 | self in bar [@foo] | CodeInjection.rb:101:3:102:5 | self in index [@foo] | provenance | |
| CodeInjection.rb:111:3:113:5 | self in baz [@foo] | CodeInjection.rb:112:10:112:13 | self [@foo] | provenance | |
| CodeInjection.rb:111:3:113:5 | self in baz [@foo] | CodeInjection.rb:112:10:112:13 | self [@foo] | provenance | |
| CodeInjection.rb:112:10:112:13 | self [@foo] | CodeInjection.rb:112:10:112:13 | @foo | provenance | |
| CodeInjection.rb:112:10:112:13 | self [@foo] | CodeInjection.rb:112:10:112:13 | @foo | provenance | |
nodes
| CodeInjection.rb:5:5:5:8 | code | semmle.label | code |
| CodeInjection.rb:5:5:5:8 | code | semmle.label | code |
| CodeInjection.rb:5:12:5:17 | call to params | semmle.label | call to params |
| CodeInjection.rb:5:12:5:17 | call to params | semmle.label | call to params |
| CodeInjection.rb:5:12:5:24 | ...[...] | semmle.label | ...[...] |
| CodeInjection.rb:5:12:5:24 | ...[...] | semmle.label | ...[...] |
| CodeInjection.rb:8:10:8:13 | code | semmle.label | code |
| CodeInjection.rb:8:10:8:13 | code | semmle.label | code |
| CodeInjection.rb:11:10:11:15 | call to params | semmle.label | call to params |
| CodeInjection.rb:11:10:11:15 | call to params | semmle.label | call to params |
| CodeInjection.rb:20:20:20:23 | code | semmle.label | code |
| CodeInjection.rb:20:20:20:23 | code | semmle.label | code |
| CodeInjection.rb:23:21:23:24 | code | semmle.label | code |
| CodeInjection.rb:23:21:23:24 | code | semmle.label | code |
| CodeInjection.rb:29:15:29:18 | code | semmle.label | code |
| CodeInjection.rb:32:19:32:22 | code | semmle.label | code |
| CodeInjection.rb:38:10:38:28 | call to escape | semmle.label | call to escape |
| CodeInjection.rb:38:10:38:28 | call to escape | semmle.label | call to escape |
| CodeInjection.rb:38:24:38:27 | code | semmle.label | code |
| CodeInjection.rb:38:24:38:27 | code | semmle.label | code |
| CodeInjection.rb:41:40:41:43 | code | semmle.label | code |
| CodeInjection.rb:78:5:78:8 | code | semmle.label | code |
| CodeInjection.rb:78:5:78:8 | code | semmle.label | code |
| CodeInjection.rb:78:12:78:17 | call to params | semmle.label | call to params |
| CodeInjection.rb:78:12:78:17 | call to params | semmle.label | call to params |
| CodeInjection.rb:78:12:78:24 | ...[...] | semmle.label | ...[...] |
| CodeInjection.rb:78:12:78:24 | ...[...] | semmle.label | ...[...] |
| CodeInjection.rb:80:16:80:19 | code | semmle.label | code |
| CodeInjection.rb:86:10:86:25 | ... + ... [element] | semmle.label | ... + ... [element] |
@@ -76,23 +20,41 @@ nodes
| CodeInjection.rb:86:22:86:25 | code | semmle.label | code |
| CodeInjection.rb:88:10:88:32 | "prefix_#{...}_suffix" | semmle.label | "prefix_#{...}_suffix" |
| CodeInjection.rb:90:10:90:13 | code | semmle.label | code |
| CodeInjection.rb:90:10:90:13 | code | semmle.label | code |
| CodeInjection.rb:101:3:102:5 | self in index [@foo] | semmle.label | self in index [@foo] |
| CodeInjection.rb:101:3:102:5 | self in index [@foo] | semmle.label | self in index [@foo] |
| CodeInjection.rb:105:5:105:8 | [post] self [@foo] | semmle.label | [post] self [@foo] |
| CodeInjection.rb:105:5:105:8 | [post] self [@foo] | semmle.label | [post] self [@foo] |
| CodeInjection.rb:105:12:105:17 | call to params | semmle.label | call to params |
| CodeInjection.rb:105:12:105:17 | call to params | semmle.label | call to params |
| CodeInjection.rb:105:12:105:23 | ...[...] | semmle.label | ...[...] |
| CodeInjection.rb:105:12:105:23 | ...[...] | semmle.label | ...[...] |
| CodeInjection.rb:108:3:109:5 | self in bar [@foo] | semmle.label | self in bar [@foo] |
| CodeInjection.rb:108:3:109:5 | self in bar [@foo] | semmle.label | self in bar [@foo] |
| CodeInjection.rb:111:3:113:5 | self in baz [@foo] | semmle.label | self in baz [@foo] |
