Java: Add flow step from startActivity to getIntent

[atorralba]

In exported activities, getIntent() is considered a remote source. Nonetheless, sometimes user input is added to intents (e.g. as extras) that are sent to non-exported activities, tainting the getIntent() calls of those as well.

This PR adds a value-preserving step to support that case.

[aschackmull]

Looks reasonable to me. If the join-order in StartActivityIntentStep::step produces something reasonable, then I'd say this looks good to merge.

[atorralba]

There isn't any bad join order metric being triggered in DCA.

By looking at some query logs manually (I had to change StartActivityIntentStep to extend Unit, otherwise I couldn't find it in the query log, I guess because it got shadowed by AdditionalValueStep tuple counts), I got the following:

[2022-05-10 11:24:11] (0s) Tuple counts for Intent::StartActivityIntentStep::step#dispred#f0820431#cpe#23#ff/2@fea16adh after 1ms:
                      1    ~0%     {0} r1 = SELECT DataFlowPrivate::TNormalReturnKinddom#a63c994e# ON 0 >= 0
                      5730 ~0%     {2} r2 = JOIN r1 WITH Expr::MethodAccess::getArgument#dispred#f0820431#fbf#cpe#13 CARTESIAN PRODUCT OUTPUT Rhs.0, Rhs.1
                      5730 ~0%     {2} r3 = JOIN r2 WITH Expr::MethodAccess::getMethod#dispred#f0820431#ff ON FIRST 1 OUTPUT Rhs.1, Lhs.1
                      
                      1    ~0%     {1} r4 = JOIN r3 WITH Intent::ContextStartActivityMethod#class#f5dac5da#f ON FIRST 1 OUTPUT Lhs.1
                      
                      533  ~0%     {2} r5 = JOIN r3 WITH #Member::Method::overrides#dispred#f0820431Plus#bf ON FIRST 1 OUTPUT Rhs.1, Lhs.1
                      4    ~0%     {1} r6 = JOIN r5 WITH Intent::ContextStartActivityMethod#class#f5dac5da#f ON FIRST 1 OUTPUT Lhs.1
                      
                      5    ~0%     {1} r7 = r4 UNION r6
                      5    ~0%     {2} r8 = JOIN r7 WITH DataFlowNodes::TExprNode#b728817f#ff ON FIRST 1 OUTPUT Lhs.0, Rhs.1 'n1'
                      5    ~0%     {2} r9 = JOIN r8 WITH DataFlowNodes::TExprNode#b728817f#ff ON FIRST 1 OUTPUT Rhs.1 'n1', Lhs.1 'n1'
                      
                      4    ~0%     {2} r10 = JOIN r9 WITH #DataFlowUtil::localFlowStep#0516ed17Plus#bf_10#join_rhs ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'n1'
                      
                      9    ~0%     {2} r11 = r9 UNION r10
                      9    ~0%     {2} r12 = JOIN r11 WITH DataFlowNodes::TExprNode#b728817f#ff_10#join_rhs ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'n1'
                      4    ~0%     {3} r13 = JOIN r12 WITH Expr::ConstructorCall::getConstructedType#dispred#f0820431#ff ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'n1', Lhs.0
                      4    ~0%     {3} r14 = JOIN r13 WITH Intent::TypeIntent#class#f5dac5da#b ON FIRST 1 OUTPUT Lhs.2, 1, Lhs.1 'n1'
                      4    ~0%     {2} r15 = JOIN r14 WITH Expr::ClassInstanceExpr::getArgument#dispred#f0820431#fff ON FIRST 2 OUTPUT Rhs.2, Lhs.2 'n1'
                      4    ~0%     {2} r16 = JOIN r15 WITH Expr::Expr::getType#dispred#f0820431#bf ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'n1'
                      1    ~0%     {2} r17 = JOIN r16 WITH Generics::ParameterizedType::getATypeArgument#dispred#f0820431#bf ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'n1'
                      72   ~0%     {2} r18 = JOIN r17 WITH Expr::MethodAccess::getReceiverType#dispred#f0820431#ff_10#join_rhs ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'n1'
                      72   ~0%     {3} r19 = JOIN r18 WITH Expr::MethodAccess::getMethod#dispred#f0820431#ff ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'n1', Lhs.0
                      
                      1    ~0%     {2} r20 = JOIN r19 WITH Intent::AndroidGetIntentMethod#class#f5dac5da#f ON FIRST 1 OUTPUT Lhs.2, Lhs.1 'n1'
                      
                      21   ~0%     {3} r21 = JOIN r19 WITH #Member::Method::overrides#dispred#f0820431Plus#bf ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'n1', Lhs.2
                      0    ~0%     {2} r22 = JOIN r21 WITH Intent::AndroidGetIntentMethod#class#f5dac5da#f ON FIRST 1 OUTPUT Lhs.2, Lhs.1 'n1'
                      
                      1    ~0%     {2} r23 = r20 UNION r22
                      1    ~0%     {2} r24 = JOIN r23 WITH DataFlowNodes::TExprNode#b728817f#ff ON FIRST 1 OUTPUT Lhs.1 'n1', Rhs.1 'n1'
                                   return r24

Nothing unreasonable as far as I can see. The cartesian product with MethodAccess::getArgument seems a bit suspicious, but I'm not sure we can avoid that. Do you think it could be a problem?

[aschackmull]

The cartesian product with MethodAccess::getArgument seems a bit suspicious, but I'm not sure we can avoid that. Do you think it could be a problem?

The left-hand side of that cartesian product is a size 1 constant, so that's completely harmless.