| CodeInjection.rb:111:3:113:5 | self in baz [@foo] | semmle.label | self in baz [@foo] |
| CodeInjection.rb:112:10:112:13 | @foo | semmle.label | @foo |
| CodeInjection.rb:112:10:112:13 | @foo | semmle.label | @foo |
| CodeInjection.rb:112:10:112:13 | self [@foo] | semmle.label | self [@foo] |
| CodeInjection.rb:112:10:112:13 | self [@foo] | semmle.label | self [@foo] |
edges
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:8:10:8:13 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:20:20:20:23 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:23:21:23:24 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:29:15:29:18 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:32:19:32:22 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:38:24:38:27 | code | provenance | |
| CodeInjection.rb:5:5:5:8 | code | CodeInjection.rb:41:40:41:43 | code | provenance | |
| CodeInjection.rb:5:12:5:17 | call to params | CodeInjection.rb:5:12:5:24 | ...[...] | provenance | |
| CodeInjection.rb:5:12:5:24 | ...[...] | CodeInjection.rb:5:5:5:8 | code | provenance | |
| CodeInjection.rb:38:24:38:27 | code | CodeInjection.rb:38:10:38:28 | call to escape | provenance | MaD:21 |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:80:16:80:19 | code | provenance | |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:86:10:86:37 | ... + ... | provenance | |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:86:22:86:25 | code | provenance | |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:88:10:88:32 | "prefix_#{...}_suffix" | provenance | AdditionalTaintStep |
| CodeInjection.rb:78:5:78:8 | code | CodeInjection.rb:90:10:90:13 | code | provenance | |
| CodeInjection.rb:78:12:78:17 | call to params | CodeInjection.rb:78:12:78:24 | ...[...] | provenance | |
| CodeInjection.rb:78:12:78:24 | ...[...] | CodeInjection.rb:78:5:78:8 | code | provenance | |
| CodeInjection.rb:86:10:86:25 | ... + ... [element] | CodeInjection.rb:86:10:86:37 | ... + ... | provenance | |
| CodeInjection.rb:86:22:86:25 | code | CodeInjection.rb:86:10:86:25 | ... + ... [element] | provenance | |
| CodeInjection.rb:101:3:102:5 | self in index [@foo] | CodeInjection.rb:111:3:113:5 | self in baz [@foo] | provenance | |
| CodeInjection.rb:105:5:105:8 | [post] self [@foo] | CodeInjection.rb:108:3:109:5 | self in bar [@foo] | provenance | |
| CodeInjection.rb:105:12:105:17 | call to params | CodeInjection.rb:105:12:105:23 | ...[...] | provenance | |
| CodeInjection.rb:105:12:105:23 | ...[...] | CodeInjection.rb:105:5:105:8 | [post] self [@foo] | provenance | |
| CodeInjection.rb:108:3:109:5 | self in bar [@foo] | CodeInjection.rb:101:3:102:5 | self in index [@foo] | provenance | |
| CodeInjection.rb:111:3:113:5 | self in baz [@foo] | CodeInjection.rb:112:10:112:13 | self [@foo] | provenance | |
| CodeInjection.rb:112:10:112:13 | self [@foo] | CodeInjection.rb:112:10:112:13 | @foo | provenance | |
subpaths
#select
| CodeInjection.rb:8:10:8:13 | code | CodeInjection.rb:5:12:5:17 | call to params | CodeInjection.rb:8:10:8:13 | code | This code execution depends on a $@. | CodeInjection.rb:5:12:5:17 | call to params | user-provided value |

View File

@@ -0,0 +1,4 @@
---
category: minorAnalysis
---
* Added a module `DataFlow::DeduplicatePathGraph` that can be used to avoid generating duplicate path explanations in queries that use flow state.

View File

@@ -851,4 +851,241 @@ module DataFlowMake<LocationSig Location, InputSig<Location> Lang> {
}
}
}
/**
* Generates a `PathGraph` in which equivalent path nodes are merged, in order to avoid duplicate paths.
*/
module DeduplicatePathGraph<PathNodeSig InputPathNode, PathGraphSig<InputPathNode> Graph> {
// NOTE: there is a known limitation in that this module cannot see which nodes are sources or sinks.
// This only matters in the rare case where a sink PathNode has a non-empty set of succesors, and there is a
// non-sink PathNode with the same `(node, toString)` value and the same successors, but is transitively
// reachable from a different set of PathNodes. (And conversely for sources).
//
pragma[nomagic]
private InputPathNode getAPathNode(Node node, string toString) {
result.getNode() = node and
Graph::nodes(result, _, toString)
}
private signature predicate collapseCandidateSig(Node node, string toString);
private signature predicate stepSig(
InputPathNode node1, InputPathNode node2, string key, string val
);
private signature predicate subpathStepSig(
InputPathNode arg, InputPathNode param, InputPathNode ret, InputPathNode out
);
/**
* Performs a forward or backward pass computing which `(node, toString)` pairs can subsume their corresponding
* path nodes.
*
* This is similar to automaton minimization, but for an NFA. Since minimizing an NFA is NP-hard (and does not have
* a unique minimal NFA), we operate with the simpler model: for a given `(node, toString)` pair, either all
* corresponding path nodes are merged, or none are merged.
*
* Comments are written as if this checks for outgoing edges and propagates backward, though the module is also
* used to perform the opposite direction.
*/
private module MakeDiscriminatorPass<
collapseCandidateSig/2 collapseCandidate, stepSig/4 step, subpathStepSig/4 subpathStep>
{
/**
* Gets the number of `(key, val, node, toString)` tuples reachable in one step from `pathNode`.
*
* That is, two edges are counted as one if their target nodes are the same after projection, and the edges have the
* same `(key, val)`.
*/
private int getOutDegreeFromPathNode(InputPathNode pathNode) {
result =
count(Node node, string toString, string key, string val |
step(pathNode, getAPathNode(node, toString), key, val)
)
}
/**
* Gets the number of `(key, val, node2, toString2)` pairs reachable in one step from path nodes corresponding to `(node, toString)`.
*/
private int getOutDegreeFromNode(Node node, string toString) {
result =
strictcount(Node node2, string toString2, string key, string val |
step(getAPathNode(node, toString), getAPathNode(node2, toString2), key, val)
)
}
/**
* Like `getOutDegreeFromPathNode` except counts `subpath` tuples.
*/
private int getSubpathOutDegreeFromPathNode(InputPathNode pathNode) {
result =
count(Node n1, string s1, Node n2, string s2, Node n3, string s3 |
subpathStep(pathNode, getAPathNode(n1, s1), getAPathNode(n2, s2), getAPathNode(n3, s3))
)
}
/**
* Like `getOutDegreeFromNode` except counts `subpath` tuples.
*/
private int getSubpathOutDegreeFromNode(Node node, string toString) {
result =
strictcount(Node n1, string s1, Node n2, string s2, Node n3, string s3 |
subpathStep(getAPathNode(node, toString), getAPathNode(n1, s1), getAPathNode(n2, s2),
getAPathNode(n3, s3))
)
}
/** Gets a successor of `node`, including subpath flow-through, but not enter or exit subpath steps. */
InputPathNode stepEx(InputPathNode node) {
step(node, result, _, _) and
not result = enterSubpathStep(node) and
not result = exitSubpathStep(node)
or
// Assuming the input is pruned properly, all subpaths have flow-through.
// This step should be in 'step' as well, but include it here for clarity as we rely on it.
subpathStep(node, _, _, result)
}
InputPathNode enterSubpathStep(InputPathNode node) { subpathStep(node, result, _, _) }
InputPathNode exitSubpathStep(InputPathNode node) { subpathStep(_, _, node, result) }
/** Holds if `(node, toString)` cannot be collapsed (but was a candidate for being collapsed). */
predicate discriminatedPair(Node node, string toString, boolean hasEnter) {
collapseCandidate(node, toString) and
hasEnter = false and
(
// Check if all corresponding PathNodes have the same successor sets when projected to `(node, toString)`.
// To do this, we check that each successor set has the same size as the union of the succesor sets.
// - If the successor sets are equal, then they are also equal to their union, and so have the correct size.
// - Conversely, if two successor sets are not equal, one of them must be missing an element that is present
// in the union, but must still be a subset of the union, and thus be strictly smaller than the union.
getOutDegreeFromPathNode(getAPathNode(node, toString)) <
getOutDegreeFromNode(node, toString)
or
// Same as above but counting associated subpath triples instead
getSubpathOutDegreeFromPathNode(getAPathNode(node, toString)) <
getSubpathOutDegreeFromNode(node, toString)
)
or
collapseCandidate(node, toString) and
(
// Retain flow state if one of the successors requires it to be retained
discriminatedPathNode(stepEx(getAPathNode(node, toString)), hasEnter)
or
// Propagate backwards from parameter to argument
discriminatedPathNode(enterSubpathStep(getAPathNode(node, toString)), false) and
hasEnter = false
or
// Propagate backwards from out to return
discriminatedPathNode(exitSubpathStep(getAPathNode(node, toString)), _) and
hasEnter = true
)
}
/** Holds if `pathNode` cannot be collapsed. */
private predicate discriminatedPathNode(InputPathNode pathNode, boolean hasEnter) {
exists(Node node, string toString |
discriminatedPair(node, toString, hasEnter) and
getAPathNode(node, toString) = pathNode
)
}
/** Holds if `(node, toString)` cannot be collapsed (but was a candidate for being collapsed). */
predicate discriminatedPair(Node node, string toString) {
discriminatedPair(node, toString, _)
}
/** Holds if `pathNode` cannot be collapsed. */
predicate discriminatedPathNode(InputPathNode pathNode) { discriminatedPathNode(pathNode, _) }
}
private InputPathNode getUniqPathNode(Node node, string toString) {
result = unique(InputPathNode pathNode | pathNode = getAPathNode(node, toString))
}
private predicate initialCandidate(Node node, string toString) {
exists(getAPathNode(node, toString)) and not exists(getUniqPathNode(node, toString))
}
private module Pass1 =
MakeDiscriminatorPass<initialCandidate/2, Graph::edges/4, Graph::subpaths/4>;
private predicate edgesRev(InputPathNode node1, InputPathNode node2, string key, string val) {
Graph::edges(node2, node1, key, val)
}
private predicate subpathsRev(
InputPathNode n1, InputPathNode n2, InputPathNode n3, InputPathNode n4
) {
Graph::subpaths(n4, n3, n2, n1)
}
private module Pass2 =
MakeDiscriminatorPass<Pass1::discriminatedPair/2, edgesRev/4, subpathsRev/4>;
private newtype TPathNode =
TPreservedPathNode(InputPathNode node) {
Pass2::discriminatedPathNode(node) or node = getUniqPathNode(_, _)
} or
TCollapsedPathNode(Node node, string toString) {
initialCandidate(node, toString) and
not Pass2::discriminatedPair(node, toString)
}
/** A node in the path graph after equivalent nodes have been collapsed. */
class PathNode extends TPathNode {
private Node asCollapsedNode() { this = TCollapsedPathNode(result, _) }
private InputPathNode asPreservedNode() { this = TPreservedPathNode(result) }
/** Gets a correspondng node in the original graph. */
InputPathNode getAnOriginalPathNode() {
exists(Node node, string toString |
this = TCollapsedPathNode(node, toString) and
result = getAPathNode(node, toString)
)
or
result = this.asPreservedNode()
}
/** Gets a string representation of this node. */
string toString() {
result = this.asPreservedNode().toString() or this = TCollapsedPathNode(_, result)
}
/** Gets the location of this node. */
Location getLocation() { result = this.getAnOriginalPathNode().getLocation() }
/** Gets the corresponding data-flow node. */
Node getNode() {
result = this.asCollapsedNode()
or
result = this.asPreservedNode().getNode()
}
}
/**
* Provides the query predicates needed to include a graph in a path-problem query.
*/
module PathGraph implements PathGraphSig<PathNode> {
query predicate nodes(PathNode node, string key, string val) {
Graph::nodes(node.getAnOriginalPathNode(), key, val)
}
query predicate edges(PathNode node1, PathNode node2, string key, string val) {
Graph::edges(node1.getAnOriginalPathNode(), node2.getAnOriginalPathNode(), key, val)
}
query predicate subpaths(PathNode arg, PathNode par, PathNode ret, PathNode out) {
// Note: this may look suspiciously simple, but it's not an oversight. Even if the caller needs to retain state,
// it is entirely possible to step through a subpath in which state has been projected away.
Graph::subpaths(arg.getAnOriginalPathNode(), par.getAnOriginalPathNode(),
ret.getAnOriginalPathNode(), out.getAnOriginalPathNode())
}
}
// Re-export the PathGraph so the user can import a single module and get both PathNode and the query predicates
import PathGraph
}
